Abnormal Sympathoadrenal Development and Systemic Hypotension in PHD3-/- Mice

Cell culture studies have implicated the oxygen-sensitive hypoxia-induciblefactor (HIF) prolyl hydroxylase PHD3 in the regulation of neuronalapoptosis. To better understand this function in vivo, we havecreated PHD3^–/– mice and analyzed the neuronal phenotype.Reduced apoptosis in superior cervical ganglion (SCG) neuronscultured from PHD3^–/– mice is associated with anincrease in the number of cells in the SCG, as well as in theadrenal medulla and carotid body. Genetic analysis by intercrossingPHD3^–/– mice with HIF-1a^+/– and HIF-2a^+/–mice demonstrated an interaction with HIF-2 ${alpha}$ but not HIF-1 ${alpha}$ , supportingthe nonredundant involvement of a PHD3-HIF-2 ${alpha}$ pathway in theregulation of sympathoadrenal development. Despite the increasednumber of cells, the sympathoadrenal system appeared hypofunctionalin PHD3^–/– mice, with reduced target tissue innervation,adrenal medullary secretory capacity, sympathoadrenal responses,and systemic blood pressure. These observations suggest thatthe role of PHD3 in sympathoadrenal development extends beyondsimple control of cell survival and organ mass, with functionalPHD3 being required for proper anatomical and physiologicalintegrity of the system. Perturbation of this interface betweendevelopmental and adaptive signaling by hypoxic, metabolic,or other stresses could have important effects on key sympathoadrenalfunctions, such as blood pressure regulation.

INTRODUCTION

Top

INTRODUCTION

In response to low oxygen tensions, organisms mount a wide-rangingadaptive response involving many cellular and systemic processes.Activation of hypoxia-inducible factor (HIF) plays a centralrole in this process, inducing transcriptional targets thatenhance oxygen delivery, better adapt cells to hypoxia, or modulatecell proliferation or survival pathways (reviewed in references16 and 37). Hypoxia may, under different circumstances, eitherpromote or protect cells from apoptosis, and HIF itself contributesto these processes both indirectly, through the defense of cellularenergy supplies, and directly, via transcriptional changes inproapoptotic or prosurvival genes. However, to generate theanatomical and physiological integrity required for oxygen homeostasisin the intact organism, these adaptive responses to hypoxiamust also be accurately interfaced with the developmental controlof growth.

Though the nature of these adaptive-developmental connectionsremains poorly understood, it is of interest that the cellularoxygen sensor PHD3 (otherwise termed EGLN3 or HPH1), one ofthree Fe(II)-dependent dioxygenases now known to negativelyregulate HIF by prolyl hydroxylation of HIF- ${alpha}$ subunits (HIF-1 ${alpha}$ and HIF-2 ${alpha}$ ) (6, 13), had been previously identified as a geneinvolved in developmental apoptosis of neurons (23-25, 40).Over 50% of neurons produced during development die throughapoptosis before adulthood (reviewed in references 10 and 31).This process is largely regulated by neurotrophic factors thatare secreted in limited amounts by target tissues such thatonly those neurons making appropriate connections survive. Duringthe investigation of these phenomena, PHD3 mRNA was identifiedas a neuronal transcript that is induced by withdrawal of theneurotrophin nerve growth factor (NGF) (24). Further studieshave demonstrated that overexpression of PHD3 in primary sympatheticneurons cultured from the developing superior cervical ganglionand in the adrenal medullary tumor cell line PC12 results inapoptosis even in the presence of saturating quantities of NGF(23, 24). This effect was not reproduced by a catalyticallyinactive PHD3 mutant (23), whereas hypoxia was able to suppressapoptosis in sympathetic neurons (44). Furthermore, knock-downof PHD3 by small interfering RNA in PC12 cells prevented apoptosiseven in the absence of NGF (23). Taken together, these observationsin cultured cells indicate that the oxygen-sensitive catalyticactivity of PHD3 has a role in the regulation of neuronal apoptosis,raising important questions about the extent of PHD3-dependentneuronal apoptosis in vivo and its role in the intact animal.

To address this, we have inactivated PHD3 (by homologous recombinationin the mouse) and have analyzed the developmental and physiologicaleffects on neuronal apoptosis. We show that PHD3-dependent modulationof NGF-dependent survival is a lineage-specific property affectingthe sympathetic nervous system and that sympathetic neuronsfrom PHD3^–/–, but not PHD2^+/– or PHD1^–/–,mice manifest increased NGF-promoted neurite growth as wellas enhanced survival. Increased numbers of sympathetic neuronssurvive to adulthood, and PHD3^–/– mice have increasednumbers of neurons in the superior cervical ganglia (SCG) andincreased numbers of chromaffin and glomus cells in the adrenalmedulla and carotid body. However, despite this increase incell number, the sympathoadrenal system was dysfunctional inPHD3^–/– mice, with reduced innervation of targetorgans and dysregulated responses, including reduced catecholaminesecretion and reduced systemic blood pressure. The findingsdemonstrate a key role for PHD3 in regulating the anatomicaland physiological integrity of the sympathoadrenal system.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Targeting vector and PHD3 inactivation. For construction of the PHD3 targeting vector, the followingfragments were cloned in a pPNTLox2 vector (from 5' to 3'):a 3.4-kb BamH1 fragment located 3.1 kb upstream of exon 1 (5'flank) and a neomycin resistance cassette in the opposite orientation,a 3.9-kb EcoR1 fragment located 1.4 kb downstream of exon 1(3' flank), and a thymidine kinase selection cassette (Fig.1A). Embryonic stem (ES) cells (129 SvEv background) were electroporatedwith the linearized targeting vector for PHD3 as described elsewhere(39). Resistant clones were screened for homologous recombinationby Southern blotting (Fig. 1B) and PCR (not shown). Correctlyrecombined ES cells were then aggregated with morula-stage embryos.To obtain PHD3^+/– germ line offspring in a 50% Swiss/50%129 SvEv background, chimeric male mice were intercrossed withwild-type Swiss female mice. PHD3 mRNA transcripts were quantifiedby RNase protection assay in embryos (Fig. 1C), using a riboprobeprotecting bp 217 to 383, 5' to 3', of PHD3 mRNA, and real-timereverse transcription-PCR (RT-PCR) in mouse embryonic fibroblasts(Fig. 1D), using the following forward and reverse primers:5'-TCCCACTCTTCCACCTTC-3' and 5'-CTGTAGCCGTATTCATTGTC-3' forglyceraldehyde-3-phosphate dehydrogenase (GAPDH); 5'-CTATGGGAAGAGCAAAGC-3'and 5'-AGAGCAGATGATGTGGAA-3' for PHD3, which amplify bp 1779to 2008, a sequence distal to the deleted region, in responseto normoxia or hypoxia (1% oxygen for 16 h). PHD3 protein levelswere also measured by Western blotting in mouse embryonic fibroblastswith a monoclonal mouse anti-PHD3 antibody, P3-188e (1), toconfirm their absence in PHD3^–/– tissues.

