Characterization of methacrylate chromatographic monoliths bearing affinity ligands

Accepted Manuscript Title: Characterization of methacrylate chromatographic monoliths bearing affinity ligands ˇ Author: Urh Cernigoj Urˇska Vidic Blaˇz Nemec Jernej ˇ Gaˇsperˇsiˇc Jana Vidiˇc Nika Lendero Krajnc Aleˇs Strancar Aleˇs Podgornik PII: DOI: Reference:

S0021-9673(16)31058-5 http://dx.doi.org/doi:10.1016/j.chroma.2016.08.014 CHROMA 357814

To appear in:

Journal of Chromatography A

Received date: Revised date: Accepted date:

19-5-2016 1-8-2016 6-8-2016

ˇ Please cite this article as: Urh Cernigoj, Urˇska Vidic, Blaˇz Nemec, Jernej Gaˇsperˇsiˇc, ˇ Jana Vidiˇc, Nika Lendero Krajnc, Aleˇs Strancar, Aleˇs Podgornik, Characterization of methacrylate chromatographic monoliths bearing affinity ligands, Journal of Chromatography A http://dx.doi.org/10.1016/j.chroma.2016.08.014 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Characterization of methacrylate chromatographic monoliths bearing affinity ligands

Urh Černigoja, Urška Vidica, Blaž Nemeca, Jernej Gašperšiča, Jana Vidiča, Nika Lendero Krajnca, Aleš Štrancara,b, Aleš Podgornik*b, c

a

BIA Separations d.o.o., Mirce 21, 5270 Ajdovščina, Slovenia COBIK, Tovarniška 26, 5270 Ajdovščina, Slovenia

b

c

Faculty for Chemistry and Chemical Technology, University of Ljubljana, Večna pot 113,

1000 Ljubljana, Slovenia

* correspondence: Aleš Podgornik, Faculty for Chemistry and Chemical Technology, University of Ljubljana, Večna pot 113, 1000 Ljubljana, Slovenia Tel.: +386 1 479 8584 E-mail: [email protected]

1

Highlights: 

pA ligand utilization decreases with increase of ligand density



pA ligand utilization is preserved for monoliths with different pore sizes



IgG binding capacity on pA ligand exceeded theoretical monolayer capacity



Binding capacity for monoliths bearing IgG ligands increases with antigen size

Abstract We investigated effect of immobilization procedure and monolith structure on chromatographic performance of methacrylate monoliths bearing affinity ligands. Monoliths of different pore size and various affinity ligands were prepared and characterized using physical and chromatographic methods. When testing protein A monoliths with different protein A ligand densities, a significant nonlinear effect of ligand density on dynamic binding capacity (DBC) for IgG was obtained and accurately described by Langmuir isotherm curve enabling estimation of protein A utilization as a function of ligand density. Maximal IgG binding capacity was found to be at least 12 mg/mL exceeding theoretical monolayer adsorption value of 7.8 mg/mL assuming hexagonal packing and IgG hydrodynamic diameter of 11 nm. Observed discrepancy was explained by shrinkage of IgG during adsorption on protein A experimentally determined through calculated adsorbed IgG layer thickness of 5.4 nm from pressure drop data. For monoliths with different pore size maximal immobilized densities of protein A as well as IgG dynamic capacity linearly correlates with monolith surface area indicating constant ligand utilization. Finally, IgGs toward different plasma proteins were immobilized via the hydrazide coupling chemistry to provide oriented immobilization. DBC was found to be flow independent and was increasing with the size of bound protein. Despite DBC was lower than IgG capacity to immobilized protein A, ligand utilization was higher.

2

Keywords: monoliths, affinity chromatography, protein A, antibodies, immobilization, protein A - IgG interaction, immunoaffinity

1 Introduction Affinity chromatography is an important chromatographic technique enabling very high product purity in a single purification step [1]. The most prominent example of its industrial utilization is protein A chromatography, which has become indispensable in production of therapeutic antibodies [2]. Immunoaffinity chromatography applications are less frequently found in industrial applications and are used, when all other chromatographic modes fail to result in high-quality product [3]. One of the main reasons for limiting occurrence of affinity chromatography in practice is high cost of the chromatographic support driven by ligand price. In order to be able to design an optimal affinity chromatographic column it is of great importance to understand the influence of immobilization procedure and the nature of chromatographic support on the chromatographic properties of the final product. The binding capacities of affinity chromatographic supports are usually lower than for ion exchangers, however the process productivity is mostly increased by shortening separation time. In diagnostic and analytical applications, high-throughput analyses are becoming indispensable due to the fast development of robotics and statistics tools, enabling analyses of large number of samples in very short time [4-6]. These reasons call for implementation of convective based chromatographic resins, such as monoliths, in affinity chromatography. Polymethacrylate monoliths afford flowrate independent binding capacity and resolution due to the convective nature of the flow, which allows relatively short analysis times compared to traditional chromatographic supports [7]. Combination of interconnected flow-through

3

channels (> 1.5 μm) modified with appropriate ligands offer binding sites that are easily accessible for large biomolecules [8]. Short column length in combination with bimodal channel structure [9] lead to a low pressure drop over the matrix bed, rendering them particularly useful for fast separation/enrichment and purification of large macromolecules, including IgG, IgM, pDNA and various viruses [10]. Convective flow properties make polymethacrylate monoliths a promising material also in different -omics applications due to the possibility of increasing the speed of analyses [11]. Having these advantages in mind we investigated properties of polymethacrylate monoliths with immobilized affinity ligands toward different blood plasma proteins. In particular, ligand density, monolith pore size, immobilization chemistry as well as nature of the ligands and antigens were varied and their influence on the chromatographic properties was studied.

