The G1 Cyclin-dependent Kinase CRK1 in Trypanosoma brucei Regulates Anterograde Protein Transport by Phosphorylating the COPII Subunit Sec31*

Transport of secretory proteins from the endoplasmic reticulum to the Golgi is mediated by the coat protein II (COPII) complex comprising a Sec23-Sec24 heterodimer and a Sec13-Sec31 heterotetramer. The mechanisms underlying COPII-mediated protein trafficking have been well defined, but the extent of regulation of this secretory machinery by cellular signaling pathways remains poorly understood. Here, we report that CRK1, a G1 cyclin-dependent kinase in Trypanosoma brucei, regulates anterograde protein trafficking by phosphorylating Sec31. Depletion of CRK1 abolished anterograde transport of the secretory protein and disrupted the localization of multiple Golgi proteins, reminiscent of Sec31 depletion. CRK1 phosphorylates Sec31 at multiple serine/threonine sites, and mutation of these phosphosites to alanine recapitulates the protein trafficking defects caused by Sec31 depletion. Mutation of these CRK1 phosphosites to aspartate restored Sec31 function. Taken together, these results uncover a novel function of CRK1 in anterograde protein trafficking and elucidate the mechanistic role of CRK1 in protein trafficking through regulation of the COPII subunit Sec31.

Protein phosphorylation has been reported to play important roles in regulating a variety of membrane trafficking processes, including the regulation of Golgi fragmentation by Cdc2-mediated phosphorylation of the cis-Golgi matrix protein GM130 during mitosis (17), the regulation of ERES disassembly by Cdc2-mediated phosphorylation of the AAA-ATPase p97 cofactor p47 during mitosis (18), the modulation of secretory cargo transport by the PCTAIRE kinase and its potential regulation of Sec23 (19), and the regulation of ER-to-Golgi trafficking by Akt kinase-mediated phosphorylation of Sec24 (20) and by casein kinase II (CK2)-mediated phosphorylation of Sec31 (21). These findings suggest the extensive regulation of protein trafficking by diverse cellular signaling pathways.
Cyclin-dependent kinases (CDKs) are crucial regulators of the cell division cycle in eukaryotes and are activated by their cyclin partners whose levels oscillate during the cell cycle. In the early branching parasitic protozoan Trypanosoma brucei, 11 CDK-related kinases, designated CRK1-CRK4 and CRK6 -CRK12, have been identified (22,23). CRK1 and CRK3 are involved in controlling the G 1 /S transition and the G 2 /M transition, respectively (24 -26). CRK1 associates with four PHO80-like cyclins, CYC2, CYC4, CYC5, and CYC7, that appear to play distinct cellular roles in addition to driving the G 1 /S transition in the procyclic form of T. brucei (27). CYC2 and CYC7 cooperate with CRK1 to regulate posterior cytoskeletal morphogenesis, whereas the CYC4-CRK1 complex appears to regulate certain unidentified nuclear processes required for the G 1 /S transition (27). Despite the essential functions of CRK1, however, its mechanistic roles remain poorly understood, mainly because its downstream substrates have not been identified.
We have set out to identify the substrates of CRK1 by chemical genetic approach, which utilizes an analog-sensitive CRK1 mutant and an engineered ATP analog to thiophosphorylate native CRK1 substrates in crude cell lysate. This allowed us to identify novel CRK1 substrates, including the Sec31 homolog, which was characterized in detail in this report. We demon-strated that CRK1 co-localizes with Sec31 at the ERES and phosphorylates Sec31 at multiple serine/threonine sites, thereby regulating anterograde protein transport. These findings elucidated the mechanistic role of CRK1 in ER-to-Golgi protein trafficking through regulating Sec31.

Results
Identification of CRK1 Substrates Using Analog-sensitive CRK1 Mutant and Bulky ATP Analog-To identify the substrates of CRK1, we employed the chemical genetic approach, which utilizes an engineered protein kinase with an enlarged ATP binding pocket to accept bulky ATP analogs (28). The CRK1 analog-sensitive (CRK1as) mutant, generated by mutating the conserved bulky gatekeeper residue phenylalanine 81 to alanine, was capable of utilizing the N 6 -(benzyl)ATP␥S to thiophosphorylate its substrates (Fig. 1, A and B), as confirmed by in vitro kinase assay with purified CRK1as in complex with CYC2 and with recombinant histone H1 as the substrate (Fig.  1C). Thiophosphorylated histone H1 was alkylated with p-nitrobenzyl mesylate (PNBM), which yields a specific thiophosphate ester (29), and then detected by Western blotting with an anti-thiophosphate ester antibody (␣-thioP). In contrast, the wild-type CRK1 was not able to utilize the ATP analog (Fig. 1C). It should be noted that CRK1as appeared to be more active than wild-type CRK1 in the presence of ATP␥S (Fig. 1C), presumably because enlargement of the ATP binding pocket enabled more efficient access of ATP␥S.
To examine whether the purified CRK1as-CYC2 complex was able to thiophosphorylate native trypanosome proteins, in vitro kinase assays were carried out by incubating recombinant CRK1as with crude trypanosome cell lysate in the presence of N 6 -(benzyl)ATP␥S. To compete off nonspecific use of N 6 -(benzyl)ATP␥S by metabolic enzymes and nucleotide exchange factors, GTP was added to the reaction. The results showed that numerous trypanosome proteins were thiophosphorylated by CRK1as (Fig. 1D). With increased concentration of GTP, nonspecific background signal was significantly decreased, and specific labeling of proteins by CRK1as was increased (Fig. 1D). These results validated the chemical genetic approach for CRK1 substrate labeling and identification.
