Nuclear gene targeting in Chlamydomonas using engineered zinc-finger nucleases


  • Irina Sizova,

    1. Institute of Biology, Experimental Biophysics, Humboldt Universität Berlin, Berlin, Germany
    2. Division of Radiation Biophysics, Petersburg Nuclear Physics Institute, Russian Academy of Sciences, Gatchina/St Petersburg, Russia
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Andre Greiner,

    Corresponding author
    • Institute of Biology, Experimental Biophysics, Humboldt Universität Berlin, Berlin, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Mayanka Awasthi,

    1. Department of Biochemistry, University of Delhi South Campus, New Delhi, India
    Search for more papers by this author
  • Suneel Kateriya,

    1. Department of Biochemistry, University of Delhi South Campus, New Delhi, India
    Search for more papers by this author
  • Peter Hegemann

    Corresponding author
    • Institute of Biology, Experimental Biophysics, Humboldt Universität Berlin, Berlin, Germany
    Search for more papers by this author

For correspondence (e-mail or


The unicellular green alga Chlamydomonas reinhardtii is a versatile model for fundamental and biotechnological research. A wide range of tools for genetic manipulation have been developed for this alga, but specific modification of nuclear genes is still not routinely possible. Here, we present a nuclear gene targeting strategy for Chlamydomonas that is based on the application of zinc-finger nucleases (ZFNs). Our approach includes (i) design of gene-specific ZFNs using available online tools, (ii) evaluation of the designed ZFNs in a Chlamydomonas in situ model system, (iii) optimization of ZFN activity by modification of the nuclease domain, and (iv) application of the most suitable enzymes for mutagenesis of an endogenous gene. Initially, we designed a set of ZFNs to target the COP3 gene that encodes the light-activated ion channel channelrhodopsin-1. To evaluate the designed ZFNs, we constructed a model strain by inserting a non-functional aminoglycoside 3′-phosphotransferase VIII (aphVIII) selection marker interspaced with a short COP3 target sequence into the nuclear genome. Upon co-transformation of this recipient strain with the engineered ZFNs and an aphVIII DNA template, we were able to restore marker activity and select paromomycin-resistant (Pm-R) clones with expressing nucleases. Of these Pm-R clones, 1% also contained a modified COP3 locus. In cases where cells were co-transformed with a modified COP3 template, the COP3 locus was specifically modified by homologous recombination between COP3 and the supplied template DNA. We anticipate that this ZFN technology will be useful for studying the functions of individual genes in Chlamydomonas.


Chlamydomonas reinhardtii is a valuable model organism for the study of important biological processes such as photosynthesis, metabolism, phototaxis, flagella assembly, circadian rhythmicity, the cell cycle and mating. In particular, Chlamydomonas is a model for human ciliopathies that result from defects in primary cilia (Snell et al., 2004). Moreover, the light-gated ion channel channelrhodopsin-2 (ChR2) is the most widely used tool in the field of optogenetics (Zhang et al., 2011). We are interested in sensory photoreceptors, and wish to decipher the overlapping functions of the 11 photoreceptors in Chlamydomonas (eight rhodopsin-related proteins, two cryptochromes and a phototropin). To achieve this, sequential deletion and modification of these photoreceptor genes is required.

The nuclear, chloroplast and mitochondrial genomes of Chlamydomonas have been completely sequenced. To fully exploit these data in terms of a functional understanding of the proteome, it is necessary to use reverse genetics, which involves gene targeting. The main obstacle for direct gene targeting is the low frequency of homologous recombination (HR) between the nuclear gene of interest and the donor DNA. In Chlamydomonas, the ratio of HR over non-homologous integration of the delivered DNA was <10−4, making isolation of homologous recombinants almost impossible (Zorin et al., 2005). To disrupt a gene of interest in Chlamydomonas, researchers currently use one of two methods: insertional mutagenesis (Gonzalez-Ballester et al., 2011) and TILLING (targeting induced local lesions in genomes), which involves chemical mutagenesis and a subsequent search for mutations of interest by heteroduplex analysis (Kurowska et al., 2011). Both methods require generation of large collections of mutants and the screening of many thousands of clones. To bypass these difficulties, knockdown approaches are widely applied, involving RNA silencing via constructs that express sense/antisense inverted-repeat RNAs or artificial microRNAs that mimic the structure of the miRNA precursor (Ebnet et al., 1999; Schroda, 2006; Cerutti et al., 2011). The silencing of over 30 Chlamydomonas nuclear genes has been reported. Although the protein level was decreased to below 5% of wild-type protein levels in several cases, these approaches have serious limitations. These include incomplete suppression of target gene function, loss of the under-expression phenotype by silencing of the suppression construct, and possible off-target effects.

