Detection of chromosomal alteration after infusion of gene-edited allogeneic CAR T cells.

A chromosome 14 inversion was found in a patient who developed bone marrow aplasia following treatment with allogeneic chimeric antigen receptor (CAR) Tcells containing gene edits made with transcription activator-like effector nucleases (TALEN). TALEN editing sites were not involved at either breakpoint. Recombination signal sequences (RSSs) were found suggesting recombination-activating gene (RAG)-mediated activity. The inversion represented a dominant clone detected in the context of decreasing absolute CAR Tcell and overall lymphocyte counts. The inversion was not associated with clinical consequences and wasnot detected in the drug product administered to this patient or in any drug product used in this or other trials using the same manufacturing processes. Neither was the inversion detected in this patient at earlier time points or in any other patient enrolled in this or other trials treated with this or other product lots. This case illustrates that spontaneous, possibly RAG-mediated, recombination events unrelated to gene editing can occur in adoptive cell therapy studies, emphasizes the need for ruling out off-target gene editing sites, and illustrates that other processes, such as spontaneous V(D)J recombination, can lead to chromosomal alterations in infused cells independent of gene editing.

Comparative systeomics to elucidate physiological differences between CHO and SP2/0 cell lines.

Omics-based tools were coupled with bioinformatics for a systeomics analysis of two biopharma cell types: Chinese hamster ovary (M-CHO and CHO-K1) and SP2/0. Exponential and stationary phase samples revealed more than 10,000 transcripts and 6000 proteins across these two manufacturing cell lines. A statistical comparison of transcriptomics and proteomics data identified downregulated genes involved in protein folding, protein synthesis and protein metabolism, including PPIA-cyclophilin A, HSPD1, and EIF3K, in M-CHO compared to SP2/0 while cell cycle and actin cytoskeleton genes were reduced in SP2/0. KEGG pathway comparisons revealed glycerolipids, glycosphingolipids, ABC transporters, calcium signaling, cell adhesion, and secretion pathways depleted in M-CHO while retinol metabolism was upregulated. KEGG and IPA also indicated apoptosis, RNA degradation, and proteosomes enriched in CHO stationary phase. Alternatively, gene ontology analysis revealed an underrepresentation in ion and potassium channel activities, membrane proteins, and secretory granules including Stxbpt2, Syt1, Syt9, and Cma1 proteins in M-CHO. Additional enrichment strategies involving ultracentrifugation, biotinylation, and hydrazide chemistry identified over 4000 potential CHO membrane and secretory proteins, yet many secretory and membrane proteins were still depleted. This systeomics pipeline has revealed bottlenecks and potential opportunities for cell line engineering in CHO and SP2/0 to improve their production capabilities.

Autoantibodies Against Unmodified and Citrullinated Human Endogenous Retrovirus K Envelope Protein in Patients With Rheumatoid Arthritis.

OBJECTIVE: Autoantibodies against proteins encoded by human endogenous retrovirus K (HERV-K) have been reported in patients with rheumatoid arthritis (RA), but their relevance, if any, has remained unresolved. We revisited this question and tested if such autoantibodies may react with citrullinated epitopes on the envelope (Env) protein of HERV-K. METHODS: Immunoblotting and ELISAs were conducted with unmodified Env protein and with Env citrullinated by protein arginine deiminase 4 (PAD4). Sera from 100 patients with RA, plasma from 32 patients with juvenile idiopathic arthritis (JIA), and healthy adult and pediatric controls were included. Antibody reactivity was evaluated for correlations with clinical and laboratory variables of the patients. RESULTS: We replicated and expanded upon published data suggesting that patients with RA or JIA have autoantibodies against HERV-K Env, some with high titers. Anti-HERV-K antibodies correlated with cigarette smoking and with circulating myeloperoxidase-DNA complexes indicative of nonapoptotic neutrophil cell death. Further, most of the patients with RA, but not those with JIA, had autoantibodies that reacted more strongly with Env that was citrullinated by PAD4. These anticitrullinated Env autoantibodies correlated with seropositivity and tended to be higher in patients with erosive disease. CONCLUSION: Our data suggest that anti-HERV-K immunity is elevated in RA and JIA and may have a connection with pathogenic protein citrullination in RA.

