System and method for lipid carriers DNA and mRNA characterization

Abstract

The present invention relates to a system and method for the comprehensive, comparative and rapid assessment of new modality therapeutics, liposomes, and encapsulated products including lipid nanoparticles (LNPs) carrying mRNA, DNA, or proteins to a patient — through a comparative analysis of acquired hyperspectral images and the associated microspectroscopic spectral data against an appropriate reference. In particular, the system and method acquires spectral images of solutions or mixtures of both a sample and a reference, and then analyzes spectral data obtained from the spectral images using: (1) Co-distribution (2D-CDS) analysis, (2) two-trace two-dimensional (2T2D) analysis, and (3) two-dimensional correlation spectroscopy (2D-COS) analysis to assess the structure, composition and stability of the new modality entity, its carrier and the encapsulated form as well as the effect of the excipient on the stability of the therapeutic. The significance extends to neural network model building and machine learning to predict the outcome of newly designed biotherapeutics.

Description

[0001] SYSTEM AND METHOD FOR LIPID CARRIERS DNA AND MRNA CHARACTERIZATION

[0002] Cross-Reference to Related Applications

[0003] This application claims the benefit of U.S. Provisional Application No. 63 / 626,917, filed January 30, 2024, the entirety of which is hereby incorporated by reference.

[0004] Statement Regarding Federally Sponsored Research None

[0005] Field of the Invention

[0006] The present invention relates to a system and method for the comprehensive, comparative and rapid assessment of new modality therapeutics-encapsulated products including lipid nanoparticles (LNPs) carrying mRNA, DNA, or proteins for delivery to a patient — through a comparative analysis of acquired hyperspectral images and the associated microspectroscopic spectral data against an appropriate reference.

[0007] Background

[0008] The COVID-19 pandemic demonstrated the therapeutic value of clinical messenger RNA (mRNA) vaccines for patients worldwide. These new therapeutic entities have complex structures that are not amenable to existing, well-established high-resolution techniques for evaluation and assessment. They are prone to nuclease degradation, thermally sensitive and are conformationally sensitive to cation interactions. Encapsulation of the mRNA using lipid nanoparticles (LNPs) has provided an efficient means for delivery into targeted cells for the expression of a disease-specific antigen protein. The LNPs are comprised of several lipid components, including an ionizable lipid, cholesterol and pegylated lipid that plays a central role in the effective delivery of the mRNA. The LNPs may also deliver DNA, proteins, or other therapeutics. The ionizable lipid is positively charged under acidic pH, allowing for the charge- driven interaction with the negatively charged therapeutic entity (mRNA). After a period of incubation, the pH of the solution is increased to achieve a neutral surface charge of the ionizable lipid. Other lipid components, such as pegylated lipid and cholesterol aid in the stabilization of the core shell of the nanoparticle. This level of complexity is key to an effective encapsulation of the mRNA for delivery whereby the internal architecture of the LNP stabilizes the mRNA, while the external architecture that contains the pegylated lipid sterically and thermally stabilize the core shell of the nanoparticle. The degree of complexity of these encapsulated systems makes it a challenge for their biochemical and biophysical characterization.

[0009] Existing analytical techniques such as Cryo-EM, X-ray scattering and zeta-potential have provided limited information regarding the size distribution of these LNP in their full (encapsulated therapeutic) and ghost (empty LNP) states. Another technique such as circular dichroism (CD) provides thermal denaturation of the purified mRNA. Other techniques such as differential scanning calorimetry (DSC) provide the overall thermal stability and thermodynamic parameters that are used to quantitate the stability of the mRNA, full (encapsulated therapeutic) and ghost (empty LNP). All of the analytical techniques presented above share the same limitation, that is a globally averaged signal, thus limiting the understanding of the inter-play during thermal stress to achieve stabilization of the LNP encapsulated mRNA. A need exists to employ a selective and sensitive biophysical technique that can provide the structural information, differentiated dynamics of the main biochemical components and the molecular interactions that drive the stability of the: (1) LNP, (2) mRNA and (3) encapsulated mRNA under varying conditions.

[0010] Summary of Invention

[0011] The system and method described herein provides an analytical platform technology for evaluation of potential mRNA, DNA, and lipid carrier platform candidates. In particular, the system and method acquires spectral images of solutions or mixtures of both a sample and a reference, and then analyzes spectral data obtained from the spectral images using: (1) Codistribution (2D-CDS) analysis, (2) two-trace two-dimensional (2T2D) analysis, and (3) two- dimensional correlation spectroscopy (2D-COS) analysis to assess the structure, composition and stability of the new modality entity, its carrier and the encapsulated form as well as the effect of the excipient on the stability of the therapeutic. These 2D algorithms provide the basis to selectively deconvolve the spectral data generated to allow for neural network (NN) model building required for machine learning (ML) and artificial intelligence (Al) to predict developability and safety. For example, the system and method may build and train machine learning algorithms using high quality data sets and direct principles to discern the minimal variables and predict or build mRNA sequences that are stable and behave as desired. As part of the 2D-CDS, 2D2T and 2D-COS analysis, plots are generated and signature peaks within the plots may be identified and analyzed. Analyzing these signature peaks provides an identification of the components in the samples.

[0012] For example, signature peaks related to the therapeutic, the carrier, and the encapsulating components may be identified. Finally, the key signature peaks identified for each component are evaluated as a function of temperature, and thermal dependence is evaluated to determine phase transition temperatures and pre-transition temperatures for the new modality therapeutic as well as the carrier and the encapsulated form to allow for stability assessment and shelf-life prediction. Furthermore, for the system and method uses an array of samples all subject to the same perturbation conditions, allowing for comparability assessment and determination of the factors needed for predictive results. For example, the multiple 2D correlation algorithms allow for the deconvolution of the spectral data, even when highly overlapped bands exist in the spectral region of interest. The deconvolution is achieved through the enhanced spectral and temporal resolution of the covariance data. The spectral deconvolution allows for the separation of the key signature peaks for each component based on their different molecular dynamics. The different molecular dynamics are inherently due to their different structural moieties and thermal transition temperatures providing a focused and scientifically driven selection process of the key variables needed for machine learning model generation.

[0013] Using the system and method described herein, a commercially available double stranded mRNA platform was comprehensively analyzed to determine the detailed molecular dynamics of the molecule and its relationship to the structure and its stability, thus demonstrating the sensitivity and selectivity of the method The system and method, as described herein, therefore provide for characterization of the mRNA, allows for small changes to be made to the mRNA platform to ensure manufacturability and achieve the desired therapeutic properties. Similarly, DNA platforms may also be analyzed. Furthermore, the analytical platform of the system and method described herein provides for the selective evaluation of different components within a therapeutic entity (i.e., the components of the LNP and the encapsulated mRNA or DNA) under thermal perturbation. In addition, the analytical platform technology may provide real-time comparability assessment for an array of samples allowing for a multiattribute analysis method. The proposed system and method take advantage of the combination of three components of the platform analytical technology to perform comparability assessments: (1) the comprehensive nature of the evaluation is based on real-time, highly selective and sensitive QCL analytical platform technology, (2) the design of the method which includes the accurate thermal perturbation of all the samples within the array and (3) by employing different 2D correlation algorithms allowing for deconvolution of the spectral data by enhancing the covariance spectral resolution and taking advantage of the temporal resolution to determine the differences in molecular dynamics among the different components of the complex system. Specifically, the method allows for the distinction of the different molecular dynamics / structural / stability relationships for each component to be established within a complex system like the LNP encapsulated mRNA when compared to pure components in the same slide cell array subject to accurate thermal control. The results of the platform provide detailed understanding towards thermal stability of the biochemical components (i.e., mRNA and varying lipids) within the LNP which is critical to the design of these new therapeutic modalities. The accurate thermal control that will allow for structural, stability evaluation and shelf-life prediction of the therapeutic modality. The method further enables focused neural network model building and machine learning to provide a competitive advantage for drug design, de-risking pre-clinical candidate selection, development and manufacturing of biotherapeutics, and increasing the rate of regulatory approval and speed to market.

