UPPSALA UNIVERSITY

(1)

UPPSALA UNIVERSITY

MASTER DEGREE PROJECT

______________________________________________________________

Designing strategies to induce miRNA strand selection

______________________________________________________________

April 2018

Author:

Cecilia Ålander

Supervisor:

Oommen Varghese

(2)

2

(3)

3

Abstract

MiRNAs are small double-stranded non-coding RNAs involved in down- regulation of genes post-transcriptionally. The miRNA duplex binds to a complex of proteins known collectively as the RISC complex where strand selection is carried out, leading to one of the miRNA strands - 3p or 5p, being fully integrated into the RISC complex, while the other strand is discarded. The strand chosen by RISC acts as a template to find mRNA targets, which will be silenced, leading to altered gene expression. Some miRNAs have shown to possess active silencing function in both strands. A miRNA with this proposed dual-strand activity is miR-34a, a miRNA involved in various important regulative cellular processes such as cell cycle arrest and apoptosis. This involvement, which is largely due to active silencing by miR-34a-5p, often leads to down-regulation of miR-34a in many cancer types. There have been attempts to shape this miRNA into a drug for treatment of cancer, however the issue is the number of potential unknown genes targeted by miR-34a-3p. If miR-34a is to be used as a drug the ability to induce selection of the 5p strand is necessary to reduce unwanted target effects.

Here we test the effects two modifications; 5'-O-methylation and nucleotide addition, on miR-34a may have on strand selection and gene silencing. The modifications proved able to induce skewed intracellular levels of 3p and 5p strands, strongly promoting RISC binding of one strand over the other. Selective silencing could also be seen in down-stream methods such as lowered protein concentration and induced cell apoptosis, indicating that both modifications are inducing strand selection.

(4)

4

List of Tables

2.1. The strand sequences of miR-34a. . . 7.1. PCR programs . . .

11 33

(6)

6

List of Figures

1.1. MicroRNA processing. . . 1.2. Structure of pre-miRNA. . . 1.3. Modifications of miR-34a. . . 2.1. miR-34a variants. . . 2.2. 3'UTR of the Renilla gene. . . 3.1. miRNA levels of (N)7-tail miR-34as. . . 3.2. miRNA levels of 5'-O-methylated miR-34as. . . 3.3. Luciferase assays of (N)7-tail miR-34as. . . 3.4. Luciferase assays of 5'-O-methylated miR-34as. . . 3.5. SIRT1 mRNA levels. . . 3.6. Western blot assay on SIRT1 and GAPDH. . . 3.7. Western blot assay on Axin2 and GAPDH. . . 3.8. Flow cytometry histograms. . . 3.9. Flow cytometry plot. . . 7.1. Additional flow cytometry histograms and dot plots. . .

8 9 10 11 13 17 18 19 19 20 21 22 23 24 33

(7)

7

1. Introduction

RNA interference (RNAi), first described by Fire and colleagues in 1998 [1], is a gene-silencing process found in eukaryotes where small non-coding RNAs approximately 22 nucleotides in length act in tandem with a complex of proteins to target and inhibit mRNA transcripts in the cytosol of cells [2, 3]. The RNAs active in this mechanism are usually divided into two groups; small-interfering RNAs (siRNAs) and microRNAs (miRNAs), based on their chemical structure [3, 4]. Since their discovery, siRNAs and miRNAs have been used to alter gene expression in various species [5, 6] and have in recent years been researched for potential therapeutic use against various diseases [7, 8].

1.1. MicroRNA biogenesis and the RISC complex

Functionally, miRNAs and siRNAs are very similar; both originate from double stranded RNA that is processed by the ribonuclease Dicer to form an RNA duplex that can be recruited by proteins to form the RNA-induced silencing complex (RISC), that is responsible for mRNA silencing [9]. SiRNAs are typically artificially designed for a specific target and are introduced to living systems exogenously.

MicroRNAs, however are generally directly encoded in genes [10] that, once transcribed into mRNAs, require several processing steps to form the mature miRNA duplex (see Figure 1.1). The transcribed miRNA mRNAs fold to form hairpin structures known as primary miRNAs (pri-miRNAs). These hairpins are shortened by the enzyme Drosha to form pre-miRNAs and are transported from the nucleus to the cytoplasm via the Exportin-5 pathway where it is shortened further by Dicer to form a final double-stranded miRNA with short 3' overhangs [11, 12, 13]

At this stage both siRNAs and miRNAs can be recruited by an Argonaute (Ago) protein, bound to one or several auxiliary proteins [14], to form the RISC complex. When the miRNA duplex binds to Ago only one of the strands is fully incorporated forming a mature RISC complex, while the other strand is discarded and subjected to degradation by nucleases in the cytosol. The strand fully bound to RISC is used as a template to finds mRNA targets, where miRNAs base-pair with the 3'UTR, while siRNA can bind anywhere on the target mRNA.

Target recognition is determined by base-pairing between the miRNA strand in RISC and the mRNA transcript [15]. Once a target is found the RISC complex binds to the mRNA, blocking ribosomal binding and recruits proteins that further induce translational repression [16].

(8)

8

Figure 1.1. MicroRNA processing and RISC-mediated gene silencing. MicroRNAs are transcribed into hairpins in the nucleus. The hairpin is cut loose from the rest of the mRNA by Drosha and transported to the cytoplasm where they are trimmed by Dicer to form mature miRNA duplex.es. The RISC complex picks up the duplex and chooses one strand to be used as a template to find mRNA targets to silence. Inhibition of the mRNA leads to translational blocking and decreased protein expression of the silenced gene.

MicroRNAs contain inherent mismatches as a result of imperfect base-pairing in the hairpin structure of pri-miRNAs [4]. As a consequence, only a part of the miRNA strand sequence base pairs with the mRNA target. This sequence is known as the seed sequence and is only around 2-8 nucleotides long [15]. Due to the limited length of the seed sequence miRNAs can have many different target genes (with similar target sequences). The mismatches can also provide the possibility of the two strands of a miRNA having distinctive target mRNAs [17].

This potential for dual-strand activity means that a single miRNA duplex can have varying effects on gene expression depending which strand is chosen by the RISC complex.

In siRNA there is usually a clear distinction in function between the two strands in the duplex leading to one strand often being called the guide or sense strand that is responsible for silencing, while the other is termed anti-sense or passenger strand [18]. Similar terminology has been used to describe miRNA function, however the uncovered possible dual-strand activity of miRNAs has

(9)

9 lead to the use of a new naming system. The strands are called miR and miR*, or 5p and 3p after their orientation in the pri-miRNA hairpin (Figure 1.2) [19]. The latter terms will be used in this project.

Figure 1.2. The hairpin structure of pre-miRNA. The side of the hairpin with a 5'-end becomes the 5p strand, and the side with a 3'-end becomes the 3p strand.

