Transcriptomic analysis of breast cancer patients sensitive ...

Page 1/26

Transcriptomic analysis of breast cancer patientssensitive and resistant to chemotherapy: Lookingfor overall survival and drug resistance biomarkers Carlos Barrón-Gallardo 

University of GuadalajaraMariel García-Chagollán 

University of GuadalajaraAndrés Morán-Mendoza 

Mexican Social Security InstituteRaúl Delgadillo-Cristerna 

Mexican Social Security InstituteMaría Martínez-Silva 

Mexican Social Security InstituteAdriana Aguilar-Lemarroy 

Mexican Social Security InstituteLuis Jave-Suárez  ( [email protected] )

Mexican Social Security Institute

Research Article

Keywords: breast cancer patients, neoadjuvant chemotherapy, resistance development, differentiallyexpressed genes (DEGs), chemotherapy response, overall survival

Posted Date: December 15th, 2020

DOI: https://doi.org/10.21203/rs.3.rs-121509/v1

License: This work is licensed under a Creative Commons Attribution 4.0 International License.  Read Full License

Version of Record: A version of this preprint was published at Technology in Cancer Research &Treatment on January 1st, 2022. See the published version athttps://doi.org/10.1177/15330338211068965.

Page 2/26

AbstractNeoadjuvant chemotherapy is one important therapeutic strategy for breast cancer with the drawback ofresistance development. Chemotherapy has adverse effects that combined with resistance couldcontribute to lower overall survival. This work aimed to evaluate the molecular pro�le of patients whoreceived neoadjuvant chemotherapy to discover differentially expressed genes (DEGs) that could be usedas biomarkers of chemotherapy response and overall survival. Breast cancer patients who receivedneoadjuvant chemotherapy were enrolled in this study and according to their pathological response wereassigned as sensitive or resistant. To evaluate DEGs, GO, KEGG, and protein-protein interactions, RNAseqinformation from all patients was obtained by next-generation sequencing. A total of 1985 DEGs werefound and KEGG analysis indicated a great number of DEGs in metabolic pathways, pathways in cancer,cytokine-cytokine receptor interactions, and neuroactive ligand-receptor interactions. A selection of 73DEGs was used further for an analysis of overall survival using the METABRIC study. Seven of thoseDEGs correlated with overall survival, of them the sub-expression of C1QTNF3, CTF1, OLFML3, PLA2R1,PODN and the over expression of TUBB and TCP1 were found in resistant patients and related to patientswith lower overall survival.

IntroductionBreast cancer ranks �rst in mortality and incidence of neoplasia in women over 20 years old worldwide1.One of the �rst therapeutic approaches to treat this disease is the use of neoadjuvant chemotherapy,which has the objective of decreasing tumoral size, increase the possibilities of conservative surgery,remove micrometastases, and improve overall survival. However, despite the bene�ts of this treatment inbreast cancer, some patients develop resistance to it, making their future therapeutic approach moredi�cult. The molecular factors involved in chemoresistance are not totally elucidated so far.Nevertheless, there is a relationship between chemoresistance and lower overall survival and disease-freesurvival2,3. In this sense, there is a growing interest and urgency for �nding molecular biomarkers usefulfor the prognosis of neoadjuvant chemotherapy response.

The use of predictive biomarkers, molecules that can be easily measured and gives clues of the behaviorof the disease, is gaining importance in the clinic. Biomarkers have been used primarily for targetedtherapy, classifying subjects, and ultimately predict chemotherapy response and overall survival. Thestudy of biomarkers in breast cancer has allowed the development of panels such as Mammaprint,Rotterdam, and OncotypeDx, which can predict the risk of metastasis, disease recurrence, and responseto tamoxifen, respectively, for hormone receptors positive patients4. Most of the biomarker panelsdeveloped to date for breast cancer have focused primarily on speci�c molecular subtypes, and there iscurrently no panel able to predict the response to neoadjuvant chemotherapy. In this work, our main goalwas to evaluate the transcriptome of breast cancer patients who were resistant or sensitive toneoadjuvant chemotherapy, to �nd differentially expressed genes (DEGs) for their potential use aschemotherapy response and overall survival biomarker.

Page 3/26

ResultsA total of 41 patients were recruited for this study, nevertheless, 18 samples had poor quality orinsu�cient quantity of RNA for the analysis, and 1 patient dead during treatment which were excludedfrom the study. Therefore, only 22 samples covered the criteria for RNAsEq. All clinical characteristics aresummarized in the supplementary table 1, brie�y, the age mean was 51.2 ± 10.18 years old. All patientswere diagnosed with breast invasive ductal carcinoma. The molecular subtype distribution among thepatients was 3 luminal A, 6 luminal B, 8 Her2, 4 triple-negative, and 1 patient with non-availableinformation. The histological grade was 1 SBRI, 10 SBRII, 9 SBRIII and 2 patients with non-availableinformation. After patients completed the 6 months of neoadjuvant chemotherapy 9 were sensitive, thatis, had a pathological complete response (pCR) and 13 of them were resistant (non-pCR)(Supplementarytable 1).

Differentially Expressed Genes (DEGs) in Resistant Patients to Neoadjuvant Chemotherapy

Analysis of DEGs was performed between resistant and sensitive patients. To accomplish thecomparisons, the sensitive group was set as the reference, so the results indicate DEGs in the resistantgroup. The data for performing the analysis was taken from the ENSEMBL project. From the 39,723genes reported in the ENSEMBL database, a total of 29,739 genes had reads that aligned with itssequence, and there were 1985 DEGs with a p-value < 0.05 (1005 DEGs overexpressed and 980subexpressed) (Fig. 1a). When �ltering the data by the log2FoldChange > |1|, the number of DEGsdiminishes to 1178 (556 overexpressed and 662 subexpressed). Besides, 73 genes (53 DEGs genes withsubexpression, and 20 with overexpression) showed a padj value < 0.05 (Fig. 1b).

