OR WAIT 15 SECS
Szabolcs Feketeis currently a scientific collaborator at the University of Geneva. He is actively investigating aspects of retention modeling and fundamental attributes of LC. He is the 2020 winner of the LCGC Emerging Leader in Chromatography Award which recognizes the achievements and aspirations of a talented young separation scientist who has made strides early in his or her career toward the advancement of chromatographic techniques and applications. He recently spoke to us about his current research work and his career aspirations.
Szabolcs Fekete, is currently a scientific collaborator at the University of Geneva. His work focuses on finding new possibilities in protein chromatography; on characterizing therapeutic proteins, on studies of liquid chromatography (LC) column technology, and on method optimization. He is actively investigating aspects of retention modeling and fundamental attributes of LC. He is the 2020 winner of the LCGC Emerging Leader in Chromatography Award which recognizes the achievements and aspirations of a talented young separation scientist who has made strides early in his or her career toward the advancement of chromatographic techniques and applications. He recently spoke to us about his current research work and his career aspirations.
In one of your studies, you experimentally determined the impact of operating pressure on the retention of large solutes (proteins) and selectivity in reversed-phase LC (1). A huge impact was observed even in gradient elution mode, which was not expected. What prompted you to specifically investigate operating pressure as a separation variable? How is what you have discovered different from what others have thought previously?
It is well-known that working at very high pressures affects various chromatographic parameters. There is a possible complication with ultrahigh-pressure liquid chromatography (UHPLC) regarding the effect of pressure and mobile phase velocity (friction) on both retention and band-broadening.
First, Giddings showed that increased pressure could induce significant changes in molecular volume and the ability of molecules to crowd together to reduce molecular volume upon adsorption. Later on, Guiochon and McCalley reported interesting results observed in isocratic mode and mostly for small-molecule separations. However, the change in retention and selectivity caused by pressure was relatively minor when eluting small solutes (in the pressure range up to 1000 bar). Our hypothesis was that, in the case of large biomolecules, the changes of molecular conformation caused by the pressure or high flow rates (thermal effects) will play a much more important role in changing retention and selectivity. Especially when considering their “bind and elute” type retention behavior. In our work, this huge conformational change was experimentally verified both by fluorescence emission measurements and by the analysis of native and reduced forms of various proteins which possess the same molecular weight but a different conformation.
Finally, a substantial effect was observed and demonstrated that the operating pressure is a useful method variable to adjust selectivity of large molecule separations. On the other hand, this extreme effect could cause serious issues when transferring previously developed conventional methods to UHPLC method.
You have also published valuable work on the retention modeling of large proteins (monoclonal antibodies and related products such as antibody-drug conjugates [ADCs]) using computer simulation (2), and suggest a generic method development approach, along with platform methods, which could be very useful for industrial pharmaceutical laboratories. How will this new information assist a broader biomedical and pharmaceutical analytical community?
We hope that this information will make the life of practicing chromatographers easier! In the common practice of analytical R&D laboratories located in most biopharmaceutical companies, when developing a new method, several method variables are typically screened over a wide range. This is time consuming and above all unnecessary in many cases. As an example, we often see that people optimize their mobile phase temperature for antibody separations, in reversed-phase mode, over a low temperature range such as 30 to 60 ºC. When they do not see peaks in their chromatogram, they ask us what is wrong with their separation. We only need to suggest that they perform one experiment at 80 or 90 ºC, after which, miraculously their protein elutes in a sharp peak.
