Standardization of allergen products: 2. Detailed characterization of GMP‐produced recombinant Phl p 5.0109 as European Pharmacopoeia reference standard

The Biological Standardization Programme of the European Directorate for Quality of Medicines and Healthcare (EDQM) aims at the establishment of well‐characterized reference standards based on recombinant allergens and validated assays for the quantification of major allergen content. The objective of this study was to examine the detailed physicochemical and immunological characterization of recombinant Phl p 5.0109, the second available allergen reference standard.

Grass pollen ranges among the most prevalent allergen sources worldwide (1). Eleven different grass pollen allergens have been identified (2). Specific immunotherapy against grass pollen allergy, the potentially only curative treatment, is currently still based on grass pollen extracts, which are inherently subject of variations regarding their representation of different grass pollen (iso)allergens (3,4). Thus, the need for standardization of allergen extracts represents an important issue for allergy diagnosis and immunotherapy (5). This has been recognized by the American Academy of Allergy, Asthma, and Immunology (6), as well as by publications of European regulatory authorities (7,8). For most allergen products, standardization of allergen extracts is still based on the determination of the IgE potency, performed by competitive IgE-binding assays and This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. skin testing, which also raises ethical concerns and safety aspects (9)(10)(11). Using these types of biological assays, the real allergen content is not determined. The resulting potency values are expressed in company-specific units and are not comparable between different manufacturers. Therefore, efforts to set up international standards for the most relevant allergen extracts were initiated by the World Health Organization/International Union of Immunological Societies (WHO/ IUIS) Allergen Standardization Subcommittee (12), and this endeavor was continued by the EU-funded CREATE project (13,14), from which two allergens were selected for the Biological Standardization Programme project BSP090 of the European Directorate for Quality of Medicines and Healthcare (EDQM). Being a clinically very relevant isoform (sensitization rate of 65-85%), rPhl p 5.0109 was selected for a clinical study using a mixture of five recombinant timothy grass pollen allergens (15,16). Thus, rPhl p 5.0109 was produced under Good Manufacturing Practice (GMP) conditions, characterized physicochemically and evaluated in a ring trial using selected sandwich ELISA systems (17,18) pursuing the aim to validate pairs of allergen plus suitable quantification system for the applicability in allergen standardization based on mass units of major allergen (19). While the first candidate biological reference preparation, recombinant Bet v 1.0101, has previously been published (20), we present here the detailed characterization of GMP-produced rPhl p 5.0109 PP5ar06007 as European Pharmacopoeia (Ph. Eur.) reference standard of Phleum pratense pollen-related allergen products.

Preparation of allergens
Allergopharma (Reinbek, Germany) has developed recombinant Phl p 5.0109 to be used in a cocktail of allergens for specific immunotherapy against grass pollinosis (21). Briefly, recombinant Phl p 5 was expressed in E. coli and purified using a sequence of chromatographic steps including hydrophobic interaction, ion-exchange, and size-exclusion chromatography. Natural Phl p 5 (nPhl p 5) was purified from water-soluble extracts prepared from Phleum pratense pollen (Allergon, Angelholm, Sweden).

Physicochemical characterization
Identity, quantity, homogeneity, folding, stability, and aggregation analyses were performed based on methodologies as described earlier (20). Additionally, high-resolution mass spectrometry and Fourier transform infrared spectroscopy analyses were performed. A detailed description can be found in the article's online Supporting Information.

Immunological characterization
Epitope integrity and equivalence of PP5ar06007 with natural Phl p 5 was determined by basophil activation test and T-cell reactivity. A detailed description can be found in the article's online Supporting Information.

Results
Several batches of recombinant Phl p 5.0109 were produced and purified under GMP conditions with a high batch-tobatch consistency (Table S1). Purity, identity, and homogeneity have been shown by physicochemical and immunological analytical methods. Batch PP5ar06007 was chosen for the further development of a reference standard. From this bulk material, final formulations of stable reference standard were prepared by adding 0.2% BSA and 0.1% trehalose, lyophilized, and stored in aliquots at the EDQM.

