Cell Biology International 33 (2009) 1253e1262www.elsevier.com/locate/cellbi

Gene networks involved in apoptosis induced by hyperthermia in humanlymphoma U937 cells

Yukihiro Furusawa b,1, Yoshiaki Tabuchi a,*,1, Ichiro Takasaki a, Shigehito Wada c,Kenzo Ohtsuka d, Takashi Kondo b

a Division of Molecular Genetics Research, Life Science Research Center, University of Toyama 930-0194, 2630 Sugitani, Toyama, Japanb Department of Radiological Sciences, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama 930-0194, Japan

c Department of Oral and Maxillofacial Surgery, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama 930-0194, Japand Laboratory of Cell and Stress Biology, Department of Environmental Biology, Chubu University, Kasugai 487-8501, Japan

Received 6 February 2009; revised 22 June 2009; accepted 25 August 2009

Abstract

To define the molecular mechanisms that mediate hyperthermia-induced apoptosis, we performed microarray and computational geneexpression analyses. U937 cells, a human myelomonocytic lymphoma cell line, were treated with hyperthermia at 42 �C for 90 min and culturedat 37 �C. Apoptotic cells (~15%) were seen 6 h after hyperthermic treatment, and elevated expression of heat shock proteins (HSPs) includingHsp27, Hsp40, and Hsp70 was detected, following the activation of heat shock factor-1. Of the 54,675 probe sets analyzed, 1334 were upre-gulated and 4214 were downregulated by >2.0-fold in the cells treated with hyperthermia. A non-hierarchical gene clustering algorithm, K-means clustering, demonstrated 10 gene clusters. The gene network U1 or U2 that was obtained from up-regulated genes in cluster I or IXcontained HSPA1B, DNAJB1, HSPH1, and TXN or PML, LYN, and DUSP1, and were mainly associated with cellular compromise, and cellularfunction and maintenance or death, and cancer, respectively. In the decreased gene cluster II, the gene network D1 including CCNE1 and CEBPEwas associated with the cell cycle and cellular growth and proliferation. These findings will provide a basis for understanding the detailedmolecular mechanisms of apoptosis induced by hyperthermia at 42 �C in cells.� 2009 International Federation for Cell Biology. Published by Elsevier Ltd. All rights reserved.

Keywords: Hyperthermia; Apoptosis; Gene expression; Gene network; Heat shock protein

1. Introduction

In medicine, hyperthermia has been a promising approachin cancer therapy. The temperatures used in heat treatments for30e60 min are in the range of 40e45 �C in the case oflocoregional treatment, and up to 42 �C in the case of whole-body hyperthermia (Hall, 2000). Regarding the molecularaspect of hyperthermia, heat shock proteins (HSPs) are highlyconserved proteins in which synthesis can be induced bya variety of stresses, especially hyperthermia exposure. HSPsbehave as molecular chaperones, which act as the primary

* Corresponding author. Tel.: þ81 76 434 7187; fax: þ81 76 434 5176.

E-mail address: ytabu@cts.u-toyama.ac.jp (Y. Tabuchi).1 These authors contributed equally to this work.

1065-6995/$ - see front matter � 2009 International Federation for Cell Biology.

doi:10.1016/j.cellbi.2009.08.009

cellular defence against damage to the proteome, initiatingrefolding of denatured proteins and regulating degradationafter severe protein damage (Georgopoulos and Welch, 1993;Lindquist and Craig, 1988). HSPs protect cells both bylimiting the effects of protein-damaging agents throughprotein chaperoning and refolding and by directly blocking thecell-death pathways (Beere, 2004). In addition, HSPs arecrucial molecules in immune reponses and HSP chaperonetumor antigens are utilized for cancer immunotherapy basedon hyperthermia (Ito et al., 2006; Menoret and Chandawarkar,1998; Srivastava et al., 1998). Although many biologicalprocesses are affected by hyperthermia, the overall responsesto hyperthermia remain unclear.

DNA microarray technique has been used to identifychanges in gene expression induced by mild hyperthermia at

Published by Elsevier Ltd. All rights reserved.

1254 Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

41 �C (Tabuchi et al., 2008) and hyperthermia at 43e44 �C(Borkamo et al., 2008; Hirano et al., 2005; Kato et al., 2003;Narita et al., 2002; Sonna et al., 2002; Wada et al., 2007; Zhouet al., 2007). Although this method is valuable in differentialgene expression, there is an increasing need to move beyondthis level of analysis. Recently, bioinformatics tools used toexplore gene ontology and networks have been utilized toanalyze functionally microarray data. Borkamo et al. (2008)have reported that gene ontology mapping can be used to sort1213 differentially expressed genes into 10 main functionalgene groups, including those related to apoptosis, transcrip-tion, and the immune system in rat glioma cells treated withhyperthermia at 43 �C. Our previous observations indicate thathyperthermia treatment at 42 �C of human leukemia U937cells induced significant apoptosis, but this response followedmild hyperthermia treatment at 41 �C (Kameda et al., 2001).More recently, DNA microarray and gene network analysesdemonstrated that significant gene networks were associatedwith cellular function and maintenance or cell morphologyand the cell cycle in U937 cells treated with mild hyperthermiaat 41 �C without inducing apoptosis (Tabuchi et al., 2008). Toclarify the detailed mechanisms by which hyperthermia at42 �C induces apoptosis in human leukemia U937 cells, wecarried out a global-scale time-course microarray analysisusing a GeneChip� analyzing system, and. explored thefunctional relationships among the candidate genes usingIngenuity Pathway Analysis tools.

