Design and Development of Artificial Zinc Finger Transcription Factors and Zinc Finger Nucleases to the hTERT Locus.

The ability to direct human telomerase reverse transcriptase (hTERT) expression through either genetic control or tunable regulatory factors would advance not only our understanding of the transcriptional regulation of this gene, but also potentially produce new strategies for addressing telomerase-associated disease. In this work, we describe the engineering of artificial zinc finger transcription factors (ZFTFs) and ZF nucleases (ZFNs) to target sequences within the hTERT promoter and exon-1. We were able to identify several active ZFTFs that demonstrate a broadly tunable response when screened by a cell-based transcriptional reporter assay. Using the same DNA-binding domains, we generated ZFNs that were screened in combinatorial pairs in cell-based extrachromosomal single-strand annealing (SSA) assays and in gene-targeting assays using stably integrated constructs. Selected ZFN pairs were tested for the ability to induce sequence changes in a Cel1 assay and we observed frequencies of genomic modification up to 18.7% at the endogenous hTERT locus. These screening strategies have pinpointed several ZFN pairs that may be useful in gene editing of the hTERT locus. Our work provides a foundation for using engineered ZF proteins (ZFPs) for modulation of the hTERT locus.Molecular Therapy - Nucleic Acids (2013) 2, e87; doi:10.1038/mtna.2013.12; published online 23 April 2013.

Introduction

Telomeres are specialized chromosomal end caps comprised of 5′-TTAGGG-3′ hexanucleotide repeats which protect human chromosomes from genomic instability and replicative attrition which is a consequence of their linear structure and the inability of DNA polymerases to completely replicate chromosomal ends.[1] Telomerase extends and maintains telomere length, which allows a cell to keep proliferative capacity. Alternatively, the absence of active telomerase results in telomere shortening with every cell cycle.[2] This becomes an important mechanism for cellular aging, lifespan, proliferative potential, and senescence. Telomere shortening is also an important tumor suppressive mechanism.[3] The human telomerase holoenzyme is a complex of two components: hTR, a single-stranded RNA template, and hTERT, a reverse transcriptase. As the limiting component, human telomerase reverse transcriptase (hTERT) expression is constitutively repressed in normal somatic cell types.[4] The expression of this gene can be upregulated to affect changes in proliferation potential, but this alone is insufficient to initiate oncogenesis.[4,5,6]

The promoter region controlling hTERT expression is densely composed of many regulatory elements. Many of these can be found within the first 300 basepairs (bp), which is considered to be the core promoter.[7,8,9] Studies have suggested that transcriptional regulation of hTERT is done in a cell context-dependent manner and the transcription factors controlling the promoter at different sites can vary between cell lines.[7] Haploinsufficiencies and mutations in hTERT have also been implicated in a number of inheritable genetic diseases that result from increasingly shortened telomere lengths. These diseases (such as idiopathic pulmonary fibrosis, aplastic anemia, and dyskeratosis congenita) exhibit attrition of fast-growing tissues, stem cell depletion, and present at earlier ages in each successive generation (anticipation).[10,11,12] The ability to modify the promoter of hTERT, therefore, would have a wide range of research and potentially therapeutic applications. In addition, systematic site-specific mutational analysis of the endogenous hTERT locus or the use of a tunable regulatory factor at the hTERT promoter across many human cell lines would yield useful insights about the regulation of this crucial gene.

Artificial zinc finger transcription factors (ZFTFs) and ZF nucleases (ZFNs) are methods to modulate the activity of specific genes. Both chimeric proteins are based on the well-characterized ZF DNA-binding protein domains (ZFPs). The polydactyl ZFP is custom engineered for a chosen target site and are commonly generated through one of four publicly available methods: (i) combining individual ZFs in a modular fashion, (ii) using randomly recombined libraries of ZFs in the bacterial-2-hybrid (B2H) selection-based OPEN platform, (iii) context-dependent assembly, or (iv) a hybrid methodology that incorporates elements of both.[13,14,15,16,17] A novel, chimeric protein with site-specific activity is then made by attaching another functional effector domain to the ZFP.

