Cancer and Metastasis Reviews 13: 209—222, 1994.
`© 1994 Klnwerncademic Publishers. Printed in the Netherlands.
`
`Modulation of tumor cell response to chemotherapy by the organ
`environment
`
`Isaiah J. Fidler, Christoph Wilmanns, Alexander Staroselsky, Robert Radinsky, Zhongyun Dong and
`Dominic Fan
`
`DepartmentofCetl Biology, HMB I73, The University of'lizxns M. D. Anderson Cancer Center, 1515 Holcombe
`Boulevard, Houston, TX 77030, USA
`
`Key words: Organ environment, drug resistance, mdrl, P—glycoprotein, metastasis, epigenetic
`
`Abstract
`
`The outcome of cancer metastasis depends on the interaction of metastatic cells with various host factors. The
`implantation of human cancer cells into anatomically correct (orthotopic) sites in nude mice can be used to
`ascertain their metastatic potential. While it is clear that vascularity and local immunity can retard or facilitate
`tumor growth, we have found that the organ environment also influences tumor cell functions such as produc-
`tion of degradative enzymes. The organ microenvironment can also influence the response of metastases to
`chemotherapy. It is not uncommon to observe the regression of cancer metastases in one organ and their
`continued growth in other sites after systemic chemotherapy. We demonstrated this effect in a series of experi—
`ments using a murine fibrosarcoma, a murine colon carcinoma, and a human colon carcinoma. The tumor cells
`were implanted subcutaneously or into different viscera] organs. Subcutaneous tumors were sensitive to dex—
`orubicin (DXR), whereas lung or liver metastases were not. In contrast, sensitivity to S—FU did not differ
`between these sites of growth. The differences in response to DXR between s.c. tumors (sensitive) and lung or
`liver tumors (resistant) were not due to variations in DXR potency or DXR distribution. The expression of
`the multidrug resistance-associated P-glycoprotein as determined by flow cytometric analysis of tumor cells
`harvested from lesions in different organs correlated inversely with their sensitivity to DXR: increased P-
`glycoprotein was associated with overexpression ofmdrl mRNA. However, the organ-specific mechanism for
`upregulatin g mdrl and P—glyeoprotein has yet to be elucidated.
`
`Introduction
`
`Despite significant improvements in diagnosis, sur-
`gical techniques, general patient care, and local and
`systemic adj uvant therapies, most deaths from can-
`cer are due to metastases that are resistant to con-
`
`ventional therapies. In a large number of patients
`with cancer, metastasis may well have occurred by
`the time of diagnosis. The metastases can be located
`in different lymph nodes and visceral organs and in
`various regions of the same organ, thus complicat—
`ing their treatment. Furthermore, the specific organ
`environment can modify the response of 3 meta—
`
`static tumor cell to systemic therapy and alter the
`efficiency of anticancer agents [1].
`The major barrier to the treatment of metastases
`is the biological heterogeneity of cancer cells in pri-
`mary and secondary neoplasms. This heterogeneity
`is exhibited in a wide range of genetic, biochemical,
`immunological, and biological characteristics, such
`
`as cell surface receptors, enzymes, karyotypes, cell
`morphologies, growth properties, sensitivities to
`various therapeutic agents, and ability to invade
`and produce metastasis [1—5].
`The search for the mechanisms that regulate the
`pattern of metastasis began in 1889 when Stephen
`
`Genentech 2101
`
`Hospira v. Genentech
`|PR2017-00737
`
`Genentech 2101
`Hospira v. Genentech
`IPR2017-00737
`
`

