Characterization of Abcc4 Gene Amplification in Stepwise-Selected Mouse J774 Macrophages Resistant to the Topoisomerase II Inhibitor Ciprofloxacin

Exposure of J774 mouse macrophages to stepwise increasing concentrations of ciprofloxacin, an antibiotic inhibiting bacterial topoisomerases, selects for resistant cells that overexpress the efflux transporter Abcc4 (Marquez et al. [2009] Antimicrob. Agents Chemother. 53: 2410–2416), encoded by the Abcc4 gene located on Chromosome 14qE4. In this study, we report the genomic alterations occurring along the selection process. Abcc4 expression progressively increased upon selection rounds, with exponential changes observed between cells exposed to 150 and 200 µM of ciprofloxacin, accompanied by a commensurate decrease in ciprofloxacin accumulation. Molecular cytogenetics experiments showed that this overexpression is linked to Abcc4 gene overrepresentation, grading from a partial trisomy of Chr 14 at the first step of selection (cells exposed to 100 µM ciprofloxacin), to low-level amplifications (around three copies) of Abcc4 locus on 1 or 2 Chr 14 (cells exposed to 150 µM ciprofloxacin), followed by high-level amplification of Abcc4 as homogeneous staining region (hsr), inserted on 3 different derivative Chromosomes (cells exposed to 200 µM ciprofloxacin). In revertant cells obtained after more than 60 passages of culture without drug, the Abcc4 hsr amplification was lost in approx. 70% of the population. These data suggest that exposing cells to sufficient concentrations of an antibiotic with low affinity for eukaryotic topoisomerases can cause major genomic alterations that may lead to the overexpression of the transporter responsible for its efflux. Gene amplification appears therefore as a mechanism of resistance that can be triggered by non-anticancer agents but contribute to cross-resistance, and is partially and slowly reversible.


Introduction
Overexpression of multidrug transporters (MDR) from the ATP-binding cassette (ABC) family is now widely recognized as a mechanism of resistance to cytotoxic drugs and is associated with therapeutic failures in patients receiving anticancer chemotherapy [1]. Several mechanisms have been described as leading to the overexpression of multidrug transporters, like induction of gene transcription (possibly caused by the drug itself ( [2] for review)), increase in mRNA stability [3], epigenetic changes [4,5], or gene amplification [6]. These mechanisms have been explored so far mainly in hepatocytes exposed to different xenobiotics, where activation of gene transcription by nuclear receptors has been well documented [7]. In cells exposed to anticancer agents, chromosomal alterations have also been reported after selection in vivo [8] or in vitro upon chronic exposure to drugs [9], but the underlying mechanisms can be much more diverse (see for a few examples [10][11][12]).
By its efflux properties, the multidrug transporter ABCC4 (MRP4) protects cells against toxicity induced by antimetabolites, such as methotrexate or analogues of purines and nucleosides, or by type I topoisomerase inhibitors, such as camptothecins [13][14][15]. ABCC4 overexpression has been reported in cancer cells, such as in prostate tumors [16] or human leukemic cells (with in vitro acquired resistance to 6-mercaptopurine [17]), and is associated with a poor clinical outcome in neuroblastoma [18]. Moreover, single nucleotide polymorphisms in ABCC4 gene have been shown to modulate the therapeutic response to methotrexate in children suffering from acute lymphoblastic leukemia [19].
Because of their broad substrate specificity, multidrug transporters can also reduce the cellular accumulation of other drugs and impair their activity if their pharmacological target is intracellular [2]. Conversely, these drugs can also trigger the overexpression of their transporters, as demonstrated for ABCC4 with the antiviral agent adefovir [20] and the fluoroquinolone antibiotic ciprofloxacin [21]. Fluoroquinolones are potent and widely used antibacterial agents that show a marked accumulation in eukaryotic cells, which explains their activity against a large array of intracellular bacteria (see [22] for review). Fluoroquinolones act by inhibiting the prokaryotic type II topoisomerase enzymes (DNA gyrase and topoisomerase IV). Although 100 to 1000-fold more active against bacterial enzymes than against their mammalian homologue topoisomerase II [23], fluoroquinolones can also cause genotoxic and clastogenic effects in eukaryotic cells at high concentrations [24], which has raised concerns about potential toxicities if used at supratherapeutic concentrations [25].
Applying to J774 macrophages a method widely used in vitro to select tumor cell lines resistant to anticancer drugs [26] and which consists in exposing cells to progressively increasing concentrations of the drug of interest, we were able to select, after about 50 passages in the presence of ciprofloxacin, cell lines in which the accumulation of this fluoroquinolone was markedly reduced [27]. This phenotype is associated with an accelerated efflux of ciprofloxacin that has been ascribed to an increased expression of Abcc4 (Mrp4) mRNA [21]. We also showed that Abcc4 protein overexpression was only slowly reversible, as more than 60 passages in the absence of ciprofloxacin were needed to obtain cells displaying a phenotype similar to that of the wild-type cell line (similar level of ciprofloxacin accumulation [27] despite a residual slight increase in Abcc4 protein content [21]). All together these data suggested that overexpression of Abcc4 could be driven through gene amplification.
The present study therefore focuses on the characterization of the progressive acquisition of multidrug resistance in J774 macrophages collected along the selection process with ciprofloxacin and examines possible genomic amplification of Abcc4 in these cells by fluorescence in situ hybridization (FISH) and multicolor FISH (mFISH). Our data show that Abcc4 gene amplification indeed occurs in the resistant cells and that it is slowly reversible. We also evidence other clonal chromosomal alterations developing along the selection process, which may reflect the genomic instability induced in eukaryotic cells when exposed to high concentrations of ciprofloxacin.

