1. Developmental Biology

Erythropoietin signaling regulates heme biosynthesis

  1. Jacky Chung
  2. Johannes G Wittig
  3. Alireza Ghamari
  4. Manami Maeda
  5. Tamara A Dailey
  6. Hector Bergonia
  7. Martin D Kafina
  8. Emma E Coughlin
  9. Catherine E Minogue
  10. Alexander S Hebert
  11. Liangtao Li
  12. Jerry Kaplan
  13. Harvey F Lodish
  14. Daniel E Bauer
  15. Stuart H Orkin
  16. Alan B Cantor
  17. Takahiro Maeda
  18. John D Phillips
  19. Joshua J Coon
  20. David J Pagliarini
  21. Harry A Dailey
  22. Barry H Paw  Is a corresponding author
  1. Brigham and Women’s Hospital, Harvard Medical School, United States
  2. Boston Children’s Hospital, Harvard Medical School, United States
  3. University of Georgia, United States
  4. University of Utah School of Medicine, United States
  5. Genome Center of Wisconsin, United States
  6. University of Wisconsin-Madison, United States
  7. Massachusetts Institute of Technology, United States
  8. Dana-Farber Cancer Institute, Harvard Medical School, United States
https://doi.org/10.7554/eLife.24767
Share this article
Cite this article
  1. Jacky Chung
  2. Johannes G Wittig
  3. Alireza Ghamari
  4. Manami Maeda
  5. Tamara A Dailey
  6. Hector Bergonia
  7. Martin D Kafina
  8. Emma E Coughlin
  9. Catherine E Minogue
  10. Alexander S Hebert
  11. Liangtao Li
  12. Jerry Kaplan
  13. Harvey F Lodish
  14. Daniel E Bauer
  15. Stuart H Orkin
  16. Alan B Cantor
  17. Takahiro Maeda
  18. John D Phillips
  19. Joshua J Coon
  20. David J Pagliarini
  21. Harry A Dailey
  22. Barry H Paw
(2017)
Erythropoietin signaling regulates heme biosynthesis
eLife 6:e24767.
https://doi.org/10.7554/eLife.24767
  2. Download BibTeX
  3. Download .RIS

Abstract

Heme is required for survival of all cells, and in most eukaryotes, is produced through a series of eight enzymatic reactions. Although heme production is critical for many cellular processes, how it is coupled to cellular differentiation is unknown. Here, using zebrafish, murine, and human models, we show that erythropoietin (EPO) signaling, together with the GATA1 transcriptional target, AKAP10, regulates heme biosynthesis during erythropoiesis at the outer mitochondrial membrane. This integrated pathway culminates with the direct phosphorylation of the crucial heme biosynthetic enzyme, ferrochelatase (FECH) by protein kinase A (PKA). Biochemical, pharmacological, and genetic inhibition of this signaling pathway result in a block in hemoglobin production and concomitant intracellular accumulation of protoporphyrin intermediates. Broadly, our results implicate aberrant PKA signaling in the pathogenesis of hematologic diseases. We propose a unifying model in which the erythroid transcriptional program works in concert with post-translational mechanisms to regulate heme metabolism during normal development.

https://doi.org/10.7554/eLife.24767.001

eLife digest

Heme is an iron-containing compound that is important for all living things, from bacteria to humans. Our red blood cells use heme to carry oxygen and deliver it throughout the body. The amount of heme that is produced must be tightly regulated. Too little or too much heme in a person’s red blood cells can lead to blood-related diseases such as anemia and porphyria. Yet, while scientists knew the enzymes needed to make heme, they did not know how these enzymes were controlled.

Now, Chung et al. show that an important signaling molecule called erythropoietin controls how much heme is produced when red blood cells are made. The experiments used a combination of red blood cells from humans and mice as well as zebrafish, which are useful model organisms because their blood develops in a similar way to humans. When Chung et al. inhibited components of erythropoietin signaling, heme production was blocked too and the red blood cells could not work properly.

These new findings pave the way to look at human patients with blood-related disorders to determine if they have defects in the erythropoietin signaling cascade. In the future, this avenue of research might lead to better treatments for a variety of blood diseases in humans.

https://doi.org/10.7554/eLife.24767.002

Introduction

Heme biosynthesis is a fundamental biological process that is highly conserved and involves eight enzymatic reactions that occur both in the cytosol and mitochondria (Severance and Hamza, 2009). In vertebrates, the most recognized role of heme is to serve as the oxygen-binding moiety in hemoglobin expressed by red blood cells (RBCs). During RBC maturation, heme metabolism genes are robustly upregulated (Chung et al., 2012; Nilsson et al., 2009; Yien et al., 2014). Not surprisingly, mutations in these genes are most commonly associated with hematologic defects in humans, underscoring the importance for a better understanding of the factors regulating heme biosynthesis. In particular, loss-of-function mutations in FECH (EC 4.99.1.1), which encodes the terminal rate-limiting enzyme in heme production, is strongly associated with the disease erythropoietic protoporphyria (EPP) (Balwani and Desnick, 2012; Langendonk et al., 2015).

The dependence of RBC biology on heme metabolism makes erythropoiesis an excellent system to gain insight into this process. Previous genetic analyses using RBCs have identified several mechanisms regulating heme metabolism most of which are transcriptional networks controlling mRNA expression of heme metabolism genes (Amigo et al., 2011; Handschin et al., 2005; Kardon et al., 2015; Nilsson et al., 2009; Phillips and Kushner, 2005; Shah et al., 2012; Shaw et al., 2006; Wingert et al., 2005; Yien et al., 2014). Currently, however, transcription-independent signaling mechanisms regulating heme production are poorly understood (Chen et al., 2009; Paradkar et al., 2009). Such mechanisms may play a critical role to couple heme metabolism to changes in the extracellular milieu, homeostasis, and development.

Here, we show that heme production is regulated by EPO/JAK2 signaling in concert with the GATA1 target, Akap10 (Fujiwara et al., 2009). During red blood cell (RBC) development, PKA expression becomes increased at the mitochondrial outer membrane (OM) through AKAP10-dependent recruitment. We found that OM PKA catalytic (PKAc) subunits become disengaged from the autoinhibitory PKA regulatory (PKAr) subunits through direct interaction with phosphorylated STAT5 downstream of EPOR activation. Furthermore, we demonstrate that FECH is a kinase target of OM PKA and its phosphorylation triggers upregulation of its activity that is required to support erythropoiesis in vivo. Our work uncovers a previously unknown facet of heme metabolism with implications on human disease.

Results

Mitochondrial PKA expression increases with erythroid maturation

To begin examining post-translational mechanisms regulating heme metabolism, we performed an unbiased comparative analysis of the changing mitochondrial proteome in maturing RBCs. Mitochondria-enriched fractions isolated from undifferentiated and differentiated Friend murine erythroleukemia (MEL) cells were analyzed by quantitative mass spectrometry (Pagliarini et al., 2008) (Figure 1A and B). MEL cells have been reliably used to dissect the molecular mechanisms underlying hemoglobin production in erythroid cells (Bauer et al., 2013; Canver et al., 2014).

Figure 1
Download asset Open asset
PKA activity regulates heme biosynthesis.

(A) A schema detailing the preparation of samples enriched for mitochondria from undifferentiated and differentiated MEL cells is shown. Following preparation of samples from both cohorts, the samples were trypsin digested and labeled with different tandem mass tags (TMTs), followed by mass spectrometry analysis. (B) Enrichment of undifferentiated (undiff) and differentiated (diff) mitochondrial samples was confirmed by western analysis for mitochondrial (PDHA1 and HSPD1) and cytosolic (GAPDH and TUBA1A) markers prior to mass spectrometry analysis. (C) The upregulation of PKA regulatory (PKAr) and catalytic (PKAc) subunits as well as previously identified heme metabolism proteins in mitochondria-enriched fractions of differentiated MEL cells but not several housekeeping proteins is presented in logarithmic scale. Please see Figure 1—source data 1 for precise changes. (D and E) Immunoblot analyses of the expression of PKA subunits in mitochondrial fractions (D) and whole cell lysates (E). All immunoblots were performed twice. Undiff-undifferentiated; Diff-differentiated; IB-immunoblot.

https://doi.org/10.7554/eLife.24767.003
Figure 1—source data 1

Changes in the mitochondrial expression of selection erythroid and housekeeping mitochondrial proteins.

Changes in the expression of proteins together with their accession numbers are shown in log2 scale. These data are depicted in the heat map in Figure 1C.

https://doi.org/10.7554/eLife.24767.004

As expected, erythroid differentiation was associated with the elevated mitochondrial expression of a number of mitochondrial proteins known to have a role in erythropoiesis such as FECH, ATPIF1, and ABCB10 while the expression of housekeeping proteins such as components of the mitochondrial transport and protein translation machinery were relatively unchanged (Figure 1C). Strikingly, we also found increased expression of PKA regulatory (PKAr; PRKAR1A, PRKAR2A, PRKAR2B) and catalytic (PKAc; PRKACA and PRKACB) subunits in mitochondria-enriched fractions (Figure 1C). We independently confirmed these results using immunoblotting with isoform-specific antibodies where we also found increased total expression of these PKA subunits in maturing erythroid cells (Figure 1D and E). We failed to detect PRKACG because it is only present in humans and not found in our murine model (Kirschner et al., 2009). In addition, we also could not detect PRKAR1B since its expression is restricted to neurons (Kirschner et al., 2009). Together, our results suggest that select PKA subunits become highly expressed in mitochondria of developing erythrocytes and that MEL cells are a good model that accurately recapitulates the expected PKA expression pattern.

Mitochondrial PKA is localized to the outer mitochondrial membrane via AKAP10

We wondered whether increased mitochondrial PKA was specific for a particular suborganellular compartment, and next, performed a series of experiments to determine their submitochondrial expression. First, intact mitochondria isolated from maturing erythroid cells were treated with proteinase K that would digest all proteins exposed on the outer mitochondrial membrane. Immunoblot analysis of untreated and treated mitochondria revealed that the majority of PKA subunits were sensitive to proteinase K activity similar to TOM20 while VDAC1, a mitochondrial outer membrane (OM) marker known to be resistant to proteinase K digestion, remained largely unaffected (Figure 2A) (Rapaport, 2003; Shirihai et al., 2000). Biochemical fractionation of the mitochondria OM, intermembrane space (IMS), and mitoplast (MP) followed by immunoblotting confirmed the predominant presence of PKA subunits in the OM fraction of maturing erythroid cells (Figure 2B).

Figure 2
Download asset Open asset
Mitochondrial PKA is localized to the outer membrane during erythropoiesis by AKAP10.

(A) Intact mitochondria isolated from maturing MEL cells (day 3) were untreated or treated with proteinase K and subsequently analyzed by immunoblotted with antibodies specific for the indicated proteins. (B) Mitochondria from day 3 maturing MEL cells were fractionated into the indicated compartments and 5 µg of protein were analyzed with immunoblotting. (C and D) A heat map demonstrating the increased mitochondrial expression of AKAP10 similar to PRKAR2B in maturing erythroid cells (C) that was confirmed using immunoblotting (D). Please see Figure 2—source data 1 for precise changes. (E and F) Proteinase K digestion assay (E) and submitochondrial fractionation (F) showed that AKAP10 is mostly localized to the OM. (G) A schematic depicting the wild-type (Akap10wt) and the two Akap10-null alleles Akap10Cas9(△ex1-2/184delTG) generated using CRISPR/Cas9 genome editing. The positions of the exon 1 (Ex1) and exon 3 (Ex3) CRISPR oligos are denoted. The introns are shown in black with exons in orange. The Akap10Cas9(ex1-2) allele has complete removal of exon 2 and truncates exons 1 and 3 to fuse exons 1’ and 3’, respectively. The Akap10Cas9(184delTG) allele has a two-nucleotide deletion in exon 3 leading to a frameshift and a premature stop codon (Stop’). Both alleles are expected to disrupt the N-terminal region encoding the mitochondrial-targeting motif. (H) The Akap10Cas9(△ex1-2) deleted allele can only be detected when genotyping was performed with primers F and R2 while the Akap10Cas9(184delTG) allele can still be detected with primers F and R1, resembling wild-type. These results were sequence confirmed. (I) Immunoblot analysis showing that neither allele gave rise to any detectable AKAP10 protein. (J and K) Loss of AKAP10 had no effect on total PKA subunit expression but reduced the amount of PKA subunits in whole mitochondria (J) as well as the OM fraction (K). All immunoblots were performed twice. Undiff-undifferentiated; Diff-differentiated; OM-outer membrane; IMS-intermembrane space; MP-mitoplast; WT-wild-type; KO-knockout; IB-immunoblot.

https://doi.org/10.7554/eLife.24767.005
Figure 2—source data 1

Change in the mitochondrial expression of AKAP10 during erythroid maturation.

The increase in AKAP10 (NP_064305) mitochondrial expression is listed in log2 scale. These data are depicted in the heat map in Figure 2C. PRKAR2B and other controls were as previously shown in Figure 1C and Figure 1—source data 1.

https://doi.org/10.7554/eLife.24767.006

A great deal of work has demonstrated that PKA is localized to subcellular compartments through interactions with a family of anchoring proteins called AKAPs (a kinase anchoring proteins) (Wong and Scott, 2004). The majority of AKAPs recruit PKA-RII subunits but not RI (Sarma et al., 2010). However, a subclass of AKAPs can bind to both RI and RII with high affinity to regulate their subcellular distribution and have been referred to as ‘dual-specificity AKAPs’ (Huang et al., 1997a, 1997b; Li et al., 2001; Sarma et al., 2010; Wang et al., 2001). In particular, PAP7, AKAP1, and AKAP10 are three such AKAPs capable of localizing to mitochondria (Huang et al., 1997a, 1997b; Li et al., 2001; Wang et al., 2001; Wong and Scott, 2004). Interestingly, although we failed to detect PAP7 and AKAP1 in our proteomics analysis, we found a pronounced increase in the mitochondrial expression of AKAP10 (Figure 2C and D). Similar to our earlier data with PKA subunits, mitochondrial AKAP10 in maturing erythrocytes is sensitive to proteinase K digestion and primarily found in the OM fraction in maturing erythroid cells (Figure 2E and F).

High-throughput expression analysis has previously shown that AKAP10 expression increases in maturing erythroid cells and is a downstream target of the GATA1 erythroid lineage master transcription factor (Fujiwara et al., 2009; Zhang et al., 2003). To date, it has no known role in erythropoiesis or heme metabolism. However, our results thus far led us to wonder if it was responsible for regulating PKA localization in maturing RBCs, and we tested this by using CRISPR/Cas9-mediated genome editing to introduce null mutations into the endogenous AKAP10 loci. Genotyping and sequencing showed that for one AKAP10 allele [AKAP10Cas9(△ex1-2)], parts of exons 1 (Ex1) and 3 (Ex3) along with all of exon 2 including the ATG start codon were deleted by our targeting strategy (Figure 2G and H). The second allele [AKAP10Cas9(184delTG)] had a 2 base-pair deletion that resulted in a premature stop codon (Stop’) (Figure 2G and H). Neither allele gave rise to full-length AKAP10 protein (KO) as shown by immunoblotting (Figure 2H and I).

Total expression of PKA subunits in maturing KO cells was similar to wild-type (WT) cells (Figure 2J). However, KO cells had reduced the levels of mitochondrial PKA subunits both in intact preparations as well as in OM-specific fractions (Figure 2J and K). These results strongly suggest that AKAP10 recruits PKA to the outer mitochondrial membrane during red cell development and connects the GATA1 transcriptional program to the PKA signaling pathway.

Mitochondrial outer membrane PKA signaling regulates hemoglobinization and erythropoiesis

Mitochondria are the site of heme production required for hemoglobin synthesis and its physiology is a crucial part of RBC maturation (Nilsson et al., 2009; Shah et al., 2012). Given the increase in mitochondrial expression of PKA subunits in maturing erythroid cells, we wondered whether PKA activity had an influence on heme production. We addressed this by first using pharmacologic agents to toggle PKA function. Compounds that activate PKA such as 8-Br-cAMP and forskolin both caused an increase in the proportion of hemoglobinized cells as shown by o-dianisidine staining, which can be blocked by PKA antagonists H-89 or PKI (14-22) (Figure 3A). An increase in the proportion of hemoglobinized cells was also observed when we treated MEL cells with dimethyl-prostaglandin E2 (dmPGE2) (Figure 3B), which is a more stable analog of prostaglandin E2 (PGE2) that has a physiologic role during multiple aspects of hematopoiesis (Goessling et al., 2009; North et al., 2007). The effects of dmPGE2 can be similarly inhibited by PKI (14-22) (Figure 3C), underscoring the specificity of dmPGE2 signaling via PKA. In contrast to PKA inhibition, the PKC inhibitor, bis-indolylmaleimide II, could not block the effects of PKA activation suggesting that the observed changes in heme synthesis are specific for PKA (Figure 3—figure supplement 1A).

Figure 3 with 1 supplement see all
Download asset Open asset
Mitochondrial outer membrane PKA signaling is required for erythropoiesis.