`

`210
`
`Pager asked ‘what is it that decides what organs
`shall suffer in a case of disseminated cancer?’ [6].
`Paget’s inquiry was motivated by the discrepancy
`between blood flow and relative frequency of me-
`tastases in different organs. He examined the au-
`topsy records of women who died of breast cancer
`
`and patients with other neoplasms and concluded
`
`that the pattern of metastasis was predictable. He
`drew attention to the frequency of ovarian metasta—
`ses and to the differences in incidence of skeletal
`
`metastases from different primary tumors. These
`findings were not compatible with the view that the
`pattern ofmetastasis was due to ‘a matter of chance’
`
`or that tissues behaved passively in determining the
`probability of clinically relevant metastases. Rath-
`er, Paget concluded that certain favored tumor cells
`(the ‘seed‘) had a specific affinity for growth in the
`milieu provided by certain organs (the ‘soil’). Me~
`tastasis resulted only when the ‘seed and soil’ were
`compatible [6].
`A modern definition of this hypothesis consists of
`three principles. First, neoplasms are heterogene-
`ous in biologic and metastatic properties [7]. Sec-
`ond, the process of metastasis is not random. Rath-
`er, it consists of a series of linked, sequential steps
`that must be completed by tumor cells if metastases
`are to develop [8]. Thus, metastatic cells must sue—
`ceed in invasion, embolization, survival in the circu—
`
`lation, arrest in a distant capillary bed, and extrava—
`sation into and multiplication in organ parenchyma.
`
`Although some of the steps in this process contain
`stochastic elements, metastasis as a whole favors
`
`the survival and growth of a few subpopulations of
`cells that preexist within the parent neoplasm [7].
`Metastases can have a clonal origin, and different
`metastases produced from the same primary neo—
`plasm can originate from the proliferation of differ—
`
`ent single cells [9, 10]. Third, the outcome of metas—
`tasis depends on the interaction of metastatic cells
`with different organ environments [11]. Thus, both
`the ‘seed’ and the ‘soil’ profoundly influence the
`outcome of systemic therapy for cancer.
`While a great deal of attention has been given to
`the heterogeneous nature of neoplasms, which in—
`cludes variations in intrinsic sensitivity to chemo-
`therapeutic agents [2, 12], less emphasis has been
`given to the influence of the organ environment,
`
`i.e., the site of growth, on tumor response to anti—
`cancer agents [5, 13, 14]. This issue is important be—
`cause it is not uncommon to observe the regression
`of cancer metastases in one organ and their contin-
`ued growth in others after systemic therapy [15, 16].
`In a classic study, Slack and Bross [17] analyzed data
`from drug-screening trials with 1687 neoplasms
`growing in 6 organ sites for primary neoplasms and
`6 organ sites for their metastases. Sixty days after
`chemotherapy, the percentage reduction in tumor
`size differed significantly among different metasta—
`ses growing in different organs, but among primary
`tumor sites it did not. With few exceptions, lymph
`node and skin metastases were more susceptible to
`chemotherapy than metastases in visceral organs
`[1?]. Differences in sensitivity to various chemo—
`therapeutic agents of experimental tumors growing
`in different organs have also been reported by sev—
`eral investigators [12, 18, 19]. Tumors in the subcutis
`were more sensitive than tumors growing in visceral
`organs, agreeing with clinical observations [14—16].
`The nutritional status of cells [20], presence of
`growth factors and other signal-transducing agents
`[21], oxygenation [20—23], pH [24—26], extent of vas-
`cular network and its functionality [27—31], and lo~
`cal immunity [32] can all contribute to the success or
`failure of cancer therapy.
`Several intrinsic properties of tumor cells can
`render them resistant to chemotherapeutic drugs
`[33]. These could include an amplification of the
`mdrl gene and overexpression of the Mr 170,000
`surface P—glyeoprotein (P-gp) [34—39], overexpres-
`sion of the Mr 22,000 calcium—binding cytoplasmic
`protein [40, 41], increased glutathione transferasc
`lcvcls‘[42], altered cellular calcium and calmodulin
`levels [43, 44], formation of double~minute chromo—
`somes [45], increased activity of protein kinase C
`(PKC) [46, 47], and lack of drug interference with
`type II topoisomcrase activity [48, 49].
`The intrinsic resistance of tumor cells to chemo-
`
`therapeutic drugs can be mediated by both genetic
`and epigenctic mechanisms. The former can devel-
`op through genc amplification (notably, the mdrl
`gene), gene rearrangements, and transcriptional,
`translational, and posttranslational events [50].
`Phenotypic changes that can modulate drug resist—
`ance are often associated with increased activities
`
`

