Cancer and Metastasis Reviews 13: 223-233, 1994. © 1994Kluwer Academic Publishers. Printedin the Netherlands.

P-glycoprotein, multidrug resistance and tumor progression Grace Bradley 1"2and Victor Ling1 1 Ontario Cancer Institute, Department o f Medical Biophysics, University o f Toronto; 2 Faculty o f Dentistry, University o f Toronto, Canada Key words: P-glycoprotein, muItidrug resistance, rat liver carcinogenesis Abstract

P-glycoprotein (Pgp) is a plasma membrane protein that was first characterised in multidrug resistant cell lines. The occurrence of Pgp in clinical tumors has been widely studied. Recent investigations have begun to focus on the relationship between Pgp detection in tumors and treatment outcome. In several types of tumors, detection of Pgp correlates with poor response to chemotherapy and shorter survival. P-glycoprotein overexpression often occurs upon relapse from chemotherapy but may also occur at the time of diagnosis. Studies of experimental rat liver carcinogenesis have shown that Pgp expression increases in late stages of carcinogenesis, suggesting that Pgp may be involved in tumor progression. While some of the Pgp isoforms are known to transport hydrophobic chemotherapeutic drugs out of tumor cells, the biologic effects of Pgp overexpression in tumor cells are not fully understood, because the spectrum of substrates for Pgp-mediated transport has not been determined. In the rat liver carcinoma model, strong expression of Pgp is associated with a highly vascular stroma, suggesting that Pgp in tumor cells may affect the connective tissue stroma. The regulation of Pgp appears to be complex, and little is known about how it is up-regulated during carcinogenesis. Further studies of the role of Pgp in malignancy may contribute to our understanding of molecular mechanisms which underlie tumor progression.

Introduction

Malignant tumors are heterogeneous in composition. Successive events are thought to occur during tumor progression, giving rise to subpopulations with increasing potential for malignant behaviour, such as widespread invasion, dissemination to distant sites and development of resistance to chemotherapy. At advanced stages of tumor progression, tumor dissemination and chemotherapy resistance are ultimately responsible for the failure of therapy to control malignant growth. There is clearly a need to understand the molecular determinants of malignant behaviour, so that malignant subpopulations of tumor cells may be identified for specific antitumor therapy at an early stage of tumor growth. One candidate for a molecular determinant of malignancy is P-glycoprotein (Pgp), since it has been linked

to resistance to multiple chemotherapeutic drugs both in vitro and in vivo. P-glycprotein is a transmembrane gtycoprotein which is overexpressed in many mammalian cells lines upon selection for multidrug resistance [1]. These cell lines are cross-resistant to many structurally and functionally unrelated drugs, due to reduced accumulation of the drugs involved. Transfection experiments using full-length cDNAs demonstrated that Pgp can confer multidrug resistance onto drug sensitive cells, and sequence analyses of cDNA clones indicated that Pgp is a tandemly-duplicated molecule with two sets of membrane spanning domains and two ATP~binding sites. Thus, Pgp may act as an ATP-dependent drug efflux pump which reduces cellular accumulation of chemotherapeutic drugs such as the vinca alkaloids, doxorub-

224 icin, daunomycin, the epipodophyllotoxins and taxol [2]. Studies of tumor tissue have demonstrated that Pgp is often overexpressed in vivo in tumors of various histologic types. Pgp overexpression may be detected after exposure of tumors to chemotherapy, or at the time of diagnosis and prior to chemotherapy. Detection of Pgp at any time during the clinical course of malignant disease appears to be prognostic of refractory disease and poor response to chemotherapy [3]. The expression of Pgp by malignant cells, and the ability of these cells to transport cytotoxic drugs and perhaps also specific metabolites may constitute one of the molecular events during tumor progression. Pgp is a membrane transporter of broad specificity. In vitro studies have shown that the spectrum of substrates for Pgp encompasses not only chemotherapeutic drugs but also some of the chemosensitizers useful in reversing multidrug resistance, fluorescent dyes and steroid hormones [4-6]. Relatively little is known about the substrates for Pgp transport in vivo, but some speculations may be made, based on the pattern of expression of the Pgp isoforms in normal tissues and on the likely functions of closely related membrane transporters. Pgp is encoded by a small multigene family. Rodent cells have three Pgp genes (encoding class I, II and III Pgp) and human cells have two (encoding class I and III Pgp) [7]. The class I and II Pgp genes confer multidrug resistance when transfected into drug sensitive cells while the class III Pgp gene has not been associated with drug resistance. The three classes of Pgp are predominantly expressed in different sets of normal tissues, supporting the idea of functional differences among them (Table 1) [8-13]. In general, the cellular localization of the Pgp isoforms is consistent with physiologic roles as membrane transporters, but the nature of the physiologic substrates for the Pgp isoforms is largely unknown. In the liver, Pgp is localized to the canalicular membrane of hepatocytes. The predominant isoform is class III, although class I Pgp is also detectable. Recently it has been shown that mice lacking in class 1II Pgp expression are unable to secrete phospholipid into bile, suggesting that class III Pgp may be a phospholipid transporter in the hepato-

cyte canalicular membrane [14]. In the intestinal lining epithelium and perhaps also in the adrenal cortex, Pgp expression appears to be'highest in the non-dividing, differentiated cells and lowest in the dividing, less-differentiated cells. However, in the bone marrow, Pgp expression is highest in the progenitor cells [12, 13]. This suggests that Pgp-mediated transport is part of the phenotype of several distinct cell types, and not always associated with terminally differentiated cells. The pgp gene family has been shown to belong to a superfamily of genes which encode ATP-binding membrane transport systems in phylogenetically diverse species [15]. Besides the pgp genes, other mammalian genes that belong to the ABC (ATPBinding Cassette) superfamily include the CFTR (cystic fibrosis transmembrane regulator) gene, the TAP (transporter of antigenic peptides) gene and, most recently, the MRP (multidrug resistance-associated protein) gene. The CFTR gene appears to encode a cAMP-regulated chloride channel [16], while the TAP gene product may be involved in Table 1. Expression of the P-glycoprotein (Pgp) isoforms in mammalian tissues.