The most downregulated genes with a p-value < 0.05 and a log2FoldChange lower than − 5 were FLG2,LCE2B, UGT1A6, KRT1, UGT2B7, PSAPL1, KRTDAP, C1orf68, LOR, SPRR2E, KPRP, LST1, AC116533.1, andAC244489.2. Meanwhile, highly expressed genes with a log2FoldChange > 5 and a p-value < 0.05 werePVALB, CRIM1, IGHV3-43, CHRNA4, HLA-DQA2, ARHGAP11A, MUC2.

Pathways Differentially Modulated in Resistance to Neoadjuvant Chemotherapy

To highlight the physiological processes that could be related to neoadjuvant chemotherapy resistance,we performed a KEGG PATHWAY and Gene Ontology (GO) enrichment analysis from DEGs with a p-value < 0.05. The GO analysis results showed that cellular components modulated in resistant patients belongto the extracellular region, including extracellular matrix, cell periphery, and intrinsic/extrinsic componentsof the plasma membrane (Fig. 2). The molecular functions of those DEGs were related to extracellularmatrix structural constituent, binding function, receptor activity, protein binding (collagen, integrin, growthfactor, calcium ion, and signal receptor binding), transmembrane receptor protein kinase activity, andmetallopeptidase activity (Fig. 2).

Besides, the biological processes with more DEGs were cell and tissue development, extracellular matrixassembly, positive regulation of epithelial to mesenchymal transition, regulation of cellular response to

Page 4/26

transforming growth factor-beta receptor signaling pathway, homophilic cell adhesion via plasmamembrane adhesion molecules, epithelium development, regulation of angiogenesis and cell migration(Fig. 2).

The KEGG pathways analysis revealed that pathways with more DEGs were related to the regulation oftranscription, cell proliferation, and signaling transduction, furthermore, there were signaling pathwaysdifferentially modulated among the �rst 15 KEGG, among those, PI3K-Akt, MAPK, Calcium, TGF-beta, Ras,cAMP and JAK-STAT signaling pathways (Fig. 3)(Supplementary table 2).

Protein-Protein Interaction Networks Functional EnrichmentAnalysis of DEGsDEGs with a value for the p-adj lower than 0.05 were further analyzed by STRING to know the possibleinteractions among their protein products. A total of 73 genes were included in this last analysis, of them,only 68 proteins were incorporated and used to build the interactions matrix. We focus our analysis onthose protein interactions with a minimum con�dence score of 0.7. The con�dence score is de�ned asthe approximate probability that a predicted link exists between two proteins in the same metabolic mapof the KEGG database (its range is from 0 to 1), with a higher score the number of interactions diminishesamong the query proteins, but those interactions are more probable to be real. A lower score results in agreat number of interactions but also more false positives. After this analysis, 18 edges (protein-proteininteraction) were reported to have a high con�dence score (0.7), with 5 clusters identi�ed. The proteins ofthe largest cluster included COL21A1, COL3A1, MMP2, DCN, VCAN, and FBLN2, and all belong to thecategory of extracellular matrix organization of the Reactome pathways. The next cluster grouped theproteins LCE2B, LOR, FLG, KRT1, KRT15, and KRT2, all of them related to keratinocyte differentiation,epithelial cell differentiation, the formation of the corni�ed envelope, and anatomical structuredevelopment. The third cluster was made up of CLDN1 and CLDN8, which are part of the plasmamembrane. HLA-A and HLA-B form a cluster that contributes to cellular response to chemical stimulus.Finally, the last cluster was composed of UGT1A7 and UGT1A6 that are part of drug metabolism(cytochrome P450). Furthermore, there are proteins that, although do not interact with each other, belongto the plasmatic membrane and anatomical structure development (Fig. 4).

DEGs in resistant patients were associated with betterOverall SurvivalTo see if there was a relationship between the DEGs observed in resistant and sensitive patients, and theoverall survival of patients with breast cancer, the previously selected DEGs were used to perform acomparative analysis against data stored in the database of cBioPortal. We selected the METABRIC,Nature 2012 & Nat Commun 2016 database that has information about the gene expression and overallsurvival data from 2509 invasive breast carcinoma patients. Nevertheless, we selected 1904 patients forthe analysis that also had Illumina sequencing data. The data about patient status (alive or dead) was

Page 5/26

taken with a cut-off of 60 months and over- or sub-expression for each analyzed gene was set up at a z-score threshold >|1|. The results showed that the group of patients with overexpression (z-score > 1 andLogRank < 0.05) of TCP1 and TUBB had lower overall survival (Fig. 5a). Additionally, subexpression (z-score < -1 and LogRank < 0.05) of C1QTNF3, CTF1, OLFML3, PLA2R1, and PODN was also related tolower overall survival (Fig. 5b). In none of the analyzed genes the percentage at risk of 50% was reached.

DiscussionNeoadjuvant chemotherapy is the standard treatment for patients with locally advanced breast cancer, itsapplication has the advantage of diminishing the volume of the principal tumor making it easier to beremoved by surgery, besides, axillary disease, micrometastases, and circulating tumor cells are expectedto be also eliminated5. The major drawback of this therapeutic approach is that a large percentage ofpatients develop chemoresistance and this complicates their treatment6. Also, there are controversialreports about the negative effect of this therapy, indicating that an increase in metastasis was related toits application7. Given the heterogeneity of the breast cancer disease, new therapeutic strategies suggestthe individualization of the treatment8; however, this has been hampered by the high cost of currentdiagnostic and prognostic panels, most of which have been developed for speci�c molecular subtypes. Inthis sense, it is important to �nd new prognostic tools that can be applied to all subtypes of breastcancer. Genomic approaches have been proven to highlight differences among patients with distinctconditions. The application of genomic strategies in breast cancer allowed to discriminate patients in theearly stages of the disease, to prognosticate metastasis to lymph nodes in triple-negative patients, and topredict the response to adjuvant chemotherapy in endocrine responsive patients9,10. Analysis of geneexpression has allowed classifying breast cancer in different molecular subtypes, in the same way,among the molecular subtypes, the expression of hormonal receptors have been of special utility for theprognosis and prediction of response to therapy8,11. In this work, the analysis of RNAseq data ofresistance and sensitive patients to neoadjuvant chemotherapy, regardless of their molecularclassi�cation, indicated a great number of differentially expressed genes. This highlights that there arenotable differences at the level of gene expression between these two conditions and there could beindicators of chemotherapy resistance and overall survival.

