The vast majority of biological processes, ranging from homoeostasis maintenance to development, cell cycle and cell differentiation, are tuned by differential gene expression. The latter is under the control of gene regulatory networks (GRNs), which consist of physical and functional interactions between DNA-binding regulatory proteins, transcription factors (TFs), and regulatory elements of their target genes (i.e. promoters, enhancers). The key function of these networks is to coordinate the progression of distinct transcription regulatory states both in space and time, a sine qua non condition for cell survival. Unfortunately, most models of GRNs are incomplete and their players far from being completely characterized. Understanding how each TF contributes to the expression output of its respective target gene in space and time will help us understand how gene regulatory networks behave, particularly under different conditions. In this regard, it would be of great interest to quantify TFs within the cell in absolute amounts linking their abundance to their transcriptional capabilities, for endogenous TF levels may thereby determine binding site occupancy.
Although qualitative information explaining TF interactions and behavior is widely available in the literature, reliable quantitative data is almost absent. This disparity is explained by TF natural low abundance in cells, which make quantitative analyses a substantial challenge considering the current technologies. To effectively understand the role and behaviour of TFs as regulatory proteins this information is of crucial importance. Moreover, quantitative data would be extremely useful for in silico modelling for computational biologists, not to mention the value that it could have in the medical context, where diagnostic and treatment of a disease is often related to the amount of a given protein biomarker. Various techniques are available for the quantification of proteins nowadays, the majority of which rely on the use of antibodies (e.g. ELISA, protein microarrays). Quantitative immunoassays are widely used because of their accuracy and the fact that can be implemented even in small laboratories due to their low cost and simplicity. Nevertheless, these techniques suffer from a certain number of drawbacks, non-linearity in quantitation and formation of unspecific reactions to name a few. Moreover, one can only analyze one protein at a time, not to mention that only a limited number of TF-specific antibodies are commercially available, limiting thereby the applicability of these methodologies to small scale studies of a hand full of TFs. In this regard, there is a strong need for a robust methodology that could bypass such limitations, possibly pushing the current limits in terms of sensitivity and specificity even farther. Until recently, mass spectrometry based methodologies lacked the necessary sensitivity to be used for the identification of low abundant proteins.
Noticeable improvements in mass spectrometer detection limits have opened the doors to the analysis of proteins that are expressed at very low levels in cells, such as TFs. Tandem mass spectrometry (MSMS) is routinely used to identify naturally occurring peptide sequences from complex mixtures of proteins, based on peptide bond fragmentation reactions. As few as two to three identified peptides are sufficient to positively reveal the presence of a particular protein in a sample. In recent years, a new technology termed Selected Reaction Monitoring (SRM) has gained popularity due to the targeted character of its approach that allows the detection and quantification of predetermined sets of proteins in complex samples with previously unseen sensitivity and specificity. In SRM precursor- to product- ion transitions are selected to build a quantitative assay. This narrow mass filtering results in high selectivity, and the non-scanning nature of this methodology grants an increase in sensitivity. The cardinal point of such an approach, when it comes to quantification, is the consistency and the uniqueness of the peptides selected. Only peptides that uniquely identify a protein of choice, and that are consistently detected in different MS runs should be utilized; such peptides are termed "proteotypic". There is no gold standard for the identification of such peptides. Nevertheless, several bioinformatic tools that guide the user in the selection of proteotypic peptide candidates (based on a set of physico-chemical properties) are currently available (e.g. Pinpoint, Thermo Scientific; Waltham, Ma., USA).
Hence, SRM appears to be particularly well suited for targeting low abundant proteins such as TFs, which are difficult to identify otherwise with conventional mass spectrometry based methodologies due to their low signal intensities in MS and to the fact that these same signals are often lost, outnumbered by the background noise belonging to their more abundant counterparts, often several orders of magnitude more significant.