Tissue and cell type

Predominant PgP isoform ~

Intestinal lining epithelium Endothelial cellsb Bone marrow progenitor cells Peripheral blood lymphocytes and natural killer cells Adrenal cortex Lining epithelium of gravid uterus Hepatocytes Cardiac muscle Striated muscle d

I I I I IY II c Ill III III

Data compiled from references 8 to 13. b Endothelial cell expression of Pgp varies among different tissues. Pgp has been detected in endothelial cells of brain, testis, skin, ovary, oviduct, uterus, vagina. c In human adrenal cortex and gravid uterus, class I Pgp is the predominant isoform. Human cells only have two Pgp genes, class I and class III. Human class I Pgp is closely related to rodent class I and II Pgps while human class III Pgp corresponds to rodent class III Pgp. d The class III Pgp isoform is expressed only in a subset of muscle fibers. a

225 membrane transport of antigenic peptides during the assembly of class I MHC molecules [17]. The MRP gene is associated with a multidrug resistance phenotype in cultured cells, which is different from Pgp-mediated multidrug resistance although there are striking similarities. The MRP gene has been detected in normal tissues but its physiologic role has not been determined [18]. Based on what is currently known about the function of various ABC superfamily members, one may speculate that the Pgp isoforms normally transport specific metabolites across cell membranes, but in addition, at least the class I and II isoforms can also transport an array of chemotherapeutic drugs out of cells and consequently cause reduced drug accumulation and multidrug resistance. In the following sections, we review some of the clinical data which suggest an association between Pgp overexpression and phenotypes associated with malignancy, and discuss the use of model systems, both in vivo and in vitro, to elucidate the possible role of Pgp in malignant progression.

P-glycoprotein in clinical malignancies Detection of P-glycoprotein in tumor samples

The in vitro studies of multidrug resistance suggest that chemotherapy of clinical malignancies may select for multidrug resistant, Pgp-positive subpopulations of malignant cells, resulting in emergence of these subpopulations and relapse of malignant disease. Thus, cancers which have failed chemotherapy are predicted to show higher expression of Pgp compared to cancers analyzed prior to chemotherapy. Furthermore, since small subpopulations of Pgp-positive cells within a tumor may be significant in determining tumor response to chemotherapy, assays for Pgp at the single-cell level, such as immunohistochemical studies, may allow more sensitive analyses of tumor samples than bulk assays such as Northern blot or slot blot. Studies of several types of malignancies, both horizontal, non-sequential studies and longitudinal studies of sequential tumor samples have indeed shown evidence of increased expression of Pgp after chemotherapy. In a

study of patients with lymphoma, i of 42 newly-diagnosed and untreated patients (2%) had detectable Pgp in the tumor cells, while 7 of 11 treated patients (64%) whose lymphoma were clinically refractory to chemotherapy had Pgp-positive malignant cells [19]. Similarly, less than 5% of patients with newly-diagnosed multiple myeloma showed Pgp in the tumor cells, compared to more than 75% of the patients who relapsed following chemotherapy that included vincristine and doxorubicin [20]. Immunohistochemical analysis of sequential tumor samples provided valuable insight into changes of Pgp expression over time in a malignant tumor which persisted in growth or relapsed following multiple regimens of chemotherapy. In a study of 30 children with soft tissue sarcoma, 5 patients had tumors which were initially Pgp-negative but became Pgp-positive after several months of chemotherapy. In each case, the appearance of Pgp in the tumor was prognostic of failure of further chemotherapy [21]. Similarly, in a study of children with neuroblastoma, among 44 patients with non-localized tumors, 6 had tumors which were Pgp-negative at diagnosis but became Pgp-positive at relapse [22]. Both of these studies analysed multiple sequential tumor specimens from several patients and revealed a significant pattern of Pgp expression, that tumors which became Pgp-positive during the course of treatment always remained Pgp-positive and often showed increasing expression of Pgp with disease progression. A study of 11 adult patients with chronic myelogenous leukemia in blast crisis showed three patients whose leukemic cells were Pgp-negative at the time of complete remission but became Pgp-positive at relapse; these patients did not respond to subsequent chemotherapy [23]. Thus there are a small number of well-documented cases of malignant tumors in which Pgp overexpression and multidrug resistance progressively developed during chemotherapy. Several studies have now demonstrated that moderate to high levels of Pgp may be expressed by tumor cells of various histologic types, where there has been no prior chemotherapy. For example, Pgp has been detected at the time of diagnosis of acute lymphoblastic leukemia [24], acute nonlymphoblastic leukemia [25], lymphoma [26], adult sarco-

