Assessment of Plant Chaperonin-60 Gene Function in Escherichia coli*

expressed separately and in combination produce three novel Escherichia coli strains: a, 8, and a& In /3 and a/3 cells, the plant gene products assemble effi- ciently into tetradecameric cpn6014 species, including novel hybrids boxylase (rubisco) in E. the the the is limited the groE operon accumulation rubisco the activity of the plant cpn60 products and on endogenous chaperonin functions. Expression of cpn-60a rubisco activity of soluble protein ex- tracts was assayed as described (18) with modifications. Final conditions in 200-pl assays were: 10-400 pg of protein, 100 mM Tris-CI, 20 mM MgCI,, 1 mM EDTA, 1 mM 8-mercaptoethanol, 25 mM NaHCO:% ('"C02 specific activity of 1.85 Bq/nmol), 0.4 mM ribulose bisphos- phate, pH 8.0. Reactions were initiated by the addition of ribulose bisphosphate and NaH'TOa and were terminated after 10 min.

Brassica napus chaperonin-60a and chaperonin-60B genes expressed separately and in combination produce three novel Escherichia coli strains: a, 8, and a& In / 3 and a/3 cells, the plant gene products assemble efficiently into tetradecameric cpn6014 species, including novel hybrids containing both bacterial and plant gene products. The levels of authentic groEL14 are reduced in these cells (Cloney, L. P., Wu, H. B., and Hemmingsen, S. M. (1992) J. Biol. Chem. 267,23327-23332).Theassemblyofcyanobacterialribulose-Pzcarboxylase (rubisco) in E. coli requires the activities of the endogenous chaperonin proteins. Furthermore, the extent to which assembly occurs is limited by the normal levels of expression of the groE operon (Goloubinoff, P., Gatenby, A. A., and Lorimer, G. H. (1989) Nature 337, 44-47). We have now monitored the accumulation of cyanobacterial rubisco in E. coli a, 8, and a8 cells to assess the activity of the plant cpn60 gene products and effects on endogenous chaperonin functions. Expression of cpn-60a alone did not enhance rubisco assembly, which is consistent with our previous observation that p60cp"-80" required the presence of p60cp"~608 for assembly into cpn6014 species. In contrast, expression of cpn-60B alone resulted in markedly enhanced rubisco assembly in cells that accumulated normal levels of both endogenous chaperonin polypeptides (groEL and groES). This demonstrates that assembled p60cp"-808 is functional as a chaperonin in E.
coli. Co-expression of cpn-60a and cpn-60B in cells with normal levels of expression of groES and groEL suppressed rubisco assembly. Increased expression of groES in cells in which cpn-6Oa and cpn-6OB were coexpressed relieved this suppression and resulted in enhanced rubisco assembly. Implications with respect to dependence of chloroplast cpn6O function on cpnlO are discussed.
Chaperonins (cpn)' are a class of molecular chaperone found in prokaryotes and in a number of compartments in eukaryotic cells (3)(4)(5). Chaperonin function is required for the correct post-translational folding and assembly of some proteins (6,7). The activities of two chaperonin proteins, a homotetradecamer of 60-kDa polypeptides (groEL,,) and a homoheptamer of 10-kDa polypeptides (groES7), are essential in Escherichia coli (8). The structures of groEL,, and of its homolog in the higher plant chloroplast are conserved (6, 9, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.. 5242; Fax: 306-975-4839. $ T o whom correspondence should be addressed. Tel.: 306-975-The abbreviations used are: cpn, chaperonin(s); rubisco, ribulose-PB carboxylase. 10); however, plant chloroplast chaperonin-60 protein (cpn6014) is distinct in that it is comprised of two related polypeptides, p60' P""" and p60'p"-60p (11,12). The roles of these polypeptides and the precise subunit composition(s) of chloroplast cpn6OI4 protein remain unknown. The extent to which chaperonin function might be specialized has not been established. Complementation of an E. coli chaperonin function by expression of a closely related heterologous chaperonin gene has been reported recently (13).