Identity
The protein sequence of PP5ar06007 was shown to be identical with the major Phleum pratense pollen allergen Phl p 5 having the Uniprot KB accession number Q84UI2 by mass spectrometry-based methods. First, intact mass measurement resulted in an average molecular mass of 28286.0 AE 0.6 Da (calibration error with GFP: 20 ppm) coinciding with the theoretical MW of Q84UI2 (N-terminal methionine removed). An additional peak at 28329.0 AE 0.6 Da with an intensity of approx. 30% was found (Fig. 1A). This side peak was identified by further mass spectrometric investigations to represent an N-terminally acetylated variant of Phl p 5.0109. Shortly, a monoisotopic mass difference of +42.0397 of the side peak compared to unmodified Phl p 5.0109 was determined using high-resolution (<1 ppm) orbitrap MS technology, and evidence for N-terminal acetylation was supplemented by top-down sequencing procedures on the intact molecule ( Fig. 1A insert). Second, nanoLC-MSMS-based peptide mapping of tryptic and V8 protease digests resulted in a merged sequence coverage of 100% (Fig. 1A insert). Additional search for modifications resulted in a positive  Table S2; unlabeled peaks are derived from self-digestion of trypsin or represented system peaks also appearing in empty runs.
(C) The protein amount was quantified by amino acid analysis to be 1.56 AE 0.03 mg/ml. Experimentally determined numbers of amino acid residues (gray bars with standard deviations from triplicate) showed a good agreement with the theoretical values (black bars).
signal for the N-terminally acetylated peptide 1-30, having a monoisotopic mass of 2722.3172 differing from unmodified peptide by +42.0031, which coincided well with the theoretical mass difference of +42.0106. Thus, PP5ar06007 was shown to consist of approximately 30% N-terminally acetylated Phl p 5.0109. Further possible modifications included deamidations of Asp 193 and Asp 194 , respectively, as well as the oxidation of Met 247 . However, these possible modifications were found just in the tryptic digests and were not reflected by the results from the intact mass measurements. Notably, such modifications can occur spontaneously during the incubation period of proteolytic digestion, as they also may represent aging effects on proteins. PP5ar06007 did not show such aging effects.

Quantity
Amino acid analysis was performed by two laboratories, one of which holding a GMP certification, for exactly quantifying PP5ar06007. Figure 1C shows the results from the first analysis. During acid hydrolysis, Cys and Trp were (partially) degraded and Asn and Gln were converted to Asp and Glu, respectively. Comparison of experimental (average with standard deviation of triplicates) with theoretical amino acid content was found to be in good agreement (standard deviations <10%) for most residues. Deviations of >10% were found for Thr, Tyr, and Phe; however, due to our experience, such deviations have to be tolerated when using this method. Results for protein quantity obtained without considering these residues stayed within the standard deviation given below. In the second analysis (GMP-certified), which was performed in duplicate, just the robust amino acids Asp, Ala, Phe, Val, Ile, and Leu were used for quantification. A concentration of 1.56 AE 0.03 mg/ml was determined.

Homogeneity
For the characterization of homogeneity of PP5ar06007, denaturing gel electrophoresis, isoelectric focusing, and sizeexclusion chromatography were applied. In SDS-PAGE, PP5ar06007 migrated as a single band at approximately 35 kD ( Fig. 2A) matching the upper band of natural Phl p 5 preparations and the usual observation of other recombinant Phl p 5 isoforms (data not shown). A migration behavior in SDS-PAGE differing from the theoretical MW has been reported previously; one reason may be that the attachment of SDS hydrophobic, acidic, basic, or glycosylated sequence stretches might be inefficient (21)(22)(23). Silver staining as well as immunoblotting with two different Phl p 5-specific monoclonal antibodies and a serum pool of grass pollenallergic patients did not reveal any further protein contamination. Under native conditions, an isoelectric point (pI) of approximately 8.6 was determined (Fig. 2B) contrasting the theoretical pI of 7.9; however, such deviations can frequently be observed for recombinant proteins under nondenaturing conditions. Isoelectric focusing serves primarily for monitoring homogeneity in regard to sample aging processes such as partial oxidation of sulfhydryl groups and deamidation of asparagine or glutamine residues (20,21). In this regard, a very high degree of surface charge homogeneity could be verified for PP5ar06007. The very faint band slightly below the main band may represent N-terminally acetylated Phl p 5, as this minor side component was identified by mass spectrometry. Accordingly, no aggregates or oligomers were detected by HPSEC light scattering (Fig. 2C). Monomeric molecules eluted as a single peak at a retention volume of 8.7 ml. Upon evaluating the right-angle light scattering, refractive index, and viscosity signals, a MW of approximately 25 kD and a hydrodynamic radius of 3.0 nm were determined. No further peaks appeared in the light-scattering chromatogram between the void (5.7 ml) and total (12.0 ml) retention volume. As this detector signal is especially sensitive to high molecular weight aggregates, PP5ar06007 was determined to consist of >99.9% homogeneous monomeric molecule.

