Rising Stars of Separation Science: Valentina D’Atri

Column, May 2022, Volume 18, Issue 05
Pages: 17–22

Q. When did you first encounter chromatography and what attracted you to the subject?

A: It was in February 2016, when I was recruited by Jean-Luc Veuthey and Davy Guillarme as a postdoctoral researcher at the University of Geneva. The goal of the project was the characterization of biopharmaceutical proteins by coupling several chromatographic methods to mass spectrometry (MS). Although I had no previous experience with chromatography, I was motivated to challenge myself in the field, and it seems that I was motivated enough for them to accept me into the group. Szabolcs Fekete gave me my first “practical lessons” in front of a high performance liquid chromatography (HPLC) system and introduced me to the elegance and the beauty of chromatographic separations. It was only after a while that I realized that there would be no better place to debut with this technique! What attracted me most about the subject was its versatility; it is simply marvellous how the results achieved with different chromatographic modes can be combined to understand the specific features of the molecules being analyzed. Although everything may seem predictable it is not just about a mobile phase passing through a column, it is about interactions, partitioning, adsorption, and the deep understanding of all these mechanisms.

Q. Can you tell us more about your Ph.D. thesis?

A: My Ph.D. thesis was focused on G-quadruplexes, some particular DNA secondary structures that have a natural propensity to self-associate in three‑dimensional scaffolds. These higher-ordered DNA structures exhibit a dramatic thermal stability and represent a suitable nucleic acid scaffold for DNA‑based nanostructures and therapeutic oligonucleotides (aptamers). During my Ph.D. research studies, I focused on both applications. I identified and characterized some DNA-based nanostructures with a fixed and tailored length, in addition to potentially therapeutic aptamers endowed with anti-HIV activity. In the latter case, the aim was to identify the structural features required for the aptamer biological activity and succeed in improving them. It was therefore necessary to exploit the synthesis and the structural characterization of quadruplex‑forming oligonucleotides, the analysis of the sequence-specific thermodynamic stability, the physical‑chemical properties and the structural features of resulting G-quadruplexes, and evaluate the binding properties to the selected proteins. I really enjoyed my Ph.D. thesis; I had the opportunity to learn different techniques and be supported by highly competent mentors.

Q. What chromatographic techniques have you worked with?

A: I started with hydrophilic interaction liquid chromatography (HILIC) and the comparison of its performance with reversed-phase LC for the analysis of biopharmaceutical proteins performed at the subunit level. Then, I moved to non‑denaturing techniques, such as size-exclusion chromatography (SEC) and ion-exchange chromatography (IEC), for performing the analysis at the intact protein level. In all cases, my goal was the direct coupling to MS for the comprehensive characterization of post-translational modifications (PTMs) and critical quality attributes (CQAs) associated with biopharmaceutical proteins.

Q. The success of therapeutic monoclonal antibodies (mAbs) has led to the development of many new antibody-based drug formats. What are some of the challenges involved in ensuring that the final product produced is the intended drug?

A: The field of antibody-based drug formats has practically exploded in recent years, with biosimilars, antibody-drug conjugates, multi‑specific antibodies, and Fc-fusion proteins filling the pipelines of the pharmaceutical industries to get into the market. To ensure that the final product is the intended drug, the challenges are virtually the same: a complete characterization of the structural features associated with the micro-heterogeneity of these formats. Unfortunately, there is no single LC mode that allows for this characterization. The use of different chromatographic methods is necessary and the complementarity between the modes is the key to overcoming these challenges.

Q. Post-translational modifications of the drug product can result in changes to the molecular surface charge distribution, with IEC and cation‑exchange chromatography (CEC) being key techniques to monitor this. In your recent paper you mention “salt‑mediated pH gradients” as a powerful tool to achieve better separation efficiency (1). Could you explain in more detail what this entails?

A: In CEC, the separation is performed with a negatively charged stationary phase that retains ionic compounds, allowing their elution through two different mechanisms: i) via salt-gradient mode, by increasing the salt concentration to weaken the ionic interactions between the protein and the stationary phase; or ii) via pH-gradient mode, by increasing the mobile phase pH while keeping a constant ionic strength to change the charge of the proteins (based on their isoelectric point [pI]). A “salt-mediated pH gradient” can be considered as a boosted pH-gradient mode that is basically obtained by using a mild salt gradient in combination with a pH-gradient mode. This approach allows the cover of a wider range of protein pI and therefore improves the separation of proteins that can be simultaneously analyzed.

