Cpf1 as a flexible tool for gene editing in different cyanobacteria species

We have previously demonstrated that Cas9 poses toxicity in cyanobacteria. We sought to circumvent the toxicity issue by employing an alternative RNA guided DNA nuclease, Fncpf1 from. We chose to utilize the FnCpf1 gene as this variant has been previously demonstrated to cleave DNA in bacterial systems. We first compared the toxicity fromto that of. Either promoterlessorwith a lac promoter were cloned into pVZ321, a replicating vector based on RSF1010. No other editing machinery or homologous repair template was included so that we could assess the toxicity of the two proteins alone. Presumably, Cas9 expression was greatly reduced relative to Cpf1 because it lacked a promoter whilehad a functional promoter. Since both genes are cloned in the same vector in the same orientation, any additional background expression should be the same in both cases and relative expression can be compared. We conjugated both constructs as well as an empty vector into2973. Only 3 colonies were obtained from the vector containing, while the vector containingyielded about half as many colonies as the empty vector (). Even with lower expression levels, a high degree of toxicity was observed fromwhileshowed far less toxicity. This suggested thatis a suitable nuclease for genome editing in cyanobacteria and therefore, we sought to develop abased editing system for these organisms. ^{15 16 16 21} Supplementary Figure 1 Francisella novicida cpf1 cas9 cas9 cpf1 cpf1 Synechococcus cas9 cpf1 cas9, cpf1 cpf1 cpf1

To facilitate rapid cloning of editing plasmids we constructed a vector, pSL2680 () based on the broad host range plasmid RSF1010. RSF1010 replicates well in most gram negative bacteria and would allow us to perform genome editing without integratinginto the chromosome. This strategy served to simplify the system and allowed us to make truly markerless mutations after the mutants are cured of the editing plasmid. The pSL2680 vector expressesfrom a lac promoter and an endogenousCRISPR array from a J23119 promoter (Biobrick #BBa_J23119). The native array had 3 repeat sequences with 27–30 nt spacer sequences separating them. The first spacer in the array is replaced withflanked by AarI sites that allowto be swapped for annealed oligos 24 nt in length; while later spacers would remain endogenous towith no target (). Using AarI, the annealed oligos are scarlessly placed between two repeats where they form the new targeting segment of a crRNA in a CRISPR array. Cpf1 will then process the CRISPR array transcript into mature crRNA which is used to target Cpf1 to a specific spot in the genome. Following the CRISPR array are unique KpnI and SalI sites to linearize the plasmids so that homologous repair templates can be inserted using Gibson assembly. We have made this base vector for genome editing available through Addgene (plasmid #85581). Supplementary Figure 2 Supplementary Fig. 3 cfp1 cpf1 Francisella novicida lacZ lacZ Francisella novicida

We performed our initial test of genome editing in2973, a cyanobacterial strain with an ability for rapid growth. We attempted all three modes of gene editing by creating a markerless point mutation, a knock-out mutation or a knock-in mutation. We elected to engineer an S264A point mutation in thegene that encodes the D1 protein of photosystem II. This specific mutation is interesting as it has been identified as the mutation that gives rise to resistance to the herbicide DCMU. As such, we would have an easy phenotype to screen for. For the knock-out we created a markerless deletion of. NblA functions to mediate degradation of the phycobilisome antenna complexes which can comprise up to 50% of the total cell protein in cyanobacteria. During conditions of nitrogen starvation, phycobilisomes are degraded to provide supplemental nitrogen for the cell. As such, this protein plays and important role in adaptation to changing environmental conditions. A knockout of thegene is easily identified due to its obvious phenotype. WT2973 bleaches under nitrogen deprivation due to the degradation of antenna complexes. When thegene is knocked out, the resulting strains remain green upon the removal of nitrate from BG11, the growth medium, because the antenna complexes are not degraded. For the knock-in, we inserted eYFP under the control of the trc promoter into NSI without the aid of antibiotic selection. Synechococcus psbA1 nblA nblA Synechococcus nblA ^{22 23 24}

