ORIGINAL ARTICLE See related commentary on pg 2317

Side Population Cells from Human Tumors Reveal Diverse Mechanisms for Chemoresistance Yuchun Luo1, Lixia Z. Ellis1, Katiuscia Dallaglio1,2, Moe Takeda1, William A. Robinson3, Steven E. Robinson3, Weimin Liu1, Karl D. Lewis3, Martin D. McCarter4, Rene Gonzalez3, David A. Norris1,5, Dennis R. Roop1,2, Richard A. Spritz6, Natalie G. Ahn7 and Mayumi Fujita1,2,5

Side population (SP) cells are identified as cells capable of excluding the fluorescent Hoechst dye and anticancer drugs, and it represents hematopoietic stem cells and chemoresistant cells from several solid tumors. In this study, we confirmed the presence of SP cells in tumors from melanoma patients. Melanoma SP cells overexpressed ATP-binding-cassette (ABC) transporters, ABCB1 and ABCB5. We generated a direct in vivo xenograft model, and demonstrated that SP cells were resistant to paclitaxel, a substrate of ABCB1, both in vitro and in vivo. However, melanoma SP cells were also resistant to temozolomide, which is not a substrate for ABC transporters, through IL-8 upregulation. In addition, profiling studies identified three signaling pathways (NF-kB, a6-b4-integrin, and IL-1) as differentially upregulated in melanoma SP cells, and there was a significant increase of PCDHB11 and decrease of FUK and TBX2 in these cells. Therefore, we provide evidence that SP is an enriched source of chemoresistant cells in human , and suggest that the selected and signaling pathways of SP cells may be a potential target for effective melanoma therapies. To our knowledge, this is a previously unreported study to isolate SP cells from melanoma patients and to investigate the gene expression profiling of these cells. Journal of Investigative Dermatology (2012) 132, 2440–2450; doi:10.1038/jid.2012.161; published online 24 May 2012

INTRODUCTION the fluorescent Hoechst 33342 dye has been employed to Malignant melanoma is the fifth most common cancer in the isolate a side population (SP) from hematopoietic cells (Goodell United States (Siegel et al., 2011) and the deadliest form of et al., 1996). This phenotype of low fluorescent staining pattern skin cancer. Despite various types of treatments, advanced results from a family of transporter known as ATP- stages of melanoma cells have proven to be intrinsically binding cassette (ABC) transporters, which are responsible for resistant to many therapies. Multiple mechanisms contribute the exclusion of Hoechst 33342 (Helmbach et al., 2001; to the development of drug resistance in melanoma, Gottesman et al., 2002). These SP cells have stem cell–like including induction of enzyme systems, reduced apoptosis, properties and are capable of self-renewal and differentiation increased DNA repair, and intra- and extra-cellular drug (Dean et al., 2005). These characteristics of SP cells have also transport (Helmbach et al., 2001; La Porta, 2007). been found in various types of tumors (Hirschmann-Jax et al., One of the well-known mechanisms of chemoresistance 2004; Grichnik et al., 2006; Dean, 2009; Dou et al., 2009; in cancers is the enhanced efflux of a broad spectrum of Zhong et al., 2010), and SP cells with a high drug efflux hydrophobic cytotoxic drugs. Recently, the ability to exclude capacity are identified as the potential cause of the chemo- resistant phenotypes of cancer stem cells in human tumors (Hirschmann-Jax et al., 2004; Dean et al., 2005). 1Department of Dermatology, University of Colorado Denver, Aurora, Colorado, USA; 2Charles C Gates Regenerative Medicine and Stem Cell SP cells have previously been isolated from human Biology, University of Colorado Denver, Aurora, Colorado, USA; 3Division of melanoma cell lines (Grichnik et al., 2006), but are yet to Medical Oncology, Department of Medicine, University of Colorado Denver, be isolated directly from patient melanoma tumors. One 4 Aurora, Colorado, USA; Department of Surgery, University of Colorado major challenge of analyzing patient samples is that Denver, Aurora, Colorado, USA; 5Denver Veterans Affairs Medical Center, Denver, Colorado, USA; 6Human Medical Genetics Program, University of substantial quantities of fresh tumor tissues are not available Colorado Denver, Aurora, Colorado, USA and 7Department of Chemistry and for isolation and investigation of SP cells that comprise only Biochemistry, University of Colorado at Boulder, Boulder, Colorado, USA a small percentage of the tumor cells. Another difficulty is Correspondence: Mayumi Fujita, Department of Dermatology, University of that experiments to further characterize biology and function Colorado, Mail Stop 8127, 12801 E. 17th Avenue, Room 4124, Aurora, of these cells are limited by the technical difficulties in Colorado 80045, USA. E-mail: [email protected] preserving and expanding tumor cells that faithfully repre- Abbreviations: ABC, ATP-binding-cassette; qRT-PCR, real-time quantitative reverse transcription–PCR; siRNA, small interfering RNA; SP, side population sent patient tumor biology. To that end, a direct in vivo Received 29 October 2011; revised 9 March 2012; accepted 22 March 2012; xenograft model provides cohorts of tumor-bearing mice published online 24 May 2012 suitable for further biological analyses and drug treatment

2440 Journal of Investigative Dermatology (2012), Volume 132 & 2012 The Society for Investigative Dermatology Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