`

`of drug-detoxifying enzymes [51], metabolism—reg—
`ulating enzymes, and efflux proteins [52]. In addi~
`tion, increased levels of PKC activity have been ob—
`served in several MDR tumor cell lines [46, 47].
`Similar effects have also been associated with de-
`
`creased activity of specific enzymes or influx of
`transmembrane proteins and production of altered
`enzymes with decreased affinity for a given drug.
`Most of the above data have been derived from
`
`examining tumor cells growing in culture. However,
`the relevance of culture conditions to the clinical re—
`
`ality is unclear. The object of this review is to sum—
`marize several relevant experimental metastasis
`systems that clearly demonstrate the profound in—
`fluence of the organ environment on the response
`of tumor cells to systemic therapy.
`
`Site-dependent differences in response of the
`UV—2237 murine fibrosarcoma to systemic therapy
`with doxorubicin
`
`To study the influence of organ mieroenvironment
`on the response of metastatic cells to systemic che—
`motherapy, we inoculated the murine U V—2237MM
`fibrosarcoma cells into different organs of synge-
`neic C3HiHeN mice and followed this with i.v. ad—
`
`ministration of DXR [53]. Focal ‘primary’ tumors
`growing in the subcutis or spleen were sensitive to
`
`DXR, whereas experimental metastases in the lung
`were not.
`
`The mechanisms that regulate this differential re-
`sponse to DXR are unclear, but tumor vascularity
`can influence the delivery of drugs to a tumor [27—
`31], and blood flow to tumors is not regulated by the
`
`same mechanisms Operative for normal tissues [15,
`18]. However, when we measured blood supply to
`tumors and organs by monitoring the distribution
`of 51Cr—labclcd RBC after iv. injection, we found no
`correlation between this measurement and re—
`
`sponse to DXR, ruling out a simple difierence in
`vascularity as a controlling factor in this process.
`Differences in accumulation of DXR in tumors
`
`growing at different sites could also account for dif—
`ferences in antiproliferativc responses. Our results,
`however, indicated that the DXR concentration in
`
`the lung was at least twice that in the skin or spleen
`
`211
`
`[53]. The higher distribution of DX'R to lung metas-
`tases than to skin tumors agreed with previous re—
`sults in which the uptake of DXR [16] or cyclophos—
`phamide [54] was increased in lung metastases pro—
`duced by intramuscularly growing rodent sarco~
`mas. Thus, the accumulation of DXR in the murine
`fibrosareoma lesions did not correlate with their
`
`sensitivity to the drug.
`The DXR resistance of fibrosarcoma lung metas—
`tases was likewise not due to the emergence of a re—
`sistant subpopulation ofcells from this heterogene—
`ous neoplasm [55]. We base this conclusion on an
`examination of DXR sensitivity under in vitro con—
`ditions. Although in some tumor systems, cells from
`lung metastases have been shown to be more sensi-
`
`tive to chemotherapy than parental cells, in the case
`of the UV—2237 fibrosareorna, DXR sensitivity was
`
`similar in both parental cells and cells isolated from
`lung metastases.
`PKC expression can be altered by tumor-promot-
`ing phorbol esters and by oneogenic transformation
`[46, 47], providing evidence that growth factors and
`related agents may serve as paracrine factors that
`alter PKC expression in particular organ environ—
`ments [56]. Since PKC activity levels correlate with
`DXR resistance in the UV-2237 fibrosarcoma cell
`
`line and its DXR—selectcd multidrug resistant varia—
`nts [27], we examined PKC activity levels in tumors
`growing at different organ sites. Similar PKC activ-
`ity levels in lung, spleen, and subcutaneous tumors
`were found, indicating that these organ environ-
`ments did not alter PKC expression in the fibrosar—
`coma cells.
`
`Resistance to chemotherapy, such as alkylating
`agents, can develop in tumors by mechanisms that
`are operative only in vivo [18]. Some of these mech—
`anisms can involve tissue pH [24—26], oxygenation
`[20—23, 57—61], local immunity, cytokincs, and other
`
`inhibitors of tumor cell growth. These may be addi~
`Live or antagonistic to one another or to chemother—
`
`apeutic agents. In particular, the role of oxygen in
`cell killing by chemotherapeutic drugs [21, 22, 57a
`61] and in cell proliferation has been extensively
`studied in different in vitro systems [58—61], includ-
`ing growth in semisolid agar. DXR was found to be
`significantly more toxic to hypoxie cells. The influ—
`ence of oxygen tension, however, appears to be tis-
`
`

