Cadence™ system: Automated semi-continuous low pH virus inactivation on a single-use mixing platform

The pH probes are a key element of the virus inactivation system. In a continuous process, we need the probes to remain precise and accurate as pH is a critical process parameter for achieving sufficient virus inactivation (2). Calibrating the probes in a single-use system can also present a challenge: it’s inconvenient to pause operations to use a probe calibration solution in the flow path of the product, especially when the product is part of a continuous process.

The Cadence™ virus inactivation system includes Mettler Toledo pH probes, which are calibrated and autoclaved before sterile installation using an aseptic bellows design. This process results in a short ‘wetting in’ period for the probe to read accurately after being autoclaved. As shown in Figure 2, this manifests as a decreasing pH offset between the pH of the solution and the probe’s reading, as the probe becomes wetted.

Fig 2. Graph showing the ‘wetting in’ time for the pH probes after autoclaving against the deviation from actual pH (pH offset). Data was tested in triplicate and the error bars represent ± SD.

In testing, the pH probe proved to be reading within tolerance after a wetting period of approximately 10 min. However, in some cases it may be necessary to adjust the pH probe to calibrate against drift in the response. To address this, we designed the system to allow a single point offset of the pH probe to improve measurement accuracy.

We recommend making this offset whilst the mixer is in the filling phase, preferably between elutions, before the first titrations take place. The system enables you to take a sample via a sampling port, and the pH transmitter software allows the offset to be taken against an earlier time point.

This means you can measure the pH of the sample offline, and apply the offset based on the time the sample was taken, even if the pH in the tank has changed since. To ensure stability, the offset measurement should be taken at least 10 min after the probe is fully covered in liquid.

The reliability of the pH probe measurement is a key requirement for a continuous VI system. To evaluate this, we tested probes performance over time in a way that simulated the actual VI process. An automated test rig was designed to assess the probes, by transferring the probes between two test solutions:

1. Human IgG (10 g/L) in 20 mM acetate, pH 3.5
2. Human IgG (10 g/L) in 20 mM Tris, pH 8.0

These simulate a typical product at the VI hold step (pH 3.5) and mimic the pH after neutralization (pH 8.0), respectively. The test rig was designed to move the probes through the following sequence, mimicking the wetting and dewetting cycles as the biocontainers empty and fill.

1. Immerse in Solution 1 (pH 3.5) for 2 h
2. Transfer to Solution 2 (pH 8.0) for 15 min
3. Expose to air for 15 min (mimicking the time the probe would be uncovered while the tank is emptied the next elution arrives)
4. Return to Step 1

Reference probes were used to monitor the pH of the two solutions over time. This experiment was performed for a total duration of 48 h. Results are shown in Figure 3.

Fig 3. Performance of the pH probe vs reference probes over 48 h, in an automated process designed to mimic the low pH virus inactivation step.

We can see from the data that the probes exhibited minimal drift through the course of the experiment. Measurements made at the beginning of the experiment, midpoint (24 h) and at the end (48 h) show that the pH probes accuracy remains well within 0.1 pH units (see Table 1).

Table 1. pH readings from test probes vs reference probes for the beginning, middle and final low pH hold step in a 48 h test

The Allegro™ 50 L single-use mixer is designed to provide high-performance mixing for a wide range of applications, from upstream processing to formulation and final fill. We offer a specific model of the Allegro™ 50 L mixer dedicated to the Cadence™ VI system, which features safety interlocks with the main control system and easy sampling.

Due to its efficient mixing and high turn-down ratio, this mixer is especially well suited to small volume formulation applications where fast, reproducible results are required.

We wanted to evaluate mixing performance, without causing foaming. To achieve this, we tested the system with 10 g/L IgG in PBS at volumes 10, 30, 40, and 50 L in the mixer. Mixing was assessed in both the clockwise (CW) and counterclockwise (CCW) directions. For each test, we held the mixer at a certain speed for 5 min to visually assess foaming. If no foaming was observed, the speed was increased and foaming reassessed. Results are shown in Figure 4.

Fig 4. Maximum mixing speed that can be used before mixer induces foaming in an IgG solution.

After we determined the maximum usable mixer speed (where foaming does not occur), we assessed the mixing efficiency for the clockwise and counter-clockwise operation at the maximum working volume (50 L) and a lower working volume (10 L). Mixing time was determined using a conductivity probe, and measuring the time it took for the conductivity to stabilize after adding NaCl via the top port of the mixer. The data is shown in Table 2.