226 ma [27], colon carcinoma [28], childhood sarcoma [21] and neuroblastoma [22]. Some of these tumors arise from tissues which normally express Pgp, for example, colon carcinoma and adrenal cortical carcinoma, while others do not, for example, sarcoma and neuroblastoma. For tumors arising from tissues which normally express detectable amounts of Pgp, such as colon carcinoma, it has been suggested that Pgp expression by the malignant cells is related to histologic differentiation towards the parent cell type [29]. This idea is contradicted by the findings of a detailed immunohistochemical study of primary colon adenocarcinomas [30]. Cells lining the lumen of malignant glands showed only occasional areas of staining for Pgp, while single cells or small cells nests at the invasive edge of the carcinoma showed the strongest Pgp staining. Those adenocarcinomas in which Pgp-positive invasive cells could be demonstrated had a significantly higher incidence of blood vessel invasion and lymph node metastases. Thus Pgp expression by the carcinoma cells may be associated with increased invasiveness into vascular and lymphatic channels, rather than with differentiation. There have been few studies of the clinical and histological characteristics which may distinguish the malignant tumors that overexpress Pgp at diagnosis from those which do not have detectable Pgp. In the studies of childhood sarcoma and neuroblastoma mentioned above, it was observed that tumors which were Pgp-positive, whether at diagnosis or at relapse, were all relatively advanced in disease stage. This was clearly illustrated by the data on the 67 children with neuroblastoma, as none of 23 patients with resected localized tumors had detectable Pgp in their tumors, while 13 out of 44 patients with nonlocalized tumors showed Pgp-positive tumors. The incidence of Pgp detection was particularly high for stage IV tumors. Another interesting observation was that metastatic lesions appeared to express Pgp at levels as high as, or higher than those seen in primary tumors [22]. Taken together, these findings suggest an association between Pgp expression and tumor progression, even though the mechanism for increased expression of Pgp during cancer development is not understood at present.

Clinical significance of P-glycoprotein expression in malignant tumors Until recently, there has been a dearth of studies which examine Pgp expression in malignant tumors in relation to subsequent clinical course. Among the studies that have been done, the conclusions have varied as to whether Pgp expression correlated with clinical outcome. At least some of the differences among clinical studies may be attributed to differences in technique for detection of Pgp in tissue samples [3]. Nevertheless, several well-documented studies have demonstrated an association between detection of Pgp in tumor samples, either at diagnosis or at relapse after initial chemotherapy, and poor response to chemotherapy and shorter survival. The most impressive correlation between detection of Pgp and poor treatment outcome was observed in childhood solid tumors. In the study of children with soft tissue sarcoma cited above, 9 patients with Pgp-positive tumors at diagnosis or at subsequent biopsy all relapsed after chemotherapy, while only i of 20 patients with Pgp-negative tumors relapsed. The probability of relapse-free survival was markedly different between the Pgp-positive patients and the Pgp-negative patients [21]. In the study of children with neuroblastoma, 12 out of 13 patients with detectable Pgp in their tumors at diagnosis relapsed after primary treatment. Of the 31 patients whose tumors did not show Pgp at diagnosis, 6 relapsed after initial treatment, and all subsequent tumor samples in these 6 patients became positive for Pgp. There was a strong correlation between detection of Pgp and a reduced probability of relapse-free survival [22]. The correlation between Pgp expression and resistance to treatment has also been examined in several studies of acute lymphoblastic leukemia (ALL) and acute nonlymphoblastic leukemia (ANLL). The largest series of ALL reported so far demonstrated a high rate of relapse after chemotherapy in both children and adults who had Pgp-positive leukemia at diagnosis. The implication of Pgp detection in tumor cells was more definitely unfavorable for adult patients because all 5 adults with Pgp-posirive leukemia relapsed after an initial remission and

227 there were no further complete remissions despite additional chemotherapy. Both children and adults with Pgp-positive leukemia had a lower probability of event-free survival than patients with Pgp-negative leukemia; again the prognostic value of Pgp detection for poor patient survival was more striking for adults than for children [24]. A study of a large series of patients with ANLL also revealed a significant correlation between detection of Pgp in leukemic cells at diagnosis and a lower rate of complete remission after intensive chemotherapy, as well as shorter survival of the patient [25]. The accumulated clinical data on Pgp expression in tumors indicate that elevated expression of Pgp may occur in different types of cancer. This need not be the result of exposure to chemotherapy and there is some evidence to suggest that Pgp expression may be higher in advanced stages of malignancy. The data from several clinical studies further indicate that increased expression of Pgp may affect the biologic behaviour of malignant cells; this is manifested clinically as lower rates of complete remission following chemotherapy and higher incidence of relapse with no subsequent remission. Thus, detection of Pgp in tumors may be an indication of poor prognosis for disease-free survival. The correlation of Pgp expression with poor outcome in the clinical studies raises a number of questions. For example, (1) at what biological stage during cancer development does Pgp expression increase? (2) What other phenotypic characteristics of the malignant cells are associated with increased Pgp expression? (3) How are changes in Pgp expression regulated in malignant cells, and are these regulatory mechanisms different from control of Pgp expression in nonmalignant cells? (4) How does Pgp expression alter the behaviour of malignant cells and result in greater resistance to therapy? The use of model systems, both in vivo animal tumor models and in vitro models using cell lines, can provide insight into some of these questions and indicate more specific methods of analyses of clinical tumor samples to eventually answer these questions.

