Inducible CRISPR/Cas9 Allows for Multiplexed and Rapidly Segregated Single-Target Genome Editing in Synechocystis Sp. PCC 6803

Establishing various synthetic biology tools is crucial for the development of cyanobacteria for biotechnology use, especially tools that allow for precise and markerless genome editing in a time-efficient manner. Here, we describe a riboswitch-inducible CRISPR/Cas9 system, contained on a single replicative vector, for the model cyanobacterium Synechocystis sp. PCC 6803. A theophylline-responsive riboswitch allowed tight control of Cas9 expression, which enabled reliable transformation of the CRISPR/Cas9 vector intoSynechocystis. Induction of the CRISPR/Cas9 mediated various types of genomic edits, specifically deletions and insertions of varying size. The editing efficiency varied depending on the target and intended edit; smaller edits performed better, reaching, e.g., 100% for insertion of a FLAG-tag onto rbcL. Importantly, the single-vector CRISPR/Cas9 system mediated multiplexed editing of up to three targets in parallel inSynechocystis. All single-target and several double-target mutants were also fully segregated after the first round of induction. Lastly, a vector curing system based on the nickel-inducible expression of the toxic mazF (from Escherichia coli) was added to the CRISPR/Cas9 vector. This inducible system allowed for curing of the vector in 25–75% of screened colonies, enabling edited mutants to become markerless.


Figures
. Evaluating the toxicity of theophylline towards S6803. (A) Single cultures were grown in BG11 supplemented with various concentrations of theophylline, or the respective amount of DMSO-carrier only. A control with neither was included. (B) Spot assay on BG11-plates supplemented with the indicated concentrations of theophylline or DMSO-carrier only. A control with neither was included. A wt S6803 culture was diluted to OD730 0.2 and used to prepare a 10x dilution series that was plated (4 µl spots). Shown is representative data from two separate experiments. Figure S2. PconII and Ptrc in combination with riboswitches B, C, and E* were compared using a Gfp-reporter. Three non-toxic theophylline concentrations were tested: 0.1, 0.25, and 0.5 mM. Samples with no added inducer were supplemented with the respective amount of DMSO-carrier: 0.05%, 0.125%, and 0.25%. Fluorescence was measured 3 days after induction by flow cytometry (Beckman Coulter CytoFLEX, FITC-channel: emission 525 nm, excitation 488 nm). 10,000 events were acquired; data analysis was done using FlowJo (FlowJo LLC). The left y-axis shows the average median fluorescence intensity (filled or empty circle); the right y-axis shows the induction ratio (induced signal divided by un-induced signal, both values were first normalized against the wild type signal). The numbers above the bars are the calculated average induction ratios. All data is presented as averages ± SD from biological duplicates. Non-visible error-bars are smaller than the data symbol.         A green line below a lane signals a fully segregated multi-edit in that screened colony. Fractions indicated the number of fully edited colonies out of the total number screened. A wt control shows how an unedited colony will appear. An "R" below a lane indicates a not fully segregated mutant that was re-streaked for a second round of induction. Figure S12. Results from comparing the editing ability of construct [B] when supplemented with PnrsB-mazF-LVA or without a curing system (control). Editing of yfp (∆20 bp), NS1 (∆30 bp), and multiplexed editing of NS1+NS4 (both ∆30 bp) were all tested. (A) S6803 transformation results after plating on selective plates, without or with 0.25 mM theophylline. (B) Induction spot assay results. 5x dilution series were plated on plates with or without 0.25 mM theophylline. Done for biological triplicates, representative data is shown. (C) Editing results for single targets yfp (∆20 bp) and NS1 (∆30 bp). (D) Editing results for multiplexed NS1+NS4 (both ∆30 bp). (C-D) A green line below a lane signals a fully edited mutant. Colonies that appear segregated but have not been marked as such, are due to them having detectable wt-bands when the gels are more closely inspected. A control ("C") shows how an unedited colony will appear. Fractions indicated the number of fully edited colonies out of the total number screened. Screening three colonies each from the transformation plates seen in (A), i.e. ones without theophylline inducer, to assess leaky editing. A green line below a lane signals a fully edited (∆458 bp) mutant. A control ("C") shows how an unedited colony will appear.  Table S2. Primers used in this study.   Benchling. 3 This score is determined by using an algorithm based on mammalian cell data, it also doesn't consider the position of the spacer in the target gene. 4 However, note that this scoring might not be directly applicable for a prokaryotic host. 5 b An "A" was added at the TSS-position of this sgRNA to accommodate for the potential need for this by PBBa_J23117; this base is thus not present in the target region. Table S4. Off-target analysis of sgRNAs used in this study, against the S6803 genome, using the CasOT software. 6 The PAM was specified to NGG only. Indicates the type of mismatch. The first number (0-2) specifies the number of mismatches in the seed region (12 nt proximal to PAM) of the binding sequence. The following number indicates the mismatches in the remaining non-seed region. Table S5. Raw data for the calculated CFUs for selected CRISPR/Cas9 editing experiments. The CFUs were calculated from spot-plates (unless otherwise noted), from spots with dispersed enough colonies. The dilution featured in those spots were then used to back-calculate to get the "un-diluted" CFUs, which was used to calculate the percentage of total plated and induced CFUs that survived (i.e. retained a healthy phenotype) and became edited. a All colonies from plates without theophylline were calculated from the 6 th spot in the dilution series (representing a 5 5 dilution). b The spot on 0.25 mM theophylline spot plates which the number of colonies were counted for is indicated in brackets. (The dilution in these spots is 5 n-1 , n being the spot number). c The two stacked values per cell is the data for technical duplicates. d This row denotes data for inductions performed using 0.5 mM theophylline. e The 0.25 mM theophylline data was not calculated from spot plates due to a lack of colonies. It was instead calculated from the full-sized induction plates. 40 µl was plated on full-sized plates, i.e. 10x more than the 4 µl for the spots on the spot plates; this was accounted for in the performed calculations. The dilution that was plated and counted is still indicated in brackets.  Topp et al. 8 , aptamer in bold, start codon underlined) TGATAAGATAGGGGTGATACCAGCATCGTCTTGATGCCCTTGGCAGCACC AAGGGACAACAAGATG Riboswitch E*: (as in Topp et al. 8 , aptamer in bold, start codon underlined) GGTGATACCAGCATCGTCTTGATGCCCTTGGCAGCACCCTGCTAAGGAGG CAACAAGATG P nrsB : (native to S6803, truncated at 5'-end to remove -10-box and TSS mapped for nrsR expression going in the opposite direction, the two mapped TSS for nrsBACD expression are shown in bold). 9 GTCTGATCTTAGCGGGGGAAGGAGATTTTCACCTGAATTTCATACCCCCTTTG GCAGACTGGGAAAATCTTGGACAAATTCCCAATTTGAGGTGGT P nrsD : (native to S6803, two mapped TSS in bold). 10 TATTCGATTCAGTACCAAGTACTATTGCGGGGACAGGACGTTTCTCAAGGCCC TCATCAATATCCCCCCTGGGGGCATAGAATAGAGATCAATTTTCTACCCCAA ACCCCCACA

