Zinc-finger-based transcriptional repression of rhodopsin in a model of dominant retinitis pigmentosa

Despite the recent success of gene-based complementation approaches for genetic recessive traits, the development of therapeutic strategies for gain-of-function mutations poses great challenges. General therapeutic principles to correct these genetic defects mostly rely on post-transcriptional gene regulation (RNA silencing). Engineered zinc-finger (ZF) protein-based repression of transcription may represent a novel approach for treating gain-of-function mutations, although proof-of-concept of this use is still lacking. Here, we generated a series of transcriptional repressors to silence human rhodopsin (hRHO), the gene most abundantly expressed in retinal photoreceptors. The strategy was designed to suppress both the mutated and the wild-type hRHO allele in a mutational-independent fashion, to overcome mutational heterogeneity of autosomal dominant retinitis pigmentosa due to hRHO mutations. Here we demonstrate that ZF proteins promote a robust transcriptional repression of hRHO in a transgenic mouse model of autosomal dominant retinitis pigmentosa. Furthermore, we show that specifically decreasing the mutated human RHO transcript in conjunction with unaltered expression of the endogenous murine Rho gene results in amelioration of disease progression, as demonstrated by significant improvements in retinal morphology and function. This zinc-finger-based mutation-independent approach paves the way towards a ‘repression–replacement’ strategy, which is expected to facilitate widespread applications in the development of novel therapeutics for a variety of disorders that are due to gain-of-function mutations.

Thank you for the submission of your manuscript 'Zinc finger-based transcriptional repression of rhodopsin in a model of dominant retinitis pigmentosa" to EMBO Molecular Medicine. We have now received the three enclosed reports on your manuscript. You will see that they find the topic of your manuscript potentially interesting but feel that the data need to be strengthened. This should be convincingly addressed in a revision.
The reviewers' reports are detailed and explicit and I will not repeat their points of concern here. I would however like to stress that referee #3 highlights the need for additional controls regarding the electroretinography and AAV infection efficiency and suggests experiments to address this question. We feel that additional insight in this regard would strengthen the findings. In their article "Zinc finger-based transcriptional repression of rhodopsin in a model of dominant retinitis pigmentosa.", Mussolino et al. describes an approach to decrease the level of mutant rhodopsin in a mouse model of an autosomal dominant form of retinitis pigmentosa (adRP) using zinc fingers to direct a transcriptional repressor to the promoter region of rhodopsin. They show in vitro that the decrease in the level of the mutant protein decreases the rate of death of ES cells differentiated towards the photoreceptor fate. The authors also demonstrate in vivo in a mouse model of adRP that zinc finger targeted repressor mediated lowering of mutant rhodopsin using adeno associated viruses leads to improved retinal function reported by ERG. This is a nice paper that demonstrates an alternative method to siRNA-based approaches. The authors use a number of clever experiments (ES cells for example) and clearly demonstrate the zinc finger based concept. The in vivo achieved decrease in mutant rhodopsin is modest and but interestingly leads to a complete rescue of the ERG signals in the time frame studied. In the future (not necessary for this paper) I would advise the authors to extend the investigations to older mice.
Referee #2 (Novelty/Model system Comments for Author): The work appears to be of high technical standard. Medium novelty: the concept of suppression replacement gene therapy has been around for a long time. However, the method used, zinc finger nucleases, has not been reported previously in mice to my knowledge. Medical impact is high: gene therapy for dominant diseases is a different proposition to that of gene replacement (ie for recessive disease). It is an important aspect which will have high medical impact if successfully and efficiently developed. The paper addresses this. The model system is adequate and appropriate Referee #2 (Remarks to the Author): The paper addresses a major issue in regard to gene therapy for autosomal dominant conditions, focusing in this instance on a degenerative disease of the retina, retinitis pigmentosa. A significant proportion of cases of this disease, generally regarded as the most important cause of blindness in the working age-group, are passed on in a dominant fashion and in those cases about 25% of patients have mutations within the rhodopsin gene. While gene therapy could, in principle, involve directing a means of suppression to a given mutation, even if this could be achieved, there are, as the authors correctly point out, a very large number of mutations within the rhodopsin gene that would require targeting. Hence, going for individual mutations is not really an option. The concept of 'suppression-replacement', or as the American groups have christened it, 'ablate-and-switch', is not a novel concept, papers addressing this concept having appeared as long as 13 years ago. However, all of these approaches so far as I can determine, have used either ribozymes, or shRNAs to effect suppression, whereas in this paper, the authors use zinc finger nuclease technology to target the DNA directly. They convincingly show in cells and in living mice that the technique has efficacy. The mouse model that they use, expressing a mutated human rhodopsin transgene, P347S, is reduced to 5 or 6 rows of photoreceptors at P70. Animals were injected with AAV expressing the zinc finger nuclease at P4. Retinas were then examined at p18 or P30 according to materials and methods. A very important figure in this paper is figure 3d, where retinal sections are shown of treated and control animals. They do not specifically say when this analysis was done, although one infers that it was P90. This should be more clearly stated. In this figure, control retinas are reduced to a single row of photoreceptor nuclei, so they must be looking well beyond P70, whereas treated animals have three, maybe up to 4 rows. In my opinion, this is an encouraging result. They do not say directly how many retinas were looked at -I presume they looked at the same number as in the ERG analysis, ie, 12, but this should be stated. Is figure 3d representative of protection attained in all of the retinas examined? I think that this is an important question and there should be some comment in the paper about it. As I said above, the innovation in this paper is not in the concept, but in the novel procedure that has been used. I do not recall seeing any other papers in which retinal protection was obtained using this suppression-replacement approach and zinc finger technology in live mice. The authors spend some time comparing the two parallel methods of doing this sort of gene therapy, ie using RNA interference to suppress transcripts or to target the DNA directly, and I agree with their comments. Their strategy has the advantage that sustained expression of shRNA is not required. This may have practical significance in that a growing body of evidence suggests that expression of shRNA in mouse retinas for long periods could result in TLR3 activation with activation of NFkß and type 1 IFs.

