`
`Rapamycin inhibits primary and metastatic tumor growth
`by antiangiogenesis: involvement of vascular
`endothelial growth factor
`
`MARKUS GUBA, PHILIPP VON BREITENBUCH, MARKUS STEINBAUER, GUDRUN KOEHL,
`STEFANIE FLEGEL, MATTHIAS HORNUNG, CHRISTIANE J. BRUNS, CARL ZUELKE, STEFAN FARKAS,
`MATTHIAS ANTHUBER, KARL-WALTER JAUCH & EDWARD K. GEISSLER
`
`Department of Surgery, University of Regensburg, Regensburg, Germany
`Correspondence should be addressed to M.G.; email: markus.guba@klinik.uni-regensburg.de,
`or E.K.G.; email: edward.geissler@klinik.uni-regensburg.de
`
`Conventional immunosuppressive drugs have been used effectively to prevent immunologic re-
`jection in organ transplantation. Individuals taking these drugs are at risk, however, for the de-
`velopment and recurrence of cancer. In the present study we show that the new
`immunosuppressive drug rapamycin (RAPA) may reduce the risk of cancer development while
`simultaneously providing effective
`immunosuppression. Experimentally, RAPA
`inhibited
`metastatic tumor growth and angiogenesis in in vivo mouse models. In addition, normal im-
`munosuppressive doses of RAPA effectively controlled the growth of established tumors. In con-
`trast, the most widely recognized immunosuppressive drug, cyclosporine, promoted tumor
`growth. From a mechanistic perspective, RAPA showed antiangiogenic activities linked to a
`decrease in production of vascular endothelial growth factor (VEGF) and to a markedly inhibited
`response of vascular endothelial cells to stimulation by VEGF. Thus, the use of RAPA, instead
`of cyclosporine, may reduce the chance of recurrent or de novo cancer in high-risk
`transplant patients.
`
`Among the most serious complications of general immuno-
`suppressive therapy in organ transplantation is the high risk
`of the recurrence of neoplastic tumors and the development
`of de novo cancer. For example, recent studies in individuals
`with bronchioloalveolar carcinoma who received a lung
`transplant showed a 57% tumor recurrence rate1. Among in-
`dividuals receiving a liver transplant for cholangiocarcinoma,
`51% have tumor recurrence within 2 years2. With regard to de
`novo malignancies, a conservative report indicates that im-
`munosuppressed organ allograft recipients have a 3–4-fold in-
`creased risk of developing cancer
`in general and a
`20–500-fold higher incidence of certain types of cancer3.
`Cancer has therefore become a major cause of death in pa-
`tients otherwise successfully treated by organ transplanta-
`tion. One approach to addressing this problem is to identify
`drugs with effective immunosuppressive but low proneoplas-
`tic, or even some antineoplastic, properties. In the present
`study we present evidence that rapamycin (RAPA) may fill
`these diverse needs.
`RAPA is a bacterial macrolide with antifungal and immuno-
`suppressant activities4,5. It forms a complex with the FK bind-
`ing protein complex (FKBP-12) that binds with high affinity to
`the mammalian target of rapamycin (mTOR). This interaction
`causes dephosphorylation and inactivation of p70S6 kinase
`which, when activated, stimulates the production of riboso-
`mal components necessary for protein synthesis and cell-cycle
`progression6. This activity, which effectively blocks IL-2 stim-
`ulation of lymphocyte division, is the basis for the recent suc-
`cessful clinical use of RAPA to prevent allograft rejection in
`
`organ transplantation7,8. In comparison, the most widely rec-
`ognized immunosuppressive drug, cyclosporine (CsA), binds
`to a different intracellular molecule, cyclophilin, ultimately
`leading to inhibition of T-cell proliferation by preventing IL-2
`transcription9. Notwithstanding the mechanistic effects of
`these drugs on T cells, concerns about the side effects of im-
`munosuppression on cancer development in transplant pa-
`tients have brought a new scientific perspective and focus to
`their study. Most evidence suggests that conventional CsA im-
`munosuppression promotes, rather than inhibits, the develop-
`ment of cancer10. Here we describe experiments indicating
`that immunosuppressive doses of RAPA have an entirely oppo-
`site effect, whereby the drug’s potent antiangiogenic activities
`strongly inhibit tumor growth in mice. We also provide evi-
`dence that the antiangiogenic effect of RAPA is linked to re-
`duced production of VEGF and to blockage of VEGF-induced
`endothelial cell signaling.