Therefore, we promote a generic method development approach which is based on the fact that proteins of the same, or similar, family (for example, mAbs or cys-linked ADCs) show very similar retention behavior. By knowing this, once you have developed a method for one mAb then it is possible to use these insights for another protein as it will–elute with similar mobile phase strength, show good recovery only at high temperature (75 to 90 ºC), and only elute in a sharp peak when using an ion-pairing agent and low pH. Considering these facts, at the end you only need to perform very few experiments (in a more limited range of two or three method variables) as input for further method optimization. The same input experiments can be performed for any mAb – no need to study again the impact of method variables. Therefore, to optimize the method, retention modeling (computer simulation) can indeed be helpful. It is true that in a wide range of method variables, the retention behavior of large proteins is complicated due to possible (either reversible or irreversible) conformational changes (caused by pressure, temperature, organic solvents, pH, and so forth) which impacts the molar volume and thus the retention. Consequently, accurate modeling of retention in the full range of method variables is most likely impossible. Fortunately, as explained previously, there is no need to model retention in a large design space, as only a limited range needs to be modeled. In that limited range, the most common models (the stoichiometric displacement model, linear solvent-strength model, adsorption model, and so forth) describe the retention of proteins very precisely. For such optimizations, any chromatographic modeling software (such as DryLab) or even some less specific software can be applied (for example, Statistica or even MS Excel).
It is worth mentioning that the effect of operating pressure and the heat developed by friction can also be modeled for protein separations, however this requires more experiments and more sophisticated models.
In other work you have carried out fundamental studies in which the effect of longitudinal temperature gradient on retention, caused by frictional heating, was experimentally dissociated from the combined effect of pressure and frictional heating (3). Through this work, the specific contributions of these effects to the overall retention were determined for both small and large solutes. Would you explain the significance and meaning of this work for the readers of LCGC North America?
The interesting thing here is that operating pressure and longitudinal temperature gradients caused by friction have contrasting effects on solute retention and it is not obvious how to distinguish these two effects. Frictional heat effects tend to decrease retention (in still-air ovens) while pressure inherently increases solute retention. In the past, pressure effects were mostly studied by varying the flow rate. However, within this we need to be careful as both the pressure and temperature gradients are strongly affected by changing the flow rate. Therefore incorrect (not accurate) conclusions have been drawn in several published studies. Our purpose was to dissociate these two contrasting effects and determine their individual contributions to the retention of solutes of various sizes. To realize that, we suggested two sets of experiments: one performed at constant inlet pressure and at varied flow rates, and the other at varied inlet pressure.
We saw that friction related effects were more important for small molecule separations whilst pressure effects were much more significant for protein separations. Insulin was an attractive example as we could clearly see the decrease of retention in constant inlet pressure mode and the increase of retention in variable inlet pressure mode when increasing the flow rate. With this information, the developed heat power and outlet column temperature could also be estimated.
What issues and problems would you define as previously ignored or neglected specifically in the field of UHPLC, and more generally in the separation sciences? What is your vision for improvements that could be made in separation instrumentation, columns, data processing algorithms, or high-speed computing power?
One of the main limitations in liquid chromatography today is still the significant extra-column volume contribution of commercial chromatographic systems to band broadening. I believe that current instruments have reached their limits and therefore one might expect some completely new integrated system designs (for example, a slot for the column, on-column injection and on-column detection) alongside developments of column technology. The adiabatic isolation of the stationary phase would probably be a huge step forward in UHPLC and supercritical fluid chromatography (SFC) and the idea and design suggested by Fabrice Gritti–called a vacuum jacketed column–is obviously marvelous. Through this the detrimental band broadening caused by frictional thermal effects could be eliminated and I would expect the commercialization of such columns to be seen in the near future.
Now it seems that column wall coatings and using frits made of inert materials are also interesting, especially for protein separations, to eliminate non-desired secondary interactions. For protein separations, it is also a hot topic today to couple non-denaturing modes, such as size-exclusion chromatography (SEC), ion-exchange chromatography (IEC or IEX), or hydrophobic interaction chromatography (HIC) to mass spectrometry (MS) detection. However, those chromatographic modes are inherently not compatible with MS and therefore there is a need for finding novel volatile mobile phase buffer systems. Very interesting recent work by Marry Wirth should also be mentioned here, namely an MS-compatible alternative to HIC called native reversed-phase liquid chromatography (nRPLC).