Plasmids to generate the three edits were constructed by inserting annealed oligos that targetor NSI into pSL2680, using golden gate assembly. A homologous repair template was synthesized as left and right fragments (or left, right and middle fragments for the eYFP knock-in) with 1 kb of homology to the upstream and downstream sequences. When making a point mutation, the editing plasmid would also be a target for Cpf1 cleavage because the homologous repair template would also contain the target of the crRNA. To prevent cleavage of the editing plasmid, we made a second silent mutation in the homologous repair template to eliminate the PAM site. This mutated PAM site will be incorporated into the chromosome upon editing and prevent further cleavage of the chromosome. This strategy will serve to drive editing to completion and force segregation. The eYFP insertion was designed such that the it would split the sequence targeted by the crRNA to prevent targeting of the editing plasmid as well as cleavage of the chromosome after eYFP has been inserted. The demonstration of markerless knock-ins and knock-outs is of special importance as markerless mutations will allow the introduction or removal of an unlimited number genes while eliminating concerns of growing antibiotic resistant bacteria. psbA, nblA

The editing plasmids were conjugated into2973 and a few colonies resulting from each experiment were collected for further analysis. Sequencing of the point mutation revealed that only 2 of the 8 colonies contained the point mutation; however, when streaked on BG11 + 5 uM DCMU, 6 of the patches showed DCMU resistance (). Apparently, selective pressure from DCMU allowed us to rescue a small subpopulation of cells that had acquired the point mutation. This suggests that some cells in each patch had received the point mutation, but editing had not gone to completion in all cells. To allow the editing process to go to completion, we repatched the colonies onto BG11 Kmthree additional times. After the three additional patchings 6 of the 8 colonies had the point mutation as determined by Sanger sequencing. The knock-in of eYFP yielded similar results as initially 2 of the 10 colonies contained the insertion of eYFP as indicated by colony PCR (). After two additional patches on BG11Kmthe number of patches containing the insertion of eYFP had risen to 6 of 10 (). These data indicate that editing is driven towards completion after a few passages on selective media. While correct colonies could be easily identified from the initial conjugation, repeated passage on selective media significantly increased the proportion of properly edited colonies. The degree of segregation of the mutants was determined by examining the deletion ofas only a fully segregated mutant would show the nonbleaching phenotype. After three generations on BG11Kmplates, 9 of the 10deletion mutants showed the nonbleaching phenotype indicating that they were fully segregated (). Together, these findings suggest that while ~20% of the colonies are edited upon first appearance, maintaining them on selective media forces editing to go to completion and drive complete segregation. Synechococcus nblA nblA Fig. 1A Fig. 1C Fig. 1D Fig. 1B 10 10 10

Patches with the insertion of eYFP could not be assayed for eYFP expression because they contained the editing plasmid which also expressed eYFP from the repair region. To clarify the issue, we cured the edited strains of the plasmid. We also cured the editing plasmid from an edited colony from thedeletion and thepoint mutation at this time as well. Toward this goal, an edited patch of each mutant strain was grown in BG11Kmto an ODof 1.0. The culture was then diluted 1:2500 into BG11 without antibiotics and grown to an ODof 1.0 to allow spontaneous loss of the editing plasmid. To obtain single cells, the culture was serially diluted and plated on BG11. 50 single colonies of each were picked and patched on BG11Kmand also BG11 to identify colonies that had become sensitive to kanamycin and thus lost the editing plasmid. We found that the editing plasmid was lost in ~8% of the cells for thedeletion, 14% of the cells for the knock-in of eYFP and 36% of the cells for the point mutation in(). Once we rid the eYFP knock-in strain of the plasmid borne copy of eYFP, we verified that eYFP was properly expressed from its chromosomal location using fluorescence microscopy (). nblA psbA nblA psbA 10 10 720 720 Supplementary Table 1 Fig. 2

