A polymorphism in the lysyl oxidase propeptide domain accelerates carcinogen-induced cancer

Abstract

The propeptide (LOX-PP) domain of the lysyl oxidase proenzyme was shown to inhibit the transformed phenotype of breast, lung and pancreatic cells in culture and the formation of Her2/neu-driven breast cancer in a xenograft model. A single nucleotide polymorphism (SNP, rs1800449) positioned in a highly conserved region of LOX-PP results in an Arg158Gln substitution (humans). This arginine (Arg)→glutamine (Gln) substitution profoundly impaired the ability of LOX-PP to inhibit the invasive phenotype and xenograft tumor formation. To study the effect of the SNP in vivo, here we established a knock in (KI) mouse line (LOX-PP^Gln mice) expressing an Arg152Gln substitution corresponding to the human Arg158Gln polymorphism. Breast cancer was induced in wild-type (WT) and LOX-PP^Gln female mice beginning at 6 weeks of age by treatment with 7,12-dimethylbenz(a)anthracene (DMBA) in combination with progesterone. Time course analysis of tumor development demonstrated earlier tumor onset and shorter overall survival in LOX-PP^Gln versus WT mice. To further compare the tumor burden in WT and LOX-PP^Gln mice, inguinal mammary glands from both groups of mice were examined for microscopic lesion formation. LOX-PP^Gln glands contained more lesions (9.6 versus 6.9 lesions/#4 bilateral). In addition, more DMBA-treated LOX-PP^Gln mice had increased leukocyte infiltrations in their livers and were moribund compared with DMBA-treated WT mice. Thus, these data indicate that the Arg→Gln substitution in LOX-PP could be an important marker associated with a more aggressive cancer phenotype and that this KI model is ideal for further mechanistic studies regarding the tumor suppressor function of LOX-PP.

Introduction

Cancer, as we now know it, is a heterogeneous disease with at least 100 different specific diseases defined by genetic changes and non-genetic alterations (1). Most human breast tumors develop in the glandular tissue and are called adenocarcinomas with ductal carcinomas comprising the majority of all breast cancers (2,3). In addition, more rare metaplastic breast carcinomas (MBCs) have been described. The first description dates to 1973, when Huvos et al. described a breast carcinoma with mixed epithelial and sarcomatoid components (4). These carcinomas are now defined as a heterogeneous group of neoplasms that exhibit mesenchymal or non-glandular components, such as spindle cells, squamous cells or mesenchymal components (5). While MBCs are rare, representing 0.25–1% of all mammary tumors, these neoplasias present as more aggressive cancers with a 5-year survival rate ranging from 49 to 68% (6,7).

Though current treatment strategies for breast cancer have met with some success, patients with a similar disease type often respond very differently to the same drug. Variations in response are likely due to heterogeneity established at the cellular level within different areas of the individual tumor, known as intra-tumor heterogeneity and/or inter-tumor heterogeneity (8–11). Tumor heterogeneity can result from somatic and/or epigenetic variations and from mutations induced by various insults (12,13). In addition, single nucleotide polymorphisms (SNPs) common among subsets of patients have been shown to be associated with variations in clinical responses (14), disease profiles and disease susceptibility (15–17). Many genome-wide association studies have been conducted, and associations of SNPs with susceptibilities have been established (18,19). Stronger dose-dependent associations are observed when several SNPs are combined (20). Thus, SNPs can be useful biomarkers for risk assessment and treatment strategies.