Altered hepatic foci

Small

Large

hyperplastic nodules

hyperplastic nodules

Hepatocellular carcinoma

I Lung metastases

Fig. 1. Protocol for experimental rat liver carcinogenesis. PH, partial hepatectom3; DMH, dimethylhydrazine, i.p., intraperitoneal, OA, orotic acid. The horizontal bar indicates the duration of the experiment in weeks and the sequential development of preneoplastic and neoplastic lesions are shown along the bar.

P-glycoprotein expression in rat liver carcinogenesis Animal tumor models allow the study of tumor development in vivo while offering the advantages of being relatively predictable and accessible for repeated analyses. Rat liver carcinogenesis models are attractive for studies of cellular and molecular events that occur during tumor development and progression, because several biological stages of tumor development have been delineated and the multistep process of carcinogenesis may be observed, from initiation of carcinogenesis to the formation of a large malignancy capable of metastasis, within a period of 12 to 18 months. In the Orotic Acid (OA) model of rat liver carcinogenesis, initiation by partial hepatectomy coupled with an injection of DMH (dimethylhydrazine), followed by 50 to 60 weeks of promotion with a 1% OA diet result in close to 100% incidence of hepatocarcinoma. Stepwise tumor development can be observed as sequential appearance of altered hepatic loci (AHF), hyperplastic nodules (HN) and liver carcinoma. Between 30 to 60% of the liver carcinoma metastasize to the lungs (Fig. 1). The expression of Pgp in these preneoplastic and neoplastic liver lesions has been examined and compared against the surrounding nonneoplastic liver in a detailed immunohistochemical study [31]. There was no remarkable increase in Pgp staining in the early lesions during liver carcinogenesis. The microscopic foci of altered hepatocytes (AHFs) and the small hyperplastic nodules were readily demonstrated on the basis of their overexpression

228

A

G

B

D

Fig. 2A. Normal liver stained for Pgp. Pgp is localized to the canalicular membrane of hepatocytes. Magnification: x 63. Altered hepatic loci do not show notable changes in the pattern of Pgp staining from normal liver. Small hyperplasticnodules typicallyexhibit a reduction in Pgp staining compared to normal liver. B. Large hyperplastic nodule showing heterogeneous expression of Pgp. Strong staining for Pgp is evident in a minority of the hepatocytes. Magnification:x 63. C. Hepatocellular carcinoma showing strong staining for Pgp in many of the cells and a trabecular-sinusoidal architecture. Magnification: x 63. D. Multiple metastases of hepatocellular carcinoma in lung are consistently positive for Pgp. Magnification: x 25.

of enzymes such as G G T (7-glutamyl transpeptidase) and GST-P (glutathione S-transferase-7.7) but there was no Pgp overexpression c o m p a r e d to the surrounding liver. Large hyperplastic nodules (1 cm or m o r e in diameter) typically a p p e a r e d after 30 or m o r e weeks of promotion. These showed a heterogeneous pattern of Pgp staining with groups of hepatocytes which stained strongly for Pgp a m o n g hepatocytes which showed little staining (Fig. 2A, B). A f t e r 50 or m o r e weeks of promotion, the liver was typically occupied by a large, irregular t u m o r with evidence of destruction of the surrounding liver and focal necrosis. Areas with m o d e r a t e to strong Pgp overexpression were observed in the liver carcinoma, and a striking finding was the correlation of increased Pgp staining with a trabecular arrangement of hepatocytes and wide intervening si-

nusoids (Fig. 2A, C). The heterogeneous appearance of the liver carcinoma probably reflected a mixture of hyperplastic nodule, reactive changes in the surrounding liver, liver carcinoma and areas of necrotic tumor. Although there was a general increase in Pgp staining with the development of liver carcinoma, the heterogeneous composition of these tumors complicated the study of any direct relationship between Pgp expression and malignant behaviour. A significant finding was that lung metastases of the liver carcinoma were consistently positive for Pgp (Fig. 2D). The association between Pgp overexpression and a trabecular architecture with widened vascular sinusoids in the liver carcinoma, and the consistent presence of Pgp-positive cells in the lung metastases suggested that Pgp m a y be involved in vascular invasion and distant metastases.

229 Overexpression of Pgp has also been demonstrated to occur relatively late during experimental liver carcinogenesis in other model systems. In rats treated with the Solt-Farber protocol for hepatocarcinogenesis, increased staining for Pgp was seen in a he~.erogeneous pattern in hyperplastic nodules and liver carcinoma but not in the preceding lesions of enzyme-altered loci [32]. In hepatitis B virustransgenic mice which developed progressive pathologic changes of liver injury, dysplasia, microscopic nodular hyperplasia and hepatocellular carcinoma over a period of 18 to 20 months, a marked increase in class I Pgp was detected only at the time of appearance of the carcinoma [33]. These findings in different model systems suggest that elevated levels of Pgp during liver carcinogenesis cannot be simply interpreted as a direct response to administration of hepatotoxins such as dimethylhydrazine or acetylaminofluorene, or to partial hepatectomy or other forms of liver cell injury. Although it has been demonstrated in several studies that liver damage by means of partial hepatectomy, carbon tetrachloride or acetylaminofluorene can increase Pgp expression in the liver, such increases have all occurred within i or 2 days with no further increase by about 4 days [34, 35]. In contrast, the increases in Pgp expression observed in long-term studies of rat liver carcinogenesis were observed several months after partial hepatectomy or administration of hepatotoxic agents. Overexpression of Pgp during liver carcinogenesis also failed to correlate with the 'Resistant Hepatocyte' phenotype which has been described for the early lesions during carcinogenesis. This phenotype includes resistance to multiple carcinogens and toxins and overexpression of enymes such as GST-R and is thought to confer proliferative advantage to precursors of hepatocellular carcinoma during tumor promotion [36]. Immunohistochemical and histochemical staining of early lesions of liver carcinogenesis, the altered hepatic foci and the small hyperplastic nodules, consistently showed uniform overexpression of enzymes such G G T and GST-P while staining for Pgp in serial sections of the same early lesions revealed no increase in Pgp staining. Instead, Pgp expression appeared to increase with the formation of a tumor mass and was also consistently detected in

