Waters Corporation (WAT) Earnings Call Transcript & Summary

Hello, everyone, and welcome to today's webinar, cracking the code analyzing mRNA and sgRNA with Advanced Enzymes, Mass Spectrometry, and Informatics. My name is Nick Pittman. I'm the Marketing Manager of the Global Biopharma business at Waters, and it's my pleasure to host today's session. Now it's fair to say that our industry is innovating quickly. It's not that many years ago that it was fully accepted that the filing of a new vaccine or a new therapeutic would take a long time around about 15 years. And the COVID pandemic and the associated vaccine development have, of course, surpassed and changed all of our industry expectations around this. But as the biopharmaceutical community explore these newer molecules, so that creates new challenges. For example, the analytical challenges posed by a continually evolving therapeutic modality pipeline. But how do we manage this? While at Waters, we work closely with our customers and our collaborators to understand those challenges. And then our application and development scientists are continually developing new and better dedicated tools for the new generation of biological modalities, meaning that users can partner with us to routinely examine different types of RNA for assessment and control of key product attributes and ultimately deliver on those changing industry expectations. On this webinar, I'll describe just that. And I'm very excited to show you the great work that's been going on in recent months. So before we start, I'm going to briefly describe the console we're going to use today. On the right, we have our Q&A box. And while you won't be able to use your microphones during this webinar, that, of course, doesn't mean that we don't want to hear from you. So please use this to ask any questions for our Q&A session, which will follow the 2 presentations today. You can, of course, see the presentation slides in the middle of the screen. You can expand those to full screen. If you want to see all of that lovely data we're going to present in a bit more detail. On the left-hand side, we have our speaker's bio, and I'd also like to ask everyone to please give us 30 seconds of your time to fill in the survey because your feedback really helps us to improve. So please be candid and let us know what we can do to deliver better webinars for your enjoyment. Okay. So today, we're going to see some examples of LC-MS methodologies the RNA therapeutics to characterize, to measure and to control key product quality attributes. After this short introduction, I'm going to hand over to my colleague, Christian, who's going to describe a comprehensive set of tools for LC-MS analysis of large RNA molecules, specifically focusing around some exciting product launches that you will have seen from Waters very, very recently. And then we will hand over to the main part of the webinar, which is a presentation from Dr. Thomas Menneteau from quality assistance talking about some insights of the implementation of these tools and solutions for their RNA analytics. We'll then move on to our Q&A section, where we can answer any questions that you might have and that you've submitted throughout those presentations. So our first presenter will be my colleague, Christian Reidy. Christian is a principal product manager based in our Milford, Massachusetts headquarters. Christian has been at Waters since 2018. He achieved his Bachelor's degree in Chemistry at the University of Arizona. And previously at Waters, he was a chemistry account manager looking after biotech labs in the Boston area. Now in Christian's current position, he has been working on tackling the challenges of developing LC-MS solutions for nucleic acid characterization, and that's exactly what he's going to describe for us today. Following Christian's talk, we're going to hear from Dr. Thomas Menneteau from Quality Assistance in Belgium. Thomas joined quality assistance in 2021. He achieved his PhD from the Universite Paul Sabatier in St. Louis. Following that, he had a postdoc position at University College of London, working on DNA repair complexes using native mass spect and cross-linking mass spect. Since joining quality assistance, he's been working on the development of LC-MS method for the analysis of biomolecules and was involved very recently in our collaborations on RNA analytics using the new analytical tools. And he's going to give us a really nice perspective on their implementation. But before we hear from Thomas, it's my pleasure now to hand over to Christian to describe the latest solutions that have been launched from the Waters table. So Christian, over to you.

