En jämförande studie av qPCR, western blot och masspektrometri för bestämning av proteinkoncentrationer

(1)

IN

DEGREE PROJECT BIOTECHNOLOGY,

SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2016,

A Comparative Study of qPCR, Western Blot and Mass

Spectrometry for the Estimation of Protein Concentrations

MARIA EDVARDSSON

KTH ROYAL INSTITUTE OF TECHNOLOGY

(2)

(3)

Sammanfattning

Inom proteomik är det av stort biologiskt och kliniskt värde att kunna precist kvantifiera proteiner i ett komplext prov. Men detta är en utmaning på grund av det stora dynamiska omfånget av proteinkoncentrationer, vilket påvisar behovet av avancerade kvantifieringstekniker. De masspektrometri- och affinitetsbaserade teknikerna med respektive reagens måste vara både känsliga och specifika, vilket måste säkerställas genom validering av dessa reagens. Dessutom används mRNA-koncentrationer för att förutsäga proteinkoncentrationer för icke detekterbara proteiner. Emellertid har korrelationen mellan mRNA- och proteinnivåer ifrågasatts. I denna studie har kvantitativ polymeraskedjereaktion, western blot och masspektrometri jämförts för kvantifiering av RBM3, ATAD2, ANLN och VIM i ett siRNA- knockdownsystem i U251-cellinjen. Jämförelsen möjliggjorde en validering av Atlas Antibodies antikroppar och QPrESTar, samt utvecklandet av en masspektrometrisk assay för målproteinerna. I allmänhet, konstaterades att metoderna korrelerade bra i sina uppskattningar av proteinkoncentrationerna för de studerade målproteinerna, vilket påvisar en reproducerbarhet hos antikropparna och QPrESTarna.

Dock måste detta resultat verifieras med en större kohort av replikat och den kvantitativa analysen optimeras. Nästa steg skulle vara att analysera metoderna i ett inducerat proteinexpressionssystem.

Abstract

It is of great biological and clinical value within proteomics to be able to accurately quantify proteins in a complex sample. However, this is a challenge due to the large dynamic range of protein concentrations, demonstrating the need for advance quantification techniques. The mass spectrometry- and affinity-based techniques with their respective reagents need to be both sensitive and specific, which has to be ensured by validation of these reagents. In addition, mRNA abundances have been used to predict protein abundances for non-detectable proteins. However, the RNA-protein level correlation has been questioned. In this study, quantitative real-time polymerase chain reaction, western blot and mass spectrometry have been compared for the quantification of RBM3, ADAT2, ANLN and VIM in a siRNA knockdown system using the U251 cell line. Via the comparison, the antibodies and the QPrESTs commercialized by Atlas Antibodies were validated and a mass spectrometry assay developed. In general, it was found that the methods correlated well in their concentration estimations, further demonstrating the reproducibility of the antibodies and the QPrESTs. However, these results need to be verified with a larger cohort of replicates and the quantitative protocol needs to be optimized. Next step would be to analyze the methods in an induced protein expression system.

(4)

1. Introduction

1.1 Aim and objectives

Before a proteomic experiment, it is of importance to consider weather the intended technique is appropriate for the proteomic question at hand. The aim of this project was to give a better insight into appropriate proteomic applications of western blot (WB), mass spectrometry (MS) and real-time quantitative polymerase chain reaction (qPCR) and the reagents used in each. The main objective was to investigate and compare qPCR-estimated mRNA concentrations, WB- and MS-estimated protein concentrations to determine how well the methods correlated in their results and if they were possibly needed in combination to achieve high-quality concentration determinations. Further, via the comparison, also to validate Atlas Antibodies monoclonal antibodies and the QPrESTs used and to develop a MS- based assay for the target proteins studied. The four target proteins that were used to compare the techniques are called RBM3 (RNA-binding motif protein 3), ANLN (Anillin), ATAD2 (ATPase family AAA domain-containing protein 2) and VIM (Vimentin), where the two former are interesting potential oncology biomarkers and the two latter are biologically and/or clinically interesting targets corresponding to two best-seller antibodies by Atlas Antibodies. The target proteins were analyzed in a siRNA knockdown system, using two siRNAs per target protein in the U251 cell line. The difference in expression of siRNA knocked down protein and normally expressed protein enabled the comparison between the techniques. The principle behind siRNA knockdown is that the activated siRNA binds the target transcript and degrades it, resulting in a lower protein expression [1].

1.2 Theory

1.2.1 Importance of the proteomics field and technical challenges

Proteomics, the study of proteins in a systemic and large-scale manner, is a fundamental practice that enables an increased understanding of a complex biological system. Knowledge about the function of proteins is critical in order to understand processes such as disease on a molecular level to develop efficient diagnostic and treatment strategies. Certain proteins present at a particular disease state can be utilized as biomarkers in for instance prognostics, diagnostics or treatment monitoring. Proteins used as biomarkers are often present in a significantly higher or lower concentration in the disease state in contrast to the healthy state. However, disease-related proteins are often present in lower abundances in comparison to the rest of the proteome, meaning more highly abundant proteins easily mask them. The large dynamic range of protein concentrations is a major challenge in the field of proteomics. The fact that the difference in protein quantity is heavily utilized within the biomarker area, demonstrates the importance of accurate protein quantification techniques. [2][3]

The proteomics field is further divided into MS- and affinity-based proteomics. The name of the subdivisions originates from the type of technique used to study proteins, each with their distinctive advantages and drawbacks. WB, which utilizes antibodies to detect and quantify proteins, has been used for decades and can be considered the state of the art tool among the affinity techniques available. [2][3]

First, proteins are separated using Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS- PAGE) before being transferred to a nitrocellulose membrane. To the blocked membrane, a primary antibody targeting the protein of interest is added and a labeled secondary antibody is used for detection [4][5][6]. Commonly, enzymatically labeled secondary antibodies are used, enabling a light signal once the chromogenic/fluorogenic/chemilumniscent substrate is added as a result from an enzymatic reaction, which is detected by the image instrument [7]. The main advantage with the technique is that it is simple and very sensitive due to the evolutionary evolved affinity of the antibodies. However, there are problems with cross-reactions and unspecific binding. The fact that the technique relies on antibodies requires that there are antibodies available for the particular target protein under study; otherwise the technique will be useless. MS-based techniques are not dependent on available antibodies, meaning that proteins lacking an affinity reagent can be analyzed. MS is dependent on the ionization of the proteins, as the resulted mass to charge ratio is used to identify and quantify them. [8][9] MS-based proteomics suffers from low sensitivity, but is superior in specificity over affinity-based proteomics. The goal within proteomics is both high sensitivity and specificity, further suggesting that the idea of combining MS- and affinity-based proteomics would partly solve the issue in the future, which is a trend seen in the proteomics field today [2][3].

