Team:Amsterdam/Export

Our newly engineered cyanobacterial factories are capable of producing fumarate directly from CO ₂ using the energy of (sun)light. However, experts in the fields pointed out to us that it is only the fumarate that is excreted from the cells that is readily available in a downstream process of a “real-world” scenario. Furthermore, if fumarate is accumulated intracellularly, it may prove to inhibit its own production, and potentially even growth. We therefore anticipate that ensuring an effective fumarate transport system is an important engineering target for our cell factories. In this module of the project the goal is to: (i) study the native fumarate transport system of Synechocystis ; and (ii) overexpress heterologous transport systems that have been shown to improve fumarate extracellular production in chemoheterotrophic organisms.

Export

Overview

A native fumarate transport system in Synechocystis is yet to be identified. However, we have successfully measured extracellular fumarate production in our engineered strains, suggesting a transport mechanism of some kind must to be present. But which?

We have carried a thorough bioinformatics homology search using sequences from known fumarate transporters from other organisms. Amongst the roughly 900 candidate genes of Synechocystis predicted to encode putative membrane proteins, we found two suitable targets that may encode components of a dicarboxylic acid transporter (such as fumarate). To test this in silico prediction we constructed single knock-out mutants of these putative genes of unknown function. Furthermore, we have also exploited the possibility of increasing fumarate production in our engineered strains through the expression of known heterologous fumarate transporters. This strategy has led to an increase in fumarate yields in E. Coli of 53%1 [1]. Two versatile BioBricks ( Bba_K2385001 and Bba_K2385000 ) have been constructed that encode fumarate transporters from Escherichia coli and Nodularia Spumigena . Finally, we performed extensive phenotyping of all the strains constructed, with particular emphasis on the impact of the genetic interventions on fumarate import and export.

Highlights

• Identification of two genes in Synechocystis , sll1103 and sll1314 , encoding putative dicarboxylate transporters
• Experimental validation that sll1314 encodes (part of) a transporter system involved in fumarate export
• Experimental validation that Synechocystis does not uptake fumarate from the environment
• Expression of neither a known fumarate transporter from E. Coli , nor one from N. Spumigena , leads to fumarate excretion in wild type Synechocystis

▶  Our quest to identify the elusive, native fumarate transport system in Synechocystis

Introduction

Our cyanobacterial cell factories produce fumarate intracellularly, but the fumarate that we measure (and could easily harvest in a future downstream process is extracellular. The efflux of fumarate that is taking place can be either the result of passive diffusion, or facilitated by a transport system. Fumarate is a dicarboxylic acid and has dissociation constants far below the intracellular pH of Synechocystis . This causes fumarate to be almost completely in its ionized form, which makes it much harder to diffuse across a hydrophobic membrane. A native fumarate transport system in Synechocystis is yet to be identified, but this strongly suggests a transport mechanism of some kind must to be present.

A transport mechanism in the form of a protein can be studied by knocking out the encoding gene and study the resulting phenotype. To this end, we are forced to first identify candidate genes that may putatively encode fumarate transport proteins within the genome of Synechocystis . Membrane transporters are notorious for being promiscuous, often displaying affinity for multiple substrates [1]. However, their specificity for certain chemical properties (e.g. presence of two carboxylic acid groups) is often sufficiently conserved across phylogenetic clades [2], making it possible to attempt the identification of genes by sequence comparisons (i.e. by homology searching). Sequence alignments unambiguously distinguish between protein pairs of similar and non-similar structure when the pairwise sequence identity is high (>40% for long alignments) [3]. Below 35%, the number of false positives increases tremendously and one can no longer infer a homologous relationship. Up to now, no C ₄ -dicarboxylate transporter has been identified in Synechocystis , which already indicates that the search will not yield an unambiguous result. We went to see an expert in the field: Anton Feenstra , an Assistant Professor Bioinformatics at Vrije Universiteit Amsterdam.

