Identification of potential glioma drug resistance target proteins based on ultra-performance liquid chromatography-mass spectrometry differential proteomics

Materials

U251 and U87 cell lines were purchased from Tianjin Genink Biotechnology Co., Ltd. (Tianjin, China), DMEM and fetal bovine serum were purchased from Gibco (Waltham, MA, USA), and the Cell Counting Kit-8 was purchased from Dojindo Laboratories, Kumamoto, Japan.

Screening of the U251 U87-TMZ cell line

Glioma cells in the logarithmic phase were added with 200 µM TMZ solution (diluted with DMEM complete medium). After 24 h, the medium was replaced, and 200 µM TMZ solution was added after 2 d of culture. When glioma cells proliferate and there is no significant change in cell morphology, glioma cells were treated with a higher concentration of TMZ according to the procedure described above. Repeat the above steps as the TMZ concentration gradually increases, such as 300, 400, 500 µM. to obtain TMZ resistant cell lines for 60 days (Lee, 2016). Viability curve of U87-TMZ and U251-TMZ for TMZ resistance compared with their parental cell lines were demonstrated in Fig. S1.

Liquid-phase MS

The type of liquid chromatography was EASY-nLC1200 (Thermo Fisher Scientific) and the tandem mass spectrometer was Q Exactive HF (Thermo Fisher Scientific). The peptides were digested with phase A (0.1% formic acid) loaded onto a self-made Trap column (100 um × 2 cm C18 packing specification is 3umline 120A). Different gradients of phase B (80% acetonitrile, 0.1% formic acid, and 19.9% water) were used to elute the trap column. The eluted peptides were separated using an analytical column (150 um × 15 cm C18 packing specification, 1.9 um120A) to form a charged spray, which was detected using MS. The B phase gradient was 0 min, 7%; 8 min, 14%; 60 min,28%; 72 min, 42%; 73 min, 95%; 78 min, 95%; and 79 min, 30%, and the flow rate was 600 nL/min. Mass spectrum parameters: The first-order detector type was an orbital trap (Orbitrap) with a resolution of 120,000; Orbitrap scan range, 300–1,400 m/z; max injection time, 80 ms; and AGC Target, 3E6. The second stage adopted a data-dependent scanning mode and the top 20 precursor ions were selected for secondary acquisition in each scan. The secondary detector was the same Orbitrap; resolution, 15,000 at 120 m/z; maximum injection time, 19 ms; AGC Target, 2E4; scan range, 200–2,000 m/z; included charge state, 2–6; collision energy duration, 15 s; mass tolerance, ±7 ppm; isolation window, 1.6 m/z; activation type, HCD; and normalized collision energy, 27%.

Data processing and statistical analysis

Microsoft Excel 2016 was used for preliminary data collation. We performed all statistical analyses using GraphPad Prism 8. We used unpaired t test for comparison of cell viability between cell lines. The results were expressed as means ± standard deviations, and P < 0.05 was considered significant. We considered p < 0.05 to indicate statistical significance *** p < 0.001 **** p < 0.0001.

MaxQuant software (version 1.6.1.0; https://www.maxquant.org/) was used to search the database of the original data collected using MS. Restriction endonuclease digestion method: trypsin/P, and a maximum of two missing sites were allowed. Variable modification: oxidation (M) and acetyl (protein N-term). Fixed modification: carbamidomethyl (C), a maximum of five variable modifications, was allowed for each peptide. The first-level quality deviation was 20 ppm. The second-level quality deviation was 20 ppm. The database was Uniprot’s SWISS database (containing 20,431 annotated proteins, released2019.07). The minimum peptide length was seven amino acids, and the maximum peptide mass was 5,000 Da. The false discovery rate (FDR) of spectrum and protein matching was set to 1% and the protein quantitative value was obtained using the iBAQ algorithm.

