Systematic comparison of wide pore size exclusion chromatography columns for the characterization of gene therapy products
Journal of Chromatography A, 1752, 2025, 465972: Fig. 2. Analysis of rAAV2 sample under standard SEC conditions using six different widepore SEC columns. Mobile phase was composed of 200 mM phosphate buffer at pH 6.2 supplemented with 250 mM KCl. Flow rate was 0.3 mL/min and injected volume was 1 µL.
A systematic comparison of wide-pore size exclusion chromatography (SEC) columns was performed for the characterization of gene therapy products, including recombinant adeno-associated viruses (rAAVs) and messenger RNAs (mRNAs). Among six SEC columns (450–700 Å) tested for rAAVs, the DNACore AAV-SEC column achieved the highest efficiency (11 000 plates), while larger-pore columns (550–700 Å) provided optimal selectivity. No single column delivered the best resolution across all serotypes, highlighting sample-dependent performance.
For mRNA analysis (700–1000 Å), the Biozen dSEC-7 LC column offered superior efficiency, especially for smaller mRNA (~1000 nt), whereas larger-pore columns like AdvanceBio SEC 1000A were better suited for longer mRNAs. Despite quantitative comparability, all columns showed limitations in resolving low- and high-molecular-weight mRNA species. These results guide column selection for gene therapy product characterization under various analytical conditions.
The original article
Systematic comparison of wide pore size exclusion chromatography columns for the characterization of gene therapy products
Mathias Buff, Alexandre Goyon, Carsten Elger, Raphael Ruppert, Markus Haindl, Kelly Zhang, Davy Guillarme
Journal of Chromatography A, 1752, 2025, 465972
https://doi.org/10.1016/j.chroma.2025.465972
licensed under CC-BY 4.0
Selected sections from the article follow. Formats and hyperlinks were adapted from the original.
Since the human genome project completion in early 2000′s, the drug discovery landscape has progressively shifted towards targeting complex biological mechanisms, previously considered as “undruggable” [1]. This has led to the development of novel classes of therapeutic modalities beyond traditional small molecules. Over the past two decades, an increasing interest for new drug modalities has been observed, first with a strong emphasis on monoclonal antibodies (mAbs) and related compounds in the 2010s, and more recently (2020s) on cell and gene therapies (CGTs). Gene therapy is defined as the repair, replacement, or deactivation of dysfunctional genes in a patient body to restore normal function [2]. Unparalleled progress and hope have been generated by CGTs, due to their potential to cure inherited diseases such as melanoma, pancreatic cancer, retinal dystrophy, spinal muscular atrophy, hemophilia B, among others. Several gene therapy products have already received regulatory approval for commercial use [2]. CGT approaches commonly involve two critical components: the drug modality and the delivery system (drug carrier).
Messenger RNAs (mRNAs) can be used to encode the protein sequence of gene editing enzymes such as Cas9. The versatility of mRNA makes it well adapted to quickly respond to urgent therapeutic needs. Additionally, platform manufacturing technologies enable the large scale production required for their global distribution. Viral vectors and lipid nanoparticles (LNPs) are the most widely used carriers to deliver the gene editing materials. Viruses possess the necessary properties for efficient cellular entry and dissemination of their genetic payloads [3]. Therefore, they can be repurposed by replacing their original genetic material with therapeutic payload. However, due to their high production cost and complex manufacturing process [4], LNPs have emerged as an alternative solution [5]. LNPs facilitates drug delivery to the body by fusing with cell membranes [6,7], ensuring efficient intracellular transport.
Thorough analytical characterization of complex gene therapies is required by health agencies prior to their administration to patients to ensure product quality, safety and efficacy. Indeed, these therapies can undergo various changes during their preparation, formulation and storage, necessitating the evaluation of various critical quality attributes (CQAs) through the use of a wide range of analytical techniques [8]. This study focuses on the characterization of mRNAs, which is a widely used as gene editing material or vaccines, and recombinant adeno-associated viruses (rAAVs), which is a prevalent gene therapy vector. Among the numerous CQAs that may need to be assessed, size variants (fragments and aggregates) are particularly critical, as they can significantly impact therapeutic efficacy of these two products [[9], [10], [11], [12]].