Recently, we described a method for disruption of the phototropin gene PHOT (Zorin et al., 2009). Chlamydomonas phototropin promotes the blue light-dependent transition from mating-incompetent pre-gametes to mating-competent gametes (Huang and Beck, 2003), but the mechanism of this response is still unknown. For inactivation of PHOT, we employed a single-stranded DNA vector that included a promoter-less fusion between the aminoglycoside 3′-phosphotransferase VIII (aphVIII) gene and a fragment of the PHOT gene that generated fewer off-target integrations than the similar promoter-less double-stranded construct. Use of ssDNA decreased the number of non-homologous recombinants more than 100-fold, and allowed us to identify phototropin deletions effectively (Zorin et al., 2009). Unfortunately, further experiments demonstrated that this fusion strategy was not applicable to inactivation of genes encoding hydrophobic photoreceptor proteins, as the fusion proteins were insoluble in most cases. Another weakness of this approach was the unchanged low level of HR.

To further develop genomic engineering in Chlamydomonas, it remains necessary to increase the frequency of recombination at loci of interest and to design the selection system in such a way that it selects for cells in which the recombination efficiency is increased.

Zinc-finger nucleases (ZFNs) are molecular tools that may be used in precisely this way. ZFNs cut the genome at specific sites, thus facilitating HR between a sequence of interest and the template DNA. ZFNs are derived from the FokI endonuclease of Flavobacterium okeanokoites (Kim et al., 1996). FokI is a bacterial type IIS restriction endonuclease consisting of an N-terminal DNA-binding domain and a non-specific DNA cleavage domain at the C-terminus. Previously, the FokI DNA-binding domain has been replaced by arrays of three to six zinc-finger domains (Shimizu et al., 2011) The synthetic proteins were named ZFNs, and allowed targeting of novel DNA sequence epitopes. To date, site-specific genome modification using ZFNs has been demonstrated for more than 11 organisms, including insects, fishes, mammals, plants and several cell culture lines (Urnov et al., 2010; Carroll, 2011; Handel and Cathomen, 2011; Kim et al., 2011). In the best cases, the rate of gene modification reached 50% (Maeder et al., 2008).

The most challenging task in developing ZFN technology is the creation of novel zinc-finger arrays with high specificity and affinity for the chosen target sites. In many cases, their properties are difficult to predict due to context-dependent neighboring effects between ZF domains. Thus, engineered ZFNs must be validated in vitro, in yeast or directly in the cells/organism of interest, using reporter genes containing the inserted target sites (Wolfe et al., 2000; Maeder et al., 2008; Ramirez et al., 2008; Lam et al., 2011; Perez-Pinera et al., 2012).

We present here a method to test the efficiency of assembled ZFNs for targeting specific DNA sequences and to modify a gene of interest in the model alga C. reinhardtii. We employ a selection system that indicates the efficiency of the expressed ZFNs, selecting transformants with active and target-specific ZFNs. Only these cells are selected for recombination events within the gene locus of interest. We demonstrate that engineered ZFNs may be successfully used for site-specific modification of a Chlamydomonas nuclear gene in which mutations are non-selectable.


COP3-specific zinc-finger nucleases

Our purpose was to develop a method for gene targeting in Chlamydomonas that is easily applicable for direct modification of any gene of interest. The COP3 gene encoding the light-gated cation channel channelrhodopsin-1 (ChR1) (Nagel et al., 2002) was chosen as an example of such a gene. RNA silencing experiments demonstrated that the suppression of COP3 expression is not lethal (Berthold et al., 2008). In the first stage, we inspected the genomic sequence of COP3 for binding sites of three-finger zinc-finger arrays with predictably good binding properties. Using the Zinc Finger Targeter (ZiFiT) online program (Sander et al., 2007), the most promising ZFN recognition site was found at position 4699–4722 in exon 13. For the right half-site, ZiFiT identified a distinct three-finger zinc-finger array (database entry OZ179) as capable of activating transcription of a reporter gene by more than 14-fold in a bacterial two-hybrid assay (Wright et al., 2005; Maeder et al., 2009). This ZFN was named ChR1-R. The complementary ZFN ChR1-L was assembled from three ZF domains that were expected to recognize three GNN nucleotide triplets (Ramirez et al., 2008; Sander et al., 2009). As the performance of such an assembled array was unknown, we created two ChR1-L versions named ChR1-L1 and ChR1-L2. The DNA recognition helix sequences provided by ZiFiT are listed in Table 1. The sequences for the full-length codon-optimized ZFNs are given in Data S1.