Together we designed a pipeline to tackle the problem from every possible angle, aiming to provide the support required for an educated selection. We made use of targeted searches, where we used sequence information from known, phylogenetically closely related transporters, but also protein motifs and conserved domains as a template for sequence alignment. Complementary, untargeted searches were used to profile large groups of sequences (including putative identifications) and find distant homologous relationships between proteins.

Figure 3.1 Sequence homology Pipeline using untargeted and targeted searches. Four complementary tracks to identify putative genes encoding a dicarboxylate transporter in Synechocystis.

The pipeline entails four complementary tracks (fig 3.1.). All sequence alignment tools are used in default settings unless stated otherwise. The e-value cut-off was set to 1.0 (as to include remote homologs and manually curate the results). To prevent profile wandering with iterative tools, the inclusion threshold (e-value) for PSI-BLAST was set to 0.05.

(1) Several proteins(families) are involved in C ₄ -dicarboxylate transport (e.g. Dcu-, Dct- and TRAP-transporters). For each of these target groups, sets of 200 transporter(domain) sequences were profiled using PRALINE ^™ , an alignment method that integrates membrane-specific scoring matrices to improve alignment of the transmembrane regions [4]. It creates a Multiple Sequence Alignment file (MSA), that is taken as a template for HHSearch, present-day, the most sensitive protein sequence alignment method [5]. HHSearch computes a Hidden Markov Model (HMM), similar to simple sequence profiles, but in addition to the amino acid frequencies in the columns of a MSA, they contain the position-specific probabilities for inserts and deletions along the alignment. The HMM is then aligned against the sequence database of Synechocystis (tax id: 1148, which is a profile-profile alignment. By aligning profile HMMs instead of simple sequence profiles, both sensitivity and alignment quality improve significantly. This allows for the identification of more distant homologous relationships.

(2) Synechococcus elongatus ( Synechococcus elongatus hereafter) and Nodularia Spumigena ( N. Spumigena hereafter) are phylogenetically closely related to Synechocystis (also cyanobacteria) and widely studied. In Synechococcus elongatus , two C ₄ -dicarboxylate transporter(domain)s have been identified: DctM and DctA. In N. Spumigena , a DcuC transporter is putatively identified based on sequence homology studies. In separate searches, each of the gene sequences was used as a template to create a Position Specific Scoring Matrix (PSSM), using PSI-BLAST on the NCBI NR db [6]. This was done in order to "enrich" the sequence with information from related sequences.. ClustalX is used to create a MSA out of the PSSM, as this is the most sensitive method that is able to take a PSSM as template [7]. The MSA is subsequently taken as a template for HHSearch.

(3) NCBI"s Conserved Domain Database is a collection of representative sequences and PSSMs, that represent evolutionarily conserved sequence fragments (or domains) [8]. It provides functional characterization based on the presence of signature sequence patterns and may serve as a starting point for functional annotation and classification. The database has annotated conserved domains for several C ₄ -dicarboxylate transport(domain)s: DctP, DctM, Dcu and DcuC. For each of the profiles, Clustal and subsequently HHSearch was used to identify remote homologs in Synechocystis .

(4) The PROSITE database consists of a large collection of biologically meaningful signatures that are described as patterns or profiles [9]. Each signature is linked to a documentation that provides useful biological information on the protein family, domain or functional site identified by the signature. The database has two family signatures (in Clustal output format) for sodium:dicarboxylate symporters. For each of the profiles, HHSearch was used to identify remote homologs in Synechocystis .

All sequence alignment tools are used in default settings unless stated otherwise. The p-value cut-off was set to 1.0 (as to include remote homologs and manually curate the results), except for iterative tools (such as PSI-BLAST) where the cut-off was set to 0.05 to prevent profile wandering.

Results & discussion

As stated in the introduction, sequence identities below 35% are not sufficient to imply a homologous relationship. Therefore, we will only highlight the genes that pop up as a hit in multiple tracks, this to ensure that a random alignment (false positive) will not be taken as putative target.

