Pinpointing Synechococcus Rubisco Large Subunit Sections Involved in Heterologous Holoenzyme Formation in Escherichia Coli

Background: Heterologous holoenzyme formation of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) have been a challenge due to limited understanding of its biogenesis. Unlike bacterial Rubiscos, eukaryotic Rubiscos are incompatible with the Escherichia coli chaperone system to fold and assemble into the functional hexadecameric conformation (L 8 S 8 ), which comprise eight large subunits (RbcL) and eight small subunits (RbcS). Our previous study reported three sections (residues 248-297, 348-397 and 398-447) within the RbcL of Synechococcus elongatus PCC6301 may be important for formation of L 8 S 8 in E. coli. Present study further examined these three sections separately, by dividing them into six sections of 25 residues (i.e. residues 248-272, 273-297, 348-372, 373-397, 398-422 and 423-447). Methods and Results: Six chimeric Rubiscos with each section within the RbcL from Synechococcus replaced by their respective counterpart sequence from Chlamydomonas reinhardtii were constructed and checked for their effect on holoenzyme formation in E. coli. Present study shows that Section 1 (residues 248-272; section of Synechococcus RbcL replaced by corresponding Chlamydomonas sequence), Section 2 (residues 273-297), Section 3 (residues 348-372) and Section 6 (residues 423-447) chimeras failed to fold and/or assemble despite successful expression of both RbcL and RbcS. Only Section 4 (residues 373-397) and 5 (residues 398-422) chimeras could form L 8 S 8 in E. coli. Conclusion: As GroEL chaperonin mediates folding of bacterial RbcL in E. coli, residues 248-297, 348-372 and 423-447 of Synechococcus RbcL may be important for interacting with the GroEL chaperonin for successful holoenzyme formation in E. coli.


Introduction
Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco, E.C 4.1.1.39) is the CO 2 xing enzyme that catalyzes incorporation of atmospheric CO 2 onto organic carbon ribulose-1,5-bisphosphate (RuBP) during the Calvin-Benson-Bassham cycle of photosynthesis to produce two molecules of 3-phosphoglycerate (3PGA) for biomass accumulation in photoautotrophs [1]. Given its pivotal role, Rubisco is accounted as the gateway for inorganic CO 2 into the biosphere and its capability to sequester CO 2 dictates the e ciency of photosynthetic CO 2 assimilation [2]. Nevertheless, Rubiscos are slow enzymes with carboxylation turnover rates (k c cat ) in a range of 1 s -1 to 13 s -1 [3]. Moreover, its inability to distinguish between CO 2 and O 2 for xation onto RuBP greatly decreases the photosynthetic e ciency as oxygenation of RuBP lead to formation of one 3PGA and one 2-phosphoglycolate (2PG), which must be recycled into 3PGA through photorespiration at the expense of high cellular energy and loss of xed carbon as CO 2 [4]. Slow catalytic rate and oxygenase activity of Rubiscos render CO 2 assimilation as one of the rate-limiting factors of photosynthesis [5]. As a result, Rubisco has been a prime target for genetic engineering to improve its catalytic performance in terms of CO 2 /O 2 speci city factor (Ω) and carboxylation rate as a means to increase photosynthetic e ciency to raise crop productivity [6,7].
Incapability of functional expression in heterologous host mainly stems from chaperone incompatibility and requirement of additional auxiliary factors for their complex biogenesis, which is a multi-step process requiring different kinds of chaperones for de novo folding and assembly [27][28][29][30]. In E. coli, folding of RbcL monomers are mediated by the endogenous GroEL-GroES chaperonin system prior to their assembly into oligomers [31,32]. The importance of the GroEL-GroES chaperonin for Rubisco holoenzyme formation in E. coli were demonstrated by increased yield of soluble Rubiscos upon overexpression of GroEL, while mutations of GroEL or GroES abolished holoenzyme formation [33,34]. In addition, studies have shown that Rubiscos are prone to aggregation and they are recognized as stringent substrates by the GroEL chaperonin, whereby in vitro reconstitution of dimeric Form II Rhodospirillum rubrum Rubisco and hexadecameric Form I cyanobacterial Rubisco require all GroEL, GroES, ATP and Mg 2+ [31,35,36]. Therefore, unsuccessful formation of eukaryotic Rubiscos in E. coli suggests that their RbcLs (and more speci cally, the amino acid sequences) are incompatible with bacterial GroEL/GroES chaperonin. Indeed, functional expression of A. thaliana Rubisco in E. coli require the co-expression of its chloroplast chaperonin Cpn60αβ/Cpn20, which cannot be replaced by the bacterial GroEL/GroES homolog [37].
Our previous study replaced sections of cyanobacterial RbcL from Synechococcus elongatus PCC6301 with their eukaryotic counterpart sequences from the green alga Chlamydomonas reinhardtii sequentially, and a few sections of Synechococcus RbcL that might be essential for successful holoenzyme formation in E. coli were reported [20]. Present study aimed to narrow down the range by further examining each half (25 residues) of three sections that were pinpointed (i.e. amino acids 248-297, 348-397 and 398-447). Therefore, six chimeric Rubiscos with 25-amino acid sections of Synechococcus RbcL (residues 248-272, 273-297, 348-372, 373-397, 398-422, and 423-447) substituted with the corresponding residues in Chlamydomonas RbcL, respectively, were constructed to examine their importance for holoenzyme formation in E. coli.

