Scalable Tangential Flow Filtration (TFF) for Downstream Processing of Viral Vectors for Gene Therapy and Vaccine Production

Use of Tangential Flow Filtration During Viral Vector Manufacturing

The production of viral vectors is a critical step in gene therapy and vaccine manufacturing. Success depends on using the right technologies to ensure an optimized, efficient, and predictable process. An essential technology used in downstream processing of viral vectors is tangential flow filtration (TFF).

During viral vector manufacturing, the TFF step is used for two important purposes:

• Volume reduction and diafiltration prior to a capture chromatography step
• Final concentration and exchange to formulation buffer prior to the final sterile filtration

Ultimately, TFF ensures the target concentration of the viral vector is reached, and the final product is in the desired drug substance matrix or buffer.

Development and optimization of TFF steps in downstream processing requires consideration of several factors.

Section overview

• Best Practices when Designing a TFF Step for Viral Vector Manufacturing
• Types of TFF Membranes and Filter Formats used in Viral Vector Purification
• TFF Operating Control Strategies and Scalability for Viral Vector Process Development
• Comparison of Pellicon^® Capsules and Hollow Fibers for TFF in Viral Gene Therapy Manufacturing
• Related Products
• Related Content

 

Best Practices when Designing a TFF Step for Viral Vector Manufacturing

Designing a TFF step begins with identifying the objective, whether it is to achieve a targeted product concentration to eliminate some downstream processing bottlenecks, or to transfer a formulation into a specific buffer and then concentrate it at the end of the purification process. It is critical to choose a system design and a device format that is scalable; in addition, the systems need to be characterized for at-scale performance with the appropriate turndown ratios to achieve the desired yields and capacities at the full scale.

An important consideration when using closed processing is that the system and the device should be designed for this approach. It is particularly helpful to have process performance and recovery data available for at-scale systems and at-scale devices.

The size and the isoelectric point of the target molecule should also be known as well as how key process variables such as temperature, shear, and mixing affect the stability of the modality. It should also be known how the modality interacts with the materials of construction, whether the membrane is polyethersulfone (PES)-based or regenerated cellulose-based. Typically, regenerated cellulose is preferred for applications requiring low molecule binding; however, the type of membrane and its interaction with the modality needs to be experimentally verified. A final consideration is how stable the modality is in a particular buffer matrix.

Types of TFF Membranes and Filter Formats used in Viral Vector Purification

Different types of membranes and filter formats are available for TFF. Membranes made of PES or regenerated cellulose with a molecular weight cut-off of 30, 100 or 300 kDa are generally recommended for viral vector production. The two most common filter formats are flat sheet cassettes or capsules and hollow fibers.

Hollow fibers, traditionally used for industrial and biomedical separations, have been adopted for many biopharmaceutical applications. They are available with modified PES and mixed cellulose ester lumens. A hollow fiber filter can be selected based on lumen number, diameter, and length. Flat sheet cassettes, available with both modified PES and regenerated cellulose membranes, are specifically designed for biopharmaceutical applications. They are very robust and offer efficient process performance and linear scalability in a compact format. 

Figure 1. A first-of-its kind spiral wound Pellicon^® Capsule format was engineered to provide high performance and linear scalability in a plug-and-play design and is particularly well-suited for use in viral vector production. Membrane pore sizes include 100 kDA and 300 kDa.

TFF Operating Control Strategies and Scalability for Viral Vector Process Development

Permeate-control TFF systems are typically used for microfiltration (MF) applications in which highly permeable membranes can lead to excessively high fluxes that degrade or destabilize performance. For tighter ultrafiltration (UF) membrane applications, such as 30 kDa for antibody retention, transmembrane pressure (TMP)-control systems are used as an alternative. In contrast, membrane cutoffs typically used in viral vector manufacturing, such as adeno-associated virus (AAV) and lentivirus (LV), are 100 and 300 kDa. The openness of these membranes falls between the tighter UF and more open MF applications.

A study was performed to evaluate both TMP and permeate control strategies during the post-clarification / pre-capture TFF step in a viral vector manufacturing workflow. The Pellicon^® Capsule was used in the experiments along with scale-down Pellicon^® XL 50 cassettes with 100 and 300 kDa Ultracel^® membrane.

For the TMP-control operation, a feed pump was used and there was free flow of permeate (Figure 1). A two-pump TFF system was used for the permeate-control operation, which included a feed pump and a permeate pump to restrict and control flow of the permeate. 

Figure 2. Schematic of TFF systems used in this study: TMP control (a) and permeate control with permeate pump (b).

