Angiogenesis is a critical element for adaptation to low levels of oxygen and occurs following long-term exposure to mild hypoxia in rats. To test whether a similar response in mice occurs, CD1, 129/Sv, C57Bl/6, and Balb/c mice were exposed to 10% oxygen for up to 3 wk. All mice showed significant increases in the percentage of packed red blood cells, and CD1 and 129/Sv mice showed increased respiration frequency and minute volume, common physiological measures of hypoxia. Significant angiogenesis was observed in all strains except Balb/c following 3-wk exposure to chronic hypoxia. CD1 hypoxic mice had the largest increase (88%), followed by C57Bl/6 (48%), 129/Sv (41%), and Balb/c (12%), suggesting that some mice undergo more remodeling than others in response to hypoxia. Protein expression analysis of vascular endothelial growth factor (VEGF), angiopoietin (Ang)-1 and Ang2, and Tie2 were examined to determine whether regulation of different angiogenic proteins could account for the differences observed in hypoxia-induced angiogenesis. CD1 mice showed the strongest upregulation of VEGF, Ang2, Ang1, and Tie2, whereas Balb/c had only subtle increases in VEGF and no change in the other proteins. C57Bl/6 mice showed a regulatory response that fell between the CD1 and Balb/c mice, consistent with the intermediate increase in angiogenesis. Our results suggest that genetic heterogeneity plays a role in angiogenesis and regulation of angiogenic proteins and needs to be accounted for when designing and interpreting experiments using transgenic mice and when studying in vivo models of angiogenesis.
- vascular endothelial growth factor
- endothelial cell
- genetic heterogeneity
the mammalian brain is exquisitely dependent on availability of oxygen, which is decreased during pathological events involving ischemia and during normal activities occurring at high altitudes. Evidence suggests that hypoxia can invoke an orchestrated response of several endothelial cell (EC) growth factors and receptors that play different but interacting roles that lead to angiogenesis and successful adaptation of the vasculature under conditions of impaired perfusion (23).
Angiogenesis refers to the formation of new blood vessels from preexisting vasculature and includes sprouting, intussusceptive, and intercalated growth (6). EC receptor tyrosine kinases and their ligands have been recognized as critical mediators of angiogenesis and include the vascular endothelial growth factor (VEGF) family of ligands and receptors, the angiopoietin (Ang) ligands, and their receptor, Tie2, among others (18, 36, 43). These molecules play key roles during the development of the embryonic vasculature (42); in pathological events and diseases involving blood vessels, including tumorigenesis (9), stroke (44), retinopathy (28), and psoriasis (5, 22); and during normal events such as menstruation (10, 16, 33), wound healing (4, 7, 20), and adaptation to alterations in oxygen availability (15, 23, 24).
Prior work completed using Wistar rats has shown that exposure to long-term hypoxia results in a robust angiogenic response mediated by temporal increases in VEGF and Ang2 (29). A similar response should be attainable in mice, allowing the experimental use of genetically altered mouse strains to determine the molecular mechanisms of hypoxic adaptation. However, whether the angiogenic response observed in Wistar rats translates to the mouse is unknown. Moreover, whether there are differences in angiogenic response due to genetic heterogeneity of specific background strains of mice is also unclear. Of translational importance is that both clinical and basic science observations suggest that genetic factors can influence the heterogeneity of responses at the behavioral, anatomic, physiological, and molecular levels in diseases involving vessel growth, such as diabetic retinopathy, cardiovascular disease, psoriasis, and tumor growth.
To identify the genetic conditions leading to robust physiological, anatomic, and molecular angiogenic responses, we tested four common background mouse strains used in the genetic engineering of transgenic mice for their reactions following long-term exposure to chronic mild hypoxia. Three inbred strains (C57Bl/6, 129/Sv, and Balb/c mice) and one outbred strain (CD1) were examined. Our findings indicate that genetic heterogeneity plays a role in angiogenic signaling and vascular remodeling following hypoxia and are consistent with the concept that strain-dependent expression of angiogenic growth factors may determine heterogeneity in the angiogenic response and potentially to angiogenesis-dependent diseases.