(6)

1.2.2 Mass spectrometry-based proteomics

MS is the most universal technique used within the field of proteomics. An MS instrument consists of three parts: the ion source, mass analyzer and the detector. The ion source allows for the ionization of the sample, the mass analyzer separates the generated gas phase ions according to their mass-to-charge ratio and the detector finally registers the separated ions. There are a number of different types of ion sources, mass analyzers and detectors. The two most common ion sources used in proteomics are the Matrix Assisted Laser Desorption Ionization (MALDI) and the Electrospray Ionization (ESI). MALDI is used for the ionization of simple peptide samples, whereas ESI is used to ionize more complex samples. ESI is often on-line coupled to a liquid chromatographic (LC) step before ionization to achieve better separation of the complex sample. Frequently used mass analyzers are time-of-flight, orbitrap, quadrupole, iontrap and fourier transform ion cyclotron resonance. Mass analyzers vary in mass limit, resolution and accuracy suggesting that a tradeoff needs to be made depending on the research question. In MS-based proteomics the sample preparation is important for the MS analysis and the final result. From a complex mixture, proteins are extracted and digested into peptides, which are easier to ionize than full-length proteins. The digestion is performed using a protease. There are different types of proteases, but commonly trypsin is used, cleaving the protein at lysine and arginine. The cleavage reaction has been demonstrated to be a major source of error. The detected peptides then need to be targeted to the right protein. Peptides with similar masses can be distinguished by the introduction of a fragmentation step, as knowledge about the peptide sequence is acquired, which is denoted as MS/MS analysis. [3][10][11][12]

In quantitative proteomics there are two types of strategies, relative quantification and absolute quantification. Relative quantification refers to the quantification of proteins between samples, whereas absolute quantification offers additional information in relation to relative quantification, namely the quantification of proteins within the same sample. In MS-based proteomics there are different types of isotope labeled internal standards that can be used for absolute targeted quantification. There are full- length protein standards, protein fragment standards and peptide standards. A problem with peptide standards is that they are spiked in after digestion; meaning digestion errors are not controlled for. Full- length proteins can be expensive to produce, but can be added before protein fractionation. The principle behind these standards is that they are heavily labeled, resulting in a mass shift between the standard and the light endogenous protein peptides (Figure 1). The intensity of the MS-peaks generated can be compared and since the concentration of the standard is known beforehand, the absolute concentration of the endogenous protein can be determined. The reason why peak intensities cannot be directly used to determine the concentration is because ionization efficiency differs between peptides. [2][3] In this study, one such type of internal standard was used, the so-called QPrEST, Quantitative Protein Epitope Signature Tag. QPrESTs are protein fragments heavily expressed in an Escherichia coli strain auxotroph for arginine and lysine, which are added as heavy isotopes to the cultivation medium. QPrESTs are not as expensive to produce as full-length protein standards and can be added before digestion, which peptide standards cannot. [2][3][13][14][15]

Figure 1. A scheme illustrating the principle of using the QPrEST standard for the quantification of an endogenous protein [15].

(7)

1.2.3 Human Protein Atlas project and Atlas Antibodies

The outcome of the Human Protein Atlas (HPA) project systematically identifying the expression of all human proteins in a tissue-based manner from a gene-centric perspective using antibody proteomics was the HPA database, launched in 2003. The major focus of the project was to identify protein expressions in normal versus cancer tissue, but expanded to also include a subcellular atlas and a cell line atlas. The project did not only result in a publically available database, but also in a valuable resource of antibodies targeting almost all human proteins that soon became interesting for other researchers. [16][17] The large interest in the antibodies generated from the project lead to the establishment of the spinoff company Atlas Antibodies in 2006, selling the validated HPA polyclonal antibodies worldwide [18].

The in-house developed large-scale production of antibodies by the HPA project has been adopted by Atlas Antibodies (Figure 2). The workflow initiates with the construction of PrEST-antigen for antibody generation, 50-150 amino acids long protein fragments identical in sequence to the endogenous protein, but unique in terms of homology to other human proteins to reduce cross-reactivity. The PrEST sequence is reverse transcribed PCR-amplified from an RNA-template and ligated into a vector containing a His6- ABP-tag, where the Histidine-tag is utilized for Immobilized Metal Affinity Chromatography (IMAC)- purification and the Albumin Binding Protein-tag for increased immunogenicity and solubility. The vector sequence is confirmed with Sanger sequencing and expressed in E.coli. The resulting PrESTs are purified and immunized into the host animal. Serum from the host animal is collected and affinity-purified on PrEST-coupled columns. Validation of purified antibodies is done using PrESTs immobilized on a microarray and WB. The antibodies are then used in different proteomic applications to study the protein expression, such as immunohistochemistry, immunofluorescence etc. [2][16][17] Validation of affinity reagents is important in order to achieve reproducibility of proteomic experiments. Unfortunately, affinity proteomics suffers from a lot of studies that cannot be replicated. The affinity reagents commercialized must be thoroughly characterized so that researchers can derive meaningful outcomes from proteomic experiments. Except for the HPA-derived antibodies, Atlas Antibodies started their own in-house production of monoclonal antibodies in 2012, some of which were used and validated in this study [18].

The major advantage with monoclonal antibodies in contrast to polyclonal antibodies is that no immunization is required once the antibody-producing hybridoma cells are created [3].

Figure 2. A flow chart describing the workflow of the HPA project. [19]

Atlas Antibodies’s newly launched QPrEST reagents originate from the light antigen used for production of antibodies in the HPA project. The QPrEST reagent with the > 99 % incorporation of heavy isotopes into the PrEST sequence is used as a multipeptide internal standard for MS-based quantification. Each

(8)

QPrEST generates a minimum of 2 unique tryptic peptides resulting in several signals that can be used for absolute quantification of proteins in a singleplex or multiplex manner. The QPrEST is spiked into the sample early in the process, reducing sample preparation-related errors, such as digestion. The QPrEST is identical to a part of the endogenous protein and about 13 000 human proteins are targeted by the QPrESTs. The tryptic peptides generated from the N-terminal Qtag enables accurate concentration determination of the QPrEST in an MS-based quantification. The light Qtag protein with its concentration determined with a highly accurate amino acid analysis is used as reference. [15][20]

1.2.4 Correlation between mRNA and protein concentrations

qPCR in contrast to MS and WB, do not measure protein levels directly but indirectly via mRNA abundances. cDNA is produced from transcripts by reverse transcription, and these are subsequently real- time qPCR-amplified. [21] Currently, there is a hot debate within the proteomics field weather mRNA levels correspond to protein levels in complex biological samples. This question has been raised due to recent advances of MS- and sequencing techniques, enabling studies of the transcriptome and the translatome, respectively, and also due to the fact that researchers commonly utilize changes in gene expression as means for predicting changes in protein abundances. [22][23][24] Generally, researchers do not agree about the mRNA-protein level correlation question. There are studies stating that there is a substantial correlation [25][26][27], but also studies indicating the opposite [28][29][30]. The combined results from the studies investigating this question are thus inconclusive, pointing towards a need for further knowledge to obtain biologically and clinically valuable proteomic data in the future. The mRNA- protein correlation seems to be greatly dependent on the type of proteomic research question, experiment and the techniques used, indicating that in certain cases it can be applicable and in certain cases it cannot.

A number of factors have been demonstrated to affect the mRNA-protein correlation; these are both technical and biological factors. Technical factors include experimental and statistical errors, whereas biological factors include regulatory RNAs and proteins, RNA secondary structure, codon bias, ribosome occupancy, protein stability etc. [22][23][24]

2. Materials and methods

2.1 Mammalian cell cultivation

Culture media was prepared according to ECACC Cell line Data Sheet for U-251 (cat no. 09063001). To Eagle’s Minimum Essential (EMEM) culture medium (Sigma Aldrich), 2 mM Glutamine (Sigma Aldrich), 1 % Non-Essential Amino Acids (Sigma Aldrich), 1 mM Sodium Pyruvate (Sigma Aldrich) and 10 % Foetal Bovine Serum (Sigma Aldrich) were added. Thawed human glioblastoma astrocytoma cells were cultivated in 20 ml cultivation medium in a T75 flask for adherent cells at 37 ℃ in 5.0 % CO2

environment and continuously subcultured at 70-80 % confluence. 1 ml 0.25 % Trypsin/EDTA (Sigma Aldrich) was used per 75 cm² surface area and incubated for 5 min at 37 ℃ to detach the cells. The seeding was performed in the interval of 2-4x10⁴cells per cm².

