To identify conserved enzyme substrates of Ofd1/OGFOD1, we performed a yeast two-hybrid screen for human OGFOD1 binding partners using a human fetal brain cDNA library as prey. We isolated uS12 (coded by the RPS23 gene) as a binding partner of OGFOD1 and confirmed that the interaction is conserved between Ofd1 and Rps23 in fission yeast (Figure 1A; Figure 1—figure supplement 1A).

Figure 1 with 2 supplements see all

Download asset Open asset

Figure 1—source data 1

SILAC mass spectrometry data.

SILAC ratios used to quantify Rps23 P62 hydroxylation shown in Figure 1F and Figure 4A–B.

https://doi.org/10.7554/eLife.28563.006

Download elife-28563-fig1-data1-v3.xlsx

To investigate whether Ofd1 binds directly to Rps23, we used an in vitro GST pull-down assay. In contrast to free GST, full-length GST-Rps23 bound Ofd1 (Figure 1B, lanes 2–3), indicating direct binding between Ofd1 and Rps23. Next, we mapped the region of Rps23 required for binding to Ofd1. Rps23 contains an evolutionarily divergent N-terminal extension domain (aa 1–46) followed by a conserved C-terminal globular domain (aa 47–143) (Smith et al., 2008). We fused GST to Rps23 truncations and found that Ofd1 binds to the N-terminal extension domain (Figure 1B, lanes 4–5). Furthermore, Rps23 aa 1–23 were necessary and sufficient to capture Ofd1 (Figure 1B, lanes 6–7). This region of Rps23 (Figure 1C, colored magenta) is buried in assembled 40S subunits and thus inaccessible to Ofd1 based on crystal structures of budding yeast ribosomes (Ben-Shem et al., 2011). Therefore, Ofd1 binds to Rps23 prior to its assembly into the 40S subunit in the nucleus. Since unassembled Rps23 represents only 2% of total Rps23 in cells (Figure 1—figure supplement 1–1B, lanes 1 and 4), we used ribosome-depleted lysates to test if Ofd1 binds Rps23 in vivo. Immunopurified wild-type Ofd1 bound unassembled Rps23 in the presence of the crosslinker DSP, but failed to bind Rps5, a ribosomal protein of similar size and charge (Figure 1D, lanes 7–8). In addition, the Fe(II)-binding mutant Ofd1 H142A D144A was unable to pull down Rps23 (Figure 1D, lanes 11–12), indicating that an intact Ofd1 active site is required to bind Rps23.

Since oxygenase function is required for Ofd1-Rps23 binding, we hypothesized that Rps23 is a substrate of Ofd1. We identified Rps23 P62 as a promising target for hydroxylation based on earlier work that reported a +16 Dalton (Da) mass shift for mammalian RPS23 aa 61–68: QPNSAIRK (Louie et al., 1996). P62 is essential and part of a highly conserved loop that projects into the decoding center of the ribosome upon anticodon binding (Sharma et al., 2007). We confirmed that the +16 Da mass shift occurred on this proline residue by incubating Ofd1 with purified 6xHis-RPS23 and performing MS/MS that targeted the QPNSAIR peptide (Figure 1—figure supplement 1–1C). In light of subsequent studies showing that Ofd1 and its fungal homologs dihydroxylate P62 (Loenarz et al., 2014), we developed a stable isotope labeling with amino acids in cell culture (SILAC) assay to quantify Rps23 P62 hydroxylation. Wild-type cells were labeled with heavy (H) lysine and ofd1Δ cells with light (L) lysine, mixed, and processed to enrich for Rps23 from assembled 40S subunits. Following digestion by LysC, H/L peptide ratios were measured by LC-MS/MS (Figure 1E). We detected only dihydroxylated P62 in wild-type cells, consistent with published findings (Loenarz et al., 2014). Analysis of the wild-type-ofd1Δ SILAC pair revealed a complete loss of dihydroxylated P62 in ofd1Δ cells, accompanied by the appearance of unmodified P62 (Figure 1F). We found similar results for the SILAC pair of wild-type and enzyme-dead ofd1 H142A D144A (Figure 1F). Collectively, these data show that Ofd1 binds unassembled Rps23 (aa 1–23) and independently demonstrate that Ofd1 catalyzes Rps23 dihydroxylation.

While Rps23 is the only known enzyme substrate of Ofd1, we previously showed that Ofd1 directly binds Nro1, a negative regulator of Ofd1 and a nuclear import adaptor (Lee et al., 2009; Yeh et al., 2011). To determine if a common Ofd1 binding site exists, we aligned the sequences of Rps23 (aa 1–23) and Nro1 (aa 1–30) required for binding to Ofd1. We identified a stretch of identical and chemically similar residues (Figure 2A) and mutated each amino acid to aspartate in the context of either GST-Rps23 FL (Figure 2B) or GST-Nro1 aa 1–30 (Figure 2C) to test if these residues are required for binding to Ofd1. Mutating the N-terminal basic residue, proline, glycine, or leucine completely abolished binding to Ofd1 for both Rps23 and Nro1, while mutating a non-conserved residue adjacent to the proline had an intermediate effect (Figure 2B–C, lanes 4–8). Downstream mutations revealed slight differences between the Ofd1 binding sequences: mutations in conserved alanine residues resulted in a severe reduction in Rps23-Ofd1 binding, while the corresponding mutations in Nro1 only partially disrupted binding (Figure 2B–C, lanes 10–11). For both Rps23 and Nro1 though, mutations in the most C-terminal amino acids displayed only mild defects in Ofd1 binding relative to upstream residues (Figure 2B–C, lanes 13–15). These similarities between the Ofd1 binding sites in Rps23 and Nro1 argue for a conserved Ofd1 binding sequence.

