Much has been learnt about fate determinants in development during neurogenesis and gliogenesis from candidate approaches, natural gene mutations and genome-wide expression analyses. To elucidate their function, and subsequently use them e.g., for direct neuronal reprogramming or therapeutic approaches, we need reliable tools for the long-term manipulation of gene expression, e.g., by seamless genomic integration. Tools for this are still relatively limited and mostly rely on viral vectors. The piggyBac transposon system allows genomic integration and long-term expression of genes of interest, providing a viable alternative as described below. These approaches are now timely to review and consider for specific pitfalls, due to the immense power of their potential for therapeutic use, e.g., by direct neuronal reprogramming in vivo (Götz and Bocchi, 2021). This applies especially as the CRISPR/Cas technology now facilitates multiplexing, targeting double-digit numbers of genes without problem (Breunig et al., 2018, 2021). This will help understanding the role of transcriptional networks in development, or to manipulate cell fate in a fine-tuned manner—e.g., to achieve correct subtype specification of glia or neurons in reprogramming—but needs to be delivered and integrated in a safe, yet reliable manner.

An important consideration in the use of integrating constructs is the similarity to endogenous transposable elements (TE), which play an important role in development. In humans, these make up the majority of the genome, with up to 69% suggested to be TE (de Koning et al., 2011; Ferrari et al., 2021). In the developing brain, the mobilization and (re-)integration of TE, especially of the endogenous retroviral LINE-1 elements, is linked to the development of neuronal subtype diversity and genomic mosaicism, and dysregulation may lead to developmental abnormalities (see e.g., Bodea et al., 2018; Misiak et al., 2019).

We will therefore review specific approaches to genome editing with an emphasis on gain-of-function and highlight a not yet reported and hitherto not yet fully explained artifact using the piggyBac transposon system in vivo.

Given the advantages of the piggyBac transposon system, we tested different transposases for gene manipulation in the developing cortex. For plasmid DNA, such as the piggyBac transposon, a favored delivery method is in utero electroporation (IUE), often performed during mid-gestation between E12 and E15 (Saito, 2006; Meyer-Dilhet and Courchet, 2020).

With the aim of later activating endogenous gene expression, we first administered the original transposase enzyme PBase (Supplementary Figures 1A,B) and a piggyBac transposon containing a fusion protein of enzymatically deactivated Cas9 and GFP (dCas9-GFP, Supplementary Figure 1B) into radial glia by IUE during mid-neurogenesis at E13 (Figure 2A). After 3 days, all of the brains showed ectopia of cells expressing the stem cell marker PAX6 and the basal progenitor (BP) marker TBR2 (Figure 2C). These markers are normally found in a distinctive band in the ventricular and subventricular zone, respectively (Götz et al., 1998; Englund et al., 2005).

To test if the observed phenotype may be caused by the expression of the foreign protein dCas9, we repeated the experiment with a simpler piggyBac transposon containing only a GFP reporter (Supplementary Figure 1B). We again observed developmental abnormalities in the electroporated brains after 3 days (Figures 2B,C). However, the variability here was much higher, ranging from some brains exhibiting no abnormalities, over the previously observed PAX6⁺/TBR2⁺ ectopia, to one embryo exhibiting severe malformations, where the cortex showed several sulcus-like indentations. We noted also a non-cell autonomous phenotype, as the folds were found not only at the electroporation site, but also at more distant positions (Figure 2D).

We then introduced a plasmid only carrying the GFP reporter, lacking the terminal repeats recognized by the transposase enzyme and thus without the possibility to integrate genomically (pCAG-IRES-GFP, Supplementary Figure 1B). In this case, no aberrant phenotype was observed (Figure 2C); a significant difference from the effects elicited by both piggyBac constructs (p_adj = 0.026 for piggyBac-GFP and p_adj = 0.005 for piggyBac-dCas9-GFP; Supplementary Table 1). Thus, the observed aberrations are dependent on the integration and transposition activity of the piggyBac system.

To determine if the phenotypes may be transient, we examined the brains later at E18, 5 days after IUE. Now all PBase-expressing brains showed an aberrant and in most cases exacerbated phenotype with anatomical malformations (Figure 2C). These could be distinguished into two levels of severity, with some cortices only showing localized hyperplasia at the electroporation site, whereas others exhibited more severe and widespread malformations, such as folded gyrus- and sulcus-like structures (Figure 2E), and, in one case, no septum with fusion of the lateral ventricles (Figure 2F).

As the PBase enzyme originates from insect cells (Cary et al., 1989), the sequence and codon usage is not adapted to the translation machinery in murine cells. The use of rare codons may influence the ribosomal translation speed and lead to ribosome collisions, a ribotoxic response (RTR) (Wu et al., 2020), and subsequently an unfolded protein stress response (UPR) (Cao and Kaufman, 2012). We thus exchanged the transposase PBase for mPB (Supplementary Figures 1A,B), the mouse codon optimized version of the same enzyme (Cadinanos and Bradley, 2007). This encodes the same amino acid sequence, generating an identical enzyme, but has an altered nucleotide sequence better suited to the murine translation machinery. When expressed together with piggyBac-GFP, only few brains exhibited PAX6⁺/TBR2⁺ ectopia at E16 and most brains did not show any abnormalities (Figures 2C,G).

At E18, we again observed a striking difference between the two versions of the transposase enzyme: Upon IUE of PBase, all brains showed aberrations, in most cases severe, while the phenotype of mPB was considerably milder and observed less frequently (Figures 2C,H), showing a significantly different overall phenotype severity between both enzymes (p = 0.017, Supplementary Table 1). Indeed, a piggyBac plasmid containing a dCas9-VPR fusion protein as well as a non-targeted gRNA (ntgRNA) and a GFP reporter (Supplementary Figure 1B), did not elicit any abnormalities 5 days after IUE, when combined with mPB (Figures 2C,I). Thus, we strongly recommend to use the codon-optimized mPB to avoid developmental aberrations.