We examined the versatility of thesystem by applying the technology to editing of two additional genus of cyanobacteria. We chose6803 and7120 because they are model organisms for the study of photosynthesis and nitrogen fixation, respectively, and genome editing has not yet been described for these strains. We constructed derivatives of the pSL2680 editing plasmid to generate a deletion mutation, a point mutation and a knock-in mutation in each strain. cpf1 Synechocystis Anabaena

In6803 we chose to delete thegene as this would generate the same non-bleaching phenotype that is observed in2973.6803 has two adjacent copies of thegene as opposed to the single gene that exists in2973, requiring a much larger deletion to eliminate both copies. In7120 we chose to delete the first 400 bp of thegene which encodes nitrogenase reductase. Thegene is the first gene in the operon that encodes the structural genes for nitrogenase and a knock-out of this gene with conventional methods would disrupt expression of the entire operon. NifH is essential for nitrogen fixation and amutant would be incapable of diazotrophic growth. We ligated oligos targeting each gene into the pSL2680 vector and cloned a region containing the deletion oforwith 1 kb of upstream and downstream sequences into the KpnI site on the vectors containing the modified CRISPR array. The resulting plasmids were conjugated into either7120 or6803 and 16 colonies were collected from each conjugation. After three rounds of repatching onto BG11Nmor BG11Km, respectively, we performed PCR from upstream oforto a region on the chromosome that is not present on the editing plasmid to determine if the deletion had occurred in the chromosome. Synechocystis nblA Synechococcus Synechocystis nblA Synechococcus Anabaena nifH nifH nifH nblA nifH Anabaena Synechocystis nblA nifH ²⁵ 20 10

In6803 we found that 7 of the 16 colonies had a segregated deletion of(); while in7120 we found that 10 of the 16 colonies had obtained the deletion ofand appeared fully segregated via PCR (). The lack of more fully segregated colonies in6803 can be attributed to the extreme level of ploidy in this organism. Ploidy is highly variable in6803 with one study demonstrating ~200 copieswhile another shows that copy number is as high as 50 in liquid culture but varies depending on growth conditions. In contrast,2973 and7120 maintain less than 10 copies of their chromosome. The high degree of ploidy likely impaired full segregation in 3 patchings in6803. To further verify segregation, we grew two colonies that contained the deletion oforand subjected them to nitrogen deprivation. Both colonies of6803 exhibited the expected non-bleaching phenotype while the WT control bleached upon removal of nitrate, indicating they were fully segregated for thedeletion (). Two7120 mutants were spotted onto BG11 -N alongside wildtype. Both edited colonies yellowed while the wildtype7120 spot grew darker and greened indicating the mutants wereand thus contained a segregated deletion of(). Synechocystis nblA1/2 Anabaena nifH Synechocystis Synechocystis Synechococcus Anabaena Synechocystis nblA nifH Synechocystis nblA Anabaena Anabaena nif- nifH Fig. 3B Fig. 3A ^{26 27 26} Fig. 3D Fig. 3C

To engineer a point mutation in7120, we made a H197G active site mutation inthat has been previously shown to abolish nitrogenase activity. Thepoint mutation would also result in a- phenotype when segregated. Such a mutation would serve as a good example of the utility of CRISPR technology for studying protein active sites. In parallel, in6803, we changed the GTG start codon of thegene to GCG to abolish translation initiation of that gene. IsiA is a light harvesting protein that associates with photosystem I and is expressed under conditions of iron limitation. IsiA also forms empty rings that are not associated with PSI where they play a protective role in energy dissipation. When expressed, IsiA generates a 77 K fluorescence emission peak at 685 nm when excited at 430 nm; thus anknockout is easily identified by the absence of this characteristic peak. Anabaena nifD nifD nif Synechocystis isiA isiA ^{28 29}

