Cancer Cell
Article
Oncogene-Targeting T Cells Reject Large Tumors while Oncogene Inactivation Selects Escape Variants in Mou Models of Cancer
Kathleen Anders,1 Christian Buschow,2 Andreas Herrmann,4 Ana Milojkovic,5 Christoph Loddenkemper,3 Thomas Kammertoens,2 Peter Daniel,5 Hua Yu,4 Jehad Charo,1 and Thomas Blankenstein1,2,*
for Molecular Medicine, 13092 Berlin, Germany of Immunology 3Institute of Pathology ´ Campus Benjamin Franklin, 12200, Berlin, Germany Charite 4Cancer Immunotherapeutics and Tumor Immunology, Beckman Rearch Institute, City of Hope Cancer Center, Duarte, CA 91010, USA 5Department of Hematology, Oncology and Tumor Immunology, Charite ´ Campus Berlin Buch, 13092 Berlin, Germany *Correspondence: tblanke@mdc-berlin.de DOI 10.1016/jr.2011.10.019
2Institute 1Max-Delbru ¨ ck-Center
SUMMARY
The genetic instability of cancer cells frequently caus drug resistance. We established mou cancer models, which allowed targeting of an oncogene by drug-mediated inactivation or monospecific CD8+ effector T (TE) cells. Drug treatment of genetically unstable large tumors was effective but lected resistant clones in the long term. In contrast, TE cells completely rejected large tumors (R500 mm3), if the target antigen was cancer-driving and expresd in sufficient amounts. Although drug-mediated oncogene inactivation lectively killed the cancer cells and left the tumor vasculature intact, which likely facilitated survival and growth of resistant clones, TE cell treatment led to blood vesl destruction and probably ‘‘bystander’’ elimination of escape variants, which did not require antigen cross-prentation by stromal cells.
INTRODUCTION One of the hallmarks of cancer is a high degree of genetic instability and the accumulation of somatic mutations. In colorectal cancers, for example, up to 10,000 somatic mutations have been detected (Stoler et al., 1999). The high mutation rate in tumors may explain the frequently obrved resistance to chemotherapy or drugs interfering with oncogene activity (Gorre et al., 2001; Knight et al., 2010; Pao et al., 2005). In the clinic, tumors can be detected at about 1 cm in diameter ($500 mm3), which corresponds to approximately 109 tumor cells (Schreiber et al., 2006; Kumar et al., 2004). Anticancer drug efficacy depends on the number of cancer cells and, thus, the n
umber of genetic variants at the time of treatment (Skipper, 1965). Drug and T cell therapy were usually analyzed against small tumors below size that can be detected in the clinic
(Schreiber et al., 2006), and their efficacy was never compared in the same tumor model. If resistance to chemotherapy or oncogene-inactivating drugs is due to lection of mutant clones caud by genetic instability, one would expect that otherwi effective adoptive T cell therapy similarly lects variants that escape T cell-mediated destruction (Liu and Bai, 2008). Antigen loss variants were found in patients with melanoma after T cell therapy (Restifo et al., 1996; Yee et al., 2000), suggesting that T cell therapy is as vulnerable to lection of escape variants as therapy with oncogene-inactivating drugs. However, in some experimental models, adoptively transferred T cells could reject large tumors (defined as R500 mm3) (Kast et al., 1989; Spiotto et al., 2004). Sufficient amounts of tumor antigen expression for cross-prentation by tumor stroma cells and T cell-derived interferon-g (IFN-g) acting on stroma hindered outgrowth of antigen loss variants (Spiotto et al., 2004; Zhang
Significance So far, the genetic instability of cancer cells impedes effective therapy with oncogene inactivating drugs as well as adoptively transferred T cells. We created ideal conditions to target the oncogene by drug-mediated inactivation or T cells, which both induced regression of large tumors. Y花与蛇1
et, only T cell therapy resulted in long-term cure, probably becau the T cells also destroyed the tumor vasculature. Becau techniques for therapy with high-avidity T cells against antigens overexpresd in human tumors have recently been developed, defining optimal conditions for T cell therapy may help improve future clinical trials.
