Ejvind Mørtz

Gene therapy is arguably the fastest developing therapeutic modality in biopharma. Following hot on the heels of classical biologics such as monoclonal antibodies (mAbs), an increasing number of gene therapies are entering clinical trials and receiving marketing approval. The development of gene therapies has taken longer than mAbs because of various reasons, not least of which were early setbacks and the need for more sophisticated delivery systems. While biologics such as mAbs can be administered directly to patients for therapeutic action, gene therapies tend to require delivery systems to safely transport the corrective nucleic acids to the required site of action. These delivery systems are most commonly modified viral vectors. The inclusion of vectors in the final drug product can introduce greater complexity to the processes required to manufacture gene therapies compared with those typically used for classical biologics like mAbs.

The biomanufacture of both mAbs and gene therapies requires the use of cellular expression systems, which produce the protein of interest for the final drug product. For mAbs, these systems can originate from various species, such as Chinese hamster ovary (CHO) or Escherichia coli (E. coli) cells. For gene therapies, it is much more common to use human cells, such as human embryonic kidney 293 (HEK-293), HeLa, or A549 lung epithelial carcinoma (A549) cells. To derive the final drug product, the product from the cells must be purified, using a series of steps where impurities are detected and monitored until their concentrations are low enough in the final drug product to meet regulatory requirements. An important class of impurities are host cell proteins (HCPs) (i.e., proteins from the cells used to express the protein of interest). HCPs need to be carefully controlled throughout the biomanufacturing process as a critical quality attribute (CQA). This is because HCPs can significantly affect the quality, efficacy, and safety of the gene therapy product, for example, by inducing or enhancing immunogenicity. Apart from process-related impurities like HCPs, product-related impurities like that derived from viral vectors also need to be carefully controlled.

Historically, the preferred method for analyzing HCPs and other impurities has been ligand-binding assays (LBAs), due to their speed, sensitivity, and ease of use. However, despite LBAs-namely enzyme-linked immunosorbent assays (ELISAs)-being the main regulatory compliant method in current use, there has been a steady increase in industry regulators requesting that submissions include data detailing HCP quantification obtained with methods orthogonal to ELISA. The approach that is rapidly emerging as the method of choice for many biopharmaceutical companies is liquid chromatography (LC) coupled with mass spectrometry (MS). In response to this growing need, Alphalyse and SCIEX have developed new LC-MS methods for HCP characterization of gene therapies and other new complex biologics (1, 2). These improved methods and workflows address key shortcomings posed by ELISAs in the analysis of HCPs for gene therapies. There are several main issues that are repeatedly encountered by industry players with the application of ELISAs for HCP analysis during the development and manufacture of gene therapies.

The common issue that frequently arises is with the specificity of the commercially available ELISAs. During the cell-based manufacture of various types of gene therapies, there exist numerous sources of potential residual contamination. Mammalian expression systems commonly used to generate gene therapy products, such as those involving HEK-293 cells, introduce a multitude of impurities that are secreted along with the product of interest into the cell culture medium. These impurities can include nucleic acids (i.e., DNA, RNA) and lipids, along with HCPs and other cellular material. Moreover, the media can also include other proteins that have been added to improve and expand its function, such as, for example, bovine serum albumin (BSA)-a common nutrient supplement for cultured cells-and benzonase-a nuclease enzyme that digests and removes unwanted DNA fragments. The biomanufacture of viral vectors as part of the final drug product is another source of impurity. Empty and partially filled viral capsids are product-related impurities that need to be removed from the final drug product because they can have an impact on the safety, efficacy, and immunogenicity of the therapy (3). Furthermore, the product of interest is often engineered to be human. The bioprocess of gene therapies can thus include impurities originating from multiple, diverse species, such as human from the HEK-293 cells, bovine from the BSA, viral particles from the vector system, and bacterial debris from the benzonase produced by E. coli. These contaminants from so many different species can be very difficult to detect using a single ELISA, particularly commercially available ELISAs.