Table 2. Mixing times with clockwise (CW) and counterclockwise (CCW) mixer direction

Mixing times were longer with CCW operation, but still completed in under 30 s without inducing foaming. CCW was chosen as the default mixing direction because its maximum mixing speed without foaming is less volume-dependent than CW operation.

The system has two mixers that are used alternately and synchronously. To maintain continuous virus inactivation, the total process must be completed before the alternate mixer tank becomes filled with incoming elutions.

Therefore, the flow rate of incoming elutions processed by the system in a given time is determined by subtracting the titration volume from the maximum mixer volume, then dividing this value by the total time required to adjust the pH (both lowering and raising), performing the low pH hold, and emptying the mixer.

To maximize efficiency, it’s necessary to empty the mixers as quickly as possible after low pH processing. Longer emptying times reduce the effective volume that the system can process. To evaluate performance, we tested the pump at increasing speeds to determine the optimal rate for mixer emptying (Fig 5).

As the pump speed increases to 80%, the mixers can be emptied at more than 500 L/h, meaning a full mixer at the maximum working volume can be emptied in under 6 min without significant pressure increase. We do not recommend using the pump above 80%, as the liquid flow enters a turbulent state. A 6 min empty time is consistent with the overall planned capacity of the system.

Fig 5. The effect of pump speed on the flow rate at which mixers can be emptied.

The Cadence™ VI system uses two mixers. Whilst one is receiving new elutions, the other mixer undertakes the titration and low pH hold, Figure 1. This setup poses a potential risk: untreated liquid may come into contact with acidified product, resulting in contamination with active virus—a common concern with top-down entry.

To minimize this risk, we designed the system with low point product entry to minimize splashing. Acid and base are also added below the liquid level and to accommodate for this a recirculation loop is used to prevent mixing with titrants during the incubation step.

Additionally, to minimize hold-ups, the system incorporates Aquasyn valve blocks, which feature a very low hold-up design (Fig 6).

Fig 6. Valve actuator without the single-use flow kit. The image highlights swept flow path within the Cadence™ VI valve block.

To validate these design choices, as well as to demonstrate that the Cadence™ VI system is free of hold-ups, we performed a series of tests. Riboflavin was used as a visual test for hold-ups, and the bacterium Brevundimonas diminuta was selected to thoroughly evaluate the system for dead-legs. Finally, we ran experiments with the bacteriophage Phi6, to mimic an enveloped mammalian virus.

Riboflavin is a standard reagent that is used to assess a system for hold-ups (3). It has ultraviolet (UV) absorbance maxima at 222, 263, and 373 nm. The 373 nm absorbance is close to the emission wavelength of commonly used blacklights (365 nm), at which riboflavin fluoresces bright yellow. Its presence after cleaning is indicative of incomplete cleaning or hold-ups. We chose riboflavin due to its very low detection limit.

To check for hold-ups, we prepared a 0.02 g/L riboflavin solution in 10× PBS—the highest soluble concentration. Mixers were filled with the solution, drained, refilled with water, mixed for 10 s and drained again. Two operators installed four different mixer biocontainers, resulting in six different tests. Riboflavin presence was visually monitored throughout the experiments using illumination at 365 nm.

No riboflavin was found above the original liquid level, or in the upper portions of the biocontainer, giving high confidence that the no splash back, low foaming design of the recirculation loop and entry points have eliminated the risk of carry-over of active virus.

In two tests, we observed a small amount of riboflavin in folds in the biocontainer after draining and before the water rinse. This is expected with single-use biocontainers, where complete draining isn’t always achievable. In terms of VI, this means a small amount of inactive elution will remain in the biocontainer and be subjected to a second treatment.

The residual volume in the biocontainer and tubing flow kit is approximately 100 mL. After refilling and subsequent emptying of the biocontainer, no riboflavin could be detected.

Additionally, we assessed potential hold-ups in the rest of the flow kit with riboflavin. Photos of the sampling port in the recirculation loop, before and after recirculation, confirmed no hold-up of the riboflavin could be detected.

Fig 7. (A) Image of sampling port with riboflavin; (B) Image of sampling port after buffer is pumped through the recirculation loop.

To further investigate potential hold-ups in the system, we selected the bacterium Brevundimonas diminuta (ATCC 19146) as a surrogate for mammalian virus. Bacterial testing includes increased sensitivity, allowing detection of one single colony forming unit (CFU) in any given volume, and enabling sampling of larger volumes by collecting bacteria (via filtration) before assaying.

