Chronic intermittent hypoxia enhances disease progression in myeloma-resistant mice

Abstract

Obesity is the only known modifiable risk factor for multiple myeloma (MM), an incurable cancer of bone marrow plasma cells. The mechanism linking the two is unknown. Obesity is associated with an increased risk of sleep apnea, which results in chronic intermittent hypoxia (CIH), and drives solid tumor aggressiveness. Given the link between CIH and solid tumor progression, we tested the hypothesis that CIH drives the proliferation of MM cells in culture and their engraftment and progression in vivo. Malignant mouse 5TGM1 cells were cultured in CIH, static hypoxia, or normoxia as a control in custom, gas-permeable plates. Typically MM-resistant C57BL/6J mice were exposed to 10 h/day CIH (AHI = 12/h), static hypoxia, or normoxia for 7 days, followed by injection with 5TGM1 cells and an additional 28 days of exposure. CIH and static hypoxia slowed the growth of 5TGM1 cells in culture. CIH-exposed mice developed significantly more MM than controls (67 vs. 12%, P = 0.005), evidenced by hindlimb paralysis, gammopathy, bone lesions, and bone tumor formation. Static hypoxia was not a significant driver of MM progression and did not reduce survival (P = 0.117). Interestingly, 5TGM1 cells preferentially engrafted in the bone marrow and promoted terminal disease in CIH mice, despite a lower tumor burden, compared with the positive controls. These first experiments in the context of hematological cancer demonstrate that CIH promotes MM through mechanisms distinct from solid tumors and that sleep apnea may be a targetable risk factor in patients with or at risk for blood cancer.

INTRODUCTION

Multiple myeloma (MM) is an incurable malignancy characterized by clonal proliferation of plasma cells in the bone marrow, bone destruction, hypercalcemia, anemia, and renal failure (49). Multiple myeloma is preceded by an asymptomatic, premalignant phase termed monoclonal gammopathy of undetermined significance (MGUS) that progresses to MM at a rate of 1% per year (29, 47). Interestingly, many of the major oncogene mutations identified in MM are also present in MGUS (34), suggesting that additional factors are important in driving malignant progression.

Unlike many mouse models of other hematological malignancies (7, 32, 63), models of MM using patient-derived genetic mutations have been largely insufficient to induce MM in mice. For example, activating mutations in Ras oncogenes is the most common single nucleotide variant detected in MM (17), yet mice expressing an activated KRas allele (KRasG12D) in germinal center cells fail to exhibit any features of MM, even when crossed with tumor-prone mice (40). Relevant to this study, the KaLwRij mouse develops a MGUS-like phenotype with age that spontaneously develops MM at a rate similar to that in humans (0.5% or mice older than 2 yr) (48). A malignant plasma cell line derived from the KaLwRij mouse (5TGM1) faithfully engrafts and causes bone lesions, gammopathy, and terminal paralysis. Although these malignant cells recapitulate many features of MM in syngeneic KaLwRij mice, they do not engraft in the closely related C57BL/6J strain of mice. These data support the model that the host’s bone marrow microenvironment is important to disease progression in MM.

Obesity is the only modifiable risk factor associated with an increased risk of MM (12, 30, 57), although the mechanism by which obesity promotes MM is unclear. In C57BL/6J mice, a high-fat diet enhances bone loss caused by 5TGM1 cells, but these animals do not appear to develop terminal disease like the KaLwRij mouse. Because nearly 45% of obese individuals also experience sleep apnea, we propose that the link between MM and obesity may be mediated by sleep apnea (59). The chronic intermittent hypoxia (CIH) that occurs with sleep apnea has emerged as a potent stimulator of solid tumorigenesis (1, 35, 41, 43), and individuals with sleep apnea are more likely to die from cancer. To date, the role of CIH in promoting blood cancer development has not been explored. Here, we tested the hypothesis that CIH promotes the proliferation of malignant 5TGM1 MM cells in culture and would allow them to engraft and proliferate in MM-resistant C57BL/6J mice.

METHODS

Experiment 1: Impact of CIH on 5TGM1 Growth in Culture

Overall design.

