Myoglobin knockout (myo-/-) mice were previously reported to show no obvious phenotype but revealed several compensatory mechanisms that include increases in cardiac capillary density, coronary flow, and hemoglobin. The aim of this study was to investigate whether severe hypoxic stress can exhaust these compensatory mechanisms and whether this can be monitored on the gene and protein level. Myo-/- and wild-type (WT) mice were exposed to hypoxia (10% O2) for 2 wk. Thereafter hemodynamic parameters were investigated by invasive measurement combined with magnetic resonance imaging. Cardiac gene and protein expression were analyzed using cDNA arrays and two-dimensional gel electrophoresis plus mass spectrometry, respectively. Hematocrit levels increased from 44% (WT) and 48% (myo-/-) to 72% in both groups. Similar to WT controls, hypoxic myo-/- animals maintained stable cardiovascular function (mean arterial blood pressure 82.4 mmHg, ejection fraction 72.5%). Cardiac gene expression of hypoxic myo-/- mice differed significantly from WT controls in 17 genes (e.g., keratinocyte lipid binding protein +202%, cytochrome c oxidase Vb +41%). Interestingly, hypoxia inducible factor-1α remained unchanged in both groups. Proteome analysis revealed reduced levels of heart fatty acid-binding protein and heat shock protein 27 both in hypoxic myo-/- and WT mice. Our data thus demonstrate that myo-/- mice do not decompensate during hypoxic stress but are surprisingly well adapted. Changes in energy metabolism of fatty acids may contribute to the robustness of myoglobin-deficient mice.

  • chronic hypoxia
  • hemodynamics
  • magnetic resonance imaging
  • gene expression
  • proteome

myoglobin, known to be an important oxygen-binding hemo-protein located in cardiac and skeletal muscle cells, was shown to facilitate oxygen transport (23). The generation of myoglobin knockout (myo-/-) mice revealed the surprising result that these mice were viable and fertile and displayed no obvious signs of cardiac or skeletal dysfunction (5, 6). However, multiple compensatory mechanisms were found to be induced in these mice, which include an increase in cardiac capillary density, coronary flow, and blood hemoglobin (6). These mechanisms together steepen the gradient of oxygen from the capillary to the mitochondria and thereby most likely contribute to the benign phenotype of myo-/- mice. Moreover, even pathological conditions like myocardial infarction caused left ventricular function to deteriorate to the same extent compared with WT hearts (15).

Additional functions of myoglobin were discovered using the myo-/- mice as appropriate control. Myoglobin was shown to be an important scavenger of bioactive nitric oxide (NO) (4) and contributes to the metabolism of oxidative radicals generated during postischemia recovery of hearts (3). Therefore myo-/- mice are likely to experience an increased oxidative and nitrosative stress. Both reactive oxygen species and NO are known to regulate gene expression (1, 14).

In view of these findings, the present study addresses two questions. First, we wanted to know whether myoglobin knockout mice show signs of dysfunction when exposed to 2 wk of chronic hypoxia (10% O2). To test whether under these conditions the compensatory mechanisms of myo-/- mice are still sufficient or become exhausted, various cardiovascular parameters were measured including left ventricular function by magnetic resonance imaging (MRI). Second, we wanted to investigate whether the hypoxia-induced functional changes were mirrored by respective changes in gene and protein expression. To this end the cDNA array technology (2,352 genes) was employed to screen for potential genetic regulatory mechanisms. In addition, cardiac proteins of mice subjected to hypoxia were analyzed by two-dimensional gel electrophoresis combined with mass spectrometry.


Animals. Myo-/- mice were generated by deletion of the essential exon-2 via homologous recombination in embryonic stem cells, as described (6). Body weight ranged from 27 to 36 g and left ventricular heart weight ranged from 108 to 138 mg, with no significant differences between all groups. Animals had to be smaller for MRI measurements and thus a reduced body weight (23-27 g). All animals were treated according to National Institutes of Health guidelines.

