Abstract

   Astrocytomas are the most common and aggressive type of primary brain
   tumors in adults. The World Health Organization (WHO) assorts them into
   grades, from I to IV, based on histopathological features that reflect
   their malignancy [35][1]. Alongside with tumor progression, comes an
   increased proliferation, genomic instability, infiltration in normal
   brain tissue and resistance to treatments. The high genomic instability
   forges tumor cells enhancing key proteins that avoid cells from
   collapsing and favor therapy resistance [36][2]. To explore genes and
   pathways associated with tumor progression phenotypes we analyzed gene
   expression in a panel of non-tumor and glioma cell lines, namely:
   ACBRI371, non-tumor human astrocytes; HDPC, fibroblasts derived from
   dental pulp; Res186, Res259, Res286 and UW467 that include grade I, II
   and III astrocytoma cell lines derived from pediatric tumors; and T98G,
   U343MG, U87MG, U138MG and U251MG, all derived from GBM (grade IV). We
   also profiled gene expression changes caused by exogenously induced
   replicative stress, performing RNA sequencing with camptothecin
   (CPT)-treated cells. Here we describe the RNA-sequencing data set
   acquired, including quality of reads and sequencing consistency, as
   well as the bioinformatics strategy used to analyze it. We also
   compared gene expression patterns and pathway enrichment between
   non-tumor versus lower-grade (LGG), non-tumor versus GBM, LGG versus
   GBM, and CPT-treated versus non-treated cells. In brief, a total of
   6467 genes showed differential expression and 5 pathways were enriched
   in tumor progression, while 2279 genes and 7 pathways were altered
   under the replication stress condition. The raw data was deposited in
   the NCBI BioProject database under the accession number PRJNA631805.
   Our dataset is valuable for researchers interested in differential gene
   expression among different astrocytoma grades and in expression changes
   caused by replicative stress, facilitating studies that seek novel
   biomarkers of glioma progression and treatment resistance.

   Keywords: RNAseq, Glioblastoma, Astrocytoma, Tumor progression, Gene
   expression profiling, Replicative stress, Camptothecin (CPT)

Specifications Table

   Subject Cancer Research
   Specific subject area Transcriptomic changes in different grade
   astrocytoma cells, comparing gene expression changes that occur in
   tumor progression or under replicative stress induced by the
   topoisomerase I inhibitor, Camptothecin (CPT).
   Type of data Transcriptomic data
   Figures
   Tables
   How data were acquired RNA sequencing: Bioanalyzer Instrument
   (Agilent), lllumina Sequencers Genome Analyzer IIx and NextSeq 500
   (Illumina Inc.).
   Software: FastQC, Trimmomatic, SGA, HISAT2, SAMtools, HTSeq-count,
   DESeq2
   Data format Raw: sra file format (repository link below)
   Analyzed: excel spreadsheet, tif figures
   Parameters for data collection Cells were grown under standard
   conditions or treated with CPT for 18 h, then RNA isolation and
   sequencing were performed.
   Description of data collection Total RNA was isolated using RNeasy mini
   kit (Qiagen), RNA quality was evaluated by Bioanalyzer (Agilent), rRNA
   was removed from samples and then samples were clustered and sequenced.
   Data source location Institution: Ribeirão Preto Blood Bank
   Ribeirão Preto, São Paulo, Brazil
   Coordinates: 21°11′18.1″S 47°48′17.3″W (−21.188357, −47.804813)
   Data accessibility Raw data is available at NCBI BioProject repository
   under the identification number PRJNA631805. Direct URL to data:
   [37]https://www.ncbi.nlm.nih.gov/bioproject/PRJNA631805
   [38]Open in a new tab

Value of the Data

     * •
       These data provide essential information on gene expression
       profiling of normal astrocytes and different astrocytoma cell
       lines, and of GBM cells submitted to CPT-induced replicative
       stress.
     * •
       Scientists who study gene expression regarding astrocytoma
       progression and its resistance to genotoxic treatments would
       benefit from this data set to expand their knowledge and apply new
       insights to the current clinical management of patients.
     * •
       The dataset generated here can be used to design new experiments
       and projects aiming to use the analyzed cell lines as a model to
       improve disease progression understanding.
     * •
       This RNA-sequencing dataset can be further explored for the
       identification of novel biomarkers of prognosis prediction and/or
       treatment responsiveness.
     * •
       This dataset can also be interrogated intending the identification
       of potential new target-genes for the development of new drugs
       and/or therapeutic approaches that sensitize tumor cells to the
       available treatments.