View larger version (23K):
[in this window]
[in a new window]

FIG. 1. PHD3 targeting strategy. (A) Targeting strategy for PHD3 inactivation. Top: wild-type PHD3 allele diagram, indicating the position of exon 1 (dark box in the genomic structure). Middle: outline of the targeting vector, specifying the genomic sequences used as 5' and 3' homology flanks, inserted on each side of a neomycin resistance (Neo) cassette. A thymidine kinase (TK) gene outside the flanking homologies allowed for negative selection against random integration events. Bottom: replacement of exon 1 by the Neo cassette after homologous recombination. Diagnostic restriction fragments are indicated with their relative sizes by the thin lines under or above the alleles. Dark bars under the genes represent the probes used for Southern blot analysis. (B) Southern blot analysis of genomic DNA from recombinant ES cells, digested with HindIII and hybridized with the 5' external probe. The 3.8-kb and 5.3-kb genomic fragments correspond to PHD3 wild-type (WT) and PHD3^null alleles, respectively. (C) RNase protection assay demonstrating PHD1, -2, and -3 mRNA levels in total RNA extracted from wild-type and PHD3^–/– embryos. U6 small nuclear RNA (snRNP) was used as an internal control. (D) Quantitative RT-PCR (left panel) and Western blot assay (right panel) showing induction of PHD3 mRNA and protein in wild-type, but not PHD3^–/–, mouse embryonic fibroblasts in response to hypoxia. N, normoxia; H, hypoxia (1% oxygen for 16 h).

Animals. Wild-type, PHD3^–/–, PHD1^–/–, PHD2^+/–,PHD3^–/–; HIF-1 ${alpha}$ ^+/–, and PHD3^–/–;HIF-2 ${alpha}$ ^+/– and HIF-2 ${alpha}$ ^+/– mice on a mixed Swiss/129SvEvgenetic background were used in the experiments. Mice from thesame litter were used for all comparisons, except for HIF-2 ${alpha}$ ^+/–versus PHD3^–/–; HIF-2 ${alpha}$ ^+/– (see Fig. 5D, below)where appropriate crosses were unlikely, by Mendelian inheritance,to produce the required genotypes within littermates. For adultmice, males of between 3 and 6 months of age were used.

View larger version (37K):
[in this window]
[in a new window]

FIG. 5. Effect of genetic inactivation of PHD3 on anatomy of SCG. (A) Bright-field images of wild-type and PHD3^–/– neonatal SCG. Bar, 100 µm. (B) Neuronal complement of neonatal SCG in wild-type and PHD3^–/– mice; counts are of viable trypsin-dissociated neurons. (C) Stereological analysis of TH-positive neurons showing increased cell numbers in the SCG from adult PHD3^–/– mice. (D) Comparison of neuronal complement of neonatal SCG in mice of the indicated genotypes. Counts are as for panel B. HIF-2 ${alpha}$ heterozygosity, but not HIF-1 ${alpha}$ heterozygosity, is associated with reduced neuronal complement.

Neuronal culture. SCG, dorsal root ganglia (DRG), and trigeminal ganglia (TG)were dissected from postnatal day 0 (P0) mouse pups, trypsinizedin 0.05% trypsin in Hanks’ balanced salt solution (Gibco)for 20 min at 37°C, and dissociated by trituration. Dissociatedcultures of SCG, DRG, and TG neurons were then plated at lowdensity (150 to 300 cells/dish) on a poly-ornithine/lamininsubstrate in 35-mm tissue culture dishes in defined, serum-freemedium, as described previously (11). A range of concentrations(0 to 250 ng/ml) of NGF or neurotrophin 3 (NT-3; both were fromR&D Systems, Minneapolis, MN) was added to support the survivalof cultured neurons.

Neuronal survival was measured by counting the number of neuronsin a 12- by 12-mm grid in the center of the dish both 3 and24 h after plating (as described in reference 11). The numberof neurons surviving after 24 h was expressed as a percentageof the initial number of neurons.

Total viable neuronal counts were measured by counting the numberof phase-bright neurons using a Neubauer hemocytometer immediatelyfollowing dissociation of neurons.

Neuritic morphology was measured by culturing neonatal SCG neuronsfor 24 h with 10, 0.4, or 0.08 ng/ml NGF, as described above.Caspase inhibitors (100 µM; Calbiochem) were added tocultures containing limiting levels of NGF (0.4 and 0.08 ng/ml)in order to avoid biased sampling of differentially survivingneurons. Neurons were then fluorescently labeled with the vitaldye calcein-AM (Invitrogen), and images were acquired usingan Axioplan Zeiss laser scanning confocal microscope. Neuriticarbors were traced using LSM510 software in order to calculatethe total neurite length and the extent of arborization by Shollanalysis. For the latter, concentric, digitally generated rings,30 µm apart, were centered on the cell soma, and the numberof neurites intersecting each ring was counted (38).

