Oxygen tension is critical for proliferation of human and murine midbrain-derived neural precursor cells (mNPCs). Here, we conditionally inactivated the hypoxia-responsive transcription factor hypoxia-inducible factor-1α (HIF-1α) in murine NPCs to determine its role in proliferation, survival, and dopaminergic differentiation in vitro as well as survival of murine dopaminergic neurons in vivo. HIF-1α conditional knock-out (HIF-1α CKO) mNPCs showed midbrain-specific impairment of survival and proliferation. Dopaminergic differentiation of HIF-1α CKO mNPCs in vitro was markedly reduced. Expression of vascular endothelial growth factor (VEGF) mRNA was reduced in HIF-1α CKO mNPCs, whereas erythropoietin signaling was not affected. Treatment of HIF-1α CKO mNPCs with 50 ng/ml VEGF partially recovered proliferation and dopaminergic differentiation in vitro. In substantia nigra (SN) of adult HIF-1α CKO mice, protein levels of dopaminergic marker molecules such as tyrosine hydroxylase (TH) and aldehyde dehydrogenase were reduced by 41 and 61%, respectively. The cell survival marker Bcl-2 was reduced by 58% while caspase-3 was activated. Nonbiased stereological cell counts of TH-positive neurons in SN of young adult HIF-1α CKO mice revealed a reduction of 31% compared with cre/wt mice (in which the wild- type Hif1a allele is expressed in parallel with the Cre recombinase allele). However, we found no impairment of striatal dopamine concentrations or locomotor behavior. In conclusion, HIF-1α seems to be a transcription factor relevant to the development and survival of substantia nigra dopaminergic neurons involving VEGF signaling.
Neural precursor cells (NPCs) offer a great promise for developing new medical treatments for disorders such as Parkinson's disease (PD) (Gage, 2003; Alvarez-Buylla and Lim, 2004). It is necessary to understand the mechanisms governing cell proliferation, dopaminergic differentiation, and senescence of NPCs to facilitate clinical therapies (Sharpless and DePinho, 2004). Reduced oxygen tension is now recognized as a common requirement for successful expansion and dopaminergic differentiation of NPCs (Ivanovic et al., 2000; Morrison et al., 2000; Studer et al., 2000; Storch et al., 2001; Milosevic et al., 2005). Low, physiologic oxygen conditions help maintain the undifferentiated state in mouse NPCs, but the molecular mechanisms underlying oxygen effects on NPCs are essentially unknown (Milosevic et al., 2005).
Many O2-responsive genes are regulated via hypoxia-inducible factor-1 (HIF-1), a heterodimer consisting of HIF-1α and HIF-1β (also called aryl hydrocarbon receptor nuclear translocatorm, or ARNT) (Semenza, 1999; Wenger, 2002). HIF-1α abundance is regulated by proteasomal degradation, and its activity is regulated by diverse posttranslational modifications. HIF-1α target genes include vascular endothelial growth factor (VEGF), a potent inducer of angiogenesis, and erythropoietin (EPO). Both have pleiotropic effects in the brain, including neurogenesis and neuroprotection (Sakanaka et al., 1998; Junk et al., 2002; Ferriero, 2005; Greenberg and Jin, 2005), as well as substantial regenerative effects on dopaminergic neurons (Studer et al., 2000; Yasuhara et al., 2005).
HIF-1α is essential for the early development of many mammalian organs including brain (Ryan et al., 1998; Tomita et al., 2003). Mice lacking HIF-1α show severe cardiac and vascular malformations, leading to embryonic lethality at approximately embryonic day 10.5 (E10.5) (Iyer et al., 1998). To study the impact of HIF-1α on NPCs and, in particular, dopaminergic differentiation of midbrain-specific NPCs (mNPCs), we conditionally knocked out HIF-1α exon 2 in NPCs using Nestin promoter-driven expression of Cre recombinase (Betz et al., 1996; Rajewsky et al., 1996). Exon 2 encodes the motif necessary for dimerization of HIF-1α and HIF-1β, binding of the HIF-1 dimer to DNA, and its function as a transcription factor (Jiang et al., 1996). Our findings indicate that HIF-1α promotes stem cell survival, growth, and dopaminergic differentiation, as well as survival of adult dopaminergic neurons.
Materials and Methods
Conditional inactivation of HIF-1α in NPCs.