The first gene that appears 2 times in tracks 2 and once in track 3 is sll1103 . It is mapped to the conserved domain of DctM (pfam06808) with an E-value of 2.5 E ^-44 , shares 31% sequence identity to the DctM subunit from Synechococcus (EHA61938.1) and 20.7% to DcuC from N. Spumigena (AHJ29173.1) (fig. 3.2). Additionally, sll1104, located in the same transcriptional unit, aligns with an E-value of 1.1 E ^-19 in track 1 when using DcuB as a query.

The second gene that appears both in track 2 and 3 is sll1314 . This gene is mapped to the conserved domain of DctP (TIGR00787) with an E-value of 3.3E ^-106 and shares 25% sequence identity to the DctP subunit from Synechococcus (WP_015168385).

Figure 3.2 Amino acid sequence alignment of the putative C ₄ -dicarboxylate transporter in N. Spumigena (DcuC) and Synechocystis ( sll1103 )

▶  Cultivation experiments to investigate the role of sll1103 and sll1314 in fumarate transport

Introduction

Now that we have made an educated guess on the proteins involved in fumarate transport, we need to experimentally validate their function. In order to find the role that these genes may play in fumarate transport, sll1103 and sll1314 were deleted in Synechocystis wild type and the derivative fumarate producing strain, ΔfumC . Genes were deleted using a markerless genetic modification method. This way, our newly designed strains will never be at risk of passing on to natural environments resistance cassettes. These strains were then characterized phenotypically in a series of different experiments in order to address our different research questions:

• We study the effect of the deletions on fumarate efflux by cultivating the double knock-out strains (ΔfumCΔsll1103 and ΔfumCΔsll1314) with ΔfumC as a control.
• The effect of the deletions on fumarate influx was studied by cultivating the single knock-out strains (Δsll1103 and Δsll1314) with the wild type as a control in the presence of exogenously supplemented fumarate.
• The postulated promiscuity of the tentatively annotated dicarboxylate transporters, sll1103 and sll1314, was tested for multiple substrates by performing high-throughput phenotyping array screens using BIOLOG plates.

Results & discussion

The targets sll1103 and sll1314 (genes encoding putative fumarate transporters) were deleted on wild type and ΔfumC background (fig. 3.3).

Figure 3.3 PCR confirmation of gene deletions on wild type and ΔfumC background of Δsll1103 (A) and Δsll1314 (B). A clean knockout gives a single DNA band at 2.023 bp, while for wild type (WT) this is 3.287 bp.

First, we studied the effect of the deletions on fumarate efflux by cultivating the double knock-out strains ( ΔfumCΔsll1103 and ΔfumCΔsll1314 ) with ΔfumC as a control for 150 hours, while monitoring cell growth and fumarate production. We hypothesized that if one of the genes encodes a protein involved in fumarate export, we should measure less extracellular fumarate compared to the control strain.

Neither of the two gene deletions led to an inhibition of cell growth (fig. 3.4A). We are mostly interested in the log phase, as this is where the fumarate is produced (remember that ΔfumC is a Synechocystis derivative producing fumarate as the cell grows). Zooming in on T = 49 (log phase) we can see that deletion of sll1103 causes a diminishable and insignificant decrease in fumarate efflux (fig. 3.4B). Deleting Δsll1314 , on the contrary, causes a significant decrease in fumarate efflux of 31 % (p-value < 0.01, ANOVA). The fact that we still see extracellular fumarate suggests that sll1314 is not the only transport system for fumarate, but this result validates that Δsll1314 is involved in fumarate transport.

Figure 3.4 Cell growth (measured at OD ₇₃₀ ) in different Synechocystis strains under constant light illumination for over 150 hours (A). Fumarate concentration in mM per OD for T = 49. Error bars indicate the 95% confidence interval. (n="3" for double knock-outs, n="2" for ΔfumC ).

