SIZ1-Mediated Sumoylation of ICE1 Controls CBF3/DREB1A Expression and Freezing Tolerance in Arabidopsis

SIZ1 Regulates Freezing and Chilling Tolerances

The siz1-2 and siz1-3 T-DNA insertion alleles, which impair SIZ1 SUMO E3 ligase function (Miura et al., 2005), caused freezing sensitivity based on survival (Figures 1A to 1C ) and electrolyte leakage assays (Figure 1D). Electrolyte leakage from unacclimated siz1 plants was approximately twice that of the wild type at 0 to −3°C (Figure 1D) even though leakage was similar at 23°C (data not shown; Lee et al., 2007). siz1 plants were impaired in cold acclimation capacity compared with wild-type plants (Figure 1D), particularly at temperatures below −3°C (Figure 1E). The siz1 plants also were chilling sensitive, which was evident by comparison with the wild type after prolonged exposure to 4°C (see Supplemental Figure 1 online). Leaf chlorosis and necrosis of siz1 plants were visible after 3 weeks of chilling, but chlorophyll content of the wild type did not decline even after 6 weeks of chilling (see Supplemental Figures 1B and 1C online). Expression of the wild-type allele Pro_CaMV35S:SIZ1:GFP suppressed the freezing sensitivity of siz1-2 plants, further confirming that SIZ1 is required for low temperature tolerance (Figure 1C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

SIZ1 Mediates Freezing Tolerance.

(A) and (B) Nonacclimated (A) or cold-acclimated (B) wild-type (Col-0), siz1-2, and siz1-3 plants were exposed for 1 h to the temperature indicated, and photographs are of plants 7 d after freezing treatment.

(C) Freezing sensitivity of siz1-2 plants is suppressed by Pro_35S:SIZ1:GFP expression. Photographs are of nonacclimated or acclimated plants 7 d after 1 h of incubation at −4 or −6°C, respectively.

(D) Electrolyte leakage from nonacclimated or acclimated (acclim) wild-type, siz1-2, and siz1-3 plants after exposure to the temperature indicated (programmed to cool at 2°C h⁻¹). Data are means ± se (n = 4 leaves, each from a different plant).

(E) Electrolyte leakage difference between nonacclimated (white bars) or acclimated (gray bars) siz1 and wild-type plants (EL_siz1-EL_wt) after exposure to the temperature indicated as in (D).

SIZ1 Controls CBF/DREB1 and CBF/DREB1 Regulon Gene Expression

In wild-type plants, low temperatures of 10°C or less induce CBF/DREB1 expression within 15 min, and transcript accumulation increases with lower temperatures (Jaglo-Ottosen et al., 1998; Chinnusamy et al., 2003). Cold (0°C)–induced CBF/DREB1 transcript accumulation in siz1-2 seedlings was less than in the wild type over a 24-h time frame (Figure 2 , middle panels). The siz1 mutation affected expression of CBF3/DREB1A more than that of CBF1/DREB1B or CBF2/DREB1C (Figure 2, middle panels).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

SIZ1 Is a Positive Regulator of CBF/DREB1 Expression, Particularly of CBF3/DREB1A.

Relative mRNA levels in wild-type (white bars) and siz1-2 (black bars) seedlings were determined by quantitative RT-PCR analyses. Ten-day-old seedlings that were grown at 23°C were incubated at 0°C for the indicated time. Transcript abundances of SIZ1; CBF1/DREB1B, CBF2/DREB1C, and CBF3/DREB1A; CBF/DREB1 regulon genes COR15A, COR47, KIN1, and RD29A; SuSy (cold induced but independent of CBF/DREB1); and CBF/DREB1 regulators ICE1, HOS1, and MYB15 are illustrated. Data are means ± sd (n = 4).

Transcript accumulation of the CBF/DREB1 regulon genes COR15A, COR47, and KIN1 began 6 to 12 h after exposure of plants to cold (Gilmour et al., 1998; Figure 2, right panels). Low temperature–induced COR15A, COR47, and KIN1 transcript accumulation was less in siz1-2 than in wild-type seedlings (Figure 2, right panels). However, expression of RD29A, another CBF/DREB1 regulon gene, was similar in plants of both genotypes (Figure 2, middle panels). This gene may be regulated by other processes, as the hos9 mutation increases RD29A expression independent of CBF/DREB1 (Zhu et al., 2004). Sucrose synthase (SuSy), which is cold regulated through a CBF/DREB1-independent pathway (Gilmour et al., 2000), exhibited similar expression patterns in wild-type and siz1-2 seedlings (Figure 2, right panels). These data indicate that SIZ1 controls cold signaling through the regulation of CBF/DREB1 expression, although not all genes of the regulon are affected. Cold-induced CBF3/DREB1A, CBF1/DREB1C, COR47, and COR15A expression also was repressed in siz1-3 plants (data not shown).