Table 1. Design of ZF modules. The helix peptide sequences for the DNA-binding motifs are shown with the corresponding base triplets. F1-F3 indicates the finger positions within the ZFN. The complete sequences are given in Data S1
ChR1-R (OZ179)

In vivo test system for zinc-finger activity in Chlamydomonas

We designed a test system for ZFNs in Chlamydomonas that is based on repair of an inactivated marker gene that was stably integrated into the genome. In tissue culture systems, a chromosomal integrated reporter (inactivated GFP) gene that harbors the ZFN target site is commonly used to test designed ZFNs (Zou et al., 2009). We combined a similar approach with our previously applied selection system, which is based on repair of a non-functional antibiotic (paromomycin) resistance marker (Sizova et al., 2001). The 24 bp ZFN target sequence of the COP3 gene was inserted into the aphVIII sequence of the ble::gfp::aphVIII test plasmid (Figure 1a). To test the relative efficiency of our system, the target sequence was flanked by the recognition sequence of the zinc-finger-containing transcription factor Zif268 fused at the N-terminus to the FokI cleavage domain (Zif268::FokI) as a control (Figure 1a) (Wright et al., 2005). The resultant plasmid p60Chop1 was transformed into the C. reinhardtii CW15-302 strain, which is known for its superior transformation rate. We have used this strain in previous studies for deletion of the phototropin gene (Zorin et al., 2009). The whole cassette including the 3′ untranslated region (UTR) was integrated in transformant ZF37. It expressed a truncated Ble::Gfp::∆AphVIII fusion protein, due to the insertion of an in-frame stop codon (Figure 1b, ORF1). In the next step, we transformed strain ZF37 with a plasmid encoding Zif268::FokI under the control of the heat-shock promoter HSP70A (Schroda et al., 2000). As a template for aphVIII repair, we co-transformed a 120 bp 5′ deleted aphVIII homolog lacking the N-terminal ATP-binding site. The construct is non-functional if integrated off-target into the genome. As the aphVIII repair process requires the expression of ZFNs, no additional selection marker was included within the ZFN plasmids. The homology region of the template and aphVIII extends from 580 bp upstream to 931 bp downstream of the target site (Figure 1b). We obtained 1–5 paromomycin-resistant (Pm-R) colonies with repaired aphVIII marker genes per transformation of asynchronously grown ZF37 culture. Control experiments with plasmids lacking either the template or the ZFNs did not result in Pm-R colonies that may result from spontaneous HR events. In our next series of experiments, using the designed ZFNs recognizing COP3, we used an optimized FokI system that does not form homodimers. This was achieved by insertion of the mutations Q486E/I499L and E490K/I538K in the respective monomers (Miller et al., 2007). These experiments showed that only when the ChR1-R module was combined with ChR1-L1 was the aphVIII gene successfully repaired, whereas the combination of ChR1-R and ChR1-L2 was non-functional. The efficiency of stimulating marker repair by ChR1-R and ChR1-L1 was similar to that of Zif268::FokI. The repaired marker protein in isolated Pm-R colonies was visualized as an enlarged protein by immunoblotting (Figure 1b, ORF2).

Figure 1.

Experimental design of the ZFN-mediated marker repair system. (a) The COP3 intron/exon distribution is shown in the header. The ZFN target sequence within exon 13 is enlarged and the recognition sites for ChR1-L1 and -L2 are shown in blue. The ChR1-R recognition sequence is shown in pink. F1-F3 indicate the position of the ZF modules from Table 1. Zif268 binding sites (green) were added upstream for control experiments. (b) Schematic representation of the model cassette after random insertion into the genome. Fusion protein expression was regulated by the HSP70A/RBCS2 (HR) promoter. The Sh ble gene conferring zeocin resistance (Zc-R, red) was used for selection of strains expressing the model cassette. The gfp sequence was inserted as a linker. The non-functional aphVIII gene was inactivated by inserting ZFN target sites from (a), represented as colored rectangles. ORF1 indicates the translation stop codon after the target site insertion. The repair mechanism is illustrated below following a ZFN-mediated DSB (indicated by scissors). The delivered repair template ∆5′aphVIII was inactivated by deleting 120 bp. Homology regions surrounding the cleavage site are indicated by arrows. Successful repair and removal of the insertion results in a functional aphVIII sequence (red), conferring resistance to paromomycin (Pm-R). ORF2 is indicating the prolonged reading frame. (c) Detection of the fusion protein using anti-Sh ble antibodies indicated an in increased molecular weight after repair of the aphVIII marker.