MATERIALS AND METHODS
C57Bl/6, 129/Sv, Balb/c, and CD1 male mice were purchased from Charles River Laboratories (Wilmington, MA) at ∼6 wk of age and allowed to acclimatize to the Case Western Reserve University animal facility for at least 1 wk before testing. Mice were caged in groups of five according to strain and treatment condition and were housed under modified barrier conditions and provided with food and water ad libitum. The present study was conducted according to the American Physiological Society's “Guiding Principles in the Care and Use of Animals,” and the protocol was approved by the Case Western Reserve University Institutional Animal Care and Use Committee. In total, 172 mice representing all four strains were examined for hypoxic ventilatory response (HVR), hematocrit, vessel density, or protein analyses following either exposure to hypoxia or normoxia (Table 1). Animals were deeply anesthetized using an intraperitoneal injection of 65 mg/kg pentobarbital sodium where required.
Long-Term Hypoxia Exposure
For mice that were exposed long term to hypoxia, cages were placed into hypobaric chambers maintained at a pressure of 380 Torr (equivalent to 10% normobaric oxygen, which results in a hypoxic exposure equivalent to half of the sea-level O2 pressure or approximates an altitude of ∼18,000 ft) for a period of 21 days. Littermate controls were kept in the same room in a similar chamber under similar conditions except that they were kept at normobaric pressure for the duration of the experiment. Body weights were taken at nine time points throughout the 21-day period. This was completed concurrently with cage cleaning and food and water replacement. Animals were returned to atmospheric pressure for a maximum of 30 min. This period of exposure does not affect results and is standard for all hypobaric studies (3, 29). In addition to the 21-day time point, a time course for protein expression analyses was completed on CD1, C57Bl/6, and Balb/c mice by exposing them to hypoxia (10% oxygen) for 1, 7, 14, or 21 days.
To examine the acute responsiveness in different strains of mice, we measured ventilation using whole body plethysmography (Buxco, Wilmington, NC). Mice were acclimatized to the chamber for 30–60 min during their regular light cycle preceding data collection and were then exposed to test gases in the following order: room air for acclimatization, room air for 20 min, hypoxia (10% O2) for 15 min. Included in the hypoxic exposure time was a 5-min period during which the hypoxic mixture replaced ambient air. Ventilatory parameters were recorded by an analog-to-digital converter connected to a computer. These parameters included tidal volume, respiration frequency, and minute volume. HVR was determined and reflects the ratio of hypoxic minute volume compared with normoxic minute volume. Animal sniffs, sighs, and grooming behaviors were eliminated from the data before their analyses. Data were collected only from the final 5 min of the 10% hypoxia exposure to measure the steady-state conditions.
Mice were weighed, and tail venous blood samples were obtained for hematocrit determination before the animals were killed. For tissues used in immunohistochemistry experiments, animals exposed to hypoxia or normoxia for 21 days were perfused transcardially with cold (4°C; 10–15 ml) PBS, followed by cold 4% paraformaldehyde in 0.1 M phosphate buffer (4°C; 20–30 ml). Brains were removed and postfixed in 4% paraformaldehyde overnight at 4°C and then cryoprotected using a 30% sucrose solution, and 30-μm-thick coronal sections were cut using a freeze-stage microtome (Leica) through the entire brain. Free-floating sections were kept in buffer containing 0.1% azide at 4°C until used.
For tissues used in the protein analyses experiments, brains from hypoxic and normoxic exposed mice were removed following 1-, 7-, 14-, or 21-day hypoxic exposure from deeply anesthetized mice. Brains were extracted, and the cortex was isolated and immediately frozen in a dry ice-ethanol bath and then stored at −80°C until ready to be used.
Immunohistochemistry was done essentially as described in the Vectastain Elite ABC Kit (Vector Laboratories) using monoclonal antibodies against platelet EC adhesion molecule-1 (PECAM-1; MEC 13.3, Pharmingen), an EC marker. The chromagen used was 3,3′-diaminobenzidene. In an effort to minimize variability due to differences in incubation times and background staining, mouse tissues were randomly assigned to one of two groups that were then processed simultaneously.
