Unbiased peptoid cell screen identifies a peptoid targeting newly appeared cell surface vimentin on tumor transformed early lung cancer cells

Abstract

To identify potential new reagents and biomarkers for early lung cancer detection we combined the use of a novel preclinical isogenic model of human lung epithelial cells comparing non-malignant cells with those transformed to full malignancy using defined oncogenic changes and our on-bead two color (red and green stained cells) (OBTC) peptoid combinatorial screening methodology. The preclinical model used normal parent lung epithelial cells (HBEC3-KT, labeled with green dye) and isogenic fully malignant transformed derivatives (labeled with a red dye) via the sequential introduction of key genetic alterations of p53 knockdown, oncogenic KRAS and overexpression of cMYC (HBEC3^{p53, KRAS, cMYC}). Using the unbiased OBTC screening approach, we tested 100,000 different peptoids and identified only one (named JM3A) that bound to the surface of the HBEC3^{p53, KRAS, cMYC} cells (red cells) but not HBEC3-KT cells (green cells). Using the JM3A peptoid and proteomics, we identified the protein bound as vimentin using multiple validation approaches. These all confirmed the cell surface expression of vimentin (CSV) on transformed (HBEC3^{p53, KRAS, cMYC}) but not on untransformed (HBEC3-KT) cells. JM3A coupled with fluorophores was able to detect and stain cell surface vimentin on very early stage lung cancers but not normal lung epithelial cells in a fashion comparable to that using anti-vimentin antibodies. We conclude: using a combined isogenic preclinical model of lung cancer and two color screening of a large peptoid library, we have identified differential expression of cell surface vimentin (CSV) after malignant transformation of lung epithelial cells, and developed a new peptoid reagent (JM3A) for detection of CSV which works well in staining of early stage NSCLCs. This new, highly specific, easy to prepare, CSV detecting JM3A peptoid provides an important new reagent for identifying cancer cells in early stage tumors as well as a resource for detection and isolating of CSV expressing circulating tumor cells.

Graphical Abstract

1. INTRODUCTION:

Targeted molecular imaging has been studied as a way to improve cancer diagnosis, both at in vitro biochemical (e.g. immunohistochemistry – IHC) and in vivo systemic (e.g. PET-CT, MRI etc.) levels. The idea here is to use a highly specific ligand as a vehicle to deliver fluorophores and imaging probes (e.g. DOTA, NOTA, etc.) onto cancer specific protein markers found in tumor tissue samples and also cancer cells in tumor microenvironment, respectively. Therefore, identifying reliable early cancer markers and highly specific ligands to target those markers play a key role in early intervention on difficult to control cancer types such as non small cell lung cancer (NSCLC). The current standard methods to identify highly specific ligands for early cancer markers mostly rely on picking an early marker from a pool of known markers and then, screening chemical libraries against those known protein markers. As an example of this approach, previously, we applied our on-bead two-color (OBTC) combinatorial cell screen to identify a high specific peptoid ligand (oligo-N-substituted glycines - a peptidomimetic [1] targeting vascular endothelial growth factor receptor-2 (VEGFR2) [2]. Next, we used this peptoid ligand to deliver imaging probes ⁶⁴Cu-DOTA in PET imaging[3] and Gd(III)-DOTA in MRI[4]. However, the sensitivity, specificity, and reproducibility issues of most of those delivering agents and less reliability of currently known early cancer markers impede the efficient utility of most of those imaging systems in clinical settings [5]. Therefore, developing novel technologies to identify reliable markers of NSCLC origination and progression, while developing ligands to target those markers at the same time would be an important new tool for early detection and treatment of NSCLC.

NSCLC initiation and development is a multistep process of specific proto-oncogene and tumor suppressor gene alterations in the cells [6]. Multiple genetic and epigenetic alterations have been identified in genome-wide analyses of lung tumors[7–9]. For example, p53 inactivation is an early mutation common in all NSCLC histological subtypes (adenocarcinoma, squamous cell and neuroendocrinal), while oncogenic KRAS mutations are found mainly in adenocarcinomas [10–12]. While targeting such key and common early mutation based markers would be ideal for cancer detection, for a variety of reasons p53 and KRAS mutation makers have not proved generally useful in detecting premalignant lesions. However, as normal cells transforming to malignant cells are driven by those key genetic alterations, it is likely there are changes (e.g. activation of new signaling cascades) occuring inside and outside of those cells responding to those key genetic alterations that could serve as such markers for the early detection of malignancy. One of the creative ways to study those new markers is to capture those “as they appear” in earliest tumor transforming cells. In recent years, we have established a protocol to transform primary human bronchial epithelial cells (HBECs) to full malignancy [13]. This protocol involves the sequential introduction of key oncogenic alterations: p53, KRAS, and cMYC in HBEC cells (Figure 1). These steps progress HBEC cells to full malignancy in a stepwise manner, resulting in immortalization (HBEC3KT) and disruption of the p53 pathway (HBEC3^p53), followed by extensive copy number changes, loss of contact inhibition, anchorage-independent growth, and tumor formation in vivo upon transplantation (HBEC3^p53,KRAS, HBEC3^{p53, KRAS, MYC}) [13]. These cell lines created at each step (HBEC3^p53, HBEC3^{p53, KRAS}, and HBEC3^{p53, KRAS, cMYC}) are expected to express early biomarkers on their cell surface responding to those key genetic alterations in tumor initiation and progression. Our hypothesis is to capture and identify those newly appearing markers on those transformed cell surfaces in an unbiased fashion, along with the identification of highly specific peptoids to target those markers. Here we apply our unique OBTC cell screening technology in an unbiased approach to target those tumor transformed cells identifying peptoid ligands that binds markers newly appeared on those cell surfaces that are absent in normal HBEC cells.

