Journal of Chromatography A, 1372 (2014) 166–173
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Thermodynamic study of the interaction between linear plasmid deoxyribonucleic acid and an anion exchange support under linear and overloaded conditions P.A. Aguilar a,b , A. Twarda a,c , F. Sousa a,b , A.C. Dias-Cabral a,b,∗ a
CICS-UBI – Health Sciences Research Centre, University of Beira Interior, 6200-506 Covilhã, Portugal Department of Chemistry, University of Beira Interior, 6200-001 Covilhã, Portugal c Faculty of Biochemistry, Biophysics and Biotechnology, Jagiellonian University, 30-387 Cracow, Poland b
a r t i c l e
i n f o
Article history: Received 8 August 2014 Received in revised form 30 October 2014 Accepted 2 November 2014 Available online 7 November 2014 Keywords: Plasmid DNA Fast flow Q-Sepharose Ion-exchange chromatography Flow microcalorimetry Adsorption isotherms
a b s t r a c t Anion-exchange chromatography has been successfully used in plasmid DNA (pDNA) purification. However, pDNA adsorption mechanism using this method is still not completely understood, and the prediction of the separation behavior is generally unreliable. Flow microcalorimetry (FMC) has proven its ability to provide an improved understanding of the driving forces and mechanisms involved in the adsorption process of biomolecules onto several chromatographic systems. Thus, using FMC, this study aims to understand the adsorption mechanism of linear pDNA (pVAX1-LacZ) onto the anion-exchange support Fast Flow (FF) Q-Sepharose. Static binding capacity studies have shown that the mechanism of pDNA adsorption onto Q-Sepharose follows a Langmuir isotherm. FMC experiments resulted in thermograms that comprised endothermic and exothermic heats. Endothermic heat major contributor was suggested to be the desolvation process. Exothermic heats were related to the interaction between pDNA and Q-Sepharose primary and secondary adsorption. Furthermore, FMC revealed that the overall adsorption process is exothermic, as expected for an anion-exchange interaction. Nevertheless, there are evidences of the presence of nonspecific effects, such as reorientation and electrostatic repulsive forces. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The potential of pDNA as a therapeutic molecule in the treatment and cure of several diseases, through DNA vaccines or gene therapy, has been sustained by innumerous recent studies [1–4]. Plasmid-based therapies require efforts from researchers to develop cost-effective technologies to produce large quantities of highly pure pDNA [5]. Anion-exchange chromatography has been successfully used in pDNA purification [1,4,6–8], however its mechanism of interaction is still not completely understood. Additionally, in industry, due to economic reasons, it is usual to perform chromatographic runs in non-linear conditions, through feed overloading in concentration or volume [9]. Under these conditions adsorption mechanism complexity is further increased. Thermodynamic parameters related to adsorption and desorption of biomolecules onto chromatographic media have helped to
∗ Corresponding author at: CICS-UBI – Health Sciences Research Centre, University of Beira Interior, 6200-506 Covilhã, Portugal. Tel.: +351 275319700; fax: +351 275319730. E-mail address: [email protected] (A.C. Dias-Cabral). http://dx.doi.org/10.1016/j.chroma.2014.11.002 0021-9673/© 2014 Elsevier B.V. All rights reserved.
elucidate complex adsorption mechanisms in liquid chromatography. These parameters can be accessed from microcalorimetric measurements, by analyzing data through van’t Hoff plots or by performing batch equilibrium experiments [10–14]. Nevertheless, these methods may not produce representative results of overloaded conditions, batch equilibrium experiments have limited resolution and the indirect method of van’t Hoff analysis may be complicated by the presence of multiple adsorption sub-processes. Conversely, calorimetric methods such as Isothermal Titration Microcalorimetry (ITM) and flow microcalorimetry (FMC) have shown their aptitude to understand the underlying adsorption mechanism of several proteins in several chromatographic media, including linear and overloaded conditions [13–17]. Despite the progress in the understanding of adsorption mechanism for proteins through thermodynamic studies [11,14–20], there is still a lack of information for plasmids adsorption onto chromatographic supports [7]. Few studies have addressed this issue. Ferreira et al. [7] studied the batch adsorption of pDNA onto anion-exchange chromatographic supports and concluded that is plausible to apply the Langmuir isotherm model. The Langmuir isotherm [21] is the simplest theoretical model usually applied to
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
describe biomolecules adsorption onto ion exchangers, however, the data obtained does not account for non-ideal effects neither the complexity of working in overloaded conditions. Tarmann and Jungbauer [22] have used van’t Hoff analysis to investigate DNA retention on Source 30 Q at different temperatures and salt concentrations. It was shown that the adsorption process was driven entropically in all studied cases. More recently, Mahut et al. [23] reach the same conclusion when reporting on the separation of pDNA isoforms on silica-based columns modified with a quinine carbamate ligand. The observed behavior was related with changes in the solvation shell (and possibly the twist of the DNA helix) upon the binding to the stationary phase. Chen et al. [24] performed a study of the interaction mechanism between single-strand (ss) and double-strand (ds) DNA with hydroxyapatite (HA) by Isothermal Titration Microcalorimetry (ITM) and static binding measurements. ITM revealed that the dehydration is the dominant step in the interaction process. Furthermore, it was concluded that the binding behavior between dsDNA and HA is mainly driven by electrostatic interactions and ssDNA binding is more complex due to the hydrophobic and – interaction between bases. Lastly, Phillips and Pinto [25], using flow microcalorimetry (FMC), performed a study that closely resembles the adsorption of DNA. They investigated the adsorption of nitrogen bases and nucleosides onto a hydrophobic interaction adsorbent. FMC simulates a chromatographic system in its operation mode, thus it is expected that its results may be representative of what happens in an actual chromatographic column [14]. Phillips and Pinto [25] study revealed that the adsorption behavior of nitrogen bases and nucleosides is a complex phenomenon highly dependent on the type of molecule. Furthermore, although hydrophobic interaction appears to be the primary mechanism for the adsorption, they observed that the measured heats of adsorption resulted from the net effect of two different types of interactions: adsorbate/adsorbent interactions and base stacking self-interactions between like molecules. Considering the already proven ability of flow microcalorimetry (FMC) in understanding the driving forces and mechanisms of biomolecules adsorption onto several chromatographic supports [14–20,26–29], the present study used FMC as a central technique to understand the adsorption mechanism of linear (ln) pDNA (pVAX1-LacZ) onto the anion-exchange support FF Q-Sepharose, considering linear and overloaded conditions. 2. Experimental methods 2.1. Plasmid production, recovery and purification The 6.05 kbp plasmid, pVAX1-LacZ (Invitrogen, Carlsband, CA, USA), was obtained by Escherichia coli (E. coli) DH5␣ fermentation as described by Sousa and Queiroz [30]. The plasmid was recovered from the cells and purified using the QIAGEN® Plasmid Maxi Kit. Isolated pDNA was analyzed by horizontal electrophoresis according to Caramelo-Nunes et al. [31]. Linear pDNA isoform was obtained using the restriction endonuclease Hind III (NZYTech, Lisbon, Portugal). Hind III was removed from pDNA samples through centrifugation, using 100 kDa Vivaspin membrane concentrators (Sartorius Stedim Biotech, Madrid). For storage and as a carrier fluid 10 mM Tris–HCl (Sigma–Aldrich, Madrid, Spain), pH 8.0 was used. 2.2. Adsorption isotherm measurements Adsorption isotherms were performed in multiwell plates, where 10 mg of dry FF Q-Sepharose have been placed in each well, adding 1 mL of a known concentration of ln pDNA in 10 mM Tris–HCl, pH 8.0, as described by Silva et al. [14]. After 24 h [7] equilibrium reached and pDNA content of mobile phase was
167
determined by spectrophotometry (UV, 260 nm). All experiments were performed in triplicates. By fitting experimental data to the Langmuir model of adsorption, maximal binding capacities qmax and association constant KA were determined. The general equation for the Langmuir isotherm is: q∗ =
qmax KA C ∗ 1 + KA C ∗
(1)
where C* is the biomolecule equilibrium concentration in solution, q* is the biomolecule solid-phase equilibrium concentration, qmax is the maximum adsorption capacity of the adsorbent and KA is the adsorption constant. 2.3. Flow microcalorimetry (FMC) Thermodynamic studies were performed in the flow microcalorimeter (Microscal FMC 4 Vi, Microscal Limited, London, UK). The flow microcalorimeter has the ability to measure the heat flow caused by interaction during the adsorption process of biomolecules onto chromatographic media. Interfaced with its cell, the microcalorimeter has two highly sensitive thermistors that are capable of detecting small temperature changes. The heat evolution or absorption during an interaction is indicated by changes in potential (imbalance in the thermistor bridge in which the two thermistors measure temperature changes in the cell). Thus, when an exothermic interaction occurs the microcalorimeter will sense an increase in energy and a positive signal will appear in the thermogram. The opposite is observed for an endothermic interaction. Experiments were performed as described by Silva et al. [14]. The system was packed with approximately 21.9 mg of dried Q-Sepharose to fill the 171 L cell. After thermal equilibrium attainment, by passing the equilibration buffer (10 mM Tris–HCl, pH 8.0) through the cell at a constant flow rate of 1.5 mL/h for 12 h, a sample of ln pDNA (in 10 mM Tris–HCl, pH 8.0) was loaded into a configurable injection loop (30, 229 or 429 L) and then injected into the cell using a constant flow rate of 1.5 mL/h. Between injections, 1 M NaOH was used as a washing solution. CALDOS 4 software (Microscal Limited, London, UK) was used to acquire, store, and process all the FMC data. Peak deconvolution was performed by the PEAKFIT software package (version 4.12, Seasolve Software Inc., San Jose, USA) using asymmetric Gaussian peaks. 3. Results and discussion The topological linear pDNA form used in this study was prepared enzymatically. High-resolution agarose gel electrophoresis was used to track the production process (Fig. 1). The use of the Qiagen Plasmid Maxi kit® yielded approximately a final concentration of 600 g mL−1 of 6.05 kbp pVAX1-lacZ pDNA (linear, open circular and supercoiled isoforms) (Fig. 1[A]). Subsequently, the pDNA was left at room temperature for three days. During this time span, an increase in the relative amount of open circular and linear pDNA and a parallel decrease in the relative amount of supercoiled pDNA was observed. In the third day, only circular and linear pDNA were present (Fig. 1[D]-(a)). After digestion with the restriction enzyme Hind III pure linear pDNA was obtained (Fig. 1[D]-(b)). Static binding capacity results of ln pDNA adsorption onto QSepharose are reported in Fig. 2. Through the analysis of the equilibrium isotherm profile, it can be seen that the adsorbed ln pDNA concentration (qpDNA ) increases from zero to a plateau region as the ln pDNA equilibrium liquid concentration (CpDNA ) increases. Within the range covered in this study, the shape of the curve fits to the Langmuir (type I) isotherm profile [12,21]. Langmuir fitted parameters, association constant (KA ) and maximum binding capacity (qmax ) were respectively 0.31 ± 0.04 mL g−1 and
168
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
Fig. 1. Agarose gel electrophoresis. Lane A: pDNA after recovery and purification with Qiagen kit; Lane B: pDNA after 1 day at room temperature after [A]; Lane C: pDNA after 2 days at room temperature after [A]; Lane D(a): pDNA after 3 days at room temperature after [A]; Lane D(b): pDNA after Hind III digestion; 1 – open circular (oc) pDNA; 2 – linear (ln) pDNA; 3 – supercoiled (sc) pDNA; 4 – 1 kb DNA Ladder (GeneRuler, Fermentas Life Science, Vilnius, Lithuania).