`
`Metastatic tumor growth in the liver
`To determine the effect of RAPA and CsA on metastatic tumor
`growth, we injected syngenic CT-26 adenocarcinoma cells in-
`traportally into BALB/c mice, simulating metastasis of colon
`cancer to the liver. Histologic analysis of hepatic tumor-re-
`placement area showed a marked decrease in the metastatic
`area in RAPA-treated mice compared to saline controls (Fig.
`1a). Livers from RAPA-treated mice had small, avascular, dis-
`seminated metastatic foci (Fig. 1b). In contrast, CsA-treated
`mice had a high mean tumor replacement area and extensive
`neovascularization within the fast-growing metastases. Thus,
`
`128
`
`NATURE MEDICINE • VOLUME 8 • NUMBER 2 • FEBRUARY 2002
`
`ARGENTUM
`Exhibit 1015
`
`©2002 Nature Publishing Group http://medicine.nature.com
`
`000001
`
`
`
`ARTICLES
`
`chambers. As expected, RAPA-mediated inhibition of angio-
`genesis was identifiable on day 11, with measurements show-
`ing a smaller vascularized tumor area (Fig. 3a) and lower
`vascular density (Fig. 3b) in treated mice than in controls.
`Promotion of angiogenesis by CsA was evident from the high
`tumor vessel density, although the percentage of tumor area
`containing vessels (vascular area) was not significantly differ-
`ent. Corresponding measurements of tumor dimensions
`showed a growth dependency on angiogenesis that was re-
`flected by a small tumor area and volume with RAPA treat-
`ment versus control (Fig. 3c and d). In contrast, the extensive
`angiogenesis in CsA-treated mice supported tumors with a
`large area and volume. Thus, results from this model indicate
`that immunosuppressive doses of RAPA, but not CsA, can sup-
`press tumor growth by inhibiting angiogenesis.
`
`Influence of rapamycin on VEGF production
`Because VEGF is one of the central regulators of vessel devel-
`opment, we were interested in testing whether RAPA could af-
`fect angiogenesis by influencing its production in vitro. An
`enzyme-linked immunosorbent assay (ELISA) of CT-26 adeno-
`carcinoma cell culture supernatants showed that RAPA inhib-
`ited the secretion of VEGF, whereas CsA did not have a
`significant effect (Fig. 4a). To rule out the possibility that the
`VEGF effect was specific to CT-26 cells, the same effect of
`RAPA was confirmed with B16 melanoma cells (Fig. 4a).
`Encouraged by these in vitro results, we tested whether the
`therapeutic concentrations of RAPA used after organ transplan-
`tation could also result in reduced VEGF in vivo. Serum samples
`taken either from non-tumor-bearing mice or from mice 10
`days after they had received an intraportal injection of CT-26
`tumor cells showed that RAPA treatment led to a reduction in
`VEGF (Fig. 4b). Therapeutic use of CsA caused VEGF to increase
`slightly, possibly as a result of production from the increased
`tumor cell mass. For the same reason, we cannot rule out the
`possibility that the decreased tumor mass in RAPA-treated mice
`could have contributed to the lowering of serum VEGF. VEGF
`concentrations were lower, however, in RAPA-treated mice
`than in mice without tumors, indicating that endogenous
`VEGF production by the host was also affected.
`To investigate whether RAPA inhibits angiogenesis through
`the downregulation of VEGF mRNA in tumor cells, or
`through upstream VEGF regulators such as, for example, hy-
`
`b
`
`*
`
`CsA
`10 mg/kg
`
` *
`
`RAPA
`1.5 mg/kg
`
`Control
`
`50
`45
`40
`35
`30
`25
`20
`15
`10
`
`05
`
`Hepatic replacement area (%)
`
`a
`
`Fig. 1
`Rapamycin inhibits, whereas CsA
`promotes, metastatic tumor growth in liver.
`Hematoxylin- and eosin-stained liver sections
`were evaluated for metastatic tumor growth
`10 d after intraportal injection of CT-26 cells
`into BALB/c mice treated daily with saline,
`RAPA or CsA. a, Intrahepatic metastases were
`measured and the hepatic replacement area
`determined. Each bar represents the mean ±
`s.e.m. from 7 mice per group. *, P < 0.05 ver-
`sus control. b, Histological sections of tumor-bearing livers on day 10
`from mice treated with saline (left), RAPA (middle) or CsA (right). The
`upper and lower panels show representative low (bar = 200 µm) and high
`magnification (bar = 50 µm) views, respectively. Asterisks indicate tumor
`vessels and arrowheads indicate tumor borders.
`
`in this liver metastasis model, RAPA and CsA had opposite ef-
`fects on metastatic tumor growth.
