ENHANCED IN VITRO INVASIVENESS AND DRUG RESISTANCE WITHALTERED GENE EXPRESSION PATTERNS IN A HUMAN LUNG CARCINOMACELL LINE AFTER PULSE SELECTION WITH ANTICANCER DRUGSYizheng LIANG

1, Lorraine O’DRISCOLL1, Susan MCDONNELL

2, Padraig DOOLAN1, Irene OGLESBY

1, Kieran DUFFY1, Robert O’CONNOR

1

and Martin CLYNES1*

1National Cell and Tissue Culture Centre National Institute for Cellular Biotechnology, Dublin City University, Dublin, Ireland2School of Biotechnology, Dublin City University, Dublin, Ireland

The human lung carcinoma cell line DLKP was exposed tosequential pulses of 10 commonly used chemotherapeuticdrugs (VP-16, vincristine, taxotere, mitoxantrone, 5-fluoro-uracil, methotrexate, CCNU, BCNU, cisplatin and chloram-bucil); resulting cell lines exhibited resistance to the selectingagents (ranging approx. 1.5- to 36-fold) and, in some cases,cross-resistance to methotrexate (approx. 1.4- to 22-fold),vincristine (1.6- to 262-fold), doxorubicin (Adriamycin, ap-prox. 1.1- to 33-fold) and taxotere (approx. 1.1- to 36-fold).Several of the variants displayed collateral sensitivity to cis-platin. A marked increase in in vitro invasiveness and motilitywas observed with variants pulsed with mitoxantrone, 5-flu-orouracil, methotrexate, BCNU, cisplatin and chlorambucil.There was no significant change in invasiveness of cells pulsedwith VP-16, vincristine, taxotere or CCNU. All of the pulse-selected variants showed elevated levels of MDR-1/P-gp pro-tein by Western blot analysis, although mdr-1 mRNA levelswere not increased (except for DLKP-taxotere). In DLKP-taxotere, MRP1 protein levels were also greatly elevated, butmrp1 mRNA levels remained unchanged. BCRP was upregu-lated in DLKP-mitoxantrone at both the mRNA and proteinlevels. Gelatin zymography, Western blot and RT-PCRshowed that DLKP and its variants secreted MMPs 2, 9 and13. MMP inhibition assays suggested that MMP-2 plays amore important role than MMPs 9 and 13 in cell invasion ofthese DLKP drug-resistant variants in vitro. These resultsindicate that drug exposure may induce not only resistancebut also invasiveness in cancer cells.© 2004 Wiley-Liss, Inc.

Key words: invasion; metastasis; chemotherapy; anticancer drug;multidrug resistance; matrix metalloproteinase; P-glycoprotein; mul-tidrug resistance protein; lung cancer

MDR1 and cancer invasion/metastasis2 are major causes ofdeath for cancer patients. So far, however, little is known about therelationship between drug resistance and cancer invasion/metasta-sis.

Reducing intracellular drug levels, often associated with over-expression of P-gp,3 members of the MRP family4 and the BCRP/MXR protein,5 is an important and frequent mechanism of MDR.

Invasion through basement membrane is believed to be a criticalstep in metastasis. There are 4 main classes of proteases involvedin proteolytic degradation of the ECM: serine-, cysteine- andaspartyl-proteases as well as MMPs.6 The MMPs are a family ofmore than 20 members produced by tumour cells, connectivetissue cells and inflammatory cells. MMPs are secreted as latentproenzymes and activated by proteolytic removal of an amino-terminal domain.7 A large body of evidence shows that MMPsplay a crucial role in cancer invasion and metastasis. Based onsequence homology and substrate specificity, MMP-2 and MMP-9are classified as type IV collagenases that degrade the majorcomponent (type IV collagen) of basement membrane;8 MMP-13,also called “collagenase-3”, belongs to a group known as intersti-tial collagenases, which degrade collagen types I, II, III, VII, VIIIand X.9

The possibility of a relationship between drug resistance andinvasion/metastasis phenotypes has been raised by 2 types ofobservation:10 firstly, some tumour cells selected for resistance to

drugs are invasive/metastatic relative to nonresistant parental cells;secondly, in some cases, secondary (more metastatic) tumours aremore resistant to chemotherapeutic drugs than their primary coun-terparts. In support of this, Osmak et al.11 reported a study ofdrug-resistant cervical and laryngeal cell lines, in which levels ofmarkers for invasion and metastasis (cathepsin D, uPA and PAI-1)were increased, in some cases, in association with drug resistance;however, the invasive/metastatic potential of these cells was notreported. In other instances reported in the literature, no correlationwas seen between drug exposure/resistance and cancer invasion/metastasis.10

We have previously demonstrated that exposure of the humannasal carcinoma cell line RPMI-2650 to melphalan, an alkylatingagent, can result in not only an MDR phenotype but also enhancedcell invasion in vitro, whereas paclitaxel (Taxol) exposure resultedin MDR without increasing cell invasiveness.12 To further explorethe association between drug resistance and cancer invasion/me-tastasis and, in particular, whether such a phenomenon can beobserved in a different kind of human tumour and with differentdrugs, we studied this phenomenon in tumour cells from a differentorgan of origin, using a broad range of chemotherapeutic drugs.We pulse-selected the poorly differentiated human lung squamouscarcinoma cell line DLKP with 10 commonly used drugs, includ-ing VP-16, VCR, TXT, MITOX, the 2 antimetabolites 5-FU andMTX and the 4 alkylating agents CCNU, BCNU, CPT and CHLO,and extensively characterised the resulting cell populations.

Abbreviations: BCNU, 1,3-bis(2-chloroethyl)-1-nitrosourea; BCRP,breast cancer resistance protein; CCNU, 1-(2-chloroethyl)-3-cyclohexyl-1-nitrosourea; CHLO, chlorambucil; cMOAT, canalicular multispecific or-ganic anion transporter; CPT, cisplatin; ECM, extracellular matrix; 5-FU,5-fluorouracil; IC50, 50% inhibitory concentration; MAb, monoclonal an-tibody; MDR, multidrug resistance; MITOX, mitoxantrone; MMP, matrixmetalloproteinase; MRP, multidrug resistance protein; MTX, methotrex-ate; MXR, mitoxantrone resistance; NSCLC, non-small-cell lung cancer;PAI, plasminogen activator inhibitor; P-gp, P-glycoprotein; SCLC, small-cell lung cancer; TXT, taxotere; uPA, urokinase plasminogen activator;VCR, vincristine; VP-16, etoposide.

Grant sponsor: Higher Education Authority.

*Correspondence to: National Cell and Tissue Culture Centre NationalInstitute for Cellular Biotechnology, Dublin City University, Dublin 9,Ireland. Fax: �353-1-7005484. E-mail: martin.clynes@dcu.ie

Received 12 March 2003; Revised 2 October 2003, 12 January 2004;Accepted 27 January 2004

DOI 10.1002/ijc.20230Published online 4 May 2004 in Wiley InterScience (www.interscience.

wiley.com).

