Introduction

Insulin-like growth factor binding proteins (IGFBPs) bind both IGF-1 and IGF-2 with high affinity and specificity, and modulate bioactivity of these growth factors [1–4]. IGFBP-3 has been identified as the major carrier of IGFs in the circulation and also plays a role in autocrine and paracrine growth control and apoptosis. It has been shown that exogenous IGFBP-3 can slow the growth of breast cancer and other tumor cells in culture by sequestration of IGFs, thus reducing their binding to receptors [5–8], and blocking their anti-apoptotic activity [9]. IGFBP-3 can also induce apoptosis and growth inhibition via IGF-1 independent mechanisms in various cell systems [3, 10–13]. IGFBP-3 was also shown to exert tumor growth inhibition by suppressing angiogenesis [14]. In addition, up-regulation of IGFBP-3 may contribute to the anti-proliferative action of some of the agents recommended for chemoprevention [15–20] or chemotherapy [3, 21, 22].

Transgenic and knockout mouse models have been developed to characterize the roles of IGFBPs [23–26]. In general, IGFBP-3 knockout mice do not exhibit a distinct phenotype under standard conditions [27], but one study using a c-myc-driven transgenic prostate cancer model has shown that in the absence of IGFBP-3, castration-induced apoptosis is attenuated and metastasis is accelerated [28].

In this study, we used an IGFBP-3 knock-out model to determine the role of IGFBP-3 in a chemically induced mammary gland carcinogenesis model and in a metastasis model.

Materials and methods

Animals

IGFBP-3 knockout (KO) mice were generated at Lexicon Pharmaceuticals Inc., The Woodlands, TX, as previously described [29]: Heterozygous male and female mice on a 129/SvEv-C57BL/6 background were bred to generate IGFBP-3 knockout and wild-type (WT) mice (which were used for the DMBA-induced mammary tumor study). To obtain IGFBP-3 KO mice on a pure Balb/c background for use in the metastasis assay, heterozygous IGFBP-3 female mice were first bred with a Balb/c male mouse for more than 10 generations of backcrossing, and then homozygous IGFBP-3 knockout and wild-type animal were generated. Primers for genotyping are shown in Table 1 as described [29]. All mice were maintained in a climate-controlled environment with conditions of 14 h light:10 h darkness, a temperature of 22 ± 2 °C, and a relative humidity of 30–60 %. Animal care and treatments were conducted in accordance with established guidelines and protocols approved by the Institute and McGill University’s Animal Ethics Committee.

Table 1 Primers for the genotyping of IGFBP-3 mice
Full size table

DMBA treatment and follow-up

To study breast carcinogenesis, we used the dimethylbenz[a]anthracene (DMBA)-induced mammary carcinogenesis model. To increase the yield of mice with DMBA-induced mammary tumors and to decrease the latency for tumor development, the protocol as described by Aldaz et al. [30] was used. Briefly, 14 IGFBP-3 wild-type (BP3+/+) and 15 knockout (BP3−/−) female mice at 6 weeks old received a subcutaneous implant in the interscapular area of two compressed pellets of medroxyprogesterone acetate (MPA), 20 mg each (Hormone Pellet press, Kansas City, KS). Then they were given DMBA by gavage (a total of 4 doses of 1 mg of DMBA in 0.1 ml of sesame oil when they were 9, 10, 12, and 13 weeks of age). Three animals of each genotype were in the control group receiving only MPA and the vehicle (sesame oil). Mice were kept for up to 13 months after the last DMBA treatment. Palpable tumors were monitored weekly and measured with a caliper. Mice were also monitored for behavioral changes, weight loss (>20 %), and dehydration. At the end of the experiment or when tumors reach 1 cm3, the animals were anesthetized, blood was collected by cardiac puncture, the animals were euthanized by cervical dislocation, and then tumors and tissue were collected and frozen until use.

Metastasis assay

Five BP3+/+ and 5 BP3−/− female mice (on a Balb/c background) between 8 and 10 weeks of age, were injected in the tail vein with 105 4T1 mouse breast cancer cells, purchased from American Type Culture Collection, Manassas, VA). 16 days post-injection, mice were anesthetized, blood was collected by cardiac puncture, and then the lungs were perfused with PBS and fixed in Bouin’s solution. The number of lung nodules, as well as their histopathology, was established by two independent and blinded pathologists. This experiment was repeated twice, with similar results.

