Proteome-wide profiling of activated transcription factors with a concatenated tandem array of transcription factor response elements

Abstract

Transcription factors (TFs) are families of proteins that bind to specific DNA sequences, or TF response elements (TFREs), and function as regulators of many cellular processes. Because of the low abundance of TFs, direct quantitative measurement of TFs on a proteome scale remains a challenge. In this study, we report the development of an affinity reagent that permits identification of endogenous TFs at the proteome scale. The affinity reagent is composed of a synthetic DNA containing a concatenated tandem array of the consensus TFREs (catTFRE) for the majority of TF families. By using catTFRE to enrich TFs from cells, we were able to identify as many as 400 TFs from a single cell line and a total of 878 TFs from 11 cell types, covering more than 50% of the gene products that code for the DNA-binding TFs in the genome. We further demonstrated that catTFRE pull-downs could quantitatively measure proteome-wide changes in DNA binding activity of TFs in response to exogenous stimulation by using a label-free MS-based quantification approach. Applying catTFRE on the evaluation of drug effects, we described a panoramic view of TF activations and provided candidates for the elucidation of molecular mechanisms of drug actions. We anticipate that the catTFRE affinity strategy will find widespread applications in biomedical research.

Almost all biological processes, ranging from cell cycle regulation to organ development, are controlled by the transcriptional regulatory system (1). In the classic cell membrane-to-nucleus signal transduction paradigm, transcription factors (TFs) are the final effectors. They are activated and bind to consensus DNA sequences to execute specific transcriptional programs in response to the signal. Thus, the ability to monitor TF activity is important for the delineation of signal transduction pathways when the cells are perturbed (e.g., treated with a drug), or when organs are under the influence of developmental cues.

Approximately 1,500 TF coding genes are reported to be in the human genome (2). TFs can be grouped into different families depending on the structure of their DNA binding domains. There are approximately 50 TF families (2), and each family prefers to bind a specific DNA consensus sequence. For example, nuclear receptors (NRs) are ligand-modulated TFs that recognize one or two hormone response element sequences such as 5′-AGAACA-3′ or 5′-AGGTCA-3′ (3). Previous studies have demonstrated the importance of linking an extracellular signaling event to the activation of TFs. For example, assigning Forkhead box (Fox) P3 to a signaling module that is crucial for regulatory T-cell development has accelerated our understanding of signal transductions and gene functions (4).

The abundance of TFs in the cells is currently inferred from mRNA profiling. Yang et al. identified 45 of 49 known NRs from several tissues in mice and linked NR expression to the circadian clock (5). Bookout et al. surveyed the expression of all 49 mouse NR mRNAs in 39 tissues (6). However, information obtained from mRNA profiling often cannot be directly translated into protein levels, let alone the activity state of the TF population.

Here, we report a method for determining DNA binding activity of multiple endogenous TFs simultaneously. By using a synthetic DNA containing a concatenated tandem array of consensus TF response elements (TFREs; catTFREs) for most known TF families, we succeeded in detecting more than 878 TFs from 11 cell types, including 400 TFs from a single cell line. We further showed that this method could quantitatively measure activated TF change in response to signaling events. We applied this method to elucidate drug effects by describing alterations of hundreds of activated TFs in response to drug treatments. We envision that this methodology will find broad applications in discovering TF activation/repression in signaling networks.

Results

Design and Characterization of catTFRE for TF Enrichment.

We referred to TF binding database JASPAR to select consensus TFREs for different TF families. To design the catTFRE construct, we used 100 selected TFREs and placed two tandem copies of each sequence with a spacer of three nucleotides in between, resulting in a total DNA length of 2.8 kb (Fig. 1C and Dataset S1). We synthesized and cloned the catTFRE sequence into a pUC57 vector and prepared the catTFRE affinity reagent by PCR amplification with biotinylated primers. We then incubated nuclear extracts (NEs) with the biotinylated catTFRE. The resulting protein–DNA complexes were digested with trypsin and analyzed with MS. Isotope-based and label-free quantification can be used in this workflow (Fig. 1 A and B).

