Mass photometry: A powerful tool for carbohydrates-proteins conjugation monitoring and glycoconjugates molecular mass determination

Abstract

Glycoconjugate vaccines are important additions to the existing means for prevention of diseases caused by bacterial and viral pathogens. Conjugating carbohydrates to proteins is a crucial step in the development of these vaccines. Traditional mass spectrometry techniques, such as MALDI-TOF and SELDI-TOF, have difficulties in detecting glycoconjugates with high molecular masses. Mass photometry (MP) is a single-molecule technique that has been recently developed, which allows mass measurements of individual molecules and generates mass distributions based on hundreds to thousands of these measurements. In this study, we evaluated the performance of MP in monitoring carbohydrate-protein conjugation reactions and characterization of conjugates. Three different glycoconjugates were prepared from carrier protein BSA, and one from a large protein complex, a virus capsid with 3.74 MDa molecular mass. The masses measured by MP were consistent with those obtained by SELDI-TOF-MS and SEC-MALS. The conjugation of BSA dimer to carbohydrate antigen was also successfully characterized. This study shows that the MP technique is a promising alternative to methods developed earlier for monitoring glycoconjugation reactions and characterization of glycoconjugates. It measures intact molecules in solution and it is highly accurate over a wide mass range. MP requires only a very small amount of sample and has no specific buffer constraints. Other MP advantages include minimal cost of consumables and rapid data collection and analysis. Its advantages over other methods make it a valuable tool for researchers in the glycoconjugation field.

Introduction

Bacterial polysaccharide antigens play a crucial role in immune recognition, but their effectiveness as immunogens is limited. They are classified as T-cell independent antigens, and the antibodies produced following immunization with carbohydrate antigens are typically insufficient to provide protection. To address this drawback, vaccines for bacterial pathogens require conjugation of carbohydrates to suitable carriers, such as proteins. This transformation converts T-cell independent antigens into T-cell dependent antigens, thereby enhancing their immunogenicity. As a result, multiple injections of glycoconjugate vaccines increase antibody titers significantly, and beyond those observed after priming with carbohydrates alone.

In general, there are two fundamentally different types of conjugation models used for making glycoconjugates [1]. When the carbohydrate antigen carriers have two or more active sites, the glycoconjugate product is cross-linked, and the preparation of such materials is difficult to characterize and reproduce. Conversely, when there is only one active group on the antigen molecule, a single-point attachment conjugate is formed, which is easier to reproduce and characterize. In addition, carbohydrates will be displayed on the carrier in a sun-burst manner, which is similar to their display on the surface of bacteria.

One of the most important physical characteristics of glycoconjugates is their molecular mass. To determine the mass, or the carbohydrate to protein ratio of glycoconjugates, several methods have been utilized, often in combination [2]. They include SDS/PAGE, various colorimetric sugar or protein assays, size-exclusion chromatography (SEC), as well as mass spectrometry techniques, such as SELDI-TOF (surface enhanced laser desorption/ionization-time of flight) and MALDI-TOF-MS (matrix-assisted laser desorption/ionization-time of flight) [3–6]. Among these methods, SELDI-TOF and MALDI-TOF MS are considered the most accurate for the determination of the molecular mass of glycoproteins.

A limitation of MALDI-TOF and SELDI-TOF is their inability to detect glycoconjugates with molecular masses above 150 kDa. Carrier protein molecules used in glycoconjugation are themselves sometimes too large, e.g. the Keyhole Limpet Hemocyanin (KLH), or they form oligomers that are larger than 150 kDa. If oligomers are present, they will also conjugate with the antigen but these oligomer glycoconjugates might not have the same carbohydrate incorporation as their monomer counterparts. Although other mass spectrometry method, for example the native mass spectrometry, has the capability to characterize large protein complex with high resolution and accuracy, such as viruses and virus-like particles [7–9]. It is important to note that this method requires sophisticated instrumentation and dedicated well-trained personnel to carry out the experiments. Therefore, developing methods for molecular mass determination based on technologies that can easily and fast detect masses beyond the 150 kDa limitation is highly desirable.