(A–C) o-dianisidine staining for hemoglobinized MEL cells treated with several pharmacologic modulators of PKA activity. (A) PKA activation with 8-Br-cAMP or high-dose forskolin (50 µM) triggers an increase in heme production that is blocked by H89 or PKI (14-22) treatment at day 3 of DMSO induction. (B and C) A similar increase in hemoglobinization was observed with dMPGE2 that was also inhibited by PKI (14-22). (D) Wild-type (WT) or AKAP10-knockout (KO) erythroid cells at day 4 of differentiation were stained with o-dianisidine. (E and F) AKAP10 expression was knocked-down using two different shRNAs (E) that lead to reduced hemoglobinization (F). (G–I) akap10-specific morpholinos (MOs) were used to inhibit akap10 expression in zebrafish embryos (G). These morphants were anemic with reduced hemoglobinization (H) and red blood cell counts (I). *p-value<0.05, Mean ± SEM, n = 3. All immunoblots were performed twice. 8-Br-cAMP-8-bromoadenosine 3’,5’-cyclic monophosphate; dmPGE2-dimethyl-prostaglandin E2; WT-wild-type; KO-knockout; shRNA-short hairpin RNA; MO-morpholino; IB-immunoblot.

https://doi.org/10.7554/eLife.24767.007

Although our results from pharmacologic experiments suggest that widespread modulation of PKA has an impact on heme production on maturing erythroid cells, they do not explicitly examine the role of mitochondrial OM PKA. The precise contributions of distinct mitochondrial pools of PKA have been a topic of controversy (DiPilato et al., 2004; Lefkimmiatis et al., 2013). Emerging evidence suggest that PKA agonists such as forskolin and cAMP cannot diffuse into the mitochondrial matrix (Acin-Perez et al., 2009; Lefkimmiatis et al., 2013). Thus far, our pharmacologic data involved the use of a high-dose of forskolin (Figure 3A–C), and when the dose was titrated down to one that was previously shown to not activate matrix PKA, we also failed to detect an effect on hemoglobinization (Figure 3—figure supplement 1B) (Acin-Perez et al., 2009).

It is difficult to rely solely on pharmacologic data to unambiguously dissect the contributions of intracellular PKA pools since dose responses are known to vary form one cell type to another (Humphries et al., 2007; Lefkimmiatis et al., 2013). However, the reduction in the levels of PKA subunits in the mitochondrial OM of AKAP10-KO maturing erythroid cells allowed us to genetically and biochemically examine the functional role of this PKA signaling compartment. Compared to wild-type controls, AKAP10-KO maturing erythroid cells exhibited a deficit in hemoglobinization (Figure 3D). This defect was also observed when AKAP10 expression was inhibited using two distinct shRNAs (Figure 3E and F).

We next examined the in vivo significance of AKAP10 and mitochondrial OM PKA signaling by using morpholinos to block akap10 expression in zebrafish (Danio rerio) embryos (termed morphants) (Figure 3G). For over two decades, the zebrafish has been an invaluable model for the study of hematopoiesis and drug discovery (Jing and Zon, 2011; Zon and Peterson, 2005). Remarkably, akap10 morphants were anemic with decreased hemoglobinization (Figure 3H, red arrowheads) compared to control embryos. We quantified these changes in red cell parameters by performing similar experiments on a transgenic zebrafish line in which all erythroid cells are marked by eGFP expression [Tg(globin-LCR:eGFP)] (Ganis et al., 2012). Flow-cytometry analysis revealed that akap10 morphants had reduced RBC counts (Figure 3I). Together, our data suggest that mitochondrial OM PKA signaling is required for proper heme production and RBC development in vivo.

The terminal heme enzyme, ferrochelatase, is directly phosphorylated by PKA

Next, we asked whether mitochondrial OM PKA signaling directly regulated heme biosynthesis by phosphorylating mitochondrial heme enzymes. Of the mitochondrial enzymes ALAS2, PPOX, CPOX, and FECH, the only enzyme with a predicted high-confidence PKA site (R/K-R/K-X-S/T-Z, where X is any amino acid and Z is an uncharged residue) is FECH at Thr116 (human FECH and Thr115 for murine FECH) (Figure 4A). This residue is evolutionarily conserved and is present on one of the lips of the active site pocket positioned in the middle of a long α-helix (Figure 4B–D). In its unphosphorylated form, the side chain hydroxyl group of Thr116 (colored fuchsia) is sandwiched between His86 (colored blue) and Leu87 on an adjacent α-helix, forming a hydrogen bond with His86 (Figure 4C and D) (Wu et al., 2001). Structural modeling suggests that the bulk of the added phosphate on the side chain of Thr116 would cause movement of the Thr116 α-helix away from the His86 α-helix, and thereby, shift the Thr116-containing α-helix closer inwards towards the active site pocket opening and the porphyrin ring (shown in red) (Figure 4C and D). Such a modification may also destabilize the structure sufficiently to allow for more efficient movement of the active site lip during catalysis.

Figure 4 with 1 supplement see all
Download asset Open asset
FECH is directly phosphorylated by PKA.

(A) The motif surrounding Thr116 in human FECH constitutes a PKA phosphorylation site with a canonical Lys (K) and Arg (R) at positions −3 and −2, respectively, and an uncharged residue at the +1 position. (B) This PKA phosphorylation motif is highly conserved in FECH proteins across vertebrate species. (C) The FECH homodimer is shown with the transparent surface in green and the subunits in solid green ribbon. PPIX is shown as a red space filling model, the [2Fe-2S] clusters as solid rust and yellow balls, and the highlighted Thr116 (site of PKA-mediated phosphorylation) shown as solid violet spheres. The α-helix in which Thr116 resides, highlighted with lemon green, forms one lip of the opening to the active site where porphyrin is bound in this structure (PDB 2QD1). (D) The structure surrounding Thr116 (shown in fuchsia), which is situated in the middle of a long α-helix, is enlarged. Thr116 is in close proximity and hydrogen bonded to His86 (dark blue) and adjacent to Val85 (dark green). The Thr116-containing α-helix is highlighted with lemon green and the protoporphyrin behind in the active site is purple. Structural modeling suggests that phosphorylation of Thr116 would result in a shift of the Thr116-containing α-helix away from the His86-containing α-helix. (E) Purified recombinant His-tagged human FECH was phosphorylated by purified PKAc only in the presence of ATP at a 1:1 ratio. The phosphorylated form of FECH was detected by immunoblotting with an anti-phosphothreonine antibody. (F) An in vitro kinase assay was also performed with wild-type and variant forms of purified His-tagged human FECH. Disruption of the lysine or arginine at the −3 and −2 positions, respectively, similarly abolishes phosphorylation in vitro as a T116A mutation as shown by immunoblotting. (G) Differentiated MEL cells were immunoprecipitated with the indicated antibodies and immunoblotting analysis was performed. Immunoprecipitated FECH can be recognized by two different anti-phosphothreonine antibodies directed against the upstream positive residues and the proline immediately following the threonine. (H) Wild-type (WT) or AKAP10-KO (KO) were lysed and immunoprecipitated with the indicated antibodies. Bound proteins were analyzed by immunoblotting. All immunoblots were performed twice. ATP-adenosine triphosphate; PPIX-protoporphyrin IX; IB-immunoblot; IP-immunoprecipitate.

https://doi.org/10.7554/eLife.24767.009

To test if FECH is directly targeted by PKA for phosphorylation, we performed an in vitro kinase assay by mixing together purified His-tagged human FECH and PKAc, followed by western analysis. This experiment showed that FECH is directly phosphorylated by PKA in an ATP-dependent fashion (Figure 4E). Using [γ-32P]-ATP labeling, we calculated that, in vitro, 9.8 ± 3.2% of purified FECH was phosphorylated after 30 min. Substitution of Thr116 with Ala (T116A) abolished this phosphorylation (Figure 4F). In addition, consistent with the preference of PKAc for positively charged residues at the −3 and −2 positions (Smith et al., 2011), mutation of either Lys113 (K113L) or Arg114 to Leu (R114L) similarly reduced FECH phosphorylation (Figure 4F), strongly indicating that Thr116 of human FECH constitutes a bona fide PKA target. We also examined FECH phosphorylation in erythroid cells by performing similar immunoblot analysis. Immunoprecipitated FECH from differentiated MEL cells was detected by two different phospho-threonine antibodies—one targeting the Lys-X-X-pThr motif and another recognizing the pThr-Pro sequence (Figure 4G). High-dose forskolin treatment also increased phosphorylation of FECH in differentiating MEL cells (Figure 4—figure supplement 1). Conversely, inhibition of OM PKA with loss of AKAP10 resulted in diminished FECH Thr115 phosphorylation (Figure 4H). In toto, our results support a model where PKA becomes localized at the mitochondria OM of maturing erythroid cells and directly phosphorylates FECH.

FECH phosphorylation is required for full activity during erythroid development

Past work has demonstrated that FECH is phosphorylated by protein kinase C (PKC) (Sakaino et al., 2009). PKC-mediated FECH phosphorylation occurs in a domain buried within an inaccessible hydrophobic fold that did not directly impact enzyme catalysis (Sakaino et al., 2009). In contrast, the position of Thr116 (Thr115 in mice) that is modified by PKA suggests that it would have a direct effect on FECH activity (Wu et al., 2001). We first examined this by using an in vitro 55Fe-based assay to measure and compare the amount of radiolabeled deuteroporphyrin-IX (DP) that can be produced by unmodified and modified FECH. DP, as a more soluble analog of the naturally occurring heme precursor protoporphyrin-IX (PPIX), is frequently employed in such measurements and is similarly metalated by FECH to generate deuteroheme (Najahi-Missaoui and Dailey, 2005). Significantly more radioactivity can be detected when purified His-tagged human FECH was added alone to the metalation reaction (Figure 5A) and the addition of purified PKAc to the reaction to catalyze the phosphorylation of FECH resulted in an approximately two-fold increase in 55Fe measurements (Figure 5A). This increase in activity is not attributable to PKAc per se, which has no ferrochelatase activity, and the reaction is completely DP-substrate dependent (Figure 5A). More detailed analysis on enzyme kinetics revealed that phosphorylation had a pronounced effect on maximum velocity (vmax) but did not significantly change the Michaelis-Menten (Km) constant (Figure 5B and Figure 5—figure supplement 1A), suggesting that it has no major influence on substrate binding. An important caveat to these kinetic measurements is that they were performed at 25°C while all other in vitro assays were performed at 37°C and, thus, may not fully reflect enzyme kinetics both in vivo as well as other single time-point experiments. We also examined FECH activity in intact mitochondria isolated from maturing erythroid cells treated with a high dose of PKA-activating forskolin. Mitochondria from differentiating MEL cells exposed to high-dose forskolin catalyzed higher DP metalation compared to mitochondria derived from untreated cells (Figure 5C). In contrast, performing the assay with N-methyl mesoporphyrin-IX (NMMP)—a PPIX analog that is an inhibitor of FECH and not subject to metalation (Dailey and Fleming, 1983)—instead of DP, resulted is very low 55Fe extraction that was refractory to forskolin treatment (Figure 5C). Conversely, FECH activity was reduced in AKAP10-KO cells that had compromised FECH phosphorylation (Figures 4H and 5D). These data indicate that phosphorylation of FECH at Thr116 by OM PKA increases FECH catalytic activity.

Figure 5 with 1 supplement see all
Download asset Open asset
Phosphorylation of FECH is required for its full activity.

(A) Single-point FECH activity was determined in vitro by measuring the amount of 55Fe incorporation into the protoporphyrin IX analog, deuteroporphyrin (DP). FECH has basal activity that is significantly increased by PKA-mediated phosphorylation. This increase in activity was dependent upon both DP and ATP, highlighting the substrate specificity of the assay for DP and the dependence on phosphorylation. (B) Kinetic analyses were subsequently performed with 0.1 µM FECH, 3 µM DP, and 0.2–100 µM 55FeCl3 at 25°C. Phosphorylation of FECH leads to a statistically significant increase in maximum velocity (vmax). There was no significant difference in the Km. Please see Figure 5—source data 1 for vmax and Km values. (C and D) FECH activity was measured in isolated intact mitochondria. Samples from high-dose forskolin (FSK)-treated differentiating MEL cells have higher FECH activity (C) (*p-value<0.05, Mean ± SEM, n = 5). In contrast, AKAP10-KO mitochondria had less FECH activity (D). Very little activity was detected in samples in which DP was substituted with NMMP. (E) A schematic showing the intron 3 and exon 4 sequences of wild-type murine Fech as well as the CRISPR oligo and the single-stranded DNA (ssDNA) that were introduced as a template for DNA repair. Intronic and exonic sequences are shown in lower and upper cases, respectively. Highlighted in yellow are the three PAM (protospacer adjacent motif) sequences closest to the T115A mutation site that facilitates the potential use of multiple CRISPR oligos. The missense mutations necessary to generate the T115A substitution are in orange. Shown in blue are synonymous substitutions designed to either disrupt the PAM sequences to prevent cleavage of the newly introduced mutant allele or to facilitate genotyping using allele-specific primers near the T115A mutation site. (F) Genomic DNA was isolated from the individual clones of MEL cells and used for PCR analysis with allele-specific primers. The parental MEL cells only had the wild-type allele. The intron 3 and exon 4 sequences of these cells were sequenced to confirm these agarose gel electrophoresis results. (G) Undifferentiated and differentiated parental and mutant cells expressing only the FECHT115A allele were lysed and subjected to western analysis to examine the induction of FECH protein during erythroid maturation. FECHThr115Ala protein had very similar up regulation with differentiation. (H) Differentiated MEL cells were lysed, immunoprecipitated with the indicated antibodies, and bound proteins were subjected to western analysis. Cells expressing only mutant FECH were phosphorylation defective at Thr115. (I–K) Mitochondria isolated from differentiated MEL cells expressing only endogenous FECHT115A (Mut), generated by genome editing, has lower FECH activity than wild-type (WT) control (I). These cells expressing non-phosphorylated FECH also have reduced hemoglobinization by o-dianisidine staining (J) and increased accumulation of PPIX substrate (K) as demonstrated by HPLC analysis (*p-value<0.05, Mean ± SEM, n = 11). (L) Wild-type or mutant differentiated MEL cells treated with vehicle (MOCK) or FSK were stained with o-dianisidine. Cells expressing mutant FECH were refractory to the effects of FSK. *p-value<0.05, Mean ± SEM, n = 3, unless otherwise specified. All immunoblots were performed twice. IB-immunoblot; vi-initial velocity; Km-Michaelis-Menten constant; FSK-forskolin; NMMP:N-methyl-mesoporphyrin-IX.

https://doi.org/10.7554/eLife.24767.011
Figure 5—source data 1

Maximum velocity (vmax) and Michaelis-Menten (Km) constants.

The vmax and Km constants for unphosphorylated and in vitro phosphorylated FECH are shown. The data here are graphically presented in Figure 5B and Figure 5—figure supplement 1. *p-value<0.05, **non-significant, Mean ± SEM, n = 3.

https://doi.org/10.7554/eLife.24767.012

EPP patients harboring FECH mutations retain residual FECH activity (Balwani and Desnick, 2012), suggesting that subtle changes in FECH function have important biological implications. Thus, to examine the in vivo implications of FECH phosphorylation under physiological conditions, we used CRISPR/Cas9-directed homology repair to knock-in a T115A substitution into the endogenous Fech gene in murine erythroid cells (Figure 5E). Genotyping and subsequent sequencing confirmed that mutant cells possessed only the FechT115A allele (Figure 5F). Compared to wild-type protein, FECHT115A mutant protein was similarly induced upon erythroid differentiation and was phosphorylation defective (Figure 5G and H). Direct measurements of enzyme activity from intact mitochondria isolated from wild-type and mutant maturing erythroid cells demonstrated that FECHT115A had diminished ferrochelatase activity (Figure 5I). Furthermore, o-dianisidine staining and high-performance liquid chromatography (HPLC) analysis revealed that erythroid cells expressing only FECHT115A protein had reduced hemoglobinization as well as lower intracellular hemin levels (Figure 5J and Figure 5—figure supplement 1B). Conversely, FECHT115A-expressing cells had concomitantly elevated accumulation of the upstream, free protoporphyrin IX (PPIX) precursor (Figure 5K). Clinically, excess erythroid PPIX accumulation is only found in EPP cases where it serves as a diagnostic marker (Balwani and Desnick, 2012; Whatley et al., 2004). The build-up of PPIX in FECHT115A-expressing cells strongly argues that this mutation specifically affects FECH function while the upstream heme biosynthetic pathway remains unaffected.

Erythropoietin signaling activates PKA to phosphorylate FECH

During normal and stress erythropoiesis, erythropoietin (EPO) signaling through its cognate receptor tyrosine kinase (EPOR) regulates survival and proliferation of erythroid progenitors (Kuhrt and Wojchowski, 2015; Testa, 2004). However, there is evidence to suggest that EPO/EPOR signaling regulates other key aspects of erythropoiesis. Unfortunately, the study of such mechanisms has been hampered by the requirement of EPOR signaling in the early stages of the erythropoietic hierarchy (Beale and Chen, 1983; Chida et al., 1999; Socolovsky et al., 1999; Testa, 2004). Nevertheless, uncovering EPOR effectors in later differentiation stages has important clinical relevance. For example, limiting erythroid iron uptake dramatically ameliorates Polycythemia Vera (PV) symptoms in a murine model expressing constitutively active V617F JAK2 (Ishikawa et al., 2015). This is likely due to STAT5-mediated transcriptional up-regulation of the transferrin receptor (Zhu et al., 2008). Despite this, neither PV murine models nor EPO overexpressing transgenic mice show any signs of iron overload (Li et al., 2011; Vogel et al., 2003), indicating that EPO signaling not only promotes iron uptake in RBCs but also coordinates its physiologic assimilation into heme without excessive iron accumulation.

EPOR activates several pathways including the janus kinase 2 (JAK2)/signal transducer and activator of transcription 5 (STAT5), mitogen activated protein kinase (MAPK), and phosphatidyl 3’-inositol kinase (PI3K) pathways (Kuhrt and Wojchowski, 2015), leading us to question whether PKA is also activated by EPO. We tested this by treating primary murine fetal liver erythroblasts with EPO to determine if EPOR activation had any effect on PKA signaling by monitoring phosphorylation of the prototypical PKA target, CREB, on Ser133 (Altarejos and Montminy, 2011; Taylor et al., 2012). As expected, EPOR activation triggered Tyr694 STAT5 phosphorylation, which was similarly observed with EPO treatment of human UT7 erythroid cells (Figure 6A–C). CREBSer133 phosphorylation but not STAT5Tyr694 phosphorylation was blocked by PKI (14-22) (Figure 6D), indicating that EPOR signaling specifically activates PKA. In contrast, co-treatment of UT7 cells with the JAK2 inhibitor, Ruxolitinib, robustly inhibited both STAT5Tyr694 and CREBSer133 phosphorylation (Figure 6E), strongly suggesting that PKA lies downstream of the EPOR/JAK2 pathway. Co-immunoprecipitation experiments further demonstrated that, following EPO exposure, PKAc dissociated from PKAr (Figure 6F), which is obligate for PKAc kinase activity (Taylor et al., 2012).