Chronic hypoxia (10% O2). Myo-/- and WT mice were exposed to hypoxia (10% O2) for 2 wk. Mice were placed in a ventilated Plexiglas chamber that was flushed with a 1:1 mixture of room air and N2. Gas was not recirculated. High inflow of gas (5,000 ml/min per kg mice) prevented accumulation of CO2 and H2O. Littermate controls were housed in an equal chamber with open top in similar conditions (room air). Food and water were provided ad libitum. Temperature was 22-24°C. The chamber was opened twice a week for 5 min for changing cages and replenishing of food and water supplies.

Blood cell parameters. For the analysis of blood cell parameters, blood was taken by cardiac puncture from urethane-anesthetized mice (0.6 ml) and collected in a pediatric tube (EDTA). Hematological parameters were determined using a fully automated SE-9000 from Sysmex (Kobe, Japan).

Morphometric procedure. The protocol for analyzing cardiac capillary density was similar to Gödecke et al. (6). In brief, the hearts were perfused via the aorta with St. Thomas solution until cardioplegia before infusion of 0.22 M saccharose and 0.1 M cacodylate for 5 min followed by 5 min of perfusion with fixative (2.5% glutaraldehyde in 0.1 M cacodylate buffer) at 100 mmHg. Hearts were excised and postfixed overnight in the same fixative. The free anterior wall of the left ventricle was cut into six to nine pieces before embedding with Spurr (Serva, Heidelberg, Germany). Semithin sections were stained with 1% toluidine blue, and the numbers of capillaries per area were counted with a computerized digital analysis system (Soft Imaging Systems, Muenster, Germany).

Hemodynamic measurements. To measure hemodynamic parameters, mice were anesthetized with urethane (1.5 g/kg) and placed on a controlled warming table (37°C). After preparation of the right carotid artery a stretched polyethylene catheter (inner diameter = 0.4 mm; filled with saline/heparin 100 U/ml) was inserted and connected to a Statham P23XL transducer. Blood pressure and heart rate were measured for a period of 10 min after steady-state levels were reached using the data-acquisition system Powerlab 800 and chart software (ADInstruments). Normoxic animals breathed room air, but hypoxic animals were supplied with a mask through which flowed 10% oxygen.

MRI. To measure left ventricular function, we used the MRI technique. Due to the necessity to obtain a prolonged stable heart and respiratory rate during the MRI measurement of the mice, we switched from intraperitoneal to inhalation narcosis with isoflurane (1.2%) together with diazepam (2.5 mg/kg mouse wt). MRI investigations were performed using a Bruker DRX 9.4-T wide-bore NMR spectrometer equipped with an actively shielded 40-mm gradient set (capable of 1 T/m maximum gradient strength and 110-μs rise time at 100% gradient switching) and a dual tuned 30-mm birdcage resonator (1H). High-resolution images of mice hearts were acquired using an ECG-triggered fast gradient echo (FLASH) cine sequence with a flip angle of 45° (17). To obtain a maximum of images over the entire cardiac cycle (∼100 ms at a heart frequency of 600 beats/min), the shortest possible echo time (1.8 ms) was applied. Using a repetition rate of 3.8 ms allowed for the acquisition of 26 frames during one R-R interval, which lead to a sufficient temporal resolution to accurately determine diastole and systole within the cardiac cycle. The resulting in-plane resolution is 117 × 117 μm2 (field of view 30 × 30 mm2, 256 × 256 matrix). The total acquisition time per slice for one cine sequence was ∼2 min. Six to eight contiguous ventricular short axis slices (slice thickness 1 mm) were acquired to cover the entire heart. Cavity and myocardial volumes in each slice were obtained by multiplication of measured component area with slice thickness. Thickness of topmost and lowest slice was corrected for overlap with nonheart structures by density measurement. Because contiguous slices without interslice gap were acquired, total volumes were calculated as the sum of all slice volumes, and no further volumetric assumptions were needed. From the MRI data measured, the following functional parameters could be derived: heart frequency, end-diastolic and end-systolic volumes (μl), stroke volume [end-diastolic - end-systolic volume (μl)], ejection fraction [stroke volume/end-diastolic volume (%)], left ventricular mass [calculated from the epicardial borders - left ventricular cavity, respectively, assuming myocardial specific gravity of 1.05 g/cm3 (mg); by definition, papillary muscle are included in the myocardial mass].