Optimized ZFN performance in Chlamydomonas

To further increase the recombination efficiency, we used two approaches. First, we inserted mutations that stimulated the catalytic activity of the nuclease domain. Second, we used synchronized cells and transformed these cells at various stages of the cell cycle. In most ZFN studies on plant or mammalian cells, cultures are usually arrested in S phase, as HR works most efficiently at this cell cycle stage. The reason for this is that HR is the only mechanism facilitating error-free repair of double-strand breaks (DSBs) during DNA replication (Symington and Gautier, 2011). As cell-cycle blockers were expected to cause undesired mutations, we synchronized Chlamydomonas cells by growth under a 14 h light/10 h dark cycle, taking into account the fact that CW15-302 cells are unflagellated and grow in a less synchronized manner than wild-type cells. Synchrony of cells used in our experiments was verified microscopically by the appearance of two- to four daughter cells at the start of the dark phase. A representative growth curve of strain ZF37 is shown in Figure S1. Under these conditions, we tested the reported amino acid substitutions in the FokI cleavage domain. Substitutions at two positions, Ser418 and Lys441 (Figure 2b), have been described as enhancers of cleavage activity (Guo et al., 2010). First, the performance of these mutations was tested on the basis of the FokI heterodimers. ZF37 cells were transformed at the start of mitosis (S/M). All subsequent mutations were integrated into both heterodimers. Similar to reported observations for heterodimers including the mutations S418P or K441E (FokI-SP or FokI-KE), we found a three- to fourfold increase in aphVIII marker repair in comparison with the original Miller FokI version (FokIM). However, combination of S418P and K441E (FokI-SP/KE) reduced the amount of Pm-R colonies more than fourfold (Figure 2a), in contrast to the findings of Guo et al. (2010), who found stimulation of mutagenesis using this combination. Our reduction may be due to cell death caused by excessive off-target nuclease activity, which cannot be completely prevented (Halford et al., 2011). Examination of the FokI crystal structure (Wah et al., 1997) shows that Ser418 forms a polar contact with a phosphate of the DNA backbone close to the active center (Asp450, Asp467, Lys469). Substitution of Ser418 by Pro resulted in removal of the polar contact between the nuclease and the DNA, which increased the enzyme activity, as found by Guo et al. (2010). In contrast, substitution of Ser418 by Arg produces an additional polar DNA contact as shown by molecular modelling of amino acid substitution S418R in the FokI cleavage domain (Figure 2b). This mutation would probably decrease the catalytic activity of the FokI cleavage domain. Next, we tested ZFNs in which Ser418 was substituted by positively charged Arg (FokI-SR), and found a significant reduction (more than threefold) in catalytic activity. However, the double-mutated nuclease FokI-SR/KE with one stimulating mutation (K441E) and another inhibiting mutation (S418R) resulted in the highest recombination-stimulating effect (Figure 2a). We considered the activity level of this nuclease to be the best compromise between activity and toxicity.

Figure 2.

Effect of FokI cleavage domain activity on the increase in HR events. (a) Changes in stimulation of HR events by mutated FokIM domains. Cells were transformed after synchronization either during the G1 phase or within the S/M phases of the cell cycle. At least two independent experiments, each including three transformations, were performed. Values are means and standard deviation from a single transformation. (b) Modeling of amino acid substitutions S418R and K441E in the crystal structure of FokI bound to DNA (PyMOL-Software; PDB entry 1FOK). The active center is represented in red, and the dimerization interface in yellow. The helix-turn-helix motif that makes contact with the DNA is shown in blue. The substitution S418R is indicated in light blue and the substitution K441E is indicated in green, and both are encircled by dashed lines.

We compared the transformation efficiencies of the optimized ZFNs in the G1 and S/M phases of the cell cycle. As expected, the number of homologous recombinants was higher for the cells at the S/M phase whenFokI-SP or FokI-KE was used. However, FokI-SR, FokI-SP/KE and FokIM did not show any preferences due to the very low number of colonies obtained from these experiments. Application of FokI-SR/KE resulted in high numbers of colonies (Figure 2a), but again with no preference for the cell-cycle phase.