ZFPs, when linked to a transcriptional effector domain, become ZFTFs that will recruit the cellular machinery necessary to activate or repress transcription and are typically developed to sequences found in the promoter regions of the target gene. Due to the ability of a ZFTF to site specifically regulate transcription in cell culture and in vivo, these artificial proteins are useful tools for drug development.[18,19] Recently, Sohn et al. (2010) developed a set of four-fingered ZFTFs that target sequences within the hTERT core promoter and demonstrated transcriptional repression of hTERT in HEK293 cells by linking the KRAB repressor domain to ZFPs made by modular assembly.[20] Several ZFTFs have also been successfully made to other target genes such as ErbB-2, VEGF, and utrophin, all of which have therapeutic potential.[21,22,23]

ZFNs are made by joining a ZFP to the nuclease domain of the FokI endonuclease and are developed in pairs to “full target sites” of the general structure: 5′-(ZFN target site 1)-spacer-(ZFN target site 2)-3′.[24,25] The binding of the two ZFNs at the cognate target half-sites allows the nuclease domain to dimerize in the spacer region and create a DNA double-stranded break (DSB).[26] If the DSB is repaired by homologous recombination using an exogenously supplied donor DNA fragment, then the ZFN-targeted locus can be edited to introduce any number of modifications from single point mutations to the installment of large transgenes. Repair of the DSB through error-prone pathways (e.g., non-homologous end joining) can also result in changes to the nucleotide content of the target site.[27,28] ZFNs have been used to perform genomic editing in human cells and are currently undergoing rapid development for gene therapy and clinical trials using ZFNs directed against the CCR5 gene that have already been initiated.[29,30,31]

In this study, we report on the progress towards engineering new ZFTFs for transcriptional activation and ZFNs that target the hTERT locus. We have identified five ZFN full sites (10 target half-sites total) within the hTERT promoter and exon-1. Multiple three-fingered ZFPs were generated for each target half-site using the OPEN platform strictly for eight target half-sites and a hybrid methodology that incorporated module fingers at the Finger 1 or Finger 2 positions in the OPEN protocols for the other two target half-sites. The resulting ZFPs were converted into ZFTFs or ZFNs by linking these DNA-binding domains to either a VP16 transactivator domain or a nuclease, respectively. ZFTFs were screened for the ability to upregulate transcriptional activity through cotransfection with an episomal hTERT promoter-driven green fluorescent protein (GFP) reporter construct in HEK293 cells. We found that the ZFTFs could not only induce GFP expression in a coarsely tunable manner, but could also be used in combination. ZFN versions of many of the same ZFPs used as ZFTFs were screened and assayed using a extrachromsomal single-strand annealing (SSA) assays, a chromosomally integrated GFP gene-targeting reporter assay, and a Cel-I mutagenesis assay on the endogenous site. Our results present promising data towards developing new tools for not only studying the hTERT locus, but also creating new methods for treating telomerase-associated genetic disease.

Results

In our search for suitable target sites, we chose to limit our focus to the hTERT core promoter and exon-1 sequences for several reasons. First, these sequences are a very GC-rich region and as such are likely to have multiple high probability ZFP-binding sites.[14,15] Second, this part of the hTERT locus is thought to be where the majority of the transcriptional control occurs and contains a high density of both known and putative regulatory sequences. Finally, for ZFN applications, higher rates of gene conversion are possible if a DSB is delivered in close proximity to the desired site of change.[32,33] Thus, if gene editing to the core promoter or 5′-coding regions were to be accomplished, then it would be an advantage to engineer ZFNs to target sites within those sequences.

