0:02 Hello, in our presentation, we want to show you a quality control method for identity and purity analysis of oligonucleotides by LC-ToF-MS.
0:12 Before I hand over to Thomas for the main presentation, I want to say a few words about A&M STABTEST.
0:19 We are AGMP certified and an FDA inspected contract service organization exclusively working for the pharmaceutical industry.
0:27 We have two sites: one in Bergheim dedicated to biopharmaceutical and ATMP testing, and another in Mainz close to Frankfurt, dedicated to small molecule analysis and inhalable drugs.
0:44 Our main goal is to offer our clients all relevant analytical parameters for small molecules, large molecules, and ATMPs under one roof.
0:52 So there's no need to send samples for complete stability studies or for release testing.
0:58 We try to offer everything under one roof to provide the best service at a reasonable price at a testing facility located in the heart of Europe.
1:11 Now, I want to hand over to Thomas to give you a presentation about LCMS applications for oligonucleotide analysis.
1:22 For oligonucleotide and intact protein analysis, we use the BioAccord system with the latest Waters Connect platform.
1:31 The scheme of the mass detector shows how the mass spectrometer works.
1:37 The flow comes from the UPLC into the Z-spray ESI source.
1:42 The solvent is nebulized and evaporated.
1:45 After entering the MS through the cone, the ions are focused and transferred to the pusher, which pushes the ions into the reflectron.
1:55 The principle of the time of flight is that all ions start at the same time point and are pushed into an opposite charge electric field. Larger ions penetrate the field deeper than smaller ions until they are rejected, resulting in longer flight times for larger ions.
2:14 In other words, the flight times are translated into mass-to-charge values or finally Dalton and molecular weight, respectively.
2:23 Since the MS is also equipped with the negative ion mode, it is theoretically possible to analyse proteins and oligos with the same instrument. However, we were careful not to destroy our performance for intact proteins by using incompatible elements and additives for oligo analysis.
2:45 After internal discussion, we agreed not to use triethylamine as usually done with oligo analysis because we expected negative effects for protein analysis with the same instrument.
3:00 Therefore, we looked for alternatives.
3:03 I found a publication from 2017 by Bassiri et al.
3:09 In this publication, the effect of different kinds of alkylamines in oligo analysis was studied.
3:17 In the case of PolyT(24), DBA (dibutylamine) indicated by the blue bar on the left side showed the highest sensitivity, and TEA (triethylamine) one of the lowest.
3:33 The same result was found with different oligos. DBA consistently showed the best or one of the best sensitivities, while TEA showed one of the lowest sensitivities.
3:49 For our method development and tests, we used commercially available standards like the OST standard from Waters, composed of five Poly thymidine oligos with chain lengths of 15 to 35 nucleotides.
4:06 We also used an RNA standard from HLND, which is more realistic, composed of four oligomers with chain lengths of 14, 17, 20, and 21 nucleotides.
4:20 Finally, we ordered several other custom oligos, including a 40mer to challenge our system and some custom thiophospho and other protected charter oligos.
4:34 Our equipment was the BioAccord system introduced at the beginning.
4:39 We tested two columns from Waters: the normal oligonucleotide BEH C18 and the Premier version of this column.
4:48 The Premier was slightly better in our hands, especially when directly out-of-the-box.
4:53 The normal column needed a few oligo injections to perform as well as the Premier one.
5:00 We started with UPLC conditions of 15 mM DBA and 25 mM hexafluoroisopropanol in both methanol and water.
5:11 The gradient started with 80% water up to 80% methanol in 10 minutes.
5:17 The column temperature was set to 60°C.
5:20 The UV chromatograms were recorded at 260 nm.
5:25 In negative mode, we set the cone voltage to -30V and the fragmentation voltage to a range from 150 to 190V.
5:34 The mass range was set to high mass, which means an m/z range from 400 to 5000 for intact oligos.
5:44 In the upper pane, you see the total ion chromatogram, and in the lower pane, the UV chromatogram of the OST standard.
5:52 With these UPLC conditions, we achieved baseline separation of the five Oligo T standards.
6:00 We will now look closer at the 25-mer, which elutes in the middle of the five peaks.
6:08 The mass spectrum of the 25-mer is displayed on the left.
6:13 The predominant ion species is the -4 protons one, but the pattern looks strange.
6:19 After deconvolution, it becomes clearer.
6:22 We see only adducts.
6:24 The major adducts are the DBA adducts indicated by the blue arrows.
6:30 In between each DBA adduct, we see sodium adducts of the DBA adduct species.
6:37 The signal of the molecular weight of the 25 T is the smallest one on the left side.
6:43 It shows a mass error of 13 ppm, which is good, but the adducts are not wanted.
6:50 One could think DBA is not the right alkyl amine, but we changed the cone voltage from -30 to -100V without any other changes.
7:02 As you can see on the left side, the mass spectrum looks much cleaner with just the three predominant ion clusters.
7:10 When we deconvolute, we see the DBA adducts are completely gone, and only sodium adducts are left, which are difficult to avoid.
7:20 The 25 T signal shows a mass error of 8ppm, which is good.
7:27 The Waters OST standard is nice to use as an SST or for method testing work, but with its artificial Poly T mers, it's not realistic.
7:38 We also wanted to test our optimised standard gradient for the RNA standard from HLND, composed of four oligomers: 14, 17, 20, and 21 mers, which is a more realistic scenario.
7:54 The mass range is between 4400 to 6600 Dalton.
8:00 We applied exactly the same conditions and parameters as for the Poly T standard.