Int. J. Cancer: 111, 484–493 (2004)© 2004 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

MATERIAL AND METHODS

ChemicalsVP-16, BCNU and CCNU were obtained from Bristol Myers

Squibb (Dublin, Ireland); VCR, MTX and 5-FU were from DavidBull Laboratories (Warwick, UK); cisplatin and mitoxantronewere from Lederle Laboratories (Gosport Hampshire, UK); TXTwas from Aventis (Antony Cedex, France); and chlorambucil wasfrom Sigma (Dublin, Ireland). All media and serum used in themaintenance of cells were obtained from GIBCO BRL (Paisley, UK).Reagents used in RT-PCR were purchased from Sigma. All 3 MMPinhibitors were purchased from Calbiochem (Nottingham, UK).

Cell cultureThe poorly differentiated human squamous cell line DLKP13

was cultured in DMEM/Ham’s F12 (1:1) supplemented with 5%FCS and 1% L-glutamine. DLKP was pulse-selected over a3-month period (usually 4 hr drug exposure at most once a week,less frequently if cells did not appear to be growing) with 10different drugs (DLKP was pulsed 4, 6, 6, 5, 4, 3, 8, 6, 5 and 8times with VP-16, VCR, TXT, MITOX, 5-FU, MTX, CCNU,BCNU, CPT and CHLO, respectively) to establish its drug-resis-tant variants. The concentrations of incubation for VP-16, VCR,TXT, MITOX, 5-FU, MTX, CCNU, BCNU, CPT and CHLO were200 ng/ml, 4 ng/ml, 2 ng/ml, 60 ng/ml, 10 �g/ml, 50 �g/ml, 10�g/ml, 6 �g/ml, 1 �g/ml and 10 �g/ml, respectively. The humanovarian 2008 cell line transfected with cDNA for MRP3 wasobtained from the laboratory of Dr P. Borst (The NetherlandsCancer Institute, Amsterdam) and cultured in MEM supplementedwith 5% FCS, 1% L-glutamine, 1% sodium pyruvate and 1%nonessential amino acid. All cell lines were free from Mycoplasmaas tested with the indirect Hoechst DNA staining method.

AntibodiesMDR-1/P-gp, MRP1, MRP2 (cMOAT) and MRP3 were de-

tected by Western blotting with C219, MRPr1, M2III-6 and

M3II-9 mouse MAbs purchased from Alexis (Nottingham, UK),respectively. BCRP was detected by immunocytochemistry withanti-BCRP MAb (Alexis). E-cadherin MAb was purchased fromR&D Systems (Abingdon, UK). Expression of cytokeratin-18 andvimentin was investigated by immunofluorescence with anticyto-keratin-18 MAb from Boehringer-Mannheim (Mannheim, Ger-many) and antivimentin MAb from Dako (Carpinteria, CA).MMP-13 was detected by Western blotting with polyclonal rabbitantihuman antibody, which was a gift from Dr M. Morgan (RoyalCollege of Surgeons, Dublin, Ireland).

Cytotoxicity assays

Toxicity to doxorubicin (Adriamycin), TXT, VCR, MTX andCPT in the DLKP cell line and its drug-resistant variants wasdetermined by the acid phosphatase method, which measures vi-able cell number after 5–7 days of continuous exposure to drugs,as previously described.14 Toxicity assays were carried out at leasttwice for each cell line and for each drug; each assay contained 8replicate wells for each drug concentration. Toxicity to MMPinhibitors was investigated using the same methods. To mimic theconditions of invasion assays, 1 � 105 cells were seeded in eachwell with 48 hr exposure to MMP inhibitors. IC50 concentrationsand fold resistance (ratio between IC50 values) were calculated foreach drug and each DLKP variant.

Western blotting

Western blotting was performed according to the method ofMoran et al.15

Immunocytochemistry

Immunocytochemistry was performed using the ABC method asdescribed by Moran et al.15

TABLE I – PRIMERS USED FOR RT-PCR ANALYSIS OF GENE EXPRESSION

Gene Primers (5�-3�) Annealingtemperature (°C)

Productsize (bp)

mdr-1 gttcaaacttctgctcctga 54 157cccatcattgcaatagcagg

mrp1 gtacattaacatgatctggtc 54 202cgttcatcagcttgatccgat

mrp2 ctgcctcttcagaatcttag 54 241cccaagttgcaggctggcc

mrp3 gatacgctcgccacagtcc 62 262cagttggccgtgatgtggctg

mrp4 ccattgaagatcttcctgg 42 239ggtgttcaatctgtgtgc

mrp5 ggataacttctcagtggg 49 381ggaatggcaatgctctaaag

bcrp agacttatgttccacgggcc 63 1113caaggccacgtgattcttcc

E-cadherin agccatgggcccttggag 55 653ccagaggctctgtcaccttc

mmp-1 tcgctgggagcaaacacatc 47 337ttcatgagccgcaacacgat

mmp-2 ctgagatctgcaaacaggacattgt 57 481ccgccctgcaggtccacgacggcat

mmp-3 ctttccagggattgactc 47 324cggccagtaaaaatacaa

mmp-7 tgtatccaacctatggaaatg 47 341cgcgcatctacagttatttac

mmp-9 atgagttcggccacgcgctgggctt 65 330gactcaaaggcacagtagtggccgt

mmp-13 gtggtgtgggaagtatcatca 51 330gcatctggagtaaccgtattg

�-actin gaaatcgtgcgtgacattaaggagaagct AR 383tcaggaggagcaatgatcttga

�-actin tggacatccgcaaagacctgtac AR 142tcaggaggagcaatgatcttga

AR, as relevant for primers with which �-actin primers were included as endogenous control.

485ENHANCED INVASIVENESS AND DRUG RESISTANCE

IMMUNOFLUORESCENCE

Cytokeratin-18 and vimentin were detected by immunofluores-cence, as outlined by Moran et al.15

RT-PCRTotal RNA from pelleted cells was extracted with TRI-reagent

(Sigma) according to the manufacturer’s instructions. The mRNAsinvestigated included those coding for MDR-1/P-gp, MRP1,MRP2 (cMOAT), MRP3, MRP4, MRP5, BCRP, E-cadherin,MMP-1, MMP-2, MMP-3, MMP-7, MMP-9 and MMP-13. �-Ac-tin was amplified in all cases as endogenous control. Forward andreverse primer sequences are listed in Table I. Plasmids containingrelevant cDNAs were included as positive controls. RNA andcDNA were replaced with water as a negative control. Plasmidsthat served as positive controls for MMP-1 and MMP-13 were akind gift from Prof. L. Matrisian (Vanderbilt University, Nash-ville, TN). The size of the resulting PCR products was determinedby comparison with the �174 DNA/HaeIII marker (Sigma). Twoseparate RNA extracts were used to provide duplicate RT-PCRexperiments. These experiments were repeated at least twice.