Lysyl oxidase (LOX) is the prototypic member of the LOX enzyme family, which oxidizes lysine and hydroxylysine residues in collagens and elastin (21). LOX is produced by stromal-derived fibroblasts as pro-lysyl oxidase (Pro-LOX) and cleaved in the extracellular space by procollagen C-proteinases to form active ~30 kDa LOX enzyme and an 18 kDa propeptide known as lysyl oxidase propeptide (LOX-PP) (22,23). Importantly, both proteins have independent roles in tumorigenesis. While the gene is not a tumor suppressor, the cleaved LOX-PP functions as a tumor suppressor (24–26), and the overexpressed LOX enzyme promotes metastasis formation in part by participating in the formation of the metastatic niche and altered ECM structures (27). In contrast, the tumor suppressor protein LOX-PP inhibits the transformed phenotype of tumor cells by attenuating cancer cell proliferation, anchorage-independent growth, migration and invasion in vitro and growth of newly established and pre-existing xenograft tumors in vivo (24,28). Interestingly, the suppressor activity is impaired in a naturally occurring variant (LOX-PP^Gln) generated by a SNP G473A (rs1800449), which is located in a highly conserved region in the LOX-PP domain (29,30). This SNP is present at a proportion of 11–26% in various populations in the HapMap database. Importantly, the SNP associates with estrogen receptor alpha negative (ERα−) breast cancer in African-American women, thus linking it to the expression of the LOX-PP^Gln in this population (25). Importantly, the increased cancer association of this SNP has been corroborated recently for breast cancer in the Chinese Han population (30) and in other solid tumors, including lung, colorectal, gastric and ovarian cancer (29,31,32), and oral submucosal fibrosis, a precancerous condition of the oral mucosa (33). Also, the LOX-PP^Gln has been associated with osteosarcoma for which higher expression levels have been found in metastatic disease (34). Overall, these data suggest an important role of the stroma-derived LOX-PP^Gln in the biogenesis of solid cancers, which has been addressed directly.

In this study, we established a knock in (KI) mouse model that expresses the LOX-PP^Gln and tested whether it alters the susceptibility to tumorigenesis induced by the polycyclic aromatic hydrocarbon 7, 12-dimethylbenz(a)anthracene (DMBA) (35). Tumors induced by this agent are fast growing, primarily MBC with squamous cell components (36). Untreated mice developed normally and grew up without noticeable health issues. Homozygous LOX-PP^Gln mice treated with DMBA developed tumors significantly faster, and more mice were moribund compared with their DMBA-treated wild-type (WT) control counterparts. These findings indicate that the Aa exchange from Arg→Gln in LOX-PP leads to the loss of a mechanism safeguarding against tumor development.

Materials and methods

Animals

Mice were housed in a pathogen-free barrier facility at the Laboratory Animal Science Center (LASC) of Boston University School of Medicine (BUSM). Animal experiments were performed in accordance with the regulations of the American Association for the Accreditation of Laboratory Animal Care (AAALAC), and the mice euthanized according to guidelines approved by the Institutional Animal Care and Use Committee (IACUC) at Boston University Medical Center. The IACUC approved protocol was listed under #AN15055.

Mouse generation and genotyping

Heterozygous LOX-PP^Gln mice in the C57Bl/6 background were generated through established embryonic stem cell technology by the UC Davis Mouse Biology Program. The LOX-PP^Gln genotype was achieved by exchange of two nucleotides in the propeptide coding region of the LOX gene. These mice were mated with C57Bl/6 WT mice (Taconic laboratories) for colony expansion and subsequently for homozygosity. Genotyping was performed by touchdown PCR. Mouse tails were digested in 250 µl of Direct PCR Tail buffer (Viagen cat # 101-T) plus 1.5 µl of Proteinase K (EMD Millipore cat # 124568) overnight. Twenty-five microliter reaction mixture was prepared by mixing 2 µl of genomic DNA, 0.5 µl of Taq polymerase (Crimson Taq DNA Polymerase, NEB cat # M0324S), 5 µl of buffer 5×, 6.5 µl of betaine 5 M, 0.325 µl of DMSO and oligonucleotides (0.4 µM each): R 5′-GCTGATGACCTCTGACTC-3′ and F 5′-GGGAGCAGGAACCGACCTGATAC-3′ (Invitrogen). Samples were heated for 5 min at 94°C followed by 10 cycles (94°C 15 s, 65°C 30 s decreasing 1º per cycle, 68°C 40 s) and 30 cycles (94°C 15 s, 55°C, 68°C 40 s). The PCR products were separated through a 1% agarose gel.

DMBA treatment

Homozygous Pro-LOX^Gln and C57Bl/6 WT nulliparous female mice were treated with DMBA as follows: Twenty-four 5-week-old Pro-LOX^Gln (KI) and Pro-LOX WT mice each was implanted with 4.5 mm diameter 90 days slow release pellets of medroxyprogesterone acetate (Innovative Research of America) at the right dorsal side (see Supplementary Figure S2A, available at Carcinogenesis Online). For this procedure, mice were anesthetized by I.P. injection of a ketamine/xylazine mixture as described previously (35). A 5 mm incision was made through the skin of the right dorsal side, and the pellet was implanted with forceps. The incised region was sealed with a 9 mm autoclip. Mice were treated with 1 mg of DMBA in 0.1 ml of sesame oil once per week by oral gavage starting at week 6 for six consecutive weeks. Mice were examined for mammary tumors by palpation twice a week. Tumor diameters were measured with calipers, and tumor volume was estimated as described (37). Mice were euthanized and underwent necropsy when the tumors reached 1.5–2 cm or when mice became moribund as indicated by excessive weight loss or difficulty breathing. The time when a mouse was euthanized is referred to as ‘end of study’ period herein. Mammary tumors were collected, measured and weighed, and portions of the tumors were fixed in Optimal Fix (American Histology Reagent Company, Inc.) or 4% formaldehyde for histology. The remaining tumor tissue was snap frozen in liquid nitrogen and stored at −80°C for molecular and biochemical studies. In addition, lung, liver, spleen and thymus tissues were collected and processed as described for the tumors.