RNA isolation. Normoxic and hypoxic mice were taken out of the chamber and were immediately cervically dislocated. Promptly hearts were perfused with ice-cold saline buffer (0.9% NaCl) before hearts were taken out and homogenized in RLT buffer (Qiagen, Hilden, Germany) with an Ultra Turrax T18 homogenizer (IKA, Staufen, Germany) at 24,000 rounds/min for 2 min. RNA isolation protocol was a modified RNeasy protocol developed for heart tissue (Qiagen). Shortly, the homogenate of tissue and RLT buffer was treated with proteinase K for 20 min before RNA was extracted on Qiagen RNeasy columns. A DNase digestion took place on-column. Quality of eluted RNA was assessed by agarose gel electrophoresis.

cDNA synthesis and array hybridization. The expression profile of 2,352 genes was studied using cDNA arrays (Atlas mouse 1.2 arrays I and II, Clontech). mRNA was isolated from cardiac total RNA using magnetic beads, before mRNA was reverse transcribed to radioactive cDNA with 32P according to the Clontech protocol. The cDNA arrays were hybridized overnight and washed the day after. cDNA membranes were exposed to imaging plates (BAS MS 2040, Raytest) for 1-7 days before radioactive signals were detected by a phosphorimager (Fujifilm fluorescent image analyzer FLA 2000) and densitometrically measured using AtlasImage software (Clontech). Signal intensities were corrected for local background and normalized to the total signal intensity of all genes on one array. Given values are the mean of four to five experiments in each group. We performed an unpaired Student's t-test to compare array data (P < 0.05). To test for type I error, we checked some interesting candidates with semiquantitative real-time PCR.

Real-time PCR with relative quantitation of gene expression. The mRNAs for actin, keratinocyte lipid-binding protein, cytochrome c oxidase Vb, and HIF-1α were measured by real-time quantitative PCR using the GeneAmp 5700 sequence detection system (PE Applied Biosystem). The sequences of forward and reverse primers were designed by Primer Express (PE ABI). The following primers were used (f = forward primer, r = reverse primer): actin-f 5′-AGG CCC AGA GCA AGA GAG GT-3′, actin-r 5′-CGT CCC AGT TGG TAA CAA TGC-3′, keratinocyte lipid-binding protein-f 5′-ATG GCC AGC CTT AAG GAT CTG GAA G-3′, keratinocyte lipid-binding protein-r 5′-CAC AGT CGT CTT CAC TGT GCT CTC-3′, cytochrome c oxidase Vb-f 5′-TCC ATG GCT TCT GGA GGT G-3′, cytochrome c oxidase Vb-r 5′-TCT CCC TCT CCA GCC CAG T-3′, HIF-1α-f 5′-TCA GAG GAA GCG AAA AAT GGA-3′, HIF-1α-r 5′-CCC GGT TGC TGC AAT AAT GT-3′.

The comparative theshold cycle (CT) method was used for relative quantitation. CT, which correlates inversely with the target mRNA levels, was measured as the cycle number at which the SYBR green fluorescent emission increases above a threshold level. Relative quantitation was performed relating the number of cycles of actin to the corresponding gene at CT 0.2 and at CT 0.4. As cycle efficiency is between 1.9 and 2 at this range, the relative amount is determined using the formula 2ΔCT, where ΔCT describes the difference between actin and the corresponding gene.