In summary, the strategy of transforming synchronized cells with the mutated S418R + K441E nuclease (FokI-SR/KE) increased the number of HR events by approximately 5–8 times relative to transformation of non-synchronized cells with FokIM. The SR/KE variant of a COP3-specific ZFN was therefore chosen for mutagenesis of the endogenous gene. Our data show that the use of publicly available databases of ZF modules and arrays for ZFN design combined with optimized SR/KE nuclease allows engineering of an enzyme that out-performs stimulation of HR by the control Zif268::FokI.

Targeting the COP3 gene

We looked for mutations within the COP3 gene among the transformants generated using the ChR1-SR/KE ZFNs. It was expected that, in ZF37 cells, the designed COP3-specific ZFNs may cut both targets: aphVIII containing the inserted zinc-finger target site and the endogenous COP3 gene. As mentioned above, antibiotic resistance served as an indicator of intracellular ZFN activity. The first search for COP3 lesions was performed among clones generated in the absence of COP3 homologous template DNA. In this case, DSB repair occurred by the non-homologous end joining (NHEJ) pathway that often leads to small deletions or insertions at the target site (Wright et al., 2005; Gonzalez-Ballester et al., 2011). Such small sequence modifications may in principle be detected by mismatch-specific nucleases such as Cel-1, described in detail for analysis of the ROSA26 locus in murine cell lines (Perez-Pinera et al., 2012). Using Cel-1, we found, among 192 Pm-R clones, two mutants with deletions of 9 and 18 bp, respectively (designated ZF37-∆9 and ZF37-∆18). Unfortunately, both were in-frame deletions. Additionally, we detected one clone (designated C11) by PCR that contained an approximately 3.5 kb insert inside the COP3 target sequence. This clone contained a partial copy of the model gene, including its promoter and the regions upstream of the ZFN cleavage site. The ChR1 protein was absent in C11, as evidenced by protein immunoblotting. Unfortunately, repair by NHEJ results in mutations that are unpredictable and the alterations are not necessarily useful, as in our mutants ZF37-∆9 and ZF37-∆18. Moreover, identification of deletions by Cel-1 is hard to visualize and detect in DNA polyacrylamide gels, especially when DNA from several clones is pooled in the initial survey (Figure S2).

As our objective was to create a robust and reliable method with predictable changes in the target locus, in our next experiments we used COP3 template DNA to promote HR. The COP3 template comprised 2 kb of the genomic COP3 sequence upstream of the target site and a homology region of 700 bp downstream. The short homology region was required to obtain permanent good PCR results for reliable recombination monitoring. Homologous regions are shaded in Figure 3(a). The target sequence for COP3 was modified in two ways: first, by insertion of a 58 bp sequence encoding a FLAG tag (Einhauer and Jungbauer, 2001), together with two in-frame stop codons (insertion template), and second, by deletion of a 228 bp fragment of the intron 12/exon 13 border, including the left ZFN target half-site (deletion template). PCR screening of the Pm-R recombinants that were generated using one of these templates identified four ∆COP3 mutants out of 192 transformants that contained the predicted modifications of the target site (Figure 3b). Two mutants (H2 and H4) had the FLAG insertion and two mutants (6E and 7G) showed the 228 bp deletion precisely copied from the donor DNA (Table 2). An additional mutant (ZF37-Ins243) was generated as a result of DSB repair through NHEJ, with a 243 bp insertion of a genomic chromosome fragment. Immunoblotting showed no residual expression of ChR1 in all five characterized mutants (Figure 3c). Table 2 lists all mutations obtained for the COP3 locus.

Table 2. List of mutants obtained from all COP3 gene targeting experiments. Each mutant is listed indicating the type of mutation, the underlying mechanism, and the regions flanking the insertions/deletions. The last column indicates whether gene alterations resulted in loss of COP3 expression. NHEJ, non-homologous end joining; HR, homologous recombination
Mutant strainType of mutationFlanking regionsMechanismCOP3 knockout
ZF37-∆9Deletion (9 bp)4709–4717NHEJNo
ZF37-∆18Deletion (18 bp)4730–747NHEJNo
ZF37°C11Insertion (3.5 kb, part of the model cassette)4709–4715NHEJYes
ZF37-Flag (H2 + H4)Insertion (58 bp, FLAG, stop)4715–4716HRYes
ZF37-∆227 (6E, 7G)Deletion4478–4714HRYes
ZF37-Ins243Deletion (8 bp) + insertion (243 bp)4478–4714NHEJYes
Figure 3.