2. Materials and methods

2.1. Cell culture

U937 cells, a human myelomonocytic lymphoma cell line,were obtained from the Human Science Research ResourceBank, Human Science Foundation (Tokyo, Japan). They weremaintained as suspension cultures in RPMI1640 medium(Wako Pure Chemical Industries, Ltd., Osaka, Japan) supple-mented with 10% fetal bovine serum (Invitrogen, Carlsbad,CA, USA) at 37 �C in humidified air with 5% CO2 and 95% air.

2.2. Hyperthermia treatment

Cells (3� 106) were collected by centrifugation and sus-pended in 3 ml of fresh medium in plastic culture tubes beforebeing exposed to 42 �C (�0.05 �C) for 0e120 min in a waterbath. The temperature was monitored with a digital thermom-eter (No. 7563, Yokogawa, Tokyo, Japan) coupled to a thermo-couple 0.8 mm in diameter during heating. After hyperthermiatreatment, the cells were incubated for 0e6 h at 37 �C.

2.3. Analysis of apoptosis

For detection of DNA fragmentation, cells (3� 106) werelysed with 0.2 ml of lysis buffer (1 mM EDTA, 0.2% TritonX-100 and 10 mM TriseHCl, pH 7.5) and centrifuged at13,000� g for 10 min. Subsequently, each DNA in the super-natant and the pellet were precipitated in 12.5% trichloroacetic

acid at 4 �C and quantified using a diphenylamine reagent afterhydrolysis in 5% trichloroacetic acid at 90 �C for 20 min. Theabsorbance at 600 nm in each sample was determined afterovernight color development with diphenylamine reagent. Thepercentage of fragmented DNA in each sample was calculated asthe amount of DNA in the supernatant (containing fragmentedDNA) divided by total DNA for that sample (supernatant pluspellet) (Kameda et al., 2001).

For detection of the expression of phosphatidylserineoutside the plasma membrane of cells, flow cytometry wasused with propidium iodide (PI) and fluorescein isothiocyanate(FITC)-labeled annexin V (annexin V-FITC kit, Immunotech,Marseille, France). Briefly, cells (5� 105) were suspended inthe binding buffer of the kit. FITC-labeled annexin V and PIwere added to the cell suspension. After incubation for 10 minat room temperature, cell suspensions were analyzed usinga flow cytometer (Kameda et al., 2001).

2.4. SDS-polyacrylamide gel electrophoresis (PAGE)and western blotting

Cells were dissolved in a lysis buffer (50 mM NaCl, 1%Nonidet P-40 and 50 mM TriseHCl, pH 8.0) containinga protease inhibitor cocktail (NACALAI TESQUE, Inc., Kyoto,Japan). SDS-PAGE and Western blotting were carried out asdescribed elsewhere (Laemmli, 1970; Towbin et al., 1979). Theprimary antibodies used were as follows: a rabbit polyclonalanti-Hsp70 antibody (MBL Co., Ltd., Nagoya, Japan); a mousemonoclonal anti-Hsp40 antibody (MBL Co., Ltd.); a goatpolyclonal anti-Hsp60 antibody (Santa Cruz Biotechnology,Inc., CA, USA); a rat monoclonal anti-Hsp90 antibody (MBLCo., Ltd.); a rabbit polyclonal anti-Hsp27 antibody (MBL Co.,Ltd.); a rabbit polyclonal anti-Hsp110 antibody (MBL Co.,Ltd.); a rabbit polyclonal anti-heat shock transcription factor-1(HSF1) antibody (Stressgen Bioreagents Co., Ann Arbor, MI,USA); and a mouse monoclonal anti-glyceraldehyde-3-phos-phate dehydrogenase (GAPDH) antibody (Organon TeknikaCo., Durham, NC, USA). Immunoreactive proteins were visu-alized using a luminescent image analyzer (LAS-4000, FujifilmCo., Tokyo, Japan) using an enhanced chemiluminescencedetection system (Tabuchi et al., 2006a, 2008). The level ofGAPDH was constant over the culture periods (supplementarydata Fig. 1S A), and all the other protein expression levels werenormalized to it.

2.5. RNA isolation

Total RNA was extracted from cells using an RNeasy TotalRNA Extraction kit (Qiagen, Valencia, CA, USA) according tothe manufacture protocol, and treated with DNase I for 15 minat room temperature to remove residual genomic DNA.

2.6. High-density oligonucleotide microarray andcomputational gene expression analyses

Gene expression was analyzed using a GeneChip� systemwith a Human Genome U133-plus 2.0 array, which was spotted

1255Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

with 54,675 probe sets (Affymetrix, Santa Clara, CA, USA).Samples for array hybridization were prepared as described inthe Affymetrix GeneChip� Expression Technical Manual.Briefly, 5 mg of total RNA was used to synthesize double-stranded cDNA with a GeneChip� Expression 30-AmplificationReagents One-Cycle cDNA Synthesis kit (Affymetrix). Biotin-labeled cRNA was then synthesized from the cDNA usingGeneChip� Expression 30-Amplification Reagents for IVTLabeling (Affymetrix). After fragmentation, the biotinylatedcRNAwas hybridized to the GeneChip� array at 45 �C for 16 h.The arrays were washed, stained with streptavidinephycoery-thrin, and scanned using a probe array scanner. The scanned chipwas analyzed using GeneChip� Analysis Suite Software(Affymetrix). The obtained hybridization intensity data wereconverted into a presence or an absence call for each gene, andchanges in gene expression level between experiments weredetected by comparison analysis. The data were furtheranalyzed using GeneSpring (Silicon Genetics, Redwood City,CA, USA) to extract the significant genes (Tabuchi et al., 2006b,2007). To examine gene ontology, including biologicalprocesses, cellular components, molecular functions, and genenetworks, the obtained data were analyzed using IngenuityPathways Analysis tools (Ingenuity Systems, Mountain View,CA, USA), a web-delivered application that enables the iden-tification, visualization, and exploration of molecular interac-tion networks in gene expression data. The gene lists identifiedby GeneSpring containing Affymetrix gene ID and the naturallogarithm of normalized expression signal ratios from Gen-eChip� CEL files were uploaded into the Ingenuity PathwaysAnalysis system. Each gene identifier was mapped to its cor-responding gene object in the Ingenuity Pathways KnowledgeBase. These so-called focus genes were used as a starting pointfor generating biological networks (Tabuchi et al., 2006b, 2007).