Quantitative RT-PCR. Dissociated SCG, DRG, and TG neurons from wild-type P0 mousepups were cultured overnight in 10 ng/ml NGF as described above.Neurons were then washed with defined culture medium and grownin either the presence or absence of NGF for 10 h (the timepoint at which PHD3 mRNA induction is maximal in the SCG [24])before harvesting the cells for RNA. Total RNA was isolatedwith the RNeasy Mini extraction kit (Qiagen, Hilden, Germany)and then reverse transcribed for 1 h at 37°C with StrataScriptreverse transcriptase (Stratagene). Reverse transcription reactionswere amplified using the Brilliant QPCR core reagent kit (Stratagene)according to the manufacturer's protocol. The PCR was performedwith the Mx3000P apparatus (Stratagene) for 45 cycles of 95°Cfor 30 s, 52°C (for PHD3) or 51°C (for GAPDH) for 1min, and 72°C for 30 s. A melting curve was obtained toconfirm that the Sybr green (Molecular Probes) signal correspondedto a unique and specific amplicon. Standard curves were generatedfor every real-time PCR run by using serial fourfold dilutionsof a reverse-transcribed RNA extract from adult heart. Primersfor PHD3 and GAPDH were as described above.

Histological and stereological analyses. To estimate the total tyrosine hydroxylase (TH)-positive cellnumber, the SCG, adrenal medulla, and carotid body from adultmice were fixed in formalin overnight and then transferred intophosphate-buffered saline (PBS) containing 30% sucrose. Twenty-µmsections were blocked for 1 h at room temperature with 10% fetalcalf serum and 1 mg/ml bovine serum albumin containing 0.1%Triton X-100 in PBS and then incubated for 16 h at 4°C witha rabbit anti-TH polyclonal antibody (Pel-Freez; diluted 1:1,000in blocking solution). The sections were washed four times inPBS-Triton before being incubated with goat anti-rabbit secondaryantibody (Envision+; Dako). Stereological estimation of thenumber of TH-positive cells was performed on sections spaced80 µm (SCG and adrenal medulla) or 40 µm (carotidbody) throughout the organ. The number of cells was estimatedby systematic random sampling using a 106,954-µm³ opticaldissector (43), excluding cells in the superficial planes ofsections. The volume of each organ was estimated according toCavalieri's principle (8). The adrenal medulla cell volume wascalculated using a rotator vertical probe. Seven randomly selectedTH-positive cells were measured in different sections per individualadrenal gland. Stereological measurements were performed usingthe CAST grid system (Olympus) with a coefficient of error of<0.09.

To detect apoptosis in the SCG from P0 mice, tissues were fixedin formalin overnight and then transferred to 70% ethanol andparaffin embedded. The terminal deoxynucleotidyltransferase-mediateddUTP-biotin nick end labeling (TUNEL) method was performed on7-µm sections of the SCG using the ApopTag peroxidasein situ apoptosis detection kit (Chemicon International) accordingto the manufacturer's directions.

To assess the density of sympathetic innervation, the eyes andsubmandibular and pineal glands from adult mice (or P5 micefor the pineal glands) were fixed and cryoprotected as above.Fifteen-µm serial sections were blocked for 1 h at roomtemperature with 10% normal goat serum containing 0.1% TritonX-100 in 10 mM PBS and then incubated for 18 h at 4°C witha rabbit anti-TH polyclonal antibody (diluted 1:200 in PBS with1% normal goat serum; Chemicon). The sections were washed threetimes in PBS before being incubated with goat anti-rabbit secondaryantibody (Alexa-Fluor, diluted 1:500 in PBS with 1% normal goatserum; Invitrogen). The outlines of the iris and submandibularand pineal glands were traced using Adobe Photoshop 7. The totaland TH-positive areas were measured using automated pixel counts,and the ratio of TH-positive area to total area was used tocalculated the TH-positive density.

Pupillometry. Adult wild-type and PHD3^–/– mice were dark adaptedfor 1 h. Subsequently, animals were removed from their homecage, immobilized by scruffing, and pupil reactions were monitored(at 0 lx followed by 150 lx of bright white light) using a commercialcharge-coupled-device camcorder (DCR-HC17E; Sony, Tokyo, Japan)attached to a dissecting microscope (Olympus, Tokyo, Japan)under infrared light-emitting diode illumination ( ${lambda}$ _max of 850nm). Pupil areas were estimated by manually fitting an ellipseto digitalized video still images using Windows Movie Maker(Microsoft, Redmond, WA) and Adobe Photoshop (San Jose, CA)software.

In vivo hemodynamics. Adult mice were anesthetized with 2% isoflurane in 100% oxygenvia a nose cone. Mice were placed supine with body temperatureregulated at 37°C and allowed to breathe spontaneously.The right carotid artery was cannulated with a 1.4 French microtippedcannula (SPR-839; Millar Instruments, Houston, TX) advancedretrograde into the left ventricular (LV) cavity under ultrasoundguidance. The right jugular vein was cannulated with polyethylenetubing for intravenous drug administration. Isoflurane was reducedto 1.5%, followed by a period of equilibration until hemodynamicindices were stable for >15 min. Pressure measurements wererecorded using a Powerlab 4SP data recorder (ADInstruments,Chalgrove, United Kingdom) at baseline in both the LV and aortaand in the LV during infusion of β₁-adrenoreceptor stimulationwith dobutamine hydrochloride (16 ng/g of body wt/min).

A subset of these mice were used to test baroreceptor reflexfunction by intravenous bolus administration of the ${alpha}$ ₁-adrenoreceptoragonist phenylephrine hydrochloride (at 50 µg/kg) to inducebrief periods of elevated blood pressure (30). Baroreceptorreflex gain was calculated as the maximal change in heart rateover the maximal change in blood pressure. Drug solutions forbolus injection were made up in 0.9% saline and a volume of5 µl. Mice were killed by cervical dislocation or removalof the heart, and the cardiac chambers blotted and weighed.

Radiotelemetry. PAC10 radiotelemeters (Data Sciences International, MN) wereimplanted in the left carotid artery of adult mice under isofluraneanesthesia as previously described (7, 29). Mice were allowedto recover for 10 days following the operation, and then continuousmeasurements of blood pressure and activity state were recorded,at 500 Hz, for three consecutive days using standard acquisitionsoftware (Data Sciences International, MN). Blood pressure wasanalyzed on a beat-per-beat basis according to the activitystate of the mouse at the time of recording (Spike 2, version6.1; Cambridge Electronic Design, Cambridge, United Kingdom).Two activity states were defined: rest, no movement of the animalas detected by the telemeter within 30 s of measurement; activity,movement of the mouse within its cage at the time of recording.

