Breast Cancer Research and Treatment 39: 69-86, 1996.
`© I996 Kluwer Academic Publishers. Printed in the Netherlands.
`
`Human breast cancer cell line xenografts as models of breast cancer —
`The immunobiologies of recipient mice and the characteristics of several
`tumorigenic cell lines
`
`Robert Clarke
`
`Vincent T. I :moardi Cancer Center, Georgetown University Medical School, Washington, DC, USA
`
`Key words: xenografts, breast cancer, cell lines, resistance
`
`Summary
`
`The ability to maintain and study human tissues in an in vivo environment has proved to be a valuable tool
`in breast cancer research for several decades. The most widely studied tissues have been xenografts of
`established human breast cancer cell
`lines into athymic nude mice. Human breast tumor xenografts
`provide the opportunity to study various important interactions between the tumor and host tissues,
`including endocrinologic, immunologic, and tumor-stroma interactions. The nude mouse is not the only
`immune-deficient recipient system in which to study xenografts. Additional single and combined mutant
`strains have been used successfully,
`including mice homozygous for the severe combined immune
`deficiency mutation (scid), both the beige (bg) and nude (nu) mutations in combination (bg/nu), and mice
`bearing the combined bg/nu/xid mutations. The differing immunobiologies are discussed, with particular
`reference to the immunobiology of breast cancer, as are the characteristics of several of the more
`frequently utilized breast cancer xenografts and cell lines. The ability of several endocrine treatments to
`modulate effectors of cell mediated immunity, e. g., estrogens and antiestrogens, and the effect of site of
`inoculation on tumor take and metastasis, also are described.
`
`Introduction
`
`The incidence of breast cancer mortality has
`continued to rise over
`the past
`thirty years,
`despite the innovation of various cytotoxic and
`endocrine therapies.
`It
`is clearly important
`to
`generate new and innovative therapies based upon
`our developing knowledge of the cellular and
`molecular factors driving breast cancer growth
`and progression.
`In this regard,the application of
`human xenograft models has much to offer both
`
`for increasing our understanding of malignant pro-
`gression, and for the development and screening
`of novel therapies.
`While there are several classes of rodent
`
`models for breast cancer, human tumor xenografts
`
`provide the unique opportunity to study the regu-
`lation of human cell growth and metastasis in an
`in vivo environment. While breast tumor biopsies
`can be established directly in nude mice, the take
`rate is generally low [1]. However, it is clear that
`a significant proportion of the established human
`
`Address for correspondence and oflprints: Robert Clarke. PhD, Vincent T. Lombardi Cancer Center and Department of
`Physiology & Biophysics, W405A The Research Building, Georgetown University Medical School, 3970 Reservoir Road NW.
`Washington DC 20007, U.S.A.; Tel: (202) 687-3755; Fax: (202) 687-7505
`
`Genentech 2053
`
`Hospira v. Genentech
`|PR2017-00737
`
`Genentech 2053
`Hospira v. Genentech
`IPR2017-00737
`
`