2.7. Real-time quantitative polymerase chain reaction(qPCR) assay

Real-time qPCR assay was performed on a Real-Time PCRsystem (Mx3000P, Stratagene Japan, Tokyo, Japan) usingSYBR PreMix ExTaq (Takara Bio, Shiga, Japan) or qPCRMaster Mix (for the use of TaqMan probes; Eurogentec, Seraing,Belgium) according to the manufactures protocols. Reversetranscriptase reaction was carried out with DNase-treated totalRNA using a random primer pd(N)6. Real-time qPCR assay wasperformed by using the specific primers listed in supplementarydata Table S1. Temperature cycling conditions for each primerconsisted of 10 s at 95 �C followed by 40 cycles for 20 s at 95 �Cand 1 min at 60 �C. The dissociation analysis was carried outfrom 55 to 95 �C by monitoring SYBR Green fluorescence andPCR-specific products were determined as a single peak at themelting curves >80 �C. In addition, the specificity of primerswas confirmed as a single band with the correctly amplifiedfragment size through an agarose gel electrophoresis of the PCRproducts. The mRNA expression level of GAPDH was constantover the culture periods (supplementary data Fig. 1S B). EachmRNA expression level was normalized to that of GAPDH.

2.8. Statistical analysis

Data are shown as mean� SD. Significance was assessedby one-way or two-way ANOVA followed by Fisher’s PLSDtest. p values <0.05 were regarded as significant.

3. Results

3.1. Effects of hyperthermia on human lymphoma U937cells

The effects of hyperthermia at 42 �C on DNA fragmen-tation and phosphatidylserine externalization, biochemicalmarkers of apoptosis, of human lymphoma U937 cells wereexamined. Incubation for 60e120 min at 42 �C significantlyincreased DNA fragmentation, the levels being 12.7� 3.3,17.3� 1.3, and 13.1� 1.0% (mean� SD) for 60, 90, and120 min, respectively (Fig. 1A). The fraction of apoptoticcells estimated as positively stained with FITC-labeledannexin V and negatively stained with PI significantlyincreased in heated cells, the levels being 8.5� 1.4,13.3� 1.8, and 16.2� 2.4% (mean� SD) for 60, 90, and120 min at 42 �C, respectively (Fig. 1C). When U937 cellswere incubated at 42 �C for 90 min followed by culture at37 �C for different periods, DNA fragmentation and phos-phatidylserine externalization were significantly and time-dependently elevated (Fig. 1B, D).

An increase in the protein expression levels of HSPsincluding Hsp27, Hsp40 and Hsp70, and the activation ofHSF1 have been detected in human neuroblastoma IMR-32cells treated with hyperthermia at 42 �C for 2 h (Yan et al.,2004) and in human lymphoma U937 cells treated with mildhyperthermia at 41 �C for 30 min (Tabuchi et al., 2008). Theeffects of hyperthermia on the protein expression levels ofHSPs and HSF1 were examined using SDS-PAGE andWestern blotting (Fig. 2 and supplementary data Fig. 2S).Treatment with hyperthermia at 42 �C for 90 min resulted ina transient mobility shift of HSF1, with the levels from 0 to 3 hincubation being increased over those of the control (Fig. 2A).It is well known that the mobility shift of HSF1 indicatesactivation of the molecule (Sarge et al., 1993). An increase inthe protein expression levels of Hsp27, Hsp40, and Hsp70 wasobserved in the cells cultured for 3 and 6 h (Fig. 2D, F, G). Incontrast, the protein expression levels of Hsp60, Hsp90, andHsp110 were almost constant over the culture periods(Fig. 2B, C, E).

3.2. Global gene expression and cluster analyses

To analyze gene-expression profiling of the cells treatedwith hyperthermia, we carried out GeneChip� analysis of cellstreated with or without hyperthermia at 42 �C for 90 min. Ofthe 54,675 probe sets analyzed, we identified 25,284 (46.2%)probe sets expressed in control cells, and 24,355 (44.5%),24,779 (45.3%), 23,616 (43.2%), and 23,952 (43.8%) probesets expressed in the cells treated with the hyperthermia fol-lowed by culturing at 37 �C for 0, 1, 3 and 6 h, respectively.

Incubation time (min)0 60 120

*

*

*

Incubation time (h)

0 1 3 6

*

*

*

DN

A f

ragm

enta

tion

(%)

0

5

10

15

20

DN

A f

ragm

enta

tion

(%)

0

5

10

15

20

Incubation time (h)

0 1 3 6

**

*

Incubation time (min)

0 60 120

Frac

tion

of c

ells

(%

)

0

5

10

15

20

Frac

tion

of c

ells

(%

)

0

5

10

15

20*

*

*

A B

C D

Fig. 1. Effects of hyperthermia on apoptosis in human lymphoma U937 cells. (A, C) U937 cells were incubated at 42 �C for 0e120 min and cultured at 37 �C for

6 h. (B, D) The cells were incubated at 42 �C for 90 min and cultured at 37 �C for 0e6 h. DNA fragmentation (A, B) or phosphatidylserine externalization assay

(C, D) was carried out. Data are presented as mean� SD (n¼ 3). *P< 0.05 vs. control (non-treated cells).