A positive regulator, ICE1, and negative regulators, HOS1 and MYB15, of CBF/DREB1 expression and freezing responses have been identified (Lee et al., 2001; Chinnusamy et al., 2003; Agarwal et al., 2006). HOS1 and ICE1 transcript accumulation was not substantially affected by cold and was similar in wild-type and siz1-2 seedlings (Figure 2, left panels). HOS1 mRNA abundance transiently decreases immediately after cold exposure but is similar to precold levels after 3 h (Lee et al., 2001), which was the first treatment time point in this experiment (Figure 2, left panels). ICE1 transcript abundance was similar in wild-type and siz1 plants even with cold treatment (Lee et al., 2005; Figure 2, left panels). On the other hand, MYB15 mRNA abundance was moderately upregulated by cold in wild-type seedlings (Agarwal et al., 2006) but was induced substantially in siz1 seedlings (Figure 2, left panels). The timing of cold-responsive MYB15 expression in siz1 seedlings resembled that of COR and KIN1 transcript accumulation (i.e., occurred later during the low temperature incubation period relative to CBF/DREB1 expression). These results indicate that SIZ1 negatively regulates MYB15 expression but does not alter appreciably the transcript level of HOS1 or ICE1.

SIZ1 Facilitates Cold-Induced Accumulation of SUMO Conjugates

Anti-SUMO1 conjugation products were accumulated in wild-type seedlings within 1 h after exposure to 0°C but to a much lesser extent in siz1-2 seedlings, at comparable times during cold treatment (Figure 3 ). Anti-SUMO1 detects both SUMO1 and SUMO2 (Murtas et al., 2003), which are the most closely related by sequence comparison among the eight family members (Kurepa et al., 2003) and may be functionally redundant (Kurepa et al., 2003). These data indicate that cold induces SIZ1-mediated SUMO1/2 conjugation in planta that is due presumably to the E3 ligase activity of SIZ1 (Miura et al., 2005). Heat shock and other environmental stresses also cause an increase in SUMO1/2 conjugation products (Kurepa et al., 2003; Miura et al., 2005; Yoo et al., 2006), implicating that sumoylation/desumoylation of protein targets may be a posttranslational regulatory process in abiotic stress signaling of plants.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Cold Stress Induces SUMO1/2 Conjugate Accumulation That Is Facilitated by SIZ1.

Shown are in planta sumoylation profiles of 10-d-old wild-type and siz1-2 seedlings that were grown on medium at 23°C (0 time) and then incubated at 0°C for the time indicated. Total protein was separated by SDS-PAGE, transferred, and detected with anti-SUMO1 that interacts with both SUMO1 and SUMO2 (Murtas et al., 2003).

SIZ1-Mediated Sumoylation of ICE1 Facilitates CBF3/DREB1A Expression

The siz1 mutation did not affect ICE1 expression (Figure 2), which infers that SIZ1 is not a transcriptional regulator of ICE1 but could regulate activity by posttranslational modification. SUMOplot (http://www.abgent.com/tool/sumoplot) predicted that ICE1 contains one SUMO conjugation motif (ΨKXE; Minty et al., 2000) with K393 as the acceptor residue. SIZ1 was necessary for sumoylation of ICE1 in vitro (Figure 4A ) but did not facilitate SUMO conjugation to the ICE1(K393R) variant protein (Figure 4A). K-to-R substitutions in the sumoylation motifs of substrate proteins effectively block isopeptide linkage of SUMO to the target protein (Gostissa et al., 1999; Sachdev et al., 2001; Lin et al., 2003; Long et al., 2004a; Gill, 2005). These results indicate that SIZ1 facilitates sumoylation of ICE1 and identifies K393 as the SUMO conjugation residue in the protein.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

SIZ1 Facilitates Sumoylation of ICE1.

(A) SIZ1 mediates in vitro sumoylation of ICE1. Affinity-purified recombinant GST-T7-ICE1 or GST-T7-ICE1(K393R) was used as a substrate in an in vitro sumoylation assay; reaction mixture contained yeast GST-ScAos1 (E1), GST-ScUba2 (E1), ScUbc9 (E2), and His-ScSmt3 (SUMO) and Arabidopsis GST-SIZ1 (E3) recombinant proteins (Takahashi et al., 2003; Miura et al., 2005), without (−) or with (+) protein. ICE1(K393R) has an amino acid substitution of the residue that is predicted to be a SUMO conjugation site in ICE1. Unconjugated and SUMO-conjugated ICE1 proteins were detected with anti-T7 (T7-ICE1). Arrowhead indicates sumoylated ICE1 and WB: T7 designates a protein gel blot detected with anti-T7.