Real-time monitoring of the progress of the conjugation reaction [10] is also a crucial step in the development of glycoconjugate vaccines. Compared to MALDI-TOF-MS, monitoring glycoconjugation reactions using SELDI-TOF-MS is more convenient because it bypasses the often-laborious purification of analytes [10]. The sample preparation can be performed directly on the SELDI-TOF chip, and the entire monitoring process can be completed within 10–15 min [10–13]. Unfortunately, the SELDI-TOF instruments are no longer commercially available. In search for a suitable alternative, we have found that MALDI-TOF in combination with micropipette tip for desalting, such as C₄ ZipTip (MilliporeSigma, MA), can also be used to monitor glycoconjugation reactions [13], but protocols involving these devices is more time-consuming due to the additional sample preparation steps.

Recently, Kukura’s group at Oxford University in UK developed mass photometry (MP) [14, 15], a single-molecule technique for the determination of the molecular mass of macromolecules. This technique is based on the interference scattering microscopy and it can detect individual molecules when they are landing on the microscope coverslip. The signal measured by MP is proportional to the intensity of light scattered by the detected molecule, and the intensity depends on both the refractive index and the mass of the molecule. Since biomacromolecules, e.g. proteins, nucleic acids, and carbohydrates, are polymers composed of similar constituents, their refractive indexes are similar, and a single refractive index value can be used for each of them [15–18]. Thus, their MP signal is proportional to the mass of the detected molecule, making an accurate determination of molecular masses possible. In a typical MP measurement, molecular masses of hundreds to thousands of molecules are measured, and therefore the mass distribution of the sample can be generated. Due to the noise floor of the MP image background, the individually measured masses tend to follow a normal distribution around their theoretical masses. Consequently, each species present in the sample manifests as a peak within the mass distribution. Ultimately, the mass of each species is determined through a Gaussian fit. MP has been successfully used to measure the molecular masses of proteins and their complexes [19], DNAs [20], as well as more complicated molecules such as membrane proteins [21], and viruses [22].

MP has several advantages making it an attractive method for the determination of the masses of molecules. First, it measures intact molecules in solution and MP samples do not require denaturation, immobilization, labeling, or other chemical modifications. Second, MP enables measurements of molecular masses over a wide range with high accuracy. The latest version of the commercially available instrument (TwoMP, Refeyn Ltd., UK) can measure protein molecules from 30 kDa to 5 MDa with a ~±2% accuracy [15]. Third, MP experiments require a low quantity of material. Typically, 10 μL of sample at about 20 nM concentration is sufficient to obtain a high-quality MP data. Additionally, there are no specific buffer requirements, and any commonly used aqueous buffer such as PBS may be used. However, high glycerol concentrations and low ionic strength buffers are not recommended [23]. Nonetheless, since low sample concentrations are typically used for MP measurements, a high sample dilution factor is often necessary. Therefore, there’s usually no need to exchange the buffer even if the sample stock is in a buffer not optimal for MP. Finally, performing MP measurements is fast; typically, the MP data collection and analysis of one sample takes less than 5 min. These attributes make MP-based technology a viable alternative for monitoring protein conjugations and for characterizing glycoconjugates.

To test the performance of MP in carbohydrate-protein conjugation, we prepared BSA conjugates from three different carbohydrate antigens: a linker-equipped lactose, a Vibrio cholerae O1 Ogawa hexasaccharide, and Vibrio cholerae O1 Inaba O-specific polysaccharide (OSP). We also used a virus capsid as a large protein complex carrier conjugated with linker-equipped lactose. The conjugation reactions were monitored by MP, and to validate the masses obtained by this method, the results were compared with data obtained by SELDI-TOF and SEC-MALS (size exclusion chromatography with multi-angle static light scattering) analysis. Overall, this report provides a critical evaluation and validation of MP as an alternative method for monitoring carbohydrate-protein conjugation and for the characterization of the conjugation products.

