A 21-kDa Chloroplast Heat Shock Protein Assembles into High Molecular Weight Complexes in Vivo and in Organelle*

The conservation of the carboxyl-terminal “heat shock” domain among small (sm) cytoplasmic and chloroplast heat shock proteins (HSPs) suggests that these smHSPs perform similar functions. Previous studies have established that cytoplasmic smHSPs are found in higher order structures in vivo (approximately 500 ma). To determine if the chloroplast smHSP is found in similar complexes, we examined the size of the 21-kDa chlo- roplast smHSP from Pisurn satiourn, PsHSP21, under non-denaturing conditions. Following sedimentation of chloroplast stromal extracts on sucrose gradients PsHSP21 is detected in fractions corresponding to 10-11 S. Upon non-denaturing gel electrophoresis, PsHSP21 was detected in two high molecular mass complexes of approximately 230 and 200 kDa, in good agreement with the sucrose gradient data. These PsHSP2l-containing particles were stable under different salt and Mg2‘ con- ditions, and their integrity was not affected by 1.0% Triton X-100 or 10 m~ ATP. To study assembly of the high molecular weight complexes containing PsHSP21, in vitro translated PsHSP21 was imported into chloro- plasts and its size was examined. Following import into chloroplasts isolated from heat-stressed plants, greater than 50% of PsHSP21 was recovered in the higher mo- lecular weight forms. In contrast, following import into chloroplasts isolated from control plants the protein was recovered exclusively in a


A 21-kDa Chloroplast Heat Shock Protein Assembles into High Molecular Weight Complexes in Vivo and in Organelle*
The conservation of the carboxyl-terminal "heat shock" domain among small (sm) cytoplasmic and chloroplast heat shock proteins (HSPs) suggests that these smHSPs perform similar functions. Previous studies have established that cytoplasmic smHSPs are found in higher order structures in vivo (approximately 500 m a ) .
To determine if the chloroplast smHSP is found in similar complexes, we examined the size of the 21-kDa chloroplast smHSP from Pisurn satiourn, PsHSP21, under non-denaturing conditions. Following sedimentation of chloroplast stromal extracts on sucrose gradients PsHSP21 is detected in fractions corresponding to 10-11 S . Upon non-denaturing gel electrophoresis, PsHSP21 was detected in two high molecular mass complexes of approximately 230 and 200 kDa, in good agreement with the sucrose gradient data. These PsHSP2l-containing particles were stable under different salt and Mg2' conditions, and their integrity was not affected by 1.0% Triton X-100 or 10 m~ ATP. To study assembly of the high molecular weight complexes containing PsHSP21, in vitro translated PsHSP21 was imported into chloroplasts and its size was examined. Following import into chloroplasts isolated from heat-stressed plants, greater than 50% of PsHSP21 was recovered in the higher molecular weight forms. In contrast, following import into chloroplasts isolated from control plants the protein was recovered exclusively in a 5 S (approximately  form. These data suggest that preexisting PsHSP21 or other heat-induced factors may be required for assembly of the higher molecular weight particles. We propose that the [10][11] S particles are the functional form of PsHSP21. Plants, like all other organisms, respond to high temperature or heat stress by synthesizing a new group of proteins designated as heat shock proteins (HSPs)' (1). The most abundant HSPs in plants are a group of small HSPs (smHSPs) which range in size from 15 to 30 kDa. Plant smHSPs belong to four multigene families and are localized to different cellular compartments including the chloroplast, cytoplasm, and endoplasmic reticulum ( 2 4 ) . Plant smHSPs exhibit a conserved car-R 0 1 GM42762 (to E. V.) and National Institutes of Health Grant F32 * Research was supported by National Institutes of Health Grant GM13953-10A1 (to K. W. 0.). The costs ofpublication of this article were therefore be hereby marked "advertisement" in accordance with 18 defrayed in part by the payment of page charges. This article must U.S.C. Section 1734 solely to indicate this fact. $ Present address: Dept boxyl-terminal "heat shock domain" (approximately 80 amino acids) that is also found in all other eukaryotic smHSPs, as well as in the vertebrate a-crystallin eye lens proteins (1,5).