Fig. 2. Equilibrium binding isotherms for ln pDNA adsorption onto FF Q-Sepharose at pH 8. At equilibrium, the adsorbed concentration, qpDNA is plotted versus the liquid phase concentration CpDNA .
0.832 ± 0.03 mg mL−1 gel. These values are of the same order of magnitude of the ones obtained by Ferreira et al. [7] with FF Qsepharose, although using a different plasmid in presence of salt (supercoiled isoform of a 4.8 kbp plasmid). As in their case, the expected high affinity of the support for the plasmid is observed, our association constant is greater than the ones reported for ion exchange of proteins and of the same order of magnitude of some constants obtained for affinity interactions [7]. Experimental maximum binding capacity, obtained from Langmuir model application to experimental adsorption isotherm, has been compared with its theoretic value. For simplification, like Tarmann et al. [22] we assumed a simple rod shape for linear pDNA and two extreme cases of orientation during adsorption, an upright and a longitudinal binding. The theoretic maximum adsorption capacity can be obtained from Ref. [32]: qm =
Am N
(2)
where Am is the support surface area, N is the Avogadro number, and is the effective area of gel surface covered by a molecule of pDNA. Once FF Q-sepharose can be assumed as nonporous media for pDNA [22], the maximum adsorption capacity has been predicted considering only the support external surface area, which was roughly estimated [33] using support particle diameter (90 m for FF Qsepharose) and extraparticle porosity (0.38 for FF Sepharose [34]). A value of 0.041 m2 mL−1 was found for the external area per unit
volume of packed bed. To know the effective area that a pDNA molecule covers in the gel surface, we also need pDNA dimensions. Based on a hydrodynamic diameter of 110 nm for a 6.05 kbp pDNA [35,36] and on a contour length of 2057 nm for its linear isoform [36], dimensions for an equivalent rod of 30 nm × 2057 nm were assumed. Calculated values for theoretical binding capacities were 0.359 and 0.003 mg mL−1 respectively considering an upright and longitudinal attachment. A good estimation was obtained with ln pDNA binding in an upright position at the outer surface. Similar results were obtained for a 4.9 kbp plasmid adsorbing on HP Q-sepharose [22]. Like in this study, our results for qmax are higher than the ones obtained experimentally. This was explained by a partial penetration of pDNA into the pores [22,37], in the present work this explanation is further supported by the known high elasticity of ln pDNA [38]. Considering the equilibrium binding isotherm data (Fig. 2), it is possible to define a linear and an overloaded zone, values of pDNA adsorbed concentration (qpDNA ) above 500 g mL−1 gel are considered in the overloaded zone. Moreover, according to the equilibrium binding isotherm shape, within the range studied, it is not expected the occurrence of multilayer adsorption, once when this phenomenon occurs, a second increase in the adsorbed concentration is observed after the first plateau in the isotherm [24]. FMC experiments were performed to better understand the interaction between ln pDNA and the Q-Sepharose, considering linear and overloaded conditions, and to investigate the role of nonspecific effects at the adsorptive process. Through the knowledge of the magnitude and chronology of thermal events during and after the biomolecule-adsorbent interaction, the adsorption mechanism can be elucidated [14,16,18,19,27–29,39,40]. Experiments were performed using different injection loops which allowed us to define two injection modes, the pulse injection for the 30 and 229 L loops and the continuous feed for the 429 L loop. Bearing in mind the 171 L of the FMC cell, the use of the 30 L loop means that the sample is introduced into the column as a pulse (30 L represents 0.2 times the cell volume). The use of the 229 L loop fulfills the FMC cell, representing 1.3 times of the cell volume. On the other hand, the use of the 429 L loop represents 2.5 times the cell volume, leading to a continuous feed of biomolecules into the cell (volume overloading), resembling the frontal analysis in chromatographic systems [33]. Figs. 3a and 4a illustrate the examples of the thermograms obtained with the 30 and 429 L injection loops respectively. In each graphic different signals resulted from different loading concentrations. In Fig. 3a, considering the ln pDNA adsorbed concentration, all thermograms where obtained in the linear zone of the isotherm. Conversely, in Fig. 4a, thermograms where obtained in the linear, transition and overloaded zone of the isotherm. All performed experiments resulted in thermograms that comprised a first endothermic peak followed by an exothermic one which may result from peak overlapping when the injection was through a continuous feed. Furthermore, in all studied cases, the first peak emerges at the exact moment when the frontal boundary of the ln pDNA solution reaches the adsorbent in the FMC cell. Heat signals obtained with the same injection mode (pulse injection, 30 and 229 L loops, or continuous feed, 429 and 1000 L loops) present similar profiles. It was noticed that by increasing the loop size (from 30 to 229 L in the pulse mode, and from 429 to 1000 L in the continuous feed) the peak profile remained similar whereas the time of the whole interaction increased, leading to a change in the magnitude of the heat signal and in the time at which each event took place (data not shown). This phenomenon may occur due to the increase in biomolecule system residence time, which gives extra time for equilibrium establishment, as was already observed by Silva et al. [14]. Whereas, comparing the heat signals obtained using different injection modes, a change in the
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
169
Fig. 3. Thermograms of ln pDNA adsorption onto FF Q-Sepharose, at pH 8. Injection loop: 30 L. Mobile phase flow rate: 1.5 mL/h. (a) Black (–) 12.6 g pDNA/g Q-Sepharose; red (– –) 53.4 g pDNA/g Q-Sepharose; blue (-..-) 111.2 g pDNA/g Q-Sepharose; (b)–(d) PEAKFIT de-convolution of thermograms for loading concentrations of (b) 12.6 g pDNA/g Q-Sepharose, (c) 53.4 g pDNA/g Q-Sepharose and (d) 111.2 g pDNA/g Q-Sepharose. Curves shown are for experimental data (red line (–)); total peak fit (black line (–)) and peaks resulting from deconvolution (blue line (...)). Vertical dashed line represents the time where the pDNA-containing plug of solution is replaced with pDNA-free mobile phase. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
heat signal profile was also observed, an extra exothermic peak seems to be present. Hereafter, data analysis will be performed considering separately the results from the different injections modes (pulse injection and continuous feed), since in each case different adsorption mechanisms may be involved. The presence of different peaks in the thermograms, Figs. 3 and 4, suggests the existence of different events during the adsorption process. These events may occur in a sequence or simultaneously. Heat signals de-convolution (Figs. 3b–d and 4b–d), using PEAKFIT software, revealed the presence of overlapped peaks. Deconvoluted thermograms were analyzed considering the mechanism proposed by Yamamoto and co-workers [24]. At the molecular level adsorption was divided into three sequential sub-processes: (i) water molecules and ions release from the biomolecules and adsorbent surfaces (endothermic heats); (ii) electrostatic attraction between the biomolecule and the ion exchanger (exothermic heats); (iii) solvation and hydration process of the bound biomolecules and its structural rearrangement if required (endothermic heats). Moreover, previous studies have revealed that endothermic heats were also related to repulsive interaction between like charged groups [13,16,27], water release due to hydrophobic interactions and biomolecules surface reorientation [14]. Thermograms in Fig. 3b–d show the de-convolution of the heat signals for experiments using the 30 L loop. From them it can be observed that the endothermic and exothermic signals overlap. Also, during the time in which the ln pDNA is flowing through the
cell (before 1200 s, time when the mobile phase plug containing ln pDNA is replaced by pDNA-free mobile phase) the endothermic peak area is larger than the exothermic one. Once in a favorable interaction, the Gibbs free energy has to be negative, here the adsorption process has to be entropically driven. Thus, considering the above assumptions (i–iii), during the interaction between the ln pDNA and the Q-Sepharose, the desolvation process, including dehydration and release of counter ions from the biomolecules and adsorbent surfaces (i), may be the main role. That is, the endothermic heat due to desolvation is compensated by the entropy gain resulting from the release of water and ions [17,24,29]. The water molecules quantification would be an interesting proof of this phenomenon, however fast performance chromatography experiments revealed no difference in the retention times using different ionic strengths, being not possible to perform the Perkins analysis [41]. Nevertheless, according to the chronological sequence of subprocesses proposed by Yamamoto and co-workers [24], the first event is endothermic and compatible with the desolvation process. Besides this, endothermic heats of adsorption were related to (1) water release due to hydrophobic interactions, (2) solvation and hydration process of bound pDNA, (3) structural rearrangements of pDNA and (4) biomolecules repulsion. Under the studied conditions hydrophobic interaction appears not to be a major contributor since we are working in absence of salt, Q-Sepharose is considered a strong anion exchanger and at pH greater than 4, DNA appears as a polyanionic molecule [6]. Also sub-processes (2) and (3) involve heat changes that were considered minimal when compared to
170
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