Mammary gland whole-mount analysis

Mammary gland whole-mount analysis was performed on four mice per time point as described (36). Briefly, the #4 inguinal mammary fat pads were spread onto microscope slides, fixed in Carnoy’s fixative overnight, hydrated and stained with carmine alum stain (Sigma–Aldrich) overnight. Subsequently, the samples were dehydrated, treated with xylene to remove the fat. Coverslips were applied and mounted with ‘Protocol’ Mount Medium (Fisher Scientific) and observed under an Olympus SZX16 stereo microscope. Pictures were taken with a QImaging camera.

Assessment of lesion development in #4 inguinal mammary glands

Whole mounts of the left and right #4 inguinal mammary glands of all mice in the LOX-PP^Gln and WT groups were prepared as described above in ‘Mammary gland whole-mount analysis.’ The processed glands were viewed under a stereo microscope. The number of microscopic lesions (nodules and abnormal structures) in each group was counted under a microscope to further assess tumor development in response to DMBA treatment.

Antibodies

Antibodies used were Ki-67 (cloneSP6; Thermo Scientific), LOX-PP (rabbit polyclonal IgG) used as described previously (27), β-actin (#A-5316; Sigma–Aldrich) and biotinylated goat anti-rabbit IgG (Vector).

Histopathology and immunohistochemical analysis

Fixed tissues were processed and embedded in paraffin by the Pathology Laboratory Services Core at BUSM. Tissue sections were stained with hematoxylin and eosin (H&E), and immunohistochemistry staining was performed on 5 μm sections using the standard ABC method (Vector Laboratories) and counterstained with hematoxylin (37). H&E slides were viewed by a pathologist based on WHO human breast histopathology criteria. Images were scanned using a TissueFAXS fluorescence slide scanner with a 10× Zeiss EC Plan-Neofluor 0.3NA objective and Baumer HXG40c 16 bit CMOS color camera coupled to a Zeiss Axio Imager Z2 upright microscope.

Protein extraction from tumor tissues and western blotting

Frozen tissues were manually pulverized under liquid nitrogen and lysed in RIPA buffer [50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 1% Nonidet P-40 (NP-40), 0.5% sodium deoxycholate and 0.1% sodium dodecyl sulfate], containing protease inhibitors (Roche). Lysates were cleared by centrifugation at 13 000×g at 4°C for 30 min. Equal amounts of protein were diluted with 5× sample loading buffer (10% SDS, 1 M Tris–HCl, 30% glycerol, 10% β-mercaptoethanol and bromophenol blue) boiled, separated through 8–10% polyacrylamide gels and transferred onto nitrocellulose membranes (Millipore). Membranes were blocked in 3% bovine serum albumin in 1× TBS, 0.1% Tween and incubated with the antibodies, as described (24).

LOX enzyme activity assay

Tumor tissues

Frozen tumor tissues were pulverized as described above and extracted in 200 μl of 4 M urea, 50 mM sodium borate, pH 8.0, supplemented with 0.25 mM phenylmethanesulfonyl fluoride and 1 μl/ml of aprotinin. Samples were incubated on ice for 10 min and centrifuged at 10 000 × g for 15 min at 4°C. Extracts were collected, and the total protein was measured in each sample using the Bradford method.

Activity assays

LOX enzyme activity was measured using the Amplex Ultra-Red Fluorescent dye as described (38). Briefly, equal amounts of protein samples were divided into two aliquots and transferred into a glass tube. β-Aminopropionitrile (0.5 mM final concentration) was added to one of the two aliquots for each sample to inhibit LOX enzyme activity followed by a master mixture resulting in final concentrations of 1.2 M urea, 50 mM sodium borate, pH 8.2, supplemented with 10 mM 1,5-diaminopentane HCl, 10 μM Amplex Red and 1 U/ml horseradish peroxidase. The samples were incubated at 37°C. The fluorescent product was measured at an excitation wavelength of 560 nm, and the emission was read at 590 nm using a fluorescence plate reader (Synergy HT; Biotech).