Editing plasmids were constructed for each point mutation in the same fashion as was done for the 2973 editing plasmids. Homologous repair templates containing the desired point mutation were included on the editing plasmid. Conjugation of the editing plasmids into the respective strains yielded hundreds of colonies, of which 16 were collected. After 3 rounds of repatching, theorgenes of 8 colonies were sequenced to verify the presence of the point mutation. Editing was more efficient in7120 and6803 compared to2973 as 7 of the 8colonies had the mutation while 6 of the 8colonies had the point mutation as determined by Sanger sequencing. Two colonies edited for theH197A mutation were examined for a- phenotype to verify segregation. Both colonies yellowed when spotted onto BG11 -N while the WT strain greened indicating that both were fully segregated for the point mutation (). 77 K fluorescence emission scans of colonies edited for thestart codon mutation were examined for the absence of the 685 nm peak to verify segregation of thestart codon mutation. All colonies did not exhibit the characteristic peak at 685 nm under conditions of iron starvation indicating that they were fully segregated for the point mutation (). nifD isiA Anabaena Synechocystis Synechococcus Synechocystis Anabaena nifD nif isiA isiA Fig. 4A Fig. 4B

For the knock-in mutations, we performed a direct gene replacement of eitherorin6803 and7120, respectively, with a promoterless eYFP. The gene replacement occurs precisely from start codon to stop codon. These knock-ins would result in a markerless direct gene replacement and place eYFP under the control of the respective promoter. This presents a new capability that cannot be done in a single step with current technology. In fact, such a replacement takes months to perform with existing techniques but can be completed in 2–3 weeks with CRISPR. To engineer the knock-ins, we employed the same crRNA that was used to make the previously discussed deletions ofor. The homologous repair template for the deletion was substituted for one in which the coding region of eitheroris replaced with that of eYFP. Conjugation of the editing plasmids into their respective strains resulted in strains in which eYFP was properly inserted intooroperons as verified by PCR between eYFP and a chromosomal region lying outside the homologous repair template (). Sanger sequencing was also used to verify that the eYFP insertions were properly aligned with respect to theorstart and stop codons. Six of the 87120 colonies had the insertion while 7 of the 8 of the6803 colonies had the eYFP insertion, after 3 rounds of patching (). nblA nifH Synechocystis Anabaena nifH nblA nifH nblA nifH nblA nifH nblA Anabaena Synechocystis Fig. 5A,B Fig. 5A,B

We cured the eYFP knock-ins of their editing plasmids in the same way that is described for2973 except that in7120 an additional sonication step was used before plating to break up the filaments so that colonies resulting from single cells could be obtained. A cured colony of each strain was examined for eYFP expression under nitrate replete and nitrate deplete conditions as both genes replaced (orare only expressed upon nitrogen deprivation. In6803, eYFP expression was observed under nitrate deplete but not nitrate replete conditions () indicating that eYFP had taken on the expression pattern of. In7120, eYFP was expressed specifically in heterocysts and only upon removal of nitrate (). Swapping coding regions in such a way allows one to replace a gene with one with altered activity while maintaining proper regulation. Such gene replacements will be a valuable tool for metabolic engineering as one can directly replace any gene with another to modify metabolic pathways. Synechococcus Anabaena nblA nifH) Synechococcus nblA Anabaena Fig. 5C Fig. 5D

2973 was grown on BG11agar plates and in BG11 liquid at 38 °C with 150 μE • m• slight.6803 and7120 were grown on BG11 agar plates and in BG11 liquid at 30 °C under 50 μE • m• slight. Exconjugates of the respective strains were grown under the same conditions on BG11 agar plates supplemented with 10 μg/mL kanamycin for6803 and2973 or 20 μg/mL neomycin for7120. Cloning was performed incoli XL1-blue strain on L-agar with 50 μg/mL kanamycin. Conjugation was performed with HB101 containing either pRL443 or pRL623with the editing plasmid on 82mm HAF Milipore filters overlayed on BG11 agar supplemented with 5% Luria Broth. Conjugation of editing plasmids into all strains was performed by mixing 100 μL of overnight cultures of HB101 pRL443 and HB101 pRL623 + editing plasmid with 200 μL of the cyanobacteria strain adjusted to an OD0.8. After 24 h incubation at 38 °C under 150 μE • m• slight,2973 conjugation filters were transferred onto BG11 supplemented with 50 μg/mL kanamycin. After 48 h incubation at 30 °C under 50 μE • m• slight,6803 and7120 conjugation filters were transferred onto BG11 supplemented with either 50 μg/mL kanamycin or 40 μg/mL neomycin. Colonies appeared within 3 days for2973 or 8 days for6803 or7120 Synechococcus Synechocystis Anabaena Synechocystis Synechococcus Anabaena E. Synechococcus Synechocystis Anabaena Synechococcus Synechocystis Anabaena ^{36 37} −2 −1 −2 −1 −2 −1 −2 −1 720

