Hypoxia-inducible factor (HIF) transcription factors implement essential changes in gene expression that enable animals to adapt to low oxygen (hypoxia). The stability of the C. elegans HIF-1 protein is controlled by the evolutionarily conserved EGL-9/VHL-1 pathway for oxygen-dependent degradation. Here, we describe vhl-1-independent pathways that attenuate HIF-1 transcriptional activity in C. elegans. First, the expression of HIF-1 target genes is markedly higher in egl-9 mutants than in vhl-1 mutants. We show that HIF-1 protein levels are similar in animals carrying strong loss-of-function mutations in either egl-9 or vhl-1. We conclude that EGL-9 inhibits HIF-1 activity, as well as HIF-1 stability. Second, we identify the rhy-1 gene and show that it acts in a novel negative feedback loop to inhibit expression of HIF-1 target genes. rhy-1 encodes a multi-pass transmembrane protein. Although loss-of-function mutations in rhy-1 cause relatively modest increases in hif-1 mRNA and HIF-1 protein expression, some HIF-1 target genes are expressed at higher levels in rhy-1 mutants than in vhl-1 mutants. Animals lacking both vhl-1 and rhy-1 function have a more severe phenotype than either single mutant. Collectively, these data support models in which RHY-1 and EGL-9 function in VHL-1-independent pathway(s) to repress HIF-1 transcriptional activity.
DURING development, homeostasis, and disease, aerobic organisms must constantly adapt to changing levels of environmental oxygen. In mammals, the hypoxia-inducible factor 1 (HIF-1), a DNA-binding transcription factor, regulates many of the critical transcriptional responses to hypoxia (Lee et al. 2004). HIF-1 target genes play central roles in cellular and systemic responses to hypoxia, including adaptive changes in angiogenesis/vascular remodeling, erythropoiesis, glycolysis, iron metabolism, cell proliferation/survival, and extracellular-matrix metabolism (Semenza 2002; Lee et al. 2004). Many common human diseases, including myocardial ischemia, stroke, peripheral vascular disease, and chronic lung disease cause acute or chronic hypoxic stress (Semenza 2000). Therapies that increase HIF-1 activity may benefit patients afflicted with these and other cardiovascular illnesses. Conversely, inhibitors of HIF-1 are being employed as cancer chemotherapeutic agents, as the survival and growth of solid tumors is dependent upon adaptation to hypoxic microenvironments (Yeo et al. 2004). It is, therefore, of central importance to elucidate the cellular and molecular networks that regulate hypoxia-inducible factor activity.
Hypoxia-inducible factors are heterodimeric transcription factors composed of α- and β-subunits. Both subunits contain basic helix-loop-helix (bHLH) and PER-ARNT-SIM (PAS) domains that mediate dimerization and DNA binding (Wang et al. 1995). The β-subunit is also termed ARNT (aryl hydrocarbon receptor nuclear translocator). ARNT is broadly expressed, and it dimerizes with other bHLH-PAS proteins. The HIFα subunit is specialized for hypoxia response (Semenza 1999). There are three HIF α-genes (HIF-1α, HIF-2α, and HIF-3α) in mammals. HIF-2α and HIF-3α exhibit tissue-specific expression patterns, and HIF-1α is expressed in most cells. Mice lacking a functional HIF-1α gene die early in embryogenesis with severe vascular defects (Iyer et al. 1998; Ryan et al. 1998). HIF-1α contains two transactivation domains, which have been shown to interact with a variety of transcriptional co-activators, including CBP/p300, SRC-1, and TIF-2 (Ema et al. 1999; Carrero et al. 2000).