Materials & Methods

Conjugation of carbohydrate antigens to BSA (Scheme 1A)

Scheme 1.

Conjugation of carbohydrate antigens to BSA and AAV

The lactose-BSA conjugates prepared previously [24] were used as the molecular mass-calibration standards. For the preparation of the other two BSA conjugates, the carbohydrate antigen squarate (Compounds 1 or 2) [11, 25, 26] and BSA (A-0281, Sigma) were dissolved in a 0.5 M pH 9.5 borate buffer to form a solution of 5 mM with respect to the antigen-squarate. The solution was stirred at room temperature for varying amounts of time. In some cases, additional amount of antigen-squarate was added. The progress of the reaction was monitored by SELDI-TOF (Protein Chip SELDI system with NP-20 chip arrays, Bio-Rad, Hercules, CA) and MP. The reaction mixture was then ultrafiltered through a Millipore Amicon Ultra (30 kDa cutoff) tube against 10 mM ammonium carbonate solution (at 4°C, 7,500 RCF), and this process was repeated eight times. The retentate, after lyophilization, yielded conjugates 3 and 4 as white fluffy solid.

1. Vibrio cholerae O1 Ogawa hexasaccharide-BSA conjugate: BSA (5.3 mg, 0.08 μmol, 1 eq.) was mixed with 1 (1.0 mg, 0.56 μmol, 7 eq.). After 20 hours, the mass was measured by both SELDI-TOF and MP. After the mass measurements, another 7 eq of 1 was added. The reaction progress was checked again using both techniques at 92 hours of the reaction time, and another 14 eq of 1 was added. The reaction was continued for another 96 hours, and was checked for the last time at 188 hours of the overall reaction time. After work-up and freeze-drying, 7.9 mg of conjugate 3 was obtained (recovery ~99%, conjugation efficiency is 68% based on 99.9 kDa average molecular weight of 3).

2. OSP-BSA conjugate: BSA (2.2 mg, 0.033 μmol, 1 eq.) was mixed with 2 (2.0 mg, 0.33 μmol, 10 eq.) and the reaction was monitored at 24, 48, and 96 hours of reaction time using SELDI-TOF and MP. After work-up and freeze-drying, 2.9 mg of conjugate 4 was obtained (recovery ~92%, conjugation efficiency is 47% based on 94.8 kDa average molecular weight of 4).

Conjugation of carbohydrate antigens to AAV (Scheme 1B)

The highly purified adeno associated virus capsid particles were obtained from AAVnergene (Rockville, MD). The capsid solution (100 μL at 4.17 mg/mL) in PBS buffer, pH 7.4, was placed in a Millipore Amicon Ultra-0.5 (30k Da cut-off) tube and ultrafiltered 5 times at 4°C against 0.5 M pH 9.5 borate reaction buffer at 7,500 RCF for 7 min/run [13]. The retentate was transferred into a 0.5 mL V-shaped reaction vial, and the volume of the solution was adjusted to 100 μL with the reaction buffer. The linker-equipped lactose squarate (5, 212 μg, 0.34 μmol, ~3000 equiv. with respect to AAV) was added into the reaction vial and the mixture was stirred at room temperature. After 96 hour, another 900 μg of 5 was added and the reaction was continued for another 96 hours. After being checked by MP, the mixture was ultrafiltered through a Millipore Amicon 0.5 (30 kDa cut-off) tube against pH 7.2 PBS buffer (6 times at 4°C, 7,500 RCF, 7min/run) to remove the unconjugated lactose squarate. A clear solution of the capsid-lactose conjugate (6) was obtained.