Figure 2

Download asset Open asset

Given that both Rps23 and Nro1 contain an Ofd1 binding site, we hypothesized that these proteins compete for binding to Ofd1 and predicted that down-regulation or deletion of one binding partner would increase binding between Ofd1 and the other partner. While Nro1 is not essential (Lee et al., 2009), Rps23 is required for cell viability due to its role in translation (Sharma et al., 2007). Two genes – rps23⁺ and rps2302⁺ – code for identical Rps23 proteins. While deletion of either gene results in a modest decrease in total Rps23 levels, levels of unassembled Rps23 are dramatically reduced by ~90% (Figure 1—figure supplement 1B). rps23Δ and rps2302Δ cells thus represent severe knock-downs with respect to this Ofd1 binding partner. To test if Nro1 and unassembled Rps23 compete for binding to Ofd1, we immunopurified Ofd1 from wild-type, nro1Δ, rps23Δ, and rps2302Δ lysates depleted of ribosomes and analyzed the bound fraction for Nro1 and Rps23. In wild-type cells treated with crosslinker, Ofd1 bound both Rps23 and Nro1, but not Rps5 (Figure 3A, lane 6). Surprisingly, Ofd1 failed to pull down Rps23 in nro1Δ cells (Figure 3A, lane 8), indicating that Ofd1-Rps23 binding requires Nro1. In addition, Ofd1 binding to Nro1 was weakened in rps23Δ and rps2302Δ cells (Figure 3A, lanes 9–10). In the absence of crosslinker, immunopurified Ofd1 bound less Nro1 in wild-type cells and the Ofd1-Rps23 interaction was barely detected (Figure 3B, lane 6). However, Ofd1-Nro1 binding was reduced in rps23Δ and rps2302Δ cells relative to wild-type cells across both conditions (Figure 3A–B; lanes 6, 9–10). Consistent with this, Nro1 bound less Ofd1 in rps23Δ cells and, to a lesser extent, rps2302Δ cells relative to wild-type when immunopurified from crosslinked samples (Figure 3C, lanes 6, 9–10). Unexpectedly, Nro1 showed strong binding to Rps23 in wild-type, ofd1Δ, rps23Δ, and rps2302Δ cells, independent of crosslinker (Figure 3C–D, lanes 6–7 and 9–10). Nro1-Rps23 binding was specific to Rps23 since Nro1 did not pull down Rps5 (Figure 3C–D, lanes 6–10). We consistently observed reduced unassembled Rps5 in rps23Δ and rps2302Δ cells, indicating that mechanisms exist to coordinate ribosomal protein expression. These results show that Rps23- and Nro1-binding to Ofd1 is interdependent rather than competitive, suggesting that they form a complex with Ofd1.

Figure 3

Download asset Open asset

To test this further, we assayed Nro1 and Rps23 binding in vitro using a GST pull-down assay. Full-length GST-Rps23 directly bound purified Nro1 (Figure 3E, lane 3). Using Rps23 truncations, we then found that Rps23 contains at least two distinct binding sites for Nro1 because both GST-Rps23 aa 1–46 and GST-Rps23 aa 47–143 bound Nro1 (Figure 3E, lanes 4 and 6). However, the Ofd1 and Nro1 binding sites on Rps23 do not overlap as GST-Rps23 aa 1–23 failed to bind Nro1 (Figure 3E, lane 5). These in vitro binding studies, together with the immunopurifications, demonstrate that Ofd1, Rps23, and Nro1 form a complex in cells.

The discovery that Nro1 is required for Ofd1 to bind to Rps23 in vivo (Figure 3A) led us to hypothesize that Nro1 functions in Rps23 hydroxylation. To test this, we examined Rps23 P62-containing peptides from wild-type and nro1Δ cells using SILAC. Interestingly, less than 10% of the detected dihydroxylated P62 signal originated from nro1Δ cells (Figure 4A), indicating that only 10% of the Rps23 is dihydroxylated in the absence of Nro1 compared to 100% in wild-type cells. Furthermore, we detected both unmodified and monohydroxylated P62 peptides with low H/L ratios, indicating that these peptides originated from nro1Δ cells (Figure 4A). To determine the relative abundance of the unmodified and monohydroxylated forms in cells lacking Nro1, we first measured the relative levels of unmodified Rps23 in nro1Δ and ofd1Δ cells using SILAC. We found that nro1Δ cells contributed 25% of the unmodified P62 signal, meaning that 32% of Rps23 P62 is unmodified in nro1Δ ribosomes. Consequently, monohydroxylated Rps23 P62 represents ~60% of the total Rps23 in these cells. (Figure 4B).

Figure 4

Download asset Open asset

Figure 4—source data 1

TMT mass spectrometry data.

TMT data used to quantify Rps23 P62 hydroxylation in vitro shown in Figure 4C.

https://doi.org/10.7554/eLife.28563.010

Download elife-28563-fig4-data1-v3.xlsx

The substantial fraction of monohydroxylated P62 in nro1Δ cells shows that the efficient dihydroxylation of Rps23 P62 requires Nro1. We tested if Nro1 participates directly in the reaction by reconstituting the reaction in vitro. Purified Rps23 and Ofd1 were incubated alone or in the presence of purified Nro1 and then analyzed by Tandem Mass Tag (TMT)-labeled LC-MS/MS to quantify P62 hydroxylation. The ratio of Ofd1:Rps23:Nro1 (1:10:10) replicated the physiological ratio of Ofd1:unassembled Rps23:Nro1 in cells (data not shown). Consistent with the in vivo results, addition of Nro1 to the in vitro reaction increased Rps23 P62 dihydroxylation by 1.6 fold (Figure 4C). The increase in dihydroxylated Rps23 was accompanied by a corresponding decrease in monohydroxylated Rps23, suggesting a role for Nro1 in facilitating the second hydroxylation event. The presence of Nro1 also resulted in a small but significant reduction in the levels of the unmodified P62 substrate compared to the reaction lacking Nro1. Together, these in vivo and in vitro results demonstrate that complete Rps23 dihydroxylation requires Nro1.

Thus far, our data shows that Ofd1 dihydroxylates unassembled Rps23 in a complex with Nro1. Since Nro1 is structurally similar to karyopherins and functions as a nuclear import adaptor for Ofd1 (Yeh et al., 2011), we tested if Nro1 plays a role in Rps23 localization. Wild-type cells that express Rps23-GFP from the rps2302 locus showed Rps23-GFP localized to the nucleus by live-cell imaging (Figure 5A). The Rps23-GFP fusion protein does not assemble into functional 40S subunits (data not shown) and localized to the nucleus rather than being exported to the cytosol. In nro1Δ cells, Rps23-GFP was excluded from the nucleus despite only a minor reduction in Rps23-GFP expression (Figure 5B), indicating that Nro1 is required for nuclear import of unassembled Rps23. However, alternate pathways for Rps23 import must exist since nro1Δ cells are viable. Furthermore, Ofd1-dependent hydroxylation of Rps23 P62 is not required for nuclear import of Rps23. In both ofd1Δ and ofd1 H142A D144A cells, Rps23-GFP localizes to the nucleus, as does Rps23 P62A-GFP in ofd1⁺ cells. (Figure 5A). These data indicate that Rps23 enters the nucleus regardless of hydroxylation state, and suggests that nuclear import of Rps23 is not blocked under hypoxic conditions.