Testing was performed over two continuous low pH cycles in a single mixer. After both cycles completed, we added sterile tryptic soy broth (TSB) to the mixer via the top port. This was aseptically collected, and 2 L (1/3 of the volume) was analyzed by membrane filtration to assess the potential for carry-over of viable organisms.

This test used a lower pH (2.5) than is typically used for VI as B. diminuta is not reliably inactivated at pH 3.0. Using this low pH ensured complete inactivation of the bacteria in direct contact with the low pH conditions. By sampling a large volume of growth media recovered post-acidification, we were able to thoroughly investigate the possibility of bacteria being held up in the system at a very low detection limit of 1 CFU/2 L.

Test method

1. Fill the mixer with 10 L of 50 mM citrate phosphate buffer spiked with B. diminuta to a target concentration of 5 ×10⁶ CFU/mL. Retain a sample to determine the initial starting concentration (input).
2. Perform VI at pH 2.5 for 60 min.
3. Neutralize by adding 1 M Tris base to pH 4.5.
4. Pump out liquid (through transfer pump pathway and standard VI method) and aseptically collect into a suitable sterile container (e.g., single-use biocontainer or equivalent).
5. Repeat Steps 1 to 3 to complete two VI cycles in one mixer.
6. Fill mixer 1 with 6 L of sterile TSB using the top port. Pump the solution through the recirculation loop.
7. Drain using the transfer pump and aseptically collect into a suitable container (post cleanliness).
8. Test for B. diminuta: Pass 2 L of the collected TSB through a sterile (0.2 µm rated) recovery membrane using vacuum filtration. Once the volume has passed, aseptically transfer the recovery membrane to a trypticase soy agar (TSA) plate and incubate at 30 ± 2°C for 7 d. After incubation, examine the recovery membrane for any colonies.

No titration control (determines viability changes due to mechanical shear or effects unrelated to acidification):
• Operate the system for a cycle, but with a ‘mock’ titration, which does not add acid.

Positive control (pH-dependent viability):

• Mix B. diminuta spiked IgG in beaker for 1 h.
• Test for viable B. diminuta by performing serial dilutions and testing undiluted volumes via membrane filtration.
• Plate onto TSA and incubate at 30 ± 2°C for 7 d.

Negative control (standing sample viability):

• Perform VI of B. diminuta spiked IgG in a beaker for 1 h.
• Test for viable B. diminuta.

Table 3. Results of the B. diminuta experiments for dead-leg testing

The experiment confirms that the system does not contain hold-ups. No B. diminuta was detected after the two VI cycles, or the post-cleanliness washout, despite thoroughly testing the complete flow kit (including the recirculation loop).

The positive control, which excluded acidification, was unaffected—indicating that mixing, pumping and time do not inactivate B. diminuta. These results confirm the system is effective for low pH inactivation and does not contain any hold-ups that can lead to contamination of inactivated with non-inactivated elutions.

To demonstrate the effectiveness of the system, we used a bacteriophage Phi6 virus analog. We chose this organism to simulate the inactivation kinetics of enveloped mammalian viruses and allow a test that closely resembles the end-user application (see Table 4). Using a bacteriophage model allows us to determine a specific log reduction value (LRV), providing a quantifiable measure of virus inactivation performance.

Table 4. Properties of the bacteriophage Phi6

*Based on genome and presence or absence of a viral envelope.

The bacteriophage testing was performed in a similar way to the B. diminuta testing. Two low pH cycles were performed, with the key differences being that the low pH inactivation for Phi6 was performed at pH 3.5, and the system washout was performed with PBS. Samples were analyzed for Phi6 using plaque assays with Pseudomonas syringiae, followed by overnight incubation at 37°C. Results are shown in Table 5.

Table 5. Properties of the bacteriophage Phi6

No Phi6 was detected after low pH virus inactivation, indicating the system is effective for inactivating bacteriophage and that there was no significant contamination of inactivated with non-inactivated elution. We performed two cycles in a single biocontainer, with no evidence of carryover of bacteriophage between cycles. Additionally, a biocontainer wash after the second inactivation cycle showed no remaining active bacteriophage.

A negative control, run with the same pump speeds and mixer speeds, but without any low pH inactivation, was performed to confirm the bacteriophage was not inactivated over time or due to shear forces from mixing and pumping. This control confirms that the observed reduction is due to the low pH step and that the system is effective for virus inactivation.

