Reduced IL-6 levels and tumor-associated phospho-STAT3 are associated with reduced tumor development in a mouse model of lung cancer chemoprevention with myo-inositol

Abstract

Several promising chemopreventive agents have for lung cancer emerged in preclinical models and in retrospective trials. These agents have been shown to modulate pathways altered in carcinogenesis and reduce markers of carcinogenesis in animal and cell culture models. Cancer-prone transgenic mice with oncogenic Kras expressed in the airway epithelium (Ccsp^Cre/+; Kras^LSL−G12D/+) were raised on diets compounded with myo-inositol. These animals form lung premalignant lesions in a stereotypical fashion over the ten weeks following weaning. Mice raised on myo-inositol containing diets showed potent reduction in the number, size, and stage of lesions as compared to those raised on control diets. myo-inositol has previously been reported to inhibit phosphoinositide 3-kinase (PI3K) signaling. However, in mice raised on myo-inositol, total PI3K signaling was largely unaffected. Proteomic and cytokine analyses revealed large reduction in IL-6 related pathways, including STAT3 phosphorylation. This effect was not due to direct inhibition of IL-6 production and autocrine signaling within the tumor cell, but rather through alteration in macrophage recruitment and in phenotype switching, with an increase in antitumoral M1 macrophages.

Introduction

Despite intense research and advances in screening and treatment, lung cancer remains the leading cause of cancer death.¹ Chemoprevention of lung cancer, especially in the pre-identified high-risk pool of patients with significant histories of smoking, could reduce mortality through reduction of new lung cancer formation and delay of premalignant transformation into aggressive cancer.

There is an emerging body of literature exploring potential chemopreventive agents in lung cancer.²² Some of the most promising agents have emerged from retrospective analyses of large clinical databases, with an attempt to identify drugs and dietary components that may modulate lung cancer risk. Many other potential agents have been studied for their roles in altering signaling pathways central to carcinogenesis. There is a wide variety of evidence for some of these agents, ranging from cell culture and animal models to small prospective clinical trials. However, despite a significant literature, no agent has been convincingly demonstrated to reduce lung cancer risk in humans.

A chemopreventive agent would likely be most effective during the initial development of premalignant lesions and transformation of these lesions into malignancy. This study utilizes a mouse model of premalignancy, Ccsp^Cre/+; Kras^LSL−G12D/+ (CC-LR), in which Kras is oncogenically activated specifically in the airway epithelium.³ While Kras is activated starting in utero, airway premalignant lesions start to become evident in week 6, progressing through bronchial dysplasia (BD) to atypical adenomatous hyperplasia (AAH) to adenomas. By week 14, cellular atypia within adenomas indicates the emergence of early adenocarcinomas. These mice die by week 30 from excessive tumor burden and do not develop metastases. The rapid and predictable evolution of premalignant lesions provides a useful platform for evaluation of chemopreventive agents. Additionally, this model recapitulates the activation of Kras in the airway epithelium, an early genetic change that occurs in approximately 15–20% of human lung adenocarcinomas.^{4, 5} The CC-LR model permits directly addressing the impact of agents on this common mutation, and thus provides additional benefit over the commonly-used benzo[a]pyrene treated A/J strain, where Kras activation often occurs spontaneously and in the context of a wide variety of genomic changes induced by benzo[a]pyrene.

myo-inositol is a sugar alcohol and a glucose isomer found in many food including grains and fruits. It is a precursor to numerous secondary messengers and was once considered a B-complex vitamin. It has been shown to inhibit lung tumorigenesis in a carcinogen-exposed A/J mouse model.^6–9 myo-inositol has been reported to inhibit phosphoinositide 3-kinase (PI3K) signaling.¹⁰ On that basis, myo-inositol was investigated in a phase IIb study to assess its chemopreventive effects in patients with bronchoscopically discovered dysplastic airway lesions.^{11, 12} Subjects receiving myo-inositol did not demonstrate a statistically significant decrease in airway lesions but did have reduced PI3K signatures.

In the CC-LR lung cancer model, myo-inositol showed a consistent effect on both tumor number and size. However, PI3K signaling seemed to be largely unaffected. Instead, proteomic analysis revealed alterations of inflammatory networks, with a particular effect on the IL-6 pathway. Alteration of the IL-6/STAT3 pathway was confirmed with cytokine assays, immunohistochemistry, and flow cytometry, which revealed that myo-inositol seemed to predominantly affect tumor associated macrophages.

Materials and Methods

Cell culture

Kras-activated and p53 mutantcell lines 393P, 344SQ (murine adenocarcinoma lung cancer cell lines), LKR-13 (murine Kras mutant lung adenocarcinoma cell line), kindly provided by Dr. Jonathan M. Kurie (MD Anderson Cancer Center, Houston, TX) were cultured in RPMI 1640 containing 10% FBS and 1% penicillin/streptomycin cocktail (Gibco).¹³ Sample collections were done after 24 h and 48 h of myo-inositol (Sigma) treatment.

MTS assays

Cytotoxicity to myo-inositol was assessed by CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega) according to the manufacturer’s instructions. Absorbance was measured at 490 nm using a microplate reader Softmax pro (Molecular Devices). IC₅₀ (50% inhibitory concentration) was determined for myo-inositol in lung cancer cell lines.