`

`70
`
`R Clarke
`
`breast cancer cell lines are tumorigenic in nude
`mice. Several have recently been determined to
`be metastatic in immune-compromised rodents.
`Human breast cancer xenografts have been widely
`used to study breast cancer progression,
`the
`effects of gene expression on tumorigenicity, and
`acquisition of antiestrogen resistance, and to
`screen new endocrine and cytotoxic agents. How-
`ever, a major restriction has been the lack of a
`diverse panel of estrogen receptor (ER) positive
`and hormone-dependent cell lines. There are only
`three major hormone-dependent cell lines avail-
`able, MCF-7 (the most widely used breast cancer
`cell line) [2], T47D [3], and ZR-75-l
`[4]. The
`
`majority of breast cancer cell lines, and almost all
`of those with a significant
`inherent metastatic
`ability,
`as opposed to transfected cell
`line
`variants, are ER negative and hormone-unrespon-
`sive [5].
`
`The development of additional breast cancer
`models is clearly of some importance. However,
`the nude mouse is not
`the only immune-com-
`promised rodent available in which to establish
`xenografts. The additional immune-compromised
`systems will be briefly introduced in the follow-
`ing sections, and an indication of their diverse
`immunobiologies provided. The effects of endo-
`crine manipulations on immunity and their impor-
`tance for the use of breast cancer xenografts
`also will be addressed. Since the immunobiology
`of the rodent systems and the immunobiology of
`breast cancer are central to the use of xenografts,
`a brief discussion of these systems also is in-
`cluded.
`It is hoped that these discussions will
`encourage investigators to broaden the use of
`different
`immune-deficient
`systems,
`and
`to
`establish novel human breast cancer xenograft
`models.
`
`Immunosurveillance of neoplastic cells:
`general comments
`
`9
`
`The precise role of immunosurveillance in breast
`and other cancers is unclear.
`In severely im-
`mune-compromised individuals, e. g., patients with
`
`AIDS, there is evidence for an increased cancer
`risk.
`In normal women, both humoral and cell-
`
`likely contribute to the
`mediated immunities
`suppression of malignancy. The humoral aspects
`are difficult
`to study in experimental animal
`models, since almost all fully immune-competent
`animals reject xenografts. However, the role of
`cell-mediated immunity (CMI) can be modeled in
`vivo, at least partly, by using the diverse immune-
`compromised rodent models described below.
`While this is not the primary use of immune-
`compromised models. an understanding of the
`role of CMI in breast cancer, and of the immuno-
`
`is im-
`biologies of the different animal models,
`portant for the use of xenograft models of breast
`cancer
`
`Immunosurveillance of neoplastic cells:
`breast cancer
`
`Cell-mediated immunity has been implicated in
`the pathogenesis of breast cancer, but its precise
`role remains to be established. For example, the
`skin window procedure, which provides an esti-
`mate of the extent of CMI, correlates inversely
`with metastatic disease [6,7].
`Low levels of
`
`natural killer (NK) cell activity are associated
`with familial breast cancer [8], and with patients
`with stage III/IV disease [9-11]. NK cell activity
`is generally low or absent in the axillary lymph
`nodes of patients with demonstrable metastatic
`disease [12,13], although lymphokine-activated
`killer (LAK) precursor cells are often present
`[13]. There appears to be an inverse relationship
`between ER expression and NK activity [14,15].
`ER-positive tumors have fewer T-cells when com-
`pared «with ER positive tumors [9]. Aminoglu-
`tethimide reduces serum estrogens and increases
`NK activity in breast cancer patients [16].
`The precise role of CMI and its effector cells
`in regulating the growth/tumorigenicity of breast
`cancer is unknown. The use of Matrigel (Colla-
`borative Research Inc., Bedford MA) to increase
`
`take rate [17,18] also suggests that
`xenograft
`an effective barrier could protect cells
`from
`
`