Fig. 4. Thermograms of ln pDNA adsorption onto FF Q-Sepharose. Injection loop: 429 L. Mobile phase flow rate: 1.5 mL/h. (a) Black (–) 1359.6 g pDNA/g Q-Sepharose; red (– –) 3599.0 g pDNA/g Q-Sepharose; blue (-..-) 5489.4 g pDNA/g Q-Sepharose; (b)–(d) PEAKFIT de-convolution of thermograms for loading concentrations of (b) 1359.6 g pDNA/g Q-Sepharose, (c) 3599.0 g pDNA/g Q-Sepharose and (d) 5489.4 g pDNA/g Q-Sepharose. (e) PEAKFIT de-convolution thermogram for 230 L loop for a loading concentrations of 1222.5 g pDNA/g Q-Sepharose. Curves shown are for experimental data (red line (–)); total peak fit (black line (–)) and peaks resulting from deconvolution (blue line (...)). Thermograms were obtained in the linear (b), transition (c) and overloaded (d) zones of the isotherm. Vertical dashed line represents the time where the pDNA-containing plug of solution is replaced with pDNA-free mobile phase. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
heat changes present in desolvation and interaction sub-processes [14,24]. Since we are still working in the linear zone of the isotherm, sub-process (4) is expected to have likewise a negligible contribution. Simultaneously to the desolvation sub-process the interaction between ln pDNA and Q-Sepharose occurs through an electrostatic attraction, leading to the appearance of an exothermic peak in the thermogram. The fact that the maximum of this peak appears after the end of the pDNA plug suggests the presence of other events. This may be due to secondary adsorption of bound pDNA (biomolecules move to a minimum-energy orientation) which occur due to the known flexibility of ln pDNA [38]. The endothermic heat change involved in the structural rearrangement of pDNA on the adsorbent surface is exceeded by the heat released resulting from the attractive interaction between pDNA and Q-Sepharose. As mentioned, structural rearrangements involve heat changes considered minimal when compared to heat changes present in attractive interactions [14,24]. Thermograms in Fig. 4b–d show the de-convolution of the heat signals for experiments using the 429 L loop, in the linear (Fig. 4b), transition (Fig. 4c) and overloaded (Fig. 4d) zones of the isotherm. The obtained heat signals are characterized by an endothermic peak followed by two overlapping exothermic peaks. As stated above, more than a change in the heat signal magnitude and length the transition from a pulse injection (Fig. 3) to a continuous feed (Fig. 4)
influences the heat signal profile, suggesting the presence, in the adsorption mechanism, of an additional sub-process. This may be related to the fact that after the first 171 L (cell volume) of the injected sample flow through the cell, another 258 L still have to pass through it, interfering with the already bounded molecules. This mode of operation approaches the volume overloading used in some preparative chromatographic applications. In this case, to analyze the thermograms, it should be also considered the interaction between adsorbed pDNA molecules and free pDNA molecules in the feed solution. Despite of some repulsion between these molecules, this interaction promotes the early reorientation and subsequent secondary adsorption of already bound biomolecules, leaving free space for pDNA in solution to adsorb. This is further supported comparing, in the linear zone of the isotherm at approximately the same equilibrium pDNA loading concentration, the thermograms obtained using the 229 L and 429 L loops (Fig. 4b and e). Considering the 429 L heat signal and observing it until around 1680 s (where the 229 L loop pDNA plug ends) it can be seen that the profile of the peak is similar to the profile obtained with the 229 L loop. After that, the adsorption enters in the volume overloading process and the heat signal magnitude increases until the end of the pDNA plug. This increase in the magnitude, as previously stated, may be due the secondary adsorption of the bound biomolecules in addition with the adsorption of more molecules.
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
171
Table 1 Heat of adsorption for ln pDNA adsorption on FF Q-Sepharose at pH 8; flow rate: 1.5 mL/h; adsorbent sample size: 21.89 mg; and temperature: 25 ◦ C. Enthalpies were determined from the deconvoluted thermograms. Loop (L)
pDNA mass feed (g)
30
0.27 1.17 2.43 4.44 9.01 10.89
229
20.68 31.72
429
57.92 87.73 203.78
pDNA loading (g g−1 )
12.6 53.40 111.2 202.8 311.5 497.6
± ± ± ± ± ±
1.7 0.02 0.1 0.2 9.4 0.1
Endothermic peak (mJ g−1 )
Exothermic peaks (mJ g−1 )
HI
HII
1.16 2.16 1.05 0.85 0.38 0.31
± ± ± ± ± ±
0.33 0.37 0.02 0.16 0.02 0.02
−3.91 −3.45 −1.84 −1.33 −0.46 −0.34
HIII ± ± ± ± ± ±
Net heat of adsorption (mJ g−1 ) HII + HIII
HTotal −2.75 −1.29 −0.79 −0.48 −0.08 −0.03
0.19 0.55 0.05 0.13 0.02 0.02
0 0 0 0 0 0
−3.91 −3.45 −1.84 −1.33 −0.46 −0.34
0 0
−0.51 ± 0.16 −0.47 ± 0.13
−0.32 ± 0.18 −0.23 ± 0.14
−0.66 ± 0.12 −0.40 ± 0.02 −0.36 ± 0.08
−0.47 ± 0.13 −0.36 ± 0.02 −0.33 ± 0.08
820.4 ± 82.9 1222.5 ± 70.8
0.20 ± 0.08 0.23 ± 0.05
−0.51 ± 0.16 −0.47 ± 0.13
1359.6 ± 132.6 3599.0 ± 111.8 5489.4 ± 152.7
0.19 ± 0.05 0.05 ± 0.01 0.03 ± 0.01
−0.29 ± 0.11 −0.32 ± 0.02 −0.29 ± 0.08
−0.36 ± 0.06 −0.08 ± 0.02 −0.08 ± 0.02
± ± ± ± ± ±
0.19 0.55 0.05 0.13 0.02 0.02
± ± ± ± ± ±
0.38 0.66 0.05 0.20 0.02 0.02
HTotal = HI + HII + HIII .
Considering heat signals obtained in the transition and in the overloaded zones of the isotherm (Fig. 4c and d) and comparing them with a signal obtained in the linear zone (Fig. 4b) it can be seen that, under linear adsorption conditions, the first exothermic
event presents a lower absolute magnitude when compared to the second event, and this tendency is reversed for loadings in the transition and overloaded zones. This may be related to the quantity of molecules present in the cell, interacting with the adsorbent at the
Fig. 5. Heat of ln pDNA adsorption onto FF Q-Sepharose obtained using the (a) 30 and 229 L loops and (b) 429 L loop. Blue squares – endothermic heat; red circles – neat heat; black triangles – 1st exothermic heat; gray diamonds – 2nd exothermic heat. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
172
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
same time. In the transition and overloaded zone, the high quantity of molecules does not favor the reorientation, reducing this process, indeed from Table 1 we can see that the enthalpy of adsorption of the first exothermic heat remains constant while its value for the second exothermic heat suffers a decrease. Fig. 5 presents the enthalpy changes for the ln pDNA adsorption onto Q-Sepharose, using the 30 and the 229 L loops (linear zone of the isotherm) (Fig. 5a) and the 429 L loop (linear, transition and overloaded zones of the isotherm) (Fig. 5b). Considering the net heat of adsorption (sum of all contribution), Table 1 and Fig. 5, since the Gibbs free energy (G = H − TS) has to be negative for a favorable interaction [10], the overall adsorption process is enthalpically driven, once it is exothermic for all the studied conditions, as expected for an anion-exchange interaction [14]. Previous studies, by van’t Hoff analysis, of supercoiled pDNA (4.9 kbp) onto Source 30 Q revealed an entropically driven process [22]. This emphasizes the widely different behaviors that systems can manifest, and the importance of characterizing each system independently, but also may indicate an extra entropic contribution arising from supercoiled pDNA alteration of conformation during adsorption, recall our study is made with ln pDNA. Under overloaded conditions, the net heat of adsorption appears to be constant (Fig. 5b), which is anticipated when the monolayer capacity is reached (Fig. 2) [14]. For the lower loading concentration, as ln pDNA loading increases the net heat of adsorption becomes more positive (Fig. 5a and b). Under these conditions, both endothermic and exothermic enthalpies decrease in magnitude. Endothermic heat major contribution was assumed to be from desolvation process. As the loading increases, each pDNA molecule foot print on the adsorbent surface may decrease in order to accommodate more molecules, as discussed pDNA upright position is preferred when binding under maximum loading conditions. Thus less water molecules and ions need to be removed in the binding process, leading to a decrease in the endothermic heat signal. Exothermic heat major contribution was assumed to result from the interaction between pDNA and Q-Sepharose and from secondary adsorption of already adsorbed molecules. The maintenance in the magnitude of the first exothermic heat (Fig. 5b) with the increase in loading concentration is expected, as a higher repulsion (endothermic contribution) between adsorbed molecules is predictable. The decrease in the magnitude of the second exothermic heat (Fig. 5b) is, as previously mention, anticipated once less reorientation results in less secondary adsorption.
4. Conclusions This study investigates the adsorption mechanism of linear pDNA onto FF Q-Sepharose considering linear, transition and overloaded adsorption conditions. Two modes of injection were used, pulse and continuous feed. Static binding capacity studies revealed that adsorption follows a type-I Langmuir isotherm. Theoretical binding capacity agreed well with experimental, indicating that interaction in upright position is preferred when under maximum loading conditions. Flow microcalorimetry shown that the overall adsorption process is exothermic, as expected for an anion-exchange interaction. Under overloaded conditions, the constant net heat of adsorption is compatible with the monolayer establishment [14]. Thermograms were analyzed considering the mechanism proposed by Yamamoto and co-workers [13]. Endothermic heat major contribution was suggested to be from water molecules and ions release (desolvation process), while major exothermic heats contributions were related to the electrostatic attraction between pDNA and Q-Sepharose and also to the secondary adsorption of already adsorbed molecules.