Purification of thiophosphopeptides for mass spectrometric analysis was carried out essentially as described (30,31). The total trypanosome proteins that had been incubated with N 6 -(benzyl)ATP␥S in the presence (experimental sample) or absence (control sample) of the CRK1as-CYC2 complex were digested to peptides with trypsin, and the products were incubated with iodoacetyl-agarose, which allows the thio-containing groups to react to form covalent bonds. Thiophosphopeptides were liberated by oxidation-based hydrolysis of the sulfur-phosphorus bond, resulting in the replacement of the thiophosphoryl sulfur atom by oxygen (30). Purified phosphopeptides were then analyzed by LC-MS/MS and searched against the T. brucei proteome database. The proteins thus identified were compared between the experimental sample and the control sample, and 17 proteins were only present in the experimental sample but not the control sample. Ten out of the 17 proteins contain the consensus CDK phosphorylation sequence Ser*/Thr*-Pro (the asterisks indicate the site of phos-FIGURE 1. Identification of CRK1 substrates by chemical genetic approach. A, procedures of the chemical genetic approach used for CRK1 substrate detection and identification. Thiophosphorylated CRK1 substrates were either detected by Western blotting after PNBM alkylation, which yields thiophosphate ester and can be detected by a thiophosphate ester-specific rabbit monoclonal antibody (anti-thioP), or trypsin-digested, affinity-purified, and analyzed by LC-MS/MS. B, alignment of the sequence surrounding the gatekeeper residue of CRK1 with its orthologs from human, budding yeast, and fission yeast. The gatekeeper residue is highlighted in red. C, in vitro kinase assays to confirm that the CRK1as, but not the wild-type CRK1, is capable of utilizing the bulky ATP analog to thiophosphorylate histone H1. CRK1 (wild-type and analog-sensitive mutant) and CYC2 were co-expressed in E. coli and co-purified. Thiophosphorylated histone H1 was detected by Western blotting with a thiophosphate ester monoclonal antibody (␣-thioP). CRK1-His 6 and CRK1as-His 6 were detected with anti-His antibody, whereas histone H1 was detected by Coomassie Blue staining. D, in vitro thiophosphorylation of T. brucei total proteins by CRK1as. Crude trypanosome cell lysate was incubated with purified CRK1-CYC2 complex and CRK1as-CYC2 complex in the presence of N 6 -(benzyl)ATP␥S. GTP (0.5 or 3 mM) was added to the kinase reaction to act as a competitor of nonspecific labeling. Thiophosphorylated proteins were detected by Western blotting with ␣-thioP antibody. Total T. brucei proteins used for in vitro kinase assay were stained with Coomassie Blue. IB, immunoblot. phorylation), suggesting that they are potential CRK1 substrates. Of the 10 proteins, seven are hypothetical proteins of unknown function, two are protein translation initiation factors (eIF4E4 and Polyadenylate-binding protein 2), and one is Sec31, a subunit of the COPII coat protein complex (Table 1). Sec31 was of particular interest and thus was further characterized in this study.
Sec31 Is an in Vivo Substrate of CRK1-Mass spectrometry identified Thr-442 in Sec31 as an in vitro CRK1 phosphosite (Table 1), but it is unclear whether this site is phosphorylated in vivo and whether Sec31 possesses additional phosphosites. To identify the in vivo phosphorylation sites in Sec31, we tagged Sec31 with a triple HA epitope at one of its endogenous loci and then immunoprecipitated Sec31-3HA for mass spectrometric analysis. This identified eight phosphosites, Thr-21, Thr-442, Thr-695, Thr-960, Thr-970, Ser-1013, Ser-1014, and Ser-1016. Except for Ser-1013, the other seven phosphosites are each followed by a proline residue, which are characteristic CDK phosphorylation sites. To test whether Sec31 is phosphorylated by CRK1 on the seven CDK consensus phosphosites, we carried out in vitro kinase assays with purified wild-type Sec31 and the Sec31-7A mutant, in which all of the seven putative CRK1 phosphosites were mutated to alanine ( Fig. 2A), and we found that immunoprecipitated CRK1 was capable of phosphorylat-ing Sec31 but not Sec31-7A (Fig. 2B), suggesting that Sec31 is an in vitro substrate of CRK1. To further confirm the in vivo phosphorylation of these seven CDK phosphosites, we overexpressed wild-type Sec31 and the Sec31-7A mutant in trypanosomes. 3HA-tagged Sec31 and Sec31-7A were immunoprecipitated with anti-HA antibody and then immunoblotted with anti-phosphothreonine-proline (␣-Thr(P)-Pro) antibody, which detects the phosphorylated threonine and serine residues followed by a proline. The results showed that Sec31, but not Sec31-7A, was detected by anti-Thr(P)-Pro antibody (Fig. 2C), further confirming that these sites are phosphorylated in vivo. Finally, to examine whether CRK1 is responsible for Sec31 phosphorylation in trypanosomes, 3HA-tagged Sec31 was expressed from its endogenous locus in the CRK1 RNAi cell line, immunoprecipitated, and then immunoblotted with anti-Thr(P)-Pro antibody. Depletion of CRK1 caused reduced phosphorylation of Sec31 in trypanosome cells (Fig. 2D). The lack of complete elimination of Sec31 phosphorylation by CRK1 depletion could be due to the incomplete knockdown of CRK1 by RNAi. Nevertheless, this result confirms that Sec31 is an in vivo substrate of CRK1.