Figure 1:

Model of in vitro malignant transformation of HBECs following the stepwise introduction of common lung cancer mutations. Cell immortalization: Step 1: HBEC cells lack independent growth in soft agar, they were first immortalized by overexpressing cdK4 and hTERT to obtain HBEC3-KT cells (1). Step 2: On HBEC3-KTcells the p53 was inactivated through knockdown to obtain HBEC3^p53 cells (2). Step 3: After p53 loss and KRAS overexpression was introduced to obtain HBEC3^{p53, KRAS} cells (3). Step 4: To the cells in step 3, overexpression of cMYC resulted in EMT, and the cells obtained were HBEC3^p53, KRAS, cMYC (4).

We first targeted fully tumor transformed HBEC3^{p53, KRAS, cMYC} cells as compared to normal parent HBEC3-KT cells and captured vimentin protein as a newly appeared marker on this tumor transformed cell surface. Vimentin is a 57-kDa protein, which is known to be universally expressed on mesenchymal cells as an intermediate filament (IF) protein. In fact, vimentin is used as a marker for the epithelial-to-mesenchymal transition (EMT) for normal development and metastatic initiation [14, 15]. A broad range of cells expresses vimentin, including neuronal, renal tubular cells, endothelial cells lining blood vessels, macrophages, neutrophils, fibroblasts, and leukocytes, where vimentin is restricted to cytoplasm help building the cellular structure. Vimentin also plays roles in cell adhesion, migration, and signaling [16–19]. It constitutively possesses a central α-helical domain, capped by non-α-helical N (head)- and C (tail)-terminal end domains. Two monomers together are associated in parallel formation to form coiled-coil dimer. It is also known to form homopolymer and heteropolymer [20–22].

Vimentin has been known for its importance in initiating and progression of cancer, which includes tumorigenesis, EMT, and metastasis [23]. Vimentin was also reported as a cancer marker in blood [24, 25] and urine [26, 27]. Roles of vimentin has been shown in inter/intracellular signaling and cell cycle control pathways through in-vitro and in-vivo studies indicate that anti-vimentin therapeutic approaches also could be possible. [23, 28–33]. As an interesting phenomenon, vimentin is also reported to be translocated onto the cell membrane. This cell surface vimentin (CSV) has been targeted in metastatic breast cancer, and CSV has emerged as a reliable prognostic tool. Mesenchymal circulating tumor cells from sarcomas tumor has been detected by CSV [34–36]. The cell surface vimentin was used to detect solid tumor [37], osteosarcoma [38], human colorectal adenocarcinomas [39], liver cancer [40], and breast cancer [41]. CSV has been used to detect lung cancer in tumor circulating cells [37, 42–46]. The overexpression of vimentin is directly correlated with cancer cell growth suggesting it can serve as a potential target to treat cancer, as small molecules [47–50], peptides [31] and antibody therapeutics were used for disruption of tumor-initiating cells by targeting CSV [51, 52].

2. RESULTS AND DISCUSSION:

2. 1. On bead two color (OBTC) combinatorial cell screen, hit identification and initial confirmation

