Adaptive laboratory evolution of cadmium tolerance in Synechocystis sp. PCC 6803

Background Cadmium has been a significant threat to environment and human health due to its high toxicity and wide application in fossil-fuel burning and battery industry. Cyanobacteria are one of the most dominant prokaryotes, and the previous studies suggested that they could be valuable in removing Cd2+ from waste water. However, currently, the tolerance to cadmium is very low in cyanobacteria. To further engineer cyanobacteria for the environmental application, it is thus necessary to determine the mechanism that they respond to high concentration of cadmium. Results In this study, a robust strain of Synechocystis PCC 6803 (named ALE-9.0) tolerant to CdSO4 with a concentration up to 9.0 µM was successfully isolated via adaptive laboratory evolution over 802-day continuous passages under cadmium stress. Whole-genome re-sequencing was then performed and nine mutations were identified for the evolved strain compared to the wild-type strain. Among these mutations, a large fragment deletion in slr0454 encoding a cation or drug efflux system protein was found to contribute directly to the resistance to Cd2+ stress. In addition, five other mutations were also demonstrated related to the improved Cd2+ tolerance in ALE-9.0. Moreover, the evolved ALE-9.0 strain was found to obtain cross tolerance to some other heavy metals like zinc and cobalt as well as higher resistance to high light. Conclusions The work here identified six genes and their mutations related to Cd2+ tolerance in Synechocystis PCC 6803, and demonstrated the feasibility of adaptive laboratory evolution in tolerance modifications. This work also provided valuable information regarding the cadmium tolerance mechanism in Synechocystis PCC 6803, and useful insights for cyanobacterial robustness and tolerance engineering. Electronic supplementary material The online version of this article (10.1186/s13068-018-1205-x) contains supplementary material, which is available to authorized users.


Background
In recent years, environmental pollution caused by heavy metals has caused serious problems, including contaminating water, entering food chain and posing threats to growth of living organisms in nature [1,2]. Among them, cadmium ion (Cd 2+ ) is one of the most dangerous heavy metals [3]. Cadmium residues mainly come from industrial products like nickel-cadmium battery and pigmenting, which inevitably pollute a large amount of water [4]. Nowadays, cyanobacteria, which have been considered as "photosynthetic microbial factories" in the biosynthesis of fine chemicals and biofuels, have attracted much attention [5]. On the other hand, as one of the most dominant prokaryotes on Earth, cyanobacteria play a pivotal role in the global carbon cycling [6], while were threatened sometimes by the unfriendly environment. For example, it was reported that the cadmium concentration could reach 0.36 ± 0.82 mg/L (3.2 ± 7.3 µM) in the industrial area of Penang, Malaysia [7], which would pose significant threat to the survival of cyanobacteria. Thus, it is essential to understand how they respond to environmental stresses such as Cd 2+ . In addition, removal of toxic metal ions such as Cd 2+ from water by cyanobacteria has been widely evaluated in recent years and is considered as a promising alternative treatment in wastewater purification [8]. For instance, a study on adsorption of Cd 2+ by Gloeothece magna suggested that they would probably be cultivated in water bodies contaminated by Cd 2+ to ameliorate its toxicity effectively [9]. Therefore,

