A dual fluorescence-based reporter assay for real-time determination of siRNA- and antisense oligonucleotide-mediated knockdown

Abstract

In recent years, many tools have been developed for targeted treatments of cancer or rare diseases including nucleic acid (NA)-based therapeutics. One of the developmental caveats is the assessment of their functionality in vivo. We therefore developed a sensitive dual fluorescence-based FluoroDetect assay for the testing of small interfering RNAs (siRNAs), antisense oligonucleotides (ASOs), and RNA-binding compounds. For a proof of principle, we inserted the noncoding HSat3 RNA as a prime target into the 3′UTR of an mCherry fluorescence protein. The FluoroDetect assay was designed to be adapted by the In-Fusion cloning with any wished target site. The assay can be used as a quick transient model for short-term experiments or as a lentiviral reporter for long-term experiments. Readout of the assay is possible with fluorescence microscopy, plate reading, or flow cytometry, and can be used to measure the cellular distribution of NA therapeutics, siRNAs, and ASOs. Thus, the FluoroDetect assay is a tool to screen NA drug candidates and facilitates the optimization and quantification of NA delivery in NA-based therapies.

Graphical abstract

cost-efficient and sensitive dual fluorescence-based FluoroDetect assay is developed to detect RNAi, screen nucleic acid (NA) drug candidates, and optimize and quantify NA delivery. Fast adaptation using In-Fusion cloning was demonstrated. Transient short-term experiments or lentiviral transduced long-term experiments can be performed using fluorescence microscopy, plate reading, or flow cytometry readout.

Introduction

Since its discovery in 1998, RNA interference (RNAi) has been established as an important tool in molecular biology harboring a large potential for targeted drug development.¹ This development is underlined by the steadily increasing number of clinical trials and approvals of small interfering RNA (siRNA)-based drugs since the Food and Drug Administration approval of Onpattro in 2018.^2,3 Up to now, there are more than 30 clinically approved RNA therapeutics and more than 1,100 are in preclinical and clinical development.^4,5,6 In particular, monogenetic, infectious, and oncological diseases are addressed with these new tools.⁴ Despite its high power, the development of siRNA therapies is a challenging and cost-intensive approach. This is mainly due to limited screening platforms for the functionality of siRNA including their proper delivery. Commonly used technologies such as western blots, immunofluorescence, and luciferase assays rely on expensive substrates or are mainly performed as endpoint measurements. This makes the capture of dynamic processes nearly impossible. Similarly, luciferase-based measurements are certainly known for their accuracy but suffer greatly from the high costs and error-prone implementation. Hence the development of an easy-to-use, more economic platform in siRNA therapy development is urgently needed.

RNAi relies on two major mechanisms: a double-stranded siRNA induces the formation of the RNA-induced silencing complex (RISC), whereas single-stranded antisense oligonucleotides (ASOs) as well as gapmers recruit Ribonuclease Hybrid (RNaseH) and sterically block the translation process.^7,8 Both mechanisms function over targeting complementary RNA sequences. Such target sequences can be inserted into the 5′ or 3′ untranslated region (UTR) of reporter genes encoding fluorescence proteins.

Kojima et al. have previously shown that fluorescence reporter assays can be used to detect siRNAs in bulk as well as in single-cell analyses.⁹ Fluorescence assays based on the red fluorescence protein mCherry (mCherry) and the enhanced green fluorescence protein (EGFP) have been established to investigate dynamic intracellular processes of receptor and ligand interactions.¹⁰ An important advantage of fluorescence measurements, as done with the FluoroDetect assay, is that the quantification can be done without lysing cells; thus, live-cell imaging is feasible.

Human Satellite III (HSat3) RNA, transcribed from pericentromeric heterochromatin under certain stress conditions, plays a crucial role in the cellular stress response by forming nuclear stress bodies.¹¹ These RNAs are significantly induced by various stressors and are involved in stress-dependent splicing and genome stability.¹² As previously shown by Kanne et al., targeting HSat3 RNAs with siRNAs modulates the stress response and enhances the efficacy of chemotherapies¹³

Using the two fluorescence proteins mCherry and EGFP, the here-developed assay aims to be more precise than a single signal-based assay. The EGFP fluorescence signal works as reference signal. The mCherry fluorescence protein is the reporter protein used to quantify the knockdown efficacy. Moreover, the siRNA 3′UTR target site of the reporter gene is flanked with two individual restriction sites (AsiSI and NsiI) for easy and fast adaptation of the assay plasmids (Figure 1A). As an internal control for advanced experimental setups, the reporter gene (mCherry) is under control of a tetracycline-deducible promotor system (TET-OFF). In addition to the single-plasmid-based approach for transient transfection, a two-plasmid-based approach was designed for the generation of lentiviral transduced stable cell lines.

Figure 1.

