Substrate priming enhances phosphorylation by the budding yeast kinases Kin1 and Kin2

Multisite phosphorylation of proteins is a common mechanism for signal integration and amplification in eukaryotic signaling networks. Proteins are commonly phosphorylated at multiple sites in an ordered manner, whereby phosphorylation by one kinase primes the substrate by generating a recognition motif for a second kinase. Here we show that substrate priming promotes phosphorylation by Saccharomyces cerevisiae Kin1 and Kin2, kinases that regulate cell polarity, exocytosis, and the endoplasmic reticulum (ER) stress response. Kin1/Kin2 phosphorylated substrates within the context of a sequence motif distinct from those of their most closely related kinases. In particular, the rate of phosphorylation of a peptide substrate by Kin1/Kin2 increased >30-fold with incorporation of a phosphoserine residue two residues downstream of the phosphorylation site. Recognition of phosphorylated substrates by Kin1/Kin2 was mediated by a patch of basic residues located in the region of the kinase αC helix. We identified a set of candidate Kin1/Kin2 substrates reported to be dually phosphorylated at sites conforming to the Kin1/Kin2 consensus sequence. One of these proteins, the t-SNARE protein Sec9, was confirmed to be a Kin1/Kin2 substrate both in vitro and in vivo. Sec9 phosphorylation by Kin1 in vitro was enhanced by prior phosphorylation at the +2 position. Recognition of primed substrates was not required for the ability of Kin2 to suppress the growth defect of secretory pathway mutants but was necessary for optimal growth under conditions of ER stress. These results suggest that at least some endogenous protein substrates of Kin1/Kin2 are phosphorylated in a priming-dependent manner.

Given the prevalence of protein phosphorylation as a regulatory mechanism in eukaryotes, it is perhaps not surprising that many proteins are phosphorylated at multiple sites. For example, according to data compiled by the PhosphoSite database, over 5000 human proteins have 10 or more sites of phos-phorylation (1). Hierarchical, or ordered, phosphorylation, frequently contributes to this phenomenon (2,3). A common mechanism for ordered phosphorylation is for a kinase to specifically recognize a phosphorylated residue within its substrates. Such "phosphorecognition" can be mediated by noncatalytic domains in the kinase, such as phosphotyrosinebinding SH2 domains (4) or phosphothreonine-binding forkhead-associated (FHA) domains (5), or by adaptor subunits, such as the cyclin-dependent protein kinase-associated protein Cks1 (6). In addition, there are several kinases that phosphorylate residues in close proximity to prior sites of phosphorylation. The first such kinases to be characterized were glycogen synthase kinase 3 (GSK3) 2 isozymes, which phosphorylate their substrates at sites four residues upstream of a phosphorylated Ser or Thr residue (7). Hierarchical phosphorylation by GSK3 is mediated by specific recognition of phosphorylated amino acids through a patch of basic residues appropriately positioned within the catalytic cleft (8,9). Several other kinases can recognize substrates primed at specific sites either upstream (casein kinase 1 (CK1)) or downstream (CK2, Cdc7, FAM20C) of their own sites of phosphorylation (2,10,11). Although most hierarchical phosphorylation has been ascribed to Ser-Thr kinases, recently several protein Tyr kinases were demonstrated to recognize pTyr-primed substrates (12,13).
Hierarchical phosphorylation can serve a number of potential functions. It can provide a mechanism for signal integration and cross-talk, because multisite phosphorylation of the substrate depends on activation of both priming and priming-dependent kinases. Dual phosphorylation can promote specific protein-protein interactions. For example, ordered phosphorylation of the adaptor protein Shc by the Tyr kinases Src and epidermal growth factor receptor mediates high affinity binding to the SH2 domain of its effector protein Grb2 (13). Dual phosphorylation of proteins within specific consensus sequences generates so-called phosphodegron motifs that are recognized by specific Skp2-cullin-F-box E3 ubiquitin ligase complexes, thus triggering ubiquitin-dependent proteasomal degradation (14). Finally, clustering of phosphorylation sites can have a large impact on local charge, affecting intramolecular interactions, as well as binding to cell membranes or other proteins (15). Such effects can be amplified by "self-priming," in which runs of appropriately spaced sites are phosphorylated in succession (3).
In this study we report that substrate priming promotes phosphorylation by the closely related budding yeast kinases Kin1 and Kin2 (Kin1/2). Kin1/2 are understudied members of a family of Ser-Thr kinases related to AMP-activated protein kinase (AMPK) (16 -18). They are most closely related to the microtubule affinity-regulating kinases (MARKs, also called Par-1) in multicellular eukaryotes, sharing an N-terminal kinase catalytic domain, as well as a C-terminal phospholipid-binding kinaseassociated 1 domain proposed to promote membrane association (19,20). Kin1/2 have been implicated in several cellular processes in budding yeast. Both kinases were identified as high-copy suppressors of secretory pathway mutants, functioning to promote exocytic vesicle fusion with the plasma membrane (21). Kin2 function in the secretory pathway appears to require its localization to the periphery of the growing bud (22). Kin2 overexpression causes defects in septin organization, polar hyperaccumulation of cell wall components, and sensitivity to cell wall disrupting agents (22). Although Saccharomyces cerevisiae kin1⌬ kin2⌬ double mutant cells have no overt cell polarity defect, deletion of the sole ortholog Kin1 in Schizosaccharomyces pombe results in substantial defects in bipolar growth, cell morphology, and organization of the cell wall (23)(24)(25)(26)(27)(28). These observations suggest that roles for the kinases in cell polarity are conserved yet more essential to fission yeast. More recently a role for budding yeast Kin1/2 was established in the ER stress-activated unfolded protein response (UPR) (29). Kin1/2 appear to promote processing of the mRNA encoding the ER stress-responsive transcription factor Hac1, and combined deletion of Kin1/2 sensitizes cells to ER stress agents. Whether the functions of Kin1/2 in the UPR and in polarized growth are mechanistically related remains to be determined.
Although a series of Kin1 substrates involved in cell polarity were recently identified through phosphoproteomics studies of fission yeast (30), to date no substrates of budding yeast Kin1/2 have been found. Here, we exploit the priming dependence of Kin1/2 and other features of its substrate phosphorylation motif to identify Sec9 as a direct substrate of the kinases in vitro and in vivo. By structure-guided mutagenesis, we designed Kin1/2 alleles defective for recognition of phosphorylated substrates, allowing us to examine the importance of substrate priming for their function in vivo. We find that substrate priming enhances Kin2 function in the ER stress response yet is dispensable for its role in promoting exocytosis. These studies suggest that recognition of phosphorylated substrates by Kin1/2 may provide a mechanism to confer robustness to cellular stress by optimizing phosphorylation of key substrates.