metastatic lesions in the lungs, suggesting an association with advanced stages in liver carcinogenesis. Further studies of the role of Pgp in experimental rat liver carcinogenesis are warranted to determine if areas with relatively high Pgp expression within the liver tumors represent tumor of greater malignancy. This may be done initially by correlating Pgp expression of small tumor samples with degree of malignancy as measured by biological assays, such as ability to form tumors in syngeneic rats at orthotopic or ectopic sites. In addition, the expression of Pgp in liver carcinoma metastatic to lungs and other tissues require further examination, in view of reports suggesting that Pgp expression by tumor cells may depend on the tissue microenvironment [37]. The observations on Pgp expression in multistage liver carcinogenesis supported the idea from clinical studies that Pgp expression may increase in more advanced stages of malignant disease. Furthermore, the studies of experimental rat liver carcinoma suggested that Pgp may be associated with aspects of tumor cell behaviour other than resistance to cytotoxic drugs. Strong expression of Pgp in liver carcinoma cells appeared to correlate with a well developed vascular supply, raising the possibility that Pgp expression may be involved in tumor cell interactions with stroma. This in turn may determine the occurrence of invasion and metastasis. In view of the clinical and laboratory evidence linking overexpression of Pgp with tumor progression, there is a need to understand how Pgp expression is regulated in tumor cells as this may contribute to the overall concept of genetic and epigenetic changes which underlie tumor progression. Few studies have been done on Pgp regulation during tumor development. However, the modulation of Pgp expression has been widely studied in cultured cells, and to a lesser extent, in normal tissues. The findings from these investigations may serve as a basis to initiate studies of Pgp regulation during malignant progression.

Regulation of P-glycoprotein expression The expression of Pgp in vivo appears to be closely regulated, and pronounced changes in Pgp content

230 can occur in a tissue over a relatively short time. In the gravid uterus of the hamster, the lining epithelium showed a rapid rise in expression of class II Pgp between approximately the 7th and ]0 th days of the 21-day gestation period and a similarly rapid fall in expression immediately postpartum [11]. Similar observations have been made in the mouse, in which it was further demonstrated that a combination of estrogen and progesterone could effect rapid changes of class II Pgp expression in the uterine epithelium [38]. Thus the physiologic expression of Pgp in the uterine epithelium appears to be regulated, at least in part, by estrogen and progesterone. It seems likely that other hormones and growth factors may regulate Pgp expression in tissues but such mechanisms have not been identified to date. Treatment of cultured cells with a variety of chemicals, as well as heat shock or X-irradiation can lead to increase in Pgp expression. Exposure of the renal carcinoma cell line HTB- 46 to heat shock or sodium arsenite for 2 to 3 hours resulted in a 7 to 8-fold increase in Pgp mRNA [39]. Treatment of the neuroblastoma cell line SK-N-SH or its neuroblastic subclone SH-SY5Y with retinoic acid resulted in approximately 10-fold increase in Pgp expression after 1 to 2 days' exposure [29]. The colon carcinoma cell lines SW620 and HCT-15 showed an increase in Pgp mRNA and protein after exposure to sodium butyrate or dimethyl sulfoxide. In SW620 cells treated with sodium butyrate, a maximal increase of 20 to 25-fold was reached after 24 hours of exposure [40]. The effects of retinoic acid, sodium butyrate and dimethyl sulfoxide were considered in terms of the use of these chemicals as in vitro differentiating agents, but direct correlation between increases in Pgp expression and indices of cellular differentiation have not been demonstrated in these studies. In all the studies mentioned above, the induction of Pgp expression in cultured cells by physical or chemical means was clearly specific to combinations of cell line and treatment employed. Two out of 9 cell lines which were given heat shock treatment responded with an increase in Pgp mRNA, while the remaining 7 cell lines did not show such a response. Four out of 8 colon carcinoma cell lines showed increased Pgp expression after sodium butyrate treatment. Dimethyl sulfoxide appeared to