1. Data Description

   In this report we present the RNA sequencing analysis of different
   grade astrocytoma cell lines. Astrocytomas are the most common and
   aggressive type of primary brain tumors in adults. The World Health
   Organization (WHO) assorts them into grades, from I to IV, based on
   histopathological features that reflect their malignancy [39][1]. Grade
   I comprise benign curable tumors that are more frequent in children.
   Grade II are considered low-grade lesions, with restrained mitotic
   activity, but showing infiltrative capacity and tendency to progress to
   higher grades. Grade III present prominent mitotic activity, nuclear
   atypia and are also prone to undergo progression to grade IV.
   Glioblastoma (GBM), the most aggressive type of astrocytoma, is
   classified as grade IV and exhibits considerably higher mitotic
   activity, atypia, likewise angiogenesis and necrosis [[40]1,[41]2]. To
   explore genes and pathways associated with tumor progression we
   analyzed gene expression in a panel of non-tumor and glioma cell lines.
   We used two non-tumor cells and 9 cell lines representative of tumor
   progression, comprising: two non-tumor cells (HDPC and ACBRI371), two
   grade I (Res186, Res286), one grade II (Res259), one grade III (UW467)
   astrocytoma cells, and 5 GBM (T98G, U343MG, U87MG, U138MG and U251MG)
   cell lines. Here we considered cell lines from grades I, II and III as
   a group representing lower-grade glioma (LGG) and the GBM cell lines as
   representative of higher-grade glioma (HGG). Additionally, we evaluated
   the impact of CPT-induced replication stress in the transcriptome of
   two GBM cells, the most resistant (U138MG) and the most sensitive
   (U251MG) (data not shown), along with non-tumor control cells. Data
   collection was obtained in two rounds of sequencing (with Genome
   Analyzer IIx and with NextSeq 500, Illumina Inc.), to increment the
   total amount of reads produced and complete the sample set to be
   studied. We generated from 27.2 to 33.6 million of reads
   (trimmed/aligned) for libraries sequenced in the first run and from
   58.2 to 86.1 million of reads for libraries sequenced in the second run
   ([42]Table 1). With this dataset we could measure the expression levels
   of 44,608 genes among all samples evaluated. To verify the consistency
   of the generated data, we made scatter plots with the number of reads
   obtained per gene whose expression was detected in each sequenced
   sample ([43]Fig. 1). Plots depicted the maximum Pearson correlation
   coefficient (PCC) when comparing different datasets (labeled A, B, C or
   D) of the same sample, in both rounds of sequencing (#1 and #2), for
   all groups of cells analyzed: non-tumor ([44]Fig. 1A), LGG
   ([45]Fig. 1B), GBM ([46]Fig. 1C) and CPT treated GBM cells
   ([47]Fig. 1D). Among non-tumor cells, we observed decreased PCC values
   when comparing ACBRI371 cells treated or not with CPT (0.87–0.88)
   ([48]Fig. 1A). HDPC datasets showed a significant lower correlation
   with ACBRI371 cells, with PCC varying from 0.45 to 0.51 for the
   different conditions evaluated ([49]Fig. 1A). In contrast, we detected
   a high degree of similarity concerning all LGG cells that showed PCC
   values varying in between 0.98 and 0.99 amongst all comparisons
   ([50]Fig. 1B). Much larger variation was observed for the GBM cell
   lines, in which PCC values remained at 0.62 or 0.64 in all comparisons
   ([51]Fig. 1C). When GBM cells were exposed to CPT, we also observed a
   reduction in gene expression correlation between treated and
   non-treated cells, similarly to ACBRI371 cells. However, differences
   were more pronounced for U138MG (PCC=0.79) than for U251MG (PCC=0.97)
   cells ([52]Fig. 1D).