Kin1 and Kin2 share an unusual phosphorylation consensus sequence including a phosphorylated residue at the ؉2 position
We recently determined phosphorylation site consensus motifs for 62 S. cerevisiae kinases by positional scanning peptide library (PSPL) analysis (31). This analysis included five AMPK-related kinases (Snf1, Gin4, Hsl1, Kcc4, and Kin1). Most of these kinases shared a common consensus motif similar to that previously described for Snf1 and mammalian AMPK (32). This motif includes strong selectivity for hydrophobic aliphatic residues at the Ϫ5 and ϩ4 positions relative to the phosphorylation site, as well as an Arg residue at the Ϫ3 position (⌽XBXXSXXX⌽, where ⌽ is an aliphatic residue, and B is a basic residue). Interestingly, Kin1 had a distinct motif that lacked selectivity for basic residues at the Ϫ3 position. To determine whether Kin2 shared this unusual motif, we purified the kinase from a yeast overexpression system and performed PSPL analysis (highly selected residues are shown in Table 1, and full quantified data are provided in Table S1). As anticipated based on their high degree of sequence similarity (93% within the catalytic domain), Kin2 had a phosphorylation consensus motif indistinguishable from Kin1. Kin1/2 shared selectivity for an aliphatic hydrophobic residue at the ϩ4 position with other AMPK-related kinases. However, the two kinases had unusual features that diverge from other kinases within the family, including strong selectivity for Asn at the Ϫ2 position. Interestingly, Kin1/2 also appeared to strongly prefer pThr at the ϩ2 position, suggesting that their activity is enhanced by priming phosphorylation of substrates. Notably, other kinases that phosphorylate primed substrates are directed by phosphorylation at other positions (2). Kin1/2 thus appeared to have a unique phosphorylation site motif distinct from previously characterized kinases.
To verify the PSPL results, we synthesized a peptide, termed KINtide, that incorporated the residues selected most strongly at the key Ϫ2 and ϩ4 positions (Asn and Leu, respectively). The KINtide sequence was derived from the Kin1/2 phosphorylation site in Sec9 (described below). In addition, we synthesized a series of KINtide variants in which specific residues were substituted with Ala, as well as an additional peptide incorporating pSer at the ϩ2 position. Kin1 or Kin2 could phosphorylate the unmodified KINtide in vitro, albeit at a modest rate (Fig. 1A). Substitution of either the Ϫ2 Asn or the ϩ4 Leu for Ala reduced the rate of phosphorylation by Kin1 substantially, confirming Table 1 Residues positively selected by Kin2 by peptide library analysis Selectivity values (shown in parentheses for each of two replicates) are quantified peptide array spot intensities normalized such that the average value at a given position is 1. Positively selected residues are shown where the average selectivity value from the two experiments was Ͼ2.

Phosphosubstrate recognition by yeast Kin1 and Kin2
that these residues are preferred by the kinase. Substitution of the intended phosphorylation site Ser residue reduced phosphorylation by ϳ24-fold, suggesting that little phosphorylation occurs at other Ser or Thr residues within the KINtide sequence. Strikingly, we found that introduction of a ϩ2 pSer residue into KINtide led a greater than 30-fold increase in phosphorylation efficiency by Kin1 or Kin2 (Fig. 1B). Although the high K m value hampered accurate determination of kinetic parameters, we estimate the K m value for Kin1 phosphorylation of KINtide ϩ2 pSer to be 490 M (S.E. ϭ 110 M; Fig. 1C). By contrast, the phosphorylation rate of unmodified KINtide was linear with respect to peptide concentration over a similar range (not shown), indicating a much higher K m value. These results confirm that Kin1/2 activity is strongly enhanced by priming phosphorylation, at least on a peptide substrate.