Fig. 1.

Outline of catTFRE pull-down strategy. (A) A tandem combination of TFRE with duplicated repeats was synthesized and amplified by PCR with biotinylated primers. Biotinylated TFRE was then incubated with cell lysate to enrich endogenous TFs. Samples were subjected to MS for measurement. (B) catTFRE pull-down coupled with label-free/based strategy. Peptides of TFs with good signal response were selected for quantification by calculating AUC. Isotope-labeled internal standard was spiked into samples, and the amount was determined after comparing peptide AUC with respective internal standard, (C) Design of catTFRE DNA. Information of consensus TF binding sequence was grabbed from JASPAR Web site. Each TF binding site was synthesized duplicated and tandemly combined with a three-nucleotide spacer. (D) Advantages of catTFRE strategy in endogenous TF enrichment. A total of 500 µg of NE was incubated with 15 pmol catTFRE or executed with trypsin digestion directly. One percent of output was loaded on MS, indentified TFs were counted, and peak areas of peptides were calculated. (E) Identifications and peptides AUC of TFs enriched by catTFRE or executed with trypsin digestion directly.

Next, we evaluated the efficiency of catTFRE DNA pull-down for enrichment of TFs from 500 μg of NE. Ninety-four TFs were identified in 1% of catTFRE pull-down eluate, whereas only 24 TFs were identified in 1% of equivalent NE input (Fig. 1D and Dataset S1), and 20 of the latter were recovered by catTFRE pull-down. Areas under the curve (AUCs) of peptides as an indication of TF abundance of the nine TFs were calculated, and showed a significant enrichment by the catTFRE (Fig. 1E). To test the sensitivity of catTFRE pull down, we carried out the experiments by using various amount of NEs ranging from 50 μg to 400 μg (Fig. 2A). We detected more than 150 TFs from 50 μg of NE, and more than 200 TFs from 400 μg of NE, demonstrating that catTFRE strategy is a sensitive and high-throughput assay for the detection of TFs.

Fig. 2.

Sensitivity and quantitative feasibility evaluation of catTFRE strategy. (A) TF identifications of serial amounts of NE incubated with 15 pmol catTFRE. TF SPCs are shown as a heat map of white to red. (B) TF enrichment and identification comparison between catTFRE and nonregulatory DNA pGEX4T2. Color density indicates total abundance of identified TFs. (C) Quantitative feasibility and linearity of catTFRE strategy evaluated by titration analysis. Serial amounts of NE were used as shown. Peptide AUC from 14 TFs and two nonspecific binding proteins were calculated. AUCs of 25 µL sample were set as 1 and others were normalized to their corresponding peptide in 25 µL sample. (D) Western blotting analysis of the catTFRE output using antibodies as indicated.

We then evaluated how effective the catTFRE pull-down is in the enrichment of endogenous TFs. We cloned a 2.8-kb DNA sequence (same length as catTFRE) from the pGEX4T2 plasmid as nonregulatory DNA control and carried out DNA pull-down experiments using the same amount of NEs and DNA. catTFRE and control DNA pull-downs identified 276 and 172 TFs, respectively; 194 TFs showed enrichment of >10 fold in catTFRE, whereas only five TFs showed enrichment of >10 fold in control DNA (Fig. 2B and Dataset S1), suggesting that catTFRE was much more specific and effective in enriching and identifying TFs by design.

