Current therapies for liver cancer have poor efficacy. In recent years, the emerging CAR-T cell therapy has demonstrated remarkable success in treating hematological malignancies, but its limited ability to infiltrate solid tumors results in suboptimal performance in liver cancer treatment. To address the issue of poor infiltration, our team initially considered adopting CAR-M cell therapy, leveraging the macrophage's natural capacity to infiltrate solid tumors. However, CAR-M therapy still faces two major challenges: antigen heterogeneity in liver cancer and the immunosuppressive tumor microenvironment. After brainstorming and consulting relevant experts, the team decided to introduce the Syn-Notch system into the macrophage chassis to overcome these challenges. The SynNotch system consists of an extracellular antigen recognition domain, a Notch transmembrane domain, and an intracellular transcription factor (e.g., Gal4-VP64) (Figure 1A). Upon recognition of a primary target antigen (INPUT), it releases the transcription factor into the nucleus, initiating the expression of user-defined downstream genes (OUTPUT). This system allows for the incorporation of desired extracellular antibodies while offering programmable output with complex logic gates.
Therefore, our overall design aim to use liver-specific (rather than liver cancer-specific) activation (Input) to address antigen heterogeneity, and subsequently maintain the anti-cancer M1 polarized phenotype (Output) to tackle the immunosuppressive tumor microenvironment (Figure 1B). In the following sections, we will present our project design and experimental results in a progressively detailed manner.
Image adapted from: Cell. 2016 Feb; Cell, 164(4):780-791.
To achieve the preliminary input design, we needed to screen for a protein meeting the following criteria:
| Criteria | Rationale |
|---|---|
| Hepatocyte Localization | Hepatocytes are the primary cell type in the liver, essential for achieving liver-specific rather than hepatocellular carcinoma -specific activation. |
| High Specificity Expression | Ensures specificity and sensitivity of the input signal. |
| Membrane Localization | Requirement of the SynNotch system. |
The goal was to identify the target protein, obtain its single-chain variable fragment (scFv), and use it to construct the signal INPUT module of our SynNotch system.
Therefore, we conducted the following phases: (I) Screening of liver-specific membrane protein SLC17A2; (II) Preparation of Anti-SLC17A2 scFv; (III) Optimization and functional validation of the Syn-M based on Anti-SLC17A2 scFv.
Screening of Liver-Specific Membrane Protein SLC17A2
1. Protein screening
Background:
We need to screen for liver-specific proteins that meet the three criteria: high/exclusive expression in the liver, a hepatocyte marker protein, and membrane localization. This is the cornerstone for constructing the signal INPUT module of our SynNotch system.
Results:
We categorized all genes into five groups using the Protein Atlas to identify proteins highly expressed in the liver. Subsequently, we filtered these proteins using our in-house software based on the criteria of hepatocyte-specific expression and cell membrane localization (for more details, please refer to our Software page).
2. Identify the extracellular domain of the candidate protein
Background:
After identifying SLC17A2 with its high liver-specific expression, and considering our foundational SynNotch system, we needed to further investigate whether this protein has a transmembrane structure and possesses an ideal extracellular domain for antibody development.
Results:
For this purpose, we performed TMHMM-2.0 prediction for transmembrane helix structure and AlphaFold structure prediction for murine SLC17A2. TMHMM-2.0 prediction indicated that murine SLC17A2 has 10 transmembrane helices. AlphaFold structure prediction revealed it possesses an ideal extracellular structure, with amino acids 48-84 being relatively conserved, representing an ideal region for antibody development (for more details, please refer to ourModel page).
Conclusion:
Combining software screening and structural predictions, we identified SLC17A2 as a suitable target for constructing the corresponding SynNotch extracellular domain to achieve liver microenvironment-specific activation.
Preparation of Anti-SLC17A2 scFv
Background:
Based on the prior screening phase identifying the liver-highly-specific target protein SLC17A2, we planned to prepare its single-chain variable fragment (scFv) as the extracellular domain of the SynNotch system to achieve liver-specific activation of macrophages.
Results:
Initially, we planned to generate high-affinity scFvs against SLC17A2 using genetic engineering techniques. However, the process of scFv development involves multiple technical challenges requiring advanced experimental skills and extensive experience, which exceeded the technical capabilities of an undergraduate team within the project timeframe. To ensure antibody quality and project timeline, we decided to outsource the preliminary scFv preparation work to a professional biotechnology company.
We provided the 48-84 amino acid sequence to Xi'an Haina Biotechnology Co., Ltd. for the preparation of Anti-SLC17A2 scFv. The company delivered three antibody clones (No.1, No.2, No.3). We sequenced these antibodies and tested their efficacy (Figures 2). Immunofluorescence results indicated that Clone No.3 performed significantly better than the other two (Figures 3).
Conclusion:
Building upon the assistance of the professional biotechnology company, we further confirmed the optimal efficacy of Clone No.3 antibody, successfully obtaining the most suitable scFv for the SynNotch system extracellular domain.
Optimization and Functional Validation of the Syn-M based on Anti-SLC17A2 scFv
1. Syn-M-INPUT Test 1 System
Background:
The SynNotch system consists of a ligand-binding domain (LBD), extracellular domain (ECD), transmembrane domain (TMD), juxtamembrane domain (JMD), and transcription factor (TF). Using the acquired Anti-SLC17A2 scFv, we designed the Syn-M-INPUT Test 1 system to verify whether the desired liver-specific activation could be achieved. In this system, the SLC17A2 scFv serves as the ligand-binding domain (LBD), GV as the intracellular transcription factor (TF), and the remaining domains retain the Notch1 structure (Figure 4).
Results:
To verify the feasibility of the Syn-M-INPUT Test 1 system, we performed immunofluorescence experiments. Fluorescence microscopy results showed that this system partially self-activated even in the absence of the SLC17A2 signal from hepatocyte surfaces (Figure 5).
Conclusion:
Syn-M-INPUT Test 1 System did not meet the design goal of liver-specific activation and required further optimization.
2. Syn-M-INPUT Test 2 System
Background:
Through a literature review, we decided to adopt a systematic modular engineering approach to modify the extracellular domain (ECD) of the existing Syn-M-INPUT Test 1 system, aiming to reduce its basal self-activation level.
Results:
We replaced the extracellular region sequence of Notch1 with the extracellular sequence of the CD8α hinge, while retaining the transmembrane domain and cleavage site of mouse Notch1, to design the Syn-M-INPUT Test 2 system (Figure 6). The optimized SynNotch receptor (Syn-M-INPUT Test 2 system) is 810 base pairs shorter than the original full-length design (Syn-M-INPUT Test 1 system). To verify the specificity of the optimized system, we performed immunofluorescence and flow cytometry assays on the Syn-M-INPUT Test 2 system.
Both immunofluorescence and flow cytometry results indicated that the Syn-M-INPUT Test 2 system did not self-activate in the absence of the SLC17A2 signal from hepatocyte surfaces. Upon receiving the correct protein signal, the Syn-M-INPUT Test 2 system exhibited a strong fluorescence signal, indicating that it could be normally and specifically activated, thus meeting the design goal (Figure 7).
Conclusion:
We replaced the ECD module component with a higher-performing element (the optimized CD8α hinge). Due to its efficient, high-fidelity activation characteristics and compact sequence length, this optimized CD8α hinge SynNotch receptor was selected for subsequent development and engineering of the macrophage chassis.
In the previous section, through multiple rounds of iteration and optimization, we successfully achieved liver-specific activation of Syn-M-INPUT Test 2 system, thereby addressing the challenge of hepatocellular carcinoma antigen heterogeneity. To address the other major challenge, the immunosuppressive tumor microenvironment, we needed to carefully design the OUTPUT signal following Syn-M-INPUT Test 2 system's liver-specific activation.
Literature review indicated that P65 is a key anti-cancer pro-inflammatory target, and the SIRPα-CD47 pathway is a crucial axis for tumor cell immune evasion. Therefore, we plan to design the output module to simultaneously overexpress P65 and SIRPα shRNA, thereby achieving upregulation of P65 and downregulation of SIRPα. Since mRNA and shRNA are transcribed by RNA Polymerase II (Pol II) and Pol III respectively, we designed a bicistronic system. The SIRPα shRNA was flanked by 5' and 3' miRNA flanking sequences, enabling it to utilize Pol II transcription linkage sites, thus allowing synchronous transcription of P65 and SIRPα shRNA.
Our final design schematic is as follows (Figure 8): the INPUT module achieves liver-specific activation based on Anti-SLC17A2 scFv, and the OUTPUT module simultaneously expresses P65 and SIRPα shRNA to function synergistically against the immunosuppressive microenvironment. We named this system "Synergy " (also be referred to as Syn-M when describing experimental results).
Therefore, after constructing Synergy based on the above design, we carried out the following phases to validate its functions: (I) Transcriptomic analysis; (II) Infiltration kinetics analysis; (III) Functional validation of P65-related effects; (IV) Functional validation of SIRPα-related effects.
Transcriptomic analysis
Background:
We co-cultured Synergy with AML12, a normal mouse liver cell line, to activate it, and then commissioned Beijing Biomarker Biotechnology Co., Ltd. to perform transcriptome sequencing to analyze the differences between it and unedited macrophages.
Results:
Pathway enrichment analysis indicated that inflammatory-related pathways such as TNF, TGF, and NFκB in activated Synergy showed significant changes. Further analysis of relevant factors revealed that the levels of factors associated with pro-inflammation, phagocytosis, T cell activation, and neutrophil regulation were all significantly elevated in activated Synergy (for more details, please refer to our Model page).
Conclusion:
Compared to unedited macrophages, activated Synergy exhibits distinct pro-inflammatory and anti-tumor properties.
Infiltration kinetics analysis
Background:
We aim to determine whether the liver cancer infiltration capacity of activated Synergy differs from that of unedited macrophages. However, such a complex experiment is clearly difficult to complete in vitro within a short timeframe. Therefore, we performed a full-process kinetic simulation of macrophage infiltration into liver cancer.
Results:
Using a multi-field coupled grid model and a gradient-based migration algorithm, we achieved dynamic simulation of macrophage behavior within the liver cancer microenvironment. The simulation results show that macrophages can migrate from vascular regions toward tumor areas along chemotactic gradients. Upon SynNotch activation, the immediate behavioral switch (P65↑, SIRPα↓) altered their migration trajectories, validating the feasibility of the engineered design (for more details, please refer to ourModel page).
Conclusion:
Compared to unedited macrophages, activated Synergy exhibits superior liver cancer infiltration capability, which may be attributed to its sustained maintenance of the pro-inflammatory, anti-tumor M1 phenotype.
Functional validation of P65-related effects
Background:
We co-cultured Synergy with the normal mouse hepatocyte cell line AML12 to activate it, followed by a series of experiments, including immunofluorescence, ELISA, and qRT-PCR, to verify whether the P65-related functions aligned with our intended design.
Results:
By co-staining the macrophage marker F4/80 and P65 via immunofluorescence, Figure 9 shows a significantly enhanced red fluorescence intensity in activated Syn-M, confirming the overexpression of the P65 gene in this system. ELISA kits were used to measure the concentrations of TNF-α, IL-6, and IFN-γ in the cell supernatant. The results indicated that, compared to unedited macrophages, the levels of TNF-α, IL-6, and IL-1β were all elevated in the Syn-M supernatant (Figure 10). After co-culturing Syn-M with AML12 for 24 hours, F4/80-positive macrophages were isolated using magnetic beads, and qRT-PCR was performed to assess the expression of P65-related factors. The results showed significantly increased expression of inflammatory cytokines, including iNOS, IL-12, TNF-α, and IL-1β (Figure 11).
Conclusion:
These findings demonstrate that activated Syn-M not only effectively overexpresses P65 but also simultaneously enhances the expression of downstream inflammatory factors of P65, thereby boosting its pro-inflammatory and anti-tumor capabilities.
Functional validation of SIRPα-related effects.
Background:
We co-cultured Synergy with the normal mouse hepatocyte cell line AML12 to activate it, followed by a series of experiments, including immunofluorescence, and phagocytosis assay, to verify whether the SIRPα-related functions aligned with our intended design.
Results:
By co-staining the macrophage marker F4/80 and SIRPα via immunofluorescence, Figure 12 shows a significantly reduced red fluorescence intensity in activated Syn-M, confirming the block expression of the SIRPα gene in this system. Figures 13 and 14 show the tumor cell phagocytosis capability of activated Syn-M. The flow cytometry results indicate that, compared to unedited macrophages, activated Syn-M can not only engulf more liver cancer cells (Hepa1-6) but also more lung cancer cells (LLC).
Conclusion:
These findings indicate that activated Syn-M exhibits enhanced phagocytic capacity against tumor cells, and this enhancement is not restricted by different tumor antigens. In other words, tumors in the liver—whether heterogeneous hepatocellular carcinomas or metastatic cancers from other origins (e.g., lung cancer liver metastases)—can all be phagocytosed by Syn-M.
Background:
We recognized that autologous macrophage-based therapies, involving ex vivo modification and reinfusion, have several limitations, including high production costs, stringent cell storage and transportation requirements, and limited sources of recipient-autologous macrophages. We needed to consider how to optimize the delivery method to better address these limitations and lower the treatment threshold. we found that Lipid Nanoparticle (LNP) delivery systems offer advantages such as excellent biocompatibility, high modifiability, and simplified storage and transportation conditions, which could compensate for several shortcomings of the autologous macrophage reinfusion strategy. Based on this, we planned to utilize LNP as a delivery vehicle to encapsulate the Syn-M related plasmids, leveraging LNP for efficient and stable in vivo delivery and gene editing.
Results:
We utilized the ionizable property of DLin-MC3-DMA and the hepatocellular carcinoma-associated macrophage targeting capability of DSPE-PEG-M2pep (Figure 15) to enhance the editing efficacy of the LNP. The LNP contains components for the in situ editing of tumor-associated macrophages (TAMs) into Syn-M, fulfilling our design objective. Furthermore, using techniques like transmission electron microscopy (TEM), we verified that the liposomes were spherical, of appropriate size (~85 nm), with uniform particle size distribution, and essentially electrically neutral (Figure 16), meeting the preliminary design requirements.
Conclusion:
Our current design for Syn-M@LNP is at the initial stage, and we plan to conduct subsequent experimental validation after the competition.