A human-codon optimized Cas9 open reading frame (ORF) containing three nucleus-localization signals (NLS) was produced by the Feng Zhang laboratory (Broad Institute, Cambridge, MA, United States) and purchased from addgene.com [pSpCas9(BB)-2A-Puro (PX459) V2.0, ID: 62988, Ran et al., 2013]. The Cas9 ORF was cloned into a pUC19 backbone using GA cloning (Gibson et al., 2009) using primers Cas9-2A-shble_1 and P2A_1 (Supplementary Material). For constitutive expression of Cas9, the FcpB/FcpA (Fucoxanthin chlorophyll a/c promoter B/Fucoxanthin chlorophyll a/c terminator A) (Bailleul et al., 2010) promoter/terminator pair were chosen (FcpB-Cas9).

The HDR-donor plasmid used in Weyman et al. (2014) for TALEN mutagenesis was adapted to a HDR-donor plasmid for targeting nitrate reductase (pKO-NR). pKO-NR contained 1 kb homology arms flanking an expression cassette coding for the sh ble gene that conferred resistance to the antibiotic phleomycin. The cassette inserted both a sh ble cassette and a STOP codon upstream of the sh ble promoter (Fucoxanthin chlorophyll a/c binding promoter F, FcpF). This was in frame with the targeted coding gene. The primers and plasmid sequence information can be found in Weyman et al. (2014).

The sRNAi promoter (chr2: 28,038-29,083) and terminator (chr2: 29,124-29,424) was amplified from Phaeodactylum genomic DNA (promoter primers = sRNAi1_1 and sRNAi1_2; terminator primers = sRNAi_3 and sRNAi_4). For Golden Gate cloning, the sRNAi terminator was shortened by PCR to 112 bp to eliminate an internal BsaI restriction site. The sRNAi promoter and terminator was cloned from Phaeodactylum genomic DNA at the locus of the most highly expressed small, non-coding, intergenic RNA called sRNAi (Rogato et al., 2014; Bielinski et al., 2017). The sequences for the sRNAi promoter and terminator can be found in the Supplementary Material.

The sgRNA backbone, which consists of the genetic fusion of the crRNA and tracrRNA (Jinek et al., 2012), was amplified from the same Addgene plasmid that contained Cas9 (primers = sRNAgU1_F and sRNAgU1_R). An sgRNA expression cassette for Phaeodactylum (sRNAi_gBbsI) was constructed using a 4-piece GA approach into an XbaI-digested pUC19 plasmid containing a bacterial expression cassette for the bla gene conferring resistance to ampicillin/carbicillin. The resulting plasmid was then PCR amplified at the 3′ end of the sRNAi and the 5′ end of the sgRNA backbone (primers = gRNA_1 and gRNA_2). This amplicon served as a linearized vector for a 2-piece GA procedure for a spacer insert that contains two flanking BbsI restriction sites (Supplementary Figure S2). The 100 bp GA insert was constructed by ligating two 60 bp oligos with homology to the sRNAi promoter and sgRNA backbone. The resulting plasmid, sRNAi_g2XBbsI, was used for BbsI-based GG cloning of CRISPR spacers (Supplementary Figure S2B).

For all Cas9 target genes, spacer sequences were obtained for nitrate reductase (Phatr3_J54983), glutamine synthetase II (GS-2) (Gene ID: Phatr3_J51092), and chloroplastic glutamate synthase (cGOGAT) (Gene ID: Phatr3_J24739) using the Phyto CRISP-Ex online package (Rastogi et al., 2016) that passed the built-in criteria of minimizing off-target Cas9 activity.

A GG cloning protocol was modified from the sgRNA expression cassette supplied by Feng Zhang laboratory. In place of the BbsI-containing spacer sequence within the sgRNA expression plasmid, a LacZ bacterial expression cassette with flanking BbsI sites was cloned (primers = LacZ_GG_1 and LacZ_GG_2) into the sRNAi expression cassette using GA. The Pt sgRNA expression plasmid previously produced (sRNAi_gBbsI) was PCR amplified (primers = LacZ_GG_3 and LacZ_GG_4) and the amplicon was gel extracted and used as a GA vector. The sgRNA cloning vector (sRNAi_gLacZ) was the assembled plasmid. This plasmid was used in all future sgRNA expression cassette cloning and assembly experiments as a cloning vector (Supplementary Figure S1B). An appropriate GG insert for sRNAi_gLacZ was assembled by ligating two 56-nt oligos together that are complimentary to each other. The oligo primer sequences for all sgRNA target sequences can be found in Supplementary Material.

