Comprehensive Profiling of Protein Particles

Example 1

1.1 Detection of Particles

On evaluation of vials in a process performance qualification (PPQ) lot (lot no. 1) comprising mAb1 formulation, about 264 vials (˜4.5%) of the PPQ lot had visible particles. Since, the acceptable limit for the PPQ lot is 0.7%, the lot was rejected.

The PPQ lot was evaluated for the presence of polysorbate level, which was found to be normal.

A particle (particle 1) from the PPQ lot was captured on a grey filter (See FIG. 9).

1.2 Isolation of the Particle

Excisions were made on the filter paper: (a) cut area 1—a 4 mm×4 mm area where the particle 1 was loaded, (b) cut area 2—a 4 mm×4 mm area where the protein solution was loaded, and (c) cut are 3—a 4 mm×4 mm area where no protein was loaded (See FIG. 10). The cut areas/samples were dissolved in 8 M urea.

The samples were incubated with 8 M urea and 5 mM dithiothreitol (DTT) at 50° C. for 30 minutes, followed by incubation with 10 mM indole-3-acetic acid (IAA) at RT for 30 minutes. The incubate samples were digested with 1 μg of rLysC for 1 hour under denaturing condition, followed by dilution with Tris-HCl. To this mixture, 1 μg of trypsin was added and the digestion was continued for 3 hours. The digestion mixture was then acidified with 20% formic acid (FA) followed by nanoLC-MS/MS analysis.

1.3 Peptide Analysis

To find out the composition of particle 1, peptide analysis of particle 1 was carried out using nanoLC-MS/MS Thermo EASY-nLC coupled to Thermo QExactive HF mass spectrometer. The identity of the proteins present, the Proteome Discoverer v. 1.4 software using the Sequest mode was used by searching mass spectral data against a CHO protein sequence database. Peptide quality scores were derived by processing against a decoy database using the Peptide Validator node within Proteome Discoverer that calculates the probability that the search algorithm incorrectly included a peptide in a sample. The false discovery rate (FDR), or the false positive rate, is a statistical value that estimates the number of false positive identifications among all identifications found by a peptide identification search. Peptides assessed with less than 5% FDR (medium and high confidence peptides) were retained, and those assessed with less than a 1% FDR (high confidence) threshold noted as high confidence. A minimum of two medium or high confidence (passing) peptides per protein were required to positively identify each HCP. The MS/MS spectra of peptides for those proteins which were identified by only one high confidence peptide were manually examined for coverage of three or more consecutive b- and y-ions and low number of abundant extraneous ions to determine their acceptance.

All the samples were analyzed for the abundance of mAb1. The cut area 1 (comprising particle 1) showed a high abundance of the mAb1 protein, followed by some presence of mAb1 in cut area 2 (protein solution without particle 1), and cut area 3 showed a low/absence of the mAb1 protein (See FIG. 11).

A host-cell protein other than mAb1-TIMP1 was found to be present in particle 1 (Table 1). However, TIMP1 was not present in the protein solution without particle 1 indicating that it was enriched in the particle.

Example 2

Four additional particles (particles 2-5) were identified in the PPQ lot (lot no. 1) containing the formulation from mAb1. These particles were captured and analyzed.

2.1 Isolation of the Particle

Each detected particle was placed on an individual Rap ID gold-coated polycarbonate membrane with a pore size of 5 μM and 40/20 nm coating as shown in FIG. 12.

Excisions were made on the membranes to isolate the particle from the protein solution, as illustrated in example 1.

2.2 Raman Analysis

Raman spectroscopy was employed to identify constituents of the particles. This method uses inelastic light scattering to generate an energy spectrum unique to each molecule, which is then compared to a reference library containing the fingerprints of various chemical structures. The Raman analysis confirmed that all four particles (particle 2, 3, 4 and 5) comprised protein.

A dark field illumination showed that the particles surrounding areas has no or very minimal protein signals, which ensures that cutting particles form the filter did not induce contamination for MS analysis (See FIG. 13).