A Cre/loxP system was used to generate HIF-1α-deficient NPCs (Le and Sauer, 2000; Nagy, 2000). Mice containing loxP-flanked Hif1a exon 2 (Hif1aloxP/loxP designated as Hif1a-floxed mice) were kindly provided by R. S. Johnson (San Diego, CA). The strategy used to obtain loxP-flanked Hif1a locus instead of the endogenous Hif1a locus was described previously (Ryan et al., 2000) (Fig. 1A). Hif1a-floxed targeted mice were crossed with mice expressing Cre recombinase under the control of the Nestin promoter, Nestin–Cre transgenic mice (JAXMICE; The Jackson Laboratory, Bar Harbor, ME). Nestin–Cre strain of mice was maintained in a C57BL/6 × SJL and Hif1a-floxed in a C57BL/6 background. Animals heterozygous for both Hif1a-floxed and Nestin–Cre alleles were mated to Hif1a-floxed homozygous animals to generate Nestin–Cre, Hif1aloxP/loxP (HIF-1αΔ) mice. To distinguish HIF-1α conditional knock-out (CKO) embryos/mice, a genomic PCR-based genotyping was applied using the following primers: floxed or wild-type (wt) HIF-1α allele, 5′-GCA GTT AAG AGC ACT AGT TG-3′ and 5′-GGA GCT ATC TCT CTA GAC C-3′, 250 bp product (floxed) or 200 bp product (wild type); Cre recombinase primers, 5′-CCT GGA AAA TGC TTC TGT CCG-3′ and 5′-CAG GGT GTT ATA AGC AAT CCC-3′, 400 bp product (Fig. 1C). NPCs were dissected from frontal (fNPCs) or mesencephalic brain regions of HIF-1α null embryos carrying the HIF-1α conditional mutation (HIF-1α CKO fNPCs and HIF-1α CKO mNPCs, respectively) as described below. Removal of HIF-1α exon 2 by Cre-induced recombination in NPCs was confirmed using the following primers: 5′-GCA GTT AAG AGC ACT AGT TG-3′ and 5′-TGT TAA ATA AAA GCT TGG AC-3′, amplifying a 650 bp fragment (Fig. 1D).
To exclude possible toxic effects of the Cre recombinase that could influence proliferation of NPCs, in all of our in vitro experiments, we compared HIF-1α CKO NPCs with the cells containing the wt Hif1a allele expressed in parallel with the Cre allele, referred to as cre/wt NPCs (Pfeifer et al., 2001).
Isolation, propagation, and differentiation of NPCs.
Mesencephalic and frontal (cortical) NPCs were dissected from mice at E14. Pregnant females were killed according to National Institutes of Health guidelines and the approval of the local animal care committee. Tissue samples were incubated in 0.1 mg/ml papain (Roche, Mannheim, Germany)/DNase solution (100 μg/ml; Roche) for 30 min at 37°C, rinsed in PBS, incubated in antipain (50 μg/ml; Roche) for 30 min at 37°C, and finally homogenized by gentle triturating using a fire-polished Pasteur pipette. The cells were expanded as a monolayer culture by plating onto polyornithine–fibronectin precoated dishes in a density of 20,000 cells/cm2 (Milosevic et al., 2005). The cells were maintained in serum-free DMEM (high glucose)/F-12 mixture (1:1) medium and supplemented with 20 ng/ml human recombinant epidermal growth factor and 20 ng/ml basic fibroblast growth factor (both from PromoCell, Heidelberg, Germany). NPCs were expanded in 3% oxygen (Storch et al., 2001; Milosevic et al., 2005).
Differentiation of NPCs was induced using defined media without mitogens but with 1% FCS and 5 μm forskolin (Sigma-Aldrich, Munich, Germany). Before electrophysiology, immunocytochemistry, or protein extraction, mNPCs were allowed to differentiate for 1 week. Electrophysiological analysis of mNPCs was also performed after a 20 d differentiation period.
In vitro VEGF administration.
HIF-1α CKO and cre/wt mNPCs were treated in vitro with 50 ng/ml recombinant human VEGF (PAN Biotech, Aidenbach, Germany), applied once for 7 d of proliferation and once during consecutive 7 d differentiation period. Values obtained for treated cultures were normalized to untreated ones.
Clonogenic survival assay.
Midbrain-derived or frontal NPCs were seeded in triplicate onto polyornithine–fibronectin precoated six-well plates (1 × 103 cells per well in 3 ml of culture medium). Two weeks after incubation at 37°C in a reduced oxygen (3%) atmosphere, the colonies were fixed in 70% ethanol and stained with 0.5% toluidine blue solution. The number of colonies obtained for cre/wt control NPCs was considered as 100%, and colonies generated by HIF-1α CKO NPCs were normalized to the control cells.