Characterization of ciprofloxacin accumulation and Abcc4 expression in cell lines resistant to different concentrations of ciprofloxacin
We showed previously that mouse J774 macrophages resistant to 200 mM ciprofloxacin were characterized by a markedly reduced accumulation of this drug [27] that was attributed to the overexpression of the ABC transporter Abcc4 [21].
To further understand the mechanisms leading to this overexpression, we have now compared the accumulation of ciprofloxacin and the expression of Abcc4 (protein and mRNA levels) in wild-type cells vs. cells resistant to ciprofloxacin concentrations of 100, 150 and 200 mM. The data illustrated in Figure 1 show that the accumulation of ciprofloxacin was decreased in parallel with the increased level of resistance. However, the process was not linearly related to the drug concentration used for selection, most of the effect being obtained only in cells resistant to the highest concentration. This reduction of ciprofloxacin accumulation was associated with a commensurate increase in the expression of Abcc4, both at the mRNA and protein levels. The lowest panel of Figure 1 shows the correlation between ciprofloxacin accumulation and protein levels of Abcc4.
Characterization of the Abcc4 locus amplification in ciprofloxacin-resistant and revertant cell lines compared to wild-type macrophages Conventional karyotype coupled to Multicolor FISH (mFISH) showed that the wild-type J774 cell line was characterized by a near-triploid karyotype ( Figure S1) and exhibited particularly a (ii) right axis: Abcc4 mRNA (grey bars) and protein (black bars) levels as a ratio to the value observed in wild-type cells (set to 1). C. Correlation between ciprofloxacin residual accumulation and Abcc4 relative expression in these cells. The curve corresponds to a best fit based on an inverse logarithmic function (y = 11.56e 20.3037 X +14.92). doi:10.1371/journal.pone.0028368.g001 derivative of Chr 14 characterized by the replacement of its telomeric end by a part of Chr 3 ( Figure 2, Chr B) in addition to the two apparently normal Chromosomes 14. The Abcc4 locus localized at Chr 14qE4 was analyzed by FISH using BAC (Bacterial Artificial Chromosome) probes, on metaphases in comparison to a more centromeric control locus (see Figure S2 for typical images and Figure 3 for a schematic representation of chromosomes with Abcc4 loci observed in the different cell lines and a quantification of the different clones).
Wild-type macrophages remained disomic for Abcc4 locus, as we observed 2 different copies of Abcc4 on two apparently normal Chromosomes 14 ( Figure 2, Chr A), in comparison to the control probe (normally located at Chr 14qA3), which exhibited a third signal on the derivative Chromosome 14 ( Figure 2, Chr B). In the intermediate resistant cell line CR100, we predominantly observed the wild-type pattern (almost 67% of the metaphases). A third copy of the Abcc4 locus was identified in about 1/3 of the cells, either on an additional apparently normal Chromosome 14 with both Abcc4 and control probes (27% of the metaphases, Figure 3), or on a marker Chromosome (7% of the metaphases). At the next step of selection with ciprofloxacin (CR150 cells), only 21% of the analyzed metaphases displayed the wild-type pattern while all the others showed different clones characterized by a low level chromosomal amplification of Abcc4 (approximately 3 copies) located either on one or on two Chromosomes 14, or associated to a marker Chromosome which did not exhibit the Chromosome 14 centromeric control probe.
In the fully ciprofloxacin-resistant cell line CR200, we observed Abcc4 high-level amplification as a homogenous staining region (hsr) in all metaphases ( Figure S2) on different marker Chromosomes that did not exhibit the Chromosome 14 centromeric control probe. Of note, all metaphases observed for the CR200 cell line displayed 2 apparently normal Chr 14 and no Abcc4 copy on the derivative Chr 14, as in wild-type cells. To identify the marker Chromosomes with Abcc4 hsr, we performed a mFISH analysis followed by Abcc4 hybridization (Figure 2 and Figure S2). Upon examination of 73 metaphases, the hsr was found with a localization that was either centromeric to a derivative Chr 5 (72.5% of the metaphases, clone I) ( Figure 2, Chr C) or telomeric to a derivative Chr 16 (10%, clone III) ( Figure 2, Chr D). Interestingly, in 13.5% and 4% of the observed metaphases, the Abcc4 amplification was found on either two (clone II) or one (clone IV) smaller marker Chromosome(s) (Figure 2, Chr M). These contained three heterochromatic regions aside of the centromere and were suspected to correspond to a derivative of Chr 13 by mFISH, although the chromosomal origin of the centromeric region remained uncertain ( Figure 2, Chr M). Other additional genomic alterations were acquired in the resistant cells, namely 2der(9)T(9;19),+13,2idic (13) in all clones, +9, in clones I and II, and +der(1)T(1;2)(H?;?) in clone III, as identified by mFISH.
In the revertant cell line Rev200 for which 31 metaphases were examined, we observed the wild-type phenotype (2 apparently normal Chr 14, and one without the Abcc4 locus) in around half of them (16)

