microRNA global expression analysis and genomic profiling of the camptothecin-resistant TALL derived cell line CPT-K 5

The clinical use of the camptothecin (CPT) derivatives, topotecan and irinotecan, has had a significant impact on cancer therapy. However, acquired clinical resistance to these drugs is common, which greatly hampers their clinical efficacy. MicroRNAs (miRNAs) is an exciting novel class of endogenous non-coding RNAs that negatively regulate gene expression of up to 50% of the protein-coding genes at the post-translational level. Abnormal expression of miRNAs is associated with pathogenesis of cancer and is also implicated in anticancer drug resistance phenotypes. We used global expression analysis to examine for differential miRNA expression between the camptothecin-resistant cell line CPT-K5 and its parental CPT-sensitive RPMI-8402. In the CPT-K5 cell line 18 miRNAs were deregulated. Fifteen of these were down-regulated and three were up-regulated. The miRNA-193a-3p, miR-130a-3p, and miR-29c-3p were the most down-regulated miRNAs at 205.9-fold, 33.9-fold and 5.5-fold, respectively, while the miRNA let-7i-5p was the most up-regulated at 3.9-fold. We used subtraction BAC-based array CGH analysis to examine for genomic copy number changes. Only for the three most down-regulated miRNAs a positive correlation was found with genomic loss of their chromosomal regions in which they are encoded. Potential functional targets of the differentially expressed miRNAs were examined by searching the miRBase and miRTarBase databases. Recurrent KEGG pathways that theoretically could be affected by the deregulated miRNAs are lysine degradation, cell cycle, PI3K-Akt-, ERbBand p53signaling pathways. We show that the intracellular levels of several miRNAs are significantly deregulated upon acquisition of CPT resistance in the T-ALL derived cell line CPT-K5, and that genomic copy number changes is not a major cause of deregulation. In addition, the most deregulated miRNAs in our study have previously been described to be involved in various types of chemotherapeutic resistance, including the chemotherapeutics CPT, gefitinib and cisplatin in other cancer and cell types. Our study adds to the current knowledge of the mechanisms of acquired CPT resistance. Specific miRNAs may prove to be future targets to reverse or inhibit development of CPT resistance thereby providing means for a more effective treatment.


Introduction
Camptothecin (CPT) specifically inhibits the nuclear enzyme DNA topoisomerase I (Top1) [1,2] .Its effect is exerted by binding to the covalent complex formed by DNA and Top1 leading to persistent DNA breaks that are turned into devastating double stranded breaks (DSBs) upon collision with the replication or transcription machinery [3] .