Structural studies of cytoplasmic smHSPs and a-crystallins indicate that these proteins exist in higher order structures i n vivo. Arrigo and Welch (6) estimated that human HSP28 has a native molecular mass of 500 kDa, and recombinant murine HSP25 has been modeled as a 730-kDa complex composed of 32 HSP25 monomers (7). The native sizes of Saccharomyces cerevisiae and Drosophila smHSPs have similarly been estimated to be on the order of 500 kDa, as determined by sucrose gradient or gel filtration analysis (8)(9)(10). Plant cytoplasmic smHSPs have been reported in complexes of 500 kDa (11) and more recently in significantly smaller sized particles of approximately 240 kDa (12). Chicken HSP24 is also found in a smaller complex (180 kDa) (13). The a-crystallins are isolated in particles in the same size range (800 kDa) or a smaller form (300 kDa), which is suggested to represent the inner core of the 800-kDa complex (14). Interestingly, HSP27 and aB-crystallin have also been detected in the same 700-kDa complex in rat kidney cells (15). It has been hypothesized that the conserved heat shock domain is responsible for the similar structural properties of these proteins (5).
The correlation between expression of smHSPs and increased cellular thermotolerance has led to the hypothesis that smHSPs protect cells from heat-induced damage (1). When mammalian or Drosophila smHSPs have been overexpressed in mammalian cell lines, some increase in thermotolerance has been observed, providing additional evidence for a specific role of smHSPs in the thermotolerance phenomenon (16)(17)(18). However, deletion or overexpression of S. cerevisiae HSP26, the only identified smHSP in this organism, failed to provide evidence for a n essential role of smHSPs during stress (19,20). Thus the function of these proteins during heat stress remains poorly defined.
Although a number of functions have been proposed for the smHSPs (1,11), recent experiments with both the a-crystallin proteins and mammalian smHSPs have sugggested that these proteins function as "molecular chaperones" (21). Molecular chaperones are proteins that alter or maintain the structure of other polypeptides and thereby facilitate proper protein folding (22)(23)(24)(25). A recombinant mammalian smHSP prevented heatinduced aggregation of other proteins i n vitro, as assessed by light scattering, and increased the half-time of heat-induced inactivation of the enzyme a-glucosidase (22). It also increased the yield of active enzyme after dilution from denaturant (22). The mechanism of these effects remains unknown.
An smHSP (HSP21) that localizes to chloroplasts has been identified in many plant species (4,(26)(27)(28)(29)(30)(31). It is a nuclearencoded protein that is posttranslationally transported into chloroplasts. Comparative sequence analysis of the chloroplast smHSP from several species has defined both the conserved heat shock domain and a "methionine bristle" domain (28 amino acids) that is unique to the chloroplast smHSP (31).

13216
This is an Open Access article under the CC BY license.
Chloroplast smHSP mRNA is not found in unstressed plants but accumulates dramatically during heat shock (4, 32). The HSP21 protein increases co~espondingly during stress to an estimated 0.02% of total leaf protein. Following stress, the smHSP i s very stable, having a half-life of 52 h (32). Under moderate stress conditions, the majority of the protein is recovered from the solubIe protein fraction of the chloroplast (32).
The conse~ation of the heat shock domain among cytoplasmic and chloroplast smHSPs suggests that these proteins serve similar functions in different cellular compartments. We sought to determine if the chloroplast smHSP from Pisum satiuum (pea), PsHSP21, is also a component of a larger protein complex, similar to the previously described cytoplasmic smKSP complexes. By examining PsHSP21 in chloroplast extracts, as well as protein imported into isolated intact chloroplasts, we demonstrate that this smHSP is found in complexes of approximately 200-230 kDa. These results further support the hypothesis that there is functional homology between the cytoplasmic and chloroplast smHSPs.