Invasion assaysInvasion assays were performed using a modification of the

method described by Albini et al.16 ECM, a reconstituted basementmembrane (Sigma), was diluted to 1 mg/ml in serum-free medium.One hundred microlitres of 1 mg/ml ECM were placed into eachinsert (Falcon, Oxnard, CA; 8.0 �m pore size), which were placedin wells of a 24-well plate (Costar, Cambridge, MA). The insertsand plate were incubated overnight at 4°C. The following day,inserts were washed with serum-free medium; then, 100 �l of the1 � 106 cells/ml suspension were added to each insert, and 250 �lof medium containing 5% FCS were added to the well underneaththe insert. Cells were incubated at 37°C for 48 hr. After this timeperiod, the inner side of the insert was wiped with a wet swab toremove the cells and the outer side of the insert was gently rinsedwith PBS, stained with 0.25% crystal violet for 10 min, rinsedagain and then allowed to dry. Inserts were then viewed under amicroscope and the cells counted. Values were expressed as meansand SDs of 30 fields (�400, 10 fields/experiment repeated 3times). Statistical significance was determined using the Microsoft(Redmond, WA) Excel Program by 2-tailed Student’s t-test. Thet-tests were carried out on all triplicate values; p � 0.001 wasconsidered significant.

All 3 MMP inhibitors were purchased from Calbiochem. Ac-cording to the manufacturer’s description, MMP inhibitor I (Cal-biochem 444250) inhibits MMPs 1, 3, 8 and 9; MMP inhibitor III(Calbiochem 444264) inhibits MMPs 1, 2, 3, 7 and 13; andMMP-2 inhibitor I (Calbiochem 444244) inhibits MMP-2. Theprocedure for MMP inhibition assays was identical to that used forinvasion assays with the exception that 200 �g/ml MMP inhibitorI, 10 �g/ml MMP inhibitor III or 10 �g/ml MMP-2 inhibitor Iwere added both into the cell suspension in the insert and to themedium in the well underneath the insert before cells were incu-bated at 37°C for 48 hr.

Motility assaysThe procedure for motility assays was identical to that for

invasion assays, with the exception that the inserts were not coatedwith ECM.17

Gelatin zymographyZymography was used to assess the level of proteolytic activity

of different proteases. Gelatin is a substrate for MMPs, serineproteases and cysteine proteases. The protein concentration of theserum-free medium in which cells were cultured was determinedby the BCA protein assay (Pierce, Rockford, IL), and 20 �g ofsupernatant protein were applied to nonreduced SDS-PAGE usinga 10% gel containing 0.1% gelatin. After electrophoresis, gelswere soaked in 2.5% Triton X-100 at room temperature with gentleshaking for 30 min and incubated in substrate buffer [50 mM

TA

BL

EII

A–

IC50

(ng/

ml)

AN

DFO

LD

RE

SIST

AN

CE

(IN

DIC

AT

ED

INPA

RE

NT

HE

SES)

OF

TH

ED

LK

PD

RU

G-R

ESI

STA

NT

VA

RIA

NT

ST

OC

OR

RE

SPO

ND

ING

SEL

EC

TIN

GA

GE

NT

CO

MPA

RE

DT

OT

HE

PAR

EN

TA

LD

LK

PC

EL

LL

INE

Cel

llin

eV

P-16

VC

RT

XT

MIT

OX

5-FU

MT

XC

CN

UB

CN

UC

PTC

HL

O

DL

KP

114

�31

1.3

�0.

21.

1�

0.07

1.3

�0.

1411

9�

3015

30�

5628

00�

1100

2000

�28

329

8�

4651

0�

85D

rug-

resi

stan

tva

rian

t21

5�

5(1

.9)

2.5

�0.

8(1

.9)

40�

3.9

(36)

7.6

�1.

5(5

.8)

253

�46

(2.1

)22

50�

71(1

.5)

8200

�30

0(2

.9)

3180

�31

1(1

.6)

480

�14

(1.6

)11

50�

212

(2.3

)

486 LIANG ET AL.

TRIS-HCl (pH 8.0), 5 mM CaCl2] at 37°C for 18–24 hr. Gels werethen stained with 2.5 mg/ml Coomassie blue for 2 hr and destainedin a mixture of acetic acid:isopropanol:distilled water (1:3:6) untilclear bands were visible. Gelatinase activity was determined asdistinct, clear bands. To determine the specific inhibition of MMPsby different MMP inhibitors after electrophoresis, gels weresoaked in 2.5% Triton X-100 with the same concentration of MMPinhibitor as mentioned above in invasion assays at room temper-ature, with gentle shaking for 30 min, followed by incubation insubstrate buffer [50 mM TRIS-HCl (pH 8.0), 5mM CaCl2], whichcontained the same concentration of MMP inhibitors at 37°C for18–24 hr. Gels were then stained and destained as describedabove. Each zymograph was done 3 times.

RESULTS

Cross-resistance profilesThe sensitivity of the DLKP drug-resistant variants to a range of

cancer chemotherapeutic agents including their own selectingagent was determined (Table IIA,B). In general, cells showedresistance to their own selecting agents and other drugs, althoughresistance was often at a low level. DLKP-TXT exhibited muchhigher levels of resistance to doxorubicin, VCR, TXT and MTXcompared to other variants. Although DLKP-MITOX cells devel-oped resistance to their selecting agent, MITOX, they were notresistant to doxorubicin, TXT or CPT and exhibited only modestlevels of resistance to VCR and MTX. DLKP-CPT and DLKP-CHLO cells exhibited low levels of resistance to CPT; DLKP-VP16, DLKP-VCR, DLKP-TXT, DLKP-MITOX, DLKP-5FU andDLKP-MTX became sensitive to cisplatin. DLKP-TXT, DLKP-VP16 and DLKP-VCR cells were highly resistant to MTX.

Detection of MDR markers by RT-PCRCompared to the DLKP parental cell line, RT-PCR (Fig. 1)

indicates that Mdr-1/P-gp mRNA was greatly increased in theDLKP-TXT cell line and that BCRP was increased in DLKP-MITOX cells; this could partially explain the relatively higherresistance to TXT and MTX, respectively, in these 2 cell lines. TheDLKP parental line expressed mRNA encoding MRP1, trace ofmrp3 and mrp5 but not mrp2 or mrp4. No apparent difference inthe expression of gene transcripts for MRPs 1–5 was observedbetween DLKP and its drug-resistant variants.

Western blot analysis of MDR markersWestern blot analysis was carried out to detect expression of a

range of well-characterised MDR markers, including MDR-1/P-gp, MRP1, MRP2 and MRP3. Compared to the DLKP parental cellline, all drug-selected DLKP variants overexpressed MDR-1/P-gp(Fig. 2), especially DLKP-TXT cells, which also overexpressedMRP1. MRP2 was not expressed in any of the DLKP variants.Low levels of MRP3 expression were observed in DLKP and someof the drug-resistant variants (DLKP-VP16, DLKP-TXT andDLKP-MITOX).