Beyond the heterogeneity, the sheer number of different proteins and other impurities poses a challenge. There are potentially thousands of protein species that would need detection. This tends to be especially true for the production of gene therapies that involve mammalian expression systems and more complex biomanufacturing steps, such as the incorporation of a viral vector, than a standard biologic. Moreover, the immunogenicity of some of the proteins, particularly human ones, could be very low, meaning that it is difficult to raise detection antibodies that provide good coverage of the spectrum of protein impurities present in a gene therapy product. Even with polyclonal antibodies, it would require the use of multiple, different types of LBA to detect anywhere near the thousands of different HCPs that may be present. A single LBA is simply insufficiently specific to detect so many proteins of different types from numerous diverse species. Another issue is that of detection sensitivity. The concentrations of the various impurities can also vary greatly, with some being present in high abundance while others are present in trace amounts. These trace impurities can be as potently important for the quality, efficacy, safety, and immunogenicity of the final product as the contaminants present in more abundant amounts. Moreover, against the background of a high-volume biological sample containing high levels of the product of interest, the detection system would need to be very sensitive and have a wide dynamic range. Again, any single commercially available ELISA would not provide the sensitivity and dynamic range needed for HCP profiling of gene therapy products.

To address these issues with commercially available ELISAs, process-specific ELISAs can be developed. However, the development of these bespoke LBAs typically takes one-and-a-half to two years. This can be too long for the development of novel gene therapies because the development of this modality tends to be fast-tracked by regulatory authorities to treat serious, life-threatening conditions. Gene therapies also tend to be one-shot-and-done curative options, the first therapy to successfully reach the market tends to be the only treatment option patients need. As such, time is of the essence when it comes to the expensive gamble of developing gene therapies.

LC–MS/MS is becoming the method of choice orthogonal to ELISAs for HCP analysis because it not only offers the ability to measure thousands of proteins practically simultaneously, it can also quantify and potentially identify these proteins. It is also a generic method that can be applied easily to a broad range of gene therapy products as well as to other biologics and new modalities. These analytical advantages make LC-MS/MS assays a beneficial choice for HCP analysis. As the need for HCP analysis using orthogonal techniques grew, Alphalyse began refining a similar assay initially developed and implemented for the analysis of vaccine products (4). This developed into what is now a detailed HCP detection analysis system. The assay involves LC to separate out the analytes in the sample before quantification using tandem mass spectrometry (MS/MS). The high-performance LC is run at a microflow gradient, which eliminates any carry over analytes as well as leverages the sensitivity phenomenon unique to low-flow gradients (2). The LC-MS/MS analytical information is then comprehensively collected using data-independent acquisition (DIA) with SWATH (7,8).

MS with SWATH acquisition occurs where a TripleTOF 6600 mass spectrometer divides the mass range into small mass windows and performs MS/MS analysis on all peptides in each window (see Figures 1 and 2) (2). As the size of the mass windows are dynamic, small windows can be used to scan areas with many peptides, and larger windows can be used to more quickly analyze areas with few peptides. Using SWATH acquisition results in low interference from the high amount of the drug substance protein because the data on the low abundance HCP peptides are acquired in data-independent mode. The proteins and peptides are identified through comparison with data stored in reference databases and ion libraries. As the SWATH acquisition records multi-dimensional data with comprehensive coverage of the HCPs present, the data can be re-analyzed repeatedly in the future. The digital records thus effectively immortalize information on each protein, facilitating the continuous improvement of the annotation of the HCP profiles alongside the updating of reference databases and ion libraries with more known compounds and their fragmentation patterns (2, 5, 6).

