Manipulation of Synechocystis sp. PCC 6803 as a platform for functional identification of genes involved in carotenoid metabolism

Carotenoids are not only essential light-harvesting/photoprotective pigments for plants, but also essential phytonutrients for human health. Although the carotenoid metabolic pathway is generally shared by photosynthetic organisms, some unique carotenoids are only synthesized in particular species. Eludication of the enzymes that catalyze their biosynthesis is a prerequisite for manipulating their production by transgenic and synthetic biological approaches. Both pigment complementation in Escherichia coli and gene overexpression in the wild-type and corresponding silencing mutants of Arabidopsis thaliana have been used for functional characterization of putative enzymes but with some limits. In this study, we engineered the carotenoid metabolic pathway of the cyanobacterium Synechocystis sp. PCC6803 by silencing its endogenous carotene hydroxylase gene (CrtO) and introducing different exogenous genes. With the zeaxanthin epoxidase gene from A. thaliana, we extended the metabolic pathway from zeaxanthin to violaxanthin in Synechocystis. We also established in Synechocystis the metabolic branch from lycopene to lutein, which does not naturally exist in cyanobacteria and some algal phyla, by transforming with the genes for lycopene ε-cyclase (Lut2) and carotene ε-hydroxylase (Lut1) from A. thaliana. Our results also confirmed that the endogenous carotene β-hydroxylase CrtR could only hydroxylate the unsubstituted β-ring. Taken together, we demonstrated that Synechocystis can be manipulated as a new platform for the functional characterization of unknown proteins that have putative functions on carotenoid metabolism. This article is protected by copyright. All rights reserved.

Carotenoids are widely distributed in nature. They function as light-harvesting/photoprotective pigments, furnish flowers and fruits with distinct colours, and benefit human health as essential phytonutrients (Rodriguez-Concepcion et al., 2018). All photosynthetic organisms synthesize lycopene and b-carotene, and higher plants share a more complicated set of carotenoids, including carotenes and their oxygenated derivatives. However, some carotenoids only exist in particular species, such as astaxanthin in the green alga Haematococcus pluvialis and lactucaxanthin in lettuce (Lactuca sativa) (Rodriguez-Concepcion et al., 2018). The elucidation of enzymes that catalyse their biosynthesis would enable the manipulation of their production by genetic engineering and synthetic biology strategies. It is usually challenging to determine the catalytic activities of these enzymes by in vitro assays using recombinant proteins in an aqueous system, because most of their substrates and products are lipid-soluble. Therefore, in vivo assays are also used for functional characterization. The pigment complementation system in Escherichia coli is able to synthesize a wide range of carotenoid substrates and has enabled the discovery of a large number of enzymes in this pathway (Cunningham and Gantt, 2007). However, the activities of some enzymes rely on additional cofactors and/or membrane structures. For example, the lycopene b-cyclase (LCYB) CruA requires a bound chlorophyll a molecule, which is absent in E. coli, for its activity (Xiong et al., 2017). Although functional complementation and/or overexpression in Arabidopsis are also widely used for characterizing new enzymes, the complicated and dynamic repertoire of endogenous carotenoids, together with the possible redundancy of enzymes, might mask the functions of the transgenes (Quinlan et al., 2007).
Considering that plants exclusively synthesize carotenoids in plastids, which have a cyanobacterial origin, we tested whether the cyanobacterium Synechocystis could be engineered as a new platform for functionally characterizing carotenoid metabolic enzymes. Compared with E. coli, Synechocystis might have an internal environment, such as the availability of cofactors, binding proteins and membrane structures, more similar to that of plastids. On the other hand, Synechocystis accumulates only four carotenoids (myxoxanthophyll, zeaxanthin, echinenone and bcarotene) that all have only b-ring in their molecules, without any carotenoids that are further modified beyond zeaxanthin or have e-rings in their structures. Therefore, Synechocystis has a much simpler carotenoid pool than Arabidopsis for identifying the functions of exogenous enzymes (Zhang et al., 2015). Moreover, the availability of the full genome sequence of Synechocystis also provides convenience in genetic manipulation.
We first cloned the kanamycin resistance gene (Kan r ) driven by the Amp r promoter (Amp r ::Kan r ) from pGBK-T7 as a selection marker. Carotenoid metabolic enzyme genes driven by the promoter of RbcL from Synechocystis to ensure a high-level expression were cloned downstream of Amp r ::Kan r . When two enzyme genes were to be expressed together, a ribosomebinding sequence (RBS, CTTTAAGAAGGAGATATACC) was used to separate their open reading frames, so that the RbcL promoter could drive their expression simultaneously (Thiel et al., 2018). Each cassette was further amplified from two ends to incorporate adaptors (GGGGAAAAATCCTCA and GTTTTCCACCAATAC) for facilitating the transformation of Synechocystis sp. PCC 6803 by homozygous recombination that replaced the endogenous CrtO gene for a b-carotene ketolase (Xiong et al., 2017). Figure 1b shows our designs. Synechocystis cells were grown on BG11 plates or in liquid medium at 28°C under a light intensity of 30 lmol photons/m 2 /s. For selecting transgenic strains, kanamycin at 50 lg/mL was used.