Referee #3 (Remarks to the Author):
This is a well-written paper describing some very interesting work on the use of Zn-finger repressors as a potential treatment for autosomal dominant retinitis pigmentosa. As the authors point out, the genetic diversity of this disease calls for a technique that will suppress the production of all endogenous rhodopsin, and its eventual replacement with a wild-type gene that is insensitive to the repressor protein. In this case, they demonstrate that the endogenous mouse rhodopsin promoter is resistant to the Zn-finger repressors they designed, so that blocking the expression of the human transgene was sufficient. Therefore, this paper represents step one in the eventual gene therapy.
The in vitro results, including the use of retinal progenitor cells derived from the transgenic mice, were carefully performed. The data from gene delivery of mice has two major shortcomings, however, that they should address either by more accurate analysis or by additional experiments. First, the electroretinography results in Fig. 3C should have compared the EGFP treated eyes with ZF-R6 treated eyes of the same mice. They are correct that variability between mice can be considerable, but the two eyes of a single untreated mouse usually exhibit the same ERG response. Second, they never discussed what extent of the retina was productively infected by AAV so that the partial preservation of photoreceptor cell nuclei seen in Fig. 3D might represent just a small fraction of the retina. A typical approach is to measure the retinal thickness along a vertical meridian, so that a pan-retinal view of protection can be obtained.
Specific comments: p. 6 Table S1 appears to be identical to Table 1. p. 7 "In vitro selection of functional ZF-Rs." "Selection" should be "screening". p. 7 They report too many significant digits for their p values. p. 8 "showed binding specificity of ZF-R2 and ZF-R6 (Fig. 1e)" Fig. 1e presents data for only ZF-R6.
p. 9 "Despite the cell death observed in retinal precursors, the levels of hRHO transcript were very low." Low levels of hRHO mRNA are to be expected if expression of mutant RHO leads to apoptosis--the interesting cells were dead or dying.
p. 10 and Fig. 3 "Quantitative real-time PCR analysis demonstrated that the EGFP positive portion of the retina treated with AAV2/8-RHOK-ZF-R6 displayed a significant ~26% reduction of the hRHO transcript" Was there a significant impact if the whole retina was measured? As ERG controls, the authors should have used the EGFP injected eyes form the same mice.
p. 10 "both scotopic and photopic ERG amplitudes were measured (Supplementary Table 1 p. 27 legend to Fig. 3: This figure should be more carefully described, given that this is not an ophthalmology journal. There are no white circles. It is inappropriate to cal this "the kinetics" of visual function loss with only 2 points: before and after. ERG is not a measure of visual function--it is a measure of the electrophysiologic response of the retina. Whether the mice can see or not requires a behavioral test that is beyond the scope of this paper. Answers to the Referees' comments:

Reviewer 1: This is a nice paper that demonstrates an alternative method to siRNA-based approaches. The authors use a number of clever experiments (ES cells for example) and clearly demonstrate the zinc finger based concept. The in vivo achieved decrease in mutant rhodopsin is modest and but interestingly leads to a complete rescue of the ERG signals in the time frame studied. In the future (not necessary for this paper) I would advise the authors to extend the investigations to older mice.
We wish to thank the Referee for these supportive comments, and as the Referee advises, to determine the stability of the treatment over time, we are currently extending the characterisation of these gene therapy effects to cohorts of older mice.

Reviewer 2: They convincingly show in cells and in living mice that the technique has efficacy. The mouse model that they use, expressing a mutated human rhodopsin transgene, P347S, is reduced to 5 or 6 rows of photoreceptors at P70. Animals were injected with AAV expressing the zinc finger nuclease at P4. Retinas were then examined at p18 or P30 according to materials and methods. A very important figure in this paper is figure 3d, where retinal sections are shown of treated and control animals. They do not specifically say when this analysis was done, although one infers that it was P90. This should be more clearly stated. In this figure, control retinas are reduced to a single row of photoreceptor nuclei, so they must be looking well beyond P70, whereas treated animals have three, maybe up to 4 rows. In my opinion, this is an encouraging result. They do not say directly how many retinas were looked at -I presume they looked at the same number as in the ERG analysis, ie, 12, but this should be stated. Is figure 3d representative of protection attained in all of the retinas examined? I think that this is an important question and there should be some comment in the paper about it.
We apologize for the confusion regarding the timing and number of animals analysed in the previous version of our manuscript. In this revised version, as indicated by the Referee, we have clearly stated the timing and numbers included. We did however erroneously indicate previously that at P60 (30 days after treatment) we were able to detect differences between zinc-finger-treated and EGFP controls. In this revised version of our manuscript we thus removed this statement. During the review of the manuscript, we had the opportunity to analyse by both ERG analysis and histology a further group of mutant animals, to extend the functional and morphological protection effects of the zinc-finger treatment. The ERG analysis has now been performed at P90 in three independent cohorts of mutant animals (in the previous version of the manuscript there were two) for a total of 16 animals (16 eyes treated with RHOK-ZF-R6 and 16 RHOK-EGFP controls). Histological analysis was performed between P90 and P100 in 9 eyes, including the novel cohort of animals. We observed protection in 7 out of 9 treated eyes. We have replaced Figure 3d with a new panel depicting the transduced area as assessed by fundus photography and low magnification microscopy. In the same panel, we have also highlighted (with a "zoom in") the protection effect that is restricted to the portion of the retina exposed to the ZF-R6 vector.