Secondly, the effect of the deletions on fumarate influx was studied by cultivating the single knock-out strains ( Δsll1103 and Δsll1314 ) with the wild type as a control in the presence of exogenously supplemented fumarate. We hypothesize that if these genes are involved in fumarate import, than we should see a difference in fumarate uptake kinetics under conditions in which cells are forced to uptake fumarate in order to grow, i.e. in the presence of the photosynthesis inhibitor DCMU.

Figure 3.5 Fumarate (mM) over time for different Synechocystis strains cultivated for 6 days.

We can see that both wild type and Δsll1103 have restricted/no capability to assimilate fumarate, under the conditions tested (fig. 3.5). Our results thus strongly hint the absence/deficiency of the fumarate import transporter in Synechocystis .

Lastly, the postulated promiscuity of the tentatively annotated dicarboxylate transporters, sll1103 and sll1314 ,was tested for multiple substrates by performing high-throughput phenotyping array screens using BIOLOG plates. We hypothesize that if the target genes are involved in the uptake of other compounds, than we should see a difference in this screen when comparing the single knock-out strains ( Δsll1103 and Δsll1314 ) with the wild type as a control. Comparisons can be performed in cells taken from cultures in the light and in the dark, since they display different expressions patterns [13].

We did not observe any differences in BIOLOG screen between wild type and the derivative knock out strains Δsll1103 and Δsll1314 . This suggests that these two genes do not encode proteins involved in the transport of the substrates tested (fig. 3.6) This however does not mean that they may not be involved in the transport of other substrates, or that their expression pattern in wild type under different conditions could not have lead to a different conclusion.

Figure 3.6 Carbon sources on the BIOLOG Ecoplate.

Conclusion

• We have experimentally validated that sll1314 encodes a transporter protein(domain) involved in fumarate transport as predicted by our bioinformatics analysis.
• We have experimentally validated that Synechocystis wild type and single gene knock-out mutants do not re-assimilate fumarate (even under carbon-limited conditions)
• We have some indication via the "metabolic fingerprinting" of the single knock-out strains that neither the protein encoded by sll1103 nor sll1314 seems to be involved in the transport of 31 additional substrates.
• A next step would be to overexpress sll1314 in ΔfumCΔzwf (our fumarate cell factory). Overexpression of a fumarate transporter may increase fumarate production even further!

▶  Investigating the effect of heterologous transporter expression on fumarate transport

Introduction

Current data on fumarate production in Synechocystis does not indicate that fumarate yields produced in our cell factories exceed the native transport capacity (at least not yet!) [14]. This indicates that overexpressing a transporter is not expected to have a great impact on fumarate productivity. However, this knowledge is based on model predictions and has not been fully experimentally validated.

Therefore, we decided to express in Synechocystis the transporter leading to the highest fumarate yield increase in E. Coli (identified by Zhang et al.)[1]. This is a C ₄ -dicarboxylate carrier, DcuC, with efflux as a major function (hereafter termed DcuC-ec). Bearing in mind that transporters are often not expressed or not correctly folded in a non-native host [1], we decided to also include a transporter from an organism that is more closely related phylogenetically. Nodularia Spumigena , also a cyanobacterium, encodes a putative transporter (hereafter termed DcuC-NS), which is similar to the one used in the Zhang study.

Both transporters are from the DcuC family, a family of transporters present only in (facultative) anaerobic bacteria capable of fumarate respiration. DcuC-ec is able to catalyze the uptake, antiport, and efflux of C ₄ -dicarboxylates (succinate, fumarate, malate) and C ₄ -dicarboxylic amino acids (aspartate), but the Kd values for the various substrates may differ [2]. It is only expressed under anaerobic growth conditions and has efflux as a major function. In the neutral pH range (pH 6-9) divalent C ₄ -dicarboxylates (e.g. fumarate ^2- ) are the substrates and can leave the cell in counter exchange with succinate (eq. 1) or through electrogenic symport (eq. 2).

fumarate _in ^2- + succinate _out ^2- ⇄ fumarate _out ^2- + succinate _in ^2- (1)

