An engineered tale-transcription factor rescues transcription of factor VII impaired by promoter mutations and enhances its endogenous expression in hepatocytes

Tailored approaches to restore defective transcription responsible for severe diseases have been poorly explored. We tested transcription activator-like effectors fused to an activation domain (TALE-TFs) in a coagulation factor VII (FVII) deficiency model. In this model, the deficiency is caused by the −94C > G or −61T > G mutation, which abrogate the binding of Sp1 or HNF-4 transcription factors. Reporter assays in hepatoma HepG2 cells naturally expressing FVII identified a single TALE-TF (TF4) that, by targeting the region between mutations, specifically trans-activated both the variant (>100-fold) and wild-type (20–40-fold) F7 promoters. Importantly, in the genomic context of transfected HepG2 and transduced primary hepatocytes, TF4 increased F7 mRNA and protein levels (2- to 3-fold) without detectable off-target effects, even for the homologous F10 gene. The ectopic F7 expression in renal HEK293 cells was modestly affected by TF4 or by TALE-TF combinations. These results provide experimental evidence for TALE-TFs as gene-specific tools useful to counteract disease-causing promoter mutations.

Little has been done to counteract the effect of promoter mutations, which represent a relevant cause of human genetic disease (http://www.hgmd.cf.ac.uk), via engineered transcription factors (eTFs).
One versatile tool is the transcription activator-like effectors (TALEs) that exhibit simple DNA-binding properties with a conserved central region composed of a series of 33-35 amino acid repetitive elements 1,2 , each containing a highly variable di-residue (RVD) at the 12 th -13 th positions that dictates the binding preference 3,4 . Therefore, the tailored mutagenesis of the RVDs and fusion to transcriptional activators (i.e., VP16, VP64, p300, p65, TET1) 5 generates TALE-TFs that can target a selected promoter region and increase gene expression by stimulating transcription 6 (Fig. 1a). Thus far, TALE-TFs have been mainly exploited to study stem-cell maintenance and differentiation [7][8][9][10][11][12][13][14] or to increase transcription of the frataxin gene in the presence of a trinucleotide (GAA) repeat expansion 15 . Single nucleotide changes severely impairing the transcription of disease genes have not yet been addressed.
Factor VII (FVII) is the serine-protease that triggers blood coagulation, and it offers a unique model of finely regulated gene expression. Therefore, FVII can be taken as model system to study the positive or negative modulation of gene transcription. Here, we chose two mutations known to cause severe FVII deficiency 16 as paradigmatic examples of mutations that impair transcriptional activity by affecting key transcription factor binding sites, which is a common pathogenic mechanism induced by promoter defects. Through reporter gene assays, we identified a single TALE-TF that was able to rescue the defective F7 promoter, and it was also active in the (a) Schematic representation of the TALE-TF structure and of the genomic region comprising F7 and the adjacent F10 genes. TALE-TF is composed of a TALE-derived modular DNA binding-domain (TALE-DBD) fused with a VP64 trans-activation domain, the latter recruiting the transcriptional machinery (Transcription Complex). (b) Representation of the reporter F7 constructs, which includes the proximal promoter region of F7 (520 bp) driving the expression of the firefly luciferase. The relative localization of the four TALE-TFs (TF1, TF2, TF3 and TF4) is reported above the scheme. The wild-type and mutated Sp1 and HNF4 binding sequences are indicated below. TSS, Transcription Start Site. (c) Promoter activity of F7 variants alone or upon co-transfection with the pTFs or the pHD, devoid of the TALE-DNA binding domain, in HepG2 cells. (d) Promoter activity of F7 variants alone or upon co-transfection with the pTF4 in HuH7 cells. Histograms report the fold expression of normalized luciferase activity over that of the pFVII-luc wt . The results are expressed as the mean ± standard deviation from at least three independent experiments. higher than that of the wild-type construct (pFVII-luc −94g , 5.9 ± 0.6-fold; pFVII-luc −61g , 25 ± 4.9-fold). The pFVII-luc wt luciferase activity was also significantly enhanced (pFVII-luc wt , 36.6 ± 17.3-fold) (Fig. 1c). Taking into account that tumour cells show different transcriptional mechanisms and profiles 20 , TF4 activity was also explored in HuH7 hepatoma cells. In this context, TF4 exhibited robust trans-activation activity on F7 promoter variants (pFVII-luc −94g , 23 ± 6.1-fold; pFVII-luc −61g , 76 ± 10.2-fold) and wild-type (17.1 ± 1.5-fold) (Fig. 1d). Conversely, TF1, TF2 and TF3, which were designed to target upstream regions (Fig. 1b), were ineffective in rescuing F7 promoter variants and seemed to exert an inhibitory effect (Fig. 1c). These results may be due to an underlying competition and/or steric hindrance with the endogenous transcription factors on their binding sites.