The plasmid containingand the nativeCRISPR array, pY002 (pFnCpf1_min), was obtained as a kind gift from Feng Zhang (Addgene plasmid # 69975). Thegene was amplified from pY002 with the cpf1 lac-L/cpf1-R primers () which also fuse a lac promoter onto. The resulting fragment was cloned into the ApaLI/EcoRI sites on pVZ321to replace the Cmcassette to generate pSL2668. Next, overlap extension PCR was used to introduce a pair of AarI sites into the first spacer in the CRISPR array of pY002 by amplifying the left and right halves using the J23119ecoL/directrepeat aarI-2 or directrepeat aarI-1/directrepeat-R primers followed by amplification using the J23119ecoL/directrepeat-R primers. The resulting PCR fragment was cloned into the EcoRI/SalI sites on pSL2668 to generate pSL2683.was amplified from the pCrispomyces-2plasmid using the lacZaarI-L/lacZaarI-R primers. The resulting fragment was then cloned into the AarI sites on pSL2683 to generate pSL2680, which served as the base plasmid for construction of editing plasmids expressing a full length pre-crRNA. Editing plasmids were constructed by cloning annealed oligos into the AarI sites on pSL2680. The following annealed oligos were ligated into the AarI sites on pSL2680: 7942nblAKOgRNAL/7942nblAKOgRNAR to yield pSL2682; 7942s264agRNAL/7942s264agRNAR to yield pSL2723; NS1gRNAL/NS1gRNAR to yield pSL2724; 6803nblAKOgRNAL/6803nblAKOgRNAR to yield pSL2726; 7120nifHgRNAL/7120nifHgRNAR to yield pSL2728; 7120nifDgRNAL/7120nifDgRNAR to yield pSL2833; and 6803isiAgRNAL/6803gRNAR to yield pSL2834. Next, PCR was used to synthesize the homology regions which were then cloned into the KpnI site on the plasmids containing the matching crRNA. The7942homology region containing the deletion ofwas synthesized from pSL2470using nblAdelRkpnI/nblAdelLkpnI and cloned into the KpnI sites on pSL2682 and pSL2684 to yield pSL2691 and pSL2689 respectively. The homology region containing the7942S264A mutation was synthesized using fusion PCR with the 7942psbAL1/7942psbAR2 and 7942psbAL2/7942psbAR1 primers followed by PCR with the 7942psbAL1/7942psbAR1 primers. The resulting PCR fragment was cloned into pSL2723 to yield pSL2796. The homology region targeting eYFP to NS1 was synthesized using fusion PCR with the pAM1303NS1L1/pAM1303NS1R2, pAM1303NS1L2/pAM1303NS1R3 and pAM1303NS1L3/pAM1303NS1R1 primers followed by PCR with the pAM1303NS1L1/ pAM1303NS1R1 primers. The resulting PCR fragment was cloned into pSL2724 to yield pSL2801. The6803homology region containing the deletion ofwas synthesized using fusion PCR with the 6803nblAdelL1/6803nblAdelR2 and 6803nblAdelL2/6803nblAdel R1 primers followed by PCR with the 6803nblAdelL1/6803nblAdelR1primers. The resulting PCR fragment was cloned into pSL2726 to yield pSL2773. The7120homology region containing the deletion ofwas synthesized using fusion PCR with the 7120nifHL1a/7120nifH R2 and 7120nifHL2/7120nifHR1 primers followed by PCR with the 7120nifHL1a/7120nifHR1 primers. The resulting PCR fragment was cloned into pSL2728 to yield pSL2739. The7120point mutation homology region was constructed in two pieced using nifDL/nifDMR or nifDML/nifDR primers. The homology template was then assembled into pSL2833 linearized with kpnI using Gibson assembly to generate pSL2839. The6803point mutation homology region was constructed in two pieces using 6803isiAL/6803isiAMR or 6803isiAML/6803isiAR primers. The homology template was then assembled into pSL2834 linearized with KpnI using Gibson assembly to generate pSL2834. The homologous repair template to insert eYFP intoof7120 was synthesized as three fragments using the primers 7120eYfplgibs/7120eYFPR1 or 7120eYFPL1/7120eYFPR2 or 7120eYFPL2/7120eYFPRgibs. The three fragments were assembled into pSL2728 linearized with KpnI using Gibson assembly to generate pSL2840. The homologous repair template to insert eYFP intoof 6803 was synthesized as three fragments using the primers 6803eYfplgibs/6803eYFPR1 or 6803eYFPL1/6803eYFPR2 or 6803eYFPL2/6803eYFPRgibs. The three fragments were assembled into pSL2726 linearized with KpnI using Gibson assembly to generate pSL2841. cpf1 Francisella novicida cpf1 cpf1 LacZ Synechococcus nblA nblA Synechococcus psbA Synechocystis nblA nblA1A2 Anabaena nifH nifH Anabaena nifD Synechocystis isiA nifH Anabaena nblA ¹⁶ Supplementary Table 2 ^{38 39 15} R