HCP profiling by LC-MS/MS can accelerate gene therapy development

LC–MS/MS with SWATH acquisition can help bring gene therapies to market faster, safer, and cheaper. In the development of the assay, we tested and assessed its capability to detect, quantify, and identify HCPs for the monitoring and control of purity, batch-consistency, and drug release testing. The assay is sensitive enough to detect proteins present at concentrations as low as 10 parts per million (ppm), with label-free absolute quantification (2). The assay is also able to measure proteins ranging in abundance from 10 ppm to 100,000 ppm, demonstrating a wide and linear dynamic range (2). The assay can also be used to successfully monitor changes in the quantity of individual HCPs through consecutive bioprocessing steps (1). Qualitative data on the physicochemical properties of each unique protein is also acquired, including the protein molecular weight (MW), hydrophobicity, and isoelectric point (pI) (1,2). This more detailed and holistic information is particularly useful for the development and optimization of biomanufacturing processes to ensure the clearance of specific HCPs. In terms of specificity and coverage, the assay demonstrated that it was to cover the full MW and pI range of the proteome of E. coli, a common gene therapy expression system and standard reference species (2). The assay based on microflow LC–MS/MS with SWATH acquisition method is highly precise (i.e., reproducible) and robust for HCP identification and quantification (2).

The additional information provided by LC–MS/MS over ELISAs on the absolute quantification and identification of the HCPs is particularly useful to facilitate the optimization of biomanufacturing processes. The comprehensive data on the HCPs enables the straightforward modification of existing processes to optimize the clearance of the HCPs and other impurities. Another popular application of the HCP detection analysis using this assay is for the evaluation of ELISAs. As ELISAs remain the main regulatory compliant method specified by regulatory authorities, the choice of which commercially available ELISAs to use for HCP detection can be supported by LC–MS/MS data. This can confirm that the specificity, sensitivity, and coverage of the ELISA is sufficient to meet regulatory standards. To further inform the selection of the ELISAs, LC–MS/MS analysis can and is used to characterize specific HCPs of concern. The new assay can be used for comparing biosimilar drug products with their originators, as well as for analyzing preclinical and clinical batches of gene therapy product (2).

Compared with the one-and-a-half to two years it usually takes to set up a new process-specific ELISA, this new LC–MS/MS assay is quick and simple to apply for each new biologic product. It typically takes six to eight weeks to optimize and establish the assay, from the start of the project to the receipt of a regulatory notice of verification. The samples can then be analyzed on an ongoing basis. The substantially shorter timeframe and improved utility of the data acquired indicates that the adoption of this LC–MS/MS assay with SWATH acquisition could accelerate the development, marketing approval, and commercial biomanufacturing of novel gene therapies. This acceleration will likely be further enhanced as advances are continually made to the assay.

Anticipating future regulatory requirements

The biopharma industry is shifting towards including HCP data from orthogonal methods to LBAs, primarily from MS assays. This trend is observed in the industry as companies increasingly request detailed HCP detection analysis using this new assay based on microflow LC–MS/MS with SWATH acquisition. It is expected that by preparing such detailed HCP data, which can continue to be better annotated over time, the chances of obtaining regulatory approval for clinical trials and marketing are improved. Moreover, it is an important component in the risk management strategies for many large companies investing in the costly and competitive development of new gene therapies. Implementing measures to minimize the need to re-run analyses later in the drug development cycle not only saves time, but could also prove more cost efficient.

“Identification and Absolute Quantification of Individual Host Cell Proteins by SWATH LC–MS,”

 presentation at the Well Characterized Biologics Conference (Rockville, MD, Oct. 24–26, 2018).
3. T. Li, Gao, et al., 

“Determination of Full, Partial, and Empty Capsid Ratios for Adeno-Associated Virus (AAV) Analysis,”

 SCIEX Technical Note, sciex.com, accessed April 2020.
4. S. Heissel, et al., 

“Host Cell Protein Analysis by Microflow-LC High Resolution SWATH-MS of Vaccine Samples Under Development,”

 poster presented at The American Society for Mass Spectrometry (ASMS) Annual Conference (San Antonio, TX, June 5–9, 2016).
5. SCIEX, “

SWATH Acquisition, Ensuring Nothing is Missed,”

 sciex.com, accessed April 2020.
6. C. Carapito, 

“Dual Data-Independent Acquisition Method using SWATH Acquisition–MS for Host Cell Proteins (HCP) Profiling and Absolute Quantification of Key Impurities during Bioprocess Development,”

 SCIEX Webinar, sciex.com, accessed April 2020.