As hepatocytes from the severely affected FVII-deficient patients with promoter mutations were unavailable, we tested TF4 on the endogenous F7 gene expression in HepG2 cells. RT-PCR analysis conducted on the total RNA extracted from pTF4-transfected cells revealed an increased amount of F7 mRNA compared to control cells, which corresponded to a 3.6 ± 0.3-fold increase upon qPCR (Fig. 2a). Intrigued by this observation, we investigated the impact of TF4 on FVII protein expression. For untransfected HepG2 cells, FVII expression is directly related to the cell number (~1 and 2 ng/ml from 0.5 × 10 6 and 1.0 × 10 6 cells, respectively) (Fig. 2b, blue bars). In medium with 0.5 × 10 6 cells transfected with pTF4, we observed approximatively a two-fold increase of FVII antigen levels (Fig. 2b, red bar), which was consistent with the extent of the increase in the F7 mRNA levels. Most importantly, the expressed FVII was enzymatically active, as measured by a very sensitive functional fluorogenic assay in plasma that evaluates the activation of factor X, the natural substrate of FVII 21 . In this assay, the activity of FVII responsible for triggering the coagulation cascade process inversely correlates with lag time 22 . As shown in Fig. 2c, in medium with 0.5 × 10 6 cells transfected with pTF4, we detected an increase of secreted protein levels with a concurrent shortening of lag times relative to untransfected cells (42 vs 50 minutes). These results were comparable to those from 1.0 × 10 6 control cells (Fig. 2c). This coherent increase in F7 mRNA, secreted protein and activity levels observed upon pTF4 treatment appears to be remarkable in the light of the very poor transfection efficiency (approximately 5%) of HepG2 cells, as estimated by FACS analysis (GFP tracer encoded by the pTF4 vector; data not shown).
Certainly, the efficacy of TALE-TFs strongly depends on the chromatin status and the accessibility of the target regions 23 , and this issue is one that cannot be properly assessed in hepatoma cells and through reporter gene assays exploiting episomal vectors. Thus, we transduced Hep10 human primary hepatocytes using adeno-associated viral vectors expressing the entire TF4 (AAV8-TF4) or a variant lacking the DNA binding domain (AAV8-Δ DBD) as a control (Fig. 2d).
The RNA analysis showed a 3.7 ± 0.33-fold induction of F7 gene expression in AAV8-TF4 versus AAV8-Δ DBD transduced Hep10 cells (Fig. 2e). FXa generation assays in medium revealed a significant shortening of lag times between AAV8-TF4 and AAV8-Δ DBD transduced cells, indicating an increase in the levels of secreted FVII protein able to promote the generation of FXa in plasma systems (Fig. 2f).
Similar to HepG2 cells, the transactivation of the endogenous F7 transcription in Hep10 cells may be limited by competition with the HNF4 transcription factor. Additionally, transactivation could be further limited by the rate of both infection and transgene expression of AAV vectors in vitro 24,25 . However, from a translational therapeutic perspective, the use of AAV is very promising [26][27][28] .
To investigate differences in the target region accessibility caused by different chromatin statuses, the effect of TFs was also explored in human embryonic kidney 293 cells (HEK293). Because HEK293 cells are of kidney origin, they lack the liver-specific HNF4 transcription factor and do not significantly express FVII. This system enabled us to provide insights into both the TF4 mechanism of transcription activation and the steric hindrance hypothesis in HepG2 cells for the TFs 1-3. Even in this experimental setting, only TF4 was able to significantly increase the transcriptional levels of the endogenous F7 gene (Fig. 3). These findings point to a mechanism in which TF4 activity is not mediated by HNF-4, in accordance with the known ability of VP64 to recruit the basal transcriptional machinery 1,29 .