We have previously developed an on-bead two-color (OBTC) combinatorial cell-screening technology that uniquely can recognize differences between two cell surfaces (Figure 2A) [2, 53–56]. This assay is based on exposing two identical cell populations, which differ only by the defined oncogenic changes introduce to progress the cells to malignancy, and which should differentially express new cell surface receptors engendered by this malignant transformation. We would use the two cell populations (red-stained for malignant derivitive) or (green stained – for normal parental cell) to identify a particular receptor expressed by the red stained but not the green stained cells using beads each carrying a multiple copies of a different peptoid. Thus, screening a peptoid library carried on beads (one-compound - multi copies per bead). If a bead binds only to cells stained red, this indicates that the peptoid on this bead binds only to that overexpressed receptor and not to any other common cell-surface molecules (Figure 2A). If it binds to any other cell surface-molecule, it should register green cells as well. In our OBTC application in this study, we wanted to target fully tumor transformed cells HBEC3^{p53, KRAS, cMYC} To capture newly appeared markers that are supposed to be absent in normal parent HBEC3-KT cells. We stained HBEC3^{p53, KRAS, cMYC} cells in red color using red Qdots (Qtracker 655) and the control HBEC3-KT cells in green color using green Qdots (Qtarcker 565). 1.0 million cells from each color were mixed in 1:1 ratio and exposed to 100,000 tentagel beads, where each bead displays a unique peptoid (with a lot of copies), totaling the 100,000 peptoid library. The beads were incubated with cell mixture for 1.0 hour at room temperature, washed out unbound cells and observed under a fluorescent microscope using the long-pass filter. We found only one bead- out of those 100,000 screened bound specifically to red cells with no green cells. That bead was picked and identified as a potential hit which is binding to a marker present only on transformed cells (HBEC3^{p53, KRAS, cMYC}) but absent in normal cell (HBEC3-KT) (Figure 2B). The bead was boiled and washed thrice to clean the cell debris. The compound was cleaved off the bead, and the peptoid sequence was identified via MALDI-TOF MS/MS sequencing, naming it as JM3A (Figure 2C). To confirm the OBTC screening results, we resynthesized JM3A on tentagel beads, where now all beads carry JM3A (as opposed to library condition, which had one bead one compound format). We then exposed red stained HBEC3^{p53, KRAS, cMYC}cells alone, (Figure 2D), green stained HBEC3-KT cells alone (Figure 2E), and the 1:1 mixture of those red and green cells (Figure 2F) to the beads containing JM3A. Only red cells bound to those beads and not green cells, confirming the findings of OBTC screening and specificity of JM3A towards HBEC3^{p53, KRAS, cMYC} cells.

Figure 2:

(A) Pictorial representation of OBTC assay. The Transformed HBEC3^{p53, KRAS, cMYC} cells were stained using red Q-dots and normal cells HBEC3-KT were stained in green using green Q-dots. One million cells of each color were taken and mixed to 1:1 ratio and were incubated for 1.0 hour with beads (100,000 beads library) containing one-bead one-compound library at room temperature. (B) We identified one red cells bound bead as a potential hit which is supposed to be binding to a marker present only on transformed cells (HBEC3^{p53, KRAS, cMYC}) and absent in normal cell (HBEC3-KT). (C) Chemical structure of the peptoid JM3A. (D) Red stained HBEC3^{p53, KRAS, cMYC} cells with Qdot 655 were exposed to beads displaying JM3A. (E) Green stained HBEC3-KT cells with Qdot 565 were exposed to beads displaying JM3A. (F) 1:1 mixture of green and red cells were incubated with beads carrying JM3A.

2.2. Proteomics-based JM3A target identification as vimentin

The OBTC assay helped us identifying JM3A peptoid that binds to a newly appeared marker on fully tumor transformed HBEC3^{p53, KRAS, cMYC} cell surface, which was absent on the surface of the untransformed normal cells. The next step is to identify that marker, and we applied standard protein cross-linking, pulldown and proteomics analysis approach as we also successfully practiced before [54]. For the pulldown assay, the JM3A was synthesized with Biotin on its C-terminal portion (to attach to magnetic beads) and benzophenone (as the photoaffinity probe) on its N-terminal portion (Supplementary Figure S3). The compound was first mounted on streptavidin magnetic beads, followed by incubation of the cells HBEC3^{p53, KRAS, cMYC} and HBEC3-KT separately. Bound cells were isolated using magnetic beads, cross-linked using UV, the cells lysed, eluted proteins separated by electrophoresis, and visualized by silver staining [57] (Figure 3). The unique band at around 60 kDa (Figure 3) in the JM3A-bound sample was not present in the whole-cell lysis of HBEC3-KT cells or the Control-PC462 compound pulldown [53, 58–60]. (This unique band was subjected to standard proteomics analysis (see Supplimentary Materials, Pages 13–16) which suggested vimentin as the targeted marker of JM3A present in HBEC3^{p53, KRAS, cMYC} cells but absent in HBEC3-KT cells.

Figure 3:

Magnetic bead JM3A pulldown of the target, separated proteins by electrophoresis and visualized these proteins by silver staining.

2.3. Confirmation of JM3A binding to Vimentin

To directly reconfirm the JM3A - vimentin interaction, an on bead protein binding assay was performed. vimentin protein (His-tag) was screened with compound JM3A (on tentagel beads) and visualized using a secondary antibody (Anti-His Alexa 647) (Figure 4A). Non-specific interactions were evaluated by incubating anti-His Alexa 647 with the beads directly, which did not show any fluorescent signal (Figure 4B). Further, the JM3A (bead displayed) bound vimentin was competed with free JM3A pre-incubation, which reduces the signal indicating JM3A-vimentin specificity (Figure 4C). We then performed direct quantitative binding study using ELISA-like assay. Vimentin protein was coated on 96-well plates, a concentration gradient of biotinylated-JM3A was introduced and probed with streptavidin-horse radish peroxidase (HRP). We found the K_d value of binding to be 19 μM (Figure 4D). We also quantified the vimentin expression in the transformed cells using qRT-PCR, and the results showed vimentin overexpression in HBEC3^{p53,KRAS,cMYC} in comparison to other cells (Figure 4E).