2

Materials and methods

2.1 Materials

All solutions were freshly prepared using purified water which meets the requirements for European Pharmacopoeia (AQUATEHNA Biro, Zgornja Kungota, Slovenia) and analytical grade reagents. Buffer solutions were filtered through a 0.22 μm PES filter (TPP, Trasadingen, Switzerland). Sodium hydroxide (NaOH), 2-amino-2-hydroxymethyl-propane1,3-diol (TRIS), sodium dihydrogenphospahte, sodium hydrogenphosphate, formic acid, glycine, sodium periodate (NaIO4), adipic acid dihydrazide, 4-morpholineethanesulfonic acid (MES), ethanolamine, ethylene glycol, sodium chloride (NaCl), hydrochloric acid (HCl), polyclonal anti-Human Serum Albumin antibody (anti-HSA) were purchased from Sigma Aldrich (St. Louis, MO, USA). 1’,1’-carbonyldiimidazol (CDI) was purchased from Tokyo

4

Chemical Industry (Tokyo, Japan), and recombinant protein A was obtained from Biomedal (Seville, Spain). 96% ethanol was from KEFO (Ljubljana, Slovenija). Human IgG (Octagam 5%, Octapharma, Lachen, Switzerland), human serum albumin (HSA) and human holotransferrin (Sigma Aldrich, St. Louis, MO, USA), human fibrinogen (Merck-Millipore, Darmstadt, Germany) and human haptoglobin (Athens Research & Technology, Athens, GA, USA) were used as target antigens in blood plasma. Monoclonal anti-Human transferin antibody was obtained from Medical University of Rijeka (Rijeka, Croatia), while monoclonal anti-Human fibrinogen and anti-Human haptoglobin antibodies were received from Biotechnical faculty, University of Ljubljana (Ljubljana, Slovenija). Amicon Ultra-4, MWCO10000 centrifugal filters were from EMD Millipore (Billerica, MA, USA).

2.2 Columns and chromatographic equipment

All chromatographic experiments were carried out using a gradient chromatography workstations, consisting of two pumps with 10 mL pump heads, optionally an autosampler with various sample loop volumes and a UV detector – Smartline (Knauer, Berlin, Germany). For data acquisition and control, Chromgate software (Knauer) was used. CIMac™ monolith columns with different pore sizes ranging from average of 770 nm to average of 2324 nm (disc dimensions: 5.2 mm (I.D.) x 5 mm length, volume 0.106 mL) were provided by BIA Separations d.o.o. (Ajdovščina, Slovenia). Hydrazide- and CDI-modified CIMac™ columns were used for the immobilization of antibodies and protein A, respectively. For quantification of protein A concentration in immobilization solutions during the immobilization procedures a strong anion exchange analytical column (CIMac™ QA) was used.

5

2.2.1 Protein A immobilization The stock solution of protein A (50 mg/mL) was diluted in 50 mM phosphate buffer with 0.5 M NaCl, pH 7.4 to the final concentration ranging from 0.1 mg/mL to 2 mg/mL. The solution (1 mL) was pumped through the CIMac™ CDI column at 0.1 mL/min and then cycled through the monolith for 24 h. The concentration of protein A in immobilization solution together with the available surface area dictated the density of bound protein A on the monolith. The deactivation of remaining CDI groups was done by treating the monolith with 2 M ethanolamine solution, pH 9.0, for 24 h at room temperature. The amount of immobilized protein A was obtained from the quantification of the protein A in the immobilization solution before and after immobilization using anion-exchange chromatographic analysis (CIMac™ QA column). The sample aliquot was 25-times diluted with 50 mM TRIS, pH 8.5. 100 µL of the sample was injected on the column, conditioned with 20 mM TRIS, pH 7.4. A 3 min long linear gradient was performed to 20 mM TRIS, 1 M NaCl, pH 7.4 at 1 mL/min. The detection wavelength was 230 nm. Quantification of protein A was performed using a calibration curve with pure protein A samples of known concentrations. A serial dilution of protein A (between 0.05 and 1 mg/mL) was prepared by diluting a stock solution (50 mg/mL) with 20 mM TRIS, pH 7.4. Each of prepared samples was additionally 25-times diluted with 50 mM TRIS, pH 8.5 before the injection on the column.