The EGLN family (also called the PHD or HPH family) of prolyl hydroxylases are key regulators of oxygen-dependent HIF-1 protein stability (Lando et al. 2003; Hirota and Semenza 2005; Kaelin 2005; Schofield and Ratcliffe 2005). In the presence of oxygen, specific proline residues (Pro-402 and Pro-564) in HIF-1α are hydroxylated by members of the EGLN family (Bruick and McKnight 2001; Epstein et al. 2001; Ivan et al. 2002). The EGLN enzymes are oxygen, iron, and 2-oxoglutarate-dependent dioxygenases (Aravind and Koonin 2001). Once hydroxylated, HIF-1α binds to the von Hippel–Lindau (pVHL) tumor suppressor, the recognition component of an E3 ubiquitin–protein ligase, and this results in ubiquitination and proteasomal degradation of HIF-1α (Cockman et al. 2000; Ivan et al. 2001; Jaakkola et al. 2001). Biochemical and physiological studies suggest that molecular oxygen availability may limit EGLN activity in some tissues. HIF hydroxylases are therefore thought to act as oxygen sensors that regulate HIF-1 activity (Kaelin 2005). Intensive study of HIF-1 regulation has revealed several hypoxia-dependent and -independent strategies by which cells regulate HIF translation, protein stabilization, nuclear localization, DNA-binding capacity, and transcriptional co-activator recruitment (Wenger 2002; Bardos and Ashcroft 2005; Brahimi-Horn et al. 2005). Recent studies also suggest that EGLN proteins may regulate HIF activity via other mechanisms, independently of their role in oxygen-dependent degradation of HIFα subunits (Ozer et al. 2005; To and Huang 2005).
The nematode Caenorhabditis elegans has proven to be a powerful model system to study evolutionarily conserved signaling pathways that regulate hypoxia response and the hypoxia-inducible factor. The C. elegans hif-1 and aha-1 genes are orthologous to mammalian HIFα and HIFβ subunits, respectively (Powell-Coffman et al. 1998; Jiang et al. 2001). The hif-1 gene encodes the HIF-1 protein, and it controls most hypoxia-induced gene expression changes in C. elegans (Shen et al. 2005). hif-1-deficient worms are less able to survive hypoxia (Jiang et al. 2001; Padilla et al. 2002). HIF-1 protein stability is regulated by the evolutionarily conserved EGL-9/VHL-1 pathway (Epstein et al. 2001). C. elegans egl-9 is orthologous to the mammalian EGLN genes, and the C. elegans homolog of the mammalian pVHL tumor suppressor is encoded by the vhl-1 gene. C. elegans carrying loss-of-function mutations in vhl-1 or egl-9 express HIF-1 protein at abnormally high levels in normoxic conditions (Epstein et al. 2001).
In this study, we investigate vhl-1-independent pathways that attenuate HIF-1 activity. We present evidence that EGL-9 functions independently of VHL-1 to repress HIF-1 activity, in addition to its well-described role in the VHL-1 pathway that controls HIF-1 stability. We also identify the rhy-1 gene, which inhibits expression of HIF-1 target genes, but has no apparent role in regulation of HIF-1 stability. Collectively, these data support models in which RHY-1 and EGL-9 function in VHL-1-independent pathway(s) to regulate HIF-1 localization or transcriptional activity.
MATERIALS AND METHODS
C. elegans strains and culture:
C. elegans were grown at 20° using standard methods (Brenner 1974). The wild-type strain was Bristol N2. CB4856 was used for single-nucleotide polymorphism (SNP) mapping. The following mutant alleles have been described previously: LGII, unc-4(e120) and dpy-10(e128); LGIII, unc-119(ed3) (Maduro and Pilgrim 1995); LGV, egl-9(sa307), egl-9(n586), egl-9(n571) (Trent et al. 1983; Darby et al. 1999), and hif-1(ia04) (Jiang et al. 2001); and LGX, vhl-1(ok161) (Epstein et al. 2001).
Genomic integration of the nhr-57∷GFP reporter:
The nhr-57∷GFP construct was a gift from Piali Sengupta (Miyabayashi et al. 1999). To integrate this transgene into the genome, we employed microparticle bombardment (Praitis et al. 2001). The bombardment was carried out using a BioRad (Hercules, CA) Biolistic PDS/HE 1000 device, 28 in. of Hg vacuum, and 650 psi rupture disc pressure. For each bombardment, 1 μg of unc-119 rescue plasmid pDP#MMO16b (a gift from Judith Austin) and 0.5 μg of nhr-57∷GFP plasmid were coupled to 0.1 mg of 0.6-μm microcarrier gold beads, and the beads were bombarded onto a layer of unc-119(ed3); vhl-1(ok161) hermaphrodites within a 35-mm diameter circle on 100-mm NGM plates without OP50 bacteria. Worms were recovered onto “opti-gro” plates with OP50 1 hr after bombardment. Non-unc rescued transformants were identified 2 weeks after the bombardment. The iaIs07 insertion [nhr-57∷GFP; unc-119 (+)] was mapped to LGIV using single-nucleotide polymorphisms (Wicks et al. 2001). The reporter then was crossed into various genetic backgrounds.