Table 1.

   RNA Sequencing metrics. The total amounts of reads produced for each
   cell line or condition analyzed were grouped into two datasets (A and
   B) for the first run (#1) or four datasets (A, B, C and D) for the
   second run (#2). Counting of the number of raw reads, timmmed reads and
   aligned reads are shown.
   First run #1
     __________________________________________________________________

   Sample Datasets* raw reads trimmed reads aligned reads total aligned
   reads per condition
   ACBRI371.A A 18,777,557 16,073,974 15,938,790 32,064,952
   ACBRI371.B B 18,812,311 16,251,936 16,126,162
   HDPC.A A 15,943,724 13,606,444 13,504,133 27,185,197
   HDPC.B B 15,979,931 13,774,367 13,681,064
   T98.A A 17,728,147 15,076,520 14,976,750 30,171,998
   T98.B B 17,785,963 15,286,117 15,195,248
   U138.A A 16,704,458 14,166,959 14,071,552 28,340,144
   U138.B B 16,744,624 14,355,064 14,268,592
   U251.A A 18,962,975 16,244,799 16,142,384 32,480,625
   U251.B B 19,015,489 16,431,033 16,338,241
   U343.A A 19,758,340 16,810,865 16,692,122 33,601,621
   U343.B B 19,791,840 17,016,541 16,909,499
   U87.A A 19,764,536 16,898,823 16,788,200 32,998,714
   U87.B B 18,881,408 16,305,409 16,210,514
     __________________________________________________________________

   Second run #2
     __________________________________________________________________

   ACBRI371 + cpt18hs.A A 21,518,098 21,095,090 20,862,592 83,405,190
   ACBRI371 + cpt18hs.B B 21,195,306 20,755,940 20,502,694
   ACBRI371 + cpt18hs.C C 21,913,608 21,476,156 21,245,612
   ACBRI371 + cpt18hs.D D 21,516,074 21,073,316 20,794,292
   R186.A A 19,595,942 19,245,970 18,417,842 73,482,118
   R186.B B 19,230,796 18,859,758 18,014,710
   R186.C C 19,960,034 19,593,568 18,755,774
   R186.D D 19,544,228 19,171,266 18,293,792
   R259.A A 18,566,902 18,175,510 16,400,986 65,617,994
   R259.B B 18,311,806 17,885,020 16,128,140
   R259.C C 18,900,300 18,492,500 16,693,170
   R259.D D 18,628,414 18,200,300 16,395,698
   R286.A A 19,534,344 19,152,002 17,501,632 69,847,434
   R286.B B 19,192,134 18,795,248 17,144,668
   R286.C C 19,876,010 19,479,954 17,808,962
   R286.D D 19,487,048 19,087,884 17,392,172
   U138 + cpt18hs.A A 22,201,702 21,832,016 14,562,900 58,244,234
   U138 + cpt18hs.B B 21,898,446 21,490,134 14,317,296
   U138 + cpt18hs.C C 22,596,730 22,211,560 14,821,870
   U138 + cpt18hs.D D 22,234,620 21,823,154 14,542,168
   U251 + cpt18hs.A A 22,296,224 21,818,672 21,557,698 86,098,692
   U251 + cpt18hs.B B 21,929,756 21,428,122 21,135,756
   U251 + cpt18hs.C C 22,702,128 22,210,046 21,951,788
   U251 + cpt18hs.D D 22,280,758 21,776,944 21,453,450
   UW467.A A 20,931,108 20,520,954 18,115,334 72,368,544
   UW467.B B 20,578,786 20,146,762 17,756,036
   UW467.C C 21,313,478 20,886,716 18,446,260
   UW467.D D 20,930,310 20,494,222 18,050,914
   [53]Open in a new tab

   *Refers to groups of reads obtained from different lanes of sequencing
   runs.

Fig. 1.

