| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Tumor Biology |
Departments of Medicine and Microbology-Immunology and the Cancer Center, University of Illinois at Chicago College of Medicine, Chicago, Illinois 60612 [J. L. C.]; Department of Medicine, National Jewish Medical and Research Center and Departments of Medicine, Immunology and the Cancer Center, University of Colorado Health Sciences Center, Denver, Colorado 80206 [T. A. M., J. M. R.]; Department of Biostatistics, City of Hope National Medical Center and Beckman Research Institute, Duarte, California 91010 [D. N. I.]; and Laboratory of DNA Viruses, Division of Viral Products, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, Maryland 20812 [A. M. L.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Studies of immune-mediated rejection of rodent cells transformed by human adenovirus (Ad) serotypes 2 and 5 (Ad2/5) led to the discovery that Ad2/5 E1A gene expression actively induces cellular susceptibility to immune-mediated killing (5, 6, 7, 8, 9, 10, 11, 12, 13, 14) , a process that is independent of cellular expression of p53 (15, 16, 17) . Other studies revealed that E1A expression also sensitizes cells to apoptotic cell death in response to chemotherapeutic agents through both p53-dependent and -independent mechanisms (15 , 18, 19, 20, 21, 22, 23, 24) . Initial E1A studies done using rodent sarcomas were followed by studies of E1A-induced susceptibility of human tumor cells to immunological, chemical, and radiation-induced injuries. These results provided the conceptual basis for the therapeutic use of E1A to genetically modify cytolytic resistant human tumor cells (20 , 22 , 25, 26, 27, 28) .
Although nude mice lack thymus-dependent, T cell-mediated immune responses, they have intact innate immunity, mediated by NK4 cells, activated macrophages, and other effector mechanisms, that conveys measurable resistance to challenge with human tumor cells (29, 30, 31) . We have reported evidence that nude mice have innate immune defenses that are sufficient to limit tumor formation by E1A-expressing rodent tumor cells (32) . These in vivo results suggest that the cellular components of the innate immune response are capable of eliminating large numbers of E1A-positive tumor cells in vivo, thereby reducing tumor formation. This concept is supported by the results of in vitro cytotoxicity experiments showing that NK cells and activated macrophages from nude mice selectively kill E1A-positive tumor cells (33 , 34) . These observations indicate that the nude mouse can be used to study correlations between E1A-induced cytolytic phenotypes observed in vitro and E1A-induced changes in tumor cell rejection by the innate cellular immune response in vivo.
Most experimental assays of tumorigenicity lack the precision that is common among in vitro assays. This is explained in part by the limitations of conventional experimental methodology for testing tumor development. The use of single doses of tumor cells for animal challenge can result in errors in estimations of the minimal numbers of cells required for tumor formation, the so-called tumor-forming end point (35) . Furthermore, standard assays using measurement of tumor cell diameter or estimated tumor volume during short periods of observation after tumor cell challenge may not accurately quantitate the tumor-inducing capacities of individual tumor lines. Previous studies showed that s.c. challenges of animals with a range of cell doses and calculation of tumor-inducing efficiency as a function of time after challenge can be used to reproducibly define two independent tumor cell traits: (a) tumor latency and (b) threshold cell dose (32 , 35) . Tumor latency is the time to first tumor appearance after tumor cell inoculation that can be quantitated using survival curve analysis with groups of animals challenged with a range of cell doses (36) . Threshold cell dose is a measure of tumor-inducing efficiency that can be estimated using 50% end points of tumor production (36) .
In the studies reported here, we tested the effects of E1A oncogene expression on the susceptibility of human breast cancer cells with a p53 gene mutation to immunological injuries and chemotherapeutic injuries in vitro. We correlated the E1A-induced in vitro cytolytic phenotypes of the cells with their tumor latencies and tumor-inducing efficiencies in untreated and chemotherapy-treated nude mice in vivo. E1A expression sensitized these tumor cells to both immune-mediated (killer cell induced and death receptor dependent) injury and apoptosis triggered by chemotherapeutic agents. Quantitative tumor induction studies showed that E1A expression caused increased tumor latency in nude mice and that chemotherapy further increased the latency period. The combined effects of host defenses and chemotherapy also reduced the tumor-inducing efficiency, but did not eliminate tumor formation, by E1A-positive cells. These results were confirmed with an E1A-expressing, p53 mutant human fibrosarcoma cell line. The data show a direct correlation between the p53-independent, E1A-induced susceptibility of human tumor cells to immune-mediated and chemotherapy-induced injuries in vitro and their tumor-inducing capacities in vivo.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Apoptosis and Cytotoxicity Assays.