Characterization of Citrullination Sites in Neutrophils and Mast Cells Activated by Ionomycin via Integration of Mass Spectrometry and Machine Learning.

Citrullination is an important post-translational modification implicated in many diseases including rheumatoid arthritis (RA), Alzheimer's disease, and cancer. Neutrophil and mast cells have different expression profiles for protein-arginine deiminases (PADs), and ionomycin-induced activation makes them an ideal cellular model to study proteins susceptible to citrullination. We performed high-resolution mass spectrometry and stringent data filtration to identify citrullination sites in neutrophil and mast cells treated with and without ionomycin. We identified a total of 833 validated citrullination sites on 395 proteins. Several of these citrullinated proteins are important components of pathways involved in innate immune responses. Using this benchmark primary sequence data set, we developed machine learning models to predict citrullination in neutrophil and mast cell proteins. We show that our models predict citrullination likelihood with 0.735 and 0.766 AUCs (area under the receiver operating characteristic curves), respectively, on independent validation sets. In summary, this study provides the largest number of validated citrullination sites in neutrophil and mast cell proteins. The use of our novel motif analysis approach to predict citrullination sites will facilitate the discovery of novel protein substrates of protein-arginine deiminases (PADs), which may be key to understanding immunopathologies of various diseases.

Development of a high yielding expression platform for the introduction of non-natural amino acids in protein sequences.

The ability to genetically encode non-natural amino acids (nnAAs) into proteins offers an expanded tool set for protein engineering. nnAAs containing unique functional moieties have enabled the study of post-translational modifications, protein interactions, and protein folding. In addition, nnAAs have been developed that enable a variety of biorthogonal conjugation chemistries that allow precise and efficient protein conjugations. These are being studied to create the next generation of antibody-drug conjugates with improved efficacy, potency, and stability for the treatment of cancer. However, the efficiency of nnAA incorporation, and the productive yields of cell-based expression systems, have limited the utility and widespread use of this technology. We developed a process to isolate stable cell lines expressing a pyrrolysyl-tRNA synthetase/tRNApyl pair capable of efficient nnAA incorporation. Two different platform cell lines generated by these methods were used to produce IgG-expressing cell lines with normalized antibody titers of 3 g/L using continuous perfusion. We show that the antibodies produced by these platform cells contain the nnAA functionality that enables facile conjugations. Characterization of these highly active and robust platform hosts identified key parameters that affect nnAA incorporation efficiency. These highly efficient host platforms may help overcome the expression challenges that have impeded the developability of this technology for manufacturing proteins with nnAAs and represents an important step in expanding its utility.

Proteogenomic Annotation of Chinese Hamsters Reveals Extensive Novel Translation Events and Endogenous Retroviral Elements.

A high-quality genome annotation greatly facilitates successful cell line engineering. Standard draft genome annotation pipelines are based largely on de novo gene prediction, homology, and RNA-Seq data. However, draft annotations can suffer from incorrect predictions of translated sequence, inaccurate splice isoforms, and missing genes. Here, we generated a draft annotation for the newly assembled Chinese hamster genome and used RNA-Seq, proteomics, and Ribo-Seq to experimentally annotate the genome. We identified 3529 new proteins compared to the hamster RefSeq protein annotation and 2256 novel translational events (e.g., alternative splices, mutations, and novel splices). Finally, we used this pipeline to identify the source of translated retroviruses contaminating recombinant products from Chinese hamster ovary (CHO) cell lines, including 119 type-C retroviruses, thus enabling future efforts to eliminate retroviruses to reduce the costs incurred with retroviral particle clearance. In summary, the improved annotation provides a more accurate resource for CHO cell line engineering, by facilitating the interpretation of omics data, defining of cellular pathways, and engineering of complex phenotypes.

Amber suppression coupled with inducible surface display identifies cells with high recombinant protein productivity.