We identified five full ZFN target sites (10 target half-sites total) either within or in proximity to the hTERT core promoter and exon-1, which are listed in . All target half-sites in the chosen full ZFN target sites follow the 5′-GNNGNNGNN-3′ sequence motif, except for the hTERT5SR target half-site, which contains a CNN triplet. Furthermore, the hTERT5 and hTERT5C target sites overlap as a 3 bp frame shift of one another. For the remainder of this report, we will narrow our scope to the hTERT5 and hTERT6 target sites and data generated at all other sites can be found in the Supplementary Tables S1–4. The hTERT5R and hTERT6L target half-sites have been targeted previously but the full target sites we describe here have not.[20,34] Both the hTERT5 and hTERT6 target sites are located in exon-1 of the hTERT locus with the hTERT6 site containing the translational start codon within the hTERT6R target half-site ().

Engineering ZFPs by modular assembly and OPEN methods

In order to create three-fingered ZFPs for the 10 target half-sites, we utilized three different methods. First, we used the B2H-based OPEN protocols to generate candidate DNA-binding domains for eight of the target half-sites, including those for hTERT5 and hTERT6 full sites ( and Supplementary Table S1). Next, for the hTERT5SL and hTERT5SR half-sites, we installed a module finger (Addgene; http://www.addgene.org) in place of one of the ZFs within the three-fingered rZF libraries used in the OPEN selections due to the lack of coverage provided for some GNN[35] and CNN sequences.[36] We have successfully done this previously in making ZFPs to the GFP gene[37] (Supplementary Table S1). In addition to those ZFPs generated through OPEN protocols for hTERT6L and hTERT6R target half-sites, we constructed ZFPs made by modular assembly (). Thus, we were able to generate multiple candidate ZFPs for each target half-site. Although there is a high degree of conservation of the residues that mediate sequence recognition, there is also significant variation of the other residues. Since there is a component of context-dependence to DNA binding, we screened variants for their activity. These results present a unique set of custom engineered, three-fingered ZFPs for the hTERT locus.

hTERT ZFTFs activate gene expression through hTERT promoter sequences

We converted many of the ZFPs for the hTERT5R, hTERT6L, and hTERT6R target half-sites into ZFTFs (23 total) by linking the VP16 transcriptional activation domain to the N-terminus of the candidate ZFP ().[38] We tested the activity of these ZFTFs using an extrachromosomal GFP reporter system. In this system, ~3 kb of the hTERT promoter and exon-1 were cloned upstream of a GFP reporter gene (pMC6). We hypothesized that the candidate ZFTF would bind to the cognate target site within this sequence and recruit the transcriptional machinery necessary to drive GFP expression (). We cotransfected the ZFTF-expressing plasmids and pMC6 into HEK293 cells (an hTERT+ cell line) and measured the changes in %GFP+ populations by flow cytometry to assess the efficacy of these ZFTFs (). We found that all of the ZFTFs acted as artificial transcription factors and increased the number of GFP+ cells in these populations, in some instances, the fold stimulation could be as high as ~35-fold or greater ().

In our screen of all ZFTFs made, we found that stimulation of GFP expression could vary significantly between individual ZFTFs (). Overall, the ZFTFs made with ZFPs from the OPEN platform performed better than those made purely by modular assembly, with few exceptions.[17] These results support the idea that even though the major contact residues are conserved, that the non-contact residues are still important in determining the overall activity of the protein ().

Of the 23 ZFTFs screened, we chose the best performing ZFTFs from each target half-site set (KW602, KW620, and KW641) and tested them for a potential dose response. We cotransfected single ZFTFs in increasing amounts with the pMC6 reporter plasmid and found that as more ZFTF was transfected, more GFP expression was generated (). We also tested whether combinations of ZFTFs could stimulate transcriptional activity and found that the individual ZFTFs did not interfere with each other. With the combinations, we found an additive response for different ZFTFs (). This data reinforces previous findings demonstrating that ZFTFs can elicit a tunable response from a target gene and suggests that this may be possible for the hTERT locus as well.[22,39]