Assays of NK cell-induced killing of tumor cells were done as described (44)
using ficoll-diatrizoate-separated human blood mononuclear cells from normal donors or the human NK cell line YT (45)
, which was provided by G. Cohen. Bone marrow-derived macrophages were obtained from the femurs and tibias of mice and cultured in DMEM supplemented with 15% granulocyte macrophage colony-stimulating factor prepared from transfected Chinese hamster ovary cells provided by T. Potter. Granulocyte macrophage colony-stimulating factor was removed from the culture medium 48 h before cytolysis assays. Twenty-four h before cytolysis assays, macrophages were activated with IFN-
(100 units/ml; R&D Systems, Minneapolis, MN) and lipopolysaccharide (1 µg/ml; Sigma). Target cells were labeled with 3H-thymidine, and 48-h cytolysis assays were performed as described (5)
. The mean percentage spontaneous radiolabel release from all types of target cells was <20%.
Assays of Fas-induced apoptosis were done using antihuman Fas antibody (BD-PharMingen, San Diego, CA) immobilized on plastic surfaces as described (46) . Assays of TRAIL-induced apoptosis were done as described (14) using recombinant human TRAIL (R&D Systems). The chemotherapeutic drugs etoposide, cisplatin, cytarabine, and hydroxyurea (Sigma) were prepared as stock solutions in DMSO and diluted into culture medium just before each apoptosis assay. Initial dose-response assays defined optimal, apoptosis-inducing drug concentrations and excluded residual DMSO in assay medium as a factor in cytotoxicity.
The apoptotic nature of the cell death response of E1A-positive cells to both chemotherapeutic and immunological injuries has been defined (47) . The number of cells exhibiting characteristic nuclear chromatin condensation was quantitated morphologically by counting cells stained with ethidium bromide (48) . The amount of radiolabeled, low molecular weight DNA released from injured cells was used as a second measure of apoptosis-related DNA fragmentation using the method of Duke and Cohen (48) . All apoptosis assays were done in DMEM containing antibiotics and 10% endotoxin-free fetal bovine serum (Sterile Systems).
Tumor Induction Assays.
Cohorts of 8-week-old female nude mice (athymic NCr-nu) were obtained from the Frederick Cancer Research Facility or Taconic Laboratories and maintained in isolator cages. Quantitative tumor induction studies were done as described (49)
, by inoculating animals s.c. between the scapulae with serial 10-fold concentrations of cells ranging from 104 to 108 cells/inoculum (depending on the experiment) in a 0.2-ml volume. Cells were prepared by trypsinization of subconfluent cell monolayers followed by prompt trypsin inactivation with serum, repeated washing, and resuspension in additive-free, serum-free DMEM for injection.
Treatment of nude mice with etoposide (Sigma) was done using an optimal dose defined previously and treatment schedule, based on LD50 data and experience with experimental therapy of other tumor lines (50 , 51) . Etoposide was prepared in a stock solution in DMSO and then diluted into HBSS at a final drug concentration of 750 µg/ml just before treatment of each mouse group. Each animal was weighed and treated with a dose calculated to deliver 30 mg/kg etoposide per i.p. injection. Treatment was started 1 day after tumor challenge and repeated approximately every 4 days throughout the period of observation. Sham-treated animals were treated with the same volumes and dosing schedules using HBSS containing an equal concentration of DMSO to that used for etoposide treatments.
Animals were checked twice weekly for tumor development and sacrificed when tumors measured >20 mm in mean diameter or at the end of the period of observation if tumor free. The 50% end points of tumor production (TPD50s) were estimated by the method of Karber (52) at each time point. TPD50s were converted into survival data to reflect the percentage of tumor-free animals as described (36) using Sigma Plot software (SPPS). Average tumor latency was calculated across all cell challenge doses using JMP software (SAS Institute) for nonlinear curve fitting and estimation of the time required to reach 50% tumor development for each cohort of mice.