Figure 5

Download asset Open asset

Because both Rps23 and Ofd1 require Nro1 for nuclear localization, we hypothesized that Nro1 imports Rps23 and Ofd1 together as a complex. Therefore, we predicted that Ofd1 might be mislocalized in rps23Δ cells due to the dramatic reduction in unassembled Rps23 and consequent reduction in complex formation (Figure 3A–D). mCherry-Ofd1 expressed from the endogenous locus localized primarily to the nucleus in wild-type cells, consistent with published data (Figure 5C) (Hughes and Espenshade, 2008). Deletion of rps23⁺ resulted in loss of nuclear localization with Ofd1 mislocalized throughout the cell (Figure 5C). This supports our hypothesis that Ofd1 and Rps23 traffic together to the nucleus with Nro1. Inhibition of Crm1-dependent nuclear export by leptomycin B (LMB) in rps23Δ cells restored Ofd1 nuclear localization, indicating that Ofd1 still traffics to the nucleus in these cells but at a reduced rate (Figure 5C). However, LMB treatment failed to restore nuclear localization of mCherry-Ofd1 in nro1Δ cells, consistent with the requirement of Nro1 for Ofd1 nuclear import (Figure 5C) (Yeh et al., 2011). Collectively, these data suggest that unassembled Rps23 is imported into the nucleus in a complex with Nro1 and Ofd1 and that Rps23 dihydroxylation likely occurs in conjunction with nuclear import.

In addition to their roles in Rps23 hydroxylation and nuclear import, Ofd1 and Nro1 are key regulators of Sre1 signaling and hypoxic adaptation in fission yeast (Hughes and Espenshade, 2008; Lee et al., 2009). Ofd1 inhibits signaling by binding Sre1N, which both prevents Sre1N DNA-binding and accelerates its degradation (Lee et al., 2011). Under low oxygen, Ofd1 binds Nro1, freeing Sre1N to activate gene expression. Given that Ofd1-Nro1 binding requires Rps23 (Figure 3A–D), we hypothesized that unassembled Rps23 functions as a positive regulator of Sre1N like Nro1.

To test if Rps23 regulates Sre1, we first analyzed growth of rps23Δ and rps2302Δ cells in the presence of the hypoxia-mimetic cobalt. In fission yeast, cobalt inhibits ergosterol synthesis, resulting in reduced ergosterol levels and the accumulation of methylated sterol intermediates (Lee et al., 2007). Cobalt likely disrupts ergosterol synthesis by causing defects in Fe-dependent enzymes since overexpression of erg25⁺, an Sre1N target gene and Fe-dependent enzyme, rescues growth on cobalt. Consequently, Sre1 activity is required for growth in the presence of cobalt and sre1Δ cells show severe growth defects on plates supplemented with cobalt chloride (Stewart et al., 2011). Growth on cobalt thus reflects the cell’s capacity to activate Sre1. Like nro1Δ cells, rps23Δ and rps2302Δ cells failed to grow on cobalt chloride (Figure 6A). Consistent with this, rps23Δ and rps2302Δ cells, like nro1Δ cells, showed decreased Sre1N activation and reduced precursor levels under low oxygen (Figure 6B, lanes 7–12). In contrast, ofd1Δ cells resemble wild-type cells with respect to Sre1N levels despite the role of Ofd1 as a negative regulator of Sre1N (Figure 6B, lanes 1–2 and 5–6). Ofd1-independent mechanisms act upstream of Ofd1 to regulate the transport and cleavage of the Sre1 precursor to generate Sre1N (Hughes and Espenshade, 2008; Stewart et al., 2011). The failure of rps23Δ and rps2302Δ cells to grow on cobalt was due to defects in Sre1N production rather than general defects in translation since expression of Sre1N from a plasmid rescued growth on cobalt (Figure 6C). Furthermore, rps23Δ and rps2302Δ cells failed to upregulate the Sre1N target gene hem13⁺ in the absence of oxygen, but hypoxic activation of the Sre1N-independent gene erg1⁺ was normal (Figure 6—figure supplement 6–1A). Finally, the rps23 mutant was the only ribosomal gene identified in both fission yeast deletion collection screens for cobalt-sensitivity (Ryuko et al., 2012; Burr et al., 2016), and we confirmed the specificity of the cobalt phenotype (Figure 6—figure supplement 6–1B). These two screens failed to identify rps2302Δ cells as cobalt-sensitive; however, we were unable to verify the deletion of rps2302⁺ by PCR in the deletion collection strain (data not shown). These data demonstrate that Rps23 is a specific positive regulator of Sre1 signaling.

Figure 6 with 1 supplement see all

Download asset Open asset

To test whether Rps23 regulates Sre1 after proteolytic activation, we used strains that express only soluble Sre1N from the sre1 locus, bypassing the requirement for cleavage (Hughes and Espenshade, 2008). Low oxygen stabilized Sre1N in wild-type cells (Figure 6D, lanes 1–2), but rps23Δ and rps2302Δ cells failed to accumulate Sre1N under low oxygen (Figure 6D, lanes 4 and 6). Deletion of ofd1⁺ in rps23Δ and rps2302Δ cells restored Sre1N expression to wild-type levels under low oxygen (Figure 6D, lanes 10 and 12), indicating that the defects in Sre1N accumulation are the result of increased inhibition by Ofd1 and further demonstrating that rps23Δ and rps2302Δ cells are not defective in Sre1N translation.

The model for Ofd1 inhibition of Sre1N predicts that Ofd1 acts on Sre1N directly, but direct binding has not been demonstrated and the Ofd1 binding site on Sre1N has not been identified. Previous yeast two-hybrid studies mapped the Ofd1 binding region to Sre1 aa 271–340 (Lee et al., 2011). Alignment of this region with the Ofd1 binding sites in Rps23 and Nro1 identified Sre1 aa 286–297 as a putative Ofd1 binding site (Figure 6E). To test if Sre1N binds Ofd1 directly, we performed an in vitro binding assay and found that GST-Sre1 aa 271–340 bound purified Ofd1 (Figure 6F, lane 3). Mutational analysis showed that Sre1N-Ofd1 binding required Sre1N residues 286–293 and 297 (Figure 6F, lanes 4–11 and 15). In contrast, individually mutating Sre1N aa 294–296 impaired Ofd1 binding but did not abolish the interaction (Figure 6F, lanes 12–14). These data indicate that Sre1N binds Ofd1 directly through a conserved motif shared with Rps23 and Nro1. Collectively, these experiments demonstrate that unassembled Rps23 functions with Nro1 as a positive regulator of Sre1N activity by binding Ofd1 and preventing its binding to Sre1N.