Mouse substantia nigra (SN) extracts or NPCs extracts were prepared as described previously (Milosevic et al., 2005). Proteins (50–100 μg) were separated by 12% SDS-PAGE and transferred to nitrocellulose membranes. Antibodies used to probe blots were as follows: mouse monoclonal anti-proliferating cell nuclear antigen (PCNA) (Santa Cruz Biotechnology, Santa Cruz, CA); mouse monoclonal anti-Bcl-2 (Santa Cruz Biotechnology), mouse monoclonal anti-pro-caspase-3 (BD Transduction Laboratories, San Jose, CA); rabbit polyclonal anti-activated caspase-3 (CM1; BD PharMingen, San Diego, CA); mouse monoclonal anti-VEGF (Abcam, Cambridge, UK); chicken polyclonal anti-aldehyde dehydrogenase (ALDH1A1) (kindly provided by N. E. Sladek, Minneapolis, MN); mouse monoclonal anti-actin (MP Biomedicals, Eschwege, Germany); mouse monoclonal anti-glial fibrillary acidic protein (GFAP) (Chemicon, Hampshire, UK); rabbit polyclonal anti-HIF-1α (Cayman Chemical, Ann Arbor, MI); rabbit polyclonal anti-β-tubulin III (anti-TUJ1; Covance, Richmond, CA); and horseradish peroxidase-conjugated secondary antibodies (Pierce, Rockford, IL).
NPCs were fixed with 4% paraformaldehyde in PBS for 10 min at room temperature, washed with PBS, and counterstained with the DNA-binding dye 4′-6-diamidino-2-phenylindole (2 μg/ml in PBS) for 15 min at room temperature, twice washed in PBS followed by incubation in blocking buffer (10% FCS and 0.2% Triton X-100 in PBS, pH 7.2) for 30 min at room temperature. After incubation with the primary antibody (1 h at room temperature in blocking buffer), the cells were incubated with fluorescent secondary antibodies, Alexa Fluor 488 conjugate or Alexa Fluor 594 conjugate (Invitrogen, Carlsbad, CA). Coverslips were mounted onto glass slides and examined under a fluorescence microscope (Axiovert 200; Zeiss, Jena, Germany). Acquisition of the cells was performed using the image-analysis software AxioVision 4 (Zeiss, Jena, Germany). The following primary antibodies were used for immunofluorescence: mouse monoclonal anti-nestin (BD Biosciences, San Jose, CA), mouse monoclonal anti-Cre recombinase (Chemicon), and rabbit polyclonal anti-β-tubulin III (anti-TUJ1; Covance).
Activated caspase-3 localization in the SN was investigated by confocal laser scanning microscopy (LSM 510; Zeiss, Oberkochen, Germany) at an excitation wavelength of 594 nm (helium/neon, red Alexa 594 immunofluorescence) and 488 nm (argon, yellow–green Alexa 488 immunofluorescence).
Immunohistochemistry and stereological cell counts.
After being fixed by cardiac perfusion (4% paraformaldehyde in phosphate buffer) both HIF-1α KO and cre/wt mice brains were postfixed for 2 h and dehydrated in 15% sucrose and 30% sucrose. Frozen brain samples were cut (40 μm) in a cryostat; 10–20 serial sections from SN levels were collected free floating in ice-cold PBS. Double immunostaining was performed with antisera for tyrosine hydroxylase (TH) (goat polyclonal anti-TH antibody, 1:100; Santa Cruz Biotechnology) and anti-activated caspase-3 (CM1, 1:2000; BD PharMingen), followed by incubation with fluorescent secondary antibodies (1:500, Alexa 488 or Alexa 594; Invitrogen). For stereology, groups of three to four transgenic mice of the same age (4 weeks old) were perfused as described above. TH-positive cells in the SN from HIF-1α CKO and cre/wt mice were selected for nonbiased quantitative stereology. Tissue sections, 40 μm thick (8–12 slices), were chosen for analysis using systematic random sampling. The SN sections were immunostained with a rabbit antibody to TH (rabbit polyclonal; Santa Cruz Biotechnology) at a 1:500 dilution, biotinylated goat anti-rabbit secondary antibody, streptavidin–HRP (Vector Laboratories, Burlingame, CA), and DAB (Sigma-Aldrich). Before dehydration of floating slices, they were processed for Nissl (30 s in 0.5% toluidine blue in PBS) to facilitate counting. The intersection interval, counting frame size, and distance between counting frames were adjusted so that, whenever possible, a reasonable number of TH-stained cells were sampled. The optical fractionator method was used to provide an unbiased estimation of the total number of dopaminergic neurons in the region of interest. Stereologic counting and estimates were done with the aid of Stereoinvestigator version 5.05.1 (MicroBrightField, Magdeburg, Germany) (Orb et al., 2004).