Two CIH profiles were tested in 5TGM1 MM cells. Profile 1 has been reported previously (10, 64) and cycled between 21% and 1.5% oxygen. Profile 2 mirrored the in vivo profile, cycling between 21% and 10% oxygen (37, 38). Cultures were exposed to CIH for 10 h/day (12 cycles/h) between 8 AM and 6 PM, followed by normoxia for the remainder of the day. Gases were balanced with 5% CO₂ to maintain culture pH and Pco₂. Cells were grown in custom Plexiglas exposure chambers, housed within a standard incubator (Coy Laboratories). Oxygen concentration within the chambers was monitored using a heated zirconium sensor. At 12 cycles/h, each CIH cycle lasted 5 min. At the beginning of the cycle, nitrogen was introduced into the chamber at a rate sufficient to lower the oxygen to the target within 30 s. This oxygen level was maintained for an additional 2 min. Oxygen was then introduced at a rate sufficient to increase the oxygen concentration to 21% within 30 s and maintained at this level for 2 min. As a control for the effect of hypoxia, cells were exposed to static 10% or 1.5% oxygen for 10 h/day. Importantly, cell culture experiments were performed in custom, gas-permeable plates. These plates had been previously validated extensively, and it has been shown that the Po₂ of the medium closely mirrors the chamber environment (44). For each CIH or static hypoxia condition, a control plate of cells was simultaneously grown in normoxia. Cells were counted daily in triplicate for 4 days, and viability was evaluated via trypan blue exclusion (56).

5TGM1 cell culture conditions.

5TGM1 murine myeloma cells were originally derived from a spontaneously sick, KaLwRij mouse (21, 48, 61) and were engineered to stably express green fluorescent protein (GFP). Cells were thawed and cultured in RPMI 1640 medium (ATCC modification: GIBCO, Grand Island, NY) supplemented with 10% fetal bovine serum (HyClone Laboratories, Logan, UT) and 1% penicillin-streptomycin solution (Mediatech, Manassas, VA). To determine the optimal cell density for cell growth experiments, 5×10⁶, 1×10⁶, 5×10⁵, and 1×10⁴ cells/ml were cultured in clear plastic plates at 37°C and 5% CO₂. Cells were counted daily for 3 days, and concentrations were noted in triplicate. Logarithmic growth over 3 days occurred with 5×10⁵ cells/ml, and this starting concentration was used in all experiments.

Experiment 2. Impact of CIH on 5TGM1 Engraftment

Overall design.

C57BL/6J mice (8-wk-old females; Jackson Laboratory, Bar Harbor, ME) and C57BL/KaLwRij mice (8-wk-old females) were housed with food and water available ad libitum and on a 12:12-h light-dark cycle. The University of Iowa Institutional Animal Care and Use Committee approved all studies.

An illustration of the experimental design is given in Fig. 1. Mice were assigned to one of three gas exposure groups: normoxia (20.9% oxygen, control), CIH, or static hypoxia (constant 10% oxygen) as a control for the effects of hypoxia. The control groups were group-housed in an exposure chamber next to the CIH group in the same room and were exposed to the same light and sound stimuli, including the clicking sound produced by the CIH chamber when the gas flow switched every 2.5 min. All mice were group-housed ≤5 mice per cage with ≤two cages per chamber. Groups were preconditioned to their assigned gas condition for 7 days, followed by intravenous injection of 1×10⁶ 5TGM1 cells or sterile saline as a negative control via a tail vein, and exposure for an additional 4 wk. Experiments with 5TGM1 cells, and normoxia with saline injection, were performed in duplicate (n = 5–6 animals per occasion), and these animals were included in statistical analyses. Given that it is theoretically possible that death could be the result of other external variables (e.g., unanticipated changes in chamber ventilation, infection, or other unknown factors) an additional group of mice were exposed to static hypoxia or CIH and injected with saline as sentinels. These animals were not included in statistical analyses of effect.

Fig. 1.

Timeline of survival experiment. Mice were preconditioned for 1 wk to 1 of 3 conditions [chronic intermittent hypoxia (CIH), static hypoxia, and normoxia]. After preconditioning, mice were injected with either 5TGM1 cells or saline, followed by 4 wk more of exposure. Mice were then returned to normal housing, where they were observed daily for signs of paralysis. Paralyzed mice were euthanized, and tissues were collected for examination.