Figure 4:

(A) JM3A displaying tentagel beads were incubated with vimentin protein (His-tag) followed by anti-His Alexa 647; vimentin showed binding with on-bead JM3A. (B) To check the non-specific interaction, JM3A displaying tentagel beads were incubated anti-His Alexa 647; the beads did not show any binding to the antibody. (C) To check the specificity of JM3A interaction with vimentin protein, the vimentin protein was pre-incubated with 10X JM3A and was added to the beads; a reduced binding of protein with beads were observed. (D) The quantitative ELISA-like binding study shows the K_d value of 19 μM. (E) qRT-PCR analysis of VIM gene showed fully tumor transformed HBEC3^{p53, KRAS, cMYC} cells over-express of vimentin in comparison to other transformed cells and normal HBEC-3KT cells. (F) Western blot analysis of magnetic-bead pulldown with JM3A and control PC462 on normal cells (HBEC3-KT) and transformed cells (HBEC3^{p53, KRAS, cMYC).} Vimentin was pulled down by JM3A only from HBEC3^{p53, KRAS, cMYC} (G) Western blot analysis of magnetic-bead pulldown assay samples with commercially available vimentin protein using JM3A and control peptoid PC462. JM3A vimentin pulldown band (lane 3) was disappeared when competed with JM3A and vimentin antibody (lanes 4 and 5, respectively). Control PC462 did not pulldown vimentin (lane 6). Lane 7 shows the repetition of JM3A pull down vimentin from HBEC3^{p53, KRAS, cMYC} cells.

To further reconfirm the JM3A-vimentin interaction, western blot analysis of samples from cross-linking and magnetic-bead pulldown assay with normal cells (HBEC3-KT) and transformed cells (HBEC3^{p53,KRAS,cMYC}) were performed (Figure 4E). Pulldown assay was performed on normal and transformed cells using JM3A and control PC462. All the pulldown fractions were run on western-blot and compared with whole-cell lysis of HBEC3-KT and HBEC3^{p53, KRAS, cMYC} cells. Figure 4E (lane 3) shows that the vimentin protein was pulldown only with JM3A in transformed cells but not in normal cells. This reconfirmed our pulldown assay and silver staining experiment shown in Figure 3.

Furthermore, JM3A-vimentin specificity was directly assessed using commercially available vimentin protein (Figure 4F). JM3A showed strong interaction and pulldown of vimentin in western-blot analysis (Figure 4F, lane 3). This JM3A bound band was disappeared when competed (blocked) with 10x free JM3A, and 5X free vimentin antibody (Figure 4F, lanes 4 and 5, respectively). The control non-binding peptoid PC462 did not show any interaction with vimentin (Figure 4F, lane 6). The results were further confirmed by running pure protein as reference band and JM3A transformed cells pulldown proteins.

2.4. In vitro biochemical diagnostic application of peptoid JM3A

Cell surface vimentin utilized as an early cancer marker, has been evaluated in the literature [31, 34–37, 42–46] and our highly specific peptoid JM3A can be applied as a tool to target vimentin. We first applied JM3A as an alternative strategy for antibody-based detection in standard immunohistochemical (IHC) applications in vitro. We performed IHC procedure using JM3A as the probe in early human cancer samples developed at MD Anderson Cancer Center. These samples are from stage I & II NSCLC patients, making the experiment relevant to the clinical specimen analyses. Biotin-JM3A was incubated with tumor tissue cross-sections compared to “normal” tissues (portions of lung without cancer) and visualized by streptavidin Q-dots 655 (red color) under a fluorescent microscope using a long-pass DAPI filter (Figure 5A). The red signal was detected in cancer tissues (TC303, TC429, TC453 and TC241) using JM3A (Figure 5A, panels 1–4), while no JM3A signal was detected in normal lung tissue TC533N (Figure 5A, panel 5). For comparison to an anti-vimentin antibody IHC, we performed the same experiment using a commercially available vimentin antibody instead of JM3A as the probe. The results were similar, with the same cancer tissues stained in red (Figure 5A, panels 6–9) and no significant staining found in the same normal tissue section of TC533N (Figure 5A, panel 10). The intensity of vimentin detection by antibody seems higher than that of JM3A and therefore more optimizations for JM3A staining need to be explored. However, peptoids are known for their high stability in room temperature with higher shelf-life, and biological amenabilities such as serum stability, tissue (tumor) penetration [61–64] and non-immunogenicity [65]. Synthesis and optimization of peptoids are much easier, rapid and economical in comparison to antibodies [66]. These inherent characteristics of peptoids will be helpful in optimizing and further development of JM3A as an alternative to antibodies in IHC type diagnostic applications in the future.

Figure 5:

Overexpression of vimentin was detected immunohistochemically (IHC) by using JM3A (panels 1–4) and commercially available vimentin antibody (panels 6–9) in early cancer tissues (TC303, TC429, TC453 and TC241). No significant staining in normal tissue TC533N was observed (panels 5 and 10, with JM3A and vimentin antibody, respectively). Detection was done using single long pass DAPI filter that simultaneously detect tissues (blue color) and Q-dots 655 (red color),.