1256 Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

The data indicate that the number of expressed genes remainedapproximately constant between these samples. The completelist of genes from all samples is available on the GeneExpression Omnibus: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc¼GSE13503.

We examined the genes that were up- or downregulated by>2.0-fold using GeneSpring software. A total of 5558 probesets were differentially expressed after hyperthermic treatment(1334 upregulated and 4214 downregulated in either point oftime); that is, 11.3% of probe sets were affected by thehyperthermia treatment. Furthermore, K-means clustering,a non-hierarchical gene clustering algorithm, was utilizedwithin GeneSpring to generate the major patterns of geneexpression during hyperthermia treatment. Ten gene clusterswere identified and clusters I-X contained 523, 671, 303, 1063,508, 557, 588, 329, 492, and 524 probe sets, respectively(Fig. 3). Clusters I, VIII, and IX contained up-regulated genes:the peak expressions of genes in clusters I, VIII, and IX were3, 6 and 1 h, respectively. Clusters II, III, IV, V, VI, VII, and Xcontained down-regulated genes: the expression levels ofgenes in clusters II, IV, and VII were gradually decreased; thebottom expressions of genes in clusters III, V, and X were 1, 3,and 0 h, respectively; the expression level of genes in clusterVI was downregulated from 0 to 3 h, and increased at 6 h(Fig. 3). List of genes of all clusters are available at ourhomepage: http://www.lsrc.u-toyama.ac.jp/mgrc/html/microarray.html.

3.3. Functional category and gene network analyses

Functional category analysis of the probe sets of the 10clusters was carried out using the Ingenuity PathwaysKnowledge Base. The top 3 functional categories of eachcluster based on the significance are shown in supplementarydata Table S2. The top 3 categories influenced by the hyper-thermia treatment in cluster I were cellular compromise,cellular function and maintenance, and drug metabolism; incluster II, the top categories were cellular growth and prolif-eration, cell cycle, and cell-to-cell signaling and interaction; incluster IX, the categories were cell death, cancer, and cellcycle (supplementary data Table S2).

Furthermore, to define the biologically relevant networks ofgenes identified here, we performed gene network analysis onthe differentially expressed genes using the Ingenuity Path-ways Knowledge Base. Several significant gene networks wereidentified (Figs. 4e6). The significant gene network U1including many HSPs such as heat shock 70 kDa protein 1B(HSPA1B), DnaJ (Hsp40) homolog, subfamily B, member 1(DNAJB1) and heat shock 105 kDa/110 kDa protein 1(HSPH1), and thioredoxin (TXN) was associated with cellularcompromise, and cellular function and maintenance in anincreased gene cluster I (Fig. 4), and the significant genenetwork U2 including promyelocytic leukemia (PML), v-yes-1Yamaguchi sarcoma viral related oncogene homolog (LYN),and dual specificity phosphatase 1 (DUSP1) was associated

HDP

AG/011ps

Hfo

oitaR

0 1 3 60

1

2

Incubation time (h)

HDP

AG/09ps

Hfo

oitaR

0 1 3 60

1

2

Incubation time (h)

HDP

AG/07ps

Hfo

oitaR

0 1 3 60

4

8

Incubation time (h)

*

* HDP

AG/06ps

Hfo

oitaR

0 1 3 60

1

2

Incubation time (h)

HDP

AG/1FS

Hfo

oitaR

0 1 3 60

Incubation time (h)

6

3

**

*

*

*

HDP

AG/04ps

Hfo

oitaR

0 1 3 60

1

2

Incubation time (h)

*

*HDP

AG/72ps

Hfo

oitaR

0 1 3 60

1

2

Incubation time (h)

*

A B C

D

G

E F

Fig. 2. Effects of hyperthermia on the protein expression levels of HSPs and heat shock transcription factor-1. U937 cells were incubated at 42 �C for 90 min and

then cultured at 37 �C for 0e6 h. SDS-PAGE and Western blotting were performed. A, heat shock transcription factor-1 (HSF1); B, Hsp110; C, Hsp90; D, Hsp70;

E, Hsp60; F, Hsp40; G, Hsp27; GAPDH (glyceraldehyde-3-phosphate dehydrogenase). Each protein expression level was normalized to GAPDH expression level.

The shadows indicate hyperthermia treatment. Data are presented as mean� SD (n¼ 4). *P< 0.05 vs. control (non-treated cells).

1257Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

with cell death and cancer in an increased gene cluster IX(Fig. 5). The gene network D1 that was obtained from down-regulated genes in cluster II contained CCNE1 (cyclin E1) andCEBPE (CCAAT/enhancer binding protein (C/EBP), epsilon)was found to be primarily associated with cell cycle, andcellular growth and proliferations (Fig. 6).

3.4. Verification of differentially expressed genes

To verify the results of gene expression analysis carried outwith the GeneChip� microarray system, real-time qPCR wasperformed. The expression levels of 12 selected genes werecomparable to those determined by microarray analysis. In 7up-regulated genes in the gene network U1, expression levelsof HSPA1B, TXN, heme oxygenase (decycling) 1 (HMOX1),HSPH1 and v-fos FBJ murine osteosarcoma viral oncogenehomolog (FOS) were gradually increased, and the peakexpressions of activating transcription factor 3 (ATF3) andFBJ murine osteosarcoma viral oncogene homolog B (FOSB)occurred from 1 to 3 h. In 3 up-regulated genes in the gene

network U2, the expression level of LYN was graduallyincreased, and the peak expressions of PML and DUSP1occurred at 3 and 1 h, respectively. The expression levels ofCCNE1 and CEBPE in the gene network D1 were graduallydecreased (Fig. 7).