Stereological and blind controls.
All tissues were processed, and sections were chosen using stereological method including controls for plane of section and randomization of start point among mice within an experimental group, to obtain an overall representation of the results from each specific brain region. This approach allows for the collection of data eliminating all known sources of methodological bias, focuses on absolute parameters rather than densities, and estimates true structural parameters and their variability, instead of sampling only regions or presenting data as a density measure. Mice were coded from the time at which they were introduced into the hypobaric chamber to ensure that the experimenter was blind to the mouse genotype and the mouse treatment condition, with the codes revealed only at the point of statistical analyses.
Blood vessel area measures.
To ensure similarity between animals, the most rostral section through the anterior commissure decussation was chosen as a landmark to randomize the start site. The initial section for each animal within a group was chosen to fall within 180 μm (6 sections) of this landmark. Approximately nine sections were chosen per animal spanning the forebrain cortexes, through to the anterior and medial aspects of the hippocampus. These sections were stained with the EC marker PECAM (CD31). Five mice per strain and treatment condition were analyzed. Images of the right and left hemispheres of the cortex encompassing the motor and sensory cortexes of each section (total of ∼18 images/mouse) were captured using a CoolSnap camera and a ×20 objective lens. The cortical boundaries were established such that the medial aspect of the genu of the corpus callosum served as the most lateral boundary and the top of the cortex itself as the superior boundary. These boundaries were chosen on the basis of their uniformity between mice. The interactive image analysis program ImagePro was used to measure total area (mm2) of the vessels within the preset field of view. Vessel area was quantified from the 18 images/mouse; means for each animal were determined. This approach allows the measurement of change in all three types of angiogenesis, including sprouting, intercalated, and intussusceptive growth. Vessel area was evaluated with four × two (strain × condition) ANOVA with the statistical package Statistica (Statsoft, Tulsa, OK). Tukey's post hoc tests were used to evaluate significant effects.
Cortexes from mice taken at 1, 7, 14, and 21 days were homogenized in RIPA lysis buffer (30 mM Tris·HCl, pH 7.4, 150 mM NaCl, 1% Igepal, 0.5% deoxycholate, 2 mM EDTA) containing sodium orthovanadate and protease inhibitors (Roche) and cleared, and protein was quantified using the BCA protein assay (Pierce). Two-hundred micrograms of whole cell lysate from CD1, C57Bl/6, and Balb/c samples were loaded into 8% SDS-PAGE (Ang1 and 2, Tie2 blotting) or 15% SDS-PAGE (VEGF blotting). Proteins were transferred to PVDF using a semidry apparatus, blocked for a minimum of 30 min in 5% milk (blotto), and probed with polyclonal VEGF (A20, Santa Cruz), polyclonal Ang1 or Ang2 (Chemicon; Sigma), monoclonal Tie2 (Chemicon), and monoclonal anti-actin A40 (Sigma) in blotto overnight at 4°C. For the monoclonal primary antibodies, the secondary detection antibody used was goat anti-mouse horseradish peroxidase (HRP) (Bio-Rad) or biotinylated goat-anti-rat (Vector) with peroxidase-labeled streptavidin (Kirkegaard and Perry Laboratories), and for the polyclonal primary antibody, the secondary used was goat anti-rabbit HRP (Bio-Rad). Blots were developed using SuperSignal West Dura Extended Duration Substrate (Pierce) according to the manufacturer's instructions. Confirmation of equal loading was completed using antibodies against β-actin on the same membranes initially probed, by either cutting the membrane in half and probing the upper half for the protein of interest and the lower half for β-actin, or stripping and reprobing. Individual results were repeated in triplicate, and three individual sets of mice for each strain were tested. Densitometry was completed using a Bio-Rad GelDoc system and Quantity One software. Data are presented as relative units, and protein expression was normalized to β-actin to account for any differences in protein loading.