Cancer Cell 20, 755–767, December 13, 2011 ª2011 Elvier Inc. 755
Cancer Cell
Overcoming Genetic Instability of Cancer
A
B
C
五大领域目标D
E
学汉字
F
Figure 1. Drug-Mediated Oncogene Inactivation in Large Tumors Induces Transient Tumor Regression南黄古道
(A) Tet-TagLuc fibrosarcoma cells were generated by infection of primary fibroblasts of a TREloxPstoploxPTagLuc transgenic mou with a Cre-encoding adenovirus (AdCre) to exci the stop castte, a Tet-off transactivator-encoding retrovirus (RvtTA), and adaptation to in vivo growth at passage 19 (p19). Expression of the TagLuc fusion gene can be regulated by dox. (B) Tet-TagLuc cells (1 3 104) in duplicates were cultured with (0.5 mg/ml) or without dox, and cell numbers were determined daily for 4 days. Error bars reprent ±SD. (C) RagÀ/À mice with established Tet-TagLuc tumors (mean ± SD, 546 ± 246 mm3 at $30 days) received dox-containing drinking water, and TagLuc expression was followed by BL imaging (1 s exposure time). The time after treatment is indicated in days (d). (D) BL signals of dox-treated tumors of individual mice (n = 8) were quantified over time. (E) Tumor growth kinetics is displayed for mice shown in (D). Results in (C–E) are reprentative for three experiments with a total of 12 analyzed mice. (F) Tumor growth kinetics of individual mice (n = 7) with small Tet-TagLuc tumors (%250 mm3) treated with dox are shown in the left panel. Time point of dox treatment is indicated. For comparison, the mice with large tumors as in
(E) are shown (right panel). The number of mice with tumor relap is indicated.
et al., 2008). The mode of tumor destruction may be different for drug and T cell therapy that, however, has not been addresd in a clinically relevant (e.g., large) tumor model. Here, we established a mou cancer model allowing direct comparison of the efficacy of drug versus T cell therapy directed against the same target protein to eradicate large established tumors. SV40 large T antigen (Tag) is a well-characterized oncogene with defined H-2b restricted epitopes (Staveley-O’Carroll et al., 2003). Tag, among other activities, inactivates the tumor suppressors p53 and retinoblastoma protein (Rb), reducing DNA repair and creating a genetically unstable phenotype (Kuerbitz et al., 1992).
RESULTS Generation of a Conditional TagLuc Expressing Tumor Cell Line in Mice To compare the therapeutic efficacy of drug-mediated oncogene inactivation and targeting the oncogene by single peptide antigen-specific CD8+ effector (TE) cells, we isolated fibroblasts from a TREloxPstoploxPTagLuc transgenic mou (Figure 1A), which contains the Tag gene fud to the firefly lucifera (Luc) gene by a linker, encoding glycine-rine (G4S)3 repeats (TagLuc). Expression of the TagLuc fusion gene in TREloxP stoploxPTagLuc mice is regulated by a tetracycline respon
756 Cancer Cell 20, 755–767, December 13, 2011 ª2011 Elvier Inc.
Cancer Cell
Overcoming Genetic Instability of Cancer多肉怎么种
A
B
C
Figure 2. Each Dox-Unresponsive Tumor Reveals a Unique Point Mutation in the Transactivator Gene
(A) Parental Tet-TagLuc cells and cells of three dox-unresponsive tumors were cultured for 5 days in the prence of dox (1 mg/ml), and TagLuc expression was analyzed by western blot analysis with an anti-Tag antibody. As loading control, b-actin was detected. (B) Relative light units (RLU) were analyzed in parental and drug-resistant Tet-TagLuc cells, cultured in the prence or abnce of dox. One of three analyzed dox-unresponsive tumors with similar results is shown. Error bars reprent ±
SD. (C) Comparison of the tTA amino acid (AA) quence from position 64 to182 of parental Tet-TagLuc cells (top) and ven dox-unresponsive tumors (tumor 2 and 6 with two mutations). Mutations are shown in bold. Mutations in the tTA leading to dox-unresponsive variants are indicated by a black circle.