2.3 Peptide Analysis

To find out the composition of the particles, the peptide analysis of the particle was carried out using nanoLC-MS/MS as illustrated in example 1. Table 2 lists the host-cell proteins identified in particle 1 through 5, along with a drug solution lot (DS), all of which comprise mAb1. The host-cell proteins were screened using the regular nanoLC-MS/MS and confirmed using targeted nanoLC-MS/MS. Particles 3 and 4 were smaller in size and had about 10-50 times less protein amount that particle 1 and particle 5, which likely explains the result of less host-cell proteins being identified.

The study confirmed the presence of host-cell proteins in the particle isolated rejected mAb1 lot no. 1.

Example 3

To study if visible particles were enriched with host-cell proteins in other lots comprising mAb1 formulation, particles in a clinical lot comprising mAb1 formulation (lot no. 2) and two other PPQ lots comprising mAb1 formulation (lot nos. 3 and 4) lots were evaluated.

The isolation of particles, Raman analysis and peptide analysis was performed as illustrated in Example 2.

Table 3 lists the host-cell proteins identified in particles from lot no. 1 (a PPQ lot), lot no. 2 (a clinical lot), lot no. 3 (a PPQ lot), and lot no. 4 (a PPQ lot). The host-cell proteins were screened using the regular nanoLC-MS/MS and confirmed using targeted nanoLC-MS/MS. Some host-cell proteins were not identified in the initial screening but were identified using confirmatory assays. Particles in lot no. 2 (clinical lot) were smaller in size and had about 15-30 times less protein amount than particles from lot no. 4 (PPQ lot), which likely explains the result of less host-cell proteins being identified. Comparing the host-cell proteins identified to be present in the particles from these lots 2-4 suggests that the proteins identified in the visible particles isolated from the lot no. 1 are comparable.

Example 4

To study if drug solution of mAb1 which could make up the PPQ lots is enriched with host-cell proteins, drug solution lots (lot nos. 5-12) were evaluated.

4.1 PLBD2 Levels

The drug solution was tested for PLBD2 using LC-MS MRM method using Hamster putative phospholipase B-like 2 (PLBD2) ELISA kit, catalog CSB-EL018125Ha from CUSABIO. This kit claims to provide quantitative determination of hamster putative phospholipase B-like (PLBD2) concentrations. A standard curve was generated showing detection of the hamster PLBL2 standard included in the kit over the range of 0.12-8 ng/ml. Drug solution from lot no.s 5-12 demonstrated PLBD2 level below the lower limit of quantitation (1 ppm i.e., 1 ng of PLBD in 1 mg mAb1)

Thus, drug solution in all of the above lots showed absence of abnormal PLBD2 levels.

4.2 HCP Profiling Using Direct Peptide Mapping

To find out the composition of the drug solution, peptide analysis was carried out using nanoLC-MS/MS as illustrated in example 1. The host-cell proteins were screened using the regular nanoLC-MS/MS and confirmed using targeted nanoLC-MS/MS. No host-cell proteins were identified in the any of the lots tested (lot nos. 5-12) (Table 4). TIMP1 and CCL13 were not identified in the initial screening of all the lots but were found to be present in the all the lots using more sensitive confirmatory assays. The relative abundance of the host cell proteins TIMP1 and CCL13 in drug solutions from lots 5-12 were plotted as normalized to relative abundance in lot no. 5 (See FIGS. 14 and 15).

Among the eight lots, lot no. 9 and lot no. 11 comprised drug solutions used to prepare the PPQ lots 1 and 3, respectively. The host cell proteins in the drug solutions in lot nos. 9 and 11 compared to the PPQ lots that they are used for suggests that there was not evaluated levels of host cell proteins observed in the lot that generated the particles compared to other drug substance lots.

4.3 HCP Profiling Using Anti-HCP Immunopurification (IP) Enrichment Followed by Peptide Mapping

To further evaluate the host cell proteins present in the drug solution lots, a more sensitive anti-HCP immunopurification was carried out. A pool of biotinylated anti-HCP antibodies was used to pull-down and enriched HCPs by removing the therapeutic proteins and other components. Following the enrichment, the HCPs were then digested by enzyme and injected on NanoLC-MS for HCP identification and quantification. The enrichment allows us to more sensitively identify and quantify HCPs.