Subcellular Localization of Sec31 and Its Co-localization with CRK1-To determine the subcellular localization of Sec31, we tagged the endogenous Sec31 with EYFP at its C terminus in FIGURE 2. Sec31 is phosphorylated by CRK1 at multiple serine and threonine residues. A, schematic drawing of the Sec31 protein, showing the seven serine and threonine residues that are phosphorylated in vivo and are putative CRK1 phosphosites. B, phosphorylation of Sec31 by CRK1 in vitro. Sec31 and Sec31-7A were purified as GST fusion protein from E. coli. CRK1-PTP was immunoprecipitated from trypanosome cell lysate. Phosphorylated GST-Sec31 (p-GST-Sec31) was detected by Western blotting with anti-phosphothreonine-proline (␣-pT-P) antibody. GST-Sec31 was detected with anti-GST antibody, whereas CRK1-PTP was detected with anti-protein A antibody (␣-ProtA). C, confirmation of in vivo Sec31 phosphorylation by Western blotting with antiphosphothreonine-proline antibody. Sec31 and Sec31-7A were overexpressed in T. brucei, immunoprecipitated (IP) with anti-HA antibody, and then immunoblotted (IB) with anti-phosphothreonine-proline (␣-pT-P) antibody to detect the phosphorylated serine and threonine residues followed by a proline in Sec31-3HA and with anti-HA antibody to detect 3HA-tagged Sec31 and Sec31-7A. D, depletion of CRK1 caused dephosphorylation of Sec31 in T. brucei. Sec31 was endogenously tagged with a C-terminal triple HA epitope in the T. brucei cell line harboring the CRK1 RNAi construct. RNAi was induced for 2 days. Sec31-3HA was immunoprecipitated from non-induced control and CRK1 RNAi cells, and immunoblotted with anti-phosphothreonine-proline (␣-pT-P) antibody to detect phosphorylated Sec31-3HA and with anti-HA antibody to detect Sec31-3HA. The wild-type 29-13 cell line served as a mock control. trypanosome cells expressing mCherry-tagged Sec13, which has previously been shown to localize to the ERES (32). Fluorescence microscopy showed that Sec31-EYFP co-localizes with Sec13-mCherry (Fig. 3A). We also examined the localization of Sec31 relative to the Golgi, and the results showed that Sec31-EYFP was juxtaposed to the Golgi, which was labeled by the Golgi stack marker TbGRASP (Golgi reassembly stacking protein) (Fig. 3B). These results confirmed that Sec31 localizes to the ERES.
We next investigated the co-localization of Sec31 with CRK1. To this end, CRK1 was endogenously tagged with a PTP epitope, and Sec31 was endogenously tagged with a triple HA epitope in the same cell line. Immunofluorescence microscopy showed that CRK1-PTP was detected in many punctate dots in the cytosol (Fig. 3C), as reported previously (27), and it co-localized with Sec31-HA at the ERES (Fig. 3C).
Sec31 Is Required for Cell Division, Cell Morphogenesis, and Golgi Protein Localization-To understand the function of Sec31 in T. brucei, we carried out RNAi in the procyclic form of T. brucei. Knockdown of Sec31 by RNAi was confirmed by quantitative RT-PCR, which showed that the mRNA level of Sec31 was reduced ϳ75% after RNAi induction for 24 h (Fig.  4A), and by Western blotting and immunofluorescence microscopy, which showed that Sec31 protein was decreased to an undetectable level after RNAi induction for 48 h (Fig. 4, B and C). This depletion of Sec31 caused severe growth inhibition and eventual cell death after RNAi induction for 3 days (Fig. 4D), suggesting that Sec31 is essential for cell proliferation and cell viability in the procyclic form.
To further characterize the cell proliferation defect, flow cytometry was performed, which detected a significant increase of cells with 4C DNA content after Sec31 RNAi for 48 h (Fig. 4, E and F), indicating that the Sec31 RNAi cells were arrested at the G 2 /M phase. To identify the specific cell cycle stage that was blocked by Sec31 RNAi, cells were stained with DAPI and counted for the numbers of nuclei and kinetoplasts. The results showed that binucleate (2N2K and 2N1K) cells were increased from ϳ3 to ϳ65% of the total cell population after Sec31 RNAi for 48 h (Fig. 4G), suggesting a cytokinesis defect. Moreover, the accumulation of two nuclei and one kinetoplast (2N1K) cell ( Fig. 4G) indicated that kinetoplast division was also inhibited. In addition to the cell cycle defect, Sec31 RNAi also disrupted cell morphology, as ϳ84% of the Sec31 RNAi cells lost typical slender morphology after RNAi induction for 48 h, becoming somewhat round in shape ( Fig. 4, C, H, and I).
Previous work reported that depletion of TbSec24.1, another COPII subunit in T. brucei, disrupted the localization of several Golgi proteins, such as the Golgi stack proteins TbGRASP and TbGolgin63, likely due to the general defect in Golgi structure (33). Depletion of TbSec24.1 also caused an accumulation of the Golgi enzyme TbGntB, an N-acetyl-D-glucosamine (GlcNAc) transferase, in the ER and a slight decrease of TbGntB in the Golgi (33). Given that Sec31 is another key component of the COPII complex, we reasoned that Sec31 RNAi might cause a similar defect in the localization of these Golgi proteins. To test this possibility, we tagged TbGntB with a triple HA epitope in the Sec31 RNAi cell line. Immunofluorescence microscopy showed that TbGRASP and TbGntB co-localize to the Golgi in control cells, and depletion of Sec31 impaired the localization of both proteins in most of the cells (Fig. 4, I and J). The protein levels of TbGRASP and TbGntB were not altered in Sec31 RNAi cells (Fig. 4K). Together, these results suggest that Sec31 is also required for TbGRASP and TbGntB localization. Unfortunately, despite numerous attempts, we were unable to tag TbGolgin63 in Sec31 RNAi cells due to unknown reasons, and thus the effect of Sec31 depletion on TbGolgin63 localization was not assessed. Given that TbGRASP is a Golgi stack protein FIGURE 3. Subcellular localization of Sec31 and its co-localization with CRK1. A, Sec31 co-localizes with Sec13 in the ERES. Cells co-expressing endogenously EYFP-tagged Sec31 and mCherry-tagged Sec13 were fixed, stained with DAPI, and examined under a fluorescence microscope. B, Sec31 was juxtaposed to the Golgi. Cells expressing EYFP-tagged Sec31 were immunostained with the anti-TbGRASP antibody, which stains the Golgi stack protein TbGRASP, and counterstained with DAPI for DNA. C, co-localization of Sec31 and CRK1. Cells co-expressing 3HA-tagged Sec31 and PTP-tagged CRK1 were co-immunostained with FITC-conjugated anti-HA mAb and antiprotein A polyclonal antibody to label Sec31-3HA and CRK1-PTP, respectively. Scale bars, 5 m. DIC, differential interference contrast.
required for Golgi stacking in T. brucei (32), the mis-localization of TbGRASP in Sec31 RNAi cells suggests that Golgi structure may be disturbed, as is the case of TbSec24.1 depletion (33).