element (TRE) and is silent in the abnce of an active transactivator (TA) (Gosn and Bujard, 2002). A loxP-flanked stop castte (between TRE and TagLuc) was excid by transient adenoviral Cre recombina (AdCre) expression in the primary cells (Figure 1A). Subquent introduction of a Tet-off transactivator (tTA) by stable gene transfer with a tTA-encoding retrovirus allowed TagLuc expression, reversible by adding doxycycline (dox) (e below). TagLuc-expressing cells at passage 19 of in vitro culture exhibited immortal growth and were adapted to tumor growth in vivo. The resulting cell line, termed Tet-TagLuc, proliferated only in abnce of dox (Figure 1B). TagLuc Inactivation Fails to Control Large Tumors in the Long Term Mice with large Tet-TagLuc tumors (546 ± 246 mm3) were treated with dox, and the kinetics of TagLuc inactivation were followed by bioluminescence (BL) imaging. After 1 day, a decrea of BL signal, declining on day 3 below detectable level at 1 s exposure time, was obrved, followed by tumor regression. Then, despite further dox treatment, the BL signal reappeared and tumors progressively grew in all cas (Figures 1C–1E). Analysis
of the efficacy of dox in treating smaller (%250 mm3) tumors showed that most tumors could still be eliminated (Figure 1F), indicating that lection of dox-unresponsive variants that likely occur at a low rate requires large numbers of tumor cells. Each Dox-Unresponsive Tumor Reveals a Unique Point Mutation in the Transactivator Gene Tumors that grew in the prence of dox were analyzed in vitro. Although dox treatment of the original Tet-TagLuc cells resulted in loss of TagLuc expression, as shown by western blot and BL analysis, the variant cell lines did not reduce TagLuc expression in respon to dox (Figures 2A and 2B), suggesting genetically acquired resistance. The data argued against the possibility that the therapy lected variant cancer cells that lost the oncogene dependence (Jonkers and Berns, 2004; Weinstein, 2002), but rather pointed to the inability of dox to bind to and inactivate the tTA. The amino acid positions in the tTA allowing dox binding are well characterized (Hillen and Berens, 1994; Hinrichs et al., 1994). DNA quence analysis of ven dox-unresponsive tumors showed in five cas a single point mutation in the tTA gene (Figure 2C). Two tumors had two point mutations each,
Cancer Cell 20, 755–767, December 13, 2011 ª2011 Elvier Inc. 757
Cancer Cell
Overcoming Genetic Instability of Cancer
A
B
Figure 3. Partial Compensation of Selection of Dox-Unresponsive Tumors by Endogenous T Cells
(A) Scheme of the experimental design. The mice, which rejected the tumor, received two albino B6 skin grafts expressing either the Luc or the rtTA transgene, both shared with the tumor cells. (B) Expansion of transferred CD8+ T cells was determined 5 and 19 days after dox treatment by determining the percentage of transferred (Vb5À) out of total CD8+ T cells (mean ± SD, 4.41 ± 1.64 versus 9.56 ± 1.9; n = 3). (C) BL signals of tumors (mean ± SD, 892 ± 237 mm3) were determined over time. (A) Spleen cell transfer and dox-treatment (n = 12); (B) spleen cell transfer without dox-treatment (n = 3); (>) dox treatment but no spleen cell transfer (n = 2). (D) Tumor growth kinetics of mice shown in (C). Numbers of mice with rejected or relapd tumors are indicated. (E) Photographs (upper panel) and pictures of BL measurement (middle panel) of Luc+ (right) and rtTA+ skin grafts (left) transplanted on either C57Bl/6 mice (left), RagÀ/À/OT-1 mice reconstituted with Tag-tolerant splenocytes that did not (middle) or did receive and reject a tumor after dox treatment (right). Pictures were acquired more than 3 months after skin transplantation. One reprentative example o
f each group is shown. Number of graft rejections/number of mice in experiment and time of graft rejection in days (d) is given.