[0014] The method is not limited to lipid nanoparticles or mRNA, but new carriers and modalities can also be characterized, such as Adeno-associated viruses as carriers of gene therapy or other types of carriers that may be developed to ensure long-term stability of the encapsulated therapeutic. A comprehensive evaluation of complex multicomponent systems is the driving force towards the biophysical and biochemical characterization of the therapeutic. The system and method described herein provide a combination product solution that employs vibrational spectroscopy apparatus, such as a quantum cascade laser microscope (QCLM), Raman amplifier, FT-IR, or light detection and ranging (LIDAR) system in the MID infrared spectral region for the fast acquisition of hyperspectral images, along with a slide cell array to allow for the comparability assessment and a series of dedicated software modules that effectively employ the three 2D-correlation algorithms required to fully discern the complex nature of the combination therapeutic (i.e., the carrier typically an LNP and the new therapeutic entity). The slide cell array may be comprised of a polymer, such as polyethylene, calcium fluoride, barium fluoride, or other polymer that is transparent within most of the spectral region of interest. The samples to be screened, along with one or more reference standards, or controls, are provided within the cells of the slide cell array, and the QCLM is used to acquire spectral images of the slide cell array. The method may also be applied to a microfluidics system where the sample is being monitored in real-time.

[0015] The analytical method allows for the achievement of the following objectives:

[0016] 1. Generate a chemical map of the HS images of the liposome, ghost LNP, encapsulated mRNA and mRNA samples.

[0017] 2. Identify the biochemical composition of each type of sample.

[0018] 3. Determine the pre-transition temperatures of the mRNA, ghost LNP and LNP encapsulate mRNA.

[0019] 4. Determine the gel-to-liquid crystalline Tmtransition temperature for the liposome samples.

[0020] 5. Determine that the liposome or LNP sample did not undergo dehydration, key to stability.

[0021] 6. Identify hydrogen bonding interaction between cholesterol and the phospholipid as well as with the surrounding aqueous environment.

[0022] 7. Identify correlations that provided internal validation of the interactions and the hydrogen bonding state within the liposomes as well as the LNP.

[0023] 8. Identify weak interactions between the mRNA and the lipidic environment.

[0024] 9. Determine the extent of double stranded (ds) mRNA in a sample.

[0025] 10. Determine the Tmtemperature for the double stranded to single stranded (ds-to-ss) transition.

[0026] 11. Determine the fraction of ds mRNA as a function of increasing temperature.

[0027] 12. Determine the stability of the mRNA sample.

[0028] 13. Determine the presence of mismatch base pairing against a standard mRNA. 14. Determine events of phosphate hydrolysis within the mRNA sample.

[0029] 15. Determine for the ds mRNA the type of helical form A or B .

[0030] 16. Determine the molecular dynamics of mRNA based on the disruption of the hydrogen bonding interactions between the bases and the weak interactions that stabilize the backbone of the mRNA.

[0031] 17. Establish the molecular differences between the ds mRNA and the ss mRNA under varying conditions including varying sequences.

[0032] 18. Determine whether certain excipients in the formulation promote the stability of the mRNA or the LNP encapsulated mRNA.

[0033] 19. Critical deconvolution of spectral data critical to neural network (NN) model building for M.

[0034] A double stranded mRNA standard was evaluated using the system and method, with the results demonstrating the selectivity and sensitivity of the method for distinguishing between the double stranded and single stranded mRNA, describing dynamic molecular events, as well as distinguishing between the A and B helical forms. The sequence dependent structural relationship for mRNA is critical to de-risk candidate selection and drug design. The presence of non-canonical nucleotide pairing known as wobbles (G-U base pair) within the mRNA sequence can destabilize the structural integrity and the stability of the mRNA. It is therefore beneficial to probe for the double stranded to single strand transitioned of the mRNA and monitor the molecular dynamic transitions associated with this process through key signature peaks within the spectral data against a reference standard as discussed below. By monitoring the temperature at which the GC syn conformer is perturbed, information regarding the content of wobbles can be determined. For example, the lower the transition temperature for the loss of the syn conformer, the higher the content of wobbles in the mRNA sequence will be. A greater content or number of wobbles indicates greater risk for hydrolysis and therefore lower double stranded content and less stability. This analysis provides both quantitative and qualitative comparability assessment for an array of mRNA pre-clinical candidates comprised of different sequences. The impact on stability is dependent on the number of wobbles present in the sequence and on the location of the same within the structure. For example, the stability of the mRNA will be differentially impacted if the location within the structure is the loop / stem or if it is the neck / end cap.

[0035] Another risk in the design of the mRNA is the hydrolysis of the phosphate backbone in regions where unpaired bases are present, impacting the backbone of the messenger RNA. The observed disproportionate loss of form A or B double helix structure and the impact on lowering the phosphate backbone transition temperature (7™) would suggest the influence of a less than desirable mRNA sequence design. In addition, establishing the Tmfor the double stranded to single stranded transition via the disruption of the hydrogen bonding interaction between the nucleotides. Internal probes may be used to assist in determining the double stranded to singlestrand transition. For example, the stretching vibrational mode v (C-N) at 1398 -1424 cm’1would serve as an internal probe for double stranded mRNA to single stranded mRNA. This strategy is also valid even in the case of synthetic unnatural nucleosides present in the sequence may become internal probes as well.

[0036] The system and method, and datasets obtained from the method, may be used for neural network model building. As discussed herein, the method was used to simulate a liposome and mRNA mixture, and the method was then used to deconvolve the spectral data for each component using the 2D correlation algorithms. Using the same method, a neural network model can be built from datasets of experimental data as well. The method provides the framework for the successful neural network model building for highly complex biotherapeutics along with their carriers. Specifically, neural network models for artificial intelligence and machine learning may be developed based on: (1) real-time spectral data with enhanced spectral and temporal resolution, (2) identification of key signature peaks specific to the component of interest within the complex biochemical system, (3) molecular dynamics associated with the thermal transition that in turn provide the evidence of the stability, (4) relationship of the nucleotide sequence and the structure for the case of the mRNA, (5) role of the excipient in stabilizing the biotherapeutic and the carrier and (6) simulation of the spectral data. These are achieved using 2D-COS, 2T2D or co-distribution algorithms. For example, the spectral deconvolution is achieved through the correlation of mRNA associated cross peaks that change during their thermal transition in a synchronous manner, allowing for the identification of structurally sensitive key signature peaks present for the complex system that are also confirmed when compared to the mRNA as a pure component within the slide cell array. Similarly, the asynchronous correlation of the cross peaks present in a separate asynchronous plot that allow for the identification of other components within the complex system, such as the liposomes or the LNPs. This can be achieved even in a highly overlapped spectral region of interest, as the separation is achieved by the differences in molecular dynamics / stability between the components of interest. The comparability assessment allows for the comparison of pure components and the mixture; any interaction whereby the changes in stability occur when compared to the pure component can also be identified. The pure components may be, for example, a known reference sample of a pure sequence of mRNA, the LPN, and / or the liposome.

[0037] Brief Description of Figures and Drawings

[0038] FIG. 1 is an illustration of a hyperspectral image output generated by the QCLM platform technology demonstrating the homogeneous distribution of the liposomes within the well at 4.5 pm spatial resolution. The image contains 223,000 QCL infrared spectra within the spectral region of 1780 - 1000 cm'1at 2 cm'1spectral resolution.

[0039] FIG. 2 is an illustration of an overlaid QCL spectral data within the spectral region of 1780 - 1000 cm'1at 2 cm'1spectral resolution within a temperature range of 30 - 46 °C with intervals of 2 °C and 4 minute thermal equilibrium periods. The peak assignments are indicated in the illustration for a cholesterol containing liposome sample.

[0040] FIG. 3A is an illustration of a QCLM overlaid covariance spectra of the full spectral region of 1770 - 1000 cm'1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C.

[0041] FIG. 3B is an illustration of a 2D-COS synchronous plot within the spectral region of 1770 - 1000 cm'1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C.

[0042] FIG. 3C is an illustration of a 2D-COS asynchronous plot within the spectral region of 1770 - 1000 cm'1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C.

[0043] FIG. 4 is an illustration of QCLM overlaid covariance spectra of the full spectral region of 1770 - 1000 cm'1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C. The illustration highlights the two spectral regions chosen associated with the lipid headgroup: (left box) lipid carbonyl (1770 -1700 cm'1) and (right box) glycerol phosphate (1200 - 1000 cm'1) modes. FIG. 5A is an illustration of overlaid QCL spectral data within the spectral region of 1290-1020 cm’1with a spectral resolution of 2 cm’1and a temperature perturbation range of 30 - 46 °C with intervals of 2 °C.

[0044] FIG. 5B is an illustration of QCLM overlaid covariance spectral data within the spectral region of 1290-1020 cm’1with a spectral resolution of 2 cm’1and a temperature perturbation range of 30 - 46 °C with intervals of 2 °C.

[0045] FIG 6A is an illustration of a comparative 2T2D synchronous plot of a liposome (sample) against a reference within the headgroup spectral region of 1290 - 1030 cm’1at 30 °C.