2.2.2 Immobilization of antibodies

The antibody stock was dissolved in 10 mM phosphate buffer (pH 7.0), containing 100 mM NaCl, to a final antibody concentration of 2 mg/mL. Then, the dissolved antibody stock was mixed with a solution of 20 mM NaIO4 in phosphate buffer at ratio of 1:1 (v/v) and

6

thermostated for 30 min at 25°C to achieve the oxidation of glycosylated moieties of the antibody to aldehyde functional groups. The reaction was quenched by addition of ethylene glycol (25 µL of ethylene glycol per mL of the antibody solution) to reaction mixture. Further, twenty-fold dilution of the antibody solution was performed with 50 mM MES buffer, pH 5.2. The immobilization solution, containing 1 mg of dissolved antibodies, was pumped through a preconditioned CIMac hydrazide column at 0.5 mL/min. The column was closed with blind stoppers and thermostated for 15 hours at 25°C. The column was finally washed using 50 mM MES, 1 M NaCl solution (pH 5.2) and stored in phosphate buffered saline (PBS) with added 0.02 % NaN3 until required. The amount of antibodies in immobilization solutions before and after the immobilization and consequently the immobilized ligand densities were determined with UV spectroscopic analysis at wavelength of 280 nm on Smart Spec 3000 spectrophotometer (Bio-Rad, Richmond, USA)

2.3 Monolith characterization

2.3.1 Specific surface area

Specific surface area of the monoliths was measured via nitrogen adsorption by TriStar II 3020 (Micromeritics Instrument Corporation, Norcross, GA, USA). Nitrogen of 99.999% purity was used. Before analyses monolith samples were dried in nitrogen flow at 70°C for 1 hour.

2.3.2 Dynamic binding capacity

7

Dynamic binding capacity (DBC) was determined with frontal analysis experiments. A protein sample, dissolved in PBS at pH 7.2, was loaded on the CIMac affinity columns at 1.0 mL/min until reaching a breakthrough. Dynamic binding capacity at 50% breakthrough (DBC50) was calculated from t50% at which UV280 was half-maximal according to the Eq. (1): 𝐷𝐵𝐶50 =

(𝑡50% ×∅−𝑉𝑑 )×𝑐0 𝑉𝑐

(1)

Φ represents the flow rate (mL/min), t50% is the time where the final absorbance value reached the 50% of the breakthrough curve, Vd is the dead volume of the system (mL), c0 is an initial protein concentration (mg/mL) and Vc the total monolith volume. Details for particular loading are described in appropriate figure caption. Although Eq. (1) is correct only for symmetric break through curves not encountered in short bed monoliths, it was used for calculation of dynamic binding capacity for consistency with data published elsewhere. This can be justified by the fact that as long as the shape of breakthrough curve is constant, what is the case for same type of molecules, constant systematic error is introduced but trend is not affected.

2.3.3 Pressure drop measurements and determination of IgG layer thickness

The differential pressure on the monolithic column during the IgG loading and elution was recorded by differential manometer (Mid-West Instruments, Sterling Heights, MI, USA). The pressure drop of an empty housing in the same buffer and at the same flow rate was subtracted from the final values in order to obtain the information of the pressure drop on the monolith itself. Measurement of pressure drop on original and loaded monolith was measured under same flow rate of 1.0 mL/min and same buffer (PBS). For calculation of adsorbed layer thickness, Eq. (2) was used [12-14].

8

  h  r  1   



 2  P PA   4  P PA  1       P PA   2  1    

(2)



where h is layer thickness, ΔP and ΔPA is pressure drop of original and monolith with adsorbed molecules respectively, r is half of pore size and ɛ is monolith porosity.

3

Results and discussion

3.1 Effect of ligand density

Effect of ligand density on dynamic binding capacity and estimation of ligand utilization is very important for optimization of column performance but also for its production costs and consequently economy of the process in which the column is used. In comparative study involving methacrylate monoliths and various particulate supports it was shown that due to an open channel structure, utilization of ligand immobilized on methacrylate monoliths is high even when no spacer is used [15], therefore also in this study ligands were immobilized directly on chemical moieties present on the monolith skeleton. To investigate the influence of a ligand density on DBC, monoliths with average pore size of 1400 nm bearing carboxyimidazole reactive moieties were used. Ligand densities from 8.3 up to 100 % of immobilized protein A were prepared (100 % represents maximal immobilized amount or maximal ligand density). Immobilized ligand was recombinant protein A containing four identical copies of an Fc region-binding domain assembled in a single 29.88 kDa polypeptide. Dynamic binding capacity was determined with human plasma IgG. Results are plotted in Figure 1A.

9

As we can see, there is a significant effect of ligand density on DBC being highly non-linear, similar as reported for particulate protein A support [16]. DBC change seems to follow saturation-like trend therefore we tested if it can be described by a Langmuir type equation. To verify this, reciprocal values of DBC against reciprocal values of ligand density (Eq. 3) were plotted (Figure 1B) resulting in high linear correlation index (R2 = 0.993). 1 𝑞

=𝑞

𝐾

𝑚𝑎𝑥

1

+𝑞 𝑆

1

𝑚𝑎𝑥

(3)

where q is DBC50 of IgG, qmax is maximal DBC, S is immobilized ligand concentration and K is equation parameter.

