In this manuscript we present a novel pEPI-vector derivative, named pEPito, with higher transgene expression levels and increased colony-forming efficiencies in vitro, and more persistent transgene expression profiles in vivo. In order to compare the novel pEPito-backbone with the backbone of the original vector, pEPI-1, a series of 13 novel vectors was constructed. Eight vectors, comprising an identical transgene transcription unit, i.e. EGFP-IRES-BSD, were generated to evaluate vector-backbone performance in vitro. These eight vector constructs were subsequently selected for BSD expression, present in the first transgene transcription unit that is part of the pEPI-vectors' episomal plasmid replicon (Figure 1). This is noteworthy, since all previous pEPI-1 based vectors were selected for the NPT gene, present in the second dispensable SV40-O/P driven transcription unit [1, 3, 4]. It is also worth mentioning, that in previous studies the expression of the first, CMV-IEP promoter driven transgene transcription unit has not been studied extensively, since main focus was laid on the analysis of the vectors' episomal replication potential, but not on transgene expression levels [3, 7–9]. In these studies partial, i.e. background, integration of few pEPI-vectors was never fully excluded, but the episomal maintenance and replication was described [1–4, 6–9]. When assaying all novel EGFP-IRES-BSD cassette containing vectors, the CpG-depleted, pEPito based constructs constantly performed better than the pEPI-1 based constructs, with respect to transient transfection efficiencies, transient EGFP expression levels, EGFP expression levels within stably selected cells, and colony-forming efficiency (Figure 3, 4). Since all experiments were performed by transfecting equal amounts of vector DNA - and not equal molar ratios of vector molecules,- to exclude side effects of stuffer DNA, some of the differences within the increased transient transfection efficiencies of the smaller plasmids are caused by this. pEPito-based vectors are on average some 25% smaller than their respective pEPI-1-based counterparts and as a consequence ~one third more pEPito vector molecules as compared to its direct pEPI-1 counterpart had been transfected initially. The differences in transient EGFP expression, the EGFP expression levels of stably selected mixed-clones, and the 7-70 fold increase in colony-forming efficiency are impressive and not caused by the different equimolar levels of the transfected DNA, which differ at a maximum of 33%. The known establishment efficiency for pEPI-1-replicons is about 1-5% . In our modified experimental setup this efficiency was lower, considering the colony-forming assay, the initial transfection rate and cell growth, and resulted in about 0,25% of the transfected and selected cells for the pEPI-1- [CMV]-construct, and - ~6 fold higher - about 1,8% for the comparable pEPito- [CMV]-construct in HEK 293.
Increased transgene expression levels and longer lasting transgene expression in vitro and in vivo have also been reported for so-called minicircles which lack any residual elements necessary for bacterial propagation [33, 34], or for CpG-depleted or CpG-free vectors that have been constructed depleted or free of any CpG motives [35, 36]. It was shown, that the CpG content of a minicircles' transgene transcription unit does not affect the epigenetic silencing of its transgene , in contrast to epigenetic elements present in the bacterial origin of replication or backbone . Chen et al. speculate that there might be a recruitment of repressive heterochromatin as inducer of a silencing complex that is initiated within bacterial vector elements , implying an important role of the chromatin structure for epigenetic silencing. As bacterial elements usually contain CpG motifs, the known effects of CpG-isles and the work of Chen et al.  are neither identical nor contradictory, but these effects might be linked at some points in our understanding. Further research has to elucidate, if and how these facts are connected. No or only weak silencing effects have been described within CpG-free plasmids [17, 21, 35]. Our novel pEPito vector, which consists of a rather CpG-rich transgene transcription unit within a CpG-depleted bacterial vector backbone, seems to exhibit similar expression profiles as the described minicircles and CpG-free vectors.