Applied strategies for template-dependent gene modifications. (a) The COP3 sequence after a ZFN-induced DSB is illustrated in the center. The design of the donor DNA is shown, and homology regions around the cleavage site are indicated by arrows. This template has the ZFN target site interspaced with a FLAG tag-encoding sequence (red rectangle). The red triangle indicates the inserted unique primer binding site, with the reverse primer (black triangle) starting outside the homology region. The ‘deletion template’ refers to a homologous template that is lacking part of the intron 12/exon 13 border. Primer positions for screening are indicated as black triangles. Homologous regions are shaded. (b) Left panel: PCR screening of Pm-R colonies using the insertion template shown in (a). Negative clones do not have insertions of the FLAG tag sequence, and consequently no PCR product was obtained. Clones ZF37-H2 and ZF37-H4 contained the inserted FLAG tag, and therefore yielded the expected PCR product. Right panel: results of PCR screening of the Pm-R colonies using the deletion template shown in (a). Clone ZF37-3B contained a deletion, resulting in a shortened PCR product (935 bp versus 1163 bp in control); clone ZF37-6E contained a random insert in the target locus, resulting in a longer PCR product (1383 bp versus 1163 bp in control) (c) Upper panel: Immunodetection of ChR1. CW15-302 and ZF37 were used as controls. ZF37-H2, -H4, -3B, -8B, and -6E show no residual ChR1 protein expression. Lower panel: Ponceau staining of the proteins used for immunodetection shown in the upper panel.

Of concern is the possibility that integrated ZFNs cut the genome at off-target loci, which may negatively influence the genome stability and phenotype of deletion strains. Therefore, we tested the genomic insertion of ZFN DNA in all ∆COP3 deletion strains. ZFN sequences inserted into the genome in only three of eight clones (Figure S3), which may indicate that prolonged ZFN expression is toxic for the cell. Thus we conclude that, in COP3 experiments, only clones which have transiently expressed the ZFNs from the plasmid DNA have survived. In clones where we detected the ZFN sequence, it is possible that just a single ZFN monomer was integrated.


We have shown that the zinc-finger nuclease technology can be used in C. reinhardtii for deletion or specific modification of a nuclear gene. The genetic manipulation is of special interest for advances in Chlamydomonas research as well as for biotechnological applications. The efficiency of the described method is sufficient for routine work as long as the individual steps of the procedure are not changed without validation.

The model system presented, based on the repair of a marker gene, enabled us to delete a non-selectable Chlamydomonas gene that encodes a membrane protein. The strategy allowed us to evaluate our designed ZFNs for target recognition in vivo, and revealed a clear preference of the COP3 specific ZFN combination L1/R over L2/R. Second, in Chlamydomonas, the mutated nuclease domain FokI-SR/KE outperformed FokI-SP/KE, several single mutants and FokIM. The intracellular ZFN activity was indicated by the repair of a defective aphVIII gene containing the same target site in the same cell. The frequency of COP3 modification was independent of the pathway for DSB repair through NHEJ or HR, and was approximately 1%.

The stable growth of the resulting Pm-R colonies for several months confirmed that transient ZFN expression has no cytotoxic effects, although we had considered the possibility of long-term off-target effects caused by continuous expression of the ZFN proteins. To keep off-target restriction to a minimum, we used a heat shock-inducible HSP70A promoter for ZFN expression (Schroda et al., 2000). Previously, heat shock induction of ZFN expression was successfully used to allow recombination events in Drosophila (Bibikova et al., 2002).

The two DNA repair pathways (NHEJ and HR) are assumed to work in an equilibrium state. NHEJ is the major pathway for repairing endonuclease-induced DSBs even during replication (S/G2 phase), and does not compete with HR. In mammalian cells, the HR frequency is a few times higher in the late S and G2 phases than in the G1 phase, whereas NHEJ is unaffected by the phase of the cell cycle (Takashima et al., 2009). As expected, our HR assay in Chlamydomonas demonstrated the superior repair efficiency for cells transformed during the cell cycle S/G2 phase (Figure 2); this is especially pronounced for ZFNs with the mutations S418P and K441E (Figure 2). The increase depends on the architecture of the FokI cleavage domain and the zinc-finger array specificity, because higher FCD activity of a low-specificity zinc-finger array may cause off-target effects, resulting in a lower survival rate. In mammalian cells, the mutations S418P and K441E additively stimulated substrate turnover and site-specific mutagenesis (Guo et al., 2010; Perez-Pinera et al., 2012). The crystal structure of the FokI domain bound to DNA as shown in Figure 2 highlights the fact that amino acid substitutions at position 418 are associated with changes in the number of polar contacts with a phosphate of the DNA backbone. The substitution S418P removes the polar contact, thereby increasing the catalytic activity, whereas the substitution S418R produces two polar contacts that probably cause the sharp decline in nuclease activity. Interestingly, we also found an increase in HR events for FokI-SR/KE during the G1 phase of the cell cycle. One of the most likely explanations is a nickase-like behavior of the FokI-SR/KE variant. Nickases, unlike nucleases, induce a single-strand cut that also stimulates HR-dependent repair but reduces the mutagenic effects caused by NHEJ, as no DSB is introduced (Ramirez et al., 2012). This may explain the higher survival rate during G1, as fewer off-target DSBs would have occurred if FokI-SR/KE was used, although HR stimulation would be comparable.