Thank you, Nick, and hello, everyone. My name is Christian Reidy, and I'm a product manager in the cell and gene therapy group at Waters Corporation. Today, in collaboration with Quality Assistance, we will be discussing new tools for characterization of RNA therapeutics. mRNA therapeutics are expanding beyond infectious disease. Companies are investing in RNA for cancer, cardiovascular and autoimmune disorders. For example, Moderna and Merck are collaborating on an mRNA mAb combo drug for treatment of melanoma. CRISPR-Cas9 is another RNA protein-based therapy that makes precise changes to the DNA for addition, removal and alternation of genes. This area is seeing significant research and development activity. 2023, the FDA approved exa-cel as the first CRISPR-Cas9 therapeutic for treatment of sickle cell anemia. There is also additional pathways for RNA therapeutics with protein replacement and mRNA-based CAR-T therapy. As these modalities get larger and more complex, the availability of new analytical tools are necessary to accelerate the development of these critical therapeutics. With this discussion, we will focus on 2 types of long RNA, single-guide RNA used for CRISPR-Cas9 gene editing, which is typically 100 nucleotides in length. mRNA for vaccines can be thousands of nucleotides long. Manufacturers must monitor molecular structure, sequence, heterogeneity and impurities of the final product. A number of critical quality attributes are listed here. LC-MS is an analytical tool commonly used to measure a number of these attributes. Digestion is performed with an enzyme to break long RNA into smaller, more manageable size for detection. In 2023, Pfizer published a paper highlighting a comprehensive oligo mapping workflow for digestion of their mRNA vaccine community. Pfizer utilized RNA's T1 for enzymatic digestion and were able to achieve approximately 56% unique sequence coverage. Excluding the poly oligonucleotides, all oligos were identified with MS with fragmentation matching. Oligo mapping involves the enzymatic treatment of a nucleic acid to produce smaller digestion products. The goal is to confirm identity, purity and perform modification mapping of the RNA. Impurities can include sequence errors or truncated RNA from incomplete transcription or synthesis. Ion-pair reversed phase with HFIP and alkylamine based mobile phases is commonly employed for separation on a C18 column. As we will see, digestion of long RNA can produce many peaks, which can be challenging to separate chromatographically. Informatics and data processing is also critical to streamline data interpretation. Manual interpretation of mRNA oligo maps can take days, if not weeks. In endoribonuclease, an enzyme that cleaves internally within the RNA at the phosphate backbone. Here, we're showing the crystal structure of MCI, one of our new enzymes for RNA digestion. The triangle denotes the region of binding and cleavage. We will refer to the enzymes listed here as RNAs throughout the presentation. Both enzymes, RapiZyme MC1 and RapiZyme Cusativin are considered endoribonuclease in the RNA's T2 family. The schematic here is an example of an enzyme digesting an RNA strand at a specific location designated by the bulging backbone in the RNA. In this case, we're highlighting dinucleotide specificity. The result is a smaller RNA product. The end where cleavage occurs can have different phosphate products depending on the enzyme's digestion properties. Not all enzymes behave the same. In fact, while we're listing different endonuclease here, they all have varying specificities and create different products during cleavage. Perhaps most common is RNase T1, which cleaves at the 3 prime end of every guanosine residue creating 3 prime end linear phosphate products. RNase 4 cleaves at uridine followed by adenosine or guanosine creating cyclic and linear phosphate products. RapiZyme MC1 targets the 5 prime end of uridine for cleavage. RapiZyme Cusativin targets the 3 prime end of cytidine, both generate predominantly cyclic phosphate products. Let's take a closer look at RapiZyme RNase specificities. Let's take a closer look at our primary cleavage points. We're denoting the 5 prime end and 3 prime end direction for an RNA sequence. For MC1, when A is followed by U, we see a high tendency for cleavage, same with U to U, and C to U. We described this as having a 5 prime end uridine specificity because once it's cleaved, we have a high abundance of 5 prime uridine products. Notably, G to U cleavage is not observed. The same for Cusativin except our specificities are C predominant C to A, C to U, and C to G. We also have U to A. Note, C to C cleavage is not observed. We can map the relative cleavage rate for MC1 and Cusativin in a table and map out the digestion hotspots. As described, MC1 cleaves at a high rate for ApU, CpU and UpU. There are also some secondary cleavages at the lower abundance highlighted in orange. Cusativin's primary cleavages are shown CpA, CpG, CpU and UpA. The p here denotes the phosphate being cleaved. It is important to note that no dinucleotide combination cleaves 100%. This can best be described as a partial digestion. Partial digestions help maintain longer RNA digestion products with diversified sequence information. Here, we're showing the detected digestion products for an sgRNA sequence for 4 RNases. This can help us compare how specificity impacts the digestion. Remember, RNase T1 cleaves after every G residue. In G rich areas, we are unable to detect the expected sequence product. Those areas are not shown below the sgRNA sequence. RNase 4 shows 2 products as it generates both cyclic and linear phosphates at a high abundance. Both RNase T1 and RNase 4 provide discrete products with no overlap. The RapiZyme MC1and Cusativin digestions create overlapping digestion products and increased redundancy due to partial digestion. In this case, MC1 provides longer products than Cusativin, which is helpful to reduce ambiguity during peak assignment at the MS1 level. Now that we better understand our RapiZyme RNases, let's discuss the procedure. Starting with our RNA sample, we have the option to denature. Denaturing can be performed at 90 degrees Celsius for 2 minutes. We would recommend denaturing when you have a higher order structure or a highly complementary sequence. For digestion, we will incubate at 30 degrees Celsius and note that we recommend pH 8 for MC1 and pH 9 for Cusativin. Once digestion