To test how a single TFRE impacted TF binding of the TF family, we made two deletion mutants named ΔNFY and ΔFox by removing nuclear transcription factor Y (NFY) or Fox binding site from the original catTFRE sequence (Fig. S1B). Deletion of the NFY binding site led to decreased binding of eight TFs to more than three fold among the 270 TFs detected (Dataset S1). NFYB and NFYC decreased by >10 fold, and NFYA decreased by seven fold (Fig. S1C). Deletion of the Fox binding site led to decreased binding of 17 TFs to more than three fold among the 270 TFs detected (Dataset S1). FOXC1, FOXD2, FOXP1, and FOXP2 decreased DNA binding more than three fold (Fig. S1D). We concluded that enrichment of TFs by catTFRE is largely dependent on their specific TF binding sites.

Label-Free and Stable Isotope Labeling by Amino Acids in Cell Culture Based Quantitative TF Screening by catTFRE Pull-Down.

To test the feasibility of label-free quantitative TF profiling, we used same amount of catTFRE DNA (15 pmol) to isolate and identify endogenous TFs by using HeLa NE in the range of 0.25 to 2 mg total protein in 250 μL of volume. As shown in Fig. 2C, all 14 selected TFs exhibit linear response to the amount of proteins used in the pull-down, whereas signals of nonspecific binding proteins, such as Actin and HSP70, remain largely unchanged (Fig. 2C and Fig. S1A). We also compared the dynamic response of three selected proteins [nuclear factor kappa b (NF-κB); nuclear receptor subfamily 2, group C, member 2 (NR2C2); and CAMP responsive element binding protein (CREB-1)] by Western blotting. As shown in Fig. 2D, the increased intensity of WB signals was consistent with the increased amount of isolated proteins as more NE was used.

Next, we tested whether this approach can be used to reveal dynamic changes of TF binding in response to extracellular stimuli. NF-κB TF is activated by various intra- and extracellular stimuli, including TNF-α (7). We treated 293T cells with 10 ng/mL of TNF-α or vehicle control for 3 h and performed TF profiling for both samples in parallel. As shown in Fig. S2 A and B, TFRE-bound NF-κB/p50 and bovine transcription factor p65 (RELA/p65) were increased five- and 13-fold, respectively, after TNF-α treatment, which is consistent with the previous reports (8, 9). Jun, a TF activated by TNF-α (10), also exhibited an increased binding by three fold to catTFRE.

The stable isotope labeling by amino acids in cell culture (SILAC)-based quantification was used to verify the label-free quantification results. We spiked the same amount of NE from SILAC-cultured HeLa cells in TNF-α–treated and vehicle control samples. The isotope-labeled TFs should bind to catTFRE with the same affinity in both samples and thereby serve as an internal standard (Fig. S2D). After normalization to isotope-labeled internal controls, SILAC quantification showed that NF-κB and Jun were increased by seven and three fold upon TNF-α stimulation, a result that is consistent with the label-free quantification (Fig. S2C).

In-Depth Analysis of TF Binding in Mammalian Cell Lines.

To uncover the potential of catTFRE in TF profiling, we used 5 mg of NE and 30 pmol of catTFRE DNA to isolate and identify TFs from different cell lines (Fig. 3A). Prefractionation of peptides into 12 fractions with isoelectric focusing resulted in the identification of 455 TFs from HeLa cells. We carried out similar experiments with 293T, H1299, HeLa, HepG2, A549, U937, MCF7, PC3, SY5Y, and MEF cells, and detected 207 to 460 TFs in these cell lines (Dataset S2). Next, we applied catTFRE pull-down to mouse liver to evaluate whether catTFRE can be used for tissue TF profiling. A total of 391 TFs were identified from mouse liver as a result (Fig. 3A). In all, we identified 878 TFs from 11 mammalian cell lines, representing more than half of all TF-coding genes in the genome. Notably, 29 of 50 Forkhead family members and 42 of 48 predicted NRs were detected. Fig. 3C summarizes the coverage for each TF family in 11 mammalian cell lines. These results also demonstrate the wide dynamic range of TF abundances, as the top 16 TFs contribute 25% of the total number of spectral counts (SPCs), an indicator of the abundance of proteins, whereas 300 TFs of lower abundance together constitute only 1% of total SPCs (Fig. 3D).

