Research article

Inositol pyrophosphates modulate hydrogen peroxide signalling

Sara Maria Nancy Onnebo, Adolfo Saiardi


Inositol pyrophosphates are involved in a variety of cellular functions, but the specific pathways and/or downstream targets remain poorly characterized. In the present study we use Saccharomyces cerevisiae mutants to examine the potential roles of inositol pyrophosphates in responding to cell damage caused by ROS (reactive oxygen species). Yeast lacking kcs1 [the S. cerevisiae IP6K (inositol hexakisphosphate kinase)] have greatly reduced IP7 (diphosphoinositol pentakisphosphate) and IP8 (bisdiphosphoinositol tetrakisphosphate) levels, and display increased resistance to cell death caused by H2O2, consistent with a sustained activation of DNA repair mechanisms controlled by the Rad53 pathway. Other Rad53-controlled functions, such as actin polymerization, appear unaffected by inositol pyrophosphates. Yeast lacking vip1 [the S. cerevisiae PP-IP5K (also known as IP7K, IP7 kinase)] accumulate large amounts of the inositol pyrophosphate IP7, but have no detectable IP8, indicating that this enzyme represents the physiological IP7 kinase. Similar to kcs1Δ yeast, vip1Δ cells showed an increased resistance to cell death caused by H2O2, indicating that it is probably the double-pyrophosphorylated form of IP8 [(PP)2-IP4] which mediates the H2O2 response. However, these inositol pyrophosphates are not involved in directly sensing DNA damage, as kcs1Δ cells are more responsive to DNA damage caused by phleomycin. We observe in vivo a rapid decrease in cellular inositol pyrophosphate levels following exposure to H2O2, and an inhibitory effect of H2O2 on the enzymatic activity of Kcs1 in vitro. Furthermore, parallel cysteine mutagenesis studies performed on mammalian IP6K1 are suggestive that the ROS signal might be transduced by the direct modification of this evolutionarily conserved class of enzymes.

  • inositol pyrophosphate
  • inositol hexakisphosphate kinase (IP6K)
  • Kcs1
  • metabolism
  • Rad53
  • reactive oxygen species (ROS)


Inositol pyrophosphates in both lower eukaryotes such as yeast as well as mammalian cells are generated by the sequential phosphorylation of IP3 (inositol trisphosphate) [1,2] (Figure 1). The IPMK (inositol polyphosphate multi-kinase; Saccharomyces cerevisiae Arg82 or Ipk2) metabolizes IP3 to IP4 (inositol tetrakisphosphate) and IP5 (inositol pentakisphosphate) [35]. Subsequent phosphorylation of IP5 by IPK1 (IP5 2-kinase) generates IP6 (phytic acid or inositol hexakisphosphate). The fully phosphorylated ring of IP6 is then further phosphorylated by Kcs1 [S. cerevisiae homologue of mammalian IP6K (inositol hexakisphosphate kinase)] and Vip1 {S. cerevisiae homologue of mammalian PP-IP5K or IP7K [IP7 (diphosphoinositol pentakisphosphate) kinase]} to generate the inositol pyrophosphates IP7 and the double-pyrophosphorylated form of IP8 [also referred to as (PP)2-IP4 (bisdiphosphoinositol tetrakisphosphate)] respectively (Figure 1) [4,68]. It is noteworthy that a triphosphate form of ‘IP8’ (referred to as PPP-IP5) has recently been identified, which is produced by the direct action of IP6Ks [9]. Inositol pyrophosphates have been linked to a variety of biological functions ranging from vesicular trafficking to apoptosis and telomere length maintenance (recently reviewed in [1,10]). The most thrilling characteristic of IP7 is its recently identified ability to pyrophosphorylate proteins in a kinase-independent manner [10]. However, only a handful of IP7 targets have been identified so far, and the signalling pathways involved remain largely unknown. It has previously been reported that expression of the mammalian IP6Ks (IP6K1/2/3) in a range of mammalian cells increases their sensitivity to DNA-damaging agents, whereas knockdown of IP6K2 resulted in protection against apoptosis [1113].

Eukaryotic cells are constantly exposed to exogenous and endogenous agents that cause DNA damage. This can lead to genomic instability, cell death or genetic anomalies such as neurodegenerative diseases and cancer. To monitor and respond to DNA damage, cells have developed sophisticated checkpoint controls as surveillance mechanisms [14]. These are evolutionarily conserved and consist of sensor proteins that recognize physical alterations in DNA and transducers that transmit the damage signal to downstream effectors [15]. When DNA checkpoints are activated by DNA lesions or replication blocks they promote transcription of DNA repair genes, delay cell-cycle progression or initiate apoptosis. In S. cerevisiae the sensor and signalling functions are mainly performed by the PI3K (phosphoinositide 3-kinase)-like kinases Mec1 and Tel1 [S. cerevisiae homologues of mammalian ATM (ataxia telangiectasia mutated) and ATR (ataxia telangiectasia mutated- and Rad3-related)]. Once activated, Mec1/Tel1 triggers a phosphorylation cascade that activates Rad53 (S. cerevisiae homologue of mammalian Chk2) [16,17]. A major source of spontaneous DNA damage is caused by ROS (reactive oxygen species), which can be formed endogenously as by-products of aerobic respiration or other cellular processes [18]. ROS is a general term that includes hydrogen peroxide (H2O2), superoxide anion (O2) and the hydroxyl radical (HO). These molecules are generally regarded as toxic compounds because of their chemical reactivity. However, it is also extensively recognized that ROS have important roles as intracellular signalling molecules [1922].