Statistical analysis

The disease-free survival (appearance of palpable tumors) and overall survival (OS) analyses of the animals were performed by analyzing a Kaplan–Meier curve, and the log-rank test was used to assess for differences in tumor curves. The number of samples per group (n) is specified in Figure 2B and C. All P-values were two-tailed, and statistical significance was accepted at P ≤ 0.05. P-values in Figure 2D and E were calculated using unpaired Student’s t-test. The P-value in Figures 4C, 5D and 6B were calculated using unpaired Student’s t-test. Error bars represent the SEM. SEM values are represented in figure legends. All statistical analyses were performed in GraphPad Prism 7.

Figure 2.

LOX-PP^Gln mice display more rapid tumor development following DMBA treatment. (A) Schematic depicting treatment regimen and observation period for tumor development for WT and LOX-PP^Gln mice. (B) Comparison of disease-free survival in WT and LOX-PP^Gln animals. P-values were calculated using log-rank test. (C) Comparison of OS in WT and LOX-PP^Gln animals. P-values were calculated using log-rank test. (D) Tumor volumes were determined by caliper measurements. Mean values for each group are indicated. (E) Tumor weights and tumor volumes were determined at time of euthanasia. Mean values for each group are indicated. n, number of mice examined. (*) P ≤ 0.05. SEM tumor weight (g): WT ± 0.05, LOX-PP^Gln ± 0.1. SEM tumor volume mm³: WT ± 104, LOX-PP^Gln ± 230.

Figure 4.

Lesion development in the inguinal mammary gland #4/bilateral. (A) Whole mount of a representative mammary gland of a WT mouse. (B) Whole mount of a representative mammary gland of a LOX-PP^Gln mouse. Black arrows point to lesions; white arrows indicate lymph node. Region in square is shown enlarged 4×. Bar on the top panels 5 mm. Bar on the bottom panels 1.25 mm. (C) A number of microscopic lesions for both groups were counted and summarized in the graph. Mean values for each group are indicated. SEM: WT ± 0.8, LOX-PP^Gln ± 1.3. n, number of mice in each group.

Figure 5.

Lymphoid infiltrations in livers and lungs from LOX-PP^Gln and WT mice. (A) H&E staining of LOX-PP^Gln and WT that shows a lymphocyte infiltration nest in localized areas. Bar, 50 µm. (B) H&E staining of a representative liver of LOX-PP^Gln and WT mice. Bar, 100 µm. White arrows point to areas with lymphocyte infiltrations. Ki67 stained nuclei. Black arrows indicate positive cells. Bar, 50 µm. (C) H&E staining of a representative lung section of LOX-PP^Gln and WT mice with massive lymphocyte infiltration. Insets are presented at ×2 magnification. Bar, 100 µm. (D) Comparison of number Ki67+ cells/40× view field in liver and lung sections in WT and LOX-PP^Gln mice. Liver Ki67 positive cells SEM: WT ± 4.5 LOX-PP^Gln ± 9.9. Lung Ki67 positive cells SEM: WT ± 17, LOX-PP^Gln ± 60.

Figure 6.

Lysyl oxidase expression and LOX enzymatic activity in tumor tissues. (A) Protein expression determined by IHC with an anti-LOX-PP antibody. Examples of adeno- and squamous cell carcinoma isolated from WT mice and of adenosquamous, SCC and SCC with fibrosis from LOX-PP^Gln mice are presented. Bar, 100 µm. Areas of ×2 magnifications are depicted on the right. (B) LOX enzyme activity in tumor tissue from WT and LOX-PP^Gln mice. The bars represent the activity in arbitrary units and SEM WT ± 285, LOX-PP^Gln ± 704 of four tumors/group. n.s., non-significant.