(Rubio-Viqueira et al., 2006). Maintaining tumor xenografts a With verapamil Without verapamil 128 128 exclusively in vivo retains fundamental genotypic features of Non-SP Non-SP the primary tumor of direct relevance to preclinical drug 96 96 testing (Rubio-Viqueira et al., 2006; Daniel et al., 2009). 64 64 Therefore, to isolate and characterize the molecular features 32 SP 32 SP of a small fraction of SP cells in patient melanoma samples, 0.01% 0.34% we generated a direct in vivo xenograft model from human 0 0 0 32 64 96 128 0 32 64 96 128 melanoma tumors. Furthermore, our gene profiling analysis Hoechst blue discovered a series of genes and pathways potentially Hoechst red involved in chemoresistance in melanoma SP cells. b Frequencies Frequencies Name Type Origin of SP (F0) of SP (F1) RESULTS MF347 Primary Skin 0.7% 0.9% Patient melanoma tumors contain SP that excludes Hoechst dye 111308 Primary Skin 0.3% NA The presence of SP cells in patient melanoma tissues was first MB947p Primary Skin 0.43% 0.17% investigated by staining single-cell suspensions with Hoechst MB929 Metastasis Skin 0.27% NA MB947m Metastasis Lymph node 0.29% NA 33342 dye. Flow cytometry plots depict Hoechst excitation MB952 Metastasis Lymph node 0.34% 0.46% at blue and red wavelengths. The cells with the least amount MB1009 Metastasis Lymph node 0.16% 0.15% of Hoechst staining, which disappear in the presence of 82908 Metastasis Adrenal gland 0.13% NA verapamil, an inhibitor of Hoechst 33342 transport, are gated as SP (Figure 1a). All of the patient melanoma tumors, c For SP analysis including three skin primary melanomas and five metastatic (MF347-F0, 111308-F0, MB947p-F0, MB929-F0, F0 Patient tumor MB947m-F0, MB952-F0, MB1009-F0, 82908-F0) melanomas, contained a small but clear SP fraction, ranging For copy number analysis from 0.13 to 0.7% of all gated cells (Figure 1b). We then (MB929-F0, MB947m-F0) generated a direct in vivo xenograft model from human For microarray melanoma specimens to characterize SP cells from patient- (MB952-F1, MB1009-F1) For qRT-PCR derived tumors (Figure 1c). Original patient tumors were (MF347-F1, MB952-F1, MB1009-F1) designated as F0 tumors, whereas those grown from F0 For in vitro drug treatment F1 1st xenograft (MB947m-F1, MB952-F1, MB1009-F1) tumors were designated as F1 tumors. F1 tumors were further For IL-8 ELISA implanted in F2 mice for in vivo drug treatment (Figure 1c). (MB952-F1) For copy number analysis After characterizing SP cells using this model, cells from (MB929-F1, MB947m-F1) HS294T were utilized for further biological and mechanistic studies because the cell line contains a larger SP fraction For in vivo drug treatment (MB952-F2, MB1009-F2) (Supplementary Figure S1 online). F2 2nd xenograft For copy number analysis (MB929-F2, MB947m-F2)

Melanoma SP cells are resistant to paclitaxel and temozolomide For copy number analysis in vitro and in vivo Fn nth xenograft (MB929-F4) SP cells expressing ABC transporters have been shown to be Figure 1. Side population (SP) is a rare population in patient melanoma resistant to multiple drugs (Duan et al., 2004; Dean, 2009). tissues. (a) A representative SP flow cytometry profile from a patient tumor Paclitaxel, a substrate of ABCB1 transporter (Duan et al., tissue (MB952-F0). Cells were stained with Hoechst 33342 dye in the 2004), has been widely evaluated and shown to have positive presence (left panel) or absence (right panel) of verapamil. SP cells are shown outcomes in phase II/III clinical trials of malignant melanoma in the lower left gated area. (b) Frequency of SP fractions in eight patient (Bedikian et al., 2004; Hauschild et al., 2009). Therefore, we melanoma specimens (F0 tumors) and selected F1 tumors. (c) Study schema. treated human melanomas with paclitaxel to examine the Clinical melanoma specimens (F0) were used for SP analysis and copy drug sensitivity of SP cells. Temozolomide, an alkylating number analysis. Tumor specimens used in each analysis are listed in parentheses. Tumors were serially passaged into mice (F1, F2,- - -, Fn) for agent for melanoma treatment (Bleehen et al., 1995), is not a further analyses. NA, not analyzed. substrate of ABC transporters and thus was used as a control. As shown in Figure 2a, melanoma cells from MB1009-F1 tumor were sensitive to 20 nM paclitaxel treatment in vitro. in tumor size (79.4±5.9% of original size) was observed This treatment resulted in a 32.5-fold increase in SP cells after 7 days, but the tumors grew thereafter (125.1±13.2% of compared with vehicle control (Figure 2b). On the other original size on day 14; Figure 2c). Temozolomide showed hand, temozolomide treatment induced less cell death, and inhibitory effects similar to paclitaxel in vivo. Tumor analysis induced moderate increase in SP cells. The increase in SP on day 14 revealed an increase in SP cells after paclitaxel cells after drug treatment despite the absence of overall cell treatment and a modest increase after temozolomide treat- increase indicates that SP cells are more resistant to the drugs ment when compared with vehicle (Figure 2d), indicating than non-SP cells. that SP cells are resistant to paclitaxel and temozolomide To further analyze drug resistance in vivo, mice with in vivo. MB1009-F2 tumors were treated with paclitaxel or temozo- This relative resistance of SP cells to drugs was further lomide. Following paclitaxel treatment, a significant decrease investigated by treating SP and non-SP cells with drugs

www.jidonline.org 2441 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

a Vehicle 20 nM Paclitaxel 50 μM Temozolomide ** 80 *** 70 60 50 40 30 20 No. of cells per

microscopic field 10 0 Vehicle Paclitaxel Temozolomide

b Vehicle 20 nM Paclitaxel 50 μM Temozolomide *** 128 128 128 6.0% *** 96 96 96 5.0% 4.0% 64 64 64 3.0% 2.0% 32 0.17% 32 5.53% 32 0.79% 1.0% 0 0 0 0.0% 03264 96 128 03264 96 128 03264 96 128 Percentage of SP cells

Hoechst blue Vehicle Paclitaxel Temozolomide Hoechst red

c 260 240 Vehicle Temozolomide (30 mg kg–1) 220 Paclitaxel (10 mg kg–1) 200 180 160 140 ***

day 0 × 100% 120 100 Volume normalized to *** 80 60 0 2 4 6 8 10 12 14 16 Days after treatment

d Vehicle 20 nM Paclitaxel 50 μM Temozolomide ** 128 128 128 0.8% **

96 96 96 0.6%

64 64 64 0.4%

32 0.09% 32 0.58% 32 0.27% 0.2%

0 0 0 Percentage of SP cells 0.0% 0 326496128 0 32 64 96 128 0 32 64 96 128 Vehicle Paclitaxel Temozolomide Hoechst blue Hoechst red

e 1.2 * 1 SP cells Non-SP cells 0.8 *

0.6

0.4

0.2

Relative viability to vechicle 0 10 nM 20 nM 100 nM 50 μM 100 μM 200 μM Vehicle Paclitaxel Temozolomide