Figure 6
Download asset Open asset
PKA links EPO signaling with heme production during erythropoiesis.

(A) Primary murine erythroblasts were cultured from the E13.5 fetal liver, starved for 2 hr and treated with EPO (50 U/mL) for the indicated times. Western analysis showed that EPO triggered increased phosphorylation of STAT5Tyr694 and CREBSer133, indicating increased JAK2 and PKA activity, respectively. (B and C) Human UT7 erythroid cells were serum-starved overnight, treated with EPO (2 U/mL) for the indicated times, and subjected to immunoblot analysis. EPO treatment increased both STAT5Tyr694 and CREBSer133 phosphorylations similar to (A). A representative blot is shown in (B) and densitometry quantification from three independent experiments (n = 3) where the phospho-CREB signal is normalized to total CREB signal is shown in (C). Time points within the first 60 min were significantly different than time 0. (D and E) UT7 cells were treated with the indicated compounds, and immunoblot analysis was performed. CREBSer133 phosphorylation can be blocked by both the PKA inhibitor, 14–22, (D) and the JAK2 inhibitor, Ruxolitinib (E). In contrast, 14–22 had no effect on STAT5 phosphorylation (D). (F) UT7 cells untreated or treated with EPO were lysed and immunoprecipitated with the indicated antibodies. Western analysis showed that EPO stimulation resulted in dissociation of PKAc from PKAr. (G) Following PHZ treatment, primary erythroblasts were harvested from the adult murine spleen, starved for 2 hr, and stimulated with EPO (50 U/mL) for 30 min. Lysates were immunoprecipitated and subjected to immunoblot analysis that showed increased FECH phosphorylation with EPO-mediated activation. (H and I) HEL cells induced to hemoglobinize by L-ALA supplementation were treated with Ruxolitinib for 2 hr. Western blot analysis demonstrated reduced CREBSer133 phosphorylation (H) and FECH phosphorylation (I) with inhibition of the constitutively active JAK2 mutant. All immunoblots were performed twice unless otherwise specified. *p-value<0.05, Mean ± SEM, n = 3. IB-immunoblot; IP-immunoprecipitate; EPO-erythropoietin; PHZ-phenylhydrazine;HEL-human erythroleukemia; L-ALA--δ-aminolevulinic acid.

https://doi.org/10.7554/eLife.24767.014

To determine if PKA-mediated FECH phosphorylation was regulated by EPO signaling, we performed anti-phosphothreonine immunoblot analysis on lysates from primary murine splenic erythroblasts (splenocytes) treated with either EPO or vehicle. Indeed, EPO stimulation resulted in an increase in FECH phosphorylation (Figure 6G). Furthermore, Ruxolitinib inhibited phosphorylation of both CREBSer133 and FECHThr116 in human erythroleukemia (HEL) cells that express the constitutively active JAK2V617F mutant protein (Figure 6H and I). Together, our data link EPO signaling to heme metabolism through PKA.

Phosphorylated STAT5 forms a complex with PKAc

The role of PKA in erythropoiesis has remained enigmatic. This is because early studies failed to detect changes in cAMP levels despite its stimulatory effect on iron incorporation during RBC maturation that is likely mediated by the synergistic effects of CREB on STAT5 transcription (Boer et al., 2002, 2003; Gidari et al., 1971; Schooley and Mahlmann, 1975). However, recent evidence indicates that PKAc activation is much more complex involving direct protein-protein interactions with other cell signaling regulators and feedback mechanisms independent of cAMP (Taylor et al., 2012; Wong and Scott, 2004; Yang et al., 2013, 1995; Zakhary et al., 2000). Thus, we next asked whether PKAc interacts with proteins in the EPOR/JAK2/STAT5 signaling pathway, leading to its activation.

Co-immunoprecipitation experiments performed using lysates from HEL cells that have constitutive JAK2V617F signaling revealed that STAT5 formed a complex with PKAc (Figure 7A). In contrast, STAT5 could not be detected when the immunoprecipitation was performed with a control antibody (Figure 7A). The STAT5-PKAc complex formation was sensitive to pharmacologic inhibition of JAK2 function with Ruxolitinib (Figure 7B). We also tested whether EPO can trigger the formation of the STAT5-PKAc complex. STAT5 only co-precipitated with PKAc in lysates harvested from EPO-stimulated human UT7 erythroid cells (Figure 7C). This fraction of STAT5 was phosphorylated on Tyr694 (Figure 7C). In contrast, STAT5 did not co-precipitate with PKAr subunits (Figure 7D), suggesting that phospho-STAT5/PKAc form a distinct complex apart from PKAr. Lastly, the phospho-STAT5/PKAc complex can only be found in cell lysates derived from the cytosol but not the mitochondria (Figure 7E) and indicates that at least a fraction of active PKAc can diffuse freely to phosphorylate nuclear CREB (Hagiwara et al., 1993; Mayr and Montminy, 2001). Based on our results, we propose a model in which PKA signaling components are localized to the outer mitochondrial membrane during erythropoiesis by AKAP10 where it becomes activated by EPO signaling to regulate heme biosynthesis (Figure 7F).

Figure 7
Download asset Open asset
Phosphorylated STAT5 forms a molecular complex with mitochondrial PKAc.

(A) HEL cells were lysed and immunoprecipitated with the indicated antibodies and subjected to immunoblot analysis. (B) HEL cells treated with MOCK or Ruxolitinib for 2 hr were subjected to similar analysis as (A). (C and D) UT7 cells starved overnight and treated with MOCK or EPO for 10 min were lysed and immunoprecipitated with the indicated antibodies. Immunoblot analysis was performed with the indicated antibodies. (E) Cytosolic and mitochondrial extracts were isolated from HEL cells and immunoprecipitated with the indicated antibodies. Bound proteins were resolved on SDS-PAGE and analyzed by immunoblotting with the indicated antibodies. (F) Model of how, during erythropoiesis, EPO signaling activates PKA at the mitochondrial OM that is localized by the GATA1-target, AKAP10. PKA phosphorylates FECH, which is required to achieve full FECH activity necessary to accommodate the vast heme demand for hemoglobin assembly. All immunoblots were performed twice. IB-immunoblot; IP-immunoprecipitate.

https://doi.org/10.7554/eLife.24767.015

Discussion

Heme metabolism genes are downstream of GATA1 during RBC development and regulation of heme production has always been thought to occur at the level of gene transcription (Fujiwara et al., 2009; Handschin et al., 2005; Phillips and Kushner, 2005). However, GATA1 is required for very early stages of erythroid development when the demand for heme and hemoglobin is still low (Fujiwara et al., 1996). This raises the possibility that heme metabolism during cell differentiation requires coordinated metabolic alterations dictated by extracellular signaling cues. EPO signaling is a critical regulator of erythropoiesis and elucidating downstream signaling pathways has been an active area of research. While it is dispensable for early erythroid specification, EPO is critical for the proliferation and survival of early erythroid progenitors (Kuhrt and Wojchowski, 2015) and promotes their differentiation by increasing iron uptake and reducing ‘stemness’ potential (Decker, 2016; Ishikawa et al., 2015; Park et al., 2016; Zhu et al., 2008). However, nothing is known regarding how EPO signaling can influence heme metabolism. Our work supports a unifying model linking the erythroid transcriptional program with a novel PKA-dependent mechanism downstream of EPO that sheds light into how heme metabolism is coupled to development.

There is a growing body of evidence that, in the absence of intrinsic apoptotic signals, mitochondrial PKA signaling within the matrix is compartmentalized due to the impermeability of the inner mitochondrial membrane (Acin-Perez et al., 2009; Lefkimmiatis et al., 2013). Under this scenario, phosphorylation of matrix proteins can only be achieved by activating signals within the matrix (Acin-Perez et al., 2009; DiPilato et al., 2004; Lefkimmiatis et al., 2013). However, our results support an alternative explanation in which proteins are modified prior to transport. The import of nuclear-encoded mitochondrial proteins requires the maintenance of these proteins in an unfolded state (Lodish et al., 2012), which would allow greater accessibility of target motifs. In support, mitochondrial membrane embedded BAX is, first, phosphorylated by OM PKA (Danial et al., 2003; Harada et al., 1999) and the requirement for protein unfolding during mitochondrial import is consistent with the low level of FECH phosphorylation in vitro (approximately 10%) that we observed. It is also very possible that binding of the FECH substrate to the PKA kinase induces conformational changes that would render the target motif more favorable to modification. This mechanism has previously been proposed as a means to prevent ‘promiscuous’ phosphorylation (Dar et al., 2005; Dey et al., 2011). Regardless the mechanism, given its exposure to the cytosol and access to proteins destined for mitochondrial localization, the OM is a prime location for such modifications to occur that would coordinate mitochondrial physiology with overall cellular behavior. Accordingly, studies have shown that OM PKA has unique signaling properties in that it is similarly responsive to cytosolic activation mechanisms but remains active much longer and is largely subject to cAMP-independent regulation (Lefkimmiatis et al., 2013). Our data where loss of AKAP10 and OM PKA signaling results in defective FECH modification and activity is consistent with this idea (Figures 2 and 3).

The regulation of PKA activity involves an intricate signaling network more complex than the canonical cAMP pathway (Lefkimmiatis et al., 2013; Manni et al., 2008; Yang et al., 2013). Our work implicates phosphorylated STAT5 as a novel PKAc binding protein that can displace it from autoinhibitory PKAr subunits (Figures 6 and 7) and is corroborated by recent work showing that phospho-STAT proteins, particularly STAT3 and STAT5, localize to mitochondria (Carbognin et al., 2016; Gough et al., 2009; Meier and Larner, 2014; Wegrzyn et al., 2009). Interestingly, in our analysis, mitochondrial expression of PKAc subunits was not as robustly increased as PRKAR2B (Figure 1C and D). Studies have demonstrated that, upon activation, mitochondrial OM PKAc begins to gradually diffuse throughout the cell (Webb et al., 2008). Thus, the non-stoichiometric increase in PKAc compared to PKAr expression is consistent with a dynamic signaling event.

To date, there have been no reported EPP-associated FECH mutations at the PKA target motif. However, the low prevalence of this disease has made discerning genotype-phenotype correlations difficult. Our findings that PKA is an effector of EPO/JAK2 signaling implicate PKA activity not only in the pathogenesis of EPP but also a spectrum of hematologic diseases (Ishikawa et al., 2015). PRKAR1A inactivating mutations are associated with metabolic syndromes in which anemia is prevalent (Stratakis and Cho-Chung, 2002). This is in agreement with murine models where PRKAR1A deletion has the most widespread effect and is the only knockout of the PKA family with embryonic lethality (Stratakis and Cho-Chung, 2002). Paradoxically, we found that it is the mitochondrial expression of PRKAR2B, and not PRKAR1A, that is most dramatically increased in maturing erythroid cells (Figure 1C–E). The complex nuances of PKA signaling make it very difficult to reconcile these findings. Regulatory subunits restrict both the localization as well as the activation of PKA (Wong and Scott, 2004), making it very challenging to distinguish the relative contributions of these two mechanisms. The dual role of PKAr is highlighted by the incomplete rescue of PRKAR1A-/--associated embryonic defects with PRKACA ablation (Amieux and McKnight, 2002). Further complicating matters is the well-documented instances of compensatory responses (Kirschner et al., 2009), raising the possibility that prominent roles for other PKA isoforms may simply be masked. Indeed, there is both genetic and biochemical evidence supporting a pivotal and specific role for PRKAR2B in blood development and disease. Global transcriptome analysis has shown that PRKAR2B mRNA is selectively high in CD71+ early erythroid cells (Su et al., 2004) and PRKAR2B binds with higher affinity than PRKAR1A to AKAP10 (Burns et al., 2003). The latter point is particularly important given the recent correlation of AKAP10 polymorphisms with human blood traits in genome-wide association studies (Gieger et al., 2011). It is also notable that AKAP10 encodes many isoforms that may localize to different subcellular compartments (Eggers et al., 2009; Huang et al., 1997a). Although our CRISPR targeting strategy was designed to specifically disrupt the N-terminal mitochondrial-targeting motif (Figure 2G), non-mitochondrial AKAP10 isoforms may function in a variety of contexts both in hematopoiesis and in other aspects of development including the cardiovascular system (Kammerer et al., 2003; Tingley et al., 2007). Our work, here, provides further evidence that perturbations in PKA signaling have significant impact on human health (Kammerer et al., 2003) including the pathogenesis of hematologic diseases that, to date, has been unappreciated and warrants further investigation.

Materials and methods

Cell culture

Request a detailed protocol

The DS19 murine erythroleukemia (MEL) subclone (RRID:CVCL_2111) was kindly provided by Arthur Skoultchi (Albert Einstein Medical College, Bronx, NY, USA). Parental human UT7 erythroid cells (RRID:CVCL_5202) were kindly provided by Meredith S. Irwin (Hospital for Sick Children, Toronto, ON, Canada). Human erythroleukemia (HEL) cells (RRID:CVCL_2481) were kindly provided by Ann Mullally (Brigham and Women’s Hospital, Boston, MA, USA). All cells are mycoplasma negative and the International Cell Line Authentication Committee lists none of them as a commonly misidentified cell line. The identities of all cells were confirmed by their labs of origin since none of them are commercially available and have no standard authentication reference sample.

DS19 MEL and primary murine erythroid progenitors from E13.5 fetal liver were cultured and differentiated as previously described (Chung et al., 2015). Differentiating DS19 MEL cells at day 3 of 2% DMSO differentiation were treated with 10 µM (low-dose) or 50 µM (high-dose) forskolin (ThermoFisher, Waltham, MA), or 50 µM 8-Br-cAMP (Sigma-Aldrich, St. Louis, MO) for 30 min or 10 µM dmPGE2 (Cayman Chemicals, Ann Arbor, MI) for 60 min and stained with o-dianisidine as described below. Inhibition with 20 µM H-98 (Tocris Bioscience, Minneapolis, MN), 100 nM PKI (14-22) (ThermoFisher Scientific), or 30 nM bis-indolylmaleimide II (Tocris Bioscience) were performed by pre-treating the cells for 30 min prior to PKA pharmacologic activation.

Bulk murine erythroid progenitors from adult spleen (splenocytes) were prepared and EPO-stimulated as previously described (Maeda et al., 2009; Socolovsky et al., 2001). After two hours of rhEPO (50 U/mL) stimulation, cells were harvested and subjected to immunoprecipitation and western analysis.

Human UT7 erythroid cells were cultured in αMEM (Gibco, Gaithersburg, MD) supplemented with 20% heat-inactivated fetal bovine serum (Serum Source International, Charlotte, NC ) and 10 ng/mL GM-CSF (Peprotech, Rocky Hill, NJ). For stimulation experiments, UT7 cells were starved overnight without GM-CSF. The next morning, rhEPO (2 U/mL) was added for the indicated times prior to harvesting.

Human erythroleukemia cells (HEL) were cultured in RPMI supplemented with 10% heat-inactivated fetal bovine serum and treated with 1 µM Ruxolitinib, which was a kind gift of Dr. Ann Mullally (Brigham and Women’s Hospital) and added to the cells for the indicated times.

shRNA experiments

Request a detailed protocol

Akap10-targeting and NT9 control shRNAs in the pLKO.1-puro vector were purchased from Sigma-Aldrich. MEL cells were electroporated and monoclonal populations expressing these shRNAs were isolated as previously described (Chung et al., 2015). The sequences of the Akap10 shRNAs were: shRNA-1, 5’-CCGGCCAAGTCATGTTGCGATCAATCTCGAGATTGATCGCAACATGACTTGGTTTTTG-3’ and shRNA-2, 5’-CCGGGCAAGAGCACTTTAGTGAGTTCTCGAGAACTCACTAAAGTGCTCTTGCTTTTTG-3’.

Isolation of mitochondria-enriched fractions

Request a detailed protocol

1 × 109 DS19 MEL cells of each condition or 5 × 108 HEL cells were collected, washed once in cold PBS, and resuspended in 1 mL of MSHE buffer (220 mM mannitol, 70 mM sucrose, 5 mM potassium HEPES pH 7.4, 1 mM EGTA pH 7.4, supplemented with Complete EDTA-free protease inhibitor tablets [Roche, Indianapolis, IN]). Samples were then dounce homogenized and pelleted by spinning at 1000 g for 10 min at 4°C. The supernatant was separated and the leftover pellet was resuspended in 200 µL of MSHE buffer. All samples were centrifuged again at 1000 g for 10 min at 4°C. All supernatants from each undifferentiated and differentiating samples were collected and combined and centrifuged again at 1000 g for 10 min at 4°C. This supernatant was then transferred to a new tube and centrifuged at 8000 g for 20 min at 4°C. The resulting supernatant containing cytosolic proteins was transferred to another tube and flash frozen. The pellet was resuspended and washed twice more in 200 µL of MSHE buffer. The final pellet, containing mitochondrially enriched membrane fractions, was flash frozen until subjected to mass spectrometry analysis. For western analyses, the pellet was lysed in NP-40 lysis buffer. For ferrochelatase activity assays, the Mitochondrial Isolation Kit for Mammalian Cells (ThermoScientific) was used according to manufacturer’s instructions. The mitochondrial pellet was resuspended in 200 µL of reaction buffer and immediately used.