`

`212
`
`sue type-dependent. Cells from pancreas and ovari-
`an carcinoma grew well in 5 % or lower oxygen at-
`mosphere, whereas lung cancer cells grew better in
`20% oxygen atmosphere [61]. In the case of ovarian
`carcinomas, incubation in a reduced oxygen atmo—
`sphere increased tumor sensitivity to DXR. These
`differences in oxygen requirement could represent
`the physiological oxygen tension for tumor cells in
`site. Although many mammalian tissues have a p02
`equivalent to a 5 % oxygen atmosphere, cells of the
`pulmonary system require higher oxygen tension
`[62].
`
`DXR and other quinoneucontainin g compounds
`are capable of reacting with molecular oxygen to
`generate various oxygen species such as superox-
`ide, hydrogen peroxide, and hydroxyl radicals [63].
`The intracellular production of these toxic radicals
`by DXR has the potential to produce cytotoxicity,
`but the effects of free radicals is neutralized by anti—
`oxidant enzymes such as supcroxide dismutase [64],
`which is activated by hyperoxia [65]. All these fac—
`tors could combine to produce the present results:
`tumors growing in the lungs are bathed by oxygen
`and are resistant to DXR, whereas tumors growing
`in the subcutis and the spleen grow under relatively
`anoxie conditions favorable to the antiproliferative
`effects of DXR [53, 66].
`These studies have shown that growth in the lung
`renders fibrosareoma cells relatively resistant to
`
`systemic administration of DXR and that variations
`in oxygenation may well be the cause. Whether
`
`growth in other visceral organs also produces these
`effects remains to be examined.
`
`The effects of the organ environment on sensitivity
`of colon carcinoma to chemotherapy
`
`Human colon carcinoma
`
`Human colon carcinomas are heterogeneous for a
`
`variety of biologic properties that include invasion
`and metastasis. The presence of a small subpopula—
`tion of cells with a highly metastatic phenotype has
`important clinical implications for diagnosis and
`therapy of cancer. For this reason, it is important to
`develop animal models for the selection and isola-
`
`tion of metastatic variants from human neoplasms
`and for testing the metastatic potential of human
`tumor cells [67].
`We have implanted human colon cancer cells
`(obtained from a surgical specimen) into different
`organs of nude mice and then recovered the tumors
`
`and established each in culture [67—69]. The colon
`cancer cells implanted into the subcutis of nude
`mice produced local tumors with only limited inva-
`siveness. This lack of invasion, as well as the conse—
`
`quent lack of metastasis, has often been associated
`with the development of a dense, fibrous capsule
`around the tumor [70]. One tumor cell property
`that is a prerequisite for metastasis is the ability to
`degrade connective-tissue extracellular matrix and
`basement membrane components that constitute
`barriers against invading tumor cells [7'1]. Metastat—
`ic tumor cells possess various proteases and glycosi-
`dascs capable of degrading extracellular matrix-de-
`grading enzymes such as type IV collagenase (gela-
`tinase, matrix metalloproteinase 2) and heparinase
`(heparan sulfate—specific cndo—B—D—glucuronidase)
`in metastatic tumor cells. We found a strong corre-
`lation between the type IV collagenase activity of
`
`human colon carcinoma cells and their ability to
`metastasize to the liver after the cells were inoculat-
`
`ed into the spleen of nude mice [72].
`We examined the influence of organ environ—
`ment on the metastasis of human colon carcinoma
`
`cells and on their extracellular matrix—degradin g ac—
`tivities using four different cell lines with different
`metastatic potentials. When the cells of each line
`were injected subcutaneously, none produced any
`visceral metastases. In contrast, when they were in-
`jected into the cecum, they metastasized to regional
`mcsenterie lymph nodes and the liver [72].
`Since the interaction of stromal fibroblasts can
`
`influence the tumorigenicity and biological behav—
`ior of tumor cells, we determined whether organ-
`
`specific fibroblasts could directly influence the in-
`vasive ability of human colon carcinoma cells [73].
`Primary cultures of nude mouse skin, lung, and co-
`lon fibroblasts were established. Invasive and meta-
`static cells were cultured alone or with the fibro—
`
`blasts. Growth and invasive properties of the cancer
`cells were evaluated as was their production of gela-
`tinase activity. Colon carcinoma cells grew on
`
`

