In vivo modifications of the maize mitochondrial small heat stress protein, HSP22.

A maize (Zea mays L.) small heat shock protein (HSP), HSP22, was previously shown to accumulate to high levels in mitochondria during heat stress. Here we have purified native HSP22 and resolved the protein into three peaks using reverse phase high performance liquid chromatography. Mass spectrometry (MS) of the first two peaks revealed the presence of two HSP22 forms in each peak which differed in mass by 80 daltons (Da), indicative of a monophosphorylation. Phosphorylation of HSP22 by [gamma-(32)P]ATP was also observed in mitochondria labeled in vitro, but not when purified native HSP22 was similarly used, demonstrating that HSP22 does not autophosphorylate, implicating a kinase involvement in vivo. Collisionally induced dissociation tandem MS (CID MS/MS) identified Ser(59) as the phosphorylated residue. We have also observed forms of HSP22 that result from alternative intron splicing. The two HSP22 proteins in the first peak were approximately 57 Da larger than the two HSP22 proteins in the second peak. MS analysis revealed that the +57-Da forms have an additional Gly residue directly N-terminal of the expected Asp(84), which had been converted to an Asn residue. These results are the first demonstrations of phosphorylation and alternative intron splicing of a plant small HSP.

The cytosolic forms of plant sHSPs have been shown to have chaperone function (18), and, recently, evidence that mitochondrial sHSPs function as chaperones has been provided by experiments of Downs and Heckathorn (19). They have shown in a number of different ways that mitochondrial HSP22 protects complex I activity during heat stress. This work was performed using apple mitochondria from control and heat-stressed fruits and with mitochondrial HSP22 purified from heat-stressed fruit. They convincingly show that complex I is the most heatsensitive component of the mitochondrial electron transport chain and that HSP22 protects its function during heat stress.
There is growing evidence for the function of HSPs in a number of different types of stress. Presumably, the HSPs function as chaperones that stabilize denatured proteins. These stresses include oxidative (20,21), ␥ irradiation (20), cold stress (22,23), and drought (24). Exposure to heat stress increases subsequent resistance of plants to other stresses (25)(26)(27)(28), which supports a general protective function for HSPs. Some sHSPs are present constitutively in vegetative tissues at very low levels, and their expression increases rapidly upon heat stress. In contrast, other sHSPs seem to be expressed highly in tissues that are embryogenic such as zygotes and somatic embryos (15, 24, 29 -33). Thus, it appears that, depending upon the structure of the promoter region, some genes are expressed developmentally, some during heat stress, or both (22,29,31). The maize mitochondrial HSP22 that we have been characterizing is expressed at very low levels constitutively in vegetative tissues and is strongly induced by heat stress (1). In this paper we describe the first analysis of a sHSP from plants using mass spectrometry. We have found that maize mitochondrial HSP22 exists in four forms that result from phosphorylation and differential intron splicing. This is the first instance where a plant sHSP has been shown to be phosphorylated in vivo, and the first time that alternative intron splicing of a plant sHSP has been demonstrated.

EXPERIMENTAL PROCEDURES
Materials-All molecular biology chemicals were from Sigma or Fisher unless otherwise noted. Taq polymerase was from Life Technologies, Inc. Restriction enzymes were from Promega (Madison, WI) or New England Biolabs (Beverly, MA). Calf intestinal alkaline phosphatase and ligase were from Roche Molecular Biochemicals.
Plant Material, Mitochondrial Isolation, and Sub-fractionation of Mitochondria-Maize (Zea mays L. inbred B73) seeds were imbibed and planted as described by Lund et al. (1) and grown at 29°C for 3 days in the dark. To induce HSP22 expression, entire trays of seedlings were placed in an incubator for 4 h at 42°C. Mitochondria were isolated from these etiolated shoots as described previously (34,35). The protein content was measured using the Lowry procedure as modified by Larson et al. (36). Isolated mitochondria were suspended in a medium consisting of 250 mM sucrose and 30 mM Mops (pH 7.2). Mitochondrial membranes were removed by sonication of 6 mg of washed mitochondria in 1 ml of 30 mM Mops (pH 8.0), followed by ultracentrifugation in a TLA-100.2 rotor (Beckman) for 30 min at 100,000 ϫ g. The supernatant contained the soluble mitochondrial proteins and the large protein complexes as described by Hayes et al. (34).