After the median normalization of the original quantitative value, log2 transformation was performed to ensure that it met the normal distribution criteria, and the quantitative value of each sample met the normal distribution criteria. Proteins that could be statistically analyzed were selected (quantification of at least three samples in one group) and the minimum value for vacancies worthy of interpolation was selected. The difference in protein was analyzed using the R software t-test and function, and the p value was corrected using the Benjamini–Hochberg algorithm. Proteins with adjusted P < 0.05, and fold change > 2 were considered differentially expressed proteins. The upregulated differential proteins were analyzed using the Cluster Profiler Package for GO and KEGG pathway enrichment analyses. Protein-protein interaction (PPI) analysis (STRING database, version 11.0) was also performed.

Screening of TMZ-resistant cell lines

U251 and U87 glioma cell lines in the logarithmic growth phase were added to the 200 µM TMZ solution in the beginning and cultured for 2 d followed by the addition of TMZ solution. The above procedures were repeated until the proliferation and morphology of glioma cells did not change significantly at this TMZ concentration. Replace higher concentration of TMZ sequentially and repeat the abovementioned steps to construct TMZ-resistant U251-TMZ and U87-TMZ cell lines. In addition to observing the morphological changes in cells to confirm the state of drug resistance, we also determined the OD value of cell-counting kit-8 (CCK8). As shown in Figs. 1A and 1B, there was no significant difference in morphology and quantity between the control and + TMZ groups of the U251-TMZ and U87-TMZ cell lines. The OD values of CCK8 in the U251-TMZ and U87-TMZ cell lines were significantly higher than those of their respective control groups. Therefore, we successfully constructed and screened TMZ-resistant glioma U251 and U87 cell lines.

Data quality control

To better analyze the differential protein expression between TMZ drug-resistant cell lines U251 and U87, the quantitative data were normalized to the median. A comparison of the results before and after normalization is shown in Fig. 2A. Normalized data were analyzed using Pearson’s correlation coefficient (Fig. 2B). The correlation coefficients between the samples were higher than those between the groups, and there were significant differences in protein expression between the two groups. Additionally, the quantitative data were also analyzed by cluster analysis, and the results showed that each group of biological repeats was clustered together, indicating high similarity and good data repeatability (Fig. 2C). Analysis of the coefficient of variation of the normalized data (Fig. 2D) showed that the coefficient of variation was low, and the quantitative stability of the protein was high.

Protein quantitative analysis

In this study, some of the 4,160 proteins detected were statistically available for analysis, including 3,308 in the U251 group and 3,621 in the U87 group. We screened the differentially expressed proteins in the four groups, particularly the upregulated proteins, and focused on screening and analyzing the differentially upregulated proteins in the U251-TMZ and U87-TMZ cell lines.

Cross-comparison of differential proteins between the U251-TMZ and U87-TMZ cell lines

A total of 1,403 differential proteins were detected in the U251 group, of which 467 were upregulated and 936 were down-regulated in the U251-TMZ cell line. A total of 676 differentially expressed proteins were detected in the U87 group, of which U87-TMZ upregulated 386 proteins and down-regulated 290 (see Data S1). Because of the heterogeneity of GBM, we only screened 28 proteins upregulated from the upregulated proteins of the U251-TMZ and U87-TMZ cell lines (three repetitive samples), and screened 67 down-regulated proteins from their respective down-regulated proteins. A total of 95 differential proteins were screened, as shown in the differential protein clustering thermogram (Fig. 3A). Owing to the lack of statistical significance of the down-regulated proteins, we did not perform follow-up bioinformatics analysis of the down-regulated proteins, but focused on the 28 proteins that were upregulated. For each quantifiable protein, the fold change was logarithmically calculated from a base of 2, and the P value was logarithmically calculated from a base of 10 to create a volcano plot (Fig. 3B).

Functional annotation enrichment analysis of protein

GO and KEGG pathway analysis of the U251-TMZ and U87-TMZ cell lines

A total of 28 proteins upregulated in the U251-TMZ and U87-TMZ cell lines were analyzed using GO and KEGG pathways, respectively. The possible pathways from biological process (BP), cellular component (CC), and molecular function (MF) were screened. As shown in Fig. 4A, these differentially expressed proteins were significantly upregulated in biological processes related to myeloid leukocyte activation, myeloid leukocyte-mediated immunity, leukocyte degranulation, and inflammation. Based on GO cell composition analysis, most of them were located in the cytoplasm and may be involved in vesicle transport in the intimal system. In addition, these proteins exhibited GO molecular functions of “catalytic activity” and “binding”, such as aminopeptidase activity and coenzyme binding. These results suggest that the development of TMZ drug resistance may be related to intracellular immune regulation and energy metabolism (Fu et al., 2021; St-Coeur et al., 2015).