Improving data processing algorithms and computing power today, in my opinion, is not required. However, it would probably be interesting for very fast separations using very short columns and improving online data processing (for deconvolution, Fourier analysis, or peak fitting). Daniel Armstrong has very recently illustrated the possibility to perform sub-second separations using 5-mm long columns on current instrumentation. For such applications it could make sense to further improve computational power.
For modeling, there is still a lack of available model parameters (or variables) that can put into model equations because they are very difficult to measure, such as the tortuosity factor, obstruction factor, diffusion coefficients, and so on. There is therefore still a need to develop experimental methodologies to obtain or derive accurate model parameters.
Based on your work in analysis of pharmaceuticals (4), where do you see the need for the most future research? Would it be in the area of component separation, speed of analysis, detection limits, automation, or other areas? What major breakthrough would you like to see for faster liquid chromatography analysis?
Honestly, I do not think there is a need for faster separations or lower detection limits than are available today, however, I believe the problem is that pharmacopeia methods are still old-fashioned and do not take into consideration the benefits of current possibilities. UHPLC was commercially introduced in 2004 (15 years ago) however in many quality control (QC) labs, old conventional high performance liquid chromatography (HPLC) methods are still used and 40 to 90-minute-long separations are routinely performed to determine 5-6 impurities of an active pharmaceutical ingredient (API), simply because those labs need to follow pharmacopeia instructions. To me, that is nonsense.
On the other hand, I think that it is not the technical possibilities but the new samples to be analyzed (the increase in complex samples such as therapeutic proteins, oligonucleotides, single cell analysis, and so forth) will shape the chromatographic needs, which is now difficult to predict.
In one of your more recent papers, you describe the potential benefits and theory of using columns packed with particles of decreasing size (particle size gradient) in liquid chromatography (5). What can you tell us about how particle size gradient affects the separation and what improvements can be made by understanding the particle size gradient effect?
That was a funny project. Some friends and colleagues (Balázs Bobály, Róbert Kormány, Krisztián Horváth) and I were together during a conference and discussing chromatographic questions. One of those was: What happens when coupling two or more columns in series, which have a different number of theoretical plates? We were aware of the earlier amazing work of Calvin Giddings on plate heights of non-uniform columns, of Leonid Blumberg’s variance of zone migration in non-uniform medium, and also of Deirdre Cabooter’s works on multi-column systems. However, we were interested in the gradient elution mode (since in practice most separations are performed in gradient mode) and in peak widths: whether the peak width changes if one of the columns that is used for the coupled systems loses a bit of its initial efficiency? If yes, then does the order of the columns impact the apparent efficiency of the system or not? These questions seem to be simple and easy at first sight but the more you think on them the more difficult they become. From here the project went on and has almost developed by itself as questions have arisen. We asked for the help of other prominent colleagues (Davy Guillarme, Santiago Codesido, Gert Desmet) and in the end, some very interesting conclusions were drawn. We found that, in the case where the later columns in the row have high enough efficiency, then the gradient band compression effect outperforms the band broadening effect and finally a “peak sharpening” will be observed during the travel of the solute along the column. That was really exciting, and it motivated us to further explore the potential gain in efficiency when sequentially placing columns according to their increasing efficiency. The next rational question was: why not apply a particle size gradient (as a limiting case of a large number of serially coupled columns). And indeed, in the best case, about 15–20% gain in efficiency could be expected at a given retention when utilizing a particle size gradient, compared to constant particle size. Conversely, when fixing efficiency, the analysis time could be decreased by about 15% with an optimal particle size gradient. In theory, applying a particle size gradient can be a good possibility to improve the quality of separations but in practice it is not easy to implement since packing columns with different particle size gradients might be challenging to achieve.
What can you share with our readers regarding your next area of interest for your research?
We are planning to work a lot with new biopharmaceutical products (mAbs, ADCs, bispecific monoclonal antibodies (bsAbs), and fusion proteins) since there are still many things to understand and develop. We are also involved in the development of new column designs and stationary phases which we believe will change the current practice of protein chromatography.