Tumors were randomly selected in each experiment for testing of E1A protein expression. Tumors were minced and trypsinized, and cells were tested by immunoblotting with E1A-specific monoclonal antibody at the first tissue culture passage.
The significance of the differences in the survival curves for mice in different cell challenge and treatment groups was estimated using the Kaplan-Meier method and Wilcoxon test with JMP software.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
50% of breast cancers with p53 mutations or deletions (38)
. MB435S also expresses low levels of both HER-2/c-erbB-2/neu-encoded p185 and mdr-1-encoded p170, both of which can cause tumor cells to be resistant to chemotherapeutic drugs (54)
. To our knowledge, the ability of p185 and p170 to influence immunologically mediated injury has not been evaluated. Use of MB435S simplified the interpretation of E1A effects on cytolytic susceptibility, because it was possible to eliminate overexpression of both p185 and p170 proteins as factors in cellular resistance to chemotherapeutic and immunologically mediated injuries. Initial studies also showed that MB435S cells are inherently resistant to apoptosis triggered by human and rodent NK cells and a variety of chemotherapeutic drugs,5
making MB435S suitable for studies of E1A-induced conversion from the cytolytic resistant to the cytolytic susceptible phenotype.
MB435S cells expressing either genomic E1A (i.e., both E1A 12S- and 13S-encoded proteins; MB435S-E1A.1, MB435-E1A.2) or only E1A 12S products (MB435-E1A12S.1) were tested for E1A oncoprotein expression by immunoblotting with an E1A-specific monoclonal antibody (Fig. 1, A and B)
.
|
and 3)
. The E1A-induced increase in susceptibility of p53 mutant MB435S cells to killing by NK cells was comparable with that observed with p53-negative Saos-2 (Fig. 2A
; Refs. 16
and 40
). E1A-induced cytolytic susceptibility of MB435S was also detected using the continuous human NK cell line YT (Fig. 2B
, left panel) and cytokine-activated, bone marrow-derived mouse macrophages (Fig. 2B, right panel). These results are consistent with our observation that there is no species restriction of NK or macrophage killing of E1A-positive cells (5
, 34)
.
|
|
. Similar E1A-induced apoptosis sensitivity was observed when cells were treated with recombinant TRAIL (Fig. 3B)
. Cell death after Fas- and TRAIL-induced signaling was confirmed to be apoptotic by two independent assays: (a) morphological assessment of nuclear chromatin condensation and (b) quantitation of DNA fragmentation (data not shown).
E1A Induces Chemosensitivity Independently of Tumor Cell Expression of p53.
The chemosensitivity of MB435S cells expressing either genomic E1A (i.e., both E1A 12S- and 13S-encoded proteins) or only E1A 12S products (Fig. 1A)
was characterized using the proapoptotic topoisomerase II inhibitor etoposide (Fig. 4A)
. In studies of E1A-expressing mouse fibroblasts, chemotherapy drug-induced apoptosis has been reported to depend on expression of p53 (16
, 55) . In contrast, p53 expression is not required for E1A-induced cytolytic susceptibility of different types of human tumor cells (15
, 21
, 23
, 47)
. In our experiments, E1A-positive MB435S, which express reduced levels of p53 because of gene rearrangement, were significantly more sensitive to the cytotoxic effect of etoposide than were E1A-negative cells (Fig. 4A)
. E1A 12S expression alone (MB435-E1A.12S) was as effective at sensitizing cells to etoposide as expression of both E1A 12S and 13S transcripts (MB435-E1A.l and MB435-E1A.2). The apoptotic nature of the cell death response by E1A-positive cells was confirmed by counting the numbers of cells exhibiting nuclear chromatin condensation and fragmentation and also by quantitating the release of low molecular weight DNA from the nuclei of injured cells (Fig. 4B)
. To further assess the range of chemotherapeutic drug injuries to which E1A-positive, p53-negative human tumor cells are susceptible, etoposide-induced apoptosis of p53-negative Saos-2 cells was compared with E1A-related sensitivity to apoptosis triggered by three other chemotherapeutic agents from different drug classes (cisplatin, an alkylating agent, and cytarabine and hydroxyurea antimetabolites; Fig. 4C
). All three classes of drugs induced apoptotic cell death of p53-negative, E1A-positive cells that was significantly greater than that observed with E1A-negative cells (P < 0.05).
