Our detailed examination of the elution profiles result from the preparative ultracentrifugation of three specific rAAV8 variants, each carrying a unique transgene. These variants included both ssDNA and scDNA forms. This approach enabled us to differentiate between full and empty viral capsids, with isolation of specific fractions for further study based on their elution patterns, which were closely monitored using optical density measurements.
The assessment of in vitro biopotency for capsids initially labeled as "empty," utilizing vector genome titer as a metric, unexpectedly showed relevant potency levels in rAAV8 harboring ssDNA and scDNA variants. During the drafting of this manuscript, McColl-Carboni A et al., (2024) also reported comparable observations for a hematopoietic stem cell-derived AAVHSC-15, harboring an ssDNA transgene of around 3.9kb. Additionally, they noted the potency in partially empty particles [28]. This led us to redefine the term from "empty capsids peak" to "partially filled capsids," more accurately reflecting the ratio of vector genome titer to rAAV8 antigen content. In our analysis, the scDNA construct was identified as generating the most truncated forms. Although the scDNA is cleverly designed to self-assemble into double-stranded DNA (dsDNA) configurations without requiring DNA synthesis within the host cell, it was notably associated with a higher occurrence of truncated forms.
The analysis of the rAAV8-FIX-ssDNA variant revealed a significant peak in the elution profile, suggesting a majority of full capsids. This was confirmed by further analytical centrifugation, which also indicated a high level of capsid purity and potential efficacy in gene delivery, as evidenced by both the vector genome per capsid ratio and biopotency measurements. The short size of the FIX-ssDNA (2.6kb) transgene likely facilitated easy packaging into the rAAV8 particles.For the rAAV8-FIX-scDNA variant, our findings pointed to notable differences in elution patterns compared to the rAAV8-FIX-ssDNA variant, with a particular emphasis on the appearance of truncated DNA forms. This variant's profile suggested a complex mixture of capsid types across the fractions, with sedimentation analysis revealing shifts in capsid composition over the elution sequence. Interestingly, even fractions enriched with truncated forms exhibited significant biopotency, hinting at their unexploited potential in gene therapy applications to rethink the appropriate cutoff for therapeutic rAAV8 preparations.
Historical datasets of the rAAV8-FVIII-ssDNA variant were included in this study due to its unusual elution behavior, which aligned with observations from other variants. Initial fractions were predominantly full capsids, transitioning to a mix that included intermediate and empty capsids. Notably, a fraction mid-elution showed the highest full/empty ratio and biopotency, challenging preconceived notions about full/empty capsid ratio purity's role throughout the UCE elution peak.
In all preparations of rAAV8 containing the different types of transgenes, we found that the empty rAAV8 capsids had 50 Svedbergs, these results align with a study recently published [28], based on datasets generated from empirical data.
The DNA strand reconstitution model
In our study, we observed a full DNA band in alkaline agarose as represented in Fig. 5Figure 4, as well as additional truncated bands, mostly in rAAV8-FIX-scDNA. This observation led us to hypothesize the potential for reconstitution of these truncated forms. Notably, the biopotency analysis of these truncated forms revealed that even partial genomes retained activity, a finding that further supported our hypothesis.
In strains with ssDNA, as proposed in Fig. 5B for FIX-scDNA and Fig. 5C for FVIII-ssDNA models, it could be possible that DNA replication begins on one side, utilizing a substantial portion of the gene already present in the genome. We hypothesize that this may lead to potential reconstitution when the strand hybridizes with its opposite counterpart, ultimately yielding a complete end copy in the target cell or organism. Such a mechanism might feasibly restore biopotency in these truncated forms Fig. 5and introduce an additional possibility, suggesting a scenario involving truncated forms of ssDNA or scDNA with relatively high biopotency. In these instances, it is reasonable to anticipate that cellular polymerase would swiftly complement any missing segments.
This finding was particularly significant in the context of working with ssDNA, as with FVIII or FIX. Our data suggests that multiple infections of rAAVs into the host cell may occur. According to the literature, these rAAVs randomly pack plus and minus strands, which could naturally hybridize and result in the reconstitution of the full gene, similar as e.g. described by Ghosh et al., 2011 for hybrid dual AAV vectors carrying bridging sequences [29]. This observation is critical, as it was previously assumed that partial genes would not exhibit biopotency.
Furthermore, the analysis of fractions F07 and F08 from the rAAV8-FIX-scDNA, revealed that even partially packed material exhibited biopotency (Fig. 3B). This is attributed to the potential for reconstitution under certain conditions. This finding highlights the importance of considering not only fully formed genetic material but also truncated forms, which can still play a crucial role in biological processes (Fig. 5B). In the case of insect cells, degradation of the strands occurs: in our historical datasets where we compared the expression systems of the baculovirus sf9 insect cells with HEK293 cells (data shown in Supplementary Fig. 2) for the production of rAAV8, as can be seen in the agarose gel, both sf9 insect cells and HEK293 were capable of packaging DNA. Interestingly, when we digested the contents of the rAAV8 in both variants, we observed that the transgene content was intact in the rAAV8 particles produced in HEK293, but in the rAAV8 particles produced in sf9 cells, the genetic material including the ITR's ends was missing, and therefore, the transgene was incomplete. Surprisingly, we noted that although strand degradation can occur, the biopotency of the vectors produced might be unaffected. This indicates that, even with partial degradation of genetic material, the critical components necessary for therapeutic effectiveness remain. This preservation of biopotency, despite strand degradation, emphasizes the reliability of host cell systems in sustaining the operational integrity of viral vectors, demonstrating their value for gene therapy applications. We believe that about 20% of DNA sequences truncated in a rAAV product might not always be regarded as impurities. Hence, rigorous precision would not be required as some truncated sequences may have the capacity to reconstitute. For instance, for genes as lengthy as FVIII, this might be a viable option. Conversely, when dealing with capsids containing short-packed genes in rAAV, the focus might be solely on the full sequences. This observation shows some agreement with the recent publication by Troxell et al., in which partial rAAV can express genomes for the EGFP [30].
This is a topic worthly for discussion. Employing ultracentrifugation in combination with a chromatography method like ion exchange could likely achieve a fine separation between full, empty, and truncated sequences. While a multistep process is conceivable, its economic viability remains in question.
The potential for truncated DNA sequences to result in the production of truncated proteins, thereby potentially enhancing immune responses, requires careful consideration. This concern is rooted in the possibility that the immune system could identify these truncated proteins as either foreign or modified entities, which could trigger an immune response. Although direct evidence to support this concern is momentarily unavailable due to technical constraints, theoretical immunological principles may suggest that any deviation from the immune system's recognition of "self" might elicit an immune reaction. For instance, modified versions of the body's own proteins, potentially arising from incomplete gene expression or synthesis inaccuracies. In addition, it is noted that immune responses are typically more closely associated with CpG islands rather than truncated sequences [31]. It is essential, therefore, to approach the implications of truncated sequences carefully, clarifying that we are addressing truncation — the absence of certain portions of the DNA template, like the polyA tail or the initial cap of the gene, rather than a mutation or alteration in the sequence itself.
This discussion underscores the importance of further investigation into the role of capsid heterogeneity in the effectiveness of gene therapies. Our findings signal a clear direction for future research to explore the ramifications of truncated sequences on immune responses more thoroughly. This knowledge could lead to improved vector selection and use, enhancing viral vector biology. It is vital for refining gene therapy to increase effectiveness and reduce immune reactions, deepening our understanding of viral vectors and leading to more sophisticated gene therapy approaches.