One-and Two-dimensional Gel Electrophoresis-One-dimensional SDS-PAGE was performed with a Bio-Rad Mini-Protean II apparatus using a 14% (w/v) resolving gel and a 5% (w/v) stacking gel. Other conditions are as described by Elthon and McIntosh (37). Molecular mass markers used were Bio-Rad low molecular weight standards. Two-dimensional isoelectric focusing/SDS-PAGE was performed as described by Barent and Elthon (38). Pharmalyte 3-10 ampholytes (Amersham Pharmacia Biotech) were used in the first dimension.
Polyclonal Antibodies, Monoclonal Antibodies, and Immunoblotting-Affinity-purified polyclonal antisera and a monoclonal antibody for maize mitochondrial HSP22 proteins were produced as described by Lund et al. (1). For immunoblots, protein gels were transferred to nitrocellulose and probed with antibodies according to Hayes et al. (34). Goat anti-mouse IgG antibodies conjugated with alkaline phosphatase were purchased from Sigma. Proteins transferred to nitrocellulose were reversibly visualized by staining with 0.2% (w/v) Ponceau S in 3% (w/v) trichloroacetic acid for 2 min, followed by rinsing with deionized distilled H 2 O. Blots were fully destained prior to antibody probing by washing with phosphate-buffered saline containing 0.3% (v/v) Tween 20.
In Vitro Phosphorylation of Mitochondrial HSP22-Maize seedlings were grown at 29°C for 3 days in the dark and heat-shocked at 42°C for 4 h. Intact mitochondria were isolated as described above; resuspended in 250 mM sucrose, 30 mM Mops, 5 mM MgCl 2 , and 1 mM CaCl 2 (pH 7.2); and stored on ice. After 10 min, [␥-32 P]ATP (Ͼ3000 Ci mM Ϫ1 ; Andotek, Irvine, CA) was added (0.05 Ci g Ϫ1 mitochondrial protein) and the mixture was incubated on ice for 5 min. The mitochondria were pelleted by centrifugation at 14,000 ϫ g in an Eppendorf microcentrifuge for 5 min at room temperature. The supernatant was aspirated, and the mitochondrial pellet was resuspended in deionized distilled H 2 O for subsequent two-dimensional SDS-PAGE analysis. The proteins in the gel were transferred to nitrocellulose (34) and exposed to X-Omat AR film for 6 days and then immunoblotted as described above.
Chromatographic Purification of HSP22-Approximately 5 mg of mitochondrial proteins (after removal of the membrane fraction) from heat stressed maize seedlings was applied to a Amersham Pharmacia Biotech fast protein liquid chromatography HR5/5 Mono-Q anion exchange column. The column was washed at a flow rate of 0.5 ml min Ϫ1 with 10 ml of 30 mM Mops (pH 8.0) and then developed with a 25-ml linear gradient to 35% (v/v) 1 M NaCl, 30 mM Mops (pH 8.0). The gradient was increased to 100% (v/v) 1 M NaCl, 30 mM Mops (pH 8.0) during the following 3 ml and then maintained at 100% (v/v) 1 M NaCl, 30 mM Mops (pH 8.0) for 5 ml. The eluate was analyzed using a UV flow cell with a 1-cm path length and an illuminated volume of 8.7 l and collected in 0.5-ml fractions. The samples were analyzed for the presence of HSP22 using SDS-PAGE immunoblots. The HSP22-containing fractions were pooled and concentrated using an Amicon device with a 10-kDa cut-off membrane (YM-10 Diaflo Ultrafilter, Amicon, Beverly, MA.). Solid NaCl was added to the pooled fraction to a concentration of 4 M and the sample applied to a hydrophobic interaction column (Amersham Pharmacia Biotech HR5/5 Phenyl-Superose) equilibrated with 4 M NaCl, 30 mM Mops (pH 8.0). Nearly pure HSP22 was collected from the column eluate prior to application of a decreasing salt gradient. The HSP22-containing fractions were again pooled and concentrated using the same Amicon device.