Cell line development (CLD) for biotherapeutics is a time- and resource-intensive process requiring the isolation and screening of large numbers of clones to identify high producers. Novel methods aimed at enhancing cell line screening efficiency using markers predictive of productivity early in the CLD process are needed to reliably generate high-yielding cell lines. To enable efficient and selective isolation of antibody expressing Chinese hamster ovary cells by fluorescence-activated cell sorting, we developed a strategy for the expression of antibodies containing a switchable membrane-associated domain to anchor an antibody to the membrane of the expressing cell. The switchable nature of the membrane domain is governed by the function of an orthogonal aminoacyl transfer RNA synthetase/tRNApyl pair, which directs a nonnatural amino acid (nnAA) to an amber codon encoded between the antibody and the membrane anchor. The process is "switchable" in response to nnAA in the medium, enabling a rapid transition between the surface display and secretion. We demonstrate that the level of cell surface display correlates with productivity and provides a method for enriching phenotypically stable high-producer cells. The strategy provides a means for selecting high-producing cells with potential applications to multiple biotherapeutic protein formats.

Ultrasmall targeted nanoparticles with engineered antibody fragments for imaging detection of HER2-overexpressing breast cancer.

Controlling the biodistribution of nanoparticles upon intravenous injection is the key to achieving target specificity. One of the impediments in nanoparticle-based tumor targeting is the inability to limit the trafficking of nanoparticles to liver and other organs leading to smaller accumulated amounts in tumor tissues, particularly via passive targeting. Here we overcome both these challenges by designing nanoparticles that combine the specificity of antibodies with favorable particle biodistribution profiles, while not exceeding the threshold for renal filtration as a combined vehicle. To that end, ultrasmall silica nanoparticles are functionalized with anti-human epidermal growth factor receptor 2 (HER2) single-chain variable fragments to exhibit high tumor-targeting efficiency and efficient renal clearance. This ultrasmall targeted nanotheranostics/nanotherapeutic platform has broad utility, both for imaging a variety of tumor tissues by suitably adopting the targeting fragment and as a potentially useful drug delivery vehicle.

T follicular helper-like cells contribute to skin fibrosis.

Systemic sclerosis (SSc) is a debilitating inflammatory and fibrotic disease that affects the skin and internal organs. Although the pathophysiology of SSc remains poorly characterized, mononuclear cells, mainly macrophages and T cells, have been implicated in inflammation and fibrosis. Inducible costimulator (ICOS), which is expressed on a subset of memory T helper (TH) and T follicular helper (TFH) cells, has been shown to be increased in SSc and associated with disease pathology. However, the identity of the relevant ICOS+ T cells and their contribution to inflammation and fibrosis in SSc are still unknown. We show that CD4+ ICOS-expressing T cells with a TFH-like phenotype infiltrate the skin of patients with SSc and are correlated with dermal fibrosis and clinical disease status. ICOS+ TFH-like cells were found to be increased in the skin of graft-versus-host disease (GVHD)-SSc mice and contributed to dermal fibrosis via an interleukin-21- and matrix metalloproteinase 12-dependent mechanism. Administration of an anti-ICOS antibody to GVHD-SSc mice prevented the expansion of ICOS+ TFH-like cells and inhibited inflammation and dermal fibrosis. Interleukin-21 neutralization in GVHD-SSc mice blocked disease pathogenesis by reducing skin fibrosis. These results identify ICOS+ TFH-like profibrotic cells as key drivers of fibrosis in a GVHD-SSc model and suggest that inhibition of these cells could offer therapeutic benefit for SSc.

A novel bicistronic gene design couples stable cell line selection with a fucose switch in a designer CHO host to produce native and afucosylated glycoform antibodies.