MATERIALS AND METHODS
Plant Growth and Heat Shock Theatment-Pea seeds (E! sativum, cv. "Little marvel") were germinated in flats of vermiculite and grown in a growth chamber under either of two conditions: 1) in a growth chamber with 22 "C day/l8 "C night temperatures at a light intensity of240 pmol m-' s-l (photosynthetically active radiation), and a 16-h photoperiod; or 2) on lighted shelves at 23-25 "C, at a light intensity of 120 pmol m-' s-', and a 16-h photoperiod. No differences were noted in experiments using plants grown under the two conditions. All plants were watered with 114 strength Hoagland's solution and grown for 9 days. Unless otherwise noted, heat stress was performed on intact plants at 38 "C according to the heat stress regime described previously (32). Briefly, the growth chamber temperature was increased gradually (4 "CAI) to 38 "C, maintained at 38 "C for 4 h, and decreased to the control temperature (22 "C) at the same rate.
Chloroplast Isolation and Fractionation-Intact chloroplasts were isolated from control or heat-stressed plants using Percoll gradients as previously described (27,32). When using heat-stressed plants, chloroplasts were isolated either at the end of the 4 h maximal temperature (38 "C) stress for in vivo native structure studies, or 18 h aRer the end of the heat shock regime for import experiments unless otherwise indicated. Chloroplasts were lysed in buffer containing 10 m M HEPES, pH 8.0, 1.0 m M 2-mercaptoethanol, 5 m~ MgCI,, 150 m~ NaCl, and protease inhibitors (5.0 m~ r-aminocaproic acid, 1.0 m~ benzamidine, 1.0 m~ phenylmethylsulfonyl fluoride, 0.1 mg ml" pepstatin A, and 0.5 mg ml-1 each antipain, aprotinin, chymostatin, and leupeptin). Variations in the lysis buffer are detailed in the results and corresponding figure legends. The chloroplast lysate was spun for 10 min at 12,000 x g to pellet the membranes. The supernatant was designated as the soluble protein fraction.
Chloroplast in Vitro Import-Protein import into isolated chloroplasts was performed basically according to Vierling et al. (27). PsHSP21 was transcribed and translated in uitro in the presence of 135SlMet 11165 Ci mmol-I; DuPont NEN) using 0.75-1.0 pCi pl-l translation mix, as described previously (4). Import reactions were performed in a total volume of 150 1. 11 containing intact chloroplasts corresponding to 100-125 pg of chlorophyll and 25-30 pl of in vitro translation mix, Import was allowed to proceed in the lighted growth chamber at 22 "C for up to 23 min (continuous import), or import was terminated after 3 min by the addition of nigencin to 1.0 p~ (30 x stock in 90% ethanol) and KC1 to 50 m~ (33). Chloroplasts were reisolated following the addition of nigericin or aRer a 3-20-min post-import incubation period. Postimport incubation was carried out in the light at 22 "C. When indicated, chloroplasts were treated after import by adding thermolysin (100 pg ml" + 1 m CaC1.J to the import reaction and incubating for 3 or 15 min at room temperature in the light. Reisolated intact chloroplasts were lysed in buffer (1.0 ml/import reaction) containing protease inhibitors as for the chloroplast fractionation studies plus 5.0 m M EDTA. Typically 9697% of the starting chlorophyll was recovered in the intact chloroplast &action following imports performed for 3 min or less as determined by measurement of chlorophyll content. Recovery of intact organelles dropped to 65-88% following longer reactions. In time-course experiments, chloroplast samples were adjusted to equal chlorophyll content before further processing. For all reactions, membranes were removed from the lysed chloroplasts by centrifugation for 10-15 min at 12,000 x g and the soluble proteins were further analyzed as described below.
Sucrose Gradient Centrifugation-Sucrose gradients (17-34% sucrose) containing the same components as the chloroplast lysis buffer were used to estimate the sedimentation coefficient of PsHSP21. For in vivo studies, the soluble protein fraction corresponding to 500-800 pg of chlorophyll was loaded onto the gradient and centrifuged at 4 "C for 21 h at 40,000 x g in a Beckman SW 40 rotor. For analysis of in vitro import reactions, soluble protein corresponding to 80-100 pg of chlorophyll was fractionated on each gradient. m e r centrifugation, 12 1-ml fractions were collected and proteins in each fraction were analyzed either by SDS-PAGE and immunoblotting for in vivo studies, or by SDS-PAGE and fluorography for in vitro import studies. For in vivo studies, 60 pl of each gradient fraction was loaded on the gel. For in vitro studies, proteins from each fraction were precipitated with trichloroacetic acid and washed with 80% acetone, after which 25% ofeach gradient fraction was loaded on the gel.
