Serum Proteomic Profiling Reveals Differentially Expressed IGHG3 and A1AG1 as Potential Predictors of Chemotherapeutic Response in Advanced Non-small Cell Lung Cancer

Materials and Methods

Subject selection. Specimens from 14 inoperable NSCLC patients were obtained for the study during the 2016-2018 period. The study protocol was approved by the Human Research Ethics Committee, Faculty of Medicine, Prince of Songkla University, Thailand (REC: 59-011-05-1). Written informed consent forms were obtained from all participants. We included patients who were newly diagnosed with pathologically confirmed NSCLC and had completed five or six cycles of first-line chemotherapy treatment using CAR plus PTX. Patients who had undergone any lung cancer-related treatment before receiving chemotherapy, or were lost to follow-up, or died during treatment were excluded from the study. A computed tomography scan of the chest was taken before treatment for staging and after treatment, at the third cycle and onwards, to assess the response to the therapy.

Treatment evaluation. Chemotherapeutic efficacy was measured according to the guidance of Response Evaluation Criteria in Solid Tumors (RECIST 1.1) (16). The evaluated response was classified into four categories: complete response (CR) defined as the disappearance of all target lesions, partial response (PR) involving a reduction of at least 30% from the sum total of target lesions, stable disease (SD) when neither lesion reduction nor progression were present, and progressive disease (PD) when a progress of at least 20% among target lesions or the presence of new lesions was observed. CR and PR were considered responders, whereas PD and SD were classified as non-responders.

Serum preparation. Blood samples were collected at the time of diagnosis and coagulated for 30 min at room temperature before centrifugation at 3,400 × g for 10 min. The serum was filtrated through a polyvinylidene difluoride (PVDF) syringe filter with a pore size of 0.22 mm (Merck Millipore, Darmstadt, Germany), and aliquots were kept at –80°C until use. On the basis of histology (adenocarcinoma, ADC and squamous cell carcinomas, SCC), individual serum samples were pooled into four pools according to each subtype, wherein two out of the four pools were responders and non-responders, respectively.

Immunoaffinity depletion. The top 14 high-abundance proteins in crude serum consisting of albumin, immunoglobulin G, immunoglobulin A, transferrin, haptoglobin, transferrin, fibrinogen, antitrypsin, alpha2-macroglobulin, alpha1-acid glycoprotein, transthyretin, apolipoprotein AI, apolipoprotein AII, and complement C3 were depleted using the Human 14 Multiple Affinity Removal System (MARS) Column with a 4.6×100 mm dimension (Agilent Technologies, Inc., Santa Clara, CA, USA) as described by Verathamjamrus et al. (17). Serum was diluted 4-fold with buffer A, and one-tenth of the total volume of the protease inhibitor (Promega, Madison, WI, USA) was added. The ensuing mixture was filtered through a 0.22 μm Spin-X centrifuge tube filter (Corning Inc., New York, NY, USA) at 16,000 × g for 1 min at 4 °C to remove particles before injection onto the MARS column. The column was equilibrated using 100% buffer A at a flow rate of 0.125 ml/min for 10 min. Then, separation was started by injecting 80 μl of pooled sera into the column, and eluted with buffer A at 0.125 ml/min for 18 min and at 1 ml/min for 2 min. The buffer was then changed to 100% buffer B at a flow rate of 1 ml/min for 7 min to elute the bound fraction. The column was finally eluted with 100% buffer A at a flow rate of 1 ml/min for 11 min and 0.125 ml/min for 1 min to regenerate the column. Absorbance was detected at 280 nm. The collected flow-through fractions were desalted using an Amicon^® Ultra-0.5 device, at a 30 kDa cutoff (Merck Millipore), and centrifuged at 15,000 × g for 30 min at 4°C. Samples were kept at –80°C for further analysis.

In-solution digestion. The total protein content was measured using the Bradford assay (Bio-Rad Laboratories, Hercules, CA, USA). For each pool, a total 5 μg of desalted serum protein was resuspended in 50 mM ammonium bicarbonate (Fluka Chemical Corp., New York, NY, USA). The pooled samples were reduced with dithiothreitol (Thermo Fisher Scientific, Waltham, MA, USA) at a final concentration of 10 mM at 95°C for 5 min, followed by alkylation with a 1:10 ratio of 200 mM iodoacetamide (Sigma-Aldrich, St Louis, MO, USA) and incubation in the dark for 30 min at room temperature. Sequencing grade trypsin (Promega) with 1:50 w/w was added to digest proteins into peptides overnight at 37°C. The digested peptide solution was cleaned up with Zip Tip C18 Pipette Tips (Merck Millipore), and the digestion was stopped with a 1% final concentration of formic acid (Sigma-Aldrich). The samples were dried using a SpeedVac Evaporator (Labconco Corp., Kansas, MO, USA).