Submitochondrial fractionation and proteinase K digestion experiments

Request a detailed protocol

Intact mitochondria were isolated as described above without protease inhibitors and resuspended in MSHE buffer containing 5 mg/mL digitonin that was prepared fresh at 4% in water immediately before each experiment. Samples were incubated on ice for 15 min with vortexing at maximum setting every few minutes for 10 s intervals. After 15 min, samples were centrifuged at 10,000 g for 10 min at 4°C. The supernatant was transferred to a second tube and the leftover pellet was re-extracted twice more with 80 µL of MSHE containing 5 mg/mL digitonin, leaving a final pellet enriched for MPs. The supernatant from the second and third extractions were discarded. The supernatant from the first extraction was then centrifuged at 144,000 g for 1 hr at 4°C. The supernatant containing the IMS was removed and kept in a separate tube. The pellet contained the OM fraction. Proteinase K protection assays were performed as previously described (Shirihai et al., 2000).

Proteomics analysis

Request a detailed protocol

Protein was extracted from purified mitochondria by dissolution in 8 M urea, 50 mM Tris pH 8.0, followed by probe sonication. Extracted protein was reduced and alkylated with dithiothreitol and iodoacetamide, respectively. Alkylated protein samples were first digested with endoproteinase LysC for 4 hr at ambient temperature with an enzyme to protein ratio of 200:1. Each sample was diluted with 50 mM Tris to 1.5 M urea and further digested with trypsin at an enzyme to protein ratio of 50:1, overnight and at ambient temperature. Peptides from each sample were desalted over a C18 solid phase extraction cartridge and dried down. Each sample was resuspended in 0.2 M triethylammonium bicarbonate pH 8.5, and labeled with tandem mass tag reagents (TMT) as previously described (Hebert et al., 2013). Labeled samples were pooled, dried, and fractionated across a strong cation exchange column (Polysulfoethyl A). Each fraction was dried, desalted, and resuspended in 0.2% formic acid.

All nano UPLC separations were performed on a nanoAcquity system. From each fraction, approximately 2 µg of peptides was injected onto a 75 µm inner diameter, 30 cm long, nano column packed with 1.7 BEH C18 particles. The mobile phases were as follows: A) 0.2% formic acid and B) 100% acetonitrile with 0.2% formic acid. Peptide were eluted with a gradient of increasing B from 0%30% over the course of 100 min, followed by a wash with 100% B and re-equilibration at 0% B. Eluting peptides were electrospray ionized and analyzed with an Orbitrap Elite mass spectrometer. The mass spectrometry analysis cycle was as follows. First a survey scan was performed with Orbitrap analysis at 60,000 resolving power at 400 m/z. Peptide precursors in the survey scan were sampled for ms/ms analysis by data dependent top 15 selection with dynamic exclusion turned on. Each peptide precursor selected for sampling was isolated in the ion trap, fragmented by higher energy collisional dissociation (HCD) at 35 NCE, followed by mass analysis of the fragments in the Orbitrap at resolving power 15,000 at 400 m/z.

All data analysis was performed in the COMPASS software suite (Wenger et al., 2011). Spectra were dssearched against a tryptic target-decoy mouse Uniprot database including protein isoforms. Methionine oxidation and TMT on tyrosine were searched as variable modifications. Cysteine carbamidomethylation, TMT on lysine and TMT on the peptide N-terminus were searched as fixed modifications. The tolerance was set to 0.01 Da for matching fragments to the database. Matching spectra were filtered to 1% FDR at the unique peptide level based on spectral matching score (E-value) and peptide precursor ppm mass error, followed by reporter ion quantitation, protein grouping according to parsimony, and filtering to 1% FDR. Reporter ion quantitation normalization was performed essentially as previously described with the following changes (Grimsrud et al., 2012). First, the dataset was annotated using the MitoCarta1.0 database (Pagliarini et al., 2008). All proteins annotated as mitochondrial for undifferentiated and differentiated samples were averaged and linear regression was performed using Microsoft Excel. The conversion factor corresponding to the slope of the linear regression was applied to all proteins in the database regardless of MitoCarta1.0 status and the log2 fold change was calculated.

CRISPR/Cas9-directed homology repair and Akap10 targeting

Request a detailed protocol

CRISPR oligos were cloned into the px335 vector as previously described (Chung et al., 2015). The sequences of the CRISPR oligos were 5’-CACCGAATTTTGGGGGTTCGGCGTT-3’ and 5’-AAACAACGCCGAACCCCCAAAATTC-3’. The single-stranded DNA oligo used to direct homology repair was 5’-TTGGAGTTTCGAAGGTGGAATAAAATCCACTCACTTATGTGTCCAATGATTTAGTAAGCTTGCACCATTCATTGCGAAGCGACGTGCGCCCAAAATTCAAGAGCAGTATCGCAGAATCGGAGGTGGATCCCCCATCAAGATGTGGACT-3’. Targeting of Akap10 (NM_019921.3) was performed as previously described with modifications (Chung et al., 2015). The CRISPR oligos were: exon-1, 5’-CTGCACTAGTCCGAAAACAG-3’ and exon-3, 5’-GCAAGGCATGATTTTTAGTG-3’.

DS19 MEL cells were electroporated and cultured as previously described (Chung et al., 2015) along with 5 µL of 10 mM ssDNA oligo. Single clones were grown in 96-well plates and screened for the presence of the FechT115A allele using genomic DNA PCR followed by antisense oligonucleotide hybridization with [γ-32P]-ATP-labeled (10 mCi/mL, specific activity = 6000 Ci/mmol, Perkin Elmer) wild-type (5’-CATCGCCAAACGCCGAACC-3’) and mutant (5’-CATTGCGAAGCGACGTGCG-3’) oligos as described elsewhere (Hildick-Smith et al., 2013). Clones harboring the mutant allele were expanded and characterized by allele specific PCR (see below) and sequenced to confirm the presence of only the mutant allele.

High performance liquid chromatography

Request a detailed protocol

HPLC analysis was performed as previously described and statistical significance was determined using two-way ANOVA (Yien et al., 2014).

Polymerase chain reaction and site-directed mutagenesis

Request a detailed protocol

Genomic DNA was isolated from DS19 MEL cells according to manufacturer’s instructions (Qiagen DNeasy kit, Germantown, MD). For screening FechT115A knock-in clones using dot blots, the following primers were used to generate the amplicon of interest: 5’-CTGTTTGGCTCTCCTTAG-3’ and 5’-GAGTCCTACTGTAACGAG-3’. For allele-specific PCRs, the latter primer from above was used with either the wild-type, 5’-CATCGCCAAACGCCGAACC-3’, or mutant, 5’-CATTGCGAAGCGACGTGCG-3’, forward primers. For Akap10 genotyping, the primers used were: forward (F) primer, 5’-GAAGGGCTCGCGGACTCG-3’; reverse-1 (R1) primer, 5’-CCCTGACAAAACCCTTGC-3’; and reverse-2 (R2) primer, 5’-CACTTGCAGTGTTTTGGGGTTT-3’.

Site-directed mutagenesis was performed using the Agilent QuikChange Lightning Multi Site-Directed Mutagensis Kit (La Jolla, CA) according to the manufacturer’s instructions. The following mutagenesis primers were used: T116A, 5’-CATCGCCAAACGCCGAGCCCCCAAGATTCAAG-3’ and 5’-CTTGAATCTTGGGGGCTCGGCGTTTGGCGATG-3’; K113L, 5’-GCACCATTCATCGCCTTACGCCGAACCCCCAAG-3’ and 5’-CTTGGGGGTTCGGCGTAAGGCGATGAATGGTGC-3’; and R114L, 5’-CCATTCATCGCCAAACTCCGAACCCCCAAGATTC-3’ and 5’-GAATCTTGGGGGTTCGGAGTTTGGCGATGAATGG-3’.

Protein purification

Request a detailed protocol

Recombinant His-tagged human FECH proteins were expressed and purified as previously described (Burden et al., 1999).

Structural modeling

Request a detailed protocol

The 2.0 Å structure of human FECH (PDB 2QD4) was visualized by PyMOL (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC).

In vitro kinase assay

Request a detailed protocol

One microgram of His-tagged wild-type or variant human FECH was mixed with purified PKAc according to manufacturer’s instructions (Promega, Madison, WI). 20% of the kinase reaction was subjected to western blot analysis. The stoichiometry of the kinase reaction was kept at 1:1. For kinase assays using [γ-32P]-ATP (specific activity 6000 Ci/mmol, Perkin Elmer, Boston, MA), the reaction was performed with an incubation time of 30 min instead of 10 min and scaled down to where 0.2 pmol of purified FECH and PKAc were used. Four-fold excess [γ-32P]-ATP was added. After 30 min, each reaction was immunoprecipitated with anti-FECH antibodies (see below). The incubation time for [γ-32P]-ATP labelling was three-times longer than for other assays to ensure maximal phosphorylation in vitro. The amount of radioactivity was quantified in a scintillation counter (Chung et al., 2015) and normalized to immunoprecipitation efficiency, which was calculated by determining the percent of FECH protein that was recovered in the immunoprecipitation relative to input using western blotting followed by densitometry with ImageJ (Schneider et al., 2012). The average and standard error was calculated from three independent experiments.

Western blotting and immunoprecipitation

Request a detailed protocol

All immunoblots were performed according to manufacturer’s instructions and as previously described except that all phosphothreonine immunoblots were performed with an HRP-conjugated protein A secondary antibody (Chung et al., 2015). Anti-PDHA1 (ab67592) and anti-AKAP10 (ab97354) rabbit polyclonal antibodies were purchased from Abcam (Cambridge, MA). Mouse monoclonal anti-TUBA1A (DM1A), rabbit polyclonal anti-STAT5 (C-17), anti-PRKACB (C-20), anti-PRKAR2A (M-20), and goat polyclonal anti-FECH (C-20) and anti-HSPD1 (K-19) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal anti-pSer133-CREB (1B6) and anti-pThr-Pro (9391S), rabbit monoclonal anti-pTyr694-STAT5 (D46E7), anti-PRKACA (D38C6), anti-SMAC (D5S4R), anti-VDAC1 (D73D12), anti-TOM20 (D8T4N), and anti-CREB (D76D11), rabbit polyclonal anti-Arg-X-X-pThr (9621S) and anti-PRKARIA (D54D9) antibodies were purchased from Cell Signaling Technology (Danvers, MA). Anti-GAPDH (MAB374) mouse monoclonal and anti-PRKARIIB (ABS14) rabbit polyclonal antibodies were purchased from Millipore (Billerica, MA). Anti-TIM23 mouse monoclonal antibody was purchased from BD Biosciences (Woburn, MA). Immunoprecipitations were performed as previously described (Chung et al., 2010). All immunoblots were performed two independent times except when otherwise specified. Densitometry was performed as previously described and analyzed by one-way ANOVA (Chung et al., 2015).

Ferrochelatase activity assay

Request a detailed protocol

Ferrochelatase activity assays using purified His-tagged wild-type human FECH was performed by using 250 ng of purified protein in assay reaction buffer (0.6 M sorbitol, 40 mM HEPES pH 7.4 pH with KOH, 50 mM KCl, 1 mM MgSO4). Ascorbic acid and NADH were then added to final concentrations of 0.4 mg/mL and 2 nM, respectively. The solution was incubated at 37°C for 5 min. After the incubation period, 0.2 mL of 55FeCl3 (38 mCi/mL, specific activity = 54.5 mCi/mol, Perkin Elmer) was added to each sample along with either 5.5 µL of 200 µM DP or NMMP. The samples were then incubated for another 10 min at 37°C and 55Fe-radiolabeled heme was extracted as previously described and counted in a liquid scintillation counter (Chung et al., 2015). For activity assays using intact mitochondria, the protocol was modified to use 50 µg of freshly isolated mitochondria. All experiments were performed in duplicate and three independent experiments were performed followed by statistical analysis using student’s t-test.

For kinetic assays, the conditions of the experiments are described elsewhere using 0.1 µM FECH, 3 µM DP, and 0.2–100 µM 55FeCl3 except the reactions were carried out at 25°C (Hunter et al., 2008). At 30, 60, 120, 180, 300, 600, and 900 s, 10% of each reaction were removed and immediately mixed with cold FeCl3 at a final concentration of 1.25 mM to stop the reaction. All samples were extracted and counted in a scintillation counter as described above. The amount of product over time was plotted and analyzed using regression analysis with GraphPad Prism v5.0 software. Michaelis-Menten analyses with the initial velocities (vi) were subsequently performed using GraphPad Prism v5.0 that calculated the maximum velocity (vmax) and Michaelis-Menten (Km) constant. Experiments were performed three times and statistical analysis was performed on GraphPad Prism v5.0.

Zebrafish experiments

Request a detailed protocol

Injections, o-dianisidine staining, and flow cytometry analysis were performed as described elsewhere (Chung et al., 2015). MOs were purchased from Gene Tools, LLC (Philomath, OR). Zebrafish embryos at the one-cell stage were injected with MOs targeting the exon-1/intron-1 (MO1) and exon-2/intron-2 (MO2) junctions of D. rerio akap10 (XM_690206). The MO sequences were as follows: MO1, 5’-TGGAGCGGCCACTTCCTTACCTTTC-3’; MO3, 5’-TTTAGCACTAGACACTTACCTTTGC-3’.

Vertebrate animal study approvals and ethics statement

Request a detailed protocol

All zebrafish (RRID:ZIRC_ZL1) and mouse experiments were performed in full compliance with the approved Institutional Animal Care and Use Committee (IACUC) protocols at Boston Children’s Hospital (Protocol #15-07-2974R) and Brigham and Women’s Hospital (Protocol #2016N000117). These studies were approved by local regulatory committees in accordance with the highest ethical standards for biomedical research involving vertebrate animals.

References

    1. Chida D
    2. Miura O
    3. Yoshimura A
    4. Miyajima A
    (1999)
    Role of cytokine signaling molecules in erythroid differentiation of mouse fetal liver hematopoietic cells: functional analysis of signaling molecules by retrovirus-mediated expression
    Blood 93:1567–1578.
    1. Dailey HA
    2. Fleming JE
    (1983)
    Bovine ferrochelatase. kinetic analysis of inhibition by N-methylprotoporphyrin, manganese, and heme
    The Journal of Biological Chemistry 258:11453–11459.
    1. Gieger C
    2. Radhakrishnan A
    3. Cvejic A
    4. Tang W
    5. Porcu E
    6. Pistis G
    7. Serbanovic-Canic J
    8. Elling U
    9. Goodall AH
    10. Labrune Y
    11. Lopez LM
    12. Mägi R
    13. Meacham S
    14. Okada Y
    15. Pirastu N
    16. Sorice R
    17. Teumer A
    18. Voss K
    19. Zhang W
    20. Ramirez-Solis R
    21. Bis JC
    22. Ellinghaus D
    23. Gögele M
    24. Hottenga JJ
    25. Langenberg C
    26. Kovacs P
    27. O'Reilly PF
    28. Shin SY
    29. Esko T
    30. Hartiala J
    31. Kanoni S
    32. Murgia F
    33. Parsa A
    34. Stephens J
    35. van der Harst P
    36. Ellen van der Schoot C
    37. Allayee H
    38. Attwood A
    39. Balkau B
    40. Bastardot F
    41. Basu S
    42. Baumeister SE
    43. Biino G
    44. Bomba L
    45. Bonnefond A
    46. Cambien F
    47. Chambers JC
    48. Cucca F
    49. D'Adamo P
    50. Davies G
    51. de Boer RA
    52. de Geus EJ
    53. Döring A
    54. Elliott P
    55. Erdmann J
    56. Evans DM
    57. Falchi M
    58. Feng W
    59. Folsom AR
    60. Frazer IH
    61. Gibson QD
    62. Glazer NL
    63. Hammond C
    64. Hartikainen AL
    65. Heckbert SR
    66. Hengstenberg C
    67. Hersch M
    68. Illig T
    69. Loos RJ
    70. Jolley J
    71. Khaw KT
    72. Kühnel B
    73. Kyrtsonis MC
    74. Lagou V
    75. Lloyd-Jones H
    76. Lumley T
    77. Mangino M
    78. Maschio A
    79. Mateo Leach I
    80. McKnight B
    81. Memari Y
    82. Mitchell BD
    83. Montgomery GW
    84. Nakamura Y
    85. Nauck M
    86. Navis G
    87. Nöthlings U
    88. Nolte IM
    89. Porteous DJ
    90. Pouta A
    91. Pramstaller PP
    92. Pullat J
    93. Ring SM
    94. Rotter JI
    95. Ruggiero D
    96. Ruokonen A
    97. Sala C
    98. Samani NJ
    99. Sambrook J
    100. Schlessinger D
    101. Schreiber S
    102. Schunkert H
    103. Scott J
    104. Smith NL
    105. Snieder H
    106. Starr JM
    107. Stumvoll M
    108. Takahashi A
    109. Tang WH
    110. Taylor K
    111. Tenesa A
    112. Lay Thein S
    113. Tönjes A
    114. Uda M
    115. Ulivi S
    116. van Veldhuisen DJ
    117. Visscher PM
    118. Völker U
    119. Wichmann HE
    120. Wiggins KL
    121. Willemsen G
    122. Yang TP
    123. Hua Zhao J
    124. Zitting P
    125. Bradley JR
    126. Dedoussis GV
    127. Gasparini P
    128. Hazen SL
    129. Metspalu A
    130. Pirastu M
    131. Shuldiner AR
    132. Joost van Pelt L
    133. Zwaginga JJ
    134. Boomsma DI
    135. Deary IJ
    136. Franke A
    137. Froguel P
    138. Ganesh SK
    139. Jarvelin MR
    140. Martin NG
    141. Meisinger C
    142. Psaty BM
    143. Spector TD
    144. Wareham NJ
    145. Akkerman JW
    146. Ciullo M
    147. Deloukas P
    148. Greinacher A
    149. Jupe S
    150. Kamatani N
    151. Khadake J
    152. Kooner JS
    153. Penninger J
    154. Prokopenko I
    155. Stemple D
    156. Toniolo D
    157. Wernisch L
    158. Sanna S
    159. Hicks AA
    160. Rendon A
    161. Ferreira MA
    162. Ouwehand WH
    163. Soranzo N
    (2011) New gene functions in Megakaryopoiesis and platelet formation
    Nature 480:201–208.
    https://doi.org/10.1038/nature10659
  53. Book
    1. Lodish HF
    2. Berk A
    3. Kaiser CA
    4. Krieger M
    5. Bretscher A
    6. Ploegh H
    7. Amon A
    8. Scott MP
    (2012)
    Molecular Cell Biology (7th Edn)
    New York: W H Freeman and Company.