Priming enhances phosphorylation of Sec9 by Kin1 and Kin2 in vitro
We searched a list of previously mapped phosphorylation sites compiled from multiple budding yeast phosphoproteomics studies (33) to identify proteins phosphorylated within the stringent Kin1/2 consensus sequence (NXpSX(S/T)X(I/L)). We found 45 occurrences of this motif in the yeast phosphoproteome (Table S2) on 42 proteins. Proteins phosphorylated within the context of this motif were significantly enriched for three gene ontology cellular component terms: site of polarized growth (26%, background frequency 3.9%, p ϭ 3.3 ϫ 10 Ϫ5 ), bud (21%, background frequency 3.5%, p ϭ 0.0015), and bud neck (19%, background frequency 2.8%, p ϭ 0.0012). Because Kin2 is reported to localize to sites of polarized growth, including the bud tip and bud neck, at least some of these proteins may be authentic direct substrates (22). In light of the preference of Kin1/2 for primed substrates, it is notable that 19 of the 45 sites were also reportedly phosphorylated at the downstream Ser or Thr residue within the motif. Among the corresponding proteins was the plasma membrane-associated t-SNARE protein Sec9, which was predicted to be phosphorylated at Ser 190 by Kin1/2 following priming phosphorylation at Ser 192 . Overexpression of Kin2 in yeast was previously shown to induce Sec9 phosphorylation (21). However, phosphorylation in vitro had been mapped to a distinct residue, Ser 315 , which did not appear to constitute the major site of phosphorylation in vivo. Our observation that Kin1 and Kin2 are priming-dependent pro-vides a potential explanation for this discrepancy, because previous assays of Sec9 phosphorylation in vitro used an unprimed substrate.
To test whether Sec9 phosphorylation by Kin1/2 was enhanced by priming phosphorylation, we needed to prepare a recombinant Sec9 substrate highly phosphorylated at Ser 192 . We noted that Ser 192 falls within the canonical Snf1/AMPK consensus sequence, suggesting that Snf1 could prime Sec9 for subsequent phosphorylation by Kin1/Kin2 ( Fig. 2A). To examine priming-dependent phosphorylation, we thus treated recombinant GST-Sec9 purified from Escherichia coli with active Snf1 in the presence of cold ATP. GST-Sec9 was then repurified on GSH resin to remove Snf1 and subjected to a radiolabel kinase assay with Kin1 (Fig. 2B). We found that although Kin1 did phosphorylate nonprimed Sec9, the level of phosphorylation increased by pretreatment with Snf1. No radiolabel incorporation was seen in the absence of Kin1, indicating that the observed increase in phosphorylation was not attributable to inefficient removal of Snf1 during repurification. To further map sites of phosphorylation by both kinases, we utilized three previously described Sec9 fragments corresponding roughly to the N-terminal (NT1), central (NT2), and C-terminal (CT) thirds of the protein (21). As anticipated, we exclusively observed priming-dependent phosphorylation of the central NT2 fragment harboring Ser 190 and Ser 192 . In the context of this fragment, a S190A mutation completely abolished both priming-dependent and priming-independent phosphorylation by Kin1 (Fig. 2C). In addition, we observed no primingdependent phosphorylation of Sec9-NT2-S192A, whereas priming-independent phosphorylation was retained. Phosphate-affinity (Phos-tag) gel analysis confirmed that Snf1 alone stoichiometrically phosphorylated Sec9-NT under these conditions as indicated by mobility shift predominantly to a single, slower-migrating species (Fig. 2D). This mobility shift was almost entirely dependent on Ser 192 . We note that under these conditions that phosphorylation of Sec9-NT2-S190A by Snf1 was reduced by approximately one-third, either because Snf1 can also phosphorylate Ser 190 or because Snf1 prefers a Ser residue at the Ϫ2 position. Nevertheless, this degree of reduction in phosphorylation by Snf1 cannot be responsible for the complete lack of priming-dependent phosphorylation of Sec9-NT2-S190A by Kin1.

Sec9 is a substrate of Kin1 and Kin2 in vivo
To determine whether Sec9 is also phosphorylated in vivo by Kin1/2, we performed quantitative MS on overexpressed TAPtagged Sec9 purified from either WT or kin1⌬ kin2⌬ yeast. For these experiments we assessed levels of Sec9 phosphorylation by multiple reaction monitoring (MRM) MS using synthetic, stable isotope-labeled, tryptic peptide standards (34,35) correspondingtotheunphosphorylatedspecies,theSer 190 -phosphorylated species, the Ser 192 -phosphorylated species, or the doubly phosphorylated species. By selecting MRM transitions unique to each peptide, we were able to quantify absolute levels of phosphorylation at Ser 190 and Ser 192 , alone or in combination. Purified Sec9 was isolated by SDS-PAGE and digested with trypsin, and phosphopeptides were enriched on a TiO 2 matrix.
We analyzed the flowthrough and bound fractions to quantify unphosphorylated and phosphorylated Sec9, respectively (  or the indicated fragments were incubated with or without active Snf1 catalytic domain in the presence of ATP and then repurified on GSH resin. Sec9 fragments were then used in Kin1 radiolabel kinase assays, fractionated by SDS-PAGE, and subjected to phosphorimaging. Autophosphorylated Kin1 migrates near the 160-kDa marker. C, the Sec9 NT2 fragment (WT or the indicated mutant) was subjected to two-stage Snf1/Kin1 kinase assay as in B. D, Sec9 NT2 fragments were subjected to phosphorylation by Snf1 catalytic domain and then fractionated by phosphate-affinity SDS-PAGE to separate phosphorylated from nonphosphorylated species followed by Coomassie staining. E, full-length Sec9 was immunoprecipitated from cultures of parental or otherwise isogenic kin1⌬ kin2⌬ yeast strains. Protein was subjected to SDS-PAGE and Coomassie staining, and the Sec9 gel band was excised, digested with trypsin, and analyzed by MRM LC-MS/MS. Levels of the indicated singly phosphorylated, Ser 190 /Ser 192 diphosphorylated, and corresponding nonphosphorylated tryptic peptides were determined using stable isotope-labeled peptide standards. Data points are shown for each biological replicate and represent the average values from all MRM transitions across triplicate MS analysis. Bars show means Ϯ S.D. (n ϭ 3). F, yeast strains harboring a stably integrated galactose-inducible WT or kinase inactive (KD) KIN2 gene were transformed with high-copy number plasmids constitutively expressing Sec9 (WT or the indicated mutant). The cells were grown to midlogarithmic phase in raffinose. Cultures were split and grown an additional 4 h in the presence or absence of 2% galactose as indicated. Protein extracts were prepared, fractionated by SDS-PAGE, and immunoblotted with Sec9 and Kin2 antisera.