Under lower loading concentrations, the decrease in magnitude of the endothermic and exothermic enthalpies, with the increase of adsorbed pDNA, were respectively associated to a reduction in pDNA foot print on the adsorbent surface (less energy is used in water release) and with the repulsion between adsorbed molecules (increase in the endothermic heat contribution). Under higher loading concentrations, the maintenance in the magnitude of the first exothermic heat, with the increase in loading concentration, was justified by a balance between adsorption and repulsion (between adsorbed molecules) and the decrease of the second exothermic heat was related with less secondary adsorption. Funding sources Work supported by FCT (Portuguese Foundation for Science and Technology), project number FCOMP-01-0124-FEDER-014750 (Ref. FCT PTDC/EBB-BIO/113576/2009). References [1] A. Ghanem, R. Healey, F.G. Adly, Current trends in separation of plasmid DNA vaccines: a review, Anal. Chim. Acta 760 (2013) 1–15. [2] S. Ginn, I. Alexander, M. Edelstein, M. Abedi, J. Wixon, Gene therapy clinical trials worldwide to 2012 – an update, J. Gene Med. 15 (2013) 65–77. [3] L. Li, F. Saade, N. Petrovsky, The future of human DNA vaccines, J. Biotechnol. 162 (2012) 171–182. [4] F. Sousa, L. Passarinha, J.A. Queiroz, Biomedical application of plasmid DNA in gene therapy: a new challenge for chromatography, Biotechnol. Genet. Eng. Rev. 26 (2009) 83–116. [5] J. Stadler, R. Lemmens, T. Nyhammar, Plasmid DNA purification, J. Gene Med. 6 (Suppl. 1) (2004) S54–S66. [6] G.N.M. Ferreira, Chromatographic approaches in the purification of plasmid DNA for therapy and vaccination, Chem. Eng. Technol. 28 (2005) 1285–1294. [7] G.N.M. Ferreira, J.M. Cabral, D.M. Prazeres, Studies on the batch adsorption of plasmid DNA onto anion-exchange chromatographic supports, Biotechnol. Prog. 16 (2000) 416–424. [8] G.N.M. Ferreira, G.A. Monteiro, D.M. Prazeres, J.M. Cabral, Downstream processing of plasmid DNA for gene therapy and DNA vaccine applications, Trends Biotechnol. 18 (2000) 380–388. [9] X. Yang, J. Wang, X. Liu, Z. Shang, Study on system identification of linear and non-linear liquid chromatography separation, Anal. Lett. 34 (2001) 803–811. [10] F. Lin, W. Chen, M.T.W. Hearn, Thermodynamic analysis of the interaction between proteins and solid surfaces: application to liquid chromatography, J. Mol. Recognit. 15 (2002) 55–93. [11] M. Rabe, D. Verdes, S. Seeger, Understanding protein adsorption phenomena at solid surfaces, Adv. Colloid Interface Sci. 162 (2011) 87–106. [12] J.C. Bellot, J.S. Condoret, Review modelling of liquid chromatography equilibria, Process Biochem. 28 (1993) 365–376. [13] F.Y. Lin, C.S. Chen, W.Y. Chen, S. Yamamoto, Microcalorimetric studies of the interaction mechanisms between proteins and Q-sepharose at pH near the isoelectric point (pI) effects of NaCl concentration, pH value, and temperature, J. Chromatogr. A 912 (2001) 281–289. [14] G.L. Silva, F.S. Marques, M.E. Thrash, A.C. Dias-Cabral, Enthalpy contributions to adsorption of highly charged lysozyme onto a cation-exchanger under linear and overloaded conditions, J. Chromatogr. A 1352 (2014) 46–54. [15] T. Blaschke, A. Werner, H. Hasse, Microcalorimetric study of the adsorption of native and mono-PEGylated bovine serum albumin on anion-exchangers, J. Chromatogr. A 1277 (2013) 58–68. [16] A. Katiyar, S.W. Thiel, V.V. Guliants, N.G. Pinto, Investigation of the mechanism of protein adsorption on ordered mesoporous silica using flow microcalorimetry, J. Chromatogr. A 1217 (2010) 1583–1588. [17] J. Korfhagen, A.C. Dias-Cabral, M.E. Thrash, Nonspecific effects of ion exchange and hydrophobic interaction adsorption processes, Sep. Sci. Technol. 45 (2010) 2039–2050. [18] A.C. Dias-Cabral, J.A. Queiroz, N.G. Pinto, Effect of salts and temperature on the adsorption of bovine serum albumin on polypropylene glycol-Sepharose under linear and overloaded chromatographic conditions, J. Chromatogr. A 1018 (2003) 137–153. [19] J. Kim, R.J. Desch, S.W. Thiel, V.V. Guliants, N.G. Pinto, Energetics of biomolecule adsorption on mesostructured cellular foam silica, Microporous Mesoporous Mater. 170 (2013) 95–104. [20] P. Raje, N.G. Pinto, Importance of heat of adsorption in modeling protein equilibria for overloaded chromatography, J. Chromatogr. A 796 (1998) 141–156. [21] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and platinum, J. Am. Chem. Soc. 40 (1918) 1361–1403. [22] C. Tarmann, A. Jungbauer, Adsorption of plasmid DNA on anion exchange chromatography media, J. Sep. Sci. 31 (2008) 2605–2618. [23] M. Mahut, A. Gargano, H. Schuchnigg, W. Lindner, Chemoaffinity material for plasmid DNA analysis by high-performance liquid chromatography with condition-dependent switching between isoform and topoisomer selectivity, Anal. Chem. 85 (2013) 2913–2920.
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173 [24] W. Chen, M. Lin, P. Lin, P. Tasi, Y. Chang, S. Yamamoto, Studies of the interaction mechanism between single strand and double-strand DNA with hydroxyapatite by microcalorimetry and isotherm measurements, Colloids Surf. A 295 (2007) 274–283. [25] J.M. Phillips, N.G. Pinto, Calorimetric investigation of the adsorption of nitrogen bases and nucleosides on a hydrophobic interaction sorbent, J. Chromatogr. A 1036 (2004) 79–86. [26] M. Ornelas, D. Loureiro, M.J. Araújo, E. Marques, C. Dias-Cabral, M. Azenha, F. Silva, Synthesis of glycylglycine-imprinted silica microspheres through different water-in-oil emulsion techniques, J. Chromatogr. A 1297 (2013) 138–145. [27] M.E. Thrash, N.G. Pinto, Characterization of enthalpic events in overloaded ionexchange chromatography, J. Chromatogr. A 944 (2002) 61–68. [28] M.E. Thrash, N.G. Pinto, Flow microcalorimetric measurements for bovine serum albumin on reversed-phase and anion-exchange supports under overloaded conditions, J. Chromatogr. A 908 (2001) 293–299. [29] M.E. Thrash, J.M. Phillips, N.G. Pinto, An analysis of the interactions of BSA with an anion-exchange surface under linear and non-linear conditions, Adsorpt. Int. Adsorpt. Soc. 10 (2004) 299–307. [30] F. Sousa, J.A. Queiroz, Supercoiled plasmid quality assessment by analytical arginine-affinity chromatography, J. Chromatogr. A 1218 (2011) 124–129. [31] C. Caramelo-Nunes, D. Bicho, P. Almeida, J.C. Marcos, C.T. Tomaz, Dynamic binding capacity and specificity of 3,8-diamino-6-phenylphenanthridineSepharose support for purification of supercoiled plasmid deoxyribonucleic acid, J. Chromatogr. A 1307 (2013) 91–98. [32] D.K. Chattoraj, K.S. Birdi, Adsorption and the Gibbs Surface Excess, Plenum Press, New York, 1984.