be similarly effective in the cell lines which responded to sodium butyrate while retinoic acid did not cause an increase in Pgp expression in these colon carcinoma cell lines. On the other hand, retinoic acid was very effective in inducing Pgp expression in neuroblastoma cell lines. It is possible that these physical or chemical treatments when administered to responsive cells result in activation of a common cellular mechanism involving overexpression of Pgp which enables cells to persist or adapt to the treatment. In a different type of experimental system, treatment of the Chinese hamster ovary cell line AuxB1 with repeated doses of lethal X-irradiation (1% cell survival) was reported to yield postirradiation sublines with increased Pgp expression. Interestingly, these sublines showed no change in sensitivity to X-rays but demonstrated a low level of multidrug resistance to drugs such as vincristine, vinblastine and colchicine [41]. Changes in Pgp expression in cultured cells appear to be regulated at many levels. Cell lines selected for multidrug resistance in vitro may elevate Pgp expression through gene amplification, increased mRNA content or increased protein [42]. The increase in Pgp mRNA expression in HTB-46 cells after heat shock or sodium arsenite treatment was abolished in the presence of actinomycin D, suggesting that new RNA synthesis was required for the response [39]. Sodium butyrate treatment of SW 620 cells increased the amount of Pgp mRNA without a corresponding increase in transcriptional activity as measured by nuclear run-off assays, suggesting that changes in mRNA stability might be responsible for the overexpression of Pgp mRNA [40]. The increase in Pgp expression in post-irradiation sCtblines of AuxB1 cells was demonstrated at the protein level but there was no comparable alteration in Pgp mRNA content or Pgp gene amplification. It was proposed that translational and/or posttranslational regulation of Pgp was altered in these sublines [41]. There is evidence to suggest that transcription of Pgp genes may be increased through the activation of specific oncogenes. The c-Ha-ras-1 oncogene and a mutant p53 gene have been shown to stimulate the activity of the human mdrl promoter in NIH 3T3 cells [43]. Other oncogenes such as c-los, c-jun or

231 c-myc did not stimulate mdrl promoter activity in this system. In another study, a mutant murine p53 gene was also found to stimulate the activity of the hamster pgpl promoter which has been transfected into Chinese hamster ovary cells [44]. Mutations of the Ha-ras oncogene and the p53 gene have been identified in many tumor types including hepatocellular carcinoma [45]. As well, it has been shown recently that mice whose p53 genes have been inactivated through gene targetting had a greatly enhanced rate of tumor progression in skin carcinogenesis studies. In these mice, absence of normal p53 gene expression did not affect skin tumor initiation by chemical carcinogen but resulted in much more rapid progression of benign skin papillomas to carcinomas [46]. Further studies are needed to determine if increased expression of Pgp in tumors is related to activation of one or more oncogenes during carcinogenesis. Thus, the studies performed to date on regulation of Pgp expression suggest several possibilities for the mechanism of Pgp overexpression during carcinogenesis. For example, malignant cells may activate, in an aberrant manner, physiologic pathways which cause elevated Pgp expression. Alternatively, changes in the physical or chemical microenvironment within a malignant tumor may trigger an increase in Pgp expression. It is also likely that activation of oncogenes or loss of tumor suppressor genes may mediate changes in Pgp expression during carcinogenesis. Studies to investigate these possible mechanisms of Pgp overexpression in tumor cells should contribute to our knowledge of molecular alterations which are associated with late stages of tumor progression.

Concluding remarks In this chapter we have presented clinical and laboratory data in support of the hypothesis that Pgp is linked to the malignant phenotype. One may speculate on how Pgp overexpression might play an important role in malignant behaviour. The ability of Pgp to exclude many lipophilic drugs should provide tumor cells that overexpress Pgp with a growth advantage during treatment with chemotherapy. In

addition, Pgp is likely to transport multiple substrates other than lipophilic drugs. The nature of these other substrates has not been defined but Pgp overexpression may provide growth advantages to tumor cells through secretion of autocrine growth factors, transport of metabolites which modify the tumor stroma, or removal of unwanted cell products. The fact that Pgp expression is highly responsive to modulation by various intrinsic and extrinsic factors further suggests that this transport system may be exploited by malignant cells during tumor progression. The challenge for the future is to identify the role played by Pgp in the development of different types of malignancies.

Key unanswered questions 1. Are increases in Pgp expression in tumors generally associated with poor response to treatment? Is Pgp the limiting factor that determines response to chemotherapy? If Pgp expression were abrogated by genetic means or antagonized by pharmacologic means at an early stage of tumor development, will tumors be more responsive to chemotherapeutic control? 2. In experimental rat liver carcinogenesis, do areas with relatively high Pgp expression within the primary liver tumors represent tumor of greater malignancy? Are distant metastases to organs other than the lungs always positive for Pgp? How does an increase in Pgp affect tumor cell behaviour? 3. What factors modulate changes in Pgp during carcinogenesis? Is the overexpression of Pgp in tumors related to activation of specific oncogenes? Do different mechanisms operate for overexpression of Pgp in tumors "de novo", prior to chemotherapy, as opposed to tumors which have been exposed to chemotherapy? 4. What substrates are transported by Pgp in malignant cells?

References 1. KartnerN, RiordanJR, LingV: CellsurfaceP-glycoprotein

232

2.