1.2. MiR-34a and induced strand selection

The miRNA of interest in this project is miR-34a, a well-studied microRNA involved in various pathways concerning cell apoptosis [20]. miR-34a is found in most tissues and is transcriptionally regulated by p53. This transcription factor has a central role in facilitating cell apoptosis in response to DNA damage and is very often inactivated in cancer cells. It is involved in the repression of countless proteins that promote tumerogenicity and metastasis such as MYC, CDK4, and generally promotes programmed cell death (apoptosis) [21]. Of all documented target gene transcripts of miR-34a most confirmed targets are 5p strand targets.

This includes most documented tumour suppressor functions of miR-34a, while the overall function of the 3p strand is largely unknown and few targets have been confirmed [22, 23].

Due to miR-34a's strong link to tumour suppression it has become a promising tool for use clinically to treat cancer. However, despite the enormous potential of miR-34a as a cancer therapeutic, an inherent issue with administrating miR-34a, or any miRNA, as a drug is the risk of unknown off-target effects performed by the other strand, which in this case concerns the less-studied strand: miR-34a- 3p. This issue leads us to the main topic of this project that is to explore how selection of a given strand, 3p or 5p, can be induced through sequence engineering and chemical modifications. Strand selection bias has been studied in both siRNA and miRNA and one trend found was that the strand with the less thermodynamically stable 5'-end is incorporated more frequently into the RISC complex [24].

The strands of miR-34a are the targets for alteration in this project and the modifications are either placed on only one of the strands to create asymmetry and selection of the unmodified strand in the duplex, or on both to see how the symmetric modifications may affect miRNA function. There are two alterations of interest here: 5'-O-methylation (Figure 1.3, A) and simple strand elongation by nucleotide addition (Figure 1.3, B).

(10)

10

Figure 1.3. Modifications of miR-34a. (A) The methylation (red) introduced at the 5'-end of miR-34a (cytosine 3p strand). The cytosine base shown is replaced by a uracil when the methylation is on the 5p strand. The 'R' in this case stands for ribose and marks the continuation of the miRNA backbone. (B) Addition of seven nucleotides at the 3' end of the 5p strand, all the nucleotides have pyrimidine bases.

It is known that the 5'-end of the strand incorporated into the RISC complex is phosphorylated [25]. Furthermore the middle domain (MID) of the Ago protein in RISC, that interacts with the miRNA strands has a phosphate binding pocket, presumably used to bind to phosphorylated miRNA strands [26]. Introducing a methyl group on the 5'-OH group would block the 5'-end from being phosphorylated. This is assumed to prevent the modified strand from being integrated into RISC. Methylation of the 5'-end is a modification that has been tested in siRNA duplexes [27], where they observed selection of one strand when the other strand was methylated, which supports this hypothesis, at the same time as showing strand selection in miRNA using 5'-O-methylation would be novel.

Nucleotide addition of miRNA strands is a known natural process where adenylation and uridylation of 3' ends is speculated to protect strands from degradation by nucleases [28]. Strand asymmetry as a general concept was been widely discussed as a driving factor for strand selection bias in both siRNA and miRNA [29, 30]. To experimentally test how asymmetry can affect strand selection seven pyrimidine nucleotides were added to the 3'-ends of miR-34a (see Figure 1.3, B).

The aim of the current study was to evaluate the efficacy of engineered miR-34a sequences in regard to functionality and strand selection. This was done by observing biological changes in Human colon cancer cells in response to the introduction of the modified miR-34a duplexes. Intracellular miR-34a strand concentrations were monitored using Stem-loop qPCR, and strand functionality was tested in luciferase assays. Direct down-stream silencing effects of the miRNAs were tested on mRNA level using RT-qPCR, protein level using Western

A B

(11)

11 Blotting and cell level using Flow cytometry. Results demonstrate clear strand bias in cells treated with all asymmetrically modified miR-34as, providing evidence that both 5'-O-methylation and nucleotide addition are able to induce strand selection in miRNA.

2. Materials and methods

2.1. MicroRNAs

The sequence for miR-34a (see Table 2.1.) was obtained from miRBase. The Single-stranded miR-34a species and the negative control scramble (scr) were procured from ChemGenes. Hybridization into the duplexes of interest (see Figure 2.1) was performed by incubation of simplex pairs dissolved in water at a ratio of 1:1 at 94°C for 2 minutes, followed by continuous temperature decline to room temperature for approximately 4 hours.

Table 2.1. The strand sequences of miR-34a.

miR-34a strand Sequence

miR-34a-3p 5'-CAAUCAGCAAGUAUACUGCCCU-3'

miR-34a-5p 3'-UGGCAGUGUCUUAGCUGGUUGU-5'

B

C

A

Figure 2.1. Schematic structures of miR-34a and its modified variants.

Normal miR-34a and strands with a tail of seven nucleotides (termed (N)7-tail) (A). Methylation at the 5'-OH end of either or both strands (B) and strands with both 5'-O-methylation and (N)7-tail (C).

(12)

12

2.2. Cell transfections

Human Colon Tumour cells (HCT-116) were used in all experiments due to previous research showing that miR-34a is down-regulated in this cell line [31].

Cells were grown to 60-80% confluency in 25ml culture flasks with high glucose Dulbecco's Modified Eagle Medium (DMEM) containing 10% Foetal Bovine Serum (FBS) and 1% Penicillin-Streptavidin antibiotics. The cells were washed with phosphate saline buffer (PBS) and flask detachment was achieved by addition of 500µl TrypLE Express Enzyme (1X) followed by 4 min incubation at 37°C. The cells were resuspended in 3ml DMEM medium and cell density was calculated using a EVE automated cell counter (VWR).

In all transfection experiments cells were seeded using a density of 10⁵ cells per millilitre in microtiter plates of varying sizes depending on experiment and incubated for 24 hours. In transfection mixes, unmodified and modified miR-34a miRNAs (see Figure 2.1) were incubated in Lipofectamine RNAiMax Transfection Reagent and Gibco Opti-MEM Reduced Serum Media for 5 minutes at room temperature prior to being added to cells. The media in the wells was replaced with fresh DMEM and cells were transiently transfected using a concentration of 100nM miRNA per well and incubated at 37°C in 5% CO2.

2.3. Stem-loop qPCR

HCT-116 cells were seeded in 12-well microtiter plates at a quantity of 10⁵ cells in 1ml DMEM per well. Transfection was done by adding a mix of 80µl Opti-MEM, 2µl 50µM miRNA and 2µl RNAiMax to cells in fresh DMEM, followed by 24 hour incubation. Total miRNA was isolated from the cells using a mirVana miRNA Isolation Kit (Thermo Fischer) and concentration was determined using a DS-11 UV-Vis Spectrophotometer. Complementary DNA (cDNA) was synthesized using a TaqMan MicroRNA Reverse Transcription Kit (Thermo Fischer) and primers specific to the two miR-34a strands and miR-21 (control gene) designed for Stem-loop quantitative polymerase chain reaction (qPCR) [32] in a CFX Connect Real-Time PCR Detection System (see Supplementary Table 7.1. for PCR programs).