Two-dimensional PAGE and mass spectrometry. The myocardial protein pattern of hearts was analyzed by two-dimensional gel electrophoresis (2D-PAGE) similar to Laussmann et al. (12). In brief, hearts were lysed (9 M urea, 2% CHAPS, 1% DTT, 10°C) and solubilized using a Potter homogenizator. Isoelectric focusing (IEF) was performed using immobilized pH gradients (IPGs) and the IPG-phor system (Amersham Pharmacia Biotech). Equilibrated IPG strips were placed on top of the gel (total acrylamide concentration 13%, gel size 1 mm × 25.5 cm × 20 cm). Second dimension electrophoresis (SDS-PAGE) was performed on a ETTAN-DALT II vertical electrophoresis unit (Amersham Pharmacia Biotech). Analytical gels were silver-stained and digitized using a UMAX PowerLook III scanner calibrated by a photographic step tablet (Kodak). Differentially expressed proteins were identified by mass spectrometry. Coomassie-stained spots were excised and the protein within the gel piece was digested by trypsin. Then the peptides were analyzed by nanospray electrospray ionization tandem mass spectrometry (ESI-MS/MS) using a SCIEX Q-STAR system (PE Sciex). Peptide sequences were identified by mascot database search (www.matrixscience.com) or de novo sequencing followed by blast database similarity search (www.ncbi.nlm.nih.gov/BLAST/).

Statistical analysis. All results are expressed as means ± SD. For comparisons between two groups, Student's t-test was used for statistical analysis. For multiple comparisons, we performed a two-way ANOVA with Bonferroni post hoc testing. MRI data were paired; all other data were unpaired. A value of P < 0.05 was considered statistically significant.


Phenotype and hemodynamic parameters. WT and myo-/- mice were investigated either under control conditions (21% O2) or after 2 wk of hypoxia (10% O2). All animals survived the exposure to 10% oxygen. During hypoxia, myo-/- mice displayed neither an obvious phenotype nor a major change in daily activity compared with WT controls. Detailed analysis revealed that after 2 wk exposure to hypoxia, WT and myo-/- animals slightly lost weight (-5.6 and -7.4%, respectively, NS), while normoxic littermates gained weight mildly (+6.8 and +15.5%, respectively, NS). Morphometric analysis revealed a density of 3,861 ± 677 and 4,234 ± 596 capillaries/mm2 in hearts of normoxic WT and myo-/- mice, respectively. Two weeks of hypoxia increased cardiac capillary density to 5,604 ± 622/mm2 in WT mice and 5,405 ± 443/mm2 in myo-/- mice (n = 4-5).

Figure 1 shows various hemodynamic parameters such as heart rate (Fig. 1A) and mean blood pressure (Fig. 1B) of normoxic and hypoxic WT and myo-/- mice. As can be seen, there are no significant differences between WT and transgenic animals nor were there differences when these animals were subjected to hypoxia. To investigate whether there are differences in left ventricular function, MRI was performed. For this purpose, animals exposed for 2 wk to 10% oxygen were analyzed and received the same gas mixture during MRI measurements. Calculated left cardiac mass was 97.5 ± 18.7 mg in the WT and 90.4 ± 2.7 mg in myo-/- mice during normoxia, whereas cardiac mass of hypoxic WT mice was unchanged (100.8 ± 9 mg) and slightly reduced in hypoxic myo-/- mice (84.3 ± 10.9 mg). Figure 2 shows the data of before and after exposure to hypoxia. Only minor differences can be observed regarding end-diastolic volumes (Fig. 2A). Left ventricular stroke volume (Fig. 2C) was not different between any of these groups. Ejection fraction (Fig. 2D) was between 72 and 81% for all four experimental groups. Heart rate was 502 ± 56 beats/min in the WT and 549 ± 42 beats/min in the myo-/-. During hypoxia no changes in heart rate were observed when animals were anesthetized with urethane. However, under isoflurane narcosis, heart rate of hypoxic myo-/- mice decreased (356 ± 68 beats/min) while heart rate remained unchanged in hypoxic WT controls (490 ± 5 beats/min).