Results

KI mouse (LOX-PP^Gln), expressing a naturally occurring LOX variant, develop normally

To elucidate in vivo the effects of loss of the tumor suppressor function of the variant LOX-PP^Gln, we generated a KI mouse that expresses LOX-PP^Gln, which contains an Arg→Gln exchange at Aa position 152, which corresponds to human residue 158 of the propeptide domain (24). In humans, the SNP rs1800449 represents a missense mutation. The codon usage in the murine sequence for Arg (AGG) differs from that for the human sequence (CGG) which required the exchange of two nucleotides AGG to CAG to achieve the same conversion from Arg→Gln in the protein sequence. Sequencing of genomic DNA of WT and LOX-PP^Gln mice confirmed that the nucleotide exchanges were correctly accomplished. PCR amplification shows the amplicon sizes of the WT, heterozygote and homozygote KI animals (Figure 1A–D). LOX-PP^Gln mice appeared healthy and were born at a normal ratio of WT, heterozygote and homozygote KI animals. Immunoblot analysis on lysates of fibroblasts isolated from lung tissues of WT and LOX-PP^Gln mice revealed that Pro-LOX was equally expressed and processed. LOX-PP and LOX-PP^Gln peptides were detected in the tissue culture supernatants of these cells. Moreover, both LOX-PP and LOX-PP^Gln had comparable expression levels (Figure 1E).

Figure 1.

Establishment of LOX-PP^Gln KI mice and comparison of mammary gland development in WT and LOX-PP^Gln mice. (A) Schematic of LOX genomic organization. Exons 1–7 are indicated, open box depicts untranslated regions. Positions of the translation start site (ATG), the SNP and the flippase recognition target (FRT) site that remained after excision of the Neo cassette are indicated. (B) Schematic of the protein structure of Pro-LOX. The signal peptide (striped box, aa 1–21), LOX-PP (aa 22–162) and, LOX-EZ (aa 163–411) and the approximate position of the Arg→Gln substitution are indicated. Cleavage of LOX-PP from LOX enzyme occurs between Gly:Asp (aa 162:163). (C) Sequence pherogram of WT and KI (LOX-PP^Gln) depicting the nucleotide exchanges introduced into the LOX gene of the LOX-PP^Gln mice. (D) Genotyping results from WT, heterozygote and homozygote animals. The size for Pro-LOX^Gln/Gln band was 451 bp and for WT mice 350 bp. (E) LOX-PP and Pro-LOX expression in supernatant and whole cell extracts (WCE), respectively, from fibroblast isolated from WT and LOX-PP^Gln mice. Blots of cell layer extracts were probed for β-actin, as a loading control. (F) Whole-mount preparations of gland number 4 of virgin animals at 6 and 10 weeks (wk) and of parous mice at day 4 of involution are presented. Bar, 5 mm. Insets are presented at ×10 magnification for each gland. Dark oval spots represent lymph nodes.

To investigate the effect of the LOX-PP^Gln on mammary gland development, we performed whole-mount analysis on #4 mammary glands of 6-week-old, 10-week-old mice and on involuting mammary glands at day 4 after weaning. Mammary gland development in virgin LOX-PP^Gln and WT mice and involution at day 4 of weaning was comparable (Figure 1F), and all mothers nursed their pups normally. No health issues or spontaneous tumor development was observed in LOX-PP^Gln mice during the entire study period, suggesting that the LOX-PP^Gln does not alter functions important for normal life and health.

Accelerated DMBA-induced tumor development in LOX-PP^Gln mice

The effects of LOX-PP^Gln on tumor development were studied in mice that were treated with 90 days slow release medroxyprogesterone acetate plus DMBA. Medroxyprogesterone acetate pellets were implanted in 5-week-old LOX-PP^Gln and WT control mice (n = 24 per group) and subsequently treated with one weekly dose of 1 mg DMBA each for six consecutive weeks starting at week 6 (Figure 2A). Appearance of mammary tumors was monitored by palpation in LOX-PP^Gln and control mice, and the onset of palpable tumors for each mouse group was plotted (Figure 2B). The first mammary tumor in LOX-PP^Gln and WT mice was detected at week 13 and 14, respectively. Median disease-free survival was 16.5 weeks for LOX-PP^Gln mice and 19 weeks for WT mice. LOX-PP^Gln mice showed a significantly enhanced susceptibility (P = 0.0028) to tumor development compared with WT mice determined by log-rank test and shown in the Kaplan–Meier curve (Figure 2B). Similarly, LOX-PP^Gln mice had significantly shorter OS (P = 0.0056) with a mean OS of 18.5 weeks for LOX-PP^Gln mice compared with 20.75 weeks for WT mice (Figure 2C). Moreover, the volumes and weights of palpable mammary tumors at the end of the study period were also significantly larger for the KI mice compared with WT-treated mice (Figure 2D and E). For example, the mean tumor volume was 1.14 versus 0.6 cm³, LOX-PP^Gln and LOX-PP, respectively, while, the mean tumor weights were 0.6 g versus 0.36 g, LOX-PP^Gln and LOX-PP, respectively. These data suggest that the tumor suppressor role of LOX-PP is compromised in the LOX-PP^Gln mice.