Mass photometry measurements

Protein conjugation samples for MP analysis were diluted with PBS (46–013-CM, Corning, NY) to a concentration of approximately 20 nM. The AAV conjugation sample was diluted to 10¹¹ particles per milliliter immediately prior to MP measurements. The current study used OneMP mass photometer and precleaned coverslips (Refeyn, UK). Following the standard protocol [23], 10 μL of filtered PBS was loaded into the sample well for MP focusing, and then 10μL of the diluted sample was added and mixed with the buffer in the same well. Data were collected for one minute immediately after the mixing using the AcquireMP software (Refeyn, UK). BSA conjugation samples were measured using the standard MP detection area, while the virus conjugation samples were measured under the expanded detection area. The MP signals in both detection modes were calibrated, respectively, using BSA (05470–1G, MilliporeSigma, MA) and the Unstained Protein Standard (LC0725, ThermoFisher, MA). MP data were processed in the DiscoverMP software (Refeyn, UK).

SEC-MALS measurements

The multi-angle light scattering coupled with size exclusion chromatography (SEC-MALS) was used to characterize the empty virus capsids and the carbohydrate-virus conjugates. The SEC-MALS system used in this study included the 1100 series HPLC system (Agilent, Santa Clara, CA), the HELEOS-II multi-angle light scattering detector, the TrEX refractive index detector, and the WTC-050N5 size exclusion column (Wyatt Technology, Santa Barbara, CA). Data obtained for the virus-lactose conjugate reaction solution ware analyzed using the ‘protein conjugate’ method in ASTRA 7 software (Wyatt Technology, Santa Barbara, CA) [27].

For the conjugate analysis of the SEC-MALS data, the averaged refractive index increments of the protein (0.185 mL/g [28]) and polysaccharide (0.15 mL/g [29, 30]) were used for the virus capsid and the linker-equipped lactose, respectively. The 280 nm UV extinction coefficient of the unconjugated virus capsid was determined by performing the unconjugated capsid sample measurement on the SEC-MALS and analyzing the data using the known protein refractive index increment (0.185 mL/g [28]). The 280 nm UV extinction coefficient of the linker-equipped lactose was determined by a spectrophotometer (HP 8453, OLIS Inc., Athens, GA).

RESULTS

MP analysis of BSA and BSA conjugates

Studies published in the recent years have shown that MP can be used to determine the molecular masses of biomolecules and their complexes [19–22]. In the current study we assessed the suitability of MP for monitoring of the conjugation reactions and the characterization of carbohydrate-protein conjugates. As illustrated in Figure 1A, the lactose-BSA conjugate molecules can be clearly detected in the MP ratiometric view. The MP signals (interferometric contrast) of a few thousand individual molecules were calculated and plotted as a histogram, represented by the blue bars in Figure 1B. For comparison, the MP signal distribution obtained for the unconjugated BSA is presented by yellow bars in the same figure. Since the MP signal is proportional to the molecular mass of each molecule, larger absolute contrast values represent molecules with higher molecular masses. The unconjugated BSA distribution shows two baseline-separated peaks at the contrast values around −0.002 and −0.004, respectively, indicating the presence of the BSA monomer and dimer. Higher-order BSA oligomers were also detected by MP and their peaks are positioned at larger absolute contrast values in the histogram. In comparison to the unconjugated BSA, the peaks corresponding to the conjugated BSA monomer and dimer are shifted to the right, indicating an increase in molecular mass due to the addition of poly/oligosaccharide to the protein. To precisely determine the peak positions, histograms were fitted with Gaussian distributions, as shown by the solid lines in Figure 1B. The contrast value for the unconjugated BSA monomer obtained from its Gaussian peak position is −0.00222, with the peak width of 0.00028, while for the conjugated BSA monomer the peak position was at −0.00288, with a slightly broader peak width of 0.00032, indicating a more heterogeneous sample. The same conjugate sample was also characterized by SELDI-TOF (Fig. 1C). As shown in this figure, the molecular mass of the conjugate is 94.1 kDa with a significantly broader peak width than that obtained for the unconjugated BSA, which confirmed the heterogeneity of the sample observed by MP.

Figure 1.