[0046] FIG 6B is an illustration of a comparative 2T2D synchronous plot of a liposome (sample) against a reference within the headgroup spectral region of 1290 - 1030 cm’1at 38 °C.

[0047] FIG 6C is an illustration of a comparative 2T2D synchronous plot of a liposome (sample) against a reference within the headgroup spectral region of 1290 - 1030 cm’1at 42 °C.

[0048] FIG 6D is an illustration of a comparative 2T2D synchronous plot of a liposome (sample) against a reference within the headgroup spectral region of 1290 - 1030 cm’1at 46 °C.

[0049] FIG 7A. is an illustration of a comparative 2D-COS synchronous plot of the liposome headgroup spectral region of 1290 - 1030 cm’1and a temperature perturbation range of 30 - 46 °C with intervals of 2 °C.

[0050] FIG 7B. is an illustration of a comparative 2D-COS asynchronous plot of the liposome headgroup spectral region of 1290 - 1030 cm’1and a temperature perturbation range of 30 - 46 °C with intervals of 2 °C.

[0051] FIG. 8A is an illustration of QCLM spectral overlays of the full spectral region of 1800 - 1000 cm’1for a commercially available engineered double stranded mRNA poly (LC) commonly used as a reference standard within the temperature range of 25 - 65 °C with temperature intervals of 4 °C.

[0052] FIG. 8B is an illustration of QCLM spectral overlays for an engineered mRNA poly (I:C) in the spectral region of 1442 - 1178 cm’1within the temperature range of 25 - 65 °C with temperature intervals of 4 °C (dark black line) corresponds to the spectrum at 65 °C.

[0053] FIG. 9 provides illustrations for mRNA poly (I:C) 2D-COS synchronous plot within the spectral region 1442 - 1320 cm’1for the temperature range of 25 - 65 °C at 3.8 pg / pL in the presence of 0.9% NaCl. FIG. 10A provides an illustration of a 2T2D synchronous plot within the spectral region 1780 - 1014 cm'1for mRNA poly (I:C) at 15.934 pg / pL in the presence of 0.9% NaCl at 65 °C against 25 °C (reference).

[0054] FIG. 10B is an illustration of a histogram plot summarizing the 2T2D synchronous cross peaks and Table 5 for mRNA poly (I:C) at 15.934 ug / uL in the presence of 0.9% (w / v) NaCl at 65 °C against 25 °C (reference) within the spectral region of 1780 - 1014 cm'1. The intensity differences for the synchronous cross peaks shown demonstrate relative stability observed for each mRNA double stranded form.

[0055] FIG. 10C provides an illustration of a 2T2D synchronous plot within the spectral region 1780 - 1014 cm'1for mRNA poly (I:C) at 23.427 pg / pL in the presence of 0.9% (w / v) NaCl at 65 °C against 25 °C (reference).

[0056] FIG. 10D is an illustration of the histogram plot summarizing the 2T2D synchronous cross peaks and Table 6 for mRNA poly (I:C) at 23.427 pg / pL in the presence of 0.9% (w / v) NaCl at 65 °C against 25 °C (reference) within the spectral region of 1780 - 1014 cm'1.

[0057] FIG. 11A is an illustration of a co-distribution synchronous plot for distribution population of mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of: 1442 - 1324 cm'1.

[0058] FIG. 11B is an illustration of a histogram plot summarizing the co-distribution synchronous cross peaks for mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of 1442 - 1320 cm'1. All cross peaks are positive.

[0059] FIG. 11C is an illustration of a co-distribution asynchronous plot cross peaks distribution population mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of 1442 - 1320 cm'1.

[0060] FIG. 11D is an illustration of a co-distribution histogram plot summarizing the asynchronous cross peaks for mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of 1442 - 1320 cm'1.

[0061] FIG. 12A provides illustration of a 2T2D synchronous plot within the spectral region 1440 - 1324 cm'1for mRNA poly (I:C) at 15.934 pg / pL in the presence of 0.9% NaCl at 46 °C against 30 °C (reference). FIG. 12B is an illustration of a histogram plot summarizing the 2T2D synchronous cross peaks for mRNA poly (I:C) at 15.934 ug / uL in the presence of 0.9% (w / v) NaCl at 46 °C against 30 °C (reference) within the spectral region of 1440 - 1324 cm'1.

[0062] FIG. 13A is an illustration of overlaid spectra in the spectral region of 1008 - 1746 cm'1at 30 °C for: simulated liposome mRNA poly (I:C) mixture (represented by the thick solid line), experimental liposome (represented by the dashed — line), and experimental mRNA poly (I:C) (represented by the thin solid line).

[0063] FIG. 13B is an illustration of overlaid spectra in the spectral region of 1008 - 1746 cm'1at 46 °C for: simulated liposome mRNA poly (I:C) mixture (represented by the thick solid line), experimental liposome (represented by the dashed — line), and experimental mRNA poly (I:C) (represented by the thin solid line).

[0064] FIG. 13C is an illustration of overlaid spectra in the spectral region of 1008 - 1746 cm'1within a temperature range 30 - 46 °C for simulated liposome mRNA poly (I:C) mixture.

[0065] FIG. 14A is an illustration of a co-distribution asynchronous plot for simulated liposome and mRNA poly (I:C) mixture in the spectral region 1436-1324 cm'1and within the temperature range of 30 - 46 °C where the cross peak contributions for both components are present.

[0066] FIG. 14B is an illustration of a histogram plot of the co-distribution asynchronous cross peak intensity changes for simulated liposome and mRNA poly (I:C) mixture in the spectral region 1436-1324 cm'1within the temperature range of 30 - 46 °C and where the cross peak contributions for both components are present.

[0067] Figure 15A is an illustration of the co-distribution asynchronous plot in the spectral region 1436-1324 cm'1where the cross peak contributions are for the pure mRNA (I:C) within the temperature range of 30 - 46 °C for the comparative analysis. Notice the lack of several cross peaks due to the temperature range which is below the Tm for mRNA (I:C).

[0068] FIG. 15B is an illustration of histogram plot of the co-distribution asynchronous cross peak intensity changes for contributions are for the pure mRNA (I:C) within the temperature range of 30 - 46 °C in the spectral region of 1436-1324 cm'1for comparative analysis against the simulated data.

[0069] FIG. 16 is a schematic workflow for the biophysical characterization of full and empty LNP and mRNA / DNA. The approach is also amenable to an array of up to 21 samples compared against a reference standard. Detailed Description of the Invention

[0070] I. Sample preparation

[0071] The multicomponent lipidic sample, LNP or mRNA have been stored under the recommended temperature conditions. An aliquot of the desired sample, reference, or negative control is applied to pre-defined sample wells of the slide cell array which is then placed in a degasser under controlled temperature conditions for set period of time, dependent on the viscosity of the sample to be analyzed. For example, an aliquot of 1-2 pL may be applied to each well within the slide cell array. The reference may be, for example, pure mRNA. Moreover, multiple references may be included in the slide cell array such that the sample being analyzed can be compared to multiple references. The negative controls may be buffers. Once the degassing process has ended the slide cell array is covered with an optically polished cover providing a fixed path length to allow for the quantitative analysis. The slide cell array may be a disposable array. The substrate for the slide cell may be salt or other IR transparent material including polymers, such as polyethylene, calcium fluoride, barium fluoride, or other polymer that is transparent within most of the spectral region of interest. The slide cell may then be assembled into a thermally controlled slide cell accessory to ensure a controlled thermal environment for hyperspectral image acquisition. The slide cell accessory may be, for example, a controllable heated chamber configured to receive the slide cell.

[0072] The system and method may allow for evaluation of drug products that have high concentrations and high doses. The system and method may also allow for evaluation of drug substances with low concentrations, including by increasing the pathlength for the analysis. This allows for both critical quality attribute assessments and product quality attribute assessments.

[0073] IL Hyperspectral (HS) image acquisition

[0074] After the slide cell array is placed in the slide cell holder, spectral images of the slide cell array are obtained. An infrared microscope, preferably a quantum cascade laser (QCL) microscope which allows for the rapid collection of a series of HS images with collecting 16x more spectral data per pixel and as a result providing an enhanced signal / noise ratio when compared to an FT-IR microscope. The QCL microscope is equipped with a heated accessory and a slide cell array, as described above, to allow for the acquisition of a series of HS images for the evaluation of both the physical (i e., phase separation) and biochemical composition of complex samples subject to a thermal perturbation. The advantage is the real-time comparability assessment of the samples of interest against a reference within the MID IR spectral region 1800 - 950 cm'1. HS images may be acquired within a desired temperature range, with predetermined equilibration periods in between acquiring HS images at different temperatures within the range. For example, the equilibration periods may be 4 minutes. Each HS image acquisition occurs within seconds and is comprised of 223,000 QCLM spectra collected at 2 cm'1spectral resolution. The entire MID IR spectral region is valid for the analysis. Further, other microscopy techniques may be applied to obtain the spectral images, such as FT-IR or Raman microscopy.

