Skip to Content
You are currently on the new version of our website. Access the old version .
MembranesMembranes
  • Article
  • Open Access

20 February 2026

12 Pages

Optimization of Tangential Flow Filtration for High-Yield, Scalable Downstream Processing of Adeno-Associated Virus

,
and
Sartorius Stedim Biotech GmbH, August-Spindler-Straße 11, 37079 Göttingen, Germany
*
Author to whom correspondence should be addressed.

Abstract

The demand for effective downstream processing of adeno-associated virus (AAV) is increasing as gene therapies advance toward broader clinical applications. Robust, efficient, and scalable ultrafiltration and diafiltration (UF|DF) operations are essential for generating high-quality AAV preparations, with tangential flow filtration (TFF) serving as a critical unit operation for vector concentration, impurity reduction, and buffer exchange while maintaining viral functionality. Development of TFF processes requires careful consideration of membrane characteristics—including chemistry, pore size or channel architecture—as these parameters directly influence vector retention, fouling behavior, and overall process efficiency. Equally important is the optimization of critical process parameters such as recirculation rate, transmembrane pressure (TMP), and total processing time, all of which govern hydrodynamic performance and product quality. This study assessed two Sartocon® Hydrosart® TFF cassette architectures—ECO-Screen and E-Screen—for the ultrafiltration and diafiltration of AAV8 clarified lysate. Through flux characterization and controlled small-scale evaluations, cassette-specific operating regions were defined. Both configurations supported high viral genome retention; however, the E-Screen geometry achieved faster processing and superior removal of host–cell protein and DNA contaminants, whereas the ECO-Screen format allowed for efficient operation under reduced pump rates and, therefore, lower shear conditions. Reproducibility assessments demonstrated minimal run-to-run variability, confirming the robustness of the optimized operating parameters. A 10-fold scale-up further validated the linearity and predictability of the UF|DF process, with consistent impurity-reduction profiles and only modest deviations in viral recovery. Collectively, these findings provide a quantitative basis for rational cassette selection in AAV purification workflows and establish a scalable, scientifically grounded UF|DF framework applicable across development and manufacturing scales.