The conserved glycosylation site Asn297 of a monoclonal antibody (mAb) can be decorated with a variety of sugars that can alter mAb pharmacokinetics and recruitment of effector proteins. Antibodies lacking the core fucose at Asn297 (afucosylated mAbs) show enhanced antibody-dependent cell-mediated cytotoxicity (ADCC) and increased efficacy. Here, we describe the development of a robust platform for the manufacture of afucosylated therapeutic mAbs by engineering a Chinese hamster ovary (CHO) host cell line to co-express a mAb with GDP-6-deoxy-D-lyxo-4-hexulose reductase (RMD), a prokaryotic enzyme that deflects an intermediate in the de novo synthesis of fucose to a dead-end product, resulting in the production of afucosylated mAb (GlymaxX™ Technology, ProBioGen). Expression of the mAb and RMD genes was coordinated by co-transfection of separate mAb and RMD vectors or use of an internal ribosome entry site (IRES) element to link the translation of RMD with either the glutamine synthase selection marker or the mAb light chain. The GS-IRES-RMD vector format was more suitable for the rapid generation of high yielding cell lines, secreting afucosylated mAb with titers exceeding 6.0 g/L. These cell lines maintained production of afucosylated mAb over 60 generations, ensuring their suitability for use in large-scale manufacturing. The afucosylated mAbs purified from these RMD-engineered cell lines showed increased binding in a CD16 cellular assay, demonstrating enhancement of ADCC compared to fucosylated control mAb. Furthermore, the afucosylation in these mAbs could be controlled by simple addition of L-fucose in the culture medium, thereby allowing the use of a single cell line for production of the same mAb in fucosylated and afucosylated formats for multiple therapeutic indications.

Tag-on-Demand: exploiting amber codon suppression technology for the enrichment of high-expressing membrane protein cell lines.

Membrane proteins play key roles in the evolution of numerous diseases and as a result have become the most dominant class of targets for therapeutic intervention. However, their poor expression and detection oftentimes prohibit drug discovery and screening efforts. Herein, we have developed an approach, named 'Tag-on-Demand' that exploits amber suppression to control the expression of 'tagged' membrane proteins for detection and selections, yet can be turned off for expression of the protein in its native form. Utilizing an engineered Chinese hamster ovary cell line capable of efficient amber suppression, we evaluated the expression of a diverse panel of model membrane proteins and demonstrated the enrichment of cells with improved expression profiles, where ~200-800% improvement in total protein expression levels were observed over pre-sorted populations after a single round of fluorescence-activated cell sorting. Furthermore, results were most striking for the typically difficult-to-express G protein-coupled receptor, CXCR2, where ~2.5-fold improvement in surface expression was observed. We anticipate that the Tag-on-Demand approach will be suitable not only for membrane protein cell line development but also for the development of intracellular and secreted protein cell lines in expression systems for which amber suppression technology exists, including bacterial, yeast, insect and cell-free expression systems.

Sequential screening by ClonePix FL and intracellular staining facilitate isolation of high producer cell lines for monoclonal antibody manufacturing.

Screening and characterization of cell lines for stable production are critical tasks in identifying suitable recombinant cell lines for the manufacture of protein therapeutics. To aid this essential function we have developed a methodology for the selection of antibody expressing cells using fluorescence based ClonePix FL colony isolation and flow cytometry analysis following intracellular staining for immunoglobulin G (IgG). Our data show that characterization of cells by flow cytometry early in the clone selection process enables the identification of cell lines with the potential for high productivity and helps to eliminate unstable cell lines. We further demonstrate a correlation between specific productivity (qP) and intracellular heavy chain (HC) content with final productivity. The unique combination of screening using ClonePix FL and the flow cytometry approaches facilitated more efficient isolation of clonal cell lines with high productivity within a 15week timeline and which can be applied across NS0 and CHO host platforms. Furthermore, in this study we describe the critical parameters for the ClonePix FL colony based selection and the associated calculations to provide an assessment of the probability of monoclonality of the resulting cell lines.

Lessons from the Hamster: Cricetulus griseus Tissue and CHO Cell Line Proteome Comparison.