`
`Tumor growth in the dorsal skin-fold chamber
`To further test how immunosuppressive drugs affect tumor de-
`velopment, we used intravital microscopy to examine the ef-
`fect of RAPA and CsA on tumor growth and angiogenesis in
`dorsal skin-fold chambers. Photomicrographs representative of
`the time-course of tumor growth and neovascularization after
`the implantation of CT-26 cells into the transparent chamber
`of control, RAPA- and CsA-treated mice are shown in Figure 2.
`In the saline-treated controls, progressive tumor growth oc-
`curred over an observation period of 11 days. During the avas-
`cular phase of tumor development, tumor cell masses became
`visible as a shadowy area by days 3 and 5 (Fig. 2 a-c). Also by
`day 5, tumors began to induce a strong angiogenic response, as
`evidenced by the development of a plexus of newly formed
`tumor vessels that continued to expand throughout the 11-day
`observation period (Fig. 2d–g). Compared to controls, RAPA-
`treated mice showed less growth and neovascularization of the
`tumors. The inhibition of tumor growth was paralleled by a de-
`crease in neoangiogenesis, which was visually evident from the
`lack of tumor vessels on days 7 and 9 (Fig. 2d and e). By day 11,
`vessel formation was visible, but predominantly in the central
`tumor area (Fig. 2f and g). The same low-magnification pho-
`tomicrograph shows that, in contrast to control tumors (Fig.
`2f), an antiangiogenic effect of RAPA was maintained in the pe-
`ripheral tumor zone. This general pattern of tumor inhibition
`by RAPA was clearly different from the effects of CsA. CsA
`treatment promoted relatively early neovascularization of tu-
`mors in mice (Fig. 2c), which continued to progress at an accel-
`erated rate between days 7 and 11 (Fig. 2d–f) as compared to
`control mice. Furthermore, by day 11, tumors from CsA-treated
`mice developed an advanced vascular network (Fig. 2f and g).
`To confirm these visual observations, we carried out multi-
`ple quantitative analyses on the tumors in dorsal skin-fold
`
`©2002 Nature Publishing Group http://medicine.nature.com
`
`NATURE MEDICINE • VOLUME 8 • NUMBER 2 • FEBRUARY 2002
`
`129
`
`000002
`
`
`
`Fig. 2 Rapamycin inhibits tumor angiogenesis, whereas CsA stim-
`ulates tumor neovascularization. Angiogenesis and tumor develop-
`ment were monitored
`in dorsal skin-fold chambers after
`implantation of CT-26 cells in BALB/c mice. a–f, Photomicrographs
`of representative tumors from mice treated with saline (left column),
`RAPA (middle column) or CsA (right column) on days 1, 3, 5, 7, 9
`and 11, respectively. In the avascular stage (day 1–3), there was
`some initial growth of tumors (shadowy areas) for all treatments.
`After day 5, tumors in saline and CsA-treated mice induced a rela-
`tively large, homogenously distributed vascular plexus compared to
`RAPA-treated mice. Dotted lines indicate the vascularized zone in re-
`lation to the entire tumor area on day 9. On day 11, an advanced hi-
`erarchy of small vessels branching into larger draining vessels was
`evident in CsA-treated mice, compared to controls or RAPA-treated
`mice. g, High-magnification views of the same tumors on day 11
`show normal, interconnected, small tumor vessels in the control
`mouse. In contrast, tumor vessels in the RAPA-treated mouse show
`blunt ends, few connections and abrupt changes in diameter. Well-
`developed vessels can be seen in the tumor from the CsA-treated
`mouse. Scale bar, 1 mm (a–f); 0.1 mm (g).
`
`tration tested (1 µg/ml, Fig. 5a). Human umbilical-
`cord vein endothelial cells (HUVECs) were very sen-
`sitive to RAPA, with a significant effect at 0.01
`µg/ml—again raising the possibility that VEGF may
`be involved in the RAPA effect. Our hypothesis,
`however, was based on the assumption that ade-
`quate VEGF was present but VEGF stimulation of en-
`dothelial cells was inhibited. To test this theory,
`HUVECs cultured under minimal serum-deprived
`conditions, were stimulated with recombinant VEGF
`with or without RAPA. Results showed that RAPA
`markedly reduced VEGF-induced HUVEC prolifera-
`tion in a dose-dependent manner (Fig. 5a). We there-
`fore examined the effects of RAPA on VEGF
`stimulation of HUVEC tubular formation morpho-
`genesis and found that VEGF-induced HUVEC tubu-
`lar
`formation was completely abrogated by
`concentrations of RAPA as low as 0.01 µg/ml (Fig.
`5b). Thus, VEGF-dependent HUVEC proliferation
`and morphogenesis seem to be very sensitive to the
`effects of RAPA.
