Skip to main content

Practical Considerations for the Preparation and MS Analysis of PEGylated Bioconjugates

Authors: Robert J. Schuster1, Jen Ahner2, Kathy Tober2, and Mark E. Hail1

Novatia1, Quanta BioDesign, Ltd.2

Abstract: FDA Regulations for investigational new drugs (INDs), including antibody-drug conjugates (ADCs), are requiring more rigorous characterization of the final product. The ability to meet these standards depends on both the raw materials and the analysis technology and methods. Many biologics incorporate polyethylene glycol (PEG) constructs, either as linkers or modifiers due to their ability to modulate solubility, stability, and pharmacokinetic properties. However, the traditional polymeric PEG reagents have inherent heterogeneity, whereas the Quanta BioDesign dPEG® constructs are manufactured as single molecules. PEGylated proteins, including antibodies, are best analyzed by ESI/LC/MS, but the ability to successfully characterize these molecules depends on the heterogeneity of the sample, mass resolution of the MS instrument, and the capabilities of the charge deconvolution software. Novatia antibody analysis services and ProMass HR technology are designed to rapidly acquire and accurately deconvolute complex ESI/LC/MS data into easy to interpret zero-charge mass spectra. Here we present practical considerations for the preparation and MS analysis of PEGylated bioconjugates, including the single-molecule dPEG® and polymeric PEG reagents. 

BACKGROUND

Antibody-drug conjugates (ADCs) include a biologic (antibody) and small-molecule drugs held together with a small-molecule linker designed to modulate in vivo pharmacokinetic properties. The FDA regulations, therefore, span the spectrum of criteria for small molecules to biologics. Thus, the analytical technology must be able to characterize the small molecules, biologic, and the conjugate.  We will focus on select areas of ADC development and analysis, where mass spectrometry and post-acquisition analysis software are shown to be the ideal tools to characterize the full range of molecules encountered during the ADC development process. 

Before conjugation can be considered, the selected monoclonal antibody (mAb) amino acid sequence, glycosylation pattern, and additional post-translational modifications (PTMs) must first be confirmed and kept relatively consistent from lot-to-lot. Additionally, characterization of impurities generated during the purification and storage, which may include fragments and aggregates of the mAb, is required. Amino acid sequence confirmation and PTM identification are best done by LC/MS/MS analysis of protease digested peptides followed by mapping to the known or expected amino acid sequence, also known as peptide mapping. Confirmation of mass, glycosylation pattern, and the presence of fragments or multimeric aggregates of the desired mAb can be rapidly determined by LC/MS analysis of the intact protein sample, followed by deconvolution of the spectra to determine the masses present. 

Chemically synthesized small molecules to be conjugated to the mAb require detailed structural characterization and profiling of impurities at or above 0.1% relative to the synthesized molecule. The elemental composition of the drug, linker, and drug-linker molecules are readily determined by high-resolution MS (HRMS), while full structural characterization is accomplished by a combined application of NMR and HRMS/MS analysis. Small molecule impurities may arise during synthesis or the drug-linker coupling process. 

The complete conjugate form of the ADC requires characterization similar to that of the unconjugated mAb with a similar application of LC/MS analysis. The ADC amino acid sequence, glycosylation pattern, and additional PTMs require confirmation that they are the same as the original unconjugated mAb. The drug load distribution, measured by the drug-antibody ratio (DAR) and identification of drug/drug-linker site(s) of attachment to the mAb, are essential to defining the structure of the ADC. The choice of mAb conjugation site(s) (lysine, cysteine, or engineered unnatural amino acid) will affect the heterogeneity of the ADC. Depending on the heterogeneity of the ADC, the LCMS method and post-acquisition parameters may have to be tuned or optimized to ensure successful analysis. LC/MS analysis of the intact ADC followed by deconvolution of the MS spectra is used to determine the change in the zero-charge mass of the mAb and identify the average number of drugs attached (i.e., DAR). Reproducible site-specific conjugation is important to the development of a consistent ADC. The identification of specific amino acids conjugated to the drug/drug-linker is performed by LC/MS/MS peptide mapping. Lastly, unconjugated drug-related and drug conjugation impurities of the final ADC, that reach a level of 0.1% may require additional structural characterization by small molecule or biologics analyses, respectively. 

The ability to meet the level of precision required for reproducible manufacturing of a biological product is dependent upon the nature of the raw materials, the process used to make the defined product, and the analytical technology used to analyze the biological materials. In the instance of an ADC, the final conjugate should have a very narrow DAR distribution that can be precisely defined, quantified, and controlled. Thus, the combination of appropriate raw materials and analytical technology is essential to meeting the requirements for approvals and should be implemented early during the R&D process.