Acrylamide Gel Electrophoresis and Immunoblotting-For analysis by SDS-PAGE, protein samples from the soluble fraction of chloroplasts or from sucrose gradients were mixed directly with 2 x SDS gel sample buffer (1 x buffer is 60 m~ Tris-HCl, pH 8.0, 60 m~ dithiothreitol, 2% SDS, 5.0 m M eaminocaproic acid, 1.0 m~ benzamidine, 15% sucrose) or were concentrated as described above by trichloroacetic acid precipitation before resuspension in 1 x SDS gel sample buffer. Samples were boiled for 2-3 min and separated by electrophoresis on 12.5% SDSp o l y a c~l a~d e gels (34). Non-denaturing gel electrophoresis was performed using the same buffer system as the denaturing gels but excluding SDS in all steps (35). For analysis of total leaf or root proteins, tissues were ground using a ground glass homogenizer in 2 x gel sample buffer (as above, but without SDS or DIT and containing 1.0 mg ml-' each antipain, aprotinin, and leupeptin) with a tissue to buffer ratio of 100 mg (fresh weight) m1-I. Homogenates were centrifuged at 12,000 x g for 15 min, and the supernatants corresponding to approximately 1.5 mg (fresh weight) for leaf tissue and 8.0 mg for root tissue were loaded on the gels. For experiments testing different chloroplast lysis buffers, the soluble fraction from lysed chloroplasts was brought to 6% glycerol and then loaded directly on the gel. When necessary, Triton X-100 was removed from the protein fraction prior to electrophoresis by passage through Sephadex G-25 (0.2-ml column bed volume) equilibrated in the standard lysis buffer without detergent. Electrophoresis on non-denaturing gradient gels (4-208 acrylamide) was carried out at 1.0 V cm" at 4 "C for 24-48 h. Molecular mass standards were bovine serum albumin (67 kDa), lactate dehydrogenase (140 M a ) , catalase (232 kDa), apoferritin (440 kDa), and thyroglobulin (dimer, 669 kDa) (Pharmacia LKB Biotechnology Inc.).
Following SDS-PAGE or native gel electrophoresis, either the gels were stained with Coomassie Brilliant Blue R, fixed, and exposed to x-ray film or proteins were transferred to nitrocellulose membrane filters for immune detection of PsHSP21. [s5SJMet-labeled PsHSP21 from in uitro import reactions was quantified by determining the radioactivity of PsHSP21 bands on the dried gels with a Betascope analyzer (Betagen). The countdmin fcpm) per band was typically 65-80, with a background of 1.0-1.5 cpm. Immunoblots were reacted with antiserum against PsHSP21, which is specific for the carboxyl-terminal segment of the protein (amino acids 131-2321 as characterized previously (36). Antibody reactions were performed as described (32) using a 15000 dilution of PsHSP21 antiserum, and antibodies were detected with lzSIprotein A (46.6 pCi pg -I; ICN Radiochemicals) or using chemiluminescence (ECL; Amersham Corp.). Western blotting results were quantified by counting the '251-labeled bands excised from the nitrocellulose filters (32). The range of cpm per excised band was 800-1000 cpm. In some experiments, proteins from chloroplast membrane fractions were also analyzed by SDS-PAGE and immunoblotting as for proteins from the soluble fractions.