DMBA-treated LOX-PP^Gln mice develop breast cancer, displayed lymphoid infiltration in lungs and livers and became increasingly moribund

By the end of the study period, 91.7% of the LOX-PP^Gln and WT mice developed tumors with two healthy mice (8.4%) surviving in the WT group and two mice (8.4%) in the LOX-PP^Gln group that developed severe dermatitis requiring euthanasia but with no signs of skin cancer. Previously, DMBA treatment of multiparous FVB mice was shown to induce mammary and lung tumors, lymphomas and skin cancer (38). Here, we found that the majority of mice developed breast cancer: 19 LOX-PP^Gln mice (79.2%) and 22 WT mice (91.6%) developed breast cancer. In addition, one WT mouse and two mice in the LOX-PP^Gln group had more than one type of malignancy. DMBA did not induce lung cancer in either of our two groups in the C57Bl/6 background.

In both groups, the most common histopathological findings of the DMBA-induced breast cancers were metaplastic carcinomas, specifically squamous, adenosquamous carcinomas and spindle cells components were present in two tumors isolated from WT-treated mice (Figure 3). In addition, most of the tumors had basaloid components. While many of the tumors were large, mixed histopathologies were common, and necrosis was present in tumors of the WT and LOX-PP^Gln group. Muscle infiltration was noted in two tumors in the LOX-PP^Gln group (Figure 3B, LOX-PPG^ln). The #4 inguinal mammary glands were examined for microscopic lesions by whole-mount analysis for differences in tumor burden as a function of genotype (Figure 4A and B) to further assess tumor formation in response to DMBA treatment. The #4 gland has several advantages: (i) it has been extensively used as model system, (ii) it can be completely dissected without interference by a neighboring gland and without contamination of skeletal muscle tissue.

Figure 3.

Histology of breast tumors from WT and LOX-PP^Gln mice. Left panels, WT mice: (A) poorly differentiated SCC, (B) poorly differentiated SCC with spindle cell components, (C) well-differentiated SCC. Right panels, LOX-PP^Gln mice: (A) adenosquamous ca., (B) SCC with extensive fibrosis, (C) well-differentiated SCC. Bar, 200 µm.

One mouse in the LOX-PP^Gln group was found dead and thus not included in this analysis. While two of the mice in the WT group did not develop macroscopic tumors, both mice did have microscopic lesions. Therefore, similarly to the accelerated tumor initiation and reduced OS, LOX-PP^Gln mice exhibited an average of 9.6 lesions/#4 bilateral glands versus 6.9 lesions/#4 bilateral glands for WT mice (Figure 4C), again suggesting the loss of LOX-PP tumor-suppressing activity in the KI mice.

To test for carcinomas (either primary or metastatic) in lungs and livers, sections of all of the livers and lungs were stained with H&E and analyzed by light microscopy. This analysis revealed no primary tumor or metastasis in either organ from the two groups of mice. However, prominent ‘lymphocyte’ infiltration was observed in 61% of the livers of LOX-PP^Gln mice compared with 29% of WT mice. These ‘lymphoid’ infiltrates presented as clusters of densely packed nuclei and individual cells and revealed infiltration of mononuclear cells in the liver parenchyma (Figure 5A and B). In addition, massive, all organ encompassing lymphoid infiltrations in both livers and lungs were found in 20.8% LOX-PP^Gln mice and in 8.3% WT mice. Mice with massive lung infiltration (Figure 5C) exhibited labored breathing. Two of the affected LOX-PP^Gln mice had enlarged livers and spleens, while two others had an enlarged thymus. Only one of the WT mice had an enlarged liver. Unfortunately, while no heparin blood samples were collected, a clear diagnosis in regard to leukemia/lymphoma was not possible at this point. Leukocytes stained positive for Ki67 suggesting that they were actively proliferating and quantification of KI67 positive cells clearly demonstrated more proliferation in LOX-PP^Gln mice (Figure 5D).