Fig. 1.

Hemodynamic parameters of wild-type (WT) and myo-/- mice in normoxia and chronic hypoxia. A: heart rate (HR). B: mean blood pressure. Mice were kept under normoxia (21% O2) or subjected for 2 wk to hypoxia (10% O2). Hemodynamic parameters did not show significant differences between the 4 groups (n = 3-5). Open gray bars, normoxic WT; open white bars, normoxic myo-/-; hatched gray bars, hypoxic WT; hatched white bars, hypoxic myo-/-.

Fig. 2.

Left ventricular function of normoxic and hypoxic WT and myo-/- hearts. A: end-diastolic volume. B: end-systolic volume. C: stroke volume. D: ejection fraction. Experimental conditions are the same as in Fig. 1. The left ventricular function was measured using magnetic resonance imaging (MRI, n = 3-4). Hypoxic animals were supplied with 10% O2 during measurement. Open gray bars, normoxic WT; open white bars, normoxic myo-/-; hatched gray bars, hypoxic WT; hatched white bars, hypoxic myo-/-. *P < 0.05; **P < 0.01.

Blood cell parameters. Two weeks of hypoxia increased hematocrit drastically from 44% (WT) and 48% (myo-/-) to 71.5% in both groups (Fig. 3A). Erythrocytes rose from 8.4 × 106/μl to 12.5 and 11.7 × 106/μl in WT and myo-/- hearts, respectively (Fig. 3B). Surprisingly, the absolute and relative reticulocyte count was lower in myo-/- animals both under normoxic and hypoxic conditions (Fig. 3C). Analysis of the erythrocyte volume [mean corpulscular volume (MCV)] revealed a small but significant increase in MCV of myo-/- mice (Fig. 3D). Hemoglobin values were 12.6 ± 0.6 and 13.7 ± 1.3 g/dl in normoxic WT and myo-/- mice, respectively (data not shown). After 2 wk hypoxia, hemoglobin was increased significantly to 20.6 ± 1.9 and 20.5 ± 0.5 g/dl in WT and myo-/- mice, respectively. Platelet count was not statistically different between normoxic WT and myo-/- animals (827 ± 248 and 674 ± 143 × 103/μl, respectively). However, chronic hypoxia decreased thrombocytes significantly both in WT and myo-/- mice (405 ± 45 and 349 ± 36 × 103/μl, respectively).

Fig. 3.

Blood cell count of normoxic and hypoxic WT and myo-/- mice. A: hematocrit. B: erythrocytes. C: reticulocytes. D: mean corpuscular volume. Experimental conditions are the same as in Fig. 1 (n = 4-5). Relative number of reticulocytes are expressed in % of erythrocytes. Open gray bars, normoxic WT; open white bars, normoxic myo-/-; hatched gray bars, hypoxic WT; hatched white bars, hypoxic myo-/-. *P < 0.05, **P < 0.01, ***P < 0.001.

Gene expression. To analyze whether WT or myo-/- mice differ in gene expression, cDNA analysis was performed using Clontech mouse 1.2 arrays I and II. Of 2,352 investigated genes, 612 could be reliably detected and measured. When comparing cardiac gene expression in myo-/- vs. WT mice under normoxic conditions, there are 15 differently expressed genes (Table 1). Note that the observed changes were generally small, but keratinocyte lipid-binding protein was increased by 425%. However, the heart-specific fatty acid-binding protein (FABP) was not significantly altered in myo-/-mice (data not shown). Note that cytochrome c oxidase Vb subunit was increased by 34%. When myo-/- mice were stressed by hypoxia, we found 17 genes differently expressed compared with the hypoxic WT response (Table 2). Again, the keratinocyte lipid-binding protein and cytochrome c oxidase Vb were increased by 202 and 41%, respectively, in hypoxic myo-/- compared with the respective WT control. Known hypoxic markers like HIF-1α were found to be unchanged both in hypoxic myo-/- and WT. To validate cDNA array data, we performed semiquantitative real-time RT-PCR on selected genes. As internal standard for RT-PCR, we used actin, which was found to be unaltered between the different groups when cDNA arrays were normalized to the total signal intensity of all genes on one array. Similar to the array data of normoxic hearts, we found cytochrome c oxidase Vb and keratinocyte lipid-binding protein in myoglobin knockout mice to be increased by 50 and 57%, respectively. PCR analysis of HIF-1α mRNA expression also revealed no significant differences between WT and myo-/- mice.

