each search algorithm and maximize the number of high confidence peptide identifications. High confidence peptides were clustered by homology to parent proteins, and the number of matched spectra per peptide and protein were recorded as a qualitative measure of relative abundance.

For independent confirmation of the peptide and protein identifications, the PEAKS Studio software platform (Bioinformatics Solutions, v7.5) was used to perform an orthogonal de novo sequencing-assisted database search.

FIGURE 1: LC-MS/MS Workflow Overview for HCP Identification and Relative Quantitation

HCP QUANTITATION: HCPs of interest were selected for assay development (e.g. HCPs with known or potential product modifying capabilities, efficient or inefficient clearance profiles, etc.). The intended purpose of the assay was for the multiplexed monitoring of the HCP levels in additional batches of in-process and DS/DP samples, using absolute quantitation.

Multiplexed LC-MRM/MS Assay Development (Absolute Quantitation mode): Bioinformatic analysis of the target protein sequences and empirical evidence from the HCP identification study were used to identify ≤ 5 surrogate peptides per target protein. The corresponding synthetic peptide standards were obtained in ‘light’ non-labeled, and ‘heavy’ stable-isotope labeled (SIL) form. The standards were HPLC-purified and quantified for accurate-weight, for use in assay development and as reference material. LC-MRM/MS assay conditions were developed using the synthetic peptide standards and optimized in sample matrix, using a nanoAcquity UPLC (Waters) coupled to a QTRAP (AB Sciex) mass spectrometer. Optimized parameters included the chromatography and retention time, the collision energy for peptide fragmentation, the best peptide fragments for monitoring each target peptide, and various assay performance characteristics were evaluated (e.g. linearity, precision and accuracy, etc.).

FIGURE 2: LC-MRM/MS Assay Development

M1030-06-041

PURPOSE • Biological drugs produced from host cell expression systems inevitably

contain host cell protein (HCP) impurities whose identity, presence and levels may determine whether or not the drug is accepted by regulatory agencies.

• Conventional HCP analysis relies on immunoassays based on polyclonal antibodies raised against the host cell system, which provide an aggregate measurement of total HCP. However, deeper characterization of HCPs using orthogonal approaches is increasingly expected by regulatory authorities.

• MS-based proteomic platforms are routinely used to analyze complex matrices with a wide dynamic range of protein abundance. Non-targeted MS enables the comprehensive identification and profiling of proteins. Targeted MS is used for the relative or absolute quantitation of specific proteins of interest.

• The purpose of this study was to use MS-based assays to obtain deeper characterization of the residual HCP content in in-process, Drug Substance or Drug Product (DS/DP) samples.

• APPLICATIONS: • Monitoring of purification process • Demonstration of HCP clearance • Comparison of culture media, process improvements and changes • Evaluation of batch reproducibility and scale up • Characterization of mock immunogen content for immunoassay

CONCLUSION MS-based HCP assays enable direct identification and quantitation of total and individual HCP impurities, with an assay sensitivity in the ~1 to 15 ppm range (DS/DP dependent). Identification and accurate quantitation of HCPs provides critical information for biosimilar vs originator comparisons, process development, optimization and monitoring, as well as for patient and product risk assessments.

OBJECTIVES • HCP IDENTIFICATION: To use non-targeted MS to comprehensively assay HCP

content (identification and relative quantitation) in in-process samples and in formulated bulk DS.

• HCP QUANTITATION: To develop a targeted MS-based assay for HCPs of interest (e.g. problematic HCPs) based on their protein sequence, and to perform sample analysis in absolute quantitation mode to track the HCP levels in ppm, at key clearance steps and in the DP.