FluoroDetect assay—Mechanism of action

(A) Schematic workflow of the FluoroDetect assay. Fluorescence signal of mCherry reporter gene can be downregulated by siRNA while the green fluorescence signal of the EGFP remains stable. (B) DNA and mRNA sequence of the target motif used in the FluoroDetect C-rich (HSat3) assay containing the flanking enzymatic cutting sites (pink) and siRNA-binding sites (blue). Additionally, the siRNA and ASO sequences are shown (m: 2′ MeO modification, ∗: thiophosphate modification). (C) Proof of concept using HeLa cells. Fluorescence microscope images showing green and red fluorescence signal in samples treated with targeting siRNA (siHSat3), untargeted siRNA control pool (siCo), and an untreated control. Images were taken 48 h after transfection.

In the following study, we describe the development, establishment, and limitations of the fluorescence-based cellular siRNA detection assay called FluoroDetect.

Results

Design of the FluoroDetect assay

FluoroDetect is a reporter assay system that measures RNA knockdown efficiencies with an internal reference signal. The reporter gene is under the control of a TET-OFF promotor system inhibited by doxycycline. The 3′UTR contains three identical siRNA target motifs, to enhance the sensitivity of the assay. For easy and fast adaptation of the assay, the target site is flanked by two individual restriction sites (AsiSI & NsiI). The addition of doxycycline or a targeting siRNA results in the loss of the reporter signal (Figure 1A). The assay plasmid was generated with hot fusion cloning starting with the pCW57.1-MCU-Flag TET-OFF plasmid (Figure S1).¹⁴

For a proof of principle, the target site HSat3 sequence was used. HSat3 is a repetitive long noncoding RNA (lncRNA) expressed under stress conditions like heat- or cytotoxic stress. It has been shown to be important in cancer progression and therapy resistance.¹³ Moreover, Valgardsdottir et al. could show that a C-rich and a G-rich lncRNA of HSat3 are transcribed.¹¹ Therefore, assays using the C-rich target motif as well as the G-rich target motif were designed and used for control experiments (Figure 1B). The response of the C-rich and the G-rich target sequence was investigated using fluorescence microscopy. For each assay (C-rich or G-rich), a respective siRNA was used. In both assays, a loss of the red reporter signal was observed 48 h after cotransfection of siHSat3 and the assay plasmid (Figure 1C). As expected, the green reference signal was unaffected by siHSat3 or an untargeted siRNA pool (siCo) treatment.

To demonstrate the adaptability of the assay, additional assays were generated: one assay containing a Heat Shock Factor 1 (HSF1) mRNA target site and one assay containing an Epidermal Growth Factor Receptor (EGFR) target site. Both target sites were created overlapping exon junctions to reduce off-target effects (Figure S2A). We could observe a strong sequence-specific knock-down in both assays (Figures S2B and S2C).

To exclude the presence of unwanted truncated, elongated, or fused mCherry and EGFP variants, a western blot was performed using a FLAG antibody and a GFP antibody to detect mCherry and EGFP. As loading control, a Histone H3-targeting antibody was used. For all three antibodies, bands of the expected size were detected (Figure S3A). This indicates that no unwanted byproducts are expressed, which could interfere with the assay measurement. Moreover, it was important to exclude unwanted fused mRNA transcripts. Therefore, a quantitative reverse-transcription PCR (RT-qPCR) experiment was performed in HeLa cells measuring the mRNA levels of mCherry and EGFP under different conditions (siHSat3, siCo, and untreated). It could be observed that mCherry mRNA is specifically reduced in the presence of HSat3 siRNA, whereas the EGFP mRNA level remained unaffected. In the siCo-treated and untreated samples, no significant changes in mRNA could be observed. A negative control was performed using wild-type HeLa cells (Empty). In all replicates, no fusion mRNA was detected. Observed changes were identical for both assays, HeLa C-rich, and HeLa G-rich (Figure S3B). EGFP and mCherry primer performance was evaluated using a dilution series of the assay plasmids. All qPCR primers pairs showed high performance (Figure S3C).

Based on the first positive evaluation using a fluorescence microscope, RT-qPCR, and western blot, the next step was the establishment of the assay in a plate reader system to facilitate the quantification of both fluorescence signals.

Plate reader analysis of siRNA knockdown

Excitation and emission for EGFP were 475 and 510/20 nm and for mCherry 575 and 610/20 nm, respectively (Figure S4A).^15,16 For all measurements, the window of detection is expected to be in the linear proportional range of cell number and fluorescence intensity. Thus, the normalized signal measured in relative fluorescence units (RFUs), given as the ratio of reporter-to-reference signal, should without siRNA treatments be stable across all datapoints in this window (Figures S4B and S4C).

In all transient experiments, the cells were first seeded in a 6-well plate. After 24 h of transfection, cells were transferred to a white-bottom tissue culture-coated 96-well plate (Figure 2A). The dilution series were performed with the cell lines HeLa, SW480, and SW620 in a range of 2,500–20,000 cells per well. Measurements of the cells were executed in 24 h intervals to investigate changes by time (Figures 2B, 2C, S5A, S5B, S6A, and S6B). The most stable signal was achieved in a range of 10,000–20,000 cells per well.

