Read time: 25 min

Lipid nanoparticles (LNPs) are widely used vehicles for oligonucleotide-based therapeutics and vaccines. However, LNPs can be susceptible to N-oxide-based impurities that can cause messenger RNA (mRNA) to lose function.

In a recent live webinar, now available on demand, Adam Crowe, Manager of Analytical Development at Precision NanoSystems Inc. (PNI), demonstrated how liquid chromatography coupled to mass spectrometry (LC-MS) using electron activated dissociation (EAD) was used to analyze an ionizable lipid, localize double bonds and saturated impurities and differentiate between different oxidated species.

While a range of questions from attendees were answered during the Q&A portion of the live webinar, not all submitted questions were addressed due to time constraints. In this blog, Adam and I answer these remaining questions to help you improve your RNA analysis.

1.     Why do you require EAD to look at N-oxidation in ionizable lipids? Doesn’t the MC3 CID data show a diagnostic fragment for the N-oxide?

Adam Crowe: It is correct that one can also get cleavage of the headgroup and information on oxygen incorporation at the tertiary amine for simpler structures, such as MC3, with collision induced dissociation (CID). The challenge we experienced, however, is that most ionizable lipids these days, such as ALC-315 from Pfizer, do not fragment in a highly predictable manner. Localizing the oxidation on the tertiary amine on more complicated structures requires you to get the m/z 61 fragment with EAD, as we have seen. Fragments obtained by CID are not enough to differentiate between different types of oxidation on more complex lipid structures.

2.     Can you elaborate on how ionizable lipids form adducts with mRNA? How prevalent are these reactions?

Adam Crowe: That is a very good question, and it’s hard to say. We looked at countless ionizable lipids and their ability to modify the mRNA. The outcome is that it is variable. We have not found an exact indicator to determine how many adducts are being formed or how radically they are being formed. It requires empirical data. Another point to consider is that many of these species are not necessarily coming from the lipids themselves. They can also be impurities introduced by client materials, which can cause these adducts. An estimation based on the information published in the suggests that adduct formation every 500 nucleotides (nt) of the mRNA can cause its inactivation. This means that it takes only a very little amount of the N-oxide to be present to inactivate a large percentage of your mRNA.

3.     Can you explain how EAD compares to ECD and/or ETD?

Matt Stone: There are a few unique differences to consider:

·        Tunability

·        Efficiency of fragmentation

·        Reaction speed

The kinetic energy of the electrons can be adjusted with EAD, which is also the basis of the work presented in the . The energy that was used for the lipid impurity work was in the order of ~12 eV. This is necessary to efficiently fragment these singly charged species and allow full structural elucidation. However, on commercially available electron capture dissociation (ECD) instruments, the kinetic energy is limited to 0–1 eV, which is a low kinetic energy. While EAD can be used to fragment at lower kinetic energies (ECD and hot ECD) and for peptides, higher energies (electron impact excitation of ions from organics, or EIEIO) can be used for efficient fragmentation of other molecules, such as singly charged small molecules or lipids. Unlike electron transfer dissociation (ETD), no transfer gas is needed for EAD, which means the reaction is fast and efficient, allowing for fast data dependent acquisition (DDA) compatible with analytical flow rates. ETD cannot fragment singly charged species and even shows limitations for doubly charged species, while EAD works well for these low-charged precursors.

4.     Have you used the ZenoTOF 7600 system to look into the other lipids in LNPs? Do you see other impurities of concern?

Adam Crowe: The ZenoTOF 7600 system comes with an atmospheric pressure chemical ionization (APCI) source, which is helpful for the analysis of other components used in LNPs, such as cholesterol and related impurities. Oxidation of cholesterol can also occur, but the compound does not ionize very well with electrospray ionization (ESI). It is something we plan to try with the system in the future. The source of your material really determines the purity. Typically, phospholipids that meet GMP standards are usually of high quality while PEG lipids are a concern because of the polydispersity. ESI can provide a good understanding of that and help you analyze all the components.

Matt Stone: We also analyzed various other ionizable lipids, including both commercially available lipids and proprietary ionizable lipids for determining impurities. We mainly focused on N-oxidation as this was a major interest. Stress samples and formulated LNPs are examples of several different impurities/degradants that are of potential concern.

5.     We are a PNI client. What analytical services does Precision NanoSystems Inc. offer to clients? Do you test for these lipid impurities?