Construction of chimeric Rubiscos
Gene fragments required for constructing chimeric rbcL-rbcS operons were ampli ed from pTrcSynLS harbouring the wild-type Synechococcus PCC6301 rbcL-rbcS operon [22] and plasmids carrying different recombinant rbcL-rbcS [20] using Pfu DNA Polymerase (Vivantis). Primers used for PCR ampli cation were designed with different restriction sites (PstI, NcoI and BsmBI) to facilitate directional cloning (Table  1; Supplementary information). Ampli ed products were digested with restriction enzymes and ligated into the vector backbone (pTrcHisB) of pTrcSynLS using T4 DNA ligase (Invitrogen) ( Table 1; Supplementary information). XL-1 Blue E. coli cells were transformed with the ligation mixtures and selected on LB plates containing 100 µg ml -1 ampicillin. Positive transformants were identi ed by colony PCR and plasmids were extracted using commercial plasmid miniprep kit for DNA sequencing.

Results
In order to examine sections of Synechococcus RbcL essential for holoenzyme formation in E. coli, six chimeric Rubiscos with sections of 25 residues (i.e. residues 248-272, 273-297, 348-372, 373-397, 398-422 and 423-447) separately changed to corresponding residues in Chlamydomonas RbcL were constructed (Fig. 1). The number of amino acid changes in the sections ranges from one to eight amino acids (Fig. 2). SDS-PAGE and Western blot analysis of denatured total cellular proteins from E. coli transformed with chimeras respectively showed that both large and small subunits of all the chimeric Rubiscos were expressed in E. coli (Fig. 3). Therefore, any undetectable assembly of L 8 S 8 was not due to non-expression. Native-PAGE and Western blot analysis showed sections 4 and 5 chimeric Rubiscos assembled into L 8 S 8 whereas sections 1, 2, 3 and 6 chimeras had no detectable complexes in E. coli (Fig.   4). It is noteworthy that the section 4 chimera showed a reduced amount of L 8 S 8 as compared to wildtype Rubisco and the section 5 chimera (Fig. 4).