We ran the system continuously for 24 h to confirm it can operate reliably without user input for prolonged periods of time. To mimic chromatographic elutions from protein A columns, we created an IgG test solution. We included 0.1× PBS as elutions from protein A columns typically ‘front’ ahead of the elution buffer and the final wash buffer is often PBS. The elution buffer is typically 50 mM acetic acid and the elution pool is normally around pH 4.0.

To mimic this, we titrated the solution to pH 4.0 with acetic acid. A pump was programmed to supply a semi-continuous feed-stream to the VI system, simulating the output of a continuous chromatography system. The pump flow rate was set to 280 mL/min, delivered cyclically with 8 min pumping followed by 4 min of not pumping, for a delivery of 10 L/h.

Preparatory experiment to determine titration setpoints

Before operating the system, we ran a test titration in order to develop a suitable titration strategy. A 100 mL mock elution was titrated down to pH 3.5 with 1 M acetic acid, and back up to pH 8.0 using 1 M Tris. We determined that around 2.5% of the original volume of acetic acid (2.5 mL) was required to achieve pH 3.5 and 5.2% of was needed to return the solution to pH 8.0.

This led us to the following titration strategy, shown in Table 6. The values are typical of the data set that you will need to generate to enter into the system when commissioning for a particular product.

Table 6. Titration strategy with 1 M acetic acid and 1 M Tris base

Transfer of titration strategy to the VI system

The titration was performed in three defined steps.

1. A bolus of titrant was added, calculated as a percentage of the weight in the mixer. For rapid pH adjustment, we added approximately half of the expected volume in this first step.
2. A smaller volume of titrant was added, and the pH checked after a period of mixing. This step was repeated until we reached the Step 2 setpoint. During acidification, the Step 2 setpoint was set relatively closely to the final desired pH, as acetic acid is a weak acid and the pH of the solution approaches that of the acid itself. When titrating back up, Tris (a much stronger buffer at pH 8.0) required more careful adjustment to avoid overshooting.
3. Smaller amounts of titrant were added, mixed, and the pH checked. Further additions were made until the final Step 3 checkpoint is reached. These additions were kept minimal to avoid overshooting the pH setpoint.

To begin the test, we installed a new flow kit and initiated the VI sequence. This opened a flow path and set the VI system to wait for elution or product pool to arrive in the mixer. pH probes were fully immersed in the elutions for approximately 10 min before a single setpoint correction was performed to calibrate the system. The 4 min interval between elutions supplied suitable time for this correction.

A sample was taken from the sampling port 30 s after the end of the elution to ensure homogeneity. A timestamp was created on the pH transmitter immediately after sampling. We measured the pH of the sample offline, and compared to the pH recorded by the system at the timestamp. This ensures the pH offset remained valid even if the system pH changed between sampling and correction.

Following this calibration, no further adjustments were needed during the 24 h operation. Samples were taken periodically to ensure that system pH readings remained accurate and represented the actual pH of samples in the mixers, with the probes staying within ± 0.05 pH units of the reading from the externally calibrated probe (see Table 7). This is within the desired accuracy of ± 0.1 pH units.

Table 7. Data comparing pH data from the internal system probes and external probe measurement of samples

pH probe performance and titration accuracy

Figure 8 shows data from the two pH probes, one in each mixer. The incoming material enters at pH 4.0, then is titrated down to pH 3.5, held for 60 min, then titrated up to pH 8.0. Once the final pH is reached, the mixer is emptied and the probe remains exposed to air until the next elution cycle begins.

The pH traces appear as expected, with no overshoots or process deviations observed. During each titration, product was pumped through the recirculation loop and the mixer was operated to maintain homogeneity. The titration time depends on the strategy used—a balance between hitting the set-point accurately and titrating quickly. In this test, we prioritized titrating quickly, with most of the titrations completed in under 5 min. All final pH values were within ± 0.07 pH units of the setpoint.

To verify probe accuracy, we took samples periodically and compared internal pH probe readings with an externally calibrated pH meter. The internal probe always read within ± 0.5 pH units of the external probe, confirming accuracy of the internal probe through the course of the 24 h experiment.

After titration down, the system was programmed to pause for 1 min before initiating the low-pH hold or emptying the tank. This confirmed pH stability, and no drift was observed during the pause. Additionally, investigations of the flow kit after the test showed no apparent damage or leaks during the 24 h of operation.

Fig 8. Data showing the operation the Cadence™ virus inactivation system via the two built-in pH probes.

1. ASTM E2888 – 12. Standard practice for process inactivation of rodent retrovirus by pH.
2. Handbook for critical cleaning: Applications, processes, and controls (CRC Press, 2011).
3. A mAb case study: CMC Biotech working group.