RT-qPCR was carried out on PCR mixes consisting of 1µl TaqMan probe primers specific to miR-34a strands and miR-21, 4µl RNase-free H2O, 5µl synthesized cDNA and 10µl TaqMan Fast Advanced Master Mix (Thermo Fischer) placed in 96-well plates. Using the same thermocycler but with a different program Real- time pPCR (RT-qPCR) was performed using TaqMan probe primers specific to miR-34a strands and miR-21 and TaqMan Fast Advanced Master Mix (Thermo Fischer). The data was analyzed using the 2^-ΔΔCT method [33] in GraphPad Prism,

(13)

13 and primer ratios were calculated. Standard error of the mean of all primer ratios was used to determine statistical variation. A Student's t-test was performed to evaluate significance with p-values noted as the following:

p ≤ 0.05 = ** and p ≤ 0.005 = ***.

2.4. Luciferase assays

The Psicheck-2 vector containing the genes encoding the luciferases Firefly and Renilla was obtained from Promega and sequences with complete complementarity to the strand sequences of miR-34a (see Table 2.1.) were cloned into the 3'-untranslated region (3'UTR) of the Renilla luciferase gene (Figure 2.2).

For the luciferase assays HCT-116 cells were seeded in a 96-well microtiter plate with 10⁴ cells in 100µl DMEM per well and incubated overnight for 24 h.

Transfection mixes containing 2µl RNAiMax, 40µl Opti-MEM and 1µl 50µM miRNA along with 600 ng dual-luciferase plasmid containing an inserted target sequence complementary to one of the two miR-34a strands (Table 2.1.) and incubated for 5 minutes at room temperature. The medium of the cells was replaced with fresh DMEM and the cells were transfected by adding 10µl of the transfection mixes to wells in triplicates and incubated for 24 hours.

Figure 2.2. Schematic view of the 3'UTR of the Renilla gene in the Psicheck-2 plasmid. The target sequence for miR-34a was inserted between the Renilla reading frame and the poly(A) stop signal.

A Dual-Glo Luciferase Assay kit (Promega) was prepared according to the manufacturer's descriptions and 75µl dual-glo reagent was added to the cell wells, the plate was incubated for 10 minutes at room temperature in darkness.

Firefly luciferase activity was determined by transferring 100µl from the wells onto a white opaque 96-well plate and measuring luminescence in a Tecan Infinite m200 plate reader. Stop-glo reagent was prepared and 75µl was added to the wells in the white plate and the white plate was incubated for 10 minutes at room temperature in darkness. Renilla luciferase activity was measured by measuring luminescence a second time on the white plate. The data was normalized to firefly luminescence and the negative control sequence (scr), and presented as percentage of knockdown. The standard deviation of the data was analyzed using GraphPad Prism. To evaluate significance, a Student's t-test

(14)

14 was performed with p-values noted as the following: p ≤ 0.05 = ** and p ≤ 0.005 = ***.

2.5. Real-Time qPCR assays

To attempt to determine changes in mRNA levels HCT-116 cells were grown in 24-well plates with 500µl DMEM and 5∙10⁴ cells per well. Transfection was carried out the same way as for Stem-loop qPCR, only the transfection mix was divided over two wells of cells in DMEM medium to create duplicates and the incubation time used was 48 hours.

The cells were washed in PBS and total RNA was isolated using a RNeasy Mini Kit (Qiagen) and quantified. Complementary cDNA was synthesized using a High- Capacity RNA-to-cDNA Kit (Thermo Fischer). RT-qPCR was performed by multiplexing two primers in each reaction well; one primer specific to β-Actin (control gene) and one for SIRT1 mRNA the same way as described earlier for Stem-loop qPCR but where 1µl H2O is replaced by 1µl of the second primer. The data was analyzed in GraphPad Prism and standard error of the mean of all datasets was used to determine statistical variation. Significance was evaluated using a Student's t-test p-values noted as the following: p ≤ 0.05 = ** and p ≤ 0.005 = ***.

2.6. Western Blot assays

To observe miRNA-induced knockdown on protein level, Western Blot assays were performed on protein content isolated from HCT-116 cells. Primary antibodies for the miR-34a targets SIRT1 and Axin2, and the control protein Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were procured from Abcam. GAPDH is a protein commonly used as a control due to its strong tendency for constant expression. The cells were seeded in 6-well plates with 2∙10⁵ cells in 2ml DMEM medium per well. Transfection was performed as described for Stem-loop qPCR but with doubled quantities and the cells were incubated for 72 hours for SIRT1 and 96 hours for Axin2.

After three days, the cells were washed with PBS twice and 200µl lysis buffer (Pierce RIPA Buffer with 1% protease inhibitors) was added to the wells, followed by scraping using cell scrapers for approximately 1 minute. The lysates were collected in microcentrifuge tubes, flash frozen in liquid nitrogen and stored at -20°C. Protein concentration was determined trough the Bradford protein assay [34] by creating a standard curve of Bovine Serum Albumin (BSA) with known concentrations of protein. The BSA protein was diluted to concentrations ranging from 0 to 2 mg/ml and 4µl of each concentration was

(15)

15 incubated with 100µl Pierce Coomassie Plus (Bradford) Assay Reagent (Thermo Fischer) in wells on a 96-well plate in darkness at room temperature for 10 minutes. Absorbance was measured at 595nm using a Tecan Infinite M200 Microplate reader, and a standard curve with absorbance as a function of protein concentration was set up using the data points. The protein samples were centrifuged in a himac CT15RE tabletop centrifuge (Hitachi Koki Co., Ltd) for 15 minutes at 35000 rpm and absorbance was determined using the using the method described for BSA. Protein concentration of the samples was determined using the linear correlation from the standard curve.

To prepare samples for SDS-PAGE, equal quantities of protein (ranging 15-30 mg depending on the lowest protein concentration) from the samples was mixed with H2O and 15µl loading dye (4x Laemmli buffer containing 10% β-mercaptoethanol) to a total volume of 60µl in microcentrifuge tubes. The samples were denatured at 95°C for 5 minutes and centrifuged briefly. A Mini-PROTEAN Tetra Cell system (Bio- Rad) was set up using Mini-PROTEAN TGX Precast Gels. Each protein sample was loaded onto the gel at a volume of 20µl. The Tetra cell system was connected to a PowerPac HV High-Voltage Power Supply (Bio-Rad) run at 100 V for approximately 40 minutes until the protein bands reach the bottom of the gel.