A variety of size-based separation techniques are currently available, among which, size exclusion chromatography (SEC) appears as one of the most straightforward technique and that can be easily implemented in quality control (QC) environment. SEC is well accepted by the scientific community as it offers excellent quantitative performance, ease of use, and broad regulatory acceptance. As a non-denaturing chromatographic technique, SEC separates molecules based on their hydrodynamic radius [13,14], ideally in absence of chemical interactions. To fully exploit the potential of SEC, selecting a suitable pore size for the column packing is crucial, and should be based on the molecular size of the analytes. Over the past decade, packing materials with pore sizes around 300 Å have been extensively used to characterize mAbs and related biomolecules [15,16]. More recently, columns with larger pore sizes (around 500 Å) have been introduced for the analysis of rAAV products [[17], [18], [19]]. Finally, ultra-wide pore SEC columns (around 1000 Å), launched in the past year have demonstrated good applicability for mRNA analysis [[20], [21], [22], [23]].
In response to the increasing interest for CGT products in the pharmaceutical industry, several chromatography providers have recently commercialized SEC columns with pore sizes ranging from 450 to 1000 Å, and particle sizes of 3 µm or smaller. The aim of this study was to evaluate the performance of 10 newly available SEC columns from both qualitative and quantitative perspectives. The evaluation was performed using various rAAV serotypes and mRNA sequences and lengths.
2. Materials and methods
2.2. Instrumentation and columns
Measurements were performed on an ACQUITYTM UPLCTM H-Class Bio System (Waters, Milford, Massachusetts, USA) equipped with a quaternary solvent delivery pump, an autosampler including a 15 µL flow-through needle (FTN) injector (rinsing solvent was ultrapure water). Since columns of 30 cm cannot fit in the standard oven of the H-class instrument, an external Temperature Control System (Waters) was integrated into the chromatography system, to heat the column to 50 °C. For mRNA products, an ultraviolet (UV) detector possessing a 5 mm path length titanium cell of 1.5 μL volume was employed. Data were acquired at 260 nm for the mRNA ladder and samples, and 280 nm for the protein mix at a 10 Hz sampling rate and 0.2 s response time. For rAAV products, a fluorescence (FL) detector possessing a 2 μL flow-cell was employed. Data were acquired using 280 nm excitation and 350 nm emission wavelengths at a 10 Hz sampling rate and 0.2 s response time. Instrument control, data acquisition, and data processing were all performed using EmpowerTM Pro 3 software.
Separations were performed on various SEC columns with varying pore sizes, as detailed in Table 1. All columns were either purchased or donated by the suppliers; and were used for the first time at the start of the study.
3. Results and discussion
3.2. Evaluation of various wide pore SEC columns for rAAV products
3.2.1. Evaluation of the qualitative performance
Six different SEC columns were evaluated for the analysis of four rAAV serotypes, namely rAAV2, rAAV5, rAAV8 and rAAV9. The different rAAV serotypes are similar in size (20–25 nm in size, with a molecular mass of about 5 MDa). In a first instance, USP plate counts were measured at a flow rate of 0.3 mL/min, corresponding to an analysis time of about 10 min. The average plate counts from three independent measurements are summarized in Fig. 1. Fig. 2, Fig. 3 show the corresponding chromatograms obtained for rAAV2 and rAAV5 samples, respectively. As shown in Fig. 1, the plate number for a given column was comparable on rAAV2, rAAV8 and rAAV9, but the efficiency values were in average reduced by about 20–30 % (except for the SRT SEC-500 column) and even up to 40 % on the GTxResolve Premier BEH SEC and Biozen dSEC-7 LC columns for the rAAV5 product. This reduction in efficiency cannot be attributed to differences in the chemical composition or size of rAAV5, but may instead result from the presence of an additional species co-eluting with the monomer. The hypothesized sample heterogeneity likely causes peak broadening resulting in a lower plate count. Interestingly, the number of plates obtained for rAAV separations varied significantly between the columns, ranging from about 900 on the SRT SEC-500 column (historically considered as a reference for rAAV characterization), to over 11,000 on the DNACore AAV-SEC column. The AdvanceBio SEC 500A, Biozen dSEC-7 LC and SurePac Bio 550 SEC Mdi columns achieved comparable efficiencies of approximately 7000 plates, while the GTxResolve Premier BEH SEC column achieved an average of 4500 plates. The significantly lower efficiency observed for the SRT SEC-500 column can be attributed to its particle size of 5 µm in comparison to 2.5 - 3 µm for the other columns. The plate count is inversely proportional to the particle size, which explains the reduced efficiency of this column. In addition, the optimal flow rate varies with particle size, and the optimal linear velocity is inversely proportional to particle size. The diffusion coefficients of the rAAVs are very small due to their large size (20–25 nm), making the optimal linear velocity quite low. The flow rate employed in this study (0.3 mL/min, corresponding to a linear velocity of about 0.4 mm/s) is well above the optimum, and this is even more true for columns packed with 5 µm particles. In the end, the low efficiency of the SRT SEC-500 column can be attributed to both its larger particle size and the excessively high linear velocity employed in this work. In contrast, the high efficiency observed for the DNACore AAV-SEC column can be attributed to the use of monodispersed silica particles and a specialized packing procedure, as stated by the manufacturer. Interestingly, the AdvanceBio SEC 500A and GTxResolve Premier BEH SEC columns packed with the smaller particle size (2.5–2.7 µm) did not provide the highest efficiency. This suggests that the average particle size alone is not the only factor influencing efficiency, and particle size distribution plays a crucial role as well. To verify our hypothesis and rule out any potential issues with the specific column batch used in this study, we reviewed the application notes provided by the column providers [[25], [26], [27], [28], [29], [30], [31]] and estimated efficiency values based on peak widths. The results reported in the application notes are in line with our findings and validate values reported in Fig. 1.