Transgenic mouse model of premalignancy

Ccsp^Cre/+; Kras^LSL−G12D/+ (CC-LR) mice were generated by crossing a mouse heterozygous for the Kras^LSL−G12D (LR) allele with a mouse containing Cre recombinase inserted into the Club cell secretory protein (Ccsp) locus.³ Animal protocols were approved by the Institutional Animal Care and Use Committee at University of Texas MD Anderson Cancer Center. Mice were maintained and scrutinized in the specific pathogen-free animal facility of University of Texas M.D. Anderson Cancer Center.

For myo-inositol containing diets, the concentration was calculated by assuming an average intake of 3 grams of diet per mouse per day, using an equivalent human dose of 4 g myo-inositol daily. This was scaled from the metabolic body weight of human (80 kg)^0.75 to mouse (20 g)^0.75 to a final concentration of 2.64 g myo-inositol/kg diet. Irradiated, color-coded experimental diets were obtained from Harlan Pharmaceuticals. Mice were placed on experimental diet within four days of weaning at 3–4 weeks.

Collection of mouse plasma, bronchoalveolar lavage fluid and tumor samples

Mice were anesthetized by IP injection of Avertin (tribromoethanol, Sigma Aldrich). Terminal blood collection by cardiac puncture was performed under anesthesia. For bronchoalveolar lavage, a cannula was inserted into the trachea and the lungs were instilled with 1 ml phosphate buffered saline twice. Bronchoalveolar lavage fluid (BALF) samples were then centrifuged for 10 min at +4°C at maximum speed. Lungs were fixed in 10% formalin and used for histopathological examination. Blood collection was done by using tubes with K2EDTA: HemogardTM Closure. Samples were centrifuged for 20 min at 1500×g at 4 °C without braking and top layer collected. Plasma samples were stored at −80°C.

Quantification of cytokines/chemokines

Cytokine/chemokine analysis was performed using a multiplex, bead-based (Luminex®) system. 4-plex (IFN-ɣ, IL-6, IL-12p40, MCSF) and 5-plex (eotaxin, IL-10, LIF, IP-10, MIG) assays were used to quantify cytokine/chemokine concentrations following the manufacturer’s protocol (EMD Millipore). To adjust for BALF concentration, all cytokine/chemokine data were normalized to total protein content in the fluid as measured by Bradford Assay (BioRad).

Histopathological examination

After fixation in 10% buffered formalin, tumor samples were embedded in paraffin blocks and sectioned at 5-μm thickness. Tumor sections were stained with hematoxylin-eosin and examined in a double-blinded fashion.

Immunohistochemistry

Immunohistochemistry for phospho-STAT3 (p-STAT3) was performed on paraffin-fixed tissue sections. After deparaffinization, sections were boiled for 15 min in ImmunoDNA Retriever Citrate buffer (pH 6.0) (Bio SB) for antigen retrieval. Endogenous peroxidases were quenched by 3% H₂O₂ for 10 mins. Sections were then blocked with TBST containing 5% goat serum for 45 min at room temperature (RT) and incubated overnight with primary antibody in TBST/5% normal goat serum. After slides were washed, for non-directly conjugated antibodies, they were incubated with secondary Ig-HRP for 45 min at RT. DAB staining was applied to visualization of protein expression. Following nuclear counterstaining with Mayer’s hematoxylin (Dako), sections were then coverslipped with DPX mountant (Sigma). For staining quantification, Leica Aperio ImageScope software was used to annotate whole lung sections and lesions and quantify strongly-positive pixels and total pixels. For each antibody, at least 16 slides, from 4 mice of each genotype and diet, were stained. Anti-CD4 (eBioscience #14-9766), -CD8 (eBioscience #14-0808), -F4/80 (Abcam #ab6640), -cleaved Caspase-3 (Cell Signaling #9661), -Ki-67 (Bethyl Lab #IHC-00375-1) were stained using standard protocols at the University of Texas MD Anderson Histology Core at Science Park.

Analysis of proteins derived from plasma and BALF by MS/MS

Plasma proteins from tumor-group and age-matched littermate controls were analyzed by large-scale high-resolution proteomic experiments. Samples were immunodepleted for 3 common plasma proteins (albumin, IgG and transferrin) by passing over immunodepletion column Mu-3 (Agilent, Part number: 5188-5218). Samples were isotopically labeled using Thermo 6-plex tandem mass tags (TMT, Thermo Scientific, Rockford, IL, USA). Extensive fractionation by nanoscale liquid chromatography and quantitative tandem mass spectrometry (nanoLC/MS) was performed as described previously.¹⁴ For plasma mass spectrometry, an independent pool of case with control (equal volume) and myo-inositol treated case with control (equal volume) were prepared from 4–6 mice. For BALF mass spectrometry, an independent pool of case with control (equal amount of protein) and myo-inositol treated case with control (equal amount of protein) were constituted from 3–8 14-week old female CC-LR and LR mice.

Real Time PCR

RNA was isolated from mouse lung tissues using the Qiagen RNeasy kit (Qiagen, Santa Clarita, CA, USA). cDNA synthesis was performed by High-Capacity cDNA Reverse Transcription Kit (Thermo Scientific, Rockford, IL, USA). Quantitative Real Time PCR was performed according to manufacturer protocol using CFX Connect™ Real time PCR Detection System (ABI, Foster City, CA). Relative gene expression data from each sample were normalized according to GAPDH and 18S rRNA. For each experiment, at least 16 samples, consisting of RNA isolated from 4 different mice of each genotype and diet combination, was analyzed in duplicate.