1. Introduction

As gene therapy continues to advance rapidly, adeno-associated virus (AAV) vectors have become essential delivery vehicles due to their versatility and the broad range of available serotypes [1]. AAV has become a widely utilized viral vector for in vivo gene therapy, as evidenced by the regulatory approval of products such as Luxturna®, Zolgensma®, Elevidys, Roctavian® and more recently, BevqezTM, which validate its clinical applicability [2,3,4]. In addition, a substantial number of gene therapy candidates are undergoing clinical evaluation, with additional products advancing through preclinical development pipelines. The predominant therapeutic indications include disorders of the central nervous system, ocular diseases, neuromuscular pathologies, lysosomal storage disorders, and hematologic conditions [5]. With the expanding use of these AAV-based therapeutics, there is increasing pressure to establish manufacturing platforms that are scalable, consistent, and economically feasible [6].
Tangential Flow Filtration (TFF) is a key operation in the downstream purification of AAV vectors, which enables vector concentration, impurity reduction, and buffer exchange, functions that are essential for producing AAV of suitable quality for clinical use [1,7]. Although the integration of ultrafiltration and diafiltration (UF|DF) steps within AAV purification workflows varies across platforms, TFF is most frequently applied before chromatographic purification and following chromatography for final formulation of the vector [8]. Compared with other viral modalities, AAV manufacturing typically involves a streamlined set of downstream unit operations centered around chromatography and TFF, which together serve as the primary drivers of product purity for both drug substance and drug product [1].
Both hollow-fiber modules and flat-sheet cassettes have been explored for AAV processing in the literature [8,9]. While sharing a common separation principle, both systems differ in their geometric configuration, which influences performance characteristics. Hollow fiber modules consist of cylindrical lumens that maintain laminar flow under typical bioprocessing conditions. This configuration allows for the handling of viscous feed streams with minimal pressure drop and exerts relatively low shear stress, making it particularly suitable for shear-sensitive materials [10,11]. Because the shear sensitivity of many emerging therapies, such as viral vectors, is not yet fully understood, there is a significant concern regarding exposure to high fluid velocities or turbulent flow. This uncertainty often encourages a conservative approach to membrane selection and process development, with hollow fiber configurations frequently preferred to reduce the risk of shear-induced product degradation [12]. In contrast, flat-sheet cassettes incorporate feed channel screens that induce turbulence, disrupting boundary layers formed by non-permeable solutes which results in improved mixing and overall filtration performance [13]. Flat-sheet formats also offer operational advantages that support large-scale bioprocess implementation, including widespread GMP adoption, regulatory familiarity, reduced hold-up volumes, straightforward scale-up, and lower fouling risk, making them particularly suitable for adeno-associated virus (AAV) purification [7]. For AAV processes, TFF membranes typically employ molecular-weight cutoffs ranging from roughly 30 to 300 kDa, with larger pore sizes offering enhanced impurity passage but also carrying some risk of vector loss [7,12,14]. Materials commonly used in TFF membranes—polyethersulfone (PES), polyvinylidene fluoride (PVDF), and regenerated cellulose (RC)—each provide distinct performance characteristics [15]. PVDF offers strong thermal and chemical resistance but can exhibit hydrophobicity-driven fouling unless surface-modified [16]. PES membranes, while widely adopted, may also display moderate hydrophobicity that contributes to non-specific interactions [17,18]. Hydrosart® membranes, based on stabilized cellulose, are highly hydrophilic and well-suited for biological applications due to low protein binding, high flux, and broad pH and temperature stability [7,18].
As the spectrum of AAV-based therapeutics expands and serotype diversity grows, optimizing TFF configurations for each vector type is becoming increasingly important. Membrane performance can be affected by parameters such as transmembrane pressure, feed composition, and flow conditions, all of which can impact fouling behavior [19]. Fouling can lead to reduced flux, longer processing durations, and greater membrane area requirements, thereby increasing system hold-up [7,14]. Process adaptations—such as post-processing washes to recover adsorbed vector or formulation components that mitigate shear stress—can help preserve AAV integrity and overall TFF yield [12].
This work aimed to establish a robust concentration and diafiltration unit operation for AAV8 using Sartocon® Hydrosart® 100 kDa tangential flow filtration cassettes. Two cassette configurations, ECO-Screen and E-Screen were compared at small scale with respect to viral genome (vg) recovery, impurity clearance—assessed by total protein and double-stranded DNA—and process efficiency, as indicated by permeate flux. The reproducibility of the selected configuration and its scalability were further evaluated to ensure consistent performance within an AAV downstream purification workflow.

2. Materials and Methods

2.1. AAV8 Production

AAV8 was generated by transiently transfecting HEK293 cells with FectoVIR®-AAV (Sartorius, Göttingen, Germany). The cell culture was performed in a 10 L Univessel® Glass stirred-tank system operated with a Biostat® B controller (Sartorius). At harvest, a Tween-20–containing lysis buffer (resulting bioreactor concentrations: 50 mM Tris, 1% [v/v] Tween 20, 2 mM MgCl2, pH 8.0) was introduced directly into the reactor to disrupt the cells and release the produced vector particles. An endonuclease treatment followed to degrade residual nucleic acids and support downstream purification efficiency. The lysate was subsequently clarified using a Sartoclear® DL75 depth filter and a Sartopore® 2 XLG membrane filter (Sartorius, Göttingen, Germany). The clarified AAV8 material was aliquoted, frozen at −80 °C, and used as feed for all experiments. The vector genome concentrations of the batches used in small-scale and scale-up studies were 3.4 × 1010 vg/mL and 4.6 × 1010 vg/mL, respectively.

2.2. Small-Scale AAV8 TFF Step Development

Ultrafiltration and diafiltration experiments were performed on a Sartoflow® Smart TFF system (Sartorius) equipped with 100 kDa Hydrosart® Sartocon® Slice 200 cassettes in both E-Screen and ECO-Screen configurations, each providing 180 cm2 of membrane surface area. Prior to processing, a permeate flux characterization was completed for every cassette to identify suitable pump settings and transmembrane pressure (TMP) for UF|DF operation. For each module, pump speed and TMP were gradually increased until a decline in permeate flux was detected, and the TMP at the inflection point was selected as the operating pressure.
All runs were carried out with a fixed starting volume of 1 L, using a constant pump rate, followed by a 10-fold concentration and five diavolumes of diafiltration with a buffer containing 20 mM Tris, 200 mM NaCl, 0.1% (w/w) poloxamer 188, and 2 mM MgCl2 at pH 7.5. After completion of UF|DF, the system and modules were rinsed twice with one hold-up volume of diafiltration buffer (50 mL), circulated for 5 min each. These rinse fractions were pooled with the final retentate.