Decision letter

  1. David Ginsburg
    Reviewing Editor; Howard Hughes Medical Institute, University of Michigan, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Erythropoietin signaling regulates heme biosynthesis." for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom, David Ginsburg (Reviewer #1), is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Ivan Dikic as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Susan Taylor (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the synthesis of all three reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

All three reviewers agreed that this is an intriguing and well written manuscript that links regulation of heme biosynthesis to Epo activation of JAK2 and PKA signaling in the mitochondria. Specifically the authors show first using an unbiased MS analysis of mitochondrial proteins that PKA isoforms are up-regulated in the mitochondria of differentiated MEL cells. They then go on to show that ferrochetalase (FECH) has a PKA phosphorylation site and that this site is selectively phosphorylated in response to high levels of forskolin and cAMP but down regulated in the presence of PKA inhibitors. They then go on to link the increased PKA signaling to EPO stimulation of JAK2. Because the mechanism for regulation of heme biosynthesis at the metabolic level in contrast to the gene transcriptional regulation is poorly understood, this is a potentially significant advance. However, the biochemical and mechanistic evidence, including the overall elucidation of PKA signaling, is rather superficial at this point, and would require considerable additional data to justify the sweeping conclusions that are made. Specific comments/suggestions are appended at the end of this letter.

Specific comments/suggestions from the reviewers:

1) Forskolin vs. cAMP. The dose of forskolin is very high and low doses show no effect although this figure ((Figure 1—figure supplement 1) is looking at the total cell extract vs. purified mitochondria. If the authors wish to promote the idea that there are localized pools of cAMP, then the requirement for high doses of forskolin, a general cyclase activator, are counterintuitive. Isoproterenol is a much more specific activator of β adrenergic signaling and should also be tested to see if there is enhanced specificity. The difference between high vs. low levels of forskolin is disconcerting and not explained or well justified. Also it is not at all established or well accepted that high doses vs. low doses of forskolin will activate mitochondrial PKA. Mitochondrial soluble cyclase is predicted to be activated by bicarbonate, not by forskolin.

2) In the Results section the authors state "We independently confirmed these results using immunoblotting analysis where we also found increased total PKA expression in maturing erythroid cells (Figure 1D and Figure 1—figure supplement 1A). These results suggest that PKA signaling may regulate heme production". Elevation in the levels of the PKA subunits does not necessarily correlate with changes in enzyme activity. The authors may wish to modify this statement to reflect this issue.

3) It is stated in the Results that “Thr116 (humanFECH) (Figure 2A). This residue is evolutionarily conserved and is present on one of the lips of the active site pocket positioned in the middle of a long α-helix". It is unusual to have a consensus phosphorylation site in the middle of a helix as this region of the secondary structure has to unravel before the phospho-transfer reaction can proceed. The authors should comment on this point.

4) Phosphorylation of FECH on the mitochondria matrix. The mitochondrial MS proteomic data shows a global enrichment of PKA signaling proteins. Specifically, the data shows an increase in RIa, RIIa, RIIb, Ca and Cb although only two of these proteins, Ca and RIIb, are actually shown in the protein blots shown in Figure 1D and the increase in the C band is actually very minimal. Although the authors state that increases in RIa and RIIa protein are also shown, in Figure 1D the levels of RIa and RIIa are not shown. The up-regulation of all of these isoforms is surprising given that there is likely to be a very specific isoform associated with PKA signaling in the matrix of the mitochondria. It is thus surprising that three of the four R-subunit isoforms are up-regulated as well as the two major C-subunit isoforms. It is extremely unlikely that all of these isoforms are in the matrix of the mitochondria as the data assumes. Perhaps several isoforms are on the outer membranes where PKA targeting is well-documented and even in the inner membrane space. But evidence supporting PKA subunits in the matrix is sparse and not independently validated in different labs. If the authors truly want to demonstrate PKA phosphorylation of FECH in the matrix of the mitochondria, they need to more rigorously validate this. Fractionation of the mitochondria could elucidate the distribution of the isoforms and show that either C or one R isoform are in the matrix. Isolated mitochondria can also be treated with cAMP to show that FECH is phosphorylated.

5) Isoform specificity. To demonstrate specificity in PKA signaling the authors need to rigorously show that one isoform, and not all three, is responsible for the phosphorylation of FECH in the mitochondria. shRNA experiments should be done to demonstrate isoform specificity. Targeting experiments can also be done. Will disruption of targeting abolish the activation of FECH? Several peptides that disrupt targeting are available. If so, can the authors discriminate between RI and RII targeting? Tools are available to carry out these experiments.

6) Mutations in PKA that lead to hematopoetic dysfunction. What are the specific mutations in PKA signaling that lead to hematopoietic dysfunction and can these also recapitulate the effects of FECH phosphorylation? These mutations will certainly also support the isoform specificity of PKA signaling.

7) Dissociation of R- and C-subunits. While high levels of cAMP can promote dissociation of the PKA subunits in cells, growing evidence suggests that unleashing of the catalytic activity of the C-subunit does not require dissociation of the R and C-subunits. These experiments with whole cells shed little light on the mechanism that is invoked here for the phosphorylation of FECH in the mitochondria. The authors show only the general dissociation of RIIb under conditions that are far from physiological. They do not address specific PKA signaling at the mitochondria, much less in the matrix of the mitochondria. PKA activity reporters are also available. Is PKA initially activated in the cytoplasm as a result of EPO binding to JAK2?

8) Total cell differences vs. mitochondrial differences. The authors need to rigorously distinguish between total cell effects vs mitochondrial effects.

9) In vitro phosphorylation of FECH. These studies carried out with recombinant expressed His-tagged FECH are also somewhat superficial. The modeling with space filling models is inadequate. The authors should at least show the individual residues and their interactions, and each residue should be clearly labeled. The phosphorylation in Figures 2D and 2E does not look robust. How long was the incubation time? Is the phosphorylation stoichiometric? Is the Km for substrate enhanced? Since recombinant proteins are available, more rigorous kinetic assays should be done, and also the authors should look to see if dimerization is influenced by phosphorylation. Is the His-tagged recombinant enzyme purified from E. coli? If so, there should be enough protein to do these experiments rigorously.

10) Phosphoproteomics. There should be a phosphoproteomic analysis of the mitochondrial fractionations and of the global proteins to see to what extent PKA substrates in general are upregulated.

11) PKC phosphorylation. If PKC is known to be an activator of FECH, then the authors should use this as a control to make sure that PKC cannot phosphorylate Thr116, which actually looks like a good PKC site, and also to demonstrate the robustness of activation and phosphorylation by PKC. Is the effect comparable for PKA?

12) Steps from JAK2 activation to FECH phosphorylation. There are many steps going from JAK2 activation at the plasma membrane to FECH phosphorylation in the matrix of the mitochondria, and many of these steps could be regulated by cAMP including not only phosphorylation of key metabolic enzymes such as FECH but also proteins involved in translocation from the plasma membrane to the matrix of the mitochondria. How does the signal from JAK2 activation by EPO get transmitted to the mitochondria? This mechanism is completely unclear. CREB phosphorylation at S133 is also very weak and not very convincing in the gels shown in Figure 4.

13) FECH disease mutations. What are the FECH mutations that cause EPP and do they correlate with the PKA regulation that is being proposed? Have any EPP mutations been identified at or near the FECH Thr116 phosphorylation site?

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for resubmitting your work entitled "Erythropoietin signaling regulates heme biosynthesis" for further consideration at eLife. Your revised article has been evaluated by Ivan Dikic (Senior editor), David Ginsberg (Reviewing editor and Reviewer #1), and two additional reviewers (Susan Taylor and John Scott).

All three reviewers agreed that the manuscript has been significantly improved. However, several key issues remain, that must be addressed satisfactorily, before acceptance for publication in eLife could be considered. The independent comments of all three reviewers are appended below. Though multiple comments/suggestions are included, the key issues that must be satisfactorily addressed in a revised submission are points 2, 4, 5 and 8 from reviewer #2 and comments #4 and 5 from reviewer #3.

Reviewer #1:

This revised manuscript is significantly improved and considerable new data have been added to address the initial critiques.

Reviewer #2:

Chung, et al. "Erythropoietin signaling regulates heme biosynthesis"

The authors have introduced a significant amount of new information to this revised manuscript and as a result both its significance and its impact have been greatly improved. While I would now recommend publication there are still some questions that the authors could address. Some are beyond the scope of this work but others should be answered.

1) The discovery that AKAP10 is the AKAP the mechanism by which PKA is targeted to the mitochondria is compelling and very interesting. In addition to the recent Geiger paper, we found early on that there was an advantageous mutation in AKAP10 when one compared older populations vs. younger populations (Kammerer et al., 2003). There were two mutations but the one in the AKAP binding site specifically reduced the affinity for RIa. We speculated at the time that this had a cardioprotective effect, but this could also be relevant for your findings.

2) There are some concerns about the in vitro phosphorylation in Figure 4, in particular 4E and 4F. This phosphorylation does not appear to be very robust, which is surprising. What is the stoichiometry? You should be able to do this easily with mass spec or radioactive ATP or using a gel shift assay that should distinguish the unphosphorylated protein from the phosphorylated protein.

3) In spite of the modest phosphorylated bands, the mutants look good and it is good that you did both the Thr116Ala mutant as well as the K mutants. However, did you try the phosphomimetic mutants (D/E)? This is an important experiment and should give an opposite phenotype to the Ala mutant?

4) The stoichiometry is also important for the kinetic assays. The Vmax effect seems to be rather modest and if the protein is not stoichiometrically phosphorylated then you would see a more modest effect. I suspect that the effect on activity would be even more pronounced if the protein were fully phosphorylated.

5) One last question that relates to the kinase assays that you used. What was the amount of PKA that you used relative to the amount of FECH? If it was sub-stoichiometric then you may have a situation where there is a single turnover but the substrate stays bound until some other signal releases it. If this were the case you would need to add near stoichiometric amounts of kinase and substrate to see full phosphorylation. Just a thought.

6) I do not think it is a concern that the motif is in a helix as this is a surface helix and it is likely that it is quite dynamic. I would guess that the helix propensity of this helix is not great especially because it has two prolines in it. This is unusual for a helix where Pro is typically a helix breaker. There was an earlier example of this with a structure of PRK tethered to its substrate eIG2alpha (Dar, Dever and Sicheri, 2005). In this case the helix in the substrate became disordered near the active site when the protein was tethered to the kinase. Sicheri suggests in a later paper a putative mechanism whereby the helix protects against phosphorylation until the substrate binds to the kinase.

7) Although it would not be necessary for this paper, I think that the helix could actually provide a good mechanism for docking onto the C-subunit using the groove that the PKI peptide uses. A peptide array might indicate that there were recognition motifs beyond the immediate site of phosphorylation. I would guess that the FI motif could be important for docking onto this surface. You could test this with an FI/AA mutant.

8) The pull down assays showing the C-subunit binding to Tyr phosphorylated STAT5 is intriguing but still preliminary. The mechanism is unclear. Is the RIIb subunit no longer associated with this complex? ie Does the binding of STAT5 induce the dissociation of the holoenzyme or does it just cause a conformational change that opens up the active site making is accessible to FECH? Does the entire complex pull down so that it is poised for binding to FECH? Are additional PKA substrates phosphorylated other than FECH or are you looking at the events in a specific signaling complex? Is the C-subunit now released into the cytoplasm and no longer tethered to the mitochondria or is it still tethered to the mitochondria? Do you trap a complex with STAT5 and FECH if you use the Ala mutant of FECH? As you do not fully dissociate the complex even in the presence of cAMP, which is consistent with the very early Johnson results, which you reference, and also with earlier mass spec data and Scott's recent papers. The high forskolin experiments would force the holoenzyme apart but this is not physiological.

9) One last point. AKAP10 has a PDZ motif at its C-terminus that immediately follow the A Kinase Binding motif. In kidney this interacts with PDZK1 but I do not know if other binding partners have been discovered. It could just be something to look for in your various data sets.

In conclusion, this is a very nice story and the biological and physiological relevance is obviously very high. If some of the above questions can be satisfactorily addressed, publication is recommended.

Reviewer #3:

This is an extensively revised version of a manuscript that was previously submitted to eLife on how erythroid transcriptional programming regulates post-translational mechanisms that influence heme metabolism. The quality of the western blot characterization of the phenomena in Figures 2 and 4 remains a weakness of the study. The most substantive change in the manuscript is the inclusion of some new and potentially interesting data on the role of AKAP10 as a mitochondrial adapter for the location of PKA subunits. I have several issues pertaining to this new data.

1) AKAP10 has two locations that could be pertinent to this work. It is found at the plasma membrane where it is believed to have some RGS function and on or in the mitochondria. The membrane-associated fraction of AKAP10 may play an important role and needs to be considered by the authors. A few comments in the discussion would suffice.

2) In subsection “Mitochondrial PKA is localized to the outer mitochondrial membrane via AKAP10” it states that AKAP10 has long been recognized as a downstream target for the GATA1 erythroid lineage transcription factor. This is not immediately apparent in the Fujiwara et al., 2009 paper (text or supplementary figures). Please clarify.

3) On the basis of other studies it is quite possible that deletion of AKAP10 results in the up-regulation of other mitochondrial anchoring proteins. This is an easily performed set of experiments. Also, the authors should incorporate analysis of WAVE1 they cite the paper that identified it at the mitochondria but never follow up on this.

4) The experiments in Figure 3A are still marred by the lack of analyses using physiological agonists of the cAMP signaling pathway. Contrary to what is stated in paragraph two of subsection “Mitochondrial outer membrane PKA signaling regulates hemoglobinization and erythropoiesis”. Forskolin is not a PKA agonist. Forskolin is a supraphysiolgical activator of adenylyl cyclases. Consequently, other cAMP responsive factors such as Epac and CNG channels will be mobilized. This issue was raised by the reviewers before and has still not be satisfactorily addressed.

5) The data linking PKA and Stat5 seems rather weak. The authors need to deftly describe how biochemical effects at the mitochondria with PKA relate to changes in nuclear CREB phosphorylation. Especially since Cam Kinases, Mist1 & Mist 2 and Akt are just as potent modulators of Ser 133 on CREB.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Erythropoietin signaling regulates heme biosynthesis" for further consideration at eLife. Your revised article has been favorably evaluated by Ivan Dikic (Senior editor), a Reviewing editor, and two reviewers.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

1) All reviewers were concerned about the quality of Figure 4, which should be replaced. Also, discussion should be included as to why the level of phosphorylation is so low, as well as an explanation of what efforts were made to increase the level of phosphoryaltion.

2) Given that the sample is only 10% phosphorylated, interpretation of the kinetic data in Figure 5 should be specifically addressed.

3) If possible, the manuscript would be strengthened by including a native gel for the in vitro phosphorylated protein, looking for a gel shift to confirm the quantification reported. Also, would it be possible to separate the phospho protein by ion exchange chromatography?

https://doi.org/10.7554/eLife.24767.017

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

All three reviewers agreed that this is an intriguing and well written manuscript that links regulation of heme biosynthesis to Epo activation of JAK2 and PKA signaling in the mitochondria. Specifically the authors show first using an unbiased MS analysis of mitochondrial proteins that PKA isoforms are up-regulated in the mitochondria of differentiated MEL cells. They then go on to show that ferrochetalase (FECH) has a PKA phosphorylation site and that this site is selectively phosphorylated in response to high levels of forskolin and cAMP but down regulated in the presence of PKA inhibitors. They then go on to link the increased PKA signaling to EPO stimulation of JAK2. Because the mechanism for regulation of heme biosynthesis at the metabolic level in contrast to the gene transcriptional regulation is poorly understood, this is a potentially significant advance. However, the biochemical and mechanistic evidence, including the overall elucidation of PKA signaling, is rather superficial at this point, and would require considerable additional data to justify the sweeping conclusions that are made. Specific comments/suggestions are appended at the end of this letter.

On behalf of all the authors, I want to thank you for your efforts in evaluating our work. We are very encouraged by the positive tone of the reviews and the in-depth analysis provided by everyone is reflective of the broad interest and provocative nature our work. Your insightful feedback has propelled us to explore new concepts. Based upon our new data, we have revised our original conclusions and developed a new unifying model where transcriptional and post-translational mechanisms work in concert to regulate heme metabolism. These new additional results have significantly strengthened our revised manuscript, which is now composed of 7 main figures instead of 4 in the original submission. In particular, our new results include:

1) in vivo evidence using an animal model that this EPO-PKA signaling cascade is important in a physiologic setting.

2) Biochemical and genetic evidence that FECH phosphorylation occurs on the outer mitochondrial membrane and not in the mitochondrial matrix.

3) A mechanism of how PKA become localized, which links the erythroid transcriptional program with EPO signaling and also explains our observed PKA- isoform-specific effects.

4) Extensive quantitative analysis including kinetic measurements of enzymatic activity of phosphorylated FECH.

5) Direct evidence of how erythropoietin signaling regulates PKA activity.

Our manuscript uncovers a novel signal transduction pathway that couples cell differentiation with heme metabolism – two fundamental cellular processes that were, previously, thought to occur independently. All reviewers agreed that our work is of significant interest and addresses this fundamental biological problem that, to date, has been unresolved. We strongly believe that our revised manuscript would strongly appeal to the broad readership of eLife, and respectfully, ask that the Editorial Office and the reviewers give our revised work consideration.