Phosphosubstrate recognition by yeast Kin1 and Kin2
tion is more abundant than Ser 190 /Ser 192 diphosphorylation, the majority of Sec9 Ser 190 phosphorylation in vivo may occur independently of substrate priming.
Overexpression of Kin2 in yeast induces Sec9 to undergo an electrophoretic mobility shift caused by phosphorylation at unknown sites (21). We examined the mobility of WT Sec9 and Sec9-S190A by SDS-PAGE upon expression of Kin2 under control of a galactose-inducible promoter. As reported previously, we observed a Sec9 mobility shift in the presence of galactose in the Kin2-expressing strain (Fig. 2F). This mobility shift was largely absent for Sec9-S190A, although we did observe a residual mobility shift suggestive of other Kin2-dependent sites of phosphorylation. Taken together with our MS data, these results indicate that Kin2 (and possibly Kin1) can phosphorylate Sec9 at Ser 190 in vivo.

Recognition of primed substrates is mediated by a cluster of basic residues in the N-terminal lobe of Kin1/2
To examine the functional importance of substrate priming by Kin1/2, we sought to generate mutant kinases specifically deficient for priming-dependent phosphorylation. To understand the structural basis for recognition of primed substrates by Kin1/2, we examined the published X-ray crystal structure of human MARK2 in complex with a pseudosubstrate inhibitor, the Helicobacter pylori CagA protein (36). We chose this structure because MARK2 is the kinase most similar to Kin1/2 (58% similarity within the catalytic domain) to have been crystallized in complex with a peptide substrate or pseudosubstrate. In this structure, the side chain of the ϩ2 residue is oriented toward the kinase ␣C helix, a conserved element of the protein kinase fold (Fig. 3A). Although MARK2 itself is not priming-dependent (37), multiple sequence alignment with fungal Kin1 homologs revealed a conserved cluster of basic residues spanning the N terminus of the helix ␣C that are predicted to be oriented toward the ϩ2 residue (Fig. 3A). To examine whether these basic residues might be involved in recognition of primed substrates by Kin1, we mutated each of them individually to Ala and assayed these mutants for phosphorylation of the KINtide and KINtide ϩ2 pSer substrates. Individual mutation of Lys 183 , Arg 187 , or Arg 190 each reduced the phosphorylation rate of KINtide ϩ2 pSer relative to the unmodified peptide, with Arg 187 mutation having the largest effect (Fig. 3B). Mutation of Arg 190 alone reduced phosphorylation of unmodified KINtide by ϳ2-fold, suggesting that it may be important for kinase activity independent of the substrate used. The combination mutantKin1-K183A,R187Aspecificallyreducedtherateofphosphorylation of the primed peptide substrate by more than 60-fold with no effect on the phosphorylation rate of unprimed KINtide. The analogous Kin2 mutant (K162A,R166A) behaved similarly (Fig. 3C). In addition to affecting priming-dependent phosphorylation of the peptide substrates, Kin1-K183A,R187A was also specifically impaired for Snf1 priming-dependent phosphorylation of the Sec9 NT2 protein fragment (Fig. 3D). PSPL analysis of Kin1-K183A,R187A and Kin2-K162A,R166A compared with their WT counterparts showed that these mutations only affected selectivity for phosphorylated residues and did not affect recognition of key residues selected at other positions ( Fig. 4 and Table S1). These results indicate that Lys 183 and Arg 187 of Kin1 (and Lys 162 and Arg 166 of Kin2) are essential for recognition of primed substrates yet dispensable for phosphorylation of nonprimed substrates.

Priming-dependent Sec9 phosphorylation is not required for Kin1 and Kin2 regulation of the late secretory pathway
Kin1 and Kin2 were previously identified as high-copy suppressors of temperature-sensitive mutants of components of the late secretory pathway, acting upstream of Sec9 (21). We examined the ability of Kin2-K162A,R166A to support growth of several mutants in this pathway at their restrictive temperatures (Fig. 5A). We found that the priming defective Kin2 mutant was equal to WT Kin2 in its ability to suppress the temperature-sensitive growth phenotype of all mutants tested. Consistent with our results indicating that Sec9 phosphorylation by Kin1 and Kin2 in vivo is largely priming-independent, these experiments suggest that substrate priming is not required for Kin2 function in exocytosis in general, at least

Phosphosubstrate recognition by yeast Kin1 and Kin2
when overexpressed. These results further confirm that mutation of Lys 162 and Arg 166 does not generally inactivate Kin2. We also found that mutation of Ser 190 , Ser 192 , and Ser 315 , alone or in combination, had no effect on the ability of overexpressed Sec9 to suppress growth defects of the same panel of mutants (Fig. S1). These results suggest that phosphorylation by Kin1/2 is not required for Sec9 function when overexpressed.