The protein bands were transferred from the gel onto a cellulose fiber blot using a Trans-Blot Turbo Transfer System (Bio-Rad). The blot was incubated in blocking solution (1x PBS with 1% Casein, Bio-Rad) on gentle rotation at room temperature for 1 hour and cut in two pieces to accommodate incubation in two different antibodies. The blot pieces were placed in blocking solution containing primary rabbit antibodies specific to SIRT1 and Axin2 diluted 1:100 and GAPDH diluted 1:5000 respectively and incubated for 24 hours at 4°C. The next day the blot s were washed in TBST (1% TWEEN 1X Tris-buffered saline buffer) under gentle rotation for 1 hour, changing the buffer every 15 minutes, followed by incubation in secondary goat-anti-rabbit antibodies diluted 1:2000 for SIRT1 and Axin2, and 1:5000 for GAPDH for 1 hour. Finally the blots were incubated in 4ml visualization solution from a Clean-Blot IP Detection Kit and taken for camera exposure using a Molecular Imager Gel Doc XR System (Bio-Rad).

2. 7. Flow cytometry

Cells were grown in 12-well plates with 10⁵ cells per well and incubated 24 hours.

The next day the cells were transfected with 25nM miRNA using the conditions described for Stem loop but with a quarter of the miRNA quantity, and incubated for 48 hours at 37°C. In this experiment, commercial miR-34a mimics from GenScript that are designed to show increased activity of one of the strands were used as controls for comparison. The cells were washed with PBS and detached using 200µl

(16)

16 TrypLE for each well. Following the instructions of the Annexin V-FITC Apoptosis Staining / Detection Kit, cells were resuspended in 500µl assay buffer and 5µl of the fluorescent dye Fluorescein isothiocyanate (FITC) and incubated in darkness for 5 minutes. The samples were measured using a CytoFLEX S Research Flow Cytometer (Beckman Coulter) with flow rate at medium, FITC light gain to 10, and Sidescatter and Forward scatter to 20. Around 10000 events were recorded and the data was analysed using the software Kaluza (Beckman Coulter) and plotted in Microsoft Excel.

3. Results

3.1. miRNA levels

As previously described, miRNAs are processed in the cytosol by the enzymes Drosha and Dicer before being picked up by the RISC complex. One of the miRNA strands is then chosen and bound to RISC while the other strand is discarded.

The cytosol contains many different RNA nucleases that quickly break down any unprotected RNAs, meaning that the discarded single-stranded miRNAs are short-lived, while the ones bound to RISC are protected. The same conditions apply to miRNAs added to cells artificially via transfection which is done in this project. Nevertheless the strand selection by the RISC complex creates a difference in abundance of the different miRNA species present in the cell, which can be quantified using Stem-loop qPCR [32]. The ratios of the miRNA strands 3p and 5p found in Figures 3.1 and 3.2 are used to describe the relation between the two strands with regard to abundance.

The strand ratios of cells transfected with unmodified miR-34a in Figure 3.1 are both close to 1 indicating that both strands are present in the cell in similar amounts. Additionally the samples where cells were treated with the symmetrical mod3b5b and modM species, which contain (N)7-tails and 5'-O- methylations on both strands respectively, show patterns almost identical to miR-34a. There is a distinct difference between strand quantities when looking at mod3b and mod5b (modified with (N)7-tails on either strand) in Figure 3.1;

the 3p strand is significantly greater in abundance in cells transfected with mod3b meaning that the modified strand of the duplex is poorly picked up by the RISC complex and appears to promote selection of 3p. Similarly there is more of the 5p strand found compared to 3p in cells transfected with mod5b.

(17)

17

Figure 3.1. The apparent miRNA levels inside HCT-cells 24 hours after transfection of (N)7-tail-modified miR-34a variations. All data points have been normalized to scr miRNA levels and are presented as ratios between the two miRNA moieties (3p and 5p). The standard deviation is based off several replicate experiments.

When looking at the data for the samples containing the mod3M duplex (Figure 3.2 below) where a 5'-O-methylation is found on the 5p-strand to promote 3p selection, the 3p/5p ratio is higher than the 5p/3p ratio indicating a higher cellular concentration of the 3p strand however the statistical variation is rather high. The mod5M-transfected sample that has a methylated 3p strand, unexpectedly has strand ratios both close to 1 like miR-34a with no apparent bias for either strand. Finally the samples transfected with miRNAs having both modifications on one strand in the duplex; mod3bM and mod5bM show clear preference for the 3p and 5p strands respectively to an even larger degree than mod3M and mod5M.

***

**

(18)

18

Figure 3.2. The apparent miRNA levels inside HCT-cells 24 hours after transfection of 5'-O-methylated miR- 34a modifications. All data points have been normalized to scr miRNA levels and are presented as ratios between the two miRNA moieties (3p and 5p).

3.2. Luciferase assays

After confirming that transient transfection of miRNAs successfully increases intracellular miRNA levels, to further investigate how the miRNAs affect gene expression we designed a dual-luciferase system on plasmids compatible with human cells where the Renilla luciferase gene had an inserted target sequence for either the 3p or the 5p strand of miR-34a in the 3'UTR. The plasmid DNA is transcribed into mRNA once inside cells. Luciferase assays are simple functional assays where the efficiency of a specific miRNA strand can be tested on a luminescent reporter gene that acts as an artificial target. To ensure a clear interaction between the miRNA and the plasmids the target sequences were designed to have complete complementarity to one of the strands of miR-34a and not only the seed sequence.

The graphs depicted in Figures 3.3 and 3.4 describe how knockdown of the Renilla luciferase luminescence (relative to Firefly luminescence) changes depending on what miRNA was transfected along with either of the plasmids.

miR-34a shows similar knockdown of both dual-reporter plasmids. Interestingly mod3b shows no difference in knockdown of either plasmid comparable to miR- 34a, despite the stem-loop data showing significantly higher intracellular levels of the 3p strand after the same amount of time post-transfection (24 hours). The mod5b and mod3b5b treated cells both show an analogous pattern of lowered knockdown of the 3p plasmid while knockdown of the 5p plasmid is equal to that of cells treated with miR-34a, hinting that the 3p strand is not as readily chosen by RISC compared to 5p.

***

** ***

***

**

(19)

19

Figure 3.3. Luciferase assays on miR-34a and (N)7-tail miRNAs. The data has been normalized to Firefly luminescence and the negative control scr.

Comparing data of samples transfected with the 5'-O-methylated miRNAs (see Figure 3.4) mod3M and mod5M both seem to show complete preference for one target plasmid, with the other showing no apparent knockdown of the Renilla luciferase. The symmetrically methylated modM shows no preference for either strand but has lower knockdown of both strands compared to normal miR-34a.

Lastly mod3bM and mod5bM again show clear preference for one target, however neither have complete absence of knockdown of either target plasmids.

Figure 3.4. Luciferase assays on miR-34a and methylated miRNAs. The data has been normalized to Firefly luminescence and the negative control scr.