RNA interference assay:
To create DNA constructs for bacteria-mediated RNA interference (RNAi), cDNAs provided by Yuji Kohara were subcloned into the L4440 double-T7 vector (Timmons and Fire 1998). When cDNAs were not available, coding sequences were PCR amplified using primers described previously (Reinke et al. 2000). The resulting constructs were transformed into the Escherichia coli strain HT115 (DE3) (Timmons and Fire 1998). RNAi analyses were performed as described previously (Kamath et al. 2001).
Forward genetic screen and mapping of ia38:
The worm strain ZG119 vhl-1(ok161); iaIs07[nhr-57∷GFP; unc-119(+)] was mutagenized with ethyl methanesulfonate (EMS) using standard methods (Jorgensen and Mango 2002). A fluorescent stereomicroscope was used to identify F2 progeny that misexpressed the nhr-57∷GFP reporter. Over 12,000 genomes were screened, and eight mutant strains were isolated. The ia38 allele is described here. The ia38 mutation was mapped by a combination of single-nucleotide polymorphisms (Wicks et al. 2001), deletion mapping, and three-point mapping. To create a strain suitable for mapping mutations relative to SNPs, the Hawaiian strain CB4856 was crossed to ZG119 vhl-1(ok161); iaIs07[nhr-57∷GFP; unc-119(+)] to generate the strain ZG160. ZG160 was then backcrossed to CB4856 three times to create the strain ZG194. ZG194 is homozygous for at least 15 SNPs from the Hawaiian strain, and these were used to map mutations to specific chromosomes. Linkage of mutations to SNPs was assayed by bulk-segregated analysis (Wicks et al. 2001). Three-point mapping placed the ia38 mutation between dpy-10 and unc-4 on LGII, at the approximate genetic map position 1.07 cM.
Generation of rhy-1 expression constructs:
To generate the rhy-1∷GFP construct pSC09, DNA was PCR amplified using the forward primer (PstI) 5′-GGCGCTGCAGCTTGGTAAGCTGACTTTCAACACG and the reverse primer (BamHI) 5′-GCGGGATCCCGTGTGGCATATTGAGATGTCAAATG. The PCR product included the rhy-1 1.7-kb upstream promoter sequence and whole genomic sequence except the sequence for the last four codons. To construct pSC15, DNA was PCR amplified using the forward primer (PstI) 5′-CACCTGCAGCAATGTATTTCAAAGAAGG and the reverse primer (XbaI) 5′-GTGTCTAGATGGCGATGATGACATATATGTC. The PCR product included a 1.8-kb upstream promoter sequence and +15 bp relative to the rhy-1 translational start site. The PCR products were cloned into the green fluorescent protein (GFP) vectors pPD95.75 and pPD95.77, respectively (provided by Andy Fire). To create transgenic animals, a solution of 20 ng/μl of pSC15 or pSC09 was injected into the gonadal syncitium, with 100 ng/μl rol-6 (+) as a co-injection marker (Mello et al. 1991). Three independent transgenic lines were examined for each construct.
L4 stage worms were collected and boiled for 5 min in 1× SDS sample buffer prepared as described (Hajdu-Cronin et al. 2004). The samples were loaded onto a 10% SDS–PAGE gel, size fractionated, and transferred onto Pure Nitrocellulose membrane (Osmonics, Minnetonka, MN). To detect nhr-57∷GFP protein, 20 L4 stage worms were used in each sample. A commercially available GFP-specific mAb (Roche Diagnostics) was used in 1:1000 dilution. Secondary antibody (anti-mouse IgG and IgM from rabbit; Pierce, Rockford, IL) was used in 1:2000 dilution. To detect HIF-1 protein, 80 μg of L4 stage worm extract was loaded into each lane. Polyclonal anti-HIF-1 sera were a gift from Peter J. Ratcliffe. These antisera were produced by rabbits immunized with a glutathione-S-transferase fusion protein that included amino acids 360–498 of C. elegans HIF-1 (Epstein et al. 2001). This anti-HIF-1 was used in 1:1000 dilution, and secondary antibody (anti-rabbit IgG from goat, Jackson ImmunoResearch, West Grove, PA) was used in 1:2000 dilution. The Western blot images were analyzed by Image J software.