Immunocytochemical detection of BCRPThe presence of BCRP, also called MXR, was investigated in

the DLKP cell line and its drug-resistant variants by immunocy-tochemistry as no antibody suitable for Western blot could besourced. Figure 3 illustrates the medium-level brown staining inDLKP-MITOX cells, indicative of BCRP expression; BCRP ex-pression was not detected in the DLKP parental cells (Fig. 3) orany of the other variants analysed (data not shown).

Immunofluorescence of cytokeratin-18 and vimentinCytokeratin-18 has been correlated to drug resistance in

epithelial cells,18 and coexpression of cytokeratin-18 and vi-mentin leads to augmentation of invasion in vitro in somecases.19 Based on the observation that DLKP cells expressvimentin but not cytokeratin-18,20 immunofluorescence analy-sis was carried out to investigate any changes in cytokeratin-18or vimentin levels. No changes were observed in the DLKPvariants (data not shown).

Detection of E-cadherin by western blot and RT-PCRIn light of the important role played by the cell adhesion

molecule E-cadherin in some cases of cell invasion in epithelialcells, E-cadherin expression was investigated by RT-PCR (Fig. 1)and Western blot (Fig. 2). Levels of E-cadherin mRNA expressiondid not differ greatly between DLKP and its variants (Fig. 1); noneof these cells expressed E-cadherin protein (Fig. 2).

Invasion assaysInvasion assays (Fig. 4a) indicated that, compared to the DLKP

line, which was poorly invasive in vitro, DLKP-MITOX, DLKP-5-FU, DLKP-MTX, DLKP-BCNU, DLKP-CPT and DLKP-CHLO cells were invasive (p � 0.001), while the differencebetween the cell invasiveness of DLKP, DLKP-VP16, DLKP-VCR, DLKP-TXT and DLKP-CCNU remained insignificant. Thedifferent responses to BCNU and CCNU were surprising. As anexample, representative results from invasion assays for 4 cellvariants are illustrated (Fig. 4b).

Motility assaysResults of motility assays showed trends similar to those of

invasion assays; as is usual, the total numbers of cells travellingthrough the pores was greater than for the corresponding invasionassays, but the rank order motility was the same as that forinvasion (data not shown).

Gelatin zymographyGelatin zymography (Fig. 5) allows assessment of the expres-

sion and activity of MMPs. DLKP and its pulse-selected variantssecreted MMP-2 and MMP-9 (HT-1080 was included as a positivecontrol for MMP-2 and MMP-9, both of which have unique m.w.among the reported gelatin zymography–detectable proteases). Athird band, of approximately 50 kDa, was detected (a few MMPs,e.g., MMPs 1, 3, 11 and 13, are approx. 50 kDa). The doublet for

TABLE IIB – PROFILES OF CROSS-RESISTANCE TO DOXORUBICIN CPT, TXT, VCR AND MTX IN THE DLKP CELL LINE AND ITS10 DRUG-SELECTED VARIANTS

IC50 (ng/ml) Doxorubicin Fold (n) CPT Fold (n) TXT Fold (n) VCR Fold (n) MTX Fold (n)

DLKP 15.3 � 1.04 1 (4) 298 � 46 1 (5) 1.1 � 0.07 1 (3) 1.3 � 0.2 1 (3) 1530 � 56 1 (4)DLKP-VP16 30 � 1.4 2 (2) 140 � 14 0.47 (3) 1.4 � 0.1 1.3 (2) 3.7 � 0.2 2.8 (2) 14120 � 441 9.2 (3)DLKP-VCR 26 � 1.8 1.7 (3) 115 � 24 0.39 (4) 1.2 � 0.1 1.1 (2) 2.5 � 0.8 1.9 (3) 6798 � 116 4.4 (3)DLKP-TXT 505 � 120 33 (3) 154 � 19 0.52 (4) 40 � 3.9 36 (3) 340 � 28 262 (3) 33970 � 870 22.2 (4)DLKP-MITOX 16 � 3.5 1.01 (2) 123 � 33 0.4 (3) 1.1 � 0.05 1 (3) 2.13 � 0.33 1.6 (3) 2150 � 71 1.4 (3)DLKP-5-FU 18 � 0.5 1.2 (2) 168 � 11 0.56 (4) 1.43 � 0.1 1.3 (3) 2.8 � 0.14 2.2 (2) 2125 � 106 1.39 (3)DLKP-MTX 18 � 5.6 1.2 (2) 168 � 18 0.56 (2) 1.3 � 0.08 1.2 (3) 3 � 0.4 2.3 (3) 2250 � 71 1.5 (3)DLKP-CCNU 51 � 6.4 3.3 (3) 316 � 21 1.06 (3) 1.3 � 0.1 1.2 (2) 3.1 � 0.14 2.4 (2) 2480 � 170 1.6 (3)DLKP-BCNU 23 � 6.4 1.5 (2) 325 � 35 1.09 (3) 1.1 � 0.04 1 (3) 2.9 � 0.5 2.2 (2) 2070 � 184 1.35 (3)DLKP-CPT 32 � 8 2.1 (2) 480 � 14 1.6 (4) 1.14 � 0.2 1.04 (3) 2.6 � 0.4 2 (3) 2107 � 414 1.38 (3)DLKP-CHLO 18 � 2.8 1.2 (2) 500 � 52 1.7 (4) 1.1 � 0.02 1 (2) 3.3 � 0.3 2.5 (3) 2435 � 374 1.59 (3)

n value indicates the number of times each toxicity assay done.

487ENHANCED INVASIVENESS AND DRUG RESISTANCE

each MMP band is believed to represent proenzyme and activeprotease. DLKP-MITOX, the most invasive of the cell lines,showed a strong total proteolysis level for all 3 MMP enzymes.After gels were incubated with EDTA, a chelating agent that canbind the zinc and calcium required for activation of MMPs, all

bands disappeared (data not shown), supporting the assumptionthat these bands are MMPs.

Detection of MMPs by RT-PCR and Western blotTo investigate the MMP family members represented by the

approximately 50 kDa band on the zymogram, Western blot anal-ysis was performed. Using an anti-MMP-13 antibody, Westernblot analysis (Fig. 6) suggested that MMP-13 is a significantcomponent of the third band shown on gelatin zymography. RT-PCR (Fig. 7) showed weak mRNA expression of MMP-2 andMMP-9 in the DLKP line and its variants. The highest levels ofMMP-2 mRNA were detected in DLKP-VP16, DLKP-MITOXand DLKP-MTX, while trace amounts were seen in other variants.MMP-9 mRNA was detected in DLKP-MITOX, DLKP-MTX andDLKP-BCNU cells, and lower levels were seen in other variants.Strong MMP-13 bands were observed in all of the cell linesanalysed. MMP-1, MMP-3 and MMP-7 mRNAs were undetect-able by RT-PCR (data not shown).

