Cancer and Metastasis Reviews 10: 201-215, 1991.
`
`© 1991 Kluwer Academic Pabiirhers. Printed in the Netherlands.
`
`Importance of orthotopic transplantation procedures in assessing the effects
`of transfected genes on human tumor growth and metastasis
`
`R.S. Kerbel, Isabelle Cornil and Dan Theodorescu
`
`Division of Cancer Research, Sunnybrook Health Science Centre, Toronto, Ontario, Canada
`Departments of Medical Biophysics, di Medical and Molecular Genetics, University of Toronto
`
`Key words: tumorigenicity, metastasis, oncogenes, suppressor genes
`
`Abstract
`
`Assessment of the function of putative dominantly-acting oncogenes or recessive tumor-suppressor genes in
`
`human tumor development and progression must ultimately involve xenografting experiments using immune
`deficient animals such as nude mice. Most human tumor xenograft experiments have employed conventional
`subcutaneous injection procedures. However, despite the simplicity of this procedure, it poses some serious
`potential drawbacks as most types of human tumor will not readily grow or metastasize from a subcutaneous
`(‘ectopic’) site of injection. In contrast, ‘orthotopic’ injection procedures will often enhance the tumorigenic
`andior metastatic ability of tumor cell populations. An example of this is summarized in the context of human
`malignant melanoma where the effects of subcutaneous versus subdermal injection are compared. Despite
`the seeming subtle and minor change in injection site, superior growth of human melanomas can be obtained
`by the latter, orthotopic-like, route of injection.
`It therefore follows that induction of tumorigenic or metastatic properties in a given human cell population
`by gene transfection may not be detected if the transfected cells are assayed in vivo only by subcutaneous
`injection procedures. An example of this is provided by experiments involving transfection of normal or
`mutated ras genes into a low-grade, well-differentiated human bladder carcinoma cell line, called RT—4.
`Thus overexpression of normal or mutated (valine 12) c-H-ras resulted in acquisition of a clinical-like
`
`invasive phenotype. However, this was clearly seen only if the cells were injected into the bladders (i.e.
`‘intravesically’) of nude mice. In contrast, conventional subcutaneous injection of the high ras expressing
`transiected RT—4 cell lines did not reveal acquisition of invasive properties: all cell lines grew locally as
`well-encapsulated tumor masses.
`It is argued that similar orthotopic injection procedures should be employed when assessing the suppres-
`sive effects of various wild-type tumor-suppressor genes on human tumor growth in vivo. Utilization of
`subcutaneous injection procedures may grossly exaggerate the growth suppressive effects of such genes. This
`could explain the paradox of why, on the one hand, alterations involving many different genes (including
`different suppressor genes) appear to be involved in human carcinoma tumorigenesis while on the other
`hand, complete suppression of tumorigenicity can be caused by transfer of a single wild-type suppressor
`gene. Such complete suppressions might be observed only after ectopic (usually subcutaneous) injection
`procedures.
`
`Introduction
`
`niques have uncovered a large arsenal of genes
`which are putatively involved in the development
`
`Recombinant DNA based molecular cloning tech-
`
`Genentech 2087
`
`Celltrion v. Genentech
`
`|PR2017-01122
`
`Genentech 2087
`Celltrion v. Genentech
`IPR2017-01122
`
`