`

`213
`
`100
`
`i9 80
`G
`.94.;
`---I
`:9
`.5!
`C.H
`
`60
`
`,c:
`
`40
`
`20
`
`0
`
`OH o
`
`S. C.
`tumor
`
`Cecum
`tumor
`
`Liver
`tumor
`
`Fig. 2. Response of human colon carcinoma cells growing in dif-
`ferent organs to DXR. Cells were injected s.c. into the spleen in
`order to produce liver metastases or into the cceal wall of nude
`mice. DXR was given on days 7 and 16 at 10 mgr‘kg. Mice with s.c.
`and liver tumors were killed 22—28 days after tumor cell injec—
`lion, and mice with cecal tumors 28—35 days after. 'Ilimors were
`weighed: livers with colon cancer metastases were weighed and
`the average weight of normal livers was subtracted to derive tu—
`mor weight. The data shown are the mean inhibition of tumor
`growth i SEM (10 micelgroup).
`
`mors growing so were most sensitive to DXR (Fig.
`1), while tumors growing in the liver were least sen-
`sitive (Fig. 2). The differences observed in viva
`were not evident in cultures established in vino. Af-
`
`ter 1w2 weeks in culture, cells derived from untreat-
`ed s.c. tumors and liver tumors were as sensitive to
`
`DXR in vitro as the parental KM12L4 cells [7’6].
`These data suggest that the organ—specific differ-
`ences in DXR sensitivity that we observed under in
`vivo conditions were not due to selection of differ—
`
`ent cell populations but to environmental factors
`that endowed tumor cells with certain properties
`that enhanced their resistance to systemic therapy
`[76].
`The distribution of DXR was lowest in tumors
`
`
`
`l
`
`1
`.‘l'
`
`20
`
`15
`
`10
`
`s
`
`c
`
`Eg
`
`-
`1)
`
`fl,
`5
`E
`:r
`[—l
`
`o
`
`slid 15+
`
`20
`
`25
`
`so
`
`35
`
`40
`
`Day
`
`Fig.1 Response of human colon carcinoma cells growing in the
`nude mouse s.e. tissue to DXR. Cells were injected in to groups of
`nude mice (n ; 10). DXR was given iv on days ”I and '16 at a dose
`olll) mgfkg (arrows). Tumors were measured every other day in
`2 diameters, and the average was taken. Control mice (0);
`DXR—treated mice (I). * P < 0.05.
`
`monolayers of all 3 fibroblast cultures but did not
`invade through skin fibroblasts. Cancer cells grow-
`ing on plastic and on colon or lung fibroblasts pro-
`duced significant levels of latent and active forms of
`type IV collagenase, whereas colon carcinoma cells
`cocultivated with nude mouse skin fibroblasts did
`not. Incubation of human colon carcinoma cells in
`
`serum—free medium containing recombinant hu-
`man interleron-B (fibroblast
`interferon) signifi-
`cantly reduced gelatinase activity [73]. These in vit—
`ro data support the in vivo data that organ-specific
`factors can influence the invasive and metastatic
`
`properties of cancer cells.
`In most patients with colon cancer, metastasis to
`regional lymph nodes or the liver is likely to have
`occurred prior to diagnosis and surgical resection of
`the primary tumor [74]. Thus, prognosis forpatients
`with advanced disease with metastases to the liver
`
`and the lungs is poor. Indeed, many chemother-
`apeutic drugs and drug combinations have pro-
`duced only marginal results [74, 75].
`Since we have been interested in obtaining a bet—
`ter understanding of the biology of colon carcinoma
`metastasis, We wished to determine whether the or-
`
`gan microenvironment could influence the re-
`
`sponse of human colon carcinoma to systemic ther-
`apy with DXR. Highly metastatic human colon can
`cer KM12L4 cells previously selected for produc—
`tion of
`liver metastases in nude mice, were
`
`implanted into 3 different organ sites of nude mice:
`the so space, the cecal wall, and the liver [76]. Tu—
`
`growing s.c., followed by tumors growing in the cc—
`cum and the liver; therefore, it cannot explain the in
`
`