To investigate generality of this finding, we applied the same equation on literature data where static binding capacity was determined for three different IgGs adsorbed on same protein A matrix, namely ProSepA Ultra High Cap Protein A media from Millipore Corporation [16] differing however in immobilized protein A density. Reciprocal values of ligand density vs. reciprocal values of static binding capacity for all three IgGs are plotted in Figure 2.

Although one cannot directly compare parameters values of fitting equations due to different axis values, linear trend for at least two of three tested IgGs indicate that phenomenon is not related to the matrix structure but it was explained by inter-ligand steric effects and the steric hindrance effects seen between the multiple binding sites on a single Protein A ligand [16]. However, it is not obvious why trend follows typical adsorption isotherm curve and can therefore be described by a Langmuir type equation. While one would expect limited maximal DBC, being determined by resin available surface area, high non-linear dependence even at low ligand density is rather unexpected.

10

Pronounced non-linear dependence indicates that ligand utilization is not constant but it is substantially affected by a ligand density. As experimental data seems to be accurately described by Langmuir type equation, its derivative can be used to estimate ligand utilization for a given ligand density: 𝑑𝑞 𝑑𝑆

𝐾

= 𝑞𝑚𝑎𝑥 (𝐾+𝑆)2

(4)

Ligand utilization is commonly defined as number of molecules bound by a single ligand molecule, however, since no absolute value for protein A immobilized amount is provided, only IgG binding capacity was converted into milimolar concentration using mw of 153 kDa for applied polyclonal human IgG while immobilized protein A density remained in %. For same reason also ligand utilization was shown in % of maximal utilization. Experimental data of IgG DBC in milimolar concentration together with estimated relative ligand utilization calculated from Eq. 4 is plotted in Figure 3.

As one can expect maximal ligand utilization is achieved at lowest ligand density, approaching value of qmax/K (limit of Eq. 4 when S approaches to zero). More surprisingly, there is a rapid decrease of ligand utilization found for both, particulate support and monolithic support attributed to inter-ligand steric effects and the steric hindrance effects seen between the multiple binding sites on a single Protein A ligand [16]. This can be expected when immobilized and/or adsorbed molecules are close nearby. Therefore, we estimated average distance between adsorbed IgG molecules. Due to affinity binding mechanics monolayer adsorption can be assumed. Monoliths with pore size 1400 nm exhibit surface area of around 3.2 m2/mL [9] and hydrodynamic diameter of 11 nm was taken for IgG [17]. Assuming the highest possible packing of IgG molecules on the surface (hexagonal packing) [18] maximal binding capacity is calculated to be only 7.8 mg IgG/mL, being therefore lower of experimentally determined values. This surprising finding might indicate multilayer

11

adsorption in affinity mode, which doesn’t seem very likely and different possible explanation will be discussed later when pressure drop data are analyzed. Based on this analysis it can be concluded that IgG molecules are tightly packed and therefore proposed explanation of ligand utilization decrease seems to be plausible. Mathematically speaking, these effects are reflected in value of K parameter (see Eqs. 3, 4) accounting for contributions of intermolecular interactions between immobilized ligand but also between adsorbed molecules. If effect of former would be prevalent, K value could be used as a measure of immobilization efficiency while dominance of interaction between IgG molecules would mainly reflect their surface properties. Data in Figure 2B demonstrates certain effect of IgG type on K values being 0.63, 0.50 and 0.90 (calculated from correlation in Figure 2 and Eq. 3), which is not however very pronounced. Further work on various protein A resins and IgG samples is needed to discriminate between these two options and estimate importance of each. Based on above discussion it is clear that amount of immobilized protein A significantly affects DBC. To minimize its effect, further experiments investigating influence of other parameters were performed with monoliths bearing highest density of immobilized protein A.

3.2 Effect of the monolith pore size

Pore size of convective chromatographic supports affects various properties such as pressure drop, surface area and also potential blocking of the chromatographic support. If pores are too small, monoliths can act like filters, preventing molecules to pass [19] or they do enter but they are reversibly retained inside the monolith due to bimodal pore size distribution [20] that can also contribute to a separation mechanism [21]. As pressure drop increases with a square of pore size decrease [22] high pressure on monoliths with small pore size can become a bottleneck on preparative scale. On the other hand, pore size also determines surface area and

12

since microscopic structure of methacrylate monoliths is preserved over various volumes [23] it was demonstrated that there is a linear correlation between a pore size and specific surface area as well as between surface area and binding capacity for ion-exchange but also hydrophobic interaction mode [24]. Therefore experimental verification of this trend on affinity monoliths seems to be natural. Monoliths with four different pore sizes were prepared by varying polymerization temperature [25] and same protein A ligand was immobilized up to maximal ligand density. Assuming that highest ligand density is independent on the pore size, immobilized amount should be proportional to the accessible surface area. Furthermore, constant ligand density should provide constant ligand utilization and consequently resulting in linear correlation between amount of immobilized ligand and bound IgG. As accessible surface is inversely proportional to a square of a pore size, it means that this should be also true for IgG’s DBC. Experimental results shown in Figure 4 confirm correctness of above assumptions, proving that maximal binding capacity is determined by accessible surface area.