hTERT ZFNs can initiate repair of an extrachromosomal GFP reporter by SSA

To further test the ZFPs engineered to site specifically target the hTERT locus, we converted a total of 39 ZFPs into ZFNs by installing a wild-type (but not codon-optimized) FokI endonuclease domain (Fn) to the C-terminus of the ZFP ( and Supplementary Table S1). Due to variations in activity seen in the ZFTFs and individual deviations from the consensus sequence in multiple ZFPs, we again designed a screening strategy to test pairing of ZFNs in a combinatorial manner for each target site. To determine the ability of a ZFN pair to recognize, bind, and cut the desired target sequence, we constructed four GFP-based g (SSA assay reporter plasmids where a full hTERT ZFN target site would be inserted between two repeated sequences within the GFP gene. The inserted ZFN target sites also include the full GFP1/2 site, which can be targeted by the GFP-ZFN1 and GFP-ZFN2 ZFN pair as a positive control and internal standard. In this assay, a GFP-based SSA reporter plasmid is cotransfected with ZFN-expressing plasmids together into HEK293 cells. After delivery of the ZFN-mediated DSB, the 5′ overhangs anneal together at the site of the repeats in the GFP gene. When repaired by the endogenous cellular SSA repair pathway, functionality is restored to the GFP reporter construct and GFP+ cells are detected by flow cytometry (). We report rates of GFP gene repair by hTERT ZFN activity as a relative percentage of the nuclease activity measured for the GFP-ZFN1/2 internal standard and positive control ( and ; Supplementary Tables S2–4).[37]

After screening all five combinatorial sets of pairings, we found that the hTERT5 and hTERT6 ZFN sets were, broadly, the most active sets of ZFNs ( and ). For the ZFNs developed for the hTERT5 target site, the KW744/KW664 ZFN pairing exhibited equal nuclease activity to the GFP-ZFN1/2–positive control pair. Also, 8 out of 16 possible pairings were shown to have >50% activity relative to the control (). The hTERT6 ZFN set produced the most active nucleases out of those tested. In this combinatorial screen, the KW633/KW609 hTERT6 ZFN pair was more active than the GFP-ZFN1/2 standard by threefold. We also observed that all but six pairs demonstrated nuclease activity that was higher than the GFP-ZFN1/2 standard (). Therefore, using the SSA assay we identified a number of highly active ZFN pairs targeting the hTERT sequences.

Gene targeting by the hTERT6 ZFNs at a chromosomally integrated GFP reporter construct

We further studied the hTERT6 ZFNs using a chromosomally integrated GFP reporter system. Based on a previously described system, we designed another GFP-based reporter construct that was rendered non-functional through the insertion of an hTERT full ZFN target site and an I-SceI homing endonuclease recognition site as an internal standard.[24] This insertion does not disrupt the endogenous GFP1/2 target site that can be found in the GFP gene and the GFP-ZFN1/2 pair was used as a positive control in this assay (). Briefly, we generated a transgenic HEK293 cell line containing the stably integrated reporter construct and cotransfected the ZFN-expressing plasmids in pairs along with a donor plasmid. Function of the GFP gene is restored if the ZFNs cut their cognate target site and stimulated homologous recombination-mediated repair with the donor plasmid. Again, ZFN activity is normalized as a relative percentage of the activity measured for the internal standard, I-SceI ().

In repeating the hTERT6 ZFN combinatorial pairing screen, we found that all hTERT6 ZFN pairs stimulated less gene targeting than the GFP-ZFN1/2–positive control. The best performing pair, KW635/KW613, produced ~40% activity relative to I-SceI (). We again observed that, in general, the ZFNs made from ZFPs with purely module fingers showed low activity in this screen when compared with nucleases from ZFPs made by OPEN methodologies.[17] The KW610 ZFN, made from the same ZFP used to make the KW600 ZFTF, produced no evidence of nuclease activity in this gene-targeting assay even though we observed measurable activity in the SSA assay screens. Despite all other hTERT6 ZFN pairs showing higher activity in the SSA activity than the GFP-ZFN1/2 pair, in this chromosomal-based assay, the hTERT6 ZFNs did not have as much activity as the GFP-ZFN1/2 pair. A reason for this discrepancy is that a chromosomal gene-targeting assay is a more stringent screen than an extrachromosomal SSA assay. Using the chromosomal GFP-based reporter, we were able to narrow the candidate ZFNs to be tested for activity at the endogenous hTERT locus.