Q. When attempting to discover the cause of the changes to molecular surface charge distribution your paper mentions that MS would be the
go‑to detection method, and in the past would be paired with reversed‑phase LC and HILIC, what has changed in this regard and why?

A: The mobile phase composition of denaturing chromatographic techniques such as HILIC and reversed-phase LC allows their direct coupling to MS, although at the expense of losing the native conformation of proteins. On the other hand, CEC is a non-denaturing technique that allows the therapeutic proteins to keep in their native folded state while separating charge protein variants. Historically, CEC was performed with a high concentration of nonvolatile salts in the mobile phases and was therefore incompatible with MS. In these conditions, apart from being able to assign a given variant as acidic or basic by reference to its elution time in relation to the main peak, it was not possible to identify which variation was responsible for the change in charge. MS analysis is essential for such identification and was obtained after performing the desalting of the CEC‑collected peaks. Among the first attempts at obtaining the coupling of CEC to MS was multidimensional chromatography (mD-LC). This could have been a solution, but the approach was not straightforward to implement and not well adapted for routine applications. For these reasons, MS-compatible mobile phases for CEC applications have emerged with the aim of obtaining a direct coupling of CEC to MS for allowing unbiased and straightforward characterization of charge variants. Different solutions have been proposed by several research groups and it is now clear that CEC–MS is a real possibility.

Q. Your recent publication details a straightforward and rapid CEC–MS method that allows the separation and identification of several charge variants (1). What challenges did you face in creating this method and what benefits does it offer over alternatives?

A: Our goal was to develop a generic CEC–MS method by using a compact benchtop time-of-flight mass spectrometer (TOF-MS). The first challenge to overcome was to find a compromise between the LC performance and the MS sensitivity. Indeed, when using mobile phases consisting exclusively of volatile salts, it should be considered that they may have a lower buffering capacity compared to mobile phases containing nonvolatile salts. In addition, the LC setup should be compatible, with a proper MS ionization process to avoid the risk of obtaining low signal intensities. Therefore, we optimized the chromatographic conditions by testing various buffers and column dimensions to select the most suitable mobile phase in terms of pH response, buffer stability over time, MS compatibility, and in terms of efficiency versus MS compatibility for the column. Then, we took care of the second challenge: the CEC–MS coupling. Honestly, we expected that it would work on the first try and that a slight optimization of the MS source conditions would be enough to achieve a good ionization of the proteins, but we had to change our minds. The first attempts at direct CEC–MS coupling were not at all satisfactory; the MS signal had very low values of intensity and resolution, and it took a moment to realize that everything depended on the way we prepared the mobile phases. Indeed, metal contaminants from the water source and the glassware, including all the laboratory glassware, mobile phase bottles, and vials, were mainly responsible for the loss of MS resolution and sensitivity. It seemed to be a small detail, but it was not obvious. We then switched to trace metal certified thermoplastics, avoided glass-bottled water, and we managed to overcome the problem of the MS sensitivity. Having solved this problem, everything was smooth and easy to perform. The optimized CEC–MS setup allowed CEC analysis in less than 10 min, and in particular the simultaneous separation and identification of several charge variants by MS.

Regarding the benefits of this CEC–MS method over alternatives, I think that a strength is that it could be easily implemented in another laboratory. Above all, it does not need an MS technology offering an exaggerated resolving power; our method has been developed on a compact benchtop MS system consisting of a TOF-MS offering a resolving power of 10,000.

Q. Another paper you published recently focused on therapeutic Fc‑fusion proteins (2). What are these exactly and what therapeutic uses could they have?

A: Fc-fusion proteins are a special mAb‑based format generated by the fusion of the fragment crystallizable (Fc) domain of an immunoglobulin G (IgG) with a biologically active domain that may be represented by a peptide, a cytokine trap, an extracellular domain (ECD) of natural receptors, or even a recombinant enzyme. The strategy behind this particular design relies on the fact that by coupling a biologically active domain to an Fc‑IgG, the size of the resulting construct can be increased above the threshold for kidney filtration (60 kDa) and therefore the circulation time of the active domain can be enhanced by avoiding the renal clearance. Therefore, the therapeutic potential of the pharmacologically active moiety would be increased, together with the general stability and solubility of the construct directly linked to the presence of the Fc‑domain. Ideally, any active domain suffering from low circulation time could be fused to an Fc-fragment to generate an Fc-fusion protein. For this reason, the therapeutic uses can be really different, from providing a replacement therapy for haemophilia A/B to the treatment of rheumatoid arthritis, plaque psoriasis, or colorectal cancer.