is completed, we will heat and activate the enzyme for 15 minutes. From there, we are ready for injection and analysis. Typically, the total prep time will be 1 to 2 hours depending on the incubation time in step 1. We've listed a table here with starting recommendations depending on your sample type, length and whether or not your RNase is modified. Thomas will show us in the next section, considerations on how to figure out the optimal digestion procedure based on time and enzyme units for your sample. This is the LC-MS data from our sgRNA digestion shown 2 slides ago. The calculated sequence coverage from each of these is included on the table on the right. As mentioned, we were unable to obtain full sequence coverage for RNase T1. Looking at the T1 tick, the digests are shorter, which can be visually determined based on retention time. Longer products will allude at later retention times. We noted that signal for RNase 4 was lower than the other RNases due to the generation of multiple phosphate products. The column load was equivalent between each injection. For RapiZyme MC-1 and Cusativin, we see a nice balance of high signal in addition to longer digestion sequences. Why is having long group digestion products so important? Let's compare the digest length x-axis to the y-axis on the left, which shows the number of identical sequences detected. We can see as the digestion product length is very short, the number of ambiguities or sequence isomers increases substantially, Sequence isomers could be very difficult to differentiate. A burdensome development for MS users trying to characterize their molecule. The optimal target zone is 10 to 25 nucleotides in length, providing LC-MS amenable molecules with unique masses. This region works best for a variety of mass spect systems. pH study was performed to determine the optimal pH for MC1 and Cusativin and ammonium acetate. Based on the results, pH 8 is optimal for MC1 and pH 9 for Cusativin. Not only did these pHs preserve our high-on target specificity but they also result in the longest digestion products compared to lower PHs. This helps us achieve digest in that 10 to 25 nucleotide regime. For this study, the average nucleotide length for MC1 was 19.5% and for cusativin was 18%. In addition to our RapiZyme RNases, we will also be launching MAP sequence for streamlined data interpretation for oligo mapping workflows. This end-to-end workflow covers sample preparation, LC-MS analysis and data processing. Sample preparation can use any number of commercially available RNases previously discussed. LC-MS analysis is performed using systems such as the BioAccord or Xevo G3 QToF. Data processing uses the new MAP sequence app, but also takes advantage of 2 micro apps, which customers can access free of charge. Let's take a look at the MAP sequence app. First, an mRNA Cleaver, we will add the RNA sequence, enzyme used for digestion and other parameters such as missed cleavages. mRNA Cleaver will generate a list of in silico predicted digestion products. MAP sequence takes these predicted fragments and confirms which ones were detected in the sample, filling in the sequence map. Coverage viewer on the right, then plots a visual map to show what parts of the sequence were detected, ambiguous or missing. A sequence coverage is reported. Next, we'll take a look at a few high-level application examples. I'm not going to delve too deeply into these as Quality Assistance will provide a more comprehensive breakdown for method development. Here is an example of an sgRNA sample digested with RapiZyme MC1. We've listed the sequence below and are highlighting the modified termini in the hairpin structure. After analyzing, the sequence coverage was determined. Confirming modifications and identity of sgRNA is necessary to ensure stability, compatibility with the Cas protein affected delivery and to reduce off-target effects. On the top left, we show an example of mRNA being digested with RapiZyme MC1. The structure highlights the 5 prime cap in the polyA tail. Analysis was performed using a traditional IPRP method. You can see that the mRNA digest chromatogram is very busy. The sequencing region has a wide range of nucleotide products of varying lengths. This area also includes a 5 prime cap. The polyA tail is the longest part of the sequence and alludes the end of the chromatogram. PolyA distribution can be further investigated in our intact mass app. Thomas will present on this workflow in more detail. Another important workflow is the 5 prime capping efficiency. This is often performed with an affinity hybridization probe to purify the 5 Prime cap away from other mRNA digestion products. The 5 prime capping probe is comprised of a 7 methyl guanosine, triphosphate section and a modified ARG base. The integrity of the cap is important to mRNA stability, translation and reduced immune response. Here, we're showing a schematic highlighting some of the impurities we can see with the capping, including no cap or incomplete cap. The sequence without a cap is preserved with MC1 digestion and was detected in the sequencing region of an mRNA analysis. Preserving the 5 Prime Cap and impurities opens the possibility for an all-in-one method for sequencing 5 prime cap efficiency and polyA tail analysis. In addition to our enzymes and software, Waters is now offering the highest purity HFIP on the market. IonHance HFIP is manufactured to have the lowest sodium potassium levels currently available. A recent ICP-MS batch test shows less than 10 parts per billion for sodium potassium content. Let's see how low metals content can impact an LC-MS run. Here, we compare the MS spectra of a 100-mer SSDNA sample prepared with IonHance HFIP to a commercially available LC-MS grade HFIP. The analysis was performed on the same system, column and with the same method conditions. The system was cleaned with the same procedure prior to analysis. We observed a decrease in sodium potassium when compared. Before we proceed to the next stage of the webinar, we ask that you complete this quick poll question on your current process for RNA digestion. Thank you. Without further ado, I'm excited to introduce you to our collaborator, Dr. Thomas Menneteau. Thomas is an R&D scientist at quality assistance and his focus is development of LC-MS methods for biomolecules. Today, Thomas will be presenting his work on RNA digestion for single guide RNA and mRNA analysis. Over to you, Thomas.