173
[33] G. Carta, A. Jungbauer, Protein Chromatography: Process Development and Scale-Up, Wiley-VCH, 2010. [34] M.C. Flickinger, Downstream Industrial Biotechnology: Recovery and Purification, Wiley-VCH, 2013. [35] A.M. Morão, J.C. Nunes, F. Sousa, M.T. Pessoa de Amorim, I.C. Escobar, J.A. Queiroz, Ultrafiltration of supercoiled plasmid DNA: modeling and application, J. Membr. Sci. 378 (2011) 280–289. [36] A. Morão, J.C. Nunes, F. Sousa, M.T.P. De Amorim, I.C. Escobar, J.A. Queiroz, Development of a model for membrane filtration of long and flexible macromolecules: application to predict dextran and linear DNA rejections in ultrafiltration, J. Membr. Sci. 336 (2009) 61–70. [37] A. Ljunglof, P. Bergvall, R. Bhikhabhai, R. Hjorth, Direct visualisation of plasmid DNA in individual chromatography adsorbent particles by confocal scanning laser microscopy, J. Chromatogr. A 844 (1999) 129–135. [38] C.R. Smith, R.B. DePrince, J. Dackor, D. Weigl, J. Griffith, M. Persmark, Separation of topological forms of plasmid DNA by anion-exchange HPLC: shifts in elution order of linear DNA, J. Chromatogr. B 854 (2007) 121–127. [39] J. Kim, R.J. Desch, S.W. Thiel, V.V. Guliants, N.G. Pinto, Energetics of protein adsorption on amine-functionalized mesostructured cellular foam silica, J. Chromatogr. A 1218 (2011) 7796–7803. [40] J. Kim, R.J. Desch, S.W. Thiel, V.V. Guliants, N.G. Pinto, Energetics of lysozyme adsorption on mesostructured cellular foam silica: effect of salt concentration, J. Chromatogr. A 1218 (2011) 6697–6704. [41] T.W. Perkins, D.S. Mak, T.W. Root, E.N. Lightfoot, Protein retention in hydrophobic interaction chromatography: modeling variation with buffer ionic strength and column hydrophobicity, J. Chromatogr. A 766 (1997) 1–14.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Thermodynamic study of the interaction between linear plasmid deoxyribonucleic acid and an anion exchange support under linear and overloaded conditions P.A. Aguilar a,b , A. Twarda a,c , F. Sousa a,b , A.C. Dias-Cabral a,b,∗ a
CICS-UBI – Health Sciences Research Centre, University of Beira Interior, 6200-506 Covilhã, Portugal Department of Chemistry, University of Beira Interior, 6200-001 Covilhã, Portugal c Faculty of Biochemistry, Biophysics and Biotechnology, Jagiellonian University, 30-387 Cracow, Poland b
a r t i c l e
i n f o
Article history: Received 8 August 2014 Received in revised form 30 October 2014 Accepted 2 November 2014 Available online 7 November 2014 Keywords: Plasmid DNA Fast flow Q-Sepharose Ion-exchange chromatography Flow microcalorimetry Adsorption isotherms
a b s t r a c t Anion-exchange chromatography has been successfully used in plasmid DNA (pDNA) purification. However, pDNA adsorption mechanism using this method is still not completely understood, and the prediction of the separation behavior is generally unreliable. Flow microcalorimetry (FMC) has proven its ability to provide an improved understanding of the driving forces and mechanisms involved in the adsorption process of biomolecules onto several chromatographic systems. Thus, using FMC, this study aims to understand the adsorption mechanism of linear pDNA (pVAX1-LacZ) onto the anion-exchange support Fast Flow (FF) Q-Sepharose. Static binding capacity studies have shown that the mechanism of pDNA adsorption onto Q-Sepharose follows a Langmuir isotherm. FMC experiments resulted in thermograms that comprised endothermic and exothermic heats. Endothermic heat major contributor was suggested to be the desolvation process. Exothermic heats were related to the interaction between pDNA and Q-Sepharose primary and secondary adsorption. Furthermore, FMC revealed that the overall adsorption process is exothermic, as expected for an anion-exchange interaction. Nevertheless, there are evidences of the presence of nonspecific effects, such as reorientation and electrostatic repulsive forces. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The potential of pDNA as a therapeutic molecule in the treatment and cure of several diseases, through DNA vaccines or gene therapy, has been sustained by innumerous recent studies [1–4]. Plasmid-based therapies require efforts from researchers to develop cost-effective technologies to produce large quantities of highly pure pDNA [5]. Anion-exchange chromatography has been successfully used in pDNA purification [1,4,6–8], however its mechanism of interaction is still not completely understood. Additionally, in industry, due to economic reasons, it is usual to perform chromatographic runs in non-linear conditions, through feed overloading in concentration or volume [9]. Under these conditions adsorption mechanism complexity is further increased. Thermodynamic parameters related to adsorption and desorption of biomolecules onto chromatographic media have helped to
∗ Corresponding author at: CICS-UBI – Health Sciences Research Centre, University of Beira Interior, 6200-506 Covilhã, Portugal. Tel.: +351 275319700; fax: +351 275319730. E-mail address: [email protected] (A.C. Dias-Cabral). http://dx.doi.org/10.1016/j.chroma.2014.11.002 0021-9673/© 2014 Elsevier B.V. All rights reserved.
elucidate complex adsorption mechanisms in liquid chromatography. These parameters can be accessed from microcalorimetric measurements, by analyzing data through van’t Hoff plots or by performing batch equilibrium experiments [10–14]. Nevertheless, these methods may not produce representative results of overloaded conditions, batch equilibrium experiments have limited resolution and the indirect method of van’t Hoff analysis may be complicated by the presence of multiple adsorption sub-processes. Conversely, calorimetric methods such as Isothermal Titration Microcalorimetry (ITM) and flow microcalorimetry (FMC) have shown their aptitude to understand the underlying adsorption mechanism of several proteins in several chromatographic media, including linear and overloaded conditions [13–17]. Despite the progress in the understanding of adsorption mechanism for proteins through thermodynamic studies [11,14–20], there is still a lack of information for plasmids adsorption onto chromatographic supports [7]. Few studies have addressed this issue. Ferreira et al. [7] studied the batch adsorption of pDNA onto anion-exchange chromatographic supports and concluded that is plausible to apply the Langmuir isotherm model. The Langmuir isotherm [21] is the simplest theoretical model usually applied to
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
describe biomolecules adsorption onto ion exchangers, however, the data obtained does not account for non-ideal effects neither the complexity of working in overloaded conditions. Tarmann and Jungbauer [22] have used van’t Hoff analysis to investigate DNA retention on Source 30 Q at different temperatures and salt concentrations. It was shown that the adsorption process was driven entropically in all studied cases. More recently, Mahut et al. [23] reach the same conclusion when reporting on the separation of pDNA isoforms on silica-based columns modified with a quinine carbamate ligand. The observed behavior was related with changes in the solvation shell (and possibly the twist of the DNA helix) upon the binding to the stationary phase. Chen et al. [24] performed a study of the interaction mechanism between single-strand (ss) and double-strand (ds) DNA with hydroxyapatite (HA) by Isothermal Titration Microcalorimetry (ITM) and static binding measurements. ITM revealed that the dehydration is the dominant step in the interaction process. Furthermore, it was concluded that the binding behavior between dsDNA and HA is mainly driven by electrostatic interactions and ssDNA binding is more complex due to the hydrophobic and – interaction between bases. Lastly, Phillips and Pinto [25], using flow microcalorimetry (FMC), performed a study that closely resembles the adsorption of DNA. They investigated the adsorption of nitrogen bases and nucleosides onto a hydrophobic interaction adsorbent. FMC simulates a chromatographic system in its operation mode, thus it is expected that its results may be representative of what happens in an actual chromatographic column [14]. Phillips and Pinto [25] study revealed that the adsorption behavior of nitrogen bases and nucleosides is a complex phenomenon highly dependent on the type of molecule. Furthermore, although hydrophobic interaction appears to be the primary mechanism for the adsorption, they observed that the measured heats of adsorption resulted from the net effect of two different types of interactions: adsorbate/adsorbent interactions and base stacking self-interactions between like molecules. Considering the already proven ability of flow microcalorimetry (FMC) in understanding the driving forces and mechanisms of biomolecules adsorption onto several chromatographic supports [14–20,26–29], the present study used FMC as a central technique to understand the adsorption mechanism of linear (ln) pDNA (pVAX1-LacZ) onto the anion-exchange support FF Q-Sepharose, considering linear and overloaded conditions. 2. Experimental methods 2.1. Plasmid production, recovery and purification The 6.05 kbp plasmid, pVAX1-LacZ (Invitrogen, Carlsband, CA, USA), was obtained by Escherichia coli (E. coli) DH5␣ fermentation as described by Sousa and Queiroz [30]. The plasmid was recovered from the cells and purified using the QIAGEN® Plasmid Maxi Kit. Isolated pDNA was analyzed by horizontal electrophoresis according to Caramelo-Nunes et al. [31]. Linear pDNA isoform was obtained using the restriction endonuclease Hind III (NZYTech, Lisbon, Portugal). Hind III was removed from pDNA samples through centrifugation, using 100 kDa Vivaspin membrane concentrators (Sartorius Stedim Biotech, Madrid). For storage and as a carrier fluid 10 mM Tris–HCl (Sigma–Aldrich, Madrid, Spain), pH 8.0 was used. 2.2. Adsorption isotherm measurements Adsorption isotherms were performed in multiwell plates, where 10 mg of dry FF Q-Sepharose have been placed in each well, adding 1 mL of a known concentration of ln pDNA in 10 mM Tris–HCl, pH 8.0, as described by Silva et al. [14]. After 24 h [7] equilibrium reached and pDNA content of mobile phase was
167
determined by spectrophotometry (UV, 260 nm). All experiments were performed in triplicates. By fitting experimental data to the Langmuir model of adsorption, maximal binding capacities qmax and association constant KA were determined. The general equation for the Langmuir isotherm is: q∗ =
qmax KA C ∗ 1 + KA C ∗
(1)
where C* is the biomolecule equilibrium concentration in solution, q* is the biomolecule solid-phase equilibrium concentration, qmax is the maximum adsorption capacity of the adsorbent and KA is the adsorption constant. 2.3. Flow microcalorimetry (FMC) Thermodynamic studies were performed in the flow microcalorimeter (Microscal FMC 4 Vi, Microscal Limited, London, UK). The flow microcalorimeter has the ability to measure the heat flow caused by interaction during the adsorption process of biomolecules onto chromatographic media. Interfaced with its cell, the microcalorimeter has two highly sensitive thermistors that are capable of detecting small temperature changes. The heat evolution or absorption during an interaction is indicated by changes in potential (imbalance in the thermistor bridge in which the two thermistors measure temperature changes in the cell). Thus, when an exothermic interaction occurs the microcalorimeter will sense an increase in energy and a positive signal will appear in the thermogram. The opposite is observed for an endothermic interaction. Experiments were performed as described by Silva et al. [14]. The system was packed with approximately 21.9 mg of dried Q-Sepharose to fill the 171 L cell. After thermal equilibrium attainment, by passing the equilibration buffer (10 mM Tris–HCl, pH 8.0) through the cell at a constant flow rate of 1.5 mL/h for 12 h, a sample of ln pDNA (in 10 mM Tris–HCl, pH 8.0) was loaded into a configurable injection loop (30, 229 or 429 L) and then injected into the cell using a constant flow rate of 1.5 mL/h. Between injections, 1 M NaOH was used as a washing solution. CALDOS 4 software (Microscal Limited, London, UK) was used to acquire, store, and process all the FMC data. Peak deconvolution was performed by the PEAKFIT software package (version 4.12, Seasolve Software Inc., San Jose, USA) using asymmetric Gaussian peaks. 3. Results and discussion The topological linear pDNA form used in this study was prepared enzymatically. High-resolution agarose gel electrophoresis was used to track the production process (Fig. 1). The use of the Qiagen Plasmid Maxi kit® yielded approximately a final concentration of 600 g mL−1 of 6.05 kbp pVAX1-lacZ pDNA (linear, open circular and supercoiled isoforms) (Fig. 1[A]). Subsequently, the pDNA was left at room temperature for three days. During this time span, an increase in the relative amount of open circular and linear pDNA and a parallel decrease in the relative amount of supercoiled pDNA was observed. In the third day, only circular and linear pDNA were present (Fig. 1[D]-(a)). After digestion with the restriction enzyme Hind III pure linear pDNA was obtained (Fig. 1[D]-(b)). Static binding capacity results of ln pDNA adsorption onto QSepharose are reported in Fig. 2. Through the analysis of the equilibrium isotherm profile, it can be seen that the adsorbed ln pDNA concentration (qpDNA ) increases from zero to a plateau region as the ln pDNA equilibrium liquid concentration (CpDNA ) increases. Within the range covered in this study, the shape of the curve fits to the Langmuir (type I) isotherm profile [12,21]. Langmuir fitted parameters, association constant (KA ) and maximum binding capacity (qmax ) were respectively 0.31 ± 0.04 mL g−1 and
168
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
Fig. 1. Agarose gel electrophoresis. Lane A: pDNA after recovery and purification with Qiagen kit; Lane B: pDNA after 1 day at room temperature after [A]; Lane C: pDNA after 2 days at room temperature after [A]; Lane D(a): pDNA after 3 days at room temperature after [A]; Lane D(b): pDNA after Hind III digestion; 1 – open circular (oc) pDNA; 2 – linear (ln) pDNA; 3 – supercoiled (sc) pDNA; 4 – 1 kb DNA Ladder (GeneRuler, Fermentas Life Science, Vilnius, Lithuania).