Figure 1 Yeast inositol pyrophosphate metabolism

IP3 is sequentially phosphorylated to IP4, IP5 and IP6 by serial enzymatic reactions carried out by IPMK and IPK1. The fully phosphorylated myo-inositol ring of IP6 is further phosphorylated by Kcs1 (S. cerevisiae homologue of mammalian IP6K) to generate the inositol pyrophosphate IP7. The double-pyrophosphate form (PP)2-IP4 of ‘IP8’ is generated by Vip1 (S. cerevisiae homologue of mammalian PP-IP5K or IP7K). Two other inositol pyrophosphate species, PP-IP4 and (PP)2-IP3 can be generated from IP5. The kinases that carry out each of these steps are shown above the arrows. A triphosphate form of ‘IP8’ has recently been identified and it is likely to be synthesized in yeast by Kcs1.

In mammalian cells overwhelming evidence indicates that the transient production of H2O2 is triggered by ligand-mediated receptor activation, including growth factors such as PDGF (platelet-derived growth factor) [23] or EGF (epidermal growth factor) [24]. These observations were instrumental in establishing H2O2 as a second messenger [19]. It is now accepted that, at low concentrations, H2O2 is an important signalling molecule which controls several physiological processes, whereas, at high concentrations, it is regarded as a highly toxic molecule [20,25]. Consequently, spatial and temporal regulation of H2O2 production and degradation ensure a positive role for this molecule as an intracellular messenger [19,26]. Cells have in fact evolved several mechanisms for removing H2O2, including catalase, glutathione peroxidase and peroxiredoxin, as a means of protecting against oxidatively induced death [27].

In yeast, ROS signalling is usually studied via the exposure to exogenous agents such as H2O2, which result in DNA base modification, single- and double-strand breaks, and the formation of apurinic/apyrimidinic lesions [2830]. Limited DNA damage will trigger a DNA-checkpoint response promoting DNA repair mechanisms. In contrast, massive DNA damage will lead to cell death. In the present study we have used yeast to characterize the IP7-mediated response to ROS. As a DNA-damaging agent, we used sublethal concentrations of H2O2, which activates the DNA-damage response pathway [31]. The results of the present study indicate that yeast mutants which are deficient in inositol pyrophosphates possess a higher threshold of resistance to the lethal effects of H2O2, but not to other DNA-damaging agents. This specific resistance to H2O2 correlates with a sustained activation of Rad53 and consequently a privileged establishment of a DNA repair mechanisms. We further report that H2O2 regulates higher inositide metabolism, causing a rapid decrease in cellular levels of inositol pyrophosphates.



Antibodies against Rad53 and Rad9 were purchased from Santa Cruz Biotechnology. The antibody against Hsp60 (heat-shock protein 60) was purchased from Stressgen. Yeast media were from Formedium, and all other reagents were from Sigma–Aldrich. NuPage 4–12% Bis-Tris gels (Invitrogen) were used for protein analyses.

Strains, plasmids and media

Yeast (S. cerevisiae) strains used in the present study are listed in Table 1 and were derived from the BY4741 (MATa, his3Δ1; leu2Δ0; met15Δ0; ura3Δ0) or the protease-deficient DDY1810 (MATa; MATa leu2-3,112 trp1901 ura3-52 prb1-1122 pep4-3 prc1-407) genetic backgrounds. Null-mutant yeast of the DDY1810 background where generated using standard homologous recombination techniques [32]. The yeast TAP (tandem affinity purification)-tagged Kcs1 strain was obtained from Open Biosystems. Mouse IP6K1 and kinase-dead IP6K1-K(226)/A [33] were sub-cloned into the pYes2NTA vector (Invitrogen). Cells were grown in SC (synthetic complete) medium containing 2% glucose at 30 °C with aeration, and were plated on SC plates unless stated otherwise. Cells containing expression plasmids were grown in SC medium lacking uracil under the same experimental conditions. Analysis of inositol polyphosphate HPLC profiles, revealed a rescue of normal inositol pyrophosphate levels when using the pYes2-IP6K1 construct in glucose-containing medium; no rescue was observed when the mutant form pYes2-IP6K1KA was utilized.

View this table:
Table 1 Yeast strains used in the present study

Colony-forming assay

To observe the effect of H2O2 in exponentially growing cultures, yeast were grown to exponential-phase in SC medium at 30 °C and were treated with various concentrations of H2O2. To assess the viability, the cultures were serially diluted and plated on to SC plates. Untreated samples were used as a standard for calculating the percentage viability.

DNA-damage response activation

Cultures (2–4 units at a D600 of 0.2–0.7) in mid-exponential phase were treated with various concentration of H2O2 for the indicated times at 30 °C. Cells were collected by centrifugation (10000 g for 5 min at 4 °C) and rapidly lysed in lysis buffer [20 mM Hepes (pH 6.8), 1 mM EGTA, 1 mM EDTA, 0.1% CHAPS, 5 mM NaF, 5 mM DTT (dithiothreitol), 200 mg/l PMSF and proteinase inhibitor cocktail at a 1:500 dilution]. For Western blot analysis, 10% of the extracts, approx. 20 μg of protein, were used for blotting with anti-Rad53 (Santa Cruz Biotechnology, sc-6748) or anti-Rad9 (Santa Cruz Biotechnology, sc-6740) antibodies. For normalization of Western blotting, the membrane was probed with an anti-Hsp60 (Santa Cruz Biotechnology, sc-13966) antibody or alternatively the blot was stained with Coomassie Blue.

