Introduction
Hydrophobic interaction chromatography (HIC) is an analytical technique that utilizes the hydrophobic properties of molecules to achieve their separation1. In this type of chromatography, sample molecules with hydrophobic portion(s) interact with and bind to the hydrophobic groups (e.g., phenyl groups) attached to the column stationary phase. A buffer with a high ionic strength (e.g., 50 mM Na3PO4 with 1.5 M (NH4)2SO4, pH 7.00) is initially applied to the column to ensure desired hydrophobic interactions between sample molecules and the hydrophobic groups. Then sample molecules are eluted with an order of increasing hydrophobicity by decreasing the salt concentration (Figure 1). Compared to reversed-phase high performance liquid chromatography (RP-HPLC), HIC is a gentle method that helps maintain the biological activities of sample molecules, leading to its wide applications for biomolecule separation, especially for protein species.
Figure 1. Schematic illustration of hydrophobic interaction chromatography. Sample molecules are retained under high salt conditions and eluted with an order of increasing hydrophobicity when salt concentration is reduced. Antibody-drug conjugates (ADCs) are an emerging class of highly potent biopharmaceutical drugs designed for targeted cancer therapy. They are constructed from three components: a monoclonal antibody (mAb) specific to a cell-surface cancer antigen with self-internalization capability, a highly potent cytotoxic drug (payload), and a linker that covalently attaches the cytotoxic drug to the mAb2. By combining the exquisite specificity of mAbs and high potency of small molecule cytotoxic drugs, ADCs offer unrivaled therapeutic potential for targeted cancer therapy compared to their counterpart mAbs or free drugs by delivering significant levels of cytotoxic drug to tumor cells while limiting damage to surrounding healthy cells. The two recently FDA-approved ADCs (Adcetris® by Seattle Genetics in 2011 and Kadcyla® by Genentech in 2013) and the growing number of ADCs under clinical development have validated the idea of these “magic bullets” and are fueling rapidly increasing research interest and investment in ADCs globally3.
Though the concept of ADCs is quite simple, successfully developing these “magic bullets” is rather sophisticated and largely depends on the correct selection of each ADC component as well as the corresponding conjugation chemistry. The conjugation technique, regardless of the amino acid site selected (cysteine vs. lysine) and process used, results in an ADC molecule that is heterogeneous with respect to drug distribution (fractions of ADC species with zero, one, two, …, and n cytotoxic drugs)4. This heterogeneity not only leads to process challenges but can also impact the clinical safety and efficacy of an ADC; therefore, developing analytical techniques capable of characterizing drug distribution is critical to ADC development and control. Since small molecule cytotoxic drugs conjugated onto mAbs tend to be hydrophobic and ADC molecules with different numbers of drug molecules have distinct degree of hydrophobicity, HIC analysis is uniquely suited to monitor drug distribution in ADCs.
In the following sections, general aspects of HIC method development will be discussed first. Then HIC methods for ADC drug distribution analysis reported in the literature will be summarized and discussed. Finally, some perspectives regarding the application of HIC for ADC drug distribution analysis will be provided.
Method Development
Fundamentally, HIC obey the same principles as any other liquid chromatography-based technique (differential partitioning), but differ significantly in mobile phase selection and elution mechanism. In a HIC separation, typical steps often include equilibration, sample injection, elution, and wash5. Following equilibration of the column with a high-salt start buffer, sample molecules are injected to the column. Exposed hydrophobic regions of the sample molecules cause the molecules to partition onto the stationary phase. As the salt concentration is reduced in the elution step, sample molecules with weaker hydrophobic interactions with the stationary phase partition into the mobile phase and elute from the column. Overall, this results in the elution of sample molecules with the lowest exposed hydrophobicity first, followed by those with more exposed hydrophobicity. After all of the sample molecules are eluted, a “saltfree” wash step is introduced to remove any hydrophobic compounds which did not elute during the elution step. Based on the mechanism of elution and sequence of steps, the following parameters are typically varied in development of a HIC separation: column, salt identity and amount, temperature, and pH6,7.