`

`locally infiltrating CMI effectors. The evidence
`from clinical studies implicates CMI effectors, at
`least
`to some degree,
`in the control of breast
`cancer growth. The diversity of immune-deficient
`models cunently available provides a potential
`means to more adequately determine,
`in an ex-
`perimental animal model, the likely contribution
`of some aspects of CMI to the control of neo-
`plasia and metastasis of human breast cancer cells
`in vivo.
`
`Immunosurveillance of neoplastic cells:
`Natural Killer (NK) and Lymphokine-
`Activated Killer (LAK) cells
`
`Both cellular and humoral immunities participate
`in the immunologic response to neoplasia. Most
`of the immune-deficient animal models used in
`breast cancer research lack T-cell-mediated im-
`
`munity. T-cell independent immunosurveillance,
`e.g., aspects of cell-mediated immunity, is thought
`to consist of a number of lymphoid cells, most
`notably NK cells, LAK precursor cells, and
`macrophages [19]. Both NK and LAK cells are
`distinct
`from cytotoxic T-lymphocytes,
`lysing
`cells lacking significant expression of the MHC
`genes. NK and LAK cells can infiltrate solid
`tumors and malignant effusions [20].
`NK cells make up approximately 125% of
`peripheral
`lymphocytes and have been widely
`demonstrated to possess antitumor activity [19].
`Since NK activity may also contribute to control
`of metastasis
`[19,21-23],
`the poor metastatic
`potential of most human xenografts growing in
`nude mice may reflect
`their elevated NK cell
`activities [19,22-24]. Some tumors appear to be
`able to suppress NK activity [25].
`LAK cells are clearly distinct from NK cells,
`a determination initially derived from studies of
`mice bearing different immuno—deficiency muta-
`tions, i.e., nu and bg [26]. LAK cells are capable
`of killing neoplastic cells, and can kill tumor cells
`resistant to NK cytolysis [27]. Some tumors pro-
`duce material capable of blocking the develop-
`ment of LAK cells from their precursors [28].
`
`Xenograft models of breast cancer
`
`71
`
`Immunosurveillance of neoplastic cells:
`macrophages
`
`Macrophages are widely observed to infiltrate
`solid tumors [9,29,30] and can kill tumor cells by
`both phagocytic and non-phagocytic processes
`[30]. Non—phagocytic cytolysis may include the
`release of
`lysosomal enzymes by exocytosis.
`Macrophages may recognize some tumors on the
`basis of their abnormal growth [31] or by surface
`modifications [30] and can produce a non-specific
`cytolysis. The tumoricidal properties of macro-
`phages are acquired following activation by con-
`tact with either the target cell and/or secreted
`products or by soluble lymphokines, e.g.
`inter-
`feron-Y [32].
`The biology of macrophage-induced cytotoxi-
`city is independent of the sensitivity of the target
`cells to lymphocyte or NK mediated cytolysis
`[33]. Tumor cells do not appear to acquire resis-
`tance to the cytotoxic effects of macrophages, in
`marked contrast to their ability to develop resis-
`tance to NK-mediated cytolysis [32-34].
`The
`limiting factor in macrophage control of neoplasia
`appears to be effectorztumor cell ratio [32]. The
`sera from some cancer patients possess macro-
`phage inhibitory activity [35], while some tumors
`secrete a macrophage colony-stimulating-like fac-
`tor [30,36]. Macrophage infiltration is associated
`with tumor progression rather than inhibition, im-
`plying that some macrophages may secrete factors
`mitogenic for tumor cells [37].
`
`Loss of immunosurveillance
`
`Tumors proliferating successfully in the presence
`of cytotoxic host cells clearly indicate that the
`cells have evaded cytolytic effectors. The ability
`to become resistant to immunologic inhibition is
`thus a central problem in cancer biology and im-
`munology.
`The precise mechanisms involved
`remain unknown, but modification or masking of
`surface antigens,
`the secretion of factors that
`inhibit NK, LAK, or macrophage activation/func-
`tion, and an altered sensitivity to the direct cyto-
`
`

`

`72
`
`R Clarke
`
`——‘
`
`by
`I
`NK cells 6%» function
`
`
`
`
`
`b9
`
`@é Cytotoxic T cells-96'"> function
`Q Helper T cells
`
`’
`
`Pre-T cells
`
`stem cell
`
`.'
`
`..............
`
`Lymphoid Q .4
`
`
`nu
`
`W
`
`I
`0.: ——>e——>
`w"
`V'r ‘
`- cel
`'9'” B
`Is
`
`
`
`.
`Pluripotent
`stem cell
`
`v";
`
`Mature B cells
`
`Interleukin
`activation
`
`t
`
`bg & xid
`I
`Macrophages fie» function
`
`LAK cells
`
`Neutrophils
`
`
`
`Myeioid
`stem cell
`
`
`
`Mast cells
`
`scid
`.92.. mucosal mast cell
`deficiency
`
`Figure 1. General representation of the likely defects in hematopoiesis associated with the different mutations. The figure is
`meant to be neither complete (e.g.. there are several myeloid lineages not shown) nor specific in terms of the precise/relative
`locations of the defects. For example, the precise origins of NK cells, and of the cells generating LAK activities. remain to
`be definitively established. The reader is referred to the text for more specific details. *The deficiencies should not necessarily
`be considered to represent a complete ablation of activity/function. e.g., not all cytotoxic T cell functions are affected by the
`bg mutation.
`
`lytic effects of effector cells, are probably in—
`volved [30]. The isolation of tumor cell variants
`
`that exhibit different biological properties in the
`presence of different
`immune effector systems
`should greatly facilitate the study of these com—
`plex tumor/immune interactions.
`
`Immune-deficient rodent models
`i
`
`There are approximately 30 loci at which muta-
`tions can alter immune function in mice [38].
`However, not all are amenable to human tumor
`
`xenograft studies. The most widely used im-
`mune-deficient rodent model is the nude mouse.
`
`mice homozygous for the nu mutation.
`
`In more
`
`recent years, other single and combined mutation
`bearing strains have become available and are
`gaining acceptance. These include mice bearing
`the beige (bg) and/or X-linked immunodeficiency
`(xid) mutations, also in combination with the nu
`mutation, and the single gene severe combined
`immunodeficiency (scid) mutation. The hemato-
`poietic defects associated with these mutations are
`summarized in Figure l.
`
`Immunobiology of mice homozygous for the
`nude (nu) mutation
`
`The nude phenotype (the mice have no hair; not
`to be confused with the hairless mutation) was
`
`