Amperometric recording of single-cell catecholamine secretion in isolated slices. Adult mouse suprarenal gland dissection, slicing, and culture,as well as the measurement of catecholamine secretion, wereperformed following previously described procedures (14). Sliceswere transferred to a recording chamber and continuously perfusedwith a solution containing 117 mM NaCl, 4.5 mM KCl, 23 mM NaHCO₃,1 mM MgCl₂, 2.5 mM CaCl₂, 5 mM glucose, and 5 mM sucrose. When40 mM or 20 mM potassium solutions were used, equimolar quantitiesof sodium were used to replace potassium. The solution was bubbledwith a gas mixture of 5% CO₂ to adjust the pH to 7.2. All theexperiments were performed in a chamber at 36°C. Secretionrate (femtocoulombs/min) was calculated as the amount of chargetransferred to the recording electrode during a given time period(32).

Plasma catecholamines. Blood from adult mice (anesthetized with Nembutal; 600 µg/kgof body weight) was collected into heparinized tubes on icethat were then centrifuged at 6,000 x g for 5 min at 4°Cto separate plasma. Plasma catecholamine levels were quantifiedby immunoassay according to the manufacturer's instructions(IBL-Hamburg).

Statistical analyses. Data are expressed as means ± standard errors of themeans (SEM), with the number of experiments indicated. Statisticalanalyses were performed using Student's t test. P values of<0.05 were considered statistically significant.

RESULTS

Top

RESULTS

Generation of PHD3-deficient mice. Mice deficient in PHD3 were generated by standard homologousrecombination procedures in ES cells. The PHD3 targeting strategyinvolved removing a genomic region encompassing the promoterand first exon, including the translational initiation site(Fig. 1A and B), and cells from PHD3^–/– homozygousanimals expressed no PHD3 transcript (Fig. 1C and D, left panel)or protein (Fig. 1D, right panel). A small reduction in theexpected number of PHD3^–/– live births was notedamong the offspring of heterozygous matings (PHD3^+/+, 77 [29%],PHD3^+/–, 145 [54%], PHD3^–/–, 47 [17%]), butotherwise PHD3^–/– mice appeared healthy and reachedadult life without obvious abnormality.

Enhanced survival of sympathetic neurons from PHD3^–/– mice in vitro. Because experimental manipulation of PHD3 expression in culturedsympathetic neurons and PC12 cells has been reported to affectcell survival (23-25, 40), we directly compared the survivalof sympathetic neurons obtained from the SCG of wild-type andPHD3^–/– mice. Heterozygous mice were mated to obtainoffspring of all three genotypes, and dissociated low-densitycultures of SCG neurons were established from newborn littermates.To minimize variability arising from technical considerations,in this and all subsequent experiments, comparisons of gangliaexplanted from mice of different genotypes were performed ina pair-wise fashion at the same experimental session. The neuronswere cultured with a range of NGF concentrations, and neuronalsurvival was estimated 24 h after plating. At nonsaturatingconcentrations of NGF, the survival of neurons from PHD3^–/–mice was consistently enhanced compared to that of neurons fromwild-type mice (Fig. 2A, left panel). Interestingly, loss ofPHD3 did not confer a survival advantage at very low NGF concentrations,suggesting that, although PHD3 reduces the capacity of neuronsto survive in the presence of NGF, it is not required for thedeath of NGF-deprived neurons.

View larger version (29K):
[in this window]
[in a new window]

FIG. 2. Survival of neurons cultured from PHD3^–/– mice. (A) NGF dose-response curves, demonstrating increased neuronal survival in neurons cultured from the SCG of P0 PHD3^–/– mice but not PHD1^–/– or PHD2^+/– mice. Neuronal survival was estimated by expressing the number of surviving neurons after 24 h in culture as a percentage of the initial number of neurons at 3 h postplating. (B) NT-3 dose-response curve, showing increased survival in neurons cultured from the SCG of P0 PHD3^–/– mice. (C) NGF dose-response curves, showing no change in survival in neurons cultured from the DRG and TG of P0 PHD3^–/– mice. (D) Quantitative RT-PCR showing induction of PHD3 mRNA after NGF withdrawal in the SCG, but not the DRG or TG, from P0 wild-type mice. Values for this and all subsequent figures are presented as means ± SEM (n = 3) for dose-response curves, or n as indicated in parentheses. *, P < 0.05 versus control; **, P < 0.01 versus control.

To test whether other prolyl hydroxylases that regulate HIFalso affect the NGF survival response, similar cultures of SCGneurons were established from PHD1^–/– and PHD2^+/–mice (2) and compared with those from their littermate controls(PHD2^–/– mice could not be used because they diein utero between embryonic day 12.5 [E12.5] and E14.5 [2, 41]).In contrast with neurons from PHD3^–/– mice, no differencesin NGF dose responses were observed (Fig. 2A, middle and rightpanels).

NGF promotes neuronal survival by binding to and activatingthe receptor tyrosine kinase TrkA (10). Because NT-3 is alsocapable of promoting the survival of neonatal SCG neurons inculture by a TrkA-dependent mechanism (12), we investigatedwhether inactivation of PHD3 affects the survival response ofSCG neurons to this neurotrophin. As with NGF, SCG neurons fromPHD3^–/– mice survived more effectively with NT-3than neurons from wild-type littermates (Fig. 2B).

In addition to sympathetic neurons, large numbers of sensoryneurons in the developing peripheral nervous system are alsodependent on NGF for survival. To ascertain whether the absenceof PHD3 also affects the NGF survival response of these neurons,we established low-density dissociated cultures from two populationsof sensory neurons that are mostly comprised of NGF-dependentneurons: those of the DRG and TG. In contrast to SCG neurons,the NGF dose responses of DRG and TG neurons from wild-typeand PHD3^–/– mice were completely overlapping (Fig.2C). Since NGF withdrawal has been reported to induce PHD3 mRNAexpression in sympathetic neurons (24), we considered whetherthis property might underlie the specificity of PHD3-dependentsurvival effects. To investigate this, we cultured SCG, DRG,and TG neurons with NGF and deprived them of this neurotrophinby extensive washing 12 h after plating. Measurement of PHD3mRNA by RT-PCR 10 h after deprivation revealed a greater-than-twofoldincrease in PHD3 mRNA relative to GAPDH mRNA in sympatheticneurons but no significant change in sensory neurons (Fig. 2D).Thus, both PHD3-dependent neuronal survival and responsivenessof PHD3 mRNA to NGF withdrawal appear to be specific to sympatheticneurons rather than a general property of NGF-sensitive neurons.

