Murine osteosarcoma primary tumour growth and metastatic progression is maintained after marked suppression of serum insulin-like growth factor I

Abstract

The insulin-like growth factor I (IGF-I) signaling pathway has been shown to play an important role in several aspects of cancer biology, including metastasis. The aim of this study was to define the contribution of serum (endocrine) and local (tumour microenvironment) IGF-I on osteosarcoma tumour growth and metastasis, a cancer that is known to be dependent on the IGF-I axis. To test this hypothesis, we evaluated the primary tumour growth and metastatic progression of K7M2 murine osteosarcoma cells injected to a genetically engineered mouse [liver-specific IGF-I deficient (LID)] in which serum IGF-I levels are reduced by 75%, while maintaining expression of IGF-I in normal tissues. We first demonstrated that IGF-I in the tumour and the tumour-microenvironment were maintained in the LID mice. Within this designed model, there was no difference in primary tumour growth or in pulmonary metastasis in LID mice compared to control mice. Furthermore, there was no difference in the number or localization of single metastatic cells immediately after their arrival in the lungs of LID mice and control mice, as analysed by single cell video microscopy. Collectively, these data suggest that marked reduction in serum IGF-I is not sufficient to slow the progression of either primary or metastatic models of osteosarcoma.

Osteosarcoma is the most common primary malignant bone tumour.^1,2 Despite effective control of the primary tumour, metastasis to the lung develops in over 30% of patients within 5 years.^2,3 Patients who present with metastatic disease or develop recurrence in the lung have very poor prognosis.^4–6 Various attempts to improve survival in these 2 populations using intensification of chemotherapy have not been successful.^4,7–12 This demonstrates the need for new treatment approaches.

Osteosarcoma occurs most often in the second decade of life. During this time, growth hormone (GH) and insulin-like growth factor I (IGF-I) levels are highest.^13,14 These observations led to the hypothesis that alterations in the IGF signaling pathway may play a role in the development or progression of osteosarcoma. In vitro and in vivo studies have suggested the importance of the IGF-I/GH pathway in osteosarcoma cells and tumours.^15,16 Osteosarcoma cells express both IGF-I and IGF-I receptor, proliferate in response to IGF-I, and demonstrate an anti-apoptotic phenotype in vitro after exposure to IGF-I.¹⁶ In mice that have undergone hypophysectomy, serum IGF-I levels are dramatically decreased and osteosarcoma tumour cell growth and metastases are significantly reduced compared to intact controls.¹⁷

Based on this, a Phase I trial was conducted using a long-acting somatostatin analog (OncoLAR) to suppress serum IGF-I in pediatric patients with osteosarcoma.¹⁸ In this trial, OncoLAR treatment of 21 osteosarcoma patients with advanced disease resulted in a 46% decrease in serum IGF-I levels with no dose limiting toxicities. However, no clinical responses were seen in this population of osteosarcoma patients.¹⁸ As part of an integrated development plan, OncoLAR combined with chemotherapy was assessed in a trial in pet dogs with naturally occurring osteosarcoma to analyse if increase tumour cell apoptosis, and therefore improved outcome could be seen in dogs receiving chemotherapy and IGF-I suppression compared to dogs treated with chemotherapy alone.¹⁹ In these pet dogs, serum IGF-I was reduced by approximately 43% after OncoLAR treatment; however, there were no differences in primary tumour necrosis, apoptosis, or survival in dogs treated with OncoLAR and chemotherapy compared to dogs treated with chemotherapy alone.¹⁹ This study suggested that the extent of serum IGF-I suppression provided by OncoLAR was not sufficient to provide a clinical benefit or that tumour or microenvironment derived IGF-I was sufficient to support osteosarcoma tumour growth and progression.