3. 4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

associated with multidrug resistance in mammalian cell lines. Science 221: 1285-1288, 1983 Endicott JA, Ling V: The biochemistry of P-glycoproteinmediated multidrug resistance. Ann Rev Biochem 58: 137171, 1989 Arceci R: Clinical significance of P-glycoprotein in multidrug resistance malignancies. Blood 81: 2215-2222, 1993 Georges E, Sharom F, Ling V: Multidrug resistance and chemosensitization: therapeutic implications for cancer chemotherapy. Adv Pharm 21: 185-220, 1990 Homolya L, Hollo Z, Germann UA, Pastan I, Gottesman MM, Sarkadi B: Fluorescent cellular indicators are extruded by the multidrug resistance protein. J Biol Chem 268: 2149321496, 1993 Ueda K, Okamura N, Hirai M, Tanigawara Y, Saeki T, Kioka N, Komano T, Hori R: Human P-glycoprotein transports cortisol, aldosterone, and dexamethasone, but not progesterone. J Biol Chem 267: 24248-24252,1992 Lee CH, Bradley G, Zhang J-T, Ling V: Differential expression of P-glycoprotein genes in primary rat hepatocyte culture. J Cell Physio1157: 392402, 1993 Thiebaut T, Tsuruo T, Hamada H, Gottesman MM, Pastan I, Willingham MC: Cellular localization of the multidrug-resistance gene product P-glycoprotein in normal human tissues. PNAS USA 84: 7735-7738, 1987 Croop JM, Raymond M, Haber D, Devault A, Arceci R J, Gros E Housman DE: The three mouse multidrug resistance (mdr) genes are expressed in a tissue specific manner in normal mouse tissues. Mol Cell Biol 9: 1346-1350, 1989 Cordon-Cardo C, O'Brien JR Boccia J, Casals D, Bertino JR, Melamed MR: Expression of the multidrug resistance gene product (P-glycoprotein) in human normal and tumor tissues. J Histochem Cytochem 38:1277-1287, 1990 Bradley G, Georges E, Ling V: Sex dependent and independent expression of P-glycoprotein isoforms in Chinese hamster. J Cell Physio1145: 398-408, 1990 Chaudhary PM, Mechetner EB, Roninson IB: Expression and activity of the multidrug resistance P-glycoprotein in human peripheral blood lymphocytes. Blood 80: 2735-2739, 1992 Drach D, Zhao S, Drach J, Mahadevia R, Gattringer C, Huber H, Andreeff M: Subpopulations of normal peripheral blood and bone marrow cells express a functional multidrug resistant phenotype. Blood 80: 2729-2734, 1992 Smit JJM, Schinkel AH, Oude Elferink RPJ, Groen AK, Wagenaar E, van Deemter L, Mol CAAM, Ottenhoff R, van der Lugt NMT, van Roon MA, van der Valk MA, Offerhaus GJA, Berns AJM, Borst P: Homozygous disruption of the murine mdr2 P-glycoprotein gene leads to a complete absence of phospholipid from bile and to liver disease. Cell 75: 451462, 1993 Childs S, Ling V: The MDR superfamily of genes and its biological implications. Important Advances in Oncology 2136, 1994 Bear CE, Li C, Kartner N, Bridges R J, Jensen TJ, Ramjeesingh M, Riordan JR: Purification and functional reconstitu-

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

tion of the cystic fibrosis transmembrane conductance regulator (CFTR). Cell 68: 809-818, 1992 Powis SJ, Townsend ARM, Deverson EV, Bastin J, Butcher GW, Howard JC: Restoration of antigen presentation to the mutant cell line RMA-S by an MHC-linked transporter. Nature 354: 528-531, 1991 Cole SPC, Bhardwaj G, Gerlach JH, Mackie JE, Grant CE, Almquist KC, Stewart EU, Kurz AMV, Duncan AMV, Deeley RG: Identification of a novel ATP-binding Cassette transporter gene overexpressed in a multidrug resistant human lung cancer cell line. Science 258: 1650-1654, 1992 Miller TR Grogan TM, Dalton WS, Spier CM, Scheper RJ, Salmon SE: P-Glycoprotein expression in malignant lymphoma and reversal of clinical drug resistance with chemotherapy plus high-dose verapamil. J Clin Oncol 9: 17-24, 1991 Grogan TM, Spier CM, Salmon SE, Matzner M, Rybski JA, Weinstein RS, Scheper RJ, Dalton WS: P-glycoprotein expression in human plasma cell myeloma: An acquired trait related to prior chemotherapy. Blood, in press Chan HSL, Thorner R Haddad G, Ling V: Immunohistochemical detection of P-glycoprotein: prognostic correlation in soft tissue sarcoma of childhood. J Clin Oncol 8: 689704, 1990 Chan HSL, Haddad G, Thorner PS, DeBoer G, Lin YR Ondrusek N, Yeger H, Ling V: P-glycoprotein expression as a predictor of the outcome of therapy for neuroblastoma. New Eng J Med 325: 1608-1614,1991 Kuwazuru Y, Yoshimura A, Hanada S, Ichikawa M, Saito T, Uozumi K, Utsunomiya A, Arima T: Expression of the multidrug transporter, P-glycoprotein in chronic myelogenous leukaemia cells in blast crisis. Br Journal of Haematology 74: 24-29, 1990 Goasguen JE, Dossot J-M, Fardel O, Le Mee E Le Gall E, Leblay R, LePrise PY, Chaperon J, Fauchet R: Expression of the multidrug resistance-associated P-glycoprotein (P-170) in 59 cases of de novo acute lymphoblastic leukemia: prognostic implications. Blood 81: 2394-2398, 1993 Campos L, Guyotat D, Archimbaud E, Calmard-Oriol P, Tsuruo T, Troncy J, Treille D, Fiere D: Clinical significance of multidrug resistance P-glycoprotein expression on acute nonlymphoblastic leukemia cells at diagnosis. Blood 79: 473476, 1992 Niehans GA, Jaszcz W, Brunetto V, Perri RT, Gajl-Peczalska K, Wick MR, Tsuruo T, Bloomfield CD: Immunohistochemical identification of P-glycoprotein in previously untreated diffuse large cell and immunoblastic lymphomas. Cancer Res 52: 3768-3775,1992 Gerlach JH, Bell DR, Karakousis C, Slocum HK, Kartner N, Rustum YM, Ling V, Baker RM: P-glycoprotein in human sarcoma: Evidence for multidrug resistance. J Clin Oncol 5: 1452-1460, 1987 Goldstein LJ, Galski H, Fojo A, Willingham M, Lai S, Gazdar A, Pirker R, Green A, Crist W, Brodeur GM, Lieber M, Cossmau J, Gottesman MM, Pastan I: Expression of a mul-

233

29.