Reviewer 3:
The data from gene delivery of mice has two major shortcomings, however, that they should Fig. 3D might represent just a small

fraction of the retina. A typical approach is to measure the retinal thickness along a vertical meridian, so that a pan-retinal view of protection can be obtained.
Following the Referee's concerns here, we have better explained how the electroretinography (ERG) experiments were performed. In connection with this, the original ERG analysis was a comparison between the EGFP-and ZF-R6-treated eyes of the same mice. In the original version of the manuscript, which was reviewed, we separated the ERG analysis of the zinc-finger-from the EGFPtreated eyes to better appreciate the differences of the electrophysiologic responses of the retina (preservation or loss of retinal function) in the same eyes, over time. We acknowledge, however, that the data showed in this way were possibly misleading and confusing. In this revised version of our manuscript, we now show in the same graph: base-line recording (post-natal day P30) of both eyes (ZF-R6 vs EGFP of the same animals) and the ERG responses of the same eyes 60 days after treatment (P90). In this way, it is possible to compare zinc-finger-and EGFP-treated eyes before and after treatment, and to appreciate the significant differences between zinc-finger and EGFP-treated eyes. In addition, since we collected further data from an additional study group, we have now included these new data with the previous data. Therefore, the data of the original graph (Figure 3c, upper and lower panels) are now merged into the graph shown below (next page), whereas the graph with all of the study groups, and hence including the novel additional one, is now shown as Figure  3c of the revised version of our manuscript. For the efficiency of the retinal transduction, we have added in this revised version of our manuscript a photograph that shows the typical transduced area following AAV vector subretinal administration, and a pan-retinal histological view of the protection. We have highlighted the effects of this gene therapy in a magnified portion of this pan-retinal view, which depicts the protection from retinal degeneration only in the transduced area.
bWave amplitudes under scotopic and photopic conditions in P347S mice. The amplitudes evoked by increasing light intensities under scotopic and photopic conditions in the P90 mice treated in one eye with EGFP (green squares, n = 12 eyes) and in the contralateral eye with ZF-R6 (red triangles, n = 12 eyes). The black circles represent the ERG responses of the same animals at P30, before treatment (n = 24 eyes [12+12]).
Specific comments: p. 6 Table S1 appears to be identical to Table 1.
We apologize for the error, and Table S1 has been removed.

p. 7 "In vitro selection of functional ZF-Rs." "Selection" should be "screening".
Selection has been replaced with screening. p. 7 They report too many significant digits for their p values.
We have added "data not shown" for ZF-R2.

p. 9 "Despite the cell death observed in retinal precursors, the levels of hRHO transcript were very low." Low levels of hRHO mRNA are to be expected if expression of mutant RHO leads to apoptosis--the interesting cells were dead or dying.
We have removed this misleading sentence. Indeed, as stated two lines below (p. 9 line 12), we normalised the human rhodopsin expression levels relative to Rho+, to discriminate between cells induced to express rhodopsin from those that were undifferentiated. p. 10 and Fig. 3

"Quantitative real-time PCR analysis demonstrated that the EGFP positive portion of the retina treated with AAV2/8-RHOK-ZF-R6 displayed a significant ~26% reduction of the hRHO transcript" Was there a significant impact if the whole retina was measured? As ERG controls, the authors should have used the EGFP injected eyes from the same mice.
As we have stated (p. 10): "significant ~26% reduction of the hRHO transcript relative to the endogenous mRho compared to contralateral EGFP-treated retinae", the comparison was performed on the same animals between the EGFP vs ZF-R6 treated mice. p. 10 "both scotopic and photopic ERG amplitudes were measured (Supplementary Table 1, Fig.  3c)" Supplementary Table S1 is the same as Table 1

and no ERG amplitudes are reported in a table.
We have now uploaded as Supplementary Materials both Supplementary Table S1and Figure S5.

p. 15. Reporter Constructs (and elsewhere): No Supplementary Materials and Methods were provided.
We have now provided a file with the Supplementary Materials.

p. 27 legend to Fig. 3: This figure should be more carefully described, given that this is not an ophthalmology journal. There are no white circles. It is inappropriate to call this "the kinetics" of visual function loss with only 2 points: before and after. ERG is not a measure of visual function--it is a measure of the electrophysiologic response of the retina. Whether the mice can see or not requires a behavioral test that is beyond the scope of this paper.
We have corrected the manuscript accordingly to these suggestions of the Referee: in the text on p. 10 and in the legend to Figure 3.