|
; MB435-E1A12S.1 = challenge cell versus MB435-E1A12S.1-nm1 and -nm2 = tumor cells from nude mice challenged with MB435-E1A12S.1). Therefore, tumor formation by E1A-expressing cells was not explained by selection for E1A-negative variants within the E1A-positive MB435 cell population inoculated into mice. The effect of E1A expression on tumor development was tested by observing nude mice for 3 months after inoculation with either E1A-positive or -negative MB435S cells. In some experiments, mice challenged with both cell types were treated with chemotherapy and compared with sham-treated controls. Previous studies using these cells (20) or other E1A-expressing human tumor cells (57) showed that nude mouse tumors induced by E1A-positive cells developed more slowly than tumors induced by E1A-negative controls during the 1st month of observation but did not reveal the long-term outcome of comparative tumor development.
To test the effects of chemotherapy on tumor development, cohorts of mice were divided into four groups of 6, and each group was inoculated with 104, 105, 106, or 107 cells/mouse using MB435S or MB435-E1A12S.1 cells and then split into two groups of 3; one group of 3 was sham treated by i.p. inoculation of HBSS, and the other group was treated with HBSS containing etoposide (30 mg/kg/mouse) as described in "Materials and Methods." Three parameters of tumor development were measured: (a) time to first detection of s.c. tumors; (b) time to 50% tumor incidence calculated across the whole cohort of challenged mice; and (c) threshold cell dose required for tumor production (final TPD50; Refs. 36 ). Each experiment was repeated in a different mouse cohort using the identical protocol.
Nude mice were significantly more resistant to challenge with E1A-positive cells than with E1A-negative cells by Kaplan-Meier analysis (Fig. 5
; solid circles versus solid triangles, respectively, and table, column 3 versus column 1, respectively; P < 0.05). The lag time between s.c. inoculation and the first tumor detected in animals challenged with 107 cells increased from 6 days for E1A-negative cells to 35 days for E1A 12S-expressing cells in sham-treated mice (Fig. 5
, table). The average tumor latency calculated across all cell challenge doses increased from 11 days for E1A-negative cells to 41 days for E1A-positive cells (Fig. 5
, table). This estimation of the average latency period provides a more reproducible measure of tumor latency than the time of first tumor appearance in a single animal, because it averages variations across the entire animal cohort challenged with a range of cell doses (35)
. E1A expression also slightly reduced the efficiency of tumor induction (threshold cell dose), as evidenced by a doubling in the number of challenge cells required to reach the final 50% end point of tumor production (TPD50 = 4.8 versus 4.5, respectively). This trend toward reduced tumor-inducing efficiency of E1A-positive MB435S cells was confirmed in a second experiment in another mouse cohort where there was a 5-fold increase in the number of challenge cells needed to reach the TPD50 end point (E1A-positive cells = 5.2, E1A-negative cells = 4.5; 0.7 log = 5-fold difference). Our experience in many nude mouse tumor induction experiments is that differences in TPD50 of <0.5 logs are usually not significant. This is consistent with the findings in these experiments in which the mean SD ± TPD50 from four different challenges of nude mice with E1A-positive MB435S cells was 5.2 ± 0.5 cells/mouse. In summary, these data showed that the cumulative innate immune defenses of nude mice delayed tumor formation by breast cancer cells expressing E1A protein as evidenced by Kaplan-Meier analysis and prolongation of the average tumor latency periods but had minimal effects on the long-term outcome of tumor development as evidenced by minor effects on net tumor-inducing efficiency (TPD50) relative to E1A-negative parental, breast cancer cells.