Analysis Product and Services

Rapid, targeted analysis of intact proteins by the application of mass spectrometry is an invaluable tool for accurate analysis at all levels of protein therapeutic development. Intact mAbs, protein subunits and multimeric species when ionized by electrospray (ESI) in positive ion mode are multiply charged. This protonation at multiple sites is a result of the propensity of basic amino acids and amino termini to carry positive charges. The resulting ESI mass spectrum of a typical mAb is a complex distribution of multiple charge states that must be “deconvoluted” or “transformed” in order to determine the uncharged mass of the mAb (Figure 1). ProMass is an automated platform for performing such deconvolutions of entire LC/MS data sets.

Figure 1: A monoclonal antibody (Murine IgG1, Intact mAb Mass Check Standard, 186006552, Waters) analyzed by ESI/LC/MS (A, top), expanded views of spectrum following noise filtering and PPL modeling (B, bottom left) and ProMass deconvolution (C, bottom right). The 4 peaks are the resolved glycosylation states of the antibody: 148223.1 Da (G0F-G0F), 148384.7 Da (G0F-G1F), 148546.8 Da (G1F-G1F), 148707.9 Da (G1F-G2F).

ProMass uses a novel deconvolution algorithm known as ZNova that produces artifact-free deconvoluted mass spectra. ZNova can be used to process data from a wide variety of biomolecules including proteins, oligonucleotides, peptides, etc. Unlike many other charge deconvolution algorithms, ZNova can accommodate complex, real-world samples and data of low signal-to-noise ratio. ProMass HR includes additional features for high-resolution data processing by incorporating the full suite of algorithms from Positive Probability Ltd (PPL). ProMass HR methods utilize the PPL algorithms that allow for monoisotopic mass determination of isotopically resolved mass spectra or enhanced resolution of isotopically unresolved mass spectra. As an example, Figure 1 shows the raw ESI mass spectrum of a murine IgG1 mAb. The noise-filtered, PPL modeled mass spectrum of the murine IgG1 mAb shown in Figure 1 has markedly enhanced resolution of the charge state peaks relative to those in the raw ESI mass spectrum (Figure 1). The PPL-enhanced charge deconvoluted mass spectrum is shown in Figure 1, which demonstrates excellent resolution of the four major glycoforms of the mAb. 

Over the years, Novatia has perfected the LCMS analysis of mAbs and ADCs through the development of highly-robust analytical methods that underpin the ProMass data processing software. Intact mAbs and ADCs are successfully analyzed in a wide array of volatile aqueous/organic buffers, using either native (neutral pH) or denaturing reversed-phase (low pH) techniques. Data acquisition of intact mAbs and ADCs, followed by deconvolution and/or DAR calculation, can be accomplished in as little as 1.5 minutes per sample. The rapid turnaround of results is highly beneficial in process development or process monitoring applications, which enables data-driven decisions to be made quickly.

In cases where an intact ADC is too heterogeneous or contains minor unknown masses, sacrificing speed for greater resolution and sensitivity is often necessary. Degylcosylation by 30-minute treatment with the enzyme PNGAse F is used to reduce the heterogeneity of the intact ADC, resulting in mass spectra of the conjugated mAb devoid of glycoforms. When knowledge of chain-specific conjugation is desired, a combination of PNGase F and disulfide bridge reduction can be used to reduce the ADC to its glycoform-free light and heavy chains. Ultimately, the nature of the conjugation and the stability of the ADC determines the types of LC/MS analysis and sample treatment protocols that are appropriate for the characterization of the ADC.

The use of discrete PEGs in the production of biological materials

PEGylation is the process of adding polyethylene glycol (PEG) to a target molecule or surface, often through covalent modification. Examples of PEGylated substances include small-molecule therapeutic drugs, peptides, proteins, the carbohydrate coats of glycoproteins, oligonucleotides, and lipids. PEGylation historically has used polymeric PEGs, which are characterized by their dispersity. This dispersity is measured by a “dispersity index” (Đ), formerly known as the polymer dispersion index (PDI). Lower molecular weight polymers have a larger Đ than those of higher molecular weights. However, no matter the dispersity index, the molecular weight species have a Poisson distribution that forms a bell-shaped curve.