On-line Reverse Phase HPLC-MS Analysis of Purified HSP22-The pooled and concentrated HSP22 peak recovered from Phenyl-Superose chromatography was applied to a Gilson HPLC (Gilson Inc., Middleton, WI) fitted with a C8 microbore (1.0 ϫ 50.0-mm Zorbax C8 300 Å) reverse-phase HPLC column. The column was developed at 50 l min Ϫ1 over 40 min with a 2-60% (v/v) acetonitrile/deionized distilled H 2 O gradient containing 0.1% (v/v) trifluoroacetic acid. Through the use of a splitting tee, 90% (v/v) of the eluate was directed through a UV flow cell (0.5 l illuminated volume) with the detector response set to 0.02-AU full scale and 215-nm wavelength. The other 10% (v/v) of the sample was analyzed on a Micromass Platform II mass spectrometer (Micromass Ltd., Manchester, United Kingdom) utilizing an electrospray ionization source and a quadrupole analyzer with a 8 s scan time from 700 -1800 m/z. The detector was calibrated using horse heart myoglobin (M r 16,951.4) and was found to be accurate to Ϯ0.01%. Mass spectra were analyzed using the MassLynx software (Micromass Ltd.).
On-line LC/MS Analysis of the HSP22 Tryptic Peptides-HSP22 peak I and peak II samples were lyophilized and each digested with freshly prepared trypsin (treated with N-tosyl-L-phenylalanine chloromethyl ketone) (T-8642 Sigma) at 25:1 HSP22/trypsin molar ratio in 0.1 M Tris pH 7.8 for 4 h at 37°C. N-Tosyl-L-phenylalanine chloromethyl ketone inhibits the chymotryptic activity (nonspecific cleavage at aromatic amino acids) of trypsin; therefore, trypsin cleaves the protein only at the C-terminal side of Arg and Lys residues. The tryptic peptides were separated and analyzed using the on-line HPLC-MS system described above but using a microbore C18 reverse phase column and developed with a gradient of 2-40% (v/v) acetonitrile/deionized distilled H 2 O containing 0.1% (v/v) trifluoroacetic acid over 40 min at a flow rate of 50 l min Ϫ1 . The Platform II mass spectrometer was calibrated with NaI using an 8-s scan time from 400 to 2000 m/z. Peptide mass determinations are accurate to Ϯ0.3 Da. Post-column splitting allowed collection of 90% of the peptide containing eluent. This eluent was fractionated manually based on A 215 peaks and the fractions stored at Ϫ80°C.
CID Tandem MS Analysis of the HSP22 Tryptic Peptides-Fractions containing the peptides of interest were further analyzed using a Finnigan LCQ electrospray ion-trap mass spectrometer (Finnigan MAT, San Jose, CA). The stored fractions were thawed and infused at 1-5 l Ϫ1 min using a syringe pump. The fractions were first analyzed while the LCQ was configured for profile data collection from 300 -2000 m/z using the following settings: capillary temperature, 135°C; maximum ion inject time, 900 ms; full scan target, 9 ϫ 10 7 ; and 3 microscans/ full scan. Once the parent ion was identified, the mass spectrometer was setup for MS/MS analysis. For CID MS/MS experiments, the instrument was configured as follows: capillary temperature, 150°C; maximum ion inject time, 400 ms; full scan target, 2 ϫ 10 7 ; 5 microscans/full scan; mass window, 1.0; and fragmentation energy was varied from 0% to 60%. CID fragmentation was found to be significantly better if the doubly charged parent ion was used; the M ϩ H ϩ ion could also yield results, but higher fragmentation energies were necessary.
Protein and Nucleotide Sequence Analysis and Comparison-All sequence analysis and comparison was performed with the Wisconsin Package Version 10.0 (Genetics Computer Group, Madison, WI). Peptide CID fragmentation ions and molecular weight determinations for proteins and peptides were calculated using the BioLynx software package (Micromass Ltd.).