Genes encoding the mature forms of Brassica napus p6WP"-@' " and p60'p"-@'p have been expressed separately and in combination in E. coli to produce three novel strains: a, 8, and cup (1). The plant polypeptides accumulate to similar high levels in each. In a cells, p6WP"-@"' accumulates but is either inefficiently assembled into cpn601, or forms a stable binary complex with grOEL14 (groEL14-p6WP""j0"). In (3 and a@ cells, the plant polypeptides assemble efficiently into cpn6O1, species, the majority of which are novel hybrid species composed of both bacterial and plant cpn60 polypeptides. The cpn601, species formed in these cells may or may not include the authentic plant and bacterial forms. As a result, B and ap cells contain (a) increased levels of total cpn6OI4, ( b ) decreased levels of authentic gr0EL14, and (c) decreased levels of groEL polypeptide present in cpn6014 species of any composition (1).
In the current study, we assessed whether the cpn6014 species containing plant polypeptides are functional as molecular chaperones in E. coli. Genetic experiments have demonstrated that the assembly of cyanobacterial rubisco (EC 4.1.1.39) in E. coli requires the functions of both gr0EL14 and groES7 and further that assembly is limited by normal levels of expression of the groE operon. Increased groE operon expression results in increased assembly and accumulation of cyanobacterial rubisco (2). We have tested the effect of expression of plant chaperonin genes on the capacity of E. coli cells to assemble cyanobacterial rubisco. Since the levels of authentic gr0EL14 are lower in cells in which efficient assembly of plant chaperonin-60 polypeptides occurs, any increase in the level of cyanobacterial rubisco assembly can be attributed specifically to activity contributed by the plant polypeptides.

Bacterial Growth Conditions-E. coli
DH5a cells (14) harboring the plasmids indicated were cultured as described (1). Induction of Ptrc and Plac was as described in the text.
Extraction of Proteins from E. coli--E. coli cells were recovered by centrifugation and resuspended in the buffer used for rubisco activity assays. Cell suspensions were passed through a French pressure cell at 6,000 p.8.i. Soluble protein extracts were defined as the supernatants recovered after centrifugation, 12,000 X g for 15 min, and desalting on Sephadex G-25. Total protein extracts were as previously described (1).
Construction of Brassica napus Chaperonin-60 Expression Vec-23333 tors"pOF39 (15) was cut with Ssp1 and ligated with a linker to incorporate NcoI and HindIII sites while maintaining the original N terminus of groES. It was cut with HindIII and BarnHI, and the fragment which carried the groES gene was ligated to pTZ19R (Pharmacia LKB Biotechnology Inc.) that had been cut with HindIII and BarnHI. A second NcoI site was introduced at the initiator Met codon of groEL by site-directed mutagenesis. The resulting plasmid was cut with NcoI, and the fragment that carried the groES gene was cloned into pKKn or pKKB (1) that had been cut with NcoI. The resulting plasmids were pKKESa and pKKESB. pKKB was cut with ScaI, EcoRI linkers were added and cut with EcoRI, and the fragment that carried the chaperonin gene was ligated to pKKESn that had been cut with EcoRI. The resulting plasmid was pKKESaB. In these constructs, the natural ribosome-binding site of groEL served as the ribosome-binding site for the plant cpn60 genes present in the synthetic groES-plant cpn60 operons. The 1acP gene, flanked by EcoRI (16), was adapted to HindIII and ligated to pACYC184 (17) that had been cut with HindIII. The resulting plasmid was pACYC. pKKn, pKK& pKKa(3, pKKES, pKKESa, pKKESB, and pKKESnB were cut with ScaI, and BarnHI linkers were added and cut with BarnHI. The fragments that carried the cpn60 genes were ligated with pACYC pACYCB, pACYCa0, pACYCES, pACYCESn, pACYCESB, and that had been cut with RarnHI. The resulting plasmids were pACYCa, pACYCESaB. pACYCESaB is illustrated (Fig. 1).