*** **

***

**

***

(20)

20

3.3. The effect on SIRT1 mRNA levels

By isolating the total amount of RNA from HCT-116 cells the abundance of specific mRNA transcripts can be determined. Silent information regulator 1 (SIRT1) is a protein that hinders cell apoptosis by deactivating p53, whose gene is directly targeted by miR-34a-5p [35]. The gene transcript for SIRT1 was attempted to be quantified using SIRT1-specific primers and the data is compiled in Figure 3.5 Normal miR-34a seems to decrease mRNA levels by around 50%

and the following four miRNAs in Figure 3.5 (mod5b, mod3b, mod3b5b and mod5M) follow a similar pattern. The last four miRNAs, mod3M, modM, mod5bM and mod3bM do not decrease SIRT1 mRNA to the same extent as the other miRNAs, moreover mod3bM show values more consistent with the negative control (scr).

Figure 3.5. RT-qPCR of SIRT1 mRNA in cells transfected with miRNAs for 48 hours. The data is based off four replicate experiments and has been normalized to the negative control scramble (scr).

3.4. Western blots

The next step was to determine potential effects the miRNAs may have on a protein level in the cells, Western blot assays were performed. Of the proteins tested, SIRT1 is a well-known target for miR-34a-5p, while few articles support miR-34a-3p-mediated suppression of Axin2. It was, nevertheless the only viable 3p target of miR-34a that was found in the literature. Various incubation times, from the moment cells were transfected, were tested to find the time point at which the greatest difference in protein quantities could be observed. From this

** *** **

***

(21)

21 process, the time-points 72 hours for SIRT1 and 96 hours for Axin2 were chosen.

The resulting immunoblots from Western have distinct bands of amassed antibody-bound SIRT1 (Figure 3.6, B) and Axin2 (Figure 3.7, B) proteins with clear differences in density. Using the programs ImageJ and Microsoft Excel the density of each SIRT1 and Axin2 band in relation to the corresponding GAPDH bands was quantified, resulting in the densitometry plots in Figures 3.6, A and 3.7, A below. According to the densitometry plot for SIRT1 silencing, addition of normal miR-34a to cells, followed by 72 hour incubation decreases the SIRT1 protein by approximately 60% compared to the negative control (scr). The protein sample silenced by the mod5b duplex, which has a (N)7-tail at the 3p strand, appears to silence SIRT1 as efficiently as miR-34a. The combination type mod5bM, which was expected to strongly select for 5p however shows less silencing compared to mod5b and mod5M. The samples transfected with duplexes that select for 3p, mod3b, mod3M and mod3bM are expected to not silence SIRT1 so the fact that all three show higher protein density than miR-34a (Figure 3.6, A) means that there might be 3p selection present. What's more, mod3bM has a density close to the negative control (scr).

Figure 3.6. Western blot assay on SIRT1 and GAPDH (A) and a densitometry analysis of the blot (B). The densitometry data has been normalized to corresponding GAPDH bands on the blot and the negative control scr. The names of the different miRNAs have been abbreviated in this figure, e.g. mod3b is 3b.

0 20 40 60 80 100 120

scr 34a 3b 5b 3bM 5bM 3M 5M

Norm. Densitometry

A

B

(22)

22 Looking at the blot that has been stained for Axin2 protein (Figure 3.7, B) miR- 34a induces silencing of the Axin2 protein by approximately 50%. Moreover mod3b and mod3bM seem to silence the protein even further meaning that it is a target for 3p, though mod3M does not appear to be as efficient. The 5p-selection samples mod5b, mod5M and mod5bM all have higher protein density than miR- 34a indicating the 5p strand does not silence Axin2. The protein density of the GAPDH band of mod5M in Figure 3.7, B is lower for some reason, which leads to the abnormally high normalized density that is greater than the negative control in the plot (Figure 3.7, A). In this blot the symmetrical miRNAs mod3b5b and modM were also tested, where the former demonstrates silencing similar to miR- 34a while the latter has a protein density closer to the negative control.

Figure 3.7. Western blot assay on Axin2 and GAPDH (A), and a densitometry analysis of the blot (B). The densitometry data has been normalized to corresponding GAPDH bands on the blot and scr. The names of the different miRNAs have been abbreviated in this figure.

3.5. Flow cytometry

Finally, to go one step further we wanted to assess the effect miR-34a and its modified variations may have on a cellular level. As mentioned before the protein SIRT1 is involved in prevention of apoptosis. This means that silencing of

0 20 40 60 80 100 120 140

scr 34a 3bM 5bM 5M 3M M 5b 3b 3b5b

Norm. Densitometry

A

B

(23)

23 SIRT1 may lead to cells undergoing apoptosis. To what degree a cell has become apoptotic can be determined by using fluorescently labelled Annexin V, a protein that binds to phosphatidylserines on cell surfaces. These serines are known markers of apoptosis [36]. After measuring the FITC stained samples on the flow cytometer the recorded events can be sorted into different populations according to size using forward scatter (FSC), granularity using side scatter (SSC) and fluorescent signal (FITC). For identifying differences in FITC signal between the different cell samples, histograms were utilized (Figure 3.8, A and B). The plots were based off a chosen set of events marked red that was picked from dot plots (Figure 3.8, C and D) with granularity (SSC) as a function of size (FSC).

Events of very low SSC and FSC are generally considered to be cell debris and were therefore omitted from the analysis (seen in grey in dot plots). A horizontal gate was placed in the histogram of the negative control and used as a baseline when analyzing increase and decrease in FITC signal across the samples. The gated events are shown in green and clear shifts in FITC signal on the x axis of Figure 3.8 between A and B can be seen by the increase in green-coloured events and in percentage total of the chosen cell population in the sample transfected with miR-34a.

Figure 3.8. Flow cytometry histograms showing cell count as a function of FITC signal for the negative control (A) and miR-34 (B). Less cells are colored green in the corresponding dot plot for the untransfected cells with FITC that acts as the negative control (C) compared to miR-34a (D).

A B

C D

(24)

24 The percentages obtained from gating all samples (See Supplementary Figure 7.2 for the remaining histograms and dot plots) were normalized to untransfected cells stained with FITC and collected in Figure 3.9 below. Unmodified miR-34a increases the apoptotic cell signal by approximately 8 times compared to cells transfected with scramble miRNA (scr). The mod3b miRNA shows effect comparable to that of the commercial mimic 3p miRNA, both of which have round half the apoptotic signal of miR-34a. Both mod5b and the 5p mimic have higher signal than normal miR-34a, albeit the 5p mimic gave an even higher signal, being close to 11 times higher than the control.

Figure 3.9. Flow cytometry data on HCT-116 cells 48 hours after transfection. Data points have been normalized to cells stained with FITC only. The mimics are used as controls to assess effectiveness of the modified miR-34as.