`

`in 1962,
`first described in Glasgow, Scotland,
`with Pantelouris
`subsequently describing the
`predominating immune deficiency [39].
`It was
`not long before Rygaard and Povlsen reported the
`ability to sustain human tumor xenografts in nude
`mice [40]. Since these early reports, the use of
`homozygous nude mice as recipients of tumor
`xenografts has become almost ubiquitous in can-
`cer research. Breast cancer research is no ex-
`
`ception, with a significant proportion of human
`breast cancer cell lines exhibiting tumorigenicity
`in nude mice [5]. While over 70 congenic strains
`of nude mice have been generated over the last 30
`years, the respective immunologies and abilities to
`support xenografts appear strain dependent [41].
`The nude mutation (nu) is located on mouse
`
`chromosome 1 l. Homozygous mice are essential-
`ly athymic [39], although a rudimentary thymus
`can be detected at necropsy in individual mice.
`The lack of a fully functional thymus is primarily_
`responsible for the poor responses to T-dependent
`antigens [42], T-lymphocytes being generally at
`or below the limit of detection. These deficien-
`
`cies may be reversed by reconstitution with T-
`cells [43-45]. B-cell maturation also is defective,
`
`at least that component dependent upon a fully
`functional
`thymus.
`The production of earlier
`forms of B—cells appears normal, since nude mice
`possess apparently normal virgin B-cells [46].
`Despite the severe immunodeficiencies appar-
`ent in homozygous nu/nu mice, they are remark-
`ably robust. This is almost certainly due to their
`remaining immune-competence.
`For example,
`nu/nu mice
`frequently exhibit an essentially
`normal response to T-cell-independent antigens
`[47]. Perhaps most notably, nude mice possess
`levels of NK cell activity that significantly exceed
`those present in normal and heterozygous nu/+
`mice of the same background [48,49]. The poor
`metastatic potential of most human xenografts
`growing in nude mice has been partly attributed
`to this elevated NK activity [19,22—24]. Spleno-
`cytes from nu/nu mice are capable of generating
`LAK cells at levels similar to normal mice [50].
`
`The levels of tumoricidal macrophages are essen-
`tially equivalent
`in nu/nu and nu/+ mice [51].
`
`Xenograft models of breast cancer
`
`73
`
`Serum IgM levels are similar to [47] or higher
`than [46]
`those in nu/+ littennates, and while
`
`there is a significant decrease in the number of
`cells making IgG and IgA, both of these immuno-
`globulins can be detected in individual animals
`[44].
`
`Clearly, nude mice are not totally immune-
`deficient. The elevated NK cell activity at least
`partly contributes to the relatively poor take rate
`of primary human breast tumors [1]. Of interest
`is the observation that tumor take rate in nude
`
`frequently increased by encapsulating
`mice is
`cells/biopsies in Mattigel [17,18]. This artificial
`basement membrane capsule could provide two
`important functions, (i) a protective barrier from
`the effects of tumoricidal macrophages, NK, and
`LAK precursor cells, and (ii) a more natural
`structural/dimensional environment for the tumor
`
`with access to several attachment and mitogenic
`molecules, e.g., laminin, fibronectin, and type IV
`collagen.
`
`The beige (bg) mutation
`
`The bg mutation is located on mouse chromosome
`
`13. Among several phenotypic modifications in-
`duced is a reduced synthesis of pigment granules
`in melanocytes which results in a light coat color.
`The general phenotype is considered analogous to
`Chédiak-Higashi syndrome in humans, a multisys-
`tem autosomal recessive disease, where afflicted
`
`individuals possess both structurally and function-
`ally defective lysosomes. The main immune-defi-
`ciency exhibited by bg/bg mice is the direct result
`of a block in NK function [52,53]. However,
`there also are
`functional defects
`in T-cells,
`
`macrophages, and granulocytes [38]. For exam-
`ple, generation of cytotoxic T-cells is impaired in
`beige mice. A further complication in these mice
`is an apparent clotting disorder resulting from
`platelet dysfunction. Ovariectomies and hormone
`pellet supplementations are frequently performed
`with breast cancer xenografts, and the loss of
`animals during or shortly after surgery occurs
`
`

