Development of a yeast model to study the contribution of vacuolar polyphosphate metabolism to lysine polyphosphorylation

A recently-discovered protein post-translational modification, lysine polyphosphorylation (K-PPn), consists of the covalent attachment of inorganic polyphosphate (polyP) to lysine residues. The nonenzymatic nature of K-PPn means that the degree of this modification depends on both polyP abundance and the amino acids surrounding the modified lysine. K-PPn was originally discovered in budding yeast (Saccharomyces cerevisiae), in which polyP anabolism and catabolism are well-characterized. However, yeast vacuoles accumulate large amounts of polyP, and upon cell lysis, the release of the vacuolar polyP could nonphysiologically cause K-PPn of nuclear and cytosolic targets. Moreover, yeast vacuoles possess two very active endopolyphosphatases, Ppn1 and Ppn2, that could have opposing effects on the extent of K-PPn. Here, we characterized the contribution of vacuolar polyP metabolism to K-PPn of two yeast proteins, Top1 (DNA topoisomerase 1) and Nsr1 (nuclear signal recognition 1). We discovered that whereas Top1-targeting K-PPn is only marginally affected by vacuolar polyP metabolism, Nsr1-targeting K-PPn is highly sensitive to the release of polyP and of endopolyphosphatases from the vacuole. Therefore, to better study K-PPn of cytosolic and nuclear targets, we constructed a yeast strain devoid of vacuolar polyP by targeting the exopolyphosphatase Ppx1 to the vacuole and concomitantly depleting the two endopolyphosphatases (ppn1Δppn2Δ, vt-Ppx1). This strain enabled us to study K-PPn of cytosolic and nuclear targets without the interfering effects of cell lysis on vacuole polyP and of endopolyphosphatases. Furthermore, we also define the fundamental nature of the acidic amino acid residues to the K-PPn target domain.

activity of its synthesizing enzyme, the NO-synthase (4). The same rules are likely to be applied to all nonenzymatic PTMs, with the selectivity dictated by the chemical and physical surroundings of the targeted amino acids and the signal strength controlled by the concentration of the specific reactive metabolite. One such PTM is the modification of lysine polyphosphorylation (K-PPn) that consists of the covalent attachment of inorganic polyphosphate (polyP), a ubiquitous linear polymer of phosphate residues, to target proteins (5). Polyphosphorylation occurs at lysine residues, creating a phosphoramidate bond (P-N) that, unlike phosphoester bonds, is highly labile under the acidic conditions typically used in mass spectrometry (MS). Lysine is an electron-rich positively-charged amino acid with a nucleophilic side chain, making it one of the most modified amino acids in nature (6). Together, these characteristics make the thorough characterization of K-PPn a challenging endeavor. Two K-PPn target proteins have been characterized in detail so far, the nuclear signal recognition 1 (Nsr1), a protein involved in ribosomal rRNA biogenesis (7), and its interacting partner topoisomerase 1 (Top1), an enzyme that unwinds supercoiled DNA. K-PPn leads to an increase in molecular weight as seen by an upwards mobility shift on NuPAGE SDS gels. The target lysine residues are located within the PASK domain (polyacidic serine and lysine(K)), a region characterized by an abundance of acidic amino acids (glutamic and aspartic acids) and lysine and serine residues (5). The importance of these amino acids for the modification is poorly understood. It has been demonstrated that mutations of all glutamic acid, aspartic acid, or serine residues within the PASK domain do not directly affect Top1 K-PPn (5). However, deletion constructs within the PASK domain of Top1 significantly reduced the mobility shift of this domain, suggesting that the three-dimensional structure of the PASK domain might play an important role in dictating the degree of the modification and/or that there are several lysine targets within the PASK domain.
All organisms from bacteria to mammals possess polyP. This simple polymer was discovered more than a century ago and for a long time was believed to be simply a phosphate storage molecule. However, recently it was revealed that polyP is a multifaceted molecule possessing several functional roles in numerous biological contexts. Among some of the functions ascribed to polyP are as follows: its chaperone-like properties for stress protection (7), its role as a blood coagulation factor (8,9), and it is a modulator of infectivity of bacterial and protozoan pathogens (10,11). The polyP biosynthetic pathway is characterized in bacteria, yeast, Trypanosoma, and Dictyostelium. In mammalian cells, despite many reports showing the existence of polyP, no homologues of PPK1 and PPK2 (polyP kinase 1 and 2, respectively) from prokaryotes or vacuolar transport complex (Vtc) from protists have been found to date. Dictyostelium possesses a PPK1 bacterial-like enzyme acquired by horizontal gene transfer (12), whereas trypanosomes possess Vtc4 yeastlike enzymology (13).
The fact that in budding yeast all the enzymology for the synthesis and hydrolysis of polyP is known makes this the preferred model in which to study K-PPn. However, this model has some drawbacks. Saccharomyces cerevisiae has very high concentrations of polyP, 80% of which is located in the vacuole (5) and therefore is not physiologically available to K-PPn cytosolic or nuclear proteins. Moreover, yeast possess four polyP-phosphatases able to hydrolyze polyP. Three are endopolyphosphatases, enzymes that hydrolyze polyP internally, namely Ppn1 (14,15), Ppn2 (16), and Ddp1 (17). A very active exopolyphosphatase is also present, Ppx1, an enzyme that hydrolyzes polyP from the terminal phosphate, to release P i and PP i as the final products (18). Ppn1 and Ppn2 have a strict vacuole localization. Ppx1, Ddp1, and Ppn1 not only target "naked" polyP but are also known to actively de-polyphosphorylate proteins (5).
The nonenzymatic nature of K-PPn predicts that the degree of this modification can be influenced by the abundance of polyP. Therefore, during cell lysis, the abundant polyP of the vacuole is released from the broken organelle and could subsequently nonphysiologically attach to target proteins. This is an issue common to most nonenzymatic PTMs where the reactive metabolite and the target protein can come in contact during cell lysis. K-PPn analysis also encounters the opposite problem, the release of the polyP phosphatases Ppn1 and Ppn2 from the vacuole. Their release in the cell lysate could result in a reduction of the degree of K-PPn. Here, we investigate these hypotheses leading to a better characterization of K-PPn, and we report on the creation of a budding yeast strain suited to study this modification in a more physiological context.