One of our ∆COP3 mutants was inactivated by a 3.5 kb insert into the target site. The insert contained a copy of the model gene with the aphVIII/ChR1 junction exactly at the cleavage position. Such a rearrangement probably resulted from repair through inter-chromosomal recombination between homologous target sequences located in different chromosomes, as described by Richardson et al. (1998) and Richardson and Jasin (2000). Although the 24 bp homology sequence was rather short, it was obviously sufficient for invasion of the disrupted COP3 end into the model gene at the heterologous chromosome.

The advantage of HR over NHEJ is that only the HR pathway for DSB repair leads to predictable modifications, as seen with the 58 bp insertion or 228 bp deletion in the mutants H2, H4, 6E and 7G. Both modifications were exactly copied from the donor DNA through a pathway best described by the synthesis-dependent strand annealing mechanism discussed previously (Bartsch et al., 2000; Carroll, 2011). Interestingly, although the ZFN target site was close to the end of the COP3 locus, the generated alterations did not result in a truncated protein. The most likely explanation is premature mRNA decay caused by the extended UTR. The described modifications facilitated the isolation of COP3 knockouts, because modified alleles could be easily screened by PCR.

The limitation of the described procedure is that CW15-302 cells are not flagellated. Transfer of the procedure to a cell wall-deficient motile strain requires some alteration of the conditions, and such optimization is ongoing in our laboratory. However, currently the method is applicable only to projects that do not involve flagella structure and physiology or on cell motility in particular. Cells of strain CW15-302 are known to have high transformation efficiency and exhibit high transgene expression levels. This high level of expression may be important for expression of the Ble::Gfp::AphVIII fusion protein.

In summary, the presented technology has the potential to modify any Chlamydomonas gene at any position, including coding regions, introns, promoters or 3′ UTRs. ZFNs are designed on the basis of available online tools and are tested in vivo by their ability to repair the aphVIII gene confering resistance to paromomycin. Simultaneous delivery of a template that is homologous to the gene of interest and is truncated around the target sequence is most convenient for identification of those clones that contain the modification of the endogenous gene. Currently, this method works reliably for strain CW15-302, and we are working on transfer of the technology to other strains that are motile.

Experimental Procedures

Plasmid construction

All constructs presented are listed in Data S1. Briefly, to construct the target sequence for the Zif268 (underlined lower-case letters) and ChR1 ZFNs (underlined capital letters), a 55 bp fragment (5′cgcccacgcGAATTCgcgtgggcgTCTAGACCCTCCGCCATGAGCGCCGGCGGC3′) was inserted into the aphVIII sequence by PCR and standard cloning. The sequence of the SacI-XhoI fragment from p60Chop1 (Figure 1b) containing the expression cassette HSP70A/RBCS2::BLE:GFP:∆APHVIII is shown in Data S1. The ∆aphVIII repair template (p∆120APH) was amplified by PCR, starting at position 120, from a functional aphVIII:PsaD UTR template.

All nucleotide sequences encoding ZFNs were codon-adapted to the Chlamydomonas nuclear genome. The coding sequence for Zif268::FokI was obtained from GeneArt/Invitrogen ( ChR1-specific ZF domains were designed using ZiFit (; the sequences were obtained from GeneArt and cloned as SmaI-NotI fragments upstream of the FokI cleavage domains. The ChR1-R, -L1 and -L2 peptides and DNA sequences are listed in Data S1. Our standard heterodimeric FokI cleavage domains included the mutations Q486E/I499L or E490K/I538K (Miller et al., 2007). The mutations S418P, K441E (Guo et al., 2010) and S418R were inserted by site-directed mutagenesis. The insertion (pFLAG) and deletion (p∆227) templates were created by overlapping PCR and cloned into p∆120APH.