CRK1 Is Also Required for Golgi Protein Localization-Given that CRK1 phosphorylates Sec31 (Fig. 2), we investigated whether depletion of CRK1 also impairs the localization of TbGRASP to the Golgi. Indeed, depletion of CRK1 by RNAi caused the mis-localization of TbGRASP in more than 80% of the cells after CRK1 RNAi induction for 3 days (Fig. 5, A and B), suggesting that CRK1 is required for TbGRASP localization. CRK1 is known to associate with four PHO80-like cyclins, CYC2, CYC4, CYC5, and CYC7 (27). However, only the deple-tion of CYC2 or CYC7 generated cells with an elongated posterior, as the depletion of CRK1 (27), suggesting that CYC2 and CYC7 play specialized functions with CRK1. Therefore, we examined whether depletion of CYC2 and CYC7 also affected TbGRASP localization, and the results showed that knockdown of CYC2 or CYC7 alone did not affect TbGRASP localization, but depletion of CYC2 and CYC7 simultaneously disrupted TbGRASP localization in ϳ65% of the cells after RNAi induction for 3 days (Fig. 5, A and B). Depletion of CRK1 or CYC2 and CYC7 did not alter TbGRASP protein levels (Fig. 5C). These results further confirmed that CYC2 and CYC7 play redundant roles and suggest that the two cyclins function together with CRK1 to regulate COPII-mediated protein transport. We next investigated the effect of CRK1 depletion on the localization of TbGntB and TbGolgin63. TbGntB and TbGol-gin63 were each tagged at their respective endogenous locus in CRK1 RNAi cell line. Immunofluorescence microscopy showed that the localizations of both TbGntB and TbGolgin63 were disrupted in CRK1 RNAi-induced cells (Fig. 5, D and E). The levels of both proteins were not changed upon CRK1 RNAi (Fig.  5, F and G). Given that CRK1 RNAi impaired the localization of the Golgi stack proteins TbGRASP and TbGolgin63 and the Golgi enzyme TbGntB, these results suggest that CRK1 depletion may cause a general defect in the Golgi structure.
CRK1 Is Not Required for Sec31 Localization and Sec31-Sec13 Complex Formation-To investigate whether CRK1-mediated phosphorylation of Sec31 is required for COPII complex formation, we carried out co-immunofluorescence microscopy and co-immunoprecipitation experiments to examine the colocalization and the in vivo interaction of Sec31 and Sec13, which were endogenously tagged with a PTP epitope and a triple HA epitope, respectively, in the CRK1 RNAi cell line. The results showed that Sec31-PTP and Sec13-3HA still co-localized in CRK1 RNAi cells (Fig. 6A) and were co-precipitated by immunoprecipitation from the CRK1 RNAi cell lysate (Fig. 6B). These results suggest that phosphorylation of Sec31 by CRK1 is not involved in Sec31 localization and Sec31-Sec13 complex formation.

Phosphorylation of Sec31 by CRK1 Is Essential for Sec31
Function-To investigate whether CRK1-mediated phosphorylation on Sec31 is required for the cellular function of Sec31, we mutated all seven CRK1 phosphosites on Sec31 to alanine (Sec31-7A) or aspartate (Sec31-7D) for expression of phosphodeficient and phosphomimic mutants in trypanosomes. Wildtype Sec31 and the two Sec31 mutants were each tagged with a C-terminal triple HA epitope and overexpressed in the procyclic form of T. brucei under a tetracycline-inducible promoter, which was confirmed by Western blotting with the anti-HA antibody (Fig. 7A). Overexpression of Sec31-7A, but not Sec31, caused severe growth inhibition and cell death after tetracycline induction for 3 days (Fig. 7B), suggesting a dominantnegative effect caused by Sec31-7A overexpression. Overexpression of Sec31-7D caused slight growth defects (Fig. 7B), suggesting that the phosphomimic mutation partially restored the function of Sec31.
To further characterize the defects caused by Sec31-7A overexpression, we analyzed which cell cycle stage was affected by counting the cells with different numbers of nuclei and kinetoplasts. The three overexpression cell lines (Sec31, Sec31-7A, and Sec31-7D) were thus compared. Overexpression of Sec31 exerted little effect on the cell cycle progression, but overexpression of Sec31-7A resulted in an accumulation of 2N2K and 2N1K cells to ϳ40 and ϳ30% of the total cell population, respectively (Fig. 7C), and disrupted cell morphology, with ϳ90% of the cells becoming rounded up after tetracycline induction for 48 h (Fig. 7D). In all of these round-shaped cells caused by Sec31-7A overexpression, TbGRASP localization was impaired (Fig. 7, E and F), although the protein level of TbGRASP was not affected (Fig. 7G). Similarly, overexpression of Sec31-7A also disrupted TbGntB localization (Fig. 7H) but did not alter the TbGntB protein level (Fig. 7I). The defects caused by Sec31-7A overexpression mimic that by Sec31 RNAi (Fig. 4). In contrast, overexpression of Sec31-7D only resulted in an increase of 2N1K cells to ϳ6% of the total cell population (Fig. 7C) and an accumulation of round-shaped cells to ϳ14% of the total population (Fig. 7D), which is consistent with the fact that overexpression of Sec31-7D caused a slight growth defect (Fig. 7B). Taken together, these results suggest that phosphorylation of Sec31 at the seven Ser/Thr residues by CRK1 is essential for Sec31 function.