2973 and6803 were inoculated into 50 mL of BG11 and grown in a MC-1000 multicultivator (Photon Systems Industries) bubbled with 3% COat 38 °C for2973 or 30 °C for6803 until late linear growth; 16 hours for2973 and 48 hours for6803. Cultures were then washed 3X with 30 mL BG11 -N and used to start fresh cultures in the multicultivator in BG11 -N to an OD720 of 0.75. After 16 hours of additional growth for2973 or 48 hours for6803, aliquots from each culture were transferred to multi well plates for analysis. Synechococcus Synechocystis Synechococcus Synechocystis Synechococcus Synechocystis Synechococcus Synechocystis 2

Wild type2973 or DCMU resistant mutants were patched onto BG11 supplemented with 5 μM DCMU and grown for 72 hours at 38 °C under 150 μE • m• slight. Synechococcus −2 −1

Cultures were grown to an ODof 0.5 at which time they were washed 1X with BG11 -N and 50 μl was spotted onto BG11 -N agar plates and grown for 72 hours at 30 °C with 50 μE • m• slight. 720 −2 −1

WT and eYFP containing mutant strains were concentrated 10-fold from mid log phase cultures. Samples were deposited onto glass slides that were coated with 2% polyethyleneimine. Cells were imaged using a Nikon Eclipse 80i microscope equipped with a Photometrics Cool Snap ES CCD camera (Roper Scientific). Filter sets (Chroma) were as follows: YFP was detected using a 480/30 nm excitation filter, a 505 nm dichroic beam splitter, and a 535/40 nm emission filter. Chlorophyll fluorescence was detected using a 560/40 nm excitation filter, a 595 nm dichroic beam splitter, and a 630/60 nm emission filter. A 100 ms exposure time was used for imaging chlorophyll fluorescence and a 1 s exposure time was used to image eYFP expression.

Log phase6803 cells were washed 3X with BG11 -iron and used to start 50 mL cultures in BG11 with or without iron. After 2 days of growth at 30 °C, 50 μE • m• slight, whole cell fluorescence was observed. Fluorescence emission spectra at 77 K were recorded on a Fluoromax-2 fluorometer (JobinYvon, Longjumeau, France) with excitation at 435 nm. Fluorescence emission curves were normalized as F/F. Synechocystis −2 −1 720