Open Access
Biotechnology for Biofuels *Correspondence: lchen@tju.edu.cn 1 Laboratory of Synthetic Microbiology, School of Chemical Engineering & Technology, Tianjin University, Tianjin 300072, People's Republic of China Full list of author information is available at the end of the article it will be of great value to decipher the tolerance mechanism to Cd 2+ in cyanobacteria.
Cellular responses of cyanobacteria to high concentration of Cd 2+ have been investigated in the past decades. In a previous study, the direct influence of Cd 2+ to photosynthetic machinery was found to be multiphase effects in model cyanobacteria Synechocystis sp. PCC 6803 (here after Synechocystis). The results showed that Cd 2+ first limited photosystem I acceptor side, and 7 h later, it disturbed photosystem II under the existence of light [10]. In addition, Cd 2+ toxicity caused the generation of reactive oxidative species (ROS) and the consumption of glutathione as well as the thiol-group containing protein [11]. Furthermore, Cd 2+ penetrated rapidly into the cells and replaced other heavy metals like Ca 2+ and Zn 2+ by competing for enzymes and disturbing membrane potential [12]. In another study, Cd 2+ triggered the integrated reprogramming of the whole metabolism in Synechocystis, which was controlled by the Slr1738 regulator [13]. Meanwhile, some genes were found involved in resistance to Cd 2+ , like smtA in Synechococcus PCC 7942 [14] and sll0649 in Synechocystis [8]. Together, cellular responses to Cd 2+ toxicity involved a variety of complex reaction mechanisms. However, up to now, the detailed mechanism of Cd 2+ toxicity to cyanobacteria has still not been fully understood. Although some algae like Phormidium ambiguum and Scenedesmus quadricauda var. quadrispina show good tolerance to Cd 2+ at the concentration of up to 0.35 mM [15], they are non-model organisms and relatively difficult to be deeply studied, while Synechocystis sp. PCC 6803 is a model organism with known genomic information [16] and relatively easily genetic operation, although it showed poor tolerance to Cd 2+ stress [8]. Therefore, it is necessary to enhance the Cd 2+ tolerance of Synechocystis and understand how the Cd 2+ tolerance is regulated, so that to guide the tolerance engineering in other algal or cyanobacterial species.
Adaptive laboratory evolution (ALE) is a strategy to improve strains via constant batch transfer under specific growth conditions [17], and has been considered as a powerful tool to generate robust strains with enhanced tolerance to multiple stresses [18]. Although it is timeconsuming for strains to accommodate, systematic modifications on genome scale could be obtained through the ALE process [19]. Predictably, ALE could also bring other consequences like trade-off in growth or cross tolerance in alternative environments [18]. In addition, mature high-throughput sequencing and genetic manipulation systems make ALE possible for mechanism research and even further phenotypes modification [20]. In Saccharomyces cerevisiae, four strategies to isolate cobalttolerant cells were performed by ALE, yielding the most resistant mutant to cobalt stress from 2.5 to 8 mM, which indicated the efficiency of ALE to improve strains [21]. In addition, ALE was employed successfully in Synechocystis to improve 1-butanol tolerance from a concentration of 0.2-0.5% (v/v) and a further metabolomic basis for rational tolerance engineering was determined [22].
In this study, the ALE strategy was employed to improve the tolerance of Synechocystis to Cd 2+ stress. As a result, after 128 continuous passages of approximately 802 days, tolerance of the evolved strain was improved from 4.6-to 9.0-µM Cd 2+ . With the aid of high-throughput re-sequencing technology, the mutations in the genome of the evolved strain compared with the wildtype strain were identified and further functionally characterized. This study demonstrated the feasibility of ALE in tolerance modifications and provided useful insights for cyanobacterial robustness and tolerance engineering.

