Engineered Photorhabdus Virulence Cassette (PVC) particles could specifically deliver payload proteins into target mammalian cells

To demonstrate that PVCs could serve as a viable target-specific protein-delivering system, we first engineered Escherichia coli to produce PVCs from P. asymbiotica ATCC 43949 (PVCpnf) using a method similar to the one published by Kreitz et al.(Kreitz et al., 2023). Similar to their design, we split the PVC system into separate structural/accessory plasmid (pPVC) and payload/regulatory plasmid (pPayload). To evaluate the target-specific protein delivery function of engineered PVCs, we first synthesized a pPVC plasmid carrying E01DARPin (pNC090, similar to the pAWP78-PVCpnf_pvc13-E01DARPin plasmid reported by Kreitz et al.) that specifically targets EGFR and a pPayload plasmid carrying Cre recombinase (pNC091, similar to the pBR322-Pdp1NTD-Cre-HiBiT plasmid reported by Kreitz et al.). Active PVCs were then purified from the lysate of E. coli cells electroporated with pNC090 and pNC091 via ultracentrifuge (Figure 1a). Negative-stain transmission electron microscopy resembled canonical PVC structures similar to the ones reported by Kreitz et al. (Figure 1b), suggesting a successful assembly of PVCs.

Figure 1. Validation of PVC Particles Specificity Delivery of Payload into Mammalian Cells. (a) Schematic representation of the Construction of Engineered PVC particles. PVCs were purified from E.coli electroporated with pNC090 (pPVC with EGFR targeting E01DARPin) and pNC091 (pPayload with Cre) via ultracentrifuge. (b) Negative-strain TEM of the purified Cre-loading, EGFR-Targeting PVCs. Scale bars: 600 nm. (c) PVC-dependent activation of gene expression in mammalian cells. Left: Schematic diagram of mammalian cell gene expression induced by Cre-delivering PVCs. Right: HEK-293T cells transfected with either pcDNA3.1(+) plasmids or pLZ362 (P_CMV-LoxP-STOP-LoxP-sNluc) were treated with Cre-carrying, EGFR-Targeting PVCs. sNluc production in the culture medium was measured 72 h after PVC administration; data shows mean±SD, n=3 independent experiments.

To evaluate the protein-delivering function of PVCs, we incubated these EGFR-targeting, Cre-delivering PVCs (

PVC _Cre^{EGFR_targeting}

) with HEK-293T cells co-transfected with pLZ362 (P_CMV-LoxP-STOP-LoxP-sNluc, a gift from Lihang Zhang, Westlake University) and either pNC089 (P_CMV-EGFR) or pcDNA3.1(+) plasmids. Secreted Nanoluc (sNluc) levels in the culture medium were then evaluated at 72 h post PVCs treatment. Results showed a significant increase in sNluc levels in EGFR-expressing cells treated with PVCs, while the cells without EGFR expression showed low sNluc levels (Figure 1c). These results demonstrated that engineered PVCs could specifically deliver payload proteins into target mammalian cells, thereby serving as a possible delivery mechanism for our purposes.

Structural-informed engineer of a UCP1-based Payload Protein

In PVCpnf, the N-terminus domain of Pdp1 (Pdp1NTD) is critical for the efficient loading of payload proteins into the PVCs. Hence, to efficiently load mitochondrial uncoupler proteins (UCP1 in our case) into the PVCs, an N-terminus Pdp1NTD domain must be supplemented. In addition, we also decided to fuse an EGFP tag to provide a potential visualization signal and improve the solubility of UCP1 in the E. coli expression system.