We also performed several very basic in vitro experiments to prototype simple diagnostic tools such as reading cells autofluorescence markers or quantifying cell numbers using MTS assay, once those vimentin expressing cells are selectively separated using JM3A. We performed the pulldown procedure using JM3A on normal HBEC3-KT, and fully tumor transformed HBEC3^{p53, KRAS, cMYC} cells. The Biotin-JM3A compound was mounted on streptavidin-coated magnetic beads, and 1.0 million cells (HBEC3-KT and HBEC3^{p53,KRAS,cMYC} cells each) were incubated for 1.0 hour, and bound cells were pulled using a magnet. The difference of number of cells pulldown between those two cell types was quantified using standard NAD(P)H (Supplementary Figure 7A) and Dityrosine (Supplementary Figure 7B) expression detection signal. Also, we added MTS reagent to the pulled down cells, and the number of cells was quantified (Supplementary Figure 7C). In all 3 experiments (Supplementary Figures A-C), the signal of HBEC3^{p53,KRAS,cMYC} cells pulled down was significantly higher than the cells pulled down from HBEC30-KT. Using these preliminary assays, we showed that JM3A could be applied to pull down early tumor transformed cells over normal cells in a quantifiable manner using simple and standard tools. These approaches can be developed to better prototype diagnostic tools in the future.

While current data in the literature indicate the significance of further studies of vimentin as a biomarker for early cancer detection (e.g. in preneoplastic lesions), further functional analyses are needed to gauge vimentin’s function in the process of tumorigenesis. As part of this will be studies of the ability of targeting vimentin to deliver therapeutic agents on tumor sites where vimentin overexpressed in tumor cells. In this study, using an isogenic preclinical model of human lung epithelial carcinogenesis we identified the expression of cell surface vimentin protein as a cancer marker, and identified a specific peptoid JM3A, which binds with cell surface vimentin.

3. CONCLUSION

Using on-bead two-color method, we identified vimentin as a biomarker on the cell surface at an early stage of cancer developed and identified a compound (JM3A) that specifically recognizes cell surface vimentin. Through on-bead and western-blot assays, we confirmed the Vimentin-JM3A interaction specificity. We showed the utility of the JM3A peptoid for identifying and quantifying of cells overexpressing vimentin in early stage non-small cell lung but not normal lung tissues from patients. This JM3A peptoid is now ready for derivative optimization and subsequent early stage tests for diagnostic capability in both studies of tumor specimens and as an aid to imaging.

4. MATERIAL AND METHODS

4.1. Chemicals and Reagents:

TentaGel MB NH₂ resin (particle size: 140–170 μm, loading capacity: 0.2–0.3 mmol/g, 520,000 beads/g) was purchased from Rapp Polymere GmbH (Tuebingen, Germany). Rink amide resin, (particle size: 100–200 mesh, loading capacity: 0.3–0.6 mmol/g) was purchased from Chem-Impex International, Inc. (Wood Dale, IL, USA). All Fmoc-protected amino acids and 2-(1H-Benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU), Hydroxybenzotriazole (HOBt), all primary amines, bromoacetic acid, N,N-diisopropylcarbodiimide (DIC), N,N-diisopropylethylamine (DIPEA), piperidine, trifluoroacetic acid (TFA), cyanogen bromide (CNBr), Triisopropylsialine (TIS), α-cyano-4-hydroxycinnamic acid, acetonitrile (ACN), hydrochloric acid (HCl), dichloromethane (DCM) and N,N-dimethylformamide (DMF), were obtained from MilliporeSigma (Burlington, MA, USA). GIBCO enzyme-free cell dissociation buffer and Qtracker Cell Labeling Kits were obtained from ThermoFisher Scientific (Waltham, MA, USA). All chemical reagents and solvents from commercial sources were used without further purification. Five-ml disposable reaction columns (CEM Corporation, Matthews, NC, USA) were used as reaction vessels for solid-phase synthesis. Syntheses of peptoids under microwave conditions were performed in a 1000 W microwave oven with 10% power. All purifications were completed on a Waters HPLC system (Waters Corporation, MA, USA). Mass spectra were recorded on an Applied Biosystems Voyager DE Pro mass spectrometer using α-cyano-4-hydroxycinnamic acid as the matrix.