Quantitative data are presented as averages ± SE and were analyzed using ANOVA followed by post hoc Tukey tests with the statistical package Statistica (Statsoft).
Different Physiological Responses Following Exposure to Hypoxia
All four mouse strains exposed to 21 days of 10% hypoxia developed polycythemia or a significant increase [F(3,30) = 6.4, P < 0.005] in the packed red blood cell volume of the blood compared with normoxic controls (hematocrit; Fig. 1). Across all four strains of mice, packed red blood cell volume was significantly elevated from 43 ± 0.7% to 71 ± 0.8% (P < 0.001). No differences between strains in the normoxic condition were observed.
Overall, comparison of weight changes across the 21-day period revealed significant main effects of hypoxia exposure [F(1,31) = 71.9, P < 0.001] and mouse strain [F(3,31) = 138.7, P < 0.001]. Post hoc analyses indicated Balb/c, C57Bl/6, and 129/Sv mice exposed to hypoxia failed to gain body weight throughout the experimental paradigm (Balb/c, 21.8 ± 0.3 g at 0 days to 21.4 ± 0.3 g at 21 days; C57Bl/6, 23.3 ± 0.4 g at 0 days to 22.9 ± 0.4 g at 21 days; and 129/Sv, 24.3 ± 0.3 g at 0 days to 24.5 ± 0.6 g at 21 days; Fig. 2). Only the CD1 mice gained significant (P < 0.0001) body weight with exposure to hypoxia (32.4 ± 0.3 g at 0 days to 34.8 ± 1.7 g at 21 days; Fig. 2). During the same period, the body weight of normoxic control mice from the same strains significantly increased (P < 0.0001) from 34.6 ± 0.69 to 43.1 ± 1.6 g for CD1, 24.2 ± 0.4 to 28.9 ± 0.7 g for C57Bl/6, 21.4 ± 0.6 to 24.6 ± 0.7 g for Balb/c, and 24.7 ± 0.9 to 29.2 ± 1.1 g for 129/Sv (Fig. 2), a natural occurrence for mice between the ages of 6 and 8 wk. The interaction between strain and condition was marginally significant [F(3,31) = 2.6, P = 0.07]. A post hoc analysis on the strain by condition indicated that CD1 mice are significantly heavier than the other strains (P < 0.001), and after 21 days of growth this continues to be true, regardless of condition.
HVR (the ratio of hypoxic minute volume compared with normoxic minute volume) was assessed using whole body plethysmography to measure ventilation frequency, minute volume of air (i.e., flow), and tidal volume (the amount of air inhaled during a normal breath). No significant differences in mouse strain were observed for HVR as a result of exposure to hypoxia, although overall tidal volume indicated a significant effect of strain [F(2,15) = 13.67, P < 0.001] and an interaction between strain and pre- vs. posthypoxia measures [F(2,15) = 3.69, P < 0.05] (Fig. 3, A and B). Post hoc data indicate that no overall change occurs in tidal volume in any of the strains as a result of hypoxia, but CD1 mice have an increased overall tidal volume relative to 129/Sv and C57Bl/6 mice (P < 0.05). A significant effect of strain [F(2,15) = 17.50, P < 0.001] was found for minute volume, the amount of air taken in. Simple effects ANOVAs to investigate strain differences indicate a strain difference [F(2,15) = 13.14, P < 0.001], where CD1 mice take in significantly more air than 129/Sv and C57Bl/6 (P < 0.05). The difference observed in both tidal and minute volume for the CD1 mice most likely reflects weight differences; therefore the ratio of the minute volumes (HVR) is a better comparative measure of hypoxic sensitivity, as it is not weight dependent.
A marginally significant strain effect is present for respiration frequency [F(2,15) = 2.82, P = 0.09]. Simple-effects ANOVAs to investigate simple strain differences before hypoxia indicate that C57Bl/6 mice have increased respiration frequency relative to CD1 and 129/Sv mice (P < 0.05). There was a significant interaction between strain and pre- vs. posthypoxia measures on respiration frequency [F(1,15) = 3.69, P < 0.05].
