LAUSR

We are grateful for the opportunity to respond to the letter [1] from anonymous employees of MeiraGTx in relation to our ‘Opinion’ article, in which we discuss gene therapy approaches for RPGR-related retinitis pigmentosa (RP) [2]. The RPGR vector being tested by MeiraGTx in human clinical trials contains a large in-frame deletion (378 base pairs), resulting in a loss of just over one-third of the entire ORF15 coding sequence [3]. The mouse is not an ideal model in which to test the efficacy of a shortened human RPGR-ORF15 vector for two reasons. First, the degeneration in the mouse is very mild compared to humans and so might be ameliorated by a weaker vector. Second, the mouse ORF15 region is shorter than the human equivalent. Hence, a deleted human ORF15 may be closer to the mouse sequence than that of the human. Since a larger inframe deletion of the ORF15 region fails to rescue even the mouse phenotype [3], then it is natural to ask whether or not the 378 base pair deleted variant is as effective as the fulllength RPGR sequence in humans. Our article is biased only in the sense that we could not find any beneficial reason for removing this critical part of the RPGR coding sequence before testing it in patients. MeiraGTx has provided no argument as to how this shortened variant might be better than or even as good as the full-length protein. The vectors developed by AGTC and Biogen both contain the full-length RPGR sequence. Whilst it is true that small in-frame deletions of up to 36 base pairs have been detected on rare occasions [4], the effect of an in-frame deletion 10 times larger than this is unknown, because deletions of this size have never been detected in the human population. There is also no logic as to how this might stabilize the RPGR coding sequence, because repetitive purine (GA) sequences will still exist flanking the deletion and the intron 15 splice donor site (which might lead to aberrant splicing [5]) has not been inactivated. Shortened RPGR coding sequences with in-frame ORF15 deletions frequently occur as a result of cloning errors due to the repetitive purine repeats in this region. Of course, we do not in any way suggest that the deleted MeiraGTx sequence arose due to a cloning error, but we do note that AGTC tested a shortened version of hRPGR (with only a 45 bp deletion within ORF15) in their preclinical program, which was claimed to have arisen during the cloning process [6,7]. AGTC decided to improve on the shortened RPGR by rational design of the coding sequence using codon optimization in order to create a full-length stable transcript. A side-by-side comparison of both transgenes in RPGRmutant dogs favored the codon optimized full-length version for their clinical trial [8]. The ORF15 region of RPGR encodes glutamate and glycine residues in the RPGR protein. This region undergoes extensive post-translational glutamylation, which is believed to be essential for cone function. Absence of RPGR glutamylation, as occurs in deficiency of the TTLL5 enzyme, leads to cone or cone-rod dystrophy [9,10]. We have also noted that RPGR patients who have the cone dystrophy phenotype have almost full-length RPGR protein but with markedly reduced glutamylation [11]. An RPGR vector containing a deletion similar to the vector developed by MeiraGTx also has significantly reduced glutamylation compared to the wildtype sequence (~70% reduction, as shown in Sun et al., 2016 [12] – Figure 5C, sample 4), and naturally this raises questions about its efficacy in cone photoreceptors. Independent improvement in cone function has not yet been shown in any of the preclinical animal models because they lack a macula and cone function will be sustained indirectly by secondary mechanisms when rod degeneration is slowed. Independent improvements in cone function following RPGR gene therapy have so far only been confirmed when applying the full-length RPGR sequence in the human macula [13]. Hence, it will be critically important to monitor the outcomes of the clinical trials to determine if the deleted RPGR vector has an equally positive effect on human cone photoreceptors. In any case, we believe that investigators and patients being recruited into clinical trials should be made aware that the MeiraGTx RPGR sequence being administered to them has a significant deletion within it and that the consequences of this are currently unknown.