`

`202
`
`of various types of cancer and malignant tumor
`progression. Most prominent among these are the
`so-called dominantly acting oncogenes and reces-
`sive ‘tumor-suppressor’ genes [1—3]. The list of
`genes which have been cloned thought to be in-
`volved in cellular functions relevant to cell invasion
`
`and metastasis (i.e. ‘malignancy‘) is also growing at
`a remarkable rate [4—9]. After such genes are iden-
`tified and mquenced their precise contributions to
`
`tumor growth and malignancy must ultimately be
`assessed by their introduction into suitable recip-
`ient cells followed by assay of tumor cell function in
`viva. For this purpose a variety of ingenious mam-
`
`malian expression vectors have been designed to
`facilitate efficient expression of transferred cloned
`gene sequences. Both plasmid based and virus (es-
`
`pecially retrovirus) based vectors have been Suc-
`cessfully used in this respect (e.g. [10]).
`Unfortunately the sophistication and ingenuity
`often associated with such recombinant DNA tech-
`
`niques is frequently compromised or negated by
`the inappropriate application of cell biological
`methods to assay in vivo the properties of trans-
`fected cell populations. For example, a survey of
`
`the literature reveals that tumor cell populations
`used in gene transfection studies are almost always
`injected subcutaneously. However, the vast major-
`ity of (non-skin) cancers neither arise in a sub-
`cutaneous location nor form metastases in this or-
`
`gan. Consequently such an ‘ectopic‘ injection of
`tumor cells will not necessarily lead to the forma-
`tion of palpable, progressively growing tumors.
`Moreover, even if such tumors do arise they fre-
`quently fail to manifest a high~grade malignant
`phenotype, i.c. they fail to form distant metastases.
`This problem may be exacerbated when one stud-
`
`ies tumor cells of human origin injected into athym-
`ic nude mice: the species difference adds further
`physiologic complexities which may have to be
`overcome to permit significant tumor take rates.
`Consider also the fact that the majority of human
`cancer surgical (biopsy) specimens will not grow
`when injected subcutaneously into nude mice [11].
`Yet most such cancers may contain a variety of
`mutations in both dominant oncogenes and tumor
`suppressor genes (e.g. [12]). Why should one ex-
`pect, therefore, that transfection of a particular
`
`dominant oncogene, or a mutant suppressor gene,
`into (for example) an immortalized but non-tumo-
`rigenic epithelial cell line will necessarily lead to its
`tumorigenic conversion? Put in another way, sub-
`cutaneous (ectopic) injection of the cells will prob-
`ably not lead to any demonstrable tumor growth,
`and from this it may be falsely concluded that the
`transfected gene is not involved in carcinoma de-
`
`transfection of a gene
`velopment. Similarly,
`thought to contribute to acquisition of invasive or
`metastatic competence may lead to a ‘false-nega-
`tive’ result if the transfected tumor cells are in-
`
`jected in a manner which effectively precludes
`metastatically-competent cells from actually me-
`tastasizing. Finally, inaccurate conclusions regard-
`ing the function of putative suppressor genes may
`likewise be obtained if inappropriate injection
`techniques are used. For example, subcutaneous
`injection of an epithelial tumor cell population
`transfected with and expressing a wild-type sup-
`pressor gene (such as p53) may lead to total or
`drastic suppression of tumor growth. But the ex-
`tent of the effect may be a consequence of the tissue
`environment in which the cells were placed: in-
`jection into a more appropriate i.e. ‘orthotopic’,
`environment might provide an altogether different
`result, e.g. partial suppression only or perhaps
`even none at all.*
`
`These speculations owe their origin to the dis-
`covery that orthotOpic injection of tumors (i.e. in-
`jection of tumors into or adjacent to their tissue of
`origin) frequently enhances their tumorigenic and:f
`or malignant properties [13—19]. This is especially
`evident in studies of human tumors in immuno-
`
`suppressed mice such as athymic nude mice [13—
`15]. Thus human retinoblastomas usually do not
`grow when injected subcutaneously into nude mice
`but may do so when injected intravocularly [20,
`21].“ Similarly, human lung cancers rarely grow
`after subcutaneous injection but frequently do SO
`after intrabronchial injection [22]. Similarly, sub-
`cutaneous injection of human advanced-stage col-
`orectal carcinoma cells into nude mice often results
`
`" Note added in proof See paper by H] . Xu at at, Intraocular
`tumor formation of RB reconstituted retinoblastoma cells. Can—
`cer Research, 51: 4481—4485, 1991.
`
`