2.2 siRNA knockdown of target proteins 2.2.1 siRNA knockdown optimization

The siRNA knockdown was optimized before the actual knockdown of all the target proteins was implemented. The siRNA knockdown optimization was performed for the target protein RBM3 in three different volumes of transfection reagent: 2, 4 and 6 µl respectively for a cell density of 4x10⁴ cells per cm² in 2 ml cultivation medium. The amount of seeding was determined before testing the different amounts of transfection reagent. The following cell densities were tested: 2x10⁴, 4x10⁴ and 8x10⁴cells/cm², of which 2x10⁴ cells/cm² was considered an adequate cell density and 1x10⁴ cells per cm²an optimal density.

Cells were transfected using Lipofectamine 2000 according to Lipofectamine 2000 DNA Transfection Reagent Protocol 2013 from Invitrogen by Life Technologies. Three different controls were used in the experiment, namely; KIF11-siRNA as a positive control, it causes mitotic arrest meaning that transfected cells appear globular rather than flat when down-regulated, “scrambled” ctrl siRNA as a negative control and untransfected cells to control for transfection-related cellular alterations. Ambion validated Silencer®

Select Pre-designed siRNA products for RBM3 were purchased from Life Technologies (RBM3: siRNA

#1 (id.no s11858)). Initially, cells were seeded to 70-90 % confluence at time of transfection

(9)

corresponding to a cell density of approximately 4x10⁴ cells per cm² in 2 ml cultivation medium. A 1:3330 Opti-MEM siRNA solution was mixed with a 1:20 Opti-MEM lipofectamine solution into a 1:1 ratio and incubated for 5 min before the DNA-lipid complexes were added to the cultured cells. The cells were incubated for approximately 3 days before transfected cells were visually inspected. Transfected cells were washed in PBS buffer (Sigma Aldrich) before RIPA buffer (50 mM Tris-HCl, pH 7.6, containing 150 mM NaCl, 1 % NP-40, 0.5 % Sodium Deoxycholate, 0.1 % Sodium Dodecyl Sulfate (SDS) and 1 tablet Complete Protease Inhibitor per 10 ml) was added. The cells were incubated for 5 min on ice before harvest and the resulting cell lysate was stored at - 20 ℃ for subsequent analysis. A Bicinchoninic acid (BCA) assay was used to determine the total protein concentration; the same used in the actual siRNA knockdown. WB was used to analyze the difference in expression of the protein in transfected cells versus untransfected cells (Appendix, Figure 7); the same used in the actual WB analysis. The siRNA knockdown optimization indicated that 2 µl of transfection reagent to a cell density of 1x10⁴ cm² was enough for efficient transfection.

2.2.2 siRNA knockdown and cell harvest

2 µl transfection reagent at a cell density of 1x10⁴ cells per cm²was used in the actual knockdown of the target proteins RBM3, ATAD2, ANLN and VIM using the same transfection protocol and conditions as for the optimization experiment, but scaled up to 20 ml cultivation medium. Ambion validated Silencer®

Select Pre-designed siRNA products were purchased from Life Technologies (RBM3: siRNA #1 (id.no s11858), siRNA #2 (id.no s11859), ATAD2: siRNA #1 (id.no s26394), siRNA #2 (id.no s26395), ANLN:

siRNA #1 (id.no s28982), siRNA #2 (id.no s28984), VIM: siRNA #1 (id.no s14798), siRNA #2 (id.no s14800)). However, the harvest of the transfected cells was achieved using trypsinization instead and the cell harvests were prepared for subsequent WB, qPCR and MS analysis, accordingly. After trypsinization, cells were centrifuged at 150 x g for 5 min in aliquots of 2-3 million cells and the medium-containing supernatant discarded. The resulting pellet was resuspended in PBS and centrifuged at the same settings.

The washing procedure was repeated twice. 200 µl and 500 µl cells resuspended in PBS were taken out per siRNA for qPCR and WB sample preparation, respectively, for every target protein. The rest of the cell pellet (250 000 cells) was kept for MS sample preparation. In the WB sample preparation, 200 µl RIPA buffer was added to the cell pellet and incubated on ice for about 5 min. The cell lysate was centrifuged for 10 min at 16 000 rpm in a tabletop centrifuge and the supernatant was stored at - 20 ℃. The total protein concentration in the WB cell lysate was determined using a BCA assay (Bio-Rad) according to manufacturer’s recommendations. The 2 mg/ml BSA standard was diluted to 1.0, 0.8, 0.6, 0.4, 0.2, 0.1 and 0.05 mg/ml in PBS for the preparation of the standard curve. Cell lysate was diluted 1:2. Concentration measurements were performed in a plate reader from Thermo at 570 nm. To the qPCR cell pellet, 350 µl RLT buffer (Qiagen) containing 40 mM Ditiotreitol (DTT) (Sigma) was added. The cell lysate was transferred to QiaShredder tubes (Qiagen), centrifuged for 30 s at 8000 x g and the flow through was stored at - 20 ℃. The MS cell pellet was resuspended in 600 µl ice cold 100 mM Tris-HCl, to which 400 µl lysis buffer (4 % SDS, 100 mM Tris, 10 mM DTT) was added and incubated at 95 ℃ for 5 min at 600 rpm in a thermo mixer. The cells were sonicated for 1 min at 50 % amp. (1 s pulse, 1 s hold) using a Bundelin Sonopuls sonicator and the cell lysate was stored at - 20 ℃. The total protein concentration in the MS cell lysate was determined using the Biorad Protein Measurement assay, BioRad DCC, according to manufacturer’s recommendations. The 16 mg/ml BSA standard was diluted to 1.6, 0.8, 0.4, 0.2 and 0.1 mg/ml in a 1:10 diluted lysis buffer. Cell lysate was diluted 1:10. Concentration measurements were performed in a plate reader from Thermo at 570 nm.

2.3 QPrEST production 2.3.1 Transformation

The QPrEST plasmid was transformed into freeze competent ∆Arg ∆Lys auxotroph Rosetta E.coli cells, where the QPrEST plasmid encodes for Kanamycin (Kan) resistance and the Rosetta plasmid for Chloramphenicol (Cam) resistance. 4 µl plasmid DNA was added to 40 µl cells and incubated for 5 min on ice, before the cells were heat-shocked for 30 s at 42 ℃ and incubated on ice for another 2 min. After 80 µl Super Optimal Broth (SOC) media (Novagen) was added, the cells were incubated at 37 ℃ at 250 rpm in a Minitron incubator (Infors) for an hour before streaking out the cells on a Kan and Cam

(10)

selective agar plate with antibiotic concentrations of 50 µl/ml and 10 µl/ml, respectively. The plate was incubated at 37 ℃ overnight.

