>>Protein Purification

We constructed a plasmid that expresses His-tagged RIG-I-CARDapaf1. The figure below shows the domains of RIG-I-CARDapaf1.

Figure3. Domain architecture of RIG-I-CARDapaf1

Following transformation of the plasmids into BL21(DE3) and subsequent protein purification, SDS–PAGE analysis revealed the expected bands for the target proteins, as shown in fig4.

Figure4. SDS–PAGE analysis of RIG-I–CARDapaf1

RIG-I–CARDapaf1 appears around 110 kDa.

>>dsRNA synthesis

Based on prior reports, template DNA for dsRNA synthesis was custom-ordered from IDT. Two complementary templates were designed such that each insert was flanked on both sides by oppositely oriented T7 promoters, enabling in vitro transcription of both strands from a single construct.

Figure5. Template DNA for in vitro dsRNA

dsRNA was generated from the template DNA using an in vitro transcription kit. As shown in the figure, a band was obtained at approximately the expected length. From a 20 uL transcription reaction, 36 ug of RNA was recovered.

Figure6. Gel electrophoresis of dsRNA

A band corresponding to 512 bp dsRNA was observed.

Results of the pull-down qRT-PCR

Purified RIG-I-CARDapaf1 and mutRIG-I-CARDapaf1 were incubated with dsRNA to observe their binding interactions. After incubation, mixtures were captured using Ni–NTA agarose pull-down. RNA was then extracted from the beads, and the recovered RNA was quantified using qRT-PCR. As a negative control, a sample containing only Ni-NTA agarose and dsRNA was prepared, and the difference was compared.

>>pull-down assay

Figure7. RT-PCR

dsRNA was precipitated using RIG-I–CARDapaf1. For samples labeled ‘mut’, see the next section for details.

As a result of RT-PCR, it was found that when the negative control prepared as beads only was reacted with dsRNA, the amount of dsRNA that precipitated fell down was the highest. From the above results, in the present experiment a significant precipitation of dsRNA by RIG-I–CARDapaf1 was not detected. Reasons can be considered, for example, that the experimental conditions are not optimized. Since it appears that a certain amount of dsRNA is precipitating even with beads only, the possibility that the washing of the beads was insufficient is high. Each sample was tested three times and error bars were attached, but the variation of the results was very large, and from this as well, our conclusion—that due to problems in experimental technique we are not able to observe the correct results—is supported.

Creating cDNA targeting dsRNA and performing qRT-PCR was a challenging task. However, as described above, the quantification itself was successful, and we expect that with optimization of the pull-down, correct conclusions will be obtained in the future.

Figure8. Measurement of the standard curve using dsRNA standard samples

A dilution series of synthesized dsRNA was prepared, and the standard curve was measured by real-time PCR.

>>Protein Purification

By following the published precedent we constructed a plasmid encoding GST-tagged CARDcasp9.

Figure10. Nucleotide sequence of the GST-CARDcasp9 expression construct

As shown in the figure, SDS–PAGE revealed a band corresponding to the target protein.

Figure11. Result of SDS-PAGE

>>pull-down assay

Figure12. Result of Western blotting

A band corresponding to the target protein appears near 110 kDa. A band near 20 kDa likely represents a degradation fragment.

Using GST-tagged fusion CARDcasp9 immobilized on glutathione–Sepharose beads, we pulled down RIG-I–CARDapaf1. The pulled-down RIG-I was detected by Western blotting using an anti-Flag antibody. The results of the Western blotting are shown in Figure12.

RIG-I–CARDapaf1 showed no co-precipitation. Because CARDapaf1 is originally a construct that has been reported to co-precipitate with GST–CARDcasp9, this was an unexpected result. Two possibilities can be considered as reasons for this. One is the possibility that the recombinant protein of RIG-I–CARDapaf1 is unstable and could not maintain sufficient interaction during the assay. The other is the possibility that because the RIG-I portion is fused unlike in the paper, steric hindrance prevented the CARDapaf1–CARDcasp9 interaction. Indeed, in the input signal, the expected 110 kDa band was faint, and an unexpectedly larger protein band was observed. This suggests the possibility that the folding of the fusion protein was not appropriate or that aggregation was occurring.

For non-CARD–RIG-I, precipitation was observed even when CARDcasp9 was not bound to the beads. We consider that this is because this fusion protein has a property of readily precipitating.

From the above results, we did not succeed in observing an interaction between the target RIG-I–CARDapaf1 and GST–CARDcasp9. We consider that this is attributable to the instability of the recombinant proteins. In COCCO, because the recombinant proteins are expressed in cells without purification, we expect that such problems can be avoided.

>>Protein Purification

Figure15. Domain architecture of mutRIG-I–CARDapaf1

The mutated amino acids were selected based on the three-dimensional structure, as described in the Modeling section. We chose residues predicted to be directly involved in the CARD–HEL2i interaction.

These plasmids were transformed into BL21(DE3), and upon protein purification, SDS–PAGE revealed bands corresponding to the target proteins, as shown in Figure16.

Figure16. SDS–PAGE of mut-RIG-I–CARDapaf1

>>dsRNA preparation

The dsRNA generated for the experiments in the preceding section was also used in this experiment.

>>Pull-down assay

As in the preceding section, we performed a pull-down assay and quantified the amount of dsRNA precipitated with mutRIG-I–CARDapaf1.

Figure17. Result of RT-PCR

Using RIG-I–CARDapaf1 or mutRIG-I–CARDapaf1 bearing mutations in HEL2i, we pulled down dsRNA.

RT-PCR analysis revealed that the negative control (beads only) recovered the highest amount of dsRNA, indicating some inadequacies, such as insufficient washing. Therefore, we cannot draw conclusions about specific protein-dsRNA binding from this experiment.

Interestingly, mutRIG-I–CARDapaf1 with mutations in HEL2i consistently yielded less recovered dsRNA than RIG-I–CARDapaf1 with wild-type RIG-I. This may indicate that introducing mutations into HEL2i, which is the dsRNA-binding domain, reduced dsRNA-binding ability. In that case, introducing mutations into HEL2i would decrease performance as a dsRNA sensor, and thus it may be more appropriate to introduce mutations into CARDapaf1 to enhance the CARDapaf1–HEL2i interaction rather than mutating HEL2i itself.

>>Protein Purification

GST-tagged CARDcasp9 was purified as described above.

Figure19. Domain architecture of GST-CARD

Multiple plasmids expressing RIG-I–mutCARDapaf1 (bearing mutations in CARDapaf1) were constructed, transformed into BL21(DE3), and the proteins were purified. As shown in the figure, SDS–PAGE revealed bands corresponding to the target protein.

Figure20. Result of SDS-PAGE

>>pull-down assay

Figure21. Result of Western blotting

Using GST-tagged CARDcasp9 immobilized on glutathione–Sepharose beads, we pulled down RIG-I–mutCARDapaf1. The pulled-down RIG-I–mutCARDapaf1 was detected by Western blotting using an anti-Flag antibody. The results of the Western blotting are shown in Figure21.

mutRIG-I–CARDapaf1 (the band near 110 kDa) included cases that, like ΔCARD RIG-I, precipitated in the tube due to aggregation regardless of the presence of GST–CARDcasp9 (mut2 and mut4), and cases that, like RIG-I–CARDapaf1, showed no precipitation under any condition (mut3, mut5).

From the above results, we did not reach the point of observing an interaction between mutRIG-I–CARDapaf1 and GST–CARDcasp9. As discussed for the wild-type protein in the previous section, we consider that this is attributable to the instability of the recombinant proteins. In COCCO, because the recombinant proteins are expressed in cells without purification, we expect that such problems can be avoided.

>>Protein Purification

The hel2i2 fragment was amplified by PCR from the plasmid expressing mutRIG-I-CARDapaf1 and assembled into pGEX-6p-1 to generate a plasmid expressing GST-tagged hel2i2.

Figure23. Sequence of GST-hel2i2

Flag-CARDapaf1 purified in the previous section was used. In the case of wild-type RIG-I, the hel2i domain interacts with its own CARD. As a positive control, a plasmid expressing the CARD domain of RIG-I with a Flag tag (Flag-CARDrig-I) was constructed.

Figure24. Sequence of Flag-CARDrig-I

These plasmids were transformed into BL21(DE3) for expression, and the recombinant proteins were purified. As shown in the figure, SDS-PAGE revealed bands corresponding to the target proteins.

Figure25. SDS-PAGE of GST-hel2i purification

>>pull-down assay

Figure26. Result of western blotting using His-Flag-CARDapaf1

Flag-CARDapaf1 was observed as a band of approximately 10–17 kDa.

Figure27. Result of western blotting using His-Flag-CARDrigI

Flag-CARDrigI was observed as a band of approximately 10–17 kDa.

Flag-CARDapaf1 was pulled down using GST-tagged hel2i immobilized on glutathione sepharose beads. As a positive control, His-Flag-CARDrigI, which is expected to interact with RIG-I hel2i, was also pulled down. The pulled-down proteins were detected by western blotting. The results of the western blotting are shown in Figure26 and 27.

His-Flag-CARDrigI bound to both wild-type and mutant hel2i. However, His-Flag-CARDapaf1 did not bind to either wild-type or mutant hel2i.

Although the interaction between the hel2i mutants and CARDapaf1 was not observed, the differences in the intensity of His-Flag-CARDrig-i bands precipitated by the hel2i mutants indicate that the mutations designed through modeling were able to modulate hel2i’s binding to CARD.

>>Protein Purification

Figure29. Domain architecture of PKR-CARDapaf1

Based on the literature, we designed a plasmid expressing DRACO (TAT-PKR-CARDapaf1) as shown in Figure29.

Figure30. SDS-PAGE of DRACO purification

The plasmid was transformed into BL21 cells to express DRACO (TAT-PKR-CARDapaf1). The recombinant protein was purified using Ni-NTA sepharose through the His tag. As shown in Figure30, a band was observed by SDS-PAGE.

>>Reproduction experiment of DRACO

Figure31. Result of reproduction experiment

In DF1 cells, the fractions of PI-positive cells and Annexin V-positive cells were measured after transfection with each of the following: DRACO stored at −80 ° + dsRNA, DRACO stored at 4 °C + dsRNA, DRACO only, dsRNA only, and non-treated control. dsRNA is 25 ng/well.

• A, Fraction of PI-positive cells in HEK293 under each condition.
• B, Fraction of Annexin V-positive cells in HEK293 under each condition.
• C, Fraction of PI-positive cells in DF1 under each condition.
• D, Fraction of Annexin V-positive cells in DF1 under each condition.

Using the prepared dsRNA and DRACO, we performed a reproduction experiment of DRACO. Specifically, both DRACO and dsRNA were transfected into HEK293 and DF1 cells using FuGENE 6 Reagent. The former is a human-derived cell line known for its high transfection efficiency, making it suitable for simple characterization of our parts. The latter is a chicken-derived cell line that was chosen because it allows observation of infection and propagation of avian influenza virus.

Twenty-four hours after transfection, we attempted to observe DRACO-induced, dsRNA-dependent apoptosis by staining the cells with PI and Annexin V and observing them under a fluorescence microscope. PI stains all dead cells, while Annexin V stains those that have undergone cell death through the apoptotic pathway. The results were evaluated by comparison with three negative controls: cells transfected with DRACO only, cells transfected with dsRNA only, and non-treated cells.

In the experiment using HEK293 cells, many cells had detached from the wells. This may have affected the results, and therefore no conclusion could be drawn as to whether the effect of DRACO was observed.

From these results, in DF1 cells, co-transfection of DRACO and dsRNA resulted in higher fractions of PI-positive and Annexin V-positive cells compared to cells transfected with DRACO alone or non-treated cells, suggesting that apoptosis may be induced by the DRACO system. However, since the highest ratios were observed in cells transfected with dsRNA alone, it is possible that the dsRNA itself exerts cytotoxic effects on the cells.

>>Generation of cell lines

In verifying COCCO, the low transfection efficiency of DF-1 cells became a problem. Since avian influenza viruses can basically replicate only in chicken cells, DF-1 cells had to be used for proof-of-concept experiments involving avian influenza virus. However, if the transfection efficiency is low, the plasmid introduction efficiency will also be low, and even if our COCCO can efficiently induce cell death, the number of apoptosis events occurring within the cell population will be small. Considering that transfection itself can induce cell death, it would be ideal to create a stable cell line that constitutively expresses COCCO, so that all cells in the population express COCCO.

For this purpose, after transfecting the plasmids into HEK293 and DF-1 cells, selection with neomycin was performed to create stable cell lines. This was because all COCCO and negative control plasmids contained a neomycin resistance gene. Normally, after screening with neomycin resistance, single cells are selected and cloned to obtain pure stable transformants. However, since this experiment was conducted in two months, there was not enough time for cloning. Therefore, we decided to conduct subsequent experiments using cell lines in which various stable transformants were mixed.

To establish stable cell lines, we used pTwist CMV Neo and designed plasmids that express the following three COCCO proteins under the control of the CMV promoter.

• RIG-I-CARDapaf1 (fig32-1）
• PKR-CARDapaf1 (fig32-2)
• PKR-ΔCaspase9 (fig32-3)

Figure32-1. Domain architecture of RIG-I-CARDapaf1

Figure32-2. Domain architecture of PKR-CARDapaf1

Figure32-3. Domain architecture of PKR-ΔCaspase9

In addition, as negative controls, we designed plasmids expressing the following four proteins. We also created a stable cell line transfected with the vector alone, without insertion of any gene encoding the proteins to be expressed.

• non CARD RIG-I (fig33-1)
• CARDapaf1 (fig33-2)
• PKR (fig33-3)
• ΔCaspase9 (fig33-4)
• vector only

Figure33. Domain architecture of non-CARD RIG-I

Figure34. Domain architecture of CARDapaf1

Figure35. Domain architecture of PKR

Figure36. Domain architecture of ΔCaspase9

Figure37. DF-1 stable cell line expressing RIG-I-CARDapaf1

We confirmed that the desired stable cell lines had been established. In subsequent experiments, all observations were performed using these stable cell lines.

>>Verification of protein expression in stable cell lines

The expression of the transgenes was confirmed by Western blotting using an anti-Flag antibody. In the HEK293 stable cell lines, recombinant expression was detected except for CARDapaf1 and PKR-ΔCaspase9, whereas in DF-1 cells, expression was not observed except for PKR-CARDapaf1. As shown below, expression of the target protein was not detected in some samples. Therefore, subsequent results focus exclusively on those with confirmed expression. As a control for Western blotting, an anti-β-actin antibody was used to determine the amount of cells, and similar intensity bands were observed in both HEK293 and DF-1 cells.

Figure38. Result of Western blotting of stable cell line

The left figure shows a Western blot using an anti-Flag antibody, and the right figure shows a Western blot using an anti-β-actin antibody.

>>Optimization of dsRNA concentration

In the previous DRACO reproduction experiment, transfection with dsRNA alone induced a certain level of cell death, which obscured the cell death-inducing effect of DRACO. Therefore, we first examined the concentration of dsRNA to determine the optimal conditions for the assay.

In the HEK293 stable cell line that was generated, dsRNA was transfected at concentrations ranging from 25 ng/well to 250 ng/well. FuGENE 6 Reagent was used as the transfection reagent.

Figure39. Result of dsRNA optimization using stable cell line

For the stable cell lines expressing COCCO and their negative controls, the fraction of PI-positive cells and the fraction of Annexin V-positive cells were determined for each concentration of transfected dsRNA.

• A, Fraction of PI-positive cells in wild-type HEK293 cells (not a stable cell line) transfected with each concentration of dsRNA.
• B, Fraction of Annexin V-positive cells in wild-type HEK293 cells (not a stable cell line) transfected with each concentration of dsRNA.
• C, Fraction of PI-positive cells in HEK293 stable cell line expressing RIG-I-CARDapaf1 transfected with each concentration of dsRNA.
• D, Fraction of Annexin V-positive cells in HEK293 stable cell line expressing RIG-I-CARDapaf1 transfected with each concentration of dsRNA.
• E, Fraction of PI-positive cells in HEK293 stable cell line expressing PKR-CARDapaf1 transfected with each concentration of dsRNA.
• F, Fraction of Annexin V-positive cells in HEK293 stable cell line expressing PKR-CARDapaf1 transfected with each concentration of dsRNA.

For the analysis, we assumed that the fluorescence intensity per cell area would separate into positive and negative populations. We created violin plots of the fluorescence intensity for each cell and used the local minima in the distribution, which were considered to represent the boundary between the positive and negative populations, as the threshold.

Figure40. Fluorescence expression per unit area and minimal values by cell type

Fluorescence intensity of Annexin and PI per cell area and the local minima in the distribution on the violin plots. For DF-1 cells, a clear distinction between the positive and negative populations could not be observed; however, for convenience, the smallest local minimum was used as the threshold.

Our objective was to identify a dsRNA dosage that activates COCCO while maintaining the viability of wild-type cells. However, the unexpected death of wild-type cells, including non-treated controls, made it difficult to determine the optimal condition. Due to time constraints precluding a repeat experiment, we fixed the dsRNA concentration at 150 ng/well based on the available data.

>>dsRNA transfection in DF-1 stable cell lines

dsRNA transfection at 150 ng/well was performed on the DF-1 stable cell lines that were generated: PKR-CARDapaf1, Vector only, and on wild-type cells. (Note: This experiment was conducted before the dsRNA concentration optimization experiment described above, so a higher concentration than the optimal was used.)

The results were observed 24 hours later. The results are shown in the following graph.

Figure41. Result of dsRNA transfection in DF-1 stable cell lines

Annexin expression levels upon transfection of 150 ng/well dsRNA in DF-1 stable cell line expressing PKR-CARDapaf1, in positive control (wild-type cells + sodium azide) and in negative controls (vector only and wild-type cells).

The graphs indicate that dsRNA transfection induced higher levels of apoptosis in the PKR-CARDapaf1 stable cell line compared to the control construct. This provides the first evidence that at least one COCCO construct can induce cell death in a dsRNA-dependent manner. Results for the other COCCO constructs remain inconclusive.

Although no statistically significant difference was observed, significance might have been detected with a larger sample size or an optimized dsRNA concentration. Additionally, since this experiment used a mixed cell population, apoptosis induction could be more pronounced if a monoclonal cell line were used.

Infection experiment in DF-1 cells

>>FACS analysis

Figure43. Result of FACS analysis at 6 hours post infection (n=3)

We investigated whether stable cell lines expressing COCCO proteins induce apoptosis upon viral infection. Specifically, DF-1 stable cell lines were infected with influenza virus (H3N8) at an MOI of 10, and cells at 3, 6, 9, 12, and 14 hours post-infection (hpi) were stained with eBioscience™ Fixable Viability Dye eFluor™ 780, a fixable dead cell stain, and Annexin V, followed by fixation with 1% formalin. After fixation, the samples were analyzed by Dr. Daiya Ohara using FACS. Cells that were negative for eBioscience™ Fixable Viability Dye eFluor™ 780 and positive for Annexin V were considered apoptotic, and the fraction of apoptotic cells was calculated. The proportion of apoptotic cells was compared between wild-type and each of stable cell lines using a t-test. The p-values were above 0.05, and no statistical significance was observed.

Virus titration

We investigated whether stable cell lines expressing COCCO proteins could suppress viral replication. Specifically, at 9 and 12 hours post-infection (hpi), viruses propagated from RIG-I-CARDapaf1, PKR-CARDapaf1, vector-only, and wild-type cells were used to infect MDCK cells in a serial dilution, and the TCID50 was measured. Comparing the PKR-CARDapaf1-expressing cell line with the wild type reveals a higher proportion of apoptotic cells in the former. However, the CARDapaf1 and PKR cell lines, for which protein expression was not confirmed, exhibited even higher rates of apoptosis.Since the CARDapaf1 and PKR lines are expected to yield results similar to the wild type or vector-only controls, this discrepancy suggests that the primary issue with this experiment is a lack of data stability.

Figure44. Result of titration experiments at 6 hours

From the results of the graph, at both 9 hpi and 12 hpi, the COCCO-expressing cells RIG-I-CARDapaf1 and PKR-CARDapaf1 showed no significant difference in viral replication compared to the controls. It cannot be excluded that 9 hpi and 12 hpi corresponded to the early stage of infection, and measuring at later time points would have been preferable.

>>Infection experiments for HEK293

Figure45. Result of microscopic analysis at 6, 9 and 12 hours post infection

The Annexin intensity threshold was defined as described previously to determine the percentage of Annexin-positive cells, excluding 12 hpi data due to elevated overall intensities. Data with excessive PI staining due to imaging artifacts were excluded. Analysis was limited to objects between 20 and 200 µm².

We investigated whether stable HEK293 cell lines expressing COCCO proteins induce apoptosis upon viral infection. Specifically, HEK293 stable cell lines were infected with influenza virus (H3N2) at an MOI of 10, and cells at 6, 9, and 12 hours post-infection (hpi) as well as non-infected controls (MOCK) were stained with Hoechst, Annexin V, and PI.

In the graph, the value at 9 hpi was lower than at 6 hpi, which was attributed to variability in cell numbers prior to infection due to experimental handling. Since the amount of virus per well was uniform, wells with higher cell numbers at 9 hpi had lower virus per cell compared to other time points. A lower MOI reduces infection efficiency.

Therefore, the data at 6 hpi and 12 hpi were primarily used for discussion of the results.

At 12 hpi, the proportion of cells with mean Annexin intensity above the threshold was very close to 1 regardless of cell type, indicating that most cells had reached apoptosis.

At 6 hpi, Compared to wild-type cells, PKR-CARDapaf1 expressing cell line showed higher proportions, suggesting that PKR-CARDapaf1expressed in these cells tend to induce apoptosis.

Nevertheless, data stability remains a significant concern in the current study. As several wells were unanalyzable, further optimization of the experimental conditions for infection is necessary.

Virus titration experiment

Viral titration was performed to investigate whether stable cell lines expressing COCCO proteins can suppress viral replication. Specifically, at 9 hours post-infection (hpi), viruses propagated from RIG-I-CARDapaf1, PKR-CARDapaf1, PKR-ΔCaspase9, CARDapaf1, vector-only, and wild-type cells were used to infect MDCK cells in a serial dilution, and the TCID50 was measured.

Figure46. Result of titration experiments with HEK 293 cell line

From the graph, it cannot be concluded that the COCCO-expressing cells RIG-I-CARDapaf1 and PKR-CARDapaf1 significantly suppressed viral replication compared to the control cell lines.