Characterization of the DnajC3 locus amplification in resistant cell lines
As DnajC3 is located close to Abcc4 at Chromosome 14qE4, and because we already detected DnajC3 protein overexpression in the

Influence of ABCC4 on ciprofloxacin accumulation in murine and human cells
As all the work presented so far had been performed with a murine cell line, we examined whether ciprofloxacin is also a substrate of the human homologue transporter. To this effect, we compared the accumulation of ciprofloxacin (a) in wild-type J774 macrophages (basal expression of Abcc4) and in J774 macrophages resistant to 200 mM of ciprofloxacin and (b) in HEK293 human cells and HEK293/4.63 transduced with the cDNA coding for human ABCC4 and expressing it to high levels [28]. The accumulation of ciprofloxacin was drastically reduced in HEK293/4.63 cells compared to HEK293 cells, and was increased by gemfibrozil, a broad-spectrum inhibitor of MRP transporters (see Figure S3), demonstrating that ciprofloxacin is also a substrate of the human homologue of Abcc4.

Discussion
The present study shows that an antibiotic is capable of inducing at large concentrations a series of clonal chromosomal alterations in an eukaryotic cell, leading, among other changes, to the overexpression of a multidrug transporter responsible for its efflux. This effect is slowly and only partially reversible at the genomic level. The demonstration remains so far limited to ciprofloxacin and the murine multidrug transporter Abcc4 (Mrp4).
However, we show here that ciprofloxacin is also substrate for ABCC4, the human homologue of this transporter (as also found by others [29]), suggesting that similar mechanisms could possibly take place in humans. This opens interesting perspectives in terms of interactions of antibacterial agents with the host in relation to potential cell toxicity, and, in a broader context, in the manner eukaryotic cells deal with exogenous compounds.
ABCC4 gene amplification as an hsr has already been described in a human T lymphoid cell line made resistant to the antiviral drug adefovir [20], another well known ABCC4 substrate. In this case, however, amplification was located at the distal end of chromosome arm 13q, which is compatible to the normal location of human ABCC4 (13q32). Hsr and double minute chromosomes are the two common patterns of gene amplification observed upon selection by anticancer agents [6]. Interestingly, we do not observe double minute chromosomes at any step of the selection, in contrast with what has been shown for Abcb1 (P-glycoprotein) in colchicine-resistant J774.2 cells [30], or for ABCG2 in mitoxantrone-resistant glioblastoma cells [31], which display double minutes at low drug concentrations, but hsr at high concentrations. In other instances, both patterns are present at the same time, as described for ABCC1 in a doxorubicin-resistant human tumor cell line [32].
We showed that the karyotypes of the fully resistant clones mainly differ from the one of the wild-type cell line by 3 different marker Chromosomes with Abcc4 high level amplification. These observations suggest a random chromosomal insertion of the 14qE4 genomic amplification as an hsr, which provides a proliferative advantage by enabling cells to resist to the strong pressure of selection and leading to clonal expansions. Of interest, gene amplification was not restricted to Abcc4, since DnajC3, another gene located at 14qE4, was co-amplified, and the corresponding protein overexpressed [Caceres et al., in preparation]. Since both Abcc4 and DnajC3 genes are constitutively expressed in the wild-type macrophages, they do not need another genetic event in addition to genomic amplification for overexpression, such as a ''switch on'' induced by a promoter capture or by insertion of retroviral sequences [33,34]. We did not observe extrachromosomal elements, such as double minutes (dmin) or episomes that are classically reported [6]. However, we cannot exclude their occurrence at the first steps of selection before integration of the amplicons within chromosomes detected in CR200 cells. In cancer cells, genomic amplification has been shown to occur at common chromosomal fragile sites, or to result from defects in DNA replication or telomere dysfunction. It is noteworthy to mention that the chromosomal band 14qE4, with Abcc4 and DnajC3 loci, has been shown to be a common fragile site that may favor genomic instability [35].