Patch-clamp analysis of mNPCs differentiated in vitro for 7 or 20 d was performed at room temperature using an inverted microscope DMIL (Leica, Bensheim, Germany) and an EPC-9 amplifier (HEKA Elektronik, Lambrecht, Germany). Recordings of voltage-gated ion channels were obtained in the whole-cell voltage-clamp mode by stepwise depolarizations with increasing amplitudes from the holding potential of −70 to 50 mV in steps of 10 mV. Series resistance values were continuously measured during all recordings. The external bath solution contained the following (in mm): 142 NaCl, 1 CaCl2, 8 KCl, 6 MgCl2, 10 glucose, and 10 HEPES, pH 7.4 (320 mOsm). Micropipettes were formed from thin-walled borosilicate glass (Science Products, Hofheim, Germany) with a Flaming Brown electrode puller P-97 (Sutter Instruments, Novato, USA) and a Micro Forge (Narishige, Tokyo, Japan). Electrodes had resistances of 2–4 MΩ when filled with the internal solution containing the following (in mm): 153 KCl, 1 MgCl2, 10 HEPES, 5 EGTA, and 2 MgATP, pH 7.3 (305 mOsm). Whole-cell currents were low-pass filtered at 2–5 kHz, digitized at 10 kHz, and analyzed using PulseFit software (HEKA Elektronik) and Prism 4 (GraphPad Software, San Diego, CA). Data were expressed as mean ± SEM. Statistical significance was considered at p < 0.05 (Student's t test, two-tailed, unpaired).
Determination of dopamine and metabolites.
For determination of dopamine, norepinephrine, and their metabolites, striatal tissue was dissected and immediately frozen in liquid nitrogen. Five striata of each genotype were processed. Dopamine, norepinephrine, 3,4-dihydroxyphenylacetic acid (DOPAC), and homovanillic acid were analyzed using standard HPLC techniques essentially as described previously (Gerlach et al., 1996).
Sensorimotor performance of HIF-1α CKO compared with cre/wt mice was evaluated using the beam-walk paradigm. Mice were trained to pass a beam for 4 consecutive days. We used a wooden beam with a length of 1 m, a diameter of 4 mm, and a rough surface to prevent slipping. Mice were motivated via the smell of the food within the target cage. On the fifth day, 10 trials per mouse were analyzed.
Spontaneous locomotion was measured for 30 min using an open-field setup. Transitions from one quadrant to another were scored. All open-field experiments were videotaped and analyzed subsequently by a blinded rater (J.L.).
RNA extraction and quantitative real-time reverse transcription-PCR analysis.
Total cellular RNA was extracted from NPCs using RNAeasy total RNA purification kit followed by treatment with RNase-free DNase (Qiagen, Hilden, Germany). Semiquantitative real-time one-step reverse transcription (RT)-PCR was performed using the Stratagene system (MX3000P; Stratagene, Heidelberg, Germany), and amplification was monitored and analyzed by measuring the binding of fluorescent SYBR Green I to double-stranded DNA. One microliter (50 ng) of total RNA was reverse transcribed and subsequently amplified using QuantiTect SYBR Green RT-PCR Master mix (Qiagen) and 0.5 μmol/L of both sense and antisense primers. The sequences for forward and reverse primers used for the target gene (TG) and the reference gene (RG) hydroxymethylbilane synthase are summarized in Table 1. The relative RNA content was determined using the formula of the comparative cycle threshold (Ct): TG/RG = 2Ct(RG) − Ct(TG) (Livak and Schmittgen, 2001). The efficiency of product formation by PCR was estimated from plots of Ct values versus serial dilutions, measured three times with different RNA samples.
Normally distributed data were subjected to statistical analyses as indicated (one- or two-way ANOVA) using the SigmaStat software package (Jandel, San Rafael, CA) and Prism 4. Results are expressed as the mean ± SEM. Statistical significance was considered at p < 0.05.