Our data indicate that Ofd1 binds independently to both a complex of Rps23-Nro1 and Sre1N, a dimeric, basic helix-loop-helix leucine zipper transcription factor (Párraga et al., 1998). Given that these proteins share a consensus Ofd1 binding sequence, Ofd1 likely forms a dimer, which is consistent with structural studies (Figure 7A) (Kim et al., 2010; Henri et al., 2010; Horita et al., 2015). Deletion of nro1⁺ reduces Ofd1-Rps23-Nro1 complex formation (Figure 3A) and inhibits Sre1N signaling (Lee et al., 2009), indicating that Rps23-Nro1 and Sre1N dimer compete for binding to Ofd1. Finally, Rps23 P62 hydroxylation by Ofd1 requires oxygen (Loenarz et al., 2014). Based on these data, we proposed a model in which Ofd1 is free to bind Sre1N and inhibit transcriptional activity under normoxia. Under hypoxia, Ofd1 hydroxylation of Rps23 is inhibited, sequestering Ofd1 in complex with Rps23-Nro1 and freeing Sre1N to activate hypoxic gene expression (Figure 7A). Consistent with this, a 90% reduction in unassembled Rps23 (rps23Δ cells; Figure 1—figure supplement 1–1B, lanes 4–5) resulted in impaired Sre1N signaling under low oxygen due to increased inhibition by Ofd1 (Figure 6D). rps23Δ cells fail to produce sufficient Rps23 to effectively sequester Ofd1 under hypoxia.

Figure 7 with 1 supplement see all

Download asset Open asset

Based on this model, we hypothesized that elevated levels of unassembled Rps23 would further sequester Ofd1 and stabilize Sre1N in the presence of oxygen. We overexpressed Rps23 in 2XSRE-ura4⁺ reporter cells and analyzed growth on plates lacking uracil (Figure 7B) (Lee et al., 2009). In this reporter strain, sre1N and sre1Δ cells failed to grow in the absence of uracil when transformed with the empty vector control (Figure 7B, spots 1 and 3). However, Rps23 overexpression in sre1N cells, but not sre1Δ cells, supported growth in the absence of uracil, indicating activation of Sre1 signaling (Figure 7B, spots 2 and 4). Furthermore, overexpression of GST-Rps23 in sre1N cells led to the accumulation of Sre1N under normoxia compared to control cells (Figure 7C). GST-Rps23 serves as a proxy for unassembled Rps23 given its ability to bind both Ofd1 and Nro1 (Figures 1B and 3E) while being unable to assemble into functional 40S subunits: cells that express GST-Rps23 from the rps2302 locus are not viable in an rps23Δ background (Figure 7—figure supplement 1A). These data support the sequestration model by demonstrating that changes in Rps23 levels affect Sre1 signaling.

Next, we tested if the Ofd1-binding sites in Nro1 and Rps23 mediate the sequestration of Ofd1. We mutated the Ofd1-binding site at several conserved residues in Nro1 and Rps23 and expressed these mutants in nro1Δ sre1N and rps2302Δ sre1N cells, respectively. These cells carrying empty vector failed to grow in the presence of cobalt (Figure 7D), consistent with the fact that both nro1Δ sre1N and rps2302Δ sre1N cells were unable to activate Sre1N under hypoxia due to Ofd1-dependent inhibition (Lee et al., 2009; Figure 6D, lanes 5–6, 11–12). We predicted that expression of wild-type Nro1 and Rps23 would rescue growth on cobalt by sequestering Ofd1, but that Nro1 and Rps23 mutants that fail to bind Ofd1 would consequently fail to restore growth.

nro1Δ sre1N cells expressing wild-type Nro1 grew on cobalt while cells expressing the Ofd1-binding mutants - Nro1 P6D, G8D, and A11D – phenocopied empty vector controls (Figure 7D). Consistent with these cobalt growth defects, cells expressing Nro1 P6D, G8D, and A11D failed to activate Sre1N under low oxygen despite equal Nro1 expression levels (Figure 7—figure supplement 7–1B, lanes 3–10). Expression of Nro1 L15D, a mutant predicted to retain Ofd1-binding (Figure 2), partially rescued both growth on cobalt and Sre1N accumulation under hypoxia (Figure 7D; Figure 7—figure supplement 1B, lanes 4,12). rps2302Δ sre1N cells expressing wild-type rps2302⁺ grew on cobalt (Figure 7D). Unexpectedly, although Rps23 P4D, G6D, and A9D are defective for Ofd1-binding in vitro (Figure 2B), rps2302Δ sre1N cells expressing these mutants showed partial to full recovery on cobalt relative to cells expressing wild-type Rps23 or Rps23 L13D, a mutant that binds Ofd1 in vitro. Furthermore, the Rps23 mutants except for G6D restored Sre1N activity under low oxygen (Figure 7—figure supplement 1C). Reduced expression of Rps23 G6D may explain its inability to fully rescue Sre1N levels (Figure 7—figure supplement 1D). These findings indicate that although the Ofd1-binding sites in Nro1 and Rps23 behave similarly in vitro, these sequences do not contribute equally to Ofd1 sequestration in vivo, with sequestration more dependent on Nro1 residues.

Our model also predicts that Ofd1 oxygenase activity is crucial for the oxygen-dependent sequestration of Ofd1 by Rps23-Nro1 and hypoxia stabilizes the Ofd1-Rps23-Nro1 complex due to the inability of Ofd1 to modify Rps23 P62. Ofd1 requires both 2OG and molecular oxygen as co-substrates for hydroxylation so we treated wild-type cells with dimethyloxalylglycine (DMOG), a cell permeable competitive inhibitor of Ofd1 that is metabolized to a 2OG analog, to test if enzyme activity affects Ofd1 binding to Rps23 and Nro1 (Jaakkola et al., 2001). Ofd1 immunopurified from cells treated with DMOG bound more Nro1 and Rps23 compared to vehicle-treated cells, in both the absence and presence of crosslinker (Figure 7E, lanes 5–8). Next, we examined whether Ofd1-Rps23-Nro1 complex stability is oxygen-dependent. Ofd1 immunopurified from wild-type cells bound more Rps23 and Nro1 in the absence of oxygen compared to cells grown in normoxia (Figure 7F, lanes 3–4). Together, these experiments show that the Ofd1-Rps23-Nro1 complex is stabilized under conditions that prevent P62 hydroxylation and support the sequestration model for Ofd1-dependent regulation of Sre1N.