MMP inhibition assaysWhen DLKP and its drug-resistant variants were incubated in

vitro with MMP inhibitor I, which inhibits MMP-9 [inhibition canbe seen in Fig. 8a(i)], cell invasion was slightly decreased com-pared to the results of invasion assays without MMP inhibitors

FIGURE 1 – RT-PCR of MDR markers and E-cadherin in the DLKPcell line and its drug-resistant variants. �-actin was amplified asendogenous control. The first lane is marker; �, positive control(relevant cloned cDNA); –, negative (H2O) control.

FIGURE 2 – Western blot of MDR-1/P-gp, MRP1, MRP2 (cMOAT),MRP3 and E-cadherin in the DLKP cell line and its drug-resistantvariants. Lysates from cell lines known to be positive for these proteinsare included as positive (�) controls, where possible.

FIGURE 3 – Immunocytochemical staining for BCRP in the DLKPparental cell line and its MITOX-selected variant (DLKP-MITOX).

488 LIANG ET AL.

[Fig. 8b(i)]. When cells were incubated with MMP-2 inhibitor I,which inhibits the function of both MMP-2 and MMP-9 [Fig.8a(ii)], cell invasion was decreased to almost nothing [Fig. 8b(ii)].When cells were incubated with MMP inhibitor III, which inhibitsMMP-2 and MMP-13 [Fig. 8a(iii)], the level of cell invasion wasintermediate between that with MMP-9 inhibitor and that withMMP-2 and MMP-9 inhibitor [Fig. 8b(iii)]. These results suggestthat MMP-2 may be particularly important in invasiveness in thissystem and that it may act synergistically with MMP-9 and, to alesser extent, with MMP-13; more specific inhibitors for each ofthe MMPs being analysed are needed to confirm this conclusion.

The concentrations of all 3 MMP inhibitors used were not toxic tothe cells (Fig. 9).

DISCUSSION

Although chemotherapy constitutes the principal therapeutictool to treat unresectable or disseminated tumours and there hasbeen a significant impact on survival in some malignancies, it isassociated with serious side effects, not all of which have beenfully characterised. We previously reported that melphalan selec-tion induced a human nasal carcinoma cell line, RPMI-2650, tobecome more invasive in vitro, whereas paclitaxel could not.12

Here, we expanded that study using a different cell line (in case theinitial results had any element of cell specificity) to investigate 10other commonly used chemotherapeutic drugs (to investigate ifsuch effects are specific to melphalan or may occur in response toother drugs).

Exposure to each of the 10 drugs induced drug resistance; thelower levels of resistance resulting from the limited numbers ofdrug pulses may more closely resemble the resistance experiencedby cancer cells during treatment in vivo than do high-level variantsselected by long-term continuous exposure. Expression of genescoding for efflux pumps was examined, even in the case ofexposure to drugs not carried by these pumps, since exposure to awide variety of drugs can lead to a range of cellular responsesresulting in cross-resistance.

TXT appeared to induce the highest level of MDR, whichappeared to be associated with overexpression of MDR-1/P-gp andMRP1 in this variant, as shown by Western blot and RT-PCR. TheMDR acquired by DLKP-MITOX cells could be, at least partially,explained by the expression of BCRP and MDR-1/P-gp. HumanBCRP was first cloned from the doxorubicin-selected and highlyresistant cell line MCF-7/AdrVp.5 However, BCRP transfectantsand mitoxantrone- or topotecan-selected cell lines that overex-pressed BCRP showed quite modest resistance to anthracyclinessuch as doxorubicin.21 Similarly, DLKP-MITOX cells showedvery little cross-resistance to doxorubicin. Cells selected withVP16, VCR and TXT (DLKP-VP16, DLKP-VCR and DLKP-TXT) were greatly resistant to MTX compared to other variants.Since MTX is a substrate for both MRP1 and MRP322 and none ofthe DLKP variants expressed MRP3, it is likely that MRP1 confersthe resistance to MTX in DLKP-VP16, DLKP-VCR and DLKP-TXT cells. For MDR-1/P-gp and MRP1, the results of RT-PCRand Western blotting do not concur in all cases, suggesting a roleof posttranscriptional regulation in expression of these proteins.For example, MDR-1/P-gp protein was overexpressed in all of theDLKP drug-resistant variants, but mdr-1 mRNA was not stronglyexpressed except in DLKP-TXT cells. Similarly, MRP1 protein,but not mRNA, was overexpressed in DLKP-TXT. The increase indoxorubicin resistance in several of the variants was not accom-panied by obvious increases in MDR-1/Pg-p and MRP-1 proteinlevels; it may reflect drug-induced alterations in topoisomerase orpump activity, e.g., due to altered phosphorylation or to othermechanisms as yet unidentified.

Since MDR-1/P-gp and MRP are important markers for drugresistance, studies have been carried out to investigate whetherthey are also markers for invasion and metastasis. Weinstein etal.23 reported an association in colon carcinomas between MDR-1/P-gp expression and enhancement of local tumour aggressive-ness. In all but one of the 95 clinical specimens of primary colonadenocarcinomas, invading carcinoma cells were present at theleading edge of the tumour. Cases were grouped on the basis of thepresence (group 1, 47 cases) or absence (group 2, 48 cases) ofMDR-1/P-gp-positive invasive carcinoma cells. There was a sig-nificantly greater incidence of vessel invasion and lymph nodemetastases in group 1 cases, indicating that MDR-1/P-gp-positiveinvasive colon cancer cells may have an increased potential fordissemination. Schneider et al.24 studied the expression of MDR-1/P-gp in specimens from 52 patients with advanced breast cancer

FIGURE 4 – Invasion assays of the DLKP cell line and its drug-resistant variants. (a) Cell invasiveness was assessed by the number ofcells that invaded through the ECM and attached to the outer side ofcell culture inserts. Compared to the DLKP parental cell line, p valuesfor the DLKP-MITOX, DLKP-5FU, DLKP-MTX, DLKP-BCNU,DLKP-CPT and DLKP-CHLO are � 0.001, indicating that theirinvasiveness is significantly greater than that of DLKP; whereas the pvalues for DLKP-VP16, DLKP-VCR, DLKP-TXT and DLKP-CCNUare 0.001, indicating that they do not differ significantly from thepoorly invasive DLKP cells. p values for DLKP-VP16, DLKP-VCR,DLKP-TXT, DLKP-MITOX, DLKP-5-FU, DLKP-MTX, DLKP-CCNU, DLKP-BCNU, DLKP-CPT and DLKP-CHLO are 0.255,0.0088, 0.057, 5.11 � 10–7, 3.96 � 10–7, 7.89 � 10–6, 0.0045, 9.49 �10–8, 1.82 � 10–5 and 4.88 � 10–6, respectively. (b) Representativemicroscopic results for DLKP and its variants (�100): (i) DLKP, (ii)DLKP-MITOX, (iii) DLKP-TXT and (iv) DLKP-BCNU.