Thank you, Christian. Hello, everyone. I'm Thomas, and today, I'll be presenting you the results we obtained at Quality Assistance with the enzyme and the data processing workflows that Christian just presented to you. All the work that we have performed has been achieved using ion-pairing reversed chromatography. So I want to start with a little explanation on how we performed it. This way, you will be able to obtain comparable results on your system when you use the workflows that we are going to present. The first step was to clean our liquid chromatography overnight using a cleaning solution containing water, isopropanol, acetonitrile and methanol in equivalent volume and 1% formic acid. This was performed to remove any potential contamination from previous experiments as we don't have a dedicated LC system for ion-pairing chromatography. We also choose to use ultrapure solvent to limitate the number of [indiscernible]. As Christian told you earlier, once released a new IonHance HFIP, which contain less metallic ions, so you will observe less additives compared to other HFIPs. Also, despite it's not really clear if there is an effect or not of the use of borosilicate bottles, we decided to be extra cautious and use HDPE bottles. We changed our solvents almost every day. Indeed, HFIP boiling temperature is quite low, so to avoid the effect of its evaporation on separation. It also has been shown that the ion-pairing reagent oxidized over time. You can find detail about those 2 points in a really nice publication for Guimaraes published in J. Am. Soc. Mass in 2023. Regarding our column, we use low absorption column as a Premier oligonucleotide as we observe fronting and tailings on other types of column from other experiments but we wanted to increase the odds to get good results. The choices are currently available in Premier column with 2 pore size 130 or 300 Angstrom. In all the experiments I present later, we used 300 Angstrom column. When you look at the data at the bottom of the screen, we obtained for the same sample. We have a better resolution with a 300 Angstrom as it's capable of separating 2 species, while we can see only 1 peak with a 100s. Column conditioning is a really important step as columns are often stored in solvent with its ion-pairing reagent. We think we have to be extra careful with this to get reproducible injection, especially at the beginning of the sequence. Finally, we inject the system suitability test to check the system. We don't have any specific criteria at the moment but as we have a mix of 5 nucleotides of various lengths, we expect to see 5 peaks with a good resolution between them. We use a homemade SST but you can find a SSDNA later available on the Waters website. Here are the LC-MS condition we use. The mobile phases use contain 0.1% DPA and 1% HFIP in water for mobile Phase A and 0.0375% DPA and 0.075% HFIP and 35% water and 65% acetonitrile for mobile phase B. The CMS parameters we use can be found in the application note you can see on the slide. We have to play a bit with the mass spec parameters as the application that has been performed on a BioAccord system but there's nothing really fancy. We analyze a sample with negative mode in MSE mode. The mass range used was from 300 to 5,000 m/z and the fragmentation energy was set to ramp from 10 to 45 electrons. Our first tests were performed on a single guide RNA called PSMD7, which is 100 nucleotides long. As Christian mentioned it earlier, we used 2 newly developed RapiZyme enzyme, MC1 and Cusativin. The protocol is really easy to set up as it takes less than an hour to obtain your digest and be ready to inject your sample. Also, most protocol are identical except for the buffer used to optimize the length of the digest obtained at the end of the sample preparation. With this step, we'll digest a guide or the mRNA, as I will talk about later, in different digestion products that I would call oligos for the rest of the presentation. So once your oligos are ready, you can inject them on the LC-MS system and you need to process the data that you've obtained. I'll talk now about the processing workflows that has been developed by Waters, which require 3 major steps. First, you need to generate in silico oligos in your sample using an app called mRNA Cleaver. The second step is to really process your data using the new waters_connect module called MAP sequence. Finally, you can visualize your data in an app called Coverage Viewer. Let's take a deeper dive in this workflow now to show you how easy this to use. As I mentioned it, the first step is to generate the oligo database using mRNA Cleaver. To do so, you need the sequence of your guide RNA and the enzyme you use. You can add some modifications but I'll come back to this later. You can add some filters so you don't generate too many oligos, for example. You must add MS parameters as we haven't used a full mass range available on our mass spect. And then there is this parameter, which is the most important from Missed Cleavages. As Christian told you, the buffer used for each enzyme has been chosen depending on the length of the generated oligos. This is not done to change the cleavage specificity but the enzymatic activity efficiency of the enzymes. For example, with the short sequence, if we digest it with MC1, we have 4 cleavage sites. If we do a search with no missed cleavage, you'll be searching for 5 species. However, if you have 1 potential missed cleavage, you can expect 5 extra oligos, which are longer, so your probability to have a unique oligo increases. With 2 missed cleavages, we have 3 new oligos and with 3 missed cleavages, 2 extra oligos. And when you look at the oligos obtained by increasing the number of missed cleavages as they are longer, you increase the probability to see unique oligos, which is better for determining your sequence. However, if you put too many missed cleavages you won't be able to observe the exact mass of your oligos but the average mass and then you lose confidence in your identification. If we go back in mRNA cleaver, now that we have set up all parameters, we can click on the calculated button and after a couple of seconds, you will obtain a CSV file containing all the potential oligos generated from the enzymatic digestion based on your sequence. The next step is to launch the data process using MAP sequence, which is a module ended in waters_connect. In the set processing parameters panel, you must have your mass tolerance, a detection threshold as well as a chromatographic section you want to process. Also, you need to approach the CSV file just generated with mRNA Cleaver. In this injection list panel, you have to choose the data you want to process, lock mass corrected or not. Then you can launch the process. And after a few minutes, you'll get your identification results. This is interface you will see once your data processed. You can see if each oligos generated with mRNA Cleaver has been found based on the [indiscernible]. If they are found, you'll see a green dots. Sometimes, mass cannot be found, and you see a red