The impact of chronic intermittent hypoxia on hematopoiesis and the bone marrow microenvironment
Obstructive sleep apnea (OSA) is a highly prevalent sleep-related breathing disorder which is associated with patient morbidity and an elevated risk of developing hypertension and cardiovascular diseases. There is ample evidence for the involvement of bone marrow (BM) cells in the pathophysiology of cardiovascular diseases but a connection between OSA and modulation of the BM microenvironment had not been established. Here, we studied how chronic intermittent hypoxia (CIH) affected hematopoiesis and the BM microenvironment, in a rat model of OSA. We show that CIH followed by normoxia increases the bone marrow hypoxic area, increases the number of multipotent hematopoietic progenitors (CFU assay), promotes erythropoiesis, and increases monocyte counts. In the BM microenvironment of CIH-subjected animals, the number of VE-cadherin-expressing blood vessels, particularly sinusoids, increased, accompanied by increased smooth muscle cell coverage, while vWF-positive vessels decreased. Molecularly, we investigated the expression of endothelial cell-derived genes (angiocrine factors) that could explain the cellular phenotypes. Accordingly, we observed an increase in colony-stimulating factor 1, vascular endothelium growth factor, delta-like 4, and angiopoietin-1 expression. Our data shows that CIH induces vascular remodeling in the BM microenvironment, which modulates hematopoiesis, increasing erythropoiesis, and circulating monocytes. Our study reveals for the first time the effect of CIH in hematopoiesis and suggests that hematopoietic changes may occur in OSA patients.
KeywordsChronic intermittent hypoxia Hematopoiesis Vascular niche Bone marrow microenvironment
Hematopoiesis has long been known to be affected by environmental hypoxia . Despite numerous reports relating hypoxia with hematopoietic modulation [3, 5, 44, 46, 48, 49, 61, 64] and the great attention currently given to hypoxia-inducible factor (HIF) , the current scientific knowledge largely relies on of studies performed under acute hypoxia stimulation in isolated systems. Therefore, the role of chronic systemic hypoxia in the bone marrow (BM) microenvironment and hematopoiesis is still unknown.
Here, we studied the role of environmental hypoxia using a clinically relevant chronic intermittent hypoxia (CIH) model, which consists of exposing the experimental animals to a paradigm of CIH for 30–35 days, as a model of obstructive sleep apnea syndrome (OSA) . OSA is an increasingly prevalent condition affecting children and adults, which is renowned as a frequent secondary cause of hypertension . There is ample evidence for the involvement of BM-derived cells in the pathophysiology of hypertension  but a relation between OSA and modulation of the BM microenvironment had not been shown and is the subject of the present study. The clinical hematological aspects of OSA are still largely unknown, with several studies reporting mainly platelet activation and increased hematocrit, but not assessing BM or circulating blood cell alterations . Furthermore, these studies typically compared groups of patients with differing disease phenotypes  or treated versus untreated patients  but not healthy controls versus patients and do not explore the mechanisms involved .
Mechanistically, we focused on the vascular compartment of the BM, because it mediates the differentiation and proliferation of hematopoietic cells, as well as their egress from the BM microenvironment . Furthermore, one of the most striking effects of hypoxia is the promotion of angiogenesis, which results from hypoxia inducible factor (HIF)-mediated increase in vascular endothelial growth factor (VEGF) expression .
The present work aims to explore for the first time whether CIH induce changes in the BM vascular compartment, which might in turn modulate hematopoiesis.
Our data suggests that CIH may promote erythropoiesis, increase the multipotential progenitor cell-derived CFUs accompanied by an increase in BM myeloid and B lymphocyte counts and a decrease BM T cells. Additionally, CIH also induces expansion of the blood monocyte compartment and perturbs the BM microenvironment by interfering with the vascular niche. Together, our results reveal hematopoietic and hematological complications of CIH which need to be validated and evaluated in a clinical setting.