View this table:
Table 1.

Differences in cardiac gene expression between myoglobin knockout and wild-type mice during normoxia

View this table:
Table 2.

Differences in cardiac gene expression between hypoxic myoglobin knockout and wild-type mice

Protein expression. Aside from the recently described differences of enzymes in fatty acid metabolism between normoxic WT and myo-/- hearts (11), there were no additional altered proteins between hypoxic WT and hypoxic myo-/- mice. However, Fig. 4 shows cardiac proteins that were significantly altered in hypoxic hearts compared with normoxic controls. Three protein spots exhibited reduced expression under hypoxia when comparing either hypoxic WT with normoxic WT or hypoxic myo-/- with normoxic myo-/- mice. Two spots were identified by mass spectrometry as two isoforms of heart FABP with shifted isoelectric point (Fig. 4, A and B) and the third spot as heat shock protein 27 (HSP27; Fig. 4C). Heart FABP was reduced under hypoxic conditions by 30 and 36% in the WT and the myo-/-, respectively. HSP 27 decreased by 59% (WT) and 55% (myo-/-).

Fig. 4.

Differences in cardiac protein expression between hypoxic and normoxic myoglobin knockout and wild-type mice. A: heart fatty acid-binding protein I (H-FABP I). B: H-FABP II. C: heat shock protein 27 (HSP27). Experimental conditions are the same as in Fig. 1. Cardiac protein expression was determined using 2-dimensional gel electrophoresis followed by mass spectrometry. Data are expressed as mean normalized absolute values (n = 5). Note that 2 of the differentially regulated protein spots represent 2 isoforms (I + II) of H-FABP with different isoelectric point. Open gray bars, normoxic WT; open white bars, normoxic myo-/-; hatched gray bars, hypoxic WT; hatched white bars, hypoxic myo-/-. *P < 0.05, **P < 0.01, ***P < 0.001.


The major finding of the present study is that myoglobin knockout mice are extremely robust and tolerate well an extended period of hypoxic stress. Apparently, the compensatory mechanisms recently reported by us and others (6, 16) are fully adequate to permit myoglobin-deficient mice to cope with 10% oxygen, similar to wild-type controls. Furthermore, a detailed genomic and proteomic analysis of myo-/- and WT hearts revealed no major divergence, which could explain the known structural differences between the two groups.

Hearts of myoglobin knockout mice are not functionally compromised in that cardiac energetics and oxygen consumption under control conditions and after β-adrenergic stimulation do not differ from wild-type controls (6). This was confirmed in the present study for the living animal: various hemodynamic parameters (Fig. 1), stroke volume, and ejection fraction measured by MRI (Fig. 2) were not different between the two groups under control conditions. Note, however, that the reticulocyte count was lower, together with a small but significant increase in mean volume of red blood cells (Fig. 3), suggesting a slightly perturbed erythrocyte formation in myo-/- mice.