[0075] III. Two trace two-dimensional (2T2D) correlation analysis

[0076] After the spectral images are acquired, a two-trace two-dimensional (2T2D) correlation1process is applied to the spectral data.. From a pair of spectra a two trace two-dimensional correlation can be applied to generate a synchronous spectrum and asynchronous spectrum The synchronous spectrum and asynchronous spectrum are given by:

[0077] The first, original spectra, s(v), corresponds to the sample and the second spectra, r(v), corresponds to the reference, respectively. A 2T2D correlation coefficient is then applied p(v1,v2) to the synchronous evaluation, and a disrelation coefficient to the asynchronous evaluation resulting in the scaled version of the 2T2D correlation spectra, as given by:

[0078] Where: (5)

[0079] This indicates the complementarity nature of the quantities.

[0080] Two contour plots are generated. These are the synchronous plot where dominant spectral components of the samples: s(v) and r(v) are observed. The diagonal is comprised of auto peaks where V-L = v2. The cross peaks are always positive. The second plot is the asynchronous ('P(v1, v2)) plot, which is more informative. In the case T* > 0, then the intensity contribution of the functional group is from vi, corresponding to the first component being more abundant. In the case T < 0, then the intensity contribution of the functional group is from V2, therefore, the second component of the sample is more abundant. Also, this indicates that peaks of the same intensity and sign correspond to the same component within the sample.

[0081] IV. Two-dimensional correlation spectroscopy (2D-COS) analysis

[0082] 2D-COS has proven useful in that it provides a detailed molecular description of the effects of the perturbation on both the side chains and, in turn, the conformational stability of the mRNA. The stability of a mRNAis dependent on its sequence, secondary and tertiary structure, post-translational modifications and formulation conditions. This correlation analysis enhances the spectral resolution to provide such information as the mRNA is destabilized, therefore serving as a true fingerprint for its identification. The 2D-COS algorithm is defined as:

[0083] Where, is the initial spectrum of the data set. Synchronous 2D correlation intensities of the covariance spectral data are defined by:

[0084] The resulting correlation intensity<t>(v1, v2) as a function of two independent wavenumber axes, vi and V2, is the synchronous plot.

[0085] Asynchronous 2D correlation intensities of the covariance spectral data are defined by: The term Nijis the element of the so-called Hilbert-Noda transformation matrix and is given by:

[0086] The cross-correlation function is applied to a covariance spectral dataset, which is a dataset obtained from subtracting an initial spectrum from subsequent spectra. For example, the initial spectrum may be acquired at low temperature, and subsequent spectra acquired at higher temperatures. The spectral changes due to temperature increase are then revelated (revealing changes in the mRNA or LNP behavior, for example), which are referred to as covariance spectral data or difference spectra. Applying the cross-correlation function to the different spectral dataset results in two separate, yet symmetrical 2D plots. The first plot is referred to as the synchronous plot. It contains positive peaks on the diagonal, known as the auto peaks, and provides the overall changes observed in the spectral dataset. The relationship established in this synchronous plot relates to the spectral intensity changes that occur synchronously, hence the name. The second 2D plot is known as the asynchronous plot. This plot relates the asynchronous intensity changes, resulting in enhanced spectral resolution, which can be used to correlate out- of-phase peak intensity changes, such as those defined for the deamidation process.

[0087] Both plots contain off-diagonal peaks, which are referred to as cross-peaks,’ these peaks correlate the spectral changes observed at the frequencies assigned to the various bands (e.g., backbone and side chain structural elements). Spectral intensity changes observed are due to the incremental thermal perturbation applied to the sample. No a priori knowledge of the system is required for interpretation of the results. The information in both plots allows for determining the sequential order of molecular events that occur during the perturbation by following Noda’s rules.2'4These plots are symmetrical in nature and for interpretation purposes, reference is made to the top triangular portion of the plot for analysis. The asynchronous plot is comprised exclusively of cross peaks that relate the out-of-phase peaks. As a result this plot reveals greater spectral resolution enhancement. The following rules can apply to establish the order of molecular events:

[0088] I. If the asynchronous cross peak V2: if positive, then V2 is perturbed prior to

[0089] II. If the asynchronous cross peak vr. if negative, then V2 is perturbed after III. If the corresponding synchronous cross peak is positive, then the order of the events is established using the asynchronous plot (rules I and II).

[0090] IV. However, if the corresponding synchronous cross peak is negative and the asynchronous cross peak is positive, the order is reversed.

[0091] V. If the synchronous plot contains negative cross peaks and the corresponding asynchronous cross peak is negative, then the order is maintained.

[0092] An order of events can be established for each peak observed on the V2 axis.

[0093] In summary, the combination of the two algorithms is powerful and provides an in-depth assessment for therapeutic authentication, such as authentication of an mRNA therapeutic: (1) a rapid comparability assessment to distinguish the mRNA component from the encapsulated LNP using the 2T2D algorithm and (2) a more dynamic fingerprint that is sensitive to the sequence, post-translational modifications, excipient presence, perturbation and formulation conditions. In addition to mRNA, DNA or proteins may also be analyzed in the same manner.

[0094] V. Two-dimensional co-distribution spectroscopy (2D-CDS)

[0095] 2D-CDS provides the distribution of the LNP encapsulated mRNA, the LNP itself, or the mRNA itself in solution and reveals the main structural events common to the components in the sample. The spectral correlations observed can be compared directly between pure component samples and the mixtures / complex samples, allowing for determinations of relative stability of each component within the sample. Used together, the methods can provide an understanding of the structural changes, the inter- and intra- molecular hydrogen bonding and the stability of the components within the LNP encapsulated mRNA. In addition to mRNA, DNA or proteins may also be analyzed in the same manner.

[0096] 2D-CDS is especially suited for evaluating the dynamic elements common to an ensemble of structures in a solution subjected to a perturbation. For a set of m spectra, are obtained as a function of the spectral variable with j = 1, 2, . .., n and some perturbation variable t1;with i = 1, 2, ... , k at defined intervals between fq and tk. In our data, the spectral variable v is the quantum cascade laser microscope wavenumber (cm4), and the perturbation variable is temperature.

[0097] The dynamic spectrum is defined as: with the average A as the reference spectrum given by:

[0098] The asynchronous 2D-CDS spectrum A(v1(v2) is defined as:

[0099] The total joint variance T is given by:

[0100] By convention, the value of Δ(v1(v2) is set to be zero, when is encountered, indicating a lack of spectral intensity at either QCLM wavenumber. The asynchronous 2D-CDS intensity is a measure of the difference in the distribution of two spectral signals along the perturbation variable axis based on the mathematical tool known as moment analysis.

[0101] The interpretation of the 2D-CDS asynchronous plot is direct. For a positive cross peak Δ > 0, the presence of the spectral signal at v1is distributed predominantly at an earlier stage, prior to v2■ In the case of a negative cross peak À < 0, the order is reversed. Finally, when ≈ 0, the average distributions of the spectral signals are similar, and therefore they co-exist.

[0102] VI. Spectral data analysis

[0103] Biotherapeutic

[0104] The focus is placed on two spectral regions of interest within 1780 - 1550 cm'1and 1320 - 1012 cm'1comprised of vibrational modes that at times overlap and hinder the analysis when using existing, known methods. The system and methods described herein overcome this, allowing for separation and distinction of these overlapping vibrational modes. These vibrational modes are crucial to determining the structure, composition, and stability of the multiple components within the complex sample. At times, some spectral overlap may occur for the same functional group arising from different components one example are phospholipid headgroup from liposome or the LNP with the phosphodiester backbone from mRNA or DNA. The system and method described herein, using the combination of two-dimensional correlation algorithms i.e., 2T2D, 2D-COS, 2D-CDS the latter also known as co-distribution, allows for the separation and distinction of the overlapping spectral peaks and their assignment to specific vibrational modes originating from a specific component within the complex sample. Hence the method provides increased selectivity of the spectral analysis which allows for understanding of the structural and hydrogen bonding network changes that define the stability of the different components within the complex sample. Each spectral region provides a window of information for critical key signature peaks associated with each type of biochemical component. Examples of the analysis of multiple sample types such as liposomes are provided below.