Moreover, these data support the hypothesis that TF4 activity depends on its position and the accessibility to the target region. In fact, in the absence of the HNF-4 transcription factor, the single TFs 1-3, which previously inhibited the activity of the FVII-luc wt construct potentially through competition with endogenous transcription factors, were unable to trans-activate F7 gene expression in HEK293 cells (Fig. 3). Noticeably, by transfecting pTF4 in combination with the other TALE-TFs, we detected an additionally increased expression of the endogenous F7 (> 20-fold), which was modest compared to that in untreated HepG2 cells. Consistently, the combination of TF1-3 was ineffective (Fig. 3).
Taken together, the observed synergistic effects note that TF1-3 could bind to their target, but they may bind too far from the transcription start site to enhance gene expression.
Target specificity is a crucial issue when proposing a new therapeutic approach. Thus, we assessed our designed TALE-TFs in multiple control experiments for off-target effects. The key role of the TALE-DNA binding domain was to guarantee specific targeting. This specificity was demonstrated in control experiments where the VP64 activation domain was expressed alone. Negligible effects on the expression of the pHD plasmid were observed in these experiments (Fig. 1c). In complementary experiments, TF4 was unable to increase the transcription of neither the pBasic construct, which does not have a promoter, or the pSlug vector, which possesses a ubiquitous and F7-unrelated Slug promoter 30 (Fig. 4a). Most importantly, the TF4 efficacy was abolished upon the partial mutagenesis (five out of the 22 nucleotides) of the target sequence (Fig. 4a, inset), either in the wild-type (pFVII-luc wtΔ5 ) or mutated (pFVII-luc −61Δ5 ) contexts, a finding that supports high specificity and a reduced probability of off-target effects (Fig. 4a). The F7 model also offered us the opportunity to test, in the context of chromatin, the specificity of TF4 for activation of F7 over the neighbouring and homologous F10 gene, which is located approximately 2.1 kb downstream of the F7 locus (Fig. 1a) and exhibits a 15-fold higher expression level than FVII. Our results with RT-PCR and qPCR analyses indicated that TF4 exerted no effect on F10 transcription in both HepG2 (Fig. 4b) and in HEK293 cell lines (Fig. 4c).
To investigate in detail the possibility of TF4 off-targeting, we tested the expression of the four genes showing the highest score of off-target prediction (supplementary Table S1). qPCR performed in pTF4-treated versus untreated HepG2 indicated no significant influence on these targets (Fig. 4d), which rules out unspecific effects on the most probable candidates.
Taken together, these results obtained with complementary approaches demonstrate the ability of TF4 to specifically enhance the F7 gene promoter function, which leads to a concurrent increase of secreted FVII protein and activity levels. Taking into account the very low therapeutic threshold for FVII deficiency 31 , this enhancement could rescue coagulation and reduce haemorrhagic diathesis in severe patients. It is worth noting that none of the promoter mutations identified so far in FVII deficient patients (http://www.hgmd.cf.ac.uk) fall within the region targeted by the active TF4. This observation implies that TF4 is applicable for all F7 promoter mutations; thus, cohorts of patients could benefit without any modification of the TALE DNA binding domain.
We believe that these novel data lay a robust foundation for the further manipulation of the TALE-TF construct to control the strength or mechanism of the activation domain (i.e., number of VP repeats 13,32 or different domains 5 ) as well as its expression levels (promoter strength) and tissue specificity (i.e., liver specific promoters 33 ). Although we provided evidence for the successful use of TF4 delivered by an AAV vector with liver tropism only in a cellular model 34 , this result could lead to optimized TF4-derivatives that could be used to increase specificity and prevent possible FVII over-expression and consequent thrombotic risks.
In conclusion, to the best of our knowledge, this study provides the first experimental insights into the ability of a single engineered TALE-TF to rescue, with negligible off-target effects, gene transcription impaired by two promoter mutations. If delivered in vivo, these results could open new therapeutic strategies for coagulation factors as well as other genetic diseases sharing similar pathogenic mechanisms.