Synchronized L4 stage worms were grown on 100-mm plates using standard methods. Total RNA was isolated using Trizol reagent. RNA samples were treated by the TURBO DNA-free kit (Ambion, Austin, TX) to remove the DNA contamination. One microgram of total RNA from each sample was converted to complementary DNA using the anchored oligo(dT) primer according to the manufacturer's protocol (EndoFree RT kit, Ambion). Fifty nanograms of total RNA from each sample was used for each real-time RT–PCR reaction using the Full Velocity SYBR Green QPCR Master Mix kit (Stratagene, La Jolla, CA). Quantitative RT–PCR was performed using the Stratagene Mx4000 multiplex PCR system. The specific primers for K10H10.2, F22B5.4, and inf-1 were designed and synthesized by Integrated DNA Technologies. The standard curve method was used to analyze the expression levels. inf-1 is not regulated by hypoxia (Shen et al. 2005) and was used as a control. The experiments were repeated in triplicate.
To develop a visual assay for HIF-1 transcriptional activity, we examined the expression of reporter constructs in which the expression of GFP was controlled by regulatory sequences from HIF-1 target genes (Shen et al. 2005). Of the reporters that we assayed, nhr-57∷GFP (Miyabayashi et al. 1999) was the most reliable reporter of HIF-1 function, and we used microparticle bombardment to integrate the nhr-57∷GFP into the C. elegans genome. This reporter was expressed at very low levels in wild-type animals in room air (Figure 1A). Hypoxia treatment increased nhr-57∷GFP levels markedly (Figure 1C). Prior studies have shown that mutants defective for vhl-1 express HIF-1 protein and nhr-57 mRNA at constitutively high levels (Figure 2A; Epstein et al. 2001; Bishop et al. 2004; Shen et al. 2005). We determined that the nhr-57∷GFP reporter was expressed at high levels in vhl-1 mutants and that nhr-57∷GFP expression required hif-1 function (Figure 1, E and G). Thus, nhr-57∷GFP was a useful visual assay for HIF-1 activity.
vhl-1-independent functions of egl-9:
We were intrigued to find that egl-9-deficient animals expressed nhr-57∷GFP at higher levels than did vhl-1 mutants. Both egl-9(sa307) and vhl-1(ok161) are strong loss-of-function alleles. While expression of the reporter was strongly enriched in the intestine of vhl-1(ok161) animals (Figure 1E), nhr-57∷GFP fluorescence was also visible in the hypodermis and pharynx of egl-9(sa307) mutants (Figure 1I). A similar pattern and level of reporter gene expression was observed with the egl-9(n586) and egl-9(n571) loss-of-function alleles (data not shown). This observation was consistent with an independent, concurrent study in which nhr-57 mRNA levels were found to be higher in egl-9 mutant animals compared to vhl-1 mutants (Bishop et al. 2004). To further characterize the egl-9 mutant phenotype, we employed real-time RT–PCR to quantitate the expression K10H10.2 and F22B5.4 mRNAs. In prior studies, we had established that these two genes were induced by hypoxia in a hif-1-dependent manner (Shen et al. 2005). As shown in Figure 2B, loss-of-function mutations in vhl-1 or egl-9 caused F22B5.4 mRNA to be overexpressed 11-fold and 42-fold, respectively. A similar pattern was seen with K10H10.2 mRNA expression. vhl-1-deficient worms expressed K10H10.2 mRNA at levels 3-fold higher than found in wild type. By comparison, egl-9(sa307) mutants overexpressed K10H10.2 mRNA 30-fold, relative to wild type. This difference between the egl-9(sa307) and vhl-1(ok161) phenotypes was highly significant (Figure 2B, P < 0.00001). These analyses provide strong genetic evidence that EGL-9 acts through both VHL-1-dependent and VHL-1-independent pathways to antagonize HIF-1.
We considered the possible mechanisms by which EGL-9 might decrease expression of HIF-1 target genes independently of VHL-1. One possibility was that once HIF-1 was hydroxylated by EGL-9, it was targeted for degradation by an E3 ligase other than VHL-1. This model predicted that HIF-1 protein levels would be significantly higher in egl-9 mutant animals compared to vhl-1 mutant animals. We employed immunoblots to test this hypothesis (Figure 2A). HIF-1 protein was expressed at similar levels in vhl-1(ok161) and egl-9(sa307) animals (averaging 4.8-fold and 4.5-fold greater than wild type, respectively). We also examined the expression of a transgene in which HIF-1 protein was tagged with GFP (Jiang et al. 2001). Strong loss-of-function mutations in vhl-1 or egl-9 resulted in similar levels of HIF-1∷GFP expression (data not shown). Thus, EGL-9 and VHL-1 act in concert to regulate HIF-1 protein levels, and EGL-9 also functions via a separate VHL-1-independent pathway to inhibit HIF-1 activity.