After performing a Gene Ontology Enrichment Analysis, we identify that most DEGs belong to cellularcomponents related principally to the membrane and the extracellular region, which are processes thatinvolve many receptors responsible for sending signals of proliferation, survival or apoptosis, and signalsto metastasis12,13. Likewise, the KEGG analysis indicated some pathways that could be involved inresistance. One of these was the PI3K/AKT pathway which can induce mTOR (mammalian Target ofRapamycin) stimulation leading to sustained proliferative signals. Activation of PI3K/AKT has beenrelated to breast cancer and currently, there is a lot of interest in the development of inhibitors of thispathway to target breast cancer14. Besides, PI3K/AKT/mTOR pathway upregulation has been linked tochemotherapy and radiotherapy resistance15 and its inhibition overcome drug resistance in breast cancercells16,17. Another pathway suggested by our analysis was the MAPK pathway. The activation of the

Page 6/26

MAPK pathway in breast cancer is promoted by the stimulation of EGFR and HER2 receptors andcontributes to drug resistance, cancer cell survival, and invasion18,19, in addition, the MAPK pathwayplays a role in the maintenance of the breast cancer stem cells and the promotion of epithelial-to-mesenchymal transition (EMT) a process necessary for tumor migration and the development of distantmetastases20. Breast cancer stem cells have been linked as responsible for the development ofchemoresistance21.

An additional pathway suggested by the KEGG analysis was the TGF-β beta pathway, which is wellknown to act as a double edge sword during cancer development. TGF-β is a potent tumor suppressor butalso enhances invasiveness and metastasis by inducing epithelial-to-mesenchymal transition (EMT)22.The loss of MED12, a regulator of the TGF-β signaling pathway, has been related to chemoresistance inlung cancer and colon cancer23, besides, activation of TGF-β compensate the action of tyrosine kinaseinhibitors (anticancer agents) by inducing the MAPK pathway24. In hepatocellular carcinoma, TGF-βsignaling contributes to drug resistance by inducing the expression of PXR25 and in squamous cellcarcinoma, the inhibition of TGF-β results in a chemosensitive phenotype26. In breast cancer, thedetermination of protein expression levels of TGF-β pathway components has been suggested to be ofutility for the prognosis and to identify patients at increased risk for disease recurrence27.

The interactome analysis showed 3 principal clusters related to extracellular matrix organization,keratinocyte differentiation, and drug metabolism. The importance of those pathways in cancer has beendocumented. Proteins of the extracellular matrix have a role in cancer activating pathways related togrowth, proliferation, and metastasis28,29. The proteins involved in keratinocyte differentiation as keratinshave been documented to play a role in invasion, and endothelial-mesenchymal transition30,31. And thethird cluster contains genes related to drug metabolism, speci�cally, overexpression of CYP450components that have been related to drug resistance in breast cancer32.

Genes downregulated in the resistance group were also observed to have low expression in the group ofpatients with lower overall survival. Among the genes related to overall survival, the CTF1 gene thatcodi�es for a cytokine of the gp130 group and that is mainly expressed in heart, skeletal muscle, kidneys,lung, and liver33, was observed with low expression in the resistant group and the lower overall survivalgroup. The functions of CTF1 have been related to activation of the JAK-STAT pathway, promotion ofangiogenesis, and activation of cellular proliferation34. Its role in cancer was related to cell growth and IL-6 positive regulation35, however, a relationship with drug response or chemotherapy resistancemechanisms has not been reported so far.

The OLFML3 (Olfactomedin-like 3) gene also had a low expression in the group of resistance as well asin the lower overall survival group. The functions and mechanisms in which this gene is involved remainunclear, although it has been related to embryonic development36. Regarding its role in cancer, it has beenproposed that the expression of this gene increases in the tumor stroma, and in the epithelial-mesenchymal transition process37,38, furthermore, it is the target of miR-155 and BRMS1. The �rst has

Page 7/26

been associated with cell cycle, proliferation and myogenic differentiation39, and the latter with theinhibition of breast cancer cell metastasis through the recruitment of LSD2/CoREST complex and theinhibition of OLFML3 expression38. In this sense, low expression levels of OLFML3 could be associatedwith less metastasis potential, we could not analyze this hypothesis since the samples included in thisstudy were not from breast cancer patients with metastasis. In addition, changes in the expression ofBRMS1 were not observed between the study groups.

The expression of PODN (Podocan) was also observed diminished in both resistant patients and thelowest overall survival group. This gene blocks proliferation in human bladder smooth muscle cells40 andits overexpression has been associated with an increase of p21 expression and a decrease of cdk2leading to the arrest of cell cycle and the inhibition of the G1 to S phase transition40,41.

An interesting gene that regulates the TGF-β pathway and that was observed with low expression inresistant patients and patients with poor overall survival was the C1QTNF3 (C1q and tumor necrosisfactor-related protein 3). This gene has been related to stimulation of cell proliferation and induction ofantiapoptotic molecules expression in prostate cancer cells by mediating the activation of the PKCsignaling pathway42. Hofmann et al. found that C1QTNF3 has anti�brotic effects by inhibiting TGF-βbeta production43, In this work, KEGG analysis points out the possible involvement of TGF-β betapathway in the mechanisms of chemotherapy resistance. Low expression of C1QTNF3 in the resistantgroup could lead to the activation of the TFG-beta pathway, which in turn could increase proliferativesignals resulting in tumor progression44.

Likewise, the PLA2R1 (Phospholipase A2 receptor 1) gene was observed with low expression in theresistant group and the group with less overall survival. This gene has tumor-suppressor activity bypromoting apoptosis and blocking transformation45, and its down-modulation has been observed in mostcancer types46. The knockdown of this gene in prostate cancer cells increases cell proliferation but doesnot affect their sensitivity to docetaxel47. In breast cancer, negative regulation of this gene was reportedin all molecular subtypes and promoter hypermethylation of this gene has been associated withaggressive subtypes48. Regarding its role in chemotherapy, PLA2R1 regulates JAK/STAT signaling andthe targeting of PLA2R1 overcomes senescence49. All those reports indicate that this gene could play animportant role in the resistance and overall survival.