(B) SIZ1-mediates SUMO1 conjugation to ICE1 in protoplasts. Wild-type or siz1-2 protoplasts were cotransformed to express T7-SUMO1 and HA-ICE1 or HA-ICE1(K393R) at 23°C for 36 h. Protoplasts were then incubated at 0°C or maintained at 23°C for 1 h and then harvested immediately thereafter. Soluble extracts were immunoprecipitated with anti-T7 Tag Agarose (Novagen) to obtain T7-SUMO1 conjugates (IP: T7). Anti-HA was used to detect T7-SUMO1-HA-ICE1 conjugates on the immublot (WB: HA; top panel). To confirm equivalent expression of HA-ICE1 in protoplasts, crude extract from protoplasts was separated by SDS-PAGE, and ICE1 was detected by protein gel blot analysis with anti-HA (WB: HA; bottom panel).

(C) Schematic representation of the effector and reporter plasmids used in the GAL4 transient expression assay (Chinnusamy et al., 2003; Tiwari et al., 2003). The ICE1 or ICE1(K393) open reading frame was inserted into the effector plasmid. GAL4 DB is the DNA binding domain of the yeast transcription factor GAL4; GUS is the reporter gene; NOS is the terminator signal of the nopaline synthase gene; and Ω is the translational enhancer of Tobacco mosaic virus.

(D) Relative GUS activities after transfection with an effector vector (vector control), ICE1, ICE1(K393R), or ARF5M (positive control; Tiwari et al., 2003). Renilla luciferase activity was used for normalization; relative GUS expression [(GUS_sample/LUC_sample)/(GUS_vector/LUC_vector)] was calculated. Data are means ± se (n = 3).

T7:SUMO1 and HA:ICE1 or HA:ICE1(K393R) cDNAs were cotransformed into protoplasts isolated from wild-type or siz1-2 plants (Jin et al., 2001) to assess if SIZ1 facilitates sumoylation in planta. SUMO1 conjugation to ICE1 was moderately induced by cold treatment in the wild type but was not evident in cold-treated siz1-2 protoplasts (Figure 4B). Also, the K-to-R mutation blocked in vivo sumoylation of ICE1 (Figure 4B). Together, these results indicate that SIZ1 facilitates sumoylation of ICE1 at the K393.

Transient expression of ICE1 but not of ICE1(K393R) in wild-type protoplasts induced CBF3/DREB1A transcript abundance in response to cold (see Supplemental Figure 2 online). Conversely, ICE1(K393R) but not ICE1 expression moderately increased MYB15 mRNA abundance (see Supplemental Figure 2 online). These results and those illustrated in Figure 2 are consistent with the premise that cold-induced, SIZ1-mediated sumoylation of K393 in ICE1 induces CBF3/DREB1A and represses MYB15 expression.

The GAL4 DNA binding system was used to evaluate the effect of K393R on ICE1 transcriptional activator function (Chinnusamy et al., 2003; Tiwari et al., 2003). The effector plasmid was constructed by fusing the cDNA encoding ICE1, ICE1(K393R), or ARF5M (positive control; Tiwari et al., 2003) to the DNA binding domain of the yeast GAL4 transcriptional activator (Figure 4C). Expression of this cassette is driven by the 35S promoter of Cauliflower mosaic virus (CaMV). The effector plasmid [ICE1, ICE1(K393R), or ARF5M], the Pro_GAL4:β-glucuronidase (GUS) reporter plasmid (Tiwari et al., 2003), and the plasmid encoding Renilla luciferase driven by the 35S promoter for normalization (Chinnusamy et al., 2003) were cotransformed into Arabidopsis protoplasts (Jin et al., 2001). Transient expression of ICE1 increased GUS reporter activity ∼20-fold (Figure 4D) as reported (Chinnusamy et al., 2003). Expression of ICE(K393R) activated GAL4-responsive transcription to an equivalent level of ICE1 (Figure 4D). Sumoylation/desumoylation seems not to regulate ICE1 transactivation activity by the criteria of this assay.

SIZ1-Dependent SUMO Conjugation to ICE1 Reduces Polyubiquitination in Vitro

Cold treatment induces HOS1 to accumulate in the nucleus, where the E3 ligase facilitates ubiquitination of ICE1, which targets the protein for proteasome degradation (Dong et al., 2006). ICE1 destabilization leads to reduced CBF3/DREB1A expression and freezing sensitivity. Recombinant T7-ICE1 was incubated together with other components of the in vitro sumoylation assay mixture, including SIZ1 as the SUMO E3 ligase (Miura et al., 2005), and then purified using anti-T7. Unsumoylated or SUMO-conjugated ICE1 was then added to an in vitro ubiquitination assay mixture that included HOS1 as the ubiquitin E3 ligase (Dong et al., 2006; Figure 5A ). Sumoylated ICE1 was less polyubiquitinated than unsumoylated ICE1 (Figure 5A). However, K393 was not the primary ubiquitination site in ICE1 (Figure 5B). These results indicate that sumoylation of K393 reduces polyubiquitination of ICE1 that could increase ICE1 stability, thereby increasing CBF3/DREB1A expression and low temperature tolerance.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Sumoylation of ICE1 Inhibits Polyubiquitination of the Transcription Factor.

