Confocal Raman Microscopy of Protein Adsorbed in Chromatographic

Jul 13, 2012 - EMD Millipore Corporation, 80 Ashby Road, Bedford, Massachusetts 01730, United States. •S Supporting Information. ABSTRACT: Confocal ...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/ac

Confocal Raman Microscopy of Protein Adsorbed in Chromatographic Particles Yuewu Xiao,* Thomas Stone, David Bell, Christopher Gillespie, and Marta Portoles EMD Millipore Corporation, 80 Ashby Road, Bedford, Massachusetts 01730, United States S Supporting Information *

ABSTRACT: Confocal Raman microscopy is a nondestructive analytical technique that combines the chemical information from vibrational spectroscopy with the spatial resolution of confocal microscopy. It was applied, for the first time, to measure conformation and distribution of protein adsorbed in wetted chromatographic particles. Monoclonal antibody was loaded into the Fractogel EMD SO3 (M) cation exchanger at 2 mS/cm or 10 mS/cm. Amide I and III frequencies in the Raman spectrum of the adsorbed protein suggest that there are no detectable changes of the original β-sheet conformation in the chromatographic particles. Protein depth profile measurements indicate that, when the conductivity is increased from 2 mS/cm to 10 mS/cm, there is a change in mass transport mechanism for protein adsorption, from the shrinking-core model to the homogeneous-diffusion model. In this study, the use of confocal Raman microscopy to measure protein distribution in chromatographic particles fundamentally agrees with previous confocal laser scanning microscopic investigations, but confocal Raman spectroscopy enjoys additional advantages: use of unlabeled protein to eliminate fluorescent labeling, ability for characterization of protein secondary structure, and ability for spectral normalization to provide a nondestructive experimental approach to correct light attenuation effects caused by refractive index (RI) mismatching in semiopaque chromatographic particles.

V

approach is helpful typically for transparent adsorbents in which protein loading is adequately high to produce a sharp adsorption front. On the other hand, confocal Raman microscopy is a wellknown nondestructive analytical technique that is able to probe the chemical and structural properties of a sample on micrometer scales. This technique, similar to other confocal approaches including CLSM, defines a small detection volume by focusing an excitation laser beam in the sample and by imaging the emitted light collected through a matched aperture. More specifically, minimization of out-of-focus contribution to the signal is achieved by utilization of a small pinhole,26 use of a small optical fiber in the imaging plane of the objective,27 or careful selection of the entrance slit to the monochromator and a specific active area of the CCD detector.28 Confocal Raman microscopy has been used in many fields including chromatography. For example, it was applied to measure distribution of protein ligands in affinity chromatographic particles29 and the surface modification of sulfopropyl groups in a cation exchanger.30 Recently, it was applied to C18 stationary-phase particles to determine the distribution of organic modifiers within the particles,31 to elucidate the wetting process of the pores when the particles were immersed in aqueous mobile phases,32 and to understand the separation mechanism involved in the C18 ion-interaction chromatography.33 In this paper, we applied confocal Raman spectroscopy to measure spatial distribution of an unlabeled protein adsorbed in

arious approaches exist to determine the depth profile of protein adsorbed in chromatographic particles. Confocal laser scanning microscopy (CLSM) has been used most. This technology, originally introduced by Ljunglöf and Hjorth,1 allows direct observation of protein uptake profiles,2 measurement of multiple protein components in one particle,3−6 and real-time in situ examination of mass transport under flow conditions.7−9 Various intraparticle protein profiles10,11 were described using different protein adsorption models ranging from the classical shrinking-core model12,13 to the homogeneous-diffusion model.14,15 These adsorption behaviors depend on the type of protein studied,6,7,16,17 the chromatographic medium used,6,16,17 and the loading conditions including pH and ionic strength.4,6,9,16−22 In spite of enormous information made possible by CLSM, one concern is that this technique typically requires fluorescent labeling of protein which might induce some artifacts.11 Unlabeled protein in chromatographic particles has also been investigated by CLSM, based on the intrinsic fluorescence of tryptophan.9,23 Fluorescence from this amino acid is relatively weak and decreases quickly when the sample is continuously irradiated by UV light.23 Another technique for unlabeled protein studies exploits a refractive index-based optical microscopy,24,25 where protein-free and protein-loaded resins may have different refractive indices. Under conditions where the adsorption front is sharp, light passing through the particle becomes concentrated at the refractive interface between the advancing protein-saturated shell and the shrinking protein-free core. A bright ring, corresponding to the position of the adsorption front, can therefore be observed with a microscope and then be used to determine the transport kinetics of protein into individual particles. As discussed by the authors,25 this © 2012 American Chemical Society