CD1 and 129/Sv mice had significantly increased respiration frequency (i.e., they took more breaths; P < 0.05) and increased minute volume (they took in more air) following exposure to hypoxia, whereas C57Bl/6 mice did not appear to respond in the same manner (Fig. 3, C and D).
To account for differences in body weights between mouse strains, we normalized the tidal volume and minute volume measures to weight (Fig. 4). Simple-effects ANOVAs on corrected tidal volume data revealed a significant main effect of hypoxia exposure [F(1,12) = 5.09, P < 0.05] and a significant interaction between strain and hypoxia [F(2,12) = 4.05, P < 0.05, Fig. 4B]. Post hoc analyses on the interaction revealed only a significant difference between the C57Bl/6 mice pre- vs. posthypoxia exposure (P < 0.05). Simple-effects ANOVAs on corrected minute volume data uncovered significant main effects of both strain [F(2,12) = 5.69, P < 0.05] and hypoxia exposure [F(1,12) = 33.94, P < 0.05, Fig. 4C]. Post hoc analyses of the strain effect indicates that C57Bl/6 have significantly greater minute volume than CD1 (P < 0.05) but not 129/Sv mice and that CD1 and 129/Sv mice also did not differ. Post hoc analyses of the hypoxic effect confirmed that overall, mice under hypoxic conditions have increased minute volumes compared with normoxic conditions (P < 0.05).
Balb/c mice were not tested due to difficulties in obtaining steady basal readings; therefore no data from these mice are presented.
Different Angiogenesis in the Brains of Hypoxic Mice
Angiogenesis was evaluated in all four mouse strains exposed to 21 days of 10% hypoxia. No statistical differences in vessel area were observed between hemispheres; therefore hemisphere data were combined and presented as vessel area per region. The four × two ANOVA revealed a significant main effect of strain [F(3,32) = 4.95, P < 0.01] and a main effect of hypoxia [F(1,32) = 25.27, P < 0.0001]. Despite the strong main effects of strain and hypoxia, an interaction between strain and hypoxia was not found [F(3,32) = 2.64, P = 0.105]. Post hoc analyses on this interaction indicated no significant differences between strains of mice under normoxic conditions. In all strains of mice except Balb/c, a trend existed for an increase in vessel area in the sensory and motor cortexes following exposure to long-term chronic hypoxia (CD1 normoxic 0.0146 ± 0.0006 mm2 vs. CD hypoxic 0.0275 ± 0.001 mm2; 129/Sv normoxic 0.0133 ± 0.0006 mm2 vs. 129/Sv hypoxic 0.0203 ± 0.0006 mm2; C57Bl/6 normoxic 0.0125 ± 0.001 mm2 vs. C57Bl/6 hypoxic 0.0185 ± 0.0005 mm2; Balb/c normoxic 0.016 ± .002 mm2 vs. Balb/c hypoxic 0.0179 ± 0.0009 mm2). Post hoc Tukey tests demonstrated that the increase in vessel area in CD1 mice alone was significant (P < 0.003), indicating angiogenesis occurred, confirming previous findings reported in the rat [Fig. 5; P < 0.05 (29)]. As the ANOVA offers a conservative analysis of multiple groups, and because of the nature of the main effects and interactions, familywise comparison t-tests with a modified P value (modified α value P < 0.05/2) were completed to further evaluate the nature of any significant effects, as the trends seen in Fig. 5, yet not represented in the ANOVA, were suggestive of more subtle significant effects. Independent t-test comparisons of the vessel areas between normoxic and hypoxic animals for the C57Bl/6 and 129/Sv animals showed a significant effect of hypoxia on both strains [t(8) = −2.95, P < 0.02, t(8) = −3.36, P < 0.01, respectively]. A strain-dependent rank order of hypoxia-induced angiogenesis existed such that CD1 > 129/Sv > C57Bl/6 > Balb/c, with an 88% (P < 0.003), 48% (P < 0.01), 41% (P < 0.02), and 12% increase (not significantly different) in vessel area compared with normoxic littermates. The CD1 hypoxic mice had significantly greater vessel area compared with both the C57Bl/6 and Balb/c hypoxic mice.