`

`74
`
`R Clarke
`
`more frequently for mice with the bg mutation
`than for many of the other immune-deficient
`models.
`
`The very low levels of NK activity in bg
`mice contrasts with the elevation in this activity
`in nu mice. The functional significance of the
`reduced NK cell and other immunologic activities
`is evidenced by the higher acceptance rate of low
`tumor cell inocula when compared with their het-
`erozygous (bg/+) litterrnates [54].
`
`Combined bg/nu mutations
`
`Since the bg mutation essentially eliminates NK
`cell activity, it was hypothesized that combining
`the bg and nu mutations would eliminate the
`problems associated with the elevation of NK cell
`activity in nude mice. Double congenic bg/nu
`mice have now been available for several years
`[48,49]. Their phenotype is largely that predicted
`by the known effects of both single gene muta-
`tions. Thus, the defective B-cell maturation and
`
`impaired T-cell dependent responses are apparent.
`While the contribution of the bg mutation signif-
`icantly reduces NK cell activity relative to homo-
`zygous nu and wild type (+/+) mice,
`the level
`remains higher than that in single mutant bg/bg
`mice [55-57].
`Splenocytes
`from these mice
`cannot produce LAK activity, an observation that
`assisted in delineating NK cells from precursors
`of LAK cells [26]. Reduced IgA and IgM levels
`can be presumed as a result of the nu mutation,
`the levels being potentially dependent upon the
`mouse background. While more immune-defi-
`cient than mice bearing only the nu mutation, the
`bg/nu mice retain the clotting disorder produced
`by the bg mutation.
`
`impaired functions contributing to the
`several
`immune-deficiencies in these mice,
`the major
`contributor is an impaired development of B-cells.
`Thus, B-cell colonies are not detected in in vitro
`
`assays [58]. A specific subset of mature B-cells
`is absent, and this appears to be the result of an
`inability of otherwise normal immature B—cells to
`respond to early activation signals.
`The general response to thymus-independent
`type-II antigens is
`low in xid mice [38,59].
`Macrophages are at least partially defective.
`In
`affected animals, macrophages produce low levels
`of IL-1 following stimulation with thymus-in-
`dependent
`type-II antigens in vitro [38].
`Im-
`munoglobulin levels, particularly IgM and IgG3,
`are low [60]. These mice, however, are not fully
`immune-compromised,
`producing
`essentially
`normal responses to thymus-independent type-I
`antigens [61].
`
`Combined bg/nu/xid mutations
`
`With the succeSSful generation of homozygous
`bg/nu mice,
`it became apparent
`that a further
`reduction in immune-competence could potentially
`be achieved by also adding the xid mutation.
`Andriole et al. [26] generated a strain designated
`NIH III, which is homozygous for all three bg/nu/
`xid mutations.
`These mice have intermediate
`
`levels of NK cells and low levels of LAK pre-
`cursor cells [26,62], and exhibit defects in the
`maturation of both B-
`and T-cells
`[63,64].
`
`activity/function may be partly
`Macrophage
`defective, due to the contribution of the xid
`
`mutation [38]. The clotting disorder (bg) also
`remains.
`‘
`
`The X-linked immunodeficiency ’(xid) mutation
`
`The severe combined immune deficiency
`(scid) mutation
`
`The xid mutation (X-chromosome) is associated
`
`with the X-linked lymphocyte regulated gene
`family [38].
`Thus, males (xid/Y) and homo-
`zygous females are affected. While there are
`
`A relatively more recent addition to the single
`gene deficiency models is the severe combined
`immune deficiency (scid) mutation [65]. The
`locus occurs on mouse chromosome 16, the equi-
`
`