2.3. AAV8 TFF Scale-Up

A duplicate scale-up UF|DF run was carried out using the Sartoflow® Advanced TFF system (Sartorius) equipped with two Sartocon® Hydrosart® Slice modules, each providing 0.14 m2 of effective membrane area for a combined area of 0.28 m2. Operating parameters—namely the pump setting (%) and transmembrane pressure—were taken directly from the TMP-scouting experiments performed at a small scale on the Sartoflow® Smart platform and transferred unchanged to the larger system. A total of 10 L of feed was processed, undergoing a 10-fold concentration step followed by five diavolumes of buffer exchange using a formulation composed of 20 mM Tris, 200 mM NaCl, 0.1% (w/w) poloxamer 188, and 2 mM MgCl2 at pH 7.5.
Upon completion of UF|DF, the TFF system and cassettes were rinsed twice with one hold-up volume each by circulating 200 mL of diafiltration buffer for 5 min. The two rinse fractions were then pooled with the final retentate.

2.4. Analytical Methods

2.4.1. Viral Genome Copies Quantification

To remove free DNA, 5 µL of cell lysate was combined with 35 µL of ddPCR dilution buffer, 5 µL of DNase I, and 5 µL of the corresponding DNase reaction buffer, followed by incubation at 37 °C for 1 h. For subsequent DNase inactivation, a mixture containing 20 µL PCR-grade water, 8 µL of 50 mM EDTA, and 1 µL of 10% SDS was prepared. A 29 µL aliquot of this mixture was added to the 50 µL DNase-treated sample.
Next, a solution composed of 14 µL PCR-grade water, 2 µL Proteinase K, and 5 µL of 20× Proteinase K buffer was prepared, and 21 µL of this mixture was added to the sample. To release encapsidated AAV genomes, the reaction was incubated at 50 °C for 1 h. Proteinase K was then heat-inactivated at 95 °C for 15 min. The resulting preparation containing purified rAAV genomes was quantified using the QX200 ddPCR platform (Bio-Rad Laboratories, Inc., Hercules, CA, USA) with an eGFP-specific primer/probe assay according to the manufacturer’s guidelines.

2.4.2. Total dsDNA Quantification

Quantification of total double-stranded DNA (dsDNA) was carried out using the Quant-iT™ PicoGreen® dsDNA Assay Kit (Thermo Fisher Scientific Inc., Waltham, MA, USA) following the protocol provided by the manufacturer. Samples and standards were prepared in duplicate and dispensed into black 96-well plates (Corning Incorporate, Corning, NY, USA) to reduce background fluorescence. Fluorescence measurements were collected on a FLUOstar® Omega plate reader (BMG LABTECH GmbH, Ortenberg, Germany). using excitation at 480 nm and emission at 520 nm. A calibration curve generated from known dsDNA standards was used to perform linear regression and calculate dsDNA concentrations for each sample.

2.4.3. Total Protein Quantification

Protein concentrations were determined using the Pierce™ Coomassie (Bradford) Protein Assay Kit (Thermo Fisher Scientific Inc., Waltham, MA, USA) according to the manufacturer’s protocol. Standards and samples were prepared in duplicate and dispensed into transparent 96-well plates (Greiner Bio-One, Kremsmünster, Austria). Absorbance at 595 nm was measured using a FLUOstar® Omega microplate reader (BMG LABTECH GmbH, Ortenberg, Germany. A second-order polynomial calibration curve generated from the standards was used to calculate the protein concentration for each sample.