   [54]Fig 1
   [55]Open in a new tab

   Scatter plots showing the correlations among datasets of reads for each
   sample analyzed. Scatter plots were generated with the number of reads
   per gene for all genes whose expression was identified in each
   sequenced sample. Plots were clustered into four different groups,
   according to the cell line origin and/or treatment condition: non-tumor
   cells, LGG cells, GBM cells and CPT-treated GBM cells. #1 refers to
   samples sequenced in the Genome Analyzer IIx, while #2 refers to
   samples sequenced in the NextSeq 500, Illumina Inc. The letters A, B, C
   and D, accompanying the name of each cell line, refer to reads obtained
   from different lanes of sequencing for the same sample.

   Differential gene expression detected in the comparisons between the
   datasets representative of astrocytes, LGG and GBM cell lines are
   illustrated by the Volcano plots in [56]Fig. 2. We identified a total
   of 2877 genes differentially expressed between the groups of cells that
   characterize the progression from LGG to GBM, of which 1698 were down
   regulated and 1179 were up regulated ([57]Fig. 2A). We also found 2466
   altered genes by comparing ACBRI371 and LGG, being 1509 down regulated
   and 957 up regulated ([58]Fig. 2B), and 1124 differentially regulated
   genes between ACBRI371 and GBM, being 250 up regulated and 874 down
   regulated ([59]Fig. 2C). Among the cell lines submitted to a
   replicative stress condition, we identified 790 genes down regulated
   and 437 up regulated in ACBRI371 ([60]Fig. 2D), 365 down regulated and
   494 unregulated genes in U138MG ([61]Fig. 2E) and 21 down regulated and
   172 up regulated genes in U251MG ([62]Fig. 2F). According to KEGG
   analysis, when considering all the altered genes found, we encountered
   enriched pathways only for: LGG versus GBM, ACBRI371 versus LGG,
   ACBRI371 versus GBM, and in ACBRI371 and U138MG cells CPT-treated
   versus non-treated ([63]Table 2 and [64]Fig. 3). For comparisons
   representative of tumor progression, the most preeminent pathways were
   pathways in cancer, PI3K-Akt signaling, neuroactive ligand-receptor
   interaction, cell adhesion molecules and calcium signaling. While for
   the comparisons evocative of responses against replication stress, the
   most enriched pathways were PI3K-Akt signaling, axon guidance,
   neuroactive ligand-receptor interaction, retrograde endocannabinoid
   signaling, p53 signaling and apoptosis ([65]Fig. 3). The genes
   uncovered in each of these pathways are shown in Supplementary Table 1.

Fig. 2.

   [66]Fig 2
   [67]Open in a new tab

   Volcano plots displaying the degree of altered gene expression among
   groups of cell lines and treated versus non-treated cells. Volcano
   plots were produced using the fold change values and p-values generated
   through the DESeq2 R package analysis to compare the mRNA expression
   changes between LGG vs GBM (A), ACBRI371 vs LGG (B), ACBRI371 vs GBM
   (C); and Control cells vs CPT treated cells for: ACBRI371 (D), U138MG
   (E) and U251MG (F). Blue dots show genes with significant p-value.
   Green dots show genes with significant fold change. Red dots represent
   genes with significance in both p-value and fold change.

Table 2.

   KEGG pathways enrichment analysis. Differentially expressed genes
   (q-values ≤ 0.0001 and log fold change > 2 or < −2) were submitted to
   KEGG evaluation. Enriched pathways found in each comparison and details
   of the analysis results are shown.
   DOWN LGG - UP GBM
     __________________________________________________________________