Fig. 3.

Differential TF expression pattern and coverage analysis of TF families among 11 cell types. (A) TF and (B) CoR profiling of 11 human cell types using catTFRE demonstrated heterogeneity in basal TF and CoR expression pattern. Normalized SPCs of TFs are shown as a heat map of white to red. FOT, fraction of total, i.e., percentage of a TF SPC to total. (C) Coverage analysis of TF families from 11 mammalian cells with catTFRE strategy. (D) Cumulative protein mass from the highest to the lowest abundance TFs.

Transcription coregulators (CoRs) cooperate with TFs to integrate diverse cellular signals and thereby mediate a coordinated transcriptional response (11). Although many CoRs do not directly bind to DNA, they can be recruited to TFREs through interaction with TFs. Considerable numbers of CoRs were identified in our TF screening, suggesting that some TF–CoR interactions are preserved in catTFRE DNA pull-down. A total of 497 CoRs were identified in the 11 tested cell types (Fig. 3B and Dataset S2). Similarly to TFs, the presence and abundance of CoRs are widely distributed, with the 32 most abundant CoRs comprising half of the CoR SPCs (Fig. S3B). Moreover, the catTFRE strategy identified 155 of 235 highly confident unconventional DNA binding proteins reported in previous work (12) (Dataset S2).

Comparison of catTFRE Pull-Down and Protein/mRNA Profiling in HeLa Cells.

We used a faster mass spectrometer (Q-Exactive; Thermo) and extended total MS measuring time to 16 h to improve the identification coverage and compared with a typical in-depth MS profiling of HeLa cells. We identified 743 TFs among 3,866 gene products identified in catTFRE pull-down and 295 TFs among 7,601 gene products from MS profiling. The enrichment of TFs by catTFRE is profound (Fig. S3C). A total of 487 TFs were identified exclusively in the catTFRE pull-down, whereas only17 TFs, such as STAT5A/B and STAT6, identified exclusively with at least three unique peptides in the MS profiling (Dataset S2). We concluded that these STATs did not bind DNA under our experimental conditions, as we detected abundant amounts of STATs in other catTFRE pull-downs (Dataset S2).

We compared the catTFRE result with mRNA-seq in the literature (13). The mRNA-seq identified 859 TFs among the 10,936 protein coding genes (fragments per kb of exon per million mapped fragments > 1), whereas catTFRE pull-down identified 743 TFs. The overlap between mRNA-seq and catTFRE is 579 TFs (Fig. S3D and Dataset S2). There are total of 1,531 genome-coding TFs, of which mRNA-seq identified 56%, whereas catTFRE identified 49% of them in HeLa cells. These data suggest that the depth of coverage for the TF subproteome by catTFRE pull-down and that of the TF subtranscriptome by mRNA-seq are comparable for HeLa cells.

Analysis of Dynamic Changes of Global TF-DNA Binding Patterns After TNF-α Treatment.

We then chose TNF-α signal transduction pathway to evaluate the potential of catTFRE in analysis of global TF alterations in response to exogenous stimulation. We performed catTFRE pull-down with 293T cells treated with TNF-α for 15, 30, and 180 min, and detected a total of 234 TFs (Fig. 4A and Dataset S3). We arbitrarily chose more than threefold intensity change as significantly changed. Overall, 20 TFs were activated by TNF-α, 13 of which were increased within 15 min, and seven TFs showed a delay in activation after 30 min. Meanwhile, binding of 15 TFs was suppressed by TNF-α, six of which were decreased within first 15 min, and another nine showed a delayed down-regulation after 30 min (Fig. 4A and Dataset S3). The remaining 199 TFs did not show significant changes upon TNF-α stimulation. Consistent with previous knowledge, TFs related to NF-κB family and JNK/P38 pathways were activated (14, 15) (Fig. S4A). In addition, several TF families that have not previously been known to be involved in TNF-α response exhibited marked changes. For example, binding of the zing finger and BTB (ZBTB) and nuclear factor of activated T cells (NFAT) family members was increased, whereas binding of high mobility group (HMG) proteins was reduced upon TNF-α treatment (Fig. S4B).