`
`ARTICLES
`
`a
`
`b
`
`c
`
`d
`
`e
`
`f
`
`g
`
`©2002 Nature Publishing Group http://medicine.nature.com
`
`poxia-inducible factor 1α (HIF-1α)11,12 and transforming
`growth factor-β (TGF-β)13–15, we cultured B16 tumor cells with
`and without 0.1 µg/ml RAPA for 10 h and isolated total RNA
`for real-time RT–PCR. Amounts of VEGF mRNA were slightly
`lower (4-fold) in RAPA-treated cells than in controls, but nei-
`ther HIF-1α nor TGF-β mRNA seemed to be up- or down-regu-
`lated in the presence of RAPA (Fig. 4c). A higher dose of RAPA
`(1 µg/ml) did not further decrease VEGF mRNA and did not
`alter HIF-1α or TGF-β mRNA concentrations (data not
`shown). Together, these studies suggest that RAPA treatment
`reduces but does not decisively block VEGF production by
`tumor cells.
`
`Effect of RAPA on proliferation and VEGF-mediated angiogenesis
`We next explored whether the antitumor activity of RAPA
`could be related to an antiproliferative effect acting either di-
`rectly on tumor cells or on vascular endothelial cells.
`Bromodeoxyuracil (BrdU) incorporation experiments showed
`that CT-26 and B16 tumor cell proliferation decreased moder-
`ately in the presence of RAPA, but only at the highest concen-
`
`Treatment of tumors with rapamycin
`We next tested the effects of RAPA on the growth of estab-
`lished subcutaneous CT-26 tumors (Fig. 6a). The tumors
`rapidly grew in untreated control mice, resulting in the death
`of all mice by 2 weeks. By contrast, mice receiving relatively
`low doses of RAPA (1.5 mg/kg/d) showed a slightly delayed in-
`crease in tumor size during the first 20 days. At this time point
`the tumors began to regress markedly and the animals had a
`100% survival rate. Later, widespread tumor necrosis occurred
`(Fig. 6d). The lowest dose of RAPA (0.15 mg/kg/d) delayed
`tumor growth slightly, but the mice died by day 23 and the tu-
`mors never entered a regression phase. Notably, a high dose of
`RAPA (15 mg/kg/d) caused a more pronounced delay in tumor
`development during the first 3 weeks, but after this the tumors
`began to grow again rapidly and the mice died shortly there-
`after. An even higher dose (30 mg/kg/d) resulted in the death
`of all mice by day 17 (data not shown). The antitumor effect of
`RAPA was not specific to CT-26 tumors or to BALB/c mice, as a
`similar treatment protocol for subcutaneous B16 tumors in
`C57BL/6J mice also caused a marked decrease in tumor vol-
`
`130
`
`NATURE MEDICINE • VOLUME 8 • NUMBER 2 • FEBRUARY 2002
`
`000003
`
`
`
`ARTICLES
`
`*
`
`Control
`
`0.30
`
`0.25
`
`0.20
`
`0.15
`
`0.10
`
`0.05
`
`0.00
`
`Tumor volume (cm3)
`
`d
`
`*
`
`*
`
`*
`
`*
`
`*
`
`1
`
`3
`
`11
`
`13
`
`25
`
`20
`
`15
`
`10
`
`05
`
`Tumor area (mm2)
`
`*
`
`c
`
` *
`
`Control
`
`200
`
`175
`
`150
`
`125
`
`100
`
`75
`
`50
`
`25
`
`0
`
`Microvascular density (cm-1)
`
`b
`
`*
`
`Control
`
`CsA
`10 mg/kg
`
`100
`
`75
`
`50
`
`25
`
`0
`
`Vascular area (% tumor area)
`
`a
`
`CsA
`10 mg/kg
`
` *
`
`RAPA
`1.5 mg/kg
`
`RAPA
`
`1.5 mg/kg
`
`RAPA
`1.5 mg/kg
`
`CsA
`10 mg/kg
`
`9
`7
`5
`Time (d)
`
`Fig. 3 CT-26 tumors in transparent chambers are markedly smaller
`and have less vascularization with rapamycin compared to CsA treat-
`ment. Tumor size and vascularization were determined after inocula-
`tion of CT-26 cells into the transparent chambers of control and
`drug-treated mice. a, Effect of RAPA or CsA on vascular area on day 11.
`b, Effect of RAPA or CsA on microvascular density on day 11. c, Tumor
`
`area in mice treated until day 13 with saline (쏔), RAPA (쎲) and CsA
`(쐽). The vertical line indicates the normal transition from prevascular
`angiogenesis-independent to angiogenesis-dependent growth. d, On
`day 13, tumors were excised and the volume determined. All results
`are the mean ± s.e.m. of 7 mice per group. *, P < 0.05 versus saline
`control.