Label-free LC-MS/MS analysis. The concentrated samples were re-dissolved in 0.1% formic acid (Merck Millipore) and analyzed with the Thermo Scientific™ Dionex™ Ultimate™ 3000 RSLC Nano System that was set up with electrospray ionization-ion trap tandem-mass spectrometry (Bruker Corp., Billerica, MA, USA) with a captive-electrospray ion source. Peptide mixtures were injected onto the Acclaim PepMap RSLC C18 column (75 μm × 150 mm dimension) connected with an Acclaim PepMap100 C18 micro-precolumn (300 μm × 5 mm dimension) (Thermo Fisher Scientific). Separation was performed at a flow rate of 0.3 μl/min at 40°C using 0.1% formic acid in water as mobile phase A and 0.1% formic acid in acetonitrile as mobile phase B. The gradient elution was as follows: mobile phase B, 1-50% at 0-70 min, increased to 90% at 70-75 min, maintained at 90% for 5 min, set back to equilibration at 1% at 90-91 min, and left at 1% for 9 min. During data acquisition, the positive ion mode with a spray voltage of 1,300 V and a heated inlet capillary temperature of 150°C were used. The mass spectra were acquired at a 400 1,400 m/z value with an ion charge count target of 400,000. The MS scans, ranging from 50-3,000 m/z and optimized at 922 m/z, were automatically recorded for MS/MS analysis using the Bruker Compass 1.4 software (Bruker Corp.).

Data analysis for protein identification and quantification. Label-free protein quantification was processed using the Progenesis QI software version 3.1 (Nonlinear Dynamics, Ltd., Newcastle upon Tyne, UK). Triplicate raw spectra files were uploaded to the software. All chromatographs were automatically aligned, and a reference run with a slight difference in MS peaks and retention time was set for all runs. Retention times and intensities were aligned and normalized to the reference values. Default operation used ‘relative quantitation using non-conflicting peptides’. All of the MS/MS data were exported as Mascot generic files (mgf) and searched against the Swiss-Prot Database using MASCOT engine database searching algorithms, version 2.4.0 (www.matrixscience.com) in the Homo sapiens taxonomy with the following search parameters: (1) enzyme specificity as trypsin; (2) only one missed cleavage; (3) variable modifications as Carbamidomethyl (C), Oxidation (M), Phospho (ST) and Phospho (Y); (4) peptide tolerance as ±1.2; (5) MS/MS tolerance as ±0.6 Da; (6) peptide charge as +2, +3 and +4; and (7) only data with more than one peptide. The database of decoy search for reversed protein sequence was set at a threshold of 1%, allowing a missed cleavage value of 1. Relative protein abundances were normalized using only unique peptides with a MASCOT score of greater than 30. Thereafter, screening ANOVA values in triplicate runs, proteins with p<0.05 were considered for subsequent analysis. The expression levels of proteins were observed in three comparisons: in all pools of responders and non-responders, within pools of ADC, and within pools of SCC. Fold change was calculated by the ratio of average intensity in responders to average intensity in non-responders. A plus (+) value denoted up-regulated expression in responders, while a minus (–) value implied a down-regulated expression in responders.

Bioinformatics analysis. The distribution of differential proteins commonly or exclusively identified in each subtype was determined using Venn diagrams made available by an online package (http://www.interactivenn.net). The expression level of significant proteins was analyzed utilizing the hierarchical clustering method. The results were graphically presented using a web tool created by Clustvis (https://biit.cs.ut.ee/clustvis/), which was built from Pearson’s correlation and average clustering values for rows and columns (18). The attributions of 53 altered proteins were queried through the web-based Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) (http://string-db.org) to construct a protein-protein interaction network at medium interaction scores of <0.4 and a false discovery rate (FDR) of <0.05. Nodes signify proteins in the constructed network, edges refer to protein interactions, edge thickness relate to the confidence of data supported and node degrees relate to the average number of interactions.