The aim of the current study was to analyse whether further reductions in serum IGF-I levels could influence osteosarcoma tumour cell growth or pulmonary metastasis or whether regional (microenvironment) sources of IGF-I were maintained and sufficient to support osteosarcoma tumour growth and progression. Towards this goal, we developed an experimental approach based on the liver-specific IGF-I-deficient (LID) mouse model.²⁰ The LID mouse has a 75% reduction in circulating IGF-I levels, with no differences in the levels of IGF-I mRNA expression in extra hepatic tissues.²⁰ To study the role of serum IGF-I in osteosarcoma tumour growth, LID mice (B6 background) were backcrossed to balb/c mice to allow for growth of K7M2 osteosarcoma cells. K7M2 murine osteosarcoma cells were then injected orthotopically into the backcrossed LID mice or their littermate controls. The K7M2 murine osteosarcoma orthotopic model has been previously described and shares many important features with human osteosarcoma including expression of IGF-I ligand and receptors.²¹

The newly derived balb/c LID mice were confirmed to have 75% reduction in serum IGF-I compared to litter-mate controls. As expected, K7M2 primary tumours and metastases expressed IGF-I ligand in both LID and littermate controls at equivalent levels. No differences in tumour growth or metastatic progression were noted between LID mice and their littermate controls. These data suggest that either (or both) sustained expression of IGF-I in the tumour microenvironment or limited serum IGF-I (less than 25%) is sufficient to support osteosarcoma primary tumour growth and metastasis in preclinical models.

Material and methods

Cell lines and media

Clonally related K7M2 and K12,²¹ and LM8²² (kindly provided by Dr. E.S. Kleinerman) murine osteosarcoma and MC38 murine colon carcinoma cell line²³ characterization and maintenance has been previously described.^21–23 All cells were maintained in vitro using Dulbecco’s Modified Eagle’s Medium (Invitrogen, Carlsbad, CA) culture media containing 10% fetal bovine serum, L-glutamine (2 mmol), penicillin (100 units/mL), and streptomycin (100 μg/mL, Invitrogen) at 37°C in a humidified 5% CO₂ incubator. For all in vitro and in vivo assays, cells were harvested using trypsin/EDTA (Invitrogen) from cultures at 70% confluence. Cell viability was assessed by trypan blue exclusion and all cell lines used were from early passages. Serum–free (SFM) medium was also used in selected in vitro experiments. All cells were verified mycoplasma free.

Animal husbandry and PCR genotyping

The creation of LID mice has been previously described.²⁰ The LID mice were backcrossed to balb/c mice and bred for 7 generations to prevent immune rejection since K7M2 osteosarcoma cells have a balb/c background. Control and LID mice express the fourth exon of the IGF-I gene flanked by 2 LoxP sites. LID mice also express the Cre-recombinase transgene, exclusively in the liver, under the control of the albumin enhancer promoter sequence. Animals were genotyped using PCR on tail DNA. As described elsewhere, LID mice have a 75% decrease in serum IGF-I levels compared to controls.²⁴ Animal care and use were in accordance with the guidelines of the NIH Animal Care and Use Committee.

RNA extraction and Rnase protection assay

RNA was extracted from cell lines using the Qiagen RNeasy Midi kit (Qiagen, Valencia, CA) according to the manufacturer’s specifications. RNA was extracted from grossly dissected primary tumours, sham surgical sites, lung metastatic nodules and normal muscle using Trizol® reagent according to the manufacturer’s instructions. Extracted RNA was quantitated by spectrophotometry and its integrity was confirmed by 1% agarose gel electrophoresis or Bioanalyzer (Agilent Technologies, Santa Clara, CA) electrophoresis. The following probes were employed: PE3-IGF-IR exon 3 and PMI4-IGF-I exon 4 as previously reported.²⁰ PE3 was amplified by Qiafilter Maxi kit (Qiagen), linearised with EcoRI and then purified with QIAquick Gel Extraction Kit (Qiagen) according to the manufacturer’s specifications. The linearised probes were biotinylated using the North2South Biotin In Vitro Transcription Kit (Thermo Scientific, Waltham, MA). RNA extraction and Rnase protection assay (RPA) was performed using the Supersignal RPAIII Chemiluminescence Detection Kit (Thermo Scientific) according to the manufacturer’s instructions.