Figure 2.

Time curve of transient transfected FluoroDetect assay

(A) Schematic structure of experiments showing the procedure of transfection and measurements. Transfection was controlled using fluorescence microscopy, and fluorescence values were measured using a plate reading system. (B and C) Linear regression of EGFP (green) signal and mCherry (red) signal and the number of seeded HeLa cells per well. Determination of relative fluorescence units (RFUs) (yellow) in dependence of HeLa cells seeded per well. (D and E) Time curve of FluoroDetect transfected HeLa cells measured in 24 h intervals. Diagrams comparing the relative fluorescence units of untreated (green), siCo-treated (blue), and siHSat3-treated (red) cell population (all graphs showing the median ± 95% confidence interval [CI]).

Furthermore, after cotransfection of target-specific siRNAs, in both, the C-rich as well as the G-rich assay, a target-specific knockdown of the mCherry signal was observed up to 120 h after transfection. Control samples were unaffected. Differences and slight variability between the control samples could be explained by the variability in growth rate and unspecific effects on the siCo pool (Figures 3D, 3E, S5C, S5D, S6C, and S6D).

Figure 3.

Strand-specific response

(A) Fluorescence microscope images of HeLa cells transfected with the FluoroDetect G-rich (HSat3) assay before (t = 0 h) and 96 h (t = 96 h) after siRNA/ASO treatment. Populations shown are treated with siHSat3, siCo, ASO C, or ASO G. For comparison, an additional untreated population is shown. (B and C) Relative fluorescence units (RFUs) of HeLa cells transfected with FluoroDetect C-rich (HSat3) or FluoroDetect G-rich (HSat3) 72 h after treatment with siRNA/ASO as indicated below the diagram (all graphs showing the median ± 95% confidence interval [CI]). Only signifivcant comparisons are shown with p value denoted as follows: ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, ∗∗∗∗ p < 0.0001. For all tests, p value significance was defined as follows: not significant (n.s.) p > 0.05; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, ∗∗∗∗ p < 0.0001.

To investigate the assay’s performance and to show that the knockdown of the endogenous target is functioning also on the endogenous mRNA level, RT-qPCR measurements of HSat3 RNA level were performed. To induce HSat3 RNA expression, a 1 h heat shock (HS) at 44 °C was performed. RNA samples were collected 6 h after HS. As the double-stranded siRNA contains a G-rich as well as a C-rich strand, the RT-qPCR result (Figure S7A) was compared with the FluoroDetect C-rich HSat3 (Figure S7B) as well as with the FluoroDetect G-rich HSat3 (Figure S7C). Both measurements showed comparable knock down intensity.

Measuring strand specificity of the assay system

Due to the double-stranded nature of siRNAs, it was of interest if the FluoroDetect assay is able to specifically detect one of the siRNA strands while not being affected by the opposite strand. This specific information can give insights about the selection and preference of the guide strand. Usually, the RISC prefers the strand with the thermodynamically less stable 5′ end.^17,18 Especially in the development and screening of siRNA/siRNA conjugates, the differentiation between guide and passenger strand can give important information for the design of the delivery modification, to enhance the efficacy and to reduce unwanted off target effects. In addition, the capability of detecting single-stranded ASOs expands the application spectrum such that the assay is not limited to RISC-mediated silencing and can also detect intracellular microRNA-mediated gene silencing.

To measure strand specificity, single-stranded ASOs were designed matching the sequences of the single strands of the HSat3-targeting siRNA. ASOs were designed as gapmers in which RNA-like sequences flank the DNA oligonucleotide and contained 2′-oxymethyl and thiophosphate backbone modifications to increase their stability and efficacy (Figure 1B).^1,19 The hybridization of the DNA bases of the ASO with the mRNA target site (Figure 1B) forms a stable DNA/RNA hybrid double strand, which recruits RNaseH leading to the degradation of the mRNA.^20,21,22

Cells were transfected with siRNAs or ASOs 24 h after the transfection of the assay plasmids. Changes in fluorescence were investigated using a fluorescence microscope 24, 48, 72, and 96 h later. In HeLa cells transfected with the G-rich assay, a knockdown of the reporter signal was observed in the samples treated with siHSat3 and ASO C 48 h after transfection of the RNAs, but no reduction of the mCherry signal could be observed with the not-matching ASO G (Figures 3A, S8A, and S8B). This is exactly what was expected, as the complementary nature of RNAi implies that G-rich sequences interact with their complementary C-rich sequences. A quantification of the system confirmed a significant strand-specific mode of action (Figures 3B, 3C, S8C, and S8D).