Isolation of the Hsp22 Gene and Vector Constructs-Based on the sequences of the cDNA clone, primers were designed for PCR amplification of a partial genomic clone containing the full coding region. The names and sequences of the primers used were: for the 5Ј end of the gene, HSP22-Nco-3F (ggaggggagagtgtcgccatggcttccattgtcgcttcc), which contains an introduced NcoI restriction site; and for the 3Ј end of the gene, HSP22-Bam-1R (gtgcccctccaatcctggatcccttaccggcacttgctt), which contains an introduced BamHI restriction site. The PCR reactions were performed using a Robocycler (Stratagene, La Jolla, CA) with the following parameters: 1 cycle of 3 min at 94°C, 35 cycles of 45 s at 94°C, 30 s at 60°C, 1.5 min at 72°C, 1 cycle of 1 min at 72°C. Amplification products were analyzed by standard agarose gel electrophoresis and purified from the gels using the Geneclean II kit (BIO 101, Vista, CA). The PCR products were digested with BamHI and NcoI and were purified by agarose gel electrophoresis and Geneclean II again. The digested products were ligated into BamHI/NcoI-digested pUC120 and transformed into electrocompetent XL1-Blue cells (Stratagene) according to the protocol of the manufacturer. Plasmid DNA was purified from individual clones using the Qiagen Spin Mini Plasmid Preparation kit (Qiagen, Inc., Valencia, CA) using the protocol of the manufacturer.
Four unique subclones containing BamHI/NcoI inserts of the predicted size were sequenced by the University of Nebraska Center for Biotechnology DNA Sequencing Core Facility using the dideoxy chain termination procedure with the Li-Cor (Li-Cor, Inc., Lincoln, NE) labeling dyes following the Li-Cor protocol and products were analyzed using the Li-Cor model 4000 or 4000L DNA sequencer.

RESULTS
HSP22 Purification-Previous studies indicated that HSP22 could be resolved using two-dimensional SDS-PAGE into two distinct Coomassie-stained spots, which were both ϳ22 kDa in mass (1). The two forms of HSP22 separated because of different isoelectric points, and we termed these spots HSP22 (acidic) and HSP22 (basic). N-terminal microsequencing of both forms indicated that the protein sequences were identical. Polyclonal and monoclonal antibodies were raised to these proteins (1) and were used to evaluate levels of HSP22 on both two-dimensional and one-dimensional immunoblots. In twodimensional immunoblots, the two HSP22 spots often appeared as four or five overlapping spots. To determine the nature of the potentially different forms of mitochondrial HSP22, we set out to purify these proteins to provide material for further analysis. Fig. 1 shows the results of anion exchange (Mono-Q) chromatography separation of mitochondrial-soluble plus complex proteins. Aliquots of chromatography fractions were analyzed by SDS-PAGE and either Coomassie-stained to reveal the total protein profile of the fractions (Fig. 1, center) or immunoblotted using the affinity-purified HSP22 polyclonal antibody to visualize the proteins (Fig. 1, bottom). Analysis of the entire elution profile (data not shown) reveals that most of the HSP22 elutes in fractions 59 -73 (Fig. 1, lower panels). Subsequent chromatography of the pooled HSP22 peak on a hydrophobic interaction column (Phenyl-Superose) removes nearly all other proteins (data not shown).

Analysis of Purified HSP22 by Reverse Phase Liquid Chromatography and Electrospray Ionization Mass Spectrometry-
The pooled HSP22 peak recovered from Mono-Q followed by Phenyl-Superose chromatography was concentrated and applied to a reverse phase HPLC. The three most prominent peaks that were observed in the UV trace of the eluate occurred at 29.2, 30, and 31.8 min retention time and they were labeled I-III (Fig. 2, top). Coomassie-stained SDS-PAGE analysis of the three peaks and the original sample that was applied to the column indicated that peaks I-III contained pure samples of 22-kDa proteins. Peak II was observed to be the most abundant, peak I was found to contain considerably less protein, and peak III appeared to contain very little protein (Fig. 2, bottom  left). Immunoblot analysis of these fractions with affinity-purified HSP22 polyclonal antibody confirmed that each of these fractions contained HSP22 (Fig. 2, bottom right). Subsequent analysis of this immunoblot with monoclonal antibody cpn60B (data not shown) revealed that the 64-kDa protein in the sample applied to the column (Fig. 2, bottom left) is the mitochondrial homolog of cpn60 (1).