Therefore, mice were distributed among groups as follows: Normoxia with 5TGM1 (n = 11), Normoxia with saline (n = 11), CIH with 5TGM1(n = 11), CIH with saline (n = 5), static hypoxia with 5TGM1 (n = 11), and static hypoxia with saline (n = 4).

After their return to normal housing, mice were monitored daily for hindlimb paralysis as a sign of myeloma development. If paralysis was observed, the mouse was euthanized, and blood, femurs and tibiae, spleen, and tail and spine were collected for flow cytometry, histology, and Micro-computed tomography (micro-CT) to confirm elevated plasma immunoglobulin levels, 5TGM1 cell engraftment, and bone damage.

Because the 5TGM1 cell line faithfully causes terminal hindlimb paralysis and spinal deformity within 50 days in KaLwRij mice, female KaLwRij mice from two litters were injected with 1×10⁶ 5TGM1 cells (n = 10) and maintained in normal housing, as a positive control.

In vivo chronic intermittent hypoxia.

Cages of female C57BL/6J mice (8 wk old) were placed into one of two Plexiglas exposure chambers (Coy Laboratories) and exposed to CIH for 10 h during the light cycle or normoxia (20.9% oxygen) as a control with water and a standard chow diet available ad libitum. Room temperature in our animal facility is maintained near 72°F. Exposures were scheduled so that the normoxic and CIH chambers were run concurrently, and chambers were arranged side by side so that mice were exposed to the same light and sound stimuli. Mice were randomly assigned to an exposure group and then group-housed within the chamber and fed a standard chow diet. Male mice were not housed in the cubicle containing the chambers.

Minimum air flow through both chambers was 2.5 l/min to maintain the CO₂ within the chambers at <0.5%. The CIH profile used here was previously reported in detail (36–38) and consisted of a 5-min cycle. In the first minute, nitrogen was introduced to lower the chamber oxygen level, which was then held for 90 s. This was followed by a 1-min reoxygenation. The chamber was then held at 21% oxygen for the remainder of the 5-min period. Oxygen levels were continuously monitored with a heated zirconium sensor (Coy Laboratories) and calibrated regularly with nitrogen and room air. Gas tensions were also verified routinely with a second sensor high-resolution gas analyzer that is sensitive ± 0.1% (Gemini, CWE). Mice were exposed to 10 h of CIH per day (10% oxygen, nadir) during the light cycle (8 AM to 6 PM). As we have described, our CIH paradigm produces a saturation profile that mimics moderately severe obstructive sleep apnea in humans (15 events/h, nadir $S{p}_{O_2}$ = 75%) (20, 37, 38). As a control for the effects of hypoxia, an additional cohort was exposed to 10 h/day constant, static hypoxia (10% oxygen) during the light cycle. Thus, our experiment included three exposure groups: normoxia, CIH, and static hypoxia.

Flow cytometry and ELISA.

Bone marrow tumor formation is a defining feature of MM. To verify that 5TGM1 cells engrafted in the bone marrow of CIH-exposed animals, we quantified spleen and bone marrow and compared this to 5TGM1 engraftment in the 5TGM1-injected KaLwRij mouse as a positive control. One femur, tibia, and half of the spleen of each euthanized mouse were used for flow cytometry. The bone marrow was flushed from the femur and tibia with PBS. The spleen was minced between two cover slides. Cells from bone marrow and spleen in phosphate-buffered saline (PBS) were passed through a disposable 40-µm cell strainer, centrifuged, resuspended in 10 ml of ammonium-chloride-potassium (ACK) lysing buffer (MK Medical, Columbia, MD), centrifuged, and then washed with PBS. Flow cytometry was performed in the University of Iowa Flow Cytometry Core, and the number of GFP⁺ cells was quantified (Becton-Dickinson LSR II; BD Biosciences, San Jose, CA). IgG2b was measured in serum with a commercial ELISA kit (Bethyl Laboratories, Montgomery, TX) with six-point standard curves run on each plate.

Histology.

Bone marrow tumors were further confirmed histologically. The other femur and tibia were decalcified in EDTA for 2 wk. The bones and the other half of the spleen were fixed and embedded in paraffin. Sections were stained with hematoxylin and eosin (H&E) by the University of Iowa Comparative Pathology Laboratory. Images were obtained at ×20 magnification and reviewed by two investigators blinded to the experimental conditions, including a board-certified pathologist (C. Galambos), with 100% agreement in their qualitative assessment of tumor presence.