Column Selection
The column is composed of hydrophobic ligands coupled to a support matrix. Many different types of columns for HIC are now commercially available from various vendors. The most commonly used hydrophobic ligands are phenyl and alkyl groups of varying chain lengths and differing hydrophobicity. For samples with low hydrophobicity, ligands with high hydrophobicity are required to ensure adequate retention. As for highly hydrophobic samples, weakly hydrophobic ligands are desirable to avoid harsh and possibly denaturing elution conditions. In addition to the ligand hydrophobicity, the ligand density also needs to be considered when developing a HIC method and a “critical hydrophobicity” (a threshold value of the hydrophobic ligand surface concentration) has to be reached before the separation6. In contrast to the hydrophobic ligands, the support matrix should be inert to minimize undesired and unpredictable interactions with sample molecules. Finally, the choice of support materials depends on the operating pressure of the system, too. Currently, kits containing small HIC columns with different hydrophobic ligand and support matrix combinations are commercially available and serve as a starting point for HIC method development.
Salt Selection
Figure 2. The Hofmeister series of cations, anions, and the relative effect of some salts on promoting hydrophobic interactions. Binding and elution in HIC are achieved by the addition and removal of salt in the mobile phase, so both salt type and concentration of the buffer play important roles in a HIC separation. Generally, large ions with low charge density (e.g., Ba2+) are strong chaotropic agents (substances capable of breaking the hydrogen bonding network between water molecules) and they can significant hinder hydrophobic interactions, whereas small ions with high charge density (e.g., PO43-) are highly anti-chaotropic and can substantially promote hydrophobic interactions5. The elution strength of an ion roughly follows Hofmeister series for cations and anions (Figure 2)5. In some cases, Ca2+ and Mg2+ may be exceptions to this as they may have specific interactions with particular proteins8. Based on the Hofmeister series, the elution strength of a salt can be predicted from its corresponding component ions (Figure 2)5. One of the most commonly used salts in stand-alone HIC separations is (NH4)2SO4 due to its high solubility (>5M in water at 20°C) and good anti-chaotropic properties. When a HIC separation immediately follows an IEC or SEC step, the salt used in the IEC or SEC can be directly used to promote hydrophobic interactions needed for HIC8. This economic design obviates tedious dialysis and buffer exchange steps. After the salt is selected, the salt concentration requires optimization. High salt concentration is desired to ensure sample molecules are sufficiently retained on the column before elution. However, the “solubility window” of the sample molecules should be assessed at various salt concentrations to avoid sample precipitation during separation. The elution step of a HIC separation can be carried out in either a gradient or a stepwise manner. Gradient elution is typically used for unknown analysis and high-resolution separations. However, after a HIC separation has been optimized using gradient elution, converting to step elution can increase efficiency (short analysis time) and reduce cost (reduced buffer consumption).
Temperature
The impact of temperature on a HIC separation is very complicated. Since hydrophobic interactions are entropy-driven, increasing temperature would improve the strength of binding and possibly achieve better resolution8. However, temperature also affects the structure, thus hydrophobicity and solubility, of sample molecules8. Taken together, it is difficult to predict how temperature impacts the separation outcome and this must be determined experimentally. In practice, working at a constant temperature (for sample, buffer, and equipment) is highly recommended and will improve reproducibility.
pH
Typically, the effect of pH is fairly small and pH values have very little significance on the final resolution of a HIC separation8. In addition, selection of buffering ions is not critical for hydrophobic interaction and phosphate buffers are most commonly used5. However, the impact of pH on proteins is sample-specific owing to their distinct charged residues and isoelectric points. Therefore, changing buffer pH values may possess the capability to modify elution profiles and achieve better separation8. When sample recovery is needed for further biological studies, the optimal pH is the one that yields the best separation and is compatible with protein stability and activity.
Antibody Drug Conjugate Drug Distribution Analysis
ADCs resulted from current conjugation techniques are typically highly heterogeneous, consisting of ADC species with different numbers of drug molecules at various conjugation sites. These species may have very different pharmacokinetic and toxicological properties and can directly affect the efficacy and safety of ADCs. ADC species with no drug attached (DAR0 species) at best are subpotent and at worst can impact the overall potency of the ADC by taking away binding sites on tumor cells from more highly drugged species. Conversely, over drugged ADCs typically exhibit higher aggregation that can be associated with increased toxicity and side effects in patients. Therefore, an analytical technique capable of characterizing and controlling ADC drug distribution is of great significance and highly desired. To this end, the use of HIC for ADC drug distribution analysis has been explored. The drug component of an ADC is typically a hydrophobic small molecule conjugated onto an exposed region of the mAb; therefore ADC species with more conjugated drug molecules have more regions of hydrophobicity, making HIC analyses uniquely suited to monitor ADC drug distribution. In practice, HIC is gaining popularity and several successful cases of using HIC for ADC drug distribution analysis have already been reported in the literature.