489ENHANCED INVASIVENESS AND DRUG RESISTANCE

before and after induction chemotherapy followed by rescue mas-tectomy. They showed that expression of MDR-1/P-gp in locallyadvanced breast cancer predicted axillary lymph node metastasesafter induction chemotherapy. An interesting finding was that, of 9cases that were MDR-1/P-gp-positive before chemotherapy, 7were still positive after treatment. These 7 patients had positivelymph nodes. In contrast, the remaining 2 MDR-1/P-gp-negativepatients had negative nodes postchemotherapy. In the case ofMRP1, higher staining intensity was observed in lymph nodemetastases compared to the corresponding primary tumours inbreast cancer patients.25

In our study, the results of invasion assays showed that MITOX,BCNU, CHLO, 5-FU, MTX and CPT exposure made DLKP cellssignificantly more invasive in vitro, whereas VP-16, VCR, TXTand CCNU did not (indicating that, in some cases, in vitro inva-

siveness is markedly enhanced in sublines generated with non-MDR drugs). Motility, like invasion, is a critical step in tumourmetastasis. In our study, all cell variants tested were more motilethan the parental cell line, DLKP. As for invasiveness, MITOX,BCNU and CHLO induced the greatest levels of motility. (Furtherstudies including anthracyclines will be useful and may generatefurther interesting results). Tofilon et al.26 reported the enhance-ment of spontaneous metastasis of a hepatocarcinoma growingintramuscularly following administration of a single dose ofBCNU. Teicher et al.27 established 4 alkylating agent-resistantvariants of a murine mammary tumour cell line, EMT-6, in vivo byrepeated treatment of tumour-bearing mice with CPT, carboplatin,cyclophosphamide or thiotepa. These drug-resistant cell lines (par-ticularly the CPT-resistant variant) were much more efficient atforming spontaneous lung metastases after s.c. injection of cells incomparison to the drug-sensitive parental line.28,29 Using themethod of Teicher et al.,27 Mitsumoto et al.30 established 2 CPT-resistant mouse epithelial tumour cell lines with enhanced meta-static capabilities associated with higher levels of cell adhesion toECM components, such as fibronectin, laminin, collagen type IVand Matrigel; enhanced MMP-2 activity; and higher levels of cellmotility. Van Putten et al.31 showed that treatment of animalsbearing a transplantable osteosarcoma with cyclophosphamide orCCNU resulted in a significant increase in lung metastases com-pared to controls. Following these initial observations, enhance-ment of metastases was also obtained in other tumour systems withcyclophosphamide32,33 and bleomycin.34,35 De Larco et al.36 de-veloped MCF-7A/F by treating the parental MCF-7 cell line se-quentially with doxorubicin and 5-fluoro-2�-deoxyuridine in vitroand found that the MCF-7A/F cells in vivo had a more rapid initialgrowth phase in situ and gave rise to spontaneous lung metastaseswithin 10 weeks, while MCF-7 cells were not able to form me-tastases.

In clinical studies, Slotman et al.37 found an increased incidenceof hematogenous metastases with induction chemotherapy for pa-tients with advanced head-and-neck squamous carcinomas com-pared to patients receiving only surgery and radiotherapy. Theyalso found an increase in metastases at sites other than the usualpulmonary site. Stefani et al.38 reported that the use of hydroxy-urea in combination with radiotherapy for the treatment of head-and-neck cancer yielded no benefit in terms of response rate orsurvival. Furthermore, the drug treatment in this protocol increasedthe incidence of distant metastases from 8% to 23%.

In our study, we examined MMPs, which are thought to play acritical role in cancer invasion and metastasis. The results of

FIGURE 5 – Gelatin zymography analysis of the DLKP cell line and its drug-resistant variants. HT-1080 is a positive control for MMP-2 andMMP-9 activity.

FIGURE 6 – Western blot analysis of MMP-13 in the DLKP cell line and its drug-resistant variants.

FIGURE 7 – RT-PCR analysis of mmp-2, mmp-9 and mmp-13 in theDLKP cell line and its drug-resistant variants. �-actin was amplified asendogenous control. The first lane is marker; �, positive control(relevant cloned cDNA).

490 LIANG ET AL.

gelatin zymography and Western blot analyses confirmed that theDLKP cell line and its variants secrete MMPs 2, 9 and 13. InDLKP-TXT cells, the activity of MMP-9 and MMP-13 was weakcompared to the other variants. DLKP-MITOX cells showedstrong total proteolytic activity of MMPs 2, 9 and 13, which maycontribute to its highest level of motility and invasiveness in vitro.Although it was demonstrated by zymography that in all cell linesMMP-2 exhibited stronger proteolytic activity than MMPs 9 and13, the results of RT-PCR showed only weak mmp-2 mRNAexpression in DLKP-VP16, DLKP-VCR, DLKP-TXT, DLKP-MITOX, DLKP-MTX, DLKP-BCNU and DLKP-CPT cells. Weakexpression of both mRNA and protein activities was observed forMMP-9. Although MMP-13 bands were not seen in DLKP-TXTand DLKP-CPT by zymography, mRNA for MMP-13 wasstrongly expressed in all of the cell lines, suggesting a role ofposttranscriptional regulation.

Results of the MMP inhibition assays indicate the importance ofMMP-2 in DLKP invasiveness, probably acting synergisticallywith MMP-9 and MMP-13. MMP-2 and MMP-9 degrade collagentype IV, a major component of ECM, whereas MMP-13 mainlydegrades fibrillar collagen and is not expected to contribute greatlyto cell invasion through basement membrane. By studying theexpression of MMP-2 and MMP-9 in 11 NSCLC cell lines and 43NSCLC clinical specimens using PCR and immunohistochemistry,Suzuki et al.39 concluded that MMP-2 plays a more important rolein invasion of NSCLC than MMP-9. Using immunocytochemistry,Bodey et al.40 reported high-intensity staining of MMPs 2, 9 and13 in human lung adenocarcinomas, which is consistent with ourresults obtained from human lung squamous carcinoma. In lungsilicosis, lung inflammation induced by exposure to silica, MMP-2,MMP-9 and MMP-13 was highlighted, suggesting that these 3MMPs play an important part in ECM remodelling.41 However,

FIGURE 8 – (a) Zymographic re-sults indicate that incubation ofDLKP and its drug-selected vari-ants with (i) MMP inhibitor I, (ii)MMP-2 inhibitor I and (iii) MMPinhibitor III leads to less intensebands indicative of (i) MMP-9, (ii)MMP-2/-9 and (iii) MMP-2/-13 ac-tivity compared to those in Figure5. (b) This decreased activity cor-responds with decreased in vitroinvasiveness of cells [shown in (i),(ii) and (iii), respectively].