Figure 2. Drug resistance of side population (SP) cells. (a, left panels) Microscopic images of MB1009-F1 cells following 48-hour treatment with vehicle, paclitaxel, or temozolomide in vitro. Bar ¼ 100 mm. (a, right panel) Average number of cells in each microscopic field (n ¼ 6). (b, left panels) SP cells from MB952-F1 tumors after in vitro treatment. (b, right panel) Percentage of SP cells (n ¼ 3). (c) Growth curve of MB1009-F2 tumors after treatment with vehicle, paclitaxel, or temozolomide in vivo (n ¼ 6). ***Po0.001 compared with vehicle. (d, left panels) SP cells from MB952-F2 tumors after in vivo treatment. (d, right panel) Percentage of SP cells (n ¼ 3). (e) Viability of SP and non-SP cells from HS294T after 72 hours of treatment with vehicle, paclitaxel, or temozolomide (n ¼ 3). All data represent mean±SE. *Po0.05, **Po0.01, ***Po0.001.

separately. When SP cells and non-SP cells from HS294T ABCB1 and ABCB5 are selectively upregulated in human were treated with paclitaxel or temozolomide, SP cells melanoma SP cells showed less sensitivity to both drugs than non-SP cells, ABC transporters, particularly ABCB1 and ABCG2, have been confirming the drug resistance of SP cells compared with associated with resistance to multiple drugs (Dean et al., non-SP cells (Figure 2e). 2001; Duan et al., 2004; Frank et al., 2005; Dean, 2009). To

2442 Journal of Investigative Dermatology (2012), Volume 132 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

identify the ABC transporters responsible for the exclusion of Chelouche-Lev et al., 2004; Pilorget et al., 2007; Boeckmann Hoechst 33342 in melanoma SP cells, we first compared the et al., 2009). The expression levels of these genes in SP and expression levels of all 48 ABC transporters between SP and non-SP cells were therefore compared in our microarray data non-SP cells (from MB952-F1 and MB1009-F1 tumors) via using patient-derived tumors, and we found that IL-8, SPHK1, microarray analysis. Results demonstrated that SP cells and MLH3 were indeed selectively expressed in SP cells expressed 13-fold more ABCB1 and ABCB5 than non-SP (Figure 5a). As IL-8 was the most differentially expressed gene cells (Figure 3a), verifying the accuracy of the SP isolation (22-fold in MB952-F1 and 55-fold in MB1009-F1), we method and suggesting that these ABC transporters, but not focused on this molecule in our study. The upregulation of ABCG2, are likely responsible for the exclusion of Hoechst IL-8 in SP cells was verified by qRT-PCR in melanoma tissues 33342. Upregulation of ABCB1 and ABCB5 in SP cells was (3.8-fold and 23.2-fold in MB952-F1 and MB1009-F1 tumors, confirmed by real-time quantitative reverse transcription–PCR respectively; Figure 5b) and HS294T (2.8-fold; Figure 5c). (qRT-PCR) using MF347-F1 and MB1009-F1 tumors (Figure This was further confirmed by ELISA analysis in melanoma 3b), and flow cytometric analysis using HS294T (Figure 3c). tumor cells (16.1-fold in MB952-F1; Figure 5b) and HS294T (5.2-fold; Figure 5c). When HS294T cells were treated with Resistance to paclitaxel in SP cells is associated with ABC paclitaxel or temozolomide, IL-8 expression was enhanced transporters 2.7-fold or 2.2-fold, respectively (Figure 5d). To determine Because of the upregulation of ABCB1 and ABCB5 in SP cells, whether this increased IL-8 expression plays a role in we hypothesized that the drug resistance is associated with temozolomide resistance, we blocked the IL-8 signaling these transporters. Consistent with the hypothesis, the expres- pathway using anti-CXCR1, a neutralizing antibody for IL-8R, sion of ABCB1 and ABCB5 was upregulated after treatment in HS294T cells, and found this blockade significantly with paclitaxel or temozolomide in vitro (in HS294T; Figure increased the sensitivity of SP cells to temozolomide, an 4a) and in vivo (in MB1009-F2 tumors; Figure 4b). effect not observed in non-SP cells (Figure 5e). We then In order to determine if drug resistance is associated knocked down IL-8 using siRNA. Successful transfection of with a drug efflux capacity, tumor cells from MB952-F1 siRNA was confirmed by reduction in mRNA at 18 hours were treated with paclitaxel or temozolomide in vitro in (69%) and secreted level at 72 hours (37%) in the presence or absence of verapamil, an efflux pump HS294T cells (Figure 5f). Compared with control siRNA, inhibitor (Figure 4c). Verapamil treatment decreased sensi- transfection of IL-8 led to a 48% reduction of SP percentage tivity to paclitaxel but not temozolomide, suggesting that at 72 hours (Figure 5g), suggesting that SP phenotype of the resistance to paclitaxel in SP cells is at least partially melanoma cells is associated with IL-8. Furthermore, cell dependent on drug efflux. viability assay revealed that transient transfection of IL-8 Next, to determine if the drug resistance of SP cells is siRNA in HS294T decreased sensitivity to temozolomide associated with drug efflux by ABC transporters, we knocked (42%) but not paclitaxel (Figure 5h), suggesting that the down ABCB1 and ABCB5 using small interfering RNA resistance to temozolomide in SP cells is at least partially (siRNA). Successful transfection of siRNA was confirmed by dependent on IL-8. reduction in mRNA at 18 hours (60% and 65% reduction of Because ABCB1, ABCB5, and IL-8 are associated with SP ABCB1 and ABCB5, respectively) and levels at phenotypes, we investigated if they are related to each other. 72 hours in HS294T cells (Figure 4d). Compared with control We found that transfection of IL-8 siRNA did not affect siRNA, transfection of ABCB1 and ABCB5 siRNAs led to 74% the expression levels of ABCB1 and ABCB5 at 18 hours and 62% reduction in SP percentage at 72 hours, respectively (Supplementary Figure S2a online), and that transfection of (Figure 4e), suggesting that SP phenotype of melanoma cells ABCB1 or ABCB5 siRNA did not affect IL-8 expression levels is associated with ABCB1 and ABCB5. Cell viability assay at 18 hours either (Supplementary Figure S2b), suggesting that revealed that transient transfection of ABCB1 and ABCB5 these two mechanisms (ABC transporters and IL-8) are not siRNAs in HS294T decreased sensitivity to paclitaxel (38% directly associated with each other. and 39% decrease with ABCB1 and ABCB5 siRNAs, respectively) but not temozolomide (Figure 4f), suggesting Gene profiling analyses via microarray and qRT-PCR that the resistance to paclitaxel in SP cells is dependent on To further delineate chemoresistant mechanisms, we con- ABCB1 and ABCB5, whereas other mechanisms are respon- ducted microarray analysis to compare differentially ex- sible for temozolomide resistance in these cells. pressed genes between SP and non-SP cells. A total of 1,729 transcripts were upregulated and 1,888 transcripts were Resistance to temozolomide in SP cells is at least partly due to downregulated in SP cells (Po0.05, fold change 42.0; IL-8 upregulation Supplementary Figure S3a online). The most common Previous studies by other researchers have shown that IL-8, biological processes of these transcripts defined by gene sphingosine kinase (SPHK), mutL homolog (MLH), mutS ontology annotations included cellular process (28.54%), homolog (MSH), postmeiotic segregation (PMS), O-6-methyl- metabolic process (17.51%), and biological regulation guanine-DNA methyltransferase (MGMT), and excision (17.38%; Supplementary Figure S3b online). Significant repair cross-complementing rodent repair deficiency, com- pathway analysis using the 3,617 differentially expressed plementation group (ERCC) 1 were related to the resistance to transcripts identified NF-kB, a6-b4-integrin, and IL-1 path- temozolomide (Gerson, 2002; Narayan and Roy, 2003; ways (Po0.05; Supplementary Figure S4 online).