2.3.2 Inoculation, overexpression and harvest

Single colonies were picked from the agar plate and inoculated in 5 ml TSB (0.3 mg/ml Tryptic Soy Broth, 5 µg/ml Yeast Extract, MQ) to which Kan and Cam had been supplemented, with the antibiotic concentrations of 50 µl/ml and 34 µl/ml, respectively. Cells were cultivated overnight at 37 ℃ at 150 rpm in a Minitron incubator (Infors). 10 µl of cultivated cells were transferred into 10 ml of QPrEST medium (500 mM Na2HPO4, 500 mM KH2PO4, 250 mM (NH4)2SO4, 5 % Glycerol, 0.5 % Glucose, 2 % Lactose, 200 mM MgSO4, 50 mM FeCl3, 20 mM CaCl2, 10 mM MnCl2, 10 mM ZnSO4, 2 mM CoCl2, 2 mM CuSO4, 2 mM NiSO4, 2 mM (NH4)6Mo7O24, 2 mM Na2SeO3, 2 mM H3BO3, 200 µg/ml Heavy isotope labeled (¹³C and ¹⁵N) lysine and arginine (Cambridge Isotope Laboratories), 200 µg/ml of each remaining light amino acid (Sigma-Aldrich)) complemented with the same Kan and Cam concentrations as for the inoculation, to start the protein overexpression and incubated for another 19-22 hours. The protein overexpression was autoinduced by lactose at glucose depletion. To harvest the cells, the cultivated cells were centrifuged at 2700 x g, 10 min, 4 ℃ in a fixed angle rotor (F13-14x50C) and the resulting medium- containing supernatant was discarded. To vortexed cell pellet, 5 ml IMAC lysis buffer (7 M Guanidiniumchloride, 47 mM Na2HPO4, 2.65 mM NaH2PO4, 10 mM Tris-HCl, 100 mM NaCl, pH set to 8.0 with NaOH) supplemented with 7 µl 20 mM β-mercaptoethanol (Sigma), was added. The cells were lysed during incubation at 37 ℃, at 150 rpm in a Minitron incubator (Infors) for 2 hours. Cell debris was discarded by centrifugation at 17 100 x g, 40 min at 4 ℃ in a fixed angle rotor (F13-14x50C).

2.3.3 Protein purification

The QPrESTs were IMAC purified, using 2 ml of mixed HisPur^TM cobalt resin slurry from Thermo Scientific, resulting in a column volume of 1 ml using PD10 flowthrough columns. 5 ml IMAC wash buffer (6 M Guanidiniumchloride, 46.6 mM Na2HPO4, 3.4 mM NaH2PO4, 300 mM NaCl, pH set to 8.0- 8.2 with NaOH) was added to dilute the lysate. The affinity gel was equilibrated with 2 column volumes of IMAC wash buffer before the lysate was loaded onto the gel. The resuspended gel was left to sediment for approximately 20 min, after which the unbound protein was allowed to flow through and the gel was subsequently washed with 100 column volumes of wash buffer. The gel was resuspended in 3 column volumes of elution buffer (6 M Urea, 50 mM NaH2PO4, 100 mM NaCl, 30 mM Acetic acid (99%), 70 mM NaAc pH 7, pH set to 4.9-5.0) and left to sediment for 10 min before eluting the QPrEST. The purified QPrEST was diluted in 15 ml PBS to a total urea concentration of 1 M. The diluted QPrEST was concentrated using a Pierce concentrator 9K MWCO (Thermo Scientific), where 12.2 ml PBS 1 M Urea was added to achieve a dead-stop volume of 1.8 ml and centrifuged at 4000 x g, 20 ℃ for 50 min using a swing bucket rotor (SORVALL Heraeus 75006445).

2.3.4 QPrEST quality control

2.3.4.1 SDS-PAGE purity estimation

Prior to the concentration measurement, the concentrated QPrEST was centrifuged at 15 000 x g, 20 ℃ for 5 min in a tabletop centrifuge to eliminate any precipitate. The QPrEST concentration was determined using a NanoDrop spectrophotometer measuring the QPrEST absorbance at 280 nm. 1:6 elution buffer/PBS solution was used as blank. The absorbance measurement was converted into a concentration value using the extinction coefficient for respective QPrEST calculated from the ExPasy ProtParam tool [31]. To assess the QPrEST purity, 1 µg QPrEST was run on an SDS-PAGE gel (Appendix, Table 4 and Figure 8). The volume of 1.5 µg QPrEST according to NanoDrop concentration measurement was prepared together with 7.5 µl 4 x Leamilli sample buffer (Bio-Rad), 1.5 µl 1 M DTT and MQ to a total volume of 30 µl. The QPrEST sample was heated for 5 min at 95 ℃ to ensure denaturation and centrifuged at 12 000 rpm in a tabletop centrifuge for 1 min to spin down any condensation. 20 µl (1 µg) sample and 10 µl molecular weight marker PageRuler^TM Plus Prestained Protein Ladder (Thermo Scientific) were loaded onto a 4-20 % Criterion^TM TGX Precast Gel (Bio-Rad) run in 1 x TGS cold running buffer (10 % 10 x TGS (Tris/Glycine/SDS (Bio-Rad)), 90 % MQ) for 40 min at 200 V. The gel was washed with deionized H2O three times 5 min on a shaker before staining with 20 ml SimplyBlue SafeStain (Life Technologies). Before imaging, using a ChemiDoc MP camera (Bio-Rad), the gel was

(11)

washed with deionized H2O two times a’ 60 min. The image analysis was performed using the ImageLab 5.1 software.

2.3.4.2 MS molecular weight determination and quantification

MS identification was performed by molecular weight determination of the QPrESTs with MS full-length analysis (Appendix, Table 2, 3, 4 and Figure 9). MS quantification was performed to determine the concentration of the QPrESTs using the QTag quantification standard (Appendix, Table 4). The molecular weight samples were prepared by adding 0.4 µl 250 mM DTT diluted in 0.1 M NH4HCO3 to 10 µl QPrEST. The samples were left to incubate for 1 hour at RT before being alkylated by the addition of 0.4 µl 400 mM Iodoacetic acid (IAA) (Sigma) diluted in 0.1 M NH4HCO3 and incubated for another 30 min in darkness due to IAA’s light sensitivity. Lastly, the samples were diluted with 90 µl 0.1 % Formic acid (FA) and transferred to a 96-well plate for subsequent MS analysis. The quantification samples were prepared by mixing 50 pmol HisABPOneStrep (Qtag) with 5 ul QPrEST protein sample in triplicate.

Proteins were diluted in 0.1 M NH4HCO3 to a final volume of 40 ul before addition of 1.25 uL 250 mM DTT followed of incubation for 1 h in RT. Proteins were alkylated by the addition of 2 µl 400 mM IAA and incubated for 30 min at RT in darkness. Lastly, 5 µl 99 % Acetonitrile (ACN) and 1 µl 100 ng/µl trypsin (T6557, Sigma) were added and incubated overnight at 37 ℃. The next day, 1 µl of 5 % FA were added to quench the digestion. The samples were diluted to 50 fmol/µl by the addition of 190 µl 0.1 % FA to 10 µl sample, before samples were transferred to a 96-well plate for subsequent MS/MS analysis.

The LC system used was DIONEX UltiMate 3000, where the nano system was used for quantification and the standard system for molecular weight determination, respectively. A 30 min gradient was used for the quantification using a 15 cm PepMap^TM C18 nanoViper^TMcolumn (Thermo Scientific) and a 15 min gradient was used for the molecular weight determination using a 5 cm ProSwift™ RP C4 column (Thermo Scientific). The MS analysis for both QPrEST molecular weight determination and quantification was executed on a Bruker Impact II Q-TOF system. The standard ESI source was used for the molecular weight determination and the Captive Spray source was used for the quantification.