ZFN-induced genomic modification at the hTERT locus

Future development of these ZFNs for gene targeting requires demonstrating efficacy for ZFN-mediated genomic modification at the endogenous hTERT locus. We chose the hTERT5 KW744/KW664 and hTERT6 KW635/KW613 ZFNs because they exhibited the best activity in our screening assays. To assay for genomic modification, we transfected the ZFN pairs into K562 cells and measured the rates of imperfect DSB repair by non-homologous end joining at the endogenous genomic locus using the Cel-I assay.[40] We found that the hTERT5 ZFNs gave ~19% allele modification activity and the hTERT6 ZFNs gave ~7% allele modification activity (). For both sets of ZFNs, using a codon-optimized nuclease domain was critical to obtain maximal activity. While our screening assays suggested that the hTERT6 ZFNs would have greater activity than the hTERT5 ZFNs, we found that the reverse was true in K562 cells. The SSA assay may not be predictive as it is a transient transfection assay in with the reporter plasmid is extrachromosomal and present in multiple copies. Similarly, the assay for the ZFTFs is also a transient extrachromosomal reporter assay and subject to the same limitations. In addition, further sources of discrepancy include: (i) that the hTERT6 ZFNs have more toxicity than the hTERT5 ZFNs (data not shown); (ii) that the chromatin differences of the target sites at the endogenous promoter in K562 cells is modulating the activity of the ZFNs; or (iii) that endogenous transcription or regulatory factor binding to the promoter is competing with ZFN binding.[7] In sum, we have demonstrated that both pairs of ZFNs have the potential to be used as tools in generating cell lines with specific hTERT gene modifications, but additional studies will be necessary to show this.

Discussion

Due to the role hTERT plays in telomere regulation, the hTERT locus is an important target for research and drug development.[41] The ability to direct hTERT transcription in a dose-dependent manner or edit the nucleotide content of the locus could yield useful insights into the regulation of this gene, cellular aging, and the oncogenic process. To this end, we have developed many ZFPs that recognize sequences within the hTERT promoter and exon-1 ( and ). We built the ZFPs using modular assembly methods, the B2H-based OPEN platform, and a hybrid method that combines the two protocols.[14,15] This resulted in several candidate ZFPs for every target half-site.[42] Therefore, we used several different screens to identify the most appropriate ZFP that has the ability to recognize and direct activity to the cognate target site, whether as a ZFTF or a ZFN ( and Supplementary Table S1).

As ZFTFs, we were able to demonstrate that many of our ZFPs linked to the VP16 effector domain could stimulate the transcription of an hTERT promoter exon-1–driven GFP reporter gene (). We were also able to demonstrate that ZFTFs developed for sites in the hTERT locus could elicit transcriptional activation in coarsely tunable manner in which incremental increases in ZFTF produced an additive response from the GFP-based reporter (). This is consistent with the findings of others using ZFTFs at other gene targets.[22,39] Our ZFTF screen also identifies some well-performing candidate ZFTFs (e.g., KW602, KW620, and KW641) that may be pursued for future research in endogenous hTERT transcriptional regulation. In comparison to the ZFTFs made by modular assembly described in Sohn et al. (2010), which were demonstrated to be effective repressors of hTERT transcription, we found the KW308 and KW309 modular assembly ZFTFs to be less effective at activating transcription than the OPEN-generated ZFTFs.[13,17,20] The differences in efficacy for modular assembly ZFPs as ZFTFs may be due to the differences in the module fingers used to construct the ZFPs, the effector domains, and design of the reporter assays.

To test the candidate ZFPs as nucleases, we screened sets of ZFN pairs in a combinatorial fashion for each hTERT target site identified using a GFP-based SSA assay (all sites) and a chromosomally integrated GFP gene-targeting assay (hTERT6 only) (). ZFNs to the hTERT5 and hTERT6 site were the most active sets of nucleases in the SSA assays ( and ; Supplementary Tables S2–4). We then selected two ZFN pairs to demonstrate mutation of the endogenous genomic sequence and were able to achieve modification frequencies up to 18.7% (). As a result, we have not only successfully custom-engineered ZFPs to sequences found in the hTERT promoter and exon-1 region, but have also shown that, as ZFNs, these artificial proteins can direct activity to the cognate target site within the genomic locus.