Western blotting. The expression levels of selected proteins in serum were determined by western blotting in individual samples, which were used in the profiling. The serum was diluted using a sample buffer, and protein concentrations were measured via the Bradford assay. Thirty micrograms of protein were denatured by boiling before loading onto a polyacrylamide gel. After resolving with 10% SDS-PAGE using a current of 10 mA for each gel, the proteins were transferred to a 0.20 μm nitrocellulose membrane (Bio-Rad Laboratories) at a constant 100 V for 1 h. Membranes were blocked with 3% bovine serum albumin (Sigma-Aldrich) and incubated overnight at 4°C with primary antibodies: rabbit monoclonal anti-alpha 1 antitrypsin (1:500, Abcam, Cambridge, UK), rabbit monoclonal anti-alpha 1 acid glycoprotein (1:5,000, Abcam), rabbit monoclonal anti-afamin (1:2,000, Abcam), rabbit monoclonal anti-S100A9 (1:1,000, Abcam), and mouse polyclonal anti-IgG3 (1:5,000, Abcam). The membranes were washed with Tris-buffered saline containing 0.1% Tween 20 (Sigma-Aldrich), followed by incubation with horseradish peroxidase-conjugated anti-rabbit or anti-mouse antibodies at a dilution of 1:2,000 (Agilent Dako, Santa Clara, CA, USA) at room temperature for 1 h. Bands were detected by WesternBright ECL (Advansta, San Jose, CA, USA), and images were visualized via the ImageQuant™ digital imaging system (GE Healthcare, Chicago, IL, USA). One of the serum samples was used as a loading control to calculate the relative expression of each protein.

Cell culture and generation of resistant cell lines. Human NSCLC cell lines of H1792 and A549 obtained from the American Type Culture Collection (Rockville, MD, USA) were cultured in a Roswell Park Memorial Institute Medium (RPMI)-1640 containing 10% (v/v) fetal bovine serum (FBS) (Gibco, Grand Island, NY, USA) and incubated in a humidified incubator supplied with 5% CO₂ at 37 °C to allow for the attachment of cells as a monolayer. The cells were seeded at 1.8×10⁶ cells in a T-75 flask (Corning Inc.) overnight to achieve a confluence of 50-60%. They were then treated with each drug; PTX (Sigma-Aldrich) at a concentration of 20 nM and CAR (TOKU-E, Bellingham, WA, USA) at a concentration of 15 μM for 3 days. The treated cells were further cultured for another 3 days in the recovery phase and the cycle was repeated by gradually increasing the concentration of each drug over a period of 6 months.

Evaluation of resistance level. The sensitivity of the cell lines to chemotherapeutic drugs was determined using a 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Gibco). In brief, 1×10⁴ cells were seeded in a 96-well plate and maintained at 37 °C in a 5% CO₂ incubator overnight. Following incubation for 72 h. with the desired range of drug concentrations, 100 μl of MTT solution at a 5 mg/ml concentration was added, and the mixture was further incubated for 2 h. The MTT solutions were then removed, and 100 μl of dimethyl sulfoxide (DMSO) (Gibco) was added. The absorbance of formazan blue crystals soluble in DMSO was measured at 570 nm subtracted with background at 650 nm using a Varioskan LUX microplate reader (Thermo Fisher Scientific). Survival curves were generated, and the inhibitory concentrations (IC₅₀) for each condition were calculated from a range of drug concentrations; PTX ranging from 3.125 nM to 200 nM, and CAR ranging from 1 nM to 400 nM. The experiments were performed in duplicate for each condition, and the calculation of resistance factors determined as the ratio of IC₅₀ values of resistant cells to the IC₅₀ values of parental cells.

RNA isolation and quantitative RT-qPCR. Total RNA from parental and resistant cells was isolated using the TRIzol^® reagent (Invitrogen, Carlsbad, CA, USA) as stated in the manufacturer’s instructions. The quantity of extracted RNA was verified using a NanoDropND-1000 UV-Vis Spectrophotometer (Thermo Fisher Scientific) at an optical density (OD) ratio of 260/280 nm and 260/230 nm. Subsequently, the purified RNA was converted to cDNA by a Thermal Cycler (Bio-Rad Laboratories) using 4 μl of iScript™ supermix (Bio-rad Laboratories) in a 20 μl final volume. The cDNA products were used to compare the relative quantification of IGHG3 and ORM1 using GAPDH as a reference gene. Primer sequences are shown in Table I. The SsoFast EvaGreen supermix (Bio-Rad Laboratories) was used, and the PCR cycling parameters were as follows: 95°C for 15 sec, followed by 40 cycles at 60°C for 30 sec and at 70°C for 60 sec. A BioRad CFX96 qPCR System (Bio-Rad Laboratories) was used for the analysis. Expression was presented as cycle threshold (Ct) values, and the experiments were performed in duplicate. Each gene was quantified using the 2^–(ΔΔCt) method (19), and the relative expression level was calculated as the ratio of each targeted gene to the GAPDH gene.

Statistical analysis. The Ct values are presented as mean±standard deviation (SD) of two independent experiments. Statistical measurements employed the unpaired t-test and were reported as significant when the p-value was <0.05 using the GraphPad Prism, Version 5 (GraphPad Software, Inc., La Jolla, CA, USA).