On the other hand, two genes were highly expressed in resistance and low overall survival patients, TUBB(tubulin-β), and TCP1 (T-complex protein 1). Tubulin-β beta together with Tubulin-α forms a heterodimerthat is the component of microtubules50,51. TUBB is necessary for microtubules polymerization anddepolymerization to spindle formation during mitosis52. It has been proposed that resistance to taxanes,microtubule-stabilizing agents, depends on tubulins mutations and the content of the tubulin isoforms53.Nokimura et al. evaluated the expression of tubulin beta by histochemistry in samples of patients thatreceived anthracycline and taxane as chemotherapy, they report that more than 10% of positivity for thisprotein was associated with taxane responders and less than 10% of positivity was a characteristic of

Page 8/26

taxane non-responders54, also, in their work, they showed evidence about the probable use of tubulin betaas a prognostic marker for overall survival. In our work, we found high expression of TUBB in resistantpatients, this contradictory result could be explained because we measured RNA levels instead of protein,high levels of RNA could be an indication of a compensatory mechanism after protein diminution,however, this hypothesis needs to be addressed. On the other hand, TCP1 is a protein involved in thefolding of cytoskeletal proteins, among those, tubulin. The interruption of the Tubulin-β/TCP-1 axis leadsto cellular apoptosis through caspase-dependent signaling55. TCP1 has been reported to play a role in thematuration of Cycline E, which in turn activates the Cdk2 to positively control the G1/S phase transition56.Overexpression of TCP-1 has been observed in multidrug-resistance uterine cancer cells and colorectalcancer cells57,58. In breast cancer patients has been reported that ampli�cation and/or overexpression ofTCP1 correlates with reduced overall survival measured at 300 months. In our analysis, it was observed asimilar pattern but with a cut-off of 60 months, a time that is more useful for clinicians 59.

In conclusion, this work highlights differences in the level of gene expression in patients resistant andsensitive to neoadjuvant chemotherapy. These differences further indicated that cellular componentsrelated to the extracellular region and plasma membrane were mainly involved. Furthermore, 73 DEGswere able to discriminate against patients resistant and sensitive to neoadjuvant chemotherapy. ThoseDEGS could be used as possible biomarkers of response to chemotherapy regardless of the molecularsubtype, and 7 of them were able to predict overall survival. More studies are needed to test theseputative biomarkers in a large number of breast cancer patients.

MethodologyPatient Eligibility and Selection

Patients aged 18 years and older with a diagnosis of breast cancer, candidates to receive neoadjuvantchemotherapy with tumor size > 2 cm and/or positive nodes, without previous therapy against cancer. Allmolecular subtypes were included in the analysis. Patients with metastatic cancer, insu�cient breastcancer biopsy tissue for pathological analysis, or RNA extraction were excluded. All participants providedwritten informed consent before enrollment. The study was approved by the Ethical and ResearchCommittee of the Instituto Mexicano del Seguro Social (IMSS) (number R-2013-785-061). All proceduresperformed in this study were in accordance with the ethical standards of the institutional and/or nationalresearch committee and with the 1964 Helsinki declaration and its later amendments or comparableethical standards.

Treatment Plan and Study Design

Patients with breast cancer candidates to receive neoadjuvant chemotherapy were recruited at theServicio de Oncología of Centro Médico Nacional de Occidente of IMSS. Patients were asked toparticipate and signed informed consent; thereafter a breast biopsy was taken for RNA extraction andsequencing. Neoadjuvant chemotherapy consisted of 8 cycles of taxanes plus anthracyclines for a period

Page 9/26

of 6 months. The chemotherapy treatment started with 4 cycles of Doxorubicin (60 mg/m2) or epirubicinplus cyclophosphamide (600 mg/m2) every 3 weeks followed by either paclitaxel (90 mg/m2) ordocetaxel (100 mg/m2) every 3 weeks for 12 weeks. After chemotherapy conclusion, another breastbiopsy was taken only when was needed as part of their treatment, this last biopsy was used by thepathology department to evaluate the response to treatment. Patients with residual tumor were assignedto the resistant group while patients with pathologic complete response (pCR), were assigned to thesensitive group. The pCR was set as the (absence of residual invasive and in situ cancer of the completeresected breast specimen and all sampled regional lymph nodes following completion of neoadjuvantsystemic therapy)

RNA extraction, QC, library preparation and sequencing.

Once the biopsy was taken, it was immediately submerged into 1 ml of RNAlater® RNA StabilizationReagent (Qiagen, Cat No. 76104) and incubated overnight in the reagent at 2–8 °C, then was transferredto − 80 °C for storage until processing. For RNA extraction the RNeasy Plus Mini Kit (Qiagen, Cat No.74136) was used according to the indications of the manufacturer. Total RNA integrity was determinedby using the Agilent 2100 Bioanalyzer Instrument and the Agilent RNA 6000 Nano Kit (Agilent, Cat No.5067 − 1511).