Polyphosphorylation mobility shift can be exacerbated during cell lysis
The nonenzymatic nature of K-PPn made us question whether upon extraction there could be an exacerbation of the modification, as measured by a mobility shift on NuPAGE. We first tested whether the polyP released from the vacuoles could interact with proteins, altering their mobility on NuPAGE, by mixing two cell cultures in a 1:1 ratio, one devoid of polyP (vtc4⌬) and the other possessing polyP. PolyP-deficient vtc4⌬ cells expressing Myc-tagged Top1 or Nsr1 (gTop1-13Myc and gNsr1-13Myc, respectively) were mixed with a second culture containing either normal (WT), high levels (vip1⌬), or no polyP (vtc4⌬). Vip1 is a yeast diphosphoinositol pentakisphosphate kinase whose deletion is known to result in elevated polyP levels (19). The proteins were immediately extracted, as described under "Experimental procedures," without detergent and using glass beads for physical homogenization. The predicted molecular masses of 130 kDa for gTop1-13Myc and 80 kDa for gNsr1-13Myc were observed, as expected, from extracts containing no polyP (vtc4⌬ ϩ vtc4⌬; Fig. 1A). However, the upward gel-mobility shift characteristic of K-PPn was observed for both proteins in extracts where polyP was present (vtc4⌬ ϩ WT; vtc4⌬ ϩ vip1⌬). This result confirms the hypothesis that K-PPn can occur during cell lysis. Next, we assessed whether K-PPn proteins could be further polyphosphorylated post-extraction. To assess this, we used polyP of two different lengths. One was synthetic polyP of an average chain length of 100 phosphate groups, herein called P100. To make polyP with an average chain length of longer than 100, we used a yeast quadruple mutant in which all the polyP phosphatases are deleted (ppn1⌬ppn2⌬ppx1⌬ddp1⌬), herein called DDY1810 QM or EDITORS' PICK: Characterization of lysine polyphosphorylation simply DQM (Fig. 1B). Protein extracts from gNsr1-13Myc vtc4⌬ (Fig. 1C, lane 3) were first incubated with P100 (Fig. 1C, lane 2), followed by incubation with extract from the DQM mutant. This experiment revealed that the mobility of gNsr1-13Myc can be further increased after the first K-PPn modification (Fig. 1C, lane 1). The same was observed when non-K-PPn gNsr1-13Myc treated with synthetic polyP of average chain length 45 (herein called P45) was subsequently mixed with DQM extract (Fig. 1C, lane 4). Importantly, gNsr1-13Myc from WT yeast, endogenously K-PPn, also shifted up after mixing with DQM extract (Fig. 1C, lanes 7 and 8). These results indicate that K-PPn proteins can be further polyphosphorylated during cell lysis.

Polyphosphorylation has an opening effect on target proteins
One question arises: how could polyP further modify K-PPn proteins? We can imagine two possibilities ( Fig. 2A): one through the lysine polyphosphorylation of new lysine residues (additional model), or one, chemically less feasible, in which there is a replacement of the shorter polyP chain with the longer one on a particular lysine residue (replacement model). To distinguish between these two models, we performed the following analysis. Non-K-PPn gTop1-GFP was purified from vtc4⌬ and kept on GFP-TRAP beads. The beads were incubated with a short synthetic polyP of an average chain length of 8 and with fluorescein (FAM) moieties at both ends, herein called bis-FAM-P8 (20). This led to the transfer of FAM to gTop1-GFP as seen by Western blotting (Fig. 2B, lanes 1 and 2). An anti-FITC antibody was used to detect FAM as they are closely-related molecules. When the same beads were further incubated with P45, we observed a mobility shift of gTop1-GFP but no reduction of the FAM signal (Fig. 2B, lane 3). This suggests that there is no replacement of bis-FAM-P8 with P45. For further confirmation, we tested whether the mobility shift of K-PPn targets could be further enhanced by denaturation, because this would make additional lysine residues available. We initially tested whether both gTop1-13Myc and Nsr1 can be Lys-polyphosphorylated under denaturing conditions. Proteins were extracted from vtc4⌬ yeast using no SDS as before or under the denaturing condition of 2% SDS, and then incubated with polyP of average chain length 65 (herein called P65) or P100 (Fig. 2C). We observed that both Nsr1 and gTop1-13Myc could still be Lyspolyphosphorylated when denatured. Unlike native proteins, denatured gTop1-13Myc or Nsr1 became readily modified at low P100 concentrations (Fig. 2D, even-numbered lanes), with this effect more prominent for Top1. These data indicate that the protein denaturation process enhances K-PPn mobility shifts by exposing extra lysine residues, therefore supporting the additional model ( Fig. 2A). Furthermore, the mobility shift increase depends on the amount of polyP indicating that an early K-PPn event allows further lysine residue(s) to be Lys-polyphosphorylated likely by inducing an opening/unfolding of the polypeptide chain.