Experiments were performed in twelve (12) male Wistar rats, aged 8–12 weeks, obtained from the NOVA Medical School animal facility. Animals were housed in polycarbonate cages, under 12-h light/dark cycles (8 am–8 pm) at a room temperature 22 ± 2.0 °C and relative humidity 60 ± 10 %. Rats were maintained one or two per cage with ad libitum access to food and water. Applicable institutional and governmental regulations concerning ethical use of animals were followed, according to the NIH Principles of Laboratory Animal Care (NIH Publication 85–23, revised 1985), the European guidelines for the protection of animals used for scientific purposes (European Union Directive 2010/63/EU), and the Portuguese Law n° 113/2013. Experimental procedures were previously approved (nr. 21/2013/CEFCM) by the Institutional Ethics Committee of the NOVA Medical School for animal care and use in research.
In vivo experiments
Rats were divided into two groups: normoxia and CIH. Animals were kept in a eucapnic atmosphere, inside of medium A-chambers (76 × 51 × 51 cm, A-60274-P, Biospherix Ltd, NY, USA) with ad libitum food and water access. The chambers were equipped with gas injectors and sensors for oxygen (O2) and carbon dioxide (CO2) levels in order to ensure the accuracy of CIH cycles. Accumulation of CO2 was prevented by the continuous flow of the gas mixtures through vent holes and the presence in the chamber of self-indicating soda lime, which absorbs the expired CO2. The CO2 levels inside the chambers never exceeded 1 %. A silica gel container was also placed inside the chambers in order to absorb water. Oxygen concentration inside the chambers was controlled using 100 % nitrogen (N2) and 100 % O2 by an electronically regulated solenoid switches in a three-channel gas mixer, which gradually lowered oxygen in the chamber from 21 to 5 % O2 (OxyCycler AT series, Biospherix Ltd, NY, USA). The chambers were infused with 100 % N2 for 3.5 min to briefly reduce the O2 concentration to 5 % and then with 100 % O2 for 7 min to restore oxygen to ambient levels of 21 %, until the start of the next CIH cycle. Each CIH cycle lasted 10.5 min, and rats were exposed during their sleep period (light phase of light/dark cycle) to 5.6 CIH cycles/h, 10.5 h/day for 32 days and analyzed 3 days after the hypoxic period. During the remaining hours of the day, the chambers were ventilated with a constant flow of room air to keep oxygen levels at 21 %. O2 was purchased as regular gas bottles (Gasin, Portugal), while N2 was generated from the air by pressure swing adsorption technology using a high-output nitrogen generator (Nitrogen 15 Plus, PSA Technology, Sysadvance, Maia, Portugal).
After exposing rats to 32 days of hypoxia followed by 3 days in normoxia, rats were sacrificed by intraperitoneal injection with medetomidine (0.5 mg/kg body weight, Domitor®, Pfizer Animal Health) and ketamine (75 mg/kg body weight, Imalgene 1000®, Mérial, Lyon, France), and cardiac puncture was performed to collect peripheral blood. Blood was collected in EDTA-coated tubes (Multivette 600, Sarstedt), and plasma sampling and complete blood counts were performed. Femur BM cells were flushed out with PBS 2 % FBS, and the total BM cell count was assessed using a Burker hemocytometer (Blau Brand).
Hypoxia quantification in bone marrow sections
Three days after the hypoxic period, the subjects were intravenously injected (via tail vein) with 60-mg/kg pimonidazole hydrochloride (Hypoxyprobe, Inc, Burlington, USA), a misonidazole-based compound, which forms adducts with thiol groups of proteins, peptides, and amino acids specifically in hypoxic cells (pO2 < 10 mmHg) (81). Two hours later, rats were euthanized using 60-mg/kg sodium pentobarbital IV (Eutasil, Ceva Santé Animale, Libourne, France) and transcardially perfused with PBS. Quantification of BM hypoxic areas was performed using the ImageJ software in 10 high-power fields (×400 magnification) per animal.
Bone marrow and peripheral blood cells were treated with red blood cell lysis buffer (Biolegend) for 15 min in the dark and were then stained for anti-CD90 (HIS51) fluorescein isothiocyanate (FITC), anti-CD11b (WT.5) allophycocyanin (APC) (BD Biosciences), anti-CD117 (2B8) APC, anti-CD19 (1D3) PE-cyanine 7 (PE/Cy7) (eBiosciences), and anti-CD3 (17A2) APC-cyanine 7 (APC/Cy7) (BioLegend). Flow cytometric analyses were carried out using an LSR Fortessa flow cytometer equipped with FACS Diva 6.2 Software (BD Biosciences). Data were analyzed with a FlowJo 9.8.2 software.