Fig. 2. Equilibrium binding isotherms for ln pDNA adsorption onto FF Q-Sepharose at pH 8. At equilibrium, the adsorbed concentration, qpDNA is plotted versus the liquid phase concentration CpDNA .
0.832 ± 0.03 mg mL−1 gel. These values are of the same order of magnitude of the ones obtained by Ferreira et al. [7] with FF Qsepharose, although using a different plasmid in presence of salt (supercoiled isoform of a 4.8 kbp plasmid). As in their case, the expected high affinity of the support for the plasmid is observed, our association constant is greater than the ones reported for ion exchange of proteins and of the same order of magnitude of some constants obtained for affinity interactions [7]. Experimental maximum binding capacity, obtained from Langmuir model application to experimental adsorption isotherm, has been compared with its theoretic value. For simplification, like Tarmann et al. [22] we assumed a simple rod shape for linear pDNA and two extreme cases of orientation during adsorption, an upright and a longitudinal binding. The theoretic maximum adsorption capacity can be obtained from Ref. [32]: qm =
Am N
(2)
where Am is the support surface area, N is the Avogadro number, and is the effective area of gel surface covered by a molecule of pDNA. Once FF Q-sepharose can be assumed as nonporous media for pDNA [22], the maximum adsorption capacity has been predicted considering only the support external surface area, which was roughly estimated [33] using support particle diameter (90 m for FF Qsepharose) and extraparticle porosity (0.38 for FF Sepharose [34]). A value of 0.041 m2 mL−1 was found for the external area per unit
volume of packed bed. To know the effective area that a pDNA molecule covers in the gel surface, we also need pDNA dimensions. Based on a hydrodynamic diameter of 110 nm for a 6.05 kbp pDNA [35,36] and on a contour length of 2057 nm for its linear isoform [36], dimensions for an equivalent rod of 30 nm × 2057 nm were assumed. Calculated values for theoretical binding capacities were 0.359 and 0.003 mg mL−1 respectively considering an upright and longitudinal attachment. A good estimation was obtained with ln pDNA binding in an upright position at the outer surface. Similar results were obtained for a 4.9 kbp plasmid adsorbing on HP Q-sepharose [22]. Like in this study, our results for qmax are higher than the ones obtained experimentally. This was explained by a partial penetration of pDNA into the pores [22,37], in the present work this explanation is further supported by the known high elasticity of ln pDNA [38]. Considering the equilibrium binding isotherm data (Fig. 2), it is possible to define a linear and an overloaded zone, values of pDNA adsorbed concentration (qpDNA ) above 500 g mL−1 gel are considered in the overloaded zone. Moreover, according to the equilibrium binding isotherm shape, within the range studied, it is not expected the occurrence of multilayer adsorption, once when this phenomenon occurs, a second increase in the adsorbed concentration is observed after the first plateau in the isotherm [24]. FMC experiments were performed to better understand the interaction between ln pDNA and the Q-Sepharose, considering linear and overloaded conditions, and to investigate the role of nonspecific effects at the adsorptive process. Through the knowledge of the magnitude and chronology of thermal events during and after the biomolecule-adsorbent interaction, the adsorption mechanism can be elucidated [14,16,18,19,27–29,39,40]. Experiments were performed using different injection loops which allowed us to define two injection modes, the pulse injection for the 30 and 229 L loops and the continuous feed for the 429 L loop. Bearing in mind the 171 L of the FMC cell, the use of the 30 L loop means that the sample is introduced into the column as a pulse (30 L represents 0.2 times the cell volume). The use of the 229 L loop fulfills the FMC cell, representing 1.3 times of the cell volume. On the other hand, the use of the 429 L loop represents 2.5 times the cell volume, leading to a continuous feed of biomolecules into the cell (volume overloading), resembling the frontal analysis in chromatographic systems [33]. Figs. 3a and 4a illustrate the examples of the thermograms obtained with the 30 and 429 L injection loops respectively. In each graphic different signals resulted from different loading concentrations. In Fig. 3a, considering the ln pDNA adsorbed concentration, all thermograms where obtained in the linear zone of the isotherm. Conversely, in Fig. 4a, thermograms where obtained in the linear, transition and overloaded zone of the isotherm. All performed experiments resulted in thermograms that comprised a first endothermic peak followed by an exothermic one which may result from peak overlapping when the injection was through a continuous feed. Furthermore, in all studied cases, the first peak emerges at the exact moment when the frontal boundary of the ln pDNA solution reaches the adsorbent in the FMC cell. Heat signals obtained with the same injection mode (pulse injection, 30 and 229 L loops, or continuous feed, 429 and 1000 L loops) present similar profiles. It was noticed that by increasing the loop size (from 30 to 229 L in the pulse mode, and from 429 to 1000 L in the continuous feed) the peak profile remained similar whereas the time of the whole interaction increased, leading to a change in the magnitude of the heat signal and in the time at which each event took place (data not shown). This phenomenon may occur due to the increase in biomolecule system residence time, which gives extra time for equilibrium establishment, as was already observed by Silva et al. [14]. Whereas, comparing the heat signals obtained using different injection modes, a change in the
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
169
Fig. 3. Thermograms of ln pDNA adsorption onto FF Q-Sepharose, at pH 8. Injection loop: 30 L. Mobile phase flow rate: 1.5 mL/h. (a) Black (–) 12.6 g pDNA/g Q-Sepharose; red (– –) 53.4 g pDNA/g Q-Sepharose; blue (-..-) 111.2 g pDNA/g Q-Sepharose; (b)–(d) PEAKFIT de-convolution of thermograms for loading concentrations of (b) 12.6 g pDNA/g Q-Sepharose, (c) 53.4 g pDNA/g Q-Sepharose and (d) 111.2 g pDNA/g Q-Sepharose. Curves shown are for experimental data (red line (–)); total peak fit (black line (–)) and peaks resulting from deconvolution (blue line (...)). Vertical dashed line represents the time where the pDNA-containing plug of solution is replaced with pDNA-free mobile phase. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
heat signal profile was also observed, an extra exothermic peak seems to be present. Hereafter, data analysis will be performed considering separately the results from the different injections modes (pulse injection and continuous feed), since in each case different adsorption mechanisms may be involved. The presence of different peaks in the thermograms, Figs. 3 and 4, suggests the existence of different events during the adsorption process. These events may occur in a sequence or simultaneously. Heat signals de-convolution (Figs. 3b–d and 4b–d), using PEAKFIT software, revealed the presence of overlapped peaks. Deconvoluted thermograms were analyzed considering the mechanism proposed by Yamamoto and co-workers [24]. At the molecular level adsorption was divided into three sequential sub-processes: (i) water molecules and ions release from the biomolecules and adsorbent surfaces (endothermic heats); (ii) electrostatic attraction between the biomolecule and the ion exchanger (exothermic heats); (iii) solvation and hydration process of the bound biomolecules and its structural rearrangement if required (endothermic heats). Moreover, previous studies have revealed that endothermic heats were also related to repulsive interaction between like charged groups [13,16,27], water release due to hydrophobic interactions and biomolecules surface reorientation [14]. Thermograms in Fig. 3b–d show the de-convolution of the heat signals for experiments using the 30 L loop. From them it can be observed that the endothermic and exothermic signals overlap. Also, during the time in which the ln pDNA is flowing through the
cell (before 1200 s, time when the mobile phase plug containing ln pDNA is replaced by pDNA-free mobile phase) the endothermic peak area is larger than the exothermic one. Once in a favorable interaction, the Gibbs free energy has to be negative, here the adsorption process has to be entropically driven. Thus, considering the above assumptions (i–iii), during the interaction between the ln pDNA and the Q-Sepharose, the desolvation process, including dehydration and release of counter ions from the biomolecules and adsorbent surfaces (i), may be the main role. That is, the endothermic heat due to desolvation is compensated by the entropy gain resulting from the release of water and ions [17,24,29]. The water molecules quantification would be an interesting proof of this phenomenon, however fast performance chromatography experiments revealed no difference in the retention times using different ionic strengths, being not possible to perform the Perkins analysis [41]. Nevertheless, according to the chronological sequence of subprocesses proposed by Yamamoto and co-workers [24], the first event is endothermic and compatible with the desolvation process. Besides this, endothermic heats of adsorption were related to (1) water release due to hydrophobic interactions, (2) solvation and hydration process of bound pDNA, (3) structural rearrangements of pDNA and (4) biomolecules repulsion. Under the studied conditions hydrophobic interaction appears not to be a major contributor since we are working in absence of salt, Q-Sepharose is considered a strong anion exchanger and at pH greater than 4, DNA appears as a polyanionic molecule [6]. Also sub-processes (2) and (3) involve heat changes that were considered minimal when compared to
170
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