Common lab reagents were obtained from either Sigma or Thermo Fisher Scientific. Oligonucleotides were provided by Integrated DNA Technologies (Coralville, IA). Heavy lysine (¹³C₆/¹⁵N₂) was purchased from Cambridge Isotope Laboratories (Tewksbury, MA). DMOG was from Frontier Scientific Inc (Logan, UT). Protease inhibitors (0.5 μM PMSF, 10 μg/ml leupeptin, and 5 μg/ml pepstatin) were included in buffers where indicated.

Strains in this study are described in Supplementary file 1 and were generated using standard techniques (Bähler et al., 1998; Alfa and Cold Spring Harbor Laboratory, 1993). All fission yeast strains were validated by PCR -sequencing and western blotting. Haploid S. pombe cells were cultured at 30°C in rich medium (0.5% [w/v] yeast extract (BD Biosciences, San Jose, CA), 3% [w/v] glucose, 225 μg/ml each of uracil, adenine, leucine, histidine, and lysine) to 1 × 10⁷ cells/ml unless otherwise indicated (Moreno et al., 1991). Minimal medium constitutes Edinburgh Minimal Medium (MP Biomedical, Santa Ana, CA) plus supplements (225 μg/ml each of uracil, adenine, leucine, histidine, and lysine).

sre1N (aa 1–440) under control of the constitutive CaMV promoter was described in Hughes et al. (2005). rps2302⁺ and rps2302 point mutants were cloned into a constitutive adh1-driven expression vector with leu2⁺ marker, derived from pART1 (McLeod et al., 1987). GST and GST-rps2302 were placed under the control of the inducible nmt1 promoter by insertion into the BamHI/NotI restriction sites of pSLF172 (Forsburg and Sherman, 1997). nro1⁺ under control of the CaMV promoter was used as a template to generate nro1 point mutants and was described in Yeh et al. (2011). Transformations were performed by electroporation unless stated otherwise.

Secondary antibodies were obtained from LI-COR (Lincoln, NE): IRDye800CW/IRDye680RD mouse and rabbit IgG, IRDye 680RD Detection Reagent. Commercially obtained antibodies include mouse anti-GST monoclonal (RRID:AB_291280, MMS-112R-500, Covance, Princeton, NJ), mouse anti-GFP monoclonal (RRID:AB_390913, 1814460, Roche Applied Science, Penzberg, Germany), mouse anti-Rps23 monoclonal (RRID:AB_2180354, SJ-K2, Santa Cruz, Dallas, TX), mouse anti-Rps5 monoclonal (RRID:AB_2713966, A-8, Santa Cruz), and mouse anti-actin monoclonal (RRID:AB_626632, C4, Santa Cruz). Lab-made antibodies are rabbit polyclonal antisera and were previously described: anti-Ofd1 (GST-Ofd1 full-length antigen; Hughes and Espenshade, 2008), anti-Nro1 (6xHis-Nro1 full-length antigen; Lee et al., 2009), anti-Dsc5 (6xHis-Dsc5 aa 251–427 antigen; Stewart et al., 2012), and anti-Sre1 (6xHis-Sre1 aa 1–260 antigen; RRID:AB_2713965, Hughes et al., 2005).

Yeast two-hybrid screen (>3 fold library coverage) was conducted at Duke University using a human fetal brain cDNA library. RPS23 represented 80% of positive clones (56/70). Confirmation of screen hits was performed following Matchmaker Gal4 Two-Hybrid System user manual (Clontech, Mountain View, CA) using the yeast two-hybrid S. cerevisiae strain AH109 (James et al., 1996). Briefly, bait (Gal4_BD empty vector, Gal4_BD-Ofd1 or Gal4_BD-OGFOD1) and prey (Gal4_AD empty vector, Gal4_AD-Rps23, or Gal4_AD-RPS23) plasmids were co-transformed into AH109 cells. Transformants were equally divided into two halves, then plated on control (SD/-Leu/-Trp) or reporter (SD/-Ade/-His/-Leu/-Trp/X-α-Gal) plates. Plates were incubated at 30°C for 8 days and images were acquired using a flatbed scanner in transmitted light mode at a resolution of 600 dpi.

Strains were cultured in rich medium to 0.5–1.0 × 10⁷ cells/ml. Cells (1 × 10⁸) were pelleted and lysed by vortexing with glass beads (0.5 mm) for 10 min at 4°C in 1% [v/v] NP-40, 50 mM Hepes-HCl pH 7.4, 100 mM NaCl, 1.5 mM MgCl₂, and protease inhibitors. Lysates were centrifuged at 20,000 × g for 1 min at 4°C and the resulting supernatant was designated the whole cell lysate (WCL). Ribosomes were pelleted from WCL by centrifugation at 80,000 rpm for 16 min at 4°C using an Optima TLX ultracentrifuge and TLA100 rotor (Beckman Coulter, Brea, CA). WCL (5 μg) and ten-fold by volume supernatant were loaded on 16% acrylamide gels. Rps23 detected in the ribosome-cleared supernatant by immunoblot was considered unassembled and normalized to Dsc5. Rps23 fractions were calculated as the mean of three biological replicates ± SEM and analyzed by paired, one-tailed t test.

GST fusion protein expression vectors were generated by cloning gene sequences into the bacterial expression vector pGEX-4T3 (GE Healthcare Life Sciences, Chicago, IL) using BamHI/NotI restriction sites. Fusion protein expression was induced in E. coli BL21-CodonPlus (DE3)-RIPL competent cells (Agilent, Santa Clara, CA) with 0.1 mM IPTG at 37°C for 2 hr in LB medium. Cells were sonicated in lysis buffer (25 mM phosphate pH 6.8, 250 mM NaCl, 2 mM DTT, and protease inhibitors) prior to the addition of 1/10 vol 10% [v/v] Triton X-100. Lysates were incubated for 10 min at 4°C then centrifuged at 20,000 × g for 10 min. MagneGST particles (1.5 μl/reaction; Promega, Madison, WI) were blocked with 1% [w/v] BSA diluted in lysis buffer for 30 min at room temperature then incubated for 1 hr at room temperature with saturating amounts of cleared cell lysates. Following 3 washes in 25 mM phosphate pH 6.8, 250 mM NaCl,1% [v/v] TWEEN 20, and protease inhibitors, the particles were resuspended in 200 μl containing 84.5 nM Ofd1 or 84.5 nM Nro1 diluted in lysis buffer plus 1% [v/v] Triton X-100 for 1 hr at 4°C. Ofd1 and Nro1 were purified under native conditions as previously described (Yeh et al., 2011). Particles were washed twice in lysis buffer plus 1% [v/v] TWEEN 20 (no DTT) and once in 50 mM Tris-HCl pH 7.5, 100 mM NaCl prior to elution with 10 mM reduced glutathione in 50 mM Tris-HCl pH 7.5, 100 mM NaCl for 30 min at room temperature. Eluates were analyzed by immunoblotting.