[0105] Liposome Analysis

[0106] The system and method may analyze liposomes to perform their biophysical characterization. The system and method evaluate two separate spectral regions that are involved in hydrogen bonding with the aqueous environment: (1) the lipid carbonyl 1770 - 1700 cm'1and the (2) phosphate headgroup region 1290 - 1030 cm'1. In the evaluation performed, the signature peaks within these two separate spectral regions correlated with each other and provided internal confirmation of the results obtained.

[0107] Hyperspectral images of a sample containing liposomes may be acquired, such as the image shown in FIG. 1. These images may be acquired using a QCLM to explore the chemical distribution of components in the sample by analysis of certain wavenumbers: 1734 cm'1for the lipid carbonyl, 1674 cm'1cholesterol second ring double bond, 1466 cm'1methylene (cholesterol and lipid acyl chains) and 1238 and 1086 cm'1associated with the phosphate vibrational modes (lipid headgroup). In general, consistent with a homogeneous distribution of the liposomes within the sample well. The hyperspectral images shown in FIG. 1 may have 4.3 pm spatial resolution, and visual inspection shows homogeneous distribution of the liposomes within the sample well. No phase separation was observed in the hyperspectral images.

[0108] Typical QCLM spectral overlays and thermal dependence evaluations were generated to show quality of the spectral data, pre-transition temperature and main gel to liquid crystalline phase transition Tmof the liposome was observed to be Tm= 43 ± 0.5 °C using two different key signature peaks (see Table 1) for lipid carbonyl and the phosphate vibrational modes the results were confirmed. FIG. 2 provides an illustration of typical QCLM overlaid spectra of the phospholipid spectral region of 1770 - 1000 cm'1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C for liposomes. A polynomial baseline correction was applied to the full spectral region of interest. The vertical dash lines in FIG. 2 represent the actual spectral position for correction. Further, the peak assignments are for peaks in the spectra are shown at the top of FIG. 2. The vibrational modes identified from the peak assignments correlate with each other, and more importantly contained functional groups that formed a hydrogen bonding network with water and between neighboring lipids, as well as intramolecular hydrogen bonding would be possible depending on the phospholipid type. If residual bulk water is present, the overlap with water bending vibrational band influences the lipid carbonyl stretching band due to the overlap. An alternate approach to verify the main thermal transition temperature is to evaluate the phosphate vibrational modes. Also, the phosphate headgroup is oriented towards the aqueous environment due to the polar nature of the glycerol phosphodiester, providing an internal probe to indicate if the phospholipid is dehydrated. Specifically, the position of the asymmetric stretch (1270 cm'1) would be indicative of this state, therefore the liposomes were never dehydrated since the asymmetric stretch was never observed to have a maximum in this position (1260 cm' *)•

[0109] Two-dimensional correlation plots are generated and used to establish the correlation between the key signature peaks as valid. Cholesterol hydrogen bonds with the lipid carbonyl and therefore the phosphate vibrational modes may be used to establish Tm. the transition temperature, within the temperature range studied. The Co-distribution plots were similar suggesting a homogeneous population distribution for each of the three samples. Further evaluation of the 2D-COS plots from sample data confirmed via the correlation between the cross peaks observed the presence of the cholesterol and phospholipid component of the liposome. Table 1 below includes information regarding phospholipids and Table 2 below includes information regarding the liposome. Purity and chemical composition were consistent of a phospholipid-cholesterol liposome having a valid Tm. Finally, the hydrogen bonding interaction between the phospholipid headgroup and water occurs within the one-two molecule thickness layer of H2O that is in the immediate surrounding of the liposome. This approach can distinguish between the inner and outer membrane of the liposome. This was also confirmed by comparing the 2D-COS asynchronous cross peaks for all three liposome sample sizes.

[0110] Further examples of the above data analysis performed with the system and method described herein are provided below.

[0111] FIGs. 3A-C illustrate typical liposome QCLM overlaid covariance spectra and 2D-COS plots of the full spectral region of 1770 - 1000 cm'1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C. FIG. 3A illustrates the covariance spectra, FIG. 3B shows the synchronous plot, and FIG. 3C shows the asynchronous plot. Correlations amongst cross peaks are evident in FIG. 3C. The higher spectral resolution obtained for this type of plot allows for the biochemical evaluation of the sample and is used to define the peak assignments.

[0112] FIG. 4 illustrates typical QCLM overlaid covariance spectra of the phospholipid spectral region of 1770 - 1000 cm’1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C for liposome. The two spectral regions chosen associated with the lipid headgroup: (left box) lipid carbonyl (1770 - 1700 cm’1) and (right box) glycerol phosphate (1200 - 1000 cm’1) modes which are sensitive to the hydrogen bonding network and are sensitive to the gel-liquid crystalline phase transition.

[0113] FIGs. 5A-B illustrate a comparative analysis of liposomes using QCLM overlaid baseline corrected spectra (as shown in FIG. 5A) and covariance spectra (as shown in FIG. 5B) of the phosphate region of 1290 - 1020 cm’1and temperature perturbation range of 30 - 46 °C with intervals of 2 °C. A linear baseline correction was applied to the spectral region of interest. This lipidic headgroup spectral region is comprised of the phosphodiester and glycerol vibrational modes. Further, as illustrated in FIG. 5B, substantive intensity changes for both the positive and negative peaks are observed, allowing for both synchronous and asynchronous evaluations to be performed.

[0114] FIGs. 6A-D illustrate comparative 2T2D synchronous plots for a liposome (sample) against a (reference) liposome within the phosphate spectral region of 1290 - 1030 cm’1. These plots were generated as a function of temperature. FIG. 6A illustrates a comparative 2T2D synchronous plots for a liposome (sample) against a (reference) liposome within the phosphate spectral region of 1290 - 1030 cm’1at 30 °C. FIG. 6B illustrates a comparative 2T2D synchronous plots for a liposome (sample) against a (reference) liposome within the phosphate spectral region of 1290 - 1030 cm’1at 38 °C. FIG. 6C illustrates a comparative 2T2D synchronous plots for a liposome (sample) against a (reference) liposome within the phosphate spectral region of 1290 - 1030 cm'1at 42 °C. FIG. 6D illustrates a comparative 2T2D synchronous plots for a liposome (sample) against a (reference) liposome within the phosphate spectral region of 1290 - 1030 cm'1at 46 °C. All cross peaks are positive; therefore, the synchronous plot is comprised of correlated cross peaks exhibiting an increase intensity change. In FIG. 6A, highlighted cross peaks represented within a dashed circle are the cross peaks that exhibit the greatest intensity change suggesting increased thermal perturbation and the actual thermal transition temperature for the specific component of the headgroup. The perturbation of the headgroup continues to progress, as illustrated in FIGs. 6B-D, including a greater number of vibrational modes. Highlighted cross peaks within solid line circles represent the continuation of the thermal perturbation phase. However, in FIG. 6C an increase in thermal perturbation is observed within all the headgroup associated cross peaks thus suggesting the gel-to-liquid crystalline thermal transition phase. In FIG. 6D maximum thermal perturbation has been reached under the conditions evaluated.

[0115] FIGs. 7A-B illustrate 2D-COS comparative analysis of the phosphate region during the liposome’s gel to liquid crystalline thermal transition. The two-dimensional correlation analysis yielded the synchronous plot shown in FIG. 7A and the asynchronous plot shown in FIG. 7B. These plots are comprised of the liposome phosphate asymmetric and symmetric stretching and the glycerol Vas(C-O) stretching modes. Table 1 below provides a summary of the MID infrared peak assignments for phospholipids as main components of LNPS. Correlations between the cross peaks associated with the phosphate asymmetric and symmetric stretching vibrations were observed in the synchronous plot. For the asynchronous plot, the cross peak associated with the phosphate asymmetric stretch was observed to transition during the thermal perturbation suggesting this mode is impacted by hydrogen bonding interaction and therefore highly sensitive to the thermal stress.

[0116] Thus, the system and method provides a valid assessment of purity, integrity and size distribution of the liposome samples provided. In addition, the gel-to-liquid crystalline thermal transition may be confirmed to be valid for a phospholipid-cholesterol containing liposome.