The phenotypic reversion is associated with a drastic regression of the Abcc4 amplification, with loss of the hsr-like amplification of Abcc4 to Chr 5 and to Chr 16. The persistence of a copy of a marker Chromosome (Chr M, with Abcc4 hsr) in nearly 30% of the revertant cells, and the time required to obtain phenotypic reversion, underline the stability of the gene amplification. As the revertant cells display a similar phenotype as wild-type cells regarding ciprofloxacin accumulation [27] and almost similar levels of Abcc4 mRNA [21], it is likely that the Abcc4 amplification observed on this marker in single copy is not associated with efficient transcription.
In vitro, gene amplification is likely to be initiated by a DNA double-strand break in cells that lack appropriate cell-cycle checkpoints [6]. Chemotherapeutic drugs targeting topoisomerase II promote DNA double-strand break, favoring thereby the development of therapy-related leukaemias. In bacteria, fluoro- quinolones form a reversible ternary complex linking together DNA and prokaryotic type II topoisomerase enzymes to impair the progression of the replication fork. This leads to several lethal damages, including SOS response induction and, possibly chromosome fragmentation, which explains their rapid bactericidal activity [36]. While no genotoxicity has been reported so far for ciprofloxacin in vivo or in cultured cells [37][38][39], chromosomal aberrations have been described in cultured human lymphocytes exposed to supratherapeutic concentrations similar to those used in this study [40,41], as well as an increase of sister-chromatid exchange [42] and in DNA single strand breaks frequencies [43] in mouse bone marrow cells. Although the concentrations needed to observe these alterations (100-200 mM, i.e. 40-80 mg/L) are well above the serum levels observed in patients receiving conventional therapies (1-4 mg/L), one needs to take into consideration that ciprofloxacin accumulates in tissues (with tissular levels reaching values 2-7 fold higher than serum levels) as well as in body fluids, with the highest concentrations (200-900 mg/L) being found in urine. This suggests that in vivo exposure may be more important than anticipated based on serum levels only.
Although pending for further investigations aimed at elucidating the mechanism leading to gene amplification, our data thus indicate that ciprofloxacin may induce in eukaryotic cells chromosomal aberrations leading to overexpression of the transporter responsible for its efflux. Whether such alterations may occur in vivo will clearly depend on the concentration of the drug, and probably also on a combination of its recognition by efflux transporters and its capacity to interact with DNAtopoisomerase complexes. Other fluoroquinolones may, indeed, be much more toxic than ciprofloxacin in this context [41,44], with many of them having been withdrawn or not accepted for registration in many countries for unsatisfactory benefit to risk ratio involving, among other untoward reactions, clastogenic effects (see [45] for an example with gemifloxacin). It is interesting to note that N-substituted piperazinyl quinolones derived from ciprofloxacin or other clinically-used fluoroquinolones show in vitro cytotoxicities that are as high as those seen with etoposide, a well known inhibitor of topoisomerase II [46,47]. These compounds are now evaluated as potential anticancer agents.