Fig. 4.

Systematical and quantitative analysis of TF profiling in TNF-α pathway. (A) Kinetic TF activation pattern of 293T cells after TNF-α stimulation. 293T cells were treated with TNF-α for different time. Relative amount of TFs compared with 0 min group are shown as a heat map of green to red that represents down-regulation and up-regulation, respectively. Accurate intensity of TFs in 0-min group was set as 1. (B) 293T cells were treated with TNF-α for 15 min in the presence or absence of PDTC. Relative amount of TFs compared with vehicle control group are shown as a heat map of green to red that represents down-regulation and up-regulation respectively. Accurate intensity of TFs in vehicle control group was set as 1.

NF-κB is known as the strongest responder to TNF stimulation. We then sought out TF changes that are associated with NF-κB activation. To this end, we blocked NF-κB activation by preincubating the cells with an inhibitor ammonium pyrrolidinedithiocarbamate (PDTC) before TNF-α treatment. As expected, TF members of NF-κB family but not JNK/P38 pathway were inhibited by PDTC (Fig. 4B and Fig. S4 C and D). Up-regulated TFs ZBTB17 and NFATs had the same response pattern as NF-κB, indicating that these TFs behave similarly as NF-κB. In contrast, the decrease in HMG family members was not affected by PDTC (Fig. 4B, Fig. S4E, and Dataset S3). This proof-of-concept study has demonstrated that catTFRE strategy is capable of systematically detecting changes in TF DNA binding activity.

Screening of TF DNA Binding Activity Change in Drug Actions.

We tested whether catTFRE can be effectively used to study the molecular effects of drug actions in K562 cells that contain the Philadelphia chromosome and the chimeric BCR-ABL1 gene. We chose phorbol myristate acetate (PMA) (16), an activator of protein kinase C (17), and imatinib mesylate (Gleevec) (18), an inhibitor of the BCR-ABL kinase, for many of their opposite effects in the regulation of the K562 cells (19, 20).

The K562 cells were treated with PMA or imatinib for 24 h, and TF DNA binding was profiled with catTFRE. The PMA treatment yielded 462 TF and 395 CoR identifications, of which 159 TFs and 92 CoRs were up-regulated and 113 TFs and 83 CoRs were down-regulated, with more than threefold change (Dataset S3). The imatinib treatment yielded 406 TFs and 371 CoRs, of which 46 TFs and 18 CoRs were up-regulated and 137 TFs and 146 CoRs were down-regulated, respectively, at more than three fold of change (Dataset S3). Analysis of the alteration of TFs with Integrated Pathway Analysis indicates that PMA and imatinib play opposite roles in differentiation, development, and cell death. PMA activates the differentiation and development programs, but suppresses the “cell death” module, whereas imatinib suppresses differentiation and development and activates the cell death module (Fig. 5 A and C). Integrated Pathway Analysis also revealed cell differentiation and BCR-AML signaling as the primary altered pathway influenced by PMA and imatinib (Fig. 5 B and D). PMA stimulated binding of TFs that are known to function in cell differentiation such as activator protein 1 (AP1), Finkel–Biskis–Jinkins osteosarcoma viral oncogene (FOS), E-twenty six (ETS), ETS domain-containing protein Elk-1 (ELK), B-cell–activating transcription factor (BATF), and runt-related transcription factor (RUNX), whereas imatinib executed the opposite program (Fig. 5E and Fig. S4 F and G).

Fig. 5.