Quantitative RT-PCR analysis

RNA was isolated using Trizol® LS reagent according to manufacturer’s instructions (Invitrogen). RNA was subjected to an additional purification step using the RNeasy spin column clean up protocol supplied by the manufacturer (Qiagen). Samples were quantitated using a Nanodrop apparatus and 1 μg of total RNA was reverse transcribed. A Universal Mouse Reference RNA (catalog #740100, Stratagene, La Jolla, CA) was used as a positive control and negative controls were included as reactions containing no cDNA template. Reverse transcription was primed using random hexamers and cDNA synthesis conducted using Moloney murine leukemia virus reverse transcriptase (Promega Corporation, Madison, WI) at 37°C for 1 hr. The reaction was halted by heating at 65°C for 10 min followed by dilution of the cDNA reaction 1:3 using nuclease-free water. Two microlitre of this diluted cDNA was used as a template for quantitative PCR using the iQ SYBR Green Supermix and assayed using an iQ5 real time thermocycler (Bio-Rad Laboratories, Hercules, CA). The following primers were used at a final concentration of 100nM for gene specific amplification: IGF-I sense primer 5′- gaccgaggggcttttacttca-3′, IGF-I antisense primer 5′-ggacggggacttctgagtctt-3′; IGF-II sense primer 5′-tttctgtttctctccgtgctgtcc-3′, IGF-II antisense primer 5′-atcgggaaatgaggtcagctgttg-3′; GAPDH sense primer 5′- ccccaatgtgtccgtcgtg-3′, GAPDH antisense primer 5′-gcctgcttcaccaccttct-3′. Thermocycling parameters were as follows: step 1 = 95°C for 1 min. 30 sec., step 2 = 95°C for 15 sec., step 3 = 63°C for 15 sec., step 4 = 72°C for 20 sec. with steps 2–4 repeated 39 additional times. Fluorescence was measured after step 4. Melt curve analysis was conducted after each run for each primer pair and condition. Quantitation was done using the formula 2^−ddCT with GAPDH as the reference gene for normalization between conditions and MC38 cell line set as the control condition so all other samples fold difference in IGF-I and IGF-II are relative to MC38.²⁵

Cell proliferation assay

Cell proliferation was assessed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) staining. K7M2 cells were plated at 10⁴ cells/well in complete media in 96-well plates. After 24 hr, the media was changed to experimental conditions: 2.5 ng/ml, 25 ng/ml or 250 ng/ml recombinant human IGF-I (R&D Systems, Minneapolis, MN) in SFM. The media was replaced every 48 hr throughout the experiment. Plates were harvested at 48, 96 and 192 hr. At each time point, the media was removed, 100 μl of MTT (0.5 mg/ml) in phenol-red free RPMI was added and the cells were incubated for a further 3–4 hr at 37°C. One hundred microlitre of isopropanol was added to each well and cell numbers were obtained after optical scanning using a plate reader (Molecular Devices, Sunnyvale, CA) with 570 and 690 nm filters, using standard curves. Experimental conditions were performed in replicates of six.

IGF-I stimulation and inhibition in K7M2 osteosarcoma cells

K7M2 cells (1.5 × 10⁶) were seeded on 100 mm plates in complete media. Twenty-four hours before IGF-I stimulation (10 ng/ml), the media was replaced with SFM after 2 washes with PBS. Inhibition studies used 0.1 or 1.0 μg/ml of α-IR3 antibody, which inhibits the IGF-I receptor.