square indicating that the species hasn't been found. Sometimes, the same mass can be observed at different retention times, and you see an orange triangle with a comment alternative peaks. If several oligos have the same mass and this mass is detected, the software shows also an orange triangle but with the comment alternative product assignment. If you click on an identified oligo, you will obtain detail about its retention time as well as the charge states observed. You can click on the exports of the injection button to export the data for visualization. Data visualization is performed in the Coverage Viewer. To make it work, you must ask the sequence of your gRNA, load the digest products that have been generated with mRNA Cleaver and the processing results from MAP sequence. The software will then compute your sequence coverage and give you details about the uniqueness of the different part of your sequencing results. These are the results we obtained for the PSMD7 guide RNA we studied with both MCI and Cusativin. As you can see, after MC1 digestion, we almost reached 100% of sequence coverage at 2 missed cleavages and the results are even better using 4 missed cleavages. For Cusativin, the results are lower at 2 missed cleavages but you can see a huge gap when you search your data with 4 missed cleavages, reaching 100% of coverage and 1 replicate. I don't know if you have noticed but there was a button in the Coverage Viewer app called include ambiguities in coverage percentage. What will change on the results if we click on it. In light colors, you can see the new coverage once the ambiguities have been added in the calculation of the sequence coverage. As you can see, for both enzyme in all replicates, we reached 100% of sequence coverage. This is really nice as it's what we want to obtain in the sequence but what are the ambiguities? Ambiguities are only goes in your gRNA with the same mass. If we take a look at the data in MAP sequence, there is an example of ambiguities. You can see that the sequences are different but their masses are exactly the same. So how can you tell if they are really all in the sample or if only 1 is present? If you remember, all mass spectrometry analysis were performed with MSC mode activated, which means that we have fragmentation information. This data can be used to sequence the oligo and confirm its identity. To do so, we used module already available in waters_connect. First, we created our 3 sequences in the Synthetic library module where the software then generates all the possible fragments that can be observed. Then, we used the Confirm sequence module to check for the fragmentation automatically. One main advantage is that you can lodge this process really fast as MAP sequence provides you with a respective retention time window to perform the search. And once the oligo has been generated in your Synthetic library database, you can reuse it on demand. That's what we did with the 3 ambiguous oligos, and here are Confirm sequence results. As you can see for the 3 of them, we have 100% of oligo coverage with nice fragmentation items identified. This means that the 3 oligos are present in the sample but they are co-eluting at the same retention time. Once ambiguities like I've just shown you have been resolved, it's possible to go back in the Coverage Viewer app to increase the normal sequence coverage. Indeed, when you look at the table on the bottom left of the screen, there is a column entitled Sequence confirmed. You can tick the box for the ambiguities that have been resolved, and this will recalculate the sequence coverage. This feature is really interesting, but this means that your ambiguous data must be semi-manually curated to include them in your final sequence coverage values. We have a modified equivalent of the PSMD7 guide RNA, where the 3 first and 3 last nucleotides are modified with 2'-O-methylation and contains phosphorothioate bonds. What we did is, we spiked 5% of this modified guide RNA into the standard PSMD7 guide RNA, and see if we could identify some oligos presenting the modification. Once injected, we can see differences between the total ion chromatograms of both samples. We first focused on the modification at the 5 terminus of the guide RNA. We generate 2 mRNA CSV files and we added the oligo specific for the modification into the standard guide CSV. Then we launched the process with MAP sequence and here are the results. As you can see, both species are identified, and you can see, as expected, a difference in terms of response between them. We went then in confirmed sequence module to confirm the sequence of the modified oligo and you can see that we obtained 100% coverage with fragment ions. When we look at the free terminus modification using the same approaches that I presented, we only identified the standard oligo. When you look at the intensity of standard oligo, you can see it's really low. Indeed, one of the cleavage site of MC1 is between 2 uridines. And as you can see, this part of the guide RNase contains only uridines. To go further, we injected a digested modified guide RNA and checked if we could see the modified oligo. As you can see, MAP sequence identified it but its intensity is really low too. We went into the Confirm sequence module and confirmed its sequence. As MAP sequence gives you the evolution time of each oligos, we used the 1 retry from the injection of the modified guide RNA to try to search for it in the spiked sample using the confirm sequence module. As you can see, the fragmentation is poor, but this is due to already low abundancy of the spike and its digestion by MC1. I think we show you nice results for guide RNA but we are also all interested in bigger spaces, especially in mRNA. As I told you, guide RNAs are only 100 nucleotide long, making them not really complex sample, as you can see on the total ion chromatogram on the top. More challenging samples are mRNA, which can have ranges from a few hundreds of nucleotides to more than the 10,000 nucleotides. At the bottom, you can see the total ion chromatogram of Cas9 mRNA, which is 4,521 nucleotides long. As mRNAs are longer than guide RNA, we had to optimize the sample preparation workflow to be sure to generate oligos with length of interest. To do so, we use the Cas9 mRNA, which is 4,521 nucleotides long and it contains a polyA tale of about 120-A. It is capped with Cap1 and all the uridines have been replaced with methoxyuridine. We tested 3 different of both MC1 and Cusativin to digest 6.7 picomoles of mRNA, 150 units, 300 units and 450 units. We incubated the reaction mixture either 30 minutes, 60 minutes or 90 minutes at 30 degrees. 