Mutation frequency assay

The CanR (canavanine resistance) assay was performed essentially as described previously by Madia et al. [34]. Briefly, yeast cultures were grown to exponential phase in SC medium at 30 °C. The cultures were treated with 1 mM H2O2 for 3 h, washed with water and approx. 1×107 cells were spread on argininedeficient plates containing 60 mg/ml L-canavanine. Serial dilutions of samples from the same cultures were spread on to plates containing only minimal medium in order to calculate the exact number of cells/ml. After 4 days of incubation, the number of colonies was counted and the mutation frequency was calculated as mutations/106 cells.

Generation of mouse IP6K1 mutants

The bacterial expression plasmid harbouring GST (glutathione transferase)–IP6K1 has been previously described [7]. The conserved cysteine mutants were generated by PCR technology using the following oligonucleotides and their complementary partners: C48A (5′-GACCACACTGTGGCCAAACCTCTCATC-3′), C187A (5′-TGGAGCCTGCGGGCCCACAAGCAGCAG-3′), C221A (5′-TTCAAGTACCCCGCGGTGCTGGACTTG-3′), C221D (5′-TTCAAGTACCCCGATGTGCTGGACTTG-3′), C248A (5′-CAGATGAGGAAGGCTGAGCAGAGCACG-3′) and C261A (5′-GGGGTCAGGGTCGCCGGCATGCAGGTG-3′). The mutagenesis was confirmed by restriction analysis or by direct sequencing. Purification of His-tagged proteins was performed as previously described [7].

IP6K assay

Purification of TAP protein was performed as previously described [35]. Briefly, yeast Kcs1–TAP cells (100 ml at a D600 of 0.8–1.2) were harvested by centrifugation (10000 g for 5 min at 4 °C) and washed with ice-cold water. Cells were homogenized using glass beads for 5 min in 5 ml of ice-cold binding buffer [50 mM Hepes (pH 7.4), 150 mM NaCl, 10 mM 2-mercaptoethanol, 1 mM magnesium acetate, 1 mM imidazole, 2 mM CaCl2 and 2 mM NaF]. The homogenate was centrifuged at 15000 g for 20 min. Calmodulin–Sepharose 4B (40 μl; Amersham) was added to the supernatant and incubated at 4 °C for 2 h. The resin was then washed twice with binding buffer containing 500 mM NaCl and then twice with 100 mM NaCl. The resin was then used to perform IP6K reactions using [3H]IP6 (Amersham) as previously described [4], and reactions were analysed by SAX (strong anion exchange)-HPLC.

Determination of yeast inositol polyphosphate levels

The studies of yeast inositol polyphosphate levels, were performed as previously described [36] with a few modifications. Yeast were grown in synthetic medium without inositol and supplemented with 75 μCi/ml [3H]inositol (PerkinElmer NEN). After drug treatment, the yeast were harvested by centrifugation (10000 g for 5 min at 4 °C), washed with ice-cold water, and lysed in 0.3 ml of ice-cold lysis buffer (1 M perchloric acid and 3 mM EDTA) with 2 min of glass bead beating. Lysates were centrifuged (10000 g for 5 min at 4 °C) and neutralized with neutralization solution (1 M potassium carbonate and 3 mM EDTA). Inositol phosphates were resolved by SAX-HPLC using a 4.6 mm×125 mm PartiSphere SAX column (Whatman). The column was eluted with a gradient generated by mixing buffer A (1 mM Na2EDTA) and buffer B [buffer A plus 1.3 M (NH4)2HPO4 (pH 3.8 with H3PO4)] as follows: 0–5 min, 0% B; 5–10 min, 0–30% B; 10–60 min, 30–100% B; 60–80 min, 100% B. Fractions (1 ml) were collected and counted using 4 ml of Ultima-Flo AP LCS-cocktail (Packard). Inositol polyphosphates species were identified by co-migration with standards using [3H]IP6 (Amersham), [3H]IP7, prepared by phosphorylation of [3H]IP6, and [3H]PP-IP4 (diphosphoinositol tetrakisphosphate) purified from metabolically labelled ipk1Δ cells as described in [7,36].