Determination of the detection and quantification limits of the assay

The limit of detection (LOD) and limit of quantification (LOQ) are important parameters in analytical methods and are calculated by dividing the product of standard deviation of the response and expansion factor by the slope of the linear regression (Figure S9A).^23,24

The difference between LOD and LOQ in this approach is the different tolerance in alpha and beta errors when defining the expansion factor. To calculate the LOD and LOQ, a dilution series of the ASOs was performed. The total amount of ASOs was 30 nM in each transfection reaction to ensure constant transfection conditions. Measurement was carried out using a plate reader (Figure 4A). The generated dataset was used to perform a linear regression, which serves as calibration curve to calculate the LOD and LOQ (Figures 4B–4D). In cases the y-intercept of the calibration curve is larger than the standard deviation, it can be used instead of the standard deviation of the response. The y-intercept represents the response when the concentration of the analyte is zero. If the y-intercept is significantly larger than the standard deviation, it indicates that the baseline noise is low and the signal is more reliable because it takes into account all systematical errors and background signals of the specific experimental setup. The LOD and LOQ were measured for the C-rich and G-rich assay in three different cell lines. The overall LOD and LOQ of the assay were calculated as the median of the three cell lines HeLa, SW480, and SW620. It was observed that the C-rich assay was slightly more sensitive with a median LOD of 6.714 nM compared to the G-rich assay, which showed a LOD of 8.555 nM (Figures S9G–S10B; Table 1).

Figure 4.

Determination of assays detection boundaries

(A) Illustration of the transfection procedure. Transfection was controlled using fluorescence microscopy, and fluorescence values were measured using a plate reading system. (B and C) Knockdown intensity measured as relative fluorescence units (RFUs) in HeLa cells 48 h after treatment with different amounts (absolute amounts: 15, 7.5, 3.75, and 1.875 pmol) of strand-specific ASOs. Only signifivcant comparisons are shown with p value denoted as follows: ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, ∗∗∗∗ p < 0.0001. For all tests, p value significance was defined as follows: not significant (n.s.) p > 0.05; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, ∗∗∗∗ p < 0.0001. (D) Linear regression of relative knockdown intensity normalized to the untreated control population of FluoroDetect C-rich (HSat3) (blue) and FluoroDetect G-rich (HSat3) (green) 48 h after treatment (all graphs showing the median ± 95% confidence interval [CI]).

Table 1.

Values used for the calculation of the limit of detection and limit of quantification

Cell line	C-rich			G-rich
	HeLa	SW480	SW620	HeLa	SW480	SW620
Slope	.176	.958	4.328	.184	.778	4.548
Y-interceptor	14.600	12.010	6.476	20.310	7.704	11.790
LOD [nM]	6.714	7.994	4.938	10.838	4.400	8.555
LOQ [nM]	20.346	24.223	14.963	32.843	13.333	25.923
Median LOD ± SD [nM]	6.714± 0.1.535			8.555 ± 3.264
Median LOQ ± SD [nM]	20.346 ± 4.651			25.923 ± 9.891

LOD, limit of detection; LOQ, limit of quanitfication.

Generating stable cell lines for screening or repetitive measurement approaches

To generate stable cell lines, HEK 293 FT cells were transfected with the packaging, envelope, and transfer plasmids to generate lentiviruses for the reference or the reporter gene. Different cell lines (HeLa, MDA MB 231, SW480, and PC9) were co-transduced with both constructs to generate C-rich and G-rich stable reporter cell lines (Figure 5A). Reporter cells were investigated with a fluorescence microscope (Figure S10). Following the pilot experiment, cells were seeded to evaluate and compare their functionality with the transient reporter assay. Fluorescence was measured at the indicated time points (Figure 5A). It could be observed that all cell lines were responsive to the siRNA as well as to the respective ASO 72 h after transfection (Figures 5B–5E). Time curves showed that siRNA knockdown is stable for up to 2 weeks. Moreover, response to the ASOs was slightly faster compared to the siRNA whereas the siRNA lasted slightly longer (Figures 5F and 5G). For a cytometer-based readout, stable cells were transfected with an siRNA-targeting HSat3 (siHSat3) or an untargeted siRNA pool (siCo). As a negative control, untransfected reporter cells were measured. The experiment was performed using the cell lines MDA MB 231 C-rich, MDA MB 231 G-rich, PC9 C-rich, and PC9 G-rich. We first defined a population called “cells” to exclude particles of an inappropriate size. Next, cell doublets were excluded using the forward scatering area atter area (FSC-A) versus forward scatter height (FSC-H) channel. Using DAPI staining as a viability dye, dead cells were excluded based on increased signal in the VioBlue channel (Figures S11A–S11C).

Figure 5.

Evaluating stable reporter cell lines

(A) Experimental design of lentiviral transduction and selection of stable transduced reporter cells. (B–E) Relative fluorescence units (RFUs) of HeLa cells and SW480 cells stably expressing the FluoroDetect C-rich (HSat3) or FluoroDetect G-rich (HSat3) 48 h after treatment with siRNA/ASO. Only signifivcant comparisons are shown with p value denoted as follows: ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, ∗∗∗∗ p < 0.0001. (F and G) Time curves showing the dynamics of RNA interference (RNAi) of the stable transduced reporter cell lines (PC9 FluoroDetect C-rich [HSat3], MDA MB 231 FluoroDetect C-rich [HSat3]) measured as relative fluorescence units in a timescale up to 336 h after treatment (all graphs showing the median ± 95% CI).