Priming-defective Kin1 and Kin2 mutants are selectively impaired for function in vivo
Examination of Kin1/2 function in exocytosis requires overexpression, which may bypass a requirement for substrate priming simply by elevating levels of kinase activity. To further investigate potential roles for substrate priming by endogenous Kin1/2, we examined their function in the UPR. As previously reported (29,38), we found that kin2⌬ cells grew more slowly than WT cells on medium containing tunicamycin and that combined deletion of KIN1 and KIN2 rendered cells highly tunicamycin-sensitive (Fig. 5B). Double mutant cells re-expressing Kin2-K162A,R166A from a low-copy plasmid under control of its own promoter grew at a slower rate than those expressing WT Kin2, an effect that was exacerbated at higher concentrations of tunicamycin (Fig. 5, C and D). We note that in these experiments kin1⌬ kin2⌬ cells re-expressing WT Kin2 grew faster on medium containing tunicamycin than corresponding WT cells transformed with a control plasmid, an effect likely mediated by modest overexpression of Kin2 from the expression plasmid. Nonetheless, these results suggest that substrate priming is required for optimal activity of Kin2 in cells in mediating resistance to ER stress.

Discussion
Here, we have shown that phosphorylation by Kin1/2 is strongly promoted by substrate priming. Recognition of pSer or pThr at the ϩ2 position in substrates is mediated by a patch of basic residues in a region comprising a portion of helix ␣C and the ␣C-␤3 loop. Basic residues in helix ␣C appear to contribute, at least in part, to recognition of primed substrates by other kinases. However, differences in the precise location and quantity of these basic residues likely underlie the positional specificity of phosphosubstrate recognition. For example, CK2 has a cluster of four basic residues located within helix ␣C, as well as three additional basic residues within its activation loop. The highly positive net local charge in CK2 is likely responsible for the strong selectivity of CK2 for acidic and phosphorylated residues at both the ϩ1 and ϩ3 positions and to a lesser extent at other downstream positions (39). In contrast Kin1/2 have a limited number of basic residues in this region, correlating with specificity for only a single phosphorylated residue. The ϩ4 pSer-binding site of GSK3 isozymes includes an Arg residue located at the N terminus of helix ␣C yet includes additional basic residues in the kinase activation loop and catalytic loop. The distinct arrangement of these residues compared with Kin1/2 likely underlies GSK3 selectivity for phosphorylated residues at the ϩ4 rather than the ϩ2 position.
To date the only other kinase reportedly primed by phosphorylation at the ϩ2 position is Golgi casein kinase, recently identified as the secreted protein FAM20C (11,40). Although FAM20C has some catalytic residues in common with canonical eukaryotic protein kinases, the overall topology of its catalytic domain is distinct (41). The X-ray crystal structure of Caenorhabditis elegans FAM20C suggests that substrates bind in a different mode or orientation compared with other kinases. Accordingly, basic residues implicated in recognition of phosphorylated substrates are located distal from helix ␣6, which is analogous to ␣C of other eukaryotic protein kinases.
In addition to uniquely recognizing primed substrates, Kin1/2 are also unusual among members of the AMPK family in other aspects of substrate recognition. Members of the calmodulin-dependent protein kinase group to which the AMPK family belongs are typically characterized as "basophilic," essentially requiring an Arg or Lys residue at the Ϫ3 position in substrates (2). Kin1/2 by contrast are largely nonselective at this position, likely attributable to a Glu to Gln substitution in a residue important for selecting basic residues (Gln 230 /Gln 209 in Kin1 and Kin2, respectively) (31). Kin1/2 are also unique within the family in selecting an Asn residue at the Ϫ2 position, a preference it shares with polo-like kinases despite sharing little sequence similarity. Aside from these differences, Kin1/2 are similar to other members of the yeast and