In vitro colony forming assay
Femur BM cells were flushed with PBS 2 mM EDTA, treated with red blood cell lysis buffer (Biolegend) for 15 min in the dark and plated onto petri dishes for 2 h. Non-adherent cells (105 cells) were collected and plated onto a semi-solid cytokine-supplemented methylcellulose medium (MethoCult GF M3434) (Stemcell Technologies). Each colony formed in this semi-solid medium is single-cell derived and represents the identity of the original progenitor cell [6, 15]. The resulting colonies were scored after 1–2 weeks of culture, according to manufacturer’s instructions.
Immunostaining and imaging
Primary antibodies used for immunohistochemistry
HIER, Tris-EDTA pH 9
HIER, Tris-EDTA pH 9
HIER, Tris-EDTA pH 9
RNA isolation and quantitative PCR
List of primers used for RT-PCR
Forward primer (5′-3′)
Reverse primer (5′-3′)
Results are expressed as mean ± standard deviation. Data were analyzed using unpaired two-tailed student’s t test. P values of <0.05 were considered statistically significant.
Chronic intermittent hypoxia does not affect BM cell number but modulates/perturbs hematopoiesis
Our data thus mimic some of the clinical aspects observed in OSA patients, as we observed no changes in platelet counts, but fails to reproduce other symptoms, such as the increase in platelet activation and aggregation, as assessed by mean platelet volume. Moreover, we report for the first time an increase in the myeloid compartment, both the BM and PB and a modulation in the percentage of B and T lymphocytes in the BM of animals exposed to CIH.
Chronic intermittent hypoxia affects the bone marrow vasculature and modulates monocyte counts
To demonstrate that there was undoubtedly an increased expansion of myeloid cells within the BM, we assessed the BM CD11b+ cell (monocyte) content. In accordance with the flow cytometric and complete blood count data, we observed an increment in the number of BM monocytes in CIH treated animals (Fig. 4e’, e”).
Together, these data show the BM vasculature and in particular the VE-cadherin and vWF-expressing vessels, and the pericyte/smooth muscle cell vessel coverage are affected by CIH exposure. This morphological change in BM vessels of CIH-treated animals is accompanied by a significant increase in the number of CD11b + monocytes.
Chronic intermittent hypoxia modulates the expression of “angiocrine genes”
Chronic intermittent hypoxia (CIH) was first reported by Fletcher and colleagues as a model for obstructive sleep apnea . In their protocol, mice were exposed to cycles of intermittent hypoxia for 7 h each day for 35 days and were found to develop a long-term hypertensive response to CIH, as it is observed in OSA patients. Such model fails to reproduce the transient hypercapnia that is observed in OSA patients, determined by airway occlusion, and mimics only the intermittent hypoxic episodes that occur chronically in these patients, allowing us to separate the mechanical component of obstruction from the effect of intermittent hypoxia itself. In fact, the main goal of our work was to explore the effects of CIH by itself in the BM vascular compartment. Additionally, over the years, several other groups have shown that this rat model of CIH mimics many aspects of the obstructive sleep apnea syndrome, such as atherosclerosis and alterations in the hematological parameters (for a review see ), and only few have manipulated the CO2 levels inside the chambers [19, 36, 58].
It remains unclear whether the partial pressure of CO2 in the arterial blood (PaCO2) is relevant in humans. Hypercapnia is not a standard parameter analyzed in polysomnographic recordings in patients and therefore there is no consensus on the impact of PaCO2 in arterial blood pressure and other parameters in patients with OSA. For instance, in clinical studies of patients with moderate OSA, the changes in PaCO2 have seemed to be irrelevant  or have shown a slight increase  during the apneic events. Combining chronic sustained hypoxia with hypercapnia was shown to restore the subcutaneous PaO2 to levels close to the normoxic rats  and to induce a smaller increase in the numbers of circulating erythrocytes . However, the effects of combined CIH with hypercapnia in the BM microenvironment and hematopoiesis were not yet described. Therefore, although some data suggest that PaCO2 may influence physiological responses to IH, further studies are needed to evaluate the combined effect of IH and hypercapnia. Only male rats were included in our study to avoid hypothetical effects of estrogens on hematologic responses to chronic intermittent hypoxia since it has been described that 17β-estradiol can influence the expression of hypoxia-inducible genes such as VEGF and endothelin-1  and decreases hypoxic induction of erythropoietin gene expression [20, 54, 80].