Even though untreated LOX-PP^Gln mice were healthy, more DMBA-treated LOX-PP^Gln animals were moribund compared with DMBA-treated WT mice. Two mice in the LOX-PP^Gln group developed severe dermatitis, and five showed weight loss and had difficulties breathing (29%) compared with two mice (8.3%) in the WT group. Macroscopically, these mice had at least one organ—liver, spleen or thymus—enlarged. Microscopically, these correlated with increased infiltration and proliferation of mononuclear cells in the lung and liver parenchyma. Collectively, these data suggest that LOX-PP^Gln has impaired tumor-suppressing activity in vivo in this carcinogen-treated mouse model.

LOX activity is comparable in tumors from both mouse lines, while expression levels are varied

To determine the status of LOX in tumor tissues, we performed immunohistochemistry on tumor tissue sections with a polyclonal antibody directed against LOX-PP. Immunoreactivity was observed in tumor tissues in the tumor stroma, in tumor cells, and overall expression was mostly comparable in LOX-PP^Gln and WT mice. However, lower staining was observed in adeno components of the LOX-PP^Gln mice (Figure 6A, LOX-PP^Gln upper panel). Of note, the cellular layer adjacent to the keratin pearls was strongly positive in differentiated SCC (Figure 6A, WT lower panel and panel 2). This was observed also for SCC in LOX-PP^Gln mice when this structure was present (data not shown). Panel 5 on the right (Figure 6A) shows strong staining of the fibrotic tissue within the tumor.

The LOX enzyme has been shown to act as a tumor promoter; thus, we determined the enzymatic activity in tumor tissues of LOX-PP^Gln and WT mice. Amplex red assays revealed that the enzyme activity in tumors from LOX-PP^Gln mice was not significantly changed from tumors isolated from WT mice (Figure 6B), suggesting that elevated LOX enzyme activity is not driving earlier tumor development in LOX-PP^Gln mice.

Discussion

The present study shows that KI mice expressing a variant Pro-LOX (LOX-PP^Gln) that has an Arg→Gln substitution in the propeptide domain are more susceptible to DMBA-induced carcinogenesis. These mice had enhanced formation of mammary tumors as well as prominent infiltration of mononuclear cells in the lungs and livers. Specifically, DMBA-induced tumor development was faster, and OS was shorter in the LOX-PP^Gln mice, and tumor weights and volumes were greater compared with the DMBA-treated WT control mice group. This accelerated tumor development appears to be independent of the enzymatic activity of LOX. These findings show for the first time the role that this Aa polymorphism plays in tumorigenesis in vivo. Importantly, these data are consistent with the observed higher susceptibility to cancer in patients resulting from the SNP rs1800449. While mice expressing the LOX-PP^Gln variant exhibited no mammary gland developmental abnormalities, they were more prone to tumor development. These findings suggest the LOX-PP^Gln mouse model will permit future longitudinal and mechanistic in vivo studies into the role of the Arg→Gln substitution in the tumor suppressor LOX-PP in cancer.

Of note, a large fraction of the murine tumors detected is classified as MBCs with predominance of squamous cell carcinomas. This is consistent with earlier findings by Currier et al. (36), who examined oncogenic signaling pathways activated in DMBA-induced tumors in WT FVB mice. Squamous and adenosquamous carcinomas were the most common histologies identified in their model. However, it is in sharp contrast to breast cancers occurring in humans, where adenocarcinomas are the most common histologies (2,3), and MBC is rare, with an estimated cases of >1% in the USA (7). Of note, several case reports for SCC of the breast have more recently been described (39–41) The discrepancy between the human breast pathologies and the high frequency of DMBA-induced SCC might depend on the use of DMBA for tumor induction in our model. On the other hand, introduction of human breast cancer-causing genes in transgenic mice or inactivation of the murine homologues of human tumor suppressor genes induced tumors with striking similar histopathologies (42), which supports the use of murine models for breast cancer studies. Hence, the induction of squamous cell carcinomas by DMBA exposure of FVB mice in different studies indicates that this carcinogen treatment might provide a useful animal model for the study of rare metaplastic breast cancer.