   Gene Set Description Size Expect Ratio p value FDR LOGFOLD > 2
   Enrichment%
   hsa05200 Pathways in cancer 526 28.24 1.8059 0.00002066 0.001347 51
   9.69581749
   hsa04151 PI3K-Akt signaling pathway 354 19.006 2.1572 2.0102E-06
   0.00021844 41 11.5819209
   hsa05165 Human papillomavirus infection 339 18.2 1.978 0.000058394
   0.0027195 36 10.61946903
   hsa04510 Focal adhesion 199 10.684 2.8079 2.13E-07 0.000069599 30
   15.07537688
   hsa04010 MAPK signaling pathway 295 15.838 1.8942 0.00051407 0.016759
   30 10.16949153
   hsa04514 Cell adhesion molecules (CAMs) 144 7.7312 2.5869 0.000078204
   0.0031868 20 13.88888889
   hsa04512 ECM-receptor interaction 82 4.4025 3.8615 1.1754E-06
   0.00019159 17 20.73170732
   hsa04933 AGE-RAGE signaling pathway in diabetic complications 99 5.3152
   3.1984 0.000017218 0.001347 17 17.17171717
   hsa04668 TNF signaling pathway 110 5.9057 2.7092 0.00023505 0.0085141
   16 14.54545455
   hsa05144 Malaria 49 2.6307 4.1813 0.000042617 0.0023155 11 22.44897959
   DOWN GBM - UP LGG No enriched pathway
   DOWN ACBRI371 - UP GBM No enriched pathway
     __________________________________________________________________

   DOWN GBM - UP ACBRI371
   hsa04080 Neuroactive ligand-receptor interaction 277 10.866 2.1166
   0.00051907 0.022 23 8.303249097
   hsa04020 Calcium signaling pathway 183 7.1789 2.9253 8.7111E-06
   0.0014199 21 11.47540984
   hsa04723 Retrograde endocannabinoid signaling 148 5.8059 2.7558
   0.00021105 0.01376 16 10.81081081
   hsa04713 Circadian entrainment 96 3.766 3.7175 0.000019704 0.0021412 14
   14.58333333
   hsa05032 Morphine addiction 91 3.5698 3.6416 0.000048805 0.0039776 13
   14.28571429
   hsa04724 Glutamatergic synapse 114 4.4721 2.9069 0.00049148 0.022 13
   11.40350877
   hsa04727 GABAergic synapse 88 3.4521 3.1864 0.00060736 0.022 11 12.5
   hsa05033 Nicotine addiction 40 1.5692 6.3729 2.2051E-06 0.00071887 10
   25
   hsa05133 Pertussis 76 2.9814 3.3541 0.00071129 0.023188 10 13.15789474
   hsa05144 Malaria 49 1.9222 4.1619 0.00056511 0.022 8 16.32653061
   DOWN ACBRI371 - UP LGG No enriched pathway
   DOWN LGG - UP ACBRI371
   hsa04080 Neuroactive ligand-receptor interaction 277 20.88 1.8678
   0.000094512 0.0034882 39 14.07942238
   hsa04514 Cell adhesion molecules (CAMs) 144 10.854 3.3166 6.42E-11
   2.09E-08 36 25
   hsa04510 Focal adhesion 199 15 2.0666 0.000076063 0.0034882 31
   15.57788945
   hsa04512 ECM-receptor interaction 82 6.181 3.3975 4.26E-07 0.000065465
   21 25.6097561
   hsa05032 Morphine addiction 91 6.8594 3.0615 2.6922E-06 0.00021942 21
   23.07692308
   hsa04012 ErbB signaling pathway 85 6.4071 2.9654 0.000013146 0.00085709
   19 22.35294118
   hsa05133 Pertussis 76 5.7287 2.9675 0.000036973 0.0020089 17
   22.36842105
   hsa05140 Leishmaniasis 74 5.578 2.8684 0.000096299 0.0034882 16
   21.62162162
   hsa05033 Nicotine addiction 40 3.0151 4.6433 6.02E-07 0.000065465 14 35
   hsa05150 Staphylococcus aureus infection 56 4.2212 3.0797 0.00020337
   0.0061666 13 23.21428571
     __________________________________________________________________

   UP ACBRI371 CPT
   hsa04115 p53 signaling pathway 72 10,218 88,078 6.71E-03 0.00021880 9
   12.5
   hsa05210 Apoptosis 136 19,301 41,448 0.00066233 0.042506 8 5.882352941
   hsa05222 Colorectal cancer 86 12,205 57,353 0.00020349 0.030728 7
   8.139534884
   hsa04064 Small cell lung cancer 93 13,199 53,036 0.00033082 0.030728 7
   7.52688172
   hsa04210 NF-kappa B signaling pathway 95 13,482 51,920 0.00037703
   0.030728 7 7.368421053
   hsa03018 TNF signaling pathway 110 15,611 44,840 0.00091270 0.042506 7
   6.363636364
   hsa04668 RNA degradation 79 11,212 53,516 0.00085155 0.042506 6
   7.594936709
     __________________________________________________________________