Folding and denaturation analysis
The CD spectrum of PP5ar06007 showed a maximum around 195 nm and a double minimum at 205 and 222 nm (Fig. 3A, left panel), typical for protein preparations with a high alpha-helical secondary structure content and similar to a corresponding CD analysis of nPhl p 5 (Fig. 3B). Upon heating, denaturation temperatures of 66°C and 67°C were determined, and a high degree of reversibility of protein unfolding was shown from the renaturation curves for the recombinant and natural preparations, respectively ( Fig. 3A and B, right panel). However, the renaturation curves showed less significant inflection points, and the CD spectra recorded immediately after renaturation showed lower maximal and higher minimal Θ mrw values. Notably, after some time at room temperature, PPar506007 gained its normal appearance in CD spectrum (data not shown), indicating that the refolding process takes longer than the experimental time. These findings were also verified using FTIR spectroscopy, where a strong sigmoidal decrease of the integral 1620-1655/cm, corresponding to the amide I vibration range of alpha helices, was observed (Fig. 3C). The decrease in alpha-helical secondary structure content was concomitant with an increase in random coil (integral 1658-1700/cm) and interchain beta-sheet (integral 1605-1626/cm) secondary structure content, of which the latter indicated heat-induced protein aggregation. Notably, some part of the material precipitated during thermal denaturation supporting the FTIR data. However, after heat denaturation, still soluble portion of PP5ar06007 did not form aggregates as investigated by DLS (Fig. 4A, dotted line).

Aggregation behavior and stability in solution and upon lyophilization with/without trehalose
The aggregation behavior in solution was investigated using DLS (Fig. 4A,  be assumed that PP5ar06007 appeared either as a mixture of monomeric and dimeric molecules or, more likely, adopted an elongated shape, which would be coinciding with the R H value obtained for monomeric PP5ar06007 by HPSEC. However, no high molecular weight aggregates were detected. Storage stability of the bulk substance at +4°C was tested after 2 months showing no significant changes (Fig. 4A,  dashed line). Moreover, a sample stored at À20°C with intermittent thawing and freezing steps was retested after >8 years upon additional thermal stressing for >3 days at >50°C (but below T m ) showing <1% aggregates (Fig. 4A, dashed dotted line). Upon routinely performed centrifugation, no precipitation was observed in any sample. Furthermore, no significant changes in the mass distribution of the DLS signals, except for a slight decrease in polydispersity, were observed: 16.5% RSD in +4°C-stored sample and 19.0% RSD in frozen/ thawed/heat-stressed sample. In summary, it can be stated that the PP5ar06007 bulk material retained its monomeric state in solution for months at +4°C and years at À20°C with intermittent thawing/freezing and was not susceptible to heat stress below T m . Using FTIR of PP5ar06007 in saline solution, a secondary structure content of 46.7% alpha helices and 8.4% intrachain beta sheets was determined (Fig. 4B), and these values remained similar upon lyophilization with and without 0.1% trehalose, a stabilizing additive in the final formulation of the Ph. Eur. reference standard (besides 0.2% BSA). Notably, no intermolecular beta-sheet formation (amide I vibration range from 1605 to 1626/cm) indicating protein aggregation was observed in any of the lyophilized preparations.
The formulated final recombinant Phl p 5.0109 Chemical Reference Substance (CRS) is stored at the EDQM at À20°C. Its stability is monitored annually by triple parallel measurement relative to a baseline reference (stored at