3.3 Pressure drop on CIMac™ protein A columns

Another important parameter to be monitored during implementation of monolithic columns is pressure drop being inversely proportional to the square of particle diameter for particulate matrixes or inversely proportional to the square of pore size in case of monoliths [9, 22, 24] Furthermore, pressure drop is further increased during molecule adsorption on the surface, due to a decrease of porosity and pore diameter, being more pronounced on matrix having smaller pores [13, 14]. Both effects have to be taken into account during process design especially on a preparative scale, to avoid operational pressure limit to be reached. Because of that we measured pressure drop of affinity monoliths before sample loading and its increase

13

for saturated monolith. Same monoliths as for measurement of pore size influence were used. Results are summarized in Table 1. Pressure drop of the monolith is expectedly inversely proportional to square of pore size as already reported [24] enabling calculation of pressure drop for given linear velocity, pore size and monolith dimensions. Also, there was expected increase of pressure during loading ranging from 0.07 bar (7.5 %) for largest pores up to 2.68 bar (26 %) for smallest pores. Higher absolute and relative increase could be anticipated when same molecules forming adsorbed layer of same thickness are loaded as thickness to pore ratio increases too. We estimated thickness of adsorbed IgG layer using Eq. 2 and similar values for all pore sizes were obtained, confirming reliability of thickness estimation from pressure drop data. However, average estimated layer thickness is only 5.4 nm, what is only half of expected IgG hydrodynamic diameter, being of around 11 nm [17]. This discrepancy could be explained by recent finding that adsorption on protein A induces a conformational IgG variant with about half the hydrodynamic size of the native IgG when eluted from the column with low pH buffer [17]. This process was demonstrated to be a consequence of binding to protein A ligand since no decrease of size was observed during sole exposure to low pH buffer, and proposed mechanism was denaturation of the upper third of the C2 domain under physiological conditions by its initial contact with protein A. As our results of adsorbed layer thickness were also obtained during exposure of IgG to a loading buffer and estimated size matches closely the values reported from DLS measurements [17] they seem to confirm proposed mechanism and this phenomenon can also explain higher packing of IgG on a surface and by this previously described discrepancy of experimentally measured DBC and upper predicted theoretical value for monolayer adsorption. Regardless the true nature of IgG affinity interaction mechanism with protein A support, experimentally obtained DBC is above the predicted value for monolayer adsorption. From

14

this point it would be meaningful to investigate opposite combination that is to immobilize antibody and capturing its antigen from the sample solution. Polyclonal human IgG was immobilized in a non-oriented way on various pre-activated supports (CDI, epoxy, aldehyde) and immobilization amount was up to 8 mg/mL matching the predicted maximal possible density. Interestingly, protein A capacity for those columns was extremely low if present at all, indicating multipoint attachment of IgG onto the surface that prevents binding of protein A to IgGs heavy chain, possibly because of lack of necessary IgG conformational changes, when interacting with protein A (see previous paragraph). While capacity for protein A is of no practical interest, immobilized IgGs against various plasma proteins might have significant importance in analytics, isolation or depletion, therefore we focused on preparation and characterization of several such columns.

3.4 Immunoaffinity monoliths

In order to determine DBC in affinity isolation of blood plasma proteins we performed immobilization of antibodies against four important plasma proteins, namely HSA, transferrin, fibrinogen and haptoglobin on monoliths with a pore size size 1400 nm. As structure and size of immobilized antibodies is very similar, comparable immobilized amount can be expected. In fact, for all IgGs same maximal ligand density up to 8 mg/mL was obtained, regardless used immobilization chemistry (CDI or hydrazide). As this value is always lower when compared to maximal binding capacity for IgG on protein A column and within expected range for monolayer binding, this is an indication that no conformational changes of IgG should occur during immobilization. Effect of immobilization chemistry on DBC was tested on monoliths with immobilized antiHSA antibody. Besides non-oriented CDI-based immobilization chemistry we also used

15

hydrazide coupling [26, 27] and, as expected, non-oriented CDI immobilization resulted in lower DBC for HSA (approximately 3-fold) in comparison to hydrazide coupling despite comparable immobilized ligand density. Hydrazide coupling enables oriented immobilization of antibodies, extending their Fab region in the lumen of monolith pores, by this facilitating affinity interaction with antigens and minimizing steric hindrances, therefore maximizing DBC. Because of that only monoliths with oriented immobilization of antibodies were further explored for DBC and data are presented in Table 2. However, none of measured DBC values is even close to the one obtained for IgG on protein A column, again confirming peculiarity of this type of adsorption. Despite similarity of immobilized IgG ligands, size of target plasma proteins differs substantially. It is therefore reasonable to assume that maximal dynamic binding capacity for antigen would differ according to their hydrodynamic radius as discussed previously for adsorbed IgG and confirmed experimentally for various ion exchange resins [28]. Clearly, this prediction assumes no conformational changes of the molecule during adsorption as this seems to be the case for IgG, indicating that each system has to be investigated experimentally. Data confirm that binding capacity indeed increases with increase of protein size, while ligand utilization shows the opposite trend. Partially this could be explained by the fact that IgG ligand is larger than HSA and transferrin, therefore no packing problem is expected. This is not the case for fibrinogen resulting in lowest ligand utilization, what could be analogized with IgG-immobilized protein A pair, where the analyte is larger than the immobilized ligand. No calculation for haptoglobin was made due to its broad range of molecular masses reported in the literature and lack of data for specific used protein.