After reviewing different methods for stable gain-of-function expression systems, we show here an unexpected artifact of the piggyBac transposon system in the developing cortex. Adverse developmental effects clearly depended on the integration of the constructs, suggesting that integration and/or transposition is important. In this regard it is interesting to note that there seems to be a connection between the size of the transposon, ranging between 3.0 and 9.6 kBp, and the severity of the phenotype, as smaller transposons elicited a more severe phenotype than larger ones (Figure 2C and Supplementary Figure 1B). It has been shown that the size of the transposon influences the rate of transposition, which starts decreasing considerably around 9 kBp (Ding et al., 2005). This may suggest that smaller constructs with potentially higher transposition rate are more prone to cause the observed abnormalities. Notably, the expression level of the transposase enzymes (Supplementary Figure 3B) seems not to affect the phenotype severity, possibly due to a saturated ratio of constantly expressed transposase enzyme to the limited number of available transposons. Once bound to the enzyme, only the length of the transposon may influence the speed or efficiency (and thereby the frequency) of the transposition process.

“Locus hopping” in the continued presence of the transposase enzyme may cause problems due to its preference for transcriptionally active sites (Ding et al., 2005; Li et al., 2013), which may disrupt their expression and alter transcription networks, the likelihood of which increases with the transposition activity and each excision and re-integration event. We would therefore suggest to develop mPB constructs that self-inactivate after integration of the transposon has been achieved, to avoid continued transposition events. Additionally, the link to endogenous transposition events has to be considered: While it has been shown that piggyBac transposase is unable to mobilize endogenous, PB-like transposons in mouse or human cells (Yusa et al., 2011; Saha et al., 2015), an endogenous human transposase highly expressed in neurons, PGDB5, has been suggested to potentially mobilize piggyBac transposons (Henssen et al., 2015).

It is intriguing that specifically the generation of PAX6⁺ and/or TBR2⁺ BPs and folding of the brain were affected, as these cells comprise an important glial subtype, the basal radial glial cells, and are particularly frequent in species with folded cortices (Borrell and Götz, 2014). Interestingly, in the developing gyrified cortex, they occur in a blockwise manner, with folds formed at the sites of expanded basal subventricular zone where basal radial glial cells are most frequent (Borrell and Götz, 2014). This organization can be mimicked even in the normally smooth murine cortex e.g., by locally lowering the levels of TRNP1, a protein involved in nuclear phase transition (Esgleas et al., 2020). Depletion of the protein, which in murine cortex is normally expressed in a salt-and-pepper fashion, in a specific region by IUE artificially causes blockwise expression, generating a region of expanded subventricular zone and basal radial glial cells, and accordingly induces cortical folding (Stahl et al., 2013; Borrell and Götz, 2014). We do not know if the ectopic BPs and folding that we observed upon PBase-mediated transposition after IUE activate a natural endogenous program, but it is intriguing to see how readily BP number and position may be influenced.

As noted above, the phenotype we observed was not strictly cell autonomous, as was also the case in the TRNP1 manipulations (Stahl et al., 2013; Esgleas et al., 2020) and other means to expand BP numbers (Rash et al., 2013; Xie et al., 2019). It will thus be very important to understand, if—and which—cell surface molecules or secreted signals may be regulated by these manipulations. Given the important communication via secreted vesicles in the cerebrospinal fluid (Marzesco et al., 2005; Morton and Feliciano, 2016), these may be involved in non-cell autonomous communication within the cortex tissue.

While the human cortex is naturally gyrified, the developmental disorder polymicrogyria (PMG) is characterized by the generation of additional, aberrant gyri in the neocortex. While PMG is commonly considered a neuronal migration disorder (Romero et al., 2018), several studies have also suggested causative progenitor and proliferation aberrations. Mutations in cyclin D2 were linked to a cortical malformation syndrome including PMG in patients, and IUE of these mutated genes caused an increased proliferation of progenitors and impaired cell-cycle exit in the developing mouse cortex (Mirzaa et al., 2014). In addition, PMG patients with mutations in the Pax6 as well as Tbr2 genes have been identified (Mitchell et al., 2003; Baala et al., 2007), which may relate to the data presented here with ectopic PAX6⁺ or TBR2⁺ cells and additional folds in this study.

While it is not immediately evident why the transposition would cause an increase in BPs, it is intriguing to consider that the naturally occurring transposition by recently integrated endogenous retroviruses (ERVs) may play a similar role, as their proportion hugely increases in the mammalian and especially the primate genome, correlating to cortex size and gyrification (Linker et al., 2017). Indeed, ERVs can regulate the expression of neighboring genes (Fasching et al., 2015; Brattas et al., 2017; Jonsson et al., 2019, 2020; Petri et al., 2019) and may thus have evolved as a novel gene regulatory machinery involved in enlarging and folding brain regions (Ferrari et al., 2021).

It may thus be interesting to follow up the artifact caused by the piggyBac transposon system mechanistically, by controlling transposition events and the generation of BPs. For the use of the piggyBac transposon system it is important to use codon optimized mPB and develop transposase removal strategies to minimize transposition after initial integration. Foremost, however, it is important to be aware of this pitfall, which has so far not been reported, and how to optimize the system as detailed here.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by the Government of Upper Bavaria.

MG and FV conceived, conceptualized the project, and wrote the perspective. FV performed all experiments and analyses shown. MK aided in preliminary experiments revealing the phenotype difference between transposase enzymes. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.