Three-dimensional X-ray.

Bone damage is another defining feature of MM. The tail and spine of paralyzed chronic intermittent hypoxia (CIH) mice were fixed in 4% formalin and imaged using a submicron three-dimensional (3-D) X-ray microscope in the University of Iowa Small Animal Imaging Core (Zeiss Xradia 520 Versa). The samples were scanned at 70 kV/6 W, with a full 360° rotation over 1,601 projections with 1-s exposure time. The projections were reconstructed into a single 3-D image (38 µm pixel size, ORS Visual software) These images were then converted to 3-D mesh [ImageJ software and BoneJ plugin (19)] for quantification of bone lesions (55) as well as to create the local thickness map and calculate trabecular bone thickness using 3-D ROI manager. Images were blindly reviewed by independent investigators who quantified the number of bone lesions (D. Hathi and M. Shokeen). Both investigators were in 100% agreement on the features of a lesion (morphology, number, texture). Once this was defined, the lesions were identified in ImageJ and quantified.

Statistical analysis.

Data are expressed as means ± SE. Kaplan-Meier’s curves were used to compare overall survival, and log-rank analysis and odds ratio were further used to test for significant between-group differences in survival, with a Bonferroni comparison for repeated comparisons. For the cell culture experiment, cell growth was compared by quantifying the area under the growth curve, and comparisons were made by analysis of variance (ANOVA). Dunnet’s test was used for post hoc comparisons to the normoxic control. For all other parameters, a normal distribution was confirmed, and groups were compared with Student’s t-test (GraphPad Prism). Significance was determined a priori at P < 0.05.

RESULTS

CIH Slows Growth and Proliferation of 5TGM1 Cells In Vitro

As expected, 5TGM1 cells grown in normoxia doubled in number within 48 h with >90% cell viability during this period (Fig. 2). Exposure to CIH or static hypoxia did not promote cell growth further. In fact, cell growth in CIH (nadir 10%) and static hypoxia (10%) was reduced over the 4-day period (AUC, P = 0.014, and P = 0.023). More extreme CIH (nadir 1.5%) decreased cell viability (P = 0.006) and cell growth (P = 0.006).

Fig. 2.

Chronic intermittent hypoxia (CIH) and static hypoxia do not promote the growth of 5TGM1 multiple myeloma cells in vitro. Top: cells were exposed to each condition for 10 h/day and CIH cycled at 12 cycles/h (n = 3 replicates per condition). Growth of 5TGM1 cells was depressed over 4 days in 2 profiles of CIH and static hypoxia. Bottom: cell growth, calculated as area under the curve over 4 days, is depressed by CIH and static hypoxia relative to control (***P < 0.05).

CIH Promotes Lethal Disease in Typically MM-Resistant Mice

Survival was generally not impacted by saline injection, CIH, or static hypoxia, with the exception of one mouse found dead of unclear causes in the normoxia group (Fig. 3). In the positive control group, 100% of KaLwRij mice injected with 5TGM1 cells (n = 10) developed terminal paralysis, demonstrating that our 5TGM1 MM cells were viable and pathogenic.

Fig. 3.

Mice exposed to chronic intermittent hypoxia (CIH) experience terminal paralysis following 5TGM1 cell injection. Top: Kaplan-Meier curve indicating that the CIH-exposed group experienced the lowest survival among C57BL/6J mice. KaLwRij mice are a positive control, receptive to 5TGM1 cells. Bottom: odds ratio for developing terminal paralysis following 5TGM1 cell injection is significantly higher in CIH-exposed mice vs. controls and static hypoxia-exposed mice.

As expected (33), C57BL/6J mice injected with 5TGM1 MM cells (n = 8) and maintained in normoxia were resistant to MM development, with only one mouse developing terminal paralysis (1 of 11, 9%). When compared with normoxia-exposed mice, the CIH-exposed C57BL/6J mice were particularly vulnerable to 5TGM1 MM cells, with 67% of mice developing terminal paralysis (odds ratio 17.5, P = 0.005 vs. normoxia control with 5TGM1). A smaller and nonsignificant subset of mice exposed to the static hypoxia profile developed terminal disease (37%) (odds ratio 5.7, P = 0.117 vs. normoxia control with 5TGM1). A power analysis revealed that an additional 22 animals would be required to observe a statistically significant difference between normoxia and static hypoxia (β = 0.05).