The adipogenic model
The terminal adipocyte differentiation is the last phase of adipogenesis, during which pre-adipocytes develop into mature adipocytes through a cascade of gene expression events. It is a natural process, consequence of normal cell turnover on one hand, as well as a necessity for fat mass storage in case of excessive weight gain. Its study has a clear medical relevance: excess fat mass, characterized by an increase in cell size and number, dramatically increases the risk of developing series of pathologies, including metabolic syndrome symptoms and cancer (Ailhaud et al., 2006). Several studies have established a basic framework of the GRN orchestrating the terminal phase of adipogenesis (Farmer et al., 2006; Rosen and McDougald, 2006). Although substantial effort has been devoted to identify TFs and co-TFs involved in adipogenesis (Oishi et al., 2005; Fujimori et al., 2010), almost no information is available on the amount of these TFs present in the nucleus or in the cytoplasm, limiting our knowledge on the subject to an almost merely qualitative one. In this regard, understanding adipogenic TFs contribution to target genes expression output during adipocyte differentiation will shed light on how the adipogenic regulatory network behaves under different physiological or pathological conditions, opening new venues for disease diagnostic and cure.
We intend to the monitor the dynamic changes in transcription factor abundance that characterize the terminal phase of the process of differentiation, and therefore build a quantitative model that will be used to infer adipogenic TF's transcriptional capabilities. To this end, the murine embryonic 3T3-L1 pre-adipocyte cell line recapitulates most of the aspects of terminal adipocyte differentiation observable in vivo (Green and Meuth, 1974). Their homogeneity, synchronous development, as well as their availability makes this cell line the perfect candidate for the development of our targeted assay.
Methodology
Implementation of SRM assays
As previously described, SRM is becoming a benchmark in targeted proteomic approaches. Generally, this technique is performed in a triple-quadrupole mass spectrometer, where the first and the third mass analysers are fixed on a precursor-, product-ion of choice respectively and the second mass analyser is used as a fragmentation chamber. SRM departs from shotgun approaches where a much wider mass window is scanned. As in other techniques that use mass spectrometry as a reference, ion intensities are indicative of peptide abundances. Quantification is generally achieved by spiking isotopically-labelled peptides in the sample at known concentrations to serve as standards. Generally, such heavy-labelled peptides are chemically synthesized and commercially available (Thermo Scientific, Waltham, Ma., USA). As little as two or three of these proteotypic peptides are in most instances sufficient for accurate quantification. Although proteotypic peptides can be selected either experimentally or predicted using bioinformatic tools (e.g. Pinpoint, Thermo Scientific), practical constraints limit the number of peptides that can be utilized (e.g. peptides containing a glutamine residue at the N-terminus cannot be synthesized). Furthermore, by shrinking the number of all possible tryptic peptides of a protein of interest to a few chosen ones, one looses valuable information from the ones that are not selected, incurring in the risk of preventing accurate measurements due to lack of candidates. Another common problem is related to the storage of peptides at low concentrations, which tend to precipitate and adsorb to tube walls. As a result, quantification accuracy can be seriously compromised. To bypass these limitations, we decided to work with entire proteins. The advantage of this technique is first that one can use all of the peptides for the identification/quantification (except C-terminus peptides). Second, the production of proteins is relatively inexpensive. Third, TFs can be produced "as you go" without the burden of storage with its inconveniences. And last, because TFs will be spikes as early as possible in the workflow, both spiked and endogenous peptides will be subject to the same modification and eventual losses during sample preparation.
Adipogenic TFs will therefore be in vitro transcribed/translated in their heavy form (isotopically labeled) using a wheat germ cell-free system (Promega, Madison, Wi., USA) to bypass cytotoxicity. A unique tag-peptide, which will serve for its quantification is fused to its N-terminus. A light version of that same tag-peptide will be used to quantify the amount of the heavy tagged TF expressed. Since the amount of the tag-peptide spiked will be known in advance, accurate quantification can be achieved by comparing ion intensities of the heavy/light tag-peptide pair. In turn, quantification of the endogenous TF is achieved by comparing TF-specific peptide signal intensities of the tagged heavy TF to their endogenous counterparts in a targeted SRM assay (Figure 1). Both measurements can be achieved at the same time in one single run. This process can be performed for each TF involved in the late stage of adipogenesis and integrated in series of comprehensive SRM assays.