Specific comments/suggestions from the reviewers:

1) Forskolin vs. cAMP. The dose of forskolin is very high and low doses show no effect although this figure ((Figure 1—figure supplement 1) is looking at the total cell extract vs. purified mitochondria. If the authors wish to promote the idea that there are localized pools of cAMP, then the requirement for high doses of forskolin, a general cyclase activator, are counterintuitive. Isoproterenol is a much more specific activator of β adrenergic signaling and should also be tested to see if there is enhanced specificity. The difference between high vs. low levels of forskolin is disconcerting and not explained or well justified. Also it is not at all established or well accepted that high doses vs. low doses of forskolin will activate mitochondrial PKA. Mitochondrial soluble cyclase is predicted to be activated by bicarbonate, not by forskolin.

Indeed, the contribution of distinct pools of cAMP and their corresponding activating signals are two significant controversies in the field. They are interrelated issues because the roles of various intracellular PKA pools have typically been extrapolated from pharmacologic studies. We thank the reviewer for bringing these key points to our attention. Over the past several months, we have worked hard to experimentally address these issues and we would like the opportunity to discuss how they address the reviewer’s queries.

A significant proportion of our new results involve the identification and characterization of AKAP10 as a mitochondrial scaffold that nucleates PKA at the mitochondrial outer membrane. Much of what we know of AKAP10 is from studies carried out by Dr. Susan Taylor (reviewer #3) over the past 20 years (Huang et al., 1997; Wang et al., 2001; and Eggers, Schafer, Goldenring and Taylor, 2009). Recently, AKAP10 has garnered a great deal of attention since its association with human cardiac function and hematologic parameters in genome-wide association studies (GWAS) (Gieger et al., 2011)

Our new work demonstrates that the majority of mitochondrial PKA in maturing red cells is actually tethered onto the outer mitochondrial membrane, which is dependent upon AKAP10. In the absence of AKAP10, we found a significant depletion of PKA in this compartment and reduced FECH phosphorylation and function (Figures 2, 4H, and 5). Inhibition of AKAP10 in vivo also results in anemia (Figure 3). This suggests that modification of FECH occurs prior to its import into the mitochondrial matrix where it has enzymatic function.

As it specifically relates to the Reviewer’s comments, we would like to begin by clarifying that our conclusion is not that cAMP is the sole activator of PKA during erythropoiesis. In fact, as we note later (see point #12 below and Figure 7 in our revised manuscript), we believe that PKA activation can be achieved in several ways. Thus, while we use pharmacologic manipulation of PKA activity many of which are cAMP agonists, we only do so in an effort to examine the relevance of the PKA pathway in principle.

When it comes to understanding the role of any compartment of PKA, we also strongly believe that our combined biochemical and genetic data offer new insight compared to previous work by other groups that rely upon extrapolation from pharmacologic studies for two reasons. First, dose-dependent and cell-type-specific effects have been documented (Humphries, Pennypacker and Taylor, 2007 and Lefkimmiatis, Leronni and Hofer, 2013), suggesting that discrepancies between independent studies may reflect technical differences. Second, the methods in which suborganellular PKA activity is monitored has been debated, making it even more difficult to reconcile differences that are observed between studies (DiPilato, Cheng and Zhang, 2004 and Acin-Perez, et al., 2009). Thus, not only do our new data uncover a novel physiologic pathway, it also sheds light into this controversy from a different scientific vantage point.

The reviewer also astutely noted that we use total cell extract instead of purified mitochondria to perform our assays, implying that the latter approach would be superior. In this case, we, respectfully, disagree. The use of purified mitochondria to study signaling pathways overlooks the dynamic nature of signal transduction. As previous studies have demonstrated, proteins (including PKA substrates, such as BAX) can be modified prior to mitochondrial transport (Harada et al., 1999 and Danial et al., 2003). Furthermore, mitochondrial outer membrane PKA is a compartment with distinct biochemical properties that can be activated by cytosolic PKA agonists such as forskolin while having vastly different kinetics (Lefkimmiatis, Leronni and Hofer, 2013). Our characterization of the role of AKAP10 is actually consistent with previous work in which case, treatment of purified mitochondria with PKA pharmacologic agents would only lead to ambiguous negative results.

In addition to including in vivo genetic and biochemical data on the role of AKAP10 into our revised manuscript, we have also included a much more thorough discussion of these topics. We are very excited that our work has elicited such a fantastic discussion, which underscores our belief that eLife is an excellent venue to publish our revised manuscript as it deserves a broad audience. We hope that the reviewer is satisfied with our responses.

2) In the Results section the authors state "We independently confirmed these results using immunoblotting analysis where we also found increased total PKA expression in maturing erythroid cells (Figure 1D and Figure 1—figure supplement 1A). These results suggest that PKA signaling may regulate heme production". Elevation in the levels of the PKA subunits does not necessarily correlate with changes in enzyme activity. The authors may wish to modify this statement to reflect this issue.

We apologize to the reviewer for our misleading statement. We have now revised this section to limit our conclusions derived from this figure only to PKA expression. This sentence now reads, “We independently confirmed these results using immunoblotting with isoform-specific antibodies where we also found increased total expression of these PKA subunits in maturing erythroid cells (Figure 1D and E).”

3) It is stated in the Results that “Thr116 (humanFECH) (Figure 2A). This residue is evolutionarily conserved and is present on one of the lips of the active site pocket positioned in the middle of a long α-helix". It is unusual to have a consensus phosphorylation site in the middle of a helix as this region of the secondary structure has to unravel before the phospho-transfer reaction can proceed. The authors should comment on this point.

This is an excellent point raised by the reviewer. It is true that access to the consensus motif is crucial for kinase phosphorylation. Our new results demonstrating the involvement of AKAP10 at the outer mitochondrial membrane corroborates this paradigm since the import of proteins across this membrane requires that proteins remain in an unfolded state (Lodish et al., 2012). This would increase accessibility of this site to PKA. Notably, this mechanism of regulation by PKA has been observed previously (Harada et al., 1999 and Danial et al., 2003). Outer membrane PKA phosphorylates BAX prior to its integration within the mitochondrial membrane (Harada et al., 1999 and Danial et al., 2003). These points are of significant interest to basic cell signaling and we have now included a discussion of this in our revised manuscript.

4) Phosphorylation of FECH on the mitochondria matrix. The mitochondrial MS proteomic data shows a global enrichment of PKA signaling proteins. Specifically, the data shows an increase in RIa, RIIa, RIIb, Ca and Cb although only two of these proteins, Ca and RIIb, are actually shown in the protein blots shown in Figure 1D and the increase in the C band is actually very minimal. Although the authors state that increases in RIa and RIIa protein are also shown, in Figure 1D the levels of RIa and RIIa are not shown. The up-regulation of all of these isoforms is surprising given that there is likely to be a very specific isoform associated with PKA signaling in the matrix of the mitochondria. It is thus surprising that three of the four R-subunit isoforms are up-regulated as well as the two major C-subunit isoforms. It is extremely unlikely that all of these isoforms are in the matrix of the mitochondria as the data assumes. Perhaps several isoforms are on the outer membranes where PKA targeting is well-documented and even in the inner membrane space. But evidence supporting PKA subunits in the matrix is sparse and not independently validated in different labs. If the authors truly want to demonstrate PKA phosphorylation of FECH in the matrix of the mitochondria, they need to more rigorously validate this. Fractionation of the mitochondria could elucidate the distribution of the isoforms and show that either C or one R isoform are in the matrix. Isolated mitochondria can also be treated with cAMP to show that FECH is phosphorylated.

In our revised manuscript, we have addressed these concerns. As the reviewer has noted, PKA refers to an enzyme complex consisting of regulatory and catalytic subunits. In humans, there are three catalytic subunits (a, b, and g) and four regulatory subunits (RIa, RIIa, RIb, and RIIb) (Kirschner, Yin, Jones and Mahoney, 2009). Mice do not have PKA-catalytic-g (Kirschner, Yin, Jones and Mahoney, 2009) and, thus, was not detected in our analysis using a murine model. In addition, the expression of RIa is restricted to neurons (Kirschner, Yin, Jones and Mahoney, 2009) and is similarly not detected in our analysis. We apologize to the reviewer for omitting these critical details, which we have now included in our revised manuscript. We have also performed a more thorough expression analysis of all PKA isoforms and included it in Figures 1D and E including submitochondrial localization (Figures 2A and B).

The points brought-up by the reviewer made us realize that we were premature in our conclusions. We performed a significant amount of new experiments that now demonstrate that FECH is phosphorylated by PKA at the outer mitochondrial membrane. To avoid being repetitive, we kindly ask the reviewer refer to our responses to points #1 and #3 regarding our new data and their implications on the PKA signaling.

5) Isoform specificity. To demonstrate specificity in PKA signaling the authors need to rigorously show that one isoform, and not all three, is responsible for the phosphorylation of FECH in the mitochondria. shRNA experiments should be done to demonstrate isoform specificity. Targeting experiments can also be done. Will disruption of targeting abolish the activation of FECH? Several peptides that disrupt targeting are available. If so, can the authors discriminate between RI and RII targeting? Tools are available to carry out these experiments.

We thank the reviewer for bringing up this point. Isoform specificity is very difficult to address in PKA signaling because of functional redundancy and compensatory responses both of which have been well documented (Kirschner, Yin, Jones and Mahoney, 2009 and Amieux and McKnight 1997). In fact, in mouse knockouts of individual PKA regulatory subunits compensation of the other isoforms is frequently found (Cummings DE, et al., Nature, 1996; Kirschner, Yin, Jones and Mahoney, 2009 and Amieux and McKnight, 1997). In addition, loss-of-function shRNA experiments do not adequately discriminate between total and mitochondrial PKA effects as noted by the reviewer in point #8 below. Thus, we, respectfully, do not believe that the approach recommended by the reviewer will yield satisfactory and unambiguous scientific answers.

We agree that aberrant targeting of PKA away from mitochondria is critical to deciphering a mitochondrial role for PKA. However, instead of using peptides, we have uncovered an adaptor protein – AKAP10 – that is a transcriptional target of GATA-1, the master regulatory erythroid transcription factor (Huang, et al., 1997 and Wang et al., 2001). Strikingly, inhibition of AKAP10 in erythroid cells as well as a zebrafish model resulted in reduced heme production and anemia (Figure 3). Moreover, as the reviewer suggested, we also examined how mislocalization of PKA away from mitochondria via AKAP10 inhibition affected FECH phosphorylation and activity. We found that loss of AKAP10 led to significantly diminished FECH phosphorylation and FECH activity (Figures 4H and 5D).

In our revised manuscript, we have included a much more thorough discussion on the interplay of the various PKA isoforms as well as included our analysis of the role of AKAP10 as a dual-specificity AKAP in heme biosynthesis. We sincerely hope that we have satisfactorily addressed the reviewer’s concerns.

6) Mutations in PKA that lead to hematopoetic dysfunction. What are the specific mutations in PKA signaling that lead to hematopoietic dysfunction and can these also recapitulate the effects of FECH phosphorylation? These mutations will certainly also support the isoform specificity of PKA signaling.

We thank the reviewer for asking these two critical questions. As the reviewer has noted, point #6 is very much related to point #5 above and, to avoid being repetitive, we kindly ask the reviewer to refer to point #5 for our response to isoform specificity.

The most prevalent PKA mutations associated with hematologic symptoms are deletions of PRKAR1A (also known as PKA-regulatory 1 α subunit). This is not to say that other isoforms do not have a role. We are aware that different PKA isoforms have different biochemical properties (for example, Cheng X, et al., J Biol Chem, 2001). However, as we noted in our response to point #5, compensatory responses has made it very difficult to address isoform specificity using classical genetic loss-of-function approaches in vivo. In addition, as noted in reviewer point #8 below, global inhibition of PKA isoforms (with RNAi or CRISPR), would be highly ineffective at delineating cytosolic and mitochondrial effects.

In contrast, our characterization of AKAP10 addresses the issue previously raised by the reviewer in an elegant and biologically relevant fashion. AKAP10-dependent recruitment of PKA requires the regulatory subunits (Wong W and Scott JD, Nat Rev Mol Cell Biol, 2004). Thus, loss of AKAP10 essentially mimics loss of PKAr specifically from mitochondria while leaving all other PKA signaling components intact. Strikingly, we found that AKAP10 deletion in erythroid cells resulted in diminished FECH phosphorylation and activity. Interestingly, recent GWAS studies have found that AKAP10 variants are associated with hematologic blood traits (Grieger C, et al., Nature, 2011). We have included all these details in the expanded Discussion section of our revised manuscript.

7) Dissociation of R- and C-subunits. While high levels of cAMP can promote dissociation of the PKA subunits in cells, growing evidence suggests that unleashing of the catalytic activity of the C-subunit does not require dissociation of the R and C-subunits. These experiments with whole cells shed little light on the mechanism that is invoked here for the phosphorylation of FECH in the mitochondria. The authors show only the general dissociation of RIIb under conditions that are far from physiological. They do not address specific PKA signaling at the mitochondria, much less in the matrix of the mitochondria. PKA activity reporters are also available. Is PKA initially activated in the cytoplasm as a result of EPO binding to JAK2?

We thank for the reviewer for raising these concerns, which has propelled us to explore new possibilities to explain our observations. Our new experiments demonstrate that activated PKA responsible for phosphorylating FECH is not in the mitochondrial matrix (please see points #1, #5, and #6).

With regards to how EPO signaling activates PKA, our new results argue that PKA is activated directly through phosphorylated STAT5 in a cAMP-independent fashion. As we have discussed, early studies in red blood cells failed to detect changes in cAMP levels throughout red cell maturation (Boer, Drayer and Vellenga, 2003; Gidari, Zanjani and Gordon, 1971; and Schooley and Mahlmann, 1975). This is not to say that PKA has no role in erythropoiesis but rather it may be cAMP-independent. In support, exogenous cAMP promotes iron uptake in red cells by phospho-CREB dependent synergism with STAT5 (Boer, Drayer and Vellenga, 2002). Thus, there is clearly cross talk between these two pathways that does not involve endogenous cAMP.

In our new Figure 7, we show using human erythroid cells, that phosphorylated STAT5 forms a complex with PKA-catalytic subunits. This supports a model whereby EPO signaling leads to phosphorylation of STAT5 that can displace and activate PKA- catalytic subunits. Our new model also explains why CREB – the proto-typical non- mitochondrial PKA target – is phosphorylated by EPO signaling. This is because the fraction of PKA that is activated in response to EPO is not matrix. Rather, it is cytosolic, which also activates mitochondrial outer membrane PKA as previously shown by other groups (Lefkimmiatis, Leronni and Hofer, 2013).

We appreciate the enthusiasm demonstrated by the reviewer in our work. However, we believe that unraveling the kinetics behind PKA activation and whether dissociation between the regulatory and catalytic subunits is absolutely required in red cells is beyond the scope of the current manuscript. The goal of our work here is not to test every nuance of PKA signaling that has been previously published; follow-up studies are more appropriate to address all these issues. Instead, the key conceptual advance is that PKA is activated by EPO and heme metabolism is regulated by this mechanism. Lastly, we would like to emphasize that, akin to EPO signaling, iron uptake induced by cAMP does not result in iron overload (Boer, Drayer and Vellenga, 2003; Gidari, Zanjani and Gordon, 1971; and Schooley and Mahlmann, 1975). This is a key point because it suggests that PKA activation can not only trigger iron import but also coordinate iron assimilation into heme. Our new data, model, and in-depth discussion are now included in our revised work.

8) Total cell differences vs. mitochondrial differences. The authors need to rigorously distinguish between total cell effects vs mitochondrial effects.

Please see our responses to points #5 and #6.

9) In vitro phosphorylation of FECH. These studies carried out with recombinant expressed His-tagged FECH are also somewhat superficial. The modeling with space filling models is inadequate. The authors should at least show the individual residues and their interactions, and each residue should be clearly labeled. The phosphorylation in Figures 2D and 2E does not look robust. How long was the incubation time? Is the phosphorylation stoichiometric? Is the Km for substrate enhanced? Since recombinant proteins are available, more rigorous kinetic assays should be done, and also the authors should look to see if dimerization is influenced by phosphorylation. Is the His-tagged recombinant enzyme purified from E. coli? If so, there should be enough protein to do these experiments rigorously.

We thank the reviewer for noting these technical issues and we apologize for not providing sufficient detail. We have now performed extensive in vitro analysis to examine enzyme kinetics (Figures 5B and Figure 5 —figure supplement 1A). In addition, we have repeated many of our immunoblots. Many of our early experiments were performed using a secondary antibody that resulted in high background signals. Recently, we switched to a protein-A sepharose HRP-conjugate that is much more effective at reducing background signals. We have now included these new experiments in Figures 4E and G.

10) Phosphoproteomics. There should be a phosphoproteomic analysis of the mitochondrial fractionations and of the global proteins to see to what extent PKA substrates in general are upregulated.

We absolutely agree with the reviewer that further confirmation of PKA activation would greatly strengthen our work. This is why we are excited to report that we have addressed this in two ways. Our response to this reviewer concern consists of unpublished and ongoing work and we would kindly request that our response be held in strict confidence and not be released for public knowledge.

First, using phospho-antibody-specific flow cytometry analysis, a colleague (Dr. Vijay Sankaran, Children’s Hospital Boston, Harvard Medical School) has independently found that PKA substrates are hyper-phosphorylated in response to EPO treatment of CD34+ human hematopoietic progenitor cells. This was presented at the recent 58th American Society of Hematology Annual Meeting in San Diego, CA (Kim AR, et al., 58th ASH Annual Meeting, San Diego, CA and https://ash.confex.com/ash/2016/webprogram/Paper93991.html). They are in the process of publishing their work and, thus, while we are excited by an independent validation of our model, we are also eager to publish our own work to capitalize on its novelty.