Detailed protocols for spacer cloning into the sRNAi_gLacZ plasmid (name GG1 cloning) are available at protocols.io (doi: 10.17504/protocols.io.4abgsan).

The Phaeodactylum episomal plasmid pPtPBR1 was obtained from R. Diner (Diner et al., 2016). First, one BsaI restriction site within the bla coding region of pPtPBR1 was removed by PCR amplification (primers = pBR322-Amp_1 and pBR322-Amp_2) and reassembly to produce plasmid pPtPBR1 (BsaI-). pPtPBR1 was then used as a template for GA cloning of a Cas9 ORF and a GG cloning BsaI acceptor site for sgRNA expression cassette cloning (Supplementary Figure S2). Cas9 was PCR amplified twice using one forward primer (Cas9-2A-sh ble_1) and two reverse primers (P2A_1 and P2A_2) so to add GA homology to the FcpF promoter (5′ of Cas9) and the sh ble ORF (3′ of Cas9). The double PCR reaction added a GSG-2A peptide DNA coding sequence to transcriptionally fuse hCas9 and sh ble (Supplementary Figure S2, GSG-P2A nucleotide sequence is in Supplementary Material). Next, a red-fluorescent protein bacterial expression cassette (mRFP) was obtained and PCR amplified (primers = Cas9-2A_sh ble-RFP_1 and 2) to produce GG acceptor sites. The remaining two GA pieces were PCR amplified from pPtPBR1 (BsaI-) to insert two GG cloning BsaI sites between the bla and tet expression cassettes and flanking mRFP, one using primers Cas9-2A-sh ble_2 and 3 (5′ – sh ble – Pt_Centromere – pMB1 – bla – 3′) and two using primer Cas9-2A-sh ble_6 and 7 (5′ – tet – OriT – FcpF – 3′). After cloning and E. coli transformation, the correctly assembled episome, pBR_Cas9-2A-sh ble, 2XBsaI-mRFP(AT) (Supplementary Figure S3A), was sequence verified.

Next, primers were designed to PCR amplify up to six sgRNA expression cassettes (cloning and construction described above) and to add BsaI-restriction and cleavage sites to the 5′ and 3′ ends of the sRNAi promoter/terminator. GG cloning is an optimal assembly platform when cloning more than one sgRNA expression cassette. For inserting one sRNAi_sgRNA cassette, primers GG-gRNA1-F and sgRNA(GG)-2 were used. For two sgRNAs, primers GG-gRNA1-F and R were used for sgRNA #1 and primers GG-gRNA3-F and sgRNA(GG)-2 were used for sgRNA #2. Between the two sgRNAs was cloned a LacZ expression cassette, which, when induced with IPTG activates beta-galactidose that digests Xgal to produce blue E. coli colonies (Supplementary Figure S3C). The sgRNA expression cassette(s) and LacZ amplicons was mixed at double the molar concentration of the BsaI-digested pBR_Cas9-2A-sh ble, 2XBsaI-mRFP(AT) vector and carried through the GG cloning reaction with BsaI-HF and a T4-ligase (Supplementary Figure S1C). The final sgRNA-LacZ-sgRNA episomal plasmid map can be seen in Supplementary Figure S3B.

Detailed protocols for sgRNA cloning into the Cas9-2A-sh ble epsiome (named GG2 cloning) is available at protocols.io (doi: 10.17504/protocols.io.4acgsaw).

Bacterial-mediated conjugation was used to introduce CRISPR components to Phaeodactylum. Diner et al., 2016 provided detailed methodology for introducing episomal plasmids to Phaeodactylum in a high-throughput manner. For all transformations, 100 uL of dense Phaeodactylum (1e8 cells/mL) was plated on conjugation-based solid agar medium (NH4/NO3-ASW, 1% agar, 5% LB in NH₄-supplemented ASW) in 6-well cell culture plates and incubated for 4 days under light (18:6) and at 18°C. Prior to transformation, pBR_Cas9-2A-sh ble, sRNAi-sgRNA episomes were transformed into recipient E. coli harboring the conjugation plasmid pTA-MOB (Karas et al., 2015; Diner et al., 2016) and selected on agar plates for both the cargo and conjugation plasmids (Figure 1). For a transformation control (positive for colonies not Cas9 cleavage), the episome pBR_Cas9-2A-sh ble, 2XBsaI_gRFP1 was used where Cas9 expression would still occur without an accompanying sgRNA. For a negative control, a similar episomal plasmid was built harboring the NAT gene that confers resistance to nourseothricin and not phleomycin. Under phleomycin selection, the transformed cells would die. Transformation efficiency for the delivery of Cas9-2A-sh ble:sgRNA episomes was 500–750 transformants per 1 × 10⁸ phleomycin-resistant cells.