Regulation of VEGF, Angiopoietins, and Tie2 Protein Expression Differs Between Mouse Strains Following Exposure to Hypoxia and May Account for Differences in Angiogenesis
CD1 mice showed the most robust angiogenic response as measured by increased vessel area (88%), and this coincided with increases in VEGF and Ang2 protein (Fig. 6, A and B), such that VEGF protein expression increased by 30% from control levels by 1 day following exposure to hypoxia (P < 0.05) and was maintained at fairly high expression levels through 7 days and then decreased to basal levels by 14 days. Ang2 protein increased by 29% with maximum levels observed 1 day after hypoxia exposure. Significant increases in Tie2 protein expression (42%, P < 0.05) were observed 21 days following initial exposure to hypoxia, consistent with the increase in angiogenesis, and perhaps reflective of an increase in the number of ECs containing Tie2. Ang1 protein increased (26%, P < 0.05) by 1 day and was found to be further increased by 14 days (38%, P < 0.05) in CD1 mice.
In the poorly angiogenic Balb/c mice, where significant angiogenesis was not observed (12% increase), VEGF levels increased in a nonsignificant manner (13%) 1 day following exposure to hypoxia, despite equivalent normoxic VEGF expression as CD1 mice (Fig. 6, A and B). By 21 days of hypoxia exposure, VEGF expression appeared to decrease by 48% less than original normoxic levels (P = 0.058). Ang2, Tie2, and Ang1 protein levels all failed to increase from basal control levels at any time point, and like VEGF, Ang2 showed a trend toward downregulation (27% decrease; P = 0.06) of protein expression.
In the moderate angiogenesis-responsive C57Bl/6 mice (41% increase in vessel area), VEGF and Ang2 protein increased by 15% (P < 0.05) 1 day following exposure to hypoxia with VEGF levels dropping back to normoxic levels by 21 days. Ang2 protein expression remained elevated throughout the 21-day period. Tie2 levels, despite a trend for increased expression (15%), did not statistically differ. Ang1 levels were found to be increased 1 day following hypoxia exposure (P < 0.05) but did not remain significantly elevated.
Strain differences are not new to biological studies but have become increasingly important when transgenic mice are studied in models where the background strain may or may not contribute to the experimental variable. We investigated the effects of genetic heterogeneity on hypoxia-induced physiological, anatomic, and molecular responses, including hematocrit, weight gain, respiratory frequency, HVR, angiogenesis in the cortex, and cellular signaling events previously shown to be involved in hypoxia-induced vascular plasticity. Strain differences were observed in respiratory frequency and minute volume, weight gain, vessel area, and temporal expression of VEGF, Ang2, Tie2, and Ang1 protein levels throughout the duration of chronic hypoxia, indicating that the genetics of different mice can contribute to hypoxia-induced responses.
Physiological Response to Hypoxia Varies with Strain
Different strains of mice respond to long-term hypoxia exposure differently, such that C57Bl/6, 129/Sv, and Balb/c mice, but not CD1 mice, failed to gain weight throughout the 21-day exposure to 10% oxygen (Fig. 2). These differences could not be attributed to a lack of hypoxic exposure, as all four strains showed a 65% increase in hematocrit levels (Fig. 1). CD1 mice were the only ones to gain weight throughout the 21-day exposure to hypoxia, suggesting that these mice may experience less stress initially to the low oxygen environment, allowing them to adapt more quickly, and initiate normal eating patterns sooner, although this is purely speculative as food quantity ingested was not quantified. This would support the initial weight loss followed by the quick rebound back to weight gain as early as 2 days following introduction to the hypobaric chamber (Fig. 2A).