Western blotting

Tissue lysates were prepared in lysis buffer (50 mM Hepes, 150 mM NaCl, 1 % Triton-X 100, 0.02 % sodium azide, 60 mM β-glycerophosphate, 1 mM DTT, protease inhibitor cocktail (Complete tablets, Roche, Laval, QC, Canada) at pH 7.2, and 1.46 µM pepstatin A (BioShop Canada Inc., Burlington, ON, Canada) added just before use). The protein concentration in the supernatant fraction was determined and the extracts were stored at −80 °C. Proteins (20 µg for tumors and 50 µg for serum samples) were resolved electrophoretically on denaturing SDS–polyacrylamide gels (10 or 12 %), transferred to nitrocellulose membranes, and probed with the following antibodies overnight at 4 °C: anti-IGFBP-3 (R&D Systems Inc., Minneapolis, MN), anti-IGFBP-2, and anti-transferrin (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-IGFBP-5 (Novus Biologicals, Littleton, CO), anti-phospho-AKTSer473, anti-AKT, anti-IGF-1 receptor, anti-poly(ADP ribose) polymerase (PARP), anti-phsopho-MAPKT202/Y204, anti-MAPK, anti-ribosomal protein S6 (S6), anti-phospho-S6S235/236, and anti-β-actin (Cell Signaling Technology, Beverly, MA) antibodies. The position of the proteins was visualized with the appropriate horseradish peroxidase-conjugated anti-rabbit, anti-goat, or anti-rat immunoglobulin antibodies (ECL, GE Healthcare LifeScience, Baie d’Urfé, QC, Canada). Four different westerns were done for each protein with tumor samples from 2 different protein lysates for all (n = 12 for BP3+/+ and n = 9 for BP3−/−). Densitometry was calculated using both a scanner and computer program according to manufacturers’ instructions. The density of the bands was normalized to the percent density of β-actin for IGFBP-5 or to total AKT protein for phospho-AKTS473 in the tumors and to the percent density of transferrin for IGFBP-2, IGFBP-3, and IGFBP-5 in the serum.

Ligand blotting

Serum samples (50 µg—equivalent of ~2 µl) were diluted in non-reducing loading buffer, heated for 5 min at 50 °C, run on 12 % polyacrylamide gel, and transferred on nitrocellulose membrane as for Western blotting. The non-radioactive ligand blotting was carried out according to the manufacturer’s protocols (#IGF005, IBT Systems, Reutlingen, Germany) and the detection of the IGFBPs was done with ECL reagents as for the Western blotting in two independent experiments (n = 4). Densitometry was calculated using both a scanner and computer program according to manufacturers’ instructions. The density of the bands for each sample was added to get the density of the total amount of IGFBPs.

Insulin, adiponectin, IGF-1, VEGF, and growth hormone (GH) ELISA

Hormone serum levels were measured in duplicate using rodent insulin (n = 13 for both BP3+/+ and for BP3−/−), VEGF (n = 7 for BP3+/+ and n = 12 for BP3−/−), and GH (n = 7 for BP3+/+ and n = 9 for BP3−/−) and mouse adiponectin (n = 12 for BP3+/+ and n = 13 for BP3−/−) immunoassays (Millipore Corp., Etobicoke, ON, Canada) and rodent IGF-1 ELISA (n = 13 for both BP3+/+ and for BP3−/−) (Diagnostic Systems Laboratories, Webster, TX), according to the manufacturer’s protocols.

Immunohistochemistry staining and histopathology

Immunohistochemistry (IHC) was performed at the Segal Cancer Centre Research Pathology Facility (Jewish General Hospital, Montreal, QC, Canada). Tissue samples were cut at 4-µm, placed on SuperFrost/Plus slides (Fisher Scientific, Ottawa, ON, Canada) and dried overnight at 37 °C, before IHC processing. The slides were then loaded onto the Discovery XT Autostainer (Ventana Medical System, Inc., Tucson, AZ). All solutions used for automated immunohistochemistry were from Ventana Medical System unless otherwise specified. Slides underwent de-paraffinization, heat-induced epitope retrieval (CC1 prediluted solution Ref: 950-124, standard protocol). Immunostaining was performed online using a heat protocol. Briefly, rabbit polyclonal anti-Ki-67 (Novus, Oakville, ON, Canada) diluted at (1:100) and rabbit monoclonal anti-ribosomal protein S6S240/244 (Cell Signaling Technology) (1:1000) in the antibody diluent (Ref: 251-018) were manually applied for 32 min at 37 °C then followed by the appropriate detection kit (OmniMap anti-Rabbit-HRP, Ref: 760-4311) for 8 min, followed by ChromoMap-DAB Ref: 760-159). A negative control was performed by omission of the primary antibody. Slides were counterstained with Hematoxylin for 4 min, blued with Bluing Reagent for 4 min, removed from the autostainer, washed in warm soapy water, dehydrated through graded alcohols, cleared in xylene, and mounted with Permount. Slides were scanned at ×20 using a ScanScope AT Turbo (Leica Biosystems, Concord, ON, Canada) with default settings. They were analyzed for phospho-S6S240/244 and Ki-67 using the Aperio ePathology Cytoplasmic v2 algorithm (Leica Biosystems) with default input parameters. Tumors (n = 11 for BP3+/+ and n = 9 for BP3−/−) and lungs (n = 5 for both BP3+/+ and for BP3−/−) were sectioned and stained with hematoxylin and eosin. Two blinded evaluators from the pathology department determined the tumor type for each sample.