Spectral libraries for the selection of transitions and identification of TFs
The conventional methodology to identify and characterize proteins in complex samples is based on the analysis of MSMS data. Popular protein identification tools such as SEQUEST, Phenyx, as well as MASCOT (Dagda et al., 2010) utilize peptide fragment fingerprinting algorithms, where experimentally obtained spectra are compared to theoretical ones. Such spectra are generated in silico from protein sequence databases. In recent years, a different approached based on the use of spectral libraries for protein identification and characterization has gained popularity, mainly due to the advantages that it presents in terms of computational time, identification rate and accuracy when compared with more traditional sequence based approaches (Lam et al., 2007; Frewen et al., 2006; Yates et al., 1998) . In this approach, experimental spectra are scored against a carefully compiled database of previously collected experimental ones. Matching experimental spectra to a set of experimentally obtained spectra instead of in silico predict ones leads to a non-negligible increase in terms of sensitivity (Ahrné et al., 2008). The reason is that the intensities of all fragment types present in the library spectrum are considered with neutral losses and miscellaneous types of fragments that may simply not be acknowledged in theoretical spectra. Additionally, the search space when dealing with spectral libraries is much smaller, which translates with an increase in precision as well as speed. Finally, peptide fragmentation patterns may vary from one the mass spectrometer to another. To overcome these issues, we intend to combine a sequence search approach with a spectral library approach, where TF spectrum libraries are compiled from the MS analysis of in vitro expressed TFs, increasing thereby the level of explanation of MSMS data overall. This composite procedure may lead to an increment in the confidence as well as in the number of TFs identified in the complex sample. A step of crucial importance in the development of SRM assays is the selection of transitions, for specific precursor- to product-ion masses will be ultimately utilized to determine the presence of a given TF in a complex sample, nonetheless to proceed to an accurate quantification. To this end, in collaboration with Erik Ahrné and Dr. Markus Mueller from the SIB in Geneva (Swiss Institute of bioinformatics), under the supervision of Dr. Lisacek, we devised a novel methodology that exploits spectral information produced in MS analyses of in vitro expression of TFs for the selection of optimal SRM transitions.
To sum up, by combining the spectral library search methodology to more traditional sequence-based search approaches, we aim at maximizing the number of interpreted spectra, increasing thereby the amount of information that can be used for quantification. Our methodology bypasses the use of predictions, which often due to the stringency of its methodology fail to bring to light the very peculiar behaviour of charged peptides in a mass spectrometer, and anyway require empirical validation at some later stage of development (Picotti et al., 2008). Thereby, we intend to compare our methodology with the ones based solely on prediction.
Current status of project
Transition selection via the Spectral library approach
Genome-wide RNA PolII occupancy data was used to identify which TFs are transcribed during adipogenesis in 3T3-L1 cells. Out of more than 1500, 757 were identified, and clustered according to their transcriptional profile over the 6 time-points of differentiation. For each cluster, the core TFs were selected, resulting in a list of about 130. These TF's ORFs were cloned using the Gateway technology into an in house modified version of the pF3A-WG vector containing barley yellow dwarf virus sequences to improve protein expression efficiency, and expressed in vitro via a wheat germ transcription-translation system. A GST tag was added at the C-terminus of the proteins to allow for purification and validation. The TFs were subsequently purified with sepharose beads coupled to glutathione. The resulting proteic compound was separated by SDS-PAGE and stained with silver nitrate for the visualization of bands (Figure 2A). Upon validation with Western Blot (WB) using anti-GST antibodies, the bands belonging to the TFs were excised and trypsinized. The extracted peptides were submitted to separation by liquid chromatography (LC) before being analyzed by mass spectrometry. Off of the 130 adipogenic TFs that we aimed to analyze in our targeted proteomic approach, we were able to clone 116. For 104 of them, we had evidence of successful in vitro expression by SDS-PAGE, results validated then by WB. In the MS analysis, peptides belonging to 100 TFs of interest (96% of the expressed TFs) were detected, thereby unambiguously identifying their presence within the excised gel fractions. Out of this vast pool of peptides, proteotypic peptide candidates were then selected based on the uniqueness, as well as the quality of their spectral features, thus creating a unique atlas of TF peptide spectra based on experimental data rather than predictions (Figure 2B). These TF peptides were then ranked by the quality of their spectra, and the top scorers chosen for selection of the transitions. Prediction of proteotypic peptides was performed using Pinpoint (Thermo Scientific) as to have further confirmation of proteotypicity. Selection of the optimal product- to fragment ion transitions was performed again by spectral analysis. We thereby obtained a pool of optimal transitions per TF, which will serve to the implementation of targeted SRM measurements. Assay development requires validation, on one hand to confirm the selectivity of the chosen transitions to uniquely identify a selected TF (distinct peptides may share similar spectral features), on the other hand to maximize its overall sensitivity (Picotti et al., 2010). We are currently evaluating the validity of the transitions that we have selected for each TF experimentally. As an example, we wanted to see if we could detect 4 top-scorers RXRα proteotypic peptides in the 3T3-L1 nuclear extract basing ourselves on the information obtained in the transition selection process. To achieve this goal, we set up a "targeted" manner utilizing an Orbitrap instrument, and interestingly were able to retrieve the 4 candidates (further examples of transition validation via SRM in section 4.iii, below).