rhy-1, an integral membrane protein that inhibits HIF-1 function:
To further define the vhl-1-independent pathway(s) by which egl-9 and other genes regulated HIF-1 activity, we employed two experimental strategies. First, we initiated a forward genetic screen to identify mutations that caused an nhr-57∷GFP expression pattern similar to that seen in egl-9 mutants (described below). Second, we used bacterially mediated RNAi to test the hypothesis that some HIF-1 target genes might be negative regulators of HIF-1 function. Prior studies had shown that C. elegans EGL-9 and the mammalian EGLN1/PHD2 and EGLN3/PHD3 genes acted in negative feedback loops to attenuate HIF activity (D'Angelo et al. 2003; Aprelikova et al. 2004; Bishop et al. 2004; Shen et al. 2005). Similarly, the mammalian CITED2 gene has been shown to be induced by HIF-1 and to feed back to block binding of HIF-1α to CBP/p300 (Bhattacharya et al. 1999). In a previous study, we identified 63 C. elegans genes that were induced by hypoxia in a HIF-1-dependent manner (Shen et al. 2005), and we will refer to these here as HIF-1 target genes. To test the hypothesis that some genes activated by C. elegans HIF-1 functioned to limit HIF-1 activity, we used bacterially mediated RNAi (Kamath et al. 2001) to reduce expression of 9 HIF-1 target genes. RNAi for W07A12.7 increased and expanded nhr-57∷GFP expression. This phenotype was similar to that exhibited by egl-9 mutant worms (data not shown; described further below). On the basis of the characterization of the loss-of-function phenotype, we named the W07A12.7 gene rhy-1 (regulator of hypoxia-inducible factor).
Concurrently, we conducted a forward genetic screen for EMS-induced mutations that increased the expression of the nhr-57∷GFP reporter in vhl-1(ok161) animals. Here, we describe the ia38 allele. Animals that were homozygous for the ia38 mutation exhibited a phenotype similar to that caused by rhy-1 RNAi. We mapped ia38 to a 1.1-cM genomic interval between dpy-10 and unc-4 on chromosome II. The mutant phenotype and genomic location suggested that the ia38 mutation might be in the rhy-1 gene. In support of this hypothesis, we determined that an extrachromosomal array containing a wild-type copy of rhy-1 was able to rescue the ia38 phenotype. We sequenced the rhy-1 gene in ia38 mutants and determined that ia38 animals contained a missense mutation that changed a conserved serine to phenylalanine (the position of Ser157 is indicated by an arrow in Figure 3, A and B). At our request, the C. elegans Knockout Consortium isolated two deletion mutations in rhy-1, which are diagrammed in Figure 3A. The ok1402 lesion deletes the region that encodes amino acids 246–429 (>36% of the predicted rhy-1 coding sequence). The rhy-1(ok1398) mutation deletes the genomic region encoding amino acids 86–337. Animals that are homozygous for either ok1398 or ok1402 overexpress the nhr-57∷GFP reporter, and this phenotype is suppressed by a strong loss-of-function mutation in hif-1 (Figure 1, M and O; data not shown). The deletion mutations fail to complement the ia38 allele. We conclude that ia38 is a mutation in the rhy-1 gene. The three rhy-1 alleles (ia38, ok1402, and ok1398) have similar phenotypes, as assessed by morphological defects and nhr-57∷GFP expression patterns.
The RHY-1 protein is predicted to be an integral membrane protein with up to 11 membrane-spanning domains and an acyltransferase-3 domain (previously annotated as domain of unknown function 33). This large motif is found in a range of bacterial enzymes that catalyze the transfer of acyl groups, other than amino-acyl, from one compound to another. The molecular functions of metazoan proteins belonging to this gene family are unknown. The serine residue mutated in the rhy-1 (ia38) allele is highly conserved. Figure 3B is an alignment of the amino acids surrounding this serine residue for related worm, human, mouse, and bacterial proteins. Computational analyses suggest that RHY-1 is localized to the plasma membrane or to the endoplasmic reticulum (http://psort.nibb.ac.jp).