Western Blotting

Mouse lung tissues and cell lines were homogenized in lysis buffer including RIPA buffer, protease and phosphatase inhibitor (Roche). Protein concentration was measured using Bradford Assay (Bio-Rad Laboratories). Western blot analysis was performed using 20 μg/lane protein according to standard procedures using polyvinylidene difluoride membranes and an enhanced chemiluminescence system (GE Healthcare). Anti-phospho STAT3 (Tyr705) (Clone D3A7) (Cell Signaling Technology, 1:2000), anti-phospho Akt (Ser473) (Cell Signaling Technology, 1:1000), anti-Akt (Cell Signaling Technology, 1:1000), anti-STAT3 (Cell Signaling Technology, 1:2000), anti-STAT1 (Cell Signaling Technology, 1:1000), anti-phospho ERK (Cell Signaling Technology, 1:2000), anti-ERK (Cell Signaling Technology, 1:1000), anti-Ras (Cell Signaling Technology, 1:1000), anti-Ras (G12D) (Cell Signaling Technology, 1:1000), anti-gp130 (CloneM-20) (Santa Cruz,1:200), anti-phospho c-Raf (Cell Signaling, 1;1000), anti-c-Raf (Cell Signaling, 1:1000), anti-Bcl-2 (Cell Signaling, 1:1000), anti-caspase-3 (Cell Signaling, 1:1000), anti-β-actin (Clone AC-15) (Sigma, 1:20000), and anti-β-tubulin (Cell Signaling, 1:2000) were used for Western blot analysis. A total of 20 mice, consisting of 5 of each combination of diet and genotype, were used for all analyses. Western blots were all repeated at least once.

Flow cytometry

Cells were stained with fluorescently labeled antibodies using anti-mouse CD45 (30-F11), anti-mouse F4/80 (BM8), anti-mouse CD11c (N418), anti-mouse Ly-6C (AL-21) and anti-mouse I-A/I-E (M5/114.15.2) and anti-mouse CD206 markers and analyzed on a Galios 561 machine. BALF from 20 total mice, consisting of 5 of each combination of diet and genotype, were used for analyses.

myo-inositol treatment of human PBMC-derived macrophages

Healthy donor derived peripheral blood mononuclear cells (PBMCs) were used for macrophage differentiation assay with and without 5 mM myo-inositol treatment. For macrophage differentiation, PBMCs were incubated in monocyte attachment medium (PromoCell, Heidelberg, Germany) at 37°C. After 1–1.5 h, non-adherent cells were removed and adherent monocytes were differentiated into M1 or M2-like macrophages using M1 or M2 Macrophage Generation Medium DXF (PromoCell, Heidelberg, Germany) with or without myo-inositol according to the manufacturer’s instructions. Following differentiation, macrophages and macrophage media were used for flow cytometric analysis and measurement of IL-6 levels, respectively at day 10. Anti-human CD68, anti-human CD86, and anti-human CD163 antibodies were used for flow cytometric identification of M1 and M2-like macrophages.

Statistical analyses

Statistical analyses were performed using the R statistical analysis software (version 3.2.2) and packages including plyr, ggplot2, limma, and reshape.^15–18 Student’s pairwise t-test, ANOVA with Tukey HSD post-hoc testing, and Mann-Whitney U test were used for inter-group comparisons. P≤0.05 was considered significant. For mass spectrometric experiments, data were log₂ transformed and quantile normalized for each tandem mass tag channel (Aguilar et al, in preparation). Data were analyzed on a peptide-based level, with imputation of protein level by averaging all individual peptides for a single protein.

Results

Mice raised on myo-inositol diet develop fewer lung tumors than those on control diets

To determine the effect of potential chemopreventive agents on premalignant lesions, CC-LR and control LR (Kras^LSL−G12D/+) mice were weaned at 3–4 weeks post-partum to cages containing diet compounded with myo-inositol or a control diet. The concentration of myo-inositol was calculated by scaling the daily therapeutic dose of the agent to the metabolic body weights of humans and mice and the approximate daily food intake of a mouse. Mice were kept strictly on these diets until sacrifice at 14 weeks. The drug-containing diet was well tolerated and there was no impact on overall body weights between the agent-containing diet and control diets (Sup. Fig. 1A).

Tumor burden was assessed at necropsy by lung surface tumor count under 5× magnification and by standardized histopathological examination of hematoxylin and eosin stained sections. myo-inositol reduced lung tumors in both stage and number (Fig. 1A–C). By lung surface tumor count, the mean number of tumors seen in CC-LR animals on a control diet was 46.6 ± 2.7, versus 22.0 ± 2.0 for those on myo-inositol (p < 0.05). Western blotting of whole lung tissue showed marked reduction in oncogenic Kras^G12D in CC-LR mice raised on myo-inositol versus those on a control diet, which likely reflects the markedly lower tumor volume. There was little effect on total Kras (Fig. 1D–F). A potential confounder could be that myo-inositol caused transcriptional repression of Ccsp, thus reducing Cre levels, leading to less oncogenic Kras signaling. This was assessed by qPCR of Ccsp, which showed no difference in animals on control or myo-inositol diet. (p=0.42, Sup. Fig. 1B) Within lesions, there was a mild reduction in cell proliferation as measured by Ki67 (Supp Fig 1C–D). However, Ki67 seemed to be localized in non-epithelial cells within the tumors. Levels of cleaved Caspase-3 (cCasp-3, Sup Fig 1D, 2A), were undetectable by immunohistochemistry or on Western blots of whole lung homogenates, indicating a low level of apoptosis.

Figure 1.