Acidic environments play an essential role in defining polyphosphorylation target regions
As shown, the three-dimensional structure of the target does not seem to play a significant major role in determining K-PPn. Nevertheless, as highlighted in the Introduction, nonenzymatic PTMs do not occur randomly; the physicochemical properties of the amino acids surrounding a specific lysine could still determine the degree of K-PPn. Therefore, we next investigated whether those properties of amino acids enriched in the PASK domain played a direct role in the modification. Previously, we demonstrated that independently mutagenizing either all glutamic or all aspartic acids within the PASK domain did not affect K-PPn (5). However, it is possible that due to their abundance the acidic nature of these amino acids compensates for each other. To test this, we concurrently mutagenized all glutamic and aspartic acids within the PASK domain to alanines and leucines, respectively ( Fig. S1; GFP-Top1(D/E-A/L)). We verified that these mutations do not affect Top1 nuclear localization; like GFP-Top1, GFP-Top1(D/E-A/L) correctly localized to the nucleus (Fig. 3A). Then we investigated the effect of these mutations on GFP-Top1 K-PPn. We expressed the proteins in both WT and vtc4⌬ and observed that GFP-Top1(D/  A, cultures from DDY1810 yeast containing different levels of polyP (no polyP-vtc4⌬, normal polyP-WT, and high polyP-vip1⌬) were mixed in a 1:1 ratio (OD 600 ϭ 0.8) with a culture from gTop1-13Myc-or gNsr1-13Myc-tagged vtc4⌬; the proteins were extracted, run on NuPAGE, and blotted with anti-Myc to detect Top1 and Nsr1 and with anti-tubulin (␣-Tub) (loading control). B, total polyP from the ppn1⌬ppn2⌬ppx1⌬ddp1⌬ quadruple DDY1810 mutant (DQM). PolyP corresponding to 5 g of total RNA was fractionated on a 30% polyacrylamide gel and stained by negative staining with DAPI. C, shift-up experiment of purified gNsr1-13Myc (vtc4⌬), kept on Myc-agarose beads, subjected to several treatments, in short, the beads were incubated for 20 min at 30°C either with 2 mM polyP100 (P100), polyP45 (P45), polyP extracted from the ppn1⌬ppn2⌬ppx1⌬ddp1⌬ DDY1810 quadruple mutant (DQM) or left untreated, washed in lysis buffer, and further incubated with polyP as described in panel C. Protein samples were run on NuPAGE and blotted with anti-Myc. The figures presented are a representation of at least three independent repeats. EDITORS' PICK: Characterization of lysine polyphosphorylation E-A/L) shows similar mobility to vtc4⌬ and WT suggesting a prevention of K-PPn (Fig. 3B). Moreover, these mutations also significantly affected the stability of the protein, as the steadystate level of the WT GFP-Top1 is ϳ5 times higher than that of the mutant (Fig. 3C). Together, these data demonstrate that the acidic environment conferred by the high number of both glutamic and aspartic acids within the PASK domain is essential for K-PPn.

Vacuolar polyphosphatases Ppn1 and Ppn2 affect polyphosphorylation
We next investigated the effect of polyP phosphatases on target mobility. Like the exacerbation of mobility shift that occurs during the extraction procedure due to the release of vacuolar polyP, it is possible that the very active polyphosphatases that reside in the vacuole might also affect the K-PPn status of nuclear and cytoplasmic targets. We observed that in EDITORS' PICK: Characterization of lysine polyphosphorylation the DDY1810 strain, which is depleted of the Pep4 protease required to proteolytically process and activate Ppn1, the mobility of Nsr1 is higher than in the BY4741 strain in which Ppn1 is active (Fig. 4A). This suggests that this polyphosphatase can act during cell lysis; as a vacuole-localized endopolyphosphatase ( Fig. S2A) (14), the presence or absence of Ppn1 should not affect the in vivo mobility of either Top1, which has an exclusive nuclear localization, or of Nsr1, which shuttles between the cytoplasm and the nucleus. We therefore decided to investigate the activity of each of the known polyP polyphosphatases. We started by engineering a strain similar to DQM but in the BY4741 background by deleting the four known polyphosphatases (ppn1⌬ppn2⌬ppx1⌬ddp1⌬ herein called BQM). As expected, and like the DQM, the polyP length in BQM is much higher than that of WT or the vip1⌬ mutant ( However, gTop1-9Myc quickly loses its mobility shift to nonmodified levels when all the phosphatases are active (i.e. WT conditions), which is not observed when all the known polyphosphatases are deleted (i.e. in BQM). Upon overnight incubation, gTop1-9Myc in BQM shows a slightly faster migration suggesting that other phosphatases are also able to hydrolyze K-PPn proteins. Analysis of Nsr1 gives slightly different results (Fig. 4C, bottom panel): in WT, extraction in lysis buffer is generally sufficient for the protein to be depolyphosphorylated; when comparing the mobility of denatured Nsr1 in WT and in BQM, we observe that Nsr1 mobility shift is higher in BQM extracts; and upon overnight incubation, Nsr1 loss of mobility shift is substantial. Taken together, these results suggest that Nsr1 is far more sensitive to polyphosphatases and to other phosphatases than Top1. Next, we analyzed the effect on polyP metabolism and on K-PPn-induced mobility shift of each polyphosphatase individually and in combination. Individually, the polyphosphatase with the greatest impact on polyP chain length was Ppn1; knockout of Ppn1 increased polyP length while no significant impact was observed upon individual deletion of the other polyphosphatases, confirming previously published results ( Fig. S2B) (21). However, this increase is considerably higher upon simultaneous deletion of Ppn2, the other vacuole-localized endopolyphosphatase ( Fig. 5 and Fig. S2, A and B) (16). Further deleting both Ppx1 and Ddp1 to create the BQM strain did not seem to impact polyP chain length (Fig. 5). These latter enzymes localize mainly to the cytoplasm and the nucleus to a lesser extent (Fig. 2C) (22). We next tested the effect of these knockouts on the mobility of both Nsr1 and Top1 (Fig. 6). When extracted under nondenaturing conditions (Fig. 6A, top  panel), Nsr1 showed extreme sensitivity to the presence of Ppn1 because whenever this endopolyphosphatase was present (all lanes except 3, 9, and 11-13), Nsr1 was completely depolyphosphorylated, demonstrating that there is some post-extraction processing by these enzymes. The highest Nsr1 mobility shift observed in the nondenaturing condition is when Ppn1 and Ppn2 are simultaneously deleted (Fig. 6A, lane 9). Under denaturing conditions where polyphosphatases are inactive (Fig. 6A, bottom panel), the trend is similar, but there is a clear difference between Nsr1 mobility in WT and vtc4⌬ that was not seen before (lanes 1 and 2, and 7 and 8). Interestingly, there is also a difference between Nsr1 mobility in BQM under native and denaturing conditions (Fig. 6A, lane 16, top and bottom, respectively) suggesting that other phosphatases are also able to A. EDITORS' PICK: Characterization of lysine polyphosphorylation hydrolyze Nsr1 K-PPn. Together, these data demonstrate that several phosphatases exert a post-extraction effect on Nsr1 mobility and degree of K-PPn. Unlike Nsr1, when extracted under nondenaturing conditions, gTop1-9Myc (Fig. 6B) showed a considerable mobility difference between WT and vtc4⌬ (lanes 1 and 2, 7 and 8, and 14 and 15), further demonstrating that K-PPn of Top1 is less sensitive to polyP phosphatases than Nsr1. However, we still observed that Ppn1 does act on Top1 post-extraction, because gTop1-9Myc mobility shift was greatest when Ppn1 is not present (Fig. 6B, lane 3). The effect of also deleting Ppn2 was negligible, if present (Fig. 6B, lane 9). However, this effect might be undetectable due to the lack of gel resolution at this high molecular weight. For the same reason, it is difficult to analyze differences between the deletion mutants following gTop1-9Myc extraction under denaturing conditions, as the K-PPn-induced mobility shift is very large for all knockout strains (Fig. S2D). In conclusion, these results indicate that polyP phosphatases, in particular Ppn1 and Ppn2, are responsible for hydrolyzing K-PPn on target proteins during the extraction procedures.