In addition, the enrichment of the 28 protein KEGG pathways upregulated in the U251-TMZ and U87-TMZ cell lines mainly included carbon metabolism, antibiotic biosynthesis, secondary metabolite biosynthesis, glyoxylic acid and dicarboxylic acid metabolism, and endocytosis (Fig. 4B).

Differential protein interaction network

In the interaction network of proteins upregulated in the U251-TMZ and U87-TMZ cell lines (Fig. 4C), the changes in these proteins are related to immune regulation and vesicular transport, as well as to energy metabolism, such as the mitochondrial stress response.

Potential GBM markers of TMZ resistance

In the MS analysis of differentially expressed proteins in the U251/U251-TMZ and U87/U87-TMZ cell lines, 28 proteins were upregulated. Through manual screening, seven proteins with the highest differential expression in the two groups were identified: UGT1A7, BCKDK, UAP1L1, RAB3GAP1, VWA8, HLA-B, and VPS26B (Table 1). At present, there is no research on the relationship between these seven proteins and TMZ resistance. Among these proteins, only UAP1L1 was related to GBM studies, and the expression of UAP1L1 in glioma tissues was significantly upregulated. Increased expression of UAP1L1 is associated with higher tumor grade and poor prognosis. UAP1L1 may promote the proliferation of glioma cells by regulating glycosylation of key proteins (Yang et al., 2021b).

Another noteworthy protein, BCKDK, has recently been found to regulate the catabolism of branched chain amino acids (BCAA) and enhance MEK/ERK signal transduction, which is a key pathway driving the growth and proliferation of cancer cells (Xue et al., 2017). Among the remaining five candidate proteins, it has been reported that UGT1A7 may affect the progression of pancreatic cancer (Yilmaz et al., 2015). Rab3GAP1 was first isolated from the synaptic soluble section of the rat brain as an Rab3A-binding protein, and it exhibits GAP activity against Rab3 (Handley & Aligianis, 2012). Rab3GAP deficiency leads to neurological disorders (Muller et al., 2011). VWA8 is expressed in organs with high energy requirements, including those with a high density of mitochondria (Luo et al., 2017). Upregulation of VWA8 expression was negatively correlated with brain metastasis associated with breast cancer (Yuan, Wang & Cheng, 2018). HLA is necessary for T-cell activation and is associated with multiple sclerosis (Hollenbach et al., 2016). VPS26 is one of the components of the cargo recognition core of the reverse transcriptase complex, which is significantly decreased in the hippocampus of patients with Alzheimer’s disease. Changes in reverse transcriptase composition are related to an increased risk of Alzheimer’s disease (Small et al., 2005; Vagnozzi & Pratico, 2019).

Verification of potential markers of TMZ resistance

To verify the results of MS analysis of the U251/U251-TMZ and U87/U87-TMZ cell lines, we performed western blotting on seven selected markers: UGT1A7, BCKDK, UAP1L1, RAB3GAP1, VWA8, HLA-B, and VPS26B. The results showed that in the four cell lines, the above seven proteins were successfully upregulated in the U251-TMZ and U87-TMZ cell lines compared to the control groups (Fig. 5). Tet-On systems for doxycycline (Dox)-inducible gene expression were applied in our two resistant cell lines. The addition of Dox reduced the expression of these potential target genes. The cell growth was inhibited after the addition of Dox at Day 4 to reduce the target gene and was restored after removement of Dox at Day 10 (Figs. 6, S2). Furthermore, the results of RT-qPCR suggested higher expression of BCKDK and UAP1L1 in TMZ resistant glioma tissues than in TMZ sensitive glioma tissues (Fig. S3). The aforementioned results verified that upregulation of these target gene possibly mediated the TMZ resistance.