4. Discussion

Although hyperthermia induces complex cellular responsesand changes in signal transduction, the overall responses tohyperthermia are difficult to explain using ordinary techniques.As a result, novel transcript-profiling technologies and compu-tational data analysis tools are being widely used in various lifescience research areas (Bredel et al., 2005; Huang et al., 2006). Inthe present study, gene networks in apoptosis elicited by hyper-thermia treatment at 42 �C in human lymphoma U937 cells wereidentified using global-scale DNA microarrays and computa-tional gene expression analysis tools. To our knowledge, this isthe first report regarding the identification of gene networksinfluenced by hyperthermia at 42 �C in cells.

Fig. 3. K-means clustering of differentially expressed probe sets by >2.0 in U937 cells treated with or without hyperthermia. Clustering was performed by

GeneSpring software. X- and y-axes indicate culture periods (h) and expression level, respectively. Cont: control (non-treated cells).

1258 Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

Our microarray analyses have demonstrated a large numberof probe sets (5558 probe sets) that were differentiallyexpressed by >2.0-fold in response to apoptosis of U937 cellselicited by hyperthermia at 42 �C for 90 min where apoptosiswas observed. In a previous study, 158 probe sets differentiallyexpressed by >2.0-fold were observed, although only mildhyperthermia (41 �C for 30 min) was used which did notinduce apoptosis (Tabuchi et al., 2008). However, the majorityof probe sets (97.9%; 5439/5558 probe sets) were specificallyexpressed in response to hyperthermia at 42 �C. We have nowbeen able to identify several significant gene networksresponsive to hyperthermia in U937 cells using the computa-tional gene expression analysis tools. Moreover, real-timeqPCR assay demonstrated that the expression levels of 12selected genes were comparable to those detected by themicroarray technique. Of particular interest was our identifi-cation of the cell death-associated gene network U2, whosecore contains PML, signal transducer and activator of tran-scription 3 (acute-phase response factor) (STAT3), and LYN(Fig. 5). Interestingly, all of the genes in U2 have beenreported to participate in apoptosis in a wide variety of cells;

that is, retinoblastoma 1 (RB1) (Day et al., 1997), PML (Xuet al., 2004), LYN (Yoshida et al., 2000), STAT3 (Bai et al.,2008), caspase 1, apoptosis-related cysteine peptidase (inter-leukin 1, beta, convertase) (CASP1) (Yamanaka et al., 2000)and caspase 4, apoptosis-related cysteine peptidase (CASP4)(Pelletier et al., 2006) all increase apoptosis of cells. Theactivation of caspase family members plays an important rolein apoptosis, and CASP4 protein is necessary for activation ofCASP1 protein (Earnshaw et al., 1999). In addition, LYNprotein shows elevate DNA binding activity and tyrosinephosphorylation of STAT3 protein (Schreiner et al., 2002).Alcalay et al. (1998) indicated that the interaction of PMLprotein and RB1 protein occurs in a cell fraction from U937cells. These cell death-associated genes in the gene networkU2 may be correlated with hyperthermia-induced apoptosis ofU937 cells. It is also noteworthy that several cell death-relatedgenes such as CASP4, LYN, and PML in U2 are not observedin mild hyperthermia-treated U937 cells without inducingapoptosis (Tabuchi et al., 2008). In addition, the cell cycle, andcellular growth and proliferation-associated gene network, D1,contained many down-regulated genes in cluster II (Fig. 6).

PP

PP

PP

PP

PP

PP

PPPP

PP

PPPP

PP

PP

PP

PP

PP

A, PP

PP

E

E, LO

E,PD

PD

E

E

E, I, PP

I, PP,RBPP, RB

PP

PP

RB

PP

PP

EE

HSPB1

HMOX1SOD2

TXN

HSPA1B

HSPA1A

DNAJA1

DNAJB1

FOS FOSB

HSPA1L

HSPA6HSPH1 KLF2

SMAD7

ATF3

PD, T

A, E, I, P, PD, RB, TPP

PP, TRE

PP

I, TPP

HSP90AA1

BAG3

STIP1

SERPINE1CLU

E

PP PP

PPPP

A, activation/deactivation

E, expression

I, inhibition

L, proteolysis

LO, localization

M, biochemical modification

RB, regulation of binding

P, phosphorylation/dephosphorylation

PD, protein-DNA binding

PP, protein-protein binding

T, transcription

TR, translocation

cytokine

enzyme

growth factor

G-protein-coupled receptor

ion channel

others

peptidase

translation factor

transcription regulator

transmembrane receptor

transporter

kinaseup-regulated

down-regulated

binding only

acts on

inhibits and acts on

direct interaction

indirect interaction

A B

A

A

B

B

Fig. 4. Gene network U1. Up-regulated probe sets of cluster I were analyzed by the Ingenuity Pathways Analysis software. The network is shown graphically as

nodes (genes) and edges (the biological relationships between the nodes).

1259Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

Cell growth or proliferation is positively regulated by CCNE1(Lin et al., 1996), BCL2 (B-cell CLL/lymphoma 2) (Gajateet al., 2000), and CEBPE (Shimizu et al., 2000) in mammaliancells, and there are relationships between BCL2 and CCNE1(Hoang et al., 1994) or CEBPE (Nakajima et al., 2006).

DUSP1

MAP2K5

LYN

PML

RB1

KHDRBS1

HBEGF

BCL2L11

PPP1R15A

BCL6 STAT3

PP

TNFAIP3

MLLPRDM2

CFLAR

PP

PP

CASP1

PRDM1

E

E

E

PP

PP

PP

PP

PPPP

PP

PP

PP

PP

P, PP

A, PP

E, I, PD, T

E, RB

RB

PD

PD, PP, T

L

LPD

PPD

RB

A, P, PP, RB

A, PP,RB

A, P, PP, TR

I

CASP4

PP

EPP

Fig. 5. Gene network U2. Up-regulated probe sets of cluster IX were analyzed

by the Ingenuity Pathways Analysis software. The network is shown graphi-

cally as nodes (genes) and edges (the biological relationships between the

nodes). For the explanation of the symbols and letters, see Fig. 4.