BRIEF REPORT
When these DSBs are unrepaired it confers apoptosis of the malignant cells and eventually eradication of the tumor [3] .
The water-soluble CPT derivatives such as topotecan and irinotecan belong to a class of chemotherapeutic drugs that have been approved for treatment of various malignancies, including colorectal, ovarian, small cell lung cancers, leukemias and lymphomas [4] .However, their clinical efficacy is greatly hampered by development of resistance against CPT and its derivatives [5] .Reversing or inhibiting CPT resistance may provide means for a more effective treatment.Preclinical studies have shown that "classical" cellular alterations, such as drug-efflux, metabolism, Top1-down regulation, TOP1 mutation and DNA damage response contribute to the CPT-resistance.However, recent lines of evidence have suggested new resistance mechanisms involving microRNA (miRNA) deregulation [6] .
More than 2,500 human miRNAs have been identified.They belong to a class of non-coding RNA that is 20 to 25 nucleotides in length.The precise mechanisms of miRNA function have not been fully clarified yet.It is known, however, that each miRNA can regulate the expression of up to hundreds of target genes simultaneously, while a single gene can also be targeted by multiple miRNAs.The miRNAs control 30-50% of the protein-coding genes [7] and are therefore involved in a multitude of signaling pathways controlling normal cell differentiation, division, and apoptosis [8,9] .There is strong evidence that dysregulated miRNAs can cause cancer and its progression [10] .However, it remains to be elucidated how they affect the response of cancer cells to chemotherapeutic treatment, in particular their involvement in acquisition of CPT resistance in leukemic cells.In one study it was shown that mRNA and miRNA expression profiles correlated with sensitivity to FdUMP [10], fluorouracil, floxuridine, topotecan, and irinotecan across a NCI-60 cell line screen [11] .In another study it was shown that 25 miRNAs were deregulated in intrinsic CPT resistance in gastric cancer derived cell lines [12] .
To explore miRNA's role in acquired camptothecin resistance we took advantage of the camptothecin resistant cell line CPT-K5 and its parental RPMI-8402 and compared their global miRNA expression.The CPT-K5 cell line was previously developed by stepwise increasing exposure of the human T-ALL derived cell line RPMI-8402 to the CPT-derivative irinotecan [13] .Later we showed that the CPT-resistance correlated with a mutation at amino acid residue 533 (p.D533G, Asp->Gly) in the DNA binding domain of the Top1 enzyme [14] .The mutant enzyme acquired altered biochemical properties, especially as to how it interacts with DNA.The changed properties resulted in a higher efficiency for recognition of specific sequences and a higher stability of cleavable complexes contributing to the cellular CPT-resistance [15] .
In the present study we show that 18 miRNAs are differently expressed between CPT-K5 and RPMI-8402 and that two miRNA's, miR-130a-3p and miRNA-193a-3p, were the most deregulated by a magnitude of 33.9-and 205.9-fold, respectively.In addition, we explored whether the deregulated miRNA expression correlated with changes in genomic copy numbers between the cell lines.

Cell cultures and purification of nucleic acids
The cell lines CPT-K5 and RPMI-8402 were cultured as described [15] .Five million cells in exponentially growth phase were used for RNA purifications with the miRNeasy Mini Kit (Qiagen Nordic, Solna, Sweden) according to manufacturer's protocol.DNA from three million cells was purified using the Gentra Puregene Blood Kit (Qiagen).
Two dual-color microarrays (version 208001V8.1)were used in a dye-swap setup with each sample (RNA from CPT-K5 and RPMI-8402) labelled with opposite fluorophores in the two hybridizations.Three micrograms of RNA was used for each labeling reaction and the matching labeling Kit from Exiqon was used according to protocol.Upon labeling the RNA was hybridized to the microarrays in an automated Tecan HS400 Pro hybridization station according to Exiqon's recommendations.The arrays were scanned at 532 and 635 nm in a GenePix 4000B scanner using the GenePix Pro 6.1 software (Molecular Devices, Sunnyvale, California, USA).The raw image files were analyzed in GenePix Pro 6.1 using a Gal-file annotated in compliance with miRBase release 11 [16] .Individual features were identified as irregular and the background subtraction was calculated using the "Morphologic opening" setting.Features with intensities below threshold (negative controls + 5SD) were excluded, and only features representing known human miRNAs where at least three of the four replicated spots had passed the above mentioned criteria were considered present and used in the subsequent analysis.The data from the two hybridizations were normalized using the global lowess algorithm in Acuity 4.0 (Molecular Devices).The normalized data from the dye-swap procedure were averaged to provide the final data.All miRNA names are reported according to the nomenclature used in miRBase 20.Exempted from this rule is data referenced from earlier publications, where miRNA names are reported according to the nomenclature used in the original publications.

qPCR analysis of miRNA expression
Specific miRNA expression were investigated by qPCR using TaqMan R MicroRNA Assays (Life Technologies Europe, Naerum, Denmark) according to protocol.The Mx3000P RQ-PCR System (Agilent Technologies, Santa Clara, CA, USA) was used for the PCR reactions and output data were analyzed by the ΔC t relative quantification model using RNU6B and RNU48 (Life Technologies Europe) as reference genes for normalization.