Phosphosubstrate recognition by yeast Kin1 and Kin2
mammalian AMPK families in having a substantial preference for a Leu residue at the ϩ4 position. In the MARK2-CagA cocrystal structure, the ϩ4 Leu residue of the pseudosubstrate occupies a hydrophobic cavity analogous to the ϩ1 residuebinding site in other kinases. Accommodation of the ϩ4 Leu residue by MARK2 is facilitated by the presence of a small Ala residue immediately upstream of the conserved APE motif at the C terminus of the kinase activation loop (the APE Ϫ1 residue). Notably, an Ala residue at the analogous position is found in Kin1/2 and all other members of the AMPK/Snf1 family, explaining their common selectivity at the ϩ4 position. Although mammalian MARKs are not dependent on substrate priming, it is notable they can be phosphorylated at sites in the N terminus of helix ␣C, which has been reported both to activate (42) and inactivate (43) the kinases. Given the importance of this region in substrate recognition, it is tempting to speculate that its phosphorylation serves to change MARK substrate specificity rather than overall catalytic activity.
Guided by the Kin1/2 phosphorylation site motif, we identified the exocytic t-SNARE protein Sec9 as a direct substrate of the kinases. Previously, Kin1/2 were found to induce Sec9 phosphorylation and dissociation from the plasma membrane, promoting its activity in exocytosis (21). Because mutation of sites phosphorylated by Kin2 in vitro had no effect on its phosphorylation in cells, it was concluded that Kin1/2 likely acted indirectly to induce phosphorylation of Sec9. Here, we identified Ser 190 of Sec9 as a Kin1/2 phosphorylation site that in vitro was largely dependent on prior phosphorylation at Ser 192 . Quantitative analysis confirmed that Sec9 phosphorylation at Ser 190 phosphorylation depended on Kin1/2 in yeast, albeit in a manner likely independent of Ser 192 priming. Notably, the Sec9 was not phosphorylated at Ser 190 to full stoichiometry, and it is possible that stimuli activating the relevant Ser 192 kinase(s) would enhance Ser 190 phosphorylation as well. Although Snf1 could phosphorylate Ser 192 in vitro, we found no increase in Ser 192 or Ser 190 phosphorylation upon glucose withdrawal (data not shown), conditions that strongly activate Snf1 in yeast. Ser 192 may therefore be phosphorylated by a distinct kinase with similar substrate specificity, perhaps within the yeast Snf1/AMPK family. We also note that mutation of the Kin1/2 phosphorylation sites in Sec9 did not apparently affect its function in the secretory pathway, at least when overexpressed. Although these experiments may reflect a consequence of overexpression, because kin1⌬ kin2⌬ strains have no overt secretory pathway defect, phosphorylation of Sec9 is unlikely to be essential for function. Similarly, we do not believe that phosphorylation of Sec9 is relevant to the function of Kin1/2 in the UPR. Neither overexpression of Sec9 nor low-copy expression of the phosphomimetic mutant Sec9-S190E,S192E could bypass the requirement for Kin1/2 in supporting growth in the presence of tunicamycin. 3 Kin1/2 may therefore promote Sec9 function in other contexts that have yet to be identified.
Using a mutant defective for recognition of primed substrates, we found that priming dependence enhanced the ability of Kin2 to function in the ER stress response, particularly at high concentrations of tunicamycin. These observations suggest that substrates important for Kin1/2 function in the UPR are at least partly dependent on priming for optimal phosphorylation. A possible explanation would be that low-level phos- or otherwise isogenic strains of the indicated genotype were spotted onto solid YPD medium in the presence or absence of 0.5 g/ml tunicamycin. The plates were incubated at 30°C until WT colonies were visible. C, BY4741 (WT) or otherwise isogenic kin1⌬ kin2⌬ strains were transformed with a low-copy plasmid expressing GFP fused to WT Kin2 or Kin2-K162A,R166A. The cultures were grown in selective medium, and then serial 5-fold dilutions were plated onto solid SD-Ura medium in the presence or absence of 0.5 g/ml tunicamycin and grown for 72 h at 30 C. D, serial dilutions of cultures were grown as for C, but at the indicated concentrations of tunicamycin. Protein was extracted from aliquots of the liquid culture and immunoblotted with antibodies to GFP to determine levels of GFP-Kin2. Homocitrate synthase (HCS) served as a loading control.

Phosphosubstrate recognition by yeast Kin1 and Kin2
phorylation of a key Kin1/2 substrate is sufficient to withstand mild ER stress but that increased phosphorylation mediated by priming is required at high levels of tunicamycin. Alternatively, increased Kin1/2 activity conferred by substrate priming might be required to counteract phosphatases activated under conditions of severe ER stress. In other systems, an optimal phosphorylation site sequence is not required for a kinase to achieve a high level of substrate phosphorylation in vivo. Rather, the presence of a "high quality" phosphorylation site appears to confer robustness to perturbation, for example by requiring lower levels of kinase activation (2). Consequently, some, but not all, substrates of a kinase tend to conform to optimal consensus sequences. Accordingly, it is almost certainly the case that additional Kin1/2 substrates exist aside from those harboring precise matches to their consensus sequence (as listed in Table S2).
Prior studies have suggested that Kin1/2 function in the UPR, as well as in cell polarity and polarized growth. The lack of overt cell polarity defects in kin1⌬ kin2⌬ cells may reflect compensation by other, perhaps related, kinases. However, the distinct phosphorylation site motif of Kin1/2 compared with their most closely related kinases suggests that they are likely to have unique roles in cell polarity. Consistent with this notion, Kin2 reportedly localizes to sites of polarized growth in budded cells (22), and a significant fraction of our potential Kin1/2 substrates share this pattern of localization (Table S2). Whether the function of Kin1/2 in cell polarity and cell wall maintenance is related to their roles in the UPR is not clear. Notably, ER homeostasis mediated by components of the UPR pathway is essential for proper organization of the cell wall (44). Furthermore, cell wall-disrupting agents activate the UPR and induce Hac1 mRNA splicing, a Kin1/2-dependent process. Because some of our candidate Kin1/2 substrates modulate sensitivity to tunicamycin and/or cell wall stress (Table S2), it is tempting to speculate that these substrates may contribute to Kin1/2 function in these processes. Future studies will be aimed at verifying these proteins as substrates and investigating the functional impact of their phosphorylation by Kin1/2.

Plasmids
Yeast high-copy plasmids (pBG1805) expressing C-terminally tandem affinity purification (TAP)-tagged full-length Kin1, Kin2, and Sec9 from a galactose-inducible promoter were from the movable ORF (MORF) collection and were obtained from the laboratory of Michael Snyder (45). The yeast highcopy plasmids expressing Kin2 and Sec9 from their respective promoters and bacterial expression plasmids for GST-tagged Sec9 fragments and Snf1 catalytic domain (Snf1-cat) were previously described (21,46,47). Sec9 fragments corresponded to the following residues: NT1, 1-150; NT2, 166 -401; and CT, 402-651. The bacterial expression vector for GST-tagged fulllength Sec9 was generated by PCR amplification of the Sec9 coding sequence from the MORF plasmid and insertion into the BamHI and NotI sites of pGEX-4T2. The low-copy vector for expressing GFP-Kin2 from its own promoter (pRS416-GFP-Kin2) included the entire Kin2 ORF plus 1 kb upstream and 250 bp downstream yeast genomic sequence, with yeast codon opti-mized GFP (from pKT0209, generated by the Kurt Thorn laboratory and obtained from Addgene) inserted between the promoter and the translation start site. All mutants were generated using the QuikChange protocol (Stratagene).