Supporting methods -Using the inducible CRISPR/Cas9-system
For an overview of the CRISPR/Cas9 target vector construction and subsequent workflow in S6803, see Figure 1 in the main text.

Theophylline inducer stock: preparation and use
A 200 mM theophylline stock, dissolved in 100% DMSO, is used. This does not dissolve at RT and requires heating before use. Theophylline is very stable 11 and repeated heating was not found to reduce the stocks performance. To dissolve it, heat in a water bath at 42-50°C, mix a few times by vortexing. When fully dissolved, use directly to make inducersupplemented BG11-plates (a final concentration of 0.25 mM is recommended. Store at 4°C between uses.

Available pPMQAK1-CRISPR/Cas9 base vectors
The best performing pPMQAK1-CRISPR/Cas9 base vectors described in this study will be submitted to Addgene. These are the constructs in order of expression strength:

Designing and constructing sgRNAs
Identification of suitable protospacers in the target area can be done with e.g. Benchling. 3 Useful guidelines are to select spacers that target, if possible, the template strand, 12 that guide Cas9 to cut as close to the edit site as possible, 13 that don't have significant offtarget binding, and that don't have extreme GC-content (>75%, <25%). In this study, spacers were also chosen to have an A at the TSS (often by adjusting spacer length by a few nts), however it is unknown if this is strictly necessary for PBBa_J23117.
To make the sgRNA piece compatible for BsaI-based Golden Gate cloning, the following overhangs are required on the 5'-and 3'-ends of the above sgRNA. BsaI-site shown in bold, created overhangs upon digestion are underlined.
For a multiple-target construct When multiplexing, an sgRNA-array needs to be constructed and this sgRNA-array must have the same BsaI-containing overhangs as described above. In this study this was done using the method described by Li et al. 2 The following template vectors built in this study were used for this purpose: pMD19-BsaI-PBBa_J23117-sgRNA-BsaI-Sp r and pMD19-Cas9_handle-S.pyogenes_terminator-PBBa_J23117-Cm r .

Designing and constructing Donor DNAs
In this study, homology arms of roughly 350 bp were used on either side of the intended edit. However, for e.g. large insertions it could be beneficial to use longer homology arms. 2 Donor DNAs are preferably constructed by overlap-PCR to join the two homology arms together to one piece. To create the desired edit, the primers used to amplify the homology regions are designed accordingly. The donor DNA must also be designed to mutate or remove the PAM and preferably also parts of the proximal seed-sequence of the protospacer. See the study by Jiang et al. for advice on which mutations are effective. 14 For larger insertions the donor DNA segments can be added as separate pieces to the Golden Gate assembly.
To make the donor DNA piece compatible for BsaI-based Golden Gate cloning, the following overhangs are required on the outermost 5'-and 3'-ends. BsaI-site shown in bold, created overhangs upon digestion are underlined.
For a multiple-target construct For multiplexed targeting the above overhangs must be used for the forward primer of the first donor DNA, and the reverse primer for the last donor DNA, respectively. The rest of the primers must be designed (e.g. with Benchling) 3 to have BsaI-overhangs that allow for assembly of the different donor-DNAs in the desired order.