There are 63 other genes in the C. elegans genome that contain acyltransferase-3 domains, including nrf-6, ndg-4, and W07A12.6. Loss-of-function mutations in ndg-4 or nrf-6 cause defects in yolk lipoprotein export from intestinal cells and confer resistance to fluoxetine (Choy and Thomas 1999; Choy et al. 2006). nrf-6 loss-of-function mutants do not mis-express the nhr-57∷GFP reporter (data not shown). The W07A12.6 gene is adjacent to rhy-1, and it is predicted to encode a protein with 78% sequence identity to the rhy-1 gene product. W07A12.6 mRNA is induced by hypoxia (Shen et al. 2005). We were not able to deplete W07A12.6 mRNA by RNAi without also targeting rhy-1, due to the high level of sequence similarity between the two genes (data not shown).
Comparison of rhy-1 and egl-9 loss-of-function phenotypes:
The morphological defects in rhy-1(ok1402) and egl-9(sa307) mutants were similar, and the two mutations resulted in comparable patterns of nhr-57∷GFP expression (Figure 1, I and M). Others have shown that the egl-9 egg-laying defect is suppressed by a strong loss-of-function mutation in hif-1 (Bishop et al. 2004). Thus, the morphological defects in egl-9-deficient worms are due, in large part, to abnormally high levels of HIF-1 function. Animals that were homozygous for any of the three rhy-1 alleles exhibited egg-laying defects and were slightly dumpy. These morphological defects were suppressed by the hif-1(ia04) deletion allele. The rhy-1(ok1402); hif-1(ia04) double mutants produced an average of only 21 self-progeny/hermaphrodite (of 45 broods scored). By comparison, the rhy-1(ok1402) mutation reduced the brood size much less dramatically, to 188 self-progeny. Wild-type hermaphrodites and hif-1(ia04) single mutants produced 251 and 253 self-progeny, respectively (of 20 broods assayed). The low fecundity of rhy-1(ok1402); hif-1(ia04) hermaphrodites was rescued by mating to wild-type males. This suggested that rhy-1(ok1402); hif-1(ia04) double-mutant hermaphrodites produced low numbers of functional sperm. In contrast, egl-9(sa307); hif-1(ia04) double mutants had normal brood sizes. These analyses suggest that while both rhy-1 and egl-9 act to inhibit expression of certain HIF-1 target genes, rhy-1 has roles in sperm development that do not require egl-9 function.
RHY-1 expression pattern:
The rhy-1-defective phenotypes suggested that rhy-1 function was required in multiple tissues. To gain further insight into RHY-1 function, we created RHY-1∷GFP reporter constructs and assayed their expression patterns. The pSC15 rhy-1∷GFP fusion included only five rhy-1 codons (Figure 4A). This reporter was expressed in several tissues. In adults, this reporter was expressed highly in the intestine and in certain sensory neurons in the head and at lower levels in body-wall muscles and socket cells (Figure 4C). During the last larval stage, GFP was visible in the vulva, in cells in the ventral nerve cord, and in cells in the tail (Figure 4D). Larval stage animals expressed pSC15 at high levels in the hypodermis (Figure 4E). Hypoxia treatment increased the levels of reporter gene expression (data not shown). Prior studies have localized egl-9∷GFP to pharyngeal muscle, body-wall muscle, vulval muscles, and sensory neurons of the head and tail (Darby et al. 1999). Thus, the expression patterns of rhy-1∷GFP and egl-9∷GFP partially overlap.
The pSC09 construct contained the predicted rhy-1 coding sequences, the three rhy-1 introns, and 1.7 kb of sequence 5′ to the start codon (Figure 4A). This fusion gene was expressed at low levels, but it rescued all three of the rhy-1 mutant alleles, as assayed by nhr-57∷GFP expression (data not shown). In animals carrying pSC09, GFP expression was perinuclear and was visible in intracellular reticular patterns. This was consistent with the prediction that RHY-1 is localized to the endoplasmic reticulum.
RHY-1 inhibits HIF-1 activity via a VHL-1-independent pathway:
As shown in Figure 1 and in Figure 5C, rhy-1(ok1402) and egl-9(sa307) animals expressed nhr-57∷GFP at high levels, relative to vhl-1(ok161) animals. To confirm that rhy-1 inhibited the expression of HIF-1 target genes other than nhr-57∷GFP, we assayed expression of two endogenous mRNAs, K10H10.2 and F22B5.4, by real-time RT–PCR. Mutation of rhy-1 significantly increased the expression of both genes (P < 0.0001 and P < 0.01, respectively), and this effect was suppressed by the hif-1(ia04) deletion mutation (Figure 5, A and B). Expression of K10H10.2 and F22B5.4 mRNAs was higher in rhy-1(ok1402); vhl-1(ok161) double mutants than in either single mutant (Figure 5, A and B). Double mutants also expressed nhr-57∷GFP protein at significantly higher levels than did either single mutant (Figure 5C, P < 0.05). Interestingly, the level of F22B5.4 mRNA in rhy-1 and vhl-1 mutants was similar, but K10H10.2 mRNA was expressed at a significantly higher level in rhy-1 mutants (Figure 5A, P < 0.00001). This suggests that the relative importance of the VHL-1 and RHY-1 pathways may vary between cell types and promoters.