Fig. 4. Thermograms of ln pDNA adsorption onto FF Q-Sepharose. Injection loop: 429 L. Mobile phase flow rate: 1.5 mL/h. (a) Black (–) 1359.6 g pDNA/g Q-Sepharose; red (– –) 3599.0 g pDNA/g Q-Sepharose; blue (-..-) 5489.4 g pDNA/g Q-Sepharose; (b)–(d) PEAKFIT de-convolution of thermograms for loading concentrations of (b) 1359.6 g pDNA/g Q-Sepharose, (c) 3599.0 g pDNA/g Q-Sepharose and (d) 5489.4 g pDNA/g Q-Sepharose. (e) PEAKFIT de-convolution thermogram for 230 L loop for a loading concentrations of 1222.5 g pDNA/g Q-Sepharose. Curves shown are for experimental data (red line (–)); total peak fit (black line (–)) and peaks resulting from deconvolution (blue line (...)). Thermograms were obtained in the linear (b), transition (c) and overloaded (d) zones of the isotherm. Vertical dashed line represents the time where the pDNA-containing plug of solution is replaced with pDNA-free mobile phase. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
heat changes present in desolvation and interaction sub-processes [14,24]. Since we are still working in the linear zone of the isotherm, sub-process (4) is expected to have likewise a negligible contribution. Simultaneously to the desolvation sub-process the interaction between ln pDNA and Q-Sepharose occurs through an electrostatic attraction, leading to the appearance of an exothermic peak in the thermogram. The fact that the maximum of this peak appears after the end of the pDNA plug suggests the presence of other events. This may be due to secondary adsorption of bound pDNA (biomolecules move to a minimum-energy orientation) which occur due to the known flexibility of ln pDNA [38]. The endothermic heat change involved in the structural rearrangement of pDNA on the adsorbent surface is exceeded by the heat released resulting from the attractive interaction between pDNA and Q-Sepharose. As mentioned, structural rearrangements involve heat changes considered minimal when compared to heat changes present in attractive interactions [14,24]. Thermograms in Fig. 4b–d show the de-convolution of the heat signals for experiments using the 429 L loop, in the linear (Fig. 4b), transition (Fig. 4c) and overloaded (Fig. 4d) zones of the isotherm. The obtained heat signals are characterized by an endothermic peak followed by two overlapping exothermic peaks. As stated above, more than a change in the heat signal magnitude and length the transition from a pulse injection (Fig. 3) to a continuous feed (Fig. 4)
influences the heat signal profile, suggesting the presence, in the adsorption mechanism, of an additional sub-process. This may be related to the fact that after the first 171 L (cell volume) of the injected sample flow through the cell, another 258 L still have to pass through it, interfering with the already bounded molecules. This mode of operation approaches the volume overloading used in some preparative chromatographic applications. In this case, to analyze the thermograms, it should be also considered the interaction between adsorbed pDNA molecules and free pDNA molecules in the feed solution. Despite of some repulsion between these molecules, this interaction promotes the early reorientation and subsequent secondary adsorption of already bound biomolecules, leaving free space for pDNA in solution to adsorb. This is further supported comparing, in the linear zone of the isotherm at approximately the same equilibrium pDNA loading concentration, the thermograms obtained using the 229 L and 429 L loops (Fig. 4b and e). Considering the 429 L heat signal and observing it until around 1680 s (where the 229 L loop pDNA plug ends) it can be seen that the profile of the peak is similar to the profile obtained with the 229 L loop. After that, the adsorption enters in the volume overloading process and the heat signal magnitude increases until the end of the pDNA plug. This increase in the magnitude, as previously stated, may be due the secondary adsorption of the bound biomolecules in addition with the adsorption of more molecules.
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
171
Table 1 Heat of adsorption for ln pDNA adsorption on FF Q-Sepharose at pH 8; flow rate: 1.5 mL/h; adsorbent sample size: 21.89 mg; and temperature: 25 ◦ C. Enthalpies were determined from the deconvoluted thermograms. Loop (L)
pDNA mass feed (g)
30
0.27 1.17 2.43 4.44 9.01 10.89
229
20.68 31.72
429
57.92 87.73 203.78
pDNA loading (g g−1 )
12.6 53.40 111.2 202.8 311.5 497.6
± ± ± ± ± ±
1.7 0.02 0.1 0.2 9.4 0.1
Endothermic peak (mJ g−1 )
Exothermic peaks (mJ g−1 )
HI
HII
1.16 2.16 1.05 0.85 0.38 0.31
± ± ± ± ± ±
0.33 0.37 0.02 0.16 0.02 0.02
−3.91 −3.45 −1.84 −1.33 −0.46 −0.34
HIII ± ± ± ± ± ±
Net heat of adsorption (mJ g−1 ) HII + HIII
HTotal −2.75 −1.29 −0.79 −0.48 −0.08 −0.03
0.19 0.55 0.05 0.13 0.02 0.02
0 0 0 0 0 0
−3.91 −3.45 −1.84 −1.33 −0.46 −0.34
0 0
−0.51 ± 0.16 −0.47 ± 0.13
−0.32 ± 0.18 −0.23 ± 0.14
−0.66 ± 0.12 −0.40 ± 0.02 −0.36 ± 0.08
−0.47 ± 0.13 −0.36 ± 0.02 −0.33 ± 0.08
820.4 ± 82.9 1222.5 ± 70.8
0.20 ± 0.08 0.23 ± 0.05
−0.51 ± 0.16 −0.47 ± 0.13
1359.6 ± 132.6 3599.0 ± 111.8 5489.4 ± 152.7
0.19 ± 0.05 0.05 ± 0.01 0.03 ± 0.01
−0.29 ± 0.11 −0.32 ± 0.02 −0.29 ± 0.08
−0.36 ± 0.06 −0.08 ± 0.02 −0.08 ± 0.02
± ± ± ± ± ±
0.19 0.55 0.05 0.13 0.02 0.02
± ± ± ± ± ±
0.38 0.66 0.05 0.20 0.02 0.02
HTotal = HI + HII + HIII .
Considering heat signals obtained in the transition and in the overloaded zones of the isotherm (Fig. 4c and d) and comparing them with a signal obtained in the linear zone (Fig. 4b) it can be seen that, under linear adsorption conditions, the first exothermic
event presents a lower absolute magnitude when compared to the second event, and this tendency is reversed for loadings in the transition and overloaded zones. This may be related to the quantity of molecules present in the cell, interacting with the adsorbent at the
Fig. 5. Heat of ln pDNA adsorption onto FF Q-Sepharose obtained using the (a) 30 and 229 L loops and (b) 429 L loop. Blue squares – endothermic heat; red circles – neat heat; black triangles – 1st exothermic heat; gray diamonds – 2nd exothermic heat. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
172
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173
same time. In the transition and overloaded zone, the high quantity of molecules does not favor the reorientation, reducing this process, indeed from Table 1 we can see that the enthalpy of adsorption of the first exothermic heat remains constant while its value for the second exothermic heat suffers a decrease. Fig. 5 presents the enthalpy changes for the ln pDNA adsorption onto Q-Sepharose, using the 30 and the 229 L loops (linear zone of the isotherm) (Fig. 5a) and the 429 L loop (linear, transition and overloaded zones of the isotherm) (Fig. 5b). Considering the net heat of adsorption (sum of all contribution), Table 1 and Fig. 5, since the Gibbs free energy (G = H − TS) has to be negative for a favorable interaction [10], the overall adsorption process is enthalpically driven, once it is exothermic for all the studied conditions, as expected for an anion-exchange interaction [14]. Previous studies, by van’t Hoff analysis, of supercoiled pDNA (4.9 kbp) onto Source 30 Q revealed an entropically driven process [22]. This emphasizes the widely different behaviors that systems can manifest, and the importance of characterizing each system independently, but also may indicate an extra entropic contribution arising from supercoiled pDNA alteration of conformation during adsorption, recall our study is made with ln pDNA. Under overloaded conditions, the net heat of adsorption appears to be constant (Fig. 5b), which is anticipated when the monolayer capacity is reached (Fig. 2) [14]. For the lower loading concentration, as ln pDNA loading increases the net heat of adsorption becomes more positive (Fig. 5a and b). Under these conditions, both endothermic and exothermic enthalpies decrease in magnitude. Endothermic heat major contribution was assumed to be from desolvation process. As the loading increases, each pDNA molecule foot print on the adsorbent surface may decrease in order to accommodate more molecules, as discussed pDNA upright position is preferred when binding under maximum loading conditions. Thus less water molecules and ions need to be removed in the binding process, leading to a decrease in the endothermic heat signal. Exothermic heat major contribution was assumed to result from the interaction between pDNA and Q-Sepharose and from secondary adsorption of already adsorbed molecules. The maintenance in the magnitude of the first exothermic heat (Fig. 5b) with the increase in loading concentration is expected, as a higher repulsion (endothermic contribution) between adsorbed molecules is predictable. The decrease in the magnitude of the second exothermic heat (Fig. 5b) is, as previously mention, anticipated once less reorientation results in less secondary adsorption.
4. Conclusions This study investigates the adsorption mechanism of linear pDNA onto FF Q-Sepharose considering linear, transition and overloaded adsorption conditions. Two modes of injection were used, pulse and continuous feed. Static binding capacity studies revealed that adsorption follows a type-I Langmuir isotherm. Theoretical binding capacity agreed well with experimental, indicating that interaction in upright position is preferred when under maximum loading conditions. Flow microcalorimetry shown that the overall adsorption process is exothermic, as expected for an anion-exchange interaction. Under overloaded conditions, the constant net heat of adsorption is compatible with the monolayer establishment [14]. Thermograms were analyzed considering the mechanism proposed by Yamamoto and co-workers [13]. Endothermic heat major contribution was suggested to be from water molecules and ions release (desolvation process), while major exothermic heats contributions were related to the electrostatic attraction between pDNA and Q-Sepharose and also to the secondary adsorption of already adsorbed molecules.