A natural physiological phenomenon in response to hypoxic environments is an increase in respiratory frequency and in the volume of air taken in within a set period of time (12, 31, 38). Differences in ventilatory responses to hypoxia attributed to strain have previously been reported in rats (13, 17, 34, 37). In our studies, although no differences in overall HVR and total volume intake (Fig. 3, A and B) were found, CD1 and 129/Sv but not C57Bl/6 mice significantly increased their frequency of respiration and their minute volume of air flow (Fig. 3). The lack of response in C57Bl/6 mice cannot be accounted for by ventilatory decline (13), as the data were collected in the final 5 min and after 10–15 min of exposure. However, the lack of response may reflect an adaptation to hypoxia via decreasing their metabolic rate, an issue that could potentially be resolved by normalizing minute ventilation to either O2 consumption or CO2 production. When we normalized to body weight, C57Bl/6 mice showed significantly increased tidal volume and minute volume despite the lack of increased respiration frequency. Body weight normalization also led to reduced minute volume in CD1 mice relative to the C57Bl/6. This suggests different mechanisms are used by CD1, 129/Sv, and C57Bl/6 mice for increasing overall air volume under hypoxic conditions. These findings stress the importance of taking multiple physiological measures to confirm exposure to hypoxia.
Angiogenic Vessel Remodeling and Cell Signaling is Strain Dependent
Strain-related differences in response to hypoxia and in angiogenesis in the rat brain have previously been found (26, 27, 40). We found different strains of mice had different angiogenesis responses following exposure to hypoxia and in the regulation of angiogenesis-related proteins. After 21 days of exposure to 10% oxygen, the outbred strain of mice, CD1, had the most significant increase in cortical vessel area (Fig. 5). The increase in vessel area confirmed a recent report showing increases in capillary density counts in the frontal cortex of CD1 mice following 4 wk of exposure to hypoxia (21) and is comparable to the increases in vascular area observed in the striatum of CD1 mice exposed to 10% oxygen for 21 days (41). Significant increases in vessel area induced by hypoxia were also observed in 129/Sv and C57Bl/6 mice but not in Balb/c mice (despite their increased hematocrit). As all four strains had similar normoxic levels of blood vessel area, the increase following hypoxic exposure may reflect different molecular responses and plasticity of the vascular components themselves, genetic heterogeneity, or both.
The current proposed paradigm of angiogenic response involves angiogenic initiation being triggered by increases in VEGF and Ang2, while microvessel maturation and stabilization is related to increases in Ang1 (14, 25). Comparison of angiogenic molecules in cortexes of mice with high (CD1), intermediate (C57Bl/6), and low (Balb/c) cortical angiogenesis showed differences in the regulation of VEGF, Ang1 and Ang2, and Tie2 protein (Fig. 6). In CD1 mice, larger increases in VEGF and Ang2 were found than in C57Bl/6 mice, although in both mice these increases were sufficient and significant enough to induce remodeling. In Balb/c mice where cortical angiogenesis remained low and insignificant, only subtle increases in VEGF were found, and by day 21, a 48% reduction was observed. Coupled with a lack of Ang2 increases, this lack of increased protein expression could potentially explain the lack of cortical angiogenesis in these mice. In CD1 and C57Bl/6 mice, both Ang1 and Tie2 protein expression increased following exposure to hypoxia in contrast to what was reported for the Wistar rat (29). The increases in Ang1 may reflect a response to neoangiogenesis and the need for vessel stabilization and the recruitment of pericytes (i.e., microvessel maturation), a role consistent with the function of Ang1 (35, 39), and may provide additional insight into why CD1 mice had the largest angiogenic response. The moderate increase in Tie2 may reflect increases in the number of ECs (and therefore Tie2), an increase in Tie2 synthesis from individual ECs, or both. These data suggest and are consistent with the idea that some mouse strains respond more robustly to angiogenic stimuli while others fail to respond at all, despite showing physiological responses (such as increases in hematocrit; Fig. 1) to the angiogenic stimulus. Although it could be hypothesized the Balb/c mice could, with time, show an angiogenic response, it seems unlikely, based on normal physiological standards.