Western blot analysis

Cells were lysed in ice-cold NP40 buffer (1% NP40, 20 mM Tris, 250 mM NaCl) with 10 mM NaF, 1 mM NaV and Proteinase inhibitor cocktail (Roche, Indianapolis, IN). The protein concentration was measured using the Bio-Rad protein assay kit (Bio-Rad Laboratories). Protein lysates (40 μg/lane) were boiled after the addition of 2X Tris-Glycine SDS Sample Buffer (Invitrogen) for 10 min, chilled on ice, and subjected to 12% SDS-PAGE electrophoresis (Invitrogen), followed by transfer to a nitrocellulose membrane (Invitrogen). Membranes were blocked with 5% nonfat dried milk in TBS-Tween-20. The membrane was then incubated with primary antibody: 1:1,000, anti-phospho-Akt, or anti-Akt (Cell Signaling, Beverly, MA) overnight at 4°C, or 1:10,000 anit-β-actin (AbCam, Cambridge, MA) for 1 hr at 37°C. After incubation, the membranes were washed and antibody binding was visualised by exposure to a 1:2,000 dilution of an anti-rabbit IgG HRP-linked antibody (Cell Signaling) or anti-mouse IgG HRP-linked antibody (Amersham Pharmacia Biotech, Piscataway, NJ) for 1 hr at 37°C and developed by SuperSignal West Pico detection (Thermo Scientific) and subsequent exposure to film.

Conditioned media and measurement of IGF-I

Two million cells were added to six-well plates in complete media. The media was replaced after 24 hr of incubation in SFM after a PBS wash. The media was collected at 24, 48 and 72 hr after changing to SFM. Collected media was concentrated using an Amicon Centricon plus-20 centrifugal filter device with a 5,000 molecular weight limit (Millipore, Bedford, MA). The experiment was performed in triplicate. Concentrations of IGF-I in conditioned media were analysed by radioimmunoassay (National Hormone and Pituitary Program, NIDDK). The sensitivity of this assay is 0.02 ng/ml.

In vivo primary tumour and spontaneous metastasis assay

For primary tumour growth assays, 2 million K7M2 tumour cells were injected intramuscularly in the left hind limb of LID and control mice as previously described.²¹ Time to detection of tumour, percent tumour take, and tumour growth were monitored. When tumours reached a diameter of 1.5 cm or greater, the tumour-bearing limb was surgically resected. Mice were then monitored for evidence of morbidity (including anorexia, dehydration, dyspnea and decreased activity and grooming behavior) associated with pulmonary metastasis. Killing of mice with presumed pulmonary metastases was based on the development of these symptoms. All mice underwent necropsy to confirm metastatic lung disease.

Single-cell metastasis imaging

K7M2 murine and LM8 human osteosarcoma cells were fluorescently labeled using 7.5 μM carboxymethylfluorescein diacetate (CMFDA) (Invitrogen), according to the manufacturer’s recommendations. At 1, 6, and 24 hr after injection of 1.0 × 10⁶ cells into the lateral tail vein, mice (n = 5 at each time point) were killed. Lungs were inflated using slow intratracheal injection of PBS, removed and then imaged ex vivo by inverted fluorescent videomicroscopy (Leica DM IRB) at 100× magnification. Ten random frames were captured and analyzed using OpenLab software (Improvision, Waltham, MA) to define and count fluorescent (green) CMFDA signals larger then 10 pixels. The mean number of metastatic cells on the surface of the lung was compared at 1, 6 and 24 hr.

Statistical analysis

Differences in tumour growth were analyzed by Welch’s corrected t-test, and differences in single cell metastasis imaging were analyzed by unpaired t-test calculated using GraphPad Prism version 4.0c. Log-rank test statistics were used for survival curves. Statistical significance was defined as a p-value of less than 0.05.