Synechocystis cultures with different transgenes showed varied colours ( Figure 1c). We further analysed their pigments by HPLC (Xiong et al., 2017). As expected, when CrtO was interrupted by Amp r ::Kan r , the transgenic strain was unable to synthesize echinenone ( Figure 1d). This further simplified the carotenoid metabolic pathway. Cyanobacteria do not naturally synthesize violaxanthin. However, when the zeaxanthin epoxidase gene from Arabidopsis (AtZEP) was introduced, the transgenic ZEP+ strain accumulated violaxanthin instead of zeaxanthin. This demonstrated that Synechocystis is suitable to support the function of ZEP.
Because violaxanthin is a substrate for neoxanthin biosynthesis in higher plants (Rodriguez-Concepcion et al., 2018), we further expressed the neoxanthin synthase from Solanum lycopersicum (SlNSY) together with AtZEP. No neoxanthin was found in transgenic NSY+ cells. This is most probably because the function of NSY in higher plants needs two additional proteins, ABA4 and NXD1, which we could not find homologs from the Synechocystis genome. However, compared with ZEP+, the NSY+ cells had a (a)  lower level of violaxanthin and a higher level of b-carotene, supporting a primitive function of NSY in catalysing the bcyclization of lycopene, as previously reported (Bouvier et al., 2000).
Carotenoids with e-rings are rarely synthesized in cyanobacteria. To test whether Synechocystis can be engineered to produce this group of carotenoids, we introduced the lycopene e-cyclase (LCYE) gene from Arabidopsis (AtLut2) and observed the production of a-carotene in the corresponding Lut2+ strain. This confirms that AtLUT2 collaborates with endogenous CruA to catalyse the eand b-cyclizations, respectively, of the two open ends of lycopene. We further incorporated the P450-type carotene e-hydroxylase gene from Arabidopsis (AtLut1), together with AtLut2, into the cassette. The transgenic LUT1/2+ cells accumulated a significant level of lutein, indicating a collaboration of endogenous carotene b-hydroxylase (CrtR) and exogenous AtLUT1 to hydroxylate both ends of a-carotene. This indicated a successful construction of the b,ebranch of carotenoid metabolism in Synechocystis and demonstrated the feasibility of utilizing this cyanobacterial platform to identify enzymes that catalyse the metabolism of lutein. It is interesting that, compared with the LUT2 strain, the LUT1/2+ cells accumulated much lower amounts of b-carotene and zeaxanthin. This revealed that the expression of additional carotene e-hydroxylase resulted in a decreased metabolic flux towards the biosynthesis of carotenoids with two b-rings in their structures.
We further cloned the b-carotene ketolase genes from Haematococcus pluvialis (HpBKT) and the bacterium Paracoccus haeundaensis (PhCrtW), both of which produce astaxanthin, and expressed each of these genes in Synechocystis. Instead of astaxanthin, canthaxanthin and its isomers were accumulated in transgenic cells of both strains. Echinenone was also found in PhCrtW-transformed cell. This showed that both HpBKT and PhCrtW preferentially use b-carotene as their substrate, and the endogenous CrtR is unable to hydroxylate canthaxanthin to produce astaxanthin, as it was recently demonstrated in rice endosperm (Figure 1a; Rodriguez-Concepcion et al., 2018;Zhu et al., 2018). Therefore, the successful production of astaxanthin in Synechocystis might need a co-expression of HpBKT (or PhCrtW) with an additional CrtZ-type carotene 3,3 0 -hydroxylase that uses canthaxanthin as a substrate (Rodriguez-Concepcion et al., 2018).
Carotenoids with two e-rings are only found in a few plants, such as lettuce that produces lactucaxanthin. Different from AtLUT2 that catalyses the e-cyclization on only one end of lycopene, the LCYE from lettuce (LsLCYE) cyclizes both ends (Cunningham and Gantt, 2007). To test its function in the cyanobacterial system, we co-expressed LsLCYE and AtLut1 together in our Lac+ strain. Similar to the LUT1/2+ strain that expresses AtLut2 and AtLut1, the Lac+ cells also accumulated only lutein. No carotenoid compounds with two e-rings were identified from the products of the Lac+ strain. This suggests that additional cofactors might be needed for the dual e-cyclization function of LsLCYE in cyanobacteria (Cunningham and Gantt, 2001).
Taken together, we demonstrated the feasibility of engineering Synechocystis as a new platform for the functional characterization of enzymes involved in carotenoid metabolism and for the screening of unknown factors that interfere with the activities of enzymes such as SlNSY and LsLCYE. Although it is difficult to compare the carotenoid production in Synechocystis with that in either E. coli or Arabidopsis, Synechocystis has the advantage of being able to grow both autotrophically and heterotrophically. This means that silencing of its genes for carotenoid biosynthesis is not as crucial as in higher plants for the growth, acclimation, and propagation, and its carotenoid productivity is also released from the restriction of the photosynthetic capability (Fiore et al., 2012). Therefore, this Synechocystis platform can also be improved as a candidate chassis for the biosynthesis of different carotenoid compounds.