3. Results and Discussion

This study assessed two cassette configurations to develop a structured workflow for AAV8 tangential flow filtration (TFF) process development. The workflow transitioned from early-stage flux characterization and small-scale evaluations to scale-up of the finalized conditions. A summary of the experimental workflow is provided in Figure 1.
Figure 1. Workflow of the study.
The Sartoflow Smart® system was employed to evaluate scale-relevant process conditions using Sartocon® Hydrosart® Slice 200 membranes with molecular weight cutoffs of 100 kDa. Two ultrafiltration cassette formats with distinct flow-channel architectures—ECO-Screen and E-Screen—were systematically assessed. The ECO-Screen cassette is engineered for efficient concentration and diafiltration of low-viscosity process streams, whereas the E-Screen cassette is intended for applications involving higher-viscosity feeds [20]. While E-Screen features wider channels and higher liquid hold-up volume, it has a lower membrane area-to-liquid volume ratio. Conversely, ECO-Screen has narrower channels and lower hold-up volume, resulting in a higher membrane area-to-liquid volume ratio. This expanded surface area in contact with the viral feed stream, reduces the volumetric flow rate required to achieve a comparable pressure drop across the system. Consequently, processes utilizing the ECO-Screen format may demand less pump rate during scale-up.

3.1. Flux Characterization

Owing to the differences in device formats, the optimal operating conditions are specific to each consumable and cannot be directly compared. From a process-development perspective, it would not be meaningful to compare the performance of both consumables under a single standardized set of operating conditions, as such conditions would not allow each device to operate within its optimal range for a given feed stream. To establish the optimal operating transmembrane pressure (TMP), permeate flux was quantified across progressively increasing TMP values at two predefined feed flow rates setpoints suitable for each cassette geometry. Consistent with prior investigations, the optimal TMP region was delineated as the point approaching the pressure-independent regime, wherein further increases in applied pressure no longer yield proportional increases in permeate flux [14]. Although a degree of operator-dependent judgment is inherently involved in final parameter selection for UF|DF operations, TMP scouting generally aims to identify conditions that maximize flux—thereby reducing overall processing time—while simultaneously maintaining sufficiently moderate hydrodynamic forces to preserve product integrity [7,14,19]. The same principle applies to the feed flow rate, or crossflow rate. During TMP scouting, the operator also identifies the maximum crossflow rate before clogging, defined as the highest flow that maintains stable flux without excessive fouling. In this critical zone, permeate flux no longer increases linearly with feed flow, indicating the threshold of the optimal operating range. Under these considerations, the adoption of a higher feed flow rate is warranted only when it provides a demonstrable enhancement in permeate flux. Moreover, selecting operating conditions that ensure adequate crossflow and turbulence can have additional benefits, including improved product quality through enhanced removal of impurities while minimizing fouling effects.
Across both consumable configurations, the resulting TMP ranges were broadly consistent, with cassette-specific differences in the feed pump selected for the ECO-Screen and E-Screen formats (Figure 2). For each module, the permeate flux was measured at increasing TMP values and at two different target feed flow rates, starting with the lowest and ending with the highest. Notably, these TMP scouting experiments were performed using modules with depolarization but without caustic cleaning between the target flow rates. Instead, between each flow rate, a five-minute depolarization step was applied, during which permeate flow was stopped and the feedstock was recirculated. Because the lower feed flow-rate TMP iterations were conducted first, we hypothesize that partial membrane fouling may have occurred under the initial operating conditions. This could result in lower permeate fluxes at the higher flow rates compared with a scenario in which the module is sanitized between TMP curves.
Figure 2. On the left—Permeate flux (LMH) measured as a function of TMP at two feed flow-rates ( 67 LMH; 167 LMH) for Sartocon® Slice 200 Hydrosart® 100 kDa ECO-Screen cassette. On the right—Permeate flux measured as a function of TMP at two different feed flow-rates ( 500 LMH; 633 LMH) for Sartocon® Slice 200 Hydrosart® 100 kDa E-Screen cassette.