In the final step, green and red fluorescence gates were defined using a fluorescence-negative sample as well as EGFP- and mCherry-expressing cells. All cell lines were measured 72 h after transfection (Figure S11D).

In all cell lines, the green fluorescence signal using the FITC-A channel remained stable (Figure 6). However, a small subset of cells appeared to lose the green fluorescence. The red mCherry signal was unaffected in the untreated and siCo-treated samples, while it was significantly reduced in the siHSat3-treated samples. This indicates that the siRNA targeting HSat3 effectively silences the expression of the target gene, leading to a reduction of mCherry signal in the transfected cells.

Figure 6.

Flow cytometer-based analysis of FluoroDetect assay

(A) Illustration showing the experimental design of the flow cytometer-based readout of the FluoroDetect assay system. Fluorescence of transduced reporter cells (MDA MB 231 C-rich & G-rich, PC9 C-rich & G-rich) was controlled using fluorescence microscopy. (B) Scatter plots of the mCherry fluorescence (PerCP-A channel) and EGFP fluorescence (FITC-A channel) of living single-cell populations. Shown are the samples treated with siHSat3 (red), treated with siCo (blue), and untreated (green). Gate showing the quarters Q1–Q4 (Q1: mCherry positive, Q2: mCherry & EGFP positive, Q3: EGFP positive, Q4: mCherry & EGFP negative). (C) Histograms of EGFP fluorescence (FITC-A channel) of gate Q2. (D) Histograms of mChery fluorescence (FITC-A channel) of gate Q2.

Discussion

For the continuous measurement of siRNA- and ASO-mediated knock-down over several hours, a dual fluorescence-based reporter assay was developed (Figures 1 and 2). The assay can be quantified using fluorescence microscopy or plate reading. It is capable of detecting strand-specific knock-down of ASOs down to a low nanomolar concentration (Figures 3 and 4; Table 1).

So far, only bioluminescence-based assays for the quantification of siRNA-mediated knockdown have been developed.^25,26 However, these assays are endpoint measurements and therefore labor and cost-intensive. The FluoroDetect assay is a dual fluorescence-based assay with EGFP as reference and mCherry as reporter. Measurements can be performed continuously, and live-cell imaging technologies can be applied. This makes the FluoroDetect particularly useful for drug development, as siRNA release processes can be precisely monitored and detected in a time-resolved manner. A first response of the siRNA cellular knockdown was measured 24 h after treatment with a peak at 72 h. This is in line with the findings of Bartlett and Davis who investigated the kinetics of siRNA-mediated gene silencing using bioluminescence.²⁷

The performance of the assay is even increased using a stable transduced assay system. This makes long-term experiments like elongated time curves and flow cytometry experiments possible. In particular, the flow cytometer-based analysis of the FluoroDetect assay gives information about subpopulations in the sample. This can be useful for pharmacological developments (Figures 5 and 6). However, for the FluoroDetect assay with EGFP and mCherry, flow cytometers with a violet (405 nm), blue (488 nm), and yellow (561 nm) laser line are recommended to ensure faithful separation of reporter and reference signal in two different laser lines.

As shown, the assay can also be used in a PiggyBac transposase-based system to generate stable reporter cell lines (Figure S3). The larger cargo capacity of PiggyBac transposase-based systems allows it to carry more complex and longer constructs. Secondly, PiggyBac poses a lower risk of insertional mutagenesis, as it tends to integrate into stable regions of the genome. Additionally, the production of PiggyBac transposons is more cost effective and simpler compared to the complex and expensive process of producing viral vectors. Another significant benefit is the ability of PiggyBac transposons not to rely on host cell factors for integration. This makes this system versatile across a wide range of cell types, including cells difficult to transduce and transfect.

The assay can be used to investigate cellular miRNA expression and can be combined with developmental or disease models. It can give insights which subpopulation of cells is expressing a specific miRNA in real time. For future experiments, cell line-derived xenograft mouse models or genetically engineered mouse models are planned to investigate the distribution of nucleic acid-based therapies in an in vivo model. In combination with the use of the fluorescence proteins from the far-red spectrum (e.g., mCherry as reference and miRFP720 as reporter), deeper tissues can be targeted in real time using an in vivo imaging system. Thus, the FluoroDetect assay can be a relevant tool for the monitoring and development of nucleic acid-based therapies.