www.jidonline.org 2443 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

a Fold change Symbol Probe ID (SP/non-SP) Function ABCB5 650376 13.1 Drug transport

ABCB1 1230048 12.9 Multidrug resistance

ABCB4 1070594 7.9 Phosphatidylcholine transport

ABCG8 2690367 7.9 Sterol transport

ABCE1 1090543 6.9 Oligoadenylate binding protein

ABCA6 5810209 4.9 NA

ABCB9 5290598 4.2 NA

ABCB3 2650156 3.5 Peptide transport Associated with paroxysmal ABCC12 6290348 3.2 kinesigenic choreoathetosis ABCB8 3990592 2.2 NA

ABCG1 4290020 2.1 Cholesterol transport

b SP cells Non-SP cells ABCB1 ABCB5 ** 0.12 0.10 *** ** 0.10 0.08 0.08 ** 0.06 0.06 0.04 GAPDH

0.04 GAPDH 0.02 Relative fold to

0.02 Relative fold to 0 0 MF347-F1 MF1009-F1 MF347-F1 MF1009-F1

c SP Non-SP 104 104 18% *** 103 103 12% ABCB1 102 102 15.4% 2.9% 6% 101 101 Percentage of ABCB1-positive cells FL-1 100 100 0% 100 101 102 103 104 100 101 102 103 104 SP Non-SP ABCB1

4 4 10 10 *** 9% 103 103 6% ABCB5 102 102 8.3% 3.1% 101 101 3% Percentage of 0 FL-1 0 10 10 ABCB5-positive cells 0% 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 SP Non-SP ABCB5

Figure 3. Expression of ATP-binding-cassette (ABC) transporters in side population (SP) cells. (a) List of ABC transporters upregulated by 42-fold in SP cells from the microarray analysis, and their fold changes and functions. The genes with a fold change of 410.0 are underlined. NA indicates that gene function is not characterized. (b) Real-time quantitative reverse transcription–PCR (qRT-PCR) validation of ABCB5 and ABCB1 expression (mean±SE) in SP cells from MF347-F1 and MB1009-F1 tumors (n ¼ 3). GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (c, left panel) Representative flow cytometry analysis of ABCB1 and ABCB5 in SP and non-SP cells from HS294T. (c, right panel) Percentage of positive cells in SP and non-SP cells. **Po0.01, ***Po0.001.

More stringent analysis (Po0.001 and fold change 415) the limited amount of SP complementary DNA template, only revealed three unregulated genes (AXL, PCDHB11, and one unregulated gene (PCDHB11) and two downregulated ZNF519) and six downregulated genes (TBX2, FUK, IL1R1, genes (TBX2 and FUK) were validated by qRT-PCR. Expres- RAB40C, ACE, and UTF1) in SP cells (Figure 6a). Because of sion of PCDHB11 in SP cells from the MB1009-F1 tumor was

2444 Journal of Investigative Dermatology (2012), Volume 132 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

a ABCB1 ABCB5 ** 0.0003 *** *** 0.012 0.0002 0.008 GAPDH

GAPDH 0.0001 0.004 Relative fold to fold Relative Relative fold to fold Relative 0 0 Vehicle 20 nM 50 μM Vehicle 20 nM 50 μM Paclitaxel Temozolomide Paclitaxel Temozolomide

b Vehicle 10 mg kg–1 paclitaxel 30 mg kg–1 Temozolomide H&E * + 0.12 ** 0.10 0.08 0.06 cells 0.04 0.02 ABCB1 Ratio of ABCB1 0 Vehicle Paclitaxel Temozolomide * + 0.35 *** 0.30 0.25 0.20 cells

ABCB5 0.15 0.10 0.05 Ratio of ABCB5 0 Vehicle Paclitaxel Temozolomide ** c ** Without verapamil With verapamil 120% 100% 80% *

× 100% 60% 40% 20% vehicle vehicle

Relative viability to Relative 0% Vehicle 20 nM Paclitaxel 50 μM Temozolomide d ABCB1 ABCB5 0.00045 *** 0.016 *** ABCB1 0.012 ABCB5 0.00030 β β -Actin 0.008 -Actin

0.00015 GAPDH GAPDH 0.004 Relative fold to fold Relative Relative fold to fold Relative 0 0 si-ctrl si-ABCB1 si-ctrl si-ABCB5

e si-ctrl si-ABCB1 si-ABCB5 128 128 128 *** *** 96 96 96 10% 8% 64 64 64 6% 4% 32 9.11% 32 2.40%32 3.53% 2% 0 0 0 0%

0932 64 96 128 64320 6128 1289664320 of SP cells Percentage si-ctrl si-ABCB1 si-ABCB5 Hoechst blue Hoechst red ** f 1.2 * 1.0 0.8 0.6 0.4

Relative viability Relative 0.2 0 si-ctrl si-ABCB1 si-ABCB5 si-ctrl si-ABCB1 si-ABCB5 20 nM Paclitaxel 50 μM Temozolomide