C
E
D
derived from two independent resistant clones. All mutations led to amino acid substitutions in positions known to be binding sites of dox or otherwi esntial for tTA function (Hinrichs et al., 1994). Importantly, each tumor had acquired the mutation at a unique tTA-inactivating position or resulting in a different amino acid replacement, showing the high instability of the cancer cells with a emingly unlimited rervoir of genetic variants in large tumors. Endogenous T Cells Only Partially Prevent Relap following TagLuc Inactivation The previous experiments were performed in RagÀ/À mice becau the C57Bl/6 (B6)-derived Tet-TagLuc cells are rejected in B6 mice as a result of the high immunogenicity of Tag. To ask whether tumor cell death by TagLuc inactivation induced endogenous T cells that counteracted the lection of drug-resistant clones, RagÀ/À/OT-1 mice bearing 22-day-old (small) tumors received naive splenocytes (Figure 3A). RagÀ/À/OT-1 mice with t
umor-unrelated transgenic (ovalbumin-specific) T cells were ud to avoid homeostatic proliferation and nonspecific T cell activation. Splenocytes from Tag-tolerant LoxP-Tag 3 Alb-Cre mice were ud, becau transfer of naive B6 splenocytes led to rejection of the tumors by spontaneously activated Tagspecific TE cells (our unpublished obrvation). However, Tet-
TagLuc cells express at least two further antigens, Luc and tTA that are foreign to the T cells and could rve as rejection antigens. Tumors in the prence of LoxP-Tag 3 Alb-Cre splenocytes progressively grew, showing that Luc and tTA are obviously too weak antigens to spontaneously induce T cells in the reconstituted mice. Following dox treatment on day 34, the tumors (R500 mm3) regresd as before, Vb5À (non-OT-1) CD8+ T cells expanded (Figure 3B), and half of the mice completely rejected the tumor, whereas in the other half, BL signals incread and the tumor resumed growth (Figures 3C and 3D). In tho mice that rejected the tumor, we analyzed whether any of the two putative tumor antigens had induced T cells becau of TagLuc inactivation-induced tumor cell death, which contributed to tumor rejection. Therefore, mice received two skin grafts, either from CAG-FLuc or rtTA-CM2 transgenic mice. In both cas, the transgene is expresd by the ubiquitous CAG promoter. For better transplant visibility, albino B6 mice were ud as transgenic skin donors. T cell-reconstituted RagÀ/À/OT-1 mice that had not received Tet-TagLuc cells long term 收条
accepted both skin grafts (Figure 3E). Naive B6 mice rejected the rtTA but long term accepted the Luc skin graft. Reconstituted RagÀ/À/OT-1 mice that had rejected Tet-TagLuc tumors after dox-induced TagLuc inactivation rejected the rtTA but not the Luc skin graft. In the mice, the rtTA skin graft was rejected faster than in naive B6 mice, suggesting that rtTA-specific memory T cells had been induced during tumor cell death (Figure 3E). The data suggested that endogeneous