3.5 Flow rate dependence

16

Finally, in order to obtain high productivity but also to perform fast analysis, immunoaffinity isolation should be performed at high flow rate. Although there are no diffusion limitations present in the monoliths due to convective mass transport, in case of large ligands, there might be some diffusional limitations for accessing their active sites due to vicinity of other ligand molecules and slower intrinsic binding kinetics that could result in decreasing binding capacity with flow rate increase. Experiments performed for anti-HSA monolith column demonstrated that DBC was independent on the flow rate, at least up to 1 mL/min (Figure 5) that corresponds to linear velocity of 282 cm/h covering the entire recommended linear velocity range for all CIM monolithic columns [23]. The first (small) breakthrough appears when sample reaches the detector (volume of the column and HPLC tubing). It is a consequence of impurities present in HSA sample (confirmed by SDS-PAGE analysis) which do not bind to the column while second break through represents column saturation with HSA.

4 Conclusions

Due to their flow unaffected properties, affinity monoliths can be implemented for HTS, PAT and also isolation, if dynamic binding capacity is sufficiently high. Investigation of immobilized protein A density revealed substantial decrease of ligand utilization with increase of ligand density. This indicates that protein A ligand density should be carefully adjusted during preparation of affinity monoliths. Interestingly, IgG DBC exceeded theoretical monolayer capacity based on estimated surface area what was explained by decreased IgG hydrodynamic diameter estimated from pressure drop data. Although this seems to be specific to protein A – IgG interaction it indicates that experimental measurements are needed to get accurate data, as extrapolations from available surface area and molecule size can be

17

misleading. Since ligand density and utilization were preserved for monoliths with different pore sizes, affinity monoliths can be optimized for particular application to exhibit optimal performance. Due to demonstrated flow independent properties, fast analysis and high throughput isolation seem to be main advantages of these supports. Additionally, immunoaffinity monoliths were successfully prepared with different immobilized polyclonal and monoclonal antibodies. Their chromatographic characterization revealed substantial and flow rate independent antigen binding capacities, making such products attractive not only for high throughput isolation, but also for some delicate downstream processes.

Acknowledgements

Sandra Kontrec, Miroslava Legiša and Maša Velikonja from BIA Separations are acknowledged for their help during the laboratory work. This work was financially supported in part by the Slovenian Research Agency (ARRS) within the National Research and Development Programme P4-0369 and project L4-7628. Support of this research by the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreements n° 312004, n° 278535 and n° 324400 is also gratefully acknowledged. The Centre of Excellence for Biosensors, Instrumentation and Process Control is an operation financed by the European Union, European Regional Development Fund and Republic of Slovenia, Ministry of Education, Science and Sport.

18

References [1] J. Sproß, A. Sinz, Monolithic media for applications in affinity chromatography, J. Sep. Sci. 34 (2011) 1958–1973. DOI: 10.1002/jssc.201100400. [2] P. Gagnon, Technology trends in antibody purification, J. Chromatogr. A 1221 (2012) 57– 70. DOI: 10.1016/j.chroma.2011.10.034. [3] L. Milá, R. Valdés, A. Tamayo, S. Padilla, W. Ferro, Application of a risk analysis method to different technologies for producing a monoclonal antibody employed in Hepatitis B vaccine manufacturing, Biologicals. 40 (2012) 118–128. DOI: 10.1016/j.biologicals.2011.12.004. [4] D. Josic, J.G. Clifton, Use of monolithic supports in proteomics technology, J. Chromatogr. A 1144 (2007) 2–13. DOI: 10.1016/j.chroma.2006.11.082 [5] M. Pucić M, A. Knezević, J. Vidic, B. Adamczyk, M. Novokmet, O. Polasek, O. Gornik, S. Supraha-Goreta, M.R.Wormald, I. Redzić, H. Campbell, A. Wright, N.D. Hastie, J.F. Wilson, I. Rudan, M. Wuhrer, P.M. Rudd, D. Josić, G. Lauc, High throughput isolation and glycosylation analysis of IgG-variability and heritability of the IgG glycome in three isolated human populations, Mol. Cell. Proteomics. 10 (2011) DOI: 10.1074/mcp.M111.010090. [6] Z. Zhang, S. Wu, D. L. Stenoien, L. Paša-Tolić, High-Throughput Proteomics, Annu. Rev. Anal. Chem. 7 (2014) 427–454. DOI: 10.1146/annurev-anchem-071213-020216. [7] M. O. Herigstad, S. Dimartino, C. Boi, G. C. Sarti,Experimental characterization of the transport phenomena, adsorption, and elution in a protein A affinity monolithic medium, J. Chromatogr. A 1407 (2015) 130–138. DOI:10.1016/j.chroma.2015.06.045. [8] A. Zöchling, R. Hahn, K. Ahrer, J. Urthaler, A. Jungbauer, Mass transfer characteristics of plasmids in monoliths, J. Sep. Sci. 27 (2004) 819–27. DOI: 10.1002/jssc.200401777. [9] I. Mihelič, D. Nemec, A. Podgornik, T. Koloini, Pressure drop in CIM disk monolithic columns, J. Chromatogr. A 1065 (2005) 59–67. DOI:10.1016/j.chroma.2004.10.054 [10] A. Podgornik, S. Yamamoto, M. Peterka, N. Lendero Krajnc, Fast separation of large biomolecules using short monolithic columns, J. Chromatogr. B 927 (2013) 80–89. DOI: 10.1016/j.jchromb.2013.02.004 [11] T. Panić-Janković, U. Černigoj, B. Nemec, U. Vidic, J. Vidič, R. Sekirnik, N. Lendero Krajnc, W.-Q. Chen, B. Mahlberg, G. Mitulović, CIMac-aHSA Monoliths Increase Throughput and Sensitivity of Proteomic Studies of IVF Media by Rapid Albumin Depletion, Chromatogr. Today. 8 (2015) 36–40. [12] V. Frankovič, A. Podgornik, N. Lendero Krajnc, F. Smrekar, P. Krajnc, A. Štrancar, Characterisation of grafted weak anion-exchange methacrylate monoliths, J. Chromatogr. A 1207 (2008) 84–93. DOI: 10.1016/j.chroma.2008.08.027.