Molecular weight determination was determined using the Biopharma Compass software and the QPrEST quantification was performed using the Hystar software for data acquisition and the Proteinscape software for data analysis.

2.4 Western blot 2.4.1 Protein separation

An SDS-PAGE gel was run to separate the proteins before protein transfer and detection. The same procedure, material and conditions as for the QPrEST purity estimation were used. However, here the gel was loaded with 10 µg protein instead.

2.4.2 Protein transfer and detection

The gel was positioned into a Trans-Blot® Turbo^TMMidi PVDF Transfer Pack (Bio-Rad) and the sandwich assembled according to manufacturer’s instructions. The transfer was run at 25 V for 10 min, after which the transferred membrane was washed in deionized H2O and left to dry at RT. The dried membrane was incubated in MeOH for less than a minute to activate it for subsequent blocking for 30 min, at RT at 50 rpm on a rocking shaker. The blocked membrane was incubated with primary antibody (anti-RBM3 mAb 1mg/ml AMAb90655, anti-ATAD2 mAb 1 mg/ml AMAb90541, anti-ANLN mAb 1 mg/ml AMAb90662 and anti-VIM mAb 0.1 mg/ml AMAb90516 (Atlas Antibodies)) in a 1:1000 dilution in blocking buffer (1 x TBST (50 mM Tris-Base, 150 mM NaCl, 0.1 % Tween 20, MQ, pH 7.4) and 5 % Semper low fat dried milk powder) for 1 hour, at RT at 50 rpm. Residual primary antibody was rinsed away using 1 x TBST buffer. An additional wash including incubation for 5 min was completed before the secondary antibody (Horseradish Peroxidase (HRP)-conjugated polyclonal goat anti-mouse immunoglobulins, P0447 (Dako)) in a 1:3000 dilution in blocking buffer was added and incubated for 30 min, at RT at 50 rpm. The same wash procedure as for the primary antibody was used to rinse away residual secondary antibody, however, with an additional wash and incubation of 5 min. WB detection substrate solutions, Immobilon Western Chemiluminescent HRP substrate (Millipore) were mixed according to the manufacturer’s instructions, in which the membrane was incubated for 1 min before it was photographed with a ChemiDoc MP Camera (Bio-Rad). Image analysis was performed using the Image software ImageLab 5.1. The entire process was repeated for the loading control, which was used to

(12)

normalize for variations in sample loading. GAPDH was used as control for the target proteins RBM3, ATAD2 and ANLN. PFN1 was used as control for the target protein VIM as GAPDH have similar molecular weight as VIM. Primary antibodies used were anti-GAPDH mAb 1 mg/ml and anti-PFN1 mAb 0.5 mg/ml (Atlas Antibodies), respectively, and the same secondary antibody was used for detection.

The membrane was dipped in substrate solution and immediately photographed without any incubation to reduce signal saturation. Triplicates were prepared for each siRNA knocked down target protein and the control.

2.5 Real-time quantitative polymerase chain reaction 2.5.1 RNA preparation

The cell lysate was transferred to a gDNA Eliminator spin column (Qiagen), centrifuged for 30 s at 8000 x g and the resulting flow through was saved, to which 350 µl 70 % ethanol was mixed. 700 µl gDNA eliminated lysate was transferred to an RNeasy spin column (Qiagen), centrifuged for 15 s at 8000 x g and the flow through was discarded. 700 µl RW1 buffer (Qiagen) was added to wash the column, centrifuged at the same settings and the flow through was discarded. In addition, 500 µl RPE buffer (Qiagen) was added to wash the column, which was repeated twice. Last centrifugation was run for 2 min instead of 15 s as previously. An additional centrifugation was run at 13 000 x g for 1 min to eliminate any residual ethanol. To elute the RNA, 50 µl RNase-free water was added to the column and centrifuged for 1 min at 8000 x g and the eluate was saved for further quality analysis. The total RNA concentration was measured using a NanoDrop spectrophotometer at an absorbance of 260 nm. RNase-free water was used as a blank.

The Agilent 6000 Nano Assay was used to analyze the quality of the RNA in an Agilent 2100 Bioanalyzer from Agilent Technologies (Appendix, Figure 10). The assay was prepared according to manufacturer’s protocol using the Agilent 6000 Nano Assay kit. The data analysis was performed in the 2100 Expert Software.

2.5.2 cDNA synthesis

The iScript synthesis kit from Biorad was used for cDNA synthesis preparation. Initially, the RNA sample was diluted to 5 ng/µl with RNase-free water based on NanoDrop concentrations. According to manufacturer’s protocol, the following was mixed: 4 µl 5x iScript reaction mix, 1 µl reverse transcriptase, 11 µl nuclease free water, which was later added to 4 µl (20 ng) RNA sample. The cDNA synthesis was performed on a PCR instrument (Bio-Rad) using the temperature cycle as follows: 25 ℃ for 5 min, 42

℃ for 30 min, 85 ℃ for 5 min and 4 ℃ on hold.

2.5.3 Quantitative analysis

In a 96 well plate, 2 µl (2 ng) cDNA sample was mixed together with 8 µl master mix containing 5 µl 2x iQ SYBR Supermix (Bio-Rad), 10 µM forward and reverse primer (Thermo Scientific) diluted in nuclease free water to a total volume of 10 µl, according to manufacturer’s recommendations. Three reference genes were used to normalize for variations in amount of template loaded and amplified between samples, namely; HMBS, SDHA, UBC. In addition, two different controls were used, a no template control, where a signal indicates contamination or primer dimer formation and a no reverse transcriptase control, where a signal indicates the presence of genomic DNA in the RNA preparation. In the no template control 2 µl water was added instead of cDNA. In the no reverse transcriptase control 2 ng RNA was added. On a real-time qPCR instrument (Bio-Rad) the following qPCR program was run: 95 ℃ for 3 min, 95 ℃ for 10 s, 60 ℃ for 10 s, 72 ℃ for 30 s, steps 2-4 were repeated in total 40 times. The temperature was raised from 65 ℃ to 95 ℃ and the signal measured after every 0.5 ℃ increase of temperature to allow for a melt curve analysis (Appendix, Figure 12). The target genes as well as the reference genes, no template control and no reverse transcriptase control were analyzed in triplicates. Data analysis was performed using the CFX Manager^TM 3.0 software (Appendix, Figure 11).

(13)

2.6 Mass spectrometry

2.6.1 Screening for proteotypic peptides 2.6.1.1 QPrEST digestion

Initially, 50 pmol of each QPrEST were pooled to achieve a total QPrEST amount of 400 pmol. DTT was added to the QPrEST pool to achieve a final concentration of 10 mM and incubated for 30 min at 56 ℃.

To alkylate the sample, IAA was added to achieve a final concentration of 50 mM and was incubated for 20 min at RT in darkness. To digest the sample, trypsin (T6567, Sigma) was added to the QPrEST pool in a 1:50 enzyme:substrate ratio and incubated overnight at 37 ℃, 300 rpm in a thermo mixer. The next day, 3 µl 10 % FA was added to achieve a final concentration between 0.1-0.5 % FA to quench the digestion and the cleaved sample was later extracted using Solid Phase Extraction.