Yeast deficient in inositol pyrophosphates are resistant to H2O2

To study whether inositol pyrophosphates influence yeast survival after exposure to H2O2, WT (wild-type) and kcs1Δ yeast (lacking the IP6K gene, and therefore not possessing inositol pyrophosphates) [37,38] (Figure 4A), were treated with 0.1 mM, 0.5 mM and 1 mM H2O2 for 3 h. We studied the effects of H2O2 using two different yeast genetic backgrounds, DDY1810 and BY4741 (Table 1). Differences in genetic background are an established cause of experimental variability [39,40] and in fact we observed diverse sensitivity to H2O2 in the two WT strains (Figures 2A and 2B show the effect of the 0.5 mM concentration). However, the survival rate evaluated by colony-formation assays revealed that the kcs1Δ yeast were significantly more resistant to higher concentrations of H2O2 when compared with the respective WT strain in both genetic backgrounds (Figures 2A and 2B). When IP7 levels were restored in kcs1Δ cells by expressing the mammalian IP6K1 enzyme, this resistance was reversed, whereas cells expressing a kinase-dead IP6K1 mutant (IP6K1KA) remained significantly more resistant to the H2O2 treatment (Figure 2A). To further address whether inositol pyrophosphate levels correlated with the survival response to H2O2, we analysed yeast lacking the IPK1 gene (ipk1Δ). As IPK1 is the enzyme responsible for IP6 synthesis, ipk1Δ yeast lacks IP6 and IP7, but contains large amounts of the inositol pyrophosphate species derived from IP5 (Figure 1): PP-IP4 and (PP)2-IP3 (bisdiphosphoinositol trisphosphate) [38]. The ipk1Δ yeast showed a sensitivity to H2O2 comparable with WT, whereas the ipk1Δkcs1Δ double mutant, which once again lacks inositol pyrophosphates, remained resistant (Figure 2B). Previous studies have shown that although ipk1Δ mutant yeast have abnormal inositol phosphate metabolism, their morphology is identical with WT yeast, suggesting that PP-IP4 and (PP)2-IP3 can carry out the functions normally performed by IP7 and IP8 [38,41]. Taken together these results demonstrated that inositol pyrophosphates play a key role in mediating H2O2-induced cell death. Notably, this function might not be restricted to IP7/IP8, but could extend to other inositol pyrophosphate species.

Figure 2 kcs1Δ mutants are more resistant to H2O2 than WT cells

Shown are graphs depicting the yeast survival rate, analysed using a colony-formation assay, after 3 h of exposure to 0 mM (black), 0.1 mM (dark grey), 0.5 mM (light grey) or 1 mM (white) H2O2 for 3 h. (A) DDY1810 genetic background WT, kcs1Δ and kcs1Δ yeast expressing either the mouse IP6K1 (kcs1Δ+IP6K1) or the kinase-dead IP6K1KA (kcs1Δ+IP6K1KA). (B) BY4741 genetic background WT, kcs1Δ, ipk1Δ and ipk1Δkcs1Δ. Values are means±S.D. from four independent experiments performed in triplicate.

Inositol pyrophosphates regulate Rad53 activation following H2O2 treatment

We next tested whether Rad53, an important component of the DNA-damage response in yeast, was a downstream target of inositol pyrophosphate signalling. Rad53 activation can be observed as an electrophoretic mobility shift caused by its trans-phosphorylation by Mec1 and Tel1 followed by Rad53 autophosphorylation [42,43]. In total, Rad53 is phosphorylated at more than 20 residues [44]. The DNA-damage checkpoint response was investigated by exposing WT and kcs1Δ mutant yeast to different concentrations (0.1 mM, 0.5 mM and 1 mM) of H2O2 for various times; proteins were then extracted and Western blotted with Rad53-specific antibodies. We observed a rapid activation of Rad53 in response to 0.5 mM H2O2 in both DDY1810 WT and kcs1Δ cells. Although we failed to observe Rad53 activation when WT cells were exposed to 1 mM H2O2, this stimulation strongly activated Rad53 in kcs1Δ mutants (Figure 3A), correlating with the increased survival of these cells. Similarly, Rad9, which acts upstream of Rad53, was only activated in kcs1Δ at high concentrations of H2O2 (results not shown). Rad9 is a large protein (148 kDa) that displays a less evident mobility shift upon H2O2 treatment. Therefore Rad53 was used in all subsequent experiments as a read-out of the activation of the DNA-damage response. Expression of mammalian IP6K1 in kcs1Δ yeast restored the WT phenotype, with rescued cells failing to activate Rad53 in response to 1 mM H2O2. In contrast, cells expressing the kinase-dead IP6K1KA still retained strong Rad53 activation (Figure 3B). These results indicate that inositol pyrophosphates, and not the Kcs1 protein itself, mediate Rad53 activation. Consistently, we did not observe Rad53 activation in the ipk1Δ mutant treated with 1 mM H2O2, although Rad53 phosphorylation occurred in ipk1Δkcs1Δ double mutants (Figure 3C).

Figure 3 kcs1Δ display sustained Rad53 activation and a decreased mutation rate

DNA-damage response activation was observed by Western blotting using an anti-Rad53 antibody. (A) Time-dependent activation of Rad53 in the DDY1810 genetic background of WT and kcs1Δ yeast after exposure to 0.1 mM, 0.5 mM and 1 mM H2O2. At the higher dose, the Rad53 pathway is only activated in kcs1Δ yeast. (B) Mammalian IP6K1 reduced the level of Rad53 activation in kcs1Δ cells. After treatment with 1 mM H2O2 for 30 min, Rad53 is activated in kcs1Δ and kcs1Δ+IP6K1KA, but not WT and kcs1Δ+IP6K1. (C) Activation of Rad53 in the BY4741genetic background. After treatment with 1 mM H2O2 for 30 min Rad53 is phosphorylated in kcs1Δ and ipk1Δkcs1Δ, but not WT and ipk1Δ. Blots were subsequently probed with an anti-Hsp60 antibody for normalization of gel loading. (D) Mutation frequency after exposure to H2O2 as determined using a CanR assay [34]. The histogram shows the average mutation frequency for three independent experiments in the DDY1810 genetic background yeast. Values are means±S.D. Also shown is the average survival rate after exposure to H2O2.