Structures were downloaded from the PDB (Berman et al., 2003) and analyzed using the PyMOL Molecular Graphics System (Schrödinger, LLC). Pairwise and multiple sequence alignments were generated using EMBOSS- WATER and T-coffee respectively (Rice et al., 2000; Notredame et al., 2000).

Ofd1 and Nro1 immunopurifications (IPs) were performed as previously described (Lee et al., 2009) with modifications. Briefly, 1 × 10⁸ cells (Nro1 IP) or 2 × 10⁸ cells (Ofd1 IP) were pelleted, washed in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄, pH 7.4 plus protease inhibitors) and resuspended in 1 ml PBS plus 2 mM DSP (Pierce, Waltham, MA) or vehicle (8% DMSO unless stated otherwise) for 5 min to crosslink proteins. The reaction was quenched by addition of 1 M Tris-HCl pH 7.5 to a final concentration of 20 mM Tris. Ribosome-cleared lysates were generated as described above and 1.0–1.5 mg protein in 600 μl volume was incubated with affinity-purified antibodies (8 ng antibody/μg protein) and 30 μl Protein A agarose beads (Repligen, Waltham, MA) for 2 hr at 4°C. Beads were washed three times and resuspended in SDS lysis buffer (10 mM Tris-HCl pH 6.8, 100 mM NaCl, 1% [w/v] SDS, 1 mM EDTA, 1 mM EGTA) plus protease inhibitors prior to boiling (95°C, 5 min) and immunoblotting.

Strains were cultured for ≥15 generations in SILAC medium (Edinburgh minimal medium plus 75 mg/l each of leucine, histidine, adenine, uracil, and 30 mg/l heavy or light lysine) as described (Fröhlich et al., 2013). Pelleted cells were resuspended in lysis buffer (50 mM Hepes-HCl pH 7.4, 100 mM NaCl, 1% [v/v] NP-40, 1.5 mM MgCl₂) plus protease inhibitors and 1 mM DTT, and lysed by vortexing with glass beads (0.5 mm). Lysates were cleared by centrifugation at 20,000 × g for 10 min and heavy-labeled or 1:1 mixtures of heavy- and light-labeled proteins were layered over a 10% sucrose cushion (20 mM Tris-HCl pH 7.2, 500 mM KCl, 5 mM MgCl₂, 10% [w/v] sucrose) plus protease inhibitors. Lysates were centrifuged at 41,000 rpm for 6 hr in the SW 41 Ti Rotor (Beckman Coulter) and ribosome pellets were solubilized in 150 μl of 6 M ultrapure urea, 3 M LiCl, 50 mM KCl, and 5 μM BME (pH adjusted to 4.5) overnight at 4°C using micro stir bars. Lysates were cleared by centrifugation at 72,000 rpm for 1 hr in the TLA100 rotor, and proteins were precipitated from the supernatant overnight with the addition of two volumes of 20% [w/v] TCA. Precipitated proteins were washed with 1:1 [v/v] ether:ethanol and air-dried prior to resuspension in 8 M ultrapure urea, 150 mM Tris-HCl pH 8.5. Samples were separated by gel electrophoresis using 16.5% acrylamide gels (Bio-Rad, Hercules, CA) and stained with Super Blue (Protea, Morgantown, WV). 15–20 kDa gel bands were excised and analyzed by LC-MS/MS.

Gel pieces were destained and rehydrated in 150 μl 0.01 µg/µl LysC (Wako Chemicals, Japan) in 25 mM triethylammonium bicarbonate (TEAB) then covered with an additional 150 μl of 25 mM TEAB and digested overnight at 37°C. Peptides were extracted from the gel pieces with 50% [v/v] acetonitrile, 0.1% [v/v] trifluoroacetic acid (TFA) and evaporated to dryness in a speed vac. Samples were rehydrated in 0.1% [v/v] TFA and loaded on Oasis HLB µElution solid phase extraction plates (Waters, Milford, MA) and desalted with two 100 μl aliquots of 0.1% [v/v] TFA followed by 100 μl of 10 mM TEAB. Each sample was then step fractionated under basic conditions with 10%, 25%, and 75% [v/v] acetonitrile in 10 mM TEAB yielding three fractions for each of the samples for LC/MS/MS analysis. These fractions were then dried and brought up in 10 μl of 2% [v/v] acetonitrile, 0.1% [v/v] formic acid.

The LC-MS/MS analysis was performed on a Q-Exactive HF mass spectrometer (Thermo Scientific, Waltham, MA) with a nanoACQUITY nano flow chromatography system (Waters). The samples were trapped at 5 μl/min then eluted onto a 20 cm x 75 µm i.d. C18 column for a 90 min gradient at 300 nl/min. The mass spectrometer settings were 240,000 resolution for MS and 35,000 resolution for MS2 with target of 3e6 for MS and 1e5 for MS2. Maximum injection times were set to 60 and 300 millisec respectively and a normalized collision energy of 28 was used. The data from the three fractions for each sample were combined and searched against the RefSeq2015 Schizosaccharomyces pombe using the Mascot (Matrix Science, London, UK) search engine running through Proteome Discoverer v.1.4 (Thermo Scientific). The precursor mass tolerance was set to 15 ppm and the fragment tolerance was set to 0.03 Da. For the search settings, the following modifications were set as variable: deamidated (N,Q), oxidation (P, M), dioxidation (P), and K8 Heavy (K). The K8 Heavy modification adds a mass of 8.014 Da, and SILAC pairs within four ppm mass were considered for calculating SILAC ratios.