The day before transformation, 4 random colonies from the selection plates were used to inoculate 3 mL of LB containing selection (Amp, Tet, Gent). After growing overnight, 200 uL of culture was used to inoculate 12.5 mL of LB. The culture was grown for 3–4 h, until an OD-600 nm of 0.8–1 was reached, pelleted, and suspended in 100uL of SOC media. Then, the E. coli was pipetted into a corresponding well containing Phaeodactylum and mixed either by a spreading loop or by rotating the 6-well plate such that the E. coli covered the entire lawn of Phaeodactylum. After E. coli was added to all wells and allowed to dry under a vacuum PCR hood, the 6-well plate was placed in the dark at 30°C for 90 min. The plate was then placed back at 18°C and exposed to light for 48 h.

Finally, each well was scraped using a sterile cell spreader, collected in a 2 mL Epitube then re-plated on 100 mm selection plates containing the appropriate selective media. The selection plates were then allowed to grow for 10–14 days or until colonies appeared. For the T7 Endonuclease I Assay reaction, half of the scraped Phaeodactylum cell lines (∼400 uL) was resuspended in 200 mL NH₄-supplemented ASW media supplemented with phleomycin and chloramphenicol. The cells were then grown under light and at 18°C for 4 days, passed into 200 mL fresh media, grown for an additional 3 days, then finally pelleted and flash frozen.

Detailed protocols for bacterial-conjugation in Phaeodactylum are publicly available at protocols.io (doi: 10.17504/protocols.io.5pvg5n6).

Micro-particle bombardment genetic transformation (Falciatore et al., 1999) was used to introduce CRISPR components to Phaeodactylum. First, Phaeodactylum was pelleted during exponential growth in liquid culture (NO₃-supplemented ASW), suspended, and plated onto agar growth plates at a cell concentration of 3e8 cells/mL (400 uL). Next, 24 ug of DNA (8 ug each plasmid; FcpB-Cas9, sRNAi-gNR-B, pKO-NR) were hybridized to tungsten beads and introduced to the plated Phaeodactylum at high velocity using the PDS-1000/He Biolistic^® Particle Delivery System (Bio-Rad, Hercules, CA). The transformed Phaeodactylum plates were then allowed to recover for 48 h at 18°C and in constant darkness. After recovery, Phaeodactylum was re-plated on agar plates containing NH₄-supplemented ASW with phleomycin to select for the presence of the pKO-NR plasmid. The selection plates were then grown under normal conditions for 21–28 day, or until Phaeodactylum colonies formed and were visible. Experimental controls were also performed where either FcpB-Cas9 and pKO-NR were co-transformed or pKO-NR was transformed alone.

To quantify the mutagenesis efficiency for both sgRNAs population-wide, the T7 endonuclease assay was used (Figure 2). While the T7 assay has been typically used to genotype a clonal cell line, this method was useful to quantify the mutagenesis efficiency of all cells after transformation and phleomycin-selection of Cas9-2A-sh ble. This was done with the intention to identify sgRNAs that may produce low rates of mutagenesis and subsequently discard them. The sgRNAs were designed to target the NR locus 55-bp apart (Figure 2A). This allowed both targeted populations (one for gNR-A and one for gNR-B) to be subjected to the same T7 assay parameters (Figure 2B). The T7 enzyme cleaves mismatched nucleotides and should cut at Cas9 target loci with NHEJ-mediated mutations. The expected band sizes after T7 digestion was 127-bp and 58-bp for gNR-A and 113-bp and 72-bp for gNR-B (Figure 2C). Homoduplexes, annealed bands that perfectly match during “heteroduplex” formation (Figure 2B), was expected to be observed in the T7 migration gel because biallelic knockouts that arise from MMEJ-mediated repair may, in theory, anneal to each other and avoid T7 cleavage. It was observed that gNR-A insufficiently targeted the NR locus (0% mutagenesis efficiency) while gNR-B mutated the NR locus at 38% (Figure 2D).