758 Cancer Cell 20, 755–767, December 13, 2011 ª2011 Elvier Inc.
Cancer Cell
Overcoming Genetic Instability of Cancer
A
B
C
(A) RagÀ/À mice with established Tet-TagLuc tumors (mean ± SD, 837 ± 287 mm3) received 1 3 106 TCR-I TE cells, and changes of TagLuc signal were followed by BL imaging (1 s exposure time). The 美味甜点
time after treatment is indicated in days (d). See also Figure S1. (B) BL signals of TE cell-treated tumors of individual mice (n = 5) were measured over time. (C) Tumor growth kinetics of mice shown in (B). Results in (A–C) are reprentative for three experiments with a total of 10 analyzed mice. (D) RagÀ/À mice with established Tet-TagLuc tumors (mean ± SD, 643 ± 82 mm3) were treated with dox, and relapd tumors (6/6) were subquently treated by TE cells (C; n = 4) or were left untreated (B; n = 2). Changes in BL signal over time of individual mice are shown. (E) Tumor growth kinetics of mice shown in (D). One reprentative of two experiments with a total of eight double-treated mice is shown.
Figure 4. Complete Eradication of large Genetically Unstable Tumors by Adoptive T Cell Therapy with Single Peptide-Specific TE cells
D
E
T cells only partially prevented tumor relap following TagLuc inactivation, even though the tumor expresd a skin graft rejection antigen. Complete Eradication of Large Tumors by Single Peptide-Specific CD8+ Effector T Cells Next, we asked whether adoptive T cell therapy with TE cells directed app下载
against the epitope I of Tag (Staveley-O’Carroll et al., 2003) also lected escape variants, when ud to treat large tumors. The epitope I region is dispensable for the transforming activity of Tag, and epitope I loss variants of murine fibrosarcoma cells could be lected in vitro by specific T cells (Mylin et al., 2007). Also, H-2 loss variants of Tag-transformed cells were found in transiently immune-suppresd mice (Gooding, 1982). Thus, escape variants of Tet-TagLuc cells under TE cell pressure appeared likely, in light of the high genetic instability and large number of tumor cells at the time of treatment. Epitope I-specific (purified TCR-I transgenic) TE cells (Figure S1 available online) were transferred into mice with large established Tet-TagLuc tumors (R500 mm3), and tumor regression was followed by BL imaging. In contrast to dox treatment, no decrea in BL signal was obrved within the first 4 days after TE cell injection, and tumors even incread in size (Figures 4A–4C). Then, between
days 5 and 6, the BL signal dramatically decread and became undetectable, accompanied by hemorrhagic necrosis of the tumor that was not en in the dox-treated tumors. TE cell-treated mice in all cas completely rejected the tumor (Figure 4C). In another experiment, mice with Tet-TagLuc tumors were treated with dox as before and, when large drug-resistant tumors had developed, were treated with TE cells, causing complete and long-term tumor rejection in all mice (Figures 4D and 4E)
. Thus, TE cells with single peptide specificity reject large tumors, even tho that had developed drug resistance, despite large genetic instability. TE Cells but Not TagLuc Inactivation Eradicates Gastric Carcinomas in Mice One cannot exclude that the effective TE cell treatment of TetTagLuc tumors was becau this cell line was generated by in vitro transformation and had not undergone in vivo evolutionary process. Previously, we had obrved in another transgenic mou model with a dormant Tag oncogene that, by stochastic rare events, sporadic tumors developed as a result of somatic mutations or epigenetic events (Willimsky and Blankenstein, 2005; Willimsky et al., 2008). Therefore, TREloxP stoploxPTagLuc mice were crosd to rtTA (tet-on) transactivator transgenic (rtTA-CM2) mice. A small cohort of double transgenic mice (with the stop castte prent) was kept on dox, and BL signals were determined over time (Figure 5A). A distinct BL signal appeared in one mou after 411 days of dox treatment that derived from a sporadic gastric carcinoma that was Luc and Tag positive (Figures 5B and 5C). Proliferation of a cell line (TC200.09) derived from this tumor depended on the prence of dox (Figure 5D). Large established TC200.09 tumors were treated with dox withdrawal or TE cells. TagLuc inactivation led
Cancer Cell 20, 755–767, December 13, 2011 ª2011 Elvier Inc. 759