Second, given that Dr. Sankaran has already validated hyper-phosphorylation of PKA substrates, we focused our efforts on understanding the functional relevance of CREB phosphorylation. We addressed this by using bioinformatics to determine if a “CREB- gene signature” can be found within the erythroid transcriptional program. We isolated previously identified genes that were direct targets of CREB (Impey, et al., Cell, 2004) and superimposed it onto gene expression databases that have been well annotated according to stages of erythroid differentiation (Zhang, Socolovsky, Gross and Lodish, 2003). In Author response image 1 is a box-and-whisker (Tukey method, GraphPad Prism) plot of our data showing changes in the expression of genes as a function of the stages of R1 to R5 stages of erythroid development. White bars represent changes in the expression of all 16,384 genes that were detected by Zhang, Scolovsky, Gross and Lodish and the red bars represent CREB target genes annotated by Impey, et al.

Author response image 1
Download asset Open asset
https://doi.org/10.7554/eLife.24767.016

At every stage of erythroid development, we found that this “CREB gene signature” was indeed up-regulated in red blood cells (p < 0.05). We are in the process of performing careful analysis on the relevance of the PKA/CREB transcriptional program in erythropoiesis the results of which are beyond the scope of the current manuscript. However, taken together, the biochemical analysis in our revised manuscript along with the analysis performed by Dr. Sankaran in human CD34+ cells and the preliminary gene expression bioinformatics analysis, we believe that we have made a strong and convincing argument that PKA is important in red blood cell development.

11) PKC phosphorylation. If PKC is known to be an activator of FECH, then the authors should use this as a control to make sure that PKC cannot phosphorylate Thr116, which actually looks like a good PKC site, and also to demonstrate the robustness of activation and phosphorylation by PKC. Is the effect comparable for PKA?

We thank the reviewer for suggesting using PKC as a specificity control. We are excited to say that we have now performed the suggested experiments (Figure 3 —figure supplement 1A). PKC inhibition using bis-indolylmaleimide II (Mahata M, et al., Mol Pharmacol, 2002) was unable to block PKA-dependent heme production suggesting that they function in two independent and parallel pathways. In addition, we have performed additional experiments using pharmacologic PKA agents on erythroid cells harboring the Thr115Ala FECH mutation (Figure 5L). When we treat these mutant erythroid cells with forskolin, we failed to detect any changes in hemoglobinization suggesting that the Thr115 site is specific for PKA activity.

12) Steps from JAK2 activation to FECH phosphorylation. There are many steps going from JAK2 activation at the plasma membrane to FECH phosphorylation in the matrix of the mitochondria, and many of these steps could be regulated by cAMP including not only phosphorylation of key metabolic enzymes such as FECH but also proteins involved in translocation from the plasma membrane to the matrix of the mitochondria. How does the signal from JAK2 activation by EPO get transmitted to the mitochondria? This mechanism is completely unclear. CREB phosphorylation at S133 is also very weak and not very convincing in the gels shown in Figure 4.

We thank the reviewer for their interest in this mechanism. We agree that it is a major question and have addressed it in our revised manuscript. To avoid being repetitive, we kindly request that the reviewer visit our response to point #7 in which we go into great detail explaining our new data.

With regards to our blots in Figure 4, we would like to first assure the reviewer that all of our immunoblots were performed at least twice. We apologize for the poor quality of some of our experiments and have addressed this concern in three independent ways. First, we have explicitly stated the replicate number for all immunoblots in the Figure Legends and the Materials and methods section. Second, for the changes in CREB phosphorylation, we have performed a third independent experiment and quantified our results using densitometry with statistical analysis (it is now new Figure 6C). Finally, we have repeated many of these experiments using reagents that further minimize background signals (please see our response to point #9). Our Materials and methods section now reflect these technical changes.

13) FECH disease mutations. What are the FECH mutations that cause EPP and do they correlate with the PKA regulation that is being proposed? Have any EPP mutations been identified at or near the FECH Thr116 phosphorylation site?

We thank the reviewer for asking these astute questions and have now included an appropriate discussion in our manuscript. In short, mutation of the phosphorylation site has yet been documented in EPP. However, this does not indicate that this pathway is not important in this disease for a number of reasons.

First, EPP is a rare disorder and because of its low prevalence, one of the major hurdles in understanding the genetics of rare diseases is finding a comprehensive genotype- phenotype correlation. For example, it took 11 years from the initial reports of GATA1 (Tsai SF, et al., Nature, 1989 and Evans T and Felsenfeld G, Cell, 1989) to the first identification of humans with GATA1 mutations (Nichols KE, et al., Nat Genet, 2000). To date, the only FECH mutation that is consistently found in EPP is a hypomorphic allele that is inherited in combination with other mutations (Gouya L, et al., Nat Genet, 2002 and Whatley SD, et al., Br J Dermatol, 2010). The search for these other genetic lesions is an active area of study. In fact, the characterization of an X-linked, dominant EPP-like disease that is associated with C-terminal mutations in ALAS2 highlights the emerging molecular and genetic heterogeneity of EPP (Whatley SD, et al., Am J Hum Genet, 2008 and Balwani M, et al., Mol Med, 2013).

Second, complete loss of FECH function has never been found in EPP and one very likely explanation is that severe FECH deficiencies are non-viable (Whatley SD, et al., Br J Dermatol, 2010). This would further complicate any attempts at finding genotype- phenotype correlations since FECH mutations must only occur in certain combinations in order for EPP to develop.

We are actively engaged in efforts to collect more clinical data from EPP patients to see if the FECH Thr116 site or the surrounding motif is mutated in EPP. In addition, we are also examining whether regulators of the EPO/PKA pathway can account for the 4% of EPP cases in which the FECH gene is not mutated (Whatley SD, et al., Br J Dermatol, 2010). We have included an appropriate discussion covering this topic in our revised work and hope that the reviewer finds it satisfactory.

[Editors' note: the author responses to the re-review follow.]

Reviewer #2:

Chung, et al. "Erythropoietin signaling regulates heme biosynthesis"

The authors have introduced a significant amount of new information to this revised manuscript and as a result both its significance and its impact have been greatly improved. While I would now recommend publication there are still some questions that the authors could address. Some are beyond the scope of this work but others should be answered.

1) The discovery that AKAP10 is the AKAP the mechanism by which PKA is targeted to the mitochondria is compelling and very interesting. In addition to the recent Geiger paper, we found early on that there was an advantageous mutation in AKAP10 when one compared older populations vs. younger populations (Kammerer et al,. 2003). There were two mutations but the one in the AKAP binding site specifically reduced the affinity for RIa. We speculated at the time that this had a cardioprotective effect, but this could also be relevant for your findings.

We thank the reviewer for bringing this study to our attention and apologize for having missed it throughout the course of preparing our manuscript. It would be interesting to see if AKAP10 mutations in older populations correlate with hematologic parameters. We are also actively searching GWAS databases to determine if there is a link between AKAP10 mutations and porphyrias (EPP in particular). We have included the Kammerer et al., 2003 reference into our discussion.

2) There are some concerns about the in vitro phosphorylation in Figure 4, in particular 4E and 4F. This phosphorylation does not appear to be very robust, which is surprising. What is the stoichiometry? You should be able to do this easily with mass spec or radioactive ATP or using a gel shift assay that should distinguish the unphosphorylated protein from the phosphorylated protein.

We thank the reviewer for suggesting this experiment. Indeed, we have used [γ-32P]-ATP labeling and determined that in vitro, the amount of phosphorylated FECH is 9.75 ± 3.18%. This result has now been included in the main text with an accompanying description of the protocol in the Materials and methods section.

3) In spite of the modest phosphorylated bands, the mutants look good and it is good that you did both the Thr116Ala mutant as well as the K mutants. However, did you try the phosphomimetic mutants (D/E)? This is an important experiment and should give an opposite phenotype to the Ala mutant?

This is an excellent suggestion. We have actually tried this experiment with the phosphomimetic mutant using in vitropurified proteins. Unfortunately, we were unable to detect any significant change in enzyme activity. This has lead us to hypothesize that the phospho-group somehow interacts with the porphyrin/iron substrates in a manner that we still do not fully understand and will require a great deal of additional structural biology experiments to decipher, which are beyond the time line for the revision.

4) The stoichiometry is also important for the kinetic assays. The Vmax effect seems to be rather modest and if the protein is not stoichiometrically phosphorylated then you would see a more modest effect. I suspect that the effect on activity would be even more pronounced if the protein were fully phosphorylated.

We agree that stoichiometry is an important consideration when interpreting the results of our kinetic assays. As we discussed in our responses to points 2 and 3, not all FECH protein is phosphorylated in vitroand, as a complementary approach, we have tried to examine the activity of His-tagged phospho-mimetic mutant FECH proteins without much success. Future work is needed to better understand the structural changes that occur with FECH upon modification that would provide important clues as to the relationship between stoichiometry and effect on catalysis.

5) One last question that relates to the kinase assays that you used. What was the amount of PKA that you used relative to the amount of FECH? If it was sub-stoichiometric then you may have a situation where there is a single turnover but the substrate stays bound until some other signal releases it. If this were the case you would need to add near stoichiometric amounts of kinase and substrate to see full phosphorylation. Just a thought.

For our in vitroassays, the ratio of PKA to FECH is approximately 1:1. We had never considered that PKA may remain bound to FECH following the modification and thank the reviewer for bringing this to our attention. We have also included a small note regarding the stoichiometry of the in vitrokinase reaction in our Materials and methods section as well as the corresponding Figure Legends.

6) I do not think it is a concern that the motif is in a helix as this is a surface helix and it is likely that it is quite dynamic. I would guess that the helix propensity of this helix is not great especially because it has two prolines in it. This is unusual for a helix where Pro is typically a helix breaker. There was an earlier example of this with a structure of PRK tethered to its substrate eIG2alpha (Dar, Dever and Sicheri, 2005). In this case the helix in the substrate became disordered near the active site when the protein was tethered to the kinase. Sicheri suggests in a later paper a putative mechanism whereby the helix protects against phosphorylation until the substrate binds to the kinase.

We thank the reviewer for the insightful feedback and providing us with their expertise. Indeed, we have considered that this phosphorylation event may be highly dynamic, which would explain its location within a helix. Furthermore, we have also wondered how stable this helical structure would remain throughout enzyme catalysis (ie. “helix propensity” as referred to by the reviewer). We have included these details and accompanying references (Dar, Dever and Sicheri, 2005 and Dey et al., 2011) into our discussion to further strengthen our manuscript.

7) Although it would not be necessary for this paper, I think that the helix could actually provide a good mechanism for docking onto the C-subunit using the groove that the PKI peptide uses. A peptide array might indicate that there were recognition motifs beyond the immediate site of phosphorylation. I would guess that the FI motif could be important for docking onto this surface. You could test this with an FI/AA mutant.

We thank the reviewer for their interest in our work. The role of PKA signaling in hematopoiesis is an emerging concept (our work and also North et al., 2007; Kim, PG, et al., J Exp Med, 2015; and Jing L, et al., J Exp Med, 2015), and we are actively exploring avenues to pursue for follow-up studies.

8) The pull down assays showing the C-subunit binding to Tyr phosphorylated STAT5 is intriguing but still preliminary. The mechanism is unclear. Is the RIIb subunit no longer associated with this complex? ie Does the binding of STAT5 induce the dissociation of the holoenzyme or does it just cause a conformational change that opens up the active site making is accessible to FECH? Does the entire complex pull down so that it is poised for binding to FECH? Are additional PKA substrates phosphorylated other than FECH or are you looking at the events in a specific signaling complex? Is the C-subunit now released into the cytoplasm and no longer tethered to the mitochondria or is it still tethered to the mitochondria? Do you trap a complex with STAT5 and FECH if you use the Ala mutant of FECH? As you do not fully dissociate the complex even in the presence of cAMP, which is consistent with the very early Johnson results, which you reference, and also with earlier mass spec data and Scott's recent papers. The high forskolin experiments would force the holoenzyme apart but this is not physiological.

We thank the reviewer for their enthusiasm and are excited to announce that we have performed additional experiments in an effort to further address the reviewer’s questions. Some of these concerns are also related to concerns raised by reviewer 3 point 5.

First, we performed additional co-immunoprecipitation experiments using antibodies directed against PKAr subunits. The goal was to examine whether phosphorylated STAT5 formed a macromolecular complex with PKAc and PKAr. In contrast to Figures 7A to 7C, STAT5 did not co-precipitate with PKAr (new Figure 7D). As it relates to the reviewer’s queries, our result suggests that the phospho-STAT5/PKAc complex is distinct from that of PKAr. It would also suggest that phospho-STAT5 may cause the PKAc/PKAr complex to dissociate. We have considered ways to further explore this latter point. However, it is technically very challenging. We do not believe that using traditional STAT5 truncation mutants would be informative since disruption of its interaction with the EPOR/JAK2 complex would similarly block activation of PKA (please see Figure 6E and I showing that JAK2 activity is required for PKA activation). It is also possible that the phospho-STAT5/PKAc complex is indirect further limiting our options regarding in vitrobinding assays with bacterially purified proteins.

The second experiment we performed is to use mitochondrial lysates for coimmunoprecipitations. This would help address the issue of whether the phospho-STAT5/PKAc complex is tethered to the mitochondria. Our results show that we could not detect a stable complex between phospho-STAT5 and PKAc using isolated mitochondria (new Figure 7E). This result is consistent with a model where AKAP10 recruits PKA to the outer mitochondrial membrane (OMM) via PKAr. Upon EPO activation, STAT5 becomes phosphorylated and phospho-STAT5 dislodges PKAc from PKAr at the OMM. This model would also explain how CREB is also phosphorylated in response to EPO since the activation of PKA is not strictly restricted to mitochondria.

Whether dissociation of the PKAr/PKAc complex is absolutely required for kinase activity is a difficult issue to address in our system. We were never able to completely disrupt the PKAr/PKAc complex with EPO stimulation (see Figure 6F). Thus, it is possible that the kinase is active even as a macromolecular complex. We hope that our latest experiments have adequately addressed the reviewer’s comments.

9) One last point. AKAP10 has a PDZ motif at its C-terminus that immediately follow the A Kinase Binding motif. In kidney this interacts with PDZK1 but I do not know if other binding partners have been discovered. It could just be something to look for in your various data sets.

We thank the reviewer for their enthusiasm for our proteomics screen. We are in the process of validating additional hits that would be of biological significance.

Reviewer #3:

This is an extensively revised version of a manuscript that was previously submitted to eLife on how erythroid transcriptional programming regulates post-translational mechanisms that influence heme metabolism. The quality of the western blot characterization of the phenomena in Figures 2 and 4 remains a weakness of the study. The most substantive change in the manuscript is the inclusion of some new and potentially interesting data on the role of AKAP10 as a mitochondrial adapter for the location of PKA subunits. I have several issues pertaining to this new data.

1) AKAP10 has two locations that could be pertinent to this work. It is found at the plasma membrane where it is believed to have some RGS function and on or in the mitochondria. The membrane-associated fraction of AKAP10 may play an important role and needs to be considered by the authors. A few comments in the discussion would suffice.

We thank the reviewer for this comment that raises an important issue regarding isoform specificity. When AKAP10 was first identified by Huang, LJ, et al., PNAS, 1997, they found multiple isoforms with different molecular weights. These isoforms may possess different N-terminal signal peptides which would direct AKAP10 to various subcellular compartments. This is a main reason why we designed our CRISPR deletion strategy to disrupt the N-terminal mitochondrial-targeting motif – to specifically address the mitochondrial role of AKAP10 in the event that multiple isoforms were expressed. Certainly, we were unable to discriminate between various isoforms in our zebrafish knockdown experiments. We have also considered the possibility that human AKAP10 mutations may have non-mitochondrial effects that would account for the observed cardiovascular phenotype (please see point 1 by reviewer #2). To address these reviewer concerns, we have, now, included a discussion in the main text of our revised manuscript as well as the corresponding Figure Legend to explicitly state that our CRISPR deletion strategy eliminates the N-terminal motif.

2) In subsection “Mitochondrial PKA is localized to the outer mitochondrial membrane via AKAP10” it states that AKAP10 has long been recognized as a downstream target for the GATA1 erythroid lineage transcription factor. This is not immediately apparent in the Fujiwara et al. 2009 Mol Cell paper (text or supplementary figures). Please clarify.

We apologize for the confusion. AKAP10 mRNA was found to be upregulated by Fujiwara, T, et al., Mol Cell, 2009 in Supplementary Table 3. There were two probe sets for AKAP10 – one found a significant induction while the other did not. Based on this, and the lack of an obvious GATA-1 binding site in the immediately upstream AKAP10 promoter, very little follow-up work was performed. We would also like to note that expression profiling of maturing erythroid cells independently performed by Zhang, Socolovsky, Gross and Lodish, 2003, also found a similar induction of AKAP10 transcript as murine red blood cells differentiated (please see raw RNA-seq data in Supplementary files in Zhang, Socolovsky, gross and LodishJ, 2003). We have now included this second reference to further support our results. In addition, we have also re-phrased that sentence to emphasize that these previous reports were high-throughput studies that require detailed analysis of raw data sets.

3) On the basis of other studies it is quite possible that deletion of AKAP10 results in the up-regulation of other mitochondrial anchoring proteins. This is an easily performed set of experiments. Also, the authors should incorporate analysis of WAVE1 they cite the paper that identified it at the mitochondria but never follow up on this.

We thank the reviewer for bringing this to our attention since it has important implications on our future work. Over the past two years, we have tried exhaustively to use CRISPR genome editing to delete akap10 in zebrafish. To date, we have failed to get germline transmission of a null allele despite evidence of a deletion event in the parental generation. As the reviewer noted, we hypothesize that there are compensatory responses (such as upregulation of other AKAPs), in murine models not found in zebrafish. While our efforts, so far, have been met with disappointment, we believe that it may actually pave the way for the use of zebrafish to study AKAP mechanisms since adaptive responses may be different than currently used cell culture and murine model systems. Our analysis is only preliminary and beyond the scope of the current work.