Mass analysis of lactose-BSA conjugates. (A) MP ratiometric image of lactose-BSA conjugate. The blue circles indicate molecules detected by MP. (B) Histograms of the MP signal distribution of BSA (yellow) and lactose-BSA conjugates (blue) with a lactose/BSA ratio of 47:1. The solid lines represent the Gaussian fits. (C) SELDI-TOF-MS of unconjugated BSA (upper) and the lactose-BSA conjugate shown in panel B (lower).

MP contrast-to-mass calibration of carbohydrate-protein conjugates

MP signals (ratiometric contrast) originate from the light scattered by individual molecules in a sample. For molecules of the same type, the scattered intensity is proportional to the molecular mass, and the contrast-to-mass calibration (equation 1) enables the conversion of contrast values to mass values for the measured molecules. To obtain the calibration values (slope and intercept) for carbohydrate antigens, MP signals of a series of standard lactose-BSA conjugates with known lactose/BSA ratios (8, 19, 27 and 47) were plotted against their molecular masses determined by SELDI-TOF. This molecular mass value represents the mass of the whole particle (mass of BSA and the carbohydrate antigen) (Fig. 2). The data points show a strong linear correlation and can be effectively modeled using linear regression analysis. The best-fit slope and intercept values shown in the figure were used as the contrast-to-mass calibration for the BSA as a carrier:

$$Contrast=slope\times Mass+{intercept}_{BSA}$$ (1)

Figure 2.

Contrast-to-mass calibration of carbohydrate-protein conjugates. The MP contrast values of BSA and lactose-BSA conjugates with different numbers of linker-equipped lactose are plotted against their masses determined by SELDI-TOF. The error bars show the standard deviation of 5 replicates and dashed line represents the best linear fit (y = −3.03e-5x – 1.977e-4, R²=0.998), which is used as the contrast-to-mass calibration.

It should be noted that equation (1) describes the measured MP contrast values for different conjugated forms of BSA, and all these molecules differ only in the mass of the conjugated carbohydrates. The linear relationship between MP contrast and the mass of molecules reflects the MP signal contribution from the carbohydrate antigen alone. Therefore, the slope reflecting the carbohydrate-only contribution will not change when carbohydrates are conjugated with different protein molecules. However, to apply equation (1) to analysis of other carrier proteins, the mass of the BSA has to be substituted with the appropriate mass of the protein under analysis. For example, to apply this calibration to a different protein carrier “X”, the MP signal of the unconjugated protein X should be measured first. The new intercept value can be determined by substituting the measured contrast value ${Contrast}_X$ and the known proteins molar mass ${Mass}_X$ into the following equation:

$${intercept}_X={Contrast}_X-slope\times {Mass}_X$$

Consequently, the new calibration for carrier X will be given by:

$$Contrast=slope\times Mass+{intercept}_X$$

Carbohydrate-BSA conjugation monitored by MP

To test the MP performance in monitoring the progress of carbohydrate-protein conjugation reactions, BSA was conjugated to two carbohydrates: V.cholerae. O1 Ogawa hexasaccharide (molecular mass ~1.75 kDa) [26], and V.cholerae.O1 Inaba OSP-core (molecular mass ~6 kDa) [11, 25]. At different time points during the reaction, 1 μL and 0.5 μL samples were withdrawn from each reaction mixture for mass spectrometry and MP measurements, respectively. Figure 3A shows the change of measured mass against time illustrating the progress of the reaction. To test the accuracy of the MP mass measurements, parallel SELDI-TOF experiments were performed. As shown in Figure 3B, the average masses measured by MP were in very good agreement with those measured by SELDI-TOF, demonstrating the MP capability to monitor carbohydrate-protein reactions. To compare the precision of the two techniques, all samples were measured 5 times on both SELDI-TOF and MP, and the standard deviations of replicate measurements were shown in Figure 3B as the horizontal and vertical error bars. The average relative standard deviation for SELDI-TOF and MP was 0.2% and 1.1%, respectively. Thus, the MALDI-TOF MS measurements of BSA conjugates are more precise. Next we tested MP performance in monitoring large conjugates, beyond the range of MS.