Statistical analyses

The data presented are mean ± SEM. For insulin, adiponectin, VEGF, GH, and IGF-1 levels, a nonparametric procedure (Mann–Whitney–Wilcoxon (MWW)) was used to perform an analysis of variance on ranks where data were not normally distributed. For tumor detection and number of tumor graphs, a one-way analysis of variance was used to determine if a significant difference was among all treatment groups. For survival graph, Log-rank test using the LifeTest procedure was used to analyze Kaplan–Meier curve and statistical significance for body weight graph was evaluated using Proc Mixed Procedure. All statistical analyses were performed using Statistical Analysis System software, version 9.2 (SAS Institute, Cary, NC), with P values < 0.05 considered significant.

Results and discussion

Female mice of each genotype (IGFBP-3 −/− and IGFBP-3 +/+) were followed for up to 13 months after the last DMBA gavage. IGFBP-3 knockout mice showed no obvious phenotype compared to wild-type mice, except that with time, the IGFBP-3 −/− mice weighed less than IGFBP-3 +/+ mice (Fig. 1a: 30.06 ± 4.19 g versus 32.79 ± 5.23 g at sacrifice time, P < 0.0001).

Fig. 1
figure 1

IGFBP-3 knockout mice develop tumors earlier than wild-type IGFBP-3 mice. a Body weight curve of both groups, determined weekly (overall significance of difference between groups: P < 0.0001); b after DMBA treatment, time of first appearance of palpable tumors was monitored twice weekly (asterisk P = 0.0144 for difference between groups); c percentage of mice without tumors as a function of time post carcinogen exposure, according to genotype. Dotted line BP3+/+, solid line BP3−/− mice (asterisk P = 0.0144 for difference between groups; d at the time of euthanasia, the number of mammary tumors was counted for each mouse. The number of tumors per mouse was not different (P = 0.5555). Square BP3+/+; open inverted triangle BP3−/− mice. e Representative H & E and immunohistochemistry of Ki-67 and phospho-S6S240/244 antibodies of breast tumors for both groups (n = 5). +/+ BP3+/+; −/− BP3−/−. Magnitude X200

Full size image

There was a major difference in the time of first appearance of mammary tumors, which was earlier in BP3−/− mice (13.9 ± 1.1 weeks versus 22.5 ± 3.3 weeks, P = 0.0144) (Fig. 1b–c). However, there was no difference in the number of tumors per mouse between the two groups (BP3−/−: 1.33 ± 0.16 versus BP3+/+: 1.54 ± 0.27 tumors, P = 0.5555) (Fig. 1d). Histopathologic analyses of the primary tumors showed that in the BP3−/− mice, 80.0 % were invasive ductal carcinomas and 20.0 % were adenosquamous carcinomas. As for tumors in BP3+/+ mice, 77.8 % were invasive ductal carcinomas, 22.2 % were poorly differentiated adenocarcinomas, and 11.1 % were defined as adenosquamous carcinomas. Representative histology photographs for both genotypes are shown in Fig. 1e. This is consistent with previous publications [30, 31].

The mitotic index was determined using Ki-67. Cell proliferation was higher in tumors from BP3−/− mice compared with BP3+/+ mice (26.92 % ± 5.23 versus 13.26 % ± 3.72, respectively; P = 0.05). However, the percentage of positive cells for phospho-S6Ser240/244 was not statistically different between both groups: 57.11 % ± 7.49 versus 75.80 % ± 5.67 (P = 0.06) (Fig. 1e).