(A) Affinity-purified recombinant GST-T7-ICE1 was incubated in an in vitro sumoylation (S) mixture (Miura et al., 2005) or not incubated (−). Proteins were immunopurified with anti-T7 Tag agarose (Novagen), washed three times, and eluted. Eluted proteins were subjected to an in vitro ubiquitination assay (Ub) (Dong et al., 2006) or not (−). Proteins were separated by SDS-PAGE and identified by immunoblot analysis using anti-T7. Arrowheads indicate sumoylated ICE1.

(B) Recombinant GST-T7-ICE1 and GST-T7-ICE1(K393R) was not or was subjected to an in vitro ubiquitination assay (Ub). Proteins were separated by SDS-PAGE and detected by immunoblot analysis with anti-T7.

Sumoylation of ICE1 Mediates Freezing Tolerance

Overexpression of ICE1 (Pro_CsV:ICE1) but not ICE(K393R) [Pro_CsV:ICE1(K393R)] increased freezing tolerance of transformed wild-type plants (Figures 6A and 6B ). Increased abundance of transgene transcripts was detected in plants of the different overexpression lines by RT-PCR (see Supplemental Figure 3 online). Freezing tolerance of ICE1-overexpressing plants obtained in this research (Figure 6A) was similar to that of ICE1-overexpressing plants reported in a prior publication (Chinnusamy et al., 2003; Figure 6A, ICE1oe). ICE1(K393R)-expressing plants were more sensitive to freezing stress than wild-type or vector control plants but less sensitive than siz1-2 plants (Figures 6A and 6B). Cold-induced transcript abundance of CBF3/DREB1A and the regulon genes COR15A and COR47 was greater in ICE1-overexpressing plants and less in ICE1(K393R) plants relative to the vector control wild type (Figure 6C). CBF1/DREB1B and CBF2/DREB1C expression was upregulated in ICE1-overexpressing plants, as reported (Chinnusamy et al., 2003), but was moderately downregulated in ICE1(K393R) transgenic plants relative to plants transformed with the vector cassette. MYB15 was moderately upregulated by ICE1(K393R) overexpression (Figure 6C). These results indicate that sumoylation of ICE1 facilitates CBF3/DREB1A expression and freezing tolerance and that substitution of K393 to R represses CBF3/DREB1A and increases MYB15 expression.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

ICE1 Sumoylation Facilitates Freezing Tolerance and Enhances CBF3/DREB1A and CBF/DREB1 Regulon Expression.

(A) Photographs are of representative plants 7 d after 1-h exposure to the indicated temperature. P_CsV:ICE1 (line 8) or P_CsV:ICE1(K393R) (line 8) was transformed into Col-0 plants. pCsV1300 transgenic lines were generated for use as a vector control. The ICE1 overexpressing line (ICE1oe) was used as a positive control (Chinnusamy et al., 2003).

(B) Survival was determined for 20 plants after freezing treatment at the indicated temperature. Data are means ± se calculated from data of five independent experiments and three independent ICE1 (3, 8, and 12) and ICE1(K393R) (6, 7, and 8) expressing lines.

(C) Relative mRNA levels in plants of ICE1 (white bars) or ICE1(K393R) (black bars) overexpressing lines or a vector control line (hatched bars) were determined by quantitative PCR. Seedlings were either untreated (0 h) or treated with low temperature (0°C) for the indicated time. Data are means ± sd (n = 4).

SIZ1 Transduces Cold Signals and Mediates Low Temperature Tolerance

SIZ1-mediated SUMO1 conjugation to ICE1 is induced by cold treatment (Figure 4B) and induces CBF3/DREB1A expression (Figure 6C; see Supplemental Figure 2 online). These results indicate that SIZ1-mediated sumoylation of ICE1 is an early event in low temperature responses of plants and induces signal cascades that regulate expression of genes that are necessary for cold tolerance. However, it is still not established how cold regulates SIZ1 E3 ligase function. SIZ1 expression is constitutive and unaffected by cold, indicating that transcriptional regulation does not directly control SIZ1 function in low temperature tolerance. Transcriptome profiling analyses reveal that genes encoding SUMO, SUMO E1, SUMO E2, and SUMO proteases are not induced by low temperature stress (Lee et al., 2005), inferring that cold-induced sumoylation of ICE1 is not due to transcriptional regulation of the SUMO conjugation/deconjugation cycle. Identification of the regulators that control sumoylation in any organism remains elusive. Thus, it remains unclear how this posttranslational control process transduces specific regulatory inputs (Hay, 2005).