Creating a model yeast strain to study endogenous polyphosphorylation
Budding yeast is currently the only model available to study K-PPn endogenously, relying on a mobility shift readout on NuPAGE gels. However, as we have demonstrated, both vacuolar polyP and polyphosphatases can influence the modification and mobility shift of K-PPn target proteins during cell lysis. It is therefore imperative to develop a model that eliminates this interference, allowing true readout of endogenous cytosolic/ nuclear K-PPn. To eliminate vacuolar polyP, we targeted the exopolyphosphatase Ppx1 to the vacuole by exogenously expressing vt-Ppx1 (Fig. 7A) (8). In DDY1810 strains we observed an 80% reduction by PAGE that was quantified by malachite green assay (Fig. 7B). Considering that vacuolar polyP accounts for 80% of the total polyP (5), this result suggests that vt-Ppx1 successfully hydrolyzed this confined polyP pool. To verify that the nuclear polyP pool had not been affected by overexpressing vt-Ppx1, nuclear polyP was isolated (Fig. 7C). We observed a reduction of the nuclear polyP by about 40% in DDY1810 WT cells overexpressing vt-Ppx1 compared with those expressing a control empty vector. Most of the missing nuclear polyP was of low molecular weight as can be observed by DAPI staining of the polyacrylamide gel (Fig. 7C, left panel). This reduction in nuclear polyP levels likely reflects post-extraction activity from vt-Ppx1 contamination of the nuclear preparations. Although the overexpressed protein was found correctly mainly in the vacuole, immunofluorescent staining revealed its presence also in the endoplasmic reticulum (Fig. 7D), an organelle known to cofractionate with nuclei. Next, we tested the effect of eliminating vacuolar polyP on Top1 and Nsr1 mobility. Top1 is a nuclear protein (24), whereas Nsr1, despite localizing mostly to the nucleus, is known to shuttle between the cytoplasm and the nucleus (25). There was no difference in degree of K-PPn for gTop1-13Myc extracted, under denaturing conditions, from cells expressing empty vector or vt-Ppx1 (Fig. 7E). This result suggests that Top1 is endogenously Lys-polyphosphorylated by nuclear polyP and is not sensitive to post-extraction addition of vacuolar polyP. The scenario is different for Nsr1. The mobility shift of Nsr1 is considerably reduced, but still detectable, in WT yeast expressing vt-Ppx1 (Fig. 7F, indicated with *). This suggests that the nuclear Nsr1 is endogenously modified by the polyP present in the nucleus. Moreover, Nsr1 is more sensitive to post-extraction contact with vacuolar polyP than Top1 (the shift labeled **). By degrading vacuolar polyP with vt-Ppx1 expression, we had eliminated one source of post-extraction skewing of results. We then considered the highly-active Malachite green assay (%) ppn1'ppn2'ppx1' EDITORS' PICK: Characterization of lysine polyphosphorylation polyphosphatases Ppn1 and Ppn2 (Fig. 6). To gain the truest representation of Nsr1 physiological K-PPn levels, we overexpressed vt-Ppx1 in a DDY1810 ppn1⌬ppn2⌬ strain, herein called DDM. We confirmed that the overexpression strain was devoid of vacuolar polyP by measuring total polyP levels (ϳ26%) when compared with the same strain expressing the empty vector (Fig. 8A). As before, we performed nuclear fractionation and quantified the nuclear polyP. Like WT, we observed that there is a 40% reduction of nuclear polyP in the DDM cells expressing vt-Ppx1 (Fig. 8B). In this case, we verified that there was some vacuolar and significant endoplasmic reticulum contamination, noted by the presence of Vph1 (subunit a of vacuolar-ATPase V0 domain) and vt-Ppx1, respectively, in the nuclear fraction, likely explaining the 40% reduction in nuclear polyP levels (Fig. 8C). We investigated Top1 and Nsr1 mobility in total and nuclear protein extracts from these cells (Fig. 8C). As for the DDY1810 WT strain, gTop1-3HA mobility was not affected in the DDM strain expressing vt-Ppx1. Con-sistent with our previous analysis, Nsr1 was more sensitive than Top1 to the presence of Ppn1 and Ppn2, as its mobility shift was greater than when Ppn1 and Ppn2 are absent (Fig. 8C) than in WT (Fig. 7F). This likely represents the physiological extent of Nsr1 K-PPn. By eliminating both vacuolar polyP and vacuolar polyphosphatases, the yeast ppn1⌬ppn2⌬ (DDM) expressing vt-Ppx1 is the ideal strain in which to study K-PPn of target proteins.