Since the UCP1-EGFP construct has been previously reported in transgenic mice (Bates et al., 2020), we started by constructing a Pdp1NTD-UCP1-EGFP construct as an initial test (Figure 2a). To validate the protein function in the best possible scenario, we transfected the HEK-293T cells with pNC087, a mammalian expression vector carrying the P_CMV-Pdp1NTD-UCP1-EGFP cassette. Cells were imaged at 48 h post transfection to validate the subcellular location of the Pdp1NTD-UCP1-EGFP fusion protein. Unfortunately, results showed that instead of localizing in the mitochondria, the Pdp1NTD-UCP1-EGFP protein was localized all over the cytoplasm and nucleus (Figure 2a), suggesting the interaction between UCP1 and key chaperone proteins enabling its mitochondria translocation was compromised by the protein fusion. Moreover, by evaluating the remaining glucose level in the culture medium, we analyzed the glucose consumption of the cells transfected with either pNC087 or pcDNA3.1(+) control plasmid, which, in a way, represented the level of cellular energy expenditure. Consistent with the failed mitochondrial localization, glucose levels in the pNC087-transfected cells showed no significant difference compared to the control group transfected with pcDNA3.1(+) vector only (Figure 2b).

To understand how the fusion of Pdp1NTD and EGFP affected the function of UCP1, we performed structural prediction using AlphaFold2. Interestingly, we observed an unexpected interaction between the Pdp1NTD domain and UCP1 (Figure 2c, red box), which could possibly change the local protein structure and affect the translocation and function of UCP1. With the same protocol, we predicted the structure of a few design options and found that we could simply solve this problem by swapping UCP1 and EGFP (Figure 2d).

To test this option, we then constructed pNC088, a mammalian cell expression vector carrying the P_CMV-Pdp1NTD-EGFP-UCP1 cassette. As expected, both wide-field fluorescent imaging (Figure 2e) and live-cell confocal imaging (Figure 2f) showed highly specific colocalization of Pdp1NTD-EGFP-UCP1 signal with mitochondria markers (MTS-mcherry, Figure 2f). Moreover, cells transfected with pNC088 showed a significantly higher glucose consumption compared to the control cells transfected with pcDNA3.1(+) vector (Figure 2g), suggesting a significantly improved energy consumption in these cells.

Figure 2. Functionality of UCP1-based Payload Protein in HEK-293T Cells. (a) Localization of UCP1-based payload protein Pdp1NTD-UCP1-EGFP in HEK-293T cells under wide-field microscopy. HEK-293T cells were transfected with pNC087 Pdp1NTD-UCP1-EGFP and imaged 48 h post transfection, scale bar: 600 μm. Data are representative image of 3 independent experiments (b) Charactrization of cellular metabolism in HEK-293T cells transfected with either pNC087 or pcDNA3.1(+). Glucose concentration in the cell culture medium concentration was measured 48 h post transfection; data shows mean±SD, n=3 independent experiments. (c) AlphaFold2 prediction of Pdp1NTD-UCP1-EGFP protein structure. The unexpected interaction between SepC and UCP1 is labeled in a red box. (d) AlphaFold2 prediction of Pdp1NTD-EGFP -UCP1 protein structure.(e, f) Localization Pdp1NTD-EGFP-UCP1 in HEK-293T cells.For wide-field microscopy in e, cells were transfected with pNC088 (P_CMV-Pdp1NTD-EGFP-UCP1).For confocal images in f, cells were co-transfected with MTS-mcherry and PNC088. Photos were taken 48 h post transfection, scale bar: 100μm for wide-field microscopy and 10 μm for confocal microscopy. Data are representative images of 3 independent experiments.(g) Charactrization of cellular metabolism in HEK-293T cells transfected with either pNC088 or pcDNA3.1(+). Glucose concentration in the cell culture medium was measured 48 h after transfection; data shows mean±SD, n=3 independent experiments.

PVC-mediated delivery of the Pdp1NTD-EGFP-UCP1 effectively altered the energy expenditure in mammalian cells in a target-dependent manner.

With the P_CMV-Pdp1NTD-EGFP-UCP1 construct as a promising payload, we then proceeded to investigate whether it can be successfully expressed in E. coli and loaded into PVCs. For such matter, we constructed pNC093, a pPayload plasmid carrying Pdp1NTD-EGFP-UCP1 payload. By electroporating E. coli cells with both pNC093 and a pPVC plasmid carrying E01DARPin (pNC090), we could then get a

PVC _{EGFR_UCP1}^{EGFR_targeting}

particle that specifically targets EGFR-expressing cells and delivers Pdp1NTD-EGFP-UCP1 protein (Figure 3a).