`

`214
`
`Table 1'. DXR distribution in organs of nude mouse and in KM12L4 tumors
`
`DXR concentration ([1ng tissue)
`
`Normal organs
`Tumors
`
`
`Plasmaa
`Cccum
`Liver
`Skin
`
`a jig DXRi'ml.
`
`0.1 i 0.0
`0.3 i 0.1
`0.6 i 0.3
`2.6 i 1.0
`5.2 i 2.0
`4.2 t 0.1
`3.3 451.1
`5.1 $0.6
`7.3:r0.4
`2.2 i 0.1
`1.0 i 0.0
`0.9 i 0.2
`
`
`—
`—
`—
`1.5 i 0.2
`2.6 i 1.1
`2.4 : 0.7
`
`1.9.+_0.4
`2.6104
`3.4 :: 0.4
`1.2 i 0.2
`1.7 :1: 0.3
`0.9 1r 0.1
`
`
`vivo differences in DXR sensitivity (Table 1). PKC
`
`for the differences observed in DXR sensitivity. It is
`
`activity was highest in samples of so origin, w_here
`tumors were most sensitive to DXR, and signifi-
`cantly lower in samples from the liver and the ce—
`cum. Downregulation of PKC has been reported,
`especially in cell systems with high intrinsic PKC ac—
`tivity [46, 77—?9]. In tumors of the colon, PKC activ-
`ity has been found to be downregulated compared
`to adjacent mucosa [79]. We suggest that downreg-
`ulation of PKC had occurred in KM12114 cells grow—
`ing in the liver or the cecum and that biliary acids
`may have contributed to downregulation at these
`
`organ sites [46, 80].
`P-glycoprotein is a transmembrane transport
`protein that mediates the efflux of naturally occur-
`ring toxic products through an active transport
`mechanism [33—39, 80—84]. The protein is physio-
`logically expressed in cells of a variety of human tis-
`
`sues including cells of the proximal tubules, the lu-
`minal surface of the colon mucosa, and the biliary
`canalicular surface of hepatocytes [85, 86]. Its ex-
`pression in these excretory organs suggests that P-
`glyeoprotcin plays a physiological role in cell clear-
`ance of extrinsic or intrinsic toxic products. Human
`
`tumors originating from these organ sites usually
`exhibit high levels of P—glycoprotein or its mRNA
`[82, til—91], indicating that the signal for P—glyco—
`protein expression can be maintained during neo-
`plastic transformation [90]. We analyzed the ex—
`pression of P—glycoprotein in KM12L4 cells harvest—
`ed from tumors growing in different organs by flow
`cytometric analyses (Fig. 3). The data show that P-
`glycoprotein expression was significantly higher in
`cells harvested from liver and cecum tumors than in
`
`cells harvested from tumors growing so and cells
`
`growing in culture. This may account at least in part
`
`of special interest that P—glycoprotein was elevated
`in organs with physiologic P-glycoprotein expres-
`sion and also in organs that are physiologically ex-
`posed to biliary acids. When cells were harvested
`from a liver metastasis and maintained under cul-
`
`ture conditions, expression of P—glycoprotein grad~
`ually diminished to the level of parental cultured
`KM12L4 cells (Fig. 4). This suggests that overex—
`
`prcssion of P—glycoprotein in cells from tumors
`growing in the liver or in the cecal wall was tran—
`
`sient, lasting only as long as they remained in the
`liver or cecum [76].
`Functional P—glycoprotein is the product of the
`mdrl and mdr3 genes. Expression of the mdrl gene
`is associated with drug resistance. The antibody
`C219 recognizes a small, highly conserved epitopc
`
`Cultured KM I 3 IA
`
`Subculis Tumor
`
`Normal Liver
`Liver Mctaslasls
`
`Normal Colon
`(Decal Tumor
`
`
`
`.
`
`.
`20
`
`0
`
`-.— -----
`40
`
`.
`60
`
`‘ * ‘
`
`‘ i
`
`.
`30
`
`[00
`
`Relative Fluorescence Unit
`tP-gJyooprotetn-borund ETC-(1219!
`
`Fig. 3’. Differential FITC-C219 antibody binding to P—glycopro-
`tein of cultured human colon carcinoma cells and of those cells
`
`grown in various environments. Cells derived from tissue cul—
`ture. normal organs, and tumor samples were reacted with FITC—
`C219. The cell—bound FI'I‘C fluorescence was analyzed by a com—
`puter and the specific expression of P-giycoprotein was present—
`ed as relative fluorescence units (RFU) normalized with the in‘
`dividual negative antibody fluorescence profile. The values are
`mean i SEM of 3 experiments.
`
`