Under lower loading concentrations, the decrease in magnitude of the endothermic and exothermic enthalpies, with the increase of adsorbed pDNA, were respectively associated to a reduction in pDNA foot print on the adsorbent surface (less energy is used in water release) and with the repulsion between adsorbed molecules (increase in the endothermic heat contribution). Under higher loading concentrations, the maintenance in the magnitude of the first exothermic heat, with the increase in loading concentration, was justified by a balance between adsorption and repulsion (between adsorbed molecules) and the decrease of the second exothermic heat was related with less secondary adsorption. Funding sources Work supported by FCT (Portuguese Foundation for Science and Technology), project number FCOMP-01-0124-FEDER-014750 (Ref. FCT PTDC/EBB-BIO/113576/2009). References [1] A. Ghanem, R. Healey, F.G. Adly, Current trends in separation of plasmid DNA vaccines: a review, Anal. Chim. Acta 760 (2013) 1–15. [2] S. Ginn, I. Alexander, M. Edelstein, M. Abedi, J. Wixon, Gene therapy clinical trials worldwide to 2012 – an update, J. Gene Med. 15 (2013) 65–77. [3] L. Li, F. Saade, N. Petrovsky, The future of human DNA vaccines, J. Biotechnol. 162 (2012) 171–182. [4] F. Sousa, L. Passarinha, J.A. Queiroz, Biomedical application of plasmid DNA in gene therapy: a new challenge for chromatography, Biotechnol. Genet. Eng. Rev. 26 (2009) 83–116. [5] J. Stadler, R. Lemmens, T. Nyhammar, Plasmid DNA purification, J. Gene Med. 6 (Suppl. 1) (2004) S54–S66. [6] G.N.M. Ferreira, Chromatographic approaches in the purification of plasmid DNA for therapy and vaccination, Chem. Eng. Technol. 28 (2005) 1285–1294. [7] G.N.M. Ferreira, J.M. Cabral, D.M. Prazeres, Studies on the batch adsorption of plasmid DNA onto anion-exchange chromatographic supports, Biotechnol. Prog. 16 (2000) 416–424. [8] G.N.M. Ferreira, G.A. Monteiro, D.M. Prazeres, J.M. Cabral, Downstream processing of plasmid DNA for gene therapy and DNA vaccine applications, Trends Biotechnol. 18 (2000) 380–388. [9] X. Yang, J. Wang, X. Liu, Z. Shang, Study on system identification of linear and non-linear liquid chromatography separation, Anal. Lett. 34 (2001) 803–811. [10] F. Lin, W. Chen, M.T.W. Hearn, Thermodynamic analysis of the interaction between proteins and solid surfaces: application to liquid chromatography, J. Mol. Recognit. 15 (2002) 55–93. [11] M. Rabe, D. Verdes, S. Seeger, Understanding protein adsorption phenomena at solid surfaces, Adv. Colloid Interface Sci. 162 (2011) 87–106. [12] J.C. Bellot, J.S. Condoret, Review modelling of liquid chromatography equilibria, Process Biochem. 28 (1993) 365–376. [13] F.Y. Lin, C.S. Chen, W.Y. Chen, S. Yamamoto, Microcalorimetric studies of the interaction mechanisms between proteins and Q-sepharose at pH near the isoelectric point (pI) effects of NaCl concentration, pH value, and temperature, J. Chromatogr. A 912 (2001) 281–289. [14] G.L. Silva, F.S. Marques, M.E. Thrash, A.C. Dias-Cabral, Enthalpy contributions to adsorption of highly charged lysozyme onto a cation-exchanger under linear and overloaded conditions, J. Chromatogr. A 1352 (2014) 46–54. [15] T. Blaschke, A. Werner, H. Hasse, Microcalorimetric study of the adsorption of native and mono-PEGylated bovine serum albumin on anion-exchangers, J. Chromatogr. A 1277 (2013) 58–68. [16] A. Katiyar, S.W. Thiel, V.V. Guliants, N.G. Pinto, Investigation of the mechanism of protein adsorption on ordered mesoporous silica using flow microcalorimetry, J. Chromatogr. A 1217 (2010) 1583–1588. [17] J. Korfhagen, A.C. Dias-Cabral, M.E. Thrash, Nonspecific effects of ion exchange and hydrophobic interaction adsorption processes, Sep. Sci. Technol. 45 (2010) 2039–2050. [18] A.C. Dias-Cabral, J.A. Queiroz, N.G. Pinto, Effect of salts and temperature on the adsorption of bovine serum albumin on polypropylene glycol-Sepharose under linear and overloaded chromatographic conditions, J. Chromatogr. A 1018 (2003) 137–153. [19] J. Kim, R.J. Desch, S.W. Thiel, V.V. Guliants, N.G. Pinto, Energetics of biomolecule adsorption on mesostructured cellular foam silica, Microporous Mesoporous Mater. 170 (2013) 95–104. [20] P. Raje, N.G. Pinto, Importance of heat of adsorption in modeling protein equilibria for overloaded chromatography, J. Chromatogr. A 796 (1998) 141–156. [21] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and platinum, J. Am. Chem. Soc. 40 (1918) 1361–1403. [22] C. Tarmann, A. Jungbauer, Adsorption of plasmid DNA on anion exchange chromatography media, J. Sep. Sci. 31 (2008) 2605–2618. [23] M. Mahut, A. Gargano, H. Schuchnigg, W. Lindner, Chemoaffinity material for plasmid DNA analysis by high-performance liquid chromatography with condition-dependent switching between isoform and topoisomer selectivity, Anal. Chem. 85 (2013) 2913–2920.
P.A. Aguilar et al. / J. Chromatogr. A 1372 (2014) 166–173 [24] W. Chen, M. Lin, P. Lin, P. Tasi, Y. Chang, S. Yamamoto, Studies of the interaction mechanism between single strand and double-strand DNA with hydroxyapatite by microcalorimetry and isotherm measurements, Colloids Surf. A 295 (2007) 274–283. [25] J.M. Phillips, N.G. Pinto, Calorimetric investigation of the adsorption of nitrogen bases and nucleosides on a hydrophobic interaction sorbent, J. Chromatogr. A 1036 (2004) 79–86. [26] M. Ornelas, D. Loureiro, M.J. Araújo, E. Marques, C. Dias-Cabral, M. Azenha, F. Silva, Synthesis of glycylglycine-imprinted silica microspheres through different water-in-oil emulsion techniques, J. Chromatogr. A 1297 (2013) 138–145. [27] M.E. Thrash, N.G. Pinto, Characterization of enthalpic events in overloaded ionexchange chromatography, J. Chromatogr. A 944 (2002) 61–68. [28] M.E. Thrash, N.G. Pinto, Flow microcalorimetric measurements for bovine serum albumin on reversed-phase and anion-exchange supports under overloaded conditions, J. Chromatogr. A 908 (2001) 293–299. [29] M.E. Thrash, J.M. Phillips, N.G. Pinto, An analysis of the interactions of BSA with an anion-exchange surface under linear and non-linear conditions, Adsorpt. Int. Adsorpt. Soc. 10 (2004) 299–307. [30] F. Sousa, J.A. Queiroz, Supercoiled plasmid quality assessment by analytical arginine-affinity chromatography, J. Chromatogr. A 1218 (2011) 124–129. [31] C. Caramelo-Nunes, D. Bicho, P. Almeida, J.C. Marcos, C.T. Tomaz, Dynamic binding capacity and specificity of 3,8-diamino-6-phenylphenanthridineSepharose support for purification of supercoiled plasmid deoxyribonucleic acid, J. Chromatogr. A 1307 (2013) 91–98. [32] D.K. Chattoraj, K.S. Birdi, Adsorption and the Gibbs Surface Excess, Plenum Press, New York, 1984.
173
[33] G. Carta, A. Jungbauer, Protein Chromatography: Process Development and Scale-Up, Wiley-VCH, 2010. [34] M.C. Flickinger, Downstream Industrial Biotechnology: Recovery and Purification, Wiley-VCH, 2013. [35] A.M. Morão, J.C. Nunes, F. Sousa, M.T. Pessoa de Amorim, I.C. Escobar, J.A. Queiroz, Ultrafiltration of supercoiled plasmid DNA: modeling and application, J. Membr. Sci. 378 (2011) 280–289. [36] A. Morão, J.C. Nunes, F. Sousa, M.T.P. De Amorim, I.C. Escobar, J.A. Queiroz, Development of a model for membrane filtration of long and flexible macromolecules: application to predict dextran and linear DNA rejections in ultrafiltration, J. Membr. Sci. 336 (2009) 61–70. [37] A. Ljunglof, P. Bergvall, R. Bhikhabhai, R. Hjorth, Direct visualisation of plasmid DNA in individual chromatography adsorbent particles by confocal scanning laser microscopy, J. Chromatogr. A 844 (1999) 129–135. [38] C.R. Smith, R.B. DePrince, J. Dackor, D. Weigl, J. Griffith, M. Persmark, Separation of topological forms of plasmid DNA by anion-exchange HPLC: shifts in elution order of linear DNA, J. Chromatogr. B 854 (2007) 121–127. [39] J. Kim, R.J. Desch, S.W. Thiel, V.V. Guliants, N.G. Pinto, Energetics of protein adsorption on amine-functionalized mesostructured cellular foam silica, J. Chromatogr. A 1218 (2011) 7796–7803. [40] J. Kim, R.J. Desch, S.W. Thiel, V.V. Guliants, N.G. Pinto, Energetics of lysozyme adsorption on mesostructured cellular foam silica: effect of salt concentration, J. Chromatogr. A 1218 (2011) 6697–6704. [41] T.W. Perkins, D.S. Mak, T.W. Root, E.N. Lightfoot, Protein retention in hydrophobic interaction chromatography: modeling variation with buffer ionic strength and column hydrophobicity, J. Chromatogr. A 766 (1997) 1–14.