4. Discussion

The issue that we aimed to solve in this project was the lack of control concerning the inherent potential dual-strand activity of miRNAs. This is a problem that hampers the study of miRNA function. In the case of miR-34a, which has the potential to be used as treatment for cancer, the unknown activity of the 3p strand raises the risk of unwanted protein targets beyond the desired function of the 5p strand.

Here we introduce simple modifications on miR-34a and assess their potential as tools to induce strand selection through a series of experiments that are designed to observe the effect of miRNAs in cell cultures. The modifications 5'-O- methylation and nucleotide addition are rather simple to achieve. This is

(25)

25 especially true for addition of nucleotides (in this case, only pyrimidines), which is not only a biologically relevant process, as there nowadays also exist a wide variety of companies that provide oligonucleotide synthesis services [37].

Consequently, should this modification prove useful to induce strand selection sufficiently, it would be accessible to most labs wishing to use it to study miRNA strand selection. While 5'-end methylation is a more unnatural modification, the key interest lies in its effect on miRNA functionality and strand selection. The fact that it has already been tested in siRNA is another driving reason [27].

All experiments began by growing human colon cancer cells (HCT-116 cells), on which the modified miR-34as would be tested, and the first step was to determine intracellular miRNA levels after transfecting the cells. This was done using Stem-loop qPCR - a method designed specifically for quantifying miRNAs, which otherwise are too small to be detected by conventional qPCR methods [32]. What the results showed were clear differences in relative strand abundance (Figures 3.1 and 3.2) between the different isolated samples. This correlates with the idea that strands from miRNA duplexes that are not incorporated into the RISC complex are degraded [19, 29]. Consequently what is seen can be interpreted as strand selection, since the more abundant strands can be assumed to have been chosen to be included in the RISC complex where nucleases cannot reach. Another factor that might be at play is asymmetry between the two strands of a duplex, which is known to influence the observed silencing function [29]. This might explain why the strand selection only appears in samples that were treated with miRNA duplexes with modifications on one of the strands. To build further, the miRNAs miR-34a, mod3b5b and modM, which all are symmetrical in structure, do not show any significant difference in strand abundance of either 3p or 5p strands. The exception to this pattern, however is mod5M, which unexpectedly does not seem to promote an increase in strand abundance of the 5p strand, for reasons unknown at this moment. An interesting observation when further comparing Figure 3.1 and Figure 3.2 is that mod3bM and mod5bM appear to produce an even stronger bias for 3p and 5p than the combined effects of mod3b and mod3M, and mod5b and mod5M respectively.

This hints to a potential synergistic effect that springs from using both modifications in the same duplex.

In contrast, the efficiency of the (N)7-tail-modified types seen in the luciferase assays are not as impressive (Figure 3.3). The strand selection efficiency of mod3b seen in the Stem-loop experiments does not appear at all when tested in the luciferase system (Figure 3.3). Only mod5b and mod3b5b show any significant difference compared to normal miR-34a, and only for 5p strand selection. Why there seems to be a decrease in 3p knockdown in the symmetrical mod3b5b sample is unclear. What is important to remember is that while stem- loop qPCR provides direct quantitative data on the apparent concentration of

(26)

26 miRNAs in the cells, the luciferase assays show data more related to "strand efficiency" i.e. how efficiently the RISC complex picks up and utilizes a certain strand to bring about translational repression. In other words, simply seeing the presence of a strand in stem-loop qPCR does not immediately imply that the strand is actively inhibiting translation.

There are two possible reasons as to why mod3b does not show less knockdown of the 5p plasmid; either the target sequences introduced on the plasmids are similar enough for both miRNA strands to bind and repress the gene, or the miRNA strands have errors in their sequences. MicroRNAs are known to interact with mRNA targets using only a part of their miRNA sequence [38], meaning that it is not impossible for the 3p strand to bind to the 5p target, especially since the target sequence is designed to be complementary to the full length of miR-34a.

However, if this was true, it would presumably show up as increased knockdown of the 5p plasmid in the other miRNA samples that select for 3p, which it does not. An error in the sequence can be harder to pinpoint, and studies have shown that a single change in the miRNA sequence can lead to altered strand selection [39]. The quickest way to investigate this is to acquire a new batch of the sequences and retest them to see if the observed pattern changes.

The methylated miRNAs do, in any case, seem to strongly induce strand selection and selective knockdown of the target plasmids in the luciferase assays. Here, a clear difference in function that could not be observed in the Stem-loop data is the clear decrease in knockdown of both strands by modM compared to miR-34a.

The modM species caused knockdown of both 3p and 5p target plasmids, indicating that phosphorylation of the 5'-end of the strands is not necessary for RISC binding and translational repression. In siRNA it has been suggested that, although it may have an enhancing effect on the degree of silencing, phosphorylation of the 5' ends in siRNA strands is not necessary for RISC binding [40]. The degree of knockdown of modM is similar to what Chen et al. reported for luciferase assays on symmetrically methylated siRNA duplexes [27]. In this article, they achieve visible strand selection of both strands of an siRNA with a predisposed strand bias, hinting to the efficacy of 5'-O-methylation. This is also reflected in the data on the other methylated strands in the luciferase assays. The very fact that methylated miRNAs appear to have no problem in targeting suggests there is no problem with the sequences as mentioned in the earlier paragraph. Perhaps the problem is the (N)7-tail; it is possible that the tail could be part-taking in target binding. Changing the sequence of the nucleotide tail might remove this issue.

Going back to assessing effects of miR-34a and its modifications, changes in SIRT1 mRNA was attempted to be quantified using RT-qPCR. This proved to be rather difficult however, as little correlation was found between the different

(27)

27 sample treatments (Figure 3.5), besides a general knockdown of SIRT1 in most samples, despite standardization attempts to increase the apparent changes by altering incubation time, miRNA concentration and other parameters.

Changes in protein content in transfected samples was quantified through Western Blots to get a more absolute verdict on the effects of miRNAs on the cells. The Western blots both show silencing of the protein targets SIRT1 and Axin2 (Figures 3.6 and 3.7). SIRT1 silencing has been tested in HCT116 cells under similar conditions using miR-34a [35], which gave comparable knockdown. Surprisingly Axin2 was also efficiently silenced by miR-34a and its modifications in a similar fashion as presented by Kim et al. [41]. The knockdown patterns are similar to the relative strand abundances seen in the Stem-loop data, implying that the observed miRNA levels 24 hours after transfection can be a good indicator of the resulting silencing patterns of the target protein isolated 48 hours later.