Would you like to acknowledge any coworkers, professors, or mentors that have been a great help to you early in your career?
During my second year at the Technical University of Budapest I met Prof JenÅ Fekete for the first time and was fascinated by his skills, knowledge and experiences. Before this encounter, I spent most of the semester on basketball courts and athletics tracks. JenÅ completely shifted my interest, refocused my attention to attending lectures and I found I enthusiastically participated in his research projects. Later on, JenÅ was my supervisor during my PhD studies and we became very good friends.
Another person who had a significant influence on my carrier was Katalin Ganzler at Gedeon Richter Plc (GR, Hungarian pharmaceutical company). I was very impressed by her talent and the way she thinks. She has a thorough global view on any topic whilst at the same time she can quickly identify all the tiny details in a way I am not able to do myself. Moreover, she always supported and encouraged me and helped me considerably in private life.
Whilst working at GR, I soon met Imre Molnár (Molnar-Institute). He was giving courses on LC method development. Thanks to him, I learnt a considerable amount about the pioneers of chromatography (Csaba Horváth, István Halász, Lloyd Snyder, and John Dolan). He often told personal and funny stories about those legends which made his courses unique and entertaining. Since then we have had a long and fruitful collaboration together.
I would also like to thank Professor Sándor Görög; who was the former director of analytical research at GR and editor of Journal of Pharmaceutical and Biomedical Analysis. We worked in different departments, however I often went to him to discuss ideas and his help was extremely valuable when writing my first journal articles. Moreover, he was the person who first contacted Professor Jean-Luc Veuthey (University of Geneva) and recommended me for a post-doc position–thanks to him I moved to my current position at the University of Geneva.
Finally, the two people I would most like to thank are Professor Jean-Luc Veuthey and Davy Guillarme at the University of Geneva. They are both well-recognized and outstanding scientists, while more importantly, these two guys are exceptional people too. They are very friendly, helpful, calm and fair. These are two persons in whom you can fully trust, who always do their best to support the colleagues, students and the group. I count myself extra-lucky and I am proud to be a member of their group.
(1) S. Fekete, J.L. Veuthey, D.V. McCalley, and D. Guillarme, The effect of pressure and mobile phase velocity on the retention properties of small analytes and large biomolecules in ultra-high pressure liquid chromatography, J. Chromatogr. A 1270, 127–138 (2012).
(2) E. Tyteca, J.L. Veuthey, G. Desmet, D. Guillarme, and S. Fekete, Computer assisted liquid chromatographic method development for the separation of therapeutic proteins. Analyst, 141(19), 5488–5501 (2016).
(3) S. Fekete, J. Fekete, and D. Guillarme, Estimation of the effects of longitudinal temperature gradients caused by frictional heating on the solute retention using fully porous and superficially porous sub-2 μm materials. J. Chromatogr. A1359, 124–130 (2014).
(4) S. Fekete, I. Kohler, S. Rudaz, and D. Guillarme, Importance of instrumentation for fast liquid chromatography in pharmaceutical analysis. J Pharm Biomed Anal87, 105–119 (2014).
(5) S. Codesido, S. Rudaz, J.L. Veuthey, D. Guillarme, G. Desmet, and S. Fekete, Impact of particle size gradients on the apparent efficiency of chromatographic columns. J. Chromatogr. A1603, 208–215 (2019).
Szabolcs Fekete, the 2020 winner of the Emerging Leader in Chromatography Award, earned his PhD degree in 2011 from Technical University of Budapest, Hungary, and is currently a scientific collaborator at the University of Geneva. His work focuses on finding new possibilities in protein chromatography; characterizing therapeutic proteins, LC column technology; LC method development, optimization, retention modeling and other fundamental studies. Fekete has published about 130 papers, with more than 4000 citations. He has also given more than 60 oral and 30 poster presentations at scientific conferences, and has won five “best poster” awards at international conferences. He also received the György Oláh award from the University of Technology of Budapest in 2011.