Biological activity
In order to assess the biological activity of PP5ar06007, basophil activation and T-cell reactivity were chosen as they   represent the best suited functional assays to compare recombinant allergens with their natural counterparts, which is of importance during the registration of new allergen products where PP5ar6007 serves as calibrator of major allergen content in natural allergen extracts. The allergenic potential of PP5ar06007 was assessed in reference to its natural counterpart using direct basophil activation in heparinized blood samples from 15 grass pollenallergic patients. From the individual titration curves (Fig. S1), the concentrations of allergen needed for half-maximal basophil activation (C50 values) were determined. The C50 values of PP5ar06007 and five other recombinant production batches were expressed in relation to nPhl p 5 as P rel values (Fig. 5B, medians: 0.91, 0.88, 0.97, 0.82, 0.82, and 0.94), demonstrating satisfying allergenic equivalence of the present reference standard with the natural allergen. Moreover, the GMP production process proved reliable and robust for the establishment of further batches that may be required in the far future.
The T-cell responses to PP5ar06007 were compared to nPhl p 5 and another recombinant production batch (PP5ar10015) using a panel of allergen-specific TCLs and TCCs derived from grass pollen-allergic individuals. Aggregation behavior and stability of rPhl p 5 PP5ar06007 in solution, lyophilized with and without 0.1% trehalose as stabilizer. (A) Dynamic light scattering of PP5ar06007 statistically weighted by the mass model for proteins using the OmniSize TM software resulted in a single fraction with a hydrodynamic radius (R H ) of 3.1 nm (solid line). Moreover, this peak appeared to be rather broad with a polydispersity of 23.6% root square deviation (RSD). Thus, it could be assumed that PP5ar06007 appeared either as a mixture of monomeric and dimeric molecules or, more likely, adopted an elongated shape, which would be coinciding with the R H value of 3.0 obtained for monomeric PP5ar06007 by SECTDA. However, no high molecular weight aggregates were detected. Dynamic light scattering after 9 weeks storage at +4°C (dashed line) revealed no significant changes in the R H (3.2 nm) and polydispersity (16.5% RSD) values. No high molecular weight aggregates were detected in PP5ar06007 within the observed storage time. Upon heating to 95°C, slow cooling, and centrifugation, a lower fraction of still soluble PP5ar06007 appeared as monomeric molecule with a R H of 3.2 nm and a polydispersity of 21.6% RSD (dotted line). Storage at À20°C with intermittent thawing and freezing and heat stressing displayed a similar R H of 3.1 nm, a slightly decreased polydispersity of 19.0% RSD, and <1% aggregates (dashed dotted line). (B) Fourier transform infrared spectroscopy of PP5ar06007 in solution (solid line), lyophilized from saline with (dotted line) and without (dashed line) trehalose revealed primarily alpha-helical secondary structure content. The amide I maxima at 1655 and 1550/cm, respectively, are characteristic for alpha-helical secondary structure content (light gray). (C) Second derivatives were computed to more precisely allocate the peak maxima of the recorded FTIR spectra in B. diagonal, indicating agreement of T-cell reactivity between recombinant and natural molecules. Few exceptions showed a higher reactivity with natural compared to recombinant Phl p 5. This can be explained as the TCLs and TCCs were raised using natural and recombinant allergen, a strategy that would not exclude T cells with higher specificity for isoforms present only in the natural mixture. Batch-to-batch consistency was monitored at the T-cell level by comparing PP5ar06007 with batch PP5ar10015 (Table S3). From all TCLs and TCCs, a median reactivity of 9.43 was determined for PP5ar06007, which was in good agreement with 10.32 for nPhl p 5 and 9.36 for PP5ar10015. The epitope specificities from most of the used TCL/TCCs were mapped covering a wide range of the allergen sequence (positions 40-246). Nine different partially overlapping epitopes were identified (25). Thus, the close correlation between PP5ar06007 and its natural counterpart evidenced the reference standard to be an excellent representative for grass pollen allergens also at the T-cell level.

Discussion
Pharmaceutical products require the highest safety profiles that can be achieved using the currently available methodology. This holds also true for allergen products determined for use in specific immunotherapy for allergic individuals (26). High production standards have to be fulfilled, which is realized by procedures adhering to the concept of GMP including state-of-the-art characterization techniques (27,28). In line with this, the process of allergen standardization has been promoted by several stakeholders in a project run by EDQM. Resulting from these efforts, Ph. Eur. reference standards for two major allergen products, birch and grass pollen, and a corresponding quantification method for Bet v 1 have become available (18). Here, we present the detailed characterization of rPhl p 5.0109 PP5ar06007 serving as calibrator for timothy grass pollen allergen products seeking marketing authorization of the European Union. As described before for rBet v 1.0101 Y0487, the main physicochemical criteria were investigated, including identity, quantity, homogeneity, folding, aggregation behavior/stability in solution, and biological activity (20). In addition, this study presents FTIR-based investigations of folding and susceptibility toward thermal denaturation for the product in solution and in trehalose-formulated lyophilized state, as this represents the Ph. Eur.-provided formulation stored at the EDQM. The FTIR analyses complemented the CD results well, where a loss in alpha-helical secondary structure content was indicated upon thermal denaturation/renaturation. Moreover, the biological activity was assessed in terms of basophil activation and T-cell reactivity compared to its natural counterpart. This is of great importance as a biological reference standard has to represent the natural molecule appearing in the allergen extracts from different angles, which in case of an active pharmaceutical ingredient for allergen immunotherapy means to be similar at the level of recognition by IgE antibodies and T cells. This similarity could be shown using human basophils from a panel of allergic donors and an array of allergen-specific T-cell lines and clones. The biological activity was furthermore established as rPhl p 5.0109 was used before in vivo showing together with P P 5 a r 0 6 0 0 7 P P 5 a r P V 0 3 P P 5 a r 0 4 0 0 2 P P 5 a r 0 5 0 0 3 P P 5 a r 0  Table S3). Stimulation indices showed a Pearson correlation of q = 0.963 (P < 0.0001). The gray dashed line represents optimal correlation in T-cell reactivity.
other recombinant major timothy grass pollen allergens sufficient clinical efficacy (15,16). Several GMP-produced batches were investigated and similar results were obtained. The physicochemical and biological assays performed certified thus a very robust GMP-based production process, which evidently represents an important prerequisite for long-term availability of a biological reference standard.
Overall, GMP-grade rPhl p 5.0109 appears highly similar to its natural counterpart from the physicochemical, structural, and biological point of view. As such, rPhl p 5.0109 is suitable as Ph. Eur. reference standard.