The CMV-IEP promoter element is frequently used within commercial plasmid vectors, since it confers robust expression of a transgene in several cell types. The expression profile of CMV-IEP, however, decreases steadily over time, which could be explained by methylation of cytosines within this promoter region . For pEPI-1 based vectors methylation of cytosines within the CMV-IEP promoter region has been reported to be inhibited by the vector encoded MARS element in CHO and HaCat cells in vitro . In contrast, Argyros et al. described silencing of a CMV-IEP promoter driven luciferase transgene within a pEPI-1 vector backbone after hydrodynamic gene delivery in mice. Within the latter study, however, the cytosine methylation of the CMV-IEP promoter element has not been investigated in detail . Taken together, no cytosine methylation of CMV-IEP promoter elements within pEPI-based vectors has been observed so far in vitro and in vivo. Therefore, a possible explanation for the silencing of transgenes regulated by the CMV-IEP promoter element within a pEPI-vector backbone might be histone deacetylation as proposed by Papapetrou et al. .
When comparing the CMV-IEP and hCMV/EF1P promoter elements within our constructs (Figure 3, 4), it is obvious that the relative promoter strength has a strong influence on vector performance. In case of stably selected HEK293 cells, CMV-IEP promoter based vectors exhibited a stronger transgene expression - as a measure of promoter strength - as vectors based on the relatively weaker hCMV/EF1P promoter element (Figure 4A). This situation, however, is inverted in stably selected NIH3T3 mixed-clones, in which the hCMV/EF1P promoter element results in higher transgene expression levels than the CMV-IEP (Figure 4B). We assume that increased transgene expression levels reflect increased transgene mRNA transcription levels, despite this has not been experimentally verified so far.
In both, HEK293 and NIH3T3 cells, the colony-forming efficiency of vectors with a functional pEPI-plasmid replicon, i.e. all MARS containing vectors, seems to partially correlate with the relative promoter strength (Figure 4). The rather weak performance of the CMV-IEP promoter element within NIH3T3 cells has been already described by Nehlsen et al. . These findings suggest, that the CMV-IEP promoter element seems to be extraordinarily susceptible for epigenetic silencing within NIH3T3 cells. For the latter cell line rapid silencing of exogenous plasmid DNA by histone deacetylation has been frequently observed [42, 43]. The hypothesis presented in Chen et al. , i.e. the formation of heterochromatin after histone deacetylation involvement [44–47] triggered by certain bacterial elements, may serve as an explanation for the better performance of pEPito based vectors with respect to transgene expression and colony-forming efficiency within stably selected NIH3T3 cells (Figure 4B, D), since the pEPI-1 vector backbones exhibit significantly higher numbers of CpG motives (Figure 2).
The positive effect of the MARS element within the different pEPI-vector backbones can be observed best within our colony-forming assays in vitro and within the expression profiles of our MF-1 mice in vivo. In all colony-forming assays the presence of a MARS as a functional component of the pEPI-vector plasmid replicon (Figure 1) increased the number of established colonies as compared to their respective control vectors without MARS (Figure 4C, D). This could be explained by the episomal replication of respective vectors, which could be further supported by bacterial rescue experiments (Table 1). The CpG depleted backbone also influences the colony forming efficiencies. As the pEPI-replicon needs the transcription into the MARS-Element to establish an episome, the higher transgene expression of the pEPito vectors might lead to a higher chance of an establishing event, resulting in higher colony forming unit numbers. In MF-1 mice the pEPito based vectors with the MARS elements resulted in prolonged transgene expression (Figure 5). Since liver cells are quiescent cells, episomal replication of pEPI-based vectors might not serve as a suitable explanation for the prolonged transgene expression. Above all, we showed by qPCR that all five hydrodynamically delivered vectors persisted at equal copy numbers of approximately 0.5 - 6 vector molecules per liver cell. The failure of rescuing of the pEPI-1 based vector could be due to the formation of heterochromatin as discussed for the in vitro results before. We favour the theory, that the prolonged transgene expression profiles of pEPito-based vectors in vivo results from a combination of (1) reduced epigenetic silencing due to the modified bacterial vector backbone [17, 20, 23, 35, 48], and (2) the presence of the MARS which might trigger the translocation of vector molecules to sites of active chromatin (Figure 1) , or enhance overall transcription levels [49, 50].