[0117] Messenger RNA analysis The MID IR spectral region of 1445 - 1000 cm’1was determined to be a spectral region of interest that would provide insights towards the structural stability of mRNA within the temperature range studied. Covariance spectra were generated to monitor the changes that occurred during the thermal perturbation. In addition, a 2D-COS algorithm was applied to the covariance spectral data to enhance temporal and spectral resolution. In addition, the codistribution analysis provides understanding regarding the distribution population of mRNA’s in each sample and their relative stability. This approach allows for the identification of key signature peaks that were correlated with each other for an in-depth understanding of the molecular behavior of mRNA due to the presence of stabilizing excipients or the presence of wobbles within the nucleotide sequence. Furthermore, the 2T2D synchronous plot enables the identification and the selection of varying nucleotide mRNA sequences, to allow for the assessment of differentiation of the therapeutic mRNA as compared to the reference standard, to predict developability. These key signature peaks were associated with the nucleotide which are sensitive to the double stranded and single stranded state. For example, GC (I:C) syn conformer is sensitive to the pairing state of the nucleotides which is lost when disruption of the hydrogen bonding between the nucleotides occurs and the associated phosphate backbone vibrational modes is sensitive to the double helix: form A and B structures, resulting in a comprehensive understanding of the molecular dynamic changes of the mRNA and the intra— molecular hydrogen bonding interactions. In addition, the asymmetric phosphate stretching modes can serve as probes for also monitoring phosphate hydrolysis, which poses a risk for efficacy of the biotherapeutic. The risk of phosphate hydrolysis increases with the increased presence of events of base pairing mismatch and is directly related to the presence of wobbles. Thus, evaluation of this 1440 - 1178 cm’1spectral region is provides a for the assessment of developability of the engineered mRNA.

[0118] Further, an overlay spectra as a function of temperature for a thermal dependence evaluation within 1780 - 1000 cm’1of the key signature peaks may be generated to allow for the evaluation of the conformational and hydrogen bonding network changes that occur within a temperature range studied, such as 25 - 65 °C, to evaluate the effects of cations such as sodium, calcium, magnesium among others on the stability of mRNA. FIGs. 8 A-B provide an illustration of QCLM spectral overlays for an engineered double stranded mRNA commonly used as a reference standard. The entire 1800 - 1000 cm'1spectral region is amenable to analysis, as shown in FIG. 8A. FIG. 8B is comprised of the mRNA QCL peaks that serve to track the transition from double stranded to single stranded RNA within the spectral region of 1442 - 1178 cm'1which contains the stretching vibrational mode v(C-N) within 1398 - 1425 cm'1sensitive to the extent of hydrogen bonding of the nucleotides when the mRNA are in a double helix structure and the GC syn conformer (Hoogsteen base pair) at 1360 and 1320 cm'1(herein for (I:C). The peaks at 1376 and 1340 cm'1) also serve as an internal probe of the double stranded structure. Further information is provided by evaluating the mRNA phosphate backbone, specifically the asymmetric stretching vibrational mode Va.s(PO2 ) at 1245 - 1235 cm'1and Va^PCh ) observed at 1225 - 1220 cm'1associated with the A and B form of the double helix, respectively. All the peaks listed within this spectral region lose intensity as temperature increases, indicative of the loss of double stranded content within the mRNA.

[0119] FIG. 9 provides an illustration of a mRNA poly (I:C) 2D-COS synchronous plot within the spectral region 1442 - 1320 cm'1for the temperature range of 25 - 65 °C at 3.8 pg / pL. These synchronous correlations establish the different structurally sensitive vibrational modes for mRNA. These structurally sensitive vibrational modes can be analyzed to provide a comprehensive characterization of the molecular structure.

[0120] FIG. 10A is an illustration of a 2T2D synchronous plot within the spectral region 1780 - 1014 cm'1for mRNA poly (I:C) at 15.934 pg / pL in the presence of 0.9% NaCl at 65 °C against 25 °C (reference). FIG. 10B is an illustration of a histogram plot summarizing the 2T2D synchronous cross peaks and Table 5 for mRNA poly (I:C) at 15.934 ug / uL in the presence of 0.9% (w / v) NaCl at 65 °C against 25 °C (reference) within the spectral region of 1780 - 1014 cm1. The intensity differences for the synchronous cross peaks shown demonstrate relative stability observed for each mRNA double stranded form. FIG. 10C is an illustration of a 2T2D synchronous plot within the spectral region 1780 - 1014 cm'1for mRNA poly (I:C) at 23.427 pg / pL in the presence of 0.9% (w / v) NaCl at 65 °C against 25 °C (reference). FIG. 10D is an illustration of the histogram plot summarizing the 2T2D synchronous cross peaks and Table 6 for mRNA poly (I:C) at 23.427 pg / pL in the presence of 0.9% (w / v) NaCl at 65 °C against 25 °C (reference) within the spectral region of 1780 - 1014 cm'1. The cross peaks shown demonstrate the strong synchronous correlations observed for mRNA, which can be a distinctive feature when evaluating the more complex LNP encapsulated mRNA. The intensity changes observed provide direct insight to the extent to which the mRNA of interest has unfolded from the double stranded to the single stranded mRNA. Furthermore, the effect of excipient can also be studied using 2T2D by evaluating spectral data in the presence and absence of the excipient at the same temperature. Tables 5 and 6 below provide summaries of the key cross peaks and their assignments. In particular, Table 5 provides a summary for mRNA poly (I:C) at 15.934 pg / pL 2T2D key synchronous cross peak coordinates, weighted intensity differences between 65 °C against 25 °C, and their assignments. And, Table 6 provides a summary for mRNA poly (I:C) at 23.427 pg / pL 2T2D key synchronous cross peak coordinates, weighted intensity differences between 65 °C against 25 °C, and their assignments. Although the data presented is within the 10 of pg / pL the method is highly sensitive to less than 0.5 pg / pL.

[0121] Briefly, the effect on the pre-Zm’s may be observed in a comparative manner for the sample in the presence and absence of cations. The impact of sodium on the structure and hydrogen bonding network of the mRNA backbone may be evident by studying phosphate asymmetric stretch peak at 1264 cm'1. Dehydration of the phosphate can be observed by a peak position shift to > 1270 cm'1evident at low temperatures (< 37 °C). This event would explain the generation of the crystals observed in the HS images for the mRNA in the presence of Na. This dehydration may be explained by changes in the hydrogen bonding network involving the phosphate backbone with its aqueous environment. An event typically observed during the crystallization. This dehydration event is not observed for mRNAs in the absence of sodium.

[0122] FIG. 11 A is an illustration of a co-distribution synchronous plot for distribution population of mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of: 1442 - 1324 cm'1. FIG. 11B is an illustration of a histogram plot summarizing the co-distribution synchronous cross peaks for mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of 1442 - 1320 cm'1. All cross peaks are positive. FIG. 11C is an illustration of a co-distribution asynchronous plot cross peaks distribution population mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of 1442 - 1320 cm'1. FIG. 1 ID is an illustration of a co-distribution histogram plot summarizing the asynchronous cross peaks for mRNA poly (I:C) at 23.427 pg / pL within the temperature range of 25-65 °C in the spectral region of 1442 - 1320 cm'1. FIG. 11A, the synchronous plot, and FIG. 1 IB, the histogram graph associated with the synchronous plot, shows the intensity changes of the correlated structurally sensitive vibrational modes observed for the distribution population of mRNA molecules within the sample. FIG. 11C is the asynchronous plot, and FIG. 1 ID is a histogram graph associated with the asynchronous plot which summarizes the cross peak intensity changes observed in the asynchronous plot. Table 7 summarizes the synchronous cross peak intensity changes and their assignments. For the synchronous co-distribution plot, shown in FIG. 11 A, different correlations can be observed such as: (1) those associated with Inosine H-bonded states (2) the I:C syn conformer state associated with the base pairing and (3) the highly structurally (double stranded to single stranded) sensitive v(C-N) correlation. For the asynchronous plot shown in FIG. 11C, stability differences are observed for Inosine, which can be linked directly to the extent of hydrogen bonding exhibited by inosine within the temperature range studied. The negative cross peaks change prior to the positive cross peaks, as summarized in the histogram of FIG. 1 ID. A greater stability of inosines is observed for the 2-hydrogen bonded state and terminal end site. Table 8 summarizes the asynchronous cross peak intensity changes observed for the distribution population of mRNA molecules within the sample.

[0123] Like in the case presented for inosine (an unnatural nucleoside) shown for the commercially available double stranded mRNA; any synthetic nucleoside may be evaluated within the sequence of an engineered mRNA therapeutic. Detection / identification of the unnatural nucleoside would involve the evaluation of functional groups that can be used for their vibrational mode assignments and cross-peak assignments through their correlations established using 2-dimensional correlation algorithms. The synthetic nucleoside may be evaluated as a pure component, within a 16 mer (short sequence) to limit complexity and yet mimic the interactions that would be typically observed in the high molecular weight (HMW) engineered mRNA sample. The approach would also inform the effects on the structure and stability of the engineered mRNA that includes synthetic nucleosides within its sequence using this comparability approach adds value to risk mitigation.