Table 5 lists the host cell proteins which were identified by performing HCP profiling using anti-HCP immunopurification enrichment.

Comparing the host cell proteins identified in lot no. 1 (PPQ lot, Table 5) to lot no. 9 (drug solution lot used to prepare the lot no. 1, Table 3), it can be concluded that no unique host cell protein is associated with the lot no. 1. This suggests that the host cell protein profile in the PPQ lot was similar to the host cell protein profile in the drug solution lot.

Example 5

The host-cell proteins could be enriched as a consequence of aggregate formation. This was studied by comparing the host-cell proteins in drug solution lot (lot no. 6) with lots enriched with high molecular weight species of mAb1 as represented in table 6 below.

The isolation of particles and Raman analysis for lots 13-15 was performed as illustrated in Example 2.

5.1 Host Cell Proteins Identified Using Direct Peptide Mapping

The samples were dried down using SpeedVac, and then dissolved by 8M Urea and 10 mM TCEP-HCl, and denatured at 50° C. for 30 minutes, followed by incubation with 10 mM indole-3-acetic acid (IAA) at RT for 30 minutes. The incubate samples were digested with 1 of rLysC for 1 hour under denaturing condition, followed by dilution with Tris-HCl. To this mixture, 1 μg of trypsin was added and the digestion was continued for 3 hours. The digestion mixture was then acidified with 20% formic acid (FA) followed by nanoLC-MS/MS analysis. Six host cell proteins were identified in the lot with enriched HMW mAb1 species, but not in the drug solution lot (Table 7).

5.2 Relative Quantitation of Host Cell Proteins Using Targeted LC-MS/NIS

The identified peptide/protein from Step 5.1 above was used to create a mass list of identified peptides in order for the mass spectrometer to look for these specific peptides to fragment to further confirm and quantify the HCPs. In doing so, the detection sensitivity was increased due to the targeted screening. Seven host cell proteins were identified in the lots with enriched HMW mAb1 species (lot nos. 13-15). The amount of the host cell proteins identified in the lots with enriched HMW mAb1 species was about 2-60 fold higher than the amount of the host cell proteins in the drug solution lot (lot no. 6). FIG. 16 shows a chart of relative quantitation of host cell proteins in the lots with enriched UMW mAb1 species.

5.3 Host Cell Proteins Identified Using Anti-HCP IP Enrichment Followed by Peptide Mapping

A pool of biotinylated anti-HCP antibodies was used to pull-down and enriched HCPs by removing the therapeutic proteins and other components. Following the enrichment, the HCPs were then digested by enzyme and injected on NanoLC-MS for HCP identification and quantification. Thirty one host cell proteins were identified in the lots with enriched HMW mAb1 species (lot nos. 13-15). Out of thirty one host cell proteins, the drug solution lot only showed the presence of two host cell proteins (Table 8).

The above results show a higher amount and enrichment of host cell proteins in lots with BMW mAb1 species than mAb1 drug solution lot. This suggests that host cell proteins are enriched BMW mAb1 species.

Example 6

Drug product comprising mAb1 was also tested for the presence of visible particles and host cell proteins, in addition to testing drug solution of mAb1, lot comprising HMW species of mAb1 and PPQ lots of mAb1.

6.1 Detection of Particles

Two vials (vial #1 and vial #2) comprising the drug product—containing a mAb1 was inspected manually to detect the presence of a particle(s).

6.2 Isolation of the Particle

For each of the vial, the detected particle was placed on an individual (Rap ID) gold-coated polycarbonate membrane with a pore size of 5 μM and 40/20 nm coating as shown in FIG. 17.