Figure 3.

Carbohydrate-BSA conjugation reaction monitored by MP. (A) The masses of V. cholerae. O1 Ogawa hexasaccharide-BSA conjugates (orange triangles) and V. cholerae.O1 Inaba OSP conjugates (blue circles) obtained from MP measurements are plotted against the conjugation reaction time. Arrows indicate time points when additional antigen was added. (B) Comparison of the masses of V. cholerae. O1 Ogawa hexasaccharide-BSA conjugates (orange triangles) and V. cholerae.O1 Inaba OSP-BSA conjugates (blue circles) measured by MP and SELDI-TOF, respectively. The diagonal dashed line indicates the mass equivalence between SELDI-TOF and MP measurements. Error bars in (A) and (B) show the standard deviation of 5 replicates.

Large carbohydrate-protein conjugates monitored by MP

Although mass spectrometers such as SELDI-TOF and MALDI-TOF are the most commonly used and reliable tools for conjugation monitoring and conjugate characterization, glycoproteins with molecular masses above ~150 kDa are outside the detection limit of most of MALDI-TOF and SELDI-TOF instruments. MP has a wider measurement range, from 30 kDa to 5 MDa, which enables characterization of large carbohydrate-protein conjugates.

Since the BSA MP mass distributions clearly show both the monomer and dimer peaks (Fig. 1B), the conjugation of BSA dimers to a linker-equipped lactose can also be simultaneously characterized. The average number of the lactose units was calculated by subtracting the theoretical mass of BSA dimer from the total mass measured by MP, and then dividing the total net lactose mass by the mass of one lactose unit. The results showed for the first time that on the molar basis, BSA dimer conjugates have the same degree of lactose incorporation throughout the reaction as the BSA monomer conjugates. (Fig. 4).

Figure 4.

Average numbers of linker-equipped lactose on BSA monomer (blue) and dimer (red). Error bars represent the standard deviation of 5 replicates.

Large protein complexes, such as virus-like particles, have also been used as carriers for vaccine development [31]. To test the applicability of the MP for monitoring conjugation of carbohydrates with large carriers, the empty adeno associate virus (AAV) capsids were conjugated with the linker-equipped lactose and the conjugation was monitored by MP. AAV capsid consists of 60 protein subunits of three different viral proteins, with a total molecular mass of approximately 3.74 MDa [32]. In a recent study we successfully characterized AAVs and their genetic cargo using MP [22]. As shown in Figure 5A, in comparison to the unconjugated capsid peak, the peak corresponding to the capsid that reacted for 6 days significantly shifted to the right, demonstrating that conjugation had taken place. The total mass of the capsid-lactose conjugates was determined using the molecular mass of the empty capsid of 3.74 MDa and applying the calibration determined using the lactose-BSA conjugates. The molecular mass of the conjugated capsid obtained by MP was 4.14 ± 0.06 MDa (five replicates), corresponding to a total of 0.40 MDa lactose (~676 lactose units) attached to the virus capsid. Figure 5B shows the progress of the lactose-virus capsid conjugation reaction monitored by MP, with the masses obtained from MP plotted against the reaction time. This result indicates that adding an additional 900 μg carbohydrate antigen into the reaction vial on the fourth day significantly boosted the reaction.

Figure 5.

Mass analysis of lactose conjugated virus capsid. (A) MP mass distribution of unconjugated virus capsid (yellow) and capsid conjugated with linker-equipped lactose after a 6-day reaction. Solid lines show the best-fit Gaussian distributions. (B) Lactose-capsid conjugation reaction monitored by MP. Error bars represent the standard deviation of 5 replicates, and the arrow shows the time point when additional antigen was added. (C) SEC-MALS analysis of capsid conjugated with linker-equipped lactose for 6 days. Plot shows the masses (left y-axis) of the whole conjugate (blue line), the protein component (red line) and the carbohydrate component (yellow line). The black line represents the light scattering chromatography profile (right y-axis).