Discussion
Unlike its eukaryotic isoform, RbcL from Synechococcus sp. PCC6301 are compatible with E. coli GroEL chaperonins for proper folding, followed by assembly with RbcS into holoenzyme without additional chaperones. Here we show that sections 4 and 5 are not essential for GroEL interaction as substitution with corresponding Chlamydomonas sequences still allowed folding and assembly into their nal hexadecameric complexes (L 8 S 8 ) in E. coli (Fig. 4), although it is worth mentioning that Chlamydomonas sequences introduced into sections 4 and 5 chimeras might have changed their kinetic properties. Notably, section 4 chimera showed reduced amount of L 8 S 8 than wild-type Rubisco and section 5 chimera. Section 4 chimera has two amino acid substitutions, S395A and V396C. One possibility is that these substitutions affect the expression level of RbcL and RbcS. But looking at the SDS-PAGE and Western blot analysis of total cellular extracts, only slight differences in the expressed level of RbcL and RbcS were observed among wild-type and chimeric Rubiscos (Fig. 3). This is consistent with previous ndings whereby most mutations in RbcL generally do not affect the steady-state mRNA level [22,34]. Another possibility is that S395A and/or V396C reduced the thermal stability of the holoenzyme, as have been reported before for other substitutions in RbcL [10,39,40].
As sections 1, 2, 3 and 6 chimeras showed no detectable L 8 S 8 (Fig. 4), sections of Synechococcus RbcL that might be important for interacting with GroEL can be narrowed down to these sections, which comprise residues 248-272 (Section 1), 273-297 (Section 2), 348-372 (Section 3) and 423-447 (Section 6). Considering that GroEL chaperonin form multivalent interaction with its substrate polypeptide, whereby substrate polypeptides bind to multiple subunits of heptameric GroEL ring for e cient binding [41][42][43], it is possible that more than one of these sections interact with GroEL. Indeed, it has been reported that Rubiscos bind to a minimum of three consecutive subunits of GroEL ring for e cient binding [41,43]. Binding of Rubisco to GroEL mutant with less than three consecutive binding-competent apical domain resulted in signi cantly reduced amounts of binary complexes of Rubisco and GroEL mutant, as compared to the wild-type GroEL [43]. In accordance with this nding, cryo-electron microscopy (cryo-EM) study of GroEL-RbcL binary complex showed the C-terminal domain of RbcL folding intermediate in contact with three consecutive apical domains while the N-terminal domain in contact with one apical domain [44]. Therefore, it may be possible that the three separate RbcL C-terminal regions (residues 248-297, 348-372 and 423-447) examined herein are involved in GroEL interaction and that loss of any favourable interactions with these three regions disrupted consecutive binding of RbcL to GroEL subunits.
Consequently, binding between GroEL and RbcL was abolished or substantially reduced to an extent that only a small number of RbcLs were captured by GroEL, properly folded and assembled, resulting in the amount of L 8 S 8 was undetectable.
Other than loss of GroEL interaction, non-assembly could be due to structural destabilizing effect of substitutions when changed to the Chlamydomonas counterpart. Therefore, despite of being captured by the GroEL chaperonin, the chimeric RbcL monomer might not fold into its native structure stably, and thus was unable to assemble into the nal hexadecamer. Sections 1, 2, 3 and 6 chimeras have four to eight substitutions (Fig. 2). In general, amino acid substitutions could contribute to bene cial, neutral or deleterious effects [45]. In the case of our chimeras, some of these substitutions might have disrupted intra-subunit interactions or even exerted disturbances to the tertiary structure. For section 1 chimera, which has ve substitutions, A269G has been reported to improve the tness of Rubisco-dependent E. coli [22]. Therefore, non-assembly is more likely caused by the remaining four substitutions. As for section 6 chimera, one of the six mutations, E444S, could cause loss of intra-subunit salt-bridge with R432 in the RbcL monomer (Fig. 5a).
In addition, as there are three states of assembly: dimeric L 2 pair, (L 2 ) 4 core complex and nal hexadecameric L 8 S 8 , mutations that affect the inter-subunit interactions of any of these assembled complexes could eventually lead to non-assembly. Of all mutations involved in non-assembly, only E348D in section 3 chimera is located at the interface between RbcL and RbcS. By comparing the Rubisco crystal structures of Synechococcus (PDB ID 1RBL) and Chlamydomonas (PDB ID 1GK8), this mutation could result in replacement of the carboxyethyl side-chain of Glu-348 by a shorter carboxymethyl sidechain of Asp, which could potentially disrupt the salt-bridge with Lys-96 in the small subunit [20] (Fig. 5b).

Conclusions
When six sections of Synechococcus RbcL are substituted by counterpart sequence from Chlamydomonas RbcL, respectively, only sections 4 and 5 chimeric Rubiscos form L 8 S 8 in E. coli. Nonassembly of sections 1, 2, 3 and 6 chimeras suggest that the substituted sections of Synechococcus RbcL might be important for interacting with GroEL chaperonin to be properly folded. Loss of these sections might have disrupted interaction with GroEL and led to misfolding or aggregation of chimeric RbcL, which precluded subsequent assembly. Yet, non-assembly could also be due to structural destabilizing effect of mutations introduced from Chlamydomonas counterpart. Although the true cause of non-assembly remains to be addressed, sections of Synechococcus RbcL that might be essential for successful holoenzyme formation in E. coli are narrowed down to residues 248-297, 348-372 and 423-447 for future dissection.