During the elution of the three HSP22 peaks, a portion of the eluate was analyzed by electrospray ionization-MS. The mass spectral data from peaks I and II were analyzed and transformed using the MassLynx software package to give the average mass values for the proteins in each peak. The scans recorded during elution of peak I were combined and transformed to reveal the presence of two proteins in the peak (Fig.  3, broken line). Peak I was found to be primarily composed of a protein with mass of 19,428 Da and less abundant component with mass of 19,508 Da. The scans corresponding to the elution of peak II were combined and transformed to reveal two proteins with the major species having a mass of 19,371 Da and the minor species having a mass of 19,451 Da (Fig. 3, solid line). The difference between the major and minor components in both peak I and peak II is 80 Da, which indicates that minor components have been modified in vivo by the addition of a phosphate group. Comparison of the two primary components of peak I (19,428 Da) and peak II (19,371 Da) and the two minor components of peak I (19,508 Da) and peak II (19,451 Da) reveals a mass difference of 57 Da. An accurate mass for the protein(s) present in peak III could not be determined, and since peak III contained little protein, we have not characterized it further.
In Vitro Phosphorylation of Maize Mitochondrial HSP22-To provide supportive evidence that the maize mitochondrial form of HSP22 can be phosphorylated, an in vitro phosphorylation assay was performed using isolated intact mitochondria (as described under "Experimental Procedures"). Autoradiographs of the transferred two-dimensional gels revealed that several proteins had become phosphorylated in both the control (data not shown) and heat-stressed samples (Fig. 4, top). Most of the 32 PO 3 Ϫ that was present on both control and heat-stressed blots appeared to be associated with spots corresponding to the 43-kDa pyruvate dehydrogenase E1-␣ subunit protein (39) which has been shown to be heavily phosphorylated. In blots of protein from HS tissue, a cluster of 22-kDa spots and a 30-kDa spot were observed to be labeled in addition to the labeled spots of control samples. Probing the blots with the HSP22 monoclonal antibody revealed that these proteins corresponded to HSP22 (Fig. 4, bottom). When the autoradiograph and immunoblot images were superimposed, the most acidic spot in the HSP22 cluster was labeled to the highest specific activity (Fig.  4, filled arrows). Some phosphorylation was observed to be associated with the more basic forms of HSP22 and with the 30-kDa basic form of HSP22, which may be HSP22 with its transit peptide (Fig. 4, open arrows), but to a lower specific activity. To determine if mitochondrial HSP22 can autophosphorylate, an aliquot of the purified protein was treated under similar experimental conditions and was not observed to become phosphorylated (data not shown). These results suggest that a kinase is likely involved in the phosphorylation of maize mitochondrial HSP22.