`
`ume after a 3–4-week period (Fig. 6b). Here also, the dominant
`feature of the B16 tumors after 4 weeks was massive tumor
`necrosis (Fig. 6e). These data indicate that RAPA significantly
`inhibits the growth of established vascularized tumors, and
`this effect is best realized with relatively low (normal im-
`munosuppressive) doses of drug.
`To test whether orthotopically placed tumors respond in a
`similar way to the effects of RAPA, we injected CT-26 tumor
`cells into the cecal wall of BALB/c mice and treated the
`mice with RAPA. Excised tumors from RAPA-treated mice
`on day 12 were 87% smaller than those from control mice
`(Fig. 6c and f). In addition, immunohistologic staining of
`tumor endothelial cells in RAPA-treated mice showed a
`marked decrease in the number of blood vessels (Fig. 6g).
`Together, these data indicate that the antiangiogenic–antitu-
`mor effect of RAPA occurs with orthotopically grown tumors
`and that this effect is not dependent on a subcutaneous
`tumor location.
`
`Discussion
`Our results indicate that RAPA and CsA have contrasting ef-
`fects on tumor development that could be crucial in address-
`ing the problem of cancer in immunosuppressed patients. In
`
`our tumor cell metastasis studies, we found that doses of RAPA
`roughly equivalent to those used in organ transplantation
`strongly inhibited tumor growth in the mouse liver. In con-
`trast to the abnormally large, vascularized tumor masses seen
`with CsA treatment, metastatic foci in the liver of RAPA-
`treated mice had few blood vessels and were too small to re-
`quire angiogenesis16,17. We considered two basic theories to
`explain the potential antitumor effects of RAPA. First and
`most simply, RAPA might directly inhibit the proliferation of
`tumor cells. Our results did not strongly support this, how-
`ever, as RAPA inhibited tumor proliferation only slightly at
`the highest concentration tested (1 µg/ml) and had no appar-
`ent effect in a more relevant therapeutic range (<1.0 µg/ml).
`The blood concentrations of the drug in mice receiving 1.5
`mg/kg/d were approximately 0.04 µg/ml at the trough and ap-
`proximately 0.8 µg/ml at 2 hours after administration (data
`not shown). RAPA has antiproliferative activity against nor-
`mal nonimmunologic cells6 and colon-38 tumor cells18, but
`only at extremely high doses of drug (100–400 mg/kg/d in-
`traperitoneally), consistent with our data. Therefore, the in-
`hibitory effect of normal immunosuppressive doses of RAPA
`on tumor growth in our model is probably not due to an an-
`tiproliferative effect on tumor cells.
`
`©2002 Nature Publishing Group http://medicine.nature.com
`
`Fig. 4 Rapamycin causes a decrease in VEGF produc-
`tion that correlates with reduced VEGF mRNA. a, CT-26
`(left) and B-16 (right) tumor cells were cultured with
`and without RAPA or CsA, and culture supernatants
`were tested for VEGF by ELISA. Each value represents
`the mean ± s.e.m. of 3 experiments performed in dupli-
`cate. *P < 0.05 versus control. b, Serum VEGF concen-
`trations in non–tumor bearing mice (no tumor) and
`CT-26 tumor-bearing mice treated with saline (con-
`trol), RAPA or CsA on day 10. Each value is the mean ±
`s.e.m. of 7 mice per group. *, P < 0.05 versus saline con-
`trols with tumor; #, P < 0.05 versus ‘no tumor’ control. c,
`To test the effect of RAPA on relative VEGF, TGF-β and
`HIF-1α mRNA concentrations in tumor cells, B16 cells
`were cultured in the absence and presence of 0.1 µg/ml
`drug for 10 h. Total RNA was used to generate cDNA
`and real-time RT–PCR was performed. RAPA treatment
`(solid line) versus no drug (dotted line) are shown for
`VEGF, TGF-β, HIF-1α and β-actin (housekeeping gene),
`and corresponding Cp values are given. Similar results
`were obtained in 3 additional experiments.