In summary, the data generated by our screening strategies provide the necessary results to move forward with using these ZFTFs and ZFNs as potential tools to modulate hTERT expression and edit the genomic locus for both research and clinical applications.

Materials and Methods

Generating ZF DNA-binding domains to target sites at the hTERT locus. Five full ZFN target sites were identified in the hTERT core promoter and exon-1 and are listed in . We used the publically available OPEN platform and protein backbone sequence to construct three-fingered ZFP arrays for each target half-site.[14,15] For the hTERT5SL and hTERT5SR target half-sites, we generated ZFPs by including a module finger at the Finger 2 and Finger 1 positions (respectively) during the creation of the three-fingered cassettes used in the stage B2H selections due to the lack of coverage provided for some GNN[35] and CNN sequences.[36] In addition to the OPEN-based ZFPs made for the hTERT6L and hTERT6R target half-sites, we constructed ZFPs by assembling module fingers at each ZF position (KW308 and KW309). For each target half-site, the OPEN protocols produced multiple candidate ZFPs, some of which were chosen for further screening ( and Supplementary Table S1).

Cloning ZFTFs and ZFNs. To create ZFTFs, chosen candidate ZFPs made by the OPEN-based methods were cloned into the PRK5.AD-GFZ3 vector provided by Toni Cathomen to link the VP16 transcriptional activator domain to the N-terminus of the ZFP.[43] ZFNs were made by cloning many of the same ZFPs into the previously characterized GFP-ZFN2-B2H vector to link a wild-type Fn domain of native sequence or codon-optimized to the C-terminus of the ZFP.[37,44] All cloning was performed using standard molecular biology techniques.

Screening hTERT ZFTFs for transcriptional activation activity. Approximately 3 kb of the hTERT promoter and exon-1 was installed upstream of a GFP reporter. In this plasmid construct (pMC6), the endogenous hTERT translational start codon was mutated from ATG to GTG, but the start codon for GFP was preserved. All ZFTFs were screened for transcriptional activation activity by cotransfecting 700 ng of ZFTF-expressing plasmid with 100 ng of pMC6 using lipofectamine into 1 × 105 HEK293 cells. GFP transcription was assayed by measuring %GFP+ cells by flow cytometry on day 2 (FACS Calibur; BD Biosciences, Franklin Lakes, NJ). pMC6 was cotransfected with a blank vector as a negative control. Then, one ZFTF from each target half-site group (three total) was cotransfected singularly in increasing amounts (200–600 ng) and in combination (200 ng each) with pMC6 to assay for any dose response. In , the x-axis plots cellular autofluorescence as “orange” (abscissa) and the y-axis plots the fluorescent intensity of GFP expressed by pMC6 as “green”. Fold stimulation of transcriptional activation was calculated as (experimental/background) and error is reported as SD.

SSA assays for screening hTERT ZFN pairs. Based on a previously described SSA strategy, reporters were cloned by inserting a hTERT full ZFN target site between repeated sequences within GFP gene (four total).[45] Each of these SSA reporters contains a single hTERT target site except for the hTERT5 reporter, which also carries the hTERT5C site due to the overlapping sequence shared (). The reporter constructs also include the GFP1/2 full ZFN target site (5′-ACCATCTTC-ttcaag-GACGACGGC-3′) as a positive control and internal standard. These extrachromosomal SSA reporter plasmids were used to assay for on-target nuclease activity for sets of combinatorial pairs of ZFNs developed for each hTERT target site. Briefly, 100 ng of each ZFN-expressing plasmid and 20 ng of appropriate reporter plasmid were cotransfected into 1 × 105 HEK293 cells using calcium phosphate and the GFP1/2 ZFN pair was used as a positive control. The repair of the GFP gene (after cutting by a pair of ZFNs) was measured by %GFP+ cells using flow cytometry on day 2. The nuclease activities of all hTERT ZFN pairs tested were normalized to the activity of the positive control and are reported as a percentage of GFP1/2 ZFN activity. Blank vector-treated cell populations were on average 0.1% GFP+ and GFP1/2 ZFN-treated cell populations were 0.62, 0.4, 0.57, 0.52, and 0.33% GFP+ on the hTERT5, hTERT5C, hTERT5S, hTERT6, and hTERT6B SSA reporter constructs, respectively.