Initial experiments were performed to characterize permeate flux of the TFF filters for both control strategies. Following this, ultrafiltration/diafiltration (UF/DF) process simulations were run to evaluate average flux, processing time, TMP, virus yield, and impurity reduction (Table 1). The process simulation goal was to concentrate four-fold (4x) in batch mode, diafilter with five diavolumes (5 DV) of HEPES buffer at constant volume, and then concentrate two-and-a-half-fold (2.5x) in batch mode for an overall volumetric concentration factor (VCF) of 10x.

The complete set of methods and experimental results can be found in the application note entitled: Evaluation of TFF Operating Control Strategies and Scalability for Viral Vector Process Development

Table 1. Study design.

Clarified Virus Model Fee		Flux Characterization		Permeate Flux Characterization Target Initial Flux for Processing
		Process Simulations UF1 - 4x DF- 5 DV UF2 - 2.5x		Pressure and Flux Trends by Control Strategy Time, Yield, Impurity Reduction by Control Strategy Impact of Feed Loading by Control Strategy
Clarified AAV2 Feed				TMP Versus Permeate Control Pellicon® Capsule Versus Pellicon® XL 50 Cassette Scaling

Comparison of TFF Control Strategies

TMP-control systems are relatively simple, the user only needs to set feed flow and the retentate valve to give a target TMP. When the operating point is set near the plateau, the flux will be close to the maximum at every point during the UF/DF process, naturally decreasing with concentration. This approach can give the shortest process time due to higher average flux. However, if the process becomes unstable or performance is too low while using TMP control, restriction of permeate flow may be needed.

Permeate-control systems can restrict permeate flow to operate well below the minimum flux of TMP-control systems and reduce polarization at the membrane, which can improve performance such as impurity reduction. However, permeate-control systems require additional control elements beyond those used in TMP-control systems (permeate pump or control valve) and could result in an exponential rise in TMP at the critical flux, potentially preventing run completion due to feed pressure or TMP limitations.

In this study, based on 300 kDa membrane comparisons with Pellicon^® Capsule and Pellicon^® XL 50 cassette, TMP-control operation demonstrated a flux (or time) advantage as high as two-fold, which can reduce the required membrane area in half. Permeate-control operation demonstrated a higher impurity removal advantage. Results for virus yield were mixed. The yield was similar for AAV2 when using both control strategies, but yield was generally higher for the permeate-control model studies with bacteriophage.

Scaling within Pellicon^® Capsule and Cassette Families

Scaling of capsules and cassettes was demonstrated by comparing flux, TMP profile, yield, and impurity reduction using a virus model feed and AAV2. The family of Pellicon^® TFF filters enables linear scaling over a wide range of feed batch sizes. Table 2 shows an example of how TFF batches up to 2,000 liters can be handled with single-use Pellicon^® Capsules. If scaling to larger, multi-use batches is required, switching to Pellicon^® 2 cassettes with the same membrane and screen combination can be done seamlessly.

Table 2. Scale-up sizing with Pellicon^® TFF filters. Example based on Application Note data (4x/5 DV/2.5x), except where estimates were extrapolated (*).

			Batch Size (L) at Loading (L/m2) of
Filter	Total Area (m2)	TFF System	35	120	150
Pellicon® XL 50 Cassette	0.0050	Cogent® Lab System	0.18	0.60	0.75
Pellicon® Capsule	0.1	Cogent® Lab System	3.5	12	15
	0.5	Mobius® TF2S System	17.5	60	75
	1	Mobius® TF2S System	35	120	150
	1.5	Mobius® TF2S System	52.5	180	225
	3	Mobius® TF2S System	105	360	450
	9	Mobius® TFF 80 System	315	1080	1350
	13.5	Mobius® TFF 80 System	472.5	1620	2025
Pellicon® 2 Cassette	80	Cogent® Process Scale System	2800	9600	12000
Permeate control process time (h)			1.8	6.2*	7.7*
TMP control process time			0.9	4.3	6.1*

Comparison of Pellicon^® Capsules and Hollow Fibers for TFF in Viral Gene Therapy Manufacturing

The following study evaluated filter performance during the TFF1 step while using a virus model feed. Pellicon^® Capsules and its scale-down filter, Pellicon^® XL 50 cassettes, were evaluated in this study and compared with two hollow fiber modules, HF-A (0.5 mm/60 cm) and HF-B (0.5 mm/20 cm), in terms of performance.