Vacuolar polyP exacerbates polyphosphorylation during protein extraction
K-PPn is a nonenzymatic PTM that results in the covalent attachment of polyP to lysine residues localized in PASK domain-containing proteins. This modification has been mostly characterized in budding yeast (5,26). Yeast possess a unique polyP physiology; it accumulates inside the acidocalci-

EDITORS' PICK: Characterization of lysine polyphosphorylation
some-like (27) vacuole, and it is actively processed by vacuolar polyP phosphatases. This distinctive polyP metabolism coupled with the nonenzymatic nature of K-PPn raises a few questions that our study aimed to address. The polymeric nature of polyP, with a size that ranges from ten to hundreds of phosphate residues in length, suggests that the covalent association of polyP to target proteins would result in an upward smear as seen, for example, with polyubiquitination. However, often this PTM results in a discrete upward mobility shift on NuPAGE, with only one band seen. K-PPn, as with virtually all nonenzymatic PTMs, occurs when the reactive metabolite, in this case polyP, is in direct contact with its targets. Importantly, 80% of yeast polyP resides in the vacuole and is not physiologically available to modify cytoplasmic and nuclear targets. Here, we confirmed that during cell lysis the polyP released from vacuoles can polyphosphorylate not only nonmodified Nsr1 but also increases the mobility shift of previously polyphosphorylated Nsr1. We showed that this occurs by modification of further lysine residues by subsequently added polyP. The initial K-PPn event may lead to an exposure and availability of further tar-getable lysine residues, perhaps suggesting that K-PPn has a denaturing effect on target proteins. This hypothesis agrees with the observation that K-PPn Top1 is inactive in its capacity to relax supercoiled DNA (5). The ability of vacuolar polyP to successively modify lysines, perhaps until the protein is saturated, could explain why K-PPn in yeast does not lead to a smear on NuPAGE; the polydisperse nature of this modification is especially difficult to define at very high molecular weights.

PASK domain's physicochemical properties are essential for polyphosphorylation
The PASK domain is characterized by an abundance of the acidic glutamic and aspartic acids, lysines, and serines. Serine residues do not seem to be directly or indirectly required for the modification to occur, because mutagenizing them all to alanine within the PASK domain did not change the capacity of Top1 to be K-PPn (5). When mutagenized individually, glutamic and aspartic acids also appeared not to be important for K-PPn. However, when mutagenized all together with the neutral amino acids alanine and leucine, respectively, Top1 was no A.

EDITORS' PICK: Characterization of lysine polyphosphorylation
longer Lys-polyphosphorylated by polyP, despite maintaining its nuclear localization. Possibly, these acidic amino acids could coordinate polyP, aided by divalent cations such as Mg 2ϩ , to position the polymer near the target lysine for K-PPn. Upon neutralization of the negative charge of the acidic amino acids through mutagenesis, this role played by the side chain negative charges no longer exists, and the protein cannot be Lys-polyphosphorylated. This resembles other modifications, such as sumoylation, for example, where the acidic residues surrounding the Ubc9 consensus region enhance the efficiency and specificity of the modification (28).

Vacuolar polyphosphatases Ppn1 and Ppn2 negatively affect polyphosphorylation of nuclear proteins upon extraction
In yeast, there are four known polyP phosphatases, Ppn1 and Ppn2 (14,16,29), Ddp1 (30), and Ppx1 (18). Both the exopolyphosphatase Ppx1 and the endopolyphosphatase Ddp1, which are present in the nucleus and the cytoplasm, have previously been shown to hydrolyze the polyP of K-PPn target proteins Nsr1 and Top1 (5). Interestingly, under native conditions, Ppx1 shows a much higher affinity for K-PPn Nsr1 than Top1, both in vitro (5) and in vivo. One possible explanation is that the conformation of K-PPn Nsr1 exposes polyP more on the surface allowing for easier access of Ppx1 than in the case of Top1. The vacuolar-localized Ppn1 requires protease processing to become active (14), and in a yeast background that has the Pep4 protease deleted, DDY1810, Nsr1, and Top1 show a highermobility shift. We therefore investigated the actual effect of the vacuolar polyphosphatases Ppn1 and Ppn2 in the hydrolysis of K-PPn targets. Despite their vacuolar localization and therefore not in direct contact with cytoplasmic and nuclearly-localized K-PPn targets in vivo, upon cell lysis, the very active Ppn1 was shown to rapidly hydrolyze target proteins. Deletion of Ppn2 alone did not dramatically change the mobility of protein targets, but its deletion in conjunction with Ppn1 had a synergistic effect on K-PPn. This demonstrates that both can hydrolyze K-PPn targets but that Ppn1 has a much higher affinity for these targets. Our results also demonstrate that besides the polyphosphatases studied here, there are others that also hydrolyze phosphorylated target proteins, as seen by loss of the downward mobility shifts of Nsr1 in the DQM under native conditions (Fig. 4B).

Generation of an ideal strain to study polyphosphorylation in yeast
We have demonstrated that in yeast, upon cell lysis, the abundant vacuolar polyP becomes available to K-PPn target proteins present in the cytoplasm and the nucleus. At the same time, there is also the release from the vacuole of the very active endopolyphosphatases, Ppn1 and Ppn2, having the opposite effect, i.e. hydrolysis of K-PPn target proteins. For these reasons, during extraction in WT, the tug-of-war between these two effects results in a mobility shift on NuPAGE of K-PPn target proteins that might not truly reflect the physiological conditions. Despite this drawback, yeast is still the best eukaryotic model in which to study K-PPn. It is a genetically tractable model organism in which the whole biosynthetic pathway for  (n ϭ 4). B, nuclear polyP of the same strains as in A. PolyP was run as in A but with polyP corresponding to 40 g of RNA (n ϭ 3 (two biological replicates in which one of the biological replicates has two technical replicates)). C, comparison of Top1 and Nsr1 mobility in total protein extracts (left panel) and nuclear protein extracts (right panel, n ϭ 2) from gTop1-3HA-tagged vtc4⌬ and ppn1⌬ppn2⌬ (DDM) overexpressing vt-Ppx1 or empty vector. Protein samples were extracted under denaturing conditions, run on NuPAGE, and blotted with anti-HA (to detect Top1), anti-Nsr1, anti-VPH1 (as a vacuolar marker control), anti-Ppx1, and anti-histone 3 (as a nuclear marker control). All yeast strains are in a DDY1810 background. The figures presented are a representation of at least three independent repeats unless otherwise stated.