Quanta BioDesign, Ltd invented the process to manufacture the single-molecule dPEG®. Like traditional PEGs, these products contain an amphiphilic backbone of repeating ethylene oxide units. However, the dPEG® products are monodisperse PEG products, manufactured using proprietary synthetic and purification processes. Because the dPEG® products are manufactured as a single molecules (monomer), not purified from a polymeric mixture, larger quantities can be manufactured for incorporation into diagnostic and pharmaceutical products.Monofunctional dPEG® reagents can be used as protein modifiers, whereas bifunctional dPEG® reagents are commonly used as linkers. The linkers provide a means to conjugate molecules, such as the bridge between the antibody and API in ADCs. The selection of the dPEG® product should be empirically determined as modest differences in the number and configuration of ethylene glycol units can modulate conjugate properties and in vivo pharmacokinetic performance. dPEG® linkers have been shown to increase water solubility, increase hydrodynamic volume, decrease immunogenicity, and modify the biodistribution and pharmacokinetic properties of therapeutic, theragnostic, and diagnostic molecules. The bifunctional dPEG® products are available in linear, branched, and Sidewinder™ conformations with a range of functionalities including amine and thiol-reactive, biotin, copper-free click chemistry, and various protecting groups.

Figure 2. Raw ESI mass spectra of the discrete m-dPEG®24-NHS ester (top; (M+NH4)+1 of 1231.6975 m/z and (M+2NH4)+2 of 624.8662 m/z) and mPEG 1K- NHS (SC) (bottom; average MW: 1000) analyzed in positive ion mode. 

HRMS characterization of discrete and polydispersed PEG molecules

As was previously stated, the characterization of the small molecules conjugated to the antibody is an important step in the development process. Depending on the instrument used for the analysis, molecules under 25kDa can be isotopically resolved allowing  chemical composition to be confirmed. The monoisotopic mass of a molecule is defined as the sum of the primary isotope mass of each element present in the molecular formula. For example, the monoisotopic mass of hydrogen is 1.0078 Da and oxygen is 15.9949 Da, therefore the monoisotopic mass of water is 18.0105 Da (1.0078+1.0078+15.9949). In this study, we confirmed the monoisotopic masses of m-dPEGâ12-NHS, m-dPEGâ24-NHS and dPEGâ4-(m-dPEGâ24)3-NHS esters and the polydispersed mPEG 1K-NHS molecules.

Since the small molecule samples were anticipated to be highly pure, the m-dPEG®24-NHS and mPEG 1K-NHS esters were analyzed by flow injection HRMS, in positive ion mode, to confirm the monoisotopic masses of the PEG molecules. The raw mass spectra of similarly sized m-dPEG®24-NHS ester (discrete PEG) and mPEG 1K-NHS ester (polydisperse PEG) appear vastly different. The m-dPEG®24-NHS ester spectrum consists of the ammonium adduct forms of a single intact PEGylated molecule (M+NH4)+1 of 1231.6975 m/z and (M+2NH4)+2 of 624.8662 m/z; while the mPEG 1K ester appears as a distribution of PEGylated species of various lengths contained within a series of bell-shaped distributions (Figure 2). To calculate the zero-charge exact mass and confirm the chemical formula, the expected chemical formula and raw spectrum of the sample were used to construct a deisotoping method in PPL ReView application of the ProMass software. The deisotoped mass spectra of m-dPEG®24-NHS ester and mPEG 1K-NHS ester (shown in Figure 3) look similar to the raw mass spectra shown in Figure 2, except for the removal of background that was obscuring low level ions in the mPEG 1K-NHS ester spectrum. It is important to note that the proton responsible for the positive charge on the molecule is located on the ammonium ion; therefore, the zero-charge species observed is the ammonia adduct. The mPEG 1K-NHS ester deisotoped mass spectrum is a complex distribution of PEG molecules of different lengths with hydrolyzed NHS ester (Ex. C56H111O30 = 1263.7140 Da) and +NH3 adduct series of the intact (Ex. C60H115O32N•NH3 = 1378.7660 Da) identified as the additional bell-shaped curves.Two additional monofunctional dPEG® constructs were analyzed by ESI/LC/MS with deisotoping using ProMass HR deconvolution software. The major masses in both the dPEG®12-NHS ester and NHS-dPEG®4-Tris(m-dPEG®24)3 ester samples are the ammonium adducts, consistent with the observations in the analysis of dPEG®12-NHS ester. Figure 4 illustrates that increasing or decreasing mass does not lead to significant dispersion of the PEG length, as observed in the polydispersed PEG species. The smaller peaks clustered around the labeled peaks are additional ammonia and acetonitrile adducts similar to those shown in Figure 3.