Adaptive laboratory evolution of Cd 2+ tolerance in Synechocystis
Wild-type (WT) Synechocystis strain was evolved by serial passaging for 128 passages (802 days) in BG11 medium supplemented with CdSO 4 , as a selective pressure to enrich population with Cd 2+ tolerance. The starting Cd 2+ concentration for WT was set as 4.6 µM as our previous study showed that WT showed a slight growth deficiency at this Cd 2+ concentration level [23]. Under normal BG11 medium without CdSO 4 , Synechocystis could achieve late exponential phase (OD 750 nm = 1.5) from an initial inoculum (OD 750 nm = 0.1) within 96 h. When CdSO 4 was added, cell growth rate decreased obviously. In this ALE process, a simple rule was established that once the evolutionary strain could reach OD 750 nm of 0.5 (from an initial inoculum of OD 750 nm of 0.1) within 96 h, the Cd 2+ concentration was increased. Thus, the Synechocystis strain was cultivated with increasing Cd 2+ concentration from 4.6 µM gradually to 9.0 µM (Fig. 1a). Figure 1b demonstrates the simplified process of the ALE experiment, as a gradually increasing concentrations of Cd 2+ process. Finally, after 128 continuous passages of 802 days, a strain that could tolerate 9.0-µM Cd 2+ was obtained, approximately 95% improvement in tolerance compared with WT. At the end of ALE process, the evolved strain was plated on BG11 agar plate supplemented with 9.0-µM CdSO 4 . Four single colonies were cultured individually under 9.0-µM CdSO 4 and the one which showed the fastest growth (data not shown; named ALE-9.0) ( Table 1) was selected for further study.
As shown in Fig. 2a, under normal BG11 medium condition without Cd 2+ , ALE-9.0 grew slightly, but not significantly, slower than WT, while under 9.0-µM CdSO 4 condition, the growth of ALE-9.0 was dramatically better than WT, as WT can hardly survive under 9.0-µM Cd 2+ , demonstrating the improved Cd 2+ tolerance in ALE-9.0. To show the effect of ALE process better, tenfold serial dilutions of WT and ALE-9.0 liquid cultures were spotted onto BG11 agar plates with different concentrations of CdSO 4 (Fig. 2b). Under normal BG11 medium, WT showed little difference in growth condition with ALE-9.0. However, once Cd 2+ was added, WT could hardly survive under 6.0-µM Cd 2+ , while ALE-9.0 still remained robust growth even under 9.0-µM Cd 2+ . In addition, ALE-9.0 showed a little yellow-green phenotype under the normal illumination of 50-μ mol photons m −2 s −1 . Full absorption spectrum indicated that ALE-9.0 had less phycocyanin at 625 nm [24] but more carotenoid around 505 nm [25] than WT in normal BG11 medium (Fig. 2c,  d).
Whole-genome re-sequencing of ALE-9.0 Notably, WT could only endure 4.6-µM CdSO 4, while ALE-9.0 could survive in medium supplemented with 9.0-µM CdSO 4 , indicating that some intrinsic genetic changes occurred during the ALE process. Wholegenome re-sequencing technology was then employed to identify the genomic differences between ALE-9.0 and WT. Genomes of ALE-9.0 and WT were re-sequenced and compared with reference from the database to check out the differences (https ://www.ncbi.nlm.nih.gov/nucco re/NC_00091 1). In all, one deletion, nine single-nucleotide polymorphisms (SNPs) and four structural variations (SVs) were detected in ALE-9.0 compared to WT after the evolution process, and then, all these mutations were individually verified by Sanger sequencing. As a result, seven SNPs and one SV were identified, and the deletion turned out to be one insertion ( Table 2). Among the confirmed nine mutations (seven SNPs, one SV, and one insertion), the insertion in slr1753 was located in a high repetition region with low mapping quality, the SNP in sll1586 was synonymous, while the others were nonsynonymous. The only SV was a 659-bp-deletion located in the ORF of gene slr0454, which resulted in an early termination of a truncated slr0454 encoding a protein of 574 amino acids (the original gene slr0454 encoding a protein of 890 amino acids). It is speculative that these mutations might lead to an increased tolerance of Synechocystis to Cd 2+ in the strain ALE-9.0.
To investigate the roles of these mutations, one strain (ALE-7.0) evolved in the middle term of this evolution  process and tolerant to 7.0-µM Cd 2+ (Fig. 1b) was also selected, cultivated, and sequenced by Sanger sequencing concerning these mutations. The results showed that six out of these nine mutations found in ALE-9.0 were present in ALE-7.0 when CdSO 4 concentration reached 7.0 µM. Thus, during the increasing concentrations of CdSO 4 from 7.0 to 9.0 µM in this ALE, only three mutations (non-synonymous SNPs in slr0721, slr1302 and the SV in slr0454) occurred in the later stage of the whole ALE process (Table 2), indicating their roles in later improvement of Cd 2+ tolerance.  slr0623 Hypothetical protein Y

Quantitative reverse transcription PCR (qRT-PCR) analysis of the mutated genes in WT and ALE-9.0
Besides the genetic differences between WT and ALE-9.0, the expression level of the mutated genes could possibly also change under CdSO 4 stress conditions. To evaluate this hypothesis, five samples (WT cultured in normal BG11 medium and under 4.6-µM CdSO 4 , ALE-9.0 cultured in normal BG11 medium, under 4.6-µM CdSO 4 and 9.0-µM CdSO 4 ) were selected for qRT-PCR analysis. The expression change was presented by the ratio of the relative expression level of the genes under stress condition to that in normal BG11 medium, respectively, and a fold change > 2.0 was used as a cutoff. As shown in Table 3, expression level of most genes did not show significant changes in WT under 4.6-µM CdSO 4 stress compared with WT cultured in normal BG11 medium, probably because that these genes did not respond to Cd 2+ under this concentration. However, transcriptional levels of most genes were increased significantly in ALE-9.0 when CdSO 4 was added. Particularly, the most significant changes were found in slr0721 and slr0798. Although the relative expression level of On the other hand, the expression of slr0454, slr0774, and slr1302 was only slightly changed in ALE-9.0 under both Cd 2+ conditions (fold changes less than 2.0). According to the above results, the expression levels of six genes, slr0721, slr0798, slr1753, sr1480, sll1586, and slr0623, were significantly induced upon the exposure to Cd 2+ after this ALE process, indicating their relevant roles in Cd 2+ tolerance of ALE-9.0.