A) Lung surface tumor count, performed under 5× magnification, was significantly reduced in mice raised on a diet containing myo-inositol (p=1.04 × 10⁻¹³). B) Bronchial dysplasia (BD) and atypical adenomatous hyperplasia (AAH were reduced in mice raised on myo-inositol (p=0.006 for BD and 0.023 for AAH) C) Representative H&E sections of whole lungs show a gross reduction in dysplastic and neoplastic lesions evident on cancer-prone mice raised on myo-inositol (bottom) than those on a control diet (top). D–F) By Western blot (F) densitometry, oncogenic Kras^G12D (D) levels were reduced in whole lung tissue of mice raised on myo-inositol versus control.Total levels of Kras were unchanged (E) (p=0.005). * indicates p < 0.05 by t-test. n=5 mice for each genotype and diet used for Western blot analyses. β-tubulin loading control in Fig 1F and 2G are identical as same filters are shown.

Phosphoinositide-3-kinase signaling is not altered in myo-inositol

myo-inositol has been reported to inhibit PI3K signaling, and this has been proposed to be the mechanism for its chemopreventive action.¹⁰ Since it was expected that myo-inositol would cause a systemic effect, PI3K signaling was assessed by measuring AKT and phospho-AKT (p-AKT) by semi-quantitative Western blotting on protein extracts of whole lung tissue. AKT and p-AKT levels were unchanged in LR versus CC-LR in either control or myo-inositol diets (Fig. 2A–C).

Figure 2.

A–C) Western blot (C) densitometry of phospho-Akt (A) and total Akt (B) levels from mouse whole lungs reveals no significant differences in animals raised on myo-inositol versus those on a regular diet. D–G) Western blot (G) densitometry on mouse whole lungs reveals that levels of phospho-ERK (p-ERK, (D)), total ERK (E), and phospho-cRAF (p-cRAF, (F)) are largely unchanged between CC-LR mice raised on control or myo-inositol diets. There is a mild reduction in total ERK levels in CC-LR mice raised on a myo-inositol diet versus those on control diets, with no significant differences observed in LR mice (p=0.0001). p-cRAF is mildly induced in LR mice raised on a myo-inositol diet (p=0.003).* indicates p < 0.05 by t-test. n=5 mice for each genotype and diet used for Western blot analyses. β-actin loading controls in Fig 2C and 4G are identical as the same filters were used for both figures. Similarly, β-tubulin loading control in Fig 1F and 2G are identical as same filters are shown.

To search for an alternative mechanism of action, semi-quantitative Western blotting was performed using a number of markers of pathways important in carcinogenesis. Kras signaling was the first pathway assessed. While oncogenic Kras^G12D was reduced in whole lung tissue, there was no apparent change in the Kras cytoplasmic effectors phospho-cRaf (p-cRAF) and phospho-ERK (p-ERK) in CC-LR mice. (Fig. 2D,F,G) There is a mild induction of p-cRAF levels in LR mice raised on myo-inositol. This could possibly be explained by heterogeneity in levels between different mice, as seen by Western blot in Figure 2G. Total ERK levels were somewhat reduced in CC-LR mice on myo-inositol, while total ERK was unchanged in LR mice on control and myo-inositol diets (Fig. 2E). Likewise, total Kras and p-ERK levels were unchanged in Kras-activated mouse lung cancer cell lines treated with myo-inositol, indicating that myo-inositol was not directly repressing Kras signaling. By p-ERK immunohistochemistry, across whole lung sections, there is a non-significant change in p-ERK levels between LR and CC-LR mice with or without myo-inositol by ANOVA with Tukey post-hoc testing (Sup Fig 1E, G). This staining appeared to be epithelial in nature. Within tumors, there was no significant differences noted between CC-LR mice raised on myo-inositol versus control diets (Sup Fig 1F, G).

In vitro assays of PI3K signaling on three mouse lung cancer cell lines, LKR-13, 393P, and 344SQ, were performed to assess whether oncogenic Kras signaling was modulating the PI3K inhibitory effect of myo-inositol. These cell lines contain a Kras^LA1 mutation which becomes activated by recombination into a Kras^G12D allele. 393P and 344SQ lines also carry a heterozygous p53^R172HΔG mutation.^{13, 19} After treatment with a range of doses of myo-inositol for 24 or 48 hours, no significant effect on cell viability was noted by MTS assay (Fig. 3) at physiologic doses of myo-inositol (1–25 mM). Levels of Kras and Kras^G12D were unchanged in 393P, 344SQ, and LKR-13 cells with myo-inositol treatment. Cleaved caspase-3 levels were undetectable, indicating very low levels of apoptosis (Sup. Fig. 2A). In 393P and 344SQ, there was no change in p-ERK levels, and in LKR-13 cells, p-AKT levels were not significantly changed by myo-inositol treatment (Sup. Fig. 2B–C).

Figure 3.

myo-inositol treatment of p53-mutated and Kras-activated mouse lung cancer cells lines (A) 393P, (B) 344SQ, or the Kras-activated p53-wild type mouse lung cancer cell line (C) LKR-13 reveals little reduction in MTS cellular proliferation assays at physiologic (1–10 mM) doses. At higher doses, especially above 50 mM, there is a small reduction in cellular proliferation.