4) The experiments in Figure 3A are still marred by the lack of analyses using physiological agonists of the cAMP signaling pathway. Contrary to what is stated in paragraph two of subsection “Mitochondrial outer membrane PKA signaling regulates hemoglobinization and erythropoiesis”. Forskolin is not a PKA agonist. Forskolin is a supraphysiolgical activator of adenylyl cyclases. Consequently, other cAMP responsive factors such as Epac and CNG channels will be mobilized. This issue was raised by the reviewers before and has still not be satisfactorily addressed.

We apologize for our oversight. We have now performed additional sets of experiments to address this issue (Figures 3B and C). It has been previously shown that prostaglandin E2 (PGE2) is a physiologic PKA agonist that promotes several aspects of hemopoiesis (North et al., 2007 and Goessling et al., 2009). We treated our erythroid cells with dimethyl-PGE2 (dmPGE2), which is a more stable analog of PGE2 and found that it also promoted hemoglobinization similar to other PKA activating agents that we have tested (Figure 3B). The effects of dmPGE2 were inhibited by PKI (14-22) (Figure 3C), providing more physiologic evidence and specificity that PKA modulation is a determinant of heme production during red cell maturation.

5) The data linking PKA and Stat5 seems rather weak. The authors need to deftly describe how biochemical effects at the mitochondria with PKA relate to changes in nuclear CREB phosphorylation. Especially since Cam Kinases, Mist1 & Mist 2 and Akt are just as potent modulators of Ser 133 on CREB.

This is a similar concern raised by reviewer 2 point 8. In our work, we use Ser133 CREB phosphorylation as a simply marker for PKA activity. We understand that other kinases have been reported to also phosphorylate the same site and our work with the PKA inhibitor, 14-22 (Figure 6D), provides strong evidence that in our work, Ser133 CREB phosphorylation results from elevated PKA activity.

As the reviewer has noted, previous work suggests that CREB is phosphorylated in the nucleus after passive diffusion of active PKAc into the nucleus (Hagiwara et al., 1993 and reviewed by Mayr and Montminy, 2001). This suggests that unlike the mitochondrial matrix that appears to have a distinct pool of PKA (Acin-Perez et al., 2009 and Lefkimmiatis, Leronni and HoferK, 2013), the interplay between cytosolic PKA activation and nuclear CREB responses is much more indiscriminate. Our new data shows that the phospho- STAT5/PKAc complex can only be detected in the cytosolic, non-mitochondrial compartment (Figure 7E), indicating that some cytosolic PKAc becomes activated that can trigger nuclear CREB phosphorylation. We hope that our new data and accompanying brief discussion (found in the Figure 7 Results section) meets the reviewers standards.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

1) All reviewers were concerned about the quality of Figure 4, which should be replaced. Also, discussion should be included as to why the level of phosphorylation is so low, as well as an explanation of what efforts were made to increase the level of phosphoryaltion.

We have now elaborated on an explanation for the low level of phosphorylation in vitroin our Discussion section. We increased incubation time from 10 minutes to 30 minutes for [γ-32P]-ATP labeling experiments to help reach maximal phosphorylation and have now included a note regarding this procedural modification in both the main text as well as the “Materials and methods” section.

We have also repeated the experiment for Figure 4F and replaced it with one that is more acceptable. We hope the reviewers are satisfied with the improved figure quality.

Overall, we have independently repeated these results several times and using complimentary approaches. We strongly believe that we have provided compelling evidence that FECH is a PKA target.

2) Given that the sample is only 10% phosphorylated, interpretation of the kinetic data in Figure 5 should be specifically addressed.

All enzyme kinetic constants are a function of temperature. Our kinetic assays were performed at room temperature while single time-point experiments were carried out at 37oC. This is a major caveat that limits the interpretation of in vitroresults. It is not technically feasible to perform a radioactive labeling assay at the desired 30-second intervals at 37oC. We have included a brief note regarding this caveat in our revised work (Results section).

Although the extent of Fech phosphorylation at the non-physiological room temperature was ~10%, this apparent modest reaction is at the terminus of a multi-step signalling cascade starting with Epo stimulation. Therefore, one would expect an “amplification effect” which would greatly magnify the apparent small changes in direct phosphorylation. Examples of such amplification effect would be the coagulation cascade or the complement system.

3) If possible, the manuscript would be strengthened by including a native gel for the in vitro phosphorylated protein, looking for a gel shift to confirm the quantification reported. Also, would it be possible to separate the phospho protein by ion exchange chromatography?

We thank the reviewers for their interest in our work. Actually, we are trying to develop a system where native phosphoproteins are produced by engineered E. coli (Park et al., 2011, Science and Lajoie et al., 2014, Science). If we are successful, we will be able to not only perform more detailed enzymatic assays but also obtain more structural data regarding how phosphorylation may influence enzymatic activity. These efforts would require more in-depth analysis that are, however, beyond the scope of the current manuscript and beyond the reasonable, allocated time frame for a revision.

https://doi.org/10.7554/eLife.24767.018

Article and author information

Author details

  1. Jacky Chung

    Division of Hematology, Brigham and Women’s Hospital, Harvard Medical School, Boston, United States
    Contribution
    JC, Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.

  2. Johannes G Wittig

    Division of Hematology, Brigham and Women’s Hospital, Harvard Medical School, Boston, United States
    Present address
    School of Biological Sciences, University of East Anglia, Norwich, United Kingdom
    Contribution
    JGW, Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0598-2897

  3. Alireza Ghamari

    Division of Hematology-Oncology, Boston Children’s Hospital, Harvard Medical School, Boston, United States
    Contribution
    AG, Data curation, Validation, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  4. Manami Maeda

    Division of Hematology, Brigham and Women’s Hospital, Harvard Medical School, Boston, United States
    Present address
    Center for Cellular and Molecular Medicine, Kyushu University Hospital, Fukuoka, Japan
    Contribution
    MM, Data curation, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  5. Tamara A Dailey

    1. Department of Microbiology, University of Georgia, Athens, United States
    2. Department of Biochemistry and Molecular Biology, University of Georgia, Athens, United States
    Contribution
    TAD, Data curation, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  6. Hector Bergonia

    Division of Hematology and Hematologic Malignancies, University of Utah School of Medicine, Salt Lake City, United States
    Contribution
    HB, Data curation, Formal analysis, Investigation
    Competing interests
    The authors declare that no competing interests exist.

  7. Martin D Kafina

    Division of Hematology, Brigham and Women’s Hospital, Harvard Medical School, Boston, United States
    Contribution
    MDK, Formal analysis, Validation, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  8. Emma E Coughlin

    Genome Center of Wisconsin, Madison, United States
    Contribution
    EEC, Data curation, Formal analysis, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  9. Catherine E Minogue

    Department of Chemistry, University of Wisconsin-Madison, Madison, United States
    Contribution
    CEM, Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  10. Alexander S Hebert

    Genome Center of Wisconsin, Madison, United States
    Contribution
    ASH, Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  11. Liangtao Li

    Department of Pathology, University of Utah School of Medicine, Salt Lake City, United States
    Contribution
    LL, Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  12. Jerry Kaplan

    Department of Pathology, University of Utah School of Medicine, Salt Lake City, United States
    Contribution
    JK, Conceptualization, Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  13. Harvey F Lodish

    Whitehead Institute for Biomedical Research, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    HFL, Formal analysis, Supervision, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  14. Daniel E Bauer

    1. Division of Hematology-Oncology, Boston Children’s Hospital, Harvard Medical School, Boston, United States
    2. Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, United States
    Contribution
    DEB, Formal analysis, Supervision, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  15. Stuart H Orkin

    1. Division of Hematology-Oncology, Boston Children’s Hospital, Harvard Medical School, Boston, United States
    2. Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, United States
    Contribution
    SHO, Supervision, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  16. Alan B Cantor

    1. Division of Hematology-Oncology, Boston Children’s Hospital, Harvard Medical School, Boston, United States
    2. Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, United States
    Contribution
    ABC, Formal analysis, Supervision, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  17. Takahiro Maeda

    Division of Hematology, Brigham and Women’s Hospital, Harvard Medical School, Boston, United States
    Present address
    Center for Cellular and Molecular Medicine, Kyushu University Hospital, Fukuoka, Japan
    Contribution
    TM, Formal analysis, Supervision, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  18. John D Phillips

    Division of Hematology and Hematologic Malignancies, University of Utah School of Medicine, Salt Lake City, United States
    Contribution
    JDP, Data curation, Supervision, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  19. Joshua J Coon

    1. Genome Center of Wisconsin, Madison, United States
    2. Department of Chemistry, University of Wisconsin-Madison, Madison, United States
    3. Department of Biomolecular Chemistry, University of Wisconsin-Madison, Madison, United States
    Contribution
    JJC, Formal analysis, Supervision, Validation, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  20. David J Pagliarini

    Department of Biochemistry, University of Wisconsin-Madison, Madison, United States
    Contribution
    DJP, Formal analysis, Supervision, Investigation, Methodology, Project administration
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0001-0087

  21. Harry A Dailey

    1. Department of Microbiology, University of Georgia, Athens, United States
    2. Department of Biochemistry and Molecular Biology, University of Georgia, Athens, United States
    Contribution
    HAD, Formal analysis, Supervision, Investigation, Methodology
    Competing interests
    The authors declare that no competing interests exist.

  22. Barry H Paw

    1. Division of Hematology, Brigham and Women’s Hospital, Harvard Medical School, Boston, United States
    2. Division of Hematology-Oncology, Boston Children’s Hospital, Harvard Medical School, Boston, United States
    3. Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, United States
    Contribution
    BHP, Conceptualization, Resources, Supervision, Writing—review and editing
    For correspondence
    bpaw@rics.bwh.harvard.edu
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0492-1419

Funding

National Heart, Lung, and Blood Institute (P01 HL032262)

  • Harvey F Lodish
  • Daniel E Bauer
  • Stuart H Orkin
  • Alan B Cantor
  • Barry H Paw

National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK070838)

  • Barry H Paw

American Cancer Society (RSG-13-379-01-LIB)

  • Takahiro Maeda

American Society of Hematology

  • Jacky Chung
  • Daniel E Bauer

Canadian Institutes of Health Research

  • Jacky Chung

National Institute of Diabetes and Digestive and Kidney Diseases (K08 DK093705)

  • Daniel E Bauer

National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK052380)

  • Jerry Kaplan

National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK090257)

  • John D Phillips

National Institutes of Health (R01 GM114122)

  • Joshua J Coon

National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK096501)

  • Harry A Dailey

National Institutes of Health (R01 GM115591)

  • David J Pagliarini

National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK098672)

  • David J Pagliarini

National Institutes of Health (P41 GM108538)

  • Joshua J Coon

National Institute of Diabetes and Digestive and Kidney Diseases (U54 DK110858)

  • John D Phillips

Diamond Blackfan Anemia Foundation

  • Barry H Paw

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to thank members of our lab as well as Eva Fast, Robert I Handin, Vijay G Sankaran, and Leonard I Zon for insightful discussions and technical assistance. We thank Arthur Skoultchi (Albert Einstein Medical College, Bronx, NY, USA) for the DS19 MEL subclone and Meredith S Irwin (Hospital for Sick Children, Toronto, ON, Canada) for the human UT7 erythroid cells. We thank Leonard I Zon (Harvard Medical School, Boston, MA, USA) for the Tg(globin-LCR:eGFP) transgenic zebrafish line. We also thank Ann Mullally (Brigham and Women’s Hospital, Boston, MA, USA) for the JAK2 inhibitor, Ruxolitinib, and the human HEL cells. We thank Eva Buy and her crew for the zebrafish animal husbandry. In vitro HPLC analyses for porphyrins were performed at the University of Utah Center for Iron and Heme Disorders (NIH U54 DK110858). This work was supported by grants from the Canadian Institutes of Health Research (CIHR Post-doctoral Fellowship, JC), the American Society of Hematology (Basic Research Fellow ASH Scholar Award, JC; Junior Faculty ASH Scholar Award, DEB), the American Cancer Society (RSG-13-379-01-LIB, TM), the Diamond Blackfan Anemia Foundation (BHP), and the National Institutes of Health (K08 DK093705, DEB; R01 DK052380, JK; R01 DK090257, JDP; P41 GM108538, JJC; R01 DK098672 and R01 GM115591, DJP; R01 DK096501, HAD; R01 DK070838, BHP; P01 HL032262, BHP, ABC, DEB, SHO, and HFL).

Ethics

Animal experimentation: In full compliance with BWH IACUC A4752-01 (Protocol #2016N000117) and BCH IACUC Protocol #15-07-2974R.

Reviewing Editor

  1. David Ginsburg, Howard Hughes Medical Institute, University of Michigan, United States

Publication history

  1. Received: December 30, 2016
  2. Accepted: May 28, 2017
  3. Accepted Manuscript published: May 29, 2017 (version 1)
  4. Version of Record published: June 20, 2017 (version 2)

Copyright

© 2017, Chung et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,307
    Page views
  • 496
    Downloads
  • 31
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Jacky Chung
  2. Johannes G Wittig
  3. Alireza Ghamari
  4. Manami Maeda
  5. Tamara A Dailey
  6. Hector Bergonia
  7. Martin D Kafina
  8. Emma E Coughlin
  9. Catherine E Minogue
  10. Alexander S Hebert
  11. Liangtao Li
  12. Jerry Kaplan
  13. Harvey F Lodish
  14. Daniel E Bauer
  15. Stuart H Orkin
  16. Alan B Cantor
  17. Takahiro Maeda
  18. John D Phillips
  19. Joshua J Coon
  20. David J Pagliarini
  21. Harry A Dailey
  22. Barry H Paw
(2017)
Erythropoietin signaling regulates heme biosynthesis
eLife 6:e24767.
https://doi.org/10.7554/eLife.24767

Categories and tags

Research organism

Further reading

    1. Cell Biology
    2. Developmental Biology

    Transcriptional drifts associated with environmental changes in endothelial cells

    Yalda Afshar, Feyiang Ma ... Luisa Iruela-Arispe
    Research Article

    Environmental cues, such as physical forces and heterotypic cell interactions play a critical role in cell function, yet their collective contributions to transcriptional changes are unclear. Focusing on human endothelial cells, we performed broad individual sample analysis to identify transcriptional drifts associated with environmental changes that were independent of genetic background. Global gene expression profiling by RNAseq and protein expression by LC-MS directed proteomics distinguished endothelial cells in vivo from genetically matched culture (in vitro) samples. Over 43% of the transcriptome was significantly changed by the in vitro environment. Subjecting cultured cells to long-term shear stress significantly rescued the expression of approximately 17% of genes. Inclusion of heterotypic interactions by co-culture of endothelial cells with smooth muscle cells normalized approximately 9% of the original in vivo signature. We also identified novel flow dependent genes, as well as genes that necessitate heterotypic cell interactions to mimic the in vivo transcriptome. Our findings highlight specific genes and pathways that rely on contextual information for adequate expression from those that are agnostic of such environmental cues.

    1. Developmental Biology

    Histone deacetylase 1 maintains lineage integrity through histone acetylome refinement during early embryogenesis

    Jeff Jiajing Zhou, Jin Sun Cho ... Ken WY Cho
    Research Article

    Histone acetylation is a pivotal epigenetic modification that controls chromatin structure and regulates gene expression. It plays an essential role in modulating zygotic transcription and cell lineage specification of developing embryos. While the outcomes of many inductive signals have been described to require enzymatic activities of histone acetyltransferases and deacetylases (HDACs), the mechanisms by which HDACs confine the utilization of the zygotic genome remain to be elucidated. Here, we show that histone deacetylase 1 (Hdac1) progressively binds to the zygotic genome from mid blastula and onward. The recruitment of Hdac1 to the genome at blastula is instructed maternally. Cis-regulatory modules (CRMs) bound by Hdac1 possess epigenetic signatures underlying distinct functions. We highlight a dual function model of Hdac1 where Hdac1 not only represses gene expression by sustaining a histone hypoacetylation state on inactive chromatin, but also maintains gene expression through participating in dynamic histone acetylation-deacetylation cycles on active chromatin. As a result, Hdac1 maintains differential histone acetylation states of bound CRMs between different germ layers and reinforces the transcriptional program underlying cell lineage identities, both in time and space. Taken together, our study reveals a comprehensive role for Hdac1 during early vertebrate embryogenesis.

    1. Developmental Biology
    2. Genetics and Genomics

    Evidence for RNA or protein transport from somatic tissues to the male reproductive tract in mouse

    Vera RInaldi, Kathleen Messemer ... Oliver J Rando
    Research Article

    The development of tools to manipulate the mouse genome, including knockout and transgenic technology, has revolutionized our ability to explore gene function in mammals. Moreover, for genes that are expressed in multiple tissues or at multiple stages of development, the use of tissue-specific expression of the Cre recombinase allows gene function to be perturbed in specific cell types and/or at specific times. However, it is well known that putative tissue-specific promoters often drive unanticipated 'off target' expression. In our efforts to explore the biology of the male reproductive tract, we unexpectedly found that expression of Cre in the central nervous system resulted in recombination in the epididymis, a tissue where sperm mature for ~1-2 weeks following the completion of testicular development. Remarkably, we not only observed reporter expression in the epididymis when Cre expression was driven from neuron-specific transgenes, but also when Cre expression in the brain was induced from an AAV vector carrying a Cre expression construct. A surprisingly wide range of Cre drivers - including six different neuronal promoters as well as the adipose-specific Adipoq Cre promoter - exhibited off target recombination in the epididymis, with a subset of drivers also exhibiting unexpected activity in other tissues such as the reproductive accessory glands. Using a combination of parabiosis and serum transfer experiments, we find evidence supporting the hypothesis that Cre may be trafficked from its cell of origin to the epididymis through the circulatory system. Together, our findings should motivate extreme caution when interpreting conditional alleles, and suggest the exciting possibility of inter-tissue RNA or protein trafficking in modulation of reproductive biology.