`

`
`
`215
`
`FluorescenceUnit
`
`Relative
`{p-gtvcnpmtein-haunasire-0219)
`
`
`0
`
`7
`
`14
`
`21
`
`28
`
`35
`
`‘b
`a“
`&’\b
`$0
`o
`4"
`.
`,5.
`$99$o§
`
`ABC D
`
`Fig. 5. Northern blot analyses of mdrl and mdr3 expression in
`human colon carcinoma cells growing in viva or in virro. Poly
`(A+) mRNA (5 ugflane) was used in all cases. The probes used
`are described in Materials and methods. Lane A, culture cells;
`lane 13. subcutaneous tumor; lane C. liver tumor; lane 11 cecal
`wall tumor grown in vitro 7 days. Densitometric quantitation is
`shown below each blot, where mdrl or mdr3 mRNA expression
`was normalized to GAPDH mRNA levels, and in each case, the
`culture cells defined as 1.0.
`
`Days Cultured on Plastic
`
`Fig. 4. Time course of FITC—C‘ZIQ antibody binding of human
`colon carcinoma cells harvested from a liver metastasis grown in
`a nude mouse after reestablishing in in vitro culture conditions.
`
`in the cytoplasmic domain of P-glycoprotein, ac-
`counting for high species cross-reactivity. Northern
`blot analyses demonstrated expression levels of
`both mdfl and mdr3 that correlated with DXR sen-
`
`sitivity and immunohistochemical and FACS data
`(Fig. 5 ). Thus, we demonstrated that DXR sensitiv-
`ity in human colon cancer cells growing in different
`organ sites in nude mice was modulated by the or-
`gan environments. Colon cancer cells growing in
`the liver and the ceeum were less sensitive to DXR
`
`than cells growing s.c., perhaps because of transient
`overexpression of P—glycoprotein. The significance
`of downregulation of PKC activity is not yet known.
`
`Murine colon carcinoma
`
`In the next set of experiments, we implanted mu-
`rine CT-Zé colon carcinoma cells into different or-
`
`gans of syngeneie BALBi'c mice [91]. We then deter-
`mined whether the organ microenvironment could
`influence the response of murine colon carcinoma
`cells to systemic therapy with DXR or S—FU. We
`found differences in sensitivity of murine colon can—
`cer cells growing in the subcutis, spleen, liver, and
`lung to DXR (a cytotoxic agent affected by the
`M I) R phenotype) and S-FU (a compound unaffect—
`ed by MDR), two structurally and pharmacologi-
`cally distinct cytotoxic agents. The sensitivity of the
`f"1126 cells to DXR was highest in the so. environ—
`ment, intermediate in the spleen and cecum, and
`
`