`

`in delectable tumor growth, but with no evidence
`of liver metastases [23]. In contrast, intra-cecal
`injection of the same cells can result in the forma-
`tion of liver metastases [23—25]. From these results
`
`it would be expected that tumorigenic transforma-
`tion of retinoblasts or lung epithelial cells by trans-
`fection of a particular gene, or combination of
`genes, may go undetected if the cells are injected
`subcutaneously as opposed to intraocularly or in-
`trabronchially, respectively. Similarly, conversion
`of an early-stage Duke’s B colorectal carcinoma to
`the more advancedfmalignant Duke‘s D ‘equiv-
`alent’ by transfection of a particular gene may also
`be missed unless the cells are injected into an ap-
`propriate orthotopic site, e. g. the caecum.
`The overall purpose of this review is to summa-
`rize recent work from our laboratory which pro-
`vides an interesting and new example of the growth
`enhancing effects of orthotopic tranSplantation of
`human cancer into nude mice, specifically human
`malignant melanoma [26]. We will then review
`work [27] which shows how implementation of or-
`thotopic transplantation techniques may be crucial
`in unmasking the function of transfected Onco-
`genes to tumor progression (in this case bladder
`cancer) and the acquisition of an invasive pheno-
`type.
`
`Orthotopic transplantation of human tumors
`in nude mice: impact on the growth and behaviour
`
`of malignant melanoma cells
`
`Giovanella and his colleagues were the first to re-
`port
`that human melanoma cell
`lines could be
`
`grown successfully in athymic nude mice [28, 29].
`Since then a large number of reports have appeared
`in the literature on the growth of such human mela-
`
`noma xenografts, as reviewed recently by Pawlow-
`ski and Lea [30]. Human melanoma, especially cell
`lines, are distinguished as being among the easiest
`of all human tumors to grow in nude mice [11]; only
`human colorectal carcinomas have a similar success
`
`rate [11, 25, 30, 31]. Unlike other types of human
`cancer
`transplanted subcutaneously, malignant
`melanomas will occasionally metastasize in nude
`mice, usually to the lungs or lymph nodes [30].
`
`203
`
`Variants having increased metastatic aggressive-
`ness can be obtained by a variety of methods, in-
`cluding by selection in viva [32—34]. In addition,
`reports appear from time to time documenting the
`
`finding of melanoma cell lines which are unusually
`aggressively metastatic in nude mice (e.g. [35— 3?]).
`Such cell lines have proven to be the exception
`rather than the rule.
`
`An extensive effort was made in our laboratory
`to isolate variants of human melanomas which
`
`would not only readily metastasize in nude mice
`but do so in a ‘clinically relevant’ fashion character-
`istic of malignant melanoma. Thus we sought to
`develop methods which could facilitate, for exam-
`
`ple, the appearance of skin and brain metastases —
`both being clinical hallmarks of advanced meta-
`
`static melanoma in man [38]. Using somewhat
`cumbersome protocols which select in vitro for so-
`called ‘lectin—resistant’ glycosylation mutants, we
`were indeed able to select for a skin andlor brain
`
`metastasizing variant (called 70-W) from a human
`melanoma cell line called MeWo [39, 40]. We were
`also able to isolate MeWo variants (such as one
`called 355) which were virtually devoid of meta-
`static competence [41]. However the results ob-
`tained with lfl-W, encouraging as they first ap-
`peared to he, were obtained only after intravenous
`inoculation of the cells [39, 40}: subcutaneous in-
`oculation of these cells showed that their ability to
`form ‘spontaneous’ lung (or other organ) metasta-
`
`ses was actually suppressed in comparison to the
`parental MeWo population [26].
`At this point we decided to evaluate the effects of
`intradermal inoculation of human melanoma cells
`
`on their relative capacity to grow and metastasize.
`The rationale for doing so stems from the fact that
`melanocytes are found in the basal layer of the
`epidermis, surrounded by keratinocytes. Thus an
`intra-epidermal or intraderrnal injection of melan-
`oma cells would supposedly constitute a more ap-
`propriate (Le. orthotopic) site than would a con-
`ventional subcutaneous injection. We reasoned
`that, based on the growth- or metastasis—enhancing
`effects of orthotopic injection in other tumor sys-
`tems (as summarized above) a similar effect might
`be obtained with melanoma. We initiated these
`studies with the MeWo cell line and the “I'D-W and
`
`