1.7 picomoles were loaded on the column for each sample generated. When you look at the results for MC1, you can see that the longer the digestion time, the lower the sequence coverage. And we observed the same phenomenon regarding the enzyme quantity used for the digestion. Similar results are obtained with Cusativin. When you compare the data obtained at 2 missed cleavages and at 4 missed cleavages, we observed an increase of about 16% on sequence coverage for MC1 and about 10% for Cusativin. Based on these results, the condition we used, which are the same than for the guide RNA are going to be 150 units for 6.7 picomoles of mRNA and 30-minute digestion. We used 3 different mRNAs of value sizes for a deeper characterization of the enzymes. Cas9 mRNA is that I already presented. Fluc mRNA, which is 1,922 nucleotide long and EPO mRNA, which is 859 nucleotide long. We tested 4 enzymes for the digestion with digestion optimized previously, MC1, Cusativin, RNase 4 and RNase T1. We also loaded more samples on the column compared to the optimization step to get better signal. And the data I'll show you now have been processed with 20 ppm mass tolerance and 4 missed cleavages. On this plot, you can see the results we obtained for the 4 enzymes on the 3 different mRNA. For the Fluc mRNA, MC1 exhibited the best results with 91% of sequence coverage, while both Cusativin and RNase 4 at 87%. RNase T1 exhibits 83% of coverage. Sequence coverage for EPO mRNA are lower and Cusativin is performing the best at 84% of sequence coverage, while MC1 is at 81%. For the Cas9 mRNA, MC1 reaches 88% of sequence coverage, while Cusativin is at 80%, RNase 4 at 78% and RNase T1 is at 56%. When we take the data obtained for the 4 enzyme on the 3 mRNAs and average it, we can see that MC1 is performing the best among the 4 enzymes and T1 the lowest. However, if we do the same including ambiguities, we see that T1 is the best enzyme due to its high occurrence of cleavage. MC1 and Cusativin gained a bit of sequence coverage once the ambiguities are added. However, RNase 4 gains almost nothing from ambiguities as it has only 2 cleavage sites and generate longer oligos. Overall, MC1 performed the best. I've talked a lot about missed cleavages and the impact of long oligos of the sequence coverage. Short oligos are bringing information regarding the sequence coverage but you have to use the ambiguity mode to take them into consideration. This would mean that you have to confirm the ambiguities using confirmed sequence, and this can be time-consuming, especially on long mRNAs. And still, you won't be sure about the origin of the short oligos in the sequence. We wanted to see if we can increase the search speed by removing the short oligos from the search. To do so, you just have to go back in mRNA cleaver and add filters on the length of the generated oligos. That's what we tested here on the Cas9 mRNA. As you can see, the influence of missed cleavages is massive on the sequence coverage obtained for both enzymes. Using the 4 missed cleavages search and removing the 3 nucleotide long oligos from the search, we saw no influence on the sequence coverage. No effect was observed until we removed the 5 nucleotides long oligos where we lost only 0.1% of sequence coverage and up to 0.8% when the 6 and 7 nucleotide long oligos were removed. Short oligos are not useful for coverage if we use high values of missed cleavage. We can also use the data obtained to determine the polyA length it originate. A polyA of approximately 120 adenosines is expected for the Cas9 mRNA we've been working with. As it is a long oligo, you can see that it elutes at around 35 minutes. To process the data, we used the intact mass module workflow. As you can see, we observed 2 distribution. Indeed, as there is no cleavage site just before the polyA tail, we can observe an extra portion of oligo. Also, as we have missed cleavages occurring during this digestion, we have 2 portions of oligo, 1 longer than the other. The results observed show a polyA tail of about 120 adenosine, which was expected. We could have imagined nothing in this processing MAP sequence but the maximum mass tolerance is 50 ppm, which is too low for a polyA tail analysis. It is also possible to confirm the cap present on the studied mRNA. Indeed, there are various modifications available in the mRNA cleaver software. This will automatically generate oligos with a modification. And once you launch your search, the algorithm will search for them. In the case of Cas9 mRNA, we were expecting the presence of Cap1, which has been found by MAP sequence. However, the mass has been found at 3 different retention times, 2 of them at around 11 minutes and one at 35 minutes. To confirm which peak is the right one, we used the Confirm sequence workflow presented earlier using the details provided by MAP sequence. If we search for the fragmentation of the oligo in the first chromatographic peak, we can see that many fragments are observed. This is the same thing with the second chromatographic peak but you can see that the retention times are really close. However, we also obtained a good coverage for the peak at 35 minutes, which seems weird. If we dive deeper into the data and we performed an extracted high end chromatogram on the [indiscernible] identified for the 3 peaks, we can see that the two 1st chromatographic peaks are relevant. However, the third one has a really low intensity. When you look at the MS spectrum at this retention time, you can see that it is really complex. And if you remember from 2 slides ago, it corresponds to the polyA tail. This means that this peak is probably a false positive identification. Also, you can use the data obtained with the intact mass module to determine new capping efficiency. To conclude this presentation, we have shown you today that 2 new enzymes, MC1 and Cusativin and are now available to perform your guide RNA and mRNA sequencing experiments, and both of them present really nice and reproducible performance. We have also introduced a new data processing workflow with easy-to-use tool and available for what is mass spectrometry uses without exporting your data out of waters_connect and it uses 3 apps, mRNA cleaver, MAP sequence and Coverage viewer. And also, you saw that inside a single injection, you can work on 3 critical quality attributes of mRNA, the sequence of the polyA tail and capping. In terms of perspective, there are still many things to test and develop. As you can remember, the condition used for the enzymes are the same except for the buffer used. So we can digest 1 sample with both enzymes in parallel and then process the data in a way to gather all the information for both enzymes. We tried it already and sequence coverage on long mRNAs are better. We haven't mentioned at all a nonspecific digestion and the evaluation of this phenomenon is usually expensive in terms of processing resource. We are currently working on it. One can imagine avoiding purification step to enrich capped oligo using that group. With all the different modifications in the mRNA Cleaver tool, we can generate in silico digest for all the different cap and then start a search with MAP sequence to try to identify the capping purity present in the sample, and that something we are currently working on at Quality Assistance. And finally, we also hope to be of help to Waters in the next evolution of the data processing tools. I'm done presenting you our results, and I'd like to thank you for your attention.