Effect of Expression of Plant Chaperonin Genes in E. coli on the Capacity to Chaperone the Assembly of Cyanobacterial
Rubisco-Plasmids designed to direct the synthesis of one or both of the mature B. napus cpn60 polypeptides (pACYCa, pACYC@, pACYCa@, and pACYC as control) were introduced into E. coli cells already carrying a plasmid-born cyanobacterial rubisco operon (pDB50 (19)). Transcription of the plasmid-encoded chaperonin and rubisco genes were inducible by isopropylthiogalactoside. The rates of synthesis of the rubisco large and small subunit polypeptides are expected to be identical in each strain. The accumulation of rubisco activity should therefore reflect the capacity of each strain to assemble Ptrc -h< . " rbs the enzyme from the available subunits.
In the absence of isopropylthiogalactoside, low levels of rubisco activity were detectable in all cells harboring pDB50. Upon addition of isopropylthiogalactoside, the rates of cell proliferation and the rubisco activity levels recovered varied depending on the plasmid present and on the stage of growth at which isopropylthiogalactoside was added. Therefore, in the following experiments, gene expression was induced when each culture reached an apparent optical density of 0.6-0.8. a, @, and a@ cells were analyzed 4 h after induction for rubisco activity (Fig. 2 A , lanes a, p, and ab), the presence of immunoreactive LxSx rubisco (Fig. ZB), and the accumulation of rubisco large subunit polypeptides (Fig. 2C). Co-expression of the plant cpn60 genes with the rubisco operon markedly affected the accumulation of active rubisco. Nondenaturing polyacrylamide gel electrophoresis analysis confirmed that the relative levels of L,Sx present were consistent with the rubisco activities measured. Thus, assessment of rubisco activity is a valid measure of the accumulation of assembled rubisco enzyme. The synthetic trp-lac promoter (Ptrc), lac repressor gene (lacI"), ribosome-binding sites (rbs), transcription terminators (term), low copy number origin of replication (p15A), chloramphenicol resistance gene (Crn"), and the orientations of the coding sequences for groES, the mature ~60'~"'""" (a), and p6OcP'"""" (B), are indicated. GroES and p60"'"~""" syntheses initiate from the lac2 rbs and p60p"~"D" initiates from the E. coli groEL rbs.
FIG. 2. Analysis of cyanobacterial rubisco gene expression in E. coli. The cyanobacterial rubisco operon and plasmid-encoded chaperonin genes were co-induced in late logarithmic phase. Four hours after induction, total and soluble protein extracts were made. A, rubisco activity expressed as nmol of C 0 2 fixed/min for a constant cell number (per A unit). R, soluble proteins resolved by nondenaturing-polyacrylamide gel electrophoresis, blotted to nitrocellulose, and probed with anti-rubisco. Positions of migration of LRSx rubisco holoenzyme and cpn60,, are indicated. C, polypeptides in total protein extracts resolved by SDS-polyacrylamide gel electrophoresis (15%), blotted to nitrocellulose, and probed with anti-rubisco. Position of migration of rubisco large subunit polypeptide (total L S U ) is indicated. Also indicated is an unidentified, immunoreactive band (band X ) as an indication of relative loadings in each lane. All strains except the negative control contained pDB50. The control strain also contained pACYC, and each of the other strains also contained the indicated chaperonin genes on pACYC-based vectors.
The levels of total large subunit polypeptide recovered showed a similar pattern of variation as did rubisco activity and immunoreactive L8Ss protein. Apparently, large subunit polypeptides that were not successfully assembled were subject to proteolysis. It has been reported that cells that assemble increased levels of cyanobacterial rubisco due to increased expression of the groE operon contain unchanged levels of rubisco large subunit polypeptides (2). From this, it was concluded that groE proteins do not influence the stability of large subunit polypeptides. Our results indicate that the plant chaperonins can influence polypeptide stability. Thus, the conditions of our experiments and those reported previously had different effects on the fate(s) of polypeptides that failed to assemble correctly.
