Oxygen sensing in Drosophila: multiple isoforms of the prolyl hydroxylase fatiga have different capacity to regulate HIFalpha/Sima

Abstract

Background: The Hypoxia Inducible Factor (HIF) mediates cellular adaptations to low oxygen. Prolyl-4-hydroxylases are oxygen sensors that hydroxylate the HIF alpha-subunit, promoting its proteasomal degradation in normoxia. Three HIF-prolyl hydroxylases, encoded by independent genes, PHD1, PHD2, and PHD3, occur in mammals. PHD2, the longest PHD isoform includes a MYND domain, whose biochemical function is unclear. PHD2 and PHD3 genes are induced in hypoxia to shut down HIF dependent transcription upon reoxygenation, while expression of PHD1 is oxygen-independent. The physiologic significance of the diversity of the PHD oxygen sensors is intriguing.

Methodology and principal findings: We have analyzed the Drosophila PHD locus, fatiga, which encodes 3 isoforms, FgaA, FgaB and FgaC that are originated through a combination of alternative initiation of transcription and alternative splicing. FgaA includes a MYND domain and is homologous to PHD2, while FgaB and FgaC are shorter isoforms most similar to PHD3. Through a combination of genetic experiments in vivo and molecular analyses in cell culture, we show that fgaB but not fgaA is induced in hypoxia, in a Sima-dependent manner, through a HIF-Responsive Element localized in the first intron of fgaA. The regulatory capacity of FgaB is stronger than that of FgaA, as complete reversion of fga loss-of-function phenotypes is observed upon transgenic expression of the former, and only partial rescue occurs after expression of the latter.

Conclusions and significance: Diversity of PHD isoforms is a conserved feature in evolution. As in mammals, there are hypoxia-inducible and non-inducible Drosophila PHDs, and a fly isoform including a MYND domain co-exists with isoforms lacking this domain. Our results suggest that the isoform devoid of a MYND domain has stronger regulatory capacity than that including this domain.

Figure 1. The fatiga gene locus.

A) Schematic representation of the fga locus. Grey boxes represent exons, black lines are introns and arrows indicate transcription initiation sites. B) fgaA, fgaB and fgaC transcripts are generated by a combination of alternative splicing and alternative initiation of transcription. Coding regions are represented by grey boxes, and untranslated regions (UTRs) are representated by white boxes. C) All three transcripts give rise to proteins containing a prolyl hydroxylase domain, whereas only FgaA has a MYND domain.

Figure 2. fgaB but not fgaA is induced in hypoxia in a Sima-dependent manner.

A) Relative expression of fgaA and fgaB during development, as measured by real time PCR, and shown as fold-change expression. Whereas expression levels of fgaA remain constant, expression of the fgaB transcript rises at the adult stage. Error bars represent standard deviations. B) fgaB but not fgaA is induced in hypoxia (5% O₂). In sima⁰⁷⁶⁰⁷ homozygous mutant embryos this induction is completely abrogated. Fold-change expression levels are relative to the expression levels in wild type (w¹¹¹⁸) embryos maintained in normoxia; error bars represent standard deviations. C) fgaB transcript levels are strongly upregulated in embryos over-expressing Sima. fgaA and fgaB mRNA expression as assessed by real time PCR; fold changes are relative to w¹¹¹⁸ expression levels.

Figure 3. The fgaB regulatory region contains a functional HIF Responsive Element.

A) fgaB but not fgaA mRNA is induced in S2 cells exposed to hypoxia (1% O₂). Real time-PCR analysis; fold-change expression levels are relative to expression in normoxia; error bars represent standard deviations. B) Schematic representation of the fgaB [−1170 −555] regulatory region. The black triangles represent putative HIF Responsive Elements (HREs). The sequence and position of presumptive HREs 1 to 7 is indicated; the HRE core consensus is highlighted in bold font. C) The [−1170 −555] DNA fragment includes functional HRE sequences. The [−1170 +1], [−555 +1] and [−1170 −555] luciferase reporter constructs were transiently co trasfected in S₂ cells, along with the Sima expression vector pAc5.1-Sima (+Sima) or along with empty vector pAc5.1-V5 (-Sima). pGl3-basic or pGl3-Hsp vectors were included as controls. Whereas the [−555 +1] fragment fails to induce Sima-dependent transcription of the reporter, the fragments [−1170 +1] and [−1170 −555] provoke strong induction of the reporter. Luciferase activity is expressed as fold induction relative to that of the corresponding empty control vector. Error bars represent the standard deviation of duplicate luciferase determinations. Each experiment was repeated at least 3 times; one representative experiment is shown. D) The only functional HRE in the [−1170 +1] interval is HRE2. S₂ cells were transiently transfected with the reporter construct [−1170 +1] containing all 7 wild type HREs or each of the mutagenized versions of the [−1170 +1] reporter (HRE1m to HRE5m and HRE6-7Δ). Each of these reporters was co-transfected along with a Sima expression plasmid (+Sima) or with an empty vector (-Sima). Expression of Sima upregulated all the reporters with the exception of HRE2m. Data are shown as in A and B. (E) HRE-dependent induction in S2 cells exposed to hypoxia: The reporter constucts [−1170 +1], [−555 +1] and [1170 +1] HRE 2m, as well as the pGl3-basic empty vector were transiently transfected in S2 cells, which were later transferred to hypoxia (1%O₂) or maintained in normoxia (21% O₂), after which luciferase activity was assessed. HRE-luc reporter was included as a hypoxia control. The [−1170 +1] reporter was strongly induced in hypoxia, while the expression of the reporters [−555 +1] and [1170 +1] HRE 2m did not increase in hypoxia. Luciferase activity is presented as fold induction relative to that of control cells maintained in normoxia.

Figure 4. FatigaA and FatigaB differentially regulate Sima.

A) Reversion of developmental viability of fga¹ homozygous mutants upon transgenic expression of FgaA or FgaB. Expression of FgaB completely reverted the fga¹ lethal phenotype, as individuals reached the adult stage. FgaA expression enabled development only to the pupal stage. (+) indicates the occurrence of individuals of the indicated developmental stage; (−) indicates absence of individuals of the mentioned stage. B) Expression of FgaB but not FgaA reverts the constitutive expression of the ldh-LacZ reporter that occurs in fga¹ mutant embryos in normoxia, as shown in the immunofluorescence micrograph after anti beta-gal staining. The arrow indicates groups of cells that express the ldh-LacZ reporter. C) Fga B is more efficient than FgaA in restoring tracheal liquid clearence in fga¹ mutant first-instar larvae. Tracheal analysis of fga¹ mutant larvae, and of fga¹ mutant larvae, expressing either FgaA or FgaB under control of a btl-Gal4 driver, was performed under a bright field microscope. Percentage of larvae with air filling defects was quantified, for whose purpose three categories were defined: tracheae full of air (white box), tracheae partially filled with liquid (grey box), and tracheae full of liquid (black box).