By employing this method prior to colony genotyping, sgRNAs with low rates of mutagenesis throughout the population can be ignored which saves on reagent costs and labor. Henceforth, gNR-A was discarded from future targeting experiments due to its low mutagenesis efficiency compared to gNR-B.

TIDE (Tracking Indels by DEconvolution) sequence analysis was used to genotype Phaeodactylum cell lines (Figure 3). To do so, each sequence of the gNR-B target locus was first curated and analyzed manually. Next, one of two paired-end sequencing files were entered into the TIDE software and compared to a wild-type sequence for each colony. TIDE outputs a bar plot with predicted indel (insertion or deletion) mutations based on the inputted electropherogram compared to the wild-type. In Figure 3A, a wild-type NR sequence read is shown above the TIDE result plot that compares wild-type to wild-type. The TIDE plot displays one bar at x-axis position “zero” and with a y-axis value of 97.1% (displayed above bar). A pink colored bar also indicates the predicted indel was statistically significant (p < 0.001). This result indicates that there is a zero-nucleotide indel mutation at the predicted Cas9 cut site that comprises 97.1% of the total sequencing reads. Clearly, the TIDE plot validates the sequencing read as a wild-type genotype.

Figure 3B shows six NR knockout mutant genotypes for cell lines 20, 21, 27, 43, 52, and 56. Cell line 20, for example, appears to have a bi-allelic mutation that is homozygous because the TIDE chart shows one pink peak that corresponds to a deletion mutation of 9-bp. Cell line 43, for example, appears to have a bi-allelic mutation that is heterozygous because the TIDE charts shows two pink peaks corresponding to a deletion genotype of 3-bp (28.5%) and 15-bp (49.4%). All six NR-KO genotypes display either homozygous or heterozygous genotypes by their TIDE charts. They also lack any peak at the “zero” position that would indicate that there was a wild-type contamination. Furthermore, cell lines with TIDE charts with only 1 or 2 indel peaks are shown here; additional peaks would indicate that the cell lines were not clonal and need to be re-streaked to find a clonal mutant.

Manual sequencing curation and TIDE analysis was performed for 80 gNR-A colonies and 64 gNR-B colonies. 2/80 (2.5%) gNR-A colonies and 26/64 (40%) gNR-B colonies contained a Cas9-mediated mutation. Here, the Cas9-sgRNA mutagenesis efficiencies of 0% and 28% calculated using T7 (Figure 2) reflects the efficiencies observed using TIDE.

Lastly, the nitrate growth assay was used to correlate the TIDE genotyping efforts to the expected NR mutant phenotype. The six mutant cell lines identified using TIDE did not grow on nitrate media compared to growth on ammonium media (Figure 3C).

A total of 28 transformed Phaeodactylum colonies and 4 wild-type colonies were picked and genotyped using manual sequencing curation and TIDE sequencing analysis (Figure 4). The genotyping was performed successively. First, all 28 colonies were screened for mutation at the GS-2 target locus. Second, GS-2 KO candidate colonies were screened for mutation at the cGOGAT target locus. Therefore, genotyping information is not provided for all 28 colonies for the cGOGAT target locus. The genotyping results using manual curation and TIDE analysis for each colony is provided in Supplementary Table S3 (GS-2) and Supplementary Table S4 (cGOGAT).

A total of 6/28 (21%) colonies contained a Cas9-mediated mutation(s) at the GS-2 target locus and 4/6 GS-2 mutant candidates also contained Cas9-mediated mutations at the cGOGAT target locus (Figure 4B). Figure 4C shows the TIDE plots for each double-knock-out cell line (3, 4, 10, and 15) for both target loci (GS-2 on the left, cGOGAT on the right). Some mutations appear bi-allelic and homozygous, such as cell line 15 at the GS-2 locus (48-bp deletion) and cell line 3 at the cGOGAT locus (24-bp deletion). Only cell line 10 at locus cGOGAT displayed a bi-allelic, heterozygous mutation of a 15-bp deletion and 12-bp deletion. The remaining genotypes were not as clearly defined and have multiple (more than two) peaks in their TIDE chart and/or they contain a trace of a wild-type sequence (both examples are shown in cell line 4 at the cGOGAT locus).