The complete set of methods and experimental results can be found in the application note entitled: Pellicon^® Capsules versus Hollow Fibers for Ultrafiltration/Diafiltration (UF/DF) in Viral Gene Therapy Manufacturing

Sizing Scenarios and Process Impact Analysis

The results of a 300 kDa permeate-control process simulation were applied to build two theoretical sizing scenarios using Pellicon^® Capsule and the hollow fibers:

• Scenario 1 assumes the same loading is used for all filters. This scenario compares run times and maximum possible VCF for a 100 L batch at 35 L/m² loading.
• Scenario 2 restricts the time required to complete the TFF process. This scenario compares filtration area required and maximum possible VCF for processing a 1000 L batch in 4 hours, which requires variable loading.

Results from Scenario 1 are shown in Table 3. When using the same loading (35 L/m²), the Pellicon^® Capsule runs 2.9x faster than hollow fiber HF-A at the same normalized crossflow rate. The capsule is still 1.7x faster even when hollow fiber HF-B uses triple the capsule crossflow rate. Although HF-B was able to save some time compared to HF-A, the higher crossflow rate reduces the hollow fiber system’s maximum possible VCF by a factor of ~4x.

Table 3. Modeling of a 100 L scale scenario for processing a TFF1 batch at 35 L/m² loading based on flux data from the 300 kDa simulation study.

Scenario I: 100 L batch, 35 L/m2, 300kDA, permeate control
Parameter	Pellicon® Capsule 5 LMM	HF-A 5 LMM	HF-B 15 LMM
Area (m2)	2.9	2.9	2.9
Run time (hr)	1.9	5.5	3.3
Run time normalized to capsule	1.0	2.9	1.7
Max VCF	33	33	8.3
Feed flow (L/min)	15.8	14.8	43.8
Mobius® TFF system	TF2S	TF2S	Flow rate too high
Note: Calculated area is before applying a safety factor. Maximum VCF is theoretically based on the system minimum recirculation volume, which depends on crossflow rate and resulting piping. The Mobius® TFF system provides permeate control with automated valve.

Scenario 2 results are shown in Table 4. To obtain the same process time (4 hours), the Pellicon^® Capsule only required 35% of the filtration area and 37% of the feed flow of the hollow fiber module when the filters operate at the same crossflow rate—i.e., compared to the HF-A case. If the crossflow rate of the hollow fiber is tripled to increase its flux (HF-B case), the capsule required only 58% of the hollow fiber filtration area and 21% of its feed flow. It should also be noted that the higher feed flow of the hollow fiber system reduces the maximum possible VCF ~4x.

Table 4. Modeling of a 1000 L scale scenario for processing a TFF1 batch in 4 hours based on flux data from the 300 kDa simulation study.

Scenario II: 100 L batch, 4-hour process, 300kDA, permeate control
Parameter	Pellicon® Capsule 5 LMM	HF-A 5 LMM	HF-B 15 LMM
Loading (L/m2)	73.2	25.4	42.3
Area (m2)	13.7	39.4	23.7
Area normalized to capsule	1.0	2.9	1.7
Max VCF	40.0	22.2	11.1
Feed flow (L/min)	75.7	204.6	362.4
Mobius® TFF system	TFF 80	Flow rate too high	Flow rate too high
Note: Calculated area is before applying a safety factor. Maximum VCF is theoretically based on the system minimum recirculation volume, which depends on crossflow rate and resulting piping. The Mobius® TFF system provides permeate control with automated valve.

Overall, the results of the study showed that hollow fiber modules give lower flux at a given shear and as a result, a much higher membrane area or processing time is required compared to Pellicon^® Capsules. While crossflow rates of hollow fibers were increased to achieve a higher production rate, flux remained lower than Pellicon^® Capsules and the resulting system may be limited in achieving high concentration targets.

In contrast, Pellicon^® Capsules achieved higher and more stable flux than hollow fibers, resulting in a more efficient TFF process while delivering a comparable virus yield. The resulting system is also smaller, which enables higher concentration targets.

Related Content

Application Note: Scalable Purification of Plasmid DNA: Strategies and Considerations for Vaccine and Gene Therapy Manufacturing
Technical Brief: Evaluation of TFF Operating Control Strategies and Scalability for Viral Vector Process Development
Application Note: Pellicon^® Capsules versus Hollow Fibers for Ultrafiltration/Diafiltration (UF/DF) in Viral Gene Therapy Manufacturing
Tech Note: Tangential Flow Filtration (UF/DF) of Plasmid DNA
Product Overview: Ultrafiltration/Diafiltration (UF/DF) of Adeno-Associated Viruses (AAV) with Pellicon^® Capsules and Pellicon^® XL 50 Cassettes
Podcast Transcript: Interview - Shear ignorance? Think again: breaking the perception of shear within viral vector manufacturing

Request Information

If interested in any of our upstream platforms above, please complete and submit the webform below for more information and begin bringing your gene therapy to life.

Fields indicated by a * are mandatory.