Received: April 13, 2012 Accepted: July 13, 2012 Published: July 13, 2012 7367

dx.doi.org/10.1021/ac300994d | Anal. Chem. 2012, 84, 7367−7373

Analytical Chemistry

Article

the cation exchanger Fractogel EMD SO3 (M)34. This medium has been characterized to have high dynamic binding capacities (DBC) for monoclonal antibodies (mAbs).35 It was also noticed that the DBC initially increases with increasing conductivity and then decreases.35 Similar observations of maximum DBC or “critical capacity”20 were also reported when a mAb, EMD Millipore mAb04, was loaded into the Fractogel EMD SO3 (M) medium at a ∼10 mS/cm conductivity.36 In order to understand the origin of the “critical capacity”, we loaded mAb04 onto the cation exchanger at 2 and 10 mS/cm and then applied confocal Raman spectroscopy to measure the protein depth profile for each case. We detected a selfsharpening protein adsorption profile for 2 mS/cm and a flat profile at 10 mS/cm, suggesting that there is a change of the dominant mass transport mechanism for protein adsorption depending on the ionic strength of the protein loading solution.

storing buffer solution. A He−Ne laser (632.8 nm), which operated at 17 mW and passed through a 100 μm diameter single-mode optical fiber and then a 60× water dipping objective lens (MRD07620 CFI Water Apo, Nikon) with a numerical aperture of 1.0, was used to excite the particle. The spectrometer used to capture the spectra is equipped with a 600-lines/mm grating and a thermoelectric-cooled backilluminated CCD camera (1024 × 127 pixels, pixel size 26 × 26 μm, Andor Technology). A total of 180 spectra from a 90 μm line scan with a 0.5 μm piezoelectric step were generated to cover the entire 60−65 μm in diameter particle; each spectrum being collected with a 4-s integration time and coaddition of 2 scans. The confocal aperture of 100 μm in diameter was controlled by the multimode collection optical fiber. Wavelength calibration was achieved using low-pressure gas discharge lines of mercury and argon.





RESULTS Raman Spectra of Protein-Free Chromatographic Particles. Fractogel EMD SO3 (M) resin has a cross-linked polymethacrylate matrix with sulfoisobutyl surface modification that is bound to the matrix with linear polymeric chains (see Supporting Information Figure S1), the so-called “tentacles”.37 Particles are characterized by sizes ranging from 40 to 90 μm in diameter, a porosity of ∼61%, an average pore size of ∼80 nm in diameter, and ligand density of ∼380 μmol/g.38 Only particles with a diameter of 60−65 μm have been selected for spectroscopic measurements in this report. Figure 1a shows