Extracted RNAs were sequenced by the Beijing Genomics Institute (BGI Genomics), brie�y, total RNA(200 ng) was puri�ed with oligo-dT beads for the obtention of mRNA, then it was fragmented into smallpieces with Fragment Buffer. mRNA was converted into cDNA using SuperScript™ III First-StrandSynthesis SuperMix (No. Cat, 18080400) (Reaction condition: 25 °C for 10 min;42 °C for 50 min 70 °C for15 min) and Second Strand Master Mix. End Repair Mix was added (30 °C for 30 min) and cDNA waspuri�ed with Ampure XP Beads (AGENCOURT). Poly-A tail addition was performed by adding an A-TailingMix (37 °C for 30 min). Adapters were added to cDNA by combining Adenylate 3'Ends DNA with RNAIndex Adapter and Ligation Mix (30 °C for 10 min). PCR ampli�cation was performed with PCR PrimerCocktail and PCR Master Mix to enrich the cDNA fragments. Then the PCR products were puri�ed withAmpure XP Beads (AGENCOURT). Validation of the library was performed in two steps, �rst, the averagemolecule length was determined using the Agilent 2100 Bioanalyzer Instrument and the Agilent DNA1000 Reagents (Agilent, Cat No. 5067 − 1504), and second, a quanti�cation of the library using TaqManProbes (Applied Bioscience) and real-time quantitative PCR (q-PCR) was done. For sequencing, thequali�ed and quanti�ed libraries were used, �rstly an ampli�cation was performed within the �ow cell onthe cBot instrument for cluster generation (HiSeq® 4000 PE Cluster Ki, Illumina). Then, the clustered �owcell was loaded onto the HiSeq4000-Sequencer for paired-end sequencing (HiSeq® 4000 SBS Kit,Illumina) with recommended read lengths 150 bp.

Analysis of Differentially Expressed Genes.

For �ltration and remotion of Illumina adapters from the reads, we used Flexbar. The alignment of theclean short reads was mapped using Kallisto and the GRCh38 version of the human genome as areference index for the generation of abundance tables. The differentially expressed genes were identi�ed

Page 10/26

using the software package DESeq2 in R 3.5.1 “Feather Spray” (R Core Team, Vienna, Austria. URLhttps://www.R-project.org/). A false discovery rate (FDR) or p-adjust < 0.05 with the Benjamini-Hochbergmethod was used for further analysis. The DEG analysis was performed based on the chemotherapyresponse setting up the sensitive group as reference. The raw and processed data were submitted toGene Expression Omnibus (GEO) repository with the number GSE162187.

Functional Gene Analysis

Gene Ontology analysis of DEGs was performed to identify characteristic biological attributes strati�edinto categories. Kyoto Encyclopedia Gene and Genome (KEGG) analysis of the DEGs was carried outonline with the bioinformatic tool KEGG Mapper. A cut-off p < 0.05 was set up as a criterion forenrichment analysis.

Functional Protein Association Analysis

Interactome analysis of p-adj < 0.5 DEGs was performed by using String version 11.0. A high con�dencescore (the approximate probability that a predicted link exists between two enzymes in the samemetabolic map in the KEGG database) > 0.7 was set up. The clusters were identi�ed by interconnectednodes. Finally, we selected the pathways among GO, KEGG, and Reactome Pathways as well as UniProtKeywords in which the evaluated proteins participate.

Overall Survival Analysis

The research was performed in cBioPortal for Cancer Genomics (https://www.cbioportal.org/) whichprovides visualization, analysis, and download of large-scale cancer genomics data sets. In this sense,we selected the dataset composed by METABRIC, Nature 2012 & Nat Commun 201660 for performingoverall survival analysis. Samples with an mRNA expression z-score threshold of >|1| were selected to beanalyzed. A cut-off of 60 months was setting up for the overall survival analysis. DEGs with an adjustedp-value < 0.05 were analyzed. A LogRank < 0.05 was taken as signi�cant.

DeclarationsAcknowledgements

This work was �nancial supported by Fondo de Investigación en Salud - IMSS(FIS/IMSS/PROT/PRIO/14/030 to LFJ-S). CAB-G is grateful for a scholarship from Consejo Nacional deCiencia y Tecnología (CONACyT)- Mexico 

Author contributions

CAB-G contributed to the caption of patients, sample processing, bioinformatic analysis, and writing ofthe paper, MG-C Contributed with sample collection, and processing, AJM-M, RD-C and MGM-Scontributed with patient recruiting, biopsies collection, biopsies analysis, and the follow up of the

Page 11/26

patients. AA-L and LFJ-S conceived the study, advised, analyzed the results and wrote and revised themanuscript.

Competing Interests Statement

The authors declare no competing interests.

References1          Bray, F. et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortalityworldwide for 36 cancers in 185 countries. CA Cancer J Clin 68, 394-424, doi:10.3322/caac.21492 (2018).

2          Guiu, S. et al. Pathological response and survival after neoadjuvant therapy for breast cancer: a30-year study. Breast 22, 301-308, doi:10.1016/j.breast.2012.07.012 (2013).

3          Krishnan, Y., Al Awadi, S., Sreedharan, P. S., Sujith Nair, S. & Thuruthel, S. Analysis of neoadjuvanttherapies in breast cancer with respect to pathological complete response, disease-free survival andoverall survival: 15 years follow-up data from Kuwait. Asia Pac J Clin Oncol 12, e30-37,doi:10.1111/ajco.12118 (2016).

4          Xin, L., Liu, Y. H., Martin, T. A. & Jiang, W. G. The Era of Multigene Panels Comes? The ClinicalUtility of Oncotype DX and MammaPrint. World J Oncol 8, 34-40, doi:10.14740/wjon1019w (2017).

5          Early Breast Cancer Trialists' Collaborative, G. Long-term outcomes for neoadjuvant versusadjuvant chemotherapy in early breast cancer: meta-analysis of individual patient data from tenrandomised trials. Lancet Oncol 19, 27-39, doi:10.1016/S1470-2045(17)30777-5 (2018).

6          Ji, X. et al. Chemoresistance mechanisms of breast cancer and their countermeasures. BiomedPharmacother 114, 108800, doi:10.1016/j.biopha.2019.108800 (2019).

7          Karagiannis, G. S. et al. Neoadjuvant chemotherapy induces breast cancer metastasis through aTMEM-mediated mechanism. Sci Transl Med 9, doi:10.1126/scitranslmed.aan0026 (2017).

8          Foulon, A., Theret, P., Rodat-Despoix, L. & Kischel, P. Beyond Chemotherapies: Recent Strategies inBreast Cancer Treatment. Cancers (Basel) 12, doi:10.3390/cancers12092634 (2020).

9          Tan, W. et al. Construction of an immune-related genes nomogram for the preoperative predictionof axillary lymph node metastasis in triple-negative breast cancer. Artif Cells Nanomed Biotechnol 48,288-297, doi:10.1080/21691401.2019.1703731 (2020).