`

`216
`
`lowest at metastatic sites such as the liver and lungs,
`whereas their sensitivity to S—FU was highest in the
`lung and intermediate in the subcutis, the spleen,
`and the cccum, and lowest in the liver. Once again,
`organ—site—associated differences in drug sensitivity
`to either DXR or S-FU were not associated with
`
`drug distribution patterns in the tumors.
`The intrinsic resistance of tumor cells to chemo—
`
`therapeutic drugs can be mediated by both genetic
`and epigenetic mechanisms. The former can devel—
`op through gene amplification (notably the mdrl
`gene), gene rearrangements, and transcriptional,
`translational, and posttranslational events [80, 92].
`Phenotypic changes that can modulate drug resist-
`ance are often associated with increased activities
`
`of drug-detoxifying enzymes [93, 94], metabolism-
`regulating enzymes [95], and efflux proteins [80, 92,
`96]. In addition, increased levels of PKC activity
`have been observed in several MDR tumor cell
`
`lines [46]. Similar effects have also been associated
`with decreased activity of specific enzymes or influx
`of transmembrane proteins and production of al—
`tered enzymes with decreased affinity for a given
`drug [97].
`Most of the above data have been derived from
`
`examining tumor cells growing in culture; however,
`the relevance of culture conditions to the situation
`
`in vivo is uncertain. We have previously reported
`that the mouse UV—223? fibrosarcoma and human
`
`KM12 colon carcinoma [98] exhibit different pat-
`terns of chemosensitivity when growing in different
`
`organs of nude mice, patterns that neither correlat-
`ed with nor were predicted from in vitro cultures.
`The present data extend these observations.
`In some experimental systems, anthracycline an—
`tibiotics such as DXR have been shown to be more
`
`effective under hypoxic conditions, which support
`the formation of free radicals [20, 22]. The DXR—
`free radical can intercalate with DNA and promote
`
`oxidation of a variety of intracellular components
`[22]. In contrast, cytotoxicity mediated by SHFU is
`not subject to intracellular redox regulation [59,
`99]. Tumors growing s.c. may be less oxygenated
`than those growing in the lungs, and this may, in
`part, explain the sensitivity of CT-26 tumors to
`DXR when implanted s.c., its resistance to DXR
`
`when growing in the lung, and its sensitivity to S-FU
`in the same organ.
`Drug metabolism may contribute to tumor re-
`sponse in different organs. The catabolic inactiva-
`tion of 5~FU occurs mainly in the liver, an organ
`with intense dihydronracil dchydrogenasc activity,
`which degrades S-FU to dihydro—S—FU [100, 101].
`From drug distribution analyses published previ—
`ously and the present data, we concluded that the
`
`sensitivity of s.c. tumors to chemotherapy is not due
`to increased accumulation of these agents. More—
`over, the S-FU sensitivity of CT—26 s.c. tumors was
`independent of tumor size and time of initial treat—
`ment. This was not the case for DXR, in which
`
`larger s.c. tumors (6 mm) containing some necrotic
`zones were more sensitive to therapy than smaller
`s.c. tumors (02 to 1.5 mm). These data suggest that
`the sensitivity of CT-26 tumors to 5—H] and DXR
`did not directly correlate with the degree of vascu-
`larization.
`
`The level of PKC activity has been directly corre-
`lated with resistance of murine UV—2237 fibrosar—
`
`coma cells to DXR, especially of those with the
`MDR phenotype [46] PKC is involved in signal
`transduction of hormones and growth factors, and
`its activity can be stimulated by a variety of tumor
`promoters that include biliary acids [102]. Biliary
`acids have in turn been implicated as promoters for
`
`colon and liver cancers [46, 102]. Since both colon
`and hepatocellular carcinomas demonstrate high
`levels of resistance to many chemotherapeutic
`
`drugs, we wished to determine whether the level of
`PKC activity was elevated in CT—26 tumors growing
`in organs exposed to biliary acids, e.g., colon and
`liver. We found that PKC activity was low in s.c.
`CT—26 tumors, higher in cecal tumors, and highest in
`liver and spleen tumors. The high level PKC activity
`in the spleen was unexpected, and its significance is
`not clearly understood.
`
`Organ-specific modulation of steady-state mdrl
`gene expression in marine colon cancer cells
`
`Most human and rodent neOplasms display hetero-
`geneity for many properties, including sensitivity to
`anticancer drugs or biologicals, and many colon
`
`

`

`cancer cells exhibit an intrinsic MDR phenotype
`
`the extracellular matrix can stimulate expression of
`
`[103—106]. The elevated expression of mdrl mRNA
`and P-gp in C1126 cells growing in the lung could
`therefore have been due to the selection of resistant
`
`the mdrl gene. Hepatocytes growing in vitro on col—
`lagen type 1 demonstrate high levels of mdrl gene
`and increased resistance to DXR, decreased DXR
`
`217
`
`variant cells. Several lines of evidence, however,
`
`suggest that the increased resistance to DXR in the
`C1126 cells in lung metastases was not due to selec—
`tion of resistant subpopulations. First, unlike most
`tumor cells selected in vitro for the MDR pheno—
`type by exposure to anticancer drugs [107, 108], the
`CT—26 cells growing in the lung did not contain am-
`
`plified mdrl. Second, once implanted into the sub-
`cutis of syngeneic mice, CT-26 cells from lung me-
`tastases produced tumors that were sensitive to
`DXR.
`In parallel studies, DXR-sensitive (71126
`cells from so tumors became resistant to the drug
`when they were inoculated i.v. and grew in the lung
`parenchyma as metastases. Steady—state mdrl
`mRNA levels directly correlated with the drug re—
`sistance phenotype in these experiments. Third, the
`increased resistance to DXR and elevated levels of
`
`mdrl mRNA and P-gp were all transient in CT—26
`growing in the lung. During growth in culture for
`> 7 days, mdrl mRNA and P-gp reverted to the
`baseline levels of CT—26 parental cells.
`The exact mechanism by which the organ envi-
`ronment, e.g., lung, regulates the expression of th