Since the virus capsids are too large to be characterized by mass spectrometry, SEC-MALS was used to verify the AAV molecular mass obtained by MP. When the UV extinction coefficients and refractive index increments of the protein and the carbohydrate are known, SEC-MALS can be used to determine not only the average total molecular masses of the carbohydrate-protein conjugates but also the mass fraction of each component of the conjugate. First, a pure empty virus capsid was analyzed by SEC-MALS and the 280 nm extinction coefficient of the capsid protein was determined to be 2.13 mL/(mg cm) based on the known refractive increment values (see the method section). The UV absorbance of the linker-equipped lactose 5 (Scheme 1) at 280nm was measured using the spectrophotometer, and the 280 nm extinction coefficient was determined to be 34.2 mL/(mg cm). Finally, 20 μL of the capsid-lactose conjugate reaction solution at 4.17 mg/mL was injected for SEC-MALS analysis. The average total molecular mass of the conjugated AAV was determined to be 4.23 MDa, of which the lactose was determined to be 0.48 MDa (Fig. 5C). The total mass difference between the MP and SEC-MALS was within the usual 3% error of SEC- MALS [33, 34].

Discussion

In this study, we conducted the first comprehensive evaluation of MP-based technique as an alternative to the characterization of carbohydrate-protein conjugates. Since MP is a novel technique, it is important to point out its advantages and limitations regarding the characterization of carbohydrate-protein conjugates. Additionally, we aimed to demonstrate how this new methodology fills gaps in the currently used methods for the characterization of glycoconjugates.

In comparison to other analytical methods, MP measurements require a much smaller amount of sample. Our data showed that sub-picomole amount of material is sufficient to generate high-quality MP data. For large particles, such as viruses, the required amounts are even smaller. In contrast, MALDI-TOF or SELDI-TOF may require tens of picomoles of material for glycoprotein analysis, and sub-nanomole amounts of sample are required for SEC-MALS measurements. Thus, MP offers a significant advantage in terms of the required sample amounts, which is particularly important when sample availability is limited, as it is often the case with complex glycoconjugates.

Performing the MP measurement is fast and undemanding. A typical data collection and analysis for a single measurement can be completed within 5 minutes, which is significantly faster than SELDI-TOF or MALDI-TOF measurements that take 10–20 minutes, or SEC-MALS that requires at least 30 minutes to complete the chromatography involved. In our facility, even new users can independently operate the MP instrument and process data after approximately 30 minutes of training. Moreover, MP can tolerate different buffers, unless the buffer has an extremely low ionic strength or contains a high concentration of glycerol [23]. Even if the buffer is not suitable for the MP measurements, due to the low working concentration, a high sample dilution factor is usually required, and therefore buffer exchange is not necessary. These advantages enable real-time monitoring of the conjugation reaction by taking a small volume of the reaction mixture (e.g. 0.5 μL or less), diluting it into an appropriate buffer, and performing the MP measurements with the diluted sample. The result can be used as a guide to either terminate the reaction, if the desired loading had been reached, or to prolong the conjugation time or add more reagent to achieve a higher loading.

MP measurements are particularly suitable for the characterization of conjugates with relatively high molecular masses that are not amenable to analysis by mass spectrometry. Using MP-based method, it is possible to monitor the glycoconjugation of protein oligomers or virus particles, and potentially to detect the crosslinking during the conjugation reaction, including the characterization of large molar mass products of these reactions. On the other hand, molecules that are smaller than 30 kDa, cannot be observed by MP. Thus, for the analysis of low molecular weight conjugates, a mass spectrometer is still the method of choice. In comparison to MP, SEC-MALS has potentially even wider mass measurement range, from 200 Daltons to 1 billion Daltons, but in practice the accessible mass measurement range is much smaller. For example, UV and refractive index signals required for MALS analysis are linearly related to the molecular mass of the sample, while the light scattering signal shows a quadratic dependence. As a result, the light scattering signal intensity from very small molecules is often too low, and conversely, the UV/RI signal intensity from very large molecules may be too low to obtain accurate SEC-MALS data. Additionally, the UV signal from large molecules may not be accurate due to the light scattering [35]. In practice, the optimum range of SEC-MALS measurements is from 10 kDa to 10 MDa, a range slightly larger than that of MP.