HIC has been used to determine the drug distribution of several ADCs containing auristatin drugs conjugated to inter-chain cysteines (mAb) and engineered cysteines (THIOMAB) through cleavable and noncleavable linkers4,9-12. Hamblett et al. from Seattle Genetics reported the use of HIC for drug distribution analysis of an ADC constructed from monomethyl auristatin E (MMAE) linked to the inter-chain cysteines of anti-CD30 mAb (cAC10) via cleavable valine citrulline (vc) linker9. An Ether-5PW column was utilized. The method consisted of a linear gradient from 100% buffer A (50 mM Na3PO4 with 2M NaCl, pH 7.00) to 100% buffer B (50 mM Na3PO4, pH 7.00/ACN/IPA, 80/10/10, v/v/v) in 50 min with the flow rate of 1.0 mL/min and column temperature of 30 °C. According to their results, ADC species with zero, two, four, six, and eight drugs per mAb were separated with high resolution. In addition, online UV/VIS was utilized to identify the separated peaks since the mAb and the drug have different Amax values. Furthermore, HIC analysis enabled the calculation of average drug-to-antibody ratio (DAR) of the ADC under study. Wakankar et al. from Genentech also presented similar results for another vcMMAE-containing ADC on a Butyl-NPR column (Figure 3)4,11.Other than these ADCs, drug distribution analysis by HIC was also demonstrated on an ADC consisting of monomethyl auristatin F (MMAF) coupled via a non-cleavable maleimidocaproyl linker to the inter-chain cysteines of anti-CD22 mAb10. In this method, a Phenyl-5PW column, buffer A (25 mM Na3PO4 with 1.5 M (NH4)2SO4, pH 6.95), buffer B (25 mM Na3PO4, pH 6.95/IPA, 75/25, v/v) were used and HIC was carried out in a gradient manner. At the same time, recent advances in ADC development have yielded ADCs with drug molecules conjugated to cysteines engineered into the mAb while the native cysteine disulfide bonds of the mAb remain intact (THIOMAB). And HIC assays for drug distribution analysis have been illustrated for this type of ADCs as well11,13.
Figure 3. Hydrophobic interaction chromatography for ADC (mAbvc- MMAE) drug distribution analysis, yielding five predominant peaks that correspond to ADC species with zero, two, four, six and eight drugs. Inset: an overly of the UV spectra of all the ADC species, normalized to 280 nm absorbance, showing the increase in 248 nm absorbance as the number of conjugated drug increases4. HIC has its own advantages and limitations with regard to ADC drug distribution analysis. Compared to RP-HPLC, HIC is gentle and non-denaturing, but unable to provide information on whether the drug is conjugated on heavy chain or light chain and incapable of separating positional isomers of ADC species with the same number of drug molecules conjugated10. In addition, HIC is typically incompatible with MS analysis due to the use of high concentrations of nonvolatile salts14. In comparison to MS, HIC allows the separation and isolation of chromatographically pure ADC species for further analysis, such as ELISA. Additionally, it is more QC-friendly and can be easily implemented and validated in a GMP laboratory. Both UV/ VIS and HIC enable the calculation of average DAR15,16, but only HIC provides information on the composition of a complex ADC sample and the exact amount of each ADC species. Besides, there are some issues associated with the determination of average DAR by UV/VIS. These issues include the absorbance of the drug overlapping with the absorbance of the mAb as well as the interference caused by the presentence of high levels of non-covalently bound free drugs, leading to over-quantitation of the drugs and relatively high average DAR values.
Perspective
HIC has been widely used in the purification and characterization of biomolecules after its development. With the introduction of ADCs in the past decade, HIC analyses have witnessed a resurgence, especially for ADC drug distribution analysis. ADCs in particular are ideally suited to analysis via HIC as they possess exposed pockets of hydrophobic moieties which serve as a mechanism of separation for HIC. Drug distribution on the ADCs is a critical parameter as DAR0 species are by definition subpotent and can even block binding sites on the tumor cells from other higher DAR species leading to reduced efficacy. Conversely, over drugged ADCs often exhibit increased aggregation and have been associated with increased toxicity. Thus HIC performs an essential analysis for ADCs, which is difficult to replicate with other analytical techniques, especially in a QC environment.