Decrease in gene expression in D1 may accelerate apoptosisinduced by hyperthermia in U937 cells.

There were only 119 probe sets containing many HSPsunder both the hyperthermia conditions in this study and themild hyperthermia condition (Tabuchi et al., 2008). Moreover,an elevation of the protein expression of Hsp27, Hsp40, and

AP2A2

RBL2

CEBPE

E, LO, P, PP

PP, RB

CUL1

SPHK1CCNE1

A, E, RB

BCR

PIK3R3

PP

IGF1R

TFDP1

CDC7

PTPN1

PP

PTK2B

EIF4E

P, PP,RB

E

A, E

EE

E

E

PP, TR

PP

DNM1

EA, E

A, M, P, PP,RB

P, PP, RB

PP, RB

PP

A, P

T

M, PP,RB

PD, T

A, LO, PP,RB

P, PP

E, PD

PP, RB, TR

BCL2

P, PP

RPS6KB1PRKCE

E

E PP

PP

E

MBP

Fig. 6. Gene network D1. Downregulated probe sets of cluster II were

analyzed by the Ingenuity Pathways Analysis software. The network is shown

graphically as nodes (genes) and edges (the biological relationships between

the nodes). For the explanation of the symbols and letters, see Fig. 4.

HDP

AG/

B1APS

Hfo

oitaR

0 1 3 60

1

2

Incubation time (h)

*

* *

* HDP

AG/

NX

Tfo

oitaR

0 1 3 60

1

2

Incubation time (h)

** *

*

HDP

AG/1

XO

MH

fooita

R

0 1 3 60

1

2

Incubation time (h)

*

**

*

**

*

HDP

AG/1

HPSH

fooita

R

0 1 3 60

1

2

Incubation time (h)

3

*

*HDP

AG/S

OFfo

oitaR

0 1 3 60

1

2

Incubation time (h)

** H

DPA

G/3FT

Afo

oitaR

0 1 3 60

1

2

Incubation time (h)

*

**

*

HDP

AG/1PS

UD

fooita

R

0 1 3 60

1

2

Incubation time (h)

3

*

*

*

*

HDP

AG/1

EN

CC

fooita

R

0 1 3 60

1

2

Incubation time (h)

* **

HDP

AG/

EPB

EC

fooita

R

0 1 3 60

1

2

Incubation time (h)

***

HDP

AG/

NY

Lfo

oitaR

0 1 3 60

1

2

Incubation time (h)

** *H

DPA

G/L

MPfo

oitaR

0 1 3 60

1

2

Incubation time (h)

5

3

4* *

*

HDP

AG/

BSOF

fooita

R

0 1 3 60

1

2

Incubation time (h)

A B C

D E F

G H I

J K L

Fig. 7. Verification of microarray results by RT-qPCR assay. U937 cells were incubated at 42 �C for 90 min and then cultured at 37 �C for 0e6 h. RT-qPCR assay

was performed. A, heat shock 70 kDa protein 1B (HSPA1B); B, thioredoxin (TXN); C, heme oxygenase (decycling) 1 (HMOX1); D, heat shock 105 kDa/110 kDa

protein 1 (HSPH1); E, v-fos FBJ murine osteosarcoma viral oncogene homolog (FOS); F, activating transcription factor 3 (ATF3); G, FBJ murine osteosarcoma

viral oncogene homolog B (FOSB); H, v-yes-1 Yamaguchi sarcoma viral related oncogene homolog (LYN); I, promyelocytic leukemia (PML); J, dual specificity

phosphatase 1 (DUSP1); K, cyclin E1 (CCNE1); L, CCAAT/enhancer binding protein (C/EBP) epsilon (CEBPE); glyceraldehyde-3-phosphate dehydrogenase

(GAPDH). Each mRNA expression level was normalized to GAPDH expression level. The shadows indicate hyperthermia treatment. Data are presented as

mean� SD (n¼ 4). *P< 0.05 vs. control (non-treated cells).

1260 Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

1261Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

Hsp70 was observed following the mobility shift of HSF1 inU937 cells under both conditions (Tabuchi et al., 2008). In thepresent study, the gene network U1 that contained up-regu-lated genes and that was associated with cellular compromise,and cellular function and maintenance was identified (Fig. 4).This network included many HSPs such as HSPA1B,DNAJB1, and HSPH1 and oxidative stress-responsive genessuch as TXN, superoxide dismutase 2, mitochondrial (SOD2),and HMOX1. Stress-induced-phosphoprotein 1 (STIP1) isregulated by cell-cycle kinases and mediates the assembly ofthe Hsp70/Hsp90 chaperone heterocomplex (Longshaw et al.,2004). A relationship between STIP and heat shock 70 kDaprotein 1A (HSPA1A) (Hernandez et al., 2002), HSPA1B(Hernandez et al., 2002), or heat shock protein 90 kDa alpha(cytosolic), class A member 1 (HSP90AA1) (Catimel et al.,2005) has been demonstrated. HSPs have been known tofunction as molecular chaperones and to possess anti-apoptoticaction (Beere et al., 2000; Mosser et al., 1997). It is thereforepossible that up-regulated U1 genes play a role in cytopro-tection from hyperthermia-induced apoptosis. The presence ofdifferentially expressed genes including HSPs has been iden-tified in a wide variety of types of cells treated with heat(Borkamo et al., 2008; Hirano et al., 2005; Kato et al., 2003;Narita et al., 2002; Sonna et al., 2002; Tabuchi et al., 2008;Wada et al., 2007; Zhou et al., 2007).