Phosphorylation of Sec31 by CRK1 Is Required for Anterograde Protein Trafficking-We next investigated whether Sec31 is required for anterograde trafficking of proteins, which are transported from the ER, via the Golgi, to the cell surface. To this end, we monitored the secretion of the secretory reporter protein BiPNAVRG-HA9 in control and Sec31 RNAi cells. BiPNAVRG-HA9 is an artificial protein consisting of 415 amino acids of the TbBiP ATPase domain fused with the C-terminal hexapeptide QPAVRG and a 9ϫHA epitope. Because of the lack of the C-terminal ER retention sequence MDDL, BiPNAVRG-HA9 is not retained in the ER but is secreted into the culture medium (34). BiPNAVRG-HA9 was stably expressed in cells harboring Sec31 RNAi construct, and RNAi was induced for 24 h. The control and Sec31 RNAi cells were first washed with fresh medium and then treated with cycloheximide for up to 4 h. BiPNAVRG-HA9 was immunoprecipitated with anti-HA antibody from cell lysates and media that were collected at different time points of cycloheximide treatment and then detected by Western blotting with anti-HA antibody. In the non-induced control cells, BiPNAVRG-HA9 was gradually secreted from the cells into the medium, with the half-time (t1 ⁄2 ) estimated to be around 90 -100 min (Fig. 8A). Strikingly, depletion of Sec31 by RNAi completely blocked BiPNAVRG-HA9 secretion into the medium (Fig. 8A), suggesting that Sec31 is required for anterograde protein trafficking in T. brucei.
We next investigated whether CRK1 is also required for anterograde protein trafficking by monitoring the secretion of BiPNAVRG-HA9. Similar to Sec31 RNAi, knockdown of CRK1 by RNAi also completely blocked the secretion of BiPNAVRG-HA9 into the medium (Fig. 8B). Furthermore, to examine whether CRK1-mediated phosphorylation of Sec31 is required for its function in anterograde protein transport, BiPNAVRG-HA9 was stably expressed in cells overexpressing Sec31, Sec31-7A, or Sec31-7D. In Sec31-overexpressing cells and Sec31-7Doverexpressing cells, BiPNAVRG-HA9 was secreted into the medium, albeit the latter had a slower rate than the former, with the half-time (t1 ⁄ 2 ) of BiPNAVRG-HA9 export estimated to be about 220 -230 min in Sec31-7D-overexpressing cells and about 90 -100 min in Sec31-overexpressing cells (Fig. 8C). However, in Sec31-7A-overexpressing cells, secretion of BiPNAVRG-HA9 into the medium was completely blocked (Fig. 8C). These results suggest that phosphorylation of Sec31 at the seven Ser/Thr sites by CRK1 is essential for its function in anterograde protein trafficking.

Discussion
In this report, we applied the chemical genetic approach to screen for substrates of CRK1, a crucial cyclin-dependent kinase involved in the G 1 /S transition in T. brucei, and we identified Sec31 as a novel substrate of CRK1. Several lines of evidence suggest that Sec31 is a bona fide CRK1 substrate. First, it was specifically thiophosphorylated in the crude cell lysate by CRK1as in the presence of N 6 -(benzyl)ATP␥S ( Fig. 1 and Table  1). Second, purified recombinant Sec31 protein was phosphorylated in vitro by CRK1 (Fig. 2B). Third, depletion of CRK1 caused significant de-phosphorylation of Sec31 in trypanosome cells (Fig. 2D). Finally, Sec31 was phosphorylated at multiple FIGURE 6. Depletion of CRK1 does not affect the localization of Sec31 to the ERES and the interaction of Sec31 with Sec13. A, effect of CRK1 depletion on Sec31 localization. Sec31 and Sec13 were endogenously tagged with PTP and a triple HA epitope, respectively, in CRK1 RNAi cell line. CRK1 RNAi was then induced for 3 days, and co-immunofluorescence microscopy was then performed with anti-protein A polyclonal antibody and FITC-conjugated anti-HA monoclonal antibody. Cells were counterstained with DAPI for DNA. B, effect of CRK1 depletion on Sec31-Sec13 complex formation, as measured by co-immunoprecipitation. Sec31 and Sec13 were tagged with a PTP epitope and a triple HA epitope, respectively, in cells containing the CRK1 RNAi construct. CRK1 RNAi was induced for 3 days, and co-immunoprecipitation was carried out in non-induced control (ϪTet) and CRK1 RNAi (ϩTet) cells. Tet, tetracycline. Immunoprecipitation (IP) was performed by incubating the cell lysate with IgG-Sepharose beads, and the immunoprecipitated proteins were separated on SDS-PAGE and immunoblotted (IB) with anti-HA antibody and anti-protein A antibody (anti-ProtA) to detect Sec13-3HA and Sec31-PTP, respectively. As negative controls, non-induced control and CRK1 RNAi cells expressing the 3HA-tagged Sec13 were used. DIC, differential interference contrast.
CDK consensus sites in vivo in trypanosomes (Fig. 2, A and B), and mutation of these phosphosites made it unable to be phosphorylated by CRK1 in vitro (Fig. 2B). Sec31 is the first CRK1 substrate identified and functionally characterized so far.
The ER-to-Golgi secretory pathway in T. brucei shares similarities with that in other model organisms (35). T. brucei expresses the orthologs of all the COPII components, including one Sec13 (32), one Sec31, and two each for Sec23 and Sec24 (33,36). All of these proteins localize to the ERES (32,33,36), which is in close proximity to the Golgi apparatus (Fig. 3B). The two Sec23 homologs and two Sec24 homologs form exclusive heterodimers, TbSec23.1-TbSec24.2 and TbSec23.2-TbSec24.1, and are all essential in the bloodstream form of T. brucei (36). The two Sec23-Sec24 heterodimers appear to be involved in transport of distinct secretory cargos (33,36). Our RNAi data demonstrated that Sec31 is essential in the procyclic form of T. brucei (Fig. 4D) and is required for localization of Golgi proteins TbGRASP and TbGntB and for anterograde transport of the secretion marker BiPNAVRG-HA9 (Figs. 4, I  and J, and 8A), similar to the defects caused by knockdown of TbSec24.1 in the procyclic form (33). These results provided compelling evidence that the T. brucei Sec31 homolog functions in the same secretory pathway as TbSec24.1.