2.6.1.2 Solid Phase Extraction on QPrEST digest

The cleaved QPrEST pool was purified using Strong Cation Exchange StageTips. To activate the membrane of a prepared StageTip with three layers of Empore Cation 47 mm Extraction Discs 2251, 50 µl 100 % MeOH was added, spun down at 3000 rpm in a tabletop centrifuge for 1 min. The membrane was equilibrated by the addition of 50 µl wash buffer (30% MeOH, 0.1 % FA, MQ) and spun down as described, before 10 µg acidified sample was applied and centrifuged. The membrane was washed two times using 30 µl wash buffer and spun down at the same settings. The sample was eluted by two additions of 20 µl elution buffer (30 % MeOH, 1.65 % NH4OH, MQ), spun down and vacuum dried for 30 min before stored at - 20 ℃.

2.6.1.3 Screening

The sample was resuspended in MS buffer (97 % MQ, 2.9 % ACN, 0.1 % FA) before subsequent LC- MS/MS-analysis. The LC-system used was DIONEX Ultimate 3000 RSLC nano system with an Easy Spray Column PepMap® C18, 75 µm diameter and 15 cm long, filled with 3 µm particles with a 100 Å bead distance. The trap column used was Acclaim PepMap®100 C18, 75 µm diameter and 2 cm long, filled with 3 µm particles with a 100 Å bead distance. A 50 min gradient was used, with the following concentration program for buffer B (90 % ACN, 4.9 % DMSO, 5 % MQ, 0,1 % FA) versus buffer A (95

% MQ, 4.9 % DMSO, 0.1 % FA) at a flow rate of 0,4 µl/min; 3 % buffer B at 5 min, 33 % at 53 min, 47

% at 55 min, 99 % at 62 min and 3 % at 63 min. Full MS scans were obtained at a resolution of 60 000 and a 300-1600 m/z range, while MS/MS scans were obtained at a resolution of 30 000 and a isolation window of 1.2 m/z. The MS analysis was executed on a QExactive HF instrument (Thermo Scientific) and data analysis was performed in the MaxQuant software with the inbuilt database Andromeda [32][33].

The peptide search was run for a minimum of six amino acids long peptides with up to two miscleavages using a false discovery rate of 0.01. The hit list (Appendix, Table 6) of proteotypic peptides was further verified (Appendix, Table 7) in a parallel reaction-monitoring (PRM) assay; the same assay later used for quantification of the target proteins. Data analysis of proteotypic peptides was completed in the Skyline software.

2.6.2 Mass spectrometry quantification of target proteins 2.6.2.1 Filter Aided Sample Preparation

1 pmol QPrEST pool to RBM3 and ATAD2, 5 pmol QPrEST pool to ANLN and 10 pmol QPrEST pool to VIM was spiked in per 250 000 cells based on the cell count obtained from a Scepter^TM 2.0 Cell Counter (Millipore). Cell lysates were digested into peptides using the Filter Aided Sample Preparation method [34]. To wash the filter units (NANOSEP 10K OMEGA, Life Sciences), 200 µl MQ was added and centrifuged at 14 000 x g for 15 min. 5-10 µg/µl protein extract was mixed with 200 µl Urea (0.1 M Tris-HCl, 8 M Urea, pH 8.5), loaded onto the filter and centrifuged using the same settings. 3 x 200 µl Urea was added and spun down. Later, 100 µl 50 mM IAA diluted in Urea was added, mixed for 1 min at 600 rpm in a thermo mixer, incubated for 20 min and centrifuged for 10 min. 3 x 100 µl Urea was added and spun down for 10 min, followed by the addition of 3 x 100 µl 50 mM NH4HCO3 diluted in MQ, spun down for 10 min. Before the addition of 100 µl NH4HCO3-trypsin solution, where the amount of trypsin corresponded to a 1:50 relation of enzyme versus substrate, the filter unit was transferred to a new

(14)

collection tube. The added NH4HCO3-trypsin solution was mixed in 600 rpm for 1 min and incubated for approximately 14 h and centrifuged for 10 min. 50 µl MQ was added and spun down for 10 min. The concentration determination was performed by NanoDrop spectrophotometer measurements at 280 nm, where 33 mM NH4HCO3 was used as blank.

2.6.2.2 Solid Phase Extraction on protein digest

The filtered protein digest was purified from salts and detergents with Solid Phase Extraction using Strong Cation Exchange StageTips. To StageTips containing 6 layers of Empore^TM Cation 47 mm Extractions Discs 2215, 50 µl 100 % MeOH was added and spun down at 3000 rpm for 1 min using a plate centrifuge (ScanSpeed 1580GR) to activate the membrane. The activated membrane was equilibrated by the addition of 50 µl wash buffer (30 % MeOH, 0.1 % FA, MQ) and spun down as described. To the equilibrated membrane, 10 µg acidified sample with a final concentration of 0.1-0.5 % FA and a pH ≤ 3 was loaded and spun down. Before elution, the membrane was washed with 2 x 50 µl wash buffer and spun down.

The sample was eluted by the addition of 2 x 30 µl elution buffer (30 % MeOH, 5 % NH4OH, MQ). The eluate was vacuum dried for 30 min and later stored at - 20 ℃ before subsequent MS analysis.

2.6.2.2 Quantification

The sample was resuspended in MS buffer (97 % MQ, 2.9 % ACN, 0.1 % FA) before subsequent LC- MS/MS analysis on a QExactive HF instrument. The same LC parameters as for the screening of proteotypic peptides were used in the optimized PRM assay. Full MS scans were obtained at a resolution of 60 000 at m/z range of 300-1600. MS/MS scans were obtained at a resolution of 30 000 and an isolation window of 1.2 m/z. The data analysis of peptide ratios was completed in the software Skyline (Appendix, Table 5).

3. Result

3.1 Western blot protein concentrations

The WB analysis was performed using monoclonal antibodies produced and commercialized by Atlas Antibodies on U251 cell lysates. HRP-conjugated antibody was used for detection.

WB results were obtained for all target proteins studied. The WB membranes presented in Figure 3 indicated that the band molecular weights corresponded to the expected molecular weights of the target proteins. The band molecular weights detected corresponded to the theoretical molecular weight of RBM3 (18 kDa), ATAD2 (160 kDa), ANLN (124 kDa) and VIM (54 kDa), respectively. This was also the case for the loading controls, GAPDH (37 kDa) and PFN1 (15 kDa). [7][35] Furthermore, the similarity in band signal between the lanes of the different loading controls suggested that the gels had been evenly loaded. In general, the relative concentrations for the siRNA knocked down proteins were lower than for the controls, which was also indicated from the band signals obtained. This further suggested a successful down-regulation of the target proteins and the specificity of the antibodies used.

To elucidate a significant difference in expression between knocked down protein and normally expressed protein and thereby a successful knockdown experiment, a one-tailed unpaired t-test with a p-value of 0.05 was considered suitable. A significant down regulation for the target proteins RBM3, ATAD2 and ANLN were obtained. Same result was obtained by visually inspecting the standard deviations in the graphs for these proteins (Figure 3A, B and C). No significant down regulation was obtained for VIM;

this was also suggested from the standard deviations seen in the graph (Figure 3D). However, this result might have been influenced by the low band signal from the last lane of the loading control for VIM. This low band signal was not expected, as the same amount of sample should have been loaded onto the gel with only small pipetting errors possible. The explanation to this could be an incomplete protein transfer from the gel to the membrane in that particular area.