Bioinformatics analysis of TF regulations induced by drugs. Functional classification of altered TFs in (A) PMA and (C) imatinib. Down-regulation groups are indicated in blue and up-regulation groups are in brown. (B and D) Volcano plot of cellular pathways responded to PMA and imatinib treatment. Z-score stands for the pathway ratio of drug-treated group to control group. (E) Alteration of differentiation, development, and proliferation-related canonical TF pathways induced by PMA and imatinib. (F) Alteration of STAT5 and E2F4 in response to imatinib treatment. Cartoon shows possible mechanism of Myc repression: In imatinib-treated CML cells, Myc’s activator STAT5 was repressed whereas Myc’s repressor E2F4 was activated, leading the competitiveness loss of Myc in interacting with Max. Intensity-based absolute quantification of protein amounts (iBAQ) is the sum of all peptide peak intensities areas divided by the number of theoretically observable full tryptic peptides. ND, not detected. (G) Regulation of oncogenes and TF components of Myc/Max/Mad network induced by imatinib. (H) Schematic model of imatinib-triggered Myc/Max/Mad network switching. Mad proteins were dramatically activated and interacted with Max to antagonize Myc/Max complexes, resulting in transcription repression.

Constitutive activation of STAT5 has been demonstrated as a mechanism for the maintenance of chronic myeloid leukemia (CML) characterized by the BCR-ABL fusion (21). We found that DNA binding activity of STAT5 is suppressed when BCR-ABL is inactivated by imatinib. In addition to known responders, imatinib dramatically activated tumor suppressors E2F transcription factor 4 (E2F4) and GATA binding protein 5 (GATA5).

Activation of v-myc myelocytomatosis viral oncogene (MYC) by BCR-AML was reported to be involved in CML progression (22) and down-regulated by imatinib (23), but the mechanism of Myc inactivation and alteration of its downstream pathway was not clear. The DNA-binding patterns of Myc activator STAT5 (24) and repressor E2F4 (25) upon imatinib treatment clearly suggest a mechanism of Myc repression (Fig. 5F).

Myc, MYC associated factor X (MAX), and mitotic arrest deficient protein (MAD) form a Myc/Max/Mad network that regulated gene activation and repression by switching between antagonistic interaction pairs of Myc–Max and Max–Mad (26). By using intensity-based absolute quantification of protein amounts (27) as an indicator of the absolute quantity of proteins, we found that Max is the most abundant protein that serves as an anchor in the Myc/Max/Mad network (Dataset S3). Mad is known to bind to Max to antagonize the Myc/Max complex (26)—the major oncogenic heterodimer—thereby inactivating Myc. We found that Mads [MAX dimerization protein 4 (MXD4) as the dominant and MXD3 as a minor variant] exhibit a dramatic increase in DNA binding when cells are treated with imatinib—they are up-regulated dramatically whereas Myc is moderately down-regulated (Fig. 5G). The result suggested a mechanism for imatinib inactivation of Myc oncogenic activity (Fig. 5H).

Discussion

TFs belong to a group of proteins that are generally of low abundance and usually underrepresented in proteome profiling experiments. In this study, we design a DNA construct of tandem TF DNA response elements, termed catTFRE, and report its applications as an affinity reagent to enrich DNA-bound TFs in mammalian cells and tissues. Combined with sensitive MS measurements, we could identify as many as 150 TFs from 50 µg of NEs in 1 h of MS measurement. Sample prefractionation and longer MS running time can further enhance the depth of TF coverage. For example, as many as 455 and 391 TFs were identified from HeLa cells and mouse liver with total MS measuring time of 12 h, respectively. Among 878 identified TFs in 11 cell types, 110 were detected in at least 10 cell types. Two hundred ninety-four TFs were identified in no more than two cell types; we consider them as cell-specific TFs (Fig. S3A).