One other possibility that has not been proposed in other hypoxia experiments reporting strain-dependent differences is that the smaller response or apparent lack of response to half-normal oxygen may reflect a greater tolerance to hypoxia. If the mice were exposed to deeper hypoxia such as the equivalent of 8 or 9% oxygen, a more robust and statistically significant response may be elicited. In other words, the strain differences might be in sensitivity to hypoxia (i.e., physiological heterogeneity) as well as genetic heterogeneity.
Our ranking of angiogenic response in the hypoxic brain (CD1 > 129/Sv > C57Bl/6 > Balb/c) differs greatly from ranking of others using different angiogenic stimuli and in different peripheral organs and organ systems. One group, using corneal micropocket assays, has shown basic fibroblast growth factor-induced angiogenesis occurs in a strain-dependent rank order with the following rankings: Balb/c > 129/Sv > C57Bl/6 > CD1, an almost reverse order from what we observed (32). Further genetic variability and ranking were observed using the same strains of mice but in an aortic ring assay (32). Moreover, our ranking of mouse response in the cortex differs from rankings of others using hypoxia as their angiogenic stimuli in the same mouse strains but in different tissues (2). Therefore, differences in neovascularization and response to angiogenic stimulus differ not only between strain and tissue/organ but also probably reflect intrinsic differences in ECs themselves, in addition to the idea that EC responsiveness can change between in vitro and in vivo approaches. That ECs maintain their own heterogeneity within different vascular beds is, irrespective of strain, significant and consistent with recent reports by McDonald's group (1, 19) where different vascular beds and the ECs within them were shown to respond differently to long-term VEGF inhibition.
Importance of Strain in Designing and Interpreting Experiments
Knockout mice are often generated using 129/Sv or other substrains of the 129 mouse ES cells, a strain that we and others have shown responds well to angiogenic stimuli. However, 129 ES cells are most often implanted into C57Bl/6 or Balb/c mice where differences intrinsic to each strain, such as robustness of molecular signaling events and plasticity to angiogenic stimuli, need to be accounted for when interpreting angiogenic data. For example, Ang2 knockout mice originally engineered on a 129/J background presented with a phenotype that included chylous ascites, defects in hyaloid vessel remodeling, defects in lymphatic vessels (11), and dysmorphogenesis of kidney cortical peritubular capillaries (30), dying early in postnatal development. After continual backcrossing into the C57Bl/6 background, survival rates increased so that >90% of mice survived and presented with only mild chylous ascites, mild vascular and lymphatic defects, and significantly reduced morbidity. When backcrossed back into the 129/J background strain, the postnatal phenotype recurred (8). This is consistent with our results suggesting that different mouse strains respond to angiogenic stimuli differently and that choice of background strain when engineering transgenic or knockout mice should be carefully considered. Similarly, when the role of angiogenesis proteins and angiogenesis inhibitors in any in vitro or in vivo assay is evaluated, whether it be primary ECs used in a tubule formation assay, an aortic ring explant, or an animal model of tumorigenesis, limb ischemia, stroke, or any type of pathology involving blood vessel plasticity, attention to the impact mouse strain can have on experimental data and interpretation may need to occur.
Our study demonstrates that genetic influences can help determine differential regulation of proangiogenic growth factors and physiological responses to a low-oxygen environment. Differences within individuals and the ability to modulate angiogenic (or antiangiogenic) events may contribute to differences in susceptibility to or severity of angiogenesis-dependent diseases and could help explain differential angiogenic tissue and patient responses during the recovery phase following ischemic or cerebrovascular injury.
This work was supported by a Summer Project for Undergraduate Research Award from the Howard Hughes Medical Institute (K. Noon), American Heart Association Grant 0435103N (N. L. Ward), National Institutes of Health Grants P30-AR-39750 (N. L. Ward) and RO1-NS-38632 (J. C. LaManna), and a grant from the Natural Sciences and Engineering Research Council of Canada (T. L. Ivanco).
We thank Xiaoyan Sun and D'Arbra Blankenship for technical support.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society