Myeloma Cells Engraft Preferentially in Bone Marrow of CIH-Exposed Mice

Our 5TGM1 cell line expresses green fluorescent protein (GFP⁺), which allows the detection and quantification of the myeloma cell burden in both the bone marrow and spleen of mice with terminal paralysis. In KaLwRij mice, the spleen was visibly enlarged at necropsy, and 5TGM1 cells were highly abundant in the bone marrow (42 ± 7% of total cells) and spleen (52 ± 7% of total cells) by flow cytometry (Fig. 4). The distribution of 5TGM1 cells between the two tissues was similar (bone marrow: spleen ratio = 0.85 ± 0.07). Although the CIH exposed mice developed the same severity of hindlimb paralysis as KaLwRij, the bone marrow tumor burden was significantly lower (5 ± 2% of total cells). Interestingly, 5TGM1 cells were six times more likely to engraft in the bone marrow than spleen of CIH mice (bone marrow/spleen ratio = 6.4 ± 3.0, P = 0.014 vs. KaLwRij). Serum IgG2b, a monoclonal tumor marker secreted by 5TGM1 cells, was also elevated to a lesser degree in paralyzed CIH mice (P < 0.001 vs. KaLwRij; Fig. 4). The bone marrow sections stained with H&E were examined histologically and showed monotonous hyperchromatic malignant plasma cells infiltrating the bone cavity and decreased normal hematopoietic cell populations and fat cells in both KaLwRij and paralyzed CIH mice (Fig. 5). In the KaLwRij mice, the tumor nearly filled the bone marrow cavity. As further evidence of an overall lower tumor burden, tumor populations were patchy in the CIH mice (Fig. 5).

Fig. 4.

Chronic intermittent hypoxia (CIH)-exposed mice have reduced tumor burden but more bone marrow tumor cell engraftment. A: representative flow cytometry showing green fluorescent protein (GFP)⁺ 5TGM1 cells in bone marrow and spleen of a paralyzed KalwRij mouse, which serves as a positive control for 5TGM1 engraftment. B: representative flow cytometry showing GFP⁺ 5TGM1 multiple myeloma (MM) cells in bone marrow and spleen of a paralyzed C57BL/6J mouse exposed to CIH, showing decreased tumor burden. C: tumor cell burden was higher in bone marrow and spleen of KalwRij mice compared with paralyzed, CIH-exposed mice. D: ratio of bone marrow to spleen 5TGM1 cells is ≅1 in KaLwRij mice, indicating equal distributions between these tissues. In paralyzed CIH-exposed mice, 5TGM1 cells preferentially engrafted in bone marrow. E: 5TGM1-specific monoclonal immunoglobulin G (IgG2b) levels were elevated in paralyzed CIH mice compared with reference levels from normal nonimmunized C57BL/6 controls (dotted lines and gray shading represent upper and lower confidence intervals 8 healthy, 8-wk-old, nonimmunized. C57BL/6J mice). Levels of IgG2b in paralyzed CIH mice were significantly lower than in positive control KalwRij mice. **P < 0.05, ***P < 0.01.

Fig. 5.

5TGM1 cell engraftment is associated with bone lesions and osteopenia in chronic intermittent hypoxia (CIH)-exposed mice. A and C: representative three-dimensional bone surface mesh image used for bone lesion counting in normoxia controls (A) and paralyzed CIH-exposed mice (C). B and D: representative bone thickness heat map images of tail and spine in a normoxia control (B) and paralyzed CIH mouse (D), used for bone thickness analysis. E: paralyzed CIH-exposed mice have more lesions per bone, relative to normoxia control. F: CIH-exposed group had decreased bone thickness compared with normoxia control. *P < 0.05.G: hematoxylin and eosin-stained slide showing normal bone marrow of a normoxia control mouse. H: hematoxylin and eosin-stained slide of the bone marrow of a paralyzed CIH-exposed mouse showing malignant plasma cells infiltrating the bone cavity.