We observed that rubisco accumulation after gene induction varied with time in a strain-dependent manner. Therefore, in an independent experiment, a, @, and a@ and control cells were harvested 2, 4, 6, and 24 h after induction, and rubisco activity assays of lysates were performed to determine the accumulation of assembled rubisco. Similar trends were apparent when rubisco activity was expressed on a per cell basis (Fig. 3 A ) or as specific or total activity (not shown).
Control cells accumulated rubisco continuously after induction (Fig. 3A). Rubisco accumulation in a cells was similar to that in control cells except after 24 h when higher levels were present in a cells. Increased rubisco accumulation after 24 h in a cells was, however, not consistently observed. Thus, p6OP""jo", when it was present alone, did not significantly increase the capacity of a cells to chaperone the assembly of rubisco. In marked contrast, rubisco accumulation in @ cells exceeded that in control cells at all times. The increase was

FIG. 3. Effect of co-expression of plant cpn60 genes in E. coli on cyanobacterial rubisco accumulation and cell growth.
The cyanobacterial rubisco operon and plasmid-encoded chaperonin genes were co-induced in late logarithmic phase cultures. The results of one representative experiment of five independent experiments are presented. A, protein extracts were made 2, 4, 6 and 24 h postinduction, and rubisco activity was assayed. Rubisco activity is expressed as nmol of CO, fixed/min for a constant cell number (per A unit). B, increase in cell numbers was monitored as apparent absorbance at 590 nm. Control cells harbored pACYC and pDB5O. All other strains contained pDB50 and the indicated chaperonin genes on pACYCbased vectors. %fold at 2 h, and 4-fold at 24 h after induction. Thus, when present alone, p60'p"-600 contributes significantly to the assembly of rubisco in E. coli. When both ~6 0 '~" " j~" and ~6 0 ' p " -~@ were present together, there was a dramatic decrease in rubisco assembly at all times. The accumulation of active cyanobacterial rubisco differed by 50-fold in p and a@ cells.

Effect of Expression of Plant Chuperonin Genes
on the Growth of E. coli Cells-It might be expected that effects on chaperonin function that resulted in decreased capacity to chaperone the assembly of cyanobacterial rubisco would be generally reflected in the overall chaperonin capacity of that cell. If so, a@ cells, which display a severe reduction in the capacity to assemble rubisco, might also be limited for chaperonin functions required for cell viability and growth. Induction of gene expression in a@ cells during late logarithmic phase growth did not result in a major change in their growth rate, although these cells did reach a slightly lower stationary phase cell density (Fig. 3B). Induction of gene expression at the time of cell inoculation, however, dramatically altered a@ cell growth (Fig. 4A). An extended lag phase was observed under these conditions. Compared with control cells, a@ cells required an additional 5 h to reach an apparent absorbance of 0.5. Thus, under these conditions of gene induction, in addition to being dysfunctional for rubisco assembly, a@ cells were deficient for some function(s) required for cell proliferation.
Since in uitro studies have implicated chaperonin function in the assembly of both chloramphenicol acetyltransferase and 0-lactamase (20,21), the effect of the presence or absence of chloramphenicol and ampicillin on the growth of a@ cells Assessment of Plant Chaperonin-60 Gene Function was tested. The extended lag phase of a@ cells when gene expression was induced at inoculation was independent of the presence of these antibiotics (not shown). The extended lag phase cannot, therefore, be attributed to the inability of these cells to chaperone the assembly or secretion of these enzymes that confer antibiotic resistance. Some unidentified activity required for cell proliferation is apparently affected by coexpression of cpn-60a and cpn-6OP under these conditions.