Okay. Fantastic. Thank you very much Thomas and Christian for your really nice presentations. Now if you would like to find out any more information about the Waters solutions that you've seen today, this is your opportunity to raise your hand and have a local specialist contact you to provide more detail. So please have a look at this quick poll question and just let us know if you would like anyone to contact you. Okay. Thank you very much. Now there's going to be a very short pause as we move to our Q&A session, and we'll then be able to go through all the questions that have come in during the great presentations today. Thank you very much for your patience. Just while we reconnected, we're now ready to start our Q&A session. I'm joined by Thomas and Christian, and I also like to welcome Matthew Gorton to the panel. Matthew is a Principal Product Manager and looks after the MAP sequence within waters_connect.

We've got quite a few questions but we've got sort of 15 minutes or so. Please do keep them coming in using that Q&A box, and we'll get through as many of them as we possibly can. So on to our first question, and I'm going to start with a really easy one, and that is can we get the slides? There will be on-demand recording of this presentation shared with everyone tomorrow. If you do want a hard copy of the Waters slides, at least we can share that with you, please just contact us using the e-mail addresses that you can see on the screen here. Okay. So on questions for our panel. And the first question that we have here is around the 5 prime cap. So do you still use antisense oligo to enrich for the cap or no cap 5 prime mRNA to have a UV signal for abiding by the USP guidelines? I guess that's a question for you, Thomas.