Array-based comparative genomic hybridization analysis
Array-based comparative genomic hybridization (aCGH) analysis was done using the CytoChip BAC-array platform (BlueGnome, Cambridge, UK) as previously described [17] .CPT-K5 DNA was labeled with Cy3 and RPMI-8402 DNA with Cy5.The GenePix 4000B laser scanner (Molecular Devices) together with GenePix Pro 6.1 software (Molecular Devices) was used to scan he microarray.The CytoChip algorithm analysis tool in the BlueFuse 3.5 software (BlueGnome) defined regions of gain or loss.Reference genome was NCBI build 36.1 (hg18).Bioinformatics analysis was performed by querying the UCSC database (http://genome.ucsc.edu).

Comparison of miRNA profiles between CPT-K5 and RPMI-8402
We examined whether differential miRNA expression between the camptothecin-resistant CPT-K5 and its parental camptothecin-sensitive RPMI-8402 cell line could be detected using a miRNA microarray platform from Exiqon.This microarray contains 5924 probes representing 429 miRNAs, and 94 miRNAs were expressed above threshold (Supplemental Table 1).Eight-teen miRNAs emerged as differently expressed by more than a 2-fold change in the CPT-K5 cell line comparied with RPMI-8402 (Table 1).Specifically, fifteen miRNAs were down-regulated and three were up-regulated in CPT-K5.The miRNA-193a-3p and miR-130a-3p, were the most down-regulated miRNAs at 205, 9-fold and 33.9-fold, respectively, while the miRNA let-7i-5p was the most up-regulated at 3.9-fold.
To validate the microarray data, four differently expressed genes were randomly chosen for validation by Taqmann MicroRNA assays.The selected miRNAs were up-regulated (miR-7-5p), down-regulated (miR-18a-5p and miR-130a-3p) or showed no change in expression (miR-223-3p) (Figure 1A).The expression levels of the miRNAs from real time PCR were consistent with results from miRNA array analysis.A comparison of expression levels between CPT-K5 and RPMI-8402 was done by real time PCR confirming the observed differences in expression of the miRNAs, miR-7-5p, miR-18a-5p and miR-130-3p, and no change of miR-223-3p as also determined by the microarray analysis (Figure 1B).

Comparison of differently expressed miRNAs and copy number changes of CPT-K5 and RPMI-8402
To evaluate a possible correlation between differently expressed miRNAs and genomic copy number changes we performed subtractive whole genome profiling to analyze for copy number changes between CPT-K5 and RPMI-8402 using a BAC-based aCGH microarray.Sixty-one genomic regions emerged as regions of copy number changes between CPT-K5 and RPMI-8402 (Figure 2A, and Supplemental Table 2).Specifically, 32 regions were gained while 29 regions showed copy number losses.We examined the copy number changes relative to the genomic positions of the differently expressed miRNAs (Table 1, Figure 2B).In the group of up-regulated miRNAs we observed for all miRNAs no copy number changes between CPT-K5 and RPMI-8402.In the group of down-regulated miRNAs nine (9/18, 50%) were located in regions of copy number losses, three (3/18, 17%) were located in regions of copy number gains and six (6/18, 33%) were located in regions of no copy number changes.
For the three most highly down-regulated miRNAs (miR-193a-3p, hmiR130a-3p, and miR-29c-3p) there were a correlation between their location in genomic regions with copy number losses while for the rest of the miRNAs there were no correlation with their differences in expression levels and copy number changes.

miRNA targets of the changed miRNAs
To identify potential functional miR-targets of the deregulated miRNAs the DIANA-miRPath (http://diana.imis.athena-innovation.gr)were searched (Table 2).Recurrent KEGG pathways that theoretically might be affected by the deregulated miRNAs are lysine degradation, cell cycle, PI3K-Akt-, ERbB-and p53-signaling pathways.