EDITORS' PICK: Characterization of lysine polyphosphorylation
the synthesis and hydrolysis of polyP is known and well-described. Next, yeast contains high levels of polyP, making the only readout so far available to detect K-PPn of targets proteins, the mobility shift on NuPAGE, possible. Higher eukaryotes not only contain much less polyP, but also the enzyme(s) responsible for its synthesis are unknown, making mammalian cell lines an unfriendly model system in which to study K-PPn events for the time being. Taking all this into consideration, an ideal model in which to study this new PTM would be a yeast strain devoid of vacuolar polyP and deleted of Ppn1 and Ppn2. Therefore, we have targeted the exopolyphosphatase Ppx1 to the vacuole in ppn1⌬ppn2⌬ DDY1810 strain and ascertained that it had lost its vacuolar polyP but retained the nuclearly-localized polyP. With only the nuclear polyP present and in the absence of the highly-reactive Ppn1 and Ppn2, we observed that Top1 mobility remains unaltered, implying that it is physiologically fully K-PPn. Interestingly, Nsr1 behavior is different; this target is a lot more sensitive to the presence of vacuolar polyP and of the vacuolar polyphosphatases. Under physiological conditions, Nsr1 might not be fully phosphorylated and upon exposure to extra polyP from the vacuole during the extraction, it might be further modified. Importantly, nuclear preparations from ppn1⌬ppn2⌬ DDY1810-expressing Ppx1 in the vacuole revealed a substantial K-PPn induced Nsr1 mobility shift, highlighting the importance of this newly-created yeast strain to study K-PPn physiological events. The demonstration that nuclear polyP is insensitive to the action of vacuole-localized Ppx1 suggests the existence of two distinct and separate polyP pools. This raises the question of the origin of the polyP nuclear pool. As mentioned in the Introduction, in yeast, polyP is synthesized by the VTC complex mostly by the subcomplex composed of the subunits Vtc1/3/4/5 (32), which is localized in the cytosolic side of the vacuole membrane (23). Upon synthesis, polyP is translocated and accumulates in the vacuole lumen (23). However, the existence of a second VTC subcomplex, Vtc1/2/4/5 (32), which localizes to the endoplasmic reticulum in regions adjacent to the nucleus, suggests a possible direct synthesis of polyP in the nucleus. Another possibility that cannot be excluded is that a portion of newly-synthesized polyP is released in the cytosol and transported into the nucleus complexed with proteins, similarly to ribonucleoprotein transport of mRNA (33). More studies are required to verify these hypotheses and to define how the nuclear pool of polyP is generated. In the meantime, our freely available strain (ppn1⌬ppn2⌬ DDY1810-expressing vt-Ppx1) will allow us and the community to study protein K-PPn of cytosolic and nuclear targets without the interfering effects, observable upon cell lysis, of the abundant vacuole polyP pool and of the active vacuole Ppn1 and Ppn2 polyphosphatases. Therefore, studies in this strain are closer to the physiological context.

Yeast strains and culture media
S. cerevisiae strains used in this study are isogenic to DDY1810 or BY4741 and are listed in Table S1. Genes were deleted by replacing the entire ORF with the marker cassettes (see Table S2 for PCR primers) (34). The DDY1810 has a dele-tion on the Pep4 protease that is coincidently required for processing and activating the endopolyphosphatase that cleaves polyP internally, Ppn1 (14). C-terminal genomic insertions for Top1, gTop1-3HA (DDY1810) and gTop1-9Myc (BY4741), were obtained by standard procedures according to Ref. 34 using primers in Table S2. Correct recombination was confirmed by PCR, and correct protein expression was confirmed by Western blotting with anti-HA and Myc antibodies, respectively.
Please note that in WT BY4741, overexpression of vt-Ppx1 did not change total polyP levels, whereas in WT DDY1810 it abolished vacuolar polyP, and therefore this strain was used for the experiments.

Expression constructs
Plasmids for yeast expression were cloned with the oligonucleotides described in Table S2 as follows. 1) pCA82-pYes3-ADH::GFP(TRP) was cloned by digesting pYes3-ADH with SpeI and NotI to remove the Trp gene and insert it into pCA58 (pYes-pADH:GFP(URA); (3)) digested with the same enzymes.

Antibodies
The antibodies used in this study were as follows:

Protein extraction
Yeast cells grown overnight in appropriate minimal media were diluted to OD 600 ϭ 0.8 and grown for 90 min at 30°C in 10 ml of fresh media. Cells were pelleted through centrifugation at 4100 ϫ g for 3 min, washed once in ice-cold MilliQ water, and frozen at Ϫ20°C. To avoid depolyphosphorylation, it is essential to work quickly, maintain cold cell extracts throughout the entire procedures, use a wide range of phosphatase inhibitors, and keep incubation times to a minimum. Unless otherwise stated, proteins were extracted in ice-cold lysis buffer (LB: 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM DTT, phosphatase inhibitor mixture 1; phosphatase inhibitor mixture 1 (PhIC1): 100ϫ: 200 mM imidazole, 100 mM sodium fluoride, 115 mM sodium molybdate, and 100 mM sodium orthovanadate; phosphatase inhibitor mixture 2 (PhIC2): 1000ϫ: 2.5 mM (Ϫ)-pbromotetramisole oxalate and 0.5 mM cantharidin; and protease inhibitor mixture (PrIC): from Sigma catalog no. P8215). Cells were vortexed in the presence of glass beads for 30 s to 1 min and homogenates centrifuged at top speed for 5 min at 4°C. Supernatants were immediately boiled in 4ϫ SDS sample buffer (0.25 M Tris-HCl, pH 6.8, 8% SDS (w/v), 30% glycerol, 0.02% (w/v) bromphenol blue) for 5 min. Proteins were run on 4 -12% NuPAGE gels (Invitrogen) and transferred to polyvinylidene difluoride membranes, and Western blotting was performed with the appropriate antibodies. When proteins were extracted under denaturing conditions, two different buffers were used: 2% SDS (LB containing 2% SDS) and 8 M urea buffer (8 M urea, 0.05 M sodium phosphate buffer, pH 7.6, and 0.14 M ␤-mercaptoethanol). Extraction procedure was as described above.
For Myc immunoprecipitations, equilibrated beads, Mycagarose (Sigma A7470), were incubated with normalized protein extracts (500 l of protein extract (1 g/l) with 50 l of 50% (v/v) beads) with head-over-tail rotation for 1 h at 4°C. The mixture was centrifuged at 369 ϫ g for 1 min and washed four times for 10 min at 4°C in LB containing 0.75% Triton X-100. Beads were either equilibrated twice in LB (without detergent) and kept at 4°C until further use (no more than a week) or were resuspended in 2ϫ SDS sample buffer and boiled. Proteins were run on 4 -12% NuPAGE gels as above.
To extract total polyP, yeast cells grown overnight were diluted to OD 600 ϭ 0.8 and grown in 10 ml of SC minimal media for 90 min at 30°C. Cells were pelleted at 4122 ϫ g for 3 min and washed twice in 1 ml of cold MilliQ water. The cell pellet was resuspended in 250 l of LETS buffer (0.1 M LiCl, 10 mM EDTA, 10 mM Tris-HCl, pH 7.4, and 0.2% SDS) and 250 l of acidic phenol, pH 4.3 (Sigma; P4682), was added together with the corresponding of 25 l of glass beads (Sigma; G8772). Cells were turbo-vortexed for 5 min at 4°C and spun at top speed for 5 min. The top layer was transferred to a fresh tube and extracted with 500 l of chloroform through turbo-vortexing (5 min at 4°C) and spun as above. The top layer was transferred to a fresh tube and precipitated with 2.5ϫ ethanol for at least 2 h at Ϫ20°C. Extract was spun as above; supernatant was aspirated, and samples were air-dried before being resuspended in 200 l of resuspension buffer (RB: 0.1% SDS, 1 mM EDTA, 10 mM Tris-HCl, pH 7.4). Quantification of RNA in polyP extract was determined by reading absorbance at 260 nm in the nanodrop. PolyP was analyzed by PAGE and visualized by negative staining with DAPI according to previously described methods (30).
A similar procedure was used to extract nuclear polyP. Purified nuclei were resuspended in 300 l of LETS buffer, passed through a 0.2-m needle four times to shear the DNA, and 300 l of acidic phenol, pH 4.3, was added. Sample was turbo-vortexed without glass beads following the same procedure as above.

Shift-up assay
The shift-up experiments were performed in different ways (refer to figure legends for details). The sources of polyP used throughout this assay were as follows: (a) purified from yeast polyP (section polyP extraction); (b) synthetic polyP of an average size of 45 (P45; Sigma, S4379), 65 (P65; Sigma, S6253), and 100 (P100; kindly provided by Dr. Toshikazu Shiba, RegeneTiss Co., Japan), and bis-FAM-P8 (20); (c) nonpurified, in this case yeast cultures from vtc4⌬ and ones containing polyP were mixed in a 1:1 ratio (OD 600 ϭ 0.8), spun at 4100 ϫ g, washed once in ice-cold water, and extracted as described above under "Protein extraction." The substrates for this assay were as follows: 1) protein extracts from the different panel of yeast mutants; and 2) unmodified purified proteins, purified from vtc4⌬ on Myc-agarose beads or GFP-trap (Chromotek-gta-10; refer to figure legends). When the assay was performed with total protein extracts, 20 l of protein extract (1 g/l) was incubated with 5 l of 10 mM P45, P65, or P100 (2 mM) for 20 min at 30°C, and 4ϫ SDS sample buffer was added, samples were boiled, resolved by NuPAGE, and immunoblotted with appropriate antibodies. When the assay was performed on Myc-agarose or GFP-trap-purified proteins beads, polyP was incubated with 15 l of bead bed volume in a total volume of 50 l of RB buffer for 20 min at 30°C with head-over-tail rotation, in the following concentrations: 5 l of 10 mM P45, P100 (2 mM), 0.5 mM bis-FAM-P8 or 2 l of total yeast polyP (1.5 g/l of RNA). After incubation, beads were washed three times in LB buffer, boiled in 2ϫ SDS sample buffer, resolved by NuPAGE, and immunoblotted with appropriate antibodies.

PolyP quantification by malachite green assay
We used the ability of malachite green, molybdate, and free orthophosphate to form a complex that absorbs light of a wavelength between 620 and 640 nm to determine polyP amounts in yeast. PolyP corresponding to 0.8 g of RNA was converted to orthophosphate by incubation by boiling in 250 l of 0.3 M perchloric acid for 30 min. Samples were cooled on ice for 2 min, and 250 l of malachite green reagent (15 mg of malachite green, 100 mg of sodium molybdate, 25 l of Triton X-100, 2.82 EDITORS' PICK: Characterization of lysine polyphosphorylation ml of 38% hydrochloric acid in 50 ml water) was added. Samples were incubated 30 min at 30°C; 150 l was dispensed in triplicate on a 96-well plate, and absorbance was measured on a Tecan plate-reader fluorimeter at 640 nm. The concentration of hydrolyzed orthophosphates was compared with a P i standard calibration curve. Each sample was measured in triplicate, and the baseline blank was always the reading from vtc4⌬. Each experiment is an average of at least three biological replicates.