HIF-2-dependent survival in sympathetic neurons from PHD3^–/– mice in vitro. Because of the known function of PHD3 as a HIF hydroxylase regulatingthe abundance of HIF-1 ${alpha}$ and HIF-2 ${alpha}$ , we investigated whether theinfluence of PHD3 on the NGF dose response of SCG neurons dependson either HIF-1 ${alpha}$ or HIF-2 ${alpha}$ . Since homozygous inactivation of HIF-1 ${alpha}$ or HIF-2 ${alpha}$ results in embryonic lethality (20, 33, 34, 42), weanalyzed the effects of heterozygous inactivation by generatingappropriate crosses with PHD3^–/– animals to allowfor littermate comparisons. Although we observed some variationin the overall survival of explanted neurons from PHD3^–/–mice at different experimental sessions, clear differences wereobserved in pair-wise comparisons of PHD3^–/– animalswith and without heterozygous inactivation of different HIF- ${alpha}$ isoforms. Heterozygous HIF-1 ${alpha}$ inactivation did not significantlyaffect the NGF dose response of cultured PHD3-deficient SCGneurons (PHD3^–/– versus PHD3^–/–; HIF-1 ${alpha}$ ^+/–littermate neonates) (Fig. 3A). In contrast, inactivating oneallele of HIF-2 ${alpha}$ caused a significant shift in the NGF dose responseof PHD3-deficient SCG neurons to higher NGF concentrations (PHD3^–/–versus PHD3^–/–; HIF-2 ${alpha}$ ^+/– littermate neonates)(Fig. 3B), suggesting that appropriate expression of HIF-2 ${alpha}$ ,but not HIF-1 ${alpha}$ , is required for the influence of PHD3 on theNGF survival dose response in culture.

View larger version (18K):
[in this window]
[in a new window]

FIG. 3. HIF dependence of the PHD3 survival effect. NGF dose-response curves for neonatal SCG neurons showing no difference in survival of neurons derived from PHD3^–/–; HIF-1 ${alpha}$ ^+/– versus PHD3^–/– mice (A) and reduced survival of neurons derived from PHD3^–/–; HIF-2 ${alpha}$ ^+/– versus PHD3^–/– mice (B).

Enhanced NGF-promoted neurite growth of sympathetic neurons from PHD3^–/– mice in vitro. In addition to supporting the survival of developing sympatheticneurons, NGF also promotes the growth of neurites from theseneurons in culture (15). For this reason, we investigated neuritearbor size and complexity in sympathetic neurons from PHD3^–/–mice. Low-density cultures of SCG neurons from P0 PHD3^–/–and wild-type mice were grown with different concentrationsof NGF, and neurite arbor size and complexity were quantified24 h after plating. At subsaturating concentrations of NGF (Fig.4B and C), but not at saturating levels (Fig. 4A), the neuritearbors of PHD3-deficient neurons were significantly longer thanthose of wild-type neurons. Sholl analysis, which provides agraphic representation of neurite branching with distance fromthe cell body, revealed that the neurite arbors of PHD3-deficientneurons were larger and more branched than those of wild-typemice in the presence of subsaturating levels of NGF (Fig. 4B and C).The typical appearance of the neurite arbors of PHD3-deficientand wild-type SCG neurons grown with subsaturating NGF are illustratedin Fig. 4D. Because the neurites in short-term SCG cultures,such as those used in our study, are exclusively axons ratherthan dendrites, we conclude that PHD3 also modulates axonalgrowth and branching in culture.

View larger version (16K):
[in this window]
[in a new window]

FIG. 4. Neurite length and arborization of sympathetic neurons cultured from PHD3^–/– mice. Increased neurite length and arborization in P0 PHD3^–/– mice at 0.4 (B) and 0.08 ng/ml NGF (C), but not at 10 ng/ml NGF (A). Neurite length is presented as the mean ± SEM on 40 to 70 neurons per genotype (similar results were obtained in three independent experiments). (D) Representative images of neurons cultured in 0.4 ng/ml NGF and fluorescently labeled with the vital dye calcein-AM. Bar, 50 µm.

PHD3-deficient mice have increased numbers of SCG neurons. To ascertain the developmental and physiological relevance ofour in vitro observations, we compared the number of neuronsin the SCG of wild-type and PHD3^–/– mice. The neuronalcomplement of the SCG of newborn animals was estimated by countingthe number of neurons in trypsin-dissociated cell suspensionsobtained from carefully dissected ganglia. In order to minimizevariability assignable to genetic background, absolute age ofneurons, and dissociation technique, these comparisons weremade between ganglia explanted at the same experimental sessionfrom mice within the same litter. Neurons were recognized anddistinguished from nonneuronal cells by their characteristiclarge, phase-bright, spherical cell bodies under phase-contrastoptics. SCG dissected from PHD3^–/– newborn miceappeared larger than those of wild-type littermates (Fig. 5A)and contained significantly more neurons (Fig. 5B). Likewise,stereological measurements carried out on serially sectionedSCG in 3- to 6-month-old mice revealed that the SCG was largerand contained significantly more neurons in PHD3^–/–mice compared with age-matched wild-type animals (Fig. 5C).As expected, the number of neurons in the SCG of wild-type adultswas lower than that in newborns as a result of ongoing programmedcell death in the immediate postnatal period. However, the numberof neurons in the SCG of PHD3^–/– mice remained substantiallyunchanged between birth and adulthood, suggesting that programmedcell death during at least the postnatal period was largelycurtailed in these mice. In keeping with this, we observed amodest reduction in the number of apoptotic or TUNEL-positivecells in the SCG from P0 PHD3^–/– mice ( $~$ 30% decrease,n = 7, P < 0.05, from 6.7 x 10^–5 to 4.9 x 10^–5TUNEL-positive cells/µm² SCG area).