By mutational analysis of the catTFRE sequence, we showed that enrichment of TFs is indeed largely dependent upon the specific DNA sequences of the TFRE. The fact that the number of TFs identified in experiments greatly exceeded the original design of 100 TF families may be explained by the following: (i) the 3-bp linkers may create additional binding sites, (ii) the tandem TFRE may also create additional binding sites, and (iii) the flexibility of TFs in TFRE recognition. We used the TF binding prediction tool PROMO (28) to computationally analyze the catTFRE sequence and find 132 “accidental” TF binding sites for human TFs and more than 300 additional TF binding sites for TFs of other species (Dataset S2). It has been known that each family of TFs binds a specific consensus sequence, but there are clear differences among members of a family (29–32). Our simplified generic design of TFRE may not reveal the subtle differences in DNA binding among members of a family. It will require a specialized design to reveal these subtle differences. Complexity and flexibility in TFRE recognition by TFs are starting to be appreciated, and our findings add more precedence to this topic.

With the newer generation of mass spectrometers and MS measurement time of approximately 16 h, catTFRE pull-down is able to detect similar number of TFs that can be detected by mRNA-seq in HeLa cells, proving a new tool for profiling TFs at protein level. catTFRE pull down provides more direct information about TFs than protein profiling and mRNA-seq, as it actually measures DNA binding “activity”; this is one step closer to profiling transcription activity of TFs and one unique advantage over protein profiling and mRNA-seq.

The quantitative nature of the catTFRE approach allows not only confirmation of TF existence in a cell, but also monitoring of their dynamic change in response to exogenous stimulation, as demonstrated by the inducible TNF-α/NF-κB pathway. By using catTFRE, we revealed alterations in binding of hundreds of TFs at the same time in addition to the well-known NF-κB factors. By using a specific NF-κB inhibitor, we were also able to classify TFs into NF-κB–dependent and -independent categories.

Drugs for various therapeutic applications frequently have “hidden phenotypes” that result from unexpected or unintended activities, as a result of their binding to unknown targets or unknown interactions between the intended drug target and other biochemical pathways. Such unknown activities may be harmful, leading to toxicity, or beneficial, suggesting new therapeutic applications. Discoveries of new and useful properties of drugs are usually made by serendipity, and the underlying mechanisms by which a drug produces an effect are often not known. We believe patterns of TF DNA binding can provide a diagnostic fingerprint of drug effects, and, in some cases, they provide hypotheses for the cellular mechanisms of drug responses.

The application of catTFRE for the treatment of K562 cells with PMA and imatinib illustrated the aforementioned points. Quantification of the changes in TF DNA binding after drug treatment quickly pointed to the distinctive effect of PMA and imatinib in cell differentiation, development, cell death, and BCR-AML signaling. In addition, simultaneously monitoring most of the TF families and their CoRs permitted us to suggest that one mechanism of imatinib inhibition of CML is through down-regulation of Myc by unbalanced DNA bind activities of STAT5 and E2F4. Decrease in Myc eventually triggered a molecular switch in the Max/Myc/Mad signaling network, whereby “off” position can results from coordinated down-regulation of Myc binding and up-regulation of Mad binding (26).

In summary, the catTFRE strategy presented here enables high-throughput identification and quantification of DNA binding activity for most cellular TFs. We envision that this technology will serve as a potent tool for elucidation of the molecular effects of drug actions, evaluation of drug efficacy, and concurrent discovery of secondary drug effects.

Materials and Methods

Material and Chemicals.

catTFRE DNA was synthesized by Genscript. Biotinylated catTFRE primers were synthesized by Sigma. Dynabeads (M-280 streptavidin) were purchased from Invitrogen.

Nano-Liquid Chromatography/Tandem MS Analysis for Protein Identification and Label-Free Quantification.

Tryptic peptides were separated on a C18 column, and were analyzed by LTQ-Orbitrap Velos (Thermo). Proteins were identified by using the National Center for Biotechnology Information search engine against the human or mouse National Center for Biotechnology Information RefSeq protein databases.