All SILAC data were filtered for unique, high confidence peptide spectrum matches (PSMs). For quantification, missing quan values were replaced with the minimum intensity and single-peak quan channels were used. To quantify Rps23 P62 hydroxylation between SILAC pairs, first the labeling efficiency ($e$) was calculated using the median H/L of 40S protein PSMs ($j=1,\dots ,m$) from the heavy sample alone,$s_j$:

Next, the relative abundance of Rps23 protein in the SILAC pair was calculated using non-P62-containing H/L ratios from Rps23 PSMs ($k=1,\dots ,q$), represented by $r_k$. These values were adjusted for labeling efficiency to give $r{a}_k$: 

P62-containing peptide ratios ($p_l$) were then separated into three groups based on P62 modification state (unmodified, monohydroxylated, and dihydroxylated) and each group was analyzed independently. Ratios were normalized for Rps23 protein level to give ${pn}_l$ using the median of (2):

Normalized values were adjusted for labeling efficiency as described above to give$p{a}_l$ and these ratios were log2 transformed. Significance was tested using Mann-Whitney, in which the $lo{g}_2\left(p{a}_l\right)$ values were compared against a theoretical set ($lo{g}_2\left({sn}_j\right)$) where H = L calculated from $s_j$values corrected for labeling efficiency (${sa}_j$):

To calculate the %H and %L for unmodified, monohydroxylated, and dihydroxylated P62, the fractional heavy ($H_l$) value of$p{n}_l$ was determined and corrected for labeling efficiency:

%H and %L were reported as the median from (5) multiplied by 100, with %L equal to 100 – (%H).

To determine the relative abundance of the unmodified and monohydroxylated forms in nro1Δ cells, we assumed that all Rps23 P62 is unmodified in ofd1Δ cells. This assumption was based on the finding that over 99.9% of the dihydroxylated P62 in the wild-type-ofd1Δ SILAC pair originated from the heavy-labeled wild-type cells (Figure 1F). In addition, analysis of the ofd1Δ alone sample detected only unmodified Rps23 P62. The percent unmodified in nro1Δ cells is therefore approximated by 100 ÷ (H/L), where H/L is the median ratio of unmodified P62 in the ofd1Δ (H) - nro1Δ (L) SILAC pair. From this analysis, we found that unmodified P62 accounts for 32% of Rps23 in nro1Δ ribosomes. We used the same logic to calculate the percentage of dihydroxylated P62 in nro1Δ cells using the wild-type-nro1Δ SILAC pair and assuming wild-type cells contain only dihydroxylated P62. Given conservation of mass and the expectation that P62 exists in one of only three states, we then estimated the relative percentage of monohydroxylated Rps23 P62 in nro1Δ cells as 100 - (% unmod) – (% di), leading to the conclusion that approximately 60% of Rps23 P62 is monohydroxylated.

S. pombe rps23⁺ cDNA was cloned into pMAL-c5X expression vector (NEB, Ipswich, MA), and the plasmid was transformed into BL21-CodonPlus (DE3)-RIPL cells. The transformants were grown to OD600 = 0.6 at 37°C and then induced with 0.4 mM IPTG overnight at 18°C. Cells were harvested by centrifugation and cell pellets were stored at −80°C. Bacterial pellets were resuspended in lysis buffer (20 mM Tris-HCl pH 8, 150 mM NaCl) and lysed by French press. Cleared lysates were loaded onto an amylose resin column (NEB) and MBP-Rps23 was eluted with 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 10 mM maltose, 1 mM EDTA, and 0.5% [v/v] Triton X-100. Fractions containing MBP-Rps23 were pooled and dialyzed in 20 mM Tris-HCl pH 8, 1 mM EDTA, and loaded onto a Mono Q anion exchanger (GE Healthcare Life Sciences). Sample was eluted with 20 mM Tris-HCl pH 8, 1 M NaCl, and purified MBP-Rps23 was collected in the flow-through, concentrated, and stored in 20 mM Tris-HCl pH 8, 150 mM NaCl.

For TMT analysis, reactions were performed in 20 μl volumes with Ofd1 (0.5 μM) and MBP-Rps23 (5 μM) diluted into lysis buffer (20 mM Tris HCl pH 7.5, 150 mM NaCl) with and without Nro1 (5 μM). Ofd1 and Nro1 were natively purified as previously described (Yeh et al., 2011). Reactions contained 20 mM Tris HCl pH 7.0, 0.1% [w/v] BSA, 0.5 mM DTT, 4 mM ascorbate, 0.15 mM FeSO₄, and 0.3 mM 2OG. After incubation at 37°C for 1 hr, the reaction was terminated by the addition of 1.25 μl 2 M HCl and analyzed by LC-MS/MS.

For the initial identification of hydroxylated P62, the reaction volume was scaled up to 50 μl. Ofd1 (5.2 μM) and purified human 6xHis-RPS23 (79 μM) were diluted into the lysis buffer along with 0.1% [w/v] BSA, 0.5 mM DTT, 4 mM ascorbate, 0.15 mM FeSO₄, and 0.3 mM 2OG. The reaction was incubated overnight at 37°C.

Reaction samples were LysC-digested and labeled with TMT reagents (Thermo Fisher Scientific) for 1 hr at room temperature. Samples were step fractionated with four basic, reversed phase fractions of 5%, 15%, 25%, and 75% [v/v] acetonitrile in 10 mM TEAB, and run on a Q Exactive HF mass spectrometer (Thermo Scientific). The mass spectrometer settings were 120,000 resolution for MS and 60,000 resolution for MS2 over a 90 min gradient, targeting 3e6 ions for MS and 1e5 for MS2. Maximum injection times were 100 millisec for MS and 200 millisec for MS2. A stepped normalized collision energy 30/35 was used for fragmentation. The isolation window was set to 1.2 Da with a 0.5 offset, and the instrument was set to fragment the top 15 peptides by data dependent analysis with a dynamic exclusion of 10 s. An inclusion list (masses in Da: 457.95331, 458.28107, 463.28510, 463.61322, 468.61679, 468.94397) was inserted to give preference to the QPNSAIRK peptide over the highest abundance peptides. The masses in the inclusion list represented the various modification states of P62 (unmodified, monohydroxylated, and dihydroxylated), as well as the TMT label and deamidated N (+0.98 Da). To increase the signal for the dihydroxylated P62 peptide, the 75% fraction was re-run with the settings ‘Do Not Pick Others.’

Data was searched against the RefSeq2015 database Schizosaccharomyces pombe using the Mascot (Matrix Science) search engine running through Proteome Discoverer v.1.4 (Thermo Scientific) with one missed cleavage allowed and a tolerance of 10 ppm MS and 0.03 Da MS2.