Chemicals
All cell culture reagents were purchased from Invitrogen (Carlsbad, CA). Ciprofloxacin HCl (potency 85%) was received from Bayer HealthCare (Wuppertal, Germany) as microbiological standard. Other chemical products were purchased from Sigma-Aldrich (St. Louis, MO).

Cell lines and culture conditions
J774 mouse macrophage-like cells [48] (referred to as wild-type cells) were cultured and maintained as already described [49]. Ciprofloxacin-resistant macrophages and their revertant were fully described in a previous publication of our group [27]. Resistant cells were obtained by chronic exposure to progressively increasing concentrations of ciprofloxacin (100 mM, 150 mM, and 200), yielding to cell lines referred to here as CR100 (used at the 5th passage), CR150 (used at the 3rd passage) and CR200 (used at the 76th passage) respectively. Revertant cells (referred to here as Rev200) were obtained by returning CR200 cells to ciprofloxacinfree and used here at their 84th passage in the absence of selective pressure. Human Embryonic Kidney HEK293 cells and HEK293/4.63 cells transduced with human ABCC4 cDNA were obtained from P. Borst (Het Netherlands Kanker Instituut, Amsterdam, The Netherlands). They were cultivated in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum [28].

Accumulation and efflux of fluoroquinolones
These experiments were performed exactly as previously described [21,27,50]. Cell-associated ciprofloxacin was assayed by fluorimetry (l exc = 275 nm and l em = 450 nm) and its cellular concentration expressed by reference to the total cell protein content as measured by the Lowry's method [51].

Quantification of Abcc4 by real-time PCR and Western blot Analysis
Abcc4 expression was assessed at mRNA and proteins levels, as described previously [21]. Real-time PCR experiments were performed starting from 1 mg of total purified RNA transcribed into cDNA, and using SYBR Green detection. Two housekeeping genes, Ywhaz and Rpl13a (mouse geNorm normalization kit, PrimerDesign Ltd., Southampton, UK) were used for normalization. The relative quantification of Abcc4 gene in cell lines of interest was done using levels measured for wild-type J774 macrophages as baseline, based on Pfaffl's equation [52]. Western-blots were performed on cell crude extracts. After electrophoresis on acrylamide gel and transfer to nitrocellulose membrane, proteins of interest were detected using anti-ABCC4 monoclonal antibody (M 4 I-10; Alexis Biochemicals, Lausen, Switzerland) or anti-actin polyclonal antibodies (Sigma-Aldrich) (dilution 1/1000), followed by appropriate horseradish peroxidase-coupled secondary antibodies (dilutions 1/ 600). Blots were then revealed by chemiluminescence assay (SuperSignal West Pico, Pierce, Thermo Fisher Scientific Inc., Rockford, IL).