PsHSP21 Is Found in High Molecular Weight Complexes in
Vzuo-Our previous studies have shown that, when heat stress is applied gradually and light intensity is moderate (250-350 PM m" s"), the majority of PsHSP21 is localized in the soluble protein fraction of the chloroplast (32). In order to examine the native size of soluble PsHSP21, total soluble chloroplast proteins from control or heat-stressed pea plants were separated on 17-34% sucrose gradients. Because the HSP2l polypeptide sucrosc) with thr peak (5'7";) in fractions 5 and 6 (24.5-26'; sucrose) (Fig. If3 ). Lrss than 2' ; of PsHSP21 \vas rccovrrrd in t h r pc>llrt fraction of the g-radicmt. Using bovine scv-um alhumin ( 6 S), catalase (11 SI, and rihulosc-hisphosphatr c:lrhoxylasr f Ruhisco) f 18 SI as standards on the gradients, a sedimentation v:llur of 10-11 S was rstimatrd for t h r fractions cnntaining PsHSP21. For a glohular protein, this valur corrc.sponds to an avrragc~ molecular mass of approximately 200 kDa. Other than PsHSI'21, an additional protein hand of 42 kI)a was also rcco p i z c d hy thv l'sFISf'21 antihodies iFig. In 1. The ahundance of this hand in vach fraction w a s always proportional to that of I'sHSP21 (Fig. 111 ). suggrsting it is a dimeric form of PsHSP21 that has not hrrn dcn:ltured hy t h r SDS snmplr huffrr.
shocked plants (Fig. 2). Therefore, the formation of t h r high molecular weight complexes containing PsHSP21 is not specific to photosynthetic organelles, hut is a process occurring in all types of plastids. These results also indicate that particle formation is not an artifact of the chloroplast isolation procrdure. since the particles arc found in whole tissue extracts as well. The total leaf and root extracts also behaved identically to chloroplast extracts on sucrose gradient analysis, with the major peak of PsHSP2l sedimenting at 10-11 S (Table I).
Stability of thr HSPZI-containing Conlplrxrs-The stahility of the PsHSP21-containing particles was investigated hy varying the chloroplast lysis conditions and analyzing the soluhle chloroplast proteins by sucrose gradient analysis (Tahle 1) or non-denaturing gel electrophoresis (Fig. 3). Conditions tested included increased or decreased NaCl and M?', or addition of EDTA, reducing agent, detergent (1'3 Triton X-100). and ATP. For the sucrose gradient analyses, the huffer components in the lysis and gradient solutions were identical. As sren in Table I, under a11 conditions tested, approximately 50('+ of PsHSP21 was detected in fractions sedimenting between 10 and 11 S (24.5-26V sucrose). None of the conditions caused apparent dissociation of the 10-11 S form or incrcased the fraction of HSP21 in the gradient pellet (<29 ). We also saw no change in protein distribution if chloroplasts were isolated the morning following the day of stress (24-h sample, Table I).
Analysis by non-denaturing gel electrophoresis of the soluble proteins from chloroplasts lysed under the same conditions listed in Table I is shown in Fig. 3 . Chloroplasts wcre incubated in all lysis solutions for 1 h a t 4 "C before removal of membranes, with the exception of the samples containing Triton X-100, in which the detergent was added to the soluble fraction after membranes wrre pelleted. None o f the conditions tested altered the pattern of immunoreactive hands; both the 230-and 200-kDa forms were detected in all samples at approximatclv the same ratio. The similar sedimentation and electrophoretic behavior of PsHSP21 under all of these conditions supports the conclusion that the high molecular weight forms are stable native complexrs containing PsHSP21. were designed to examinr thr s i z r of P.;llSWI ftrllowinK import into isolated chloroplasts. Assrmhly of chloropln.;t-sol~ll~l(~ and membrane protein complexrs has hrrn shown to procrrtl fnithfully in an isolatrd chloroplast systrm ( X , 381. Following t h r assembly of PsHSP21 into high molrculnr rornplrxrs in such n system was thrreforc, undcrtakrn to providr indtapc>ndrnt confirmation of thr native sizr of PsHSP21 and to dvfinr. condition.; required for asscmhly of thr PsHSP2l-cnntaining rompl(.xc~q. I t was of particular interrst to drtrrminc. if fartors produccd dllring heat strrss wrrr rrquircd for PsHSP'Ll nsscbmhly. :IS trstrd hy examining PsHSP21 nssrmhly in rhloroplnsts i . ; o l : l t c d from control or heat-strrssrd plants.