Generation and characterization of HIF-1α CKO mNPCs
Hif1a-floxed targeted mice were crossed with mice expressing Cre recombinase under the control of the Nestin promoter (Nestin–Cre mice) to obtain Nestin-Cre; HIF-1αloxP/loxP (HIF-1αΔ) mice referred to HIF-1α CKO mice (Fig. 1A,B). NPCs were dissected from both midbrain and frontal part of each embryonic brain and expanded in vitro. After genotyping, only NPCs containing the wt Hif1a allele that were cre-positive (cre/wt), as well as NPCs containing the Hif1α-floxed allele that were cre-positive (HIF-1α CKO) were kept in culture for additional analysis (Fig. 1C,D). HIF-1αΔ/Δ homozygous mutant myocyte enhancer factors (MEFs) and wt MEFs were used as a negative and positive control for the genotyping (Fig. 1D). Immunocytochemical staining confirmed expression of Cre recombinase in Nestin-positive NPCs (Fig. 1E).
Wild-type murine NPCs (cre/wt) and NPCs lacking exon 2 of the Hif1a gene, designated as HIF-1α CKO dissected from at least four different embryos were expanded in 3% oxygen and checked for expression of HIF-1α protein. In cre/wt NPCs, HIF-1α protein was stabilized in 3% oxygen revealing a 120 kDa band as seen by immunoblotting. This band was not present in HIF-1α CKO cells (Fig. 2A). As soon as successful Cre-mediated excision was confirmed in NPCs, these cells were considered as HIF-1α knock-outs.
HIF-1α is important for survival and proliferation of midbrain-derived NPCs
mNPCs generated from at least four different embryos were first expanded for 2 weeks in 3% oxygen and then split for additional expansion for 2 weeks, in either 20 or 3% oxygen. Protein extracts were probed with the proliferation marker PCNA and prosurvival marker Bcl-2. Compared with cre/wt cells, expression of both proteins was significantly reduced in HIF-1α inactive NPCs when they were expanded in 3% oxygen. As shown in Figure 2A, PCNA expression was reduced to 61 ± 13% (n = 4; p = 0.035, one-way ANOVA), whereas Bcl-2 levels were only 34 ± 11% in HIF-1α CKO mNPCs (n = 4; p = 0.015). When HIF-1α was eliminated by 20% oxygen, PCNA and Bcl-2 expression did not significantly differ in HIF-1α CKO mNPCs compared with cre/wt mNPCs (Fig. 2B).
Clonal assays confirmed the necessity of HIF-1α for survival and proliferation of mNPCs. Mesencephalic NPCs derived from HIF-1α CKO embryonic brains in 3% oxygen produced 20 ± 4% colonies normalized to 100% colonies formed by cre/wt NPCs (Fig. 2C,D).
HIF-1α is not necessary for survival and proliferation of frontal NPCs
To test region-specific effects of HIF-1α deletion in NPCs, we performed the same experimental procedure on frontal NPCs. Proliferation (PCNA protein expression) and survival (Bcl-2 protein expression) assessed by Western blotting did not reveal any significant differences in HIF-1α null compared with cre/wt NPCs in either 3 or 20% oxygen (Fig. 3A,B). Frontal NPCs derived from HIF-1α CKO embryonic brains in 3% oxygen produced a similar number of colonies compared with cre/wt samples (98 ± 3%) (Fig. 2D).
HIF-1α is important for in vitro neuronal and dopaminergic differentiation of midbrain-derived NPCs
Whole-cell extracts for Western blotting were acquired from NPCs after 4 weeks of proliferation and after 1 week of differentiation. Extracts were separated and probed with the neuronal marker (TUJ1), the glial marker (GFAP), or TH, a marker for dopaminergic neurons. As shown in Figure 4A, neuronal and glial markers were not significantly changed in fNPCs, whereas differentiated HIF-1α CKO mNPCs exhibited a significant reduction in TH expression, 35 ± 4% of wt (p < 0.001). Neuronal differentiation and morphology of NPCs from frontal cortex was not affected by HIF-1α deletion as confirmed by immunocytochemical analysis of frontal NPCs (Fig. 4B). In both KO and cre/wt differentiated fNPCs, extensive neuronal processes were observed. Furthermore, protein expression of TUJ1 and GFAP did not vary between two different cell types (Fig. 4C). However, TUJ1 staining on cultured mNPCs 1 week after differentiation in 3% oxygen showed an altered morphology in HIF-1α CKO mNPC compared with cre/wt NPCs. Cre/wt neurons were more mature, exhibiting longer neurites (Fig. 4D).
Electrophysiological analyses of voltage-gated ion channels showed significantly decreased maximal sodium inward currents of HIF-1α CKO mNPCs (116.9 ± 33.9 pA) 7 d after in vitro differentiation compared with cre/wt NPCs (318.2 ± 77.7 pA; n = 12 for each genotype; p = 0.032, t test), whereas no significant reduction was calculated for potassium outward rectifying currents (Fig. 5). After 20 d of mNPC differentiation, values for sodium and potassium currents were increased in both HIF-1α CKO and cre/wt cells without altering the difference between the two genotypes (data not shown). In contrast, there was no significant difference in maximal amplitudes of sodium and potassium currents between HIF-1α CKO and cre/wt frontal NPCs 7 d after in vitro differentiation (data not shown).