4.2. On bead two color binding assay for combinatorial library screen:

100,000 peptoid library beads were washed twice in KSFM medium containing 3% BSA (media) and then incubated in 1.0 mL (KSFM + 3% BSA) for 1.0 hour in a polypropylene tube. HBEC3^{p53, KRAS, cMYC} cells (transformed cells) and HBEC3-KT cells (normal cells) were removed from culture plates with GIBCO enzyme-free cell dissociation buffer 2.0 mL per plate for 20 minutes at 37°C. Cells were washed and suspended in KSFM media. Cells were counted and distributed in 1.5 ml microcentrifuge tubes with 1.0 × 10⁶ cells in 1.0 ml of media. Next, the cell labelling procedure was conducted as follows: 1.0 μL each of Qtracker reagents A and B were mixed in 1.5 mL microcentrifuge tubes and incubated for 5.0 minutes at room temperature. 0.2 mL of media was added to each tube and vortexed for 30 seconds. 1.0 × 10⁶ cells were added to each tube containing the labelling solution and incubated at 37°C for 60 minutes. HBEC3^{p53, KRAS, cMYC} cells were labelled with Qtracker 655 (red color) (ThermoFisher Scientific, Cat# Q25021MP) and HBEC-3KT cells were labelled with Qtracker 565 (green color) (ThermoFisher Scientific, Cat# Q25031MP). Cells were washed twice and suspended in 1.0 mL of KSFM + 3% BSA media. Labelled cells were visualized with the long-pass filter of the BX-53F fluorescence microscope (Olympus, PA) with a color camera. Both cell types were mixed in 1:1 ratio and pipetted up and down several times to break the clumps. 2.0 mL of cell suspension mixture (1.0 million each of red and green cells) was added to the tube containing 100,000 beads and incubated at room temperature with gentle shaking for 1.0 hour. During incubation, cell binding to the beads were checked from time to time at about 15 minutes intervals to make sure not to over equilibrate, which could increase the non-specific binding of cells to the beads. The beads were gently washed two times with KSFM + 3% BSA media and visualized under the fluorescent microscope using the long-pass filter.

4.3. Isolation and preparation of beads for sequencing:

Single bead containing fluorescently tagged red cells was identified using a fluorescent microscope under 10X objective magnification and removed manually with a 20 μl pipette with medium size pipette tips. Selected beads were washed three times with 1.0% SDS and boiled in the same solution for 10 minutes to strip off bound cells and proteins. Finally, the beads were washed three times with water. To cleave the compound from the bead and prepare it for MS/MS sequencing. To prepare the cleaving solution: 30 μl of cyanogen bromide (CNBr) (5.0 M in Acetonitrile (ACN)) was added to 1.0 mL of 0.1 N HCl. 50 μl of cleaving solution was added to the 1.5 mL tube, which contained the single isolated bead. The tube was incubated at 25°C for 4.0 hours. The solution was evaporated using a freeze dryer (SP Scientific, NY, USA), and the cleaved compound was suspended in 20 μL of water. MS/MS sequencing data was obtained using AB Sciex TOF/TOF 5800 machine.

4.4. Validation of on bead two color binding screening results:

After identifying the compound (JM3A) with MS/MS sequencing, it was resynthesized on TentaGel MB NH₂ beads. Three tubes of 25,000 beads with each compound (containing JM3A compounds) were prepared by washing and incubating for 1.0 hour in KSFM + 3% BSA. Two million cells each of HBEC3^{p53, KRAS, cMYC} cells were stained red in color using Qtracker 655, and HBEC-3KT cells were stained in green color using Qtracker 565. One million HBEC3^{p53, KRAS, cMYC} (red cells) were suspended in 1.0 mL of KSFM + 3% BSA media and were added to another 25,000 beads containing tube. One million HBEC-3KT cells (green cells) were suspended in 1.0 mL of KSFM + 3% BSA media and were added to 25,000 beads containing tube. To make a mixture of cells, 0.5 × 10⁶ red cells and 0.5 × 10⁶ green cells were mixed together and suspended in 1.0 mL of KSFM + 3% BSA media were added to the third tube containing 25,000 beads. The cells were incubated with the beads for 1.0 hours at room temperature. The beads were gently washed twice with KSFM + 3% BSA media and visualized under the fluorescent microscope using the long-pass filter.

4.5. Synthesis of JM3A:

JM3A was synthesized on Rink amide resin (particle size: 100–200 mesh, loading capacity: 0.3–0.6 mmol/g)/TentaGel MB NH₂ resin. 100 mg of resin was taken in 5 mL reaction column, the resin was swelled in dimethylformamide (DMF) for 1.0 hour prior to use, and the Fmoc group was deprotected by treating the resin with 2.0 mL of 20% piperidine solution in DMF twice for 10 minutes each. The resin was first coupled to Fmoc-Met-OH (5 equiv.) using 5.0 equiv. HBTU and 5.0 equiv. HOBt as coupling reagents in the presence of 10.0 equiv. of DIPEA for 2 hours. Fmoc was removed with the method described above. Subsequent amino acid Fmoc-Lys(Boc)-OH was introduced using the same peptide-coupling protocol (HBTU/HOBt/DIPEA), washing 10 times with DMF between each reaction. After removing the Fmoc group as described above, six peptoid residues were then coupled using a two-step peptoid coupling procedure (acylation and amination) under a microwave assisted synthesis protocol[66–68]. For the acylation step, beads were treated with 1.0 M bromoacetic acid (1.0 mL) and 1.5 M DIC (1.0 mL), and microwaved at 10% power (2 × 15 seconds) with gentle shaking in between for 30 seconds. After washing with DMF, beads were treated with 1.0 mL of 2-methoxyethylamine, (2.0 M), and coupling was performed by shaking at 25 °C for 2 hours. The procedure was repeated to attach the remaining five residues: piperonylamine, isobutylamine, N-Boc-1,4-butanediamine, N-Boc-1,4-butanediamine and isobutylamine, respectively. In the end, beads were washed with dichloromethane (DCM) and dried under vacuum before cleavage. Beads were then treated with a cleaving cocktail of TFA/H₂O/TIS (95%/2.5%/2.5%) for 2.0 hours. The crude compound was then purified using HPLC (Waters Corporation, MA, USA) and analyzed by MALDI-TOF (Applied Biosystems Voyager DE Pro mass spectrometer).