Test strain creation

Cells of Chlamydomonas strain CW15-302 (now listed as CC-4350 in the Chlamydomonas Center, were grown in Tris-acetate-phosphate (TAP) medium under 20 W m−2 cool fluorescent white light at 120 rpm in an orbital shaker at 25°C. Seven days after inoculation, cells were used for transformation with glass beads (Kindle, 1990). Approximately 1–2 μg p60Chop1 linearized with ScaI was used to generate the model strains. After overnight incubation, the cells were collected by centrifugation at 2000 g for 10 min and plated on TAP agar containing 10 μg ml−1 zeocin (Invivogen). Zeocin-resistant colonies were inoculated into TAP medium, grown, and used for total DNA isolation. DNA was analyzed by PCR using a Ble forward and a PsaD terminator reverse primer (5′CCGAGATCGGCGAGCAGCCGTGG3′ and 5′GATGCTGCATGTGCACAGTCACGCTGTCTCC3′). A clone containing a single complete model cassette was designated ZF37 and used in the following experiments.

Test strain transformation

ZF37 cells were grown in TAP medium under 20 W m−2 cool fluorescent white light at 120 rpm orbital shaker (Innova 44, New Brunswick) with a 14 h (25°C)/10 h (18°C) light/dark cycle for 7–10 days after inoculation, cell culture was continuously diluted to maintain rapid growth. Thirty to forty minutes after the switch from light to dark, cells were harvested by centrifugation at 2000 g for 10 min at 20°C. The cell pellet was resuspended in TAP medium to a concentration of 3 × 108 cells ml−1, transferred to a 15 ml−1 Falcon tube, and incubated horizontally for 35–40 min in a water bath at 40°C. Each transformation reaction included 300 μl cell suspension, 5 μg ChR1-R DNA and 10 μg ChR1-L1 DNA. In control experiments, 10 μg Zif268::FokIDNA was used. Five to ten micrograms of p∆120APH/Flag or p∆120APH/∆227 template DNA were used in every reaction.

Screening of targeted mutations in the COP3 gene

Isolated clones containing targeted mutations in the COP3 gene (Pm-R colonies) were inoculated into TAP medium in 96-well plates and grown for a few days. Genomic PCR was performed as described by Cao et al. (2009) using 1 m betaine and Phusion polymerase (Fermentas, in GC buffer # F-521S. For the Cel-1 mismatch detection assay, we used a SURVEYOR mutation detection kit (Transgenomic, with forward primer 5′CATGCAGCCCATGCAGCAGGCTATG3′ and reverse primer 5′GAGGTACCGACTGGCCTGAGCTGTGTTGCG3′. For colonies obtained using the template pFLAG, PCR amplification was performed using one primer located within the FLAG sequence (5′GACTACAAGGACGACGACGACAAGGTGTAATGA3′) and the other one outside the homologous region (5′GTTAAAGCTTCGCCCTTGTTCCTGCCACCA3′). For screening colonies obtained with the p∆228 donor DNA, one primer was located before the deletion (5′CTTTTCTTGGAACTTGTTGCGAACCTGCATGTCA3′) and the reverse primer was the same as for pFLAG.

Antibody generation

A C-terminal (Ct) fragment of ChR1 (amino acids 320–712) was cloned into the pET-SUMO (small ubiquitin-like modifier) expression system (Invitrogen) that possesses an N-terminal 6 × histidine tag. Expression of this protein in Escherichia coli BL21(DE3) produced a recombinant protein in which the Ct fragment of ChR1 was fused to the SUMO protein. The affinity-purified ChR1-Ct fusion protein was used as an antigen for generating polyclonal antibodies in rabbit Bangalore Genei (Merck India, The immune serum was affinity-purified using protein A–Sepharose beads (Sigma-Aldrich, followed by immobilized recombinant SUMO protein fragments.

Protein immunoblotting

Proteins were separated by 10% Laemmli polyacrylamide gel electrophoresis. Immunoblotting was performed under standard conditions using antisera against ChR1 (1:10 000) or anti-sh-Ble Sh ble is a binding protein with strong affinity for antibiotics of the phleomycin family (1:2000, Invivogen). Alkaline phosphatase antibodies (anti-rabbit; A3687; Sigma-Aldrich) were used for detection at a dilution of 1:2000.


This work was supported by the Deutsche Forschungsgemeinschaft (FOR1261, to P.H.) and the Russian Foundation for Basic Research (11-04-00944-a, to I.S.). We thank Pete Lefebvre Department of Plant Biology, University of Minnesota for discussion and critical reading of the manuscript.