In contrast to the effect of PHD3 deletion on SCG neuron number,there were no significant differences between the numbers ofneurons in the SCG of either PHD1^–/– or PHD2^+/–neonates compared with wild-type littermates (data not shown).

To investigate the in vivo significance of HIF- ${alpha}$ gene dosageon the survival of cultured neurons from PHD3^–/–mice, we compared the number of neurons in the SCG of PHD3^–/–versus PHD3^–/–; HIF-1 ${alpha}$ ^+/– littermate neonatesand PHD3^–/– versus PHD3^–/–; HIF-2 ${alpha}$ ^+/–littermate neonates. Consistent with the studies of survivalin culture, these studies revealed that heterozygous inactivationof HIF-2 ${alpha}$ , but not HIF-1 ${alpha}$ , reduced the SCG neuronal populationin PHD3^–/– animals (Fig. 5D). Interestingly, thiseffect was not observed in the PHD3-positive background (comparisonof wild-type versus HIF-2 ${alpha}$ ^+/– mice) (Fig. 5D). Furtherexperiments designed to test the effect of PHD3 inactivationin the context of HIF-2 ${alpha}$ heterozygosity (comparison of PHD3^–/–;HIF-2 ${alpha}$ ^+/– versus HIF-2 ${alpha}$ ^+/– mice) revealed no significantdifferences in SCG cell numbers (Fig. 5D), suggesting that PHD3-dependenteffects on cell number were affected by integrity of the HIF-2 ${alpha}$ pathway and vice versa. Taken together with our in vitro survivaldata, these in vivo observations suggest that the selectiveincrease in SCG neurons in PHD3^–/– mice resultsfrom reduced cell loss during the phase of programmed cell deathin the perinatal period as a consequence of the enhanced sensitivityof PHD3-deficient neurons to NGF and that this effect is atleast partially dependent on HIF-2 ${alpha}$ .

PHD3-deficient mice have increased numbers of cells in the adrenal medulla and carotid body. NGF-dependent cells of the sympathoadrenal axis extend to theneural crest-derived chromaffin and glomus cells of the adrenalmedulla and carotid body. Our findings that PHD3^–/–mice have significantly more sympathetic neurons than wild-typemice prompted us to examine cell number elsewhere in the sympathoadrenalsystem. Stereological analysis of adult mice (3 to 6 monthsold) revealed significantly more TH-positive cells in both theadrenal medulla and the carotid body (Fig. 6). Interestingly,though total numbers of cells were increased in both organsin PHD3^–/– mice, overall organ volume was increasedfor the carotid body, but decreased for the adrenal medulla,in PHD3^–/– mice (Fig. 6). The latter phenomenonmay be explained by a reduction in chromaffin cell volume (from1,743 µm³ in wild-type to 1,195 µm³ in PHD3^–/–mice; P < 0.05; n = 7), consistent with the significant increasein TH-positive cell density in the adrenal medulla (Fig. 6A).

View larger version (31K):
[in this window]
[in a new window]

FIG. 6. Effect of genetic inactivation of PHD3 on the anatomy of other sympathoadrenal tissues, the adrenal medulla (A) and carotid body (B). Stereological analysis of TH-positive neurosecretory cells demonstrating increased cell numbers in the adrenal medulla (A) and carotid body (B) of adult PHD3^–/– mice.

Sympathetic innervation of target tissues. Given that inactivation of PHD3 in vivo results in increasednumbers of SCG neurons surviving to adulthood and that NGF ismore effective in promoting neurite growth from cultured PHD3-deficientSCG neurons, we asked whether sympathetic innervation densityis increased in PHD3^–/– mice. The SCG innervatesseveral anatomically discrete structures, including the iris,submandibular gland, and pineal gland, whose innervation densitycan be relatively easily estimated by quantifying the area occupiedby TH-positive nerve fibers in tissue sections. Surprisingly,this analysis revealed significantly fewer TH-positive fibersin the iris, submandibular gland, and pineal gland of PHD3^–/–mice compared with wild-type animals (Fig. 7A, B, and C).

View larger version (17K):
[in this window]
[in a new window]

FIG. 7. Sympathetic innervation of SCG target tissues from PHD3^–/– mice: immunohistochemistry demonstrating TH-positive neurons in SCG target tissues. Shown are representative images of TH-stained (bright red) neurons in the iris. Measured as the ratio of the TH-positively stained area over the total area of the SCG target tissue, there was decreased sympathetic innervation density of the iris (A), submandibular gland (B), and pineal gland (C). (D) Average pupil sizes in conscious, adult PHD3^–/– mice and wild-type controls under normal illumination (150 lx of bright white light) and after 1 h of dark adaptation (0 lx).

Physiological effects on the sympathetic nervous system. We next sought to assess the integrity of physiological responsesthat are dependent on the sympathetic nervous system. Becauseof the reduced sympathetic innervation of the iris in PHD3^–/–mice (Fig. 7A), we tested light-to-dark pupillary responses.While no differences in pupil diameter were seen under normallighting conditions (150 lx) (Fig. 7D), pupil diameter was significantlydecreased in the dark-adapted eye from PHD3^–/– mice(0 lx) (Fig. 7D). This implies decreased tone in the sympatheticallyinnervated dilator pupillae muscle fibers of PHD3^–/–mice.

Since one of the most important roles of sympathoadrenal functionis in the control of blood pressure, we went on to perform aseries of measurements of systemic blood pressure regulation.First, a series of resting measurements were performed underanesthesia. These revealed a modest but significant reductionin blood pressure in PHD3^–/– mice versus wild-typelittermates (Table 1) but no significant differences in heartrate (Table 1). Consistent with reduced arterial pressure, wefound that left ventricular weights were significantly reducedin PHD3^–/– mice versus wild-type mice (Table 1).

View this table:
[in this window]
[in a new window]

TABLE 1. Cardiovascular function in anesthetized mice^a

To pursue the differences in blood pressure further, measurementsof aortic blood pressure were repeated in conscious mice usingradiotelemetry. These confirmed that, at rest, PHD3^–/–mice were hypotensive compared to wild-type mice, with reducedsystolic blood pressures (Fig. 8). Furthermore, the differencein systolic blood pressure was enhanced when the mice becameactive, with as much as 20 mm Hg difference in systolic pressurebetween PHD3^–/– mice and littermate controls (Fig.8).