4.6. Synthesis of Biotin-JM3A-BP:

Synthesis of Biotin-JM3A-BP was done using the similar protocol described for JM3A. Sequence for amino acids residues for Biotin-JM3A-BP were Fmoc-Cys(Trt)-OH, Fmoc-Met-OH and Fmoc-Lys(Boc)-OH. After removing the Fmoc, six peptoid residues and benzophenone were then coupled using a two-step peptoid coupling procedure (acylation and amination) under a microwave assisted synthesis protocol. The sequence of residues is: 2-methoxyethylamine, piperonylamine, isobutylamine, N-Boc-1,4-butanediamine, N-Boc-1,4-butanediamine, isobutylamine and 4-aminobenzophenone, respectively. The compound was cleaved from the beads by treating with TFA/H₂O/TIS (95%/2.5%/2.5%) for 2.0 h. Cysteine attached JM3A-BP was obtained by purifying the mixture using HPLC. Biotin-maleimide [N-Biotinoyl-N′-(6-maleimidohexanoyl)hydrazide] was added to the purified portion of Cysteine attached JM3A-BP in 1:1 equivalent ratio in water and the pH the solution was adjusted to 7, the mixture was allowed to stir for overnight at 4 °C, and the compound was purified using HPLC to obtain Biotin-JM3A-BP.

4.7. Synthesis of Biotin-JM3A:

Synthesis of Biotin-JM3A was done using the similar protocol described for JM3A. Sequence for amino acids residues for Biotin-JM3A were Fmoc-Cys(Trt)-OH, Fmoc-Met-OH and Fmoc-Lys(Boc)-OH. After removing the Fmoc, six peptoid residues were then coupled; sequence of residues is: 2-methoxyethylamine, piperonylamine, isobutylamine, N-Boc-1,4-butanediamine, N-Boc-1,4-butanediamine and isobutylamine, respectively. The compound was cleaved and after purifying the mixture using HPLC; Biotin-maleimide [N-Biotinoyl-N′-(6-maleimidohexanoyl)hydrazide] attached as described above to obtain Biotin-JM3A.

4.8. Synthesis of Biotin-PC462-BP:

Synthesis of Biotin-PC462-BP was done using the similar protocol described for JM3A. Sequence for amino acids residues for Biotin-PC462-BP were Fmoc-Cys(Trt)-OH, Fmoc-Met-OH, Fmoc-Lys(Boc)-OH and Fmoc-Gly-OH. After removing the Fmoc, five peptoid residues and benzophenone were then coupled using a two-step peptoid coupling procedure (acylation and amination) under a microwave assisted synthesis protocol. The sequence of residues is: allylamine, 2-methoxyethylamine, allylamine, 2-methoxyethylamine, allylamine and 4-aminobenzophenone, respectively. The compound was cleaved from the beads by treating with TFA/H₂O/TIS (95%/2.5%/2.5%) for 2.0 h. Cysteine attached PC462-BP was obtained by purifying the mixture using HPLC. Biotin-maleimide [N-Biotinoyl-N′-(6-maleimidohexanoyl)hydrazide] was added to the purified portion of Cysteine attached PC462-BP in 1:1 equivalent ratio in water and the pH the solution was adjusted to 7, the mixture was allowed to stir for overnight at 4 °C and the compound was purified using HPLC to obtain Biotin-PC462-BP.

4.9. On-bead target pulldown:

The compound Biotin-JM3A-BP and control peptoid Biotin-PC462-BP were first mounted on streptavidin magnetic beads (Dynabeads™ MyOne™ Streptavidin T1 (Thermo Fisher)). 5 μL of streptavidin magnetic beads was added to a solution of Botin-JM3A-BP/Biotin-PC462-BP (10 μM, 1 mL) and incubated the solution for 30 min with gentle shaking. The magnetic beades were washed 3 times with PBS and blocked for 30 mins using KSFM + 3% BSA media. 1.0 million cells HBEC3^{p53, KRAS, cMYC} and HBEC3-KT were incubated for 1.0 hour (KSFM + 3% BSA media). Bound cells were separated using magnetic beads, washed three times with PBS and cross-linked the benzophenone to the bound protein using UV (λ = 354 nm). After cross-linking, the cells were lysed using NP-40 cell lysis buffer (in the presence of protease inhibitor). The bound protein was separated magnetically from the lysate, and beads were washed to remove any unbound debris. The magnetic beads were boiled in 1.0% SDS (100 μL) to denature the streptavidin release of the proteins. The released proteins were run in SDS-PAGE gel electrophoresis and visualized these proteins by silver staining.