HIF-1α supports the development and/or survival of dopaminergic neurons
SN was dissected from 4-week-old HIF-1α CKO mice and their cre/wt littermates. At the time of analysis, these animals were macroscopically identical. Tissue extracts were analyzed via immunoblotting. Neuronal β-tubulin III (TUJ1) expression did not significantly differ between genotypes, whereas TH protein was reduced to 59 ± 5% in HIF-1α CKO mice (n = 5; p = 0.008, t test). The expression of ALDH1A1, another early marker of SN dopaminergic neurons, was reduced to 39 ± 19% (n = 5; p = 0.002). Prosurvival protein Bcl-2 was reduced to 43 ± 12% (n = 5; p = 0.03). Pro-caspase-3 was cleaved in HIF-1α CKO mice SN extracts, resulting in reduced levels of 20 ± 15% (n = 5; p = 0.008), but activated caspase-3 was increased because it was not detectable in extracts from cre/wt mice. VEGF protein was reduced in SN taken from HIF-1α CKO compared with cre/wt mice (Fig. 6A). Double-immunofluorescence labeling of brain slices in combination with confocal laser scanning microscopy indicated that the activated, cleaved form of the caspase-3 was predominantly present in dopaminergic (TH-positive) cells in the SN of HIF-1α CKO mice (Fig. 6B). In the SN of cre/wt mice, we could not find any cells positive for both caspase-3 and TH.
Finally, stereological non-biased counts of TH-positive cells in the SN of HIF-1α CKO mice (n = 4) showed a robust and significant reduction compared with cre/wt SN (total number of TH-positive neurons in the SN, 7489 ± 477 vs 10,817 ± 903, respectively, or 69 ± 6% of cre/wt) (Fig. 6C). However, the number of TH-positive cells in another brain region, ventral tegmental area (VTA), remained unaffected (n = 4) (Fig. 6C). Nissl staining of both cre/wt and HIF-1α CKO brain slices revealed no difference in the SN (50,096 ± 246 vs 48,913 ± 2006, respectively; n = 3), dentate gyrus (375,897 ± 61,210 vs 364,628 ± 30,580), and medial habenula (84,560 ± 6141 vs 79,318 ± 5740). In addition, biochemical measurements of dopamine, norepinephrine, and their metabolites (DOPAC and HVA) did not reveal any significant differences between genotypes (data not shown). These findings are well in line with other mouse models of Parkinson's disease, in which moderate losses of dopaminergic neurons within SN are readily compensated (Orb et al., 2004).
An evaluation of potential sensorimotor deficits using analysis of beam walking and spontaneous locomotion failed to show any difference between genotypes (data not shown). Open-field behavior of HIF-1α CKO and cre/wt mice was quantified for 30 min. Statistical analysis by two-tailed t test revealed no statistical differences between HIF-1α CKO and cre/wt mice (n = 7 of both genotypes).
Expression analysis of HIF-1α-regulated genes revealed VEGF and VEGF receptors as major downstream genes of HIF-1α in mNPCs
The expression of VEGF and VEGF receptor (VEGFR) subtypes as HIF-1α regulated genes was explored and quantified in expanded mNPCs and fNPCs by real-time RT-PCR. VEGF expression was reduced in HIF-1α CKO mNPCs compared with both cre/wt mNPCs and fNPCs (n = 7; p = 0.036) (Table 2, Fig. 7A) but showed no change between both fNPC subtypes. Expression levels of VEGFR-1 (Flt-1) were significantly higher in mNPCs compared with fNPCs independent of HIF-1α expression (Table 2). VEGFR-2 (Flk-1) expression was 2.7- to 3.8-fold increased in HIF-1α CKO compared with cre/wt NPCs in both mesencephalic and frontal NPCs and 31- to 44-fold higher in mNPCs compared with fNPCs independent of HIF-1α expression (n = 4; p = 0.018) (Table 2, Fig. 7A). VEGFR-3 (Flt-4) expression did not differ between HIF-1α CKO and cre/wt mNPCs. We did not detect differences in mRNA levels of the EPO receptor (EPOR) in HIF-1α CKO and cre/wt mNPCs and fNPCs (data not shown). Additionally, there were no detectable mRNA levels of EPO in all four cell types. Significantly lower expression of neuronal glucose transporter 3 (Slc2a3) in both fNPC subtypes and in HIF-1α CKO mNPCs compared with cre/wt mNPCs were seen (Table 2). In contrast, we did not detect significant differences in mRNA levels of various other non-neuronal HIF-1α-regulated genes with respect to the four NPC subtypes (Table 2).