10        Albain, K. S., Paik, S. & van't Veer, L. Prediction of adjuvant chemotherapy bene�t in endocrineresponsive, early breast cancer using multigene assays. Breast 18 Suppl 3, S141-145,doi:10.1016/S0960-9776(09)70290-5 (2009).

Page 12/26

11        Brufsky, A. M. & Dickler, M. N. Estrogen Receptor-Positive Breast Cancer: Exploiting SignalingPathways Implicated in Endocrine Resistance. Oncologist 23, 528-539, doi:10.1634/theoncologist.2017-0423 (2018).

12        Lugo-Cintron, K. M. et al. Breast Fibroblasts and ECM Components Modulate Breast Cancer CellMigration Through the Secretion of MMPs in a 3D Micro�uidic Co-Culture Model. Cancers (Basel) 12,doi:10.3390/cancers12051173 (2020).

13        R, M. B.-C. et al. Extracellular Matrix Derived from High Metastatic Human Breast Cancer TriggersEpithelial-Mesenchymal Transition in Epithelial Breast Cancer Cells through alphavbeta3 Integrin. Int JMol Sci 21, doi:10.3390/ijms21082995 (2020).

14        Costa, B., Amorim, I., Gartner, F. & Vale, N. Understanding Breast cancer: from conventionaltherapies to repurposed drugs. Eur J Pharm Sci 151, 105401, doi:10.1016/j.ejps.2020.105401 (2020).

15        Wanigasooriya, K. et al. Radiosensitising Cancer Using Phosphatidylinositol-3-Kinase (PI3K),Protein Kinase B (AKT) or Mammalian Target of Rapamycin (mTOR) Inhibitors. Cancers (Basel) 12,doi:10.3390/cancers12051278 (2020).

16        Mei, Y., Liao, X., Zhu, L. & Yang, H. Overexpression of RSK4 reverses doxorubicin resistance inhuman breast cancer cells via PI3K/AKT signalling pathway. J Biochem 167, 603-611,doi:10.1093/jb/mvaa009 (2020).

17        Hu, Y. et al. Effects of PI3K inhibitor NVP-BKM120 on overcoming drug resistance and eliminatingcancer stem cells in human breast cancer cells. Cell Death Dis 6, e2020, doi:10.1038/cddis.2015.363(2015).

18        Burotto, M., Chiou, V. L., Lee, J. M. & Kohn, E. C. The MAPK pathway across different malignancies:a new perspective. Cancer 120, 3446-3456, doi:10.1002/cncr.28864 (2014).

19        Yousefnia, S. et al. Mechanistic Pathways of Malignancy in Breast Cancer Stem Cells. Front Oncol10, 452, doi:10.3389/fonc.2020.00452 (2020).

20        Leontovich, A. A. et al. Raf-1 oncogenic signaling is linked to activation of mesenchymal toepithelial transition pathway in metastatic breast cancer cells. Int J Oncol 40, 1858-1864,doi:10.3892/ijo.2012.1407 (2012).

21        De Angelis, M. L., Francescangeli, F. & Zeuner, A. Breast Cancer Stem Cells as Drivers of TumorChemoresistance, Dormancy and Relapse: New Challenges and Therapeutic Opportunities. Cancers(Basel) 11, doi:10.3390/cancers11101569 (2019).

22        Katsuno, Y., Lamouille, S. & Derynck, R. TGF-beta signaling and epithelial-mesenchymal transitionin cancer progression. Curr Opin Oncol 25, 76-84, doi:10.1097/CCO.0b013e32835b6371 (2013).

Page 13/26

23        Huang, S. et al. MED12 controls the response to multiple cancer drugs through regulation of TGF-beta receptor signaling. Cell 151, 937-950, doi:10.1016/j.cell.2012.10.035 (2012).

24        Brunen, D. et al. TGF-beta: an emerging player in drug resistance. Cell Cycle 12, 2960-2968,doi:10.4161/cc.26034 (2013).

25        Bhagyaraj, E. et al. TGF-beta induced chemoresistance in liver cancer is modulated by xenobioticnuclear receptor PXR. Cell Cycle 18, 3589-3602, doi:10.1080/15384101.2019.1693120 (2019).

26        Brown, J. A. et al. TGF-beta-Induced Quiescence Mediates Chemoresistance of Tumor-PropagatingCells in Squamous Cell Carcinoma. Cell Stem Cell 21, 650-664 e658, doi:10.1016/j.stem.2017.10.001(2017).

27        de Kruijf, E. M. et al. The prognostic role of TGF-beta signaling pathway in breast cancer patients.Ann Oncol 24, 384-390, doi:10.1093/annonc/mds333 (2013).

28        Insua-Rodriguez, J. & Oskarsson, T. The extracellular matrix in breast cancer. Adv Drug Deliv Rev97, 41-55, doi:10.1016/j.addr.2015.12.017 (2016).

29        Pickup, M. W., Mouw, J. K. & Weaver, V. M. The extracellular matrix modulates the hallmarks ofcancer. EMBO Rep 15, 1243-1253, doi:10.15252/embr.201439246 (2014).

30        Kim, H. J., Choi, W. J. & Lee, C. H. Phosphorylation and Reorganization of Keratin Networks:Implications for Carcinogenesis and Epithelial Mesenchymal Transition. Biomol Ther (Seoul) 23, 301-312,doi:10.4062/biomolther.2015.032 (2015).

31        Karantza, V. Keratins in health and cancer: more than mere epithelial cell markers. Oncogene 30,127-138, doi:10.1038/onc.2010.456 (2011).

32        Azzariti, A., Porcelli, L., Quatrale, A. E., Silvestris, N. & Paradiso, A. The coordinated role of CYP450enzymes and P-gp in determining cancer resistance to chemotherapy. Curr Drug Metab 12, 713-721,doi:10.2174/138920011798357042 (2011).

33        Lopez-Yoldi, M., Moreno-Aliaga, M. J. & Bustos, M. Cardiotrophin-1: A multifaceted cytokine.Cytokine Growth Factor Rev 26, 523-532, doi:10.1016/j.cytogfr.2015.07.009 (2015).