To address whether RHY-1 regulates HIF-1 expression, HIF-1 activity, or both, we assayed hif-1 mRNA and HIF-1 protein levels. The rhy-1(ok1402) mutation resulted in a slight increase in hif-1 mRNA levels (1.8-fold) (Figure 5D) and HIF-1 protein (2.2-fold) (Figure 5E). By comparison, HIF-1 protein levels in vhl-1(ok161) and egl-9(sa309) animals were 4.8-fold and 4.5-fold greater than in wild-type animals (Figure 5E), while hif-1 mRNA levels were not changed (Figure 5D). This is consistent with a model in which RHY-1 has a minor effect on HIF-1 protein expression and functions primarily in a VHL-1-independent pathway to limit HIF-1 function.
If EGL-9 and RHY-1 acted in distinct pathways, then we would expect that the expression of HIF-1 target genes might be higher in the double mutant than in either single mutant. To test this, we assayed the expression of nhr-57∷GFP protein, K10H102 mRNA, and F22B5.4 mRNA in rhy-1(ok1402), egl-9(sa307) and double-mutant animals. For all three genes assayed, expression in rhy-1(ok1402); egl-9(sa307) double mutants was not enhanced relative to expression in egl-9 single mutants (Figure 6, A and B). While not conclusive, these data leave open the possibility that RHY-1 and EGL-9 function in the same pathway.
The transcription factor HIF is a central regulator of physiological and pathological adaptation to hypoxia. The regulation of HIF activity occurs at the multiple levels, and the cellular networks that control HIF include negative and positive interactions (Chun et al. 2002; Dery et al. 2005). We have employed genetic strategies to further understand the regulatory circuits that control transcriptional responses to hypoxia.
Dual functions for EGL-9:
We conclude that EGL-9 inhibits HIF-1 function via VHL-1-independent pathway(s), in addition to its well-described role in controlling HIF-1 stabilization. The EGL-9/VHL-1 pathway has been studied intensively. EGL-9 hydroxylates HIF-1 in an oxygen-dependent manner. This modification results in VHL-1 binding and subsequent ubiquitination and proteasomal degradation of HIF-1 (Epstein et al. 2001). In this study, we show that egl-9 mutants express HIF-1 target genes at higher levels than do vhl-1 mutants. Moreover, we show that this difference in HIF-1 function is not attributable to differences in HIF-1 protein levels. These results provide strong support for models in which EGL-9 represses the activity of HIF-1 protein.
Two recent studies in mammalian cell lines provide further insight into mechanisms by which EGLN proteins might regulate HIF activity independently of VHL. First, the candidate tumor suppressor ING4 was shown to form a complex with EGLN1/HPH-2. Additionally, EGLN1/HPH-2 and ING4 were shown to associate with DNA-binding sites for HIF (Ozer et al. 2005). This study provides support for a model in which HPH-2/EGLN1 recruits ING4 and transcriptional repressors to the HIF complex, thereby inhibiting the expression of HIF target genes. Another group has presented further evidence that EGLN1 inhibits HIF-1α transcriptional activity independently of the VHL pathway that regulates HIF-1α protein stability (To and Huang 2005). When considered with these studies, the C. elegans data are consistent with the following model: EGL-9 binds HIF-1 at the conserved LXXLAP motif in the ODD domain. When the oxygen concentration is sufficiently high, EGL-9 hydroxylates HIF-1, resulting in binding to VHL-1 and subsequent degradation of HIF-1 (Epstein et al. 2001). EGL-9 also binds HIF-1 in the nucleus, and EGL-9 recruits certain transcriptional repressors to the regulatory regions of HIF-1 target genes. Clearly, many aspects of this model remain to be tested. However, it is interesting to note that C. elegans EGL-9 and mammalian EGLN1 each include an N-terminal MYND-type zinc-finger domain and that this protein motif is present in certain transcriptional repressors (Lutterbach et al. 1998; Masselink and Bernards 2000; Gottlieb et al. 2002).