|
, open symbols and table, columns 2 and 4). Etoposide treatment significantly delayed tumor development by E1A-expressing cells (open circles) when compared with tumor development in all nude mice challenged with E1A-negative cells by Kaplan-Meier analysis (P = 0.02). This E1A-induced sensitivity to chemotherapy also was reflected in an increased time to first tumor appearance (50 versus 35 days) and average tumor latency calculated across the cohort (59 versus 41 days; Fig. 5
, table). In contrast, etoposide treatment had no significant effect on tumor development by E1A-negative cells (Fig. 5
open versus closed triangles and table, columns 1 and 2). Chemotherapy further reduced the tumor-inducing efficiency of E1A-positive cells by 2.5-fold, compared with sham-treated mice (increased TPD50 from 4.8 to 5.2), but had no effect on the tumor-inducing efficiency of E1A-negative cells (TPD50 = 4.5 in both cases). A similar, E1A-related enhancement of the etoposide effect on the tumorigenicity of MB435S cells was observed in a second experiment (increased average tumor latency from 29 to 51 days and TPD50 from 5.1 to 5.5). Similar studies of sham- and etoposide-treated nude mice were repeated using H4 fibrosarcoma cells. E1A expression does not change the in vitro growth rate of H4 (41) but does convert them from a cytolytic resistant to a cytolytic susceptible phenotype when challenged by NK cells or TRAIL (12 , 14) . E1A effects on the chemosensitivity of H4 cells followed the predicted pattern of E1A-induced susceptibility. The mean ± SE percentage of killing in seven cytotoxicity assays with etoposide at 25 µM was: E1A-negative H4 = 12.9 ± 3.3% versus E1A-positive H4 = 59.1 ± 4.1% (P < 0.001).
Quantitative tumor induction studies showed that E1A-induced changes in H4 tumorigenicity were similar to those observed with MB435S cells (Fig. 6)
. E1A expression significantly delayed H4 tumor development in sham-treated nude mice compared with E1A-negative H4 (Fig. 6
, solid circles versus solid triangles, respectively; P = 0.03 by Kaplan-Meier analysis) and was associated with a doubling of the average tumor latency of the challenged cohort (15 versus 7 days, respectively). E1A-positive cells also had a 10-fold lower tumor inducing efficiency than E1A-negative H4 cells (TPD50 = 5.2 versus 4.2, respectively; Fig. 6
, table, column 3 versus column 1, respectively). Etoposide treatment significantly delayed tumor development in animals challenged with E1A-positive H4 compared with mice challenged with E1A-negative cells (Fig. 6
, open circles versus open triangles, P = 0.03 by Kaplan-Meier analysis) and further prolonged the average tumor latency of the treated cohort (29 days versus 15 days, respectively) and reduced the tumor-inducing efficiency of E1A-positive H4 cells (TPD50 = 5.8) compared with E1A-negative H4 cells (TPD50 = 4.5) by 20-fold (Fig. 6
, table, column 4 versus column 2, respectively).
|
and 6)
. E1A expression did not, however, eliminate tumor induction by either human tumor cell line, if the cells were inoculated at a sufficient dose and the animals were observed for prolonged periods of time after challenge. The question of whether E1A-expressing human tumor cells can induce tumors in nude mice has been a source of debate. It has been suggested that E1A expression in primary human cells may induce limited cell cycle progression but may be unable to transform cells to a fully neoplastic and tumorigenic phenotype (58) . The related postulate is that E1A induces human cells to exhibit a relatively normal epithelial-like phenotype, thus eliminating their tumorigenic potential (59) . There are, however, a few reports that E1A-expressing, human tumor cells or immortalized human cell lines are tumorigenic in immunodeficient mice (60, 61, 62) .
The studies presented in this report suggested a possible explanation for these discrepancies. The results indicated that E1A expression might both prolong the time for tumor development and reduce the tumor-inducing efficiency of a human tumor cell in a host like a nude mouse with an intact innate immune response. These observations predicted that E1A-immortalized human cells might form tumors very inefficiently in nude mice, in part because of E1A-induced cytolytic susceptibility and the innate immune defenses of the animals. The results of our previous studies also predicted that the level of NK cell response of the challenged animal might determine the success of the tumor challenge with E1A-positive human cells (32 , 33 , 63) , e.g., nude rats that have higher levels of NK activity than nude mice (33) might be more resistant to challenge with NK-susceptible, E1A-expressing human cells.