8:06 To our surprise, we obtained only two signals in UV and MS, but not even separated, instead of four separated signals.
8:17 Obviously, we had to adjust the UPLC conditions for this mixture.
8:24 We started the gradient at 85% A instead of 80% and quickly went to 55% A, then in 9 minutes to 20% A.
8:37 We slightly increased the flow from 0.2 to 0.25 mL/min.
8:48 With these relatively small changes, we saw a dramatic effect.
8:53 Not only could we see our four oligos, but they were also very well separated, even the 20 and 21 mers.
9:03 This shows that a little adjustment to the gradient can have a significant effect, more so than directly using a different column or changing the elements.
9:13 We challenged our system by using the conditions for the OST standard to analyze our custom 40mer.
9:22 The 40mer was used as delivered, with no purification from our side.
9:27 The gradient worked well for that oligomer, and the retention time was in the expected range.
9:35 In the mass spectrum of the 40mer, the predominant ion species was the M -5 protons at m/z 2452.
9:46 The deconvoluted mass showed an error of 8 ppm compared to the theoretical value, which is good.
9:54 We used the 40mer to test the range of linearity.
10:00 We injected the 40mer from 0.1 to 200 ng/µL.
10:06 The UV linearity is excellent.
10:08 For the MS, we used the extracted ion chromatogram of the most intense ion species, the M -5H at m/z 2452.
10:20 We had to apply a quadratic transformation for the areas, and then it showed good linearity.
10:27 For impurity analysis, we spiked a smaller oligomer at 0.1% into the 40mer.
10:35 We distinguish between non-target analysis, where all peaks above a threshold of 0.05% are considered, and targeted analysis, where we know the impurity and extract an ion chromatogram.
10:51 The major impurities in the 40mer were derived from the synthesis of the oligomer, mainly the n-1 and n-2.
11:00 The spiked impurity at 0.1% was readily detected as well.
11:06 For oligo stability, we dissolved 2'-methoxyethyl phosphorothioate oligonucleotide in water and analysed it on day one.
11:18 There was only one signal in MS and UV observed for the intact oligo.
11:25 The sample was stored for 30 days at 5 to 8°C.
11:33 The originally single peak split into several peaks, which were not fully resolved, although the gradient was adjusted.
11:42 The intact oligo was observed at 6.2 minutes UV retention time.
11:48 The other peaks showed slightly weaker retention on the column.
11:51 To understand what happened, we looked into the MS traces from 5.7 to 6.4 minutes in the red box.
12:01 The total ion chromatogram was combined into one spectrum.
12:06 On the left side, we see the mass spectrum of the combined split peak after deconvolution.
12:12 The intact oligo showed a mass error of 5 ppm.
12:16 The mass difference between the intact thiophospho oligo and the peak at lower retention time showed a mass difference of 16, proving the stepwise desulfuration of thiophosphates under these storage conditions.
12:33 During the 30 days, at least four significant desulfurations took place, substituting four sulfurs with oxygen.
12:44 The intact oligo had only 23% abundance.
12:48 The dominant species was the single desulfuration product at 34%, followed by 25% for the double, 13% for the triple, and 5% for the four-fold desulfuration product.
13:04 With the BioAccord, it is possible to analyse intact masses and perform sequence analysis in parallel.
13:11 Fragmentation patterns of oligos are complex because they have multiple fragmentation sites, resulting in various observed fragments.
13:23 Two main fragment paths are observed.
13:28 The upper one includes the 3' terminal, with different breaking points called W, X, Y, and Z, indicating the exact breaking points.
13:41 The second includes the 5' terminal, with breaking points called A, B, C, and D. Additionally, there is another cleavage point where the base is cleaved off.
13:54 In reality, more than one fragmentation pathway takes place, making manual sequence analysis very difficult.
14:02 The higher the modification of the oligos, the fewer fragmentation pathways and cleaner spectra are obtained.
14:09 Returning to our thiophospho oligo, the most abundant precursor is the M -3H at m/z 2374.
14:21 Using a cone voltage range between 150 to 190V, we achieved fragmentation of the thiophospho oligo, resulting in 100% sequence coverage and confirmation.
14:35 As seen in the dot chart on top, we observe multiple fragmentation species for some ions, such as W, X, Y, and Z, and in most cases, more than one species.
14:50 This is the corresponding deconvoluted spectrum for the fragmentation.
14:54 Manual interpretation is very difficult, but the Waters Connect platform has an integrated oligo sequencing application.
15:06 In conclusion, DBA (dibutylamine) has better sensitivity than triethylamine, as stated in the literature.
15:17 DBA adducts are suppressed by a cone voltage of -80V, and no in-source fragmentation is observed.
15:30 25 mM hexafluoroisopropanol plus 15 mM DBA are sufficient for electrospray ionization and chromatographic separation efficiency.
15:43 The column temperature has no significant influence on the separation.
15:51 For back pressure reasons, we recommend 60°C, and the mass spectrometer RDA showed sufficient linearity to quantify impurities.
16:06 The mass precision after deconvolution with BayesSpray is usually better than 10ppm, which is sufficient for identity confirmation.
16:21 Sequence analysis by in-source fragmentation, especially for larger and highly modified oligonucleotides, is possible, but there is no general method that can be applied to oligonucleotides.
16:35 At least the gradient, flow, and fragmentation cone voltage must be adjusted to the target molecule.
16:47 Thank you, Thomas, for showing us what is possible with LCMS analysis and the wonderful spectra and chromatograms, and the amount of information we can gain from this single technology to characterize smaller oligonucleotides.
17:00 Thank you again.