[0124] FIG. 12A illustrates a 2T2D synchronous plot for mRNA (I:C) at 46 °C against 30 °C in the spectral region of 1442 - 1324 cm’1. This provides the minimal key signature peaks associated with the evaluations of the sequence structure relationship. FIG. 12Bis a histogram graph providing the weighted intensity differences for mRNA (I:C) when comparing the spectra at different temperatures. Further, Table 9 below summarizes the weighted intensity differences for the key signature cross peaks and their assignments.

[0125] LNP encapsulated mRNA

[0126] The system and method may further provide characterization of both mRNA and LNP samples. The use of 2T2D and co-distribution analysis of segmented spectral data representing pre- and post Tm for each component within the system allows for the augmented differentiation due to the molecular dynamic differences between the components. This strategy effectively allows for both the spectral and temporal resolution to impact the differentiation of the biochemical components within a complex system. The different temperature ranges chosen are based on the thermal transition temperatures for each pure component, which allows the system and method to distinguish each pure component within the complex system. This effectively deconvolutes the spectral data for the encapsulated LNP mRNA against the pure mRNA spectra under the same temperature conditions. This analysis is provides for generating neural network models that allow for deep machine learning to predict the outcome of a newly designed biotherapeutic. Results like those presented for the simulated data below with associated tables have been generated for a liposome and mRNA sample to demonstrate feasibility. A summary of the key signature cross peaks and their assignments are used to identify the contributions for each component within the mixture / system. Consequently, the method described herein may make unequivocal peaks assignments to the mRNA which are distinct from those assigned to the LNP directly, without the use of exogenous labels.

[0127] Simulated spectra of a mixture commonly implemented as a second phase of the model building used for neural networks were generated and evaluated.

[0128] FIGs. 13 A,B illustrate overlaid spectra in the spectral region of 1008 - 1746 cm'1for a simulated liposome mRNA poly (I:C) mixture, an experimental liposome, and an experimental mRNA poly (LC) to demonstrate the capability of the 2D correlation algorithms to deconvolve the spectral contributions associated with different biochemical components for neural network model building and machine learning. FIG. 13 A is an illustration of overlaid spectra in the spectral region of 1008 - 1746 cm'1at 30 °C for: simulated liposome mRNA poly (I:C) mixture (represented by the thick solid line), experimental liposome (represented by the dashed — line), and experimental mRNA poly (EC) (represented by the thin solid line). FIG. 13B is an illustration of overlaid spectra in the spectral region of 1008 - 1746 cm'1at 46 °C for: simulated liposome mRNA poly (EC) mixture (represented by the thick solid line), experimental liposome (represented by the dashed — line), and experimental mRNA poly (EC) (represented by the thin solid line). Further, FIG. 13 C illustrates simulated overlaid spectra for a liposome mRNA poly (EC) mixture in the spectral region of 1008 - 1746 cm'1within a temperature range 30 - 46 °C to demonstrate the capability of the two-dimensional correlation algorithms to deconvolve complex biochemical systems based on the correlation of the intensity changes and their different thermal transition temperatures.

[0129] FIG. 14 A illustrates a co-distribution asynchronous plot for a simulated liposome and mRNA poly (EC) mixture within the spectral region 1436-1324 cm'1and within the temperature range of 30 - 46 °C. The cross peak contributions from both the simulated liposome and mRNA poly (EC) are present. FIG. 14B is a histogram graph of the co-distribution asynchronous cross peak intensity changes where the cross peak contributions for both components are present. The positive cross peaks are perturbed first and are all associated with the liposomes undergoing a gel-to-liquid crystalline phase transition (acyl chains undergoing gauche rotamer transitions), while the negative peaks are perturbed later and are all associated with the mRNA (EC) which has not undergone its thermal transition from double stranded to single stranded. Each biochemical component may be monitored within the spectra demonstrating the value of a label free analysis of biochemically engineered complex systems. Table 10 below proves a summary of the distribution population of simulated liposome + mRNA poly (EC) at 23.427 pg / pL codistribution asynchronous cross peak coordinates, assignments and intensity changes in the spectral region of 1442 - 1324 cm'1and temperature range of 30 - 46 °C. FIG. 15A is a codistribution plot for experimentally derived pure mRNA (EC) in the spectral region 1436-1324 cm'1within the temperature range of 30 - 46 °C (pre-Z™). Because the temperature range is below the Tm for mRNA (EC), several cross peaks are not present. FIG. 15B. is an illustration of histogram plot of the co-distribution asynchronous cross peak intensity changes for contributions are for the pure mRNA (EC) within the temperature range of 30 - 46 °C in the spectral region of 1436-1324 cm'1. The results are used to allow for comparative evaluation of the simulated spectral data and the experimental mRNA (EC) spectral data. The absence of correlated cross peaks associated exclusively with the liposome associated acyl chain vibrational modes is evident, thus proving the usefulness of the analysis above. Table 11 provides a summary of the distribution population of experimental mRNA poly (I:C) at 23.427 pg / pL co-distribution asynchronous cross peak coordinates, intensity changes and assignments in the spectral region of 1442 - 1324 cm'1and temperature range of 30 - 46 °C.

[0130] The approach may further include lipid nanoparticle in the system / sample to be analyzed, as the bases for the spectral deconvolution relies on the differences in correlation of the vibrational modes and stability of the biochemical components, which will lead to the identification and analysis of each component within the system. Thus, this approach is beneficial for drug development to ensure optimization of drug and carrier design to ensure safety and efficacy of new biotherapeutics.

[0131] A schematic representation of a workflow for the system and method is shown in FIG. 16. As shown in FIG. 16, the system and method first acquires a series of hyperspectral images of the slide cell array. The slide cell array includes wells containing the sample being analyzed, and one or more reference samples or controls. Spectral data is obtained from the hyperspectral images and exported for analysis. The analysis includes the application of three algorithms to the data, as well as an evaluation of thermal dependence. In particular, a 2T2D algorithm is applied, and 2T2D synchronous and asynchronous plots and weighted difference spectra are generated. Signature peaks in the asynchronous plot along with the weighted difference spectra are analyzed to determine the components in each sample well. A 2D-COS algorithm is also applied, and 2D- COS synchronous and asynchronous plots are generated. Cross peaks in the 2D-COS plots are evaluated and compared to define assignments for each peak. The identification of the key signature peaks for the comparative analysis are based on the correlations observed using the 2D- COS plots for each sample type (i.e., ghost LNP, mRNA and the encapsulated mRNA samples). The 2D-COS analysis also allows for the confirmation for the peak assignments. Tables 1-4, included below, summarize the peak assignments for each biochemical component of interest. Further, the detailed molecular dynamic fingerprint generated during thermal perturbation may be evaluated for the defined 2D-COS cross peaks through their assignments.

[0132] A 2D-CDS algorithm is also applied, and 2D-CDS synchronous and asynchronous plots are generated. Cross peaks in the co-distribution plots are evaluated and compared, which provide indications of the relative stability of the components in the samples and can be used to evaluate the complex nature of the distribution population of therapeutic entity. The therapeutic entity may comprise, for example, the LNP encapsulated mRNA, pure (or ghost, empty) LNP, or pure (or naked) mRNA. The cross peaks and their intensity changes inform the molecular stability of most of the molecules within the sample. In a comparability assessment this evaluation extends to the identification of the optimal conditions for encapsulation of the mRNA, determining the optimal sequence to achieve the highest content of double stranded RNA or DNA in the structural form desired or simply optimizing the excipient to ensure shelf-life. These new modalities are subject to degradation and are observed to have differences in stability due to their formulation and storage conditions. Similarly, the 2T2D algorithm is ideal for the direct comparison of two spectra. Thus, the system and method provides a comprehensive analysis that is both qualitative and quantitative. That is, the system and method allow for both identifying the components in the sample and checking the stability of each sample. The 2T2D evaluation allows for confirmation that the components have been correctly distinguished. Structural changes of each component of the encapsulated product can be determined, giving an evaluation of their stability. If one component is determined to be unstable, the system and method allow for its identification such that the encapsulated product may be redesigned.

[0133] The system and method described above achieve the following:

[0134] • Qualitative and quantitative evaluation hyperspectral images of the mRNA Na samples.

[0135] • The capability to determine thermal dependence spectral data of all samples.

[0136] • High selectivity and sensitivity of the analysis methods.

[0137] • For quantitative analysis, the ability to focus on spectral regions of interest, such as the spectral region of 1178-1018 cm'1where many sugar and phosphate signature peaks are located.