An important function of IGFBP-3 is to act as a carrier protein for IGFs in the circulation by the formation of a stable complex with IGF-1 and the acid labile subunit (ALS) protein, an interaction which prolongs the half-life of the growth factors in the circulation. We measured by ELISA circulating total IGF-1 levels and found that they were not significantly different between the two groups: 700.2 ± 65.9 pg/ml versus 881.8 ± 101.7 pg/ml, respectively, for BP3−/− and BP3+/+ mice (Fig. 2a). The trend toward higher IGF-1 levels in the control mice is likely due to reduced serum half-life of IGF-1 in the absence of IGFBP-3. We cannot exclude the possibility that the proportion of free-IGF-1 uncomplexed with binding proteins in the tissue microenvironment of primary tumor sites could be higher in the knockout animals compared with wild-type. Levels of insulin (0.456 ± 0.073 versus 0.421 ± 0.071 ng/ml), adiponectin (11.02 ± 1.05 versus 14.52 ± 1.59 ng/ml), and VEGF (71.60 ± 5.64 versus 58.76 ± 4.24 pg/ml) were also similar in BP3−/− as in BP3+/+ mice, respectively (Fig. 2b–d). The levels of circulating GH were also measured (Fig. 2e). No statistical significant difference was observed, although a trend toward higher GH levels was observed in BP3−/− compared with BP3+/+ mice (2.41 ± 0.89 versus 1.44 ± 0.44 ng/ml). However, since it is well-known that GH levels vary in a pulsatile manner, measurements at a single time point provide limited information concerning growth hormone profiles over time.

Fig. 2
figure 2

Circulating hormone levels are similar between the two groups. a IGF-1 levels; b insulin levels; c adiponectin levels; d VEGF levels; e growth hormone levels. Blood was collected at the time of euthanasia from non-fasting BP3−/− and BP3+/+ mice. Differences between the two groups were non-statistically significant. Data presented are the mean ± SEM

Full size image

To better understand the mechanisms responsible for the difference in time for tumor appearance, western ligand blots were performed to determine if circulating levels of other IGFBPs increased to compensate for the absence of IGFBP-3. Figure 3a shows that in some IGFBP-3 knockout mice the amount of IGFBP-1, 2, or 5 was increased, but that there was considerable variability. As for IGFBP-4, no increase was observed in the knockout mice. Densitometry was done to determine whether total IGFBPs levels were similar between both groups. Total IGFBPs levels were lower in BP3−/− than in BP3+/+ mice (Fig. 3b). There is evidence for some compensatory increase in other IGFBPs. Since by western ligand analysis, one cannot differentiate between IGFBP-1, 2, and 5, conventional western blots were used to assess the intensity of individual binding proteins in the serum of each genotype (Fig. 3c). No difference was observed between BP3+/+ and BP3−/− mice for IGFBP-2; however, there was a trend for IGFBP-5 band intensity to be higher in BP3−/− than in BP3+/+ mice, suggesting that IGFBP-5 level may increase to compensate to some extent for the loss of IGFBP-3 (Fig. 3d).

Fig. 3
figure 3

IGFBP-3 knockout: IGFBPs do not compensate for the absence of IGFBP-3. a The distribution of the IGFBPs in both genotypes was determined by western ligand blots of serum. b Densitometry of ligand blot to determine the total IGFBPs levels for both BP3+/+ and BP3−/−. asterisk P = 0.0368. Data presented are the mean ± SEM. c Western analysis was performed to evaluate individually the levels of IGFBPs (anti-IGFBP-2, anti-IGFBP-3, anti-IGFBP-5, and anti-transferrin (loading control) antibodies) in serum. d Densitometry of (C) asterisk P = 0.0036. Data presented are the mean ± SEM. Representative western and western ligand blots are presented. Numbers 1–4 correspond each to an individual animal sample

Full size image

In the tumor samples, IGF-1R downstream signaling was analyzed and there was a significant increase in phosphorylation of AKTSer473 in BP3−/− mice, but no obvious difference in levels of total AKT or total IGF-1R proteins (Fig. 4). Interestingly, in the tumors, there seems be an increase in IGFBP-5 in the IGFBP-3 knockout compared with the wild-type mice. No difference was observed in the phosphorylation levels of MAPKT202/Y204 and S6S235/236 proteins, nor in the levels of the cleaved PARP protein, marker of apoptosis, between the 2 groups (data not shown).