It is plausible that low temperatures directly or indirectly induce posttranslational modification of SIZ1 that alters the regulatory function of the protein. The human PIASy is sumoylated at K35, which is necessary for PIASy E3 ligase activity that activates the transcription factor Tcf-4 (Ihara et al., 2005). Since Arabidopsis SIZ1 contains predicted sumoylation motifs, it is possible that SUMO conjugation regulates the function of the protein. TGF-β activates the p38 MAP kinase cascade that stabilizes the PIASxβ protein and enhances PIASxβ gene expression (Ohshima and Shimotohno, 2003). PIASxβ facilitates sumoylation of Smad4, which is necessary for transactivation function (Ohshima and Shimotohno, 2003).

Posttranscriptional control of SIZ1 function could be a result of subnuclear targeting. SIZ1 is located in nuclear speckles at 23°C (Miura et al., 2005), which is analogous to PIAS proteins that localize to mammalian promyelocytic leukemia (PML) nuclear bodies (Sachdev et al., 2001; Seeler and Dejean, 2003). The precise function of PML nuclear bodies is not clear, but these are implicated in both gene activation and repression because transcription factors and transcriptional coregulators, including chromatin-modifying proteins, colocalize to PML bodies (Ching et al., 2005). Cytochemical analyses and genetic evidence indicate that PML bodies interact with specific genes and are associated with transcriptionally active chromatin (Wang et al., 2004; Ching et al., 2005). Recently, subnuclear compartmentalization processes have been implicated in gene activation (Sharrocks, 2006). It is possible that low temperature affects subnuclear localization of SIZ1 and/or of ICE1 that alters CBF/DREB1 expression.

SIZ1-Mediated Sumoylation of ICE1 May Facilitate and Stabilize Transcription Factor Activity

ICE1 preferentially induces CBF3/DREB1A expression within minutes after exposure of plants to low temperatures (Baker et al., 1994; Chinnusamy et al., 2003; Gilmour et al., 2004). The paradigm is that cold stimulates ICE1 transcription factor activity, by some yet unknown processes, which facilitates CBF3/DREB1A expression (Chinnusamy et al., 2003; Dong et al., 2006; Figure 7 ). SUMO conjugation can enhance or repress transcription factor activity (Gill, 2005; Hay, 2005). For instance, sumoylation of NFAT-1 (for nuclear factor of activated T cells), p53 tumor suppressor, Smad4, and Tcf-4 is necessary for transcription factor activity (Ohshima and Shimotohno, 2003; Gill, 2005; Hay, 2005; Ihara et al., 2005). Likewise, sumoylation of transcription factors represses activity through processes that involve transcriptional coregulatory complexes (Johnson, 2004; Gill, 2005; Hay, 2005).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

A Model of SIZ1 Regulation of Cold Signaling and Tolerance.

SIZ1-dependent sumoylation of ICE1 at K393 regulates expression of CBF3/DREB1A and cold tolerance. SUMO conjugation to ICE1 induces CBF3/DREB1A expression, and then CBF3/DREB1A induces cold-responsive genes, such as COR15A, COR47, and KIN1 (Figures 2 and 6C), and cold tolerance (Figures 1 and 6). Sumoylation may enhance protein stability to block polyubiquitination of ICE1 (Figure 5). Furthermore, SUMO conjugation to ICE1 represses MYB15 expression (Figure 6C). Sumoylation of ICE1 is a key regulatory process in cold signaling and tolerance.

Our results indicate that the ICE1(K393R) substitution does not alter transactivation activity of the MYC factor in the GAL4 assay (Figure 4D), even though expression of CBF3/DREB1A is downregulated in ICE1(K393R)-overexpressing plants (Figure 6C). Interestingly, the ice1 mutation (R236H) does not affect ICE1 transactivation activity but represses CBF3/DREB1A expression (Chinnusamy et al., 2003). Mammalian PIAS3 interacts with the coactivator p300/CBP and the transcription factor Smad3 (Long et al., 2004b). Recruitment of the p300/CBP coactivator to chromatin stimulates PIAS3-mediated transcriptional activation of Smad3 in response to TGF-β (Long et al., 2004b). Therefore, it is possible that SIZ1 may be involved in recruitment of coactivator complex subunits to the CBF3/DREB1A chromatin to facilitate induction of expression by ICE1.