Screening of the mutated genes related to Cd 2+ stress via knockout analysis
Gene knockout analysis was then used to evaluate the relationship of the nine mutated genes revealed by genome re-sequencing with Cd 2+ tolerance. Relevant genes were replaced with chloramphenicol-resistance cassettes in WT and ALE-9.0 by homologous recombination, respectively (the related names of mutants were shown as ∆X WT and ∆X ALE-9.0 ), and knock of neutral site slr0168 in WT (WT-C) and ALE-9.0 (ALE-9.0-C) was selected as controls (Table 1). 4.0-or 8.0-µM CdSO 4 was added as additional stress for knockout strains derived from WT and ALE-9.0 to eliminate the intolerance to Cd 2+ caused by poor growth.
As shown in Fig. 3, ∆slr1302 WT showed significantly poor growth than WT-C in normal BG11 medium (Fig. 3d), while the others showed similar growth as WT-C, suggesting that only the knockout of slr1302 affected the growth of WT under normal BG11 condition. Under this circumstance, it cannot be told whether slr1302 was involved in Cd 2+ tolerance or not. For the remained eight genes, under 4.0-µM CdSO 4 condition, ∆slr0454 WT , ∆slr0721 WT , ∆sll1586 WT , and ∆slr1753 WT did not show significant difference from WT-C (Fig. 3a, c, h, i), while ∆ssr1480 WT , ∆slr0623 WT , ∆slr0774 WT , and ∆slr0798 WT demonstrated to be more sensitive to Cd 2+ compared with WT-C (Fig. 3b, d, e, g), indicating that these four genes might be involved in Cd 2+ tolerance in the WT strain. Among them, ∆slr0798 WT could hardly grow under 4.0-µM CdSO 4 (Fig. 3e), suggesting clearly the importance of slr0798 to Cd 2+ tolerance in WT.
The nine mutated genes were also knockout individually in ALE-9.0. As shown in Fig. 4, ∆slr1302 ALE-9.0 showed the same growth pattern with ∆slr1302 WT (Figs. 3f, 4f ), suggesting that it also affected the growth of ALE-9.0; thus, no conclusion can be made whether slr1302 was involved in Cd 2+ tolerance or not. For the remained eight knockout strains, under 8.0-µM CdSO 4 condition, ∆ssr1480 ALE-9.0 and ∆sll1586 ALE-9.0 did not show significant difference compared with ALE-9.0-C (Fig. 4g, h). Notably, the other six mutants, i.e., ∆slr0454 ALE-9.0 , ∆slr0623 ALE-9.0 , ∆slr0721 ALE-9.0 , ∆slr0774 ALE-9.0 , ∆slr0798 ALE-9.0, and ∆slr1753 ALE-9.0 were demonstrated to be more sensitive to Cd 2+ stress than ALE-9.0-C (Fig. 4a-e, i), indicating their vital roles in Cd 2+ tolerance in the evolved strain ALE-9.0. Considering the combined results of knockout analysis in both WT and ALE-9.0, slr1302 was involved in the growth of both relative strains under normal BG11 condition though the relative mutants were more sensitive than WT or ALE-9.0 under Cd 2+ stress condition (Figs. 3f, 4f ). Therefore, it remained to be investigated whether it was also involved in Cd 2+ tolerance or not, because the growth deficiency would also result in the sensitive phenotype under Cd 2+ stress condition. For the remained eight genes, their involvement in Cd 2+ tolerance did not fully match each other in WT and ALE-9.0, probably due to the different concentrations of Cd 2+ stress, indicating the complexity of Cd 2+ tolerance regulation mechanism in Synechocystis.