`

`valent human mutation occurring at chromosome
`
`alternatives to the nude mouse, with several
`
`Xenografi‘ models of breast cancer
`
`75
`
`8qu [66]. The scid mutation produces mice with
`significantly smaller lymphoid organs [65,67].
`There is a clear defect
`in the differentiation]
`
`maturation of lymphocytes [68], and both pre-B
`and B-cells are undetectable. The few remaining
`T-cells appear non-functional.
`In contrast,
`the
`entire myeloid lineage appear normal [68,69].
`The scid mutation produces a deficiency in the
`rearrangement of genes encoding antigen-specific
`receptors on both B- and T-cells [67]. Several
`scid models are “leaky”, generating small num-
`bers of functional B- and T—cells. However, the
`
`majority of scid mice become “leaky” with age,
`i.e., by 10-14 months of age [70].
`It appears that
`the precursor lymphocytes are unable to adequate-
`ly join the cleaved variable region segments as
`catalyzed by the immunoglobulin V(D)J recom-
`binase [71].
`In general, immunoglobulin levels
`are at or below the limit of detection.
`IgGs 2a,
`2b, and 3a [38], IgM, and IgA [65,69] are rarely
`detected. However, in common with all immune—
`deficient models, there is individual variation, and
`
`some mice may produce detectable levels of two
`or more IgG isotypes and/or IgM [65]. Macro-
`phages, NK cells [72-75], and LAK precursor
`cells [50] are essentially normal
`in scid mice.
`However, the mucosal mast cells appear deficient,
`perhaps due to a lack of specific lymphokines
`required 'for their development from progenitors
`[76].
`
`In general, scid mice bearing this mutation are
`more severely immune-compromised than any of
`the other viable single gene mutation models.
`Nevertheless,
`the standard SPF environment
`is
`
`usually considered sufficient for their mainten-
`ance.
`It seems likely that, as these mice become
`more readily available and less expensive, they
`will be more widely utilized.
`
`l
`
`Impact of the different immunobiologies on
`xenografting
`
`immunobiologies of these other
`The different
`rodent models provide viable and important
`
`occasionally supporting the growth or metastasis
`of xenografts that are considered either “non-
`tumorigenic” or “non-metastatic” based on their
`lack of growth in nude mouse models. The issue
`of which immune-deficient model is most appro-
`priate for maintaining xenografts is somewhat
`controversial [77]. There is no clear and com-
`
`pelling evidence that scid or bg/nu/xid mice have
`a reproducibly higher overall take rate than nu/nu
`mice, with most
`investigators
`finding these
`models essentially equivalent [77-81]. However,
`the overall take rate can be a misleading indicator
`when an investigator wishes to evaluate the tum-
`origenicity or metastatic potential of specrfic cell
`lines/biopsies. An individual cell
`line may be
`tumorigenic or metastatic in one immune-deficient
`model and not in another, an observation that may
`or may not be a direct reflection of the different
`immunobiologies of the recipient mice. While
`this seems most likely to occur when assessing
`metastatic potential
`[70,80,82],
`individual vari-
`ability among cell
`lines for tumorigenicity also
`cannot be easily discounted. We have suggested
`that, particularly where in vivo growth is assessed
`as part of the characterization of a new cell line,
`more than one model be used [83]. While this
`
`may be less important for cell lines of mammary
`origin than for other cell lines, the designation of
`a cell
`line as “non-tumorigenic in the nude
`mouse” or “non-metastatic in the nude mouse”
`
`seems most appropriate, at least until its lack of
`tumorigenicity/metastatic potential is confirmed in
`other models [83].
`
`Use of immune-deficient rodent models for
`
`endocrinologic studies
`
`Endocrinologic studies of hormone-responsive
`xenografts are frequently performed with breast
`cancer xenografts. Mice can be purchased ovari-
`ectomized directly from each of the major ven-
`dors,
`including Taconic
`(Germantown NY),
`Charles River (Wilmington MA), Harlan Sprague
`Dawley (Frederick MD) and Jackson Laboratories
`
`