RNAi of Sec31 also arrested cytokinesis and distorted cell morphology (Fig. 4, E-H), suggesting that Sec31 plays additional roles in cell division and cell morphogenesis. The requirement of Sec31 in cell cycle progression was also observed in the fission yeast, in which Sec31 null mutation caused accumulation of bi-nucleate cells (37). Importantly, this cell cycle defect was not attributed to a general defect in protein secretion (37). This notion is also supported by the finding that TbSec24.1 and TbSec24.2 RNAi did not cause cell cycle arrest, despite the protein secretion defect (33). It should be noted that depletion of Sec31 also caused defective kinetoplast duplication/segregation, producing 2N1K cells (Fig. 4G), which was due to inhibition of basal body segregation (data not shown). Although inhibited basal body segregation can lead to cytokinesis arrest and disrupts cell morphology in T. brucei (38), it is unlikely to contribute directly to the defects in cytokinesis and cell morphology in Sec31 RNAi cells, as the Sec31-deficient 2N2K cells were similarly arrested in cytokinesis and lost cell morphology (Fig. 4, C and G). The precise mechanisms under- lying the role(s) of Sec31 in cell division and cell morphogenesis remain to be explored. Given that Sec31 forms the outer layer of the COPII complex with Sec13, it would be interesting to investigate whether depletion of Sec13 causes similar defects in cell division and cell morphology.
CDKs are known to also play important roles in a vast array of cellular processes that, in the traditional viewpoint, are not under cell cycle control. These include transcription, mRNA processing, translation, etc. (39). Our work uncovered a new function for CRK1, a G 1 CDK in T. brucei, in the regulation of COPIImediated anterograde protein trafficking, which is supported by the findings that CRK1 phosphorylates Sec31 in vivo in trypanosomes (Fig. 2) and mutation of CRK1 phosphorylation sites on Sec31 disrupted the function of Sec31 in protein secretion (Fig. 8C) and that depletion of CRK1 abolished protein secretion (Fig. 8B). This new function for CRK1 appears to be independent of its role in the G 1 /S cell cycle control, as the Sec31 phospho-deficient (Sec31-7A) mutant was arrested at cytokinesis but not at the G 1 /S boundary of the cell cycle (Fig. 7C).
The mechanistic role of CRK1-mediated Sec31 phosphorylation is still unclear. Our data showed that CRK1 is neither required for Sec31 localization to ERES nor for Sec31-Sec13 complex formation (Fig. 6). Human Sec31 is phosphorylated by casein kinase II, and this phosphorylation is also not required for Sec31 localization, but the non-phosphorylatable form of Sec31 becomes less dynamic at ERES in HeLa cells (21). Surprisingly, phosphorylation of Sec31 by casein kinase II appears to decrease its binding to Sec13, despite that the depletion of casein kinase II caused ER-to-Golgi trafficking defects (21). It is not clear whether CRK1 phosphorylation of Sec31 alters the dynamics of Sec31 at the ERES, but clearly it does not have any impact on Sec31-Sec13 interaction. Given that the Sec31-Sec13 complex is involved in vesicle fission from the ER membrane (40) and in accommodating large cargo molecules (41), CRK1 phosphorylation might affect Sec31 function in assembly of the outer layer Sec13-Sec31 coat to incorporate larger cargo molecules or in deformation of ER membrane into vesicles, which needs to be further explored.
Sec31 from Saccharomyces cerevisiae was also identified as a substrate of Cdk1 (42), but whether Cdk1-mediated phosphorylation of Sec31 is required for protein trafficking in the budding yeast was not investigated in that study. Cdk1 phosphorylates yeast Sec31 at Ser-836 and Ser-980 (42), both of which do not have equivalent sites in T. brucei Sec31. Conversely, the seven CRK1 phosphosites in T. brucei Sec31 (Fig. 2) also do not have equivalent sites in yeast and human Sec31 proteins. It should be noted that Sec31 homologs from T. brucei, budding yeast, and humans only share an overall sequence identity of ϳ20 -25%. Therefore, the lack of equivalent Cdk1 phosphorylation sites among these Sec31 homologs does not necessarily reflect any functional distinction. It is not known whether human Sec31 is a substrate of CDK1, but a CDK-related PCTAIRE protein kinase interacts with Sec23 and regulates ERto-Golgi trafficking (19). Nevertheless, the identification of Sec31 from S. cerevisiae and T. brucei, two diverged eukaryotes, as a CDK substrate suggests a likely conserved function for CDK in regulating ER-to-Golgi protein trafficking by modulating the function of Sec31 via phosphorylation. RNAi construct (A), the CRK1 RNAi construct (B), or the overexpression construct containing either wild-type Sec31, Sec31-7A, or Sec31-7D (C) were stably transfected with a construct encoding the secretory protein BiPNAVRG-HA9. After induction with tetracycline for 24 h (Sec31 RNAi and overexpression of wild-type and mutant Sec31 proteins) or 48 h (CRK1 RNAi), cells were washed with fresh medium, resuspended in fresh medium, and treated with cycloheximide (CHX) for 4 h. Samples from cells and the medium were collected at the indicated times. BiPNAVRG-HA9 was immunoprecipitated from cell lysate and culture medium by anti-HA antibody, fractionated by SDS-PAGE, and detected by anti-HA antibody. Band intensity was determined using the ImageJ software, plotted against the time of cycloheximide treatment, and presented below the Western blots. Blue lines indicate the BiPNAVRG-HA9 protein detected in cells (non-secreted protein), and red lines indicate the BiPNAVRG-HA9 protein detected in the medium (secreted protein). Error bars indicate S.D. calculated from three independent experiments.

Experimental Procedures
Expression and Purification of Recombinant CRK1, Analogsensitive CRK1, and CYC2 in Escherichia coli-The full-length coding sequence of CYC2 was PCR-amplified and cloned into the pGEX-4T-3 vector for expression of GST-tagged CYC2. Full-length coding sequences of CRK1 and the budding yeast Cak1p, the CDK activating kinase, were PCR-amplified and cloned into pCDF-Duet1 vector (Novagen) for co-expression of His 6 -tagged CRK1 and Cak1p. To generate the analog-sensitive CRK1 mutant, site-directed mutagenesis was performed to mutate Phe-81 to alanine. The pGEX-CYC2 plasmid and the pCDF-CRK1-Cak1p plasmid were transformed into E. coli BL21 Rosetta strain (Novagen). The strain with the transformed plasmids was cultured in LB medium supplemented with ampicillin, spectinomycin, and chloramphenicol.