Methods
Design and cloning of the expression vectors. To create the pFVII-luc wt vector, the F7 proximal promoter region (520 bp) was amplified with primers 5′ AAAAAGCTTCTGTGGCTCACCTAAGAAACCAG 3′ and 5′ AAAAAGCTTGATGAA ATCTCTGCAGTGCTGC 3′ and cloned upstream of the firefly luciferase sequence into the pGL3 Basic Vector (Promega, Madison, USA) using the Hind III restriction sites (underlined).
TALE-TFs (named pTF1-4) were designed using the TAL Effector Targeter webtool 19 entering the 150 bp of the minimal promoter region of the F7 gene, scanning for repeat arrays of 18-26 TALE and T as an upstream base and unchecking all the other options.
Four TALEs binding to the minimal promoter sequence and selected upon the minimal number of off-targets were created as described 32 .
Cell cultures and reporter gene assays. The FVII studied in this work is of liver origin, so the reporter gene assays were conducted in hepatoma HepG2 and HuH7 cells. Cells were seeded into 12-well plates (0.2 × 10 6 cells/well) and transfected using Lipofectamine 3000 (Life Technologies, Carlsbad, USA) with 1 μ g of each pFVII-luc variant and/or 1 μ g of each pTF or pHD, expressing the VP64 alone, as a control. To normalize for the transfection efficiency, the cells were co-transfected with 100 ng of pRluc (Promega) expressing Renilla luciferase. Forty-eight hours post-transfection, cells were lysed to evaluate the luciferase activity by the Dual-Luciferase assay (Promega).
To evaluate transfection efficiency, FACS analysis was conducted in HepG2 cells transfected with TALE-TFs vectors, which express GFP fused to the 3 ′ of the TALE-TF genes by a self-cleaving P2A linker.
AAV Plasmid construction and packaging. TF4 or the Δ DBD (TF without the DNA binding domains) backbones were cloned in the AAV vector plasmid under the liver specific chimeric promoter of human alpha-1-antitrypsin and cer/hepatic locus control region 25 and packaged by ViGene Biosciences, Inc., as AAV8 strain. The titers reuslted 3.61 × 10 14 GC/ml for the AAV8-TF4 and 3.15 × 61 × 10 14 GC/ml for the AAV8-Δ DBD.
Primary hepatocyte transduction. Hep10 cells (human hepatocytes from ten donors) were obtained from Thermo Fisher Scientific (Waltham, USA) and were cultured according to the manufacturer's instructions. Briefly, one vial of Hep10 cells was thawed, diluted in Cryopreserved Hepatocytes Recovery Medium (CHRM), centrifuged at 100 rcf for 10 minutes at 4 °C and resuspended in Williams' Medium E (1x, no phenol red) supplemented with the Hepatocyte Maintenance Supplement Pack. Twenty-four hours later, cells were counted and seeded in 12-well plates at a density of 0.3 × 10 6 cells/ml in Williams' Medium E and the Supplement pack as described above, with the addition of Vitamin K 5 μ g/ml and the AAV8-TF4 or pAAV8-Δ DBD at 1000 MOI. Forty-height hours later, the cells and medium were collected for the analysis of RNA and FVII activity levels.

Evaluation of FVII protein levels and activity.
FXa generation assays were analysed by the statistical software GraphPad Prism 5 (GraphPad Software, San Diego, USA). The specific parameter lag time (expressed in minutes) in FXa generation assays was extrapolated from the first derivative of relative fluorescence units (RFU) as a function of time (minutes).
Off-target analysis. TALE-TFs off-target analysis was performed by entering the target region for each TALE-TF in the Target Finder software 19 with default options. The first ten off-target regions for each TF are reported with the prediction score in the supplementary Table S1. The off-target sites for the TF4 were analysed by qPCR as described above, with the following primers: ABR_F, CTGCTGAGGAAACACCACAG; ABR_R, CCCCACAACTTGCTCCTAAG; LRRC3B_F, GCCCAAGCAAGGAAAGAAAT; LRRC3B_R , GGTAAAGGT TCCAGCCCTGT; ARID1B_F, GCGTGTGCAGGAGTTCAATA; ARID1B_R, GGAATTTCCATCTTGCTCTCA; and MYO5C_F, TCCTAAATAGCCGGGAGGAT; MYO5C_R, CTGCTGGGGATCTGAATCAT.