Confirmation of the roles of the mutated genes in Cd 2+ tolerance by gene overexpression in the WT strain
To evaluate the roles of these genes, especially their mutations in modifying the tolerance of Cd 2+ , the original gene in WT and their mutated genes after evolution were, respectively, overexpressed in the WT strain using an integrative vector pCP3031 with a strong promoter Pcpc 560 [26]. The constructed strains were named OE-X WT and OE-X ALE-9.0 . Meanwhile, the empty vector of pCP3031 was overexpressed in WT and the resulted strain OE-C was set as control ( Table 1). The growth patterns of these constructed strains were monitored under corresponding Cd 2+ stress condition.
Among these nine genes, overexpression of four genes, slr0454 WT , slr0623 WT , slr0721 WT  slr0798 WT , as well as their corresponding mutated genes, slr0454 ALE-9.0 , slr0623 ALE-9.0 , slr0721 ALE-9.0 , and slr0798 ALE-9.0 , successfully improved the Cd 2+ tolerance compared with OE-C, while all these engineered strains demonstrated similar growth as OE-C in normal BG11 medium (Fig. 5a-d), further demonstrating that their expression levels were important for the Cd 2+ tolerance. On the other hand, the overexpression of one mutated gene, slr0454 ALE-9.0 exhibited better growth compared with the expression of slr0454 WT , suggesting that the truncated Slr0454 ALE-9.0 was probably more effective than Slr0454 WT . Therefore, the results showed that both the expression level of slr0454 and the activity of protein Slr0454 were important for Cd 2+ tolerance, indicating the vital role of this gene in Cd 2+ tolerance.
According to the NCBI annotation, slr0454 encodes a cation or drug efflux system protein belonging to Acriflavin-cation resistance (Acr) family. Consistent with our results, this kind of protein was reported previously to help Gram-negative bacteria to keep the intracellular homeostasis under high metal concentrations [27].
In addition, the result of protein BLAST showed that Slr0454 was homologous to inner membrane transporter, AcrB of Escherichia coli (E. coli), which has been found mediated resistance to ions including nickel, cadmium, and cobalt [28].
For the remained five genes, none of them showed positive effect on Cd 2+ tolerance under Cd 2+ stress condition when the WT or mutated genes were overexpressed (Additional file 1: Fig. S1a-e) of three genes (i.e., slr0774 WT , slr0774 ALE-9.0 , ssr1480 WT , ssr1480 ALE-9.0 , slr1753 WT , and slr1753 ALE-9.0 ) even had negative effects on the tolerance to Cd 2+ , probably due to that the expression of these genes was already saturated, or their overexpression has brought extra metabolic burden to WT or ALE-9.0.

Cross tolerance to other stresses obtained in ALE-9.0
Possible cross tolerance of ALE-9.0 to other stresses was also investigated, including ZnSO 4 , CoCl 2 , CuSO 4 , high light, ethanol, and H 2 O 2 . The results showed that compared to WT, ALE-9.0 grew better under stress conditions of ZnSO 4 and CoCl 2 (Fig. 6a), while grew worse in other metal stresses like CuSO 4 (Additional file 1: Fig. S2).
Notably, ALE-9.0 showed better acclimation to higher illumination intensity at 200-μmol photons m −2 s −1 than WT (Fig. 6b). Although ALE-9.0 grew slower than WT at the very beginning, it was able to catch up with WT at OD 750 nm at day 5 and keep growing for almost 8 days, with the final OD 750 nm greater than WT. In addition, WT showed a bleaching phenotype, while ALE-9.0 was still yellow-green after cultured for 8 days (Fig. 6c). H 2 O 2 and biofuel ethanol stress were also investigated, while no enhanced tolerance (actually decreased tolerance) were observed between WT and ALE-9.0 (Additional file 1: Fig. S2). Together, the results showed that along with the improved Cd 2+ tolerance in ALE-9.0 during the ALE process, cross tolerance to Zn 2+ , Co 2+ , and high light (at late growth phase) were also obtained.