19

[13] N. Lendero Krajnc, F. Smrekar, A. Štrancar, A. Podgornik, Adsorption behavior of large plasmids on the anion-exchange methacrylate monolithic columns, J. Chromatogr. A 1218 (2011) 2413–2424. DOI:10.1016/j.chroma.2010.12.058. [14] F. Smrekar, M. Ciringer, A. Štrancar, A. Podgornik, Characterisation of methacrylate monoliths for bacteriophage purification, J.Chromatogr. A 1218 (2011) 2438–2444. DOI:10.1016/j.chroma.2010.12.083. [15] R. Hahn, E. Berger, K. Pflegerl, A. Jungbauer, Directed immobilization of peptide ligands to accessible pore sites by conjugation with a placeholder molecule, Anal. Chem. 75 (2003) 543–548. [16] S. Ghose, B. Hubbard, S.M. Cramer, Binding capacity differences for antibodies and Fcfusion proteins on protein A chromatographic materials, Biotechnol. Bioeng. 96 (2007) 768– 779. DOI: 10.1002/bit.21044. [17] P. Gagnon, R. Nian, D. Leong, A. Hoi, Transient conformational modification of immunoglobulin G during purification by protein A affinity chromatography, J. Chromatogr. A 1395 (2015) 136–142. DOI: 10.1016/j.chroma.2015.03.080. [18] H.N. Endres, J.A.C. Johnson, C.A. Ross, J.K. Welp, M.R. Etzel, Evaluation of an ionexchange membrane for the purification of plasmid DNA, Biotechnol. Appl. Biochem. 37 (2003) 259–266. DOI: 10.1042/BA20030025. [19] K. Benčina, M. Benčina, A. Podgornik, A. Štrancar, Influence of the methacrylate monolith structure on genomic DNA mechanical degradation, enzymes activity and clogging, J. Chromatogr. A 1160 (2007) 176–183. DOI: 10.1016/j.chroma.2007.05.034. [20] H. Koku, R.S. Maier, K.J. Czymmek, M.R. Schure, A.M. Lenhoff, Modeling of flow in a polymeric chromatographic monolith, J. Chromatogr. A 1218 (2011) 3466–3475. DOI: 10.1016/j.chroma.2011.03.064. [21] B. Gabor, U. Černigoj, M. Barut, A. Štrancar, Reversible entrapment of plasmid deoxyribonucleic acid on different chromatographic supports, J. Chromatogr. A 1311 (2013) 106– 114. DOI: 10.1016/j.chroma.2013.08.075. [22] M. Barut, A. Podgornik, M. Merhar, A. Štrancar, Short monolithic columns – rigid discs, in: F. Švec, T.B. Tennikova, Z. Deyl (Eds.), Monolithic materials: preparation, properties, and applications, Elsevier, Amsterdam, 2003, pp. 51. [23] A. Podgornik, J. Jančar, M. Merhar, S. Kozamernik, D. Glover, K. Čuček, M. Barut, A. Štrancar, Large-scale methacrylate monolithic columns: design and properties, J. Biochem. Biophys. Methods 60 (2004) 179–189. DOI: 10.1016/j.jbbm.2004.01.002. [24] A. Podgornik, V. Smrekar, P. Krajnc, A. Štrancar, Estimation of methacrylate monolith binding capacity from pressure drop data, J. Chromatogr. A 1272 (2013) 50–55). DOI: 10.1016/j.chroma.2012.11.057.