Ejvind Mørtz, PhD, is co-founder and COO of Alphalyse. He has more than 20 years of experience from several biotech companies, including the development of protein analysis methods in research and chemistry, manufacturing, and control development of protein biologics, and business collaborations with contract manufacturing organizations/contract researche organization.

Articles

This method is expected to help bring gene therapies to market faster, safer, and cheaper.

A Novel Method for Host Cell Protein Analysis

	Gene therapy is arguably the fastest developing therapeutic modality in biopharma. Following hot on the heels of classical biologics such as monoclonal antibodies (mAbs), an increasing number of gene therapies are entering clinical trials and receiving marketing approval. The development of gene therapies has taken longer than mAbs because of various reasons, not least of which were early setbacks and the need for more sophisticated delivery systems. While biologics such as mAbs can be administered directly to patients for therapeutic action, gene therapies tend to require delivery systems to safely transport the corrective nucleic acids to the required site of action. These delivery systems are most commonly modified viral vectors. The inclusion of vectors in the final drug product can introduce greater complexity to the processes required to manufacture gene therapies compared with those typically used for classical biologics like mAbs.

	The biomanufacture of both mAbs and gene therapies requires the use of cellular expression systems, which produce the protein of interest for the final drug product. For mAbs, these systems can originate from various species, such as Chinese hamster ovary (CHO) or Escherichia coli (E. coli) cells. For gene therapies, it is much more common to use human cells, such as human embryonic kidney 293 (HEK-293), HeLa, or A549 lung epithelial carcinoma (A549) cells. To derive the final drug product, the product from the cells must be purified, using a series of steps where impurities are detected and monitored until their concentrations are low enough in the final drug product to meet regulatory requirements. An important class of impurities are host cell proteins (HCPs) (i.e., proteins from the cells used to express the protein of interest). HCPs need to be carefully controlled throughout the biomanufacturing process as a critical quality attribute (CQA). This is because HCPs can significantly affect the quality, efficacy, and safety of the gene therapy product, for example, by inducing or enhancing immunogenicity. Apart from process-related impurities like HCPs, product-related impurities like that derived from viral vectors also need to be carefully controlled.

	Historically, the preferred method for analyzing HCPs and other impurities has been ligand-binding assays (LBAs), due to their speed, sensitivity, and ease of use. However, despite LBAs-namely enzyme-linked immunosorbent assays (ELISAs)-being the main regulatory compliant method in current use, there has been a steady increase in industry regulators requesting that submissions include data detailing HCP quantification obtained with methods orthogonal to ELISA. The approach that is rapidly emerging as the method of choice for many biopharmaceutical companies is liquid chromatography (LC) coupled with mass spectrometry (MS). In response to this growing need, Alphalyse and SCIEX have developed new LC-MS methods for HCP characterization of gene therapies and other new complex biologics (1, 2). These improved methods and workflows address key shortcomings posed by ELISAs in the analysis of HCPs for gene therapies. There are several main issues that are repeatedly encountered by industry players with the application of ELISAs for HCP analysis during the development and manufacture of gene therapies.

	The common issue that frequently arises is with the specificity of the commercially available ELISAs. During the cell-based manufacture of various types of gene therapies, there exist numerous sources of potential residual contamination. Mammalian expression systems commonly used to generate gene therapy products, such as those involving HEK-293 cells, introduce a multitude of impurities that are secreted along with the product of interest into the cell culture medium. These impurities can include nucleic acids (i.e., DNA, RNA) and lipids, along with HCPs and other cellular material. Moreover, the media can also include other proteins that have been added to improve and expand its function, such as, for example, bovine serum albumin (BSA)-a common nutrient supplement for cultured cells-and benzonase-a nuclease enzyme that digests and removes unwanted DNA fragments. The biomanufacture of viral vectors as part of the final drug product is another source of impurity. Empty and partially filled viral capsids are product-related impurities that need to be removed from the final drug product because they can have an impact on the safety, efficacy, and immunogenicity of the therapy (3). Furthermore, the product of interest is often engineered to be human. The bioprocess of gene therapies can thus include impurities originating from multiple, diverse species, such as human from the HEK-293 cells, bovine from the BSA, viral particles from the vector system, and bacterial debris from the benzonase produced by E. coli. These contaminants from so many different species can be very difficult to detect using a single ELISA, particularly commercially available ELISAs.