Discussion
In the present study we showed that 18 miRNAs are deregulated during acquisition of resistance to the camptothecin derivative irinotecan in the CPT-K5 cell line.By comparing the miRNA differences with genomic copy number changes between the two cell lines the expression of the most down--fold decrease) correlated with identified genomic losses.The most deregulated miRNAs were miR-193a-3p, miR-130a-3p, and miR-29c-3p exhibiting a 205.9-, 33.9-, and 5.5-fold decrease in expression, respectively.For the remaining 15 deregulated miRNAs no correlation between their relative expression level and genomic copy number status was found.These findings indicate that high expression changes are more likely associated with copy number changes compared to moderate or minor expression changes.This notion is supported by the findings that massive up-regulation of MYC expression is associated with high-level amplification of the MYC oncogene [18] , and that massive down-regulation is associated with gene deletions [19] .When moderate or minor expression changes are observed other factors than copy number changes may influence their expression level [10,20] .The miRNA expression is mainly regulated in a tissue-specific and disease state-specific fashion by different transcription factors while some miRNAs are regulated by  tumor suppressor or oncogene pathways such as TP53, MYC and RAS [21] .Deregulated miRNA expression can also result from changes in epigenetic regulation, such as methylation status of miRNA genes [22] or may result from mutations in miRNA genes [23] .
Altered intracellular levels of miRNAs interfere with the chemoresponses in a variety of cancer cells [11,12,[24][25][26] .In a study utilizing the NCI-60 cell line panel it was shown that mRNA and miRNA expression profiles correlated with sensitivity to FdUMP [10], fluorouracil, floxuridine, topotecan, and irinotecan [11] .In a study of the CPT-resistant colon cancer cell line SW1116/HCPT, seventy-seven miRNAs were differentially expressed compared to its parental SW1116 (30 miRNAs were down-regulated and 47 were up-regulated) [25] .Among these miRNAs the miR-548d was the highest up-regulated (124.55-fold) while miR-641 was the most down-regulated (60.20-fold).The authors also performed differential gene expression showing that -regulated (404.1-fold), and that it is one of the target genes of miR-506, which was overshown to regulate some MDR proteins, which uses ATP to extrude chemotherapeutic agents from the cells [27] .Intrinsic drug resistance to hydroxycamptothecin was studied in six gastric cancer cell lines and it was shown that 25 miRNAs were deregulated in the resistant cells, including up-regulated miR-196a, miR-338, miR-126, miR-98, let-7g, and down-regulated miR-200 family, miR-31, and miR-7 [12] .In our study we found that let-7i-5p and miR-7-5p were the two most up-regulated and that miR-193a-3p, miR-130a-3p, and miR-29c-3p were the three most down-regulated miRNAs.Comparing our differential expression miRNA results with the previous studies on CPT resistance in various cell lines and cell types it is only the miR-7-5p that is a recurrent miRNA involved in CPT resistance.We found that miR-7-5p was up-regulated while Wu et al. found it to be down-regulated [12] .
The disparate results between the different studies may be explained by variations in the molecular pathways in the different cancer cells.It should be noted that miRNAs execute their biological function via repression of many different protein-coding genes involved in a multitude of signaling pathways.Another reason could be the effect of local tumor microenvironment, which is well known to be modulated via a variety of signaling networks [28] .
The most deregulated miRNA in our study was the 205.9-fold down-regulation of miR-193a-3p in the acquired CPT resistance of the CPT-K5 cell line.Its aberrant expression has been reported in all the of cancer examined, including colorectal cancer [29] , non-small cell lung cancer (NSCLC) [30] , myeloid leukemia [31] and Wilms' tumor blastema [32] .The transcription factors XB130 [33] , and p63 [34] have been implicated in the regulation of miR-193a expression as well as the DNA methylation state of its promotor region [35] .A tumor-suppressor role of miR-193a-3p has been reported in NSCLC [30] and epithelial ovarian cancer cells [36] .Conversely, miR-193a-3p can also  1.
promote both in vivo growth and chemo-resistance of hepatocellular carcinoma [35] and bladder cancer cells [37] .The bladder cancer cell line 5637 is chemosensitive when the promotor of miR-193a is hypermethylated while the chemoresistant bladder cancer cell line is drug resistant because of a hypomethylated promotor [38] .Four direct target genes of miR-193a-3p were identified (PLAU, HIC2, SRSF2 and LOXL4) conveying bladder cancer multi-chemoresistance against various chemotherapeutic drugs, including etoposide, carboplatin, cisplatin, 5-fluorocil, and doxorubicin.Resistance against CPT has not prior to our study been associated with deregulated miR-193a-3p expression.
The second-most down-regulated miRNA in our study was miRNA-130a-3p exhibiting a 33.9-fold down-regulation.miR-130a plays a crucial role in tumor biology with different functions in various cancers acting as both an oncogene (NSCLC, cervical cancer and colorectal cancer) and as a tumor suppressor gene (glioblastoma, prostate cancer and leukemia) [39] .The reasons for the apparently contradictory roles of miR-130a are not clear.Recently, it was shown that miR-130a under-expression leads to gefitinib resistance in NSCLC whereas overexpression increases sensitivity to gefitinib [40] .The MET gene was shown to be a direct target of miR-130a and that MET amplification leads to gefitinib resistance by activating the ERBB3 signaling pathway.In ovarian cancer it was shown that under-expression conferred cisplatin-resistance by targeting X-linked inhibitor of apoptosis [41] .In hepatocellular carcinoma (HCC) miR-130a increases drug resistance by regulating RUNX3 and wnt signaling in cisplatin treated HCC [39] .Taken together these findings indicate that a common drug resistance mechanism may be ascribed to miR-130a.
We observed that the miR-29c-3p was down-regulated by 5.5-fold.miR-29c has been identified as a tumor suppressor in several human cancers [42] .Low expression of miR-29c was recently shown to be positively associated with therapeutic resistance in 159 nasopharyngeal carcinoma cases against ionizing radiation and cisplatin [43] .The authors also showed that expression of the anti-apoptotic factors, MCL-1 and BCL-2 in NPC tissues and cell lines were repressed by miR-29c.