Determination of the HSP22 Phosphorylation Site-Native HSP22 was purified from heat-stressed maize seedlings and used to identify the site(s) of protein phosphorylation. Anal-ysis of the mature HSP22 protein sequence using the Biolynx software identified nine tryptic peptides that contain the potentially phosphorylatable residues Ser, Thr, and Tyr. These peptides are shown in Table I along with the expected mass of the unphosphorylated peptide. Further analysis of the mature HSP22 protein sequence with NetPhos 2.0 software indicated five potential serine phosphorylation sites, with poor probabilities for phosphorylation of the Thr and Tyr residues. The purified HSP22 peak II sample was digested with trypsin and the peptides analyzed by on-line LC/MS as described under "Experimental Procedures." Analysis of the LC/MS data indicated that eight of the nine peptides of interest were present in the unmodified form (Table I). Peptide T3 was the only peptide that was found to have detectable levels of ϩ80 Da modification that is indicative of phosphorylation. Peptide T3 contains three potentially phosphorylatable residues, Tyr 54 , Ser 59 , and Ser 63 . Ser 59 and Ser 63 were predicted by the NetPhos 2.0 software to have high probability for phosphorylation. To definitively assign one of these residues as the site of phosphorylation, the LC/MS samples that contained the highest amount of the phosphorylated T3 peptide (mass ϭ 1435.6 Da) and the unmodified T3 peptide (mass ϭ 1355.6 Da) were subjected to CID MS/MS in a quadrupole ion trap mass spectrometer. The results with the phosphorylated T3 peptide are presented in Fig. 5 and show that Ser 59 is the only amino acid that is phosphorylated. These results prove that the ϩ80 Da shifts in mass (Fig. 3), observed during mass spectrometric analysis of peaks I and II from reverse phase HPLC separation of purified mitochondrial HSP22, are due to phosphorylation of Ser 59 .
Alternative Intron Splicing in Maize Mitochondrial HSP22-An explanation for the ϩ57-Da modification of mitochondrial HSP22 became apparent after sequencing of a region of the genomic clone that was isolated through PCR as described under "Experimental Procedures." The region that was sequenced corresponded to the entire coding region and was found to contain a 134-base pair intron starting at position 253 FIG. 3. Mass spectrometric analysis of HSP22 reverse phase HPLC peaks I and II. The mass spectra from each of the HPLC peaks (data not shown) was combined and transformed using the MassLynx software package to determine the average molecular mass for the protein components in each peak. The mass shown for each peak is given as the 80% centroid value. Peak I is shown with the dashed line, and peak II is shown as the solid line. from the first base of the start codon. Data base comparisons indicated that the Arabidopsis mitochondrial hsp22 has a similar intron (40). Analysis of the 5Ј splice site indicated that if an alternative splice site just 3 base pairs downstream is used, then the original Asp 84 (115 Da) would be replaced by a glycine (57 Da) and an asparagine (114 Da), the difference being ϩ56 (Fig. 6). To evaluate this possibility, we used CID tandem mass spectrometry to analyze the T6 peptide, which corresponds to the potentially spliced region of the protein. The results shown in Fig. 7 indicate that the alternatively spliced T6 peptide is present in peak I, whereas the normal T6 peptide is present in peak II. These results prove that the alternative intron splicing event occurs in vivo and that the model presented in Fig. 6 is correct. To further evaluate the event of intron splicing, we re-evaluated sequencing results from 25 HSP22 cDNAs that were previously sequenced in an effort to obtain a full-length cDNA (1). Of the 25 cDNAs sequenced, 4 showed alternative splicing (data not shown).

DISCUSSION
Using purified native protein, we have shown that native mitochondrial HSP22 exists in four forms that result from phosphorylation and alternative intron splicing. We can estimate the amounts of the different forms of HSP22 through analysis of our chromatographic data. The results of Fig. 2 showing separation of fast protein liquid chromatography-purified HSP22 on a C8 reverse phase column indicate that peak II, the normally spliced form of HSP22, is the most prevalent TABLE I Identification of potentially phosphorylated tryptic peptides in purified maize mitochondrial HSP22 (peak II) The MS data from the LC/MS analysis of the HSP22 tryptic peptides were analyzed by searching for the m/z values ((M ϩ nH ϩ )/n) where M ϭ the monoisotopic mass of each peptide containing a Thr, Ser, or Tyr residue, and where n ϭ 1, 2. The tryptic peptide names and sequences are given, and the amino acid position given is relative to the start of the HSP22 protein with transit peptide intact. To determine if these peptides were also present in a phosphorylated state, the data were searched for the phosphorylated m/z values ((80.0 ϩ M ϩ nH ϩ )/n), where n ϭ 1, 2.   form of HSP22. Estimation of the area under these peaks indicates that peak II constitutes 86% of the HSP22 pool (peak I plus peak II). As indicated previously, we could not determine the mass of the protein(s) in peak III and did not characterize it further. The normally spliced form of HSP22 (86% of peak I plus peak II) is 36% phosphorylated, based upon estimation of the area under the peaks of Fig. 3. Based upon a similar estimation, the alternatively spliced form of HSP22 (14% of peak I plus peak II) is 35% phosphorylated. These results are indicative of mitochondrial HSP22 isolated from 3-day-old etiolated maize seedling shoots that had been heat-shocked at 42°C for 4 h. The relative amounts of the different forms of HSP22 in vivo in different tissue types, under other physiological conditions, in different developmental stages, or under other stress conditions is currently not known.