Construction of an Atlas of TF peptide spectra
The novel development of an open source public repository for peptide tandem mass spectrometry data, the Global Proteome Machine Database (GPMDB), has made it possible to retrieve a list of proteotypic peptides for a limited number of species, based on experimental observations. This repository contains more than four million annotated peptide mass spectra, contributed by many laboratories. In spite of this large amount of spectra, low abundant proteins are poorly represented, due to the difficulties presented above. The number of TF-specific spectra in such databases is rather small and the peptide coverage rather low. To this end, the information contained in our TF-specific peptide atlas can complement and most interestingly complete existing information available in such databases, for a large number of TF-specific peptides have never been detected to date. Our methodology has therefore yielded to the creation of a peptide Atlas that contains more than 1000 TF peptide spectra, representing some 800 unique peptides belonging to approximately 100 TFs. These numbers are constantly growing for TF addition results in an accumulation of new data.
We obtained therefore a high quality peptide database, built from a collection of multiple screenings of in vitro expressed TFs. In this regard, a consensus spectrum specific to each peptide was created by merging spectra, and will be utilized for identification purposes. This vast effort is the first of its kind; we aim at creating a comprehensive TF-specific spectral repository to make available to the public. TF-specific data can be utilized as a guide for future SRM assays or to improve protein identification.
Absolute quantification of in vitro expressed heavy tagged TFs
The development of an assay towards the absolute quantification of endogenous TFs can be seen as a two step process. First, the amount of in vitro expressed heavy TFs has to be accurately determined via the use of a peptide-tag. This information will subsequently allow us to proceed to the quantification of the endogenous TF by comparing the signal intensities of signature peptides (light versus heavy). To this end, we are testing two distinct peptide tags: Flex (amino acid sequence: TENLYFQGDISR), derived from the FLEXIQuant approach for protein quantification (Steen and co-workers, Harvard Medical School, USA), and SH-quant (amino acid sequence: AADITSLYK) derived from the SH-quant based quantification technique (Gstaiger and co-workers, ETHZ, Switzerland). Both tags were cloned into our in house modified version of the pF3A-WG vector upstream the recombination Gateway site. Upon cloning utilizing a TF-specific ORF, the final protein construct presents a GST tag at its C-terminus, and either a Flex or a SH-quant tag at its N-terminus (the Flex tag is accompanied by a Strep tag II) as it can be seen in Figure 3.A. We successfully cloned and in vitro expressed two master regulators of adipogenesis (validation by WB), PPARγ and RXRα, in these two constructs in their light and heavy (isotopically labeled) form (the latter will be used for quantification) (Figure 3.B). Peaks belonging to the Flex and SH-quant peptides were observed in MS1 spectra, and the MS2 analysis allowed for their unambiguous identification (Figure 3.C for SH-tag).