Adam Crowe: We like to view ourselves as an end-to-end solutions provider. We have experts on the chemical side, the analytical side and all the way through to the cell biology/cell potency side. So, the short answer is, yes, we will do these analytics for you. From the lipid perspective, we have extensive capabilities, including looking at impurities, solubility and stability and developing methods for routine analysis and more related to both the drug product and the drug substance.

6.     Do regulatory agencies require assessment of mRNA lipidation as a release test for mRNA LNP vaccines?

Adam Crowe: The simplest answer is, not yet. I have some involvement with the guides on analytics related to mRNA LNPs put out by the National Institute of Standards and Technology (NIST) and the ISTN. There is a great deal of discussion around this topic. Although there is guidance available now, there could be regulatory requirements in the future. It is still very early in the field, and the behavior of the adducts is complicated. How much of a concern these adducts constitute is yet to be determined.

7.     Can you comment on the levels at which N-oxides are impacting mRNA efficacy?

Adam Crowe: It is our current understanding that once mRNA is modified by these lipid impurities, the translation of the mRNA is impacted. The ribosome will not be able to recognize the modified nucleoside and will not complete the expression of the protein of interest. So, it is directly proportional: 10% of modified mRNA means a 10% drop in potency. For certain types of medications, such as vaccines, this can be a very significant drop.

8.     How much optimization of the EAD method and how much sample were needed to achieve that kind of fragmentation? Is the method applicable to other lipid species, or is an optimization required for each compound individually?

Matt Stone: The amounts needed are comparable to CID analysis and are not a concern. For the data presented, we used 0.4 μg on column, which enabled us to find, efficiently fragment and elucidate impurities well below 0.1%. The EAD method does not require optimization in terms of fragmentation parameters for each component individually. The fragmentation of these singly charged species is very efficient when using high electron kinetic energies >10 eV. A user can specify target precursor m/z of interest. However, one can also acquire the data with DDA to get fragmentation data on a wealth of components in the sample. The speed of EAD enables compatibility with DDA at analytical flow conditions, which is particularly helpful when the components in a sample are not yet well known.

9.     Which HPLC method was used to assess the ionizable lipid purity?

Adam Crowe: The work that was presented during the is based on an analytical flow ExionLC AD system under reversed-phase conditions. The Phenomenex Kinetex Phenyl Hexyl is the column of choice with dimensions of 100 mm x 2.1 mm. Mobile phases A and B were water and methanol/acetonitrile, respectively. While in this case 10 mM ammonium acetate was used as a modifier, 0.1% formic acid can also be used as a modifier. The tertiary amine of the ionizable lipid can be prone to tailing, which can be reduced by adding an ion-pairing agent, such as TFA.

10.  What is the current level of expectation of the FDA for the industry on adducts data for new ionizable lipids?

Adam Crowe: No guidance has yet been issued. However, the US Food and Drug Administration (FDA) is well aware of this issue and is monitoring the field for direction. In general, we have started to see increased questions about impurities in LNPs during investigational new drug (IND) filings that may indicate a change in the FDA’s future views on the adducts.

11.  Could you explain how the lipid library was made?

Adam Crowe: At a high level, a list of key parameters for iL activity was compiled and an in-depth review of the IP space was completed. The resulting predicted structures underwent multiple rounds of optimization, followed by extensive screening.

12.  Would lipid impurities be a problem for other payloads, such as sgRNA?

Adam Crowe: Yes, these reactive species can also pose a risk for single guide RNA (sgRNA) used for CRISPR/Cas9 applications. One interesting question is, why didn’t we see these adducts earlier? Small interfering RNA (siRNA) LNPs have been around for many years now. The indicates that the rate of formation of these adducts is very low and is largely determined by length. If we estimate 1 adduct for every 500 nt, that would mean only a subfraction of the short siRNA is affected. For mRNA, a very small amount of lipidation events can inhibit its function. You may see such adducts forming with sgRNA, which are usually in the order of 100 nt. However, because of the size, it likely will not be a huge driver of the quality of your material.

13.  What approaches have you been taking to control the N-oxide impurities?

Adam Crowe: When possible, heat and exposure to air (oxygen) should be avoided. The range of N-ox formation is highly dependent on structure, and empirical data is required to assess the risk.

14.  Are there indications that any of these impurities given to adducts might be forming during the formulation processes?

Adam Crowe: This is a complicated question as it depends on several factors. For example:

·        What is the source of the adducting species (N-ox iL or other impurity)?

·        What is the propensity of the unique iL to cause adducts?

·        What formulation conditions are used?

When the first two are well controlled, the last becomes less significant. In fact, adduct formation over time can be measured in the drug product during stability testing.