To test whether the prolonged activation of Rad53 in kcs1Δ mutants was associated with a increased activation of DNA repair mechanisms, a CanR mutation frequency assay was performed [45]. This assay measures mutation of the CAN1 gene which encodes the arginine permease involved in the uptake of arginine and its cytotoxic analogue canavanine. Cells that have acquired mutations that inactivate CAN1 are capable of growing in medium containing canavanine. Mutation frequency can be easily measured by counting the number of colonies that are able to grow in canavanine-containing medium [34]. The basal rate of mutations in kcs1Δ yeast is comparable with WT (Figure 3D). Exposure of WT cells to 1 mM H2O2 for 30 min increased the mutation frequency of WT yeast more than 20-fold, whereas longer treatment (3 h) induced massive cell death. In contrast, kcs1Δ yeast showed a much lower mutation frequency after exposure to H2O2 for 30 min. The kcs1Δ yeast after 3 h treatment, a condition that was associated with a similar survival rate to WT yeast treated for 30 min, showed a still lower increase of mutation frequency (Figure 3D). These results indicate that the privileged activation of Rad53 in response to DNA damage in yeast deficient in inositol pyrophosphates results in the subsequent activation of a DNA-damage response mechanism that ultimately supports the survival of the cell.

Figure 4 vip1Δ yeast accumulate IP7 and are more resistant to H2O2

(A) HPLC analysis of [3H]inositol phosphates in DDY1810 WT, kcs1Δ, vip1Δ and kcs1Δvip1Δ yeast, revealed the accumulation of IP7 in vip1Δ yeast. (B) Vip1Δ is more resistant to H2O2 than WT. The histogram shows the average survival of WT, kcs1Δ, vip1Δ and kcs1Δvip1Δ cells after 3 h exposure to 0 mM (black) 0.1 mM (grey) or 1 mM (white) H2O2. Values are means±S.D. for three independent experiments. (C) DNA-damage response activation as observed by Rad53 phosphorylation. After treatment with 1 mM H2O2 for 30 min, Rad53 is activated in kcs1Δ, vip1Δ and kcs1Δ vip1Δ.

vip1Δ mutants accumulate large amounts of IP7, are resistant to H2O2 and display prolonged Rad53 activation

Vip1 was recently identified as an additional IP6K in yeast [6]. However, subsequent analysis of the mammalian form suggested that it is an enzyme that more probably converts IP7 into IP8 [46,47]. HPLC analysis of the vip1Δ single-mutant yeast revealed a substantial accumulation of IP7 (Figure 4A and Table 2), similar to what has been recently reported [48]. In yeast mutants for inositol kinases, such as IPMK and IPK1, the intracellular accumulation of specific inositol polyphosphate species provides the genetic proof that they are the physiological substrates of the deleted kinases [38,49]. The HPLC analysis of vip1Δ single mutants therefore indicates that cellular IP7 is produced by Kcs1, whereas Vip1 is likely to be responsible for the synthesis of (PP)2-IP4 (Figures 1 and 4A, and Table 2). Interestingly, we found that vip1Δ mutants displayed a substantial resistance to H2O2, although vip1Δkcs1Δ double mutants were likewise resistant to cell death to an extent similar to kcs1Δ yeast (Figure 4B). Moreover, both vip1Δ and vip1Δkcs1Δ yeast also exhibit Rad53 activation in response to 1 mM H2O2 (Figure 4C), suggesting that it is the double-pyrophosphate form of IP8 [(PP)2-IP4] (Figure 1) that plays a role in mediating the H2O2 response.

View this table:
Table 2 Inositol pyrophosphate levels expressed as a percentage of IP6

The values represent the means±S.E.M. from two independent experiments run in duplicate, of the inositol pyrophosphate levels expressed as a percentage of IP6 in DDY1810 WT, kcs1Δ, vip1Δ and kcs1Δvip1Δ yeast.

kcs1Δ cells show normal cell-cycle-dependent Rad53 activation and actin cytoskeleton depolarization after treatment with H2O2

A previous report by Leroy et al. [31] demonstrated that Rad53 activation after H2O2 treatment only occurred during the S-phase of the cell cycle [31]. To study whether the different levels of Rad53 activity observed in kcs1Δ and WT cells were due to abnormal Rad53 activation throughout the cell cycle, cells were arrested in G2/M-phase by nocodazole or in S-phase by HU (hydroxyurea) and treated with H2O2. In both WT and kcs1Δ yeast, Rad53 was predominantly activated during the S-phase of the cell cycle, indicating that increased Rad53 activation and survival of kcs1Δ cells is not cell-cycle-dependent (Figure 5A).

Figure 5 Normal cell-cycle-dependent Rad53 activation and actin cytoskeleton depolarization in kcs1Δ yeast

(A) DNA-damage response during the cell cycle. DDY1810 genetic background WT and kcs1Δ cells were arrested either in the S-phase of the cell cycle with HU or in the G2/M-phase with nocodazole, and were subsequently treated with 0.5 mM H2O2 for 15 min. In both WT and kcs1Δ, Rad53 is predominantly activated during the S-phase of the cell cycle. (B) Actin staining before and after exposure to H2O2. WT, kcs1Δ and kcs1Δ+IP6K1 were treated with 1 mM H2O2 for 3 h and the actin cytoskeleton was visualized with rhodamine-labelled phalloidin. Unchallenged kcs1Δ cells appear to have a less polymerized actin arrangement. However, after H2O2 treatment the actin cytoskeleton depolarizes equally well in kcs1Δ as in WT and kcs1Δ+IP6K1 yeast.