	Beyond the heterogeneity, the sheer number of different proteins and other impurities poses a challenge. There are potentially thousands of protein species that would need detection. This tends to be especially true for the production of gene therapies that involve mammalian expression systems and more complex biomanufacturing steps, such as the incorporation of a viral vector, than a standard biologic. Moreover, the immunogenicity of some of the proteins, particularly human ones, could be very low, meaning that it is difficult to raise detection antibodies that provide good coverage of the spectrum of protein impurities present in a gene therapy product. Even with polyclonal antibodies, it would require the use of multiple, different types of LBA to detect anywhere near the thousands of different HCPs that may be present. A single LBA is simply insufficiently specific to detect so many proteins of different types from numerous diverse species. Another issue is that of detection sensitivity. The concentrations of the various impurities can also vary greatly, with some being present in high abundance while others are present in trace amounts. These trace impurities can be as potently important for the quality, efficacy, safety, and immunogenicity of the final product as the contaminants present in more abundant amounts. Moreover, against the background of a high-volume biological sample containing high levels of the product of interest, the detection system would need to be very sensitive and have a wide dynamic range. Again, any single commercially available ELISA would not provide the sensitivity and dynamic range needed for HCP profiling of gene therapy products.

	To address these issues with commercially available ELISAs, process-specific ELISAs can be developed. However, the development of these bespoke LBAs typically takes one-and-a-half to two years. This can be too long for the development of novel gene therapies because the development of this modality tends to be fast-tracked by regulatory authorities to treat serious, life-threatening conditions. Gene therapies also tend to be one-shot-and-done curative options, the first therapy to successfully reach the market tends to be the only treatment option patients need. As such, time is of the essence when it comes to the expensive gamble of developing gene therapies.

	LC–MS/MS is becoming the method of choice orthogonal to ELISAs for HCP analysis because it not only offers the ability to measure thousands of proteins practically simultaneously, it can also quantify and potentially identify these proteins. It is also a generic method that can be applied easily to a broad range of gene therapy products as well as to other biologics and new modalities. These analytical advantages make LC-MS/MS assays a beneficial choice for HCP analysis. As the need for HCP analysis using orthogonal techniques grew, Alphalyse began refining a similar assay initially developed and implemented for the analysis of vaccine products (4). This developed into what is now a detailed HCP detection analysis system. The assay involves LC to separate out the analytes in the sample before quantification using tandem mass spectrometry (MS/MS). The high-performance LC is run at a microflow gradient, which eliminates any carry over analytes as well as leverages the sensitivity phenomenon unique to low-flow gradients (2). The LC-MS/MS analytical information is then comprehensively collected using data-independent acquisition (DIA) with SWATH (7,8).

	MS with SWATH acquisition occurs where a TripleTOF 6600 mass spectrometer divides the mass range into small mass windows and performs MS/MS analysis on all peptides in each window (see Figures 1 and 2) (2). As the size of the mass windows are dynamic, small windows can be used to scan areas with many peptides, and larger windows can be used to more quickly analyze areas with few peptides. Using SWATH acquisition results in low interference from the high amount of the drug substance protein because the data on the low abundance HCP peptides are acquired in data-independent mode. The proteins and peptides are identified through comparison with data stored in reference databases and ion libraries. As the SWATH acquisition records multi-dimensional data with comprehensive coverage of the HCPs present, the data can be re-analyzed repeatedly in the future. The digital records thus effectively immortalize information on each protein, facilitating the continuous improvement of the annotation of the HCP profiles alongside the updating of reference databases and ion libraries with more known compounds and their fragmentation patterns (2, 5, 6).