30.

31.

32.

33.

34.

35.

36.

37. 38.

tidrug resistance gene in human tumors. J Natl Canc Inst 81: 116-124, 1989 Bates SE, Mickley LA, Chen Y, Richert N, Rudick J, Biedler JL, Fojo AJ: Expression of a drug resistance gene in human neuroblastoma cell lines: modulation by retinoic acid-induced differentiation. Mol Cell Biol 9: 4337-4344, 1989 Weinstein RS, Jakate SM, Dominguez JM, Leibowitz MD, Koukoulis JK, Kuszak JR, Klusens LF, Grogan TF: Relationship of the expression of the multidrug resistance gene product (P-glycoprotein) in human colon carcinoma to local tumor aggressiveness and lymph node metastasis. Cancer Res 51: 2720-2726, 1991 Bradley G, Sharma R, Rajalakshmi S, Ling V: P-glycoprorein expression during tumor progression in the rat liver. Cancer Res 52: 5154-5164, 1992 Nakatsukasa H, Evarts RR Butt RK, Nagy R Thorgeirsson SS: Cellular pattern of multidrug-resistance gene expression during chemical hepatocarcinogenesis in the rat. Mol Carcinogenesis 6: 190-198, 1992 Kuo MT, Zhao J-Y, Teeter LD, Ikeguchi M, Chisari FV: Activation of multidrug resistance (P-glycoprotein) mdr3/ mdrla gene during the development of hepatocellular carcinoma in hepatitis B virus transgenic Mice. Cell Growth & Diff 3: 531-540, 1992 Gant T, Silverman JA, Bisgaard HC, Burt RK, Marino PA, Thorgeirsson SS: Regulation of 2-acetylaminofluorene- and 3-methylcholanthrene-mediated induction of multidrug resistance and cytochrome P450IA gene family expression in primary hepatocyte cultures and rat liver. Mol Carcinogenesis 4: 499-509, 1991 Brown PC, Thorgeirsson SS, Silverman JA: Cloning and regulation of the rat mdr2 gene. Nucleic Acids Res 21: 38853891, 1993 Farber E: Cellular biochemistry of the stepwise development of cancer with chemicals. Cancer Res 44: 5463-5474, 1984 Fidler IJ: Critical factors in the biology of human cancer me_ tastasis. Cancer Res 50: 6130-6138, 1990 Arceci RJ, Baas F, Raponi R, Horwitz SB, Housman D,

39.

40.

41.

42.

43.

44.

45.

46.

Croop JM: Multidrug resistance gene expression is controlled by steroid hormones in the secretory epithelium of the uterus. Mol Reprod Develop 25: 101-109, 1990 Chin KV, Tanaka S, Darlington G, Pastan I, Gottesman MM: Heat shock and arsenite increase expression of the multidrug resistance (MDR1) gene in human renal carcinoma cells. J Biol Chem 265: 221-226, 1990 Mickley LA, Bates SE, Richert ND, Currier S, Tanaka S, Foss F, Rosen N, Fojo AT: Modulation of the expression of a multidrug resistance gene (mdr-1/P-glycoprotein) by differentiating agents. J Biol Chem 264: 18031-18040, 1989 Hill BT, Deuchars K, Hosking LK, Ling V, Whelan RD: Overexpression of P-glycoprotein in mammalian tumor cell lines after fractionated X irradiation in vitro. J Natl Canc Inst 82: 607-612, 1990 Bradley G, Naik M, Ling V: P-glycoprotein expression in multidrug-resistant human ovarian carcinoma cell lines. Cancer Res 49: 2790-2796, 1989 Chin K-V, Ueda K, Pastan I, Gottesman MM: Modulation of activity of the promoter of the human MDRI gene by Ras and p53. Science 255: 459-462, 1992 Zastawny RL, Salvino R, Chen J, Benchimol S, Ling V: The core promoter region of the P-Gtycoprotein gene is sufficient to confer differential responsiveness to wild-type and mutant p53. Oncogene 8: 1529-1535, 1993 Scorsone KA, Zhou Y-Z, Butel JS, Slagle BL: p53 mutations cluster at codon 249 in hepatitis B virus-positive hepatocellular carcinomas from China. Cancer Res 52: 1635-1638, 1992 Kemp CJ, Donehower LA, Bradley A, Balmain A: Reduction of p53 gene dosage does not increase initiation or promotion but enhances malignant progression of chemically induced skin tumors. Cell 74: 813-822, 1993

Address for offprints: Victor Ling, Ontario Cancer Institute, 500 Sherbourne St., Toronto, Ontario, Canada M4X 1K9