   DOWN ACBRI371 CPT
   hsa05033 Neuroactive ligand-receptor interaction 277 11,126 26,065
   0.0000017977 0.000058604 29 10.46931408
   hsa04723 Retrograde endocannabinoid signaling 148 59,446 45,420
   2.30E-07 3.75E-05 27 18.24324324
   hsa04724 Glutamatergic synapse 114 45,789 48,046 5.75E-06 6.24E-04 22
   19.29824561
   hsa04727 Dopaminergic synapse 131 52,617 38,010 2.20E-03 0.000010239 20
   15.26717557
   hsa05032 Cell adhesion molecules (CAMs) 144 57,839 34,579 0.0000010483
   0.000042717 20 13.88888889
   hsa04713 Nicotine addiction 40 16,066 11,826 1.11E-12 3.62E-10 19 47.5
   hsa04728 GABAergic synapse 88 35,346 50,925 8.42E-05 6.86E-03 18
   20.45454545
   hsa04514 Morphine addiction 91 36,551 49,246 1.47E-04 9.61E-03 18
   19.78021978
   hsa05150 Circadian entrainment 96 38,559 46,681 3.56E-04 0.0000019319
   18 18.75
   hsa04080 Staphylococcus aureus infection 56 22,493 5335 0.0000016155
   0.000058518 12 21.42857143
   UP U138 CPT No enriched pathway
     __________________________________________________________________

   DOWN U138 CPT
   hsa04151 PI3K-Akt signaling pathway 354 73,938 24,345 0.00039531
   0.021461 18 5.084745763
   hsa04360 Axon guidance 175 36,551 49,246 1.93E-04 0.0000062837 18
   10.28571429
   hsa04015 Rap1 signaling pathway 206 43,026 37,187 0.0000055099
   0.00089812 16 7.766990291
   hsa04510 Focal adhesion 199 41,564 36,089 0.000015772 0.0017139 15
   7.537688442
   hsa04020 Calcium signaling pathway 183 38,222 36,628 0.000025978
   0.0021172 14 7.650273224
   hsa04540 Gap junction 88 18,380 48,966 0.000083868 0.0054682 9
   10.22727273
   hsa05146 Amoebiasis 96 20,051 39,899 0.00084677 0.029830 8 8.333333333
   hsa04720 Long-term potentiation 67 13,994 50,022 0.00046081 0.021461 7
   10.44776119
   hsa04971 Gastric acid secretion 75 15,665 44,686 0.00091504 0.029830 7
   9.333333333
   hsa05143 African trypanosomiasis 35 0.73102 68,397 0.00072934 0.029720
   5 14.28571429
   UP U251 CPT No enriched pathway
     __________________________________________________________________

   DOWN U251 CPT
   hsa05330 Allograft rejection 38 0.020351 98,276 0.00015027 0.020956 2
   5.263157895
   hsa05332 Graft-versus-host disease 41 0.021957 91,085 0.00017519
   0.020956 2 4.87804878
   hsa04940 Type I diabetes mellitus 43 0.023029 86,849 0.00019285
   0.020956 2 4.651162791
   hsa05320 Autoimmune thyroid disease 53 0.028384 70,462 0.00029377
   0.023756 2 3.773584906
   hsa05416 Viral myocarditis 59 0.031597 63,297 0.00036436 0.023756 2
   3.389830508
   hsa04612 Antigen processing and presentation 77 0.041237 48.5
   0.00062109 0.033746 2 2.597402597
   [68]Open in a new tab

Fig. 3.

   [69]Fig 3
   [70]Open in a new tab

   Enriched KEGG pathways in each collection of genes presenting altered
   expression in the indicated comparisons. All genes with q-values ≤
   0.0001 (adjusted p-value set to avoid identification of false positive
   enrichments) and log fold change > 2 or < −2 of each comparison were
   subjected to pathway analysis by KEGG. Comparisons that revealed
   enriched pathways are shown. A False Discovery Rate (FDR) ≤ 0.05 were
   used as threshold to select significant pathways. Graphs were plotted
   with GraphPad Prism 4.0 software.