EXPERIMENTAL SECTION Reagents and Materials. The cation exchanger, Fractogel EMD SO3 (M) (EMD Millipore, Billerica, MA), was supplied in 20% ethanol with 0.15 M NaCl. The resin was extensively washed with Milli-Q water and then equilibrated with a loading buffer of pH 5, 15 mM acetic acid and sodium acetate, with conductivity adjusted with a saturated NaCl solution to 2 mS/ cm (∼20 mM [Na+]) or 10 mS/cm (∼100 mM [Na+]) measured using an Oakton CON 11 standard conductivity meter (Eutech Instruments). Monoclonal antibody used for this work, EMD Millipore mAb04, was expressed in Chinese Hamster Ovary cell culture and purified by Protein A affinity chromatography. Protein purity was ∼99% measured by Protein A HPLC using a POROS A HPLC (Applied Biosystems) column on an Agilent 1260 HPLC platform. Prior to use, the protein solution was dialyzed against the respective loading buffer and then diluted to a concentration of 2 mg/mL. Protein concentration was determined by absorbance at 280 nm with an extinction coefficient of 1.532 mL mg−1cm−1. Human γ-globulin, from Cohn Fractions II and III with approximately 99% purity determined from electrophoresis, was acquired from Sigma. It has often been used in protein chromatography studies and was measured in this report as a reference protein to compare with the mAb04 protein adsorbed in the chromatographic particles for conformational studies. Protein Loading. About 40 μL of gravity settled Fractogel EMD SO3 (M) particles were suspended in 35 mL of mAb04 solution (2 mg/mL, pH 5, 2 mS/cm or 10 mS/cm). Sufficient amount of protein solution was used, so that there would be no significant concentration change during the loading process. Protein and the chromatographic particles were gently mixed at room temperature in a reaction tube by end-over-end rotation. Samples were removed from the reaction tube at defined times, centrifuged to remove protein supernatant, diluted by the loading buffer, and centrifuged again. Three cycles of centrifugation−dilution were carried out to remove free protein from the particles. The final protein-loaded Fractogel EMD SO3 (M) particles were stored in the respective loading buffer. Confocal Raman Microscopy Measurement. Confocal Raman spectroscopy was carried out on an alpha-300R confocal Raman microscope (WITec, Germany), which is an f/4 lensbased spectrometer with a focal length of 300 mm. Prior to each measurement, Fractogel EMD SO3 (M) particles were allowed to completely settle onto a hydrophobic Teflon membrane (0.22 μm, FGLP02500, EMD Millipore) from the

Figure 1. (a) Confocal Raman spectra of the Fractogel EMD SO3 (M) particle settled onto a Teflon substrate from an aqueous buffer solution. The line scan of 90 μm was carried out to cover the entire ∼60 μm diameter of the particle. Spectra were recorded in 0.5 μm intervals, but each spectrum stacked in 2.5 μm increment was an average of five spectra. (b) Three spectra extracted from panel (a) corresponding to the buffer solution, particle, and Teflon substrate. See Discussion in text for the labeled bands.

confocal Raman spectra from 90-μm line scan of the proteinfree Fractogel EMD SO3 (M) particle fixed in an aqueous buffer solution (pH = 5, 15 mM acetate, 2 mS/cm). These spectra were recorded covering the entire particle from 5 to 10 μm above the particle surface to ∼20 μm underneath the particle. Spectra were recorded at 0.5 μm depth intervals and displayed as average of 5 spectra at 2.5 μm increments for clarity. 7368

dx.doi.org/10.1021/ac300994d | Anal. Chem. 2012, 84, 7367−7373

Analytical Chemistry

Article

Raman measurement. The fixation was mainly due to gravity of the particle and van der Waals interaction between the particle and the substrate, with minor contribution from the optical trapping effects described by transferring the momentum of the incident radiation to the particle.41−45 From Figure 2a, a nearly linear drop of the Raman intensity could be observed with depth, the signal near the bottom interface being about one twentieth of that at the top boundary. This is mainly due to some mismatching in refractive indices, ∼1.5 for the particle vs ∼1.3 for water filled in pores. In the following discussion, we will use only the spectra corresponding to the top half of the particle to calculate protein distribution. The first derivative of the average depth profile was calculated and shown in Figure 2b. The derivative approach has been used to locate the exact interface position in confocal Raman studies of a layered film and provides an approximate way to determine the thickness of each layer.46 The positive peak in the first derivative curve in Figure 2b appears at the frontier surface of the particle, and the negative peak corresponds to the back surface. The distance between the two peaks is ∼63 μm corresponding to the particle size. We define the frontier interface as the zero depth with the (+) depth implying movement toward the particle. First derivative of the depth profile could also provide information about depth resolution.46 The full-width-at-half-maximum (fwhm) of the positive peak in Figure 2b, ∼5.2 μm, could be regarded as an approximate estimation of the depth resolution that will be used for discussion of protein distribution in the chromatographic particle. This depth resolution is quite acceptable for real samples, although it is larger than the calculated theoretical value of ∼1.2 μm with the 632.8 nm laser for excitation and the water dipping objective with a numerical aperture of one for imaging. Conformation of Protein Adsorbed in Chromatographic Particles. Raman spectroscopy is a well-known technique to analyze the secondary structure of proteins, mainly based on the amide I vibration, which is a property of the peptide group involving the CO stretching vibration with minor contributions from the out-of-phase CN stretching vibration, the CCN deformation, and the NH in-plane bending.47 The α-helix and β-sheet structures produce amide I bands at ∼1656 and ∼1667 cm−1, respectively, in their Raman spectra.48,49 Figure 3a shows Raman spectrum of the Fractogel EMD SO3 (M) particle with protein loaded at 2 mS/cm for 24 h. Raman spectrum of the protein-free particle has also been included in gray for comparison. Both spectra were obtained by averaging about 60 spectra recorded in 0.5 μm depth intervals covering the top half of the particles. The similarity of the two spectra suggests that the predominant spectral features are generated by the polymeric material of the particle. However, some new bands including those at 1615, 1555, and 1003 cm−1 and more Raman intensity at 1667 cm−1 can easily be found in the spectrum of the protein-loaded particle. The difference spectrum (protein-loaded particle minus protein-free particle), shown in Figure 3b, represents the protein adsorbed in the chromatographic particle. The spectrum of human γ-globulin is shown in Figure 3 for comparison with the adsorbed protein, which was recorded in solid state with a 30 s integration time. The agreement between the two protein spectra of Figure 3b,c is quite high. It is also worth mentioning that, in the difference spectrum of Figure 3b, the amide I frequency at 1667 cm−1 is characteristic for the β-sheet conformation of protein.48,49