3 H ⁺ _in + fumarate _in ^2- ⇄ 3 H ⁺ _out + fumarate _out ^2- (2)

The transporters were turned into versatile BioBricks ( Bba_K2385001 and Bba_K2385000 ) that have the potential to increase the yields of any organic acid producing cell factory. Standard biological parts (like these) and molecular biology tools have been developed for widely studied organisms as E. Coli . These parts can also be used in other organisms, but characterization of parts in a non-native host, reveals differences in performance. This emphasizes the importance of detailed part-characterization for a specific biological chassis. Good characterization and documentation would open up the possibility of enhancing the productivity of traditional (algal) compounds and producing new ones through metabolic engineering, bringing us one step closer to a biobased economy.

We overexpressed the genes separately in wild type and in the ΔfumC strain. In order to investigate the effect on fumarate production, we cultivate the newly designed strains and monitor cell growth and fumarate production.

Results & discussion

We overexpressed DcuC-ec and DcuC-NS (ec = E. Coli , NS = N. Spumigena , cyanobacterium) separately in wild type and in the ΔfumC strain. Unfortunately, we were only able to get the strains on wild type background ready in time (fig. 3.7). Nonetheless, we can still study the effect on fumarate efflux by cultivating the expression strains (DcuC-ec and DcuC-NS) on the wild type background, while using ΔfumC as a positive control, while monitoring cell growth and fumarate production for 150 hours.

Figure 3.7 PCR confirmation of gene expressions on wild type background of DcuC-ec and DcuC-NS. A clean expression gives a single DNA band around 3.120 bp, while for wild type (NC) this is approximately 312 bp.

After 150 hours, no extracellular fumarate is being produced by neither the DcuC-ec nor DcuC-NS expression strains (fig. 3.8A). The ΔfumC is used as a control for fumarate production and produces roughly 1 mM of fumarate after 150 hours. This indicates that under these conditions heterologous expression of DcuC alone, irrespective if it is from E. Coli or N. Spumigena , is not sufficient for fumarate to be secreted above our detection limit (~0.05 mM). To our surprise, we measure extracellular lactate being produced by DcuC-NS with a maximum of 5 mM (fig. 8B). Lactate often indicates that there is a contaminant, but extensive contamination checks did not show contamination of any kind.

Figure 3.8 Extracellular fumarate (A) and lactate (B) production in different Synechocystis strains under constant light illumination for over 150 hours. (n="3" for expression strains, n="2" for ΔfumC). The blue dots (in A) and the red dots (in B) are at the zero baseline underneath the other strain.

Conclusion

• We were able to integrate the heterologous transporters DcuC-ec and DcuC-NS into the genome of Synechocystis. When cultivating for 150 hours, no extracellular fumarate was produced by the DcuC-ec or DcuC-NS expression strain.
• DcuC-NS is a putative C4-dicarboxylate transporter, but yet no experimental evidence can support this. Therefore, more extensive experiment have to be done on this transporter to confirm its function.
• The transporters were turned into versatile BioBricks ( Bba_K2385001 and Bba_K2385000 ) that have the potential to increase the yields of any organic acid producing cell factory.
• A next step would be to express DcuC-ec and DcuC-NS on the ΔfumC background that is already producing fumarate to see if this could further boost fumarate production. This will test simultaneously whether intracellular fumarate was the limiting factor in the experiment described above; and whether it is inhibiting its own production, and potentially even growth, in the strain with the ΔfumC genotype.

▶  How we made the magic happen

Materials & Methods

Cultivation conditions

Cultures during plasmid construction in E. Coli were grown on Luria-Bertani (LB) plates (1,5% agar) or LB liquid medium at 37℃ in an incubator with a shaking speed of 200 rpm. Synechocystis was grown in BG11 standard medium, together with 10 mM TES-NaOH (pH = 8.0). Appropriate antibiotics were used, alone or in combination, in the following concentrations: 50 μg mL ^-1 kanamycin, 100 μg mL ^-1 ampicillin and 30 μg mL ^-1 nickel.