In this study, we report that CIH induces a deviation from the normal body weight gain observed in normoxia-exposed rats. These results are consistent with those obtained in a parallel study, where the effects of chronic intermittent hypoxia on body weight of male Wistar rats from the NOVA Medical School colony were evaluated. We observed (unpublished data) that rats exposed to 35 days of CIH weighed significantly less compared with age-matched healthy male Wistar rats kept in normoxia. Several authors reported a weight loss of CIH-exposed rats when compared with control rats [69, 70, 75, 85]. Such alteration in body weight might be explained by the production and release of leptin into the circulation as a response to hypoxia [1, 39, 51, 53, 76, 81]. Leptin is coded by an hypoxia inducible gene that acts upon the hypothalamus to control body weight by reducing food intake and increasing energy expenditure [31, 59, 81]. Moreover, leptin-deficient mice exposed to CIH have a normal weight gain, compared to normoxic mice , indicating that this is a specific leptin-dependent mechanism rather than just a stress response. In our study, the lower body weight may be due to a higher metabolic activity in the hypoxic rats, as we did not observe significant differences in caloric consumption between the two groups of animals, but this assumption needs further validation.
The increased risk of CIH and OSA patients to develop cardiovascular complications led us to hypothesize that a systemic mechanism might be involved in the pathophysiology of such co-morbidities. The recruitment of bone marrow-derived cells has been amply demonstrated to be involved in the onset and progression of cardiovascular diseases . The involvement of BM cells has, for instance, been shown to be important in the setting of atherosclerosis  and has also been implicated in hypertension . Therefore, in the present study, we explored the hypothesis that CIH might affect the BM microenvironment and therefore affect hematopoiesis.
Here, we show a significant increase in BM and PB myeloid (monocyte) counts in animals exposed to CIH, accompanied by an increase in the total monocyte and granulocyte progenitor cell-derived CFUs (CFU-M and CFU-G). Several studies reported an increase in circulating granulocytes both in acute and chronic hypoxia [14, 61] and increased neutrophil and lymphocyte counts in the upper airway mucosa of OSA patients [29, 68]. Interestingly, a significantly lower percentage of macrophages was found in the mucosa of these patients . Nevertheless, chronic intermittent hypoxia increases the amount of pro-inflammatory pulmonary macrophages . No reports of changes in circulating monocyte or in BM myeloid counts as a response to hypoxia exposure were found. However, Roiniotis and his colleagues showed that hypoxia can have a pro-survival effect both in monocytes and macrophages  and Yoon described decreased myelopoiesis in Hif1a−/− embryos , suggesting a positive correlation between hypoxia exposure and expansion of the myeloid compartment. Our data also shows that CIH can modulate the BM lymphocyte content, a finding that had not been described either in CIH-exposed rats or OSA patients. Domagala-Kulawik and her colleagues described a decrease in circulating B cells and an increase in several T lymphocyte subsets in OSA patients, but they addressed only peripheral blood and not BM lymphocyte content . Furthermore, our data is consistent with previous reports showing that stabilization of Hif1α and Hif2α in thymocytes resulted in a remarkable increase in thymocyte apoptosis [4, 13] and that HIF1α deficient chimeric mice have impaired B cell development with decreased proliferation of B cell progenitors .
In addition, we report an increment in circulating erythrocytes and in blood hemoglobin and hematocrit, which correlates with the higher numbers of erythroid colonies (BFU-E) derived from BM progenitors. This increase in erythroid colonies was also described in rats exposed to chronic sustained hypoxia (4 weeks). However, the authors of that study failed to detect any remarkable alterations in granulocyte-macrophage progenitor numbers . Previous studies have shown that acute and chronic sustained  or intermittent hypoxia promoted erythropoiesis [5, 46, 49, 50, 64]. Furthermore, although there is a general lack of healthy controls in most studies, clinical data of OSA patients also suggest an increase in hemoglobin levels [12, 30] and in hematocrit [22, 32], and diurnal variations in erythropoietin levels  which together seem to be correlated with the severity of OSA [12, 77]. The effects of hypoxia in platelet parameters are dependent on the hypoxia administration. In detail, short-term chronic sustained hypoxia (1–4 days) was reported to promote thrombocytosis . However, after 4–5 days of exposure, platelet counts returned to normal and thereafter rapidly declined between the fifth and the ninth days of hypoxia, leveling off at half their normal value [44, 48]. Studies in chronic intermittent asphyxia, however, have shown it does not affect platelet count, but instead increases platelet activation and aggregation , an effect that is correlated with the severity of the disease in OSA patients . These results deserve further studies but emphasize the differences between sustained and intermittent hypoxia.