Proteomic analysis shows alteration of inflammatory networks

For further pathways discovery, high sensitivity multiplexed nanoLC/MS of pooled plasmas collected from LR and CC-LR mice raised on myo-inositol and control diets was conducted. Output data was sorted for peptides that increased in cancer-prone CC-LR versus littermate control LR mice, with a magnitude greater in mice on a control diet than on a myo-inositol diet ((CC-LR – LR)_control > (CC-LR – LR)_myo). Complement and coagulation factors as well as mouse-specific peptides such as murinoglobulin were filtered out of the final list, yielding 434 peptides representing 123 proteins. Filtering this list for proteins with a greater than 0.5 log₂ normalized intensity increase in cancer vs control yielded a list of 50 proteins. (Supplemental Data) Network analysis performed using Ingenuity Pathway Analysis (IPA, QIAgen) revealed induction of numerous pathways related to inflammatory cell infiltration. (Table 1A, Sup. Table 1).

Table 1A.

Altered genetic networks in plasma proteomics

Molecular and Cellular Functions	Molecules	p-value range
Cellular Movement	20	6.66×10−03 – 2.95×10−10
Cell-To-Cell Signaling and Interaction	16	6.66×10−03 – 1.26×10−07
Lipid Metabolism	17	6.66×10−03 – 4.43×10−06
Molecular Transport	15	6.66×10−03 – 4.43×10−06
Small Molecule Biochemistry	18	6.66×10−03 – 4.43×10−06
Physiological System Development and Function	Molecules	p–value range
Hematological System Development and Function	15	6.66×10−03 – 1.26×10−07
Immune Cell Trafficking	14	5.99×10−03 – 1.26×10−07
Cell-mediated Immune Response	7	4.89×10−03 – 6.31×10−07
Tissue Development	19	6.66×10−03 – 6.37×10−06
Connective Tissue Development and Function	14	6.66×10−03 – 2.39×10−05

nanoLC/MS on BALF collected at 14 weeks was conducted using a similar pipeline as to that previously used for plasma samples. Initial analysis revealed 701 peptides representing 275 proteins. Of these, 158 peptides representing 23 proteins were increased in CC-LR versus LR on control diet and decreased in CC-LR on a myo-inositol diet versus those on control diet (Sup. Table 2). IPA of this list revealed enrichment in cell growth, movement, and proliferation networks (Table 1B).

Table 1B.

Altered genetic networks in BALF proteomics

Molecular and Cellular Functions	Molecules	p-value range
Cell Morphology	15	2.23×10−02 – 1.63×10−06
Cell-To-Cell Signaling and Interaction	15	1.96×10−02 – 6.21×10−06
Cell Death and Survival	16	2.10×10−02 – 1.00×10−05
Cellular Movement	11	2.17×10−02 – 2.17×10−05
Cellular Growth and Proliferation	17	2.17×10−02 – 5.66×10−05
Physiological System Development and Function	Molecules	p–value range
Organismal Development	17	2.17×10−02 – 2.77×10−06
Nervous System Development and Function	8	1.86×10−02 – 6.21×10−06
Cardiovascular System Development and Function	11	2.17×10−02 – 8.15×10−06
Organ Morphology	13	1.86×10−02 – 8.15×10−06
Connective Tissue Development and Function	9	2.10×10−02 – 2.17×10−05

Cytokine profiling shows amelioration of IL-6 mediated inflammation

To provide confirmatory evidence of changes in the inflammatory milieu as detected by plasma mass spectrometry, bead-based cytokine profiling of BALF was performed on cancer-prone and control animals raised on both diets. These assays revealed an expected rise in IL-6 in cancer-prone CC-LR animals raised on a control diet. Littermate control LR animals raised on a myo-inositol diet also had a smaller yet significant increase in IL-6 levels. In CC-LR animals, IL-6 was significantly reduced in animals raised on myo-inositol. (Fig. 4A). Animals raised on myo-inositol also had reductions in the levels of the IL-6-class cytokine Leukemia Inhibitor Factor (LIF, Fig. 4B).

Figure 4.

A) Bead-based immunoassays on bronchoalveolar lavage fluid (BALF) reveal a significant decrease in (A) IL-6 and (B) the IL-6 related cytokine LIF in CC-LR mice raised on myo-inositol versus those on a control diet (p=0.053 for LR in IL-6, p=0.019 for CC-LR in IL-6 and 0.036 for CC-LR in LIF). A mild independent induction of IL-6 was noted in LR mice raised on myo-inositol versus those on a control diet. Assays were normalized to total BALF protein concentration as measured by a Bradford assay. C–G) The reduction of IL-6 in CC-LR mice raised on myo-inositol diet is also reflected in decreased levels of phospho-STAT3 (C, p=0.01) and Gp130 (E, p=0.018) seen by whole lung Western blot (G). Also noted was a small induction in p-STAT3 in LR mice raised on myo-inositol (p=0.139). D) Total STAT3 levels are unchanged between CC-LR mice raised on control or myo-inositol diet, where myo-inositol does increase STAT3 protein levels in LR mice (p=0.018). There were no significant changes seen in phospho-STAT1 (F). * indicates p < 0.05 by t-test. n=5 mice for each genotype and diet used for all analyses. β-actin loading controls in Fig 2C and 4G are identical as the same filters were used for both figures.