3.2. Small-Scale AAV8 TFF Step Development

3.2.1. UF|DF Performance

Once the optimal operating transmembrane pressure (TMP) and feed flow rate values had been established for each ultrafiltration cassette type, these conditions were subsequently applied to assess cassette performance during the concentration and diafiltration of AAV clarified lysates.
For both cassette configurations, the optimal TMP values were selected within the same operational range, although higher feed flow rates were applied to the E-Screen cassette (Table 1). Comparison of the two cassette configurations revealed that the system-regulated pump rate was 2–3-fold lower for the ECO-Screen format than for the E-Screen format. The E-Screen configuration achieved higher fluxes and shorter processing times, but at the cost of increased pump rates. Given that feed viscosity remained stable (~1–1.2 cP) throughout all UF|DF trials, the differing flow behaviors and fouling propensities are more accurately attributed to the differences in processing time between the E-Screen and ECO-Screen membrane formats, arising primarily from their distinct membrane geometries and corresponding TFF operating regions, which result in different fluid dynamics. ECO-Screen features narrow flow channels and is operated at defined crossflow velocities to achieve a target pressure drop (ΔP), directly linked to pump rate and crossflow. In contrast, due to its wider internal channel structure and larger liquid hold-up volume, E-Screen requires higher pump rates and crossflow to reach the same ΔP (i.e., operating region). Hypothetically, if ECO-Screen would be operated at the same pump rates as E-Screen, the resulting ΔP would be substantially higher, leading to increased crossflow and permeate flux compared to E-Screen. Indeed, at equivalent feed flow rates and crossflow, ECO-Screen demonstrates higher flux values with protein-based solutions. However, operating ECO-Screen at such elevated ΔP is generally feasible only with cleaner feeds as those protein solutions. Complex feeds, such as viral vector solutions, are more prone to aggregation, fouling and solute layer formation. Under these conditions, the higher crossflow rates that define the optimal operating range of E-Screen cannot be applied to ECO-Screen. Consequently, direct comparison of processing times between ECO-Screen and E-Screen formats is not straightforward.
Table 1. Parameters and results of small-scale AAV8 UF|DF runs performed with Sartocon® Slice 200 Hydrosart® ECO-Screen and E-Screen.
Overall, the more open structure of E-Screen enables faster processing because it allows for higher pump rates to operate within its optimal range. The resulting higher shear and improved surface sweeping enhance fouling resistance and sustain flux. Nevertheless, ECO-Screen remains advantageous in specific scenarios: it enables efficient AAV processing when pump capacity is limited—particularly at larger scales—and allows operation at lower shear rates [14,21]. Characterizing the observed variations in flux decline and distinct polarization and fouling phenomena in depth was beyond the scope of the present study; future work should incorporate flux-decline modeling to more rigorously assess how device architecture and TFF parameters shape overall UF|DF performance.
With respect to functional performance metrics, viral genome recovery was largely consistent among the two cassette formats (Figure 3). Both configurations achieved nearly complete retention of viral genome, with 98% and 93% for the ECO-Screen and E-Screen recovery in the retentate, respectively, with no virus detected in permeate samples. Protein clearance was higher using the E-Screen cassette geometry compared with ECO-Screen (85% versus 75%, respectively), whereas DNA removal increased significantly while using E-Screen configuration, compared with the ECO-Screen format (40% versus 7%, respectively) (Figure 3). These differences in impurity-removal performance bear direct relevance for process design, as contaminant burden substantially influences flux behavior; elevated protein and DNA concentrations promote fouling and increase overall membrane resistance, thereby reducing productivity over time [22]. Nevertheless, the enhanced protein-removal capabilities represent a critical asset in AAV vector purification workflows, and the exploration of larger-pore membranes in a further study may provide a viable strategy to improve impurity clearance, provided that viral retention remains within acceptable thresholds. Based on the lower pump rate requirement and possibly slightly higher product recovery, the ECO-Screen cassette format was selected to further evaluate the AAV8 UF|DF.
Figure 3. Viral genomes recovery, dsDNA and protein removal of AAV8 ultrafiltration and diafiltration performed with Sartocon® Slice 200 Hydrosart® 100 kDa ECO-Screen and E-Screen TFF cassettes.