15.  What percentage of potency losses do you usually see due to the adducts?

Adam Crowe: Our initial findings corroborate those in the , where high adduct levels correspond to low or loss of potency in vitro. Data are still limited on this issue, and it is complicated to study due to the concurrent degradation of mRNA with adduct formation.

16.  Could you comment on ELSD compared to CAD for use in lipid QC?

Adam Crowe: We use several different HPLC detectors at PNI for different applications. In general, CAD is preferred over ELSD due to its higher sensitivity, linear and quadratic response and response consistency between different lipids. Therefore, CAD is preferred for impurity/purity assessment, raw material testing and total lipid quantification. However, ELSD can be effectively employed for lipid content in drug product when combined with orthogonal assays to test for impurities.

17.  Is there any focus on possible breakdown products of ionizable lipids when considering the design to be used—i.e., homology with phosphatidylinositol (PIP3) and other cell signaling species?

Adam Crowe: Ionizable lipid design is not my specialty, but at a high level, the need for increased metabolism and expression has fueled interest in the biodegradable family of iLs. We are seeing more ether and amine linkages with esters and amides to promote rapid clearance in the aim of reducing potential toxicity concerns.

18.  Did you ever see a reduction of the N-oxide under EAD conditions?

Matt Stone: We have not seen a reduction of the N-oxide under EAD conditions based on our initial data evaluation. We are happy to discuss this topic further with you directly. Feel free to reach out to your local SCIEX contact with this request.

19.  You mentioned Precision NanoSystems Inc. has over 100 ionizable lipids. Are those your property or just a general library of available lipids?

Adam Crowe: Those 100 ionizable lipids were all developed by PNI and are available for license through PNI. For clients at an early pre-clinical phase, sourcing ionizable lipids can be a challenge. The commercially available lipids, which were used in the recent COVID-19 mRNA vaccines, work very well but can be difficult to obtain and take to the clinic and can also be very expensive to license. PNI offers lipids that are already pre-screened for different application areas, such as cell or gene therapy or vaccine applications, and ready to be used to provide a clear path for clients going into clinical studies. We can also use non-proprietary lipids and help determine the best options after discussions with you.

20.  How many mRNAs fit in one LNP? Is there an assay to find this?

Adam Crowe: This is a complex question as LNPs do not exist as a homogeneous population, and mRNA sizes range from <1,000 kb to >15,000 kb. In fact, even the most optimized formulations can contain a distribution of sizes from 20 nm to 200 nm. The particle volumes therefore differ, and the mRNA or particle will vary across sizes. Similarly, the presence of empty LNPs will also bias any average result. As a result, particles require separation by size, density or charge prior to assessing mRNA content. There are limited options available, with field-flow fractionation coupled to multi-angle light scattering (FFF-MALS) being the most recommended in the LNP field. With this option, LNPs are separated chromatographically by size and the mass contribution of the mRNA and lipids is measured for each LNP population to derive an mRNA/ particle value.

21.  Are the impurities in MC3 from the synthesis step (like reactants or solvents) or after synthesis and during storage?

Adam Crowe: I suspect the impurities are from both. We have seen some iL batches in which a reactive impurity accounts for the majority of adduct formation. Assuming 1–2 adduct events per mRNA is enough to inhibit translation, stoichiometry implies that trace levels of reactive impurities could significantly impact LNP potency. However, logically, we see impurities becoming less impactful with higher purity materials. As a general rule, iLs with >90% relative purity by CAD/MS generally show less adduct formation than incompletely purified materials.

22.  How long does it take for MC3 to form MC3-OH adduct? Does freshly prepared MC3 contain this adduct? Will the adduct amount increase along with storage duration?

Adam Crowe: Rates are dependent on iLs and storage. While iLs are, generally speaking, quite stable at cryo temperatures, certain solvents and an elevated temperature can lead to significant increases of oxidative species in a short period of time. It’s important to note that the N-oxide itself is modifying the degradation product, not the mRNA. And yes, the adduct formation will increase over time if it is in the solution and in the presence of oxygen. This occurs during storage when it is not frozen. Adducts appear less problematic at cryo temperatures.

23.  Can you comment on how much signal improvement you can obtain when using the Zeno trap? What about signal to noise?

Adam Crowe: We observed that an improvement of signal is obtained throughout the entire mass range of the MS/MS spectrum. More signal improvement can be expected at the lower mass range versus the higher mass range, with a 5–15x improvement depending on the fragment. This signal gain also translates into signal-to-noise gain.

Related to RUO-MKT-11-15013-A