Results

K7M2 cells express IGF-I receptor and proliferate in response to IGF-I stimulation

K7M2, K12 and LM8 murine osteosarcoma cells expressed IGF-I mRNA by quantitative RT-PCR (Fig. 1a). However, IGF-II mRNA levels were very low in osteosarcoma cells (Fig. 1b). Western blot analysis for the IGF-I receptor in K7M2, K12 and LM8 murine osteosarcoma revealed both pro IGF-IR and IGF-I receptor β (IGF-IRβ). Interestingly, the more aggressive K7M2 cells expressed higher proIGF-I receptor and phosphorylated Akt(Ser473) levels compared to the clonally related less aggressive K12 cells (Fig. 1c). A similar pattern of IGF-IR mRNA was found by Northern blot (data not shown). We then examined whether K7M2 cells responded to IGF-I stimulation in vitro. Cell proliferation was assessed by MTT analysis. As shown in Figure 1d, the addition of IGF-I resulted in increased proliferation of K7M2 cells in a dose dependent manner. Furthermore, assessment of conditioned media taken from cultured K7M2 cells suggested autocrine production of IGF-I (data not shown).

Figure 1 –

K7M2 cells express IGF-I receptor and proliferate in response to IGF-I stimulation. (a) Quantitative RT-PCR analysis of IGF-I steady-state mRNA levels in tumour cell lines. One microgram of total RNA was reverse transcribed and analyzed by real-time PCR for expression of GAPDH (invariantly expressed control gene) and IGF-I. IGF-I mRNA expression was normalized to GAPDH between samples and expressed as normalized fold expression using MC38 murine colon carcinoma as the control condition (set at a value of 1). The commercial murine reference RNA is a mixture of RNA including reproductive, embryonal, lymphoid, endocrine and connective tissues. (b) IGF-II mRNA expression was normalized to GAPDH as in (a), but the murine reference RNA was set as the control condition because of the fact that IGF-II levels are undetectable in MC38 cells. (NS *) no signal above background. (c) Both K12 and K7M2 osteosarcoma cells express IGF-I receptor as analysed by Western blot analysis. Expression of the proIGF-I receptor levels and phosphorylated Akt were greater in the highly metastatic K7M2 cells when compared with the less metastatic K12 cells. (d) K7M2 osteosarcoma cells proliferate after addition of escalating concentrations of IGF-I (0, 2.5, 25 or 250 ng/ml) as analysed by MTT analysis at various time points.

IGF-I increases Akt phosphorylation in K7M2 cells, in part by autocrine production of IGF-I

Upon treatment with IGF-I, K7M2 cells exhibited an increase in Akt phosphorylation (Ser473), but no changes in total Akt protein in serum free conditions (Fig. 2a). Interestingly, increased basal phosphorylation of Akt (Ser473) was seen in serum free conditions (Fig. 2b). This phosphorylation was partially inhibited using IGF-I receptor antibody blockade (αIR3), suggesting that autocrine release of IGF-I was in part responsible for this basal phosphorylation of Akt (Ser473). These data support the relevance of the K7M2 murine osteosarcoma model to investigate the role of host derived IGF-I on tumour growth and progression, since human osteosarcoma cell lines and tissues also express IGF-I and its receptor.²⁶

Figure 2 –

The IGF-I-Akt pathway is intact and responsive in K7M2 cells. (a) Western blot analysis shows increased levels of phospho-Akt after exposure to IGF-I at various time points. Total Akt levels remain unchanged. (b) The basal phosphorylation of Akt, noted in serum free conditions, was partially inhibited through IGF-I receptor blockade using 1.0 or 0.1 μg/ml of αIR3 (IGF-I antibody). MOPC21 = an isotype-matched control antibody.

Genotypes and phenotype analysis of LID balb/c backcrossed mice

To confirm the genotype of LID balb/c backcrossed mice PCR analysis was performed using primers to detect either LID mice (200 bp) or non-LID mice (400 bp). Over 200 offspring mice were genotyped and all of these mice expressed the appropriate construct expected of a loxP-flanked igf1 allele (LL)/Cre- mouse and 44/77 mice expressed the constructs expected of a LL/Cre+ (LID) mouse (Fig. 3a). Serum IGF-I levels in balb/c backcrossed mice were 75% lower than that of their control littermates as analysed by radioimmunoassay (Fig. 3b). These data confirm the stability of the LID mouse phenotype, characterized by 75% decrease in serum IGF-I, after backcross of mice to a balb/c background.