Materials and methods

Vector design

All inserts were designed according to the hot fusion protocol described by C. Fu et al.¹⁴ The plasmid pCW57.1 Empty (N-FLAG, TetOFF) was created inserting a synthetic multiple cloning site oligo in the pCW57.1-MCU-Flag-TetOFF (Addgene plasmid #185657, RRID: Addgene 185657) generously provided by Dipayan Chaudhuri. A synthetic sequence of C-rich HSat3 flanked by AsiSI and NsiI restriction site sequences was cloned into the pmCherry-C3 (Clontech) generating pmCherry-C3 (HSat3 C-rich). The EGFP and mCherry (HSat3 C-rich) gene sequences were amplified by PCR from the plasmids pEGFP C1 and pmCherry-C3 (HSat3 C-rich). The generated PCR products were cloned into the pCW57.1 Empty (N-FLAG, TetOFF) generating the pFluoroDetect (C-rich HSat3). Oligos containing the HSat3 G-rich target motif (ordered from Integrated DNA Technologies, Inc. (IDT)) were ligated into the digested pFluoroDetect (C-rich HSat3) to generate pFluoroDetect (G-rich HSat3). For lentiviral transduction, single-color transfer plasmids were used. The transfer plasmids pCW57.1 TET-OFF mCherry (C-rich HSat3) and pCW57.1 TET-OFF mCherry (G-rich HSat3) coding the reporter gene were generated cutting the plasmids pFluoroDetect (C-rich HSat3) and pFluoroDetect (G-rich HSat3) to exclude the EGFP. A transfer plasmid coding only EGFP was cloned by inserting an EGFP sequence into the pCDH-3xFLAG-TERT (Addgene plasmid #51631, RRID: Addgene 51631) generously provided by Steven Artandi. To generate the transposon plasmids, a modified version of the PiggyBac transposon vector system (System Biosciences, REF PB513B-1) was used. The FluoroDetect (HSat3 C-rich and HSat3 G-rich) gene castes were transferred into transposon backbone flanked by a 5′ITR (Inverted Terminal Repeat)and a 3′ ITR. Plasmid maps of the used plasmids are contained in the supplemental material (Figures S1 and S12–S25).

Molecular cloning

For In-Fusion cloning, primer pairs were designed matching the desired inserts, as described by C. Fu et al. in 2014. Each primer had a 5′ homology overlap of 17–30 bp flanking the insert sites of thr destination backbone. Insert PCR products were separated on a 1% agarose gel (Carl Roth GmbH + Co. KG, Art. Nr. 3810.3) in Tris-Acetate-EDTA (TAE) buffer, followed by cutting out the band of the desired product size and cleaning up following the manual of the NucleoSpin gel and PCR cleanup kit (MACHEREY-NAGEL GmbH & Co. KG, ref. 740609.50). Linearization/restriction digest of the acceptor plasmid was performed using the appropriate restriction enzyme. All restriction enzymes were ordered from NEB. In-Fusion was performed in a final volume of 20 μL, in 0.2 M Tris pH 7.5, 0.2 M MgCl2, 2 mM dNTP’s, 0.2 M DTT, 5% PEG-800, 0.15 U T5 Exonuclease, and 1 U Phusion Hot Start Flex DNA polymerase. For all In-Fusion reactions, the molar ratios of 1:1, 1:2, and 1:4 of digested plasmid to insert were tested, and a total amount of 200 ng DNA was used. In-Fusion reaction was performed 1 h at 50°C followed by a ramp down to 20°C with a ramping speed of 0.1°C per second. In-Fusion reaction was stored at −20°C or immediately used for bacterial transformation. The plasmids were isolated using NucleoSpin plasmid mini kit (MACHEREY-NAGEL GmbH & Co. KG, ref. 740588.50) or NucleoBond plasmid Xtra Midi EF kit (MACHEREY-NAGEL GmbH & Co. KG, ref. 740420.50), when used for cell culture experiments. All generated plasmids were validated by sequencing Microsynth Seqlab GmbH (Microsynth). All primers and cloning oligos were ordered from IDT.

Cell culture

HeLa (RRID: CVCL0030), SW480 (RRID: CVCL_0546), SW620 (RRID: CVCL_0547), MDA MB 231 (RRID: CVCL_0062), HEK 293 FT (RRID: CVCL6911), and PC9 (RIDD: CVCL_B260) were used in this study.

The human cervical cancer cell line HeLa, lung cancer cell line PC9, and the colon cancer cell lines SW480 and SW620 were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (PAN-Biotech, P04-01500) supplemented with 10% (v/v) fetal calf serum (FCS) (Sigma-Aldrich, F7524), 100 U/mL penicillin, 100 μg/mL streptomycin (Sigma-Aldrich, P4333-100ML), and a final concentration of 2 mM L-Glutamine (Sigma-Aldrich, G7513-100ML). The human embryonic kidney cell line HEK 293 FT was cultured in DMEM GlutaMAX (Thermo Scientific, 10569010) supplemented with 10% (v/v) FCS (Sigma-Aldrich, F7524), 100 U/mL penicillin, and 100 μg/mL streptomycin (Sigma-Aldrich, P4333-100ML). All cell lines were tested negative for Mycoplasma contamination using the Mycoalert Plus Mycoplasma detection kit (Lonza, LT07-710). Cells were cultured at 37°C, 95% humidity, and 5% CO₂.