`

`204
`
`
`
`Fig. I. Histological appearance of MeWo human melanoma
`injected subcutaneously into an athymic nude mouse. Note
`location of tumor in deep subcutaneous tissue below subcutane-
`ous muscle. Hematoxylin and eosin stain (H+ E); original
`magnification, X 18
`
`385 lectin-resistant variants. The MeWo cell line
`
`was originally established in culture from a pig-
`mented malignant
`lymph node metastasis [31].
`Subcutaneous injection of these cells into 6—12
`week old nude mice resulted in slightly melanotic
`
`tumors which overtly metastasized to the lungs af-
`ter a long period but which rarely metastasized to
`lymph nodes. Thus some of the properties manifes-
`ted in the autochthonOus host appeared to have
`been lost with the establishment of MeWo cells in
`
`culture and their growth in nude mice. Moreover, a
`total of between 300,000 and 500,000 cells was
`
`required to give a 100% tumor take rate [26, 32],
`Le. it was ‘moderately’ tumorigenic in nude mice.
`Because nude mouse epidermis and dermis are
`so thin, it was not possible to directly deposit a
`bolus of cells intradermally. We found instead that
`
`the cells were actually deposited in the most super-
`ficial layer of the subcutis (i.e. ‘subdermally‘). The
`subdermal injections were performed with a 305’;
`gauge needle in ether-anesthetized nude mice.
`Variable numbers of cells were injected into the
`
`mouse subdermis of the middle of the lateral right
`flank. The needle was carefully introduced as su-
`perficially as possible into the mouse skin so that
`the bevel of the needle was visible by transparency
`through the first skin layer [26]. We injected varia—
`ble numbers of cells into the mouse subcutaneous
`
`fascia of the middle of the lateral right flank [26].
`
`The animals used for these particular experiments
`were female NIH Swiss athymic nude mice at 6—10
`weeks of age [26].
`As shown in Tables 1 and 2, the results we ob-
`
`tained in respect to both tumorigenicity and metas-
`tasis were striking. Thus as few as 20,000 cells was
`all that was required to give a 100% take rate after
`
`subdermal inoculation — highly unusual for a tumor
`
`xenograft. In contrast, 500,000 cells were required
`to give a 100% take rate after subcutaneous in-
`
`oculation (Table 1). Moreover the tumors could be
`
`easily visualized early in the subdermal location as
`
`they were deeply pigmented, in striking contrast to
`the ‘deep‘ subcutaneous counterpart [26]. Visible
`tumors appeared approximately seven days after
`subdermal
`injection of 5 X 105 cells and three
`weeks after 2 X 10‘1 cells were injected subcutane-
`ously; it took three weeks for tumors to first appear
`after subcutaneous injection of 5 X 105 cells [26].
`Three months after injection,
`the mice were
`killed and their organs were examined for macro-
`
`scopic metastases. In the mice with MeWo tumors
`implanted subdermally, lymph node involvement
`
`and lung metastases Were significant (Table 2), but
`the 355 cells gave rise to very few metastases even
`
`after subdermal injection. This is not surprising
`and is actually in accord with the deficiency in
`organ-colonizing ability observed with 385 cells af-
`ter intravenous inoculation [39—41]. The 70-W vari-
`ant, on the other hand, was selected for its greater
`ability to colonize multiple organs (e. g. lungs, skin
`and brain) [39, 40]. Despite this more aggressive
`
`Table l. Comparison of tumorigenicity of MeWo human melan—
`oma cells in nude mice after subcutaneous or subdermal in-
`
`jection
`
`Cell number injected
`
`Frequency of Tumor takes"
`
`Subcutaneous
`Subdermal
`
`
`5 X 105
`3 X 10‘
`105
`5 X 10“
`2 X 10"
`
`SIS
`BIS
`215
`05
`04"5
`
`515
`SIS
`515
`SIS
`5:"5
`
`* Number of mice injectedinumber of mice with progressively
`growing tumors at site of implantation.
`
`