Figure 3 –

Genotype and phenotype analysis of LID balb/c backcrossed mice. LID (B6) mice were backcrossed to either Cre (+) or Cre (−) balb/c mice for 7 generations resulting in a total of approximately 200 offspring, 5 of which (samples 4–8) are depicted. (a) PCR reaction using specific primers for either wild type (WT) or “floxed” IGF-I gene (loxP-flanked igf1 allele; LL) locus in balb/c backcrossed mice. Primers detect either LID mice (LL; 200 bp) or non-LID mice (WT; 400 bp). M = 200 bp ladder, WL = a control mouse from a very early backcross with both WT and LL IGF-I, B = control balb/c mouse with WT IGF-I. All seventh generation backcross mice expressed the LL IGF-I alleles. (b) PCR reaction using specific primers for Cre recombinase (Cre) in balb/c backcrossed mice. M = 200 bp ladder, WL = a control mouse from a very early backcross expressing Cre, B = control balb/c mouse with no Cre expression, Cre = Cre recombinase. Mice 4, 5, 6 and 8 expressed Cre (199 bp) and were part of the experimental group. Mouse 7 did not express Cre and was used as a littermate control. (c) Serum IGF-I levels in balb/c backcrossed mice were 75% lower than that of their control littermates as analysed by radioimmunoassay.

Serum IGF-I suppression does not influence expression levels of IGF-I or the IGF-IR in the tumour

Initial studies using K7M2 tumour cells delivered to LID mice (n = 5) resulted in both primary tumour growth and metastasis. Based on the fact that LID mice retain tissue IGF-I levels and that the K7M2 model express both the IGF-I ligand and receptor, we asked if IGF-I ligand expression was maintained in the primary and metastatic tumour microenvironment. Using Rnase protection assays we found no differences in the mRNA levels of IGF-I or IGF-IR in primary tumour sites. These included tumours evaluated at day 2 or day 21 of tumour growth from LID mice and control littermates (Fig. 4a). Muscle taken from the contralateral gastrocnemius muscle were also examined as controls and showed no differences in IGF-I levels at day 2 or day 21 (Fig. 4b). To analyse if surgical manipulation was contributing to the local levels of IGF-I, mice that underwent operation, but not implantation of tumour fragments were also examined, and showed no differences between LID and control mice (Fig. 4c). Similar studies of normal lung and metastatic lung lesions also did not reveal differences in IGF-I levels (data not shown). The inability of systemic IGF-I suppression to modulate IGF-I or IGF-IR expression at the primary tumour or secondary metastatic sites allowed us to then ask if local production of IGF-I was able to influence growth of the K7M2 tumours and metastasis in mice with suppressed serum IGF-I.

Figure 4 –

Tumour and tumour microenvironment IGF-I is not influenced by changes in serum IGF-I in LID mice. The levels of IGF-I mRNA in primary tumour tissue (a), normal muscle (b), and surgical site without implantation of tumour fragments (c). Mice used included LID mice, control littermates or balb/c mice. The bars represent IGF-I mRNA levels normalized to β-actin expression.

There were no significant differences in osteosarcoma primary tumour growth or pulmonary metastasis between LID mice and their control littermates

LID mice and their control littermates were orthotopically injected with K7M2 cells. There were no differences in primary tumour growth between LID mice and their control littermates (Fig. 5a). Primary tumour formation in LID mice and control littermates were similar (72 versus 81%) and shared an equal time of latency (15 days for both groups). All mice that developed primary tumours underwent amputation of the affected extremity, and were monitored for signs of morbidity associated with pulmonary metastases. All of the mice, both LID and controls, succumbed to metastasis to the lungs. There was no difference in overall survival between LID mice and their control littermates as shown in the Kaplan-Meier survival curve (p = 0.65) (Fig. 5b). These data suggest that suppression of serum IGF-I do not influence primary tumour growth or metastasis in murine K7M2 osteosarcoma.