`
`*
`
`*
`
`*
`#
`
`No tumor
`
`Control
`
`
`
`RAPA
`1.5 mg/kg
`
`CsA
`10 mg/kg
`
`100
`90
`80
`70
`60
`50
`40
`30
`20
`10
`0
`
`Serum VEGF (pg/ml)
`
`*
`
`*
`
`b
`
`*
`
`*
`
`500
`
`400
`
`300
`
`200
`
`100
`
`Control
`0.1 µ g/ml
`1 µ g/ml
`0.1 µ g/ml
`RAPA
`1 µ g/ml
`CsA
`RAPA
`CsA
`
`0
`
`
`
`Control
`0.1 µ g/ml
`RAPA
`RAPA
`1 µ g/ml
`
`800
`
`700
`
`600
`
`500
`
`400
`
`300
`
`200
`
`100
`
`0
`
`VEGF secretion (pg/ml)
`
`a
`
`c
`
`NATURE MEDICINE • VOLUME 8 • NUMBER 2 • FEBRUARY 2002
`
`131
`
`000004
`
`
`
`*
`
`ARTICLES
`
`Serum
`deprived
`
`VEGF
`25 ng/ml
`
`28
`
`)
`
`24
`
`20
`
`16
`
`12
`
`048
`
`)
`
`Microvascular density (cm-1
`
`b
`
`*
`
`*
`
`*
`
`*
`
`*
`
`*
`
`RAPA, 0.01 µg/ml
`
`RAPA, 0.1 µg/ml
`
`
`RAPA, 1 µg/ml
`
`110
`100
`90
`80
`70
`60
`50
`40
`30
`20
`10
`0
`
`BrdU incorporation (% of control)
`
`a
`
`*
`
`*
`
`RAPA
`RAPA
`RAPA
`
`1 µg/ml
`0.01 µg/ml
`0.1 µg/ml
`+ VEGF (25 ng/ml)
`
`ever, rule out the possibility that RAPA dis-
`turbs angiogenesis-dependent intracellular
`pathways downstream of these molecules, or
`that the mRNA translation or protein degra-
`dation rates of HIF-1α or TGF-β are modified
`by RAPA. These arguments notwithstanding,
`we suspected that the rather modest decrease
`in VEGF production by tumor cells could not
`fully account for the potent antiangiogenic
`effect. In addition, because it can be argued
`that many tumor cells produce excessive
`amounts of proangiogenic factors, including
`VEGF16,22, a moderate reduction in VEGF pro-
`duction may not be sufficient to impede
`neoangiogenesis fully. Therefore, assuming
`adequate amounts of VEGF, we tested
`whether RAPA could be inhibiting angiogen-
`esis at a second level, where receptor-medi-
`ated stimulation of vascular endothelial cells
`occurs. This hypothesis was based on evi-
`dence that the PI3K-p70S6 kinase intracellu-
`lar signaling pathway is required for VEGF
`stimulation of endothelial cells23,24. In this
`regard, one study25 has shown that RAPA in-
`hibition of p70S6 kinase blocks PMA-induced tube formation
`in three-dimensional HUVEC cultures. VEGF receptor-2 sig-
`naling in pig endothelium has also been linked to the activa-
`tion of the PI3K-p70S6 pathway26. Indeed, our data indicating
`that RAPA markedly inhibited VEGF-dependent HUVEC pro-
`liferation and completely abrogated VEGF-induced tubular
`formation favors this mechanistic explanation. RAPA concen-
`trations that caused these in vitro effects on endothelial cells
`were well within the range measured in serum after RAPA
`treatment for tumors and also are similar to those used in
`organ transplantation to prevent allograft rejection.
`The convincing antiangiogenic effects of RAPA in our earlier
`studies led us to test if this drug had antitumor effects in a
`more clinically relevant situation. The potency of RAPA in this
`respect was revealed by its ability to control and effectively
`shrink established subcutaneous and orthotopic tumors in
`mice. The hypothesis that RAPA affects tumor growth primar-
`ily through antiangiogenesis is supported by our data showing
`fewer CD31-positive blood vessels in orthotopic tumors in
`mice treated with RAPA compared to untreated controls. The
`theory is further supported by our observation of an atypical
`dose-response effect on established tumors. A relatively low,
`non-cytotoxic dose of RAPA caused the tumors to regress only
`after they became increasingly dependent on angiogenesis.
`This observation is consistent with the low-dose, delayed ef-
`fects on tumors typically seen with other reported antiangio-
`genic treatment protocols27–30. The greater delay in the early
`development of established tumors subjected to high doses of
`RAPA may be explained by a modest antiproliferative effect, as
`has been reported with very high in vivo doses of RAPA18. We
`speculate that antiproliferative effects associated with these
`high RAPA concentrations may slow early tumor growth but
`also weaken the general condition and immunity of the
`mouse to a point where tumor growth has the advantage.
`Together, these experiments suggest that immunosuppressive
`doses of RAPA may have long-term antiangiogenic effects that
`control tumor growth.