METHODSHCP IDENTIFICATION: For each study sample (harvested cell culture fluid, in-process samples, DS), a starting amount of up to 500 μg was digested with trypsin enzyme to generate peptides for analysis. For selected samples, an additional peptide prefractionation step was performed using SCX and HPRP to enhance the detection and coverage of low-abundance HCPs. Peptides were analyzed on a nanoAcquity UPLC (Waters) coupled to a high-resolution accurate mass Q Exactive mass spectrometer (Thermo). The equivalent of 2.5 μg protein was used per injection. The acquisition method alternated between LC-MS for peptide ion detection and LC-MS/MS to obtain sequence information from peptide fragments. A process-specific protein sequence database was built based on all sequences known or assumed to be present in the study samples (e.g DS, host proteome, process-related protein additives, etc). The acquired MS/MS data were searched against the custom database to match the acquired peptide sequence data to corresponding database sequences, and any expected or unexpected post-translational protein modifications were also identified. Three database search engines (Mascot, OMSSA, and X!Tandem) were used to leverage the strengths of

UPS2 (Protein_ID)

Spiked Level (ppm)

Without peptide fractionation

With peptide fractionation

ALBU_HUMAN 626.01 Yes YesCAH2_HUMAN 274.67 Yes YesCAH1_HUMAN 271.11 Yes YesLEP_HUMAN 152.43 Yes YesHBB_HUMAN 149.69 Yes YesHBA_HUMAN 142.70 Yes YesUBIQ_HUMAN 99.97 Yes YesCO5_HUMAN 80.78 Yes Yes

CATA_HUMAN 56.25 Yes YesSUMO1_HUMAN 36.62 Yes YesNQO1_HUMAN 29.00 Yes YesPRDX1_HUMAN 20.73 Yes YesPPIA_HUMAN 19.03 Yes YesMYG_HUMAN 16.09 Yes YesCYB5_HUMAN 15.12 Yes YesEGF_HUMAN 5.99 No Yes

SYHC_HUMAN 5.49 No YesKCRM_HUMAN 4.07 No YesNQO2_HUMAN 2.44 No YesRETBP_HUMAN 1.99 No NoUBC9_HUMAN 1.70 No YesLYSC_HUMAN 1.39 No Yes

LALBA_HUMAN 1.33 No YesNEDD8_HUMAN 0.86 No Yes

Diges�on of proteins to pep�des(Trypsin enzyme)

Extensive pep�de frac�ona�on (OPTIONAL: 2 Orthogonal methods X 8 frac�ons each)

Harvest In-process step 1

In-process step 2 … In-process

step N DS

• Process blank

• UPS2 (Sigma), a proteomics dynamic range standard set comprised of 6 mixtures of 8

Process Quality Controls (PQC)

• Monoclonal Ab• Recombinant protein• Fusion proteins• Pep�de drugs••

VaccinesAdenovirus

LC-MS/MS analysis(High resolu�on Q Exac�ve mass spectrometer)

1. Electronic Data Report: List of iden�fied HCPs, i l di t l t f l � �t �

Iden�fica�on of HCP (Match detected pep�des to corresponding sequences in process-specific HCP database)

pproteins with a dynamic range of 5 orders of magnitude (50 pmoles to 500 amoles)

• UPS2 spiked with DS

Monitoring of general MS instrument sensi�vity, reproducibility and stability

Assessment of impact of DS on sensi�vity

Rou�nely obtain LOD range 1-10 ppm Custom Database Client process-specific sequences

DS sequencesHost proteome (ex: CHO, E.coli, human, yeast)Process-specific addi�ves

Caprion-specific sequencesPQC protein mixture and associated addi�vesCommon laboratory contaminants

including spectral counts for rela�ve quan�ta�on. 2. Wri�en Report: Summary study design, methods,

results discussion and conclusions.