Transfection and lentiviral transduction

The transfection of the assay plasmids (pFluoroDetect HSat3 C-rich; pFluoroDetect HSat3 G-rich) and cotransfection of plasmids and siRNA/ASOs was performed using Lipofectamine 2000 (Invitrogen Inc., #11668027) according to the manufacturer’s recommendations. The medium was changed 24 h after transfection.

The used Piggy Bac transposon system was transfected using Lipofectamine 2000 (Invitrogen Inc., #11668027) according to the manufacturer’s recommendations. The medium was changed 24 h after transfection. Mass ratio of transposon plasmid to transposase plasmid was 2:1. 72 h after transfection, cells were selected using with puromycin dihydrochloride (5 μg/mL, Sigma-Aldrich, P8833-25MG) and G418 (400 μg/mL, InvivoGen, ant-gn-1). Selection medium was changed in 96 h intervals.

Additionally, single transfections of siRNAs and ASOs were performed using Lipofectamine RNAiMAX reagent (Invitrogen Inc., #13778030) according to the manufacturer’s recommendations. The transfected ASOs shared the sequence with the siRNA strands but were chemically modified. The sequences of siRNA/antisense-oligos targeting HSat3 sequences are provided in Figure 1. In the case of ASO dilution series, it is important to dilute the targeted siRNA/ASO with an untargeted control RNA to minimize systematic errors caused by different strong repulsion rates between the samples. Sequences of ASOs targeting HSF1 or EGFR are given in Table S1.

For viral transductions the plasmids, psPAX2 (gifted from Didier Trono, Addgene plasmid #12260; http://n2t.net/addgene:12260; RRID: Addgene_12260) and pMD2.G (Gifted from Didier Trono, Addgene plasmid #12259; http://n2t.net/addgene:12259; RRID: Addgene_12259) were used as packaging plasmid and transfected with PEI MAX (Polysciences, #24765(1), CAS Number: 49553-93-7). The packaging plasmids were cotransfected with the transfer plasmid pmCherry HSat3 C-rich, pmCherry HSat3 G-rich or pCDH EGFP (Neo). The transfection media was replaced with fresh media after 24 h. The lentiviruses were harvested 48 and 72 h after transfection. Lentivirus containing media was centrifuged at 500 × g for 10 min followed by passing through a filter membrane with a 0.45 μm pore size (SARSTEDT, REF: 83.1826). Virus-containing medium was stored at −80°C and supplemented with a final concentration of 8 μg/mL polybrene (Merck, TR-1003-G) directly prior to transduction. Cells were incubated 72 h with the transduction mix followed by a selection with puromycin dihydrochloride (5 μg/mL, Sigma-Aldrich, P8833-25MG) and G418 (400 μg/mL, InvivoGen, ant-gn-1). The selection medium was changed in 96 h intervals.

Fluorescence imaging

For fluorescence imaging, the cells were transferred to the microscope at the indicated time point in a transparent bottom and tissue culture-coated 6-well plate (TPP Techno Plastic Products AG, Prod. No.: 92006). Imaging times were tried to be as short as possible to avoid irregularities in culture conditions. All fluorescence images of transiently transfected cell lines were taken by Keyence BZ-X800E/BZ-X810 (Keyence Deutschland GmbH) and the CFI Plan Apo λ4x objective (NA0.20 WD20.00 mm) (Keyence, 972030). Fluorescence filter sets were BZ-X filter GFP (Ex.470/40, Em.525/50) (Keyence, OP-87763) for green channel and BZ-X filter TRITC (Ex.545/25, Em.605/70) (Keyence, OP-87764) for the red channel.

All fluorescence pictures of cells in Biosafety Level 2 (S2) were taken by using the Nikon ECLIPSE fluorescence microscope.

Plate reading

The cells were seeded and incubated for 24 h at 37°C, 95% humidity, and 5% CO₂. Cells were transfected as described earlier and incubated for 24 h. For measurement with the Tecan Infinite 200 PRO micro plate reader (Tecan Germany GmbH), cells needed to be transferred to a tissue culture coated 96 well flat bottom assay microplate (Falcon, ref. 53296). Therefore, cells were washed once with 1 mL PBS and trypsinized. Indicated number of cells were seeded in 100 μL cell culture medium per well in a 96-well flat-bottom assay microplate.

Flow cytometry

The cells were seeded and incubated for 24 h at 37°C in a humidified incubator with 5% CO₂. Cells were transfected as described earlier. For measurement with the MACSQuant Analyzer 16 flow cytometer (Miltenyi Biotec B.V. & Co. KG) cells were suspended into FACS buffer (2% FCS, 1 mM ethylene glycol-bis-(2-aminoethyl)-tetra-acetic-acid [Carl Roth, Product No.: 3054.3], 4 mM ethylenediaminetetraacetate [EDTA, Carl Roth, Product No.:8040.2] in PBS). For this purpose, adherent cells were washed once with PBS followed by 10 min incubation at 37°C with 500 μL FACS buffer per well. The detached cell suspension was passed through a CellTrics 30 μm cell strainer (Sysmex, REF: 04-0004-2326) in a 5 mL polystyrene round-bottom tube (Falcon, REF: 352054) to isolate single cells. The cells were chilled on ice till analysis. Right before measurement, cells were incubated with a final concentration of 10 ng/mL DAPI (Thermo Fisher Scientific Inc, REF: 62248) to stain dead cells.