Consistent with decreased levels of IL-6, Western blotting for phospho-STAT3 (p-STAT3) showed a marked reduction in cancer-prone animals raised on a myo-inositol diet (Fig. 4C, G). Total STAT3 levels were unchanged (Fig. 4D, G). As observed with IL-6, LR animals raised on myo-inositol had a higher level of p-STAT3, yet myo-inositol raised cancer-prone CC-LR animals have a markedly lower level of p-STAT3 than CC-LR animals on a control diet. Activation of STAT3 has numerous roles in promoting carcinogenesis, cancer cell survival, and tumor growth.²⁰ Expression of the Gp130 protein, which is required for JAK/STAT3 activation, was reduced in CC-LR mice raised on myo-inositol (Fig. 4E, G). One of the major functions of STAT3 in lung cancer is to promote epithelial-to-mesenchymal transition (EMT), in which it is antagonized by STAT1.^{13, 21} Phospho-STAT1 levels were low in all animals assessed, suggesting that EMT is not a major causative pathway (Fig. 4F, G).

Kras-activated mouse lung cancer cells were used to assess if myo-inositol directly repressed IL-6 and STAT3 activation. After 24 and 48 hour exposure to myo-inositol at a range of concentrations, there was no effect noted on STAT3 activation. (Sup. Fig. 2B) However, these cells had low endogenous levels of p-STAT3. Kras, p-STAT3, and STAT3 levels are affected by cell density and confluency, and levels of both increased from 24 hrs to 48 hrs, however, this change was not affected by myo-inositol. To test whether myo-inositol could modulate the intracellular response to IL-6, LKR-13 cells were treated with IL-6 (100 ng/ml) in the presence of exogenous myo-inositol (5 and 10 mM). Western blot analysis of these experiments showed no change of p-STAT3, suggesting that myo-inositol did not act directly on the IL-6 response within a cancer cell.

Immunohistochemistry was used to measure the effect of myo-inositol on intra-tumoral inflammatory cells. Total macrophage staining, as assessed by F4/80 positivity, was reduced in sections from CC-LR animals raised on myo-inositol (p=0.004, Sup. Fig. 4A, D). CD4 and CD8 staining, measuring major T-lymphocyte subsets, showed very low positivity and were not significantly changed (Sup. Fig. 4B–D)

IL-6 modulates the lung cancer microenvironment, promoting a pro-tumoral type 2 and suppressing an anti-tumoral type 1 macrophage response, which is reflected in BALF macrophage phenotypes.^22–24 BALF macrophage phenotypes were assessed using flow cytometry. More M1-like macrophages (CD45⁺F4/80⁺Ly6c⁻MHCII⁺) were detected, with no significant change in M2-like macrophages (CD45⁺F4/80⁺Ly6c⁻CD206⁺) in BALF of the CC-LR mice raised on myo-inositol vs those on a control diet (Fig. 5A–B, Sup. Fig. 5).

Figure 5.

CC-LR mice show a fall in (A) M1-type antitumoral macrophages and a small rise in (B) protumoral M2-type macrophages. With myo-inositol, there is a recovery of the fall in M1 macrophages with no significant change in M2 macrophages. For M1 macrophages, p=0.002 for LR vs CC-LR on control diet, p=0.03 for LR vs CC-LR on myo-inositol, p=0.008 for CC-LR on control vs myo-inositol diet; for M2 macrophages, p=0.019 for CC-LR vs LR on control diet and p=0.029 for CC-LR vs LR on myo-inositol. C) myo-inositol at 5 mM does not affect healthy PBMC differentiation into an M1-like phenotype (CD68^hiCD86⁺CD163⁻) but increases M2-like differentiation (CD68^hiCD86⁺CD163⁺). In M2 differentiation media, PBMCs that do not become CD163⁺ show an M2b-like phenotype. The percentage of these cells decrease with myo-inositol. D–F) myo-inositol decreases IL-6 secretion in M1-like (D, p=0.010), M2-like (E, p=0.001), and M2b-like (F, p=0.022) cells. * indicates p < 0.05 by t-test. n=5 mice for each genotype and diet used.

Hypothesizing that myo-inositol altered IL-6 secretion in macrophages, IL-6 levels were measured from the media of healthy PBMCs differentiated into M1-like or M2-like macrophages with and without 5 mM myo-inositol (Fig. 5C, Sup. Fig. 6). The percentage of cells positive for CD68, a pan-macrophage marker, did not change in the presence of myo-inositol. There was a marked increase in CD86⁺CD163⁺ M2-like macrophages in the presence of 5 mM myo-inositol. In these cultures, however, the M2b-like CD86⁺CD163⁻ macrophages were reduced. M2b macrophages are known to secrete IL-6.²⁵ IL-6 secretion, measured from M1-like, M2-like, and M2b-like differentiated macrophages, was significantly reduced in cells grown with 5 mM myo-inositol (Fig. 5D–F). This suggests that myo-inositol has roles in both promoting anti-tumoral macrophage differentiation while also reducing IL-6 secretion from all major subtypes of macrophages.

Intratumoral p-STAT3 signaling is suppressed by myo-inositol

Immunohistochemical staining for p-STAT3 revealed a similar pattern to that seen by Western blotting. CC-LR mice on a control diet showed significantly more p-STAT3 than LR mice, and CC-LR mice on a myo-inositol diet had a near complete abrogation of p-STAT3 staining (Fig. 6A–C).

Figure 6.

A) Immunohistochemical staining for p-STAT3 shows a rise in levels in CC-LR mice that is significantly decreased on a myo-inositol diet (p=0.023). B) p-STAT3 levels within tumors are higher, and are significantly reduced in CC-LR mice treated with myo-inositol (p=2.21 × 10⁻⁶). C) Immunohistochemistry shows reduced pSTAT3 staining in adenomas from mice raised on myo-inositol versus control diets Scale bar = 200 μm, micrographs at 10× magnification with a 20× inset. n=4 mice for each diet and genotype used for immunohistochemical analyses.