	HCP profiling by LC-MS/MS can accelerate gene therapy development

	LC–MS/MS with SWATH acquisition can help bring gene therapies to market faster, safer, and cheaper. In the development of the assay, we tested and assessed its capability to detect, quantify, and identify HCPs for the monitoring and control of purity, batch-consistency, and drug release testing. The assay is sensitive enough to detect proteins present at concentrations as low as 10 parts per million (ppm), with label-free absolute quantification (2). The assay is also able to measure proteins ranging in abundance from 10 ppm to 100,000 ppm, demonstrating a wide and linear dynamic range (2). The assay can also be used to successfully monitor changes in the quantity of individual HCPs through consecutive bioprocessing steps (1). Qualitative data on the physicochemical properties of each unique protein is also acquired, including the protein molecular weight (MW), hydrophobicity, and isoelectric point (pI) (1,2). This more detailed and holistic information is particularly useful for the development and optimization of biomanufacturing processes to ensure the clearance of specific HCPs. In terms of specificity and coverage, the assay demonstrated that it was to cover the full MW and pI range of the proteome of E. coli, a common gene therapy expression system and standard reference species (2). The assay based on microflow LC–MS/MS with SWATH acquisition method is highly precise (i.e., reproducible) and robust for HCP identification and quantification (2).

	The additional information provided by LC–MS/MS over ELISAs on the absolute quantification and identification of the HCPs is particularly useful to facilitate the optimization of biomanufacturing processes. The comprehensive data on the HCPs enables the straightforward modification of existing processes to optimize the clearance of the HCPs and other impurities. Another popular application of the HCP detection analysis using this assay is for the evaluation of ELISAs. As ELISAs remain the main regulatory compliant method specified by regulatory authorities, the choice of which commercially available ELISAs to use for HCP detection can be supported by LC–MS/MS data. This can confirm that the specificity, sensitivity, and coverage of the ELISA is sufficient to meet regulatory standards. To further inform the selection of the ELISAs, LC–MS/MS analysis can and is used to characterize specific HCPs of concern. The new assay can be used for comparing biosimilar drug products with their originators, as well as for analyzing preclinical and clinical batches of gene therapy product (2).

	Compared with the one-and-a-half to two years it usually takes to set up a new process-specific ELISA, this new LC–MS/MS assay is quick and simple to apply for each new biologic product. It typically takes six to eight weeks to optimize and establish the assay, from the start of the project to the receipt of a regulatory notice of verification. The samples can then be analyzed on an ongoing basis. The substantially shorter timeframe and improved utility of the data acquired indicates that the adoption of this LC–MS/MS assay with SWATH acquisition could accelerate the development, marketing approval, and commercial biomanufacturing of novel gene therapies. This acceleration will likely be further enhanced as advances are continually made to the assay.

	Anticipating future regulatory requirements

	The biopharma industry is shifting towards including HCP data from orthogonal methods to LBAs, primarily from MS assays. This trend is observed in the industry as companies increasingly request detailed HCP detection analysis using this new assay based on microflow LC–MS/MS with SWATH acquisition. It is expected that by preparing such detailed HCP data, which can continue to be better annotated over time, the chances of obtaining regulatory approval for clinical trials and marketing are improved. Moreover, it is an important component in the risk management strategies for many large companies investing in the costly and competitive development of new gene therapies. Implementing measures to minimize the need to re-run analyses later in the drug development cycle not only saves time, but could also prove more cost efficient.

	1. R.R. Lund, K. Pilely, and E. Mørtz,  

, June 30, 2019. 
	2. R.R. Lund, et al., 

 presentation at the Well Characterized Biologics Conference (Rockville, MD, Oct. 24–26, 2018).
	3. T. Li, Gao, et al., 

 SCIEX Technical Note, sciex.com, accessed April 2020.
	4. S. Heissel, et al., 

 poster presented at The American Society for Mass Spectrometry (ASMS) Annual Conference (San Antonio, TX, June 5–9, 2016).
	5. SCIEX, “

 sciex.com, accessed April 2020.
	6. C. Carapito, 

	Ejvind Mørtz, PhD, is co-founder and COO of Alphalyse. He has more than 20 years of experience from several biotech companies, including the development of protein analysis methods in research and chemistry, manufacturing, and control development of protein biologics, and business collaborations with contract manufacturing organizations/contract researche organization.