VEGF stimulates proliferation and dopaminergic differentiation of HIF-1α CKO mNPCs in vitro
Because VEGF seemed to be reduced in HIF-1α CKO mNPCs and potentially mediate the effects of HIF-1α in these cells, we tried to rescue HIF-1α CKO mNPCs by supplementing culture media with VEGF. Proliferation, analyzed by PCNA expression, was significantly increased in HIF-1α CKO mNPCs after application of VEGF (50 ng/ml) (47 ± 5% in untreated vs 75 ± 4% in VEGF-treated cells; p < 0.05, two-way ANOVA). In addition, TH protein was elevated from 26 ± 5% of control in untreated HIF-1α CKO mNPCs to 68 ± 6% after VEGF treatment (n = 4; p < 0.05) (Fig. 7B,C). Proliferation of cre/wt cells did not significantly change after VEGF treatment. The prosurvival protein Bcl-2 and the neuronal marker TUJ1 were not altered by VEGF treatment in either cell type (Fig. 7B,C). Thus, only partial recovery of HIF-1α CKO mNPCs proliferation and differentiation was induced by VEGF in vitro administration.
Here we demonstrate for the first time that HIF-1α is important for proliferation, survival, and differentiation of murine tissue-specific midbrain-derived NPCs. Thus, HIF-1α may be an essential mediator with respect to the beneficial effects of lowered oxygen tension on NPCs (Studer et al., 2000; Storch et al., 2001; Milosevic et al., 2005). Adult HIF-1α-deficient mice also showed deficits of SN dopaminergic neurons, indicating that HIF-1α is not only important for proliferation and differentiation of mNPCs in vitro but may also mediate regeneration or survival in vivo.
Gas phase oxygen concentrations of 1–5% correspond to the physiological environment in embryonic tissue and adult brain (Silver and Erecinska, 1998). These “physiological” concentrations seem to be required for stem cell maintenance (Gustafsson et al., 2005; Covello et al., 2006). Recently, we demonstrated that “room air” damages murine midbrain-derived NPCs, initiating a variety of cellular events (Milosevic et al., 2005). However, it was not clear whether 3% oxygen affects NPC proliferation and differentiation via an HIF-1-dependent signaling pathway or via another mechanism.
In an attempt to elucidate the role of HIF-1α in 3% oxygen-stimulated survival, proliferation, and differentiation in murine NPCs, we conditionally targeted HIF-1α in NPCs using a Cre/loxP-based system (Ryan et al., 2000). Tomita et al. (2003) created mice with neural-cell-specific HIF-1α deficiency, exhibiting hydrocephalus, neuronal loss, and impairment of spatial memory. In the present paper, we focused on the midbrain dopaminergic system, investigating possible alterations of dopaminergic markers (ALDH1A1, TH, dopamine, and metabolites) and expression of major HIF-1α downstream proteins, such as VEGF and EPO. Deletion of HIF-1α in mNPCs had an impact on neuronal morphology in vitro accompanied by a significant reduction of sodium inward currents elicited by depolarizations, suggesting an altered morphological and functional differentiation compared with cre/wt mNPCs. In this regard, cortical NPCs remained unaffected. Midbrain-derived NPCs also showed decreased levels of TH after differentiation. In contrary to the major finding by Tomita et al. (2003), which was apoptosis in the cortical plate resulting in cortical atrophy in HIF-1α null mutant mice, we did not observe any significant changes regarding proliferation, survival, and neuronal differentiation in frontal (cortical) NPCs. In addition, we did not notice any alterations in cortical brain regions in 4-week-old mice, which, according to Tomita et al. (2003), at that age should have been visible. Our HIF-1α CKO mice were macroscopically indistinguishable from cre/wt mice but exhibited morphological alterations in the SN. However, our mice did not show major locomotor deficits or dopamine deficiency within striatal tissue, suggesting that the moderate loss of dopaminergic neurons at that age was well compensated. HIF-1α functional mutants developed by Tomita et al. (2003) were created also using Cre/loxP technology but excising exons 13–15 of the HIF-1α gene while we excised exon 2. HIF-1α mutant protein in the report by Tomita et al. (2003) lacked the transactivation domain, whereas our HIF-1α mutant mice lack the DNA binding domain. Thus, the HIF-1α mutant protein was either absent or functionally inert (Jiang et al., 1996; Ema et al., 1999). At this point, it is not clear whether the differences seen between two HIF-1α CKO mice originate as a consequence of a different genetic manipulation, differences in genetic background, or other reasons.