2. Experimental Design, Materials and Methods

2.1. Cell culture and treatment

   ACBRI371 is a non-tumor human astrocyte cell line that was kindly
   donated by Prof. Dr. Elza Tiemi Sakamoto Hojo (São Paulo University,
   Ribeirão Preto, SP, Brazil). HDPC (Human Dental Pulp Cells) is a
   primary culture of fibroblasts isolated from dental pulp of a 5 years
   old boy in the laboratory of Dr. C. Costa and Dr. J. Henbling, who
   gently provided these cells to our laboratory. HDPC were cultivated in
   standard conditions, with α-MEM (Minimum Essential Medium Eagle)
   supplemented with 10% of fetal bovine serum and 100 U/mL penicillin and
   0.23 mg/mL streptomycin. HDPC was used as an outside cell culture, to
   be representative of non-brain expression patterns. The cell lines
   Res186 (grade I), Res286 (grade I), Res259 (grade II) and UW467 (grade
   III) were all derived from pediatric tumors that were first established
   by Dr. Michael Bobola (University of Washington, Seattle, WA) and
   kindly donated to our group by Dr. Fausto Rodriguez (Johns Hopkins
   University, Baltimore, MD). T98G, U343MG, U87MG, U138MG and U251MG are
   commercially available GBM cell lines and were obtained from the
   American Type Culture Collection. The non-tumor and GBM cell lines were
   grown in high-glucose DMEM, while LGG cells were grown in DMEM-F12. All
   media used were supplemented with 10% fetal bovine serum and 1%
   penicillin/streptomycin. All cellular stocks were kept in liquid
   nitrogen before thawed with the appropriate medium. They were all
   cultured up to a maximum of 75% confluence before and after plating for
   RNA isolation.

   To identify differentially expressed genes associated with tumor
   progression, we simply cultured the panel of cell lines representative
   of different grade astrocytoma and compared their RNA-sequencing
   results. For the replicative stress study, we developed preliminary
   experiments to choose adequate CPT treatment conditions and the most
   appropriate cell lines for sequencing, caring to induce maximum
   replicative stress yet keeping viable proliferating cells, and
   selecting one highly CPT-resistant and another CPT-sensitive GBM cell
   line. In summary, we conducted: (1) MTT dose-response curve to identify
   the CPT IC[50] for each GBM cell line and pinpoint the most resistant
   and sensitive cells, which were U138MG (IC[50]=3.425 nM) and U251MG
   (IC[50]=0.05 nM), respectively; ACBRI371 astrocytes presented an
   intermediate IC[50] (1.041 nM) and were also used as control cells
   (data not shown). (2) H2AX phosphorylation was also accessed in 9
   different time points of CPT-treatment at the IC50 for U251MG and
   U138MG. The peak of H2AX activation was reached around 18 h of
   treatment, which was then selected as the appropriate time point of
   analysis (data not shown). Therefore, to evaluate the impact of
   replicative stress induction on gene expression of GBM cells, we
   performed RNA-sequencing of U138MG, U251MG and ACBRI371 cell lines
   treated or not with CPT at the IC[50] for 18 h.