`

`colonizing potential, 70-W cells had no significant
`metastatic ability three months following subcuta-
`neous injection. Thus the properties of intense
`melanin production and capacity for lymph node
`metastasis were not lost by the MeWo cell line, as
`we had originally thought. Rather they lay dor-
`mam, ready to be ‘resurrected’ provided the cells
`were placed into an appropriate tissue environ—
`ment.
`
`The histologic characteristics of subcutaneous or
`subdermal implanted tumors Were found to be
`quite different, as described previously [26]. These
`characteristics were studied in sections of tissue
`
`stained with hematoxylin and eosin, fixed in forma-
`lin, and embedded in paraffin. The subcutaneous
`tumor was established below the subcutaneous
`
`muscle into the fascia in the deepest layers of the
`subcutaneous tissues (Fig. 1). The margins of the
`small
`tumor mass were circumScribed by con-
`
`densed fibroblasts and occasional inflammatory
`cells, which formed a pseudocapsule [26]. Masson-
`Fontana staining revealed small, dark areas of mel-
`anin production centrally located in the tumor mass
`[26]. As the tumor size increased, the deep sub-
`cutaneous tumor disrupted the subcutaneous mus-
`cle in which it was still contained and reached the
`
`subcutaneous fat tissue above the muscle, where
`
`we noted the absence of the fibrous sheath [26].
`We observed that the subdermally implanted tu~
`
`205
`
`
`
`Fig. 2. Extension of subderrnal MeWo tumor abutting on over-
`lying epidermis. H + E; original magnification, x 25 (taken
`from reference [26]).
`
`mors had been deposited in the most superficial
`portion of the subcutaneOus tissue just below the
`
`tumors grew, they expanded downward into the
`superficial subcutaneous tiSsue; they also readily
`
`dermis (Fig. 2). No fibrous capsule was detected at
`the tumor periphery. Masson-Fontana staining re-
`vealed large areas of pigmented cells diffusely dis-
`tributed throughout the tumor masses [26]. As the
`
`grew upward into the dermis, where they abutted
`the overlying surface of the squamous epithelium
`[26]. Finally, they penetrated the epithelium, form-
`ing an ulcer. Thus, these tumors were initially sub-
`
`Table 2. Metastatic potential of MeWo, ill-W and 355 cells after subcutaneous or subdermal injection in nude mice’
`
`Cell line
`
`Route of Injection
`
`Lung Metastases
`Number of mice with lymph node
`involvement
`Number of nodules
`
`
`Number of mice
`
`MeWo
`
`TO—W
`
`s.c.
`s.d.
`s.c.
`
`335
`
`s.d.
`s.c.
`s.d.
`
`
`1
`7
`0
`
`0
`l
`2
`
`sc = subcutaneous; sd = subdermal; taken from Cornil er a1. [26]
`
`5
`8
`0
`
`1,2.6,8,18
`1,4,9.l3.15.16,26.60
`0
`
`1.5.10
`3
`0
`0
`1.2
`2
`_—_—_—“
`
`

`

`206
`
`
`
`Tumormiflihtor-ml)
`
`0.500
`am
`
`04300
`0.200
`
`0.!00
`0.000
`
`
`
`
`
`Growth curves In nude mice of human melanomas
`
`from advanced lMET+l or early {MET‘)
`stage. of tumor progression
`
`star +
`
`MET"
`
`1.200
`
`neon
`
`0.400 0.000
`
`0.300
`
`WM mlhludv V6?)
`o——o 5.0.
`I—. 8.6.
`
`0.260
`
`0.100
`
`0.000
`
`5.000
`4.000
`3.000
`
`2.000
`L000
`
`0.000
`
`Melts altos Inlocfion
`
`Fig. 3. Comparative growth curves of metastatically incompetent (met ‘) early stage melanomas or metastatically-competent (met +)
`advanced human melanomas after subcutaneous or subdermal injection of 10“ cells into NIH Swiss nude mice. SPl—RAS is a
`ras-transfecled mouse mammary carcinoma used as a negative [non-melanoma) control.
`
`dermal but very quickly grew to become intrader-
`
`competent for metastasis;
`
`their removal,
`
`there-
`
`mal. At equal tumor size, the subdermal tumors
`became superficial and ulcerated more readily than
`the subcutaneous tumors. an observation that is
`
`incidentally associated with a more serious progno-
`sis in ‘vertical growth phase“ melanomas.
`We next asked whether the growth enhancing
`effects of subdermal inoculation could be repro-
`duced by other melanoma cell lines. To answer this
`question we evaluated the growth characteristics of
`ten different independent melanoma cell lines ob-
`tained from different stages of melanoma progres-
`sion. Some of the lines were obtained from the
`
`radial growth (R0?) or early (‘thin‘) vertical
`growth phase (VGP) of primary melanoma pro-
`gression. Such tumors. with few exceptions. are not
`
`fore, usually results in cure of the patient [42, 43].
`In
`contrast,
`thicker VGP primary
`lesions
`(> 0.76 mm) carry a much worse prognosis and are
`generally competent for metastasis as are, by defi-
`nition, distant metastases. Most established human
`melanoma cell lines are derived from the latter two
`
`stages, especially metastases. Advanced VGP pri-
`mary lesions appear to consist predominantly, or
`entirely. of ‘metastatically-competent’ tumor cells
`which are indistinguishable from melanoma cells
`populating distant metastases [44, 45].
`We were fortunate to have acquired a number of
`human melanoma cell lines from these various
`
`stages of progression through the generosity of Dr.
`Meenhard Herlyn of the Wistar Institute Philadel~
`
`