The present results indicate that the differentially expressedgenes and gene networks identified here are probably involvedin apoptosis induced by hyperthermia at 42 �C. The currentdata will provide a better understanding of the mechanism ofapoptosis induced by hyperthermia at 42 �C in cells.

Acknowledgement

This study was supported in part by a Grant-in-Aid fromthe Japanese Ministry of Education, Culture, Sports, Science,and Technology.

Appendix. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.cellbi.2009.08.009.

References

Alcalay M, Tomassoni L, Colombo E, Stoldt S, Grignani F, Fagioli M, et al.

The promyelocytic leukemia gene product (PML) forms stable complexes

with the retinoblastoma protein. Mol Cell Biol 1998;18:1084e93.

Bai Y, Ahmad U, Wang Y, Li JH, Choy JC, Kim RW, et al. Interferon-gamma

induces X-linked inhibitor of apoptosis-associated factor-1 and Noxa

expression and potentiates human vascular smooth muscle cell apoptosis

by STAT3 activation. J Biol Chem 2008;283:6832e42.

Beere HM. ‘‘The stress of dying’’: the role of heat shock proteins in the

regulation of apoptosis. J Cell Sci 2004;117:2641e51.

Beere HM, Wolf BB, Cain K, Mosser DD, Mahboubi A, Kuwana T, et al.

Heat-shock protein 70 inhibits apoptosis by preventing recruitment of

procaspase-9 to the Apaf-1 apoptosome. Nat Cell Biol 2000;2:469e75.

Borkamo ED, Dahl O, Bruland O, Fluge O. Global gene expression analyses

reveal changes in biological processes after hyperthermia in a rat glioma

model. Int J Hyperthermia 2008;24:425e41.

Bredel M, Bredel C, Juric D, Harsh GR, Vogel H, Recht LD, et al. Functional

network analysis reveals extended gliomagenesis pathway maps and three

novel MYC-interacting genes in human gliomas. Cancer Res 2005;65:

8679e89.

Catimel B, Rothacker J, Catimel J, Faux M, Ross J, Connolly L, et al.

Biosensor-based micro-affinity purification for the proteomic analysis of

protein complexes. J Proteome Res 2005;4:1646e56.

Day ML, Foster RG, Day KC, Zhao X, Humphrey P, Swanson P, et al. Cell

anchorage regulates apoptosis through the retinoblastoma tumor

suppressor/E2F pathway. J Biol Chem 1997;272:8125e8.

Earnshaw WC, Martins LM, Kaufmann SH. Mammalian caspases: structure,

activation, substrates, and functions during apoptosis. Annu Rev Biochem

1999;68:383e424.

Gajate C, Barasoain I, Andreu JM, Mollinedo F. Induction of apoptosis in

leukemic cells by the reversible microtubule-disrupting agent 2-methoxy-

5-(20,30,40-trimethoxyphenyl)-2,4,6-cycloheptatrien-1-one: protection by

Bcl-2 and Bcl-X(L) and cell cycle arrest. Cancer Res 2000;60:2651e9.

Georgopoulos C, Welch WJ. Role of the major heat shock proteins as

molecular chaperones. Annu Rev Cell Biol 1993;9:601e34.

Hall EJ. Hyperthermia. In: Radiobiology for the radiologist. 5th ed. New York:

Lippincott Williams & Wilkins; 2000. p. 495e520.

Hernandez MP, Sullivan WP, Toft DO. The assembly and intermolecular

properties of the hsp70-Hop-hsp90 molecular chaperone complex. J Biol

Chem 2002;277:38294e304.

Hirano H, Tabuchi Y, Kondo T, Zhao Q-L, Ogawa R, Cui Z-G, et al. Analysis

of gene expression in apoptosis in human lymphoma U937 cells induced

by heat shock and the effects of a-phenyl N-tertbutylnitrone (PBN) and its

derivatives. Apoptosis 2005;10:331e40.

Hoang AT, Cohen KJ, Barrett JF, Bergstrom DA, Dang CV. Participation of

cyclin A in Myc-induced apoptosis. Proc Natl Acad Sci U S A 1994;91:

6875e9.

Huang Y, Yan J, Lubet R, Kensler TW, Sutter TR. Identification of novel

transcriptional networks in response to treatment with the anticarcinogen

3H-1,2-dithiole-3-thione. Physiol Genomics 2006;24:144e53.

Ito A, Honda H, Kobayashi T. Cancer immunotherapy based on intracellular

hyperthermia using magnetite nanoparticles: a novel concept of ‘‘heat-

controlled necrosis’’ with heat shock protein expression. Cancer Immunol

Immunother 2006;55:320e8.

Kameda K, Kondo T, Tanabe K, Zhao QL, Seto H. The role of intracellular

Ca2þ in apoptosis induced by hyperthermia and its enhancement by

verapamil in U937 cells. Int J Radiat Oncol Biol Phys 2001;49:1369e79.

Kato N, Kobayashi T, Honda H. Screening of stress enhancer based on anal-

ysis of gene expression profiles: enhancement of hyperthermia-induced

tumor necrosis by an MMP-3 inhibitor. Cancer Sci 2003;94:644e9.

Laemmli UK. Cleavage of structural proteins during the assembly of the head

of bacteriophage T4. Nature 1970;227:680e5.

Lin J, Reichner C, Wu X, Levine AJ. Analysis of wild-type and mutant

p21WAF-1 gene activities. Mol Cell Biol 1996;16:1786e93.