Sampling and Fumarate measurements

Cell culture samples were harvested at selected time points for OD and fumarate analysis. 1 ml of sample was centrifuged at 15.000 rpm for 10 min. Then 500 μl supernatant was taken and filtered (Sartorius Stedin Biotech, minisart SRP 4, 0.22 μm) for sample preperation. Fumarate concentration was measured by HPLC-UV/VIS (LC-20AT, Prominence, Shimadzu), with ion exclusion Rezex ROA-Organic Acid column (250x4.6 mm; Phenomenex) and UV detector (SPD-20A, Prominence, Shimadzu) at 210 nm wavelength. 10 μL of the HPLC samples were injected through an autosampler (SIL-20AC, Prominence, Shimadzu), with 5 mM H ₂ SO ₄ as eluent at a flow rate of 0.15 ml min ^-1 and column temperature of 45 ℃.

Deleting putative transporter encoding genes in Synechocystis

sll1103 (BAA18318.1) and sll1314 (BAL35144.1) were deleted in wild type Synechocystis and in the fumarate producing ΔfumC strain. Gene deletion was done according to the markerless genetic modification method, with kanamycin as a positive selection marker (50 μg/ml). Homologous regions were amplified using primers up_fwd and up_rev (H1) and down_fwd and down_rev (H2). Segregation was checked using (colony)PCR with primers up_fwd and down_rev. All primers used are listed in table 1. The resulting, markerless deletion strains are listed in table 2.

Table 1. Primers used in this study.

Table 2. Strains constructed in this study.

Overexpressing transporters to increase fumarate yield

For the expression of heterologous transporters in Synechocystis , a C ₄ -dicarboxylate transporter gene DcuC (NP_415154) was amplified from the genome DNA of Escherichia coli . Homologous regions were amplified using primers DcuC_ec_up_fwd and DcuC_ec_up_rev (H1) and DcuC_ec_down_fwd and DcuC_ec_down_rev (H2). DcuC-NS (AHJ29173.1) from Nodularia Spumigena was synthesized at IDT as a gene-block. The genes were integrated into the genome of wild type Synechocystis and ΔfumC , directly downstream of the psbA2 promotor, as described elsewhere [15]. Segregation was checked using (colony)PCR with primers on the expression vector. All primers used are listed on in table 1. The resulting expression strains are listed in table 2.

Characterisation

We study the effect of gene modifications on growth and fumarate transport in a multi-cultivator . The cultivator consists of eight vessels, each containing a Synechocystis strain.

Experiment 1 - The effect of gene deletions on fumarate efflux

We study the effect of the gene deletions on fumarate efflux by cultivating the double knock-out strains ΔfumC Δsll1103 (in triplicate) and ΔfumC Δsll1314 (in triplicate), with ΔfumC (produces fumarate in a stable fashion) as a control (in duplicate). Sampling was done once a day, as described above.

Experiment 2 - The effect of gene deletions on fumarate influx

We study the effect of the gene deletions on fumarate influx by cultivating the single knock-out strain Δsll1103 (in duplo) and wild type Synechocystis (in duplo) with and without external fumarate supplemented (5mM) to the cultures. Sampling was done once a day, as described above.

Experiment 3 - The effect of gene deletions on the "metabolic fingerprint" of Synechocystis

Further characterization of the deletion strains was done making a so called "metabolic fingerprint" on a Biolog Ecoplate ^TM . This was done for the single knock-out strains Δsll1103 and Δsll1314 , with wild type as a control. The Biolog Ecoplate ^TM comprises of 96 wells, containing 31 of the most useful carbon sources in triplicate [16]. Formation of purple color occurs when the microbes can utilize the carbon source and begin to respire. The respiration of the cells in the community reduces a tetrazolium dye that is included with the carbon source. One plate was inoculated in singleton with wild type, Δsll1103 and Δsll1314 at OD ₇₃₀ 0.5 from a culture that had been growing in continuous light. The plate was incubated at 30 ℃ and OD was measured daily with a BMG FluoStar Optima plate reader, using emission at 600 nm. Another plate was inoculated according to the same scheme, after 24 hours of dark incubation.

