Development of Salt Tolerant Interaction Chromatography (STIC)
A user requirements specification for a new standard in flow-through polishing must consider the base matrix and the ligands,
both of which have to be optimized individually before identifying an optimal format for the technology. Suitable models are
particularly important when specifying the design space.
Initial analysis of the factors limiting the performance of first-generation Q membranes (Sartobind Q) showed that the grafted
hydrogel layer on the macroporous support collapses at high salt concentrations and can no longer be accessed by macromolecules
and viruses.2 Therefore, we developed a second-generation base support matrix comprising a cross-linked, regenerated cellulose membrane
with ultrapores, providing a novel double-porous structure. The new matrix had a significantly higher binding capacity at
high salt concentrations and was less sensitive to increasing salt concentrations than standard Q membranes.
A recent study has shown that the salt tolerance of an anion exchange matrix is determined by the net charge of the ligand,
its molecular structure, and the immobilization density.3 It was also shown that the number of free primary amine groups significantly influences the ability of a matrix to work
in high salt concentrations, by compensating capacity-limitation through increased charge density.4 Polycations with multiple NH2-groups are efficient ligands and have been used to remove pathogens from blood.5 We therefore developed a polyallylamine ligand covalently coupled to the double-porous membrane described above, and investigated
its contaminant removal performance at different pH and conductivity values (Figure 1).
Figure 1. Structure of the covalently attached polyallylamine on a Sartobind STIC membrane
STIC Performance Data
Although acidic contaminants such as nucleic acids and endotoxins can be removed easily under most process conditions, this
is not necessarily the case for host cell proteins and viruses, especially those with a more neutral or even basic isoelectric
point. It was therefore important to identify suitable "worst case" models to mimic the problem of early breakthrough under
physiological conditions and investigate the behavior of the new material in spiking trials. The conductivity-related phenomenon
of virus and HCP breakthrough in AEX chromatography has been described for viruses, bacteriophages, and HCP.6–8
Figure 2. Log reduction value (LRV) of a standard Sartobind Q membrane (three layers, 15 cm2 total membrane area) for bacteriophages PP7 (A) and ΦX174 (B). Load 4 x 107 pfu/mL in 25 mM Tris-Cl, pH 8.0 at a flow rate of 20 mL/min as a function of an increasing NaCl concentration of 0, 50, and
150 mM (1.4, 6.7, and 16.8 mS/cm).
To investigate the effect of increasing conductivity on both the traditional Q membrane and the new polishing chemistry, we
processed phage spiking solutions comprising 4 x 107 pfu/mL in 25-mM Tris-Cl (pH 8.0) and NaCl concentrations of 0, 50, and 150 mM (1.4, 6.7, and 16.8 mS/cm). The membrane device
had a three-layer configuration with a 15 cm2 membrane area (0.4 mL membrane volume, 0.75 mm bed height). The flow rate was maintained at 20 mL/min (contact time 0.9
s) in all experiments. Sample loads of up to 1,200 mL were applied per device. In the first set of experiments, we compared
the binding of the bacteriophages PP7 and ΦX174 on standard Q membranes as a function of increasing conductivity at neutral
pH. Although the bacteriophages have a comparable diameter (24–33 nm), their isoelectric points are distinct, with PP7 being
more acidic (pI 4.3–4.9) than ΦX174 (pI 6.4–6.7).9 PP7 demonstrated no loss in binding capacity in 150-mM NaCl (16.7 mS/cm) whereas ΦX174 only bound when there was no additional
salt in the Tris buffer (Figure 2). We therefore chose ΦX174 as the low-binding model virus for our study, confirming earlier
Figure 3. Log reduction value (LRV) of Sartobind Q and Sartobind STIC membranes (three layers, 15 cm2 total membrane area) for the bacteriophage ΦX174. Load 4 x 107 pfu/mL in 25 mM Tris-Cl, pH 8.0 at flow rate of 20 mL/min as a function of an increasing NaCl concentration of 0 (A), 50 (B),
and 150 mM (C).
In a second series of experiments, the new salt tolerant chemistry was compared with traditional Q chemistry when challenged
with ΦX174 under conditions of increasing salt concentration and sample load. Although the salt tolerant chemistry achieved
a log reduction value (LRV) of five at all salt concentrations up to and including physiological conditions, the performance
of the Q membrane dropped sharply from its initial value of 4 LRVs at 1.4 mS/cm when challenged with higher salt concentrations
or higher loads (Figure 3A–C). To confirm that these results were also applicable to actual model viruses, an MVM spiking
study was carried out under high salt conditions (16.8 mS/cm), showing that no breakthrough occurred when using STIC (Table
1). The method was similarly robust when dealing with DNA and BSA (Table 1) and experiments to test the method's performance
for HCP removal under typical process conditions are in progress.
Table 1. Binding capacity and log reduction values (LRV) for Satobind Q and Sartobind STIC prototype membranes