Chinese hamster ovary cells represent the dominant host for therapeutic recombinant protein production. However, few large-scale data sets have been generated to characterize this host organism and derived CHO cell lines at the proteomics level. Consequently, an extensive label-free quantitative proteomics analysis of two cell lines (CHO-S and CHO DG44) and two Chinese hamster tissues (liver and ovary) was used to identify a total of 11 801 unique proteins containing at least two unique peptides. 9359 unique proteins were identified specifically in the cell lines, representing a 56% increase over previous work. Additionally, 6663 unique proteins were identified across liver and ovary tissues, providing the first Chinese hamster tissue proteome. Protein expression was more conserved within cell lines during both growth phases than across cell lines, suggesting large genetic differences across cell lines. Overall, both gene ontology and KEGG pathway analysis revealed enrichment of cell-cycle activity in cells. In contrast, upregulated molecular functions in tissue include glycosylation and lipid transporter activity. Furthermore, cellular components including Golgi apparatus are upregulated in both tissues. In conclusion, this large-scale proteomics analysis enables us to delineate specific changes between tissues and cells derived from these tissues, which can help explain specific tissue function and the adaptations cells incur for applications in biopharmaceutical productions.

Development of a fluorescent reporter system for monitoring ER stress in Chinese hamster ovary cells and its application for therapeutic protein production.

Mammalian cell expression systems have become a workhorse for the production of biotherapeutic proteins. As such, there is an ever increasing demand for higher productivity from these expression platforms to reduce manufacturing costs. While great advances have been made in the optimization of culture conditions and cell line selection to improve productivity, protein mis-folding remains a common limitation to high levels of production of therapeutic proteins. Accumulation of mis- and unfolded protein in the endoplasmic reticulum (ER) causes ER stress and initiates the unfolded protein response (UPR) that results in an activation of protein folding machinery, translation attenuation in an effort to proper folding of the newly synthesized peptides or may even lead to apoptosis if the correct folding is not restored. As a result, UPR associated apoptosis often results in lower protein expression. To better understand the molecular mechanisms in these pathways, we developed a reporter construct that detects Inositol-requiring enzyme 1 (IRE1)-alpha mediated splicing of X-box binding protein 1 (XBP1) to monitor the course of UPR activation in cell lines expressing monoclonal antibodies. Using this reporter we observed a clear activation of UPR in cells treated with known ER stress causing pharmacological agents, such as Tunicamycin (Tm) and Thapsigargin (Tg), as well as in stable IgG expressing cells during fed-batch cultures. Furthermore, we developed a stress metric that we term as ER stress index (ERSI) to gauge basal ER stress in cells which we used as a predictive tool for isolation of high IgG expressing cell lines. This reporter system, with its ability to monitor the stress involved in recombinant protein expression, has utility to assist in devising engineering strategies for improved production of biotherapeutic drugs.

Statistical Models for the Analysis of Isobaric Tags Multiplexed Quantitative Proteomics.

Mass spectrometry is being used to identify protein biomarkers that can facilitate development of drug treatment. Mass spectrometry-based labeling proteomic experiments result in complex proteomic data that is hierarchical in nature often with small sample size studies. The generalized linear model (GLM) is the most popular approach in proteomics to compare protein abundances between groups. However, GLM does not address all the complexities of proteomics data such as repeated measures and variance heterogeneity. Linear models for microarray data (LIMMA) and mixed models are two approaches that can address some of these data complexities to provide better statistical estimates. We compared these three statistical models (GLM, LIMMA, and mixed models) under two different normalization approaches (quantile normalization and median sweeping) to demonstrate when each approach is the best for tagged proteins. We evaluated these methods using a spiked-in data set of known protein abundances, a systemic lupus erythematosus (SLE) data set, and simulated data from multiplexed labeling experiments that use tandem mass tags (TMT). Data are available via ProteomeXchange with identifier PXD005486. We found median sweeping to be a preferred approach of data normalization, and with this normalization approach there was overlap with findings across all methods with GLM being a subset of mixed models. The conclusion is that the mixed model had the best type I error with median sweeping, whereas LIMMA had the better overall statistical properties regardless of normalization approaches.

CD6 as a potential target for treating multiple sclerosis.