Figure 5 –

Suppression of serum IGF-I does not influence primary tumour growth or metastasis in murine K7M2 osteosarcoma. (a) LID mice and their control littermates were orthotopically implanted with K7M2 tissue fragments. There were no differences in primary tumour growth between LID mice (n = 32) and their control littermates (n = 22) as analysed in 2 separate experiments. (b) As seen in the Kaplan-Meier survival curve, there was no difference in overall survival after amputation of the primary tumour between LID mice and control littermates. (*) Two mice were killed due to recurrence of the tumour at the amputation site, 1 control mouse on day 56 and 1 LID mouse on day 87.

Suppression of serum IGF-I did not influence the number of metastatic cells that arrive and are retained in the lung

Since serum IGF-I suppression in our model was not sufficient to suppress primary tumour growth or metastatic progression we asked whether a more subtle effect could be observed in circulating metastatic cells early after their arrival in the lung. Using a single cell fluorescent imaging technique, we found that the numbers, morphology and localization of cells that arrested in the lung after the intravenous injection of cells were similar in control and LID mice. Again, there were no significant differences in the number of K7M2 cells present on the surface of the lung in either group of mice at 1 hr (control 64.6 ± 2.4 versus LID 51.2 ± 10.4, p = 0.07), 6 hr (control 19.4 ± 3.6 versus LID 19.8 + 5.3, p = 0.89) or at 24 hours (control 5.9 + 5.0 versus LID 4.5 ± 2.0, p = 0.59) (Fig. 6a). In addition, there were no significant differences in the number of LM8 cells present on the surface of the lung in either group of mice at 1 hr (control 37.1 ± 2.2 versus LID 34.6 + 4.0, p = 0.60), and at 6 hr (control 8.0 ± 0.3 versus LID 9.7 + 2.6, p = 0.55) (Fig. 6b). These data suggest that suppression of serum IGF-I do not influence retention of metastatic osteosarcoma cells early after their arrival in the lung.

Figure 6 –

Suppression of serum IGF-I does not influence retention of metastatic cells early after their arrival in the lung. There were no significant differences in the number of tumour cells on the lung surface at 1, 6 or 24 hr after tail-vein injection of K7M2 cells (a) and LM8 cells (b) in LID mice or their control littermates (n = 15) by in vivo/ex vivo experimental metastasis single-cell fluorescent imaging.

Discussion

The IGF-I signaling pathway has been shown to play an important role in several cancers^23,27–29 and plays a significant role in the biology of human osteosarcoma cells.¹⁶ Before the development of IGF-I and IGF-I receptor specific therapy, one option to modulate this important pathway in cancer was to suppress circulating levels of the ligand. Preclinical studies suggested the potential role of serum IGF-I suppression in several cancers, including, breast and colon cancer.^23,30 However, clinical studies of human patients and pet dogs with osteosarcoma using the somatostatin analog, OncoLAR, did not lead to clinical responses.^18,19 These trials demonstrated that IGF-I was suppressed by approximately 45% in these patients and led us to ask the question whether IGF-I expression may be maintained in the primary tumour and metastatic microenvironments, despite dramatic reduction in serum IGF-I and if present whether this maintained microenvironment expression of IGF-I could support primary tumour growth and progression.

To test this hypothesis, we used LID transgenic mice that had suppression of circulating IGF-I. LID mice have tissue-specific knockout of IGF-I in the postnatal liver through an albumin-Cre IGF-I loxP transgene. Mice expressing this transgene (LID) have a 75% reduction in serum IGF-I (liver-derived) but no significant alterations in body weight, survival, or reproductive ability.²⁰ Furthermore, despite suppression of serum IGF-I in the LID mice, normal tissues maintain wild-type levels of IGF-I. In the successfully backcrossed mice, our data demonstrated the expected 75% suppression of serum IGF-I with the maintenance of both normal tissue and tumour/tumour-microenvironment IGF-I in LID mice. Using this mouse model, we found that LID mice and littermate controls had similar primary tumour growth rates despite the 75% reduction in host serum IGF-I in LID mice. After amputation of the tumour bearing extremity, there was no difference in the development of metastatic lung nodules in LID mice and control mice. Furthermore, serum IGF-I suppression did not have an effect on single metastatic cells early after their arrival in the lung.