A novel function for Rad53 in regulating cell morphology and the actin cytoskeleton has been previously described [50]. Phalloidin staining of rad53Δ mutants was similar to WT cells, displaying a polarized actin cytoskeleton and actin cables extending from the mother cell into the bud [50]. Treatment of WT yeast with HU arrested the cells with large buds and a depolarized actin cytoskeleton. In contrast, the actin cytoskeleton of rad53Δ mutant cells remained polarized upon HU treatment. These results suggested that Rad53 plays a role in regulating the actin cytoskeleton polarization upon exposure to some cell stressors [50]. To study whether abnormal Rad53 signalling in kcs1Δ yeast might affect the polarity state of actin after exposure to H2O2, yeast were stained with rhodamine-labelled phalloidin. In kcs1Δ cells the actin cytoskeleton appeared predominately diffuse and not organized in filaments when compared with WT cells. Interestingly, the diffuse phenotype was reversed by introducing mammalian IP6K1 (Figure 5B). The actin behaviour in WT and kcs1Δ yeast was also analysed following exposure to 1 mM H2O2 for 1 h. We observed that, in both WT and kcs1Δ-treated yeast, actin became depolarized to a similar extent (Figure 5B). This result suggests that Rad53-mediated regulation of the polarity state of the actin cytoskeleton is not influenced by inositol pyrophosphates.

Yeast deficient in inositol pyrophosphates are sensitive to phleomycin treatment

The normal Rad53 activation during the cell cycle and the unaffected ability to polymerize actin in kcs1Δ cells in response to H2O2 indicated that inositol pyrophosphates only specifically alter Rad53 signalling in relation to cell survival. To test whether inositol pyrophosphate signalling plays a wider role in affecting cell survival, we performed cell survival and Rad53 activation assays following exposure of yeast to the DNA-damaging agent phleomycin. The glycopeptide antibiotic phleomycin belongs to the bleomycin family and directly binds to DNA creating double-strand breaks [51,52]. WT and kcs1Δ yeast were exposed to 10 μg/ml phleomycin and the survival rate was assessed by colony-formation assays (Figure 6A). We observed that kcs1Δ yeast was more sensitive to phleomycin, when compared with WT yeast. The hypersensitivity of kcs1Δ yeast to phleomycin was reversed by introducing the mammalian IP6K1 enzyme (Figure 6A). Rad53 activation analysis did not reveal significant differences between WT and kcs1Δ cells in response to different doses of phleomycin (Figure 6B). Similarly, a time course of Rad53 activation using 10 μg/ml phleomycin failed to reveal any difference between WT and kcs1Δ yeast (Figure 6C). Our results indicated that kcs1Δ yeast are more sensitive to phleomycin treatment than WT. Therefore it is likely that inositol pyrophosphate signalling is not directly sensing DNA damage.

Figure 6 Sensitivity to phleomycin treatment of yeast deficient in inositol pyrophosphates

(A) Analysis of WT, kcs1Δ and kcs1Δ yeast expressing mouse IP6K1 (kcs1Δ+IP6K1) exposed to 0 (black) or 10 μg/ml (white) phleomycin for 1 h revealed the sensitivity of kcs1Δ to this drug. Values are means±S.D. for three independent experiments. (B) Rad53 phosphorylation analysis did not reveal significant differences between WT and kcs1Δ cells exposed for 1 h to different doses of phleomycin. (C) Similarly, a time course of Rad53 activation using 10 μg/ml phleomycin, did not reveal significant differences between WT and kcs1Δ.

Exposure to H2O2, but not to phleomycin, causes a rapid decrease in inositol pyrophosphates

The opposite behaviour of kcs1Δ yeast to phleomycin and H2O2 treatment suggested a differential modulation of inositol pyrophosphate signalling. To study whether either H2O2 or phleomycin treatment affected intracellular levels of IP7 and other inositol pyrophosphates, WT and ipk1Δ yeast were labelled with [3H]inositol and the cellular inositol phosphate levels were analysed by SAX-HPLC [36] following 20 min of exposure to 1 mM H2O2 or 10 μg/ml phleomycin. Treatment with phleomycin did not affect the levels of inositol pyrophosphates in either WT or ipk1Δ yeast (Figures 7A and 7B).

Figure 7 H2O2 decreases the levels of inositol pyrophosphates in WT and ipk1Δ yeast

Representative HPLC analysis of the [3H]inositol polyphosphate profile (left-hand panels) of early exponentially growing WT (A) and ipk1Δ (B) yeast untreated or treated with 1 mM H2O2 or 10 μg/ml phleomycin for 20 min. The average inositol pyrophosphate levels are expressed as a ratio of their precursor IP6 for WT and IP5 in ipk1Δ yeast (right-hand panels). Values are means±S.D. for three independent experiments. (C) Effect of H2O2 on Kcs1–TAP enzymatic activity. Purified proteins were incubated with [3H]IP6 in the presence of 0, 0.1, 1 or 10 mM H2O2 or 10 mM DTT. The histogram represents the percentage of substrate converted into inositol pyrophosphates after a 15 min reaction. Values are means±S.D. for three independent experiments.