While ADCs may be ideal for HIC chromatography, the HIC method development is still challenging particularly when applied to a complex system like ADCs. Many parameters, including the column (hydrophobic ligands and support matrix), salt (type and concentration), temperature, and pH, should be taken into consideration when developing a HIC method for ADC analysis. In addition, due to limited understanding about the exact mechanism(s) of HIC and the structure diversity of ADC samples, method development for HIC is still largely empirical. Continued advancement of the understanding of HIC separation mechanism and increased availability of commercial columns will lead to wider applications of HIC, for example on lysinelinked ADCs, which have yet to be explored. At the same time, HIC will be increasingly coupled to other analytical techniques (e.g., SEC) to make a multi-dimensional chromatographic system with the potential to tackle the high heterogeneity and complexity of ADCs.
References
- Jennissen, H.P., Hydrophobic Interaction Chromatography, in eLS. 2001, John Wiley & Sons, Ltd.
- Chari, R.V.J., M.L. Miller, and W.C. Widdison, Antibody–Drug Conjugates: An Emerging Concept in Cancer Therapy. Angewandte Chemie International Edition, 2014. 53(15): p. 3796-3827.
- Guo, J., et al., Assessment of Physical Stability of an Antibody Drug Conjugate by Higher Order Structure Analysis: Impact of Thiol-Maleimide Chemistry. Pharmaceutical Research, 2014. 31(7): p. 1710-1723.
- Wakankar, A., et al., Analytical methods for physicochemical characterization of antibody drug conjugates. mAbs, 2011. 3(2): p. 161-172.
- GE Healthcare, Hydrophobic Interaction and Reversed Phase Chromatography. Handbook 11-0012-69 AA: p. 7-51.
- Jennissen, H.P., Hydrophobic Interaction Chromatography. II/Affinity Separation/Hydrophobic Interaction Chromatography, 2000: p. 8.
- Hjertén, S., Some general aspects of hydrophobic interaction chromatography. Journal of Chromatography A, 1973. 87(2): p. 325-331.
- O’Farrell, P.A., Hydrophobic Interaction Chromatography, in Molecular Biomethods Handbook, R.R. John M. Walker, Editor. 2008, Humana Press. p. 731-739.
- Hamblett, K.J., et al., Effects of Drug Loading on the Antitumor Activity of a Monoclonal Antibody Drug Conjugate. Clinical Cancer Research, 2004. 10(20): p. 7063-7070.
- Stephan, J.-P., et al., Anti-CD22-MCC-DM1 and MC-MMAF Conjugates: Impact of Assay Format on Pharmacokinetic Parameters Determination. Bioconjugate Chemistry, 2008. 19(8): p. 1673-1683.
- Xu, K., et al., Characterization of the drug-to-antibody ratio distribution for antibody–drug conjugates in plasma/serum. Bioanalysis, 2013. 5(9): p. 1057-1071.
- Hamilton, G.S., Antibody-drug conjugates for cancer therapy: The technological and regulatory challenges of developing drug-biologic hybrids. Biologicals, 2015: p. 15.
- Junutula, J.R., et al., Site-specific conjugation of a cytotoxic drug to an antibody improves the therapeutic index. Nat Biotech, 2008. 26(8): p. 925-932.
- Lichen Xiu, S.G.V., Andrew J. Alpert, Song Jin, and Ying Ge, Effective Protein Separation by Coupling Hydrophobic Interaction and Reverse Phase Chromatography for Top-down Proteomics. Analytical Chemistry, 2014. 86: p. 8.
- Chen, Y., Drug-to-Antibody Ratio (DAR) by UV/Vis Spectroscopy, in Antibody-Drug Conjugates, L. Ducry, Editor. 2013, Humana Press. p. 267-273.
- Ouyang, J., Drug-to-Antibody Ratio (DAR) and Drug Load Distribution by Hydrophobic Interaction Chromatography and Reversed Phase High-Performance Liquid Chromatography, in Antibody-Drug Conjugates, L. Ducry, Editor. 2013, Humana Press. p. 275-283.