S. cerevisiae nuclei fractionation
Nuclei fractionation was adapted from the Hahn laboratory at Fred Hutchinson Cancer Research Center. In summary, yeast cells were grown in 2 liters of YPD (Formedium), supplemented with 3% (w/v) glucose at 30°C until OD 600 ϭ 3-5 (ϳ30 g of cells). When yeast cells had been transformed with pGDP-vt-Ppx1, cells were grown the same way but in minimal media containing 3% (w/v) glucose and without uracil. Cells were pelleted at 4100 ϫ g for 10 min, resuspended in 35 ml of 50 mM Tris-HCl, pH 7.5, and 30 mM DTT, and incubated 15 min at 30°C. Cells were pelleted at 4100 ϫ g for 10 min and resuspended in 20 ml of YPD containing 1 M sorbitol (YPDS), and an additional 15 ml of 2 M sorbitol was added. Cells were transformed into spheroblasts with 200 l of zymolyase (USBiological, Zymolyase 100T-Z1004) and incubated at 30°C with gentle occasional swirling. When 80% of the cells were transformed into spheroblasts, 100 ml of YPDS was added, and cells were pelleted at 4100 ϫ g for 10 min. Cells were resuspended in 250 ml of YPDS and allowed to recover at 30°C for 30 min. Cells were pelleted at 4100 ϫ g for 10 min and washed twice in 200 ml of ice-cold YPDS. Cells were pelleted at 4100 ϫ g at 4°C for 10 min and resuspended in 250 ml of ice-cold 1 M sorbitol. Cells were pelleted as above and resuspended in 100 ml of Buffer N (25 mM K 2 SO 4 , 30 mM Hepes, pH 7.6, 5 mM MgSO 4 , 1 mM EDTA, 10% glycerol, 0.5% Nonidet P-40, 7.2 mM spermidine, 3 mM DTT, and 1:500 PrIC). Cells were subjected to 4 -6 strokes in a type A "tight" Dounce on ice, transferred to 50-ml centrifuge tubes, and pelleted at 230 ϫ g for 10 min at 4°C. Supernatant was recovered and spun again at 390 ϫ g for 10 min at 4°C. Supernatant was transferred to a clean centrifuge tube and spun at 1910 ϫ g for 10 min at 4°C. Supernatant was further pelleted at 4303 ϫ g for 10 min at 4°C to recover the nuclei. Nuclei were frozen at Ϫ80°C.

FACS
Cells were grown in SC minimal medium to logarithmic phase. Just before measurement, 50 l of the cell suspension was added to 950 l of TBS, vortexed, and immediately analyzed using an LSRII flow cytometer (BD Biosciences). 20,000 events were recorded for each sample, and the mean fluorescence value was used to estimate protein expression levels.

Immunofluorescence and fluorescence microscopy
To determine vt-Ppx1 localization, in ppn1⌬xppn2⌬ gTop1-3HA-tagged cells expressing GPD-vt-Ppx1 we performed immunofluorescence with anti-Ppx1 antibody. Cells were grown overnight in Sc media without uracil at 30°C, diluted in fresh media to OD 600 0.6, and grown for 90 min. Cells were spun at 4100 ϫ g for 3 min, resuspended in 2 ml of softening buffer (50 mM Tris-HCl, pH 7.5, and 30 mM DTT), and incubated for 15 min at 30°C. Cells were pelleted at 4100 ϫ g for 3 min, resuspended in 2 ml of a solution containing 2:1.5 of YPDS and 2 M sorbitol and 20 l of zymolyase (USBiological, Zymolyase 100T-Z1004), and incubated at 30°C with occasional gentle swirling. When 80% of the cells were transformed into spheroblasts, formaldehyde was added to 4%, and cells were further incubated for 60 min at 30°C. Cells were spun at 390 ϫ g for 2 min, carefully resuspended in 0.5 ml of Buffer A (100 mM Tris-HCl, pH 8, 1 M sorbitol) with 0.1% (v/v) SDS, and incubated at room temperature for 10 min. Cells were spun as above, gently washed twice in Buffer A, resuspended in 0.5 ml of Buffer B (50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.1% (v/v) Tween 20, 1% (w/v) nonfat dried milk, 0.5 mg/ml BSA), and incubated at room temperature 15 min to block. After this period, 1 l of anti-Ppx1 (1:500) antibody was added, and cells were incubated for 90 min at room temperature with head-to-tail rotation. Cells were spun at 390 ϫ g for 2 min and washed five times in Buffer B. After the last wash, 0.5 l of mouse Alexa 488 antibody was added to 500 l of cell suspension and incubated for 60 min in the dark at room temperature with head-over-tail rotation. Cells were washed as above with two extra washes with PBS, incubated in PBS (2.5 g/ml Hoechst) for 20 min in the dark, and washed twice again in PBS. Cells were mounted on microscope slides and observed in a spinning disc confocal microscope-Ultra VoxII with a ϫ100 CFI Plan Apochromat VC 1.4 lens and a Hamamatsu C9100 -13 EMCCD camera. Pictures were collected using Velocity software (PerkinElmer Life Sciences) and treated with ImageJ software.
For observation of the polyP phosphatase localizations, cells were grown overnight in Sc minimal media at 30°C, diluted to OD 600 0.6 in fresh media, grown for 60 min, after which DAPI (2.5 g/ml) was added, and cells were grown for an extra 30 min before being harvested by centrifugation. Cells were resuspended in 50 l of the same media and mounted onto a microscopy slide. Cells were observed on a widefield microscope, Zeiss Axio Imager with an EC Plan Neofluar ϫ100 1.3 NA lens, and an QImaging Retiga EXi grayscale CCD camera. Pictures were collected using OpenPlan software and treated with ImageJ software.

Statistics
Tests for significance of observed differences were performed by an unpaired Student's t test, and error bars represent the standard deviation. p values differences higher than 0.1 were considered as not significant.