Fig. 4
figure 4

Effects on signaling downstream of the IGF-1 receptor in breast tumors. a At the end of the experiment, tumors were collected and snap frozen. Cell lysates were prepared and proteins were processed by Western blot analysis with anti-IGFBP-5, anti-phospho-AktSer473, anti-total Akt, anti-total IGF-1R, and anti-β-actin (as loading control) antibodies. b Densitometry graphs of (a). asterisk P = 0.0038. Representative western blots are presented. Data presented are the mean ± SEM. Numbers 1–4 correspond each to an individual animal sample

Full size image

To determine whether IGFBP-3 has a role in breast cancer cell metastasis to the lungs, we used the murine 4T1 mammary cell line model, as described in the “Materials and methods” section. Figure 5a shows that the number of metastatic nodules on lungs of IGFBP-3 −/− mice was significantly higher as compared to the number of nodules on IGFBP-3 +/+ lungs (88.3 ± 14.0 versus 49.2 ± 12.9 nodules, respectively, P = 0.0159). With respect to nodule size, BP3−/− mice had a higher percentage of nodules larger than 1 mm compared to BP3+/+ mice (80.6 % versus 75.6 %, respectively). Histopathology showed ductal carcinoma for both BP3−/− and BP3+/+ hosts (Fig. 5b). Proliferation was assayed using Ki-67 marker: a non-significant trend toward higher Ki-67 in BP3−/− mice was observed (2.86 % ± 0.31, versus 2.62 % ± 0.40, P = 0.659). However, results obtained for phospho-S6S240/244 labeling showed higher activation for the BP3−/− mice (29.39 % ± 6.89 versus 8.09 % ± 4.25, respectively; P = 0.0281 (Fig. 5b).

Fig. 5
figure 5

Effects of IGFBP-3 on lung metastasis of 4T1 cells. 4T1 cells were injected in the tail vein to both BP3−/− and BP3+/+ mice. a Nodules on lungs were counted 16 days post-injection, distinguishing between nodules smaller or larger than 1 mm, asterisk P = 0.0159. Data presented are the mean ± SEM. b Representative H & E and immunohistochemistry of Ki-67 and phospho-S6S240/244 antibodies of lung metastasis for both groups (n = 5). +/+ BP3+/+; −/− BP3−/−. Magnitude ×200; insert ×400

Full size image

The difference in signaling downstream of the IGF-1 receptor in tumors of BP3 −/− and BP3+/+ mice at the time of sacrifice was modest in magnitude relative to the difference in number of metastasis. This raises the possibility that IGFBP-3 knockout may influence early steps in the metastatic process. The results from the metastasis assay suggest that IGFBP-3 has a role as metastasis suppressor gene, and that any increase in IGFBP-5 was not sufficient to abrogate the effect of the absence of IGFBP-3 on the metastasis endpoint. Our results are consistent with previous published data for metastasis models of other tissues: in a prostate cancer model, using the transgenic PB-Hi-Myc mice crossed with an IGFBP-3 knockout mouse model [28] showed that the absence of IGFBP-3 increased metastasis to distant organs. In other in vitro models, IGFBP-3 inhibited migration of endometrial cancer cells [32] and was involved in inhibiting anchorage-independent growth [33], which could favor a metastatic process.

Although IGFBP-3 knockout does not lead to major phenotypic differences compared to wild-type mice under control conditions, our results, for the first time, reveal that the IGFBP-3 knockout mice differ from controls in terms of mammary tumor growth following exposure to DMBA and in a model of breast tumor cell metastasis. The small decrease in body weight we observed in the absence of IGFBP-3 implies that at the whole-organism level, IGFBP-3 serves to increase IGF activity, perhaps by prolonging serum half-life. This result is consistent with prior reports [27, 34]. Previous studies have shown that a decrease in GH/IGF-1 levels reduced tumor development in several models [34–37]. On the other hand, in the context of DMBA-induced carcinogenesis and of metastasis, our data demonstrate that knockout of IGFBP-3 increases tumor aggressiveness, without a major change in GH or IGF-1 serum levels. Our observations thus support the hypothesis that IGFBP-3 influences neoplasia by reducing IGF-1 bioactivity and/or by direct IGF-independent actions [1–11].

Clinically, it is well-known that there is considerable person-to person heterogeneity in IGFBP-3 serum levels, and some of this variability is attributable to polymorphisms in the IGFBP-3 promoter [38, 39]. This variation has not been consistently associated with breast cancer risk or outcome. A likely explanation for this is that the variation within the normal range would be expected to have more subtle effects on neoplasia than the extreme variation used in the IGFBP-3 knockout model.