Myo-/- hearts were reported to exhibit an increased coronary flow, coronary reserve, and hematocrit, as well as a significant increase in capillary density (6, 15), which together steepen the diffusion gradient for oxygen from the capillary to the mitochondria, thereby permitting adequate delivery of oxygen despite the lack of myoglobin. More recently, we have shown that myoglobin serves two additional functions: it is an important intracellular scavenger of NO (4) and can effectively detoxify free oxygen radicals formed during ischemia/reperfusion (3). Thus, from a biochemical point of view, loss of cardiac myoglobin is characterized by problems in oxygen delivery together with elevated nitrosative and oxidative stress. All these factors are known to influence gene expression (1, 14, 19, 21), which might have been responsible for the observed phenotype. However, the analysis of 2,352 genes revealed only minute differences between the two groups (Table 1). Noteworthy is that the keratinocyte lipid-binding protein, also known as epidermal type FABP, which belongs to the multigene family of intracellular FABPs (18), was more than fourfold elevated (Table 1). The mRNA of this protein was also twofold higher in hypoxic myo-/- hearts compared with hypoxic WT controls (Table 2).

The free fatty acid uptake by the heart involves high-affinity binding to membrane proteins and passive flip-flop across the phospholipid bilayer followed by intracellular trafficking via cytosolic FABP to the mitochondria, the site of metabolic disposition. We recently made the unexpected observation that several enzymes of the β-oxidation pathway of free fatty acids are substantially reduced in myo-/- hearts, suggesting a metabolic shift from fatty acid to glucose utilization when myoglobin is lacking (11). The present study extends this observation: at the mRNA level, the keratinocyte FABP was increased in myo-/- hearts both under normoxia and hypoxia. In addition the adipocyte FABP mRNA was upregulated in myo-/- under normoxia (Table 1). Taken together these findings may be interpreted to indicate a compensatory upregulation of two FABPs (keratinocyte and adipocyte FABP) under conditions of reduced β-oxidation. Whereas RT-PCR analysis confirmed the different expression levels found by the array analysis, it is possible that due to a large type II error some differently expressed genes have been overlooked by the screening procedure. Thus it is possible that the number of genes with minor but relevant changes in expression levels is larger than assumed on the basis of our array analysis.

It is surprising how well adapted myo-/- mice are even under conditions of prolonged hypoxic stress. We found no differences in heart rate, mean arterial pressure, and stroke volume; also hematocrit values increased to the same extent in WT compared with the myo-/- mice. Although we could not detect an increased mortality of myo-/- mice during gestation (see below), exposure to hypoxia in utero could make myo-/- mice rather robust for a later hypoxic stimulus. Under the hypoxic conditions chosen, WT and myo-/- mice equally well reached a new steady state. However, we cannot exclude the possibility that under conditions of a more severe hypoxic stress, a reduced potential to adapt and a higher mortality could be observed in myo-/- mice.

At the protein level, we found that hypoxia caused a decrease in the concentration of heart FABP and HSP27 (Fig. 4). At the mRNA level, heart FABP was also found to be decreased but did not reach the level of significance (HSP27 was not present on the gene array). The additional differences in fatty acid metabolism recently reported (11) remained unaffected during hypoxia.

In the course of our MRI measurements, we made the observation that myo-/- mice anesthetized with isoflurane exhibited a reduced heart rate when subjected to hypoxia. However, when mice were anesthetized with urethane, there was no effect on heart rate. Similarly, when a conscious myo-/- mouse instrumented by telemetry was kept at 10% O2 there was no effect on heart rate after 2 wk hypoxia (unpublished observation). Therefore isoflurane must have altered the heart rate under the specific condition when myoglobin was lacking and numerous compensatory mechanisms were activated. Because isoflurane is widely used in MRI experiments (the anesthesia can conveniently be controlled), caution should be exercised when working with transgenic mice in which a defined genetic intervention may have altered the response of the entire system.

