Here we conduct the first side-by-side comparison of CK30PEG NPs and AAVs carrying matched expression cassettes. In our dose-response assay we determined that NPs are less efficient per vg than AAVs, but that NPs produce gene expression levels similar in scale to those generated by AAV. Although self-complementary AAV vectors bypass the requirement for viral second-strand DNA synthesis, and therefore have a faster onset of gene expression than conventional AAVs, NPs still had earlier onset of expression (PI-2 vs. PI-7). Expression levels were similar for AAVs and NPs at PI-14, but at subsequent timepoints, GFP levels from AAVs were higher than those from NPs. However, if dose equivalency had been chosen based on dose-response results from PI-30, then no difference in PI-120 expression levels between these delivery systems might have been observed. Overall, our data from the dose-response studies indicate that NP-driven gene expression can be modulated by altering the dose and suggest that higher levels of NP-based expression can be achieved if needed. Importantly, the NP doses that we use, although higher than the doses of AAV, are stable, and easy to manufacture 
. Although future studies may examine the immune response to nanoparticles in more depth, importantly, we have demonstrated that the doses we use here are well tolerated and do not induce an inflammatory response after delivery to the eye 
. In addition, we have shown that they are well tolerated even after repeat injection 
, a key feature if multiple dosing is therapeutically required.
Interestingly, although AAV2-CBA-GFP was expressed in all retinal layers while AAV5-MOP-GFP expression was limited to photoreceptors, these two treatments generated similar levels of GFP message and protein. This may be because AAV5 was reported to drive photoreceptor gene expression at higher levels than AAV2 
, thus leading to similar expression levels in animals treated with photoreceptor specific (AAV5-MOP) and ubiquitously expressed (AAV2-CBA) AAVs. However, we also observed that at PI-14 and PI-30 NP-CBA-GFP and NP-MOP-GFP exhibited similar levels of GFP expression suggesting that promoter strength also contributes to expression levels independent of delivery strategy.
Although persistent gene expression is critical for successful ocular gene therapy and has been elusive for non-viral vectors, here we show that NPs do not suffer from transient expression. Neither NP-MOP-GFP, AAV2-CBA-GFP, nor AAV5-MOP-GFP exhibit significant decreases in gene expression between PI-30 and PI-120 and fundus images and retinal sections examined at PI-360 demonstrate long-term expression in NP-MOP-GFP and AAV treated animals. However, it is critical to observe that the vector content/delivery vehicle affected the persistence of gene expression. AAV2-CBA-GFP injected animals express GFP for up to one year, while NP-CBA-GFP was silenced by PI-90. It is not clear why the CBA promoter drives persistent gene expression when delivered in an AAV, but transient expression when delivered as a NP, but several possibilities exist. Firstly, the DNA content of the two delivery systems is different. Although both the NPs and the AAV were generated from matched ITR plasmids, the NPs contain the entire plasmid while AAV production results in virions carrying only the expression cassette. Prokaryotic plasmid backbone element can influence gene expression 
, and may promote NP-CBA-GFP silencing. A second contributing factor may be differential methylation or epigenetic state between AAV and NP DNA arising from the biological source of the DNA: bacteria for NPs and HEK cells for AAV. Finally, NPs remain episomal, but AAVs can be episomal or integrated 
. Some of the AAVs may have integrated into the genome thereby promoting different regulation of gene expression than the NPs. Despite these differences between the delivery systems, our observation that NP-MOP-GFP can generate persistent retinal gene expression indicates that the delivery method per se is not responsible for the lack of long-term gene expression from NP-CBA-GFP.
These data emphasize the importance of proper selection and testing of vector elements, and it has been well established that the same gene with different promoters may have different therapeutic effects and safety profiles 
. For example, we previously demonstrated that NPs carrying pZeo-CMV-GFP had high expression in the eye and lung at PI-2 but were silenced thereafter 
. In contrast, in keeping with what we show here with NP-MOP-GFP, we have also observed persistent gene expression from other vector/promoter combinations. NPs carrying pcDNA-MOP-NMP (NMP-normal mouse peripherin/retinal degeneration slow) drove photoreceptor gene expression for up to 10 months (longest timepoint examined) 
, and our ongoing work has shown photoreceptor and RPE expression for up to 15 months and 2 years, respectively (unpublished data) with a variety of vectors. In the past, tissue-specific promoters have been successfully used not only in NPs but also in AAVs to target multiple retinal cell types 
_ENREF_51. The need for this tissue specificity in gene therapy vectors is highlighted by our observation that subretinally injected AAV leads to GFP expression in the visual pathways of the brain only when a ubiquitous promoter is used. It is difficult to conclusively say whether NPs could travel to the brain and drive gene expression since brain expression from AAV2-CBA-GFP was not observed until PI-90 days, a timepoint at which NP-CBA-GFP was silenced. However, our results from DNA amplifications () suggest that subretinally injected NPs do not leave the eye. In contrast to AAV2-CBA-GFP DNA which was detected in the eye, ON, and optic chiasm at PI-60 days, we detected NP-CBA-GFP DNA only in the eye.
The issue of AAV2-based brain expression after subretinal injection has been extensively studied yet remains controversial. Although subretinal delivery of rAAV2-RPE65 did not lead to vector or gene expression in the visual pathways in the brain of RPE65-mutant dogs, that study was only maintained for three months 
, a timepoint at which we just began to observe GFP expression in the mouse brain. In contrast, other groups showed that intravitreal or subretinal injection of AAV drives expression in the brain along the visual pathway in rat, dog, mouse, and pig 
_ENREF_53_ENREF_55. Delivery of AAV to the brain by ocular injection has even been used to mediate improvements in mouse models of lysosomal storage disease 
. Here we observe gene expression consistent with anterograde axonal transport from retinal ganglion cells (i.e. expression in ON, optic chiasm, and LGN) as well as expression consistent with trans-synpatic transport of the virus (expression in the VC). While others have also observed trans-synaptic transport of recombinant, replication deficient AAVs 
, the mechanisms that underlie this process are not understood. Ectopic expression and transmission of the virus from the target cell to other cells may be harmful 
and are undesirable from a regulatory standpoint.
In this side-by-side comparison study of AAVs and NPs we demonstrate that CK30PEG NPs can safely drive persistent gene expression (up to 1 year) after subretinal injection in adult mice. In addition, in contrast to AAVs, which were detected in the visual pathways of the brain, NPs remained in the eye. These NPs have several benefits for intraocular use; not only are they safe and non-toxic to the eye 
, but they have a much larger vector capacity than AAVs 
, features that make them a highly clinically relevant complement to AAV for ocular gene therapy and an excellent option for the delivery of genes that are too large for AAV.