HIF-1α might be protective in some neurological disorders (Soucek et al., 2003). Neuroprotective effects of EPO on dopaminergic neurons suggested a possible positive effect of hypoxia via HIF-1α and subsequent EPOR expression (Demers et al., 2005). In line with recent studies suggesting a pivotal role of HIF-2α in regulating EPO expression (Chavez et al., 2006), we did not find relevant EPO expression in our cell system and no differences of EPOR expression in HIF-1α CKO versus cre/wt mNPCs, suggesting that EPOR signaling is not involved in HIF-1α-dependent survival, proliferation, and differentiation of mNPCs. Conversely, VEGF as another downstream target gene of HIF-1α acts as a direct neurotrophic or even neuroprotective factor (Matsuzaki et al., 2001; Sun et al., 2003). We detected a notable reduction in VEGF mRNA expression in HIF-1α CKO compared with cre/wt mNPCs and a most likely compensatory upregulation of VEGFR-2. SN of HIF-1α CKO mice expressed less VEGF protein compared with cre/wt. We further demonstrated that proliferation and dopaminergic differentiation of HIF-1α CKO mNPCs were partially recovered after in vitro administration of VEGF, indicating that other target genes besides VEGF are involved. The lack of HIF-1α coincided with apoptotic cell death in dopamine-producing, TH-positive cells in the SN of adult mice. Procaspase-3 was cleaved in the SN of HIF-1α CKO mice confirmed by in situ caspase-3 activation. Expression of the anti-apoptotic protein Bcl-2 in SN of HIF-1α CKO mice was decreased, likely as a consequence of an effector caspase activation (Milosevic et al., 2003). However, HIF-1α CKO mice exhibited absence of gross locomotor and sensorimotor behavioral deficits, which is in line with other mouse models with moderate dopaminergic deficits (Masliah et al., 2000; Orb et al., 2004).
HIF-1α activates the expression of hypoxia-inducible genes that contain a hypoxia response element located in the promoter or enhancer regions. Hypoxia induces TH mRNA expression in rat mesencephalic cultures (Leclere et al., 2004), whereas HIF-1α contributes to induction of TH transcription in PC12 cells (Schnell et al., 2003). Our findings indicate that HIF-1α represents an important factor for in vitro neuronal and dopaminergic differentiation, as well. Conditional knock-out of HIF-1α affected SN neurons in young adult mice. In addition to the decline of TH expression, we observed a reduction in another neuronal marker enriched in SN, ALDH1A1. ALDH1A1 is shown to be expressed in A9 dopaminergic neuronal group, the most vulnerable site in PD (Chung et al., 2005). We showed a prominent reduction in the number of dopaminergic (TH-positive) neurons in the SN of HIF-1α-deficient mice. Many animal models with degeneration of dopaminergic neurons show specificity for dopaminergic neurons in SN compared with VTA (Liss et al., 2005; Maingay et al., 2006). Accordingly, HIF-1α-deficient mice have no deficit in A10 dopaminergic neurons. Moreover, as revealed by Nissl staining, other neuronal types (e.g., GABAergic neurons in gyrus dentatus or SN, catecholaminergic neurons in medial habenula) likely also remained unaffected in HIF-1α CKO, because examined brain regions did not exhibit difference in neuronal cell numbers when compared with cre/wt mice.
Our data represent first evidence for a role of HIF-1α for dopaminergic development and survival. We also identified an important downstream gene: VEGF. However, because VEGF supplementation only partly antagonizes the lack of HIF-1α, future studies need to identify other downstream targets of HIF-1α that mediate this effect and may therefore represent potential drug targets for regeneration or protection of SN dopaminergic neurons, the cell type selectively affected in Parkinson's disease.
This work was supported in part by Interdisziplinäres Zentrum für Klinische Forschung Leipzig Grant TP C27 and the Research Program of the Medical Faculty Carl Gustav Carus of the Technical University of Dresden. We thank Ute Roemuss, Annett Brandt, Sylvia Kanzler, Rainer Burger, and Sabine Gehre for stem cell preparation and/or excellent technical assistance. We thank Dr. Randall S. Johnson for providing HIF-1α-floxed mice.
- Correspondence should be addressed to Dr. Javorina Milosevic, Department of Neurology, Max-Bürger-Forschungszentrum, Johannisallee 30, 04103 Leipzig, Germany.