Journal of Chromatography A, 1752, 2025, 465972: Fig. 1. Average efficiency values obtained for the monomer peak of four different rAAV serotypes on six different SEC columns with wide pores (450 - 700 Å). Error bars correspond to the RSD values calculated from three independent measurements. Mobile phase was composed of 200 mM phosphate buffer at pH 6.2 supplemented with 250 mM KCl. Flow rate was 0.3 mL/min and injected volume was 1 µL.
Using a single descriptor, such as efficiency, is clearly not sufficient to assess the overall quality of a SEC column for a specific application. Recently, Fekete et al. introduced a more comprehensive metric to properly describe SEC chromatogram quality, the dimensionless separation quality factor (Qs) [32]. This metric incorporates five SEC specific parameters: peak-to-valley ratio, elution window width, peak widths, peak-positioning and recovery. This approach was first applied to the rAAV2 chromatograms reported in Fig. 2. Four of the parameters defined by Fekete et al. (excluding recovery) and the individual values for each factor are provided in Table S1 of the supplementary material. These values, which are dimensionless and range from 0 to 1, allow to better understand which properties are advantageous or not for a given column, with higher values indicating better performance. As summarized in Table S1, the separation quality between the monomer and dimer (Q1 term) is not a highly discriminative quality factor, as most columns performed exceptionally well, with Q1 values of 0.99. The only exception was the SRT SEC-500 column, which has a slightly lower Q1 value of 0.92. This reduction is clearly attributed to the lower efficiency generated by this column, resulting in lower resolution between the monomer and dimer, as shown in Fig. 2. Similarly, the kinetic efficiency expressed as Q3 term, was lower for the SRT SEC-500 column (0.75) compared to the other columns, which all possess values of 0.84–0.85. While this result is expected for the SRT SEC-500 column, it was more surprising for the other columns, considering the significant differences in efficiency observed among them. In the calculation method employed for Q3, a normalized efficiency was considered instead of an absolute metric (peak width or efficiency). This normalized efficiency was calculated based on the peak width and the difference between the column dead time and the total exclusion time, rather than the “classical” elution time, so a large peak widths can be compensated by a large SEC elution window. The elution window utilisation (Q2 term), was also comparable across all SEC columns, with values consistently low, ranging from 0.16 to 0.19. These results indicate that the eluted species exhibit a relatively narrow size range, compared to the full separation potential of the columns. Only the SRT SEC-500 column demonstrated again slightly lower performance, with a Q2 term of 0.14. Among the evaluated parameters, the most discriminative was the pore size match (Q4 term), as the separation in SEC is strongly influenced by the column pore size. Optimal selectivity in SEC is typically achieved when analytes elute near the centre of the elution window, corresponding to an equilibrium constant, KSEC of 0.5 [13,33]. As shown in Table S1, significant variations were observed in Q4 values among the columns, ranging from 0.33 to 0.87. The SRT SEC-500 column had the lowest Q4 value of 0.53, intermediate values were obtained for the GTxResolve Premier BEH SEC and DNACore AAV-SEC columns (Q4 = 0.63–0.66), and the highest values were obtained for Biozen dSEC-7 LC, AdvanceBio SEC 500A and SurePac Bio 550 SEC Mdi columns (Q4 = 0.71–0.72). These differences highlight the strong correlation between pore size and Q4 term. The SRT SEC-500 and GTxResolve Premier BEH SEC columns, with pore sizes of 450 - 500 Å appear inadequate for the rAAV samples analysed in this study. In contrast, the larger pore size of the SurePac Bio 550 SEC Mdi and Biozen dSEC-7 LC columns 550 - 700 Å explain their higher Q4 values. Interestingly, the AdvanceBio SEC 500A column was an outlier, achieving one of the best Q4 values, despite its average pore size of only 500 Å. This discrepancy might be due to a wide pore size distribution or an inaccurate pore size specification mentioned by the manufacturer. In the end, when combining all the terms into Qs, representing the overall quality of column for the rAAV2 sample, the