Yeast strains and media
Single deletion mutants on the BY4741 (S288c) genetic background were purchased from Open Biosystems. The kin1⌬ kin2⌬ strain (BTY008) was generated by transforming kin1⌬ cells with a previously described targeting construct including the LEU2 gene flanked by genomic KIN2 sequence (18). Clones were selected on SC-Leu medium, and correct targeting of the KIN2 locus was confirmed by PCR from genomic DNA using primers flanking both ends of the insertion site. Temperaturesensitive secretory pathway mutant strains and strains expressing Kin2 from an integrated galactose-inducible cassette were previously described (21). The host strain for expression of MORF constructs (Y258) was obtained from Michael Snyder. Yeast medium components were from BD Biosciences except for complete supplement dropout mixtures (MP Biomedicals).

Protein expression and purification
TAP-tagged full-length Kin1 and Kin2 used for in vitro kinase assays were overexpressed in yeast and purified as described (45) with the following modifications. Yeast (strain Y258) transformed with pBG1805-Kin1 or Kin2 were grown overnight at 30°C in SD-Ura medium. Cultures were expanded into 400 ml of SC-Ura ϩ 2% raffinose, grown ϳ16 h to midlogarithmic phase (A 600 ϭ 0.6 -0.7), and induced by adding 200 ml of 3ϫ YEP-Gal medium (3% yeast extract, 6% peptone, 6% galactose). After shaking 6 h at 30°C, the cells were collected by centrifugation and washed twice with double-distilled H 2 O, and pellets were frozen at Ϫ80°C. The cell pellets were thawed on ice, suspended in 2 ml of lysis buffer (50 mM Tris, pH 7.5, 1 mM EDTA, 1 mM EGTA, 0.1% Triton X-100, 1 mM DTT, 0.4 mM Na 3 VO 4 , 1 mM NaF, 5 mM ␤-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, 2 g/ml pepstatin A, 10% glycerol) containing 150 mM NaCl and an equal volume of acid-washed 0.5-mm glass beads (Biospec) and lysed by vortexing. Following centrifugation, the supernatant was reserved, and insoluble debris was re-extracted by vortexing in 2 ml of lysis buffer containing 650 mM NaCl. The glass beads were pelleted, and the combined supernatants were diluted with 9 ml of lysis buffer. After adding 100 l of IgG-Sepharose beads (GE Life Sciences), the slurry was tumbled for 2 h at 4°C. The beads were pelleted and washed three times with wash buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Triton X-100, 10% glycerol) and three times with cleavage buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM DTT, 0.1% Tween 20, 25% glycerol). Kinase was released from the beads by overnight treatment with GST-3C protease in 300 l of cleavage buffer. Following centrifugation, the supernatant was incubated with GSH-Sepharose 4B (GE Life Sciences) to remove the protease. Kinase preparations were snap-frozen in aliquots and stored at Ϫ80°C.
E. coli strains Rosetta (DE3) (for expression of full-length Sec9) or BL21(DE3) (for all other constructs) were purchased from EMD Millipore. Transformed cells were grown at 37°C to

Phosphosubstrate recognition by yeast Kin1 and Kin2
an A 600 of 0.8 -1.0 in 200 ml of terrific broth with ampicillin, and expression was induced with 0.2 mM isopropyl ␤-D-thiogalactopyranoside. After 2 h, the cells were pelleted, washed once with cold PBS, and frozen in dry ice/EtOH. Pellets were thawed on ice and resuspended in 5 ml of 20 mM Tris, pH 7.5, 140 mM NaCl, 1 mM EDTA, 1 mM DTT, 10 g/ml leupeptin with 0.2 mg/ml lysozyme. After 15 min of incubation on ice, 70 l of 10% sodium deoxycholate and 25 l of 200 mM phenylmethylsulfonyl fluoride were added and rotated 15 min at ambient temperature. DNase I (150 units) and 65 l of 1 M MgCl 2 were added, and mixing was continued for another 15 min. Insoluble material was removed by centrifugation, and the supernatant was incubated with 0.25 ml of GSH-Sepharose 4B resin with rotation for 1-2 h, 4°C). The beads were washed three times with GSH wash buffer (50 mM Tris, pH 8.0, 50 mM NaCl, 1 mM DTT, 0.01% Igepal CA630, 10% glycerol). Sec9 constructs were eluted into 0.5 ml of GSH wash buffer containing 10 mg/ml reduced GSH, and eluates were dialyzed overnight at 4°C into GSH wash buffer. Snf1-cat was cleaved from the GST moiety rotating the beads overnight in 0.25 ml of GSH wash buffer containing 1 unit of thrombin, 2.5 mM CaCl 2 , and 5 mM MgCl 2 . Following centrifugation and removal of the supernatant, the beads were incubated with an addition 0.25 ml of GSH wash buffer with rotating for 15 min. The beads were pelleted, and the combined supernatants were incubated with benzamidine-agarose (Sigma-Aldrich) to remove thrombin. The protein samples were aliquoted, frozen in dry ice/EtOH, and stored at Ϫ80°C.