Figure 4. Mechanisms for paclitaxel resistance in side population (SP) cells. (a) Real-time quantitative reverse transcription–PCR (qRT-PCR) of ABCB1 and ABCB5 in HS294T after in vitro drug treatment (n ¼ 3). GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (b, left panels) Hematoxylin and eosin (H&E), ABCB1, and ABCB5 staining of M1009-F2 tumors after in vivo treatment. Bar ¼ 100 mm. (b, right panel) Percentage of positive cells (n ¼ 3). (c) Viability of MB952-F1 melanoma cells following treatment, with or without verapamil in vitro (n ¼ 3). (d) qRT-PCR and western blot analysis of ABCB1 and ABCB5 in HS294T after small interfering RNA (siRNA) transfection of scrambled (si-ctrl), ABCB1 (si-ABCB1), or ABCB5 (si-ABCB5) (n ¼ 3). (e, left panels) SP cells in HS294T after siRNA transfection. (e, right panel) Percentage of SP cells (n ¼ 3). (f) Viability of siRNA-transfected cells after 72 hours of treatment with paclitaxel or temozolomide (n ¼ 9). All data represent mean±SE. *Po0.05, **Po0.01, ***Po0.001.

www.jidonline.org 2445 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

abSP cells Non-SP cells 3.0 *** 1,000 ***

2.5 ) 800 –1 2.0 ** 600 1.5

SP-MB952-F1 SP-MB1009-F1 Non-SP-MB952-F1 Non-SP-MB1009-F1 400 1.0

IL-8 IL-8 (pg ml 0.5 200 SPHK1 0 Relative fold to GAPDH 0 MLH3 MB952-F1 MB1009-F1 MB952-F1 SPHK2 c MLH1 SP cells Non-SP cells MSH2 0.16 *** 600 MSH3 Scale 0.14 3-fold above mean value )

2.75 0.12 –1 MSH6 2.5 400 *** 2.25 0.10 PMS1 2.0 1.75 0.08 1.5 0.06 PMS2 –1.5 GAPDH 200 –1.75 0.04 –2.0 IL-8 (pg ml MGMT –2.25 Relative fold to 0.02 –2.5 0 0 ERCCT –2.75 –3.0-fold below mean value SP Non-SP SP Non-SP

d e SP cells 1.2 Non-SP cells IL-8 1.0 ** ** 0.3 *** 0.8 0.6 0.2 0.4

0.1 Relative viability 0.2

0 0 Relative fold to GAPDH Vehicle 20 nM 50 μM Anti- Anti- Vehicle Vehicle Paclitaxel Temozolomide CXCR1 CXCR1 Vehicle 50 μM Temozolomide

fgsi-ctrl si-IL-8 *** 128 128 0.25 4.5 * )

0.20 –1 96 96 3.0 0.15 64 64 0.10

GAPDH 1.5

0.05 IL-8 (pg ml 32 9.11% 32 4.77% Relative fold to 0 0 si-ctrl si-IL-8 si-ctrl si-IL-8 0 0 0 32 64 96 128 096432 6128 Hoechst blue h Hoechst red 1.2 *** *** 10% 1.0 8% 0.8 6% 0.6 4% 0.4 2%

Relative viability 0.2 0%

0 Percentage of SP cells si-ctrl si-IL-8 si-ctrlsi-IL-8 si-ctrl si-IL-8 20 nM 50 μM Paclitaxel Temozolomide

Figure 5. Mechanisms for temozolomide resistance in side population (SP) cells. (a) Heatmap of temozolomide resistance genes. Red/green colors represent high/low expressions. (b, c) Real-time quantitative reverse transcription–PCR (qRT-PCR; left panels) and ELISA (right panels) of IL-8 in (b) patient-derived tumors and (c) HS294T. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (d) qRT-PCR of IL-8 in HS294T after treatment. (e) Viability of SP and non-SP from HS294T cultured with or without temozolomide, and treated with vehicle or anti-CXCR1. (f) qRT-PCR (left panel) and ELISA (right panel) of IL-8 after small interfering RNA (siRNA) transfection of scrambled (si-ctrl) or IL-8 (si-IL-8) in HS294T. (g, upper panel) Representative SP cells after IL-8-siRNA transfection in HS294T. (g, lower panel) Percentage of SP cells after transfection. (h) Viability of IL-8-siRNA-transfected HS294T after treatment. All data represent mean±SE. (n ¼ 3 except h where n ¼ 9). *Po0.05, **Po0.01, ***Po0.001.

2446 Journal of Investigative Dermatology (2012), Volume 132 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

a Upregulated in SP Fold change Symbol Probe ID Full name P-value Function (SP/non-SP) AXL receptor AXL 6480703 67.8 1.49E–04 Signal transducer tyrosine kinase

Protocadherin Calcium-dependent cell PCDHB11 2120504 24.2 7.60E–04 beta 11 adhesion protein Zine finger protein ZNF519 6760612 19.1 1.12E–04 Transcriptional regulation 519 Downregulated in SP Fold change Symbol Probe ID Full name P-value Function (non-SP/SP) Required for mesoderm TBX2 2360753 T-box 2 131.0 3.68E–04 differentiation FUK 6180204 Fucokinase 93.6 6.99E–04 Cell-cell interactions Interleukin 1 receptor Immune and inflammatory IL1R1 240274 55.2 7.72E–04 type I responses RAB40C, member Ubiquitination and RAB40C 7050193 RAS oncogene 27.5 9.05E–05 proteasomal degradation family of target proteins

Angiotensin I ACE 3830398 23.4 9.13E–04 Vasoconstrictor activity converting enzyme 1 Undifferentiated Transcriptional coactivator UTF1 3180341 embryonic cell 15.6 5.64E–04 of ATF23 transcription factor 1

b PCDHB11 FUK TBX2 ** * ** 0.08 0.04 0.03 0.06 0.03 0.02 0.04 0.02 0.01 to GAPDH 0.02 to GAPDH 0.01 Relative fold Relative Relative fold Relative to GAPDH 0 0 fold Relative 0 SP Non-SP SP Non-SP SP Non-SP

c Tumor Xenograft Genes ABCB1 ABCB5 IL-8 PCDHB11 FUK TBX2 F0 2 2 2 1 2 2 F1 2 2 2 1 2 2 MB929 F2 2 2 2 1 2 2 F4 2 2 2 1 2 2 F0 2 2 2 2 1 2 MB947m F1 2 2 2 2 1 2 F2 2 2 2 2 1 3

d Transporter resistance ABC transporters

Substrate of ABC transporters (paclitaxel)