Acknowledgments
This work was supported by a grant from the EDQM, the Land Salzburg, and the Christian Doppler Association (CD Lab for Allergy Diagnosis and Therapy). The work of MH was supported by grant P18820-B13 of the Austria Science Funds (FWF) and Biomay AG, Vienna, Austria. We thank Brigitta Arlt, Britta Friggemann, and Kerstin Joppich-R€ ohm (Allergopharma) for their excellent technical assistance in performing the basophil activation test and T-cell proliferation assay.

Conflict of interest
Dr. Nandy reports personal fees from Allergopharma GmbH & Co. KG, during the conduct of the study; Dr. Kahlert reports personal fees from Allergopharma, during the conduct of the study; Dr. Thilker reports personal fees from Allergopharma, during the conduct of the study; Dr. Neubauer reports grants from European Union, during the conduct of the study; Dr. Klysner reports personal fees from Allergopharma, during the conduct of the study; Dr. van Ree reports grants from European Commission, grants from Dutch Science Foundation, grants from Netherlands Lung Fund, outside the submitted work; and Consultancy for HAL Allergy BV, Leiden, The Netherlands; Dr. Vieths reports personal fees from Food Allergy Resource and Research Program, Lincoln, NE USA, personal fees from Medical University of Vienna, Austria, grants from Monsanto Company, personal fees from American Academy of Asthma, Allergy and Immunology, personal fees from Deutsche Dermatologische Gesellschaft, personal fees from Westdeutsche Arbeitsgemeinschaft fu¨r pa¨diatrische Pneumologie und Allergologie e.V., Ko¨ln, Germany, personal fees from Gesellschaft fu¨r pa¨diatrische Allergologie und Umweltme-dizin, personal fees from Ä rzteverband Deutscher Allergologen, personal fees from Swiss Society for Allergy and Immunology, personal fees from Schattauer Allergologie Handbuch, personal fees from Elsevier Nahrungsmittelallergien und Intoleranzen, personal fees from Karger Food Allergy: Molecular Basis and Clinical Practice, non-financial support from German Research Foundation, non-financial support from Federal Institute for Risk Assessment, non-financial support from European Directorate for the Quality of Medicines and Health Care, non-financial support from European Academy of Allergy and Clinical Immunology, non-financial support from Deutscher Allergieund Asthmabund, non-financial support from Association Mone´gasque pour le Perfectionnement des Connaissances des Me´dicins, non-financial support from Federal Office of Consumer Protection and Food Safety, non-financial support from German Chemical Society (GDCh), non-financial support from AKM Allergiekongress, non-financial support from International Union of Immunological Societies, outside the submitted work; Dr. Fereira reports personal fees from AllergenOnline Database, from Indoor Biotechnologies, from SIAF, Davos, and from HAL Allergy, outside the submitted work. The other authors have no conflicts of interest to disclose.

Author contributions
MH, ANa, HK, MT, MS, and PB performed experimental and preparative work; MH, ANa, HK, MT, MS, PB, and FF evaluated data; SK, KHB, and SV provided material; ANe, SK, and FF provided experimental/instrumental infrastructure; MH, RvR, KHB, SV, and FF conceived the study; MH, ANa, HK, and FF wrote and edited the manuscript; MH, RvR, SV, SK, KHB, ANe, and FF provided funding.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Data S1. Methods in greater detail. Figure S1. Individual results of basophil activation test. Table S1. Quality assessment of six production batches of rPhl p 5.0109. Table S2. Peak list of MS-based peptide mapping upon tryptic digestion. Table S3. T-cell reactivity of PP5ar06007 vs nPhl p 5 and recombinant batch PP5ar10015.