(15)

Figure 3. WB-estimated protein concentrations. Each figure represents data for the studied proteins, where A. RBM3; B. ATAD2;

C. ANLN and D. VIM, respectively. The first row illustrates the relative normalized protein concentration for each target protein, derived from analyzed band intensities from the WB membranes showed in the second row. Each individual bar diagram, show the relative concentration of knocked down protein using two different siRNAs in relation to normally expressed protein termed as the control. The bar diagram includes standard deviations for the relative concentrations illustrated as error bars. The individual WB membranes illustrate the visual signals obtained for each triplicate, as well as the loading control located beneath each blot.

The loading pattern for each blot was as follows when lanes are counted from the left corner: lane 2 - siRNA #1, lane 3 - siRNA

#2, lane 4 - control. This pattern was repeated twice for the remaining 6 lanes, resulting in triplicates for each siRNA and control.

3.2 Real-time quantitative polymerase chain reaction mRNA concentrations

U251 cell lysates were used in the qPCR analysis. RNA purified from genomic DNA was used as template for the reverse transcriptase-mediated cDNA synthesis. cDNA template was mixed with forward and reverse primers and real-time qPCR-amplified. Resultant amplification curves were used to analyze gene expression.

qPCR results were obtained for one of the target proteins studied, namely RBM3. ATAD2, ANLN and VIM were not detected, most probably due to suboptimal primer annealing temperatures for these target proteins. Figure 4C indicates that the correct DNA fragments have been amplified for the target gene and the reference genes. Moreover, that no primer-dimer formation or other unspecific amplification occurred. This is evident from the fact that the melt peak curves were showing a sharp single peak and the melting temperature was the same for each gene and all samples. Thus the qPCR analysis functioned as a verification analysis ensuring that the protein studied in WB and MS was the correct protein. Though, Sanger sequencing would have completely determined if the right fragment had been amplified. For the reference gene SDHA, there was one melting curve deviating from the rest, belonging to the no reverse transcriptase control sample. This result suggested the presence of genomic DNA, however the signal appeared late in the cycling, thus it was considered a background signal (Appendix, Figure 4).

Figure 4B demonstrates that the amount of template amplified differed between the reference genes, which was unexpected. The expected outcome was that the amount of template should have been equal between the samples. The unexpected result suggested an uneven loading of template in the cDNA synthesis or qPCR reaction. This might have influenced the normalized relative concentration of RBM3 to be higher than expected, which is seen in Figure 4A, particularly for siRNA #1. Generally, it can be concluded from the same figure that the relative normalized concentration for the two siRNAs was lower than for the control. A one-tailed unpaired t-test with a p-value of 0.05, revealed a significant protein down-regulation for siRNA #2, but not for siRNA #1. This was further verified from the standard deviations presented in the graph.

(16)

Figure 4. qPCR-estimated mRNA concentrations. A. Bar diagram illustrating the relative normalized mRNA concentration for the knocked down protein versus the normally expressed protein termed as the control. Two different siRNAs were tested to knock the protein, termed as siRNA #1 and siRNA #2. B. Bar diagram illustrating the relative mRNA concentration for the target gene as well as for the reference genes: HMBS, SDHA and UBS, used for normalization. C. Line chart illustrating the melt peaks for the target gene and the reference genes. The different colors represent the different genes, where red is the target gene RBM3, and orange is the reference gene HMBS, green is SDHA and blue is UBC. The melt peak describes the temperature at which the synthesized DNA fragment denatures from the cDNA template, representing the melting temperature Tm of the double stranded DNA product. The temperature is plotted against the negative regression of fluorescence, - d(RFU)/dt.

3.3 Mass spectrometry protein concentrations

The MS analysis was accomplished using QPrESTs derived from the HPA project, commercialized by Atlas Antibodies on U251 cell lysates. QPrESTs were expressed in the presence of heavy isotopes in an arginine and lysine auxotroph E.coli strain. QPrESTs were quality controlled by a MS molecular weight determination and quantification. QPrEST purity was estimated by SDS-PAGE. Initially, a shotgun analysis was performed in order to identify as many proteotypic peptides as possible. These were verified, generating a final hit list of tryptic peptides. The amount of QPrEST that was spiked in to cell lysate was optimized for the list of tryptic peptides analyzed in a PRM assay.

MS results were obtained for two of the four target proteins included in this study, namely ANLN and VIM. RBM3 and ATAD2 were not detected at all in the MS analysis probably due to complexity issues. In addition, one of the two siRNAs used for VIM knockdown was detected, namely siRNA #2. siRNA #1 was not detected, possibly due to sample preparation errors. A number of two and ten peptides were selected based on the ratios obtained from the MS analysis for the quantification of ANLN and VIM, respectively. Eight more peptides were generated for VIM in contrast to ANLN, which could be explained by a larger number of tryptic peptides in the two VIM-specific QPrEST sequences used in relation to the two ANLN-specific QPrEST sequences (Table 1).

The absolute quantification of ANLN for the two individual peptides is seen in Figure 5C. When the absolute concentration obtained for the two different peptides for ANLN was compared they did not seem to differ substantially; the opposite was observed for VIM (Figure 5D). The identified absolute concentration for ANLN was in the range of 160 000-190 000 copies per cell for the normally expressed protein, whereas the absolute concentration for the knocked down protein was in the range of 30-60 000

(17)

copies per cell; 100-130 000 copies per cell lower in relation to the normally expressed protein. The two different siRNAs resulted in marginally dissimilar copy numbers per cell, where siRNA #1 resulted in a 25 000 lower copy number per cell in relation to siRNA #2, suggesting a more efficient knockdown. A larger cohort of peptides would have been needed in order to study the variation between peptides for ANLN.

For all ten peptides used for the quantification of VIM, a down-regulation of the protein using siRNA #2 was not obvious (Figure 5D). However, a small difference in copy number per cell for the knocked down protein versus the normally expressed protein was identified for all ten peptides. The copy number per cell obtained in this study was hundreds of times higher in comparison to previous studies that have determined it to 20 million copies per cell [20]. Also, the copy number per cell varied between the ten peptides, most likely due to the low quality of the ratios obtained.

The relative quantification of ANLN (Figure 5A) and VIM (Figure 5B) was calculated as an average from the peptides generated for each protein. Here, the standard deviations illustrated as error bars served as an indication for a significantly down-regulated protein. The determination of a significant protein down- regulation using a t-test would have been possible for the MS analysis if the amount of valid replicates had been larger. A comparison based on the standard deviations between siRNA knocked down protein concentrations and the normally expressed protein concentrations, suggested that the target protein ANLN was significantly down-regulated, however not VIM.

Figure 5. MS-estimated protein concentrations. A. Relative normalized concentration of siRNA #1 and siRNA #2 knocked down ANLN, based on the average concentration calculated from the generated peptides used in the MS analysis. B. Relative normalized concentration of siRNA #2 knocked down VIM, based on the average concentration calculated from the generated peptides used in the MS analysis. C. Copy number per cell for individual peptides generated for ANLN, where each peptide is numbered P1-P2. D. Copy number per cell for individual peptides generated for VIM, where each peptide is numbered P1-P10.

Table 1. List of the individual peptides used to determine the absolute and relative concentration of the target proteins ANLN and VIM. The tryptic peptides are highlighted in red to show the QPrEST sequence location.