Gene-targeting assay for hTERT6 ZFNs. The hTERT6 ZFNS were tested for the ability to stimulate homologous recombination at a chromosomal target by using a previously described GFP gene-targeting assay.[24] This assay uses a GFP reporter that has been mutated by the insertion of an hTERT full ZFN target site and an I-SceI recognition site as a positive control and internal standard. Within the GFP gene, the GFP1/2 full ZFN target site can be found as well (see above). This reporter construct was electroporated into HEK293 cells and a monoclonal cell line was derived. ZFN-expressing plasmids were cotransfected by calcium phosphate into the monoclonal cell line in pairs (100 ng each) for a combinatorial screen with 300 ng of repair donor plasmid. Nuclease-mediated gene-targeting events were measured as %GFP+ cells by flow cytometry on day 3. The nuclease activities of all ZFN pairs tested were normalized to the I-SceI–positive control and are reported as a relative percentage of that activity.

Cel-I assay to measure ZFN-induced genomic modification. Imperfect repair of a ZFN-induced DSB can result in mismatches in the DNA helix that bulge, creating the substrate for the Cel-I nuclease. Screening bulk genomic DNA for these mismatches at the hTERT locus allowed us to quantify percent modification after the application of our ZFNs into K562 cells. In brief, 1 × 106 K563 cells were nucleofected according to the manufacturer's instructions (Lonza, Basel, Switzerland) and genomic DNA was isolated from cell populations that had been nucleofected with the hTERT5 ZFNs (KW744/KW664) or hTERT6 ZFNs (KW635/KW613) that possessed a wild-type Fn domain of native sequence or was codon-optimized. Then, the hTERT promoter exon-1 sequence that contains both the hTERT5 and hTERT6 sites was PCR-amplified using the AccuPrime PCR kit (Invitrogen, Carlsbad, CA) and a melting temperature of 60 °C. When the primers 5′-GCCCGAGTTTCAGGCAGCGCTGCGTCCTG-3′ and 5′-TGATGTGCCTGCGCTGCTCTCCGCATGTCG-3′ were used, a 373 bp product was formed, which we gel-purified from agarose. A total of 200 ng of the purified PCR product was then treated with the Surveyor (Cel1) nuclease kit (Transgenomic, Omaha, NE). If the genomic hTERT locus had been modified by the hTERT5 or hTERT6 ZFNs, then digest products of 248 + 125 bp or 308 + 65 bp would appear, respectively. We loaded Surveyor digest products onto a 10% acrylamide gel and imaged our results using ethidium bromide/ultraviolet. Analysis was performed using ImageJ software and calculations described by Miller et al. (2007) after normalizing for nucleofection efficiency (~70%).[46]

Table S1. Zinc finger DNA-binding domains generated for the hTERT5C, hTERT5S, and hTERT6B target sites. Table S2. Combinatorial screen of hTERT5C ZFN pairs in the SSA assay. Table S3. Combinatorial screen of hTERT5S ZFN pairs in the SSA assay. Table S4. Combinatorial screen of hTERT6B ZFN pairs in the SSA assay.

Table 1

hTERT target sites and sequences

Table 2

Zinc finger DNA-binding domains generated in this study for the hTERT5 and hTERT6 target sites

Table 3

Combinatorial screen of hTERT5 ZFN pairs in the SSA assay

Table 4

Combinatorial screen of hTERT6 ZFN pairs in the SSA assay

Table 5

Combinatorial screen of hTERT6 ZFN pairs in the gene-targeting assay