Conclusions
Our present study demonstrates that the intracellular levels of certain miRNAs are significantly deregulated upon acquisition of camptothecin resistance.We show a positive correlation of genomic losses and miRNAs down-regulated by ≥5.5-fold.Interestingly, all major deregulated miRNAs described in our study have previously been described in other cancer and cell types to be involved in various types of chemotherapeutic resistance, including CPT, gefitinib and cisplatin.Our study adds to a better understanding onto which miRNAs that might have a role in resistance to CPT and its derivatives.Resistance to CPT limits its clinical efficacy and by knowing which miRNAs that are involved in the resistance mechanism future experiments with direct miRNA targets may be designed to examine whether it'll be possible to reversing such resistance or to avoid its development.

Supplemental
Table1.Detailed summary of miRNA expression differences.

Figure 1 .
Figure 1.Analysis of miRNA expression.Panel A. Comparison of microarray-based and qPCR-based measurements of selected miRNAs.Log2 values for expression of miR-130a-3p, miR-18a-5p, miR-223-3p, and miR-7-5p in CPT-K5 relative to RPMI-8402 are shown.Dark and light grey bars represent data from microarray and qPCR measurements, respectively.Panel B. Fold difference in expression of indicated miRNAs measured by qPCR.Means from triplicate experiments are indicated.Dark and light grey bars represent data from RPMI-8402 and CPT-K5, respectively.

Figure 2 .
Figure 2. Analysis of genomic copy number changes.Panel A. CPT-K5 genome chart view of BAC-based aCGH profile where RPMI-8402 is used as reference genome.Chromosomal position is specified on the x-axis and log 2 ratio on the y-axis.Panel B. Chromosome view of called regions.Blue bars next to the chromosomal ideograms indicate regions of gain and red bars indicate regions of losses.Chromosomal positions of miRNAs that are differently expressed in CPT-K5 relative to RPMI-8402 are indicated by open (up-regulated) or closed (down-regulated) circles where the numbers in the circles refer to the miRNAs listed in Table1.

Table 1 . miRNA expression differences and copy number changes in CPT-K5 relative to RPMI-8402 microRNA name a Pre-microRNA Cytoband Genomic position (bp) Mean (log2) Fold change Copy number change Up-regulated
a microRNAs are named according to miRBase 20.b microRNAs with more than one genomic location although it is only possible to measure the overall expression of the respective microRNAs.NC refers to no copy number change.bp refers base pair.