34        Yang, Z. F. et al. Cardiotrophin-1 enhances regeneration of cirrhotic liver remnant afterhepatectomy through promotion of angiogenesis and cell proliferation. Liver Int 28, 622-631,doi:10.1111/j.1478-3231.2008.01687.x (2008).

35        Robledo, O. et al. Regulation of interleukin 6 expression by cardiotrophin 1. Cytokine 9, 666-671,doi:10.1006/cyto.1997.0220 (1997).

Page 14/26

36        Zeng, L. C., Han, Z. G. & Ma, W. J. Elucidation of subfamily segregation and intramolecularcoevolution of the olfactomedin-like proteins by comprehensive phylogenetic analysis and geneexpression pattern assessment. FEBS Lett 579, 5443-5453, doi:10.1016/j.febslet.2005.08.064 (2005).

37        Torres, S. et al. Proteome pro�ling of cancer-associated �broblasts identi�es novelproin�ammatory signatures and prognostic markers for colorectal cancer. Clin Cancer Res 19, 6006-6019,doi:10.1158/1078-0432.CCR-13-1130 (2013).

38        Qiu, R. et al. BRMS1 coordinates with LSD1 and suppresses breast cancer cell metastasis. Am JCancer Res 8, 2030-2045 (2018).

39        Zhao, S. et al. OLFML3 expression is decreased during prenatal muscle development andregulated by microRNA-155 in pigs. Int J Biol Sci 8, 459-469, doi:10.7150/ijbs.3821 (2012).

40        Sun, Y. et al. MiR 3180-5p promotes proliferation in human bladder smooth muscle cell bytargeting PODN under hydrodynamic pressure. Sci Rep 6, 33042, doi:10.1038/srep33042 (2016).

41        Shimizu-Hirota, R., Sasamura, H., Kuroda, M., Kobayashi, E. & Saruta, T. Functionalcharacterization of podocan, a member of a new class in the small leucine-rich repeat protein family.FEBS Lett 563, 69-74, doi:10.1016/S0014-5793(04)00250-9 (2004).

42        Hou, Q. et al. RankProd Combined with Genetic Algorithm Optimized Arti�cial Neural NetworkEstablishes a Diagnostic and Prognostic Prediction Model that Revealed C1QTNF3 as a Biomarker forProstate Cancer. EBioMedicine 32, 234-244, doi:10.1016/j.ebiom.2018.05.010 (2018).

43        Hofmann, C. et al. C1q/TNF-related protein-3 (CTRP-3) is secreted by visceral adipose tissue andexerts antiin�ammatory and anti�brotic effects in primary human colonic �broblasts. In�amm Bowel Dis17, 2462-2471, doi:10.1002/ibd.21647 (2011).

44        Katz, L. H. et al. Targeting TGF-beta signaling in cancer. Expert Opin Ther Targets 17, 743-760,doi:10.1517/14728222.2013.782287 (2013).

45        Bernard, D. & Vindrieux, D. PLA2R1: expression and function in cancer. Biochim Biophys Acta1846, 40-44, doi:10.1016/j.bbcan.2014.03.003 (2014).

46        Vindrieux, D. et al. PLA2R1 mediates tumor suppression by activating JAK2. Cancer Res 73, 6334-6345, doi:10.1158/0008-5472.CAN-13-0318 (2013).

47        Quach, N. D. et al. Role of the phospholipase A2 receptor in liposome drug delivery in prostatecancer cells. Mol Pharm 11, 3443-3451, doi:10.1021/mp500174p (2014).

48        Mitwally, N., Yousef, E., Abd Al Aziz, A. & Taha, M. Clinical Signi�cance of Expression Changes andPromoter Methylation of PLA2R1 in Tissues of Breast Cancer Patients. Int J Mol Sci 21,doi:10.3390/ijms21155453 (2020).

Page 15/26

49        Griveau, A., Wiel, C., Ziegler, D. V., Bergo, M. O. & Bernard, D. The JAK1/2 inhibitor ruxolitinib delayspremature aging phenotypes. Aging Cell 19, e13122, doi:10.1111/acel.13122 (2020).

50        Parker, A. L., Teo, W. S., McCarroll, J. A. & Kavallaris, M. An Emerging Role for Tubulin Isotypes inModulating Cancer Biology and Chemotherapy Resistance. Int J Mol Sci 18, doi:10.3390/ijms18071434(2017).

51        Downing, K. H. & Nogales, E. Crystallographic structure of tubulin: implications for dynamics anddrug binding. Cell Struct Funct 24, 269-275, doi:10.1247/csf.24.269 (1999).

52        Gadadhar, S., Bodakuntla, S., Natarajan, K. & Janke, C. The tubulin code at a glance. J Cell Sci 130,1347-1353, doi:10.1242/jcs.199471 (2017).

53        McGrogan, B. T., Gilmartin, B., Carney, D. N. & McCann, A. Taxanes, microtubules andchemoresistant breast cancer. Biochim Biophys Acta 1785, 96-132, doi:10.1016/j.bbcan.2007.10.004(2008).

54        Norimura, S. et al. Candidate biomarkers predictive of anthracycline and taxane e�cacy againstbreast cancer. J Cancer Res Ther 14, 409-415, doi:10.4103/jcrt.JCRT_1053_16 (2018).

55        Liu, Y. J., Chang, Y. J., Kuo, Y. T. & Liang, P. H. Targeting beta-tubulin/CCT-beta complex inducesapoptosis and suppresses migration and invasion of highly metastatic lung adenocarcinoma.Carcinogenesis 41, 699-710, doi:10.1093/carcin/bgz137 (2020).

56        Won, K. A., Schumacher, R. J., Farr, G. W., Horwich, A. L. & Reed, S. I. Maturation of human cyclin Erequires the function of eukaryotic chaperonin CCT. Mol Cell Biol 18, 7584-7589,doi:10.1128/mcb.18.12.7584 (1998).