Nonsubstrate of ABC transporters (temozolomide)

Intrinsic resistance

SP cell survives Non-SP cell dies

Figure 6. Differentially expressed genes in side population (SP) cells. (a) A list of microarray genes that most discriminated SP from non-SP cells (Po0.001, fold change 415). (b) Real-time quantitative reverse transcription–PCR (qRT-PCR) expression levels (mean±SE) of PCDHB11, FUK, and TBX2 in SP and non-SP cells (n ¼ 3). *Po0.05, **Po0.01. (c) Copy number analysis of genes analyzed in this study from F0 to Fn tumors from two patient-derived tumors. (d) Two different drug resistance systems exist in SP cells. ABC transporters contribute to transporter resistance by pumping out drugs that are their substrates. Intrinsic resistance represents inherent survival systems in SP cells, such as induction of certain signaling pathways, enzyme systems, reduced apoptosis, and increased DNA repair. Non-SP cells are sensitive to drug exposure because they lack the resistant systems.

www.jidonline.org 2447 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

0.073-fold relative to that of glyceraldehyde-3-phosphate chemoresistance to temozolomide in SP cells. IL-8 signaling dehydrogenase but its expression in non-SP cell was below activates a set of cell survival and proliferation pathways the detection level (o1 107–fold; Figure 6b). Also, the including MAPK, PI3-K, JAK2-STAT3, and b-catenin signaling non-SP/SP ratio was 91.7-fold for FUK, as measured by qRT- (Waugh and Wilson, 2008); thus, it comes as no surprise that PCR, and 16.0-fold for TBX2. These results were further this molecule plays an important role in the intrinsic confirmed using the MB952-F1 tumor. chemoresistance of SP cells. Previous studies using a direct in vivo xenograft model of In addition to ABC transporters and IL-8, melanoma pancreatic cancer have shown that the status of key genes in SP cells expressed genes and pathways associated with pancreatic cancer is stable and drug susceptibility is a stable development, stemness, and metabolic process. For example, trait after passaging tumors into mice (Rubio-Viqueira et al., PCDHB11 belongs to the protocadherin family, and 2006). To address whether passaging melanoma tumors in the lack of protocadherins causes massive apoptosis of mice alters genes in this study, we determined the copy spinal interneurons in a mouse model (Lefebvre et al., number of six genes (ABCB1, ABCB5, IL8, PCDHB11, FUK, 2008). TBX2 is necessary for cell proliferation and suppres- and TBX2) in F0, F1, F2, and Fn tumors from two patient- sion of senescence in melanoma (Vance et al., 2005). FUK is derived samples. As shown in Figure 6c, we observed high associated with fucosylation, and reduced fucosylation in concordance of gene copy number between original tumors cancer induces immunoevasion and metastasis (Moriwaki and xenografted tumors. et al., 2009). Three pathways identified in melanoma SP cells DISCUSSION may be closely related to chemoresistance. For example, This study isolated SP cells not only from the original patient NF-kB protects cells from apoptosis following DNA damage, melanoma tissues but also from in vivo xenograft mouse and is a principal mechanism for chemoresistance (Yama- models. Melanoma SP cells were shown to be resistant moto and Gaynor, 2001). Notably, NF-kB was shown to two melanoma drugs, paclitaxel and temozolomide, both to induce ABCB1 expression (Zhou and Kuo, 1997; in vitro and in vivo. Melanoma SP cells overexpressed Bentires-Alj et al., 2003). In addition, both a6-b4-integrin ABCB1 and ABCB5, and their resistance to paclitaxel and IL-1 signaling pathways induce resistance to apoptosis was at least partly attributed to ABC transporters. Further- by activation of NF-kB in tumors (Arlt et al., 2002; Weaver more, IL-8 plays a role in chemoresistance to temozolomide et al., 2002). in SP cells. Microarray and qRT-PCR studies demonstrated Therefore, it appears that chemoresistance of SP cells upregulation of PCDHB11 and downregulation of FUK and derives from at least two different types of mechanisms TBX2 in SP cells, and identified NF-kB, a6-b4-integrin, and (Figure 6d). One is the ABC-transporter-conferred resistance IL-1 signaling networks differentially expressed between SP due to the efflux of drugs. The other is the intrinsic resistance and non-SP cells. to cytotoxic drugs, possibly by induction of certain signaling ABCB1, previously known as MDR1 or P-glycoprotein, pathways and enzyme systems, changes in cytoskeleton, has been demonstrated to confer resistance to multiple reduced apoptosis, and increased DNA repair. hydrophobic drugs (Gottesman et al., 2002). Primary cultures In conclusion, we provide evidence that SP is an enriched of human melanoma contain ABCB1-positive cells that source of chemoresistant cells in human melanoma tumors. coexpress ABCB5 and ABCC2 (Keshet et al., 2008). ABCB5 These data will pave the road for further mechanistic studies was recently shown to mediate chemoresistance in human of melanoma SP cells, and provide potential targets for future melanoma (Frank et al., 2005) and identify melanoma- melanoma therapies. initiating cells (Schatton et al., 2008). ABCB1, ABCB4, and ABCB5 are closely homologous (Tusnady et al., 2006), and MATERIALS AND METHODS ABCB5 and ABCB1 have overlapping substrate specificities Clinical melanoma specimens (Frank et al., 2005), implying their cooperative functions in Human melanoma tissues were obtained from surgical specimens mediating multidrug resistance in SP cells. Furthermore, with written patient consent under institutional review board- recent studies indicate that in addition to serving as drug approved protocols at the University of Colorado Hospital, adhering efflux pumps, ABC transporters also play fundamental roles in to the Declaration of Helsinki Principles. Human melanoma tissues melanogenesis (Chen et al., 2009), tumor biology (Fletcher were cut into 3-mm3 pieces, and processed for cell isolation and et al., 2010), and immunomodulation (Schatton et al., 2010). direct in vivo xenograft. Our results are consistent with the role that ABC transporters play in melanomas and the chemoresistance signature that Direct in vivo xenograft characterizes SP cells. Animal experiments were performed under the institutional guide- The resistance of SP cells to temozolomide was observed lines for the use of laboratory animals. Under isofluorane anesthesia, in a glioma study (Bleau et al., 2009), but the specific patient tumors were implanted into 6-week-old female athymic (nu/ mechanisms were not previously studied. Interestingly, we nu) mice (NCI, Bethesda, MD) by a small incision and subcutaneous found that IL-8 was upregulated in melanoma SP cells, which pocket made in each side of the lower back (Rubio-Viqueira et al., is consistent with data from ovarian cancer and endometrial 2006). These mice were designated as F1 mice. F1 tumors were SP cells (Moserle et al., 2008; Cervello et al., 2010). In harvested when reaching a size of 1,500 mm3 and implanted into F2 addition, we found that IL-8 upregulation is responsible for mice. Tumors were serially passaged into mice.