We are currently utilizing the information obtained with our spectral library methodology to implement a series of preliminary SRM assays, utilizing in vitro expressed tagged PPARγ and RXRα as trail blazers. The scope of this sort of assays is to test the validity of the transitions that we have selected in identifying TFs, as well as to optimize the parameters that will be used in a comprehensive SRM experiment with the complex sample (3T3-L1). SH-Quant and Flex tagged PPARγ and RXRα constructs were expressed as previously described, purified, separated by molecular weight, and stained for detection. Bands belonging to the 4 constructs were excised, peptides extracted after trypsinization and submitted to SRM. The 5 best responsive peptides (in terms of overall quality of the spectra) were selected (we decided to limit to 5 the number of peptides) and will be used in the later stage of the project for the quantification of endogenous PPARγ and RXRα (Figure 4). Each precursor ion was coupled to several fragment ions. As one may notice, at each step of the selection process precautions have to be taken and problematic ions are singled out; the number of both precursor and fragment ions therefore declines. This meticulous selection/validation coupled iterative process ensures that only the best responsive ions are ultimately selected, increasing or confidence level on the subsequent quantification.
After having validated that the PPARγ and RXRα proteotypic peptides we have selected (including the SH-Quant and Flex peptide tags) were well behaved, we proceeded to the quantification phase. The two TFs were in vitro expressed with the SH-Quant tag on their C-terminus in their heavy form, purified, separated by SDS-PAGE and the gel was stained with Coomassie Blue. Bands were excised and peptide extracted prior to SRM (Figure 5.A, PPARγ as example). In parallel we built an external calibration curve for the SH-Quant peptide tag (utilizing SH-Quant light peptide specifically synthesized and accurately quantified) spiking increasing amounts of the peptide in the mass spectrometer, ranging from 6 to 60 fmol, in technical triplicates. The resulting calibration curve returned an R squared of 0.977 (Figure 5.B). Utilizing the external standard curve we were able to accurately quantify the amount of expressed proteins via SRM analysis. By knowing the fraction of sample injected in the mass spectrometer, one can easily calculate the total amount of protein present in the gel band (Figure 5.A and C, PPARγ as example). In Figure 5, as a proof of principle, we showed that the gel band contained 1.419 pmol of SH-PPARγ-GST. Furthermore, the ratio of incorporation of heavy amino acids during protein in vitro expression can be calculated (Figure 5.D) and a correction can be applied to reflect the exact amount of protein produced. We can now therefore accurately quantify the amount of any in vitro expressed TF by applying this methodology. We now possess all of the necessary tools to quantify endogenous TFs in 3T3-L1 cells, the very next and ultimate step in our workflow.
Proteomic analysis of 3T3-L1 nuclear extract at day 4 of differentiation
In order to assess the sensitivity of the methodology, and define the dynamic range at which low abundant TFs could be detected, we decided to implement a classic two-dimensional LC-MS experiment. Although 3T3-L1 total nuclear protein extract was collected at each of the 6 time-points during the differentiation process, the implementation of our SRM pipeline as well as preliminary tests were performed using nuclear protein extract at day 4 of differentiation (highest level of expression of PPARγ and RXRα) obtained from 1.5 x 10e7 cells; 250 μg of nuclear proteins were thereby purified. Proteins were digested with trypsin prior to LC-MSMS. In this regard, 3 different types of sample fractionation were tested. The first one relied on the separation of peptides by SAX (strong anion exchange) into 6 fractions. The second one relied on the separation of peptides by SCX (strong cation exchange). Finally, the third method was based on the separation of peptides by their pI (isoelectric point) using an "Off-gel" electrophoretic system. The aim of testing different fractionation procedures was to identify the one that provides an optimal degree of separation of the sample, allowing the distinction of peaks belonging to our TFs of interest from the complex background. By taking these two master regulators of adipogenesis as a proof of principle, PPARγ and RXRα, 2 peptides belonging to PPARγ and 2 peptides belonging to RXRα were detected with different methods of separation. In this regard, The SAX analysis of the total nuclear proteome permitted the detection of one peptide belonging to PPARγ. Interestingly enough the peptide was predicted to be proteotypic by Pinpoint, and detected in the in vitro PPARγ expression experiment as well. On the other hand, sample fractionation by "Off-gel" electrophoresis allowed for the highest detection rate (2 peptides belonging to RXRα).