Lytic Bone Lesions and Osteopenia in CIH Mice

Vertebral bones from paralyzed, 5TGM1-injected CIH mice were evaluated by micro-CT to further confirm that terminal paralysis was associated with bone damage, a key feature of MM in humans. Paralyzed mice from the CIH group had more lytic bone lesions (P = 0.003) and reduced bone thickness (P = 0.0005) compared with healthy C57BL/6J mice with no 5TGM1 injection (Fig. 5).

DISCUSSION

Using both in vitro and in vivo models, we demonstrate that CIH does not promote MM cell proliferation directly, but promotes disease aggressiveness in vivo. While there is a growing body of evidence in animals and humans demonstrating the link between solid tumor aggressiveness and CIH (5, 8, 9, 13, 16, 23, 25, 26, 31, 50, 65), this is the first work to demonstrate that CIH may be a potentially modifiable contributor in blood cancers as well. We found that CIH causes preferential tumor cell engraftment in the bone marrow, and a lethal disease phenotype, despite a lower overall tumor burden.

In other blood cancers, patient-relevant mutations are sufficient to recapitulate malignant disease in mice (32, 63). An important and longstanding question in MM is, why do patient-relevant, plasma cell mutations fail to convey a malignant phenotype in mice? We have previously categorized several patient-derived genetic mutations that are insufficient to induce MM in mice [KRAS, p53, RB1, etc. (24, 39)]. Indeed, we and others have suggested that there is an important “inflammatory” or “angiogenic” switch that is required for the engraftment of malignant cells (59).

CIH does not promote 5TGM1 cell proliferation in culture, supporting the idea that our observations are the result of CIH altering the bone marrow microenvironment and/or altering the interaction of myeloma cells with the host immune system. Our in vitro data are consistent with cell culture experiments in melanoma and lung epithelial tumor cell lines in which CIH does not induce cell proliferation directly (2). Tumor aggressiveness in these models occurs via indirect effects on tumor-associated macrophages. It is possible that macrophages are also involved in the pathogenesis of our CIH model. In fact, macrophage polarization is a key feature of the KaLwRij bone marrow. These macrophages support B cell proliferation (4). CIH is also a potent stimulator of interleukin (IL)-6 expression, an inflammatory cytokine involved in T cell proliferation, B cell differentiation, and monocyte maturation (53). IL-6 is expressed by MM cells and is involved in the pathogenesis of MM (6, 28, 45). Dechow et al. and Rutsch et al. (18, 51) have demonstrated that the IL-6 signaling pathway can collaborate with the oncogene MYC to induce a MM-like phenotype in mice. It is possible that CIH allowed the engraftment of 5TGM1 cells in our model by creating a proinflammatory bone marrow environment, which we will test in future studies.

CIH also promotes solid tumor growth, in part by promoting angiogenesis (62). Hypoxia plays a major role in angiogenesis through hypoxia-inducible factor (HIF), which is a key transcriptional factor for several genes involved in the angiogenesis process in both normal physiological and neoplastic settings (46, 54). HIF upregulates the transcription of angiogenic growth factors like vascular endothelial growth factor leading to the proliferation, migration, assembly, and lumen acquisition of endothelial cells and increased angiogenesis (58). Particularly relevant to MM, angiogenesis in the bone marrow is associated with disease progression and poor prognosis (22). In the only investigation of the effects of CIH on bone marrow published to date, Alvarez-Martins et al. (3) found that CIH itself causes remodeling of the bone marrow vasculature in otherwise healthy mice. That said, in a breast cancer model, CIH increased tumor aggressiveness not by simply increasing tumor blood supply but by promoting clonal diversity, upregulating metastasis-associated genes, and increasing stem cell-like markers (13). This is an area that warrants further exploration, particularly in the context of blood cancer where these questions have not been explored.

CIH may allow the engraftment of 5TGM1 cells by altering host immunity. Although the C57BL/6J and KaLwRij mouse lines are closely related (4), we cannot rule out that CIH may promote disease in our model by inhibiting T cell-mediated immunity. Although our study establishes the first link between CIH exposure and the development of a hematological malignancy, further studies are needed to fully investigate and understand the underlying mechanisms.