2448 Journal of Investigative Dermatology (2012), Volume 132 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

Tumor cell isolation and sorting of SP cells from tumor cells Colorado Cancer Center (UCCC) and the Skin Disease Research Center Mechanically minced tumor tissues were digested with 235 U ml1 (SDRC) (Alistaire S. Acosta and Karen Helm) for helping with FACS sorting, 1 the Molecular Genetics Core from the SDRC and microarray core from UCD collagenase (Sigma-Aldrich, St Louis, MO) and 850 U ml hyalur- for helping with copy number analysis, and the Mind Research Network onidase (Sigma-Aldrich) for 2 hours at 37 1C with intermittent vortex, from New Mexico (Marilee Morgan and Kent Hutchison) for helping followed by sequentially passing through 70- and 40-mm filters to with microarray analysis. We thank Garrick Talmage, Tracey Ferrara, and Katherine Gowan for assisting experiments. This work was supported by the obtain single-cell suspension. Red blood cells were lysed using 1 US National Institutes of Health grants P30CA046934 (to NGA and MF and Red Blood Cell Lysis Buffer (eBioscience, San Diego, CA). Cells were the Flow Core of the UCCC), R03CA125833 (to MF), and P30AR057212 washed and re-suspended in DMEM with 2% fetal bovine serum and (Flow Core and the Molecular Genetics Core of the SDRC), Wendy Will Case 1% HEPES at a concentration of 1 106 cells per ml, and stained Cancer Fund (to MF), and the Tadamitsu Cancer Research Fund (to MF). with Hoechst 33342 dye (5 mgml1, Sigma-Aldrich) in the presence or absence of Verapamil (50 mM, Sigma-Aldrich) at 37 1C for SUPPLEMENTARY MATERIAL 90 minutes with intermittent vortex. Propidium iodide (1 mgml1) Supplementary material is linked to the online version of the paper at http:// was used to remove dead cells. Human stromal cells were www.nature.com/jid eliminated from human melanoma tumors using phycoerythrin- cy7-labeled anti-human CD45 (eBioscience) and anti-human CD31 REFERENCES (eBioscience) antibodies. Mouse stromal cells were eliminated Arlt A, Vorndamm J, Muerkoster S et al. (2002) Autocrine production of from xenografted tumors using phycoerythrin-cy7-labeled anti- interleukin 1beta confers constitutive nuclear factor kappaB activity mouse CD45 (eBioscience) and anti-mouse CD31 (eBioscience) and chemoresistance in pancreatic carcinoma cell lines. Cancer Res antibodies. Human melanoma cells were then positively selected 62:910–6 by allophycocyanin-labeled anti-human CD147 (eBioscience). Bedikian AY, Plager C, Papadopoulos N et al. (2004) Phase II evaluation of paclitaxel by short intravenous infusion in metastatic melanoma. SP and non-SP cells were sorted by MoFlo (DakoCytomation, Melanoma Res 14:63–6 Carpinteria, CA) and the results were analyzed with Summit software Bentires-Alj M, Barbu V, Fillet M et al. (2003) NF-kappaB transcription factor (DakoCytomation). induces drug resistance through MDR1 expression in cancer cells. Oncogene 22:90–7 Sorting of SP cells from cell lines Bleau AM, Hambardzumyan D, Ozawa T et al. (2009) PTEN/PI3K/Akt Human metastatic melanoma cell line, HS294T, was obtained from pathway regulates the side population phenotype and ABCG2 activity in glioma tumor stem-like cells. Cell Stem Cell 4:226–35 the American Type Culture Collection (Manassas, VA). Cells were Bleehen NM, Newlands ES, Lee SM et al. (1995) Cancer Research Campaign maintained in RPMI-1640 medium (Mediatech, Manassas, VA) phase II trial of temozolomide in metastatic melanoma. J Clin Oncol supplemented with 10% fetal bovine serum (Mediatech), 100 units 13:910–3 1 per ml penicillin, 0.1 mg ml streptomycin, and 2 mML-glutamine. Boeckmann L, Schirmer M, Rosenberger A et al. (2009) Effect of DNA repair Single cells were prepared using 1 mM EDTA (Sigma-Aldrich) for host factors on temozolomide or dacarbazine melanoma treatment in 15 minutes at 37 1C and stained with Hoechst 33342 dye for SP analysis Caucasians. Pharmacogenet Genomics 19:760–9 and sorting. Single cells were also used for biological analysis. Cervello I, Gil-Sanchis C, Mas A et al. (2010) Human endometrial side population cells exhibit genotypic, phenotypic and functional features of somatic stem cells. PLoS One 5:e10964 Copy number analysis of melanoma cells Chelouche-Lev D, Miller CP, Tellez C et al. (2004) Different signalling DNA from human melanoma cells was isolated using QIAamp DNA pathways regulate VEGF and IL-8 expression in breast cancer: implica- Mini Kit (Qiagen, Valencia, CA) and quantified by PicoGreen tions for therapy. Eur J Cancer 40:2509–18 dsDNA assay (Invitrogen, Grand Island, NY). Melanoma DNA Chen KG, Valencia JC, Gillet JP et al. (2009) Involvement of ABC transporters (250 ng) was hybridized to SNP probes (Illumina, San Diego, CA) at in melanogenesis and the development of multidrug resistance of melanoma. Pigment Cell Melanoma Res 22:740–9 the University of Colorado Denver microarray core. Probes consisted Daniel VC, Marchionni L, Hierman JS et al. (2009) A primary xenograft model of 595 markers: 293 for ABCB1 (chr7:87132948-87342611); 249 for of small-cell lung cancer reveals irreversible changes in gene expression ABCB5 (chr7:20654830-20816658); 6 for IL-8 (chr4:74606223- imposed by culture in vitro. Cancer Res 69:3364–73 74609433); 6 for PCDHB11 (chr5:140579183-140582618); 31 for Dean M (2009) ABC transporters, drug resistance, and cancer stem cells. FUK (chr16:70488498-70514177); and 10 for TBX2 (chr17: J Mammary Gland Biol Neoplasia 14:3–9 59477257-59486827). Probe copy number was analyzed using Dean M, Fojo T, Bates S (2005) Tumour stem cells and drug resistance. Nat CNApartition V3.1.6 (Illumina). Rev Cancer 5:275–84 Dean M, Hamon Y, Chimini G (2001) The human ATP-binding cassette (ABC) transporter superfamily. J Lipid Res 42:1007–17 Other methods Dou J, Wen P, Hu W et al. (2009) Identifying tumor stem-like cells in mouse Further information about the materials and methods used in this melanoma cell lines by analyzing the characteristics of side population work are provided in the Supplementary Materials and Methods cells. Cell Biol Int 33:807–15 online section. Duan Z, Brakora KA, Seiden MV (2004) Inhibition of ABCB1 (MDR1) and ABCB4 (MDR3) expression by small interfering RNA and reversal of paclitaxel resistance in human ovarian cancer cells. Mol Cancer Ther CONFLICT OF INTEREST 3:833–8 The authors state no conflict of interest. Fletcher JI, Haber M, Henderson MJ et al. (2010) ABC transporters in cancer: more than just drug efflux pumps. Nat Rev Cancer 10:147–56 ACKNOWLEDGMENTS Frank NY, Margaryan A, Huang Y et al. (2005) ABCB5-mediated doxorubicin We thank the University of Colorado Denver (UCD) melanoma tissue bank for transport and chemoresistance in human malignant melanoma. Cancer providing human melanoma samples, the Flow Core of the University of Res 65:4320–33