`

`phia [43, 45]. Figure 3 shows some of the results we
`obtained in terms of growth rate. Thus we found
`
`that melanomas grew better after subdermal in-
`jection (compared to subcutaneous injection) —
`provided they were derived from metastatically-
`
`competent primary VGP lesions or from distant
`metastases. In contrast, this growth enhancement
`was not observed when early-stage primary RGP
`or VGP derived cell lines were tested. It would
`
`therefore appear that metastatic melanoma cells
`are able to utilize signals from the dermal me-
`senchyrne in a way that provides them with a
`growth advantage in contrast to their non-meta-
`static counterparts, which appear unable to do so.
`We have recently obtained preliminary evidence
`that dermal fibroblasts secrete a diffusible growth
`
`207
`
`0r melanoma cells may be necessary to uncover
`phenotypic changes relevant to tumorigenicity or
`metastasis induced by certain genetic alterations.
`Unfortunately we cannot provide an example of
`this principle in the context of human melanoma as
`
`no consistent genetic alteration (of a specific onco-
`gene or suppressor gene) has yet been identified
`that is thought to be involved in human melanoma
`
`development or progression. However we have ob-
`tained such an example of this principle in the
`context of human bladder cancer, and this is sum-
`marized below.
`
`Orthotopic injection of res oncogene transfected
`human bladder cancer cells reveals a role for the nut
`
`factor which inhibits the growth of early-stage
`
`gene in the acquisition of an invasive phenotype
`
`metastatically-ineompetent human melanoma cell
`whereas more advanced melanomas are resistant
`
`to this inhibitor [46]. In contrast, another separable
`growth factor is released by dermal fibroblasts
`which stimulates the growth of advanced-stage,
`metastatic melanoma cells while failing to have
`such an effect on early-stage melanoma cells [46].
`Whether these results are relevant to the in vivo
`
`growth results described in Fig. 3 is unclear. But it
`seems reasonable to suggest they help explain how
`the microenvironment may facilitate the over-
`growth of initially rare metastatically-competent
`tumor cell variants within primary tumors, includ-
`ing melanomas — a phenomenon we have termed
`‘clonal dominance’ of primary tumors by meta-
`statically-competent tumor cells [44, 47].
`in summary, our results provide yet another ex-
`ample of the potential importance of employing
`orthotopic transplantation techniques when assess-
`ing the growth and malignant characteristics of hu-
`man tumors in nude mice. The effects can some-
`
`times be quite remarkable even when a seemingly
`subtle change in injection site is involved — in this
`case from deep subcutaneous to subdermal. Fur-
`
`thermore, when it is considered how large a num-
`ber of MeWo melanoma cells (5 x 105) are re-
`quired to give a tumor take after subcutaneous
`injection compared to subdermal injection (2 x
`10‘) it becomes clear why utilization of orthotopic
`transplantation of genetically altered melanocytes
`
`About 80% of bladder cancer patients present with
`a non-invasive (superficial) form of the disease.
`
`Most of these patients are cured by surgical remOv-
`al of the tumor, but a certain proportion — about
`20—25% — have a recurrence of their disease in the
`
`invasive form. The genetic basis of this progres-
`sion, or ‘conversion’, of superficial
`to the life-
`threatening,
`invasive form of bladder cancer is
`
`therefore a subject of considerable clinical impor-
`tance. In this respect several genetic alterations
`have been described which appear to be correlated
`with bladder cancer deve10pment andror progres
`sion. Among these are changes in the expression of
`the ms oncogene (reviewed in [27]) and perhaps
`mutations in the p53 tumor suppressor gene [48].
`Other phenotypic changes have been noted as well,
`such as up-regulation of epidermal growth factor
`(EGF) receptors in invasive bladder cancer (e.g.
`[49—51]).
`How does one go about establishing whether
`such ‘correlations’ are actually involved in the pro-
`gressionlconversion of non—invasive to invasive
`bladder cancer? An obvious answer would be to
`
`transfer mutant ms or p 53 genes into non-invasive
`bladder cancer cells and then determine whether
`
`they then acquire a high grade invasive phenotype
`in viva. However if the transfected tumor cells are
`
`injected subcutaneously into nude mice the mani-
`festation of an invasive phenotype — even if induced
`
`