Interestingly, we observed that treatment with H2O2 led to a rapid decrease of inositol pyrophosphate levels. The H2O2 effect was particularly evident in the ipk1Δ yeast, as they contain high levels of the pyrophosphates PP-IP4 and (PP)2-IP3 (Figure 7B). These results suggest that IP7/IP8, and possibly other inositol pyrophosphates, play an important role in mediating the cellular responses to H2O2. The decrease of inositol pyrophosphates in WT yeast suggests that the cells may be trying to decrease inositol pyrophosphate-mediated signalling, mimicking the situation in kcs1Δ yeast, and thus becoming less sensitive to H2O2. These yeast data are complementary to the mammalian system where several reports indicate that an increase in inositol pyrophosphate cellular concentration sensitizes the cell to apoptotic treatment [1113].

We then asked whether H2O2 directly affected the enzymatic activity of Kcs1. Endogenous TAP–Kcs1 was purified from yeast and subjected to an enzymatic reaction in vitro. H2O2 treatment at concentrations as low as 0.1 mM dramatically reduced the enzymatic activity of Kcs1 (Figure 7C) and completely blocked its activity at higher doses. In contrast, exposure to the reducing agent DTT at high concentrations enhanced TAP–Kcs1 enzymatic activity (Figure 7C).

The main biochemical effect of H2O2 is believed to be cysteine oxidation [20]. To identify the possible cysteine residues targeted by H2O2, we analysed the mouse IP6K1 enzyme because Kcs1 is a polypeptide of 1059 amino acids and its large size complicates the generation and analysis of recombinant enzyme. Kcs1 belongs to an highly evolutionarily conserved family of enzymes and it is usually acknowledged that mammalian IP6Ks and yeast Kcs1 share the same biochemical characteristics. In fact, in vitro, mouse IP6K1 activity is negatively affected by H2O2 and enhanced by DTT in a similar fashion to the TAP–Kcs1 enzyme (results not shown). For this reason, to identify the possible H2O2 target sites, we have utilized the mouse IP6K1 enzyme that possesses 12 cysteine residues, of which only five are evolutionarily conserved with Kcs1 and the other two mammalian isoforms (IP6K2 and IP6K3).

Initially, we mutated the five conserved cysteine residues to alanine (Table 3). Analysis of these mutants revealed that the mutation of Cys221 to alanine completely abolished enzymatic activity. Interestingly this residue is located in close proximity to the domain important for the catalysis of this class of enzymes [53], as revealed by the crystal structure of the IP3 3-kinase [54] in which the aspartic acid in position 224 has been show to co-ordinate ATP. Cysteine-to-alanine mutation, while blocking the role of cysteine in protein functions, does not mimic oxidized cysteine that usually evolves to sulfonic acid. Consequently we decided to mutate Cys221 to aspartic acid. Interestingly the mutant C221D eliminated the enzymatic activity of this class of enzyme (Table 3). These results indicate that it is likely that the cysteine residue (amino acid 221 in mouse IP6K1, amino acid 773 in Kcs1), localized inside the inositol-binding domain [53,54], is responsible for the redox regulation of this class of enzymes. However, the analysis of other mutants reveals an important role for another three of the five evolutionarily conserved cysteine residues. The mutation Cys187 to alanine generated an enzyme only able to convert IP6 into IP7, even after a long time of incubation (Table 3). In contrast, the mutations Cys48 or Cys261 to alanine were more prone to convert IP6 into a higher inositol pyrophosphate, such as IP8 (Table 3). It is possible that this class of enzymes possess an intramolecular disulfide bond and that its destruction generates a more catalytically flexible enzyme. However, further work is required to verify this hypothesis.

View this table:
Table 3 Activity of mouse IP6K1 cysteine mutants

Recombinant IP6K1 proteins (10 ng) were incubated in the presence of 100 μM IP6 and trace amounts of [3H]IP6 in a 10 μl reaction volume for the indicated time. The reactions were stopped by adding EDTA and analysed by SAX-HPLC as described in [36]. Values are the percentages of the inositol pyrophosphates generated and represent the means±S.D. from three independent experiments. mIP6K1, mouse IP6K1.

Nevertheless, these results suggest that the evolutionarily conserved IP6K enzymes are sensors of the cellular redox status, probably through a cysteine residue located in their catalytic site. Oxidative stress signals might be in part directly transduced by inositol pyrophosphates.


In the present study we used genetically modified yeast with modulated levels of inositol pyrophosphates as a model to investigate the role of inositol pyrophosphates in the oxidative stress response. We found that IP7/IP8 and possibly other pyrophosphates derived from IP5 are required to confer sensitivity to H2O2 treatment. Cells that were depleted of inositol pyrophosphates showed resistance to this DNA-damaging agent and exhibited increased Rad53 activation which ultimately resulted in the avoidance of apoptosis and a dramatic decrease in the mutation rate. However, yeast mutants that do not possess inositol pyrophosphates are more sensitive to DNA damage caused by phleomycin, indicating that inositol pyrophosphates are not involved in directly sensing DNA damage. Instead, there was a rapid decrease of cellular inositol pyrophosphate levels following exposure to H2O2, indicating that inositol pyrophosphates act as signalling molecules in response to this genotoxic agent. The inhibitory effect of H2O2 on the enzymatic activity of TAP–Kcs1 in vitro suggests that the ROS signal might be transduced by the direct modification of Kcs1 enzyme. The comprehensive mutagenesis analysis of the evolutionarily conserved cysteine residues of this class of enzyme identified a specific residue (in mouse IP6K1 Cys221) localized in the previously characterized inositol-binding domain [53,54], which is a likely target of H2O2 oxidation. This evolutionarily conserved enzymatic feature suggests that, also in mammalian cells, H2O2 will elicit a decrease of inositol pyrophosphates. In fact, we observed a reproducible reduction of inositol pyrophosphate levels in HeLa cells after 30 min of incubation with 1 mM H2O2 (A. Saiardi, unpublished work). However, the response to oxidative stress in the mammalian system might be cell-type-dependent because no difference in IP8 levels was observed in DDT1 MF-2 cells, although a lower dose of 0.3 mM H2O2 was used in these cells [55].