To validate whether Pdp1NTD-EGFP-UCP1 protein could be correctly expressed, we performed SDS-PAGE analysis on purified PVCs, which showed a strong and clear band on approximately 69 kDa, which resembles Pdp1NTD-EGFP-UCP1 (Figure3b). In the meantime, negative-stain transmission electron microscopy on purified PVCs (Figure 3c) showed similar structures to the Cre-carrying PVCs in Figure 1b, suggesting that the Pdp1NTD-EGFP-UCP1 protein could be correctly loaded into the PVCs. Additionally, by incubating these PVC particles with HEK-293T cells transfected with either pNC089 (P_CMV-EGFR) or pcDNA3.1(+) plasmids. we demonstrated these

PVC _{EGFR_UCP1}^{EGFR_targeting}

particles could selectively enhance the energy expenditure in EGFR-expressing cells (Figure 3d). Altogether, our findings demonstrate a successful engineer of a PVC-based strategy to boost cellular energy expenditure by specifically deliver UCP1 into target cells.

Figure 3. Delivery of the Fat Burning Payload through PVC _{EGFR_UCP1}^{EGFR_targeting} Particles to HEK-293T Cells.(a) Schematic representation of the construction of PVC _{EGFP_UCP1}^{EGFR_targeting} particles.(b, c) Charactrization of the assembled PVC _{EGFP_UCP1}^{EGFR_targeting} particles by SDS-PAGE(b) and negative-strain TEM(c). Scale bars, 100nm.(d) Charactrization of cellular metabolism in PVC _{EGFP_UCP1}^{EGFR_targeting} treated HEK-293T cells transfected with either pNC089 or pcDNA3.1(+) plasmids. Glucose concentration in the cell culture medium was measured 48 h after PVC _{EGFP_UCP1}^{EGFR_targeting} administration; data shows mean±SD, n=3 independent experiments.

Structure-informed design of adipose-targeting PVC Tail Fiber Protein

To enable the PVC-based delivery of UCP1 into white adipose tissue, we then decided to equip the tail fiber protein of PVC (PVC13) with an adipose tissue-targeting 9-mer peptide developed by Kolonin et al. in 2004 (Kolonin et al., 2004). To ensure the efficient exposure of the targeting peptide on the tail fiber, we predicted the structure of PVC13 trimers inserted with adipose tissue targeting peptide flanked by different linkers (Figure 4a). We were unable to peruse validating these constructs due to the limited time before the jamboree and the practical difficulties in working with such a huge plasmid (~25kb), while we did make some effort in simplifying the cloning process by designing a new PVC13 part which allows subsequent Golden-gate-based cloning of the targeting sequence into the pPVC plasmid (Figure 4b, see also the second part of our Engineering page for more information).

Figure 4. AlphaFold2-guided Design of Adipose Cell-targeting PVC Coat Expressing Plasmids. (a) AlphaFold2 based prediction of engineered PVC tail fiber trimer structure. Structure of adipose-targeting CKGGRAKDC peptide-presenting PVC tail fiber with indicated linkers were shown.(b) Schematic diagram of the newly-designed pvc13 part allowing Golden Gate cloning of targeted sequence into the pPVC plasmid.

Reference

Bates, R., Huang, W., & Cao, L. (2020). Adipose Tissue: An Emerging Target for Adeno-associated Viral Vectors. Mol Ther Methods Clin Dev, 19, 236-249. https://doi.org/10.1016/j.omtm.2020.09.009
Kolonin, M. G., Saha, P. K., Chan, L., Pasqualini, R., & Arap, W. (2004). Reversal of obesity by targeted ablation of adipose tissue. Nat Med, 10(6), 625-632. https://doi.org/10.1038/nm1048
Kreitz, J., Friedrich, M. J., Guru, A., Lash, B., Saito, M., Macrae, R. K., & Zhang, F. (2023). Programmable protein delivery with a bacterial contractile injection system. Nature, 616(7956), 357-364. https://doi.org/10.1038/s41586-023-05870-7