To test these possibilities, nude mice and rats were challenged s.c. with large numbers (106, 107, or 108) of E1A-expressing human 293 cells and observed for 3 months for tumor development (Fig. 7)
. 293 is an immortalized human cell line that was derived by transfection of normal human embryonic kidney cells with sheared adenoviral DNA but that may be of neural cell origin (64)
. 293 express high levels of E1A oncoprotein and are highly NK sensitive (5)
. Small tumors were first apparent 3–4 weeks after 293 cell inoculation into both types of animals. Nude mice developed tumors only after inoculation of 107 or 108 cells, with a calculated TPD50 of 6.5 cells. Nude rats were slightly more resistant to tumor formation, with a TPD50 of 7.2 cells. In both types of animals, the last tumors were detected 6 weeks after cell inoculation. Therefore, a challenge with 3–15 x 106 293 cells and an observation period of ≥4–6 weeks was required to detect tumor formation in these experiments. These data suggested that, like E1A expression in human breast cancer and fibrosarcoma cells, E1A expression during immortalization of primary human cells may limit the tumorigenicity of these cells in NK competent animals but that this limitation can be overcome with high cell challenge numbers. The significance of the apparent relative resistance of nude rats versus nude mice to 293 challenges could not be determined in this limited study.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Available studies indicate that transient or stable expression of the E1A oncogene can directly or indirectly repress human tumor cell growth in vivo (20 , 25, 26, 27 , 41 , 56 , 57 , 66, 67, 68) . However, it is difficult to quantitate these in vivo effects using existing tumorigenicity assays. In the present study, we characterized E1A-induced changes in cytolytic susceptibility to immunological and chemotherapeutic drug injuries in vitro and used a quantitative tumor production assay to test the correlations among these in vitro data with patterns of tumorigenicity in nude mice. The results show that E1A-induced cytolytic susceptibility of human tumors as measured in vitro is predictive of measurable changes in two parameters of tumorigenicity in the context of nude mouse innate immunity, tumor latency and tumor inducing efficiency, and may predict increased tumor cell sensitivity to chemotherapy in vivo.
These results confirm previous observations that both tumor cell inoculation dose and the period of observation after tumor cell challenge can be critical to the experimental outcome (36)
. One example is shown in the experiments comparing E1A-positive and -negative breast cancer cells in sham- and etoposide-treated mice (Fig. 5)
. If animals challenged with E1A-positive breast cancer cells had been observed for <1 month after tumor cell challenge, it could have been concluded that E1A-positive cells were nontumorigenic in untreated nude mice or that chemotherapy eliminated the tumorigenicity of E1A-positive cells. Prolongation of the period of observation to 3 months after tumor cell inoculation revealed that the E1A-positive cells were highly tumorigenic in nude mice, albeit with a slightly reduced efficiency (increased TPD50; Fig. 5
, table), contrasted with E1A-negative cells. The longer observation period also revealed that chemotherapy delayed and reduced the efficiency of, but did not eliminate, tumor formation by E1A-positive cells. Similar observations were made with E1A-expressing H4 fibrosarcoma cells (Fig. 6
and table).
The greatest E1A-related differences in tumor induction for both MB435S (Fig. 5)
and H4 (Fig. 6)
cells were observed when contrasting the survival curves representing tumor development by E1A-negative cells in sham-treated mice (closed triangles) with the curves for E1A-positive cells in etoposide-treated mice (open circles). For both types of tumor cells, there was an E1A-related prolongation in tumor latency depicted by the more slowly falling survival curve and a reduction in the tumor-inducing efficiency depicted by the higher TPD50 at the end of the 2–3-month observation period (Figs. 5
and 6
tables). These observations suggest that the E1A-induced sensitivity of human cells to nude mouse innate immune responses and to chemotherapy may be additive.