Discussion
ALE has been demonstrated as an effective approach to obtain desired biological properties of the evolved strain. The titer of d-lactic acid produced by the evolved strain has been increased 2.0-fold than the original strain in Leuconostoc mesenteroides [29]. In addition, adaptive evolution under thermal stress not only increased the survival temperature from 33 to 41.5 °C, but also conferred cross tolerance to isobutanol in Corynebacterium glutamicum [30]. In this study, ALE was also applied to enhance Cd 2+ tolerance of Synechocystis and an evolved strain tolerant to 9.0-µM CdSO 4 was isolated after 802 day series passages. The slightly slower growth of ALE-9.0 compared with WT under normal BG11 condition (Fig. 2a, b) could be ascribed to the trade-off character to balance tolerance of higher additional stress [31]. Considering changes in color and full absorption spectrum (Fig. 2c, d), since signal near 625 nm of full absorption spectrum was measured as phycocyanin content [32] and degradation of the phycocyanin caused a color change of cyanobacterial cultures from blue-green to yellow-green [33], ALE-9.0 demonstrated less phycocyanin than WT. In addition, phycocyanin was the major part of phycobilisomes for harvesting light and causing energy migration toward photosystem reaction centers [34]. Meanwhile, early study showed that lack of phycocyanin would result in poor light-harvesting ability [24]. Thus, we speculated that the light-harvesting ability of ALE-9.0 became weaker after long-time exposure to Cd 2+ .
ALE always brings about global changes at wholegenomic, transcriptional, and metabolomic levels [35]. In a previous research with Synechocystis, an acid-evolved strain identified 11 mutations in the genome, and the transcriptional differences were demonstrated by qRT-PCR [36]. A recent study improved isobutanol tolerance from 2 to 5 g/L with combinatorial malfunctions of three genes [37]. Consistently, in this study, nine mutations were identified by the whole-genome re-sequencing. Combined with knockout and overexpression analysis, out of these nine mutations, only the SV in slr0454 was demonstrated to improve the tolerance directly (Fig. 5a). As mentioned above, Slr0454 was homologous to AcrB. It was reported that AcrB cooperated with a membrane fusion protein AcrA and an outer membrane channel TolC to form an AcrAB-TolC system to export drugs [38]. In this study, the results also showed that the truncated Slr0454 contributed to tolerance of Cd 2+ in Synechocystis, consistent with a previous study in Synechococcus sp. PCC 7942 showing the truncated form of the bacterial heat shock protein ClpB contributed to development of thermotolerance [39], probably by transforming the protein structure into a more effective conformation.
On the other hand, five mutated genes, slr0623, slr0721, slr0798, slr0774, and slr1753, were demonstrated to be involved in the Cd 2+ tolerance in ALE-9.0 via knockout analysis (Fig. 4b-e, i). Among these five genes, although no positive effects were shown in the overexpression analysis between the WT genes and mutated ones, the increased expression of slr0623, slr0721, and slr0798 was demonstrated to contribute to the increased Cd 2+ tolerance in ALE-9.0 (Table 3, Fig. 5b-d). slr0623 encodes thioredoxin (Trx) TrxA, which is a class of small redox proteins known to be present in most microorganisms. Consistent with our results, research in E. coli showed Trx was inhibited by Cd 2+ and posed a positive role in protection from Cd 2+ [40]. According to this study, Cd 2+ directly bound to Trx by forming a chelator and decreased thiol-disulfide transferase activity, and this kind of Cd 2+ sink might help against Cd 2+ stress. the decarboxylating NADP-dependent malic enzyme, participating in tricarboxylic acid cycle. A study on malic acid and Cd 2+ stress in Miscanthus sacchariflorus proved exogenous addition of malic acid could alleviate Cd 2+ toxicity through enhancing photosynthetic capacity and restraining ROS accumulation [41]. Even though this phenomenon was widely discovered in plants [42,43], we supposed that similar tolerance mechanism might also be utilized in cyanobacteria, while further evidences are still needed. It was noteworthy that expression level of slr0721 .0 in ALE-9.0 did not show significant change under 4.6 µM, but was increased significantly under 9.0-µM CdSO 4 (Table 3). Meanwhile, Sanger sequencing showed that the mutation of slr0721 occurred in later stage after 7.0 µM ( Table 2), indicating the possibility of different response mechanism that this gene was involved to Cd 2+ stress. Finally, Slr0798 is an SmtB-like repressor concerning zinc-transporting P-type ATPase involved in zinc tolerance [44]. Consistent with the previous study in Synechocystis, overexpression of slr0798 gene with a replicative vector pJA2 could significantly enhance Cd 2+ tolerance [23]. Unlike SmtB, Slr0798 triggered excess Zn 2+ expulsion by via Slr0798-mediated efflux into the periplasm, which we supposed was the same mode for Cd 2+ . According to the qRT-PCR analysis (Table 2), the expression of slr0798 ALE-9.0 was significantly up-regulated under CdSO 4 stress, suggesting the importance of expression level of this gene in Cd 2+ tolerance. The other two mutated genes, slr0774 and slr1753 shared similar functions related to membrane protein. These two genes showed involvement in Cd 2+ tolerance in the knockout analysis, but neither could improve tolerance directly according to the overexpression analysis, probably due to that the expression of these genes was already saturated for Cd 2+ tolerance. slr0774 encodes SecD, a part of Sec protein. The general secretory (Sec) pathway was considered as a major translocation process of protein from cytosol across the cytoplasmic membrane in bacteria [45] and SecD acted as an auxiliary component to enhance translocation efficiency [46]. It is then speculative that the mutation of slr0774 could lead to different efficiency of SecD or help SecD interact with other membrane protein better, leading to high Cd 2+ tolerance. In addition, slr1753 was found as an outer membrane fraction for its homology to a cell-surface glycoprotein in Clostridium thermocellum [47]. Besides glycoprotein interacted selectively and non-covalently with carbohydrate and increase of EPS production enhanced Cd 2+ resistance [48], it is thus supposed that the mutation of slr1753 possibly was able to help optimize the constitution and content of saccharide on cell-surface, leading to different Cd 2+ tolerances. Since the ALE-9.0 could tolerate 9.0-µM Cd 2+ , except for the function of each single gene, the combinations of these genes and mutations to work together most probably also exist [30] which is still yet to be determined. Mutations occur randomly during the course of ALE and are selected naturally when a particular mutation enhances the activity of a protein and/or thereby the better tolerant or survival [37]. For Cd 2+ tolerance, some genes including slr0649, slr0946, slr1738, and sll1598 were also demonstrated to be related to Cd 2+ tolerance in the previous studies [8]. However, these genes were not found to mutate during our ALE process, which might be due to, we supposed, the randomness of ALE experiment, and meanwhile, on the other hand, it was also possible that although these genes did not mutate, their expression level might change, which remains to be investigated.
As for cross tolerance, the evolved ALE-9.0 also obtained cross tolerance to Zn 2+ , Co 2+ , and high illumination intensity (Fig. 6a, b). Zn 2+ belongs to the same group as Cd 2+ , which have the same chemical properties, consistent with that some mutated genes in ALE-9.0 like slr0798 also showed involvement in Zn 2+ tolerance [8]. In addition, the previous studies have found that Cd 2+ and Co 2+ shared some similar toxicity mechanism [49] and Acr family showed resistance to Co 2+ too [28], consistent with the finding that slr0454 involved in Co 2+ tolerance. In addition, this was also confirmed in our experiments by the result that a higher tolerance to Co 2+ was exhibited in ALE-9.0 (Fig. 6a). For high light, ALE-9.0 was hypothesized less sensitive to strong light than WT because of the poor light-harvesting ability [22]. At first, WT grew quickly as a result of enough light and ample carbon resource for downstream reaction, but later, the accumulation of ROS emerged became dominant [50] and caused the photo damage [24], while ALE-9.0 could absorb enough light only for growth and strong light did less harm to it, which also resulted in larger biomass of ALE-9.0 than WT (Fig. 6b). In addition, our results showed that ALE-9.0 did not demonstrate enhanced tolerance (actually even decreased tolerance, Additional file 1: Fig. S2) to H 2 O 2 . Although Cd 2+ induces stresses including oxidative stress, tolerance of strains to Cd 2+ involves many aspects like ion efflux and chelation [51], so it can be supposed that during this ALE process, the enhanced Cd 2+ tolerance of ALE-9.0 may not involve oxidative tolerance. Besides, the ALE-9.0 grew slower than WT under normal condition, which may also result in the sensitiveness to most of other unrelated stresses such as H 2 O 2 .