View larger version (19K):
[in this window]
[in a new window]

FIG. 8. Aortic blood pressures in conscious, adult PHD3^–/– mice, as measured by radiotelemetry. (A) Pooled frequency histogram of systolic blood pressure over the 3-day recording period (using 1 mm Hg bins) from four (wild-type) or five (PHD3^–/–) resting (solid lines) and active (dotted lines) mice. (B) Average blood pressure recordings over a 3-day recording period. Decreased systolic and diastolic blood pressures were observed in PHD3^–/– mice. These differences were enhanced when the mice were active.

If reduced blood pressure were due to sympathoadrenal dysfunctionrather than a primary cardiac or vascular defect, we arguedthat responses to exogenous adrenoreceptor agonists should bepreserved or even exaggerated. Under anesthesia, we measuredresponses to infusions of the ${alpha}$ ₁-agonist phenylephrine and theβ₁-agonist dobutamine. Results are shown in Table 1. Inresponse to the exogenous ${alpha}$ ₁-adrenoreceptor agonist phenylephrine,the maximal rise in blood pressure was exaggerated in PHD3^–/–mice versus wild-type littermates. Interestingly, reflex bradycardiawas blunted in PHD3^–/– animals, suggesting thatbaroreceptor reflex sensitivity, a measure of endogenous autonomicfunction, was reduced. In response to the exogenous β₁-adrenoreceptoragonist dobutamine, PHD3^–/– mice also showed anexaggerated increase in heart rate and contractility, consistentwith reduced intrinsic sympathetic function rather than a primarycardiac or vascular cause of hypotension.

Finally, we tested adrenal medullary function both by measuringindividual chromaffin cell catecholamine secretion rates byamperometry in explanted adrenal slices and by measuring totalcirculating catecholamine levels in intact animals. Amperometricallymeasured basal secretion of catecholamines was reduced by approximately50% in chromaffin cells from PHD3^–/– mice (Fig.9A). Secretion in response to potassium-induced depolarizationwas similarly reduced (20 mM potassium) or reduced to a smallerextent (40 mM potassium) in PHD3^–/– mice (Fig. 9A),consistent with reduced secretory capacity. Plasma catecholaminelevels were also lower in PHD3^–/– mice (Fig. 9B),suggesting that the observed increase in chromaffin cell numberwas not sufficient to compensate for the decrease in chromaffincell functionality under physiological conditions.

View larger version (22K):
[in this window]
[in a new window]

FIG. 9. Catecholamine secretion in PHD3^–/– mice. (A) Representative amperometric recordings of basal catecholamine release, as well as responsiveness to induction by 20 mM and 40 mM potassium (20K and 40K), of adrenal slices isolated from adult mice. The table shows the secretion rate (in fC/min) under basal conditions, as well as with 20 mM and 40 mM potassium (20 and 40 K). (B) Circulating catecholamines (adrenaline and noradrenaline) from anesthetized, adult mice. Both adrenal slice and circulating catecholamines are reduced in PHD3^–/– mice.

DISCUSSION

Top

DISCUSSION

Our findings show that the HIF prolyl hydroxylase PHD3 has animportant role in regulating the development of the sympathoadrenalsystem and that its ablation has substantial physiological consequencesthat extend into adult life.

We observed significantly more neurons in the SCG of newbornPHD3^–/– mice compared with wild-type littermates,and this elevated number of SCG neurons was maintained to adulthood.The apparent failure of the neuronal complement of the SCG todecrease in PHD3^–/– mice postnatally, when naturallyoccurring programmed cell death ordinarily matches the numberof sympathetic neurons to the requirements of their targets,suggests that the elevated number of neurons in the SCG of PHD3-deficientmice is due to reduced cell death, in keeping with the observedshift in the NGF survival dose response of PHD3-deficient SCGneurons in culture to lower NGF concentrations. Interestingly,the effect of PHD3 inactivation on NGF responsiveness in theperipheral nervous system appears to be restricted to neuronsof the sympathetic lineage, as neural crest-derived, NGF-dependentsensory neurons from PHD3^–/– mice responded normallyto NGF. It is unclear whether dysregulated apoptosis might occurin regions within the central nervous system in PHD3^–/–mice. However, these mice are viable as adults, without obviousneurological abnormalities. In addition, brains from PHD3^–/–mice were not obviously abnormal and were not significantlydifferent in weight (0.43 ± 0.01 g versus 0.43 ±0.00 g in PHD3^–/– and wild-type males; 0.42 ±0.01 g versus 0.44 ± 0.01 g in PHD3^–/– andwild-type females). Nevertheless, these observations do notpreclude dysregulation of apoptosis that is compensated foror might be revealed in pathological situations.

As well as enhancing the sensitivity of SCG neurons to the survival-promotingeffects of NGF, deletion of PHD3 also makes these neurons moreresponsive to the neurite growth-promoting effects of NGF inculture. Because target-derived NGF is not only required forsympathetic neuron survival during development in vivo but isalso responsible for the terminal growth and branching of sympatheticaxons in their targets, we expected to find increased sympatheticinnervation density in PHD3^–/– mice. Surprisingly,we observed the opposite in several SCG target tissues in theseanimals. One possible explanation for this apparent paradoxis that the increased number of sympathetic neurons innervatingtarget tissues results in elevated uptake and removal of NGFfrom these tissues by retrograde axonal transport, resultingin lower ambient levels of NGF in the targets. Whereas retrogradetransport of NGF in signaling endosomes to the cell bodies ofsympathetic neurons is required for survival, the extent ofaxonal growth and branching in target tissues is governed bythe ambient level of NGF in these tissues (15).

In addition to possessing elevated numbers of sympathetic neurons,PHD3^–/– mice have increased numbers of chromaffinand glomus cells in the adrenal medulla and carotid body. Togetherwith sympathetic neurons, these cells are derived from the sympathoadrenalneural crest lineage (22). Whether the increase in chromaffinor glomus cell number in PHD3^–/– mice results fromenhanced survival or from enhanced proliferation and/or differentiationof their precursors is not known.