Later in the cold episode, ICE1 activity decreases, at least in part, are attributable to proteolysis facilitated by HOS1, a RING finger ubiquitin E3 ligase, which is a negative regulator of CBF/DREB1 expression (Dong et al., 2006). Degradation of ICE1 presumably contributes to lower CBF3/DREB1A expression that occurs after the initial induction by low temperatures. SUMO and ubiquitin can interact competitively or cooperatively on the same substrate to regulate protein function and biological processes (Hay, 2005; Ulrich, 2005). Sumoylation of IκBα, Smad4, or NEMO reduces ubiquitin-mediated degradation of these proteins (Ulrich, 2005). Sumoylation of ICE1 at K393 attenuates polyubiquitination of the transcription factor in vitro (Figure 5A). Perhaps SUMO conjugation to ICE1 prevents access of the ubiquitination complex, which may include HOS1, to a target residue(s) in the protein (Pichler et al., 2005) and thereby reduces proteasome degradation (Dong et al., 2006). Thus, SUMO or ubiquitin conjugation to ICE1 results in either transcription factor activity or destabilization (i.e., inactivation) causing opposing effects on CBF3/DREB1A expression.

SUMO Modification of ICE1 Represses MYB15 Expression

Sumoylation of ICE1 also negatively affects MYB15 transcript abundance, which presumably facilitates CBF3/DREB1A expression (Figure 6C; see Supplemental Figure 2 online). MYB15 represses CBF/DREB1, particularly CBF3/DREB1A, expression (Agarwal et al., 2006). It is yet to be determined whether SUMO-conjugated ICE1 directly or indirectly represses MYB15. Sumoylation of the same transcription factor can result in induction or repression of different target genes (Gill, 2005; Hay, 2005). SUMO modification of p53 enhances transcription of p21, an inhibitor of cyclin-dependent kinase (Gostissa et al., 1999). Interestingly, sumoylation of p53 represses expression of the human ribosomal gene cluster gene that inhibits DNA replication (Schmidt and Müller, 2002). The overall effect of p53 sumoylation, however, is to mediate p53-dependent G1 cell cycle arrest (Müller et al., 2000; Megidish et al., 2002). Also, PIAS-mediated sumoylation of Smad4 causes activation or repression of specific genes (Ohshima and Shimotohno, 2003; Long et al., 2004a). Cold-induced MYB15 expression occurs much more rapidly and intensively in siz1 seedlings than in the wild type. Sumoylated ICE1 may negatively regulate MYB15, but ICE1(K393R) expression enhances MYB15 transcript abundance (Figure 6C; see Supplemental Figure 2 online).

CBF/DREB1 gene function is tightly regulated both transcriptionally and posttranscriptionally (Thomashow, 1999; Lee et al., 2001; Chinnusamy et al., 2006), presumably because uncontrolled CBF/DREB1 activation may be disadvantageous in particular environments. In fact, constitutive CBF/DREB1 overexpression results in dwarfism (Gilmour et al., 2004), and hyperexpression of CBF/DREB1 caused by a hos1 mutation results in freezing sensitivity (Ishitani et al., 1998). The results presented herein confirm that ICE1 is a key regulator of CBF/DREB1 expression and cold tolerance and that sumoylation is a critical process in determining ICE1 activity (Figure 7). SIZ1-mediated sumoylation may facilitate CBF/DREB1 expression, ICE1 stability, and MYB15 repression. SUMO1/2 conjugation to K393 of ICE1 is the apparent biochemical basis for these positive and negative control processes. Presumably, the relative affect of cold exposure on these positive and negative control processes can provide a fine-tuned balance of phenotypic responses that enable fitness to an environment that experiences different low temperature episodes.

Plant Freezing Assay

Wild-type (ecotype Col-0), siz1-2, and siz1-3 plants were grown for 3 weeks (from sowing) in soil at 23°C under a long-day photoperiod (16 h light/8 h dark). Plants grown under these conditions were considered nonacclimated. For cold acclimation, 3-week-old plants were incubated at 4°C for 7 d under a long-day photoperiod. Plant freezing tests were performed as described (Zhu et al., 2004). Briefly, plants were exposed to low temperatures in a controlled temperature chamber (model Tenney-JR; Tenney Engineering). For Figure 1, plants were exposed to 4°C for 0.5 h, then 0°C for 1 h, and then the programmed cycle of temperature reduction (−2°C over 30 min and then maintenance of that temperature for 1 h). The cycle was repeated until the desired temperature was reached. At 0°C, ice chips were spread over the plants to facilitate uniform nucleation. For Figure 6, cold-acclimated (4°C) plants were exposed to 0°C for 0.5 h, and ice chips were sprinkled over these plants before the chamber was programmed to cool at −1°C h⁻¹. After a freezing treatment, plants were incubated at 4°C for 1 d and then returned to 23°C, and survival was determined 7 d later.