`

`208
`
`— may not manifest itself. This is because, as dis-
`
`cussed above and elsewhere [13—15, 26] human
`tumors injected subcutaneously into nude mice
`usually grow as well-encapsulated masses with lit-
`tle evidence of local invasion or distant metastases
`
`(see also Fidler, this volume). This is true even for
`tumors which were highly invasivefmetastatic in
`the patient from whom the tumor was originally
`isolated.
`
`A possible solution to this problem in the context
`of bladder cancer was recently provided by Ahrling
`et el'.
`[52]. They found that orthotopic (i.e.
`in-
`travesical) injection of cells from a low grade well-
`differentiated human bladder carcinoma cell line
`
`called RT—4 resulted in locally non-invasive blad-
`der cancers in nude mice, i.e. the cells remained
`
`confined to the bladder mucosa similar to super-
`ficial transitional cell carcinomas [52]. In contrast,
`cells from the human ‘EJ’ bladder carcinoma cell
`
`line — which was obtained from a high grade, in-
`vasive poorly differentiated bladder cancer— were
`found to mimic their clinical invasive behaviour in
`
`nude mice with remarkable fidelity — provided the
`cell's were injected intrevesr'ceit'y into the bladder
`[52]. This leads logically to the following question:
`would transfer of a gene clinically suspected of
`contributing to bladder carcinoma progression into
`RT-4 cells endow the cells with high grade invasive
`(and perhaps metastatic) characteristics? By in-
`jecting the transfected cells orthotopically into
`nude mice, i.e. intravesically, an answer to this
`question might be obtained.
`Below we summarize our attempts to utilize this
`approach in an attempt to determine the role of res
`genes in human bladder tumor progression. First, a
`short summary is provided on res expression and
`bladder cancer as a background to these studies.
`
`Ras gene expression and human biedder cencer
`
`Dominantly acting cellular transforming genes be-
`longing to the res family of oncogenes have now
`been detected in a wide spectrum of animal and
`human cancers by DNA-mediated gene transfer
`experiments in which immortalized non-neoplastic
`cells are used as recipients. By employing such
`
`assays in combination with gene cloning and se-
`quencing analysis, it has been estimated that ap-
`proximately 5—10% of human transitional cell blad-
`
`der cancers (TCC) contain activatedfmutated res
`oncogenes (reviewed in [27]). Moreover, of the
`three known res family members, by far the most
`common found to be mutated in urothelial malig-
`nancies is c-Ha-res [27]. This raises the question of
`whether the presence of activated res oncogenes is
`causally associated with the acquisition of a degree
`of invasiveness of such tumors. The various studies
`
`cited above suggest that based on prevalence com-
`parisons of TCC’s with mutated res and invasive
`
`TCC’s that this is probably not the case. However
`in view of the low frequency of occurrence of acti-
`vated res genes combined with the relatively small
`number of bladder tumors analyzed it is difficult to
`rule out the possibility that patients having tumors
`with an activated res oncogene constitute a distinct
`clinical subgroup of invasive tumors.
`The possible relationship of res gene expression
`to bladder cancer development and progression
`has also been analyzed by immunohistochemical
`techniques (reviewed in [2?]). These studies have,
`in the main, concentrated on estimating the level of
`the res gene protein product, p21, in tumors of
`various stages. The results have shown that in gen-
`eral, there is a correlation between levels of p21
`and the degree of tumor invasiveness similar to
`
`what has been observed in some other types of
`tumor. Detailed staining for p 2] in normal bladder
`tissue has revealed that the basal (progenitor) cells
`of the multilayered transitional epithelium stain
`with the highest intensity while more superficial
`(differentiated) compartments stain to a much less-
`er degree. Thus the level of normal res protein
`diminishes considerably with differentiation and
`
`c-Ha-res over-expression per se is not restricted to
`the malignant state in bladder tissue. It is thus
`conceivable that in the context of malignant dis-
`ease, a deregulation of res gene expression, or
`expression of a mutant protein may occur as a
`“second hit’, and when combined with an earlier
`cellular lesion or lesions, results in the induction of
`invasive bladder cancer. The literature, taken to-
`
`gether, is therefore suggestive of a role for altered
`res gene expression in the progression of human
`
`