3.2.2. Reproducibility of Established AAV8 TFF Step

Three independent UF|DF runs of the harvested AAV8 lysate were conducted using the Sartocon® Slice 200 Hydrosart® 100 kDa ECO cassettes on the Sartoflow® Smart TFF system to evaluate the reproducibility of the established process. The operating parameters applied in these confirmatory studies correspond to those listed in Table 1 (for the ECO-Screen configuration). Overall, viral recovery and impurity clearance were consistent with the outcomes of the initial development experiments (Table 2). High run-to-run reproducibility was observed, with an average of 121% recovery and standard deviation of 18% for viral genome recovery. In this case, quantitative assays such as ddPCR require sample dilution to fall within the assay’s dynamic range. Retentate samples typically require higher dilution factors, and assay variability is therefore amplified during back-calculation, leading to greater error propagation and potentially explaining recoveries above 100%. No viral particles were detected in the permeate under any experimental condition, with all measurements below the assay’s limit of detection. Therefore, the recovery values are reported only for the retentate. Similarly, dsDNA and protein removal demonstrated strong consistency across replicates, with average values of 79% and 13% and standard deviations of 3% and 6%, respectively. Overall, the data demonstrate that the UF|DF process established for AAV8 using the Hydrosart® 100 kDa ECO cassette is both robust and efficient, delivering highly reproducible performance across independent runs.
Table 2. Viral genomes recovery (% mean + StDev; N = 3) and dsDNA and protein removal (% mean + StDev; N = 3) of AAV8 ultrafiltration and diafiltration performed with Sartocon® Slice 200 Hydrosart® 100 kDa ECO-Screen.

3.3. Scale-Up Performance of Established AAV8 TFF Process

The scalability of the established UF|DF process was subsequently evaluated. The scale-up run was performed using a 10-fold increase in harvested AAV8 lysate volume, corresponding to the maximal working volumes of the respective feed vessels (from 1 to 10 L). Two Sartocon® Slice cassettes were implemented to provide the membrane surface area required for scaling from the Sartocon® Slice 200 cassette format. However, due to differences in these available formats and membrane areas of the TFF cassettes for small- and large-scale operations, the membrane area was increased by a factor of 15.6 (from 180 cm2 to 0.28 m2). As a result, the loading density per membrane area was lower in the scale-up run (approximately 36 L/m2) compared with the small-scale trials (approximately 55 L/m2), partially explaining the lower running time experienced in the scaled-up run. Transmembrane pressure (TMP), pump rate, and associated operating conditions were scaled to maintain an equivalent feed flux to that used in the Sartocon® Slice 200 studies (Table 3). At the increased scale, a modest reduction in viral genome recovery was observed (90% vs. 121% vg; Figure 4). In fact, no definitive conclusions can be drawn as this difference falls within the expected variability range of the quantification assay. Besides this analytical uncertainty, assuming the potential lower recovery observed upon scale-up of the UF|DF process, the variance can also be attributed to factors such as the increased system hold-up volume and challenges in maintaining identical hydrodynamic conditions (especially when operating with different systems at small and larger scales), both well-known scale-dependent effects. Lastly, contaminant clearance performance remained largely consistent between scales, with comparable removal of host–cell DNA (89% vs. 79% reduction) and protein impurities (20% vs. 13% reduction; Figure 4). Collectively, the alignment of recovery and impurity removal metrics across scales demonstrates that the AAV8 ultrafiltration and diafiltration process exhibits predictable scalability from the Sartocon® Slice 200 format to the Sartocon® Slice configuration.
Table 3. Parameters of small-scale and scaled-up AAV8 UF|DF runs performed with either Sartocon® Slice 200 Hydrosart® ECO-Screen (small-scale) or 2× Sartocon® Slice Hydrosart® 100 kDa ECO-Screen (scaled-up).
Figure 4. Viral genomes recovery, dsDNA and protein removal from an AAV8 ultrafiltration and diafiltration operation comparing Hydrosart® 100 kDa ECO-Screen cassettes at small scale (Sartocon® Slice 200, mean from N = 3) and scaled-up (2× Sartocon® Slice, mean from N = 2).