It is useful to consider what these data suggest about the limits of the antitumor effects of E1A on human cells. The E1A-related changes in the tumor survival curves indicate that E1A expression can slow tumor appearance from an s.c. site and reduce, but not eliminate, tumor development (e.g., solid circles versus triangles in Figs. 5
and 6
). In the absence of chemotherapy, the reduction in tumor-inducing efficiency ranged from 2-fold for breast cancer cells (Fig. 5
table, sham-treated comparison) to 10-fold for fibrosarcoma cells (Fig. 6
table, sham-treated comparison). This indicates that ≤50–90% of the tumor cells may be destroyed at the site of inoculation by the innate host defense. If the cell inoculum for a given animal were 107 cells and the TPD50 were 4.8 (i.e., 104.8 cells required to cause 50% tumors) as for MB435 cells, however, it would be likely that the animal would still die as a consequence of the tumor challenge even if 90% of the cells were destroyed (0.1 x 107 = 106 cells, which is >15 times as many cells as would be required to form a tumor in 50% of the animals). These data for MB435S and the comparable data for H4 cells indicate that, even when expressed at high levels in human tumor cells, E1A is unable to eliminate tumor growth of a large population of cells in nude mice. It cannot be excluded that this limitation in the tumor rejection effect of E1A might result from its inability to overcome "pro-tumorigenic" traits of these cells derived through selection for progressive tumor development in humans. In addition, we have reported that the innate immune response of nude mice is not as effective as an intact cellular immune defense (e.g., as would be present in a euthymic host) against E1A-expressing tumor cells (32)
. Therefore, it is likely that a greater tumor rejection response would be observed with these human cells if tested in a host with an intact cellular immune response.
The 293 cell data may provide some insight into this question. In contrast to E1A-expressing MB435 and H4 tumor cells, E1A-expressing 293 cells were derived by immortalization of normal, primary human embryonic kidney cells. Despite this normal cell background and the E1A-related sensitivity of 293 to killing by NK cells (5) , 293 are still tumorigenic in NK-competent nude mice and nude rats challenged with high cell doses. This is consistent with the original report of 293 tumorigenicity in nude mice (60) . These observations indicate that E1A-induced tumor rejection must be considered in the context of the balance that exists between the cell dose and antitumor effects (e.g., innate immune defenses and chemotherapy). This suggests that there are limits to the extent of E1A-related tumor rejection in animals that have intact innate immunity but lack the complementary effect of T cell-dependent, antitumor responses.
Multiple factors, in addition to innate immunity, that may be difficult or impossible to measure in vitro could affect survival of E1A-expressing tumor cells in vivo. Among these are cellular stimulation by growth factors (69 , 70) , synergistic interactions between host cellular defenses and chemotherapeutic agents (22) , limitations on achievable drug concentrations at the site of tumor formation (71) , and altered responses of E1A-positive cells to extracellular biological matrices (41) . This complexity underscores the importance of measurements that can be used to quantitate the cumulative in vivo determinants of tumorigenicity when comparing tumor development between control and test tumor cell populations. The data from this study indicate that quantitative in vivo assessment of tumor latency and threshold cell dose can be used for this purpose.
The independence of average tumor latency and threshold cell dose as parameters of tumorigenicity (36)
defines four possible outcomes of the comparative tumor production assays as depicted in Fig. 8
. These theoretical curves of tumor production depict time as a measurement of tumor latency on the X axis and tumor-inducing efficiency as represented by the percentage of tumor-free animals on the Y axis (36)
. Comparison of two different tumor production curves might show no change in the tumorigenic phenotypes as represented in Fig. 8A
. In the present studies, the observations of no significant change in tumorigenicity between sham- and drug-treated nude mice challenged with E1A-negative breast carcinoma or fibrosarcoma cells represent this pattern (Figs. 5
and 6
, closed versus open triangles). A second theoretical pattern of comparative differences in tumor production could show an increase in tumor latency without a change in the final tumor-inducing efficiency (TPD50) as depicted in Fig. 8B
. The comparison of E1A-negative and -positive MB435S cells in sham-treated mice most closely approximates this pattern (Fig. 5