-
Effects of groES Gene Dose-It is anticipated that a cochaperonin (cpnlO) exists in the plastid, but the gene encoding it has not yet been identified (22). Thus, it is not possible to assess the effect of the presence of a plant cpnlO homolog on plant cpn6O function. I n vitro studies of chaperonin function suggest that the capacity to assemble rubisco might be affected by the relative levels of cpn6014 and cpn107 in the cell. Therefore, to assess the effect of different levels of cpn107 on the ability of plant cpn6O to chaperone the assembly of rubisco in E. coli, a second series of plasmids was designed to include a copy of the E. coli cpnlO gene, groES. Each of these plasmids (pACYCES, pACYCESa, pACYCESp, andpACYCESaP) was introduced into cells that contained pDB50. Expression of the plasmid copies of groES were inducible by isopropylthiogalactoside (not shown). The accumulation of rubisco in these cells is presented in Figs. 2 and 3.
Increased levels of groES in ES cells led to a decrease in their capacity to assemble rubisco, as compared with control cells. Increased levels of groES in ESP cells had little or no effect on rubisco assembly at early times as compared with P cells. At later times, however, ESP cells had accumulated much less rubisco than / 3 cells. Increased levels of groES in ESa cells resulted in a greatly decreased capacity to assemble rubisco at all times as compared with a cells. Remarkably, increased levels of groES in ESaP cells relieved the suppression of cyanobacterial rubisco assembly that had been observed in a0 cells. Furthermore, ESaP cells had a greater capacity to assemble rubisco at all times than control cells. Thus, p60PR"j0" and p60'p"-60" when present together can assemble into active cpn6OI4 species that chaperone the assembly of cyanobacterial rubisco, but a co-ordinate increase in the level of groES is required for this activity.
Although the suppression of rubisco assembly seen in a@ cells was relieved by elevated expression of groES as in ESaP cells (Fig. 3A), the extended lag phase observed when gene expression was co-induced at the time of inoculation (Fig. 4B) was not relieved. Thus, although this is a convenient assay for chaperonin function in vivo, the capacity to assemble and accumulate cyanobacterial rubisco does not appear to represent the full range of normal chaperonin functions in E. coli.
Implications-Assessment of the assembly of active cyanobacterial rubisco in E. coli has provided a sensitive in vivo assay for heterologous chaperonin function. Greatly enhanced chaperonin function attributable to the activity of assembled plant cpn6O polypeptide has been demonstrated in /3 cells. Since the majority of cpn6OI4 species in these cells contain both plant and bacterial gene products, it would seem that the enhanced activity is attributable to the hybrid species. The hybrid species also appear to be active with respect to normal groELI4 functions required for cell viability and growth. These two observations suggest that some functions of p60'p"-60u and groEL polypeptides may be interchangeable. The inability of p60'p""0" subunits to assemble in E. coli in the absence of p60'p"-60p, however, argues for at least some degree of functional specialization of p60Cp"~608.
There is evidence that cpn6OI4 species formed in P and a@ cells have different requirements for groES function. GroES function is limiting for cyanobacterial rubisco assembly in ab cells but not in ESaa cells. This suggests that the cpn6OI4 species that form in E. coli when cpn-60a and cpn-6OP are coexpressed require groES function for their activity. In contrast, the activity of the cpn6OI4 species that form in P cells is apparently not limited by the levels of groES polypeptide present. In fact, an increase in groES expression over that present in / 3 cells leads to decreased rubisco assembly, as in ESP cells. This suggests that ~60'~""jO@ function may not require groES function. It is tempting to suggest that multiple cpn6014 species may exist in the chloroplast and that the requirements for the function of chloroplast groES homologues may differ between them. Differential requirements for groES function have been described for the folding of different polypeptides both in vivo and in vitro (20,(22)(23)(24)(25)(26).