Three types of spectra (Figure 1b) could be extracted from Figure 1a corresponding to the buffer solution, particle, and Teflon substrate. They were obtained for the respective species by averaging a number of spectra carefully selected to avoid interfacial regions. Of particular interest are the Raman bands assigned to the polymethacrylate39 matrix of the particle including those at 1718 (CO stretching), 1454 (CH3/CH2 deformation), 1252 (C−C degenerate stretching of CC4), 1122 (C−O stretching coupled with methyl rocking), and 966 cm−1 (main chain C−C stretching). Additionally, there are a few note-worthy points. First, the buffer solution generates weak Raman signals in the spectral region of interest (400−2000 cm−1). Only the broad band around 1632 cm−1, assigned to O− H bending of the water molecules, might produce interference because of overlapping with the amide I vibration of protein as discussed below. Second, spectral quality of the Teflon substrate underneath the particle suggests that, using a He− Ne laser for excitation and a water dipping objective for imaging, we are able to measure confocal Raman spectra of the semiopaque chromatographic particle of ∼60 μm in diameter. Third, some spectral features of the Teflon substrate, such as the band at 733 cm−1, appear in the spectrum of the particle or even in the spectrum of the buffer solution due to out-of-focus contribution, an effect well studied for layered polymeric films.40 Figure 2a shows confocal depth profiles of the Fractogel EMD SO3 (M) particle calculated on the basis of the intensity at 1454 cm−1 in the Raman spectra (Figure 1a). Four confocal Raman measurements were carried out within 2 h on the same particle, and their depth profiles are shown in gray with the average in black. Perfect overlap of four depth profiles suggests that the particle was well fixed on the Teflon substrate during

Figure 2. Depth profiles calculated on the basis of the Raman intensity at 1454 cm−1 for the Fractogel EMD SO3 (M) particle settled onto a Teflon substrate from an aqueous buffer solution. Four confocal Raman measurements on the same particle are shown in gray, and their average is in black (a); the first derivative of the average depth profile determines the particle size and depth resolution (b). 7369

dx.doi.org/10.1021/ac300994d | Anal. Chem. 2012, 84, 7367−7373

Analytical Chemistry

Article

Figure 3. (a) Raman spectra of the Fractogel EMD SO3 (M) particle with protein loaded at 2 mS/cm for 24 h (in black) and the proteinfree particle (in gray); (b) Raman difference spectrum of two particles (protein-loaded particle minus protein-free particle); (c) Raman spectrum of the human γ-globulin protein measured in solid state as an IgG reference. Band assignments are based on refs 48 and 49. Abbreviation: ν for stretching mode.