www.jidonline.org 2449 Y Luo et al. Gene Expression Profiling of SP in Human Melanoma

Gerson SL (2002) Clinical relevance of MGMT in the treatment of cancer. Pilorget A, Demeule M, Barakat S et al. (2007) Modulation of P-glycoprotein J Clin Oncol 20:2388–99 function by sphingosine kinase-1 in brain endothelial cells. J Neurochem Goodell MA, Brose K, Paradis G et al. (1996) Isolation and functional 100:1203–10 properties of murine hematopoietic stem cells that are replicating Rubio-Viqueira B, Jimeno A, Cusatis G et al. (2006) An in vivo platform for in vivo. J Exp Med 183:1797–806 translational drug development in pancreatic cancer. Clin Cancer Res Gottesman MM, Fojo T, Bates SE (2002) Multidrug resistance in cancer: role 12:4652–61 of ATP-dependent transporters. Nat Rev Cancer 2:48–58 Schatton T, Murphy GF, Frank NY et al. (2008) Identification of cells initiating Grichnik JM, Burch JA, Schulteis RD et al. (2006) Melanoma, a tumor based human melanomas. Nature 451:345–9 on a mutant stem cell? J Invest Dermatol 126:142–53 Schatton T, Schutte U, Frank NY et al. (2010) Modulation of T-cell Hauschild A, Agarwala SS, Trefzer U et al. (2009) Results of a phase III, activation by malignant melanoma initiating cells. Cancer Res 70: randomized, placebo-controlled study of sorafenib in combination with 697–708 carboplatin and paclitaxel as second-line treatment in patients with Siegel R, Ward E, Brawley O et al. (2011) Cancer statistics, 2011: the impact unresectable stage III or stage IV melanoma. J Clin Oncol 27:2823–30 of eliminating socioeconomic and racial disparities on premature cancer Helmbach H, Rossmann E, Kern MA et al. (2001) Drug-resistance in human deaths. CA Cancer J Clin 61:212–36 melanoma. Int J Cancer 93:617–22 Tusnady GE, Sarkadi B, Simon I et al. (2006) Membrane topology of human Hirschmann-Jax C, Foster AE, Wulf GG et al. (2004) A distinct ‘‘side ABC proteins. FEBS Lett 580:1017–22 population’’ of cells with high drug efflux capacity in human tumor cells. Vance KW, Carreira S, Brosch G et al. (2005) Tbx2 is overexpressed and plays Proc Natl Acad Sci USA 101:14228–33 an important role in maintaining proliferation and suppression of Keshet GI, Goldstein I, Itzhaki O et al. (2008) MDR1 expression identifies senescence in melanomas. Cancer Res 65:2260–8 human melanoma stem cells. Biochem Biophys Res Commun 368:930–6 Waugh DJ, Wilson C (2008) The interleukin-8 pathway in cancer. Clin Cancer La Porta CA (2007) Drug resistance in melanoma: new perspectives. Curr Med Res 14:6735–41 Chem 14:387–91 Weaver VM, Lelievre S, Lakins JN et al. (2002) beta4 integrin-dependent Lefebvre JL, Zhang Y, Meister M et al. (2008) gamma-Protocadherins regulate formation of polarized three-dimensional architecture confers resistance neuronal survival but are dispensable for circuit formation in retina. to apoptosis in normal and malignant mammary epithelium. Cancer Cell Development 135:4141–51 2:205–16 Moriwaki K, Noda K, Furukawa Y et al. (2009) Deficiency of GMDS leads to Yamamoto Y, Gaynor RB (2001) Therapeutic potential of inhibition of the NF- escape from NK cell-mediated tumor surveillance through modulation of kappaB pathway in the treatment of inflammation and cancer. J Clin TRAIL signaling. Gastroenterology 137:188–98, 98 e1-2 Invest 107:135–42 Moserle L, Indraccolo S, Ghisi M et al. (2008) The side population of ovarian Zhong Y, Guan K, Zhou C et al. (2010) Cancer stem cells sustaining cancer cells is a primary target of IFN-alpha antitumor effects. Cancer the growth of mouse melanoma are not rare. Cancer Lett 292: Res 68:5658–68 17–23 Narayan S, Roy D (2003) Role of APC and DNA mismatch repair genes in the Zhou G, Kuo MT (1997) NF-kappaB-mediated induction of mdr1b expression development of colorectal cancers. Mol Cancer 2:41 by insulin in rat hepatoma cells. J Biol Chem 272:15174–83

2450 Journal of Investigative Dermatology (2012), Volume 132