Figure 3. Deisotoped mass spectra of the discrete m-dPEG®24-NHS ester adducts (top; zero-charge masses of (M•NH3) = 1230.6910 Da and (M•2NH3) = 1247.7180 Da) and mPEG 1K- NHS (SC) (bottom; hydrolyzed NHS ester Ex. C56H111O30 = 1263.7140 Da) and +NH3 adduct series of the intact (Ex. C60H115O32N•NH3 = 1378.7660 Da).
Figure 4: ProMass HR deisotoping of ESI spectra of additional monodisperse m-dPEG®-NHS ester products m-dPEG®12-NHS ester (top: C30H55NO16•NH3: zero-charge exact mass = 702.377 Da), m-dPEG®24-NHS ester adduct (middle: zero-charge masses of (M•2NH3) = 1247.7180 Da) and NHS-dPEG®4-(m-dPEG®24)3-ester (bottom: C180H350N6O88•3NH3: zero-charge exact mass = 4055.387 Da.

LC/MS characterization of dPEG constructs conjugated to mouse IgG1 antibody

It is important to appreciate early in the product development stage, that the complexity or heterogeneity of each of the raw materials leads to compounded complexity and limited resolution in the analysis of the final product. In order to accurately deconvolute the observed ESI mass spectrum of mouse IgG1 mAb conjugated with dPEG® constructs, we first developed a model of the deglycosylated mAb (data not shown) mass spectrum, using the PPL algorithm in ProMass HR. By removing the N-linked glycans, the observed mass spectrum could be re-plotted following the removal of the background signal and more accurately represent the protein m/z peaks of the charge envelope. The deglycosylated peak model is incorporated into the ProMass HR parameter method used to deconvolute the raw ESI mass spectra of intact IgG1 mAb (shown in Figure 1) and the conjugation of IgG1 with m-dPEG®12-NHS ester (shown in Figure 5).

Figure 5. The raw ESI spectrum of the dPEG®12-labeled mouse IgG1 mAb (top) and the ProMass HR deconvoluted spectrum (bottom). No labeled states were detected below 4 or above 10 labels.

The mAb was labeled with 10 molar excess of m-dPEG®12-NHS ester using standard amine-reactive labeling conditions. The intact antibody conjugate was analyzed by LC/MS, leaving the N-linked glycans in place (Figure 5). We expected that labeling the available amines on a monoclonal antibody with 10 molar excess of the NHS-ester dPEG® reagents would lead to a similar distribution of labels per antibody observed in Figure 5 and Table 1. ProMass deconvolution resolved the labeled peaks despite the presence of the glycans on the intact antibody and maintained the same glycosylation pattern observed in the unconjugated IgG1 mAb (shown in Figure 1). The deconvoluted mass spectrum of the m-dPEG®12-labeled antibody appears as a series of 3 peak clusters of the most abundant glycosylation states and increased by 570.5 Da (Figure 5 and Table 1). The use of single-molecule dPEG® constructs with functionalities designed for site-specific conjugation, when analyzed by ProMass HR, allows for each population of molecules to be measured, including unlabeled, a single label, and multiple labels. An additional feature of the latest version of ProMass HR is the ability to automate average DAR value calculations by simply inputting the mass of the mAb and the expected mass addition of a single conjugation.

Table 1: Conjugate mass addition and average number of conjugates (DAR) for dPEG®12-labeling of mAb.

A comparable conjugation of 10 molar excess m-1K PEG – NHS (SC) labeled IgG1 mAb, using identical conditions to the m-dPEG®12-labeled antibody, was produced. The intact antibody conjugate was analyzed by LC/MS, leaving the N-linked glycans in place. The resulting polydisperse m-1KPEG-labeled IgG1 was too complex to deconvolute the entire elution peak. Instead of deconvoluting the entire peak of elution, 10-second slices of the peak were deconvoluted and summed to represent the whole peak. As shown in Figure 6, the most abundant peak was a broad distribution representing a population of ~5-7 labels per antibody. Due to the distribution of mass with a polymer, it is not possible to resolve to the same level that can be achieved with the single-molecule dPEG® reagents. The resulting complexity would be exacerbated when additional drug molecules are incorporated at the other functional end of the PEG chain.In the final conjugation examples, IgG1 mAb samples were labeled with 10 molar excess of either m-dPEGâ24-NHS or dPEGâ4-(m-dPEGâ24)3-NHS esters. Once again, the conjugated antibodies were analyzed intact, with glycans remaining in place. The monodispersed dPEG® labeled conjugate raw ESI mass spectra were noticeably more complex than the dPEGâ12-labeled antibody, but less complex than that of the m-1KPEG polydispersed labeled mAb. ProMass HR easily deconvoluted the spectra using identical parameter method settings to that of dPEG®12-labeled sample. Figure 7 shows the deconvoluted mass spectra of the dPEG® construct labeled antibodies and shows the average DAR values. The same 3 peak cluster of labeled peaks increasing by 1100 and 3892.6, respectively, were observed in both the m-dPEGâ24-NHS and dPEGâ4-(m-dPEGâ24)3-NHS labeled antibodies. In addition, we observed that the average DAR value decreased as the overall mass and complexity of the label increased. Additional experiments are required to understand this observation fully.