Expression of recombinant proteins was induced with 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside for 6 h at 27°C. Cells were lysed in BugBuster reagent (Novagen) supplemented with 1 mM ATP and 10 mM MgCl 2 , and the cleared cell lysate was incubated with glutathione-Sepharose 4B beads for 30 min at 4°C. Beads were washed with 60 column volumes of lysis buffer (50 mM Tris-Cl, pH 7.5, 0.3 M NaCl, 0.1% Triton X-100, and 0.01% glycerol) supplemented with 1 mM ATP and 10 mM MgCl 2 . Proteins bound to the beads were eluted with 2 column volumes of elution buffer (50 mM Tris-Cl, pH 7.5, 10 mM reduced glutathione) and concentrated and dialyzed against the storage buffer (50 mM HEPES, pH 7.5, 0.1 M NaCl, 1 mM DTT, 0.1 mM EDTA, 0.01% Tween 20, and 30% glycerol). Purified proteins were analyzed by SDS-PAGE followed by Coomassie Blue staining and Western blotting with anti-GST and anti-His antibodies. Western blotting showed that recombinant CRK1-His 6 (and CRK1as-His 6 ) was co-purified with GST-CYC2, due to binding to the latter.
In Vitro Kinase Assay-For in vitro kinase assay to test whether CRK1as can use N 6 -(benzyl)ATP␥S, the CRK1-CYC2 complex and the CRK1as-CYC2 complex purified above were used, and recombinant histone H1 was used as the substrate (Upstate). 400 ng of purified kinase-cyclin complex was incubated with 1 g of histone H1 in the kinase buffer (50 mM HEPES, pH 7.5, 0.1 M NaCl, 10 mM MgCl 2 ) containing 1 mM ATP␥S or N 6 -(benzyl)ATP␥S for 1 h at room temperature. Kinase reaction was stopped by adding 2 mM EDTA. Subsequently, 2.5 mM PNBM (Epitomics) was added to the kinase reaction and incubated for 1 h at room temperature. Samples were run on an SDS-polyacrylamide gel and transferred onto a PVDF membrane. Thiophosphorylation was detected by Western blotting with a thiophosphate ester rabbit monoclonal antibody (Clone 51-8, Epitomics).
For the in vitro kinase assay to confirm that CRK1 can phosphorylate Sec31, Sec31 was cloned into the pGEX-4T-3 vector. Site-directed mutagenesis was carried out with pGEX-Sec31 vector to mutate all seven CRK1 phosphosites in Sec31 to alanine, thus generating pGEX-Sec31-7A vector. The two plasmids were transformed into E. coli BL21 strain, and recombinant GST-Sec31 and GST-Sec31-7A were purified through binding to glutathione-Sepharose beads. CRK1-PTP, which was immunoprecipitated from T. brucei cell lysate, was mixed with purified GST-Sec31 and GST-Sec31-7A in the kinase buffer (see above) in the presence of 1 mM ATP. Phosphorylated GST-Sec31 was detected by Western blotting with anti-Thr(P)-Pro antibody (Cell Signaling Technologies), which detects phosphorylated serine and threonine followed by a proline.
Identification of CRK1 Substrates Using Analog-sensitive CRK1 and Bulky ATP Analog-Kinase assay and purification of thiophosphopeptides were carried out according to the published protocol (31). T. brucei cells (5 ϫ 10 8 ) were lysed by sonication in RIPA buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40), and cell lysate was cleared by centrifugation. The cleared cell lysate was split into four fractions, and one fraction was incubated with 1 g of purified CRK1as-CYC2 complex or CRK1-CYC2 complex in the presence of N 6 -(benzyl)ATP␥S. To compete off nonspecific labeling, 0.5 or 3 mM GTP was added to the kinase reaction. After incubating for 1 h at room temperature, proteins were digested with trypsin. The peptide mix was incubated with iodoacetyl-agarose beads for 16 h at room temperature. Beads were then washed successively with 80% acetonitrile, 50 mM Tris-Cl, pH 7.5, 50% acetonitrile, 50 mM Tris-Cl, pH 7.5, 50 mM Tris-Cl, pH 7.5, 5 M NaCl, 5% formic acid. Thiophosphopeptides bound to the beads were eluted with 1 mg/ml oxone and then cleaned with a C-18 ZipTips. Peptides were analyzed on an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) interfaced with an Eksigent nano-LC 2D Plus chipLC system (Eksigent Technologies) at the Proteomics Core Facility of the University of California at San Francisco.
Data analysis was performed according to our previous publications (43,44). Raw data files were searched against the T. brucei genome database (version 4) using the Mascot search engine. The search conditions used peptide tolerance of 10 ppm and MS/MS tolerance of 0.8 Da with the enzyme trypsin and two missed cleavages.
RNA Interference-To knock down Sec31 by RNAi, a 500-bp DNA fragment corresponding to the N-terminal coding region of Sec31 was PCR-amplified from the genomic DNA and cloned into the Stem-Loop vector pSL (43). The resulting plasmid, pSL-Sec31, was linearized with NotI and then transfected into the 29-13 cell line by electroporation according to our published procedures. The successful transfectants were selected with 2.5 g/ml phleomycin in addition to 15 g/ml G418 and 50 g/ml hygromycin and then cloned by limiting dilution in a 96-well plate containing SDM-79 medium supplemented with 20% fetal bovine serum and all three antibiotics. RNAi was induced with 1.0 g/ml tetracycline.
In Situ Epitope Tagging of Proteins-For C-terminal epitope tagging of Sec31 at one of its endogenous loci, a 400-bp fragment corresponding to the C-terminal coding region of Sec31 was PCR-amplified from the genomic DNA and cloned into pC-3HA-PAC and pC-EYFP-PAC vectors. The resulting plasmids were linearized and electroporated into the wild-type 427 cell line and the cell line harboring the Sec31 RNAi construct pSL-Sec31. Transfectants were selected with 1 g/ml puromycin and cloned by limiting dilution in a 96-well plate.