`
`Fig. 5
`Rapamycin inhibits tumor cell proliferation and VEGF-dependent HUVEC proliferation
`and tubular formation. a, The effect of RAPA on the proliferation of CT-26 (쏔) and B16 (쐽)
`tumor cells and HUVEC cells (쐽) determined by BrdU incorporation (shown as % of control
`BrdU incorporation). An additional experiment was performed with HUVECs whereby serum-
`deprived cells were stimulated with recombinant VEGF (
`). BrdU incorporation was mea-
`sured and is expressed as the percent of control (VEGF-stimulated cultures). All data represent
`the mean ± s.e.m. from 3 experiments. *, P < 0.05 versus control. Compared to serum depriva-
`tion, optimal culture supplementation increased BrdU incorporation by 79%, and recombinant
`VEGF in serum-deprived medium increased incorporation by 61%. b, The effect of RAPA on
`VEGF-induced HUVEC tubular formation as tested by culturing HUVECs in serum-deprived
`medium only or in serum-deprived medium containing recombinant VEGF ± RAPA; vascular
`tube density was measured 8 h later. Results represent the mean ± s.e.m. of 3 experiments. *, P
`< 0.05 versus VEGF with no drug.
`
`Our second theory about the RAPA antitumor effect was
`based on our early observation that small liver metastases in
`RAPA-treated mice had few blood vessels. From this, we pre-
`dicted that RAPA could mediate its effect through inhibition
`of tumor vascularization. Using a dorsal skin-fold chamber
`model, which is well established for the specific study of
`tumor neoangiogenesis16,17,19, we found that in vivo treatment
`with RAPA strongly inhibits tumor angiogenesis and associ-
`ated tumor growth. Also consistent with the liver metastasis
`experiments, CsA treatment caused a completely opposite ef-
`fect, where relatively large tumors formed and were sup-
`ported by advanced neovascularization. One consideration in
`the interpretation of data from the dorsal skin-fold chambers
`is that tumors developed in an ectopic location. To elucidate
`possible host microenvironmental influences20,21, it would be
`useful to carry out angiogenesis-specific studies of tumors in
`an orthotopic location. In regard to this question, we note
`that CD31 staining of vascular endothelial cells in orthotopi-
`cally placed CT-26 tumors showed a significant reduction of
`tumor vessels with RAPA treatment. Data presented here pro-
`vide the first clear evidence that normal immunosuppressive
`concentrations of RAPA and CsA have divergent effects on
`angiogenesis and that RAPA has potent tumor inhibiting an-
`tiangiogenic effects.
`From a mechanistic perspective, our study suggests the an-
`tiangiogenic effect of RAPA is related to VEGF antagonism at
`two separate levels. At one level, VEGF production is reduced.
`This is evidenced by in vitro experiments showing that RAPA
`diminishes VEGF secretion by tumor cells, and is further cor-
`roborated by the reduced VEGF mRNA concentrations mea-
`sured in the same cells. This general observation was also
`evident in vivo, where serum VEGF was lower in mice treated
`with RAPA. Although we speculated that the decrease in VEGF
`production could be related to the reported upstream regula-
`tory effects of either HIF-1α11,12 or possibly TGF-β13–15, we found
`no evidence for quantitative changes in these molecules at the
`mRNA level in the presence of RAPA. Our results do not, how-
`
`132
`
`NATURE MEDICINE • VOLUME 8 • NUMBER 2 • FEBRUARY 2002
`
`©2002 Nature Publishing Group http://medicine.nature.com
`
`000005
`
`
`
`ARTICLES
`
`d
`
`Control
`
`
`
`Rapamycin
`1.5 mg/kg/d
`
`900
`800
`700
`600
`500
`400
`300
`200
`100
`0
`
`Tumor Volume (mm3)
`
`c
`
`20,000
`18,000
`16,000
`14,000
`12,000
`10,000
`8,000
`6,000
`4,000
`2,000
`0
`
`Tumor Volume (mm3)
`
`b
`
`900
`800
`700
`600
`500
`400
`300
`200
`100
`0
`
`Tumor volume (mm3)
`
`a
`
`0
`
`5 10 15 20 25 30 35 40 45
`Time (d)
`
`5
`
`10
`
`20
`15
`Time (d)
`
`25
`
`30
`
`Fig. 6 Rapamycin treatment controls the growth of tumors
`in mice. Mice were injected subcutaneously with CT-26 or
`B16 cells and tumors were allowed to grow for 7 d before ini-
`tiation of treatment with 0 (saline; 쏔), 0.15 (왓), 1.5 (쎲), 4.5
`(앳) or 15 (왕) mg/kg/d RAPA. Another group of experiments
`involved orthotopic (cecum)
`injection of CT-26 cells.