Finally the effects miR-34a may have on cell viability were tested. Being a miRNA that silences SIRT1 which is involved in repressing apoptosis, miR-34a was expected to induce some proapoptotic effects on the cancer cells. It was hard to predict how minute the differences between the different modified duplexes would be, but the hope was of course that strand selection would be visible by increased or decreased apoptotic signal. This signal was quantified using Annexin V labelled with FITC and cells were sorted using flow cytometry according to size, granularity and FITC intensity. Looking at the change in FITC intensity in the histograms proved to be a good way to determine differences between the samples and the resulting graph (Figure 3.9) shows the pattern expected from SIRT1 knockdown [35]. The 3p-promoting duplexes mod3b and the 3p mimic should in theory not induce apoptosis since they select for the non- silencing 3p strand. What is seen in Figure 3.9 is that these samples only gave an apoptotic signal approximately twice the size of the FITC signal intensity of the controls, suggesting that 3p selection is occurring in these samples, which also reconfirms that SIRT1 is not silenced by miR-34a-3p. Furthermore the mod5b and 5p mimic samples produced an even higher apoptotic signal than miR-34a, suggesting that selection of 5p led to an increased silencing of SIRT1, which consequently led to an increase in apoptotic tendency in the cells.

All experiments combined, it is clear that there is strand selection at work in the samples with the modified miR-34as. The fact that the data suggests unmodified miR-34a apparently lacks bias for any strand underlines the need for strand selection. 5'-O-methylation provides insight in miRNA - RISC binding and can be a potent minute change in the miRNA that creates profound differences in strand selection. Nucleotide addition too, showed significant changes in selection and has the potential to be altered in various ways to achieve different outcomes,

(28)

28 perhaps even more efficient strand selection. In combination with its availability and its potential for sequence and length variability, it has the potential to be a great method to induce strand selection in both miRNA and siRNA.

5. Acknowledgements

I would like to thank the Polymer group here at Uppsala University for their kindness, support and all the intriguing scientific discussions we had during my 6 month stay in the group. My supervisor Oommen Varghese never failed to provide answers to all my questions. And Sandeep Kadekar took the time to teach me all about the methods I used in this project. I would also like to thank Ganesh Nawale for his insights on RNA chemistry as well as all the remaining members of the group. I wish you all the best.

(29)

29

6. References

1. Xu, S., Driver, S.E., Kostas, S.A., Fire, A., Montgomery, M.K. & Mello, C.C.

1998, "Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans", Nature, vol. 391, no. 6669, pp. 806-811

2. Hannon, G.J. 2002, "RNA interference", Nature, vol. 418, no. 6894, pp. 244- 251

3. Heigwer, F., Port, F. & Boutros, M. 2018, "RNA Interference (RNAi) Screening in Drosophila", Genetics, vol. 208, no. 3, pp. 853-874

4. Wu, H., Ma, H., Ye, C., Ramirez, D., Chen, S., Montoya, J., Shankar, P., Wang, X.A. & Manjunath, N. 2011, "Improved siRNA/shRNA functionality by mismatched duplex", PloS one, vol. 6, no. 12, pp. e28580

5. Arguel, M., Jaouannet, M., Magliano, M., Abad, P. & Rosso, M. 2012, "siRNAs Trigger Efficient Silencing of a Parasitism Gene in Plant Parasitic Root- Knot Nematodes", Genes, vol. 3, no. 3, pp. 391-408

6. de Felippes, F.F., Wang, J. & Weigel, D. 2012, "MIGS: miRNA‐induced gene silencing", The Plant Journal, vol. 70, no. 3, pp. 541-547

7. Peng, Y. & Calin, G.A. 2018, "Crucial role of non-coding RNAs in disease", Cancer Letters, vol. 420, pp. 127

8. Garber, K. 2015, "Alnylam's RNAi therapy targets amyloid disease", Nature biotechnology, vol. 33, no. 6, pp. 577

9. Ha, M. & Kim, V.N. 2014, "Regulation of microRNA biogenesis", Nature reviews. Molecular cell biology, vol. 15, no. 8, pp. 509

10. Carthew, R.W. & Sontheimer, E.J. 2009, "Origins and Mechanisms of miRNAs and siRNAs", Cell, vol. 136, no. 4, pp. 642-655

11. Schanen, B.C. & Li, X. 2011, "Transcriptional regulation of mammalian miRNA genes", Genomics, vol. 97, no. 1, pp. 1-6

12. Han, J., Choi, H., Kim, V.N., Kim, J., Yim, J., Lee, J., Rådmark, O., Lee, Y., Kim, S., Ahn, C. & Provost, P. 2003, "The nuclear RNase III Drosha initiates microRNA processing", Nature, vol. 425, no. 6956, pp. 415-419

13. Lund, E., Güttinger, S., Calado, A., Dahlberg, J.E. & Kutay, U. 2004, "Nuclear Export of MicroRNA Precursors", Science, vol. 303, no. 5654, pp. 95-98

(30)

30 14. Pare, J.M., Tahbaz, N., López-Orozco, J., LaPointe, P., Lasko, P. & Hobman, T.C. 2009, "Hsp90 regulates the function of argonaute 2 and its recruitment to stress granules and P-bodies", Molecular biology of the cell, vol. 20, no. 14, pp. 3273

15. Lewis, B.P., Burge, C.B. & Bartel, D.P. 2005, "Conserved Seed Pairing, Often Flanked by Adenosines, Indicates that Thousands of Human Genes are MicroRNA Targets", Cell, vol. 120, no. 1, pp. 15-20

16. Iwakawa, H. & Tomari, Y. 2015, "The Functions of MicroRNAs: mRNA Decay and Translational Repression", Trends in Cell Biology, vol. 25, no.

11, pp. 651-665

17. Hashimoto, Y., Akiyama, Y. & Yuasa, Y. 2013, "Multiple-to-multiple relationships between microRNAs and target genes in gastric cancer", PloS one, vol. 8, no. 5, pp. e62589

18. Hägerlöf, M., Hedman, H., Elmroth, S.K.C., Biochemistry and Structural Biology, Lund University, Lunds universitet & Biokemi och Strukturbiologi 2007, "Platination of the siRNA sense-strand modulates both efficacy and selectivity in vitro", Biochemical and Biophysical Research Communications, vol. 361, no. 1, pp. 14-19

19. Meijer, H.A., Smith, E.M. & Bushell, M. 2014, "Regulation of miRNA strand selection: follow the leader?", Biochemical Society transactions, vol. 42, no. 4, pp. 1135

20. Slabáková, E., Culig, Z., Remsík, J. & Soucek, K. 2017, "Alternative mechanisms of miR-34a regulation in cancer", Cell Death and Disease, vol.

8, no. 10

21. Misso, G., Di Martino, M.T., De Rosa, G., Farooqi, A.A., Lombardi, A., Campani, V., Zarone, M.R., Gullà, A., Tagliaferri, P., Tassone, P. & Caraglia, M. 2014, "Mir-34: a new weapon against cancer?", Molecular therapy.