Q. What are the analytical challenges that therapeutic Fc-fusion proteins present?

A: Given the diverse nature of the biologically active domain constituting the Fc-fusion proteins, the analytical challenges can indeed be sample-dependent, even though the main goal is generally to obtain a complete characterization of the PTMs. Among the PTMs, I think that the most challenging one would be the characterization of the glycan profile. Indeed, some Fc-fusion proteins, such as etanercept, abatacept, and belatacep, contain up to six N-glycosylation sites, aflibercept can have up to 10, and conbercept up to 14. In addition, O-glycans may also be present, as in the case of etanercept, which can have up to 26 O-glycans. It is therefore easy to imagine the complexity of these glycosylation profiles and how they can be tricky to analyze compared to mAbs, which generally contain only two N-glycosylation sites and no O-glycans.

Q. You recently published a paper on bispecific antibodies (bsAbs) (3). What exactly are bsAbs and what do they offer that other biopharmaceuticals
do not?

A: Bispecific antibodies (bsAbs) combine the antigen recognition sites of two (or more) antibodies in a single protein construct, and therefore allow the targeting of two (or more) different epitopes either on the same or on different antigens. As an example, emicizumab is a bsAb with a mAb‑shaped structure consisting of two identical light chains (LCs) and two different heavy chains (HCs) able to recognize two different targets, namely the activated coagulation factor IX (FIXa) and the coagulation factor X (FX). By using this double recognition, emicizumab can mimic the function of the plasma clotting factor VIII (FVIII) that is generally missing in patients with haemophilia A. So, without having any homology to the native FVIII, emicizumab acts as cofactor mimetic and is used for the routine prophylaxis of patients missing FVIII.

Q. What unique analytical challenges do bsAbs present?

A: The correct chain-association is the most critical challenge to monitor during bsAbs development and production. The higher the number of antigen recognition sites, the higher the probability of mismatched chains. Therefore, beyond the canonical analytical characterization that is applied to mAb-related formats, the correct chain‑association is the dedicated challenge related to bsAbs.

Q. What are you currently working on?

A: I am currently having fun with “exotic” mAb-related formats including multi‑specific antibodies and immunocytokines. In parallel, drug formats as therapeutic oligonucleotides and viral vectors for gene therapy have attracted our attention because they represent the pharmaceutical applications of a future that is already here. From an analytical point of view, these new research fields come with completely different and exciting challenges compared to the field of biopharmaceutical proteins. Therefore, beyond mAb-related formats, I am also currently working on the development of LC–MS analytical methods for the characterization of DNA/RNA oligonucleotides and viral vectors, and given the speed at which these fields are evolving, I imagine that I will not have time to get bored!

References

  1. A. Murisier, B.L. Duivelshof, S. Fekete, J. Bourquin, A. Schmudlach, M.A. Lauber, J.M. Nguyen, A. Beck, D. Guillarme, and V. D’Atri. J. Chroma. A 1655, 462499 (2021).
  2. B.L. Duivelshof, A. Murisier, J. Camperi, S. Fekete, A. Beck, D. Guillarme, and V. D’Atri, J. Sep. Sci. 44, 35–62 (2020).
  3. B.L. Duivelshof, A. Beck, D. Guillarme, and V. D’Atri, Talanta 236, 122836 (2022).

Valentina D’Atri is a Research and Teaching Fellow in the Analytical Sciences Department of the School of Pharmaceutical Sciences at the University of Geneva in Switzerland. She studied pharmaceutical biotechnology sciences (M.Sc. and B.Sc.) and obtained her Ph.D. in industrial and molecular biotechnologies (2013) from the University of Naples Federico II, Italy. She had her first postdoctoral experience (2013–2015) in the group of Valérie Gabelica (INSERM, Bordeaux, France), where she was involved in projects related to the structural characterization of DNA structures by linking ion mobility–mass spectrometry to molecular modelling. In 2016, she joined the group of Jean-Luc Veuthey and Davy Guillarme at the University of Geneva, where she is now currently working. Her interests and research activities focus on the development of cutting-edge LC–MS analytical workflows for the detailed characterization of innovative therapeutic drugs, such as biopharmaceutical proteins (monoclonal antibodies, antibody-drug conjugates, bi/tri-specific antibodies, Fc-fusion proteins), therapeutic oligonucleotides (ASO, siRNA), and viral vectors (AAV). She has currently authored over 50 peer-reviewed contributions, including articles and book chapters.