[0138] • Quantitative analysis of mRNA structurally related peaks to ascertain the extent of double stranded content in a given sample.

[0139] • Detailed biochemical information assessed for all samples.

[0140] • Comparability between same sample types. • The capability to determine pre-transition temperatures for samples.

[0141] • The capability to rank order each type of sample based on thermal stability.

[0142] • The capability to establish the presence of mRNA encapsulated in an LNP.

[0143] • The capability to provide evaluation using a third correlation algorithm for both positive and negative controls.

[0144] Table 1. Summary of the MD infrared peak assignments for phospholipids as main components of LNP's.

[0145] Assignment Vibrational mode Peak position Comment Table 2. Summary of the MID Infrared peak assignments for cholesterol

[0146] Assignment Vibrational mode Peak position Comment alkene stretching v(C=C) 1S67 second ring of the choiesteroi

[0147] Methylene scissoring mode S.(CH2j 1464

[0148] Methyl bending 8(CH3) 1373 ring deformation 1054

[0149] Table 3. Summary of peak assignments for pegylated lipid nano particle (PEG-LN?)

[0150] Assignment Vibrational mode Peak position Comment

[0151] Table 4. Summary of peak assignments for mRNA in the presence and absence of Na*

[0152] Assignment Vibrational mode Peak position Comment*

[0153] Table 5. Summary for mRNA poly (kC) at 15.934 pg / ul 2T2.D key synchronous cross peak coordinates, weighted intensity differences between 65 °C against 25 °C and their assignments.

[0154] Weighted

[0155] Cross peak intensity Cross peak

[0156] Coordinates difference Assignment

[0157] Table 6. Summary for mRNA poly (l:C) at 23.427 pg / uL 2T2D key synchronous cross peak coordinates, weighted intensity differences between 65 °C against 25 °C and their assignments.

[0158] Weighted

[0159] Cross peak Intensity cross peak

[0160] Coordinates difference assignment

[0161] (cm1, cm1) (A.Uj Table 7. Summary of the distribution population of mRNA poly (I:C) at 15.934 pg / pL codistribution synchronous cross peak coordinates, assignments and intensity changes for the spectral region of 1442 - 1324 cm'1and temperature range of 25 - 65 °C.

[0162] Cross peak Cross peak Intensity

[0163] Coordinates Assignments change

[0164] (cm1, cmX) (A.U.)

[0165] (1396. 1332) v(C-N), 1 syn conformer 9.55E-09

[0166] (1400. 1332) v(C-N), 1 syn conformer 1.17E-08

[0167] (1408. 1332) v(C-N), 1 syn conformer 1.43E-08

[0168] (1415. 1332) v(C-N), 1 syn conformer 1.90E-08

[0169] (1422. 1332) v(C-N), 1 syn conformer 1.30E-08

[0170] (1374. 1332) l:C syn conformer 1.19E-08

[0171] (1358. 1332) 1 non-H bond, 1 2-H bond state 4.23E-09

[0172] (1350. 1332) 1 1-H bond, 1 2-H bond state 5.32E-09

[0173] (1415. 1350) v(C-N) body, 1 syn conformer 1.11E-08

[0174] (1408. 1350) v(C-N), 1 syn conformer 8.31E-09

[0175] (1400. 1350) v(C-N), 1 syn conformer 6.73E-09

[0176] (1396. 1350) v(C-N), 1 syn conformer 4.85E-09

[0177] (1374. 1350) l:C syn conformer 6.91E-09

[0178] (1358. 1350) 1 non-H bond, 1 1-H bond state 2.16E-09

[0179] (1374. 1358) l:C syn conformer 5.48E-09

[0180] (1396. 1358) v(C-N), 1 syn conformer 3.90E-09

[0181] (1400. 1358) v(C-N), 1 syn conformer 5.39E-09

[0182] (1408. 1358) v(C-N), 1 syn conformer 6.60E-09

[0183] (1415. 1358) v(C-N) body, 1 syn conformer 8.77E-09

[0184] (1415. 1374) v(C-N), C syn conformer 2.38E-08

[0185] (1408. 1374) v(C-N), C syn conformer 1.84E-08

[0186] (1400. 1374) v(C-N), C syn conformer 1.44E-08

[0187] (1396. 1374) v(C-N), C syn conformer 1.09E-08

[0188] (1415, 1396) v(C-N), v(C-N) 1.78E-08

[0189] (1415, 1400) v(C-N), v(C-N) 2.30E-08

[0190] (1415, 1408) v(C-N), v(C-N) 2.94E-08

[0191] (1422, 1415) v(C-N), v(C-N) 2.67E-08 Table 8. Summary of the distribution population of mRNA poly (I:C) at 15.934 pg / pL codistribution asynchronous cross peak coordinates, assignments and intensity changes for the spectral region of 1442 - 1324 cm'1and temperature range of 25 - 65 °C. Table 9. Summary for mRNA poly (i:C) at 15.934 pg / pL 2T2D synchronous cross peak coordinates, weighted intensity differences between 46 °C against 30 °C and their assignments.

[0192] Weighted

[0193] Cross peak intensity Cross peak

[0194] Coordinates Difference Assignments

[0195] (cm1, cm1) (A.U.)

[0196] (1415, 1339) 6.46E-04 v(C-N), 1 syn conformer

[0197] (1374, 1339) 6.33E-04 l:C syn conformer

[0198] (1415, 1374} 2.72E-03 v(C-N), C syn conformer

[0199] (1396, 1350) v(C-N), 1 syn conformer

[0200] Table 10. Summary of the distribution population of simulated liposome +■ mRNA poly (UC) at 23.427 ug / uL co-distribution asynchronous cross peak coordinates, assignments and intensity changes in the spectral region of 1442 - 1324 cm ’ and temperature range of 30 - 48 °C.

[0201] Cross peak Intensity Cross peak

[0202] Coordinates change Assignments

[0203] S / midadon mRA / 4 po / y(W) + bposome thermal dependence

[0204] Table 11. Summary of the distribution population of experimental mRNA poly (I:C) at 23.427 pg / pL co-distribution asynchronous cross peak coordinates, intensity changes and assignments in the spectral region of 1442 - 1324 cm'1and temperature range of 30 - 46 °C.

[0205] Cross peak Intensity Cross peak

[0206] Coordinates change Assignments

[0207] ( cm'1, cm1) (A.U.)

[0208] (1415. 1352) 3.83E-12 1 syn conformer, v(C-N)

[0209] (1403. 1352) 5.02E-12 1 syn conformer, v(C-N)

[0210] (1374. 1352) 4.07E-12 EC syn conformer

Claims

Claims1. A method for identifying the biochemical composition of a sample, comprising: a) providing the sample in a slide containing at least one sample well and at least one reference well; b) acquiring at least one spectral image of the sample and at least one spectral image of the at least one reference using a quantum cascade laser microscope under controlled temperature conditions; c) identifying and selecting, in at least one of the acquired first spectral images, a region of interest; d) obtaining spectral data for the sample and at least one reference for the region of interest; e) applying a baseline correction to the spectral data for the sample and the reference for the region of interest; f) applying a two-trace two-dimensional correlation to the baseline corrected sample and reference spectral data to generate a synchronous spectrum(t>(v1, v2) and asynchronous spectrum 'P(v1, v2), and identifying components of the sample based on wavenumbers of signature peaks in the plots; g) applying a two-dimensional correlation analysis to the baseline corrected data and reference spectral data to generate a synchronous correlation plot and an asynchronous correlation plot, and identifying components of the sample based on wavenumbers of signature peaks in the plots; and h) applying a two-dimensional co-distribution analysis to the baseline corrected data to generate a synchronous correlation plot and an asynchronous co-distribution plot, and evaluating cross peaks within the co-distribution plots to determine stability of the identified components in the sample.

2. The method of claim 1, wherein identification of the components includes an identification of a therapeutic, a carrier, and an encapsulating component.

3. The method of claim 2, further comprising evaluating signature peaks for each component as a function of temperature, and performing a thermal dependence evaluation to determine phase transition temperatures and pre-transition temperatures of the therapeutic, carrier, and encapsulating component.

4. The method of claim 3, further comprising assessing the structure, composition, and stability of therapeutic, carrier and encapsulating component.

5. The method of claim 1, further comprising deconvolving spectral data for each component and identifying elements and variables to build a neural network model.

6. The method of claim 5, further comprising applying the neural network model to a machine learning algorithm to predict outcomes for designed therapeutics.

7. The method of claim 1, wherein the sample includes synthetic or unnatural nucleotides.

8. The method of claim 1, wherein the sample includes synthetic or unnatural lipids.

Images