Cytogenetic analysis
Metaphase chromosomes were obtained according to standard methods for mouse cell lines [53]. Cells were grown for 48 h into 6well plates (initial density of 5610 4 cells per cm 2 ) and fed with fresh medium 4 h before addition of a mitotic spindle inhibitor (KaryoMAXH Colcemid TM Solution, Invitrogen) at a final concentration of 0.02 mg/ml. After 45 min of incubation at 37uC, cells were washed with PBS, detached by trypsinization and collected by centrifugation (900 rpm, 10 min). They were then submitted to hypotonic choc by resuspension in KCl 75 mM and incubation for 10 min at room temperature. After centrifugation (900 rpm, 5 min), cell pellets were resuspended in Carnoy's fixative (methanol/glacial acetic acid, 3/1) and incubated 15 min at room temperature. This fixation step was repeated three times. Metaphase chromosomes were obtained after spreading and air-drying of fixed cells onto microscope glass slides. Metaphase cells were banded with trypsin denaturation followed by a Wright staining. Metaphases were analysed with Ikaros Imaging System (MetaSystems, Altlussheim, Germany). Karyotypes were described according to The Rules for Nomenclature of Chromosome Aberrations (Revised: January 2005) from the International Committee on Standardized Genetic Nomenclature for Mice (http://www.informatics.jax.org/ mgihome/nomen/anomalies.shtml). A total of 8 and 21 metaphasic cells were analyzed, for wild-type and resistant macrophage (CR200) cell lines, respectively.

Multicolour FISH (mFISH)
Fresh metaphase spreads were pretreated with RNase A (100 mg/ml)/2xSSC for 45 min at 37uC, then denatured as previously described. The multicolour probe kit (21XMouse, MetaSystems) was denatured according to manufacturer's instructions, and then hybridized to metaphases at 37uC for 3 to 4 days. Post-hybridization washes and counterstaining were performed according to manufacturer's instructions, using DAPI/antifade (MetaSystems). Ten metaphases were analyzed for wild-type macrophage cells, and more than 20 for resistant cells CR200. When mFISH was combined with Abcc4 probe hybridization, we compared metaphases hybridized first with mFISH probes, then washed and hybridized again with the Abcc4 probe. Figure S1 mFISH karyotype of wild-type J774 macrophages. Chromosomes are displayed with false colors, as indicated by rounds; squares indicate the combination of true colors given by the probes. (TIF) Figure S2 Abcc4 FISH and mFISH analysis in wild-type J774 macrophages, in cells resistant to increasing concentrations of ciprofloxacin, and in revertant cells. A: Metaphase spreads of J774 wild-type (upper left panel), CR100 (upper right panel), CR150 (middle panel) and Rev200 (lower panel) cells were subjected to FISH analysis with an Abcc4 BAC probe (green) and a control BAC probe located on Chr 14 (red). Chromosomes were counterstained with DAPI. Representative metaphases of the different clones observed are shown. Green arrow indicates Chr 14 with Abcc4 copy, red arrow points to the control BAC probe (red) located on Chr 14, green circle indicates Abcc4 amplification, and green square Abcc4 additional copy. B: Metaphase spreads of the main three clones (I, II, III) observed in CR200 cells hybridized first with mFISH probes and subsequently with the Abcc4 BAC probe (green) and the control BAC probe (red) located on Chr 14 (chromosomes counterstained with DAPI). (TIF) Figure S3 Comparison of ciprofloxacin accumulation in murine and human cells with basal or overexpression of Abcc4/ABCC4. Cellular accumulation of ciprofloxacin in J774 mouse macrophages (WT or CR200) and in human embryonic kidney cells (HEK293, parental cells; HEK293/4.63 transduced with the human cDNA coding for ABCC4 and overexpessing the transporter to high levels [28,29]). Cells were incubated during 2 h with an extracellular concentration of 20 mg/L (50 mM) of ciprofloxacin in the absence of in the presence of 500 mM gemfibrozil. Data are expressed in percentage of the value measured in control condition in the parental cell line and are the mean 6 SD of 3 independent determinations. (TIF)