Conclusion
In this study, an evolved strain ALE-9.0 of Synechocystis that could tolerate up to 9-µM CdSO 4 after 802 day ALE process was obtained. The mutations in the genome of ALE-9.0 compared with WT were identified by genome re-sequencing. One mutation of slr0454 was demonstrated capable of improving Cd 2+ tolerance directly and five mutated genes, slr0623, slr0721, slr0798, slr0774, and slr1753, were demonstrated involved in the Cd 2+ tolerance in ALE-9.0. In addition, the evolved ALE-9.0 also obtained cross-tolerance ability to Zn 2+ , Co 2+ , and high light. Our work here identified six genes related to Cd 2+ tolerance and demonstrated the feasibility of ALE in tolerance modifications. This work also provided valuable information to decipher the cadmium tolerance mechanism in Synechocystis and useful insights for cyanobacterial robustness and tolerance engineering.

Bacterial growth conditions
The wild-type Synechocystis, laboratory-evolved, and constructed strains were grown on BG11 agar plate or in BG11 medium (pH 7.5) under a light intensity of approximately 50-μmol photons/m 2 /s in an illuminating or shaking incubator of 130 rpm at 30 °C (HPX-9162 MBE, BOXUN, China, HNY-211B Illuminating Shaker, Honour, China) [52]. Proper antibiotic was added to maintain the stability (i.e., 20-μg/mL spectinomycin, 20-μg/mL chloramphenicol) of the constructed strains. Cell optical density and full absorption spectrum were monitored by a UV-1750 spectrophotometer (Shimadzu, Japan) at 750 nm. E. coli strain DH 5α was used for constructing and collecting plasmids. E. coli was grown on LB agar plate or in LB liquid medium in incubator at 37 °C or shaking incubator at 200 rpm supplemented with appropriate antibiotic (i.e., 50-μg/mL spectinomycin, 50-μg/ mL chloramphenicol, and 200-μg/mL ampicillin).

Adaptive laboratory evolution of Cd 2+ tolerance
Adaptive laboratory evolution of Cd 2+ tolerance was carried out in 20 mL liquid BG11 medium in a 100 mL shake flask. The CdSO 4 stock solution was prepared with CdSO 4 ·8/3H 2 O of analytical pure, purchased from Aladdin (Shanghai, China). The initial WT stain was cultured with 4.6-µM CdSO 4 from an inoculum of OD 750 nm 0.1. Cd 2+ concentration in BG11 medium was increased by 0.3-0.4 µM when the culture reached OD 750 nm of 0.5 within 96 h. The simplified process is shown in Fig. 1b. To exclude the potential effects of the residual Cd 2+ in the last passage, the culture was centrifuged and transformed into fresh BG11 medium during evolution. Serial adaptation passages were conducted until the final tolerance to CdSO 4 achieved 9.0 µM. After confirming that the strain can survive under 9.0-µM CdSO 4 , the evolved strain was screened on BG11 agar plate with 9.0-µM CdSO 4 , and four clones were isolated and re-cultured in BG11 liquid medium. After re-confirmation of the tolerance, one clone showing greatest growth state was selected for further analysis (Fig. 1a).

Whole-genome re-sequencing
Isolation of genomic DNA was carried out as described previously [53]. Total DNA obtained was subjected to quality control by agarose gel electrophoresis and quantified by Qubit. The genome of WT strain and the evolved strain ALE-9.0 were sequenced with MPS (massively parallel sequencing) Illumina technology by a paired-end library with an insert size of 350 bp. This 350-bp library was sequenced using an Illumina HiSeq4000 by PE150 strategy. Original figure data obtained by high-throughput sequencing were transformed into raw sequenced reads (raw data, or raw reads). Then, sequenced data were filtered and the sequence of adapter and low-quality data were removed, resulting in the clean data used for subsequent analysis. Variation information of the sample and the reference is obtained by aligning the sample reads with the designated reference (https ://www.ncbi. nlm.nih.gov/nucco re/NC_00091 1). Final results involved SNP (single-nucleotide polymorphism), InDel (insertion and deletion of small fragments in the genome), and SV (insertion, deletion, inversion, and translocation of the large segments in the genome level).

Sanger sequencing
To validate the SNP, InDel, and SV revealed by resequencing, Sanger sequencing was performed. Primers used to amplify gene fragments are listed as "primers for Sanger sequencing" in Additional file 2: Table S1. The gene fragments were then ligated to EZ-T ™ (GENSTAR, Beijing, China) by original TA cloning kit and the plasmids obtained were sent for sequencing.

Strains' construction
Strains and plasmids used in this study are listed in Table 1. Among them, E. coli DH5α was used for vector construction and amplification. For knockout of relative original and mutated genes, the plasmids' framework was obtained from above plasmid used for sequencing by PCR with primers for gene knockout (Additional file 2: Table S1), and then ligated with chloramphenicol-resistance cassette (amplified from a plasmid pACYC184). Then, the constructed plasmid was transformed into Synechocystis by natural transformation [54].
For genes overexpression, an integrative vector pCP3031 with spectinomycin-resistant cassette was used [26]. Relative genes were first obtained by PCR. Primers for gene overexpression are listed in Additional file 2: Table S1. Afterwards, target genes were ligated into pCP3031. The constructed plasmid was finally transformed into Synechocystis by natural transformation [54].