Under hypoxic conditions rhy-1 mRNA expression is induced by HIF-1, and RHY-1 feeds back to attenuate HIF-1 activity. In this report, we demonstrate that RHY-1 inhibits expression of HIF-1 target genes via a mechanism that is distinct from the VHL-1-mediated pathway that controls HIF-1 protein levels.
The molecular function(s) of RHY-1 are unknown. The RHY-1 protein is predicted to include up to 11 membrane-spanning domains, and it appears to be localized to the endoplasmic reticulum or to the plasma membrane. Analysis of the RHY-1 sequence reveals a single motif that may be informative to its biochemical function: the acyltransferase-3 domain. The acyltransferase-3 domain arose before the evolution of the hypoxia-inducible factor, and it is predicted to catalyze the transfer of acyl groups, other than amino-acyl, from one compound to another. Thus, RHY-1 may have roles in the synthesis, metabolism, or transport of bioactive lipids that modulate HIF-1 activity. Interestingly, while the C. elegans genome contains 64 genes with acyltransferase-3 motifs, the human, mouse, and Drosophila genomes each contain only one member of this gene family. It will be interesting to elucidate how RHY-1 modulates HIF-1 regulatory circuits and to determine whether the mammalian acyltransferase-3 gene family members have roles in hypoxia response.
There are at least three alternative models to describe how RHY-1 might attenuate expression of HIF-1 target genes. The simplest model is that RHY-1 binds HIF-1 directly and inhibits nuclear translocation of the transcription factor. A second possibility is that RHY-1 has a role in the metabolism or transport of bioactive lipids, which then control HIF-1 localization or transcriptional activity. This model would be consistent with the critical role of Ser157. The serine residue that is mutated in the rhy-1(ia38) allele is predicted to be in the second transmembrane domain of RHY-1 and could be involved in putative transporter functions for RHY-1. In further support of this model, the C. elegans NRF-6 acyltransferase-3 protein has been shown to interact with NRF-5, a predicted lipid-binding protein (Choy et al. 2006). Third, RHY-1 might function indirectly by modulating other cellular processes that ultimately control HIF-1 function. Two of these models invoke as-of-yet-unidentified sensors of lipids or other molecules regulated by RHY-1, and these sensors are postulated to control HIF-1 function. Clearly, it will be informative to identify other genes that interact with RHY-1 to control HIF-1. RHY-1 has an essential role in a negative feedback loop that limits expression of HIF-1 target genes, and we favor models in which regulating HIF-1 is a primary function for RHY-1.
Both EGL-9 and RHY-1 act in negative feedback loops to limit transcription of HIF-1 target genes, but it is not yet clear whether these two proteins act in concert. Interestingly, the repressive function of EGL-9 appears to be greater than that of RHY-1. As described above, EGL-9 might recruit transcriptional repressors to the HIF-1 complex. If this is true, then RHY-1 might regulate the localization or association of HIF-1, EGL-9, and putative repressors.
When considering these models, it is interesting to note that hypoxia treatment resulted in a nhr-57∷GFP expression pattern that was similar to that seen in vhl-1 mutants, but less intense than that seen in rhy-1- or egl-9-defective animals (see Figure 1). Prior studies have established that hypoxia inhibits EGL-9 enzymatic activity and VHL-1-dependent degradation of HIF-1 (Epstein et al. 2001). However, it is possible that EGL-9 oxygenase activity is not required for the VHL-1-independent functions of EGL-9. Further studies will address these intriguing hypotheses, as they will ultimately inform our understanding of the regulatory circuits that control the hypoxia-inducible factor during development, homeostasis, and disease.
We thank Piali Sengupta, Peter Ratcliffe, Ann Rose, and Andy Fire for providing plasmids or antibodies used in these studies. We are grateful to Abhijeet Shah, Kelly Gillette, and Mae Young for their roles in isolating and mapping the rhy-1(ia38) allele. Mutant strains were obtained from the Caenorhabditis Genetics Center, which is supported by the National Institutes of Health National Center for Research Resources. An American Heart Association Established Investigator award to J.A.P.C. provided funding for this work.
Communicating editor: K. Kemphues
- Received July 17, 2006.
- Accepted September 5, 2006.
- Copyright © 2006 by the Genetics Society of America