Nucleic acids, vol. 3, pp. e194

22. Guennewig, B., Roos, M., Dogar, A.M., Gebert, L.F.R., Zagalak, J.A., Vongrad, V., Metzner, K.J. & Hall, J. 2014, "Synthetic pre-microRNAs reveal dual- strand activity of miR-34a on TNF-α", RNA (New York, N.Y.), vol. 20, no. 1, pp. 61

23. Cha, Y.H., Kim, N.H., Park, C., Lee, I., Kim, H.S. & Yook, J.I. 2012, "MiRNA-34 intrinsically links p53 tumor suppressor and Wnt signaling", Cell Cycle, vol. 11, no. 7, pp. 1273-1281

24. Khvorova, A., Reynolds, A. & Jayasena, S.D. 2003, "Functional siRNAs and miRNAs Exhibit Strand Bias", Cell, vol. 115, no. 2, pp. 209-216

(31)

31 25. Lau, N.C., Lim, L.P., Weinstein, E.G. & Bartel, D.P. 2001, "An Abundant Class of Tiny RNAs with Probable Regulatory Roles in Caenorhabditis elegans", Science, vol. 294, no. 5543, pp. 858-862

26. Elkayam, E., Kuhn, C., Tocilj, A., Haase, A.D., Greene, E.M., Hannon, G.J. &

Joshua-Tor, L. 2012, "The structure of human argonaute-2 in complex with miR-20a", Cell, vol. 150, no. 1, pp. 100-110

27. Chen, P.Y., Weinmann, L., Gaidatzis, D., Pei, Y., Zavolan, M., Tuschl, T. &

Meister, G. 2008;2007;, "Strand-specific 5'-O-methylation of siRNA duplexes controls guide strand selection and targeting specificity", RNA (New York, N.Y.), vol. 14, no. 2, pp. 263-274

28. Katoh, T., Sakaguchi, Y., Miyauchi, K., Suzuki, T., Kashiwabara, S., Baba, T.

& Suzuki, T. 2009, "Selective stabilization of mammalian microRNAs by 3' adenylation mediated by the cytoplasmic poly(A) polymerase GLD-2", Genes & development, vol. 23, no. 4, pp. 433-438

29. Schwarz, D.S., Hutvágner, G., Du, T., Xu, Z., Aronin, N. & Zamore, P.D. 2003,

"Asymmetry in the Assembly of the RNAi Enzyme Complex", Cell, vol. 115, no. 2, pp. 199-208

30. Suzuki, H.I., Katsura, A., Yasuda, T., Ueno, T., Mano, H., Sugimoto, K. &

Miyazono, K. 2015, "Small-RNA asymmetry is directly driven by mammalian Argonautes", Nature structural & molecular biology, vol. 22, no. 7, pp. 512

31. Gao, J., Li, N., Dong, Y., Li, S., Xu, L., Li, X., Li, Y., Li, Z., Ng, S.S., Sung, J.J., Shen, L. & Yu, J. 2015, "miR-34a-5p suppresses colorectal cancer metastasis and predicts recurrence in patients with stage II/III colorectal cancer", Oncogene, vol. 34, no. 31, pp. 4142-4152

32. Chen, C., Ridzon, D.A., Broomer, A.J., Zhou, Z., Lee, D.H., Nguyen, J.T., Barbisin, M., Xu, N.L., Mahuvakar, V.R., Andersen, M.R., Lao, K.Q., Livak, K.J.

& Guegler, K.J. 2005, "Real-time quantification of microRNAs by stem-loop RT-PCR", Nucleic acids research, vol. 33, no. 20, pp. e179-e179

33. Livak, K.J. & Schmittgen, T.D. 2001, "Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method", Methods (San Diego, Calif.), vol. 25, no. 4, pp. 402

34. Markwell, J. 2009, "Fundamental laboratory approaches for biochemistry and biotechnology, 2nd edition", Biochemistry and Molecular Biology Education, vol. 37, no. 5, pp. 317-318

35. Yamakuchi, M., Ferlito, M. & Lowenstein, C.J. 2008, "miR-34a Repression of SIRT1 Regulates Apoptosis", Proceedings of the National Academy of Sciences of the United States of America, vol. 105, no. 36, pp. 13421- 13426

(32)

32 36. Rimon, G., Bazenet, C.E., Philpott, K.L. & Rubin, L.L. 1997, "Increased surface phosphatidylserine is an early marker of neuronal apoptosis", Journal of neuroscience research, vol. 48, no. 6, pp. 563-570

37. Hughes, R.A. & Ellington, A.D. 2017, "Synthetic DNA Synthesis and Assembly: Putting the Synthetic in Synthetic Biology", Cold Spring Harbor Perspectives in Biology, vol. 9, no. 1, pp. a023812

38. Mullany, L.E., Herrick, J.S., Wolff, R.K. & Slattery, M.L. 2016, "MicroRNA Seed Region Length Impact on Target Messenger RNA Expression and Survival in Colorectal Cancer", PloS one, vol. 11, no. 4, pp. e0154177 39. Angart, P.A., Carlson, R.J., Adu-Berchie, K. & Walton, S.P. 2016, "Terminal

Duplex Stability and Nucleotide Identity Differentially Control siRNA Loading and Activity in RNA Interference", nucleic acid therapeutics, vol.

26, no. 5, pp. 39-317

40. Gregory, R.I., Chendrimada, T.P., Cooch, N. & Shiekhattar, R. 2005, "Human RISC Couples MicroRNA Biogenesis and Posttranscriptional Gene Silencing", Cell, vol. 123, no. 4, pp. 631-640

41. Kim, N.H., Cha, Y.H., Eun Kang, S., Mi Lee, Y., Lee, I., Young Cha, S., Kyung Ryu, J., Min Na, J., Park, C., Yoon, H., Park, G., Yook, J.I. & Kim, H.S. 2013,

"p53 regulates nuclear GSK-3 levels through miR-34-mediated Axin2 suppression in colorectal cancer cells", Cell Cycle, vol. 12, no. 10, pp.

1578-1587

(33)

33

7. Supplementary

Table 7.1. PCR programs

Program

No. Name Steps

1 2 3

1 Stemloop cDNA 30 min

at 16°C

30 min at 42°C

5 min at 85°C 2 Taqman fast miRNA 2 min at

95°C 20 sec.

at 95°C (1 sec. at 95°C followed by 20 sec. at 60°C) x 40 cycles

3 mRNA to cDNA 1 h at

37°C

5 min at 95°C

4 Taqman fast 20 sec.

at 95°C (1 sec. at 95°C followed by 20 sec. at 60°C) x 40 cycles

Figure 7.1. Flow cytometry histograms and dot plots

A

B

(34)

34 C

D

E

Figure 7.1. The remaining histograms and dot plots, from which the percentages for Figure 14 were obtained. The gated events are shown in orange. From top to bottom the histograms and dot plots belong to scramble (scr) (A), mod3b (B), mod5b (C), mimic 3p (D) and mimic 5p (E).

UPPSALA UNIVERSITY