FIGURE 9 – Cytotoxicity of the 3inhibitors used in our study. For eachcell line, cell viability in the absenceof any MMP inhibitor was countedas 100%.

491ENHANCED INVASIVENESS AND DRUG RESISTANCE

although MMP-2 is essential in invasion of DLKP variants, italone is not sufficient to cause invasion. DLKP and DLKP-CCNUcells secrete MMPs 2, 9 and 13 but are poorly invasive, suggestingthat other mechanisms determine the drug-induced transition to theinvasive phenotype. Despite the important role of MMP-2 inNSCLC, it is interesting to note that MMP-2 was not found inhuman SCLC,42 which may indicate distinct strategies of invasionbeing used by these 2 types of lung cancer. Future studies inves-tigating a potential role for the histone deacetylase metastasis-associated (MTA1) gene product,43 as well as other metastasis-associated genes (e.g., mst1, nm23, KiSS1, MKK4, galectin-3, etc.)among the DLKP variants, may help to fully elucidate their mech-anism(s) of drug-induced in vitro invasiveness.

In conclusion, the results presented here (summarised in TableIII) provide further support to our earlier findings that exposure tochemotherapeutic drugs can result in not only resistance but alsoinvasive properties but that this does not always occur; thus, acausal link between resistance and invasiveness cannot be assumedbased on these results. However, the cellular model system de-

scribed here may be useful in studying further the time course andmolecular mechanism of induction of invasiveness and the degreeto which it is drug- and dose-specific. It is important to characterisethis effect further and to identify molecular markers that could beused to assess the relevance of this effect in vivo. It is obviouslyimportant to apply such methods to establish the extent to whichinduction of invasiveness might occur in cancer patients duringchemotherapy and to devise strategies (whether by drug schedul-ing or inhibition of the process) to diminish it. An increasedunderstanding of drug resistance, cell invasion and their relation-ship should ultimately lead to better cancer treatment regimens thatavoid induction of the invasive phenotype and identify moleculartargets for circumvention of drug resistance and invasion/metas-tasis simultaneously.

ACKNOWLEDGEMENTS

This work was supported by funding from the Higher EducationAuthority’s PRTLI Cycle 3 program.

REFERENCES

1. Clynes M, ed. Multiple drug resistance in cancer. Amsterdam: Klu-wer, 1998.

2. Woodhouse EC, Chuaqui RF, Liotta LA. General mechanisms ofmetastasis. Cancer 1997;80:1529–37.

3. Sauna ZE, Smith MM, Muller M, Kerr KM, Ambudkar SV. Themechanism of action of multidrug-resistance-linked P-glycoprotein.J Bioenerg Biomembr 2001;33:481–91.

4. Borst P, Evers R, Kool M, Wijnholds J. A family of drug transporters:the multidrug resistance–associated proteins. J Natl Cancer Inst 2000;92:1295–302.

5. Doyle LA, Yang W, Abruzzo LV, Krogmann T, Gao Y, Rishi AK,Ross DD. A multidrug resistance transporter from human MCF-7breast cancer cells. Proc Natl Acad Sci USA 1998;95:15665–70.

6. DeClerck YA, Imren S, Montgomery AM, Mueller BM, Reisfeld RA,Laug WE. Proteases and protease inhibitors in tumor progression.Adv Exp Med Biol 1997;425:89–97.

7. DeClerck YA. Interactions between tumour cells and stromal cells andproteolytic modification of the extracellular matrix by metalloprotein-ases in cancer. Eur J Cancer 2000;36:1258–68.

8. Giannelli G, Antonaci S. Gelatinases and their inhibitors in tumormetastasis: from biological research to medical applications. HistolHistopathol 2002;17:339–45.

9. Brinckerhoff CE, Rutter JL, Benbow U. Interstitial collagenases asmarkers of tumor progression. Clin Cancer Res 2000;6:4823–30.

10. Liang Y, McDonnell S, Clynes M. Examining the relationship be-tween cancer invasion/metastasis and drug resistance. Curr CancerDrug Targets 2002;2:257–77.

11. Osmak A, Niksic D, Brozovic A, Ambriovic Ristov A, Vrhovec I,Skrk J. Drug resistant tumor cells have increased levels of tumormarkers for invasion and metastasis. Anticancer Res 1999;19:3193–8.

12. Liang Y, Meleady P, Cleary I, McDonnell S, Connolly L, Clynes M.Selection with melphalan or paclitaxel (Taxol) yields variants withdifferent patterns of multidrug resistance, integrin expression and invitro invasiveness. Eur J Cancer 2001;37:1041–52.

13. Law E, Gilvarry U, Lynch V, Gregory B, Grant G, Clynes M.Cytogenetic comparison of two poorly differentiated human lungsquamous cell carcinoma lines. Cancer Genet Cytogenet 1992;59:111–8.

14. Heenan M, O’Driscoll L, Cleary I, Connolly L, Clynes M. Isolationfrom a human MDR lung cell line of multiple clonal subpopulationswhich exhibit significantly different drug resistance. Int J Cancer1997;71:907–15.

15. Moran E, Cleary I, Larkin AM, Nic Amhlaoibh R, Masterson A,Scheper RJ, Izquierdo MA, Center M, O’Sullivan F, Clynes M.Co-expression of MDR-associated markers, including P-170, MRPand LRP and cytoskeletal proteins, in 3 resistant variants of the humanovarian carcinoma cell line, OAW42. Eur J Cancer 1997;33:652–60.

16. Albini A, Iwamoto Y, Kleinman HK, Martin GR, Aaronson SA,Kozlowski JM, McEwan RN. A rapid in vitro assay for quantitatingthe invasive potential of tumor cells. Cancer Res 1987;47:3239–45.

17. O’Driscoll L, Linehan R, Liang Y, Joyce H, Oglesby I, Clynes M.Galectin-3 expression alters cell adhesion, motility and invasion in alung cell line (DLKP) cell line, in vitro. Anticancer Res 2002;22:3117–25.

18. Bauman PA, Dalton WS, Anderson JM, Cress AE. Expression ofcytokeratin confers multiple drug resistance. Proc Natl Acad Sci USA1994;91:5311–4.

19. Hendrix MJC, Seftor EA, Chu YW, Trevor KT, Seftor REB. Role ofintermediate filaments in migration, invasion and metastasis. CancerMetastasis Rev 1996;15:507–25.

20. McBride S, Meleady P, Baird A, Dinsdale D, Clynes M. Human lungcarcinoma cell line DLKP contains 3 distinct subpopulations withdifferent growth and attachment properties. Tumor Biol 1998;19:88–103.