4. Conclusions

This study provides a comprehensive evaluation of two Hydrosart® TFF cassette geometries—ECO-Screen and E-Screen—for ultrafiltration and diafiltration of AAV8, establishing a structured and scalable workflow for TFF process development. Through flux characterization studies, small-scale performance assessments, and verification of run-to-run reproducibility, it was demonstrated that both cassette formats deliver high viral genome recovery and effective contaminant removal, with each offering distinct operational advantages. The E-Screen format enables faster processing by operating at higher pump rates, which increases shear and enhances surface sweeping, thereby improving fouling resistance, maintaining high flux, and supporting efficient impurity removal. In contrast, ECO-Screen is advantageous when pump capacity is limited—particularly at larger scales—and allows operation at lower shear rates, making it well suited for processing sensitive products. Importantly, scalability of the optimized UF|DF process was confirmed across a 10-fold increase in feed volume, with consistent impurity reduction and only minimal impact on viral recovery. These findings validate the robustness and transferability of the established operating parameters and highlight the suitability of Hydrosart® TFF cassettes for AAV8 purification across development scales.
In conclusion, this work underscores the critical influence of cassette architecture on TFF performance and provides practical guidance for selecting filtration formats based on process constraints, desired throughput, and product quality considerations. The demonstrated scalability and reproducibility support the broader applicability of this workflow to additional AAV serotypes and offer a strong foundation for future optimization efforts, including fouling characterization and membrane selection strategies tailored to vector-specific biophysical properties. It is important to note that, while AAV activity is critical for final process validation, this study intentionally addresses an earlier phase of process development. The objective was to characterize and compare filtration performance of TFF cassettes under defined engineering conditions rather than to assess vector functionality. As such, activity measurements were not conducted at this stage. The results provide a necessary process development guide framework that will inform subsequent studies incorporating biological activity and infectivity assays.

Author Contributions

Conceptualization, S.C. and F.B.; methodology, S.C. and F.B.; formal analysis, S.C. and F.B.; resources, A.T.; writing—original draft preparation, S.C.; writing—review and editing, S.C. and F.B. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author(s).

Acknowledgments

The authors would like to kindly thank Magda Tomala and Suneetha Nunna for viral feed stream production, Axel Thiefes and Lucas Herbst for harvest and clarification processing and Katja Heinrichs for the analytics execution support. The authors also thank Yvonne Volz, Martin Leuthold and Marc Jenke for their fruitful discussions.