One interesting observation in bone marrow sections of animals in CIH was the significant increase in VE-cadherin-expressing vessels and in smooth muscle cell coverage, accompanied by a decrease in the vWF-positive vessels. However, we did not observe an increase in total vessel number assessed by CD105 expression. These findings highlight the heterogeneity of the vascular content of the BM microenvironment (similar findings, in a different context were reported in Remedio et al. 2012)  and demonstrate that different vascular markers should be used concomitantly, to avoid misinterpretation of single marker-staining patterns.
These morphological changes in bone marrow vessels upon CIH exposure are also indicative of a molecular process which appears to be favoring vascular stability. vWF is associated with activated and thus less stable vessels, since it is upregulated in endothelial cells treated with FGF2 and VEGF (potent angiogenic inducers) . Contrastingly, VE-cadherin expression and smooth muscle coverage have been associated with increased vessel stability , usually induced after an active angiogenic (generating new vessels) process . HIF1α, that we found to be upregulated in hypoxic rats, is one of the major inducers of angiogenesis, as it upregulates Vegf expression, ultimately leading to vessel permeability and instability [23, 41, 47, 71]. This process is tightly coupled with a decrease of VE-cadherin in the endothelial tight junctions [37, 47], in a VEGF-dependent manner. Our data suggest that the BM sinusoids may not be responding to the proangiogenic effects of VEGF and instead become more stable upon hypoxia exposure. Additionally, the reported increase in SMA-positive vessels in CIH-exposed rats is suggestive of vessel stabilization. This is in line with the findings that hypoxia promotes endothelial cell activation which will lead to the release of mitogenic factors for smooth muscle cells [33, 52].
An understanding of the role of vessels (and of endothelial cells that comprise them) in organ function and recovery following injury has dramatically changed in the last years. It is now accepted that endothelial cells within each organ express a different subset of trophic growth factors, known as “angiocrine factors,” that will satisfy the function and metabolic demands of that specific organ. Moreover, ECs play an active role in organ recovery, through an adaptation of the expression of these trophic factors, supporting the regeneration and proliferation of stem and progenitor cells in the affected tissues, in a paracrine manner. [8, 9, 57]. In the present study, we reasoned the vascular changes seen in the bone marrows of animals exposed to CIH might affect the production of specific angiocrine factors; identification of such factors could in turn explain the alterations in hematopoiesis seen in CIH animals. We observed significant changes in the expression of Csf1, which explains the increase in BM and PB myeloid compartment , and in the levels of Vegfa and Angpt1. Vegf is an hypoxia-inducible gene and is most likely upregulated in response to the increased levels of Hif1α in hypoxic rats. Angiopoietin 1 (Angpt1) in particular has been shown to modulate vessel stability by promoting the chemoattraction of smooth muscle cells to newly formed vessels , usually in response to augmented VEGF levels . Additionally, Angpt1 protects blood vessels from VEGF-induced permeability by inhibiting internalization of VE-cadherin which leads to an increase in VE-cadherin expression and vessel stabilization [26, 27]. Together, these molecular findings correlate with the vascular changes observed in the bone marrow of animals exposed to CIH.
Taken together, our data obtained from an animal model of OSA, reveal that the systemic effects of CIH result in modulation of the bone marrow microenvironment, namely, the bone marrow vasculature, which in turn might be perturbing hematopoiesis. Our results pave the way for pre-clinical and clinical studies aimed at validating these findings in OSA patients.
The authors thank the Neo-vascularization Lab members for their input and suggestions.
This study was supported by Fundação para a Ciência e Tecnologia (FCT, Portuguese Government) grants and fellowships.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.