•

•

• Prioritize list of HCPs • Select ≤5 surrogate

peptides per HCP

MSAIQAAWPSGTECIAKYNFHGTAEQDLPFCKGDVLTIVAVTKDPNWYKAKNKVGREGIIPANYVQKREGVKAGTKLSLMPWFHGKITREQAERLLYPPETGLFLVRESTNYPGDYTLCVSCDGKVEHYRIMYHASKLSIDEEVYFENLKMQLVEHYTSDADGLCTRLIKPKVMEGTVAAQDEFYRSGWALNMKELKLLQTIGKGEFGDVMLGDYRGNKVAVKCIKNDATA…

• Use synthetic isotope-labeled peptides to develop assay conditions

inte

nsity

RT

• Final optimized assay monitors 2 fragments (transitions) per peptide

HCP protein sequence, trypsin digestion

Peptide detection (MS) Fragmentation Fragment detection (MS/MS)

14.92

18.68

109.19

336.02

966.77Prot A

In-proc1

In-proc2

In-proc3

From Bulk

0 200 400 600 800 1000 0% 20% 40% 60% 80% 100%

[B] HCP Composition & Abundance

79 HCP*

13 HCP

6 HCP

[A] Total HCP

* Top 50 of 79 HCP shown

2 HCP

2 HCP

(ppm) (Relative abundance)

No light pep�de spike Spike light pep�des Spike light pep�des

Diges�on

Study samples(Individual DS)

Calibra�on curves(BSA or pooled DS)

QC samples(pooled DS)

No light pep�de spike (endogenous only)

Spike light pep�des: (Low, Mid, High)

Desalt / MRM analysis

Endogenous levels(Concentration back-calculated

using calibration curve)

Spike light pep�des:(≥7 non-zero standards)

Calibra�on curve(Peak area ratio as a function of

nominal concentration)

Precision and Accuracy (Concentration back-calculated using

calibration curve)

Spike SIL peptides (�xed concentration)

info@caprion.com

Absolute Quantitation of HCPs in Study Samples: Study samples, QC and calibration standards were processed and analyzed in the same analytical run using the workflow in Figure 3. A starting amount of 25 μg protein was used per sample, and the equivalent of 4 μg protein was injected per LC-MRM/MS run. MRM peak signals were integrated and the ratio of the ‘light’ transitions (from the endogenous peptides or non-labeled standards) over the ‘heavy’ transitions (from the SIL internal standards) were calculated. The endogenous study sample and QC sample concentrations were back-calculated from the calibration curve, and the levels of each HCP in each study sample were reported in ppm.

FIGURE 3. Workflow for Absolute Quantitation of ≤ 15 HCP Using Multiplexed LC-MRM/MS (Relative Quantitation Mode: ≤325 proteins)

RESULTS

Chart 1: Sensitivity of LC-MS/MS for HCP Identification in the Presence of DS

A test sample mimicking a DS with protein impurities in the ~1 ppm to 600 ppm range was analyzed in triplicate, using LC-MS/MS either with or without peptide prefractionation. The test sample consisted of SILu Lite Mab Universal Standard antibody (Sigma) spiked with the UPS2 dynamic range protein standard set (Sigma), using a ratio of 500:1. Results show a protein-dependent assay sensitivity of ~1 to 15 ppm for the different mock HCP impurities (UPS2 proteins).

FIGURE 4: Identification and Relative Quantitation of Total and Individual HCP X Using LC-MS/MS

Comparative LC-MS/MS analysis of in-process samples and formulated bulk DS. [A] Direct HCP measurements shown in aggregate, in ppm. Overall clearance is shown, similar to information provided by immunoassays. [B] Identification and relative abundance of the individual constituent HCPs, including reported problematic HCPs such as PLBL2, pyruvate kinase, etc. Clearance of individual HCPs is shown, as well as the residual individual HCP levels in the formulated bulk DS.

FIGURE 5: Absolute Quantitation of HCP of Using LC-MRM/MS

[A] Calibration curve and representative chromatograms. [B] The two peptides per protein used for monitoring provided consistent results and enabled the precise and accurate determination of HCP X concentrations in ppm in the final DS.

[A]

[A] Total HCP

[B]

[B] HCP Composition & Abundance

Mass Spectrometry Platforms for High Coverage Process-Specific Host Cell Protein Identification and QuantitationChristina Bell, John Babetas, Laetitia Cortes, Vanessa Diniz Atayde,

Rudolf Guilbaud, Stéphane Parent, Michael Schirm and Lorella Di Donato

Caprion Biosciences Inc.

HCP X