So at the moment, no, I never really like to enrich with antisense oligos, so -- but I always do my experiments with mass spect coupled to LC with a UV system but yes, I don't use oligos enrichments. And so we have currently a technique to just study the cap moieties on mRNAs. So using mRNA transition and everything, but yes, what I really want to develop with this approach is to try to search for all the possible capping moieties within the digestion products and see if we can get nice quantification with it.

Okay, cool. Thank you very much for clarifying that. The next question is also about enrichment. So maybe that's not one that we can answer but I'll put it to the panel. And that's talking about if we do use antisense or DNA oligo for enrichment, does that protect the cleavage site of RapiZyme that are in that duplex that's formed? Maybe one for Christian.

Yes. Because RapiZyme or RapiZyme RNases cleavage single-stranded nucleotides, we typically wouldn't use an antisense probe to protect it. Although if we did, then maybe it cleaves for their upstream. So we haven't further downstream. So we haven't quite investigated that. But we believe that we can digest with the enzymes and then from there, maybe do a biotinylated cleanup approach.

Okay. Cool. Thank you very much. The next question, is there a maximum limit of the size of mRNA studies? And what about siRNA. One for you, perhaps, Thomas?

So, at the moment, I haven't been above 4,500, but one can imagine, yes, working with siRNA, the thing would be to optimize gradient a bit do it do show a longer gradient. So rather than 50 minutes runs maybe something like 100-minute runs and also maybe optimize the digestion condition because in our case, it's enough. But if we don't put enough enzyme, we might have too long oligo so we won't be able to get nice identification results.

Yes. That makes sense. To optimize the digestion, optimize the separation and give yourself the best chance to get the best sequence coverage of those bigger oligos. Okay. Next question, n - 1 impurities for sgRNA product cause additional unpredictable of target and hence unpredictable risk for patients. Any update on resolution of n - 1 impurities by HPLC. Anyone want to take this one? Or maybe this is for -- okay go for, Christian?

I can jump in there. Yes, this was alluded to, I think, in Thomas' data in terms of n - 1 resolution and it being improved with the 300 Angstrom column. So certainly, tweaking the mobile phase and optimizing those conditions is important but we also see the best resolution when using larger pore size. So that would be the 300 Angstrom, Beh C18 over, say, the 130 Angstrom.

Brilliant. Thank you. I've got a couple of questions here on modified basis and how the RapiZyme enzymes deal with that. So any comments on notifications like methyluridine, 5 prime methylcytosine, modified pentoses base modifications. Do we have sort of a feeling on any biases that are introduced there?

So at the moment, I've tested with some modified mRNA. So we have methoxyuridines with -- on mRNAs. We have also phosphorothioate groups. So yes, but we plan to go deeper and see if it can digest properly the different guide RNase and RNase.

It sounds like a bit of ongoing work there with Quality Assistance. Christian, are you going to comment on that?

Yes, just to reiterate. So both enzymes do cleave pseudouridine and then Cusativin, which is more cytidine focused, we saw the 3 prime and cleavage cytidine, that one would do methylcytidine. So maybe just to summarize, we do see specific cleavage at modified basis.

Okay. Cool. Thank you. Now one on to multiple enzyme strategies. So can mRNA be digested by dual or multiple enzymatic digestion if in case 2 long nucleotides are being corrected by a single enzyme. Any thoughts on multiple enzyme digestion strategies?

Yes, I can jump in there as well. There's no reason why we couldn't do multiple enzymes. I think at the moment, we were looking at, say, MC1 and Cusativin in separate workflows, getting the sequences and then combining them but for very long mRNA, maybe we do want to combine, if you're worried about, say, longer digestion products. But we haven't looked at that yet, so it is something that can be investigated.

Just quickly from a -- sorry, just quickly from an informatics point of view, I would echo what Christian has said is that it tends to be harder to pull apart the different digestions if they're in 1 test tube. So ideally, yes, we would have them as separate digestion, separate samples and then combine the results afterwards. But it is possible. It's just a little bit more complex on the informatics side.