2.2. RNA isolation and sequencing

   Total RNA was isolated with RNeasy Mini Kit (Qiagen) following the
   manufacturer's instructions. The RNA isolation for each cell line and
   treatment condition was performed only once (one biological replicate).
   The density and purity of RNA samples were accessed by 260/280 nm
   absorbance ratios, measured with NanoDrop spectrophotometer (Thermo
   Fisher Scientific). The RNA quality was evaluated by electrophoresis in
   the Bioanalyzer Instrument (Agilent), and samples with an RNA Integrity
   Number (RIN) ≥ 7 were subsequently utilized. For the library
   construction, 300 ng of high-quality RNA and the TruSeq Stranded Total
   RNA LT Sample Prep Kit (Illumina Inc.) were applied. Firstly, the
   RiboZero technology (Illumina Inc.) was used to remove rRNA and
   preserve only poly(A) and other non-poly(A) transcripts, which were
   then fragmented. RNA fragments sized between 200 and 500 bp were
   utilized to generate the sequencing libraries. Clustering was performed
   in an automated system cBot (Illumina Inc.) and samples were sequenced
   with TruSeq SBS kit v5, single-read 72 cycles. We have produced two
   datasets, one sequenced in Genome Analyzer IIx (GAIIx, Illumina Inc.)
   and another sequenced in NextSeq 500 (Illumina Inc.) ([71]Table 1).
   These two datasets are technical replicates obtained from the same RNA
   extraction. All reagents were utilized following the manufacturer's
   protocols.

2.3. Reads mapping and normalization

   The two raw datasets obtained were qualitatively analyzed with FastQC
   [72][3]. For each RNA sample analyzed, the GAIIx sequencing generated 8
   single-read fastq files obtained from 8 lanes, while the NextSeq 500
   dataset yielded 8 paired-end fastq files obtained from 4 lanes. Raw
   reads quality filtering was accomplished by Trimmomatic [73][4],
   removing the Illumina adaptor sequences, low quality bases (phred score
   quality > 20), and reads shorter than 35 bp. Subsequently, the read
   error correction was performed by SGA in the preprocess mode (set to -q
   25 -f 20 -m 35), followed by index and then correct mode [74][5]. To
   evaluate trimming and quality control, the processed reads were
   inspected by FastQC. Furthermore, the total human transcriptome
   coverage was assessed for each single-read fastq file or paired-end
   duo. To achieve a similar coverage distribution for the two different
   sequencing runs, the GAIIx reads (8 lanes) were randomly grouped in two
   distinct replicates (A and B), whereas the NextSeq dataset (4 lanes)
   was kept as four paired-end duos fastq files, each lane representing
   one replicate (A, B, C and D). This organization was taken forward into
   the mapping step, in which reads were aligned with the Genome Reference
   Consortium Human Build 38 (GRCh38) using HISAT2 [75][6], according to a
   previously described optimization [76][7]. [77]Table 1 shows the
   statistical information for each step.

2.4. Differential gene expression

   In order to assess differential gene expression, the number of reads
   for each transcript was calculated by the HTSeq-count algorithm,
   settings were operated as -f bam -r pos -s no -a 10 -t exon -i gene_id
   -m intersection-nonempty [78][8]. We conducted a Pearson correlation
   assessment for all output data using the R function core. Then, the
   data were directed to the DESeq2 R package for differential expression
   analysis [79][9]. Genes were considered differentially expressed when
   showing an expression change greater than 2-fold, with a p-value cutoff
   of 10e^−6.

2.5. Pathway enrichment analysis

   The genes that showed with q-values ≤ 0.0001 (adjusted p-value set to
   avoid identification of false positive enrichments) and log fold change
   > 2 or < −2 for each comparison were subjected to pathway analysis
   using the KEGG database ([80]www.webgestalt.org). A False Discovery
   Rate (FDR) ≤ 0.05 was used as a threshold to select significant
   pathways. Pathway enrichment charts were plotted with GraphPad Prism
   4.0 software.

Ethics Statement

   The authors declare that this study did not involve any human or animal
   subjects.

CRediT Author Statement

   Juliana de Sousa: Investigation, Validation, Writing - Original draft
   preparation, Writing - Reviewing & Editing; Patrick da Silva: Formal
   analysis, Validation, Data curation, Visualization, Writing - Original
   draft; Rodolfo Serafim: Formal analysis, Data curation, Visualization;
   Ricardo Nociti: Formal analysis; Cristiano Moreira: Supervision; Wilson
   Silva Jr: Supervision, Resources; Valeria Valente: Conceptualization,
   Resources, Supervision and Data analysis, Project administration,
   Funding acquisition, Writing - Original draft and Reviewing.

Declaration of Competing Interest

   The authors declare that they have no known competing financial
   interests or personal relationships, which have, or could be perceived
   to have, influenced the work reported in this article.

Acknowledgments