21. Schinkel AH, Allen JD, Brinkhuis RF, Jackson SC. Murine breastcancer resistance protein BCRP mediates 2 distinct MDR phenotypes.Proc Am Assoc Cancer Res 2001;42:279.

22. Zeng H, Chen ZS, Belinsky MG, Rea PA, Kruh GD. Transport of

TABLE III – SUMMARY OF KEY RESULTS FROM THIS MULTIPLE CORRELATION STUDY SHOWING FOLDRESISTANCE OF THE DLKP DRUG-RESISTANT VARIANTS TO CORRESPONDING SELECTING AGENTS COMPARED

TO THE PARENTAL DLKP CELL LINE

Cell line Fold resistance toselection agent Invasive MDR-1/P-gp MRP1 MMP-2 MMP-9 MMP-13

DLKP N/A N �� � � �DLKP-VP16 1.9 N � � ��� � ��DLKP-VCR 1.9 N � � ��� � ��DLKP-TXT 36 N ��� ��� � � �DLKP-MITOX 5.8 Y � � �� �� ���DLKP-5FU 2.1 Y � � �� �� ��DLKP-MTX 1.5 Y �� � �� �� ��DLKP-CCNU 2.9 N �� �� �� � ��DLKP-BCNU 1.6 Y �� �� �� � �DLKP-CPT 1.6 Y � �� �� �DLKP-CHLO 2.3 Y �� � �� ��

In vitro invasiveness (N, no; Y, yes) and expression levels of MDR-1/P-gp, MRP1, MMP-2, MMP-9and MMP-13 (scored as , �, �� and ��� to indicate levels). N/A, not applicable.

492 LIANG ET AL.

methotrexate (MTX) and folates by multidrug resistance protein(MRP)3 and MRP1: effect of polyglutamylation on MTX transport.Cancer Res 2001;61:7225–32.

23. Weinstein RS, Jakate SM, Dominguez JM, Lebovitz MD, KoukoulisGK, Kuszak JR, Klusens LF, Grogan TM, Saclarides TJ, RoninsonIB, Coon JS. Relationship of the expression of the multidrug resis-tance gene product (P-glycoprotein) in human colon carcinoma tolocal tumor aggressiveness and lymph node metastasis. Cancer Res1991;51:2720–6.

24. Schneider J, Gonzalez-Roces S, Pollan M, Lucas R, Tejerina A,Martin M, Alba A. Expression of LRP and MDR1 in locally advancedbreast cancer predicts axillary invasion at the time of rescue mastec-tomy after induction chemotherapy. Breast Cancer Res 2001;3:183–91.

25. Zochbauer-Muller S, Filipits M, Rudas M, Brunner R, Krajnik G,Suchomel R, Schmid K, Pirker R. P-glycoprotein and MRP1 expres-sion in axillary lymph node metastases of breast cancer patients.Anticancer Res 2001;21:119–24.

26. Tofilon PJ, Basic I, Milas L. Prediction of in vivo tumor response tochemotherapeutic agents by the in vitro sister chromatid exchangeassay. Cancer Res 1985;45:2025–30.

27. Teicher BA, Herman TS, Holden SA, Wang Y, Pfeffer MR, CrawfordJW, Frei E. Tumor resistance to alkylating agents conferred by mech-anisms operative only in vivo. Science 1990;247:1457–61.

28. Teicher BA, ed. Drug resistance in oncology. New York: MarcelDekker, 1993. 115–26.

29. Kerbel RS, Kobayashi H, Graham CH. Intrinsic or acquired drugresistance and metastasis: are they linked phenotypes? J Cell Biochem1994;56:37–47.

30. Mitsumoto M, Kamura T, Kobayashi H, Sonoda T, Kaku T, NakanoH. Emergence of higher levels of invasive and metastatic properties inthe drug-resistant cancer cell lines after the repeated administration ofcisplatin in tumor-bearing mice. J Cancer Res Clin Oncol 1998;124:607–14.

31. Van Putten LM, Kram LKJ, Van Drerendonck HHC, Smink T, FuzyM. Enhancement by drugs of metastatic lung nodule formation afterintravenous tumour cell injection. Int J Cancer 1975;15:588–95.

32. Carmel RJ, Brown JM. The effect of cyclophosphamide and other

drugs on the incidence of pulmonary metastasis in mice. Cancer Res1977;37:145–51.

33. Moore JV, Dixon B. Metastasis of a transplantable mammary tumourin rats treated with cyclophosphamide and/or irradiation. Br J Cancer1977;36:221–6.

34. Stewart RAN, Hacker MP. Enhancement of B16 melanoma pulmo-nary colonization in mice following bleomycin or hyperoxia. Proc AmAssoc Cancer Res 1984;25:60.

35. Orr FW, Adamson IYR, Young L. Promotion of pulmonary metastasisin mice by bleomycin-induced endothelial injury. Cancer Res 1986;46:891–7.

36. De Larco JE, Wuertz BRK, Manivel JC, Furcht LT. Progression andenhancement of metastatic potential after exposure of tumor cells tochemotherapeutic agents. Cancer Res 2001;61:2857–61.

37. Slotman GJ, Mohit T, Raina S, Swaminathan AP, Ohanian M, RushBF. The incidence of metastases after multimodal therapy for cancerof the head and neck. Cancer 1984;54:2009–14.

38. Stefani S, Eells RW, Abbate J. Hydroxyurea and radiotherapy in headand neck cancer. Radiology 1971;101:391–6.

39. Suzuki M, Iizasa T, Fujisawa T, Baba M, Yamaguchi Y, Kimura H,Suzuki H. Expression of matrix metalloproteinases and tissue inhib-itor of matrix metalloproteinases in non-small-cell lung cancer. Inva-sion Metastasis 1998;18:134–41.

40. Bodey B, Bodey B Jr, Groger AM, Siegel SE, Kaiser HE. Invasionand metastasis: the expression and significance of matrix metallopro-teinases in carcinomas of the lung. In Vivo 2001;15:175–80.

41. Perez-Ramos J, de Lourdes Segura-Valdez M, Vanda B, Selman M,Pardo A. Matrix metalloproteinases 2, 9 and 13, and tissue inhibitorsof metalloproteinases 1 and 2 in experimental lung silicosis. Am JRespir Crit Care Med 1999;160:1274–82.

42. Michael M, Babic B, Khokha R, Tsao M, Ho J, Pintilie M, Leco K,Chamberlain D, Shepherd FA. Expression and prognostic significanceof metalloproteinases and their tissue inhibitors in patients with small-cell lung cancer. J Clin Oncol 1999;17:1802–8.

43. Nicolson GL, Nawa A, Toh Y, Taniguchi S, Nishimori K, MoustafaA. Tumor metastasis-associated human MTA1 gene and its MTA1protein product: role in epithelial cancer cell invasion, proliferationand nuclear regulation. Clin Exp Metastasis 2003;20:19–24.

493ENHANCED INVASIVENESS AND DRUG RESISTANCE