CD6 was established as a marker of T cells more than three decades ago, and recent studies have identified CD6 as a risk gene for multiple sclerosis (MS), a disease in which autoreactive T cells are integrally involved. Nevertheless, the precise role of CD6 in regulating T-cell responses is controversial and its significance in the pathogenesis of various diseases remains elusive, partly due to the lack of animals engineered to alter expression of the CD6 gene. In this report, we found that CD6 KO mice showed decreased pathogenic T-cell responses, reduced spinal cord T-cell infiltration, and attenuated disease severity in experimental autoimmune encephalomyelitis (EAE), an animal model of MS. CD6-deficient T cells exhibited augmented activation, but also significantly reduced survival and proliferation after activation, leading to overall decreased Th1 and Th17 polarization. Activated CD6-deficient T cells also showed impaired infiltration through brain microvascular endothelial cell monolayers. Furthermore, by developing CD6 humanized mice, we identified a mouse anti-human CD6 monoclonal antibody that is highly effective in treating established EAE without depleting T cells. These results suggest that (i) CD6 is a negative regulator of T-cell activation, (ii) at the same time, CD6 is a positive regulator of activated T-cell survival/proliferation and infiltration; and (iii) CD6 is a potential new target for treating MS and potentially other T-cell-driven autoimmune conditions.

Guiding bispecific monovalent antibody formation through proteolysis of IgG1 single-chain.

We developed an IgG1 domain-tethering approach to guide the correct assembly of 2 light and 2 heavy chains, derived from 2 different antibodies, to form bispecific monovalent antibodies in IgG1 format. We show here that assembling 2 different light and heavy chains by sequentially connecting them with protease-cleavable polypeptide linkers results in the generation of monovalent bispecific antibodies that have IgG1 sequence, structure and functional properties. This approach was used to generate a bispecific monovalent antibody targeting the epidermal growth factor receptor and the type I insulin-like growth factor receptor that: 1) can be produced and purified using standard IgG1 techniques; 2) exhibits stability and structural features comparable to IgG1; 3) binds both targets simultaneously; and 4) has potent anti-tumor activity. Our strategy provides new engineering opportunities for bispecific antibody applications, and, most importantly, overcomes some of the limitations (e.g., half-antibody and homodimer formation, light chains mispairing, multi-step purification), inherent with some of the previously described IgG1-based bispecific monovalent antibodies.

A CD80-Biased CTLA4-Ig Fusion Protein with Superior In Vivo Efficacy by Simultaneous Engineering of Affinity, Selectivity, Stability, and FcRn Binding.

Affinity- and stability-engineered variants of CTLA4-Ig fusion molecules with enhanced pharmacokinetic profiles could yield improved therapies with the potential of higher efficacy and greater convenience to patients. In this study, to our knowledge, we have, for the first time, used in vitro evolution to simultaneously optimize CTLA4 affinity and stability. We selected for improved binding to both ligands, CD80 and CD86, and screened as dimeric Fc fusions directly in functional assays to identify variants with stronger suppression of in vitro T cell activation. The majority of CTLA4 molecules showing the largest potency gains in primary in vitro and ex vivo human cell assays, using PBMCs from type 1 diabetes patients, had significant improvements in CD80, but only modest gains in CD86 binding. We furthermore observed different potency rankings between our lead molecule MEDI5265, abatacept, and belatacept, depending on which type of APC was used, with MEDI5265 consistently being the most potent. We then created fusions of both stability- and potency-optimized CTLA4 moieties with human Fc variants conferring extended plasma t1/2 In a cynomolgus model of T cell-dependent Ab response, the CTLA4-Ig variant MEDI5265 could be formulated at >100 mg/ml for s.c. administration and showed superior efficacy and significantly prolonged serum t1/2 The combination of higher stability and potency with prolonged pharmacokinetics could be compatible with very infrequent, s.c. dosing while maintaining a similar level of immune suppression to more frequently and i.v. administered licensed therapies.

High-Throughput Lipidomic and Transcriptomic Analysis To Compare SP2/0, CHO, and HEK-293 Mammalian Cell Lines.

A combined lipidomics and transcriptomics analysis was performed on mouse myeloma SP2/0, Chinese hamster ovary (CHO), and human embryonic kidney (HEK) cells in order to compare widely used mