Scientist Stories: Yizhou Dong, RNA Engineering for Therapeutics, Genome Editing & Cell Therapy

So hello everybody, and welcome to this month's ERCC webinar. I'm Roger Alexander, the Scientific Outreach Coordinator for the Extracellular RNA Communication Consortium. We're in a second phase of the consortium now. The initial phase was focused on understanding the biology of extracellular RNA, as well as diagnostic and therapeutic applications. And in the second phase, we're doing more technology development focusing on being able to analyze single extracellular vesicles and characterize exRNA carriers. Sort of in that vein, we're pleased to have with us Yizhu Dong of The Ohio State University. Today he's going to talk to us about nanomaterials and RNA engineering for mRNA therapeutics, genome editing, and cell therapy. Before you start, Yizhu, If people have questions throughout the talk, they can put their questions in the chat, and I'll read those first at the end, and then I'm sure we'll have time for people to unmute themselves and ask live questions. So, Yuzhu, look forward to hearing about your work.

Thanks, Roger, for the kind invitation. It's really my great pleasure to share with you some of our recent developmental nanomaterials and the RNA engineering for various applications. So first, I would like to give you an overview about the research in our lab. We are focusing on developmental technology platforms which can be applied to diverse diseases. The first direction is lipid or lipid-derived nanoparticles. We size different type of lipids or lipid derivatives and then we can formulate them into lipid nanoparticles, and we can assemble different type of therapeutic cargoes such as siRNA, messenger RNA, or CRISPR system. The second platform is to engineer different type of RNA such as messenger RNA or genome editing components. The goal is to increase the protein production or improve the gene editing efficiency. The third direction we recently work on is adoptive cell therapy. For example, we can engineer macrophages for treating infectious diseases. So today I'll introduce you several examples from each direction. So methadone has become a very promising class of drugs, and especially during the current COVID-19 pandemic, there are two messenger RNA-based vaccines that have been used in the, in many countries against SARS-CoV-2. And meanwhile, messenger RNA has been used for many therapeutic indications such as cancer immunotherapy, protein replacement therapy, and also genome editing. Although mRNA has become a very promising class of drug, delivery is still a key challenge for the broad application of this type of new agents in humans. So there are many biological barriers need to be overcome. So for this group of audience, I don't think I need to give a lot of background about this. And to overcome all these biological barriers, we really need to develop a new type of biomaterials. And based on our previous experience on the delivery of different type of RNA, we designed this type of lipid-like compounds or lipid derivatives shown here, and it has a long chemical name. We just give it a short name, TT-Lipids. These TT-Lipids are composed of a phenyl ring, 3 amide linkers, and multiple lipid chains. And we can tune the carbon chain here from 2 carbons to 8 carbons or even longer, and also we can tune the end lipid chains as well. Through this very simple one-step reductive amination reaction, we can synthesize these TT lipids in large scale. And this slide shows the overall strategy to optimize the lipid-like nanoparticles for mRNA delivery in vivo. The newly synthesized PT lipids can be formulated with helper lipids such as phospholipids, cholesterol, and PEG lipid, and here the therapeutic cargo is messenger RNA. Through this microfluidic-based device, we can assemble them into very nice nanoparticles. Then we can characterize the particle properties such as particle size, their potential entrapment efficiency, and also mRNA translation efficiency. We utilize the orthogonal experiment design, which can help us identify the optimal formulation with significantly reduced experiment load. And after that, we can test the optimal formulation in different animal models. And here's the in vitro screening. We formulated the TT lipids with mRNA encoding firefly luciferase,, and then we treated a liver cancer cell line, Hep3B cells, and on the left, the y-axis, that's luminescence intensity, and the x-axis, that's different TT lipids. We can see TT3 is better than other TT lipids, also much better than the control. Next, we studied the effects of phospholipids on the delivery efficiency. We chose three widely used phospholipids, DSPC, DOP, and POP. You can see DOP-formulated TT3 is much better than DSPC or POP-formulated TT3. So we chose DOP-formulated TT3 for further optimization. So here is orthogonal experiment table. As I just mentioned, we utilize this orthogonal experiment design to optimize the formulation ratio of these TT lipids. And in this formulation, we have 3 components— 4 components: TT3A, DOP, cholesterol, and PEG lipids. And we assign the 4 levels for each component. And these orange highlights, that's original the formulation ratio. Because we don't know the effect of each component on the delivery efficiency, we just arbitrarily assign the 4 levels for each component. So 4 levels and 4 components give overall 256 combinations. But using this orthogonal experiment design, we only need to do 16 experiments, which represent all these 256 combinations. And after analysis based on the results from these 16 experiments, we can predict the optimal formulation from these 4 levels and the 4 components. And then here on the right, this is orthogonal experiment table. And from number 1 to number 16, That's 16 different formulation ratios for each component and each level of all these components. And the next, we did the same in vitro study in Hep3B cells. The y-axis is luminescence intensity, x-axis, that's 16 different formulation ratios. And then we analyze the results. In this experiment, we utilize the same Firefly Luciferase So that means the higher luminescence intensity, the higher delivery efficiency. And then in this right table, so the K' number, that's the total luminescence intensity for each component at each level, and we average that. That's a K' number. And so you can see TT3, the level 1, is best. TOP, level 4, cholesterol type O2 and the pack lipid type O1. And at the bottom here is the predicted optimal formulation. More importantly, this experiment can tell us the trend for each component on the delivery efficiency. And all these y-axis, that's the average luminescence intensity. X-axis, that's molar ratio for each component. You can see the lower molar ratio, the TT3, the higher delivery efficiency. The higher DOP, the higher delivery efficiency. Plasma molar ratio between 20 to 40, and the less panglipid, higher delivery efficiency. So this information is very important, which enable us to further optimize the formulation ratio. Then we can did several rounds of optimization and validation, eventually identify the optimal formulation for the PT lipids. To save some time, I'll just escape that process. And then we selected a model disease. Here is hemophilia. Hemophilia is a genetic disorder caused by missing or mutation of factor VIII or factor IX. Patients suffer from life-threatening bleeding or serious complications such as from hemorrhage or from bleeding. So in this study, we formulated TT3 with mRNA encoding human factor IX, and then we did intravenous injection first in wild-type mice. On the left, the y-axis, that's the human factor IX concentration in plasma, and the x-axis, different treatment groups. And you can see it showed a dose-dependent recovery of the factor IX. In mouse plasma. Similarly, we did the experiments in Factor IX knockout mice, and this is the hemophilia B mouse model. You can see the trend is the same. Then we are wondering if these Factor IX proteins are active or not. We performed a Factor IX activity assay, and this is the y-axis, that's human factor IX activity in MIU per mL, access at different treatment groups. And here the trend is same. It also showed that dose-dependent recovery of the factor IX activity. At 1.1 milligram per kilogram, factor IX activity is around 800 MIU per mL. In normal plasma, the factor IX activity is between 500 to 1,500 MIU per mL in humans. So that means that TT3 is able to fully recover the factor IX activity to a normal range at this dose in these factor IX knockout mice. And recently, we collaborated with the groups of Ron Weiss and Darryl Irvine, and we formulated a TT3 with self-replicating RNA encoding IL-12 for cancer immunotherapy. And here is an illustration of the strategy. First, we formulated the nanoparticles with the replica, and then we did intratumoral injection, and it triggered ICD. It's immunogenic cancer cell death, and then subsequently triggered the antitumor immunity. They performed the in vivo experiments in many mouse tumor models. Here are some examples. After a single injection of the lipid nanoparticle formulation, you can see in 3 different mouse tumor models, B16F10, CD26, and UMR1.7, This formulation significantly extended the mouse survival. They also performed many maximum action studies, and here is one example. We can see if we knock out MID88 or STEAM, the therapeutic efficacy is significantly compromised. Meanwhile, we did a lot of chemical modifications of the TT lipids. We installed the carbon chains, branched ester chains, or linear ester chains on the scaffold, and then we produced these functionalized lipid-like compounds. And then we formulated them with mRNA encoding firefly luciferase, and we injected intravenously in mice and screened their mRNA delivery efficiency in vivo. So on the left here, the y-axis, that's a fold-over increase normalized by TD3, x-axis different FTD lipids. We found that FTD5 is much better than other FTD lipids and also better than TD3. And on the right here is the characterization of the particle properties. We can see the particle size is around 100 nanometers and the PDI is between 0.1 to 0.3. We also studied the biodegradability of these different lipids. We chose two representative examples. One is the FTT lipids with linear ester chain on the right and the other one is FTT lipids with branched ester chain on the left. You can see both particles showed a very low concentration in the mouse plasma after intravenous injection, but in the liver, the lipids with linear ester chain is cleared very fast. For the other lipids with branched ester chain, the degradation rate is much slower. So we can tune the chemistry of the TT lipids and then adjust the degradation rate of these nanoparticles. Then we collaborated with a group of Dr. Dennis Sabatino at the Children's Hospital Philadelphia. We formulated FTD5 with mRNA encoding factor VIII, and then we did the treatment in both wild-type mice and hemophilia A mice, and we noticed that both the factor VIII protein and the activity recovered to almost a normal range compared with wild-type mice. We also performed a preliminary analysis of the histology from the untreated mice and also the particle-treated mice. We found that there's no significant change in these major organs heart, liver, spleen, lung, and kidney. Then we collaborated with Beam Therapeutics, which is a leading company for base editing, and we formulated FTD5 with mRNA encoding base editor and also guide RNA targeting PCSK9 gene. And on the left, the y-axis, that's base editing, efficiency A to G, X has different doses. It also showed a dose-dependent base editing efficiency. And on the right, that is one piece of the deep sequencing results at the dose of 0.125 milligrams per kilogram. We can see here the base editing efficiency is still over 50%. So we are very excited about these results, and our collaborators and also us, we think is the most efficient delivery system for base editing reported in literature. And currently, these materials have been licensed by multiple biotech companies and further developed them for different type of diseases. Next, I would like to share with you our efforts on the RNA engineering. And the first example is to engineer the CRISPR-Cpf1 system for genome editing. And Cpf1 was first reported by the group of Feng Zhang at MIT, and this is a type 5 CRISPR-Cas effector endonuclease. And on the left in Figure A, this illustration of the Cpf1, it also termed as Cas12a, and the purple is the Cpf1 protein, It is functional through a single guide RNA. It can recognize the T-rich PAM region and induce the double-stranded DNA breaks. And on the right here is the illustration of the guide RNA. It's a relatively short RNA composed of 5' handle, also called the direct repeat, and also a 3' spacer or guide segment. And this also includes a seed region and the 3' permanent. So we know there are a lot of chemically modified nucleotides, and here are some examples such as phosphosylate, 2'-methoxy, 2'-fluorine, unlocked nucleotides, locked nucleotides. These chemically modified nucleotides have been used for modifications of siRNA or other type of RNA. So we are thinking maybe we can incorporate these chemically modified nucleotides into our guide RNA. And here we designed various chemically modified guide RNA, such as phosphosylate at the full length, or just the 5' end and 3' end, or only 5' end, or only at the seed region, or 3' termini. So we synthesized these guide RNA, then we tested the gene editing efficiency. So we utilize T7E1 assay, which is widely used in the gene editing field. And here the CRWT, that's wild-type guide RNA. So all other guide RNA, the gene editing efficiencies normalized with the wild-type guide RNA. We can see several guide RNA completely lost their gene editing efficiency. But quite a few of them showed a comparable gene editing efficiency, and particularly one guide RNA with 5' 2' fluorine modification at the 3' end significantly increased the gene editing efficiency to about 120 to 130%. Encouraged by these results, we think we can further modify the messenger RNA encoding the Cpf1. And on the left here is natural uridine, and on the right these are widely used chemically modified U, such as 5'-methyl-5'- and also 5'-molybdenum. So we produced this messenger RNA, and we utilize the AsCpf1 plasmid as a control. And this is a generous gift from the group of from John, and we found one particular RNA with 5 modifications is significantly better than plasmid and also better than other messenger RNA. Then we are thinking maybe we can combine the chemically modified guide RNA and the chemically modified messenger RNA to further enhance the gene editing efficiency. And so we tested this combination in multiple cells at multiple gene loci. You can see the red column, that's the combination of chemically modified guide RNA and chemically modified messenger RNA. It showed a consistent increase of gene editing efficiency in 3 different gene loci and in 3 different human cell lines. And then we performed a similar study in another Cpf1 family member called LbCpf1., and we found the increase is more profound. And for the combination of the chemically modified mRNA and the guide RNA, it showed dramatic gene editing efficiency. But the combination of the wild-type guide RNA and the plasmid, we cannot detect obvious gene editing. To further confirm the gene editing efficiency, we performed the targeted deep sequencing. And on the left here is on-target gene editing. On the right, that's the analysis of off-target effects. So we can see the chemically modified combination showed at least threefold better than the original combination with the wild-type guide RNA and plasmid., and also we didn't notice any significant increase of the off-target effects. So encouraged by these results, we think we can further engineer the sequences of the RNA, and in this study we think we already know like the untranslated region of the messenger RNA is very important for the regulation of the mRNA properties, So we think maybe we can engineer the 5' and 3' level, the UTR, and then improve the protein production. So that's our study rationale for this, and we utilize two approaches. One is to analyze the expression of the endogenous genes, and the other approach is the de novo design of new type of the 5' and 3' UTR. And after several years of the optimization and analysis of results, then we identified a lead UTR. We call that NASA UTR. And then, like last year, the COVID-19 outbreaks occurred. We decided to utilize these NASA mRNA to encode SARS-CoV-2 antigens.. And then we did the vaccination of the wild-type mice. Then we can analyze the antigen expression and also the specific production of the antibodies. And then here's the initiation of the study. We think maybe we can analyze endogenous genes that we found in the literature, over 4,000 different genes. And there's results we can analyze how many protein copies or amino acids produced from a single copy of messenger RNA, and then we can analyze and rank all these UTRs. We found that in particular one family that's called RPS27A, and it's a part of the ribosome 40S unit.. And in this family, there are two members. We give a short name S27A-45 and S27A-44. So this family, they have different 5' UTR, but they share the same 3' UTR. So we produced these mRNA transcripts encoding the firefly luciferase. We also selected several control UTR reported in the literature, such as CYBA, AG, and also some engineered UTR, AG+G, 5' AG+G, and also no UTR. And then we found that we also like analyze the 5' UTR sequence. We found that there are some help region is a terminal oligopyrimidine region, which was reported inaccurately correlated with the translation. So we removed those regions from the 5' UTR. We generated this S27A-45', and also S27A-44'. And we found these messenger RNAs showed comparable expression of the luciferase with the best UTR reporting in the literature. In our study, we tested CYBA and AG+G. And the next, we studied the 3' UTR. We chose some, like, abundant UTR from different origins, such as the same, this RPS27A, or some from some viral proteins or transpiring. And we found that these UTRs, they showed a similar level of luciferase production. And then we explored our second approach, that's a de novo design. We set up several criteria such as minimum secondary structures in the 5' UTR, or like only one like a start codon in the sequence AUG, and also we incorporate a codon sequence in the 5'-UTR. And based on the sequences, we designed the 5'-UTR with different lengths from 10 nucleotides to 90 nucleotides. And from the left figure, you can see when we increase the length of the UTR and the protein production efficiency is increased. But if it's too long, in our case, the production is reduced when it's reached 19 nucleotides. And we also know the combination of this AUGC is very important for the translation. So we also studied the different combination of this AUGC, but we keep the length constant as 70 nucleotides. Sorry. So that's Figure B. And the Figure C, we try to remove any, like, microRNA binding sites, and we found for this particular UTR NCA-7 and NCA-7D both are pretty good. 7D is slightly better than 7. So we chose this NCA-7D as our 5' UTR. And then next, we try to optimize the 3' UTR. In the literature, there are many, like, functional motif reported. Which can bind with different type of RNA-binding proteins. So we identified quite a few sequences, then we inserted the sequences after the 3' UTR. So here are a few examples shown on the left, and we found one particular functional motif we give name R3U, it's much better than other motif and also better than the original, like, positive control CYBA, AG+G, or without motif. So after all this optimization process, we selected this NC837D as 5' UTR, S27A plus this functional motif as the 3' UTR, and we gave it a short name like NASA UTR. And then we found that there are more and more UTR reported in the literature, and we selected two UTRs, MOD1 and MOD2, as another two sets of positive control. And then we tested the luminescence intensity. We found that after all this optimization process, NASA UTR is over tenfold better than our start point. That's the original, like, the first UTR we tested from endogenous gene, and also it's significantly better than the two UTRs reported in the literature. And the next, we plan to utilize this messenger RNA to encode SARS-CoV-2 antigens And on the left, here is an illustration of the potential SARS-CoV-2 antigens, including the spike protein, nucleocapsid, membrane protein, envelope protein, and also this receptor binding domain. And we first produced this NASA mRNA encoding the receptor binding domain. We would like to know if it still can increase protein production. We found that consistently it's better than CYBA and better than the model 2 reported in the literature. Then we realized the expression of these SARS-CoV-2 antigens. For all these 5 different potential antigens, we can see they're like always production in vitro. Next, we would like to test the in vivo activities. We then chose two different delivery vehicles. One is MC3, that is FDA-approved lipid formulation used for siRNA delivery. And also there are quite a lot of preliminary preclinical studies utilize MC3 for delivery of different type of mRNA. And another delivery vehicle is TT3 we just mentioned. And after intramuscular injection, we found TT3 is much better than MC3. And then we formulated these two lipids with spike mRNA, and we only find the specific serum IgG. We found TT3 induced over 300-fold higher specific antibodies than the M3 formulated formulation. Next, we compared two different administration routes, intramuscular injection and subcutaneous injection. In our study, we found intramuscular injection is better than subcutaneous injection. Lastly, we did those dependency study. We found that for both spike mRNA and RBD mRNA, at high dose there is saturation effect. At the lower dose, there is dose-dependent production of the serum IgG. We also found that at lower dose, spike mRNA induced higher level of the serum IgG than RBD. It's also consistent with the literature report. Lastly, I would like to switch gears to our recent studies on nanoparticles-enabled adoptive cell therapy for treating sepsis. So when we design new type of materials, we try to learn from the nature. When we walk into the pharmacy stores, we saw many different type of vitamins. And the vitamins are very important components. They have essential roles for many cellular pathways and functions. And the chemically, vitamins also possess very unique chemical groups such as hydroxy groups, charges, or like hydrophobic groups. And here, for example, is orange, there is vitamin C, and we are thinking maybe we can engineer these vitamins and develop vitamin-derived nanoparticles and then deliver messenger RNA to engineer macrophages, which can be applied to treat infectious diseases. Such as sepsis. So here, like, sepsis is a life-threatening condition. This disease affects over 30 million people worldwide. It remains the number one cause of death in hospitals. Although there's significant advances in intensive care and antibiotics treatment, the mortality rate is still pretty high, around 25 to 30% for this disease. And about 70 to 80% of sepsis deaths is related to persistent infection. And also, multidrug-resistant bacteria significantly increase the challenge for treating this disease. Meanwhile, for sepsis patients, these infections severely paralyzed the immune system of the patients, such as the macrophages. And so here are many challenges for treating this type of deadly disease. This slide shows our strategy to treat sepsis, especially the immunocompromised sepsis. And we are thinking we can produce different type of vitamin-derived lipid nanoparticles then deliver antimicrobial components into macrophages. Then we can transfer these macrophages into the sepsis hosts, which can enhance the innate immunity and also eliminate all the drug-resistant bacteria. To test our concept, we first synthesized a series of vitamin-derived lipids. And here are some examples, vitamin B3, vitamin C, vitamin D, vitamin E, vitamin H. And we have an amino chain here which can interact with messenger RNA under acidic conditions. And we performed a similar screening assay. And here is in a macrophage cell line. It is RAW264.7 cells. And from the left figure, you can see vitamin C is much better than other vitamins, also better than Lufectamine, this commercially available transfection agent, and also better than electroporation. And then we also found that the protein production reached the peak around 12 hours. Then we did several rounds of optimization, as I mentioned, using orthogonal experiment design. After all these optimization steps, we identified one particular formulation. Here is C5. It is over 7-fold better than the original formulation we tested for the vitamin C, and it is over 70-fold better than Lefractamine 3000 over 300-fold better than electroporation. On the right here is a cryo-EM image of the nanoparticles. You can see these particles are relatively homogeneous in the spherical morphology. We also studied the endocytosis pathway of the vitamin C-derived lipid nanoparticles. We selected 3 widely used endocytosis pathways for particles, including clathrin-mediated endocytosis, macropinocytosis-mediated endocytosis, and also caveolin-mediated endocytosis. We first treated the cells with inhibitors for each pathway respectively, and then we treated nanoparticles encapsulating fluorescent-labeled RNA. We found that if we inhibit heavy oleic pathway, the uptake of the particles significantly reduced over 95%. So we think heavy oleic is a major pathway to uptake these VCLNPs. And the next, we need to figure out some therapeutic cargoes for these VCLMP to overcome the bacterial invasion. And here is a natural pathway. One macrophage meets bacterial, macrophage will uptake the bacterial and form phagosome. And then phagosome fuse with lysosome to form phagolysosome. Many bacteria, they figure out some, like, strategies to overcome the bacterial killing effects in macrophages. They can secrete different types of toxins or different types of agents. So, we are wondering if we can block the immune invasion of these bacteria and enhance the antimicrobial activity of these macrophages. So we designed this therapeutic cargo. We call that AMP-CATB mRNA. This mRNA is composed of 3 components. One is antimicrobial peptide, and the second one is enzyme-sensitive linker, and the third one is Caspase B, CATB. So the strategy is to utilize the natural translocation of the CATB to bring the whole protein into lysosome., and in the lysosome, the enzyme-sensitive linker is cleaved and then releases antimicrobial peptide. In the lysosome, antimicrobial peptide can help to eliminate the drug-resistant bacteria. First, we would like to confirm if this CADDAB is able to translocate the whole protein into lysosome. So we produced a reporter mRNA encoding the enhanced GFP. And here is our confocal imaging studies. And the green signal is GFP-CATB, and the red signal, that's a commercially available lysosome tracker. And on the right here is a merge. We can see there's a really nice merge of these two signals. The Pearson's correlation coefficient is around 0.91. So that means these two signals has great overlap. And after that, we try to understand the antimicrobial activity of this strategy in vitro, and we used multi-drug resistant S. aureus, and we have several groups here. Including PBS, free mRNA, empty nanoparticle, and also we have the particle encapsulated mRNA, but we also treat the cells with CatB inhibitor. And the last one is particles encapsulated the empty CatB mRNA. We can see at all different time points tested, there's a significant antimicrobial activity observed. And if we inhibit the CatB activity, if we cannot have the CatB transport the protein to the lysosome, or like if they cannot cleave the linker, then the antimicrobial activity is significantly reduced. Encouraged by the in vitro results, we moved to in vivo studies. We first established the immunocompromised sepsis mouse model according to the method reported in the literature. We first gave the mice 3 doses of the cyclophosphamide to suppress the immune system. And then we collected blood from the mice to quantify the markers.. And after that, we infected mice by IP injection with the bacterial. And then this is an immunosuppressed sepsis model. And here on the left, that's white blood cells. On the right, that's lymphocytes. You can see here is a normal level. After the model setup, there's a significantly reduced white blood cells and also reduced lymphocytes. So this is consistent with this sepsis model reported in the literature. And then we give different treatments for these mice, including PBS and the PBS-treated raw cells, and also our macro cells injected IP and macro cells injected IP plus IV.. And we found that these macrophage cells injected IP plus IV significantly reduced the bacterial load in the mouse blood. And more importantly, there's a significant extension of the mouse survival here, probably over 60%. And for all the surviving mice, their body weight continued to And after 20 days, we further quantified the bacterial load from these mice. We found several mice, they showed no detectable bacteria, but 4 mice, they still have persistent infections. So for those 4 mice, we did a repeat administration of this adoptive cells. And after another 10 days, we found that there's no detectable bacteria in their blood. And for all these survived mice, after 30 days, their white blood cells returned to the normal level, and also their lymphocytes returned back to the normal levels. And the next, we tried to mimic the sepsis patients by infecting these mice with mixed multidrug-resistant bacteria. In this case, that's S. aureus and E. coli. And we did similar experiments, and we quantified the bacterial load, and also we monitored the survival. You can see there's a significant reduction of the bacterial load, and also there's a lot of mice, they can survive after the treatment. Consistently, their body weight continued to increase, and these white blood cells, lymphocytes, are back to the normal range. So in conclusion, we talk about all these three directions. The first one is to develop different type of lipid-like nanoparticles or lipid derivatives, and we show the proof of concept of delivery of multiple types of mRNA in different disease models such as hemophilia, for gene editing, or for cancer treatment. And for the second part, we talk about RNA engineering. We are able to increase the gene editing efficiency at least threefold We also developed some, like, gene editing inhibitors. Due to the time limit, I didn't show the results here. And for the second part of the RNA engineering, we were able to engineer UTR of the messenger RNA and significantly increase the protein production. And we utilize this mRNA to produce mRNA-based vaccines against SARS-CoV-2. And lastly, we talk about adoptive macrophage therapy. We were able to develop vitamin C-derived lipid nanoparticles, which is able to deliver mRNA very efficiently in these macrophages. And this adoptive cell therapy is able to improve the survival rates in septic mice. They were able to clear the bacteria from these mice and restore their body conditions. So for the next 5-10 years, I hope we can integrate our specialties together with other collaborators in several fields, including biomaterials, RNA engineering, genome editing, immunoengineering, and cell therapy. Our ultimate goal is really to translate our knowledge and the technology into more effective treatments for the patients in the clinic. Lastly, I would like to acknowledge my lab members. There's a group of talented postdocs, graduate students, undergraduate students, and a few high school students. And thanks to our collaborators for their significant contributions for different projects. Thanks to funding agencies supporting our projects as well. Thank you very much for your attention.

I would like to take any questions. Thank you. Uh, thanks very much, Yiju. Let's jump into questions. Um, Anna Krzyzewski asks about targeting of the LNPs.

Anna, do you want to ask that? Um, hi, sure. So, uh, thanks first of all for a really great talk. Uh, so, you know, uh, question is about the distribution of your LNPs and, um, you know, different variants. If what can you tell us about the, you know, the broad versus targeted tissue distribution?

If you can expand on this, would be really wonderful. Sure. Generally, the biodistribution can be tuned by the particle properties. Some particles are more preferred to liver, other particles more preferred to spleen or lung or some other, like, organs. So it's really case by case. For the first part, the particles we tested, it's really great for the liver applications. And also for some particles are pretty good for intramuscular or other, like, organs. But generally, I think the particle properties is very important for their eventual, like, distribution, and also for, like, in different organs, there are also several types of the cells.

They can target different cell types. I guess to follow up on that, in your, in your cancer immunotherapy work, I think you did in situ injection, you know, so that the LNPs are right there. I'm used to thinking of vesicles having, you know, signaling proteins attached to the surface. It's— I need to follow up on your comment about different head groups trafficking to different tissues, but I wonder if you can, if you could fold in adding a surface molecule to your LNP creation.

Yes, yeah, yeah, in fact we can do that very conveniently with a different type of chemistry. So, um, yes, we have some ongoing projects.

We can target different type of the cells using their like specific ligands. Um, Zeljka McBride asks how you picked vitamin C, and I'll sort of piggyback on that. Am I correct that the vitamins that are head groups as an alternative to TT, and in your original orthogonal experimental design, I believe you looked at some other head groups, but was it, was it, was it just the design of experiments that led you to TT, or which other head groups did you take a look at?

Yeah, we did check, check, check, like, different type of, like, either head groups or, like, core groups that can install lipids around the core groups. And for the, for that project, we, the rationale is for a lot of cells, they really need vitamins for different type of their cell functions. So we think maybe we can incorporate these vitamins into the delivery vehicles. So we produced all these vitamin-derived lipids, and then we formulated them and tested in different cell types, and we found vitamin C is particularly good for delivery in macrophages.

Design all the cargoes and all the studies following that. I really appreciated your use of orthogonal experimental design. In a previous life in aerospace engineering, I used it to— you know, it really enables you to explore a space of design parameters efficiently. I'm wondering if Um, well, first of all, one specific question. Um, what is the entrapment efficiency parameter that you're optimizing? Is that the RNA getting into the lipid or some kind of lipid targeting to a tissue?

What, what, what is that parameter?

Um, oh, so that's like, uh, how much percentage of mRNA loaded into the particles. Okay. And my second question about the method is if it weren't luciferase, would you need to redo the exploration given a different mRNA or a different disease with a different tissue target, or is there—

or can you sort of— or not? Normally, if the RNA has a similar length, so we don't really need to do the optimization again. We will validate the delivery efficiency. If we found that there's a difference of the delivery efficiency, we may like further explore that. But normally, if like two different RNAs, they have similar lengths, such as the luciferase with another like mRNA, they have quite similar length, we optimize with luciferase and then we utilize for another mRNA, normally the delivery efficiency are quite comparable.

And then your discussion of optimizing 5' and 3' UTRs, in some of your work. One question is, are you targeting translational efficiency or increased stability of the RNA or both? And you said you were guided by literature for choosing the R3U and the NASR UTRs.

Do you know the mechanism, like which RBPs they're binding to, what the mechanism is that makes them successful? So we are still exploring that. For those components that's already well studied in the literature, there are a lot of results. Like for the functional motif, those are normally like from viral components. They can bind with RNA binding proteins. And for the 5' UTR, that's The lead UTR is based on the normal design, and generally that's based on several criteria. So we, I think it would be important to have more studies really like uncover like if it's increased translation efficiency or improve the half-life or stability.

Those type of things, we need to do more studies to understand that. Thanks very much. Other questions? I'm looking down my list. Let's see. I think, I think I asked all. Um, very much. That's quite a lot of work. And I, it's interesting that the looks like you're doing a good job getting collaborators to, um, you know, get these things out into the world. So thanks very much. Oh, my pleasure. Thank you for having me. Um, I'd like to let everybody know that next Next month's speaker is going to be Stephen J. from the University of Maryland College Park, and he's going to be talking about overcoming obstacles to EV-based therapies. Oh, so we do have one more question.

Zeljko, do you want to just ask live, or I can read it if not? Probably better if you, if you read it, but anyway, I'm just, I think it's very exciting that you found actually a targeting molecule from vitamin.

So I'm just curious, is there anything else that can be that you identified targeting to certain cell types? Yeah, that's a very interesting question. So we are trying to now explore more and more targeting molecules, so we hope we can identify some new molecules that can target different cell populations. Yeah, those studies are still ongoing. That's wonderful. Thank you. That's exciting.

My pleasure. Thank you. The Holy Grail. Exactly.

Me too. Thanks very much for the talk today. Thank you. Take care.

Bye.