20

[25] F. Švec, J.M.J. Frechet, Kinetic control of pore formation in macroporous polymers. The formation of "molded" porous materials with high flow characteristics for separations or catalysis, Chem. Mater. 7 (1995) 707–715. [26] P. Brne, Y.P. Lim, A. Podgornik, M. Barut, B. Pihlar, A. Štrancar, Development and characterization of methacrylate-based hydrazide monoliths for oriented immobilization of antibodies, J. Chromatogr. A 1216 (2009) 2658–2663. DOI: 10.1016/j.chroma.2008.11.005. [27] I.A. Tarasova, A.A. Lobas, U. Černigoj, E.M. Solovyeva, B. Mahlberg, M. Ivanov, T. Panić-Janković, Z. Nagy, M.L. Pridatchenko, A. Pungor, B. Nemec, U. Vidic, J. Gašperšič, N. Lendero Krajnc, J. Vidič, M.V. Gorshkov, G. Mitulović, Depletion of human serum albumin in embryo culture media for in vitro fertilization using monolithic columns with immobilized antibodies, Electrophoresis, accepted for publication, DOI: 10.1002/elps.201500489. [28] S. Yamamoto, A. Kita, Rational Design Calculation Method for Stepwise Elution Chromatography of Proteins, Food Bioprod. Process. 84 (2006) 72–77. DOI: 10.1205/fpb.05180. [29] J. K. Armstrong, R. B. Wenby, H. J. Meiselman, T. C. Fisher, The Hydrodynamic Radii of Macromolecules and Their Effect on Red Blood Cell Aggregation, Biophys. J. 87 (2004) 4259–4270. DOI: 10.1529/biophysj.104.047746.

21

Figure captions:

Figure 1: A) IgG dynamic binding capacity as a function of immobilized protein A density. B) Reciprocal value of dynamic binding capacity (mmol/mL) vs. reciprocal value of protein A ligand density (1/%) for monolith (data are taken from Figure 1A) Stationary phase: CIMac protein A (1400 nm average pores size) with different ligand densities. Binding buffer: PBS, pH 7.2. Elution buffer: 0.1 M glycine, pH 2.0. Sample: polyclonal human IgG (0.5 mg/mL). Q = 1.0 mL/min; λ = 280 nm.

Figure 2: Reciprocal value of dynamic binding capacity vs. reciprocal value of protein A ligand density for three different IgG’s on particulate protein A matrix Data obtained from reference [16].

Figure 3: IgG DBC and relative ligand utilization as a function of relative protein A density.

Figure 4: Correlation between relative amount of immobilized protein A and adsorbed IgG vs. reciprocal value of a pore size diameter square. Monolith with pore size distribution maximum of 770, 1000, 1450 and 2324 nm were used. Chromatographic conditions used are the same as in described in Figure 1.

Figure 5: Dynamic binding capacity (DBC) measurements using CIMac hydrazide columns with immobilized anti-HSA at different flow rates. Applied analyte: HSA (0.25 mg/mL) in PBS buffer. Binding buffer: PBS, pH 7.2. Elution buffer: 0.1 M glycine buffer, pH 2.0. Q = 0.1, 0.4 and 1 mL/min. λ = 280 nm.

22

23

Fig.1

24

Fig.2

25

Fig.3

26

Fig.4

27

Fig.5

28

Tables:

Table 1: Pressure drop and estimated IgG adsorbed layer thickness for Protein A monoliths of different pore size Average

Pressure

Pressure

Absolute

Relative

Estimated

pore size

drop on

drop on

pressure

pressure

IgG

(nm)

original

monolith

drop

drop

adsorbed

monolith

with

increase

increase

layer

[bar]

adsorbed

[bar]

[%]

thickness

IgG [bar]

[nm]

770

10.8

13.48

2.68

24.8

6.1

1000

6.05

6.8

0.75

12.4

4.2

1450

4.03

4.45

0.42

10.4

5.2

2324

0.93

1

0.07

7.5

6.0

29

Table 2: DBC50 for different plasma proteins using immunoaffinity chromatographic monoliths. Columns: CIMac hydrazide columns with immobilised anti-HSA, anti-fibrinogen, antitransferrin and anti-haptoglobin antibodies. Pore size 1440 nm. Binding buffer: PBS, pH 7.2. Elution buffer: 0.1 M formic acid, pH 2.4. Sample: 0.1 mg/mL of antigen (HSA, fibrinogen, transferrin, haptoglobin) in PBS. Q = 1.0 mL/min, λ = 280 nm Adsorbed

M (antigen)

Hydrodynamic DBC50

Ligand

target molecule

(kDa)

radius [29]

utilizationa

(mg/mL)

(%) HSA

66

3.51

1.35

57

Fibrinogen

340

10.95

4.5

37

Transferrin

80

3.72

3.5

122

Haptoglobin

between 100 -

between 4.7-

2.2

n.c.

900

25

a

1:1 molar ratio between antibody and antigen was used for ligand utilization calculation

despite the fact that each antibody has two antigen binding sites.

30