, closed triangles versus closed circles, respectively). There was a 4-fold E1A-related increase in the period of average tumor latency (11 versus 41 days, respectively) but only a 0.3 log (2-fold) E1A-related increase in the number of cells required for the 50% end point of tumor production (TPD50 = 4.5 versus 4.8 cells, respectively; Fig. 5
table). Previous studies of different clones of DNA virus-transformed rodent cells have shown a similar pattern of marked differences in tumor latency but little or no difference in threshold cell dose (36)
. A third theoretical pattern of comparative differences in tumor production could be the absence of a change in the period of tumor latency but a significant decrease in the tumor-inducing efficiency (increased TPD50) as represented in Fig. 8C
. This pattern of tumor production has been reported for DNA virus-transformed mouse sarcoma cells (36)
. The trend toward reduced tumor-inducing efficiency by 293 cells in nude rats compared with nude mice without a change in the rate of tumor development most closely represents this pattern in the present study (Fig. 7)
. A fourth pattern of difference in tumor production could be a combination of an increase in tumor latency and a decrease in tumor-inducing efficiency as depicted in Fig. 8D
. This pattern is most clearly demonstrated in the present studies by comparing the curves shown in Fig. 6
for tumor development by E1A-negative H4 cells in sham-treated mice (closed triangles) with the curves for E1A-positive H4 cells in etoposide-treated mice (open circles). There was both an E1A-related prolongation in tumor latency (8)
depicted by the more slowly falling survival curve and a reduction in tumor-inducing efficiency as evidenced by the 20-fold higher TPD50 at the end of the 2-month observation period (Fig. 6
table). This quantitative assay of tumor production makes it possible to define and dissect these relationships between tumor cell phenotypes identified in vitro versus tumor latency or threshold tumor cell dose.
|
In summary, these results demonstrate that E1A-induced cytolytic susceptibility of human tumor cells to both immunological and chemotherapeutic injuries as measured in vitro correlates with E1A-induced reduction in two distinct parameters of tumor development in nude mice: (a) tumor latency and (b) tumor-inducing efficiency. The use of this quantitative tumor induction assay adds increased precision to the measurement of E1A-related phenotypes and provides a basis for future translational studies of oncogene effects, effects of modulation of host innate cellular immune defenses, and therapeutic interventions.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported by Public Health Service Grants CA86727 (to J. L. C.) and CA76491 (to J. M. R.) and Department of the Army Grant DAMD-17-98-1-8324 (to J. L. C.). 
2 Supplementary data for this article are available at Cancer Research online (Internet address: http://cancerres.aacrjournals.org).
3 To whom requests for reprints should be addressed, at Section of Infectious Diseases, MC-735, Department of Medicine, University of Illinois at Chicago, 808 South Wood Street, Chicago, IL 60612. Phone: (312) 996-6732; Fax: (312) 413-1657; E-mail: JLCook{at}uic.edu
4 The abbreviations used are: NK, natural killer; TRAIL, tumor necrosis factor-related apoptosis-inducing ligand; TPD50, log10 number of cells required to produce tumors in 50% of mice.
5 J. L. Cook and J. M. Routes, unpublished data.
Received 2/20/03. Accepted 4/ 4/03.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. M. Routes, S. Ryan, K. Morris, R. Takaki, A. Cerwenka, and L. L. Lanier Adenovirus serotype 5 E1A sensitizes tumor cells to NKG2D-dependent NK cell lysis and tumor rejection J. Exp. Med., December 5, 2005; 202(11): 1477 - 1482. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Routes, K. Morris, M. C. Ellison, and S. Ryan Macrophages Kill Human Papillomavirus Type 16 E6-Expressing Tumor Cells by Tumor Necrosis Factor Alpha- and Nitric Oxide-Dependent Mechanisms J. Virol., January 1, 2005; 79(1): 116 - 123. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. A. Miura, H. Li, K. Morris, S. Ryan, K. Hembre, J. L. Cook, and J. M. Routes Expression of an E1A/E7 Chimeric Protein Sensitizes Tumor Cells to Killing by Activated Macrophages but Not NK Cells J. Virol., May 1, 2004; 78(9): 4646 - 4654. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. M. Frisch E1A as a Tumor Suppressor Gene: Commentary re S. Madhusudan et al. A Multicenter Phase I Gene Therapy Clinical Trial Involving Intraperitoneal Administration of E1A-Lipid Complex in Patients with Recurrent Epithelial Ovarian Cancer Overexpressing HER-2/neu Oncogene. Clin Cancer Res 2004;10:2905-2907. Clin. Cancer Res., May 1, 2004; 10(9): 2905 - 2907. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