Discussion

A major reason that lung cancer is the leading cause of cancer mortality is that diagnosis is often made at a late stage. Chemoprevention of lung cancer, especially in high-risk populations, for instance those undergoing early detection screening by low-dose CT scan or those with a previous history of lung cancer, has the potential to increase early detection and reduce mortality. Here, the potential chemopreventive agent myo-inositol was tested in an animal model that recapitulates the progression from airway premalignant lesions to lung adenocarcinoma through oncogenic Kras activation in the airway epithelium. myo-inositol showed a strong chemopreventive effect in this model, reducing both tumor number and stage.

myo-inositol is a well-tolerated sugar alcohol found in a number of fruits and grains. Numerous previous studies have shown that myo-inositol has a chemopreventive effect in a tobacco-exposed mouse model of lung cancer. Its ability to repress PI3K in vitro provides a plausible mechanism of action for this chemopreventive effect. Supporting this, a PI3K gene expression signature was identified in the proximal airways of high-risk smokers. In high-risk smokers who had regression of these lesions while treated with myo-inositol, the PI3K signature was decreased. Building on this, a recent randomized, double-blind, placebo-controlled phase IIb trial was conducted to assess the effect of myo-inositol on airway premalignant lesions in high-risk smokers. This trial showed non-significant decreases in airway premalignant lesions between patients treated with placebo and myo-inositol. Patients on myo-inositol did have reduction in IL-6 measured in BALF, however, and those who did show complete response did have a reduction in PI3K activation in cytologically normal airway. This suggested a heterogeneous or modest effect of myo-inositol. An alternate explanation is that repression of PI3K in vivo is secondary to an unseen primary effect of myo-inositol, perhaps related to the decrease in IL-6.

This study confirmed that myo-inositol demonstrated a potent chemopreventive effect in a non-mutagen exposed Kras model of lung premalignancy and early cancer. However, in this model, PI3K and ERK signaling was largely unchanged by whole lung Western blotting and immunohistochemistry. Proteomic profiling revealed changes in numerous inflammatory pathways, as well as those associated with lung defense mechanisms and host response. This was confirmed by cytokine profiling of BALF, which demonstrated large reductions in IL-6 and the IL-6 related cytokine LIF. A similar reduction in IL-6 was seen in the aforementioned phase IIb trial. Correlating with the reduction in IL-6 is marked reduction in p-STAT3, seen both in Western blotting of whole lung homogenates and within tumors by IHC. Reduction in IL-6 seems to be altering the immune microenvironment, as demonstrated by an increase in anti-tumoral M1 macrophages. No effect was seen on other inflammatory cell subsets, including CD4 and CD8-positive lymphocytes. The link described here is correlative, but plausible. Obvious future work will be to test this mechanism, first in co-culture models and then by directly modulating IL-6 in animals raised on myo-inositol.

Secretion of IL-6 is induced by oncogenic Kras activation and promotes Ras-mediated tumorigenesis. IL-6 has well described roles in the initiation and growth of lung cancer and in modulating the tumor microenvironment. Early in tumorigenesis, Kras and PI3K signaling promote an autocrine cytokine circuit, leading to epithelial expression of IL-6.^{26, 27} This early expression leads to a pro-inflammatory tumor microenvironment and promotes recruitment of macrophages and class-switching of these macrophages into a pro-tumoral M2 phenotype. The role of IL-6 in inducing JAK/STAT pathways in tumor-associated macrophages (TAMs) is now well described in a variety of cancers. These include promotion of human hepatocellular carcinoma precursor cells, colorectal cancer, small cell lung cancer, and non-small cell lung cancer.^28–30 In numerous situations, TAM-derived IL-6 leads to activation of STAT3 for the promotion of tumorigenesis through induction of proliferation and suppression of apoptosis.³¹ Targeting TAMs through blockade of chemokines and cytokines, including those in the IL-6 response axis, is an area of intense investigation.³²

Chronic obstructive pulmonary disease (COPD) also induces IL-6 inflammation, perhaps mechanistically explaining the additional lung cancer risk and poor outcome to therapy seen in COPD patients.³³ Members of the IL-6 signaling pathway thus provide attractive targets for intervention, especially in patients with COPD. Indeed, deletion of IL-6 or its receptors, or blockade of IL-6 using monoclonal antibodies leads to smaller and fewer lung cancers in Kras mice models.^{24, 34, 35} While initial clinical trials on myo-inositol as a lung cancer chemopreventive did not yield positive results, this study presents a new opportunity to revisit myo-inositol’s chemopreventive effect in high-risk COPD patients.

Supplementary Material

Supp FigS1

Supplemental Figure 1: A) Diets with chemopreventive agents were well tolerated, with no significant differences in mouse body weight measured at analysis at 14 weeks B) There was no effect on CCSP levels by myo-inositol, as assessed by qPCR. C–D) Intralesional Ki67 was slightly but significantly reduced by immunohistochemistry, indicating reduced cellular proliferation (p=0.014). Ki67-positive cells seemed to be non-epithelial and are likely inflammatory (D, top panels, inset). Cleaved caspase-3 (cCasp-3) was not detected within lesions (D, bottom panels). E–G) By immunohistochemistry, phospho-ERK (p-ERK) was non-significantly reduced by myo-inositol treatment. Levels of p-ERK between LR and CC-LR mice was equivalent across whole lungs (E) or within lesions (F). Scale bar = 200 μm. n=16–20 total mice used for each genotype and diet for weight and tumor count analyses. n=4 mice for each diet and genotype used for immunohistochemical analyses.