H2O2 is formed naturally through biological processes by the action of oxidases (e.g. in peroxisomes) and by the action of the superoxide dismutase as a by-product of the respiratory chain. Furthermore, the transient production of H2O2 is triggered by receptor activation for ligands such as PDGF [23] or EGF [24]. Therefore H2O2 plays an important role as a second messenger in signal transduction and amplification [19,56,57]. H2O2 signalling cascades require the activation of kinases as well as inhibition of phosphatases (for a review see [20,58]). The rapid decrease of inositol pyrophosphates after H2O2 treatment suggests that inositol pyrophosphates are acting downstream of H2O2 in one or several pathways. Previously, IP7 was demonstrated to mediate a novel form of non-enzymatic protein pyrophosphorylation, suggesting a potentially pre-eminent role for the molecule in post-translational modification of proteins [35,59]. The differences observed between WT and kcs1Δ yeast may thus reflect altered IP7-mediated protein pyrophosphorylation; indeed H2O2 treatment alters the IP7 phosphorylation profile (S.N.M. Onnebo and A. Saiardi, unpublished work). The identification of the protein substrates of IP7 could provide additional insights into the signalling responses triggered by H2O2 as well as improve our knowledge of the physiological relevance of IP7-mediated phosphorylation.

The analysis of vip1Δ yeast revealed an accumulation of IP7 (Figure 4), demonstrating that this protein represents the physiologically relevant IP7 kinase. It is important to note that in the cell there are two isomers of IP8, (PP)2-IP4, a double-pyrophosphorylated form generated by the action of IP6K and Vip1, and a second isomer, the triphosphate PPP-IP5 generated directly by IP6K [9]. The discovery that Vip1 physiologically synthesizes (PP)2-IP4, and that in vip1Δ yeast there is accumulation of IP7, leads us to reinterpret a recent report indicating that the IP7 synthesized by Vip1 regulates a cyclin–CDK (cyclin-dependent kinase)–CKI (cyclin-dependent kinase inhibitor) transcriptional complex [60]. It is more likely that the absence of the double-pyrophosphorylated form of IP8 in vip1Δ yeast or the large accumulation of Kcs1-synthesized IP7 is responsible for cyclin–CDK–CKI complex regulation.

In yeast the DNA-damage sensing functions are mainly executed by Mec1 and Tel1 (S. cerevisiae homologues of mammalian ATM and ATR). Once activated, Mec1/Tel1 trigger a phosphorylation cascade that activates Rad53 (S. cerevisiae homologue of mammalian Chk2) [16,17,61]. Inositol pyrophosphates have previously been identified as negative regulators of Tel1 which, apart from playing an important role in the DNA-damage response, also regulates telomere length, as seen by the shorter telomeres in tel1Δ yeast [41,62]. However, Mec1/Tel1 activation assayed by the degree of Rad53 phosphorylation after phleomycin treatment appears not to be affected by inositol pyrophosphates. This suggests the presence of normal DNA-damage response machinery in the yeast with altered inositol pyrophosphate metabolism, further indicating a specific and definite role of this class of inositols in modulating ROS signalling.


Sara Onnebo and Adolfo Saiardi intellectually developed the project. Sara Onnebo performed the majority of the experimental work. Adolfo Saiardi performed the yeast inositol polyphosphate analysis and the IP6K1 enzymatic reactions. Sara Onnebo and Adolfo Saiardi wrote the manuscript.


This work was supported by the Medical Research Council funding of the Cell Biology Unit; and by Marie Curie International Reintegration Grants [grant number 014827].


We thank Rashna Bhandari, Adam C. Resnick and Antonella Riccio for suggestions and helpful comments, and members of the Saiardi laboratory for discussion.

Abbreviations: ATM, ataxia telangiectasia mutated; ATR, ataxia telangiectasia mutated- and Rad3-related; CanR, canavanine resistance; CDK, cyclin-dependent kinase; CKI, cyclin-dependent kinase inhibitor; DTT, dithiothrietol; EGF, epidermal growth factor; Hsp60, heat-shock protein 60; HU, hydroxyurea; IP3, inositol trisphosphate; IP4, inositol tetrakisphosphate; IP5, inositol pentakisphosphate; IP6, inositol hexakisphosphate; IP7, diphosphoinositol pentakisphosphate; IP8, [also referred to as (PP)2-IP4], bisphosphoinositol tetrakisphosphate; IP6K, IP6 kinase; IPK1, IP5 2-kinase; IPMK, inositol polyphosphate multi-kinase; Kcs1, Saccharomyces cerevisiae homologue of mammalian IP6K; PDGF, platelet-derived growth factor; PP-IP4, diphosphoinositol tetrakisphosphate; (PP)2-IP3, bisdiphosphoinositol trisphosphate; ROS, reactive oxygen species; SAX, strong anion exchange; SC medium, synthetic complete medium; TAP, tandem affinity purification; Vip1, Saccharomyces cerevisiae homologue of mammalian IP7K (diphosphoinositol pentakisphosphate kinase); WT, wild-type


View Abstract