Our findings in osteosarcoma are similar to those recently published in a murine prostate cancer model, where tumour growth and progression was similar between LID mice and controls.³¹ The importance of cancer-specific assessment of this biology is underlined by distinct results in murine models of colon and mammary cancer. In these studies, there were decreases in murine colon cancer growth and metastasis,²³ and delayed incidence of chemically and genetically induced mammary tumours in LID mice.³⁰ It is currently unclear what differences in these cancers account for the disparate results. We hypothesize that the maintenance of high levels of IGF-I in the tumour microenvironment in LID mice, despite suppression of serum IGF-I is sufficient to support tumour growth and metastasis in osteosarcoma where autocrine production of IGF-I is established. It is also possible that in osteosarcoma, unlike other cancer types assessed in LID mice, is able to effectively use the 25% of IGF-I levels that persist in LID mice. It is possible that further suppression of serum IGF-I, beyond 75%, may be required to modulate primary tumour growth or metastatic progression in osteosarcoma. It also may be argued that the effect of serum IGF-I suppression may have been more evident in a less aggressive model of osteosarcoma. For example, studies in hypophysectomised mice demonstrate significant modulation of osteosarcoma growth where serum IGF-I levels are decreased by 85%.¹⁷ However, removal of the pituitary gland results in suppression of many growth factors in addition to IGF-I. As such, a causal relationship with serum IGF-I was not possible. In work by Scotlandi et al, the dependence of osteosarcoma on the IGF-I axis has been suggested to be lower than other pediatric sarcomas (i.e., Ewings sarcoma).³² Genetic approaches to more completely suppress serum IGF-I in mice are complicated by the fact that total IGF-I knockout is lethal in the neonatal period.^33,34 In a murine prostate cancer model reported by Majeed et al, a germ line mutation that reduces circulating levels of GH and IGF-I, delayed prostate cancer progression and prolonged survival.³⁵ This study suggests that GH in addition to IGF-I may collectively contribute to cancer development, in this case in the Transgenic Adenocarcinoma of Mouse Prostate (TRAMP) prostate cancer model. It should be noted that murine studies of this pathway are limited by the fact that IGF-II is expressed in the prenatal but not the postnatal mouse. Conversely, in humans IGF-II levels are maintained and in some cases are several fold higher than IGF-I levels. Therefore, the potential role of circulating IGF-II is minimized in these murine studies. More broadly, the endocrine effects of IGFs in mice cannot always be directly translated to the human.

In summary, despite a substantial suppression of serum IGF-I (75% decrease), tumour microenvironment IGF-I levels were maintained in both osteosarcoma primary tumours and metastatic sites. In this context of suppressed serum IGF-I and sustained IGF-I in the microenvironment, we found the aggressive biology of osteosarcoma primary tumour growth and progression to be maintained. These data suggest against the utility of endocrine approaches to modulate IGF-I in osteosarcoma. The importance of the IGF-I pathway in osteosarcoma and the persistence of high levels of IGF-I and IGF-IR expression in tumours despite serum suppression do however support the value and potential utility of newly targeted therapeutics, including monoclonal antibodies directed at IGF-I and IGF-I receptor,³⁶ as potential treatments for osteosarcoma.

Abbreviations:

IGF-I

insulin-like growth factor I

IGF-IR

insulin-like growth factor-I receptor

LID

liver-specific IGF-I-deficient

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide

OncoLAR

a long-acting somatostatin analog

RPA

RNA extraction and Rnase protection assay

SFM

serum-free media