Introduction

The engineering of a mitochondrial uncoupling payload protein and the reprogramming of PVCs to target adipocytes are two critical steps in the development of ARROW, our weight-loss drug (see the detailed design of ARROR in the Description and Design pages).
Through multiple design-build-test-learn cycles, we have successfully engineered the fat-burning payload, a UCP1-based fusion protein that can be effectively expressed in bacteria and packaged into PVCs while maintaining its ability to enhance cell metabolism upon delivery into target cells. Simultaneously, we have been working on constructing an adipocyte-targeting PVC coat by modifying its tail fibers. In this section on engineering success, we will go through the engineering cycles we went through to develop the fat-burning payload and adipocyte-targeting tail fibers. (Figure 1).

Figure 1. An Adipocyte-targeting UCP1-delivering PVC Particle.

Engineering of the Fat-burning Payload Module

Figure 2. An Schematic Diagram of the Fat-burning Payload Module Engineering.

The First Engineering Cycle

Design:

In order to efficiently package UCP1 into PVCs, a Pdp1NTD protein domain must be fused to its N-terminus. Additionally, we decided to fuse an EGFP tag onto UCP1 to improve solubility and provide a visible tag. Hence, it would be crucial to test whether this gene fusion affects the subcellular localization and function of UCP1.
It has been reported that a C-terminus fusion of EGFP does not affect the mitochondrial localization and function of UCP1 (Bates et al., 2020). Hence, we first designed the Pdp1NTD-UCP1-EGFP construct (Fig. 3a) and validated its localization and function in HEK-293T cells.

Build and Test:

To assess the function of the fusion protein, we transfected HEK-293T cells with pNC087, which encodes a P_CMV promoter-driven Pdp1NTD-UCP1-EGFP protein expression cassette. Cells were imaged at 48 h post transcription to identify the cellular localization of the fusion protein. Cellular glucose consumption was then evaluated by measuring the remaining glucose levels in the cell culture medium. Results showed that instead of localizing in the mitochondria, the Pdp1NTD-UCP1-EGFP protein was localized all over the cytoplasm and nucleus (Fig. 3b). 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 (Fig. 3c). Interestingly, Alphafold2-based prediction of Pdp1NTD-UCP1-EGFP structure revealed an unexpected interaction between the Pdp1NTD domain and the UCP1 domain, which might be the cause of the UCP1 function loss in this design (Fig. 3d).

Figure 3. The design and characterization of the 1st Generation Fat-burning payload. (a) Design of the 1^st Generation of Fat Burning Payload Fusion Protein (Pdp1NTD-UCP1-EGFP). (b) Localization of Pdp1NTD-UCP1-EGFP payload protein in HEK-293T cells under wide-field microscopy. HEK-293T cells were transfected with pNC087 (P_CMV-Pdp1NTD-UCP1-EGFP) and imaged at 48 h post transfection, scale bar: 100 μm. Data are representative image of 3 independent experiments. (c) Bar Chart Illustrating Glucose Consumption in pNC087 Transfected HEK-293T Cells. Glucose concentration in the cell culture medium concentration was measured 48 h post transfection; data shows mean±SD, n=3 independent experiments. (d) AlphaFold2-based Structural Prediction of Pdp1NTD-UCP1-EGFP Fusion Protein.

Learn:

According to the structural prediction, the failure of the Pdp1NTD-UCP1-EGFP construct was probably due to the undesired interaction between the Pdp1NTD and UCP1, which affects the localization and function of the protein. This problem could be corrected by altering the linker length or changing the overall protein design.

The Second Engineering Cycle

Design:

Based on the knowledge gained from the initial engineering cycle, we then tried multiple design options with AlphaFold2. Interestingly, structural prediction showed that by simply swapping the UCP1 and EGFP domain, the Pdp1NTD-EGFP-UCP1 configuration could effectively reduce the interaction between Pdp1NTD and EGFP (Fig. 4a). We then tested this construct in HEK-293T cells.

Build and Test:

Similar to the first engineer cycle, we transfected HEK-293T cells with pNC088, a Pdp1NTD-EGFP-UCP1 expressing plasmid, and evaluated the cellular localization and function of the fusion protein at 48 h post-transfection. As expected, both wide-field fluorescent imaging (Fig. 4b) and live-cell confocal imaging (Fig. 4c) showed a highly specific colocalization of Pdp1NTD-EGFP-UCP1 signal with mitochondria (labeled by MTS-mcherry). Moreover, cells transfected with pNC088 showed a significantly higher glucose consumption compared to the control cells transfected with pcDNA3.1(+) vector (Fig. 4d), suggesting a significantly improved energy consumption in these cells.

Figure 4. The design and characterization of the 2nd Generation Fat-burning payload. (a) AlphaFold2-assisted Prediction of the Structure of 2^nd Generation of Payload Fusion Protein. (b, c) Localization Pdp1NTD-EGFP-UCP1 in HEK-293T cells. For wide-field microscopy in b, cells were transfected with pNC088 (PCMV-Pdp1NTD-EGFP-UCP1). For confocal images in c, cells were co-transfected with MTS-mcherry and PNC088. Photos were taken 48 h post transfection, scale bar: 100 μm for wide-field microscopy and 10 μm for confocal microscopy. Data are representative images of 3 independent experiments. (d) 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.

With the Pdp1NTD-EGFP-UCP1 construct as a promising payload, we then proceeded to investigate whether it can be successfully expressed in E. coli and assembled into PVCs. Therefore, we electroporated our payload plasmid pNC093 (expressing Pdp1NTD-EGFP-UCP1) together with an EGFR-targeting pPVC vector (pNC090, similar to the pAWP78-PVCpnf_pvc13-E01DARPin plasmid reported by Kreitz et al.) into E.coli to generate the $P V C_{E G F P - U C P 1}^{E G F R -targeting}$ particle (Fig. 5a). The particle then purified by ultracentrifuge. The expression of Pdp1NTD-EGFP-UCP1 protein was then validated by SDS-PAGE (Fig. 5b). Importantly, TEM imaging of the purified PVC fraction showed the existence of syringe-like structures similar to the previously published paper (Fig. 5c), suggesting the Pdp1NTD-EGFP-UCP1 payload could be efficiently packed into PVCs. Additionally, by incubating these PVC particles with HEK-293T cells transfected with either EGFR-expressing plasmid or empty vector control, we demonstrated these $P V C_{E G F P - U C P 1}^{E G F R -targeting}$ particles could selectively enhance the energy expenditure in EGFR-expressing cells (Fig. 5d).

Figure 5. Delivery of the Fat Burning Payload through

P V C_{E G F P - U C P 1}^{E G F R -targeting}

Particles to HEK-293T Cells. (a) Schematic representation of the construction of

P V C_{E G F P - U C P 1}^{E G F R -targeting}

particles. (b, c) Charactrization of the assembled

P V C_{E G F P - U C P 1}^{E G F R -targeting}

particles by SDS-PAGE (b) and negative-strain TEM (c). Scale bars, 100nm. (d) Charactrization of cellular metabolism in

P V C_{E G F P - U C P 1}^{E G F R -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

P V C_{E G F P - U C P 1}^{E G F R -targeting}

administration; data shows mean±SD, n=3 independent experiments.

Learn:

Through the aforementioned two cycles, we have successfully generated a UCP1-based payload protein that works as we expected. Once overexpressed in mammalian cells, it is correctly located in mitochondria and is able to enhance cellular energy expenditure. It can also be effectively expressed in E. coli and loaded into PVCs. Additionally, PVCs carrying these payload proteins could effectively enhance cell metabolism in a target-dependent manner.

Engineering of the adipose-targeting PVC Module

Figure 6. Schematic diagram of the adipose-targeting PVC coat module engineering

The First Engineering Cycle

Design:

To facilitate the targeted delivery of UCP1 into white adipose tissue using PVC, our design involved modifying the tail fiber protein of PVC (PVC13). We achieved this by attaching a previously described 9-mer peptide that specifically targets adipose tissue (Kolonin et al., 2004). In order to ensure optimal exposure of the targeting peptide on the tail fiber, we used protein structure prediction deep learning model AlphaFold2, to examine the trimeric structure of PVC13(as shown in Figure 7a) via AlphaFold2. By considering different types of linkers to connect PVC and the 9-mer peptide (as depicted in Figure 4a), we identified three potential candidates for our initial trials: 1GS, 3GS, and 1EAAAK. These candidates were chosen based on their ability to individually and effectively expose the 9-mer peptide.
Our initial cloning strategy employed traditional infusion cloning methods, utilizing SphI and SalI enzymes for the double enzyme digestion of the pNC090 vector (PAWP78-PVCpnf pvc13-E01DARPin). We then performed PCR amplification for the subsequent fragments, fNC177 and fNC178, which were specifically designed for the insertion of targeting peptide and linker sequences. To maximize the abundance of infusion products, these two fragments were overlapped during the cloning process.

Build and Test:

Figure 7. Design and Construction of Adipose-targeting Tail Fiber Proteins. (a) Alphafold2-based design of adipose-targeting peptide-presenting tail fibers. (b) Agarose gel electrophoresis image showing the digestion of pNC090 (PAWP78-PVCpnf pvc13-E01DARPin) with SphI and SalI restriction enzymes. (c) Overlapping PCR Analysis of fNC177 and fNC178.

During the experiments, we encountered challenges with our plasmid constructs. Specifically, when digesting pNC090 with SphI and SalI restriction enzymes, we observed a band around 6 kb, resulting in a lower than expected concentration after gel purification(Figure 7b). Additionally, during the overlap PCR of inserting fragments fNC177 and fNC178, the desired band at approximately 6 kb did not appear, despite our effort to optimize reaction conditions (Figure 7c).

The Second Engineering Cycle

Design and Future Work

During our interview with Mr.Deyang Zhou, renowned in the field of mammalian cell synthetic biology, we discussed the challenges encountered in cloning the 25 Kbp plasmid. To streamline the plasmid construction process, he suggested exploring the utilization of the Golden Gate Assembly method. Surprisingly, upon devising a specific cloning strategy for each plasmid, we observed a significant reduction in workload. In particular, we designed a novel PVC 13 part (BBa_K4960024) to facilitate subsequent cloning of the targeting sequence into the pPVC plasmid. The engineered model plasmid, pNC092 (pAWP78-PVCpnf_pvc13-2*BsaI), features this PVC13 part and can serve as a template for future teams seeking to modify PVCs (Figure 8).
Unfortunately, due to time constraints, we could not complete the construction of the intended plasmids before the WIKI FREEZE deadline. However, we are currently undergoing construction using the Golden Gate Assembly method. Once these constructs are finalized, we will promptly proceed with testing the fat-burning payload on 3T3-L1 cells, enabling us to continue the development of the adipose-targeting PVC coat Module.

Figure 8. Schematic Diagram of Golden Gate Assembly Strategy for New pvc13-based pPVC Constructs

The aforementioned engineering process outlines the comprehensive efforts involved in building, testing, and optimizing the UCP1-delivering adipocyte-targeting PVC particle, ARROW, from scratch. This intricate process can be categorized into two key modules: the fat-burning payload module and the adipocyte-targeting PVC coat module. Thorough characterization of the fat-burning payload has undeniably demonstrated its efficacy. For more detailed insights into our future developments, we encourage you to refer to our Implementation and Integrated Practices pages.

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