**1. Introduction**

Complex generic and biosimilar drug products (DPs) are increasingly developed and comprehensive analysis of these DPs is the foundation for their regulatory approval [1–4]. The active pharmaceutical ingredient (API) or drug substance (DS) in protein DPs ranges in size from short peptides to large monoclonal antibodies (mAbs). The native folding of proteins, heterogeneity, dynamic exchange between conformations, oligomerization and aggregation profile in a formulation are collectively called the higher order structure (HOS) properties of protein therapeutics and are typically critical for efficacy and safety [5].

**Citation:** Wang, D.; Zhuo, Y.; Karfunkle, M.; Patil, S.M.; Smith, C.J.; Keire, D.A.; Chen, K. NMR Spectroscopy for Protein Higher Order Structure Similarity Assessment in Formulated Drug Products. *Molecules* **2021**, *26*, 4251. https://doi.org/10.3390/ molecules26144251

Academic Editor: Robert Brinson

Received: 21 May 2021 Accepted: 8 July 2021 Published: 13 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Protein HOS is stabilized by weak hydrogen bonding, electrostatic and hydrophobic forces, which are solvent dependent, and, consequently, formulation differences affect HOS [6–9]. In addition, proteins can be chemically modified either purposely, e.g., pegylation, or unintentionally, e.g., oxidation, which could introduce variability to protein HOS [10–12]. All these factors and the accompanying sensitivity to solution conditions necessitate characterizing the protein chemistry and HOS with minimal perturbation to the formulation by ideally using DPs [13]. In addition, the analytical means to assess protein HOS in a formulation are desired for generic and biosimilar drug developers that mostly only have access to the marketed originator DPs that are usually deemed as the reference DPs.

With the development of higher field strength magnets and cryogenic probes, modern high-resolution NMR spectroscopy is a non-invasive and sensitive method for protein molecular structure characterization [14–18]. However, several assumptions among stakeholders have limited the application of NMR on formulated DPs. The first is that strong excipient signals in a DP would interfere with weak DS signals such that NMR spectra would be dominated by the peaks of the excipients and would not be useful for protein HOS assessment. Indeed, NMR for protein HOS characterizations [19] has been applied on proteins extracted from DP [20–22], proprietary DS or non-marketed DP [23–25], which aimed to demonstrate the applicability of modern heteronuclear NMR to characterize proteins with 15N and 13C nuclei at natural abundances.

Second, the lack of acceptable metrics for similarity assessment means that most comparisons have been made at a visual level. The question of the level of similarity that is practically measurable remains to be answered quantitively. Previous attempts were made to collect NMR spectra on a DS enriched formulation of filgrastim [26] and DP formulations of insulin [27]. A combined chemical shift difference of 8 ppb or less was proposed as the threshold for experimental precision in 2D-NMR comparisons of biosimilars using data between the US and Indian marketed filgrastim DPs [26]. The principal component analysis (PCA) of the insulin DP NMR spectra revealed the practically achievable similarity threshold expressed in Mahalanobis distance (DM) to be 3.3 or less [27]. These values were achieved when 600 MHz spectrometer with room temperature probe was used, therefore, the derived metrics were practical (in terms of the availability of instruments) and could be useful in establishing the acceptance criteria for a certain DP before and after a manufacturing change and for the comparison between a generic or biosimilar protein and the reference DP. However, their validity has not been further tested.

Third, the type of HOS properties reliably measured from DPs using modern NMR is not entirely clear. Herein, 1D 1H NMR spectra were acquired on a range of marketed DPs with protein molecular weights ranging from 3 kDa to 145 kDa and with the protein concentration as low as 0.01 mM. The protein HOS properties of folding, intermediate exchange and oligomerization were all reflected in the NMR spectral patterns. Using rituximab DPs, the proposed DM similarity metric was verified again. Using insulin glargine DPs, the methyl peak profile method showed that both chemical shift and relative peak height can be used to derive practically achievable similarity metrics. Finally, the sensitive 1H-15N sofast HMQC experiment was demonstrated to be a valuable NMR method to characterize the protein backbone HOS.

#### **2. Results**

The peptide and protein drug products (DPs) listed in Table 1 were sourced from the US market except Reditux®, which was sourced from India. All DPs are the reference drugs except Basaglar® and Reditux®, which are follow-on products to Lantus® and Rituxan®, respectively. All 1D 1H, 2D 1H-13C and 2D 1H-15N NMR spectra were collected on formulated DPs with minimal dilution of adding 5% D2O (*v*/*v*).


**Table 1.** Drug Products studied.

<sup>1</sup> Based on the equivalence between 1 mg and 6000 I.U. per USP NF.; <sup>2</sup> based on the equivalence between 0.0347 mg and 1 USP unit per USP NF and Eu. Pharm.; n/a: not available from the drug label.