4.10. Target identification:

The pulldown proteins were run in SDS-PAGE gel electrophoresis. The gel was stained using standard silver staining protocol[57]. The unique band in JM3A pulldown of HBEC3^{p53, KRAS, cMYC} was submitted to the Proteomics and Metabolomics Facility at MD Anderson Cancer Center for MS/MS protein sequencing analysis.

4.11. On-bead target pulldown and western blotting:

On-bead target pulldown was performed as described above, and collected proteins were using SDS-PAGE gel (Mini-PROTEAN TGX Gels) (Bio-Rad, CA, USA) and transferred onto nitrocellulose membranes. The membrane was blocked for 1.0 hour using blocking buffer. Membranes were then probed with anti-Vimentin (Vimentin Rabbit PolyAB, Proteintech, IL, USA) (1:1000) primary antibodies and visualized using appropriate secondary antibody: the membrane is washed with the wash buffer and incubated with 1:5000 Eu-labelled anti-rabbit antibody (Eu-Anti-Rabbit IgG, Molecular Devices) for 1.0 hour. The membrane was washed and visualized using the SpectraMax i3 spectrophotometer (Molecular Devices, CA, USA).

4.12. Magnetic bead binding assay:

The compound Biotin-JM3A was first mounted on streptavidin magnetic beads (Dynabeads™ MyOne™ Streptavidin T1 (Thermo Fisher)), followed by incubation of the 1.0 million cells HBEC3^{p53, KRAS, cMYC} and HBEC3-KT each for 1.0 hour (KSFM + 3% BSA media). Bound cells were separated using magnetic beads. The cells were quantified in three ways (A) The bound cells HBEC3^{p53, KRAS, cMYC} and HBEC3-KT were lysed and quantified by cells autofluorescence marker NAD(P)H with Excitation (Ex) at 340 nm and Emission (Em) at 450 nm. (B) Also, the bound cells were quantified using another cell autofluorescence marker Dityrosine (Ex-325 nm and Em-400 nm). (C) Ratio of bound HBEC3^{p53, KRAS, cMYC} and HBEC3-KT cells with JM3A was also quantified using MTS assay, to the bound portions of cells MTS reagent (Promega, WI, USA) was added and incubated for 2 hours, the absorbance at 490nm was recorded using the SpectraMax i3 spectrophotometer.

4.13. ELISA-like binding assay:

The 96-well nickel coated white plates (Pierce ™) were incubated with 50μL of 1μg/mL his-tagged human vimentin protein (MilliporeSigma) for 1 h in Tris-buffered saline with 0.05% tween 20 (TBST) at room temperature. After washing three times with TBST, the 100μL of StartingBlock blocking buffer (Thermo Fisher Scientific) was added for blocking. 50μl of biotinylated JM3A with different concentrations from 100nM to 100μM were prepared in blocking buffer and added in each well for incubating 1 h at room temperature. After removing solutions and washing with TBST, the 50μl of streptavidin-horse radish peroxidase (HRP) (Biolegend) was prepared at a 1:1000 dilution in blocking buffer and was added for 1 h incubation. Following the similar washing step, the 50 μL of SuperSignal ELISA Pico Chemiluminescent substrate (Thermo Fisher Sicentific) was added and the luminescence signal was detected at all wavelength using the microplate reader (Molecular Device). The result of JM3A binding vimentin was statistically analyzed using Graphpad Prism 8 software.

4.14. Immunohistochemistry:

The parafilm coated tissue slides were heated for 10 mins, the slides were next incubated in Xylene for 5 minutes (Twice), next slides were washed with 100% ethanol (EtOH) for 10 minutes (mins). Followed by 95% ethanol (in dd water) wash for 10 mins. Next, slides were washed with 75% ethanol (in dd water) wash for 10 mins, followed by washing with dd H₂O for 10 mins. Next, the slides were placed in citrate buffer solution (pH 6) and were heated to 95–98 °C for 10 mins, slides were allowed to cool till room temperature in citrate buffer solution. The slides were washed with ddH₂O three times (5 mins each) and incubated in 3% hydrogen peroxide for 10 min, washed the slides dd water two times (5 mins each) followed by two times with wash buffer (PBST) for 5 min each. The tissue section was blocked with 300 μl of blocking solution (2% BSA in PBS) for 1.0 hour at room temperature in a hydration chamber. The blocking solution was removed, and 300 μl of Biotin-JM3A (25μM) compound was added and incubated for 1.0 hour at room temperature in the hydration chamber. The tissue sections were washed 5 times with wash buffer Streptavidin Q-dots 655 (1/200 dilution in PBS) (Thermo fisher) and incubated for 1.0 hour at room temperature in the hydration chamber. Sections were washed 5 times with wash buffer for 5 min each and covered the sections using cover slides. The slides were visualized under a fluorescent microscope (Olympus BX-51) using long-pass DAPI filter. Basic clinical details (age, gender, histology, stage, smoking status) of the patients are in Supporting table ST1.