Further confirmation of the β-sheet secondary structure is available by examining the amide III vibration at ∼1240 cm−1. The amide III mode is the in-phase combination of the NH bending and the CN stretching vibration with small contributions from the CO in-plane bending and the CC stretching vibration.47 It shows characteristic Raman bands depending on the secondary structure, for example, at ∼1298 cm−1 for α-helix and ∼1240 cm−1 for β-sheet.48,49 Therefore, it could be concluded unambiguously from the Raman study that, after the mAb04 protein was loaded into the Fractogel EMD SO3 (M) particles at pH 5 and 2 mS/cm, its original conformation has well been maintained. The same conclusion could be drawn for the protein loaded at 10 mS/cm (see Supporting Information Figure S2). Depth Profile of Protein Adsorbed in Chromatographic Particles. When confocal Raman spectroscopy is applied to measure protein-loaded Fractogel EMD SO3 (M) particles, the confocal detection volume contains polymeric material of the particle, protein, and the buffer solution. All components will produce Raman signals in one spectrum. Assuming homogeneous pore structures throughout the particle, spectral normalization based on its characteristic bands makes it possible to correct light attenuation effects caused by refractive index (RI) mismatching in the semiopaque particle, as well as to alleviate concern caused by variation of the excitation laser power during the Raman measurement. Figure 4 shows normalized confocal Raman spectra of the protein-loaded Fractogel EMD SO3 (M) particles. Protein was loaded for 5 min at pH 5 and a 2 mS/cm (Figure 4a) or 10 mS/ cm (Figure 4b) conductivity. Spectral normalization was based on the Raman intensity at 1718 cm−1, which has a flat baseline on its right side and is close to the protein amide I band at 1667 cm−1. The zero depth, corresponding to the frontier interface of the particle, was defined using the approach described earlier. Each spectrum in Figure 4a is an average of 2 spectra recorded at 0.5 μm intervals; while each in Figure 4b is an average of 9 spectra. Therefore, the first spectrum marked zero-depth in Figure 4b measures the shell of 4.5 μm from the particle surface, and the last spectrum of 27 μm depth detects the region covering the center of the particle. Protein distribution in the particle could be examined on the basis of the amide I intensity at 1667 cm−1 in the confocal

Figure 4. Confocal Raman spectra of the Fractogel EMD SO3 (M) particle with protein loaded at 2 mS/cm (a) or 10 mS/cm (b) for 5 min. All spectra were normalized on the basis of the 1718 cm−1 band. Each spectrum stacked in 1 μm intervals (a) is an average of 2 spectra recorded in 0.5 μm intervals, and each spectrum stacked in 4.5 μm intervals (b) is an average of 9 spectra.

Raman spectra. It decreases quickly with depth (Figure 4a) for protein loaded at 2 mS/cm. Significant protein signal is observed in the zero-depth spectrum, but little could be detected in the spectrum of 5 μm depth. This suggests that a sharp protein adsorption front has been formed. Therefore, the shrinking-core model12,13 is suitable to describe the protein adsorption behavior at 2 mS/cm. This model assumes that only the nonadsorbed analyte molecules inside the pores are free to diffuse, while in equilibrium with the immobilized adsorbed molecules.50 In contrast to those in Figure 4a for protein loaded at 2 mS/cm, Raman spectra in Figure 4b for 10 mS/cm do not show significant intensity change at 1667 cm−1 with depth. At 10 mS/cm, protein was able to reach the center of the particle within 5 min suggesting that the homogeneous-diffusion model14,15 is applicable to describe the mass transport mechanism in this case. This model assumes that all analyte molecules in the particle, adsorbed and free in pores, are mathematically indistinguishable and are able to diffuse with the flux as a function of the overall concentration.51 Figure 5 shows depth profiles of protein loaded into the chromatographic particles at 2 mS/cm (Figure 5a) and 10 mS/ cm (Figure 5b) for different loading times. They are calculated on the basis of the Raman intensity at 1667 cm−1 in the difference spectra (protein-loaded particle minus protein-free particle) after all spectra were normalized on the basis of the 1718 cm−1 band. The zero-depth corresponding to the particle surface was determined again using the procedure described earlier. In Figure 5a for 2 mS/cm, a sharp protein adsorption 7370