Experiment 4 - The effect of heterologous gene expression on fumarate efflux

We study the effect of the heterologous gene expressions on fumarate efflux by cultivating the DcuC-ec (in triplicate) and DcuC-NS (in triplicate), with ΔfumC (produces fumarate in a stable fashion) as a control (in duplicate). Sampling was done once a day, as described above.

References

1. IG Janausch, E Zientz, QH Tran, A Kröger, and G Unden. C 4-dicarboxylate carriers and sensors in bacteria. Biochimica et Biophysica Acta (BBA)-Bioenergetics, 1553(1):39-56, 2002.
2. Paulsen, I. T., Nguyen, L., Sliwinski, M. K., Rabus, R., & Saier, M. H. (2000). Microbial genome analyses: comparative transport capabilities in eighteen prokaryotes. Journal of molecular biology, 301(1), 75-100.
3. Rost, B. (1999). Twilight zone of protein sequence alignments. Protein engineering, 12(2), 85-94.
4. Pirovano, W., Feenstra, K. A., & Heringa, J. (2008). PRALINE™: a strategy for improved multiple alignment of transmembrane proteins. Bioinformatics, 24(4), 492-497.
5. Söding, J. (2004). Protein homology detection by HMM-HMM comparison. Bioinformatics, 21(7), 951-960.
6. Altschul, S. F., Madden, T. L., Schäffer, A. A., Zhang, J., Zhang, Z., Miller, W., & Lipman, D. J. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic acids research, 25(17), 3389-3402.
7. Thompson, J. D., Gibson, T., & Higgins, D. G. (2002). Multiple sequence alignment using ClustalW and ClustalX. Current protocols in bioinformatics, 2-3.
8. Marchler-Bauer, A., Derbyshire, M. K., Gonzales, N. R., Lu, S., Chitsaz, F., Geer, L. Y., ... & Lanczycki, C. J. (2014). CDD: NCBI"s conserved domain database. Nucleic acids research, 43(D1), D222-D226.
9. Bairoch, A., Bucher, P., & Hofmann, K. (1997). The PROSITE database, its status in 1997. Nucleic Acids Research, 25(1), 217-221.
10. Forward, J. A., Behrendt, M. C., Wyborn, N. R., Cross, R., & Kelly, D. J. (1997). TRAP transporters: a new family of periplasmic solute transport systems encoded by the dctPQM genes of Rhodobacter capsulatus and by homologs in diverse gram-negative bacteria. Journal of bacteriology, 179(17), 5482-5493.
11. Kaneko, T., Sato, S., Kotani, H., Tanaka, A., Asamizu, E., Nakamura, Y., ... & Kimura, T. (1996). Sequence analysis of the genome of the unicellular cyanobacterium Synechocystis sp. strain PCC6803. II. Sequence determination of the entire genome and assignment of potential protein-coding regions. DNA research, 3(3), 109-136.
12. Quintero, M. J., Montesinos, M. L., Herrero, A., & Flores, E. (2001). Identification of genes encoding amino acid permeases by inactivation of selected ORFs from the Synechocystis genomic sequence. Genome Research, 11(12), 2034-2040.
13. Du., W. unpublished.
14. Savakis, P. E., Angermayr, S. A., & Hellingwerf, K. J. (2013). Synthesis of 2, 3-butanediol by Synechocystis sp. PCC6803 via heterologous expression of a catabolic pathway from lactic acid-and enterobacteria. Metabolic engineering, 20, 121-130.
15. Choi, K. H., & Dobbs, F. C. (1999). Comparison of two kinds of Biolog microplates (GN and ECO) in their ability to distinguish among aquatic microbial communities. Journal of Microbiological Methods, 36(3), 203-213.

▶  Supplementary Material