Description

Several bipartite systems have been described for use with C. elegans (Wei et al., 2012; Wang et al., 2017; Wang et al., 2018; Mao et al., 2019; Nonet, 2020). However, little manipulation of the drivers or reporters has been performed to optimize them for C. elegans. Mao. et. al. 2019 demonstrated that the QF activation domain is a more potent activator than either VP64 or the hybrid VP64-p65-Rta tripartite activatorVPR. Here, I describe modifications of linker region between the activation domains of both GAL4_SK and TetR drivers that increase the activity of these reporters. Although these studies are not comprehensive, I opted to describe them herein because the modification of the GAL4_SK driver increases the activity of this driver substantially such that it is now on par with the LexA, TetR and QF2 drivers.

I used an efficient RMCE protocol (Nonet, 2020) to create the transgenic animals. Modified versions of a mec-4 promoter GAL4_SK-QF_AD, GAL4_SK-VP64 and TetR-QF_AD constructs were created in an RMCE integration vector using a Golden Gate cloning approach, then integrated on Chr IV using a standard injection protocol. After outcrossing to the appropriate reporter, the expression level of GFP at steady state in PLM and ALM soma of L4 animals was quantified (Figure 1).

Previously, I described drivers consisting of C. elegans codon optimized synthetic GAL4_SK, TetR and LexA DNA binding domains fused to the QF activation domain based on the observations of Mao. et al. 2019. Although the GAL4_SK construct was modestly active, the LexA and TetR construct were incapable of activating the reporter (supplemental methods of Nonet, 2020). Replacement of a 49 amino acid portion of central domain of QF, which separates the TetR and LexA DNA binding domains and the QF activation domain in the original constructs, with a 12 amino acid flexible linker converted both into much stronger drivers than the GAL4_SK-QF_AD driver (Nonet, 2020). Here I show that insertion of a similar linker also greatly increases the potency of the GAL4_SK-QF_AD driver. I also extended the linker of the TetR-L-QF_AD and this further improved activity of this driver. In Nonet, 2020, I also tested the functionality of a GAL4_SK-VP64 construct (Wang et al. 2017) in single copy and found it was incapable of expressing the GFP reporter. To test if the failure of GAL4_SK-VP64 was also the result of insufficient domain separation, I inserted a 40 amino acid flexible linker in between GAL4 and VP64. Although the linker containing driver activates transcription, it does so extremely poorly in comparison to the GAL4_SK-QF_AD drivers. I speculate this is likely due to the loss of the MED25 subunit of the mediator complex in the nematode lineage (Grants et al., 2015). VP64 (a 4X VP16) is known to activate transcription in part through interaction with MED25 (Mittler et al., 2003) and the loss of this interaction could account for the observed weak activation properties of VP64 and related activators in worms (Mao et al. 2019 and herein).

In addition to the modification of the linker domain of the transgenes I characterize herein, some of the transgenes also differ in other ways that could theoretically impact my conclusions. First, some driver transgenes contain a tbb-2 3′ UTR and others use an act-4 3′ UTR. I consider it very unlikely these differences impact GFP expression for two reasons. First, my lab has previously demonstrated that mec-4promoter driven GFP-C1 transgenes employing the tbb-2 3′ UTR and the act-4 3′ UTR express at very similar levels in ALM and PLM (Dour and Nonet, 2021). More importantly, I previously demonstrated that the activity of the both a GAL4-QF_AD driver and a LexA-L-QF_AD driver are insensitive to dosage of the driver (Figure 5 of Nonet, 2020). Specifically, GFP levels observed in GAL4_SK/+; UAS::GFP ~= GAL_SK; UAS::GFP. Thus, the level of expression of the driver is unlikely to be determining the GFP signal level. Rather, I speculate that GAL4_SK is saturating the UAS binding sites in all transgenes and that the inherent activation properties of the driver determines the expression level.

Second, all drivers were integrated at jsTi1493 except for the inactive linkerless TetR-QF_AD driver. I present data on integration of NMp3730 (mec-4p tetR-QF_AD tbb-2 3′ UTR) at jsTi1490 (Nonet, 2020). I also previously created insertions of the same plasmid at jsTi1493 IV and jsTi1492 II (jsSi1531 [mec-4p tetR-QF_AD tbb-2 3′] IV and jsSi1534 [mec-4p tetR-QF_AD tbb-2 3′] II; Table S1 and Table S5 in Nonet, 2020). The GFP expression was undetectable in jsSi1531/+; jsSi1519 [7X tetO ∆pes-10p GFP]/+ and jsSi1534/+;jsSi1519/+ double heterozygote animals. However, only the jsTi1490 transgene was homozygosed with the reporter. Thus, I quantified that insertion. However, this difference is highly unlikely to impact my findings as the other two insertions are qualitatively completely inactive.

The improvements to GAL4_SK-QF_AD by insertion of a flexible linker is an important addition to the C. elegans bipartite expression toolkit since the GAL4 system is so extensively developed in Drosophila. Using multi-copy lines and a VP64 activator, Wang et al. (2018) have already shown that the split GAL4_SK system is functional in worms. Incorporation of a similar flexible linker and a QF activation domain into those tools should permit development of a robust single copy split-GAL4_SK system. Other GAL4 system tools previously developed for Drosophila such as GAL80ts and GAL4-PR tools (Caygill and Brand, 2016) which provide temporal control in addition to spatial control could also easily be incorporated into the worm toolbox.

In addition, further manipulation of the size and properties of the linker domain separating the DNA binding domain and activation domains could yield drivers with even stronger activation proper which would likely also be applicable to the LexA, QF and Tet ON/OFF driver/reporter systems.