(18)

3.4 Comparison of real-time quantitative polymerase chain reaction, western blot and mass spectrometry concentrations

When the WB-estimated concentrations were compared with the MS-estimated concentrations for ANLN (Figure 6A) it was observed that the WB-estimated concentration for the knocked down protein for both siRNAs were more than 20 % higher than for the MS-estimated concentration for the knocked down protein for both siRNAs, in relation to the control. However, this trend was not observed for the knocked down VIM (Figure 6B), probably due to the large variation between the peptides generated as previously explained. The same tendency was found when the WB-estimated protein concentrations were compared with the qPCR-estimated mRNA concentrations for the target protein RBM3 (Figure 6C). Here, the WB- estimated concentration for the knocked down protein for both siRNAs were approximately 15 % higher than the qPCR-estimated concentration for the knocked down protein for both siRNAs, in relation to the control. This deviation might be due to inherent technical limitations with the WB method, further discussed under Discussion and future perspectives.

Both the WB analysis and the MS analysis for the target protein ANLN resulted in a significant down regulation for both siRNAs used, as described in respective result sections for each analysis. Also, both methods resulted in the conclusion that no significant down regulation was present for VIM regarding siRNA #2. The fact that both techniques generated the same result signifies their correlation in protein concentration measurements. This result was also seen in the comparison of the WB and qPCR concentrations obtained for RBM3. Both methods suggested a significant down regulation of the protein using siRNA #2. However, in contrast to WB, no significant down regulation for siRNA #1 was observed using qPCR. As described in the qPCR result section, a possible explanation to this might be an uneven template loading resulting in the high variation for this particular siRNA.

Figure 6. Comparison of WB, MS and qPCR-estimated concentrations. A. Bar diagram presented with respective standard deviations illustrating WB versus MS-estimated relative protein concentrations for the target protein ANLN. B. Bar diagram presented with respective standard deviations illustrating WB versus MS-estimated relative protein concentrations for the target protein VIM. C. Bar diagram presented with respective standards deviation illustrating WB-estimated relative protein concentrations versus qPCR-estimated relative mRNA concentrations for the target protein RBM3.

(19)

4. Discussion and future perspectives

In this study, a comparison of qPCR, WB and MS for the estimation of protein concentrations has been performed. The comparative study was enabled through the development of a laboratory protocol for the quantitative analysis of the target proteins included in the study. The development included optimization of siRNA transfection. Also, an MS-based assay for the two target proteins ANLN and VIM was established. The QPrESTs used in the MS experiment and the monoclonal antibodies used in the WB experiment have via this comparative study been thoroughly validated. This has gained valuable insights into appropriate proteomic applications for the reagents commercialized by Atlas Antibodies, which can be used to market their QPrESTs and antibodies for proteomic research.

The combined results from this study confirmed that no noteworthy difference regarding the estimation of the relative protein concentrations for the knocked down target proteins ANLN and VIM in relation to the control seem to exist between WB and MS, indicating a good correlation between the methods (Figure 6A and B). This result further demonstrates the reproducibility of the QPrESTs and the monoclonal antibodies used. Even without verifying the correct target genes with qPCR for these proteins, it is evident enough from using WB and MS, two orthogonal methods, that the correct genes have certainly been studied. For these two target genes the primer annealing temperature needs to be optimized to be able to analyze the RNA-protein correlation.

RBM3 and ATAD2 could not be detected in the MS assay developed. This might be due to a complexity issue; that those higher abundant proteins like ANLN and VIM masked the lower abundant proteins RBM3 and ATAD2. The next step in the project would therefore be to develop a more sensitive assay.

This would include the optimization of several parameters such as increased protein separation using longer gradients or reduce the complexity by protein fractionation. Also, the combination of affinity proteomics and MS-based proteomics, where antibodies are used to enrich the low abundant proteins from a complex mixture could be an alternative. Thus utilizing the sensitivity of the antibodies and the specificity of the QPrESTs. However, an immuno-MS assay for cell lines is generally not needed.

MS-estimated protein concentrations could have been more accurate if more peptides had been used for the quantification. A solution to this would have been to increase the number of QPrESTs used for each target protein or using higher collision energy, thereby increasing the number of proteotypic peptides that can be targeted. The large variation of the absolute concentration for knocked down VIM and normally expressed VIM (Figure 5D) could be explained by the high ratios obtained. The ratios obtained were in the range of 50-200; optimally one would like a ratio as close to 1 as possible to obtain reliable and high- quality concentration determinations. The same situation was for ANLN, but in the opposite direction;

low ratios of approximately 0.001 were used (Appendix, Table 5). The large variation could also be due to variations in the amount of cells that the QPrESTs were spiked into. Optimally, these variations need to be normalized for, e.g using histone PrESTs proportional to the number of cells in the sample. [36]

In this study it has been indicated that WB suffers from several inherent technical problems; such as incomplete protein transfer, variations in the detection method, background noise, image analysis etc., which could explain the overall higher relative protein concentration in contrast to MS (Figure 6A and B).

Optimizing the WB method by using an internal standard as in MS would have likely solved these issues.

Though, this was not studied here. The conclusion that can be drawn from this study is that the MS method seems to be the most accurate method, with its unmatched specificity combined with the usage of internal standards. This enables precise absolute quantification of the protein concentrations; only relative values were obtained using WB. The different methods would probably correlate less well if absolute protein concentrations were compared.

It was also concluded from this study that no noteworthy difference in down-regulated relative RNA- protein concentrations in relation to the control seem to exist between qPCR and WB for the target protein RBM3 (Figure 6C), suggesting a linear relationship of the RNA and protein levels in the studied cell line. However, this needs to be verified using MS as an orthogonal method to WB. Nothing could be concluded about the RNA-protein correlation for the other target proteins studied. Although, the results obtained for RBM3, might indicate that there exists a linear relationship even for these proteins. It is however possible that the RNA-protein correlation for the other proteins are not linear. Studies have

(20)

indicated that studying smaller sets of proteins tend to reveal a non-linear relationship between RNA- protein levels. However large-scale sets of proteins studied will even out these deviations and the correlation tends to be more linear. [22]

A larger cohort of both biological and technical replicates used in the different experiments would have improved the validity of the results obtained from this study. The more replicates included, the higher degree of statistical correctness of the concentrations measured. Furthermore, the results are restricted to the U251 cell line used in the study. This means that the results might not be reproduced in a more complex sample. Cell lines are more homogenous and not as complex as tissue samples, suggesting that the results obtained in this study will most probably differ for the heterogeneous tissue from which the cell line originated. However, in research, cell lines are commonly used making the results still useful.

The next step in the project would be to analyze the methods in an induced protein expression system, e.g studying the c-Fos protein induced by phorbol 12-myristate 13-acetate (PMA) in the HeLa cell line. This would enable the comparison of the correlation between the methods when protein concentrations vary over time. Further, this would allow the study of the linearity between the methods and thereby give additional information in what range of protein concentrations correlation exist and not, which was out of scope for this project.

5. Acknowledgements

I hereby would like to acknowledge Atlas Antibodies for giving me the opportunity to perform my master thesis at their company. There exists a genuinely welcoming atmosphere at the company and I always felt like a part of it. I would like to give thanks to all employees at the R&D department for always answering my questions and for helping me out with the laboratory work. Especially, I would like to thank my three external supervisors, Oscar Ljungqvist, Tove Boström, and Henrik Johannesson for having created this interesting master thesis project and for guiding me through it, continuously keeping me inspired and motivated. Also, many thanks to Fredrik Edfors at Science for Life Laboratory assisting me with MS- related laboratory work and questions. In addition, I would like to thank my KTH supervisor Torbjörn Gräslund and my examiner Maria Humble for giving constructive feedback and for being supportive throughout the project.