There are numerous genes that are known to be induced by hypoxia or nitrosative stress (1, 14, 19, 21). Surprisingly, however, the expression of such a prominent protein as HIF-1α is not altered by 2 wk of hypoxia either in myo-/- or in WT animals as confirmed by PCR. Similarly, Ladoux and Frelin (10) reported unchanged HIF-1α mRNA levels in hypoxia, while others observed an increase in HIF-1α when rats were exposed to hypoxia for 3 wk (9). Possibly HIF-1α protein expression peaks early after the hypoxic perturbation and then declines to low expression levels (20). Alternatively, it is conceivable that HIF-1α is regulated at the post-mRNA level (22). However, we did not find evidence for this in our proteome studies. Another explanation for our results might be that the increased hematocrit and capillarity in myo-/- hearts might have buffered the hypoxic stimulus so well that capillary Po2 was unaffected. However, only a detailed time course study of the expression of HIF-1α together with tissue Po2 measurements can resolve this question.

That mice lacking myoglobin have no obvious phenotype was already surprising (5, 6); that these mice show no impairment of cardiac function even when stressed by severe hypoxia as in the present study came quite unexpected. Contrary to our findings, Garry and colleagues (13, 16) reported that normoxic myo-/- mice show an induction of HIF-1α gene expression and display ventricular dysfunction in response to hypoxia. Additionally, Garry and coworkers (16) could not detect a change in fatty acid utilization between WT and myoglobin knockout mice. Moreover, they have reported that most embryos succumb to heart failure at midgestation (16), whereas our group did not find an increased incidence of malformation or death during gestation. These divergent findings are likely due to the different genetic background used to generate myo-/- mice. While Garry et al. (5) used a C57BL/6 genetic background (personal communication), we generated myo-/- mice on a NMRI background (6). The surprising observation of a decreased difference in capillarity between WT and myo-/- mice in this study in contrast to our previous study (6) may also be related to a drift in genetic background (NMRI vs. mixed NMRI and 129 background, respectively).

Phenotype variation depending on different genetic background has been reported (2, 8). Dansky et al. (2) showed that mice lacking apolipoprotein E (ApoE) displayed a different degree of atherosclerosis depending on the mouse strain. ApoE-deficient mice on C57BL/6 background had fivefold increased atherosclerotic lesions despite lower serum cholesterol levels compared with the FVB/N strain. Additionally, mice lacking the mitochondrial superoxide oxidase (Mn SOD2) showed a different phenotype depending on the background (8). On CD1 background, SOD2-deficient mice showed cardiomyopathy and died within the first week of life, but on a different background (129/Sv and C57BL/6), these transgenic mice displayed rather neurological disorders and longer survival. These experiments indicate that genetic background can profoundly affect the phenotype and imply the existence of genetic modifiers. Therefore different genetic regulatory networks compensating for loss of myoglobin could also explain these divergent findings.

No-phenotype knockouts have puzzled biologists for many years, and the phenomenon is often described as genetic buffering due to redundancy of regulatory networks (7). Little is known, however, about the principles of gene interactions and to which extent changes in genetic background alter the phenotypic appearance. Because myoglobin is a major cardiac protein constituting ∼1% of the cardiac proteome and its loss is without functional consequences, myo-/- mice might constitute a suitable model to study in more detail gene and protein interaction in a mammalian system which brings about this robustness. One also should be aware, however, that with present state of the art genomic and proteomic analysis, we do not get a complete overview of the mouse genome or transcriptome. Two-dimensional gel electrophoresis usually can resolve about 600-800 proteins in view of probably more than 200,000 protein species in a typical cell. Due to sensitivity problems, low abundant proteins such as transcription factors are likely to go undetected. Similarly, present gene arrays cover about one fourth of the entire information present in the mouse genome.


This work was supported by the Biological-Medical Research Center (Biologisch-Medizinisches Forschungszentrum) of the Heinrich-Heine-University Düsseldorf.


We thank D. Haubs for excellent technical assistance. We also thank Dr. K. Zanger for help with histological procedures and Dr. S. Metzger for protein identification by ESI-MS/MS.


  • 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.


View Abstract