Supp FigS2

Supplemental Figure 2: A) In Kras-activated p53-mutant mouse lung cancer cell lines, treatment with myo-inositol for 24 or 48 hours did not alter total or cleaved Casp-3 levels.) B) Western blots of Kras, Kras^G12D, p-STAT3, STAT3, p-ERK, and ERK in two Kras-activated and p53 mutant mouse lung cancer cell lines, 393P and 344SQ. No significant differences were noted. C) Western blots showing levels of oncogenic Kras^G12D, total Kras, p-Akt, and Akt from the Kras-activated LKR-13 mouse lung cancer cell line. There was no change in these proteins at a range of myo-inositol doses with treatment for 24 or 48 hours.

Supp FigS3

Supplemental Figure 3: myo-inositol does not directly alter STAT3 phosphorylation in Kras activated lung cancer cells. LKR-13 cells were either co-treated with recombinant IL-6 and myo-inositol (top) or pretreated with myo-inositol before IL-6 (bottom). LKR-13 cells not treated with IL-6 have low endogenous levels of p-STAT3. Levels of p-STAT3 were unchanged at two different doses of myo-inositol with IL-6 treatment.

Supp FigS4

Supplemental Figure 4: A) Within lesions of CC-LR mice, macrophages, as quantitated by immunohistochemistry for F4/80, are reduced in animals raised on a myo-inositol diet (p=0.004). B) Very few CD4-positive lymphocytes were observed with lesions, with no significant differences between animals raised on a control and myo-inositol diet. C) Few CD8-positive lymphocytes were observed within lesions. Treatment with myo-inositol did not induce significant changes. D) Representative lesions for CC-LR mice raised on control (left panels) or myo-inositol diets (right panels) and stained for F4/80 (top), CD4 (middle), and CD8 (bottom). Scale bar = 200 μm, micrographs are at 10× with 20× insets. n=4 of each diet and genotype for immunohistochemical analyses.

Supp FigS5

Supplemental Figure 5: Gating method for flow cytometric analysis of BALF macrophages. Live singlet cells were identified by forward- and side-scatter properties. Macrophage subsets were quantitated from the subset of CD45⁺/F4–80⁺ double positive cells. M1 cells were identified as Ly6C⁻/MHCII⁺ and M2 as Ly6C⁻/CD206⁺. n=5 mice of each diet and genotype used for analyses.

Supp FigS6

Supplemental Figure 6: Gating method for flow cytometric analysis of M1-like and M2-like macrophages derived from peripheral blood mononuclear cells (PBMCs). Live singlet PBMCs grown in M1- or M2-promoting differentiation media expressing high levels of the pan-macrophage marker CD68 were considered differentiated. myo-inositol did not alter the percentage of CD68⁺ cells. M1-like macrophages were identified as CD86⁺/CD163⁻. M2-like macrophages were identified as CD86⁺/CD163⁺ where M2b-like macrophages were CD86⁺/CD163⁻.

Novelty and Impact.

myo-inositol is a sugar alcohol that has been previous studied as a potential lung cancer chemopreventive agent. Previous work has focused on modulation of the phosphoinositide 3-kinase pathway as a mechanism of action for myo-inositol. Here we report that the chemopreventive effect of myo-inositol in a Kras-activated mouse model of lung premalignancy is instead associated with a reduction in IL-6 and phospho-STAT3 signaling, likely mediated through tumor-associated macrophages.

List of abbreviations

Fig.

figure

Sup. Fig.

supplementary figure

IL-6

interleukin-6

STAT

Signal transducer and activator of transcription gene family

Kras

Kirsten rat sarcoma viral oncogene homolog

Ccsp

Club cell secretory protein

PI3K

phosphoinositide 3-kinase

CC-LR

Ccsp^Cre/+ Kras^LSL−G12D/+LR Kras^LSL−G12D/+BD bronchial dysplasia

AAH

adenomatous atypical hyperplasia

IC₅₀

50% inhibitory concentration

BALF

bronchoalveolar lavage fluid

IFNɣ

interferon-ɣ

IL-12p40

interleukin-12 p40 subunit

MCSF

macrophage colony stimulating factor

IL-10

interleukin-10

LIF

leukemia inhibitory factor

IP-10

interferon-ɣ induced protein 10

MIG

monokine induced by gamma-interferon

TBST

Tris-buffered saline with 0.1% Tween-20

DAB 3

3′-diaminobenzidine

TMT

tandem mass tags

MS

mass spectrometry

MS/MS

tandem mass spectrometry

nanoLC/MS

nanoscale liquid chromatography and tandem mass spectrometry

PCR

polymerase chain reaction

cDNA

complementary DNA

RIPA

radioimmunoprecipitation assay buffer

Akt

RAC-alpha serine/threonine-protein kinase

ERK

extracellular signal–regulated kinase

c-Raf

Virus-induced Rapidly Accelerated Fibrosarcoma proto-oncogene serine/threonine-protein kinase

CD

cluster of differentiation gene family

PBMC

peripheral blood mononuclear cells

ANOVA

analysis of variance

Tukey HSD

Tukey’s Honest Significant Differences post-hoc analysis

MHC

major histocompatibility complex

NF-κB

nuclear factor κB

IPA

Ingenuity Pathway Analysis

JAK

Janus kinase

EMT

epithelial-to-mesenchymal transition

CT

computerized tomography

IHC

immunohistochemistry

COPD

chronic obstructive pulmonary disease