RT-qPCR

RNA was isolated using the NucleoSpin RNA kit (MACHEREY-NAGEL GmbH & Co. KG, ref. 740955). Subsequently the cDNA was synthesized using SuperScriptIV reverse transcriptase (Thermo Fisher Scientific Inc., ref. 18090200) with random hexamers (Metabion, customized) and HSat3-specific primer pairs as described by Valgardsdottir et al..¹¹ All cDNA reactions were incubated with RNase H (New England Biolabs GmbH, REF: M0297L) to degrade remaining RNA templates. Each qPCR reaction contained 5 ng of cDNA and 300 nM gene-specific qPCR primers and was performed in 1x GoTaq qPCR MasterMix (Promega GmbH, REF: A6002) at a final volume of 10 μL. Cycling parameters were chosen according to the manufacturer’s protocol. qPCR was performed using QuantStudio5 (Thermo Fisher Scientific Inc.). For all samples an RT-minus control sample was measured to exclude DNA contamination. The cDNA and qPCR primer sequences are provided in Table S1. All primers were ordered from IDT.

Western blot

Lysis of cells was performed using Pierce lysis buffer (25 mM Tris–HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% NP-40, and 5% glycerol) containing a protease inhibitor cocktail (cOmplete, Roche Diagnostics, ref. 11836170001). The proteins were separated by SDS-PAGE using a precast polyacrylamide gel (mPAGE 4–12% Bis-Tris, Merck KGaA, REF: MP41G15) with a MOPS SDS running buffer (Merck KGaA, REF: MPM0PS) at 200 V for 36 min. Transfer on nitrocellulose membrane was performed in mPAGE transfer buffer (Merck KGaA, REF: MPTRB) at 70 V for 2 h and 4°C. Afterward, the membrane was blocked in 5% skim milk-TBST for 1 h at room temperature (RT). Incubation with primary antibodies (ANTI-FLAG M2, Merck KGaA, REF: F1804, RRID: AB_262044; Anti-GFP-Antibody Clone 3F8.2, Merck KGaA MAB1083, RRID: AB_1587098, Histone H3 Clone D2B12 XP, Cell Signaling Technology, Inc., ref. 4620S, RRID: AB_1904005) was performed at 4°C over night. Subsequently, the membrane was washed three times (10 min each with TBS-0.1% Tween [Tween 20, Carl Roth GmbH + Co. KG, REF: 9127.1]) and incubated with the horseradish peroxidase-conjugated secondary antibody (anti-mouse immunoglobulin G [IgG], Cell Signaling Technology, Inc., REF: 7076, RRID: AB_330924; Anti rabbit IgG, Cell Signaling Technology, Inc., REF: 7074, RRID: AB_2099233) for 1 h at RT. Chemiluminescence was performed using Western Lightning Plus-ECL Substrate (Perkin Elmar LAS, REF: NEL104001EA). Used protein ladder was ordered from Thermo Fisher Scientific Inc. (REF: 26620).

Vector maps and digital cloning

Vector maps and digital cloning was planned and performed using the software package SnapGene (SnapGene, RRID: SCR_015052). Plasmid maps are available in the supplemental material.

Statistical analysis

Statistical analyses were performed with the software package GraphPad Prism 9 (GraphPad Software, RRID: SCR_002798). The type of statistical analyses, parameters, and the number of replicates are indicated in the figure legends. For all tests, p value significance was defined as follows: not significant (n.s.) p > 0.05; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Schematic figures

All schematic figures were created with BioRender and Adobe Illustrator.

Data availability

The data supporting the findings of this study are available in the Supplemental Material and additional details upon request from the corresponding author, Michal R. Schweiger.

Acknowledgments

We thank all the reviewers for constructive criticism and tedious reviewing. We thank the research group of Prof. Roman Thomas, especially Nina Wobst for her support and the discussion during the developmental process as well for the usage of the TECAN Infinite 200 PRO plate reader & MACSQuant Analyzer flow cytometer. Moreover, we thank Prof. Ines Neundorf for her supervision in biochemical aspects. All schematic figures were created with BioRender.

The study was funded by the German Research Foundation: SFB 1530 to M.R.S..

Author contributions

F.H. and M.R.S. designed the study and coordinated experiments and data analysis; F.H. and E.W.-Z. performed the experiments; F.H. designed the assay system, assay plasmids, and the antisense oligonucleotides; F.H., C.G., and M.R.S. analyzed the data; F.H. and M.R.S. wrote the manuscript with all comments from all authors; M.O. gave supervision in lentiviral transduction.

Declaration of interests

The authors declare no competing interests.