Electrolyte Leakage from Leaves

Electrolyte leakage from fully developed rosette leaves of 3-week-old plants was assessed as described (Zhu et al., 2004), with modifications. Each leaf (5th or 6th rosette leaf) was placed into a tube containing 200 μL deionized water, an ice chip was added to initiate nucleation, and the tube was incubated in a refrigerated circular bath (NESLAB RTE-120). The temperature of the bath was programmed for a temperature decrease from 0 to −8°C by a continuous reduction of −1°C over 30 min until the desired temperature was reached. Each tube was removed from the refrigerated bath and then placed immediately into ice to facilitate gradual thawing. Deionized water (14 mL) was added to the sample that was then shaken overnight, after which the conductivity of the solution was determined with the Accumet model 20 pH/conductivity meter (Fisher Scientific). The tube was then incubated at 90°C for 1.5 h and cooled to room temperature, and conductivity of the solution was determined (i.e., 100% conductivity). Conductivities of four leaves, each from a different plant, were evaluated at the different temperature minima.

Quantitative RT-PCR

Total RNA was isolated using TRIZOL reagents (Invitrogen) according to the manufacturer's protocol. Three micrograms of RNA were used as template for first-strand cDNA synthesis using Superscript II reverse transcriptase (Invitrogen) and an oligo(dT₂₁) primer. Primer pairs for real-time PCR (see Supplemental Table 1 online) were designed as described (Miura et al., 2005). Real-time PCR amplification, validation experiments, and calculation of relative differences in expression level were performed as per instructions (User Bulletin 2 for ABI PRISM 7700 sequence detection system; Applied Biosystems) (Miura et al., 2005). Briefly, real-time PCR amplification was performed with 20 μL of reaction solution, containing 5 μL of 50-fold–diluted cDNA, 0.3 μM of each primer, and 1× QuantiTect SYBR Green PCR Master Mix (Qiagen). Analysis was performed using the Applied Biosystems PRISM 7700 sequence detection system. Relative transcript abundance was calculated using the comparative C_T method (User Bulletin 2 for ABI PRISM 7700 sequence detection system). For a standard control, expression of Actin2 was used. After calculation of ΔC_T (C_{T,gene of interest} − C_T,actin2), ΔΔC_T [ΔC_T − ΔC_{T,WT(0 h)}] was calculated as instructed. The relative expression level was calculated as 2^−ΔΔCT. A 2^−ΔΔCT value for the wild type without cold treatment (0 h) was normalized to 1[2^{−ΔΔCT(ΔCT,WT(0 h)− ΔCT,WT(0 h))} = 2⁰ = 1].

In Vivo Analysis of Sumoylation Profiles

Seeds were sown onto 1× Murashige and Skoog salts medium containing 3% sucrose (Miura et al., 2005). Ten-day-old seedlings were incubated at 0°C for the times indicated prior to harvesting. Total protein was extracted as described (Murtas et al., 2003, 2005). Total protein (200 μg) was separated by SDS-PAGE and then blotted. The gel blot was probed with anti-SUMO1 (kindly provided by G. Coupland; Murtas et al., 2003) and was detected using ECL plus (Amersham).

Purification of Recombinant Proteins and in Vitro SUMO Conjugation Assay

To introduce the T7 tag (MASMTGGQQMG) into pGEX-5X-1 (Amersham), pET-21a (Novagen) and pGEX-5X-1 were digested with NdeI and EcoRI, respectively, and then incubated with T4 DNA polymerase for blunt-end ligation. The blunt-ended vectors then were digested with XhoI. The DNA fragment containing the T7 sequence was separated on an agarose gel, collected using the QIAquick gel extraction kit (Qiagen) and ligated into the digested pGEX-5X-1. The resulting construct was named pGEX-5X-T7. The ICE1 open reading frame was amplified from cDNA (synthesized from RNA of 3-h cold-treated [0°C] seedlings) using pfu DNA polymerase (Stratagene), and the primers ICE1-T7F and ICE1-HAR (see Supplemental Table 2 online). The PCR product and pGEX-5X-T7 were digested with EcoRI and XhoI and then ligated. The construct was designated as pGST-T7-ICE1. To make an ICE1(K393R) protein, AA₁₁₇₈G was replaced with AGG by site-directed mutagenesis using the primers ICE1K393RF and ICE1K393RR (see Supplemental Table 2 online) (Miura et al., 2005). The PCR product was cloned into pGEX-5X-T7. The resulting plasmid construct was named pGST-T7-ICE1(K393R). ICE1 recombinant protein constructs were transformed into Escherichia coli BL21(DE3). E. coli cells containing pGST-T7-ICE1 or pGST-T7-ICE1(K393R) were incubated at 37°C until cells reached mid-log phase of growth. Then, 0.1 mM isopropylthio-β-galactoside was added, and cells were incubated at 18°C for 16 h. Crude extracts were purified using Glutathione Sepharose 4 Fast Flow (Amersham). Yeast proteins and Arabidopsis thaliana SIZ1 protein were purified as described (Takahashi et al., 2003; Miura et al., 2005).