Conflicts of Interest

Author Sara Cardoso, Franziska Bollmann and Alexander Tappe were employed by the company Sartorius Stedim Biotech GmbH. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Jungbauer, A.; Wheelwright, S. Downstream processing of AAV-based gene therapy vectors. Sep. Purif. Technol. 2025, 368, 133051. [Google Scholar] [CrossRef] [Scilit]
  2. Gene, Cell, + RNA Therapy Landscape Report Q4 2025. American Society of Gene & Cell Therapy and Citeline. 2025. Available online: https://www.asgct.org/news-publications/landscape-report (accessed on 24 November 2025).
  3. Wang, J.; Gessler, D.J.; Zhan, W.; Gallagher, T.L.; Gao, G. Adeno-associated virus as a delivery vector for gene therapy of human diseases. Signal Transduct. Target. Ther. 2024, 9, 78. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  4. Karasawa, H.; Mizoguchi, H.; Choi, B.; Hsieh, C.; Miyano, M.; Moriya, Y.; Muthusamy, S.; Takakura, K.; Suzuki, K.; Tsurumaki, Y.; et al. Regulatory insights on advanced CAR-T cell products, AAV-based gene therapies, and medical care/practice in cell and gene therapies: Report from the 6th Asia Partnership Conference of Regenerative Medicine, April 20, 2023. Regen. Ther. 2024, 26, 967–980. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  5. Zhao, Z.; Anselmo, A.C.; Mitragotri, S. Viral vector-based gene therapies in the clinic. Bioeng. Transl. Med. 2021, 7, e10258. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  6. Jiang, Z.; Dalby, P.A. Challenges in scaling up AAV-based gene therapy manufacturing. Trends Biotechnol. 2023, 41, 1268–1281. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  7. Mendes, S.; Faria, T.Q.; Nascimento, A.; Noverraz, M.; Bollmann, F.; Nestola, P.; Roldão, A.; Peixoto, C.; Silva, R.J.S. Accelerated development of AAV purification process using a high-throughput and automated crossflow system. Separations 2024, 11, 73. [Google Scholar] [CrossRef] [Scilit]
  8. Arunkumar, A.; Singh, N. Ultrafiltration behavior of recombinant adeno-associated viral vectors used in gene therapy. J. Membr. Sci. 2020, 620, 118812. [Google Scholar] [CrossRef] [Scilit]
  9. Miyaoka, R.; Tsunekawa, Y.; Kurosawa, Y.; Sasaki, T.; Onodera, A.; Sakamoto, K.; Kakiuchi, Y.; Wada, M.; Nitahara-Kasahara, Y.; Hayashita-Kinoh, H.; et al. Development of a novel purification method for AAV vectors using tangential flow filtration. Biotechnol. Bioeng. 2023, 120, 3311–3321. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  10. Arunkumar, A.; Singh, N.; Schutsky, E.G.; Peck, M.; Swanson, R.K.; Borys, M.C.; Li, Z.J. Effect of channel-induced shear on biologics during ultrafiltration/diafiltration (UF/DF). J. Membr. Sci. 2016, 514, 671–683. [Google Scholar] [CrossRef] [Scilit]
  11. Wang, S.; Godfrey, S.; Ravikrishnan, J.; Lin, H.; Vogel, J.; Coffman, J. Shear contributions to cell culture performance and product recovery in ATF and TFF perfusion systems. J. Biotechnol. 2017, 246, 52–60. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  12. McCarney, L.; Ravichandran, A.G.; Tansey, S.; Dango, M.; Marchand, N. A blueprint for TFF format selection: Hollow fibers and flat sheets for ultrafiltration of AAV. Sep. Purif. Technol. 2025, 370, 133064. [Google Scholar] [CrossRef] [Scilit]
  13. Perry, C.; Rayat, A.C.M.E. Lentiviral vector bioprocessing. Viruses 2021, 13, 268. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  14. Cardoso, S.; Krueger, S.; Engelhardt, J.; Tappe, A. Development of a robust large-scale downstream processing of lentiviral vectors using tangential flow filtration. Sep. Purif. Technol. 2025, 378, 134568. [Google Scholar] [CrossRef] [Scilit]
  15. Nestola, P.; Peixoto, C.; Silva, R.R.J.S.; Alves, P.M.; Mota, J.P.B.; Carrondo, M.J.T. Improved virus purification processes for vaccines and gene therapy. Biotechnol. Bioeng. 2015, 112, 843–857. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  16. Keskin, B.; Ormancı-Acar, T.; Türken, T.; Imer, D.Y.; Koyuncu, I. Effect of wetting agent on the dye filtration performance of ultrafiltration membrane. Water Sci. Technol. 2020, 82, 577–586. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  17. Madsen, E.; Kaiser, J.; Krühne, U.; Pinelo, M. Single-pass tangential flow filtration: Critical operational variables, fouling, and main current applications. Sep. Purif. Technol. 2022, 291, 120949. [Google Scholar] [CrossRef] [Scilit]
  18. Zydney, A.L. New developments in membranes for bioprocessing—A review. J. Membr. Sci. 2021, 620, 118804. [Google Scholar] [CrossRef] [Scilit]
  19. Cardoso, S.; Oender, S.; Engelhardt, J.; Tappe, A. Downstream workflows for intermediate purification of lentiviral vectors using tangential flow filtration and AEX membrane chromatography. Eng. Life Sci. 2025, 25, e70056. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  20. Sartorius. Hydrosart® Ultrafiltration Cassettes; Datasheet; Sartorius: Göttingen, Germany, 2021. [Google Scholar]
  21. Rosenberg, E.; Hepbildikler, S.; Kuhne, W.; Winter, G. Ultrafiltration concentration of monoclonal antibody solutions: Development of an optimized method minimizing aggregation. J. Membr. Sci. 2009, 342, 50–59. [Google Scholar] [CrossRef] [Scilit]
  22. Oetomo, B.; Luo, L.; Qu, Y.; Discepola, M.; Kentish, S.E.; Gras, S.L. Controlling tangential flow filtration in biomanufacturing processes via machine learning: A literature review. Digit. Chem. Eng. 2025, 14, 100211. [Google Scholar] [CrossRef] [Scilit]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.