`

`bladder cancer, but clearly more direct evidence
`would be required to establish a mechanistic rela-
`tionship.
`
`Orthotopic injection of human bladder cancer cells
`into nude mice
`
`We therefore undertook a series of experiments in
`which RT-4 bladder TCC cells were transfected
`
`with constructs containing the normal form of the
`c—H—ras gene or an activated ms mutated at the
`codon coding the valine 12 position in the protein
`[2?]. We then tested a series of transfectants which
`did, or did not, overexpress the normal or mutated
`forms of ms mRNA [27], as shown in Figure 4.
`
`The oncogene constructs were expressed in the
`Homer 6 vector and contained genomic Ha-ras
`oncogenes generously provided by Dr. N. Wilkie.
`These constructs contain either a 6.4 kb normal
`
`cellular Ha-ras (pH06N1) gene or a 6.6 kb, valine
`12 mutated form (pH06T1), both under the control
`
`of a Moloney LTR promoterienhancer and SV40
`
`enhancer sequences producing a 1.2 kb mRNA
`transcript indistinguishable in size from the endo—
`genously produced transcript. The plasmid DNA
`used for transfection was prepared by the standard
`cesium chloride gradient method followed by
`transfection of the RT—4 cells by the polybrenei’
`DMSO shock method [2?]. Control cell line trans—
`fectants were generated by transfection with the
`plasmid pSVzneo. Selection of colonies which had
`stably integrated the plasmid DNA was done by
`their continued growth for two Weeks in medium
`containing 500 ,ug/ml of G418 [27]. With respect to
`injection procedures, some mice were giVen sub-
`cutaneous injections in the anterior flank alternatev
`ly with 4 x 10‘ or 1 X 10" tumor cells. The animals
`were sacrificed when the skin over the tumor be-
`
`came necrotic, which occurred approximately 3
`months after inoculation. If no tumor appeared,
`the mice were observed for at least 3 months. The
`
`primary tumor, lung, liver, and spleen were hist—
`ologically examined by r0utine hematoxylin-eosin
`staining [2?].
`Other groups of nude mice were injected by the
`intravesical (orthotopic) route. The implantation
`
`209
`
`
`
`o
`1-th1—
`8:..LLLL
`:tlJooEE
`I
`Ilil
`TEESE‘EE
`mccmcemcc
`
`
`
`Fig. 4. Northern analysis for ms and pectin expression of vari-
`ous c— Ha—ras transfected RT-4 cell lines. Total RNA was isolat-
`
`ed from the RT-4 cell clones as previously described. For North-
`ern blot analyses, equal amounts of total RNA (10-15 ,ug} were
`electrophoresed on 1% agarose gels containing 0.66 M formal—
`dehyde and transferred to Gene Screen. Blots were hybridized
`using ’2 P—Iabeled probes of A) c—Ha—ms gene fragment isolated
`after Sac I digest and B) the B—actin gene. Taken from Theodo
`rescu et at. [2?]. ‘ct‘ stands for control ms while ‘mr’ stands for
`mutated ms.
`
`technique used was similar to that of Ahlering er of.
`[27,