Excisions were made on the membranes: (a) particle sample—a 3 mm diameter circular area with the particle(s) was excised and (b) negative control—a 3 mm diameter circular area without the particle was excised (FIG. 18). The particle sample and negative sample were dissolved in 8 M urea. Three particle samples and two negative controls were isolated from the membranes for the two vials as represented in Table 9.

6.3 Raman Analysis

Raman spectroscopy was performed on the particles as illustrated in example 2. The Raman spectra of one of the particle 1 as recited in table 9 confirmed that the particles were proteinaceous (See FIG. 19). The green trace as seen in FIG. 19 is the protein reference spectrum and the red trace as seen in FIG. 19 is the particle sample spectrum.

6.4 Peptide Analysis

Peptide analysis was performed on the particles as illustrated in example 2.

All the particle samples and negative samples were analyzed for the abundance of mAb1. The particle samples showed a high abundance of the mAb1 protein and negative samples showed absence of the mAb1 protein (See FIG. 20). The abundance peaks were estimated based on the averaged peak areas of the top 2 most abundant peptides in mAb1.

Proteins other than mAb1 obtained from the peptide analysis are represented in table 10. The mAb1 abundance in the particle samples and negative samples was normalized to 1E6 to relatively quantify the other proteins in ppm in Table 10. The term ND in table 10 is to represent that the protein was not detected.

The presence of protein in the particle samples and negative samples is shown in table 11.

The peptide analysis revealed the presence of host-cell proteins in the particle isolated from drug product containing mAb1.

Example 7

Detection of sub-visible particles was also performed using a Formulated Drug Substance (FDS) comprising mAb1.

In order to isolate sub-visible particles, 1.5 ml of the FDS solution was pipetted on a gold-coated polycarbonate membrane with a pore size of 5 μM and 40/20 nm coating. This was then washed with milli-Q water. A particle free negative control was established for comparison by initially pipetting 1.5 ml of the FDS solution on a 0.2 μM filter (Millipore) and the flow-through was collected. The flow-through was then pipetted on a gold-coated polycarbonate membrane with a pore size of 5 μM and 40/20 nm coating and washed with milli-Q water.

Raman analysis and peptide analysis was performed as described in Example 2. The particles comprising protein were identified using the Raman spectroscopy. The dark-filled images of the particles comprising proteins identified by the recited method is shown in FIG. 21. The peptide analysis of the sub-visible particles demonstrated the presence of a host cell protein—Metalloproteinase inhibitor 1 along with mAb1, whereas the particle free negative control showed only the presence of mAb1. The absence of the host protein Metalloproteinase inhibitor 1 in the particle free negative control suggests that the host cell proteins were enriched in the sub-visible particles only (Table 12).

Example 8

To evaluate the presence of host cell proteins in bioprocess, harvested cell culture fluid (HCCF) from Preclinical Manufacturing and Process Development (PMPD) for mAb2 was tested. The HCCF for mAb2 showed presence of particles after sterile filtration and on freeze-thaw cycle. During the HCCF sterilizing filtration step at Industrial Operations and Product Supply (TOPS), high back pressure was experiences, indicating some non-soluble materials were clogging the filter.

The isolation of particles, Raman analysis and peptide analysis was performed as illustrated in Example 2. FIG. 22 shows a dark filled image and a Raman image of the particles. Other than mAb2, 743 host cell proteins were identified in one of the particles. The top ten HCPs (based on sequence coverage) were L-lactate dehydrogenase (UniProt Accession: Q06BU8), Actin, cytoplasmic (UniProt Accession: G3GVD0), Transketolase (UniProt Accession: G3GUU5), Rab GDP dissociation inhibitor beta (UniProt Accession: G3GR73), Glyceraldehyde-3-phosphate dehydrogenase (UniProt Accession: P17244), V-type proton ATPase catalytic subunit A (UniProt Accession: G3H066), Transgelin (UniProt Accession: G3H7Z2), Glutathione S-transferase Mu 5 (UniProt Accession: G3ILF1), Chloride intracellular channel protein 4 (UniProt Accession: G3HMU4), and Leukotriene A-4 hydrolase (UniProt Accession: G3HBI9).