We are currently performing in-depth analysis to shed light on the advantages of each technique, in order to select the one that optimizes peptide separation, increasing thereby chances of detection in 3T3-L1 cells. The detection of peptides belonging to TFs by a conventional LC-MSMS approach is definitely promising, and increases our confidence in the robustness of our SRM based methodology, orders of magnitude more sensitive. The positive results in terms of TF peptide detection achieved with conventional mass spectrometry techniques pushed us to move towards a more selective and sensitive approach, such as SRM. The awareness that we are dealing with protein species that sit at the horizon of detectability, drove us to test other fractionation techniques that further enrich the total nuclear proteic extract for the TFs we were interested in monitoring. In this regards, 3T3-L1 nuclear extract was separated on an SDS-PAGE gel and stained for band pattern visualization. The gel area in which we expected to recover PPARγ and RXRα was excised (note that the bands are not visible, due to the extremely dense band patterning). Because the two TFs molecular weight differs only by 6 kDa, it was possible to recover both proteins in one single gel fraction. WB validation was used as a reference during the cutting (Figure 6.A), confirming band positioning on the gel. This procedure simplified the search space by a great extent, for the majority of nuclear protein species present in the nuclear extract won't simply be taken into account. Various protein species are nevertheless present in the excised fraction. Subsequent to excision, peptides are extracted from the gel band and submitted to SRM. The 5 proteotypic peptides selected belonging to PPARγ and RXRα peptide were indeed detected (Figure 6.B, RXRα as an example). Parent-ion fragmentation resulted into the detection of fragment-ion species. The most intense ones will therefore be used for endogenous TF quantification in the next step (Figure 6.C, RXRα GLSNPAEVEALR peptide as an example).
These latest SRM results utilizing 3T3-L1 nuclear extract confirm the validity to utilize information obtained from in vitro expression of TFs to select peptide transition for SRM. It provides useful empirical information that may otherwise be left in the dark when dealing with predictors.
Multiplexing: Designing SH-Quant and Flex-tag variants
The number of endogenous TFs that are quantifiable in one single SRM run is limited by the number of peptide tags available, for one tag can track one TF only. Up until now, we have considered two different tags, the SH-Quant (AADITSLYK), and the Flex (TENLYFQGDISR), which have allowed us to establish the methodology. We now aim at expanding the number of TFs that can be analyzed in one assay by creating a series of tag variants (Figure 7.A). In order to preserve the phisico-chemical properties that confer to the two tags their proteotypic character, we interfered as little as possible in the peptide design. We replaced only one or maximum two amino acids in the sequence of the tags at defined positions (Figure 7.B). Such sequence modifications may though perturb the peptide behavior in the mass spectrometer. To assess the proteotypic properties of the newly designed candidates are not perturbed, we decided to have the tag variants chemically synthesized, and will test their performance in a series of SRM runs. Optimal candidates will therefore be retained, and used for the high-throuput implementation of the methodology, reducing thereby the time and the costs associated with it.
Participants
Due to the multi-disciplinary nature of the above mentioned project that encompasses molecular biology, bioengineering, biophysics, chemistry, as well as programming, database-design, and statistics, it was of utmost importance to find the right person that possessed the necessary knowledge and capabilities to successfully bring to completion this study. To this regards, the latter was assigned to a bright PhD student with a vast multi-disciplinary background, Jovan Simicevic, who envisioned and devised this research project from the very beginning. This study is being carried on in strict collaboration with the Proteomics Core Facility of Dr. Marc Moniatte at the EPFL (Dr. Adrien Schmid, Dr. Florence Armand, Romain Hamelin). The PIG group of the Swiss Institute of Bioinformatics (SIB) in Geneva, lead by Dr. Lisacek (Erik Ahrné, Dr. Markus Mueller) developed and implemented the spectral library search approach.
Figure 1:
Figure 2:
Figure 3:
Figure 4:
Figure 5:
Figure 6:
Figure 7:
Figure 1. Workflow of the SRM assay for the absolute quantification of adipogenic TFs in 3T3-L1 cells. First, the total nuclear protein extract is obtained from 3T3-L1 cells. A heavy tagged in vitro expressed TF is spiked in the sample, together with a light tag-peptide (its amount is known in advance). The canonical SRM pipeline is resumed. Quantification of the heavy tagged in vitro expressed TF is achieved by comparing light/heavy tag-peptide ion intensities. At the same time, quantification of the endogenous TF is achieved by comparing heavy tag-peptides (obtained from the heavy tagged in vitro expressed TF) ion intensities to their light endogenous counterparts.