`

`76
`
`R Clarke
`
`(Bar Harbor ME). Adrenalectomized and hypo-
`
`physectomized mice can generally also be ob-
`tained in this manner, although viability of the
`mice requires supplementation with glucocorti-
`coids/mineralocorticoids.
`
`relatively straightforward
`There are several
`approaches for studying additive endocrine treat-
`ments. We have routinely used the sustained re-
`lease pellets provided by lnnovative Research of
`America (Toledo, OH). For hormone-dependent
`cells including MCF-7, we routinely use the 60-
`day release 0.72 mg 17B-estradiol pellets. These
`are easily placed s.c. between the shoulder blades
`using a sterile 10 or 12 gauge trocar. Many
`experimental agents also can be administered by
`implantation of silastic pellets or mini osmotic
`pumps. We have found the Alzet osmotic pumps
`from Alza Corp. (Palo Alto, CA) effective for
`administering growth factors and other agents.
`The smaller sizes (100 pl and 200 pl) can be
`introduced into a small s.c. pocket, although we
`also have introduced these i.p.
`into NCr nu/nu
`mice with no adverse effects. The larger pumps
`are for use only in rats, and can be used success-
`fully in nude rats (mu/mu).
`
`The hypogonadal/scid (hpg/scid) mouse as an
`ovarian-ablation model
`
`Ovariectomies are frequently performed on nude
`mice to provide an endocrinologic environment
`equivalent to that of postmenopausal women [84].
`This produces additional cost to the investigator
`and stress to the animal. The Jackson Labora-
`
`tories have recently generated a novel model that
`could alleviate some of these concerns. Mice
`
`(hpg) mutation [85]
`bearing the hypogonadal
`express a non-functional (truncated) LHRH pro-
`tein [86]. Thus, these mice are hypogonadic, with
`almost undetectable levels of LH’ and FSH, and
`
`have serum estrogen and progestin levels essen-
`tially equivalent
`to those detected in ovari-
`ectomized mice [87]. By combining the hpg and
`scid mutations, it has been possible to generate an
`immune-compromised strain with a postmeno-
`pausal endocrinologic environment [88]. It seems
`
`likely that this model may become more widely
`used in the future, both as a potential model of
`postmenopausal breast cancer, and to support the
`growth of breast tumor xenografts.
`
`Effect of endocrine manipulations on the
`immune response in rodents
`
`Estrogen and antiestrogen supplementation of
`rodents has been widely used to study the endo-
`crine regulation of breast cancer xenografts.
`However, several endocrine agents are known to
`influence those effectors of CMl that remain in
`
`For example,
`immune—compromised rodents.
`pharmacological but not physiological concentra-
`tions of E2 can inhibit NK activity in athymic
`nude mice [89-92]. This has been invoked as a
`
`potential explanation for an apparent ability of E2
`supplementation to increase the growth of estro-
`gen receptor negative breast cancer xenografts.
`Clearly, it is important to consider the ability of
`endocrine agents to perturb several effectors of
`cell-mediated immunity in study design.
`It could be argued that the apparent endocrine
`responsiveness of xenografts is actually an im-
`munologic,
`rather
`than direct endocrinologic,
`phenomenon. This seems unlikely for several
`reasons. Firstly, it is widely acknowledged that
`estrogens tend to produce a biphasic effect on NK
`cell activity, where NK levels rise for the first 30
`days with an overall reduction in NK activity not
`observed until
`later [89-92]. However, most
`
`MCF-7 xenografts produce palpable tumors, and
`can frequently be seen to grow, during this initial
`period when NK cell activity is rising above the
`already elevated levels in nude mice. Secondly,
`the concentrations of estrogens reported to sup-
`press NK activity [89-92] appear higher than
`those required to support MCF-7 xenografts [84,
`93-96]. Approximately 300 pg/ml/day of estra-
`diol
`is released from the widely used 60-day
`release 0.72 mg estrogen pellets produced by
`Innovative Research of America, Toledo OH [97].
`
`estrogen
`require
`still
`Finally, MCF-7 cells
`supplementation for growth in bg/nu/xid mice (R.
`Clarke, unpublished observations), despite the
`
`