For co-localization of Sec31-EYFP and Sec31-mCherry, the cell line harboring the pC-Sec31-PAC vector was transfected with pC-Sec13-mCherry-NEO. For co-localization of Sec31-3HA with CRK1-PTP, the cell line harboring the pC-Sec31-3HA-PAC vector was transfected with pC-CRK1-PTP-NEO. Both transfectants were selected with 40 g/ml G418 in addition to 1 g/ml puromycin and then cloned by limiting dilution as described above. For endogenous tagging of TbGntB and TbGolgin63, the PCR-based epitope tagging method was used according to previous publication (46).
For endogenous tagging of Sec31 and Sec13 with PTP and the triple HA epitope in the CRK1 RNAi cell line, a 500-bp fragment corresponding to the C-terminal coding region of Sec31 was cloned into the pC-PTP-PAC vector and electroporated into the CRK1 RNAi cell line. Transfectants were selected by 1 g/ml puromycin and cloned by limiting dilution. Subsequently, a 500-bp DNA fragment corresponding to the C-terminal coding region of Sec13 was cloned into the pC-3HA-BSD vector, and the resulting construct was electroporated into the CRK1 RNAi cell line harboring the pC-Sec31-PTP-PAC vector. Transfectants were further selected with 10 g/ml blasticidin and cloned by limiting dilution.
Co-immunoprecipitation-Control and CRK1 RNAi-induced cells (3 days) co-expressing endogenously PTP-tagged Sec31 and 3HA-tagged Sec13 or expressing Sec13-3HA alone were harvested by centrifugation, lysed in 1 ml of cell lysis buffer (25 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1 mM DTT, 1% Nonidet P-40, and protease inhibitor mixture), and cleared by centrifugation at the highest speed in a microcentrifuge. The supernatant was then incubated with 50 l of IgG-Sepharose 6 fast flow beads (GE Healthcare) at 4°C for 1 h. Beads were then washed six times with the cell lysis buffer, and bound proteins were eluted with 10% SDS. Eluted proteins were then separated on SDS-PAGE, transferred onto a PVDF membrane, and immunoblotted with anti-HA antibody and anti-protein A antibody to detect Sec13-3HA and Sec31-PTP, respectively.
Overexpression of Phosphodeficient and Phosphomimic Mutants of Sec31-Full-length coding sequence of Sec31 was PCR-amplified and cloned into the pLew100 -3HA vector. Subsequently, site-directed mutagenesis was then performed to mutate all of the seven CRK1 phosphorylation sites on Sec31 to alanine (Sec31-7A) and aspartate (Sec31-7D). The three plasmids were linearized with NotI digestion and electroporated into the 29-13 cell line. Successful transfectants were selected with 2.5 g/ml phleomycin in addition to 15 g/ml G418 and 50 g/ml hygromycin and then cloned by limiting dilution. Overexpression of Sec31 and its mutants were induced by adding 0.1 g/ml tetracycline into the culture medium, and cell growth was monitored by counting the cells every day for 3 days.
Immunofluorescence Microscopy-Cells were washed once with PBS, adhered to the coverslips for 30 min, and fixed and permeabilized with cold methanol (Ϫ20°C) for 30 min. Cells were then rehydrated with PBS for 10 min. Cells were blocked with 3% BSA in PBS for 45 min at room temperature and then incubated with the primary antibody for 1 h at room temperature. The following primary antibodies were used: FITC-conjugated anti-HA antibody (1:400 dilution, Sigma); anti-protein A antibody (1:400 dilution, Sigma); and anti-TbGRASP antibody (1:400 dilution) (32). Cells were washed three times with PBS and then incubated with FITC-conjugated anti-mouse IgG (1:400 dilution, Sigma) or Cy3-conjugated anti-rabbit IgG (1:400 dilution, Sigma) for 1 h at room temperature. Cells on the coverslips were washed three times with PBS, mounted with DAPI-containing VectaShield mounting medium (Vector Laboratories), and imaged using an inverted fluorescence microscope (Olympus IX71) equipped with a cooled CCD camera (model Orca-ER, Hamamatsu) and a PlanApo N 60 ϫ 1.42-NA lens. Images were acquired using the Slidebook 5 software.
Secretion of the BiPNAVRG-HA9 Reporter-The pXS5-BiPNAVRG-HA9 plasmid (34) was linearized and electroporated into cell lines harboring the Sec31 RNAi construct (pSL-Sec31), the CRK1 RNAi construct (pZJM-CRK1), and each of the three overexpression constructs (pLew100-Sec31-3HA, pLew100-Sec31-7A-3HA, and pLew100-Sec31-7D-3HA) for constitutive expression of BiPNAVRG-HA9 fusion protein. For BiPNAVRG-HA9 secretion assays, non-induced control cells and tetracycline-induced cells were washed three times with fresh medium and then incubated with 100 g/ml cycloheximide to inhibit protein synthesis. Time course samples of equal volume (5 ml of cell culture) immediately before (0 h) and after cycloheximide treatment (2 and 4 h) were collected, and cells were centrifuged to separate cell pellet and culture medium. Cell pellets were then lysed in immunoprecipitation buffer (25 mM Tris-Cl, pH 7.6, 100 mM NaCl, 1 mM DTT, 1% Nonidet P-40, and protease inhibitor mixture) of the equal volume (5 ml) of the culture medium. Protease inhibitor mixture was also added to the culture medium to prevent the degradation of secreted proteins. Both the crude cell lysate and the culture medium were incubated with anti-HA antibody for 4 h at 4°C with gentle rotation, and then incubated with protein G-Sepharose beads for 1 h at 4°C. Immunoprecipitates were eluted, separated on SDS-PAGE, transferred onto a PVDF membrane, and immunoblotted with anti-HA antibody to detect BiP-NAVRG-HA9 in both the cells and the medium. Protein band intensity was determined using ImageJ software.
Statistical Analysis-Statistical analysis was performed using the t test provided in the Microsoft Excel software. For immunofluorescence microscopy experiments, images were randomly taken, and all of the cells in each image were counted.