PSPL analysis
Phosphorylation sequence specificity of WT and mutant Kin1 and Kin2 was determined by PSPL analysis (31). The peptide library (Anaspec) consisted of 200 peptide mixtures having the general sequence YAXXXXX(S/T)XXXXAGKK(biotin), where eight of the nine residues indicated by X were an equimolar mixture of the 17 amino acids (excluding Cys, Ser, and Thr), where "S/T" indicates an equimolar mixture of Ser and Thr, and where the biotin group was conjugated to the C-terminal Lys side chain via an aminohexanoic acid spacer. The remaining X position was fixed as one of the 20 unmodified amino acid residues, pThr, or pTyr. Kinase reactions (2 l) were performed in 1536-well plates in kinase assay buffer (50 mM HEPES, pH 7.4, 10 mM MgCl 2 , 1 mM DTT, 0.1% Tween 20) containing 50 M peptide substrate, and 50 M ATP containing 0.03 Ci/ml [␥-33 P]ATP (PerkinElmer Life Sciences). Following incubation at 30°C for 2 h, 200-nl aliquots of each reaction were transferred to a streptavidin-coated membrane (Promega), which was washed and dried as described (31). Radiolabel incorporation into peptides was quantified by phosphorimaging. The data were normalized so that the average signal for the 20 unmodified amino acids at a given position was assigned to 1, and results show the means of two separate runs. Heat maps were generated using Microsoft Excel. Candidate substrates were identified from published phosphoproteomics studies (33) that conformed to the consensus sequence NX(pS/pT)X-SX(I/L). Gene Ontology (GO) term enrichment among the resulting 42 genes was performed using the Saccharomyces Genome Database GO term finder tool (version 0.86) (48).

In vitro kinase assays
Kinase substrate peptides (KINtides) were prepared by standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) solid-phase synthesis procedures and purified by reversed-phase HPLC prior to use. Peptide identity was verified by MALDI-MS. Peptide kinase assays were performed in 75 l of kinase assay buffer containing the indicated concentration of peptide and were initiated by adding ATP to 50 M (with 0.08 Ci/ml [␥-33 P]ATP) and transferring to 30°C. Aliquots (20 l) were removed at 5-min intervals and spotted onto P81 phosphocellulose filters (GE Healthcare), which were immediately quenched in 75 mM phosphoric acid. Filters were washed three times in the same solution, air-dried, and analyzed by scintillation counting. Phosphate incorporation into peptides was calculated from standards consisting of varying amounts of reaction mixtures spotted onto filters that were left unwashed. The K m value was estimated by fitting reaction rates at varying substrate concentration to the Michaelis-Menten equation using Prism 7 (GraphPad software).
To perform in vitro priming assays, Snf1-cat (6 g) was initially activated by incubating with recombinant Elm1 (49) in 30 l of Snf1 activation buffer (50 mM Tris, pH 8.0, 50 mM NaCl, 10 mM MgCl 2 , 0.01% Igepal CA-630, 1 mM DTT, 1 mM ATP, 10% glycerol) for 1 h at 30°C alongside control reactions lacking Snf1. Snf1 reactions were then diluted 10-fold into 75 l of 10 mM HEPES, pH 7.4, 100 mM NaCl, 10 mM MgCl 2 , 1 mM DTT, 0.2 mM benzamidine, 1 mM ATP containing 7 g of GST-Sec9 (full-length or fragment) and incubated 30 min at 30°C. To purify phosphorylated Sec9, reaction tubes were chilled on ice and diluted with 100 l of GSH wash buffer (50 mM Tris, pH 8.0, 100 mM NaCl, 1 mM DTT, 0.01% Igepal CA-630) containing 25 l of GSH-Sepharose beads. After rotating 1 h at 4°C, the beads were pelleted and washed three times with 0.5 ml of GSH wash buffer. GST fusion proteins were eluted with 35 l of 6 mg/ml reduced GSH in GSH wash buffer. A portion of the repurified GST fusion proteins were retained for nonradioactive SDS-PAGE. For the second stage of the assay, 10 l of repurified GST fusion protein was diluted 2-fold into kinase assay buffer containing ϳ5 ng of Kin1 and 10 M ATP with 0.1 Ci/l [␥-33 P]ATP and incubated a further 30 min at 30°C. The reactions were quenched by adding 4 l of 6ϫ SDS-PAGE loading buffer (final concentrations: 2% SDS, 25 mM Tris, pH 6.8, 10% glycerol, 5% ␤-mercaptoethanol) and heating to 95°C for 5 min. The samples were fractionated by SDS-PAGE (10% acrylamide), and gels were stained with Coomassie Blue, destained, dried, and exposed to a phosphor storage screen to detect radiolabel incorporation. Nonradioactive samples were fractionated by Phos-tag SDS-PAGE (containing 7.5% acrylamide, 2.5 M Phos-tag acrylamide (Wako Chemicals), 5 M MnCl 2 ), and gels were stained with Coomassie Blue.

Analysis of Sec9 phosphorylation in vivo
Sec9-TAP was expressed in 100-ml cultures of the indicated strains and purified on IgG-Sepharose (25 l) as described above for Kin1/2-TAP, except that protein was eluted by suspending in 50 l of 50 mM Tris, pH 6.8 with 1% SDS and heating to 95°C for 2 min. The beads were centrifuged, and the super-