A Control Chloroolasts
was allowed to proceed continuously for up to 23 min or, to facilitate assessment of assemhly. was terminated after 3 min hy the addition of nigcvicin followed hv a n additional postimport incuhation period of 3-20 min. After import and postimport incubation, chloroplasts were treated with or without the protease thermolysin before rc.isolntion and analysis. Protease treatment is routinely used to demonstrate that imported proteins are protected within the organelle. Expcrimcmts were performed in parallel using chloroplasts isolated from either control or heat-stressed plants.
The amount of PsHSP21 recovered over time during the import experiment is shown in Fig. 4. When PsHSP21 was imported into control chloroplasts without adding nigericin to block import and without treating with protease, the amount of PsHSP21 imported into chloroplasts continued to increase throughout the 23-min incuhation, with a maximal rate in the first 10 min. When the ionophore nigericin was added to the import reaction after 3 min, no increase in imported PsHSP21 was ohserved between 3 and 23 min. Therc~forc~, nigericin ~f f c ctively uncoupled chloroplast photophosphorylation ( 3 3 ) and terminated the import. When the protease thcrmolysin was used to treat control chloroplasts, the same level of PsHSP21 was recovered from thermolysin-trcated and untreated chloroplasts (Fig. 4A ). This is consistent with thc previous ohservation that PsHSP21 is well protected from protease digestion after import into chloroplasts from control p1ant.s ( 4 , 2 7 1 .
Thermolysin treatment also effectively blocked further import; samples treated with and without thermolysin showed the same total imported cpm, although incubation in the light was continued during the thermolysin treatment. Presumahly thct protease immediately degrades the availahle precursor and therehy terminates import.
For experiments with heat-stressed chloroplasts, chloroplasts were isolated from plants 18 h following the end of the stress treatment because sufficient yields of intact organelles for import experiments were difficult to obtain directly following the stress. Our previous studies have shown that chloroplast PsHSP21 is present at ahout 82r; of the maximum level a t this time (32). Results showed that PsHSP21 import kinrtics were similar, hut not identical.
t o th;lt of thc. rontrol chloroplasts (Fig. 4n). Within 20 min thrb total amount of l'sFIS1'21 imported into chloroplasts from control or he:lt-strc-ssctl plants was similar, constituting 18-20"; of the total I'sHSI'2I used in the. import reaction. As shown for control chloropl:lsts, nigc-ricin effectively hlockrd import of protcsins into chloroplasts from heat-stressed plants. However. in contr:lst t o control chloroplasts. protease treatment had a dramatic c+Twt on thv rwovcv-y of PsHSP2l from chloroplasts of heat-strrsscd plants. :Is shmvn in Fig. 4n, only ahout half as much I'sFIS1'21 \ v a s rvcoverrd from t h r thcv-molysin-treated chloropl:rst< rnmparr-ri t o the untreated chloroplasts in hoth continuorts and niprricinterminated import exprrimcmts. In the vxpvrimcmt tvith nigcbricin. the amount of PsHSP21 rc-covcrcd from th(*rmolysintreated chloroplasts increnscd during thc firqt 10 min of postimport incuhation. Thrrr ~~a s no cvirtrbncr th:lt thv th(brmolysin treatment resulted in general protrolysis of many chloroplast proteins as assessed on Coomassic Rlrtr,-st:linc.d p 1 s 1 not shown I .
In import rxperimrnts using chloroplasts from heat-stressed plants, the scdimentation pattrrn of PsHSP21 was very diffrrc n t from that seen in control chloroplasts (Fig. 5 8 ). Similar to control chloroplasts, when no post-import incubation was performed the majority (approximately 6Y"' ) of PsHSP21 was recoverrd in fractions 2 and 3 ( 5 S ) . However, after 20 min of post-import incubation, the labeled PsHSP21 sedimented in two main peaks. one in the 5 S region (fractions 2 and 3 , . and anothrr new peak in the 10-11 S region (fractions 5 and 6 ) (Fig.  5 8 ). This new high molecular weight prak of importrd protrin cofractionatrd with thr endogenous PsHSP21 as cirtrrmined hy Western hlot analysis, indicating that the 10-11 S form of the importcd PsHSP21 \vas corrrctly assrmhled (not shown). Quantitative analysis of thr distrihution of PsHSP21 in the gradients rrvralrd that at least 50'; of importrd PsHSP21 assemhled into thr 10-11 S form within 20 min following import (Fig. 58 ).