Although MP has several advantages for monitoring conjugation and characterizing carbohydrate-protein conjugates, it has some limitations compared to other methods. Specifically, conjugate characterization is usually complicated by their heterogeneity, with the number of carbohydrate antigens attached to the carrier normally distributed around the mean loading value. The resolution of SELDI-TOF mass spectrometer allows to distinguish the different numbers of carbohydrate units attached to the protein molecule even for a relatively large antigens (Fig. 6B). In contrast, the MP resolution is about 25 kDa for small molecules or approximately 10% of the total mass for larger molecules, with the measurement accuracy of approximately ±2% [15]. Our results also showed that mass spectrometer has better reproducibility than MP, with a relative standard deviation of SELDI-TOF data about 0.2%, while the deviation of MP results is about 5 times larger (Fig. 3B). Therefore, MP cannot resolve mass differences of conjugates with different numbers of carbohydrate-antigen units and can only determine the average mass obtained from the Gaussian fit of the MP distribution histogram (Fig.6A). However, the heterogeneity of conjugation can be qualitatively estimated based on the width of the MP distribution peak. Although in most cases the average mass of the entire conjugate provided by MP is the only important information for conjugate characterization and conjugation monitoring, in some cases the size of the antigen is unknown and the resolution of mass spectrometer may help with the antigen characterization. In comparison to MP, the resolution of SEC-MALS is even lower and depends on the size exclusion column. Typically, SEC-MALS is only able to resolve carrier protein monomers and dimers.

Figure 6.

Mass analysis of V. cholerae. O1 Ogawa hexasaccharide-BSA conjugate. (A) Mass distribution obtained by MP. The solid line shows the best-fit Gaussian distribution. The figure shows the best-fit molecular masses (kDa) and the peak widths. (B) SELDI-TOF analysis of the same sample. The figure shows the masses (Da) of BSA conjugated with different numbers of carbohydrate antigens.

It is important to note that MP is a light scattering-based technology and, as such, indiscriminately detects objects that scatter the incident light. The lack of signal source discrimination makes it impossible to perform MP analysis on samples that contain large amounts of impurities, especially when the molecular mass of the co-solute is similar to that of the target species. Therefore, when MP is used as a method to characterize conjugates, the carrier protein for the conjugation reactions should be sufficiently pure, showing narrow peak(s) in the MP mass distribution, as it is difficult to quantify the mass shift if the peaks are broad or not resolved. However, this lack of MP signal discrimination allows us to assess the sample quality determinants, including sample purity, aggregation or degradation. Similarly, the samples for mass spectrometry also must be relatively pure. Large amount of impurities can result in a poor resolution and accuracy of the mass spectra, as well as increased background noise and reduced signal-to-noise ratio. In contrast, the advantage of SEC-MALS is that the impurities are separated from the target species on the size exclusion column, unless the impurities and target species have similar hydrodynamic radii. Therefore, the signal collected by SEC-MALS is from the molecules that are already purified.

In conclusion, the data obtained in this study demonstrate that MP can accurately and precisely characterize carbohydrate–protein conjugates within a wide range of molecular masses, including characterization of large conjugates that are not amenable to the analysis by MALDI- or SELDI-TOF. MP measurements are fast and use only a small amount of material, often requiring no sample preparation beyond simple dilution. These qualities make MP a highly viable alternative for the biophysical characterization of glycoproteins in both vaccine industry and academic settings.

Availability of Data and materials:

The data set generated during the current study are available from the corresponding author upon reasonable request.