The in vitro sumoylation assay was performed as described (Takahashi et al., 2003; Miura et al., 2005). Briefly, 3 μg of GST-T7-ICE1 or GST-T7-ICE1(K393R) was added to a reaction mixture containing 1.8 μg GST-ScUba2, 0.85 μg GST-ScAos1, 0.17 μg ScUbc9, 3 μg His-ScSmt3, 2 μg GST-AtSIZ1, 50 mM Tris-HCl, pH 7.4, 10 mM ATP, 2 mM DTT, and 5 mM MgCl₂. The reaction mixture (30 μL) was incubated at 37°C for 1.5 h. Proteins were separated by SDS-PAGE, and immunoblot analysis performed with anti-T7 antibody (Bethyl Laboratories) was performed.

In Vitro Ubiquitination of ICE1

Arabidopsis MBP-HOS1 (cDNA clone was kindly provided by Jian-Kang Zhu; Dong et al., 2006) was purified using amylose resin (New England Biolabs) as instructed. Purified GST-T7-ICE1 was sumoylated in vitro as described above. ICE1 was purified using T7-Tag antibody agarose (Novagen). The ICE1 was washed three times and eluted according to the manufacturer's instructions. ICE1 protein was added to the ubiquitination buffer containing 0.2 μg of human E1 and 0.4 μg of E2 UbcH5b (Boston Biochemicals), 1 μg of E3 (MBP-HOS1), 10 μg of bovine ubiquitin (Sigma-Aldrich), 1 mM MgCl₂, 2 mM ATP, and 2 mM DTT. The mixture was incubated at 30°C for 4 h, and samples were separated by electrophoresis on 8% SDS gels.

Protoplast Purification and Immunoprecipitation Analysis

The ICE1 coding region was amplified with pfu DNA polymerase and the primers ICE1-HAF and ICE1-HAR (see Supplemental Table 2 online). The BamHI- and XhoI-digested PCR product was inserted into the plasmid p326-HAN (Jin et al., 2001) to produce a chimeric HA-ICE1 or HA-ICE1(K393R) fusion under the control of the 35S promoter. These plasmids were purified using the Qiagen plasmid maxi kit according to the manufacturer's instructions. Arabidopsis protoplasts were prepared from 2-week-old wild-type or siz1-2 seedlings with Cellulase Onozuka R-10 and Macerozyme R-10 (Yakult Pharmaceutical) as described (Jin et al., 2001). The fusion constructs were introduced into Arabidopsis protoplasts by polyethylene glycol–mediated transformation (Jin et al., 2001). After transformation with T7:SUMO1 and HA:ICE1 or HA:ICE1(K393R), protoplasts were incubated for 36 h at 23°C. Cells were then maintained at 23°C or incubated at 0°C for 1 h and then immediately harvested for protein isolation.

Immunoprecipitation was performed essentially as described (Jin et al., 2001). T7-Tag antibody agarose (Novagen) was added to the lysate. After incubation at 4°C for 3 h, the agarose was washed three times. After elution from T7 antibody agarose, immunoprecipitated proteins were separated by SDS-PAGE and detected by immunoblotting using anti-HA.

Transgenic Plants

The ICE1 coding region was amplified with pfu DNA polymerase and the primers ICE1-expF and ICE1-EGR (see Supplemental Table 2 online). pGEX-T7-ICE1 or pGEX-T7-ICE1(K393R) cDNA clone was used as a template for PCR. The binary vector pCsV1300 (promoter from cassava vein mosaic virus) was digested with BamHI. The BamHI-digested PCR product was inserted to construct pCsV1300-ICE1 or pCsV1300-ICE1(K393R). Each binary vector was mobilized into Agrobacterium tumefacians GV3101 and transformed into wild-type plants by the floral dip method (Miura et al., 2005). Hygromycin-resistant plants were selected, and diagnostic PCR was performed with the primers ICE1K393RF and NOS-R (see Supplemental Table 2 online).

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: SIZ1, AAU00414; ICE1, AAP14668; and HOS1, AAB87130.

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure 1. siz1 Plants Are Chilling Sensitive.

• Supplemental Figure 2. Sumoylation of ICE1 Induces CBF3/DREB1A and Represses MYB15 Expression.

• Supplemental Figure 3. ICE1 or ICE1(K393R) Expression in Transgenic Wild-type Plants.

• Supplemental Table 1. Primer Sequences Used to Detect SIZ1 or Genes Involved in Cold Signaling by the Quantitative PCR Method.

• Supplemental Table 2. Primer Sequences Used for Construction of Vectors.