Figure 2. (A) TFs are expressed using a plasmid template that allows cell-free protein expression using wheat germ lysate, affinity-purified, separated from other lysate proteins using SDS-PAGE and visualized with silver nitrate. (B) Experimentally detected peptides of the adipogenic master regulator PPARï§ were compared with predicted proteotypic peptides using the Pinpoint software, revealing proteotypic peptide candidates which were then ranked by spectral analysis. The top scorers will be used to develop SRM adipogenic assays. Based on our spectral library of TF peptides we now aim at designing optimized targeted SRM assays. Synthetic proteotypic peptides will be spiked in 3T3-L1 total nuclear protein extracts at different stages of differentiation in known amounts to identify and ultimately quantify endogenous TFs in nuclear extracts in absolute terms. The absolute protein quantity could then be inferred from the heavy/light ratios of the respective TF peptides within the spiked sample.
Figure 3. (A) Drawing explaining the structure of our fusion proteins used for quantification. A GST tag is fused on the C-terminus of the TF, and a SH-quant (top) or a Flex tag (including a Strep-tag, bottom) is fused at its N-terminus. (B) Silver-stained SDS-PAGE gel of PPARγ and RXRα in their SH-tag construct (left) and Flex construct (right), both in their light and heavy form. Bands belonging to these constructs are clearly visible (validation performed by WB). (C) MS1 (top) and MS2 (bottom) spectra used to identify the SH-tag peptide.
Figure 4. (A) Drawing explaining the SRM workflow. PPARγ and RXRα are in vitro expressed both in the SH-quant or a Flex construct, purified, separated by SDS-PAGE and stained with Coomassie Blue. Bands belonging to the 4 fusion proteins are clearly visible in the red squares (top left). Subsequently bands are excised, peptides extracted after trypsinization, and submitted to LC-SRM (top right). (B) SRM chromatograms of SH- PPARγ-GST (left) and SH-RXRα-GST (right). The top lane represents the extracted total ion count of the 5 selected proteotypic peptides altogether. The 6 lanes below show the monitoring of the 5 selected peptides per protein construct (PPARγ on the left and RXRα on the right) plus the SH-Quant peptide tag. Their peaks are clearly visible and represented in different colors.
Figure 5. (A) SH-PPARγ-GST was in vitro expressed, purified, separated by SDS-PAGE and stained with Coomassie Blue. Its band is clearly visible and marked by the red arrow. (B) SH-Quant standard curve obtained by spiking increasing amounts of the SH-Quant synthetic peptide in technical triplicates in SRM. (C) Relative intensities of SH-Quant peptide analyzed by SRM spiked in increasing amounts from 6 to 60 fmol in technical triplicates. The blue peak represents levels of the SH-Quant peptide derived from spiked SH-PPARγ-GST obtained from A). The amount of SH-PPARγ-GST injected in the mass spectrometer was quantified at 28.38 fmol. By knowing the fraction of material spiked in the mass spectrometer (1/50th), the total amount of protein present in the excised band can be calculated: 1.419 pmol. (D) SRM chromatogram of the SH-Quant tag showing that the incorporation of heavy arginine and lysine is not complete (~95%). Basal levels of the light peptide are observable.
Figure 6. (A) 3T3-L1 NE was separated SDS-PAGE and stained with silver nitrate (left). The area around 50 kDa (red square) in which PPARγ and RXRα were expected to be found as confirmed by WB (right) was excised and peptides extracted prior to SRM analysis. (B) SRM chromatograms of 4 selected RXRα proteotypic peptides. (C) SRM chromatograms of the RXRα proteotypic peptide (precursor ion) GLSNPAEVEALR (top). Only well-responsive fragment ions were selected for quantification (circled in red).
Figure 7. (A) Drawing explaining how tag variants will allow for multiplexing, increasing the number of TFs that can be quantified in a single SRM run. SH-quant or Flex tag variants are as usual fused at the N-terminus of a pool of TFs to be quantified. Application of our methodology will allow for the absolute quantification of endogenous TFs. (B) Drawing explaining the design of tag variants. Amino acid residues that are mutated are colored in red, both on the Flex-tag (left) and the SH-Quant-tag (right). The tables contain the sequence of the 25 tag variants.