Maize mitochondrial HSP22 is the first plant sHSP shown to be phosphorylated. Suzuki and co-workers (7) were unable to find any evidence for phosphorylation of the pea cytosolic or chloroplastic sHSPs using H 3 32 PO 4 in vivo feeding studies. This supported the findings of Nover and Scharf (41), who studied suspension cultures of tomato cells and found no evidence of mitochondrial sHSP phosphorylation. Sorghum, millet, and barley have all been examined for the presence of phosphorylated sHSPs, but no phosphorylation was detected (42). In light of these findings, it has been proposed by Waters et al. (5) that the lack of sHSP phosphorylation observed in plants is a distinguishing feature of plant sHSPs with respect to the mammalian orthologs, which are phosphorylated and contain multiple sites for this type of modification.
In animal systems, phosphorylation of HSP22 is correlated with a reduction in the oligomerized state. It has been suggested, but not proven, that this reduction in oligomerization increases HSP22 chaperone activity. It would likely increase the accessibility of HSP22 to substrate proteins, particularly membrane-associated proteins. In this light, Downs and Heckathorn (19) have proven that mitochondrial HSP22 protects mitochondrial complex I from heat stress. In animal systems that have been well characterized, phosphorylation of the sHSPs is regulated by a mitogen-activated protein kinase scheme (8,9,(43)(44)(45). Whether a system like this is functional with any plant mitochondrial HSP22s is not known. However, maize is a heat-tolerant plant, and further characterization of the maize mitochondrial system will reveal the nature of the kinase that may be involved and any potential similarity to the animal systems that have been described. The effect of phosphorylation on mitochondrial HSP22 chaperone activity and oligomerization state remains to be investigated.
In addition to phosphorylation of maize mitochondrial HSP22, we have shown that the gene is susceptible to alternative intron splicing. The maize and Arabidopsis genes contain single introns of very similar size and in analogous positions. Both genes contain a single intron with the common GT motif at the 5Ј end of the intron and the AG motif at the 3Ј end of the intron. Unlike the Arabidopsis gene, the maize HSP22 gene also has an alternative splice site 3 base pairs downstream of the A nucleotide in the 5Ј motif. Through mass spectrometry and sequence analysis of multiple cDNAs, we have established that the alternatively spliced form is expressed at the translational level in vivo. The significance of alternative splicing to HSP22 function is not known, but the intron splice site is not near either of the two highly conserved regions, consensus regions I and II, that define the sHSPs. On the other hand, the alternatively spliced form of maize mitochondrial HSP22 contains an additional glycine residue and an aspartate residue is replaced by an asparagine residue. These changes could alter the local environment because it results in the conversion of an acidic Asp residue to a neutral Asn residue and insertion of an additional neutral amino acid glycine. These modifications change the calculated isoelectric point for the mature protein (transit peptide removed) from 5.38 to 5.52. There are numerous examples in which single amino acid substitutions dramatically alter protein function (13), and there are also several examples in which a substitution in regions far from the active site affects function (16). The changes in maize mitochondrial HSP22 change a region, which is predicted by the Garnier-Osguthorpe-Robson method (17) of secondary structure prediction to be in a ␤-sheet conformation, to a region predicted to form a turn. According to the Chou-Fasman method of secondary structure prediction, the unaltered region is not predicted to have secondary structure, but the modified region is predicted to form a turn. Whether or not the changes associated with alternative intron splicing result in changes in maize mitochondrial HSP22 function is an important question that remains to be investigated.