`a, CT-26 subcutaneous tumor volume over 4–6 wk, shown as
`the mean ± s.e.m. from 6–8 mice. b, B16 subcutaneous tumor
`volume (obtained as in a). c, Tumor volume on day 12 for or-
`thotopically placed CT-26 tumors (n = 8 per group, P = 0.003,
`RAPA versus control). d, Photographs of the subcutaneous
`tumor site from CT-26-injected BALB/c mice that had received
`either saline (day 14) or RAPA (day 35). Top, control; bottom,
`+1.5 mg/kg RAPA. e, Photographs of the tumor site 28 d after
`subcutaneous injection of B16 cells into saline- (control) or
`RAPA-treated C57BL/6J mice. f, Photographs of orthotopically
`placed CT-26 tumors from a control and a RAPA-treated
`mouse on day 12. g, Tumors were stained for endothelial cells
`using anti-CD31 antibody (10× objective view). Quantitative
`analysis of 2 tumors from each group shows 564 ± 10 ves-
`sels/mm2 in control versus 240 ± 5 vessels/mm2 in RAPA-
`treated mice (P = 0.0005).
`
`e
`
`B16
`
`f
`
`CT-26 orthotopic
`
`1 cm
`
`Control
`
`+ 1.5 mg/kg
`RAPA
`
`+ 4.5 mg/kg
`RAPA
`
`Control
`
`+1.5 mg/kg
`RAPA
`
`g
`
`Although our study focused primarily on the effects of
`RAPA, our findings with CsA are also of interest as a contrast
`to reports indicating that CsA has antiangiogenic effects. One
`report indicated that endothelial-cell functions, including
`proliferation, are inhibited by CsA in chick embryo chorioal-
`lantoic membranes31. Corneal neovascularization in rats can
`also be inhibited by CsA32. In the latter study, however, CsA
`acted in an inflammatory situation created by a surgical
`trauma, which might have overstated its role in pathological
`angiogenesis. In contrast, our experiments showed that CsA
`had tumor-promoting activity. Consistent with our findings, a
`recent report33 showed that CsA promotes the invasive charac-
`teristics of adenocarcinoma cells, and thus cancer progression,
`through a TGF-β-dependent mechanism. We used CsA con-
`centrations similar to those used by these researchers, whereas
`the chick embryo experiments involved higher concentra-
`tions. These findings conform to previous reports indicating
`that pro- or antiangiogenic effects can depend on the concen-
`tration of TGF-β15,30. We did not, however, observe TGF-β
`mRNA up- or downregulation in the presence of RAPA.
`Whether RAPA influences TGF-β intracellular signaling crucial
`to angiogenesis is not known. Our findings combined with
`those of the adenocarcinoma cell study33 suggest, however,
`that CsA may promote cancer development by acting on at
`least two physiologic levels—by promoting angiogenesis and
`by increasing the invasiveness of tumor cells.
`In summary, we have shown in different mouse models that
`the immunosuppressive drug RAPA may have inhibitory ef-
`fects on the development of cancer. The tumor inhibition me-
`
`diated by RAPA seems to be based on antiangiogenic activity,
`which correlates with impaired VEGF production and block-
`age of VEGF-induced vascular endothelial cell stimulation.
`These findings are of considerable interest because treatment
`with immunosuppressive drugs is generally thought to pro-
`mote rather than inhibit cancer development and progression,
`as we found with CsA in our experiments. Because the devel-
`opment and recurrence of cancer are among the most serious
`side effects of traditional immunosuppression, RAPA might be
`an effective alternative to conventional CsA-based therapies
`in high-risk patients, particularly in transplant medicine
`where de novo and recurrent neoplasms are a significant cause
`of morbidity and mortality.
`
`Methods
`Mouse and tumor-cell culture. Male 6–8-week BALB/c and C57BL/6J
`mice (Charles River, Sulzfeld, Germany) were used for all experiments
`and each procedure was approved by the regional authorities according
`to German animal-care regulations.
`CT-26 cells used were derived from a mouse BALB/c colon adenocar-
`cinoma34; B16-F10 melanoma cells were derived from C57BL/6J mice
`(D. Männel, University of Regensburg). Cells were maintained by cell
`culture in RPMI with 10% FBS.
`Administration of RAPA and CsA. RAPA (Wyeth Pharma, Münster,
`Germany) was normally administered at doses of 1.5 mg/kg/d intraperi-
`toneally, beginning on day 0 or 7 relative to tumor implantation. In
`dose-response experiments, the drug was given intraperitoneally at
`doses from 0.15 to