57        Lin, Y. F., Tsai, W. P., Liu, H. G. & Liang, P. H. Intracellular beta-tubulin/chaperonin containing TCP1-beta complex serves as a novel chemotherapeutic target against drug-resistant tumors. Cancer Res 69,6879-6888, doi:10.1158/0008-5472.CAN-08-4700 (2009).

58        Coghlin, C. et al. Characterization and over-expression of chaperonin t-complex proteins incolorectal cancer. J Pathol 210, 351-357, doi:10.1002/path.2056 (2006).

59        Guest, S. T., Kratche, Z. R., Bollig-Fischer, A., Haddad, R. & Ethier, S. P. Two members of the TRiCchaperonin complex, CCT2 and TCP1 are essential for survival of breast cancer cells and are linked todriving oncogenes. Exp Cell Res 332, 223-235, doi:10.1016/j.yexcr.2015.02.005 (2015).

60        Pereira, B. et al. The somatic mutation pro�les of 2,433 breast cancers re�nes their genomic andtranscriptomic landscapes. Nat Commun 7, 11479, doi:10.1038/ncomms11479 (2016).

Figures

Page 16/26

Figure 1

Distribution of DEGs by p-value and Log2-FoldChange. The DEGs were plotted in the graph in which the x-axis represents the Log2 Fold Change and the y axis the Log10 of the p-value. The sensitive group wasset up as a reference for comparison. a) DEGs with a p<0.5. b) DEGs �ltered by a padj<0.05. Red colorindicates overexpressed genes, blue color indicates subexpressed genes. C1orf68 and LCE2B genes arenot showed in the plot since its Log2FoldChange is lower than the limit of the x-axis.

Page 17/26

Figure 1

Distribution of DEGs by p-value and Log2-FoldChange. The DEGs were plotted in the graph in which the x-axis represents the Log2 Fold Change and the y axis the Log10 of the p-value. The sensitive group wasset up as a reference for comparison. a) DEGs with a p<0.5. b) DEGs �ltered by a padj<0.05. Red colorindicates overexpressed genes, blue color indicates subexpressed genes. C1orf68 and LCE2B genes arenot showed in the plot since its Log2FoldChange is lower than the limit of the x-axis.

Page 18/26

Figure 2

GO analysis of DEGs. A total of 1985 DEGs with a p<0.05 were analyzed in the Pantherdb tool. The light-gray bars show the total genes reported for each of the GO terms, dark-gray bars are the number of DEGs,and the colored bars indicate the expected number of genes for each GO term in normal conditions. TheGO categories of biological process, cellular component, and molecular functions are colored by green,

Page 19/26

blue, and red respectively. The y-left axis lists the enrichment terms (FDR<0.05) for the 3 GO categories,the x-axis represents the Log2 of the DEG numbers.

Figure 2

GO analysis of DEGs. A total of 1985 DEGs with a p<0.05 were analyzed in the Pantherdb tool. The light-gray bars show the total genes reported for each of the GO terms, dark-gray bars are the number of DEGs,and the colored bars indicate the expected number of genes for each GO term in normal conditions. The

Page 20/26

GO categories of biological process, cellular component, and molecular functions are colored by green,blue, and red respectively. The y-left axis lists the enrichment terms (FDR<0.05) for the 3 GO categories,the x-axis represents the Log2 of the DEG numbers.

Figure 3

KEGG analysis of DEGs. A total of 1985 DEGs with a p<0.05 were analyzed in the KEGG mapper tool. Onthe x-axis is represented the number of DEGs, on the y-axis the name of the pathways. Red color meansoverexpressed genes and blue color means subexpressed genes in the chemotherapy resistance group.

Page 21/26

Figure 3

KEGG analysis of DEGs. A total of 1985 DEGs with a p<0.05 were analyzed in the KEGG mapper tool. Onthe x-axis is represented the number of DEGs, on the y-axis the name of the pathways. Red color meansoverexpressed genes and blue color means subexpressed genes in the chemotherapy resistance group.

Page 22/26

Figure 4

Interactome analysis. A total of 73 DEGs with a padj<0.05 were analyzed in the STRING tool. Every node(circle) represents the protein encoded by every gene. The colors indicate the functional enrichments ofthe node in GO, KEGG, or Reactome pathways. The edges (lines) represent the type of evidence reported inthe literature for de interaction of proteins. A con�dence value of 0.7 was used to establish theinteractions.

Page 23/26

Figure 4

Interactome analysis. A total of 73 DEGs with a padj<0.05 were analyzed in the STRING tool. Every node(circle) represents the protein encoded by every gene. The colors indicate the functional enrichments ofthe node in GO, KEGG, or Reactome pathways. The edges (lines) represent the type of evidence reported inthe literature for de interaction of proteins. A con�dence value of 0.7 was used to establish theinteractions.

Page 24/26

Figure 5

Overall Survival (OS) Analysis of DEGs. A total of 73 DEGs with a padj<0.05 were analyzed in thedatabase of (METABRIC, Nature 2012 & Nat Commun 2016) with a cut off time of 60 months and a z-score expression >|1|. From the query genes, only 7 genes had a log-rank<0.05. a) OS of TCP1 and TUBB;b) OS of C1QTNF3, CTF1, OLFML3, PLA2R1, and PODN. The blue color indicates the group of patients

Page 25/26

that have the gene downregulated and the red color indicates the group of patients with the geneupregulated.

Figure 5

Overall Survival (OS) Analysis of DEGs. A total of 73 DEGs with a padj<0.05 were analyzed in thedatabase of (METABRIC, Nature 2012 & Nat Commun 2016) with a cut off time of 60 months and a z-score expression >|1|. From the query genes, only 7 genes had a log-rank<0.05. a) OS of TCP1 and TUBB;

Page 26/26

b) OS of C1QTNF3, CTF1, OLFML3, PLA2R1, and PODN. The blue color indicates the group of patientsthat have the gene downregulated and the red color indicates the group of patients with the geneupregulated.

Supplementary Files

This is a list of supplementary �les associated with this preprint. Click to download.

Supplementarytable1.pdf

Supplementarytable1.pdf

Supplementarytable2.pdf

Supplementarytable2.pdf