ranking predominantly reflects pore size match (Q4 term), as it was the most discriminant descriptor. The Biozen dSEC-7 LC, AdvanceBio SEC 500A and SurePac Bio 550 SEC Mdi columns achieved the best separation performance. However, when visually inspecting the chromatograms (Fig. 2), only the SRT SEC-500 behaved differently from the other five columns which provided comparable results. The AAV5 sample highlighted the importance of the sample serotype on the column performance and ultimately their ranking (Fig. 3). In particular, three columns (DNACore AAV-SEC, AdvanceBio SEC 500A and SRT SEC-500) were able to separate an additional high molecular weight species (HMWS), demonstrating a better performance for this particular sample. In contrast, this additional impurity peak could not be observed with the Biozen dSEC-7 LC and SurePac Bio 550 SEC Mdi columns and only a minimal amount of the HMWS was observed with the GTxResolve Premier BEH SEC column. These findings indicate that the optimal column choice can depends on the sample being analysed. Even the SRT SEC-500 column, which has the lowest plate number and Qs value, can provide interesting results on specific samples, such as rAAV5. The results highlight the importance of screening multiple SEC columns during method development to ensure optimal separation performance and maximal coverage of rAAV size variants.
Journal of Chromatography A, 1752, 2025, 465972: Fig. 2. Analysis of rAAV2 sample under standard SEC conditions using six different widepore SEC columns. Mobile phase was composed of 200 mM phosphate buffer at pH 6.2 supplemented with 250 mM KCl. Flow rate was 0.3 mL/min and injected volume was 1 µL.
Journal of Chromatography A, 1752, 2025, 465972: Fig. 3. Analysis of rAAV5 sample under standard SEC conditions using six different widepore SEC columns. Mobile phase was composed of 200 mM phosphate buffer at pH 6.2 supplemented with 250 mM KCl. Flow rate was 0.3 mL/min and injected volume was 1 µL.
4. Conclusions
The aim of this study was to evaluate the practical capabilities and limitations of various recently commercialized wide-pore and ultra-wide-pore SEC columns for the characterization of two gene therapy products, namely rAAVs and mRNA.
SEC columns with pore sizes around 500 Å were obtained from six manufacturers. They were compared for the separation of size variants for various rAAV serotypes. Significant differences in kinetic efficiency were observed between the columns, with the DNACore AAV-SEC achieving the highest efficiency (N = 11,000 plates), likely due to the use of monodisperse silica, whereas the SRT SEC-500 column exhibited much lower efficiency (N = 1000 plates), attributed to its larger 5 µm particles. Based on pore size match calculations, optimal selectivity for rAAVs, which is typically achieved when analytes elute near the centre of the elution window, was obtained with columns having larger pore sizes ranging from 550 to 700 Å. Despite variations in kinetic efficiency and selectivity between columns, the optimal column choice remained highly sample dependent. For example, an additional HMWS peak was observed for the rAAV5 using the DNACore AAV-SEC, AdvanceBio SEC 500A, and SRT SEC-500 while the Biozen dSEC-7 LC, AdvanceBio SEC 500A and SurePac Bio 550 SEC Mdi columns demonstrated superior performance for rAAV2. Interestingly, even the SRT SEC-500 column, despite its lower efficiency, provided relevant results for specific samples, such as rAAV5. Moreover, all tested SEC columns can be successfully used for stability-indicating assays of rAAVs, as comparable degradation trends were observed.
The separation of mRNA size variants was evaluated using ultra wide pore SEC columns (700–1000 Å) obtained from five manufacturers. The Biozen dSEC-7 LC column achieved the highest efficiency for all mRNA samples and provided the best separation performance for small mRNAs such as the eGFP mRNA (< 1000 nt). Conversely, columns packed with 1000 Å particles such as the AdvanceBio SEC 1000A were more suitable for the separation of size variants from larger mRNA products (Cas9 mRNA, about 4500 nt). Overall, similar amount of impurities eluting before and after the main peak were obtained across all tested columns, with only a few exceptions. Increasing the column temperature to 50 °C impeded quantitative assessments, while the addition of 10 mM MgCl2 did not offer any significant advantage for the samples studied.