dx.doi.org/10.1021/ac300994d | Anal. Chem. 2012, 84, 7367−7373

Analytical Chemistry

Article

particles is that the Raman signal from the silica support is weak and generally does not interfere with the vibrational spectra of the C18 chain or organic modifier, allowing high-quality Raman spectra to be recorded using a short integration time. When confocal Raman spectroscopy was applied to measure chromatographic particles made of polymeric materials, a long integration time was used to acquire high-quality spectra.29,30 In this report, we used a 4 s integration time in the Raman measurement. The spectrometer that was used, alpha-300R from WITec, has some unique features that allow for fast spectral acquisition, which may have sacrificed other features such as flexibility to install a NIR laser on the same system.65 Lenses used in the lens-based spectrometer usually result in high spectral throughput, as they have a transparency of approximately 98% compared with the 75−80% reflectivity of the mirror used in the mirror-based Czerny−Turner spectrometer.65 Three mirrors typically used in a C-T instrument may have been optimized for operation from the UV up to the NIR impacting the reflectivity in the visible range. In addition, the back-illuminated CCD camera used in the alpha-300R instrument has a quantum efficiency of ≥90%,65 higher than what could be achieved by a front-illuminated CCD detector. The presence of an aqueous buffer solution surrounding the Fractogel EMD SO3 (M) particles allows for imitation of the real application conditions in the confocal Raman measurement. This is made possible using a water dipping objective, which was designed to work without a cover glass and could be dipped directly in the buffer solution. The objective also helps deep depth profiling, because the polymeric material of the particle has a refractive index (∼1.5) closer to that of the water (∼1.3) than the air (∼1.0) filled in pores. Some mismatching in refractive indices has contributed to the Raman intensity decrease with depth (Figure 2a). Previous confocal measurements66 indicate that relatively small refractive index mismatches (∼0.2) can cause significant signal loss. Due to the decreased collection efficiency of Raman signals (Figure 2a), a sufficient amount of protein with loading times of ≥20 min is required to produce quality spectra for the back half of the particle to calculate the overall protein depth profile (see Supporting Information Figure S3). Comparison of CLSM and Confocal Raman Microscopy in Protein Chromatography. Fluorescence spectroscopy is an extremely sensitive analytical technique allowing CLSM to detect trace amounts of protein in the chromatographic particles. In fact, it is so sensitive that dilution of the fluorescent labeled protein in the order of 1:100 to 1:10 by the unlabeled protein is needed, in order to maintain a linear relationship between the fluorescence intensity and the protein concentration in the particle.16−22 Another benefit of CLSM is that various dyes could be used to label different proteins allowing multicomponent adsorption to be examined in one particle.3−6 One concern for CLSM is that it typically requires fluorescent labeling of protein, which might induce artifacts.11 First, dye attached on the protein may change protein properties in terms of surface charge, hydrophobicity, and molecular weight. Second, the outcome of fluorescent labeling strongly depends on experimental conditions including pH, type and concentration of the protein and dye, reaction time, and temperature. As the bioconjugation reaction is often nonsite specific, fluorescent labeling may produce heteroge-

Figure 5. Time series of confocal depth profiles of protein loaded into the Fractogel EMD SO3 (M) particles at 2 mS/cm (a) and 10 mS/cm (b). Calculation was based on the amide I intensity at 1667 cm−1 in the Raman difference spectra (protein-loaded particle minus proteinfree particle, after spectral normalization based on the 1718 cm−1 band). Profiles times are 5, 10, 20, 60, 300, and 1440 min.

front could be observed moving with time into the interior of the particle, agreeing with the CLSM studies on the Fractogel EMD SO3 (M) particles with a mAb loaded at 4 mS/cm.52 We notice that there is no perfect overlap in the intensity profiles on the top side of the particles. One explanation is that, for short loading times, the protein adsorption layer has a thickness less than ∼5 μm, the depth resolution of the confocal Raman measurement for the Fractogel EMD SO3 (M) particle (Figure 2b). Under this condition, confocal Raman microscopy cannot accurately draw the picture that a uniform saturated protein concentration has reached throughout the