Lindquist S, Craig EA. The heat-shock proteins. Annu Rev Genet 1988;22:

631e77.

Longshaw VM, Chapple JP, Balda MS, Cheetham ME, Blatch GL. Nuclear

translocation of the Hsp70/Hsp90 organizing protein mSTI1 is regulated

by cell cycle kinases. J Cell Sci 2004;117:701e10.

Menoret A, Chandawarkar R. Heat-shock protein-based anticancer immuno-

therapy: an idea whose time has come. Semin Oncol 1998;25:654e60.

Mosser DD, Caron AW, Bourget L, Denis-Larose C, Massie B. Role of the

human heat shock protein hsp70 in protection against stress-induced

apoptosis. Mol Cell Biol 1997;17:5317e27.

Nakajima H, Watanabe N, Shibata F, Kitamura T, Ikeda Y, Handa M.

N-terminal region of CCAAT/enhancer-binding protein epsilon is critical

for cell cycle arrest, apoptosis, and functional maturation during myeloid

differentiation. J Biol Chem 2006;281:14494e502.

Narita N, Noda I, Ohtsubo T, Fujieda S, Tokuriki M, Saito T, et al. Analysis of

heat-shock related gene expression in head-and-neck cancer using cDNA

arrays. Int J Radiat Oncol Biol Phys 2002;53:190e6.

1262 Y. Furusawa et al. / Cell Biology International 33 (2009) 1253e1262

Pelletier N, Casamayor-Palleja M, De Luca K, Mondiere P, Saltel F, Jurdic P,

et al. The endoplasmic reticulum is a key component of the plasma cell

death pathway. J Immunol 2006;176:1340e7.

Sarge KD, Murphy SP, Morimoto RI. Activation of heat shock gene tran-

scription by heat shock factor 1 involves oligomerization, acquisition of

DNA-binding activity, and nuclear localization and can occur in the

absence of stress. Mol Cell Biol 1993;13:1392e407.

Schreiner SJ, Schiavone AP, Smithgall TE. Activation of STAT3 by the Src

family kinase Hck requires a functional SH3 domain. J Biol Chem 2002;

277:45680e7.

Shimizu K, Kitabayashi I, Kamada N, Abe T, Maseki N, Suzukawa K,

et al. AML1-MTG8 leukemic protein induces the expression of gran-

ulocyte colony-stimulating factor (G-CSF) receptor through the up-

regulation of CCAAT/enhancer binding protein epsilon. Blood 2000;96:

288e96.

Sonna LA, Gaffin SL, Pratt RE, Cullivan ML, Angel KC, Lilly CM. Effect of

acute heat shock on gene expression by human peripheral blood mono-

nuclear cells. J Appl Physiol 2002;92:2208e20.

Srivastava PK, Menoret A, Basu S, Binder RJ, McQuade KL. Heat shock

proteins come of age: primitive functions acquire new roles in an adaptive

world. Immunity 1998;8:657e65.

Tabuchi Y, Kuribayashi R, Takasaki I, Doi T, Sakai H, Takeguchi N, et al.

Overexpression of heat shock protein 70 restores the structural

stability and functional defects of temperature-sensitive mutant of

large T antigen at nonpermissive temperature. Cell Stress Chaperones

2006a;11:259e67.

Tabuchi Y, Takasaki I, Doi T, Ishii Y, Sakai H, Kondo T. Genetic networks

responsive to sodium butyrate in colonic epithelial cells. FEBS Lett 2006b;

80:3035e41.

Tabuchi Y, Takasaki I, Suto A, Kondo T, Suzuki Y, Obinata M. Genetic

networks in nonpermissive temperature-induced cell differentiation of

sertoli TTE3 cells harboring temperature-sensitive SV40 large T-antigen.

Cell Biol Int 2007;31:1231e6.

Tabuchi Y, Takasaki I, Wada S, Zhao QL, Hori T, Nomura T, et al. Genes and

genetic networks responsive to mild hyperthermia in human lymphoma

U937 cells. Int J Hyperthermia 2008;24:613e22.

Towbin H, Staehelin T, Gordon J. Electrophoretic transfer of proteins from

polyacrylamide gels to nitrocellulose sheets: procedure and some appli-

cations. Proc Natl Acad Sci U S A 1979;76:4350e4.

Wada S, Tabuchi Y, Kondo T, Cui ZG, Zhao QL, Takasaki I, et al. Gene

expression in enhanced apoptosis of human lymphoma U937 cells treated

with the combination of different free radical generators and hyperthermia.

Free Radic Res 2007;41:73e81.

Xu ZX, Zhao RX, Ding T, Tran TT, Zhang W, Pandolfi PP, et al. Promyelo-

cytic leukemia protein 4 induces apoptosis by inhibition of survivin

expression. J Biol Chem 2004;279:1838e44.

Yamanaka K, Tanaka M, Tsutsui H, Kupper TS, Asahi K, Okamura H, et al.

Skin-specific caspase-1-transgenic mice show cutaneous apoptosis and

pre-endotoxin shock condition with a high serum level of IL-18. J

Immunol 2000;165:997e1003.

Yan D, Saito K, Ohmi Y, Fujie N, Ohtsuka K. Paeoniflorin, a novel heat shock

protein-inducing compound. Cell Stress Chaperones 2004;9:378e89.

Yoshida K, Weichselbaum R, Kharbanda S, Kufe D. Role for Lyn tyrosine

kinase as a regulator of stress-activated protein kinase activity in response

to DNA damage. Mol Cell Biol 2000;20:5370e80.

Zhou M, Zhang A, Lin B, Liu J, Xu LX. Study of heat shock response of

human umbilical vein endothelial cells (HUVECs) using cDNA micro-

array. Int J Hyperthermia 2007;23:225e58.