Despite the increased number of cells in parts of the sympathoadrenalsystem of PHD3^–/– mice, we observed impaired homeostaticand physiological responses regulated by this system, includingreduced blood pressure, impaired light-to-dark pupillary dilation,and reduced catecholamine secretion from adrenal chromaffincells, with lower plasma catecholamine levels in PHD3^–/–mice. Sympathetic function is a key determinant of systemicblood pressure regulation (reviewed in reference 17), and elevatedsympathetic activity is present in many forms of human hypertension.Apart from the findings reported here, we have not so far observedother basal abnormalities in PHD3^–/– mice. Thus,though PHD3 could have other, as-yet-unknown effects on bloodpressure regulation, the evidence for sympathetic hypofunction,reduced catecholamine secretion, reduced responses to activity,and preserved responses to exogenous adreno-agonists all supporta causal link between sympathoadrenal dysregulation and bloodpressure dysregulation in PHD3^–/– animals.

Though previous cellular studies have implicated oxygen-dependentcatalytic activity of PHD3 in the regulation of neuronal survival,these studies have given conflicting data on the role playedby the known substrates HIF-1 ${alpha}$ and HIF-2 ${alpha}$ (23, 44). In the currentwork we found that heterozygous inactivation of HIF-2 ${alpha}$ , but notHIF-1 ${alpha}$ , had major effects on sympathetic neuronal survival andthe neuronal complement of the SCG.

Interestingly, neonatal sympathetic failure has been observedin HIF-2 ${alpha}$ ^–/– mice, and HIF-2 ${alpha}$ mRNA has been reportedto be strongly expressed in sympathoadrenal tissues of the mouse(42). Though difficulties in obtaining background-free immunohistochemicalsignals precluded assessment in the mouse, we have also foundthat PHD3 is strongly expressed in the rat sympathoadrenal system,suggesting that the two proteins are coexpressed in this lineage(B. Wijeyekoon and C. W. Pugh, unpublished data). Interestingly,studies of small interfering RNA-mediated suppression of PHDsin tissue culture have indicated that PHD3 appears to exertgreater regulatory effects on HIF-2 ${alpha}$ than HIF-1 ${alpha}$ (1). Taken together,the simplest synthesis of these findings is that PHD3 inactivationpromotes sympathoadrenal neuronal survival, at least in part,through upregulation of HIF-2 ${alpha}$ . Nevertheless, consistent witha recent report of pharmacological PHD inhibition in sympatheticneurons (26), HIF-2 ${alpha}$ protein levels in the SCG of these normoxicanimals were below the detection threshold for immunohistochemistry,so this could not be confirmed. It should also be noted thatthis explanation does not preclude the existence of other PHD3substrates that may contribute to the observed effect. Indeed,the lack of effect of HIF-2 ${alpha}$ heterozygosity on a PHD3-positivebackground suggests that the relationship between HIF-2 ${alpha}$ andneuronal survival is not simple and depends on PHD3 status,perhaps implying the existence of other substrates in additionto HIF-2 ${alpha}$ . Thus, the importance of PHD3, as opposed to PHD1 orPHD2, in neuronal apoptosis might reflect the existence of adiscrete, non-HIF, PHD3 substrate in this process, or the relativeabundance of PHD3 in these cells coupled to the preferred targetingof HIF-2 ${alpha}$ by PHD3, or both.

Whether and how adult sympathetic hypofunction in PHD3^–/–mice relates to dysregulation of HIF-2 ${alpha}$ is also unclear. Togetherwith the finding of sympathetic failure in association withcomplete inactivation of HIF-2 ${alpha}$ in some but not all studies (33,42), our findings suggest that although an intact PHD3/HIF-2 ${alpha}$ system is necessary for proper sympathoadrenal development,there is no simple relationship of hyper- or hypofunctionalityto predicted effects of these genotypes on HIF-2 ${alpha}$ levels.

While we have demonstrated physiological deficits in adult PHD3^–/–mice, it is not clear whether deficits arise purely as a consequenceof PHD3 deficiency during development or whether there is anongoing PHD3 requirement for sympathetic function in the adult.A great deal of interest has been generated by the possibilitythat environmental effects on development might influence medicallyimportant aspects of adult physiology, such as blood pressureregulation (reviewed in references 3 to 5 and 28). However,there are as yet few mechanistic clues as to how such processesmight operate. The dependence of PHD activity on oxygen andcofactors, such as the Krebs cycle intermediate 2-oxoglutarate,Fe(II), and ascorbate (35), and inhibition of PHDs by metabolicintermediates, such as succinate, fumarate, pyruvate, and oxaloacetate(9, 18, 19, 21, 27, 36), raises the interesting possibilitythat environmental influences on PHD3 activity could perturbsympathetic development in a manner that might have importanteffects on cardiovascular control.

In conclusion, our findings reveal an important role for PHD3in the developing sympathoadrenal system. It is an intriguingpossibility that this connection between hypoxia pathways andthe developing sympathoadrenal system might be a means by whichchanges in oxygen tension could influence key aspects of anatomicaland physiological maturation of this system. Whether this providesa paradigm for environmental influences affecting developmentwill be of interest in the future.

ACKNOWLEDGMENTS

This work was supported by the British Heart Foundation, theWellcome Trust, and the EU Framework 6 Pulmotension Consortium.

We thank P. van Wesemael, M. Dewerchin, S. Wyns, E. Gils, B.Hermans, L. Kieckens, and L. Notebaert (Katholieke UniversiteitLeuven, Belgium), as well as D. Adlam, S. Neubauer, R. Foster,and K. Buckler (University of Oxford, United Kingdom), for theircontribution.

We have declared that no conflict of interest exists.

FOOTNOTES

* Corresponding author. Mailing address: The Henry Wellcome Building for Molecular Physiology, University of Oxford, Headington Campus, Roosevelt Drive, Oxford OX3 7BN, United Kingdom. Phone: (44) 1865 287 990. Fax: (44) 1865 287 992. E-mail: pjr{at}well.ox.ac.uk

Published ahead of print on 10 March 2008.

T.B. and D.G. contributed equally to this work.

REFERENCES

Top