When chloroplasts from heat-stressed plants werr treated with thermolysin. the distrihution of PsHSP21 recovered from the gradients was changcd significantly. As shown in Fig. 5 8 , the 5 S form of PsHSP2I was digested hv thermolysin. while the 10-11 S form was not affrctrd hy the protease treatment.
Quantitativr analvses suggrst that digrstion of the 5 S form of PsHSP21 (ahout 54V of total importrd PsHSP21) would account for the drcreasrd rrcovcry of protrin ohservrd in t h r thermolysin-treated chloroplasts from hrat-strrssed plants. Increased assrmhlv of t.he thrrmolvsin-resistant 10-11 S form would also account for increased HSP21 rrcovrry with time, as seen in Fig. 5R (snrnple 8).
The non-denaturing gel data indicate that the 10-11 s form of the protein comprises two separatr high molecular wright species that are not resolved hy sucrose kp-adient analysis. To determine if hoth of these sprcirs are formed during in r,itro import, the 3-min import sample and the 20-min post-import incubation samplr from experiments using control or hratstressed chloroplasts were examined hv nativr grl clectrophoresis and autoradiography (Fig. 6). In the 3-min sample, I'sHSP21 importcd into control chloroplasts is resolved as a low molecular wright species with an apparent mass of42 kDa. The amount of this species rrmains unchanged after a suhsequent 20-min incubation period, with no accumulation of high mo- Results of this study drmonstratr that. in its nativr statr, tho chloroplast smHSP is a compnncnt of high molrcular mass particles. LJndrr a varirty of chloroplast lysis cnnditions PsHSP21 exhibited the same sedimentation behavior on sucrose gradients and was resolved as the same two molecular mass forms of 200 and 230 kDa by non-denaturing gel electrophoresis. The integrity of the PsHSP2l-containing particles in the presence and absence of salt or divalent cations, as well as detergent, suggests that strong ionic interactions and hydrophobic forces stabilize these higher order structures. Particle size was also unaltered by the addition of ATP, indicating there is no energy dependent change in particle conformation or composition. The observation of the same high molecular weight forms in total root extracts demonstrates that formation of the particles does not require photosynthetic functions but occurs in all types of plastids. The presence of chloroplast PsHSP21 in a higher order complex is consistent with structural studies of cytoplasmic smHSPs and cy-crystallin proteins which share a conserved carboxyl-terminal domain with PsHSP21 (39, 40). We hypothesize that the PsHSP21 particles we have characterized are functionally equivalent to previously characterized cytoplasmic smHSP particles. A majority of studies have reported that the cytoplasmic smHSPs and a-crystallins are found in complexes of approximately 500 kDa. Although we have never observed PsHSP21 in discrete soluble particles larger than 230 kDa, we cannot rule out the possibility that the 200-and 230-kDa complexes are stable forms of a partially dissociated larger species. Furthermore, the methods we have used to estimate protein mass, density gradient analysis and non-denaturing gel electrophoresis, are both based on comparisons to globular protein standards. If the PsHSP21-containing particles deviate from a globular structure then their actual mass may be significantly different (41,42).
The composition of the cytoplasmic smHSP complexes is not well characterized. The complexes are generally thought to be oligomeric structures of the HSPs themselves, as is the case for the a-crystallin complexes (43). However, the smHSP complexes have been purified t o biochemical homogeneity only from HeLa cells, and in this case a high molecular weight polypeptide copurified with the smHSP (6). Although recombinant murine HSP25 formed a high molecular weight structure (7), this complex has not been directly compared to the native protein. We do not know the composition of the 200-and 230-kDa particles in the chloroplast, other than that they contain PsHSP21. Further work is needed to purify these complexes; however, in the absence of a defined assay for smHSP function, it is not yet possible to determine the structure of the active form.