Figure 6. The raw ESI spectrum of the mPEG 1K-labeled mouse IgG1 mAb (top) and the ProMass HR deconvoluted spectrum (bottom). The broader deconvoluted peaks contain an abundance of masses which correspond to the average conjugation range shown in the spectrum (bottom), this is due to the polydispersive character of the mPEG 1K molecules. Low level polydispersed peaks lead to the production of deconvolution artifacts shown in clusters to the right of the “5-7 label peak”; based on the molar ratio for conjugation it is unlikely that these peaks represent real conjugates.
Figure 7. The ProMass HR deconvoluted mass spectra of the m-dPEGâ12– (top), m-dPEGâ24– (middle) and dPEGâ4-(m-dPEGâ24)3– (bottom) labeled mouse IgG1 mAb are shown above. All 3 conjugates show the same 3 peak cluster (inset of each spectrum) of labeled glycosylated species of IgG1 increasing by 570.5 (m-dPEGâ12-), 1100 (m-dPEGâ24-) or 3892.6 (dPEGâ4-(m-dPEGâ24)3-) Da, respectively.

Conclusion 

FDA regulations require rigorous characterization of INDs. The composition of these drugs is highly complex and the ability to properly characterize the final product is impacted by the homogeneity of the starting materials and the power of the analytical methods. Here we demonstrate that the Novatia LC/MS-based analysis of the intact antibody can resolve the glycosylation state and distribution of labeling as representative of the DAR due to the purity of Quanta BioDesign dPEG® reagents. Furthermore, we demonstrate that the glycosylation states and DAR can be resolved with the addition of small m-dPEG®12-NHS ester (MW: 685.75) to larger linear m-dPEG®24-NHS ester (MW: 1214.39) and even to the complex branched NHS-dPEG®4-Tris(m-dPEG®24)3 ester (MW 4006.69) products. In contrast, conjugation with a polymeric PEG limits the level of quantification to an average DAR without the ability to resolve glycosylation states or the distribution of DAR. Incorporation of the single-molecule dPEG® reagents is applicable for those who are developing linkers for ADCs as well as wide range biologic conjugates, all of which can be best characterized using LC/MS based analysis.

Two of the most critical considerations for gaining regulatory approval of INDs are limiting heterogeneity of the raw materials and establishing the appropriate analytical methodologies early in the development process. The scientists at Quanta BioDesign and Novatia are dedicated to helping our clients address both obstacles at all stages of development. Our products and services are available across the spectrum of Research and Development, from benchtop science to manufacturing at clinical scale. 

Materials and Methods 

The monodisperse dPEG® reagents m-dPEG®12-NHS ester (MW: 685.75, PN10262), m-dPEG®24-NHS ester (MW: 1214.39, PN10304), and NHS-dPEG®4-Tris(m-dPEG®24)3 ester (MW 4006.69, PN10454) were manufactured by Quanta BioDesign (Plain City, OH). The polymeric PEG reagent mPEG 1K- NHS(SC) (PG1-SC-1k-1) was purchased from Nanocs. The representative monoclonal antibody used for conjugation was Murine IgG1, Intact mAb Mass Check Standard (186006552, Waters). To prepare the conjugates, PEG reagents were first solubilized in a stock of dried DMAC. The antibodies were dialyzed using a 20K MWCO Slide-A-Lyzer G2 (ThermoFisher) and then adjusted to the concentration of 4mg/ml IgG in 50mM borate buffer, pH 8.5. The amine-reactive PEG constructs were added at a 10 molar excess per mole of IgG, for 1h at room temperature. Excess unconjugated PEG products were removed and buffer exchanged to 10mM ammonium acetate using Ultracel 0.5ml, 50K MWCO centrifuge filters (Amicon). Each of these reagents was analyzed individually and as the respective antibody-PEG conjugate. The monoclonal antibody, PEG reagents, and antibody-PEG conjugates were analyzed by Novatia using reversed-phase LC/MS or native SEC/ MS methods on the Waters Xevo Q-TOF mass spectrometer followed by deconvolution and analysis using the ProMass HR for MassLynx software.