Previous in vitro studies have shown that LOX-PP has tumor suppressor activities. Specifically, LOX-PP suppresses the Ras signaling pathway as we have shown for ras-transformed fibroblasts and also for carcinoma cells (24,26,43,44). LOX-PP reverses the invasive phenotype of cells in culture, attenuates fibronectin-stimulated integrin signaling and migration and inhibits growth and promotes apoptosis of pre-existing breast cancer xenografts (24,28,45). In contrast, ectopic expression of the LOX-PP^Gln, in cells in culture and in a xenograft model, has impaired ability to inhibit the transformed phenotype. Importantly, a more aggressive breast cancer cell phenotype as measured by increased phospho Erk1/2, vimentin and invasive colonies in Matrigel was induced by expression of LOX-PP^Gln in the context of the proenzyme Pro-LOX (25). The earlier onset of tumors in this LOX-PP^Gln KI mouse model supports the notion that LOX-PP^Gln plays a role in tumor initiation/onset of tumor development. To more precisely identify its role in this process, future studies should aim at analyzing early stages in tumor development. While the published in vitro findings suggest that the tumor suppressor activity of LOX-PP may counteract the tumor-promoting activity of the LOX enzyme, it is not clear whether the earlier tumor onset in the LOX-PP^Gln mouse model is due only to the loss of the tumor suppressor function or whether tumor-promoting activities may also be involved. Nonetheless, in our studies, we did not find evidence that the activity of LOX enzyme is elevated in tumors from LOX-PP^Gln animals.

Pro-LOX is secreted as a ~50 kDa proenzyme from which the ~18 kDa LOX-PP is released after processing by procollagen C-proteinases. LOX-PP is produced predominantly by activated stromal fibroblasts and secreted into the tumor stroma. Excessive activation of fibroblasts occurs in a variety of pathologies including fibrosis and cancer through inflammatory and/or fibrogenic signals such as TGFβ, which promotes fibrosis and tumor progression. Our data that KI mice display a median ~3.5 weeks earlier tumor onset compared with DMBA-treated WT control mice may suggest that LOX-PP secreted from stromal-derived fibroblasts plays a role in the prevention of tumor initiation. It is likely that the tumor suppressor function exerted by LOX-PP WT delays tumor onset and that this control mechanism is lost in LOX-PP^Gln-expressing mice. This is of significance because tumor epithelium induced changes in the tumor stroma to promote tumor development opens a window for the use of the rs1800049 SNP as diagnostic marker in those tumors where an association has been established. This concept of tumor interdependence with stroma in tumor progression is increasingly appreciated (46,47). We demonstrated with this mouse model a loss of the tumor suppressor activity of LOX-PP^Glnin vivo. Together with the data from the human studies that associated this SNP with disease progression of various carcinomas makes a strong argument that this SNP qualifies as a prognostic marker. Moreover, the Aa exchange introduced by the SNP replaces a positively charged Aa (form often salt bridges) with a polar Aa Gln (involved in hydrogen bonds formation) (24). Data presented here suggest that this substitution significantly alters the properties of LOX-PP^Gln. It will be important to identify how this Arg in LOX-PP contributes to its tumor suppressor function.

Conclusion

By generating a KI mouse model for a naturally occurring LOX-PP variant LOX-PP^Gln and analyzing carcinogen-induced tumor development in these mice, we were able to demonstrate that the tumor suppressor function of WT LOX-PP is compromised in this variant protein. The accelerated tumor development and enhanced tumor weights indicate more aggressive tumor growth in mice expressing the LOX-PP^Gln variant. In humans, the SNP rs1800449 responsible for the Arg→Gln substitution has been shown to be associated with increased cancer incidence in several solid cancers. Therefore, these studies have further established this SNP and the resulting Arg→Gln substitution as potential biomarkers. Our findings indicate that the Arg→Gln substitution in LOX-PP could be an important marker associated with a more aggressive cancer phenotype and that this KI model is ideal for further mechanistic studies regarding the tumor suppressor function of LOX-PP. Overall, these data suggest an important role of the stroma-derived LOX-PP^Gln in the biogenesis of solid cancers.

Supplementary material

Supplementary materials can be found at Carcinogenesis online.

Funding

These studies were supported by National Institute of Health (NIH) grant CA143108.

Conflict of Interest Statement: None declared.

Abbreviations

Aa

amino acid

Arg

arginine

DMBA

7, 12-dimethylbenz(a)anthracene

Dox

doxycycline

Gln

glutamine

H&E

hematoxylin and eosin

KI

knock in

LOX

lysyl oxidase

LOX-PP

lysyl oxidase propeptide

MBCs

metaplastic breast carcinomas

OS

overall survival

Pro-LOX

pro-lysyl oxidase

SNP

single nucleotide polymorphism

WT

wild-type