In order to compare the relative abundance of Rps23 P62 hydroxylation in the absence and presence of Nro1, first the relative abundance of Rps23 protein between the two samples was calculated using non-P62-containing peptides. PSMs with unique QuanResultIDs were used for quantification. The intensities from each PSM ($j=1,\dots ,m$), corresponding to the samples with and without Nro1, represented by $s_1$ and $s_2$ respectively, were summed and divided by one another to give a correction factor ($f$) for Rps23 protein levels:

Next, the P62-containing PSMs ($k=1,\dots ,q$) ($p_{1,k}$ and $p_{2,k}$) were separated into three groups based on P62 modification status (unmodified, monohydroxylated, and dihydroxylated) and each group was analyzed independently. The ratios $\raisebox{1ex}{$p_{1,k}$}\!\left/ \!\raisebox{-1ex}{$p_{2,k}$}\right.$ were normalized to Rps23 protein abundance and log2 transformed:

Finally, the Wilcoxon signed rank test was used to determine if the median of these ratios differed significantly from zero, the null hypothesis.

Indicated strains were cultured in rich medium to 4–7 × 10⁶ cells/ml. When indicated, cells were incubated with 92.5 nM Leptomycin B (Cell Signaling, Danvers, MA) and vehicle (0.5% [v/v] ethanol) in rich medium for 1 hr. For live-cell imaging, 5 × 10⁶ cells were pelleted, resuspended in 15 μl rich medium, and 1 μl spotted on untreated glass slides immediately prior to imaging. Cells were imaged on a Zeiss Axio Imager M2 upright fluorescence microscope (Oberkochen, Germany). Images were captured using a Hamamatsu ORCA-ER digital camera (Japan) and iVision software (BioVision, Milpitas, California), and brightness/contrast settings were adjusted using ImageJ such that the range was identical across the strains under comparison.

Yeast strains were cultured and processed as previously described (Hughes et al., 2005) with hypoxic conditions maintained using an In vivo₂ 400 workstation (Biotrace, Inc, Leeds, UK). Briefly, cells grown to 1 × 10⁷ cells/ml were resuspended in oxygenated or deoxygenated rich medium at 5 × 10⁶ cells/ml for 3 hr then harvested and flash froze in liquid nitrogen. Cell pellets were lysed in 27 mM NaOH, 1% [v/v] 2-mercaptoethanol followed by TCA precipitation. Protein pellets were resuspended in SDS lysis buffer, and 20–100 μg protein was loaded for gel electrophoresis and immunoblotting. For alkaline phosphatase (Roche, Basel, Switzerland) treatment, 20 μg protein was diluted in 50 mM Tris-HCl pH 8.5 and incubated at 37°C for 1 hr prior to gel loading.

Immunoblotting was performed as described in Hughes et al. (2005) with modifications. Unless indicated otherwise, samples were processed as described for low oxygen assays. Following electrophoresis, gels were transferred to nitrocellulose membranes using Trans-Blot Turbo transfer system (Bio-Rad). Membranes were blocked with 5% [w/v] milk in PBS-T (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄, 0.05%(v/v) Tween 20, pH 7.4) then incubated in primary antibody followed by an appropriate secondary (IRDye800CW or IRDye680RD mouse or rabbit IgG) and scanned using LI-COR Odyssey CLx imaging system. For immunopurification samples, IRDye 680RD Detection Reagent was added to primary antibody incubation.

Strains grown on rich medium agar for two days were diluted in water to 1.6 × 10⁶ cells/ml and 3 μl were spotted on plates containing rich medium and rich medium supplemented with 1.6 mM CoCl₂. Plates were incubated at 30°C for the indicated time. For random spore analysis, strains were mated for two days on malt-extract plates at room temperature prior to glusulase treatment (0.5% [v/v]) for 16 hr. Spores were then diluted to 1.6 × 10⁶ cells/ml and spotted on rich medium or rich medium plus nourseothricin (100 µg/ml) and G418 (100 µg/ml).

Total RNA preparation from S. pombe and RT-qPCR analysis have been described previously (Shao and Espenshade, 2014). Briefly, total RNA was isolated using RNA STAT-60 (amsbio, Cambridge, MA). cDNA was synthesized using oligo d(T)₂₃VN primers (NEB). The tested genes were quantified by real-time PCR using SYBR Green qPCR master mix (Promega). tub1⁺ served as the internal control to calculate the relative expression across different samples. Error bars represent ± SEM of fold changes from three biological replicates.

Human RPS23 cDNA was cloned into the pProEX HTb bacterial expression vector (Invitrogen) and protein expression was induced in E. coli BL21-CodonPlus (DE3)-RIPL competent cells (Agilent) with 0.6 mM IPTG at 37°C for 2 hr in LB medium. Cells were pelleted and solubilized in lysis buffer (6 M Guanidine HCl, 0.5 M NaCl, and 20 mM Tris-HCl pH 8.0) (10 ml/L bacterial culture) at room temperature for 2 hr. Lysates were cleared by centrifugation in a JA20 rotor (Beckman Coulter) at 16,000 rpm for 40 min and incubated with Ni-NTA agarose (1 ml resin/10 ml lysate; Qiagen, Hilden, Germany) on a rotator for 2 hr at room temperature. Lysates plus resin were transferred to columns and washed 3x with lysis buffer. 6xHis-RPS23 was eluted 3x with lysis buffer adjusted to pH 3.5 and dialyzed overnight into 25 mM phosphate pH 6.8, 0.25 M NaCl, 2 mM DTT, and 1 mM PMSF.

In vitro hydroxylation reaction with Ofd1 and 6xHis-RPS23 was digested with 1 μg trypsin (0.2 μg/μl in 50 mM acetic acid) for 2 hr at 37°C and quenched with the addition of TFA to a final concentration of 0.1% [v/v]. Peptides were analyzed using LTQ Orbitrap Velos MS (Thermo Fisher Scientific) and parent masses (393.22 Da and 401.21 Da) were selected by SIM scan. Peptides were searched against the RefSeq 2012 database Homo sapiens using the Mascot V2.2.6 (Matrix Science) search engine running through Proteome Discoverer v1.3 (Thermo Scientific) with fragment match tolerance set to 0.03 Da.

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

We thank the members of the Espenshade lab for their advice and acknowledge the Duke Screening Center for performing the yeast two-hybrid and Bob O’Meally from the JHMI Mass Spectrometry Core for analyzing the SILAC and TMT samples.

1. Received: May 12, 2017
2. Accepted: October 19, 2017
3. Accepted Manuscript published: October 30, 2017 (version 1)
4. Accepted Manuscript updated: November 1, 2017 (version 2)
5. Version of Record published: November 16, 2017 (version 3)

© 2017, Clasen 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.