`

`in these mice
`
`very low NK activity apparent
`[52,53].
`Also of relevance is the observed ability of
`tamoxifen (TAM) to stimulate NK activity in vivo
`[62]. Prolonged treatment of MCF-7 xenografts
`produces a TAM and estrogen dependent pheno-
`type [98,99], as does transfection with FGF4
`[100] despite the opposing effects of these agents
`on NK cell activity (estrogens inhibit, anti-
`estrogens stimulate). An ability of estrogen to
`apparently stimulate MDA-MB-231 (ER-negative)
`xenografts in nude mice has been specifically
`attributed to a reduction in cell loss [101].
`
`Breast cancer cell lines as xenografts
`
`lines currently
`Many of the breast cancer cell
`available are tumorigenic in the nude mouse.
`These cell lines most frequently produce adeno-
`carcinomas in vivo, with the degree of glandular
`differentiation generally greater in the ER-positive
`lines. As with many human tumor xenografts, the
`characteristics of the human breast tumors are
`
`generally faithfully reconstituted in in vivo ex-
`perimental systems. Thus, the ER positive/nega-
`tive, PGR positive/negative, hormone dependent/
`independent, antiestrogen responsive/unresponsive,
`and drug responsive/resistant characteristics of
`cell lines observed in vitro are reflected in their
`
`respective in vivo growth responses. The charac-
`teristics of many of the breast cancer cell lines
`have recently been reviewed in detail [5].,
`
`Breast cancer cell lines as xenografts:
`hormone-dependence and acquisition of
`hormone-independence
`
`The three most widely used hormone-dependent
`human breast cancer xenografts are the MCF-7
`[2], ZR-75-1 [4], and T47D cell lines [3]. These
`
`cell lines are require some degree of estrogenic
`supplementation for tumorigenesis in nude mice.
`The estrogen-induced growth of each is inhibit-
`ed by an appropriate dose of antiestrogen. While
`ER-positive tumors frequently invade locally and
`
`Xenograft models of breast cancer
`
`77
`
`metastasize in patients, these xenografts are poor—
`ly invasive and rarely, if ever, are metastatic in
`nude mice.
`
`The requirement for estrogen is surprising,
`since all
`three cell
`lines were derived from
`
`malignant effusions in postmenopausal women.
`We wished to determine whether the application
`of appropriate selective pressures could generate
`variants that no longer require E2 for growth in
`vivo. This was readily achieved by selecting for
`growth in the mammary fat pads of ovariec-
`tomized nude mice [96,102]. These mice have
`
`steroid hormone levels approximately equivalent
`to those found in postmenopausal women [84,
`103].
`
`Cells selected by one passage (MCF7/MIII)
`[96] or two passages (MCF7/LCC1) [102] (Table
`I), retain ER and FOR expression, sensitivity to
`antiestrogens [96,102], and sensitivity to LHRH
`analogues [104]. Both variants exhibit a signifi-
`cant increase in metastatic potential [105]. Since
`tumors with these characteristics arise frequently
`in women, we have suggested that the phenotypes
`exhibited by the MCF7/MIII and MCF7/LCC1
`
`the major ER/PGR
`cells more closely reflect
`positive, antiestrogen responsive phenotype found
`in postmenopausal women with breast cancer
`[106].
`
`Breast cancer cell lines as xenografts:
`acquisition of antiestrogen resistance
`
`While >70% of ER/PGR positive tumors respond
`to antiestrogens, the great majority of these will
`ultimately acquire a resistant phenotype [107].
`The modeling of this progression has generally
`taken two forms,
`in vitro selection either in a
`
`stepwise manner [108] or by treatment with a
`high dose of drug [109], or in vivo selection
`against a continuous drug exposure [98,99]. A
`prolonged in vivo selection almost exclusively
`generates cells that are TAM dependent, and at
`least initially E2 inhibited [98,99].