Studies in solid tumors have observed that CIH increases tumor aggressiveness, and metastatic properties (1, 42). Surprisingly, CIH-exposed mice developed the same paralysis phenotype as KaLwRij mice, with one-tenth the bone marrow tumor burden and a tendency for 5TGM1 cells to engraft particularly in bone marrow. Decreased tumor burden is further supported by decreased MM-specific serum IgG2 paraprotein and patchy tumor infiltration observed histologically. In KaLwRij mice, 5TGM1 cells uniformly distributed through the bone marrow and their expansion would translate to uniform stress on the bone. Patchy growth 5TGM1 cells in CIH mice could put focal stress on the bone, causing fracture and paralysis despite an overall lower tumor burden. Indeed, we noted areas of patchy infiltration and bone damage in the CIH-exposed mice. It is also possible that CIH changes the gene expression and malignant potential of the 5TGM1 cells, resulting in a tumor that is particularly osteolytic and damaging to bone. Future experiments are required dissect the impact of CIH on the 5TGM1 cell itself, and to determine whether preconditioning 5TGM1 cells with CIH improves their ability to engraft in the host marrow.

Limitations and Future Directions

In our study, we used a fully transformed tumor cell containing many mutations, as opposed to a single driver mutation model. Future studies should evaluate the impact of CIH in other models, including models with KRAS, p53, and RB1 (24, 39) mutations and Myc overexpression (51).

In our study, 1.5% O₂ (11 mmHg) and 10% O₂ (71 mmHg) blunted cell growth in the 5TGM1 line, supporting our hypothesis that these cells do not grow better in hypoxia. Still, it is important to note that, whereas the KaLwRij/5TGM1 model has been frequently used in the myeloma field, the 5TGM1 cells behave like a cell line in culture; that is to say, they grow optimally in 21% oxygen and less well at oxygen tensions more similar to human bone marrow (mean 52 mmHg) (15, 27). Therefore, we used 21% O₂ as our optimal growth condition and cycled back to this level in CIH experiments to evaluate proliferation. Future experiments of cell signaling and physiological function in human primary cells should be conducted in the range of Po₂ values present in bone marrow (11).

In our survival studies, sample sizes for groups receiving 5TGM1 injections ranged from 8 to 11 mice, depending on litter size. Despite this sample size, the low incidence of terminal disease in the normoxic and static hypoxia groups means that we were unable to make comparisons to the CIH groups. We are, thus, unable to characterize the disease phenotype of the few animals in the normoxia and static hypoxia groups that developed disease. Instead, we report comparisons between CIH and normal C57BL/6J mice and KaLwRij and demonstrate clear differences.

Perspectives and Significance

In this first report of the impact of CIH in a model of hematological malignancy, we demonstrate that CIH promotes the engraftment and expansion of highly malignant plasma cells specifically in the bone marrow of typically resistant mice. Unique to this model, CIH promotes terminal disease, with key features of multiple myeloma, despite a relatively low tumor burden. Because CIH is a model of sleep apnea, we propose that sleep-disordered breathing could be a unique, targetable risk factor in multiple myeloma. If CIH promotes a pro-tumor bone marrow microenvironment generally, it may also affect the development and progression of additional marrow-resident cancers.

GRANTS

This work was supported by Grant IRG-15-176-40 from the American Cancer Society (M. L. Bates), administered through The Holden Comprehensive Cancer Center at The University of Iowa and Holden Comprehensive Cancer Center (P30CA086862).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.A., D.H., M.S., H.X., F.Z., M.H.T., and M.L.B. conceived and designed research; M.A., S.K., D.J.D., C.T., H.X., J.Y., M.H.T., and M.L.B. performed experiments; M.A., S.K., C.G., D.H., S.J., M.S., H.X., J.Y., M.H.T., and M.L.B. analyzed data; M.A., S.K., D.J.D., C.G., D.H., S.J., M.S., H.X., F.Z., M.H.T., and M.L.B. interpreted results of experiments; M.A., D.H., M.S., J.Y., M.H.T., and M.L.B. prepared figures; M.A., M.H.T., and M.L.B. drafted manuscript; M.A., D.J.D., C.G., D.H., S.J., M.S., C.T., H.X., J.Y., F.Z., M.H.T., and M.L.B. edited and revised manuscript; M.A., S.K., D.J.D., C.G., D.H., S.J., M.S., C.T., H.X., J.Y., F.Z., M.H.T., and M.L.B. approved final version of manuscript.