In the in vitro import experiments PsHSP21 assembly is time-dependent and appears to lack a requirement for ATP. The application of the ionophore nigericin to the import reaction after 3 min did not abolish particle formation. These data indicate that the assembly process is not dependent on the presence of a proton potential across the thylakoid membrane after 3 min of import. The assembly of the PsHSP21-containing particles in the absence of high ATP in the chloroplast also suggests that chaperonin 60 (the Rubisco-binding protein or GroEL) is probably not involved in the assembly of PsHSP21containing particles. This is consistent with the observation that no PsHSP21 was found at the position of chaperonin 60 on the sucrose gradients or non-denaturing gels, as is the case for other proteins that interact with chaperonin 60 (44).
The inability of PsHSP21 to assemble into the high molecular weight complexes in control chloroplasts could be explained in several ways. One possibility is that in chloroplasts from heat-stressed plants a pre-existing pool of unassembled PsHSP21 molecules could accelerate the kinetics of particle assembly. Because there is no PsHSP21 in control chloroplasts (32), assembly in control chloroplasts would be mass-limited, because the level of in vitro imported PsHSP21 is extremely low. Another implication of pre-existing PsHSP21 in chloroplasts from heat-stressed plants is that the appearance of in vitro imported PsHSP21 in the assembled particle could result from monomer exchange between imported and endogenous PsHSP21 molecules. Failure to assemble in control chloroplasts could also be due to the need for other heat-induced factors or to posttranslational modifications specific to the heat-stressed chloroplasts. Osteryoung et al. (45) have observed that when the chloroplast smHSP from Arabidopsis thaliana, AtHSP21, is constitutively overexpressed in Arabidopsis plants it will assemble into the native, higher order structure even in plants which have never experienced heat stress. This result suggests that additional heat-induced factors are not required for PsHPS21 assembly. However, it is possible that overexpression of AtHSP21 causes other alterations in the chloroplast that then favor assembly. We currently favor the hypothesis that assembly in control chloroplasts is mass limited. To test this hypothesis, we are developing vectors for overexpression of the PsHSP21 precursor in Escherichia coli to enable isolation of large quantities of precursor which can then be used to increase the mass of PsHSP21 imported into control chloroplasts.
In isolated chloroplasts, about 50-70% of the total imported PsHSP21 assembled into the 200-and 230-kDa forms and the rest remained in a form at approximately 5 S on sucrose gradients and 42 kDa on non-denaturing gels. In nigericin-treated chloroplasts the 5 S/42-kDa form behaved as a precursor to the 200-and 230-kDa forms; the 5 S/42-kDa form decreased concomitant with the increase in the higher molecular weight forms. Whether the small form is a dimer of PsHSP21 that is an assembly intermediate in vivo remains to be tested. The protease sensitivity of the 5 S form in chloroplasts from heatstressed plants that otherwise appear intact is curious. We suggest this results from a heat-induced alteration of the chloroplast envelope that allows penetration by thermolysin. However, previous studies have indicated that the chloroplast envelope is quite heat-stable in vitro (46), and the protein import functions of the membranes are clearly intact. An alternative explanation is that CaC1, added with the thermolysin activates an endogenous chloroplast protease that accumulates or is activated following heat stress. We have not tested this possibility, nor are we aware that such a protease activity is found in chloroplasts.
Recent data suggest that the a-crystallins and smHSPs are molecular chaperones that act in an ATP-independent fashion to prevent aggregation of other proteins (25). Current models propose that they function by binding denatured substrates, thereby reducing their concentration and allowing free molecules to undergo folding. There is no evidence that they actively alter protein conformation as do the chaperonin 60s (47). The presence of smHSPs in three plant cell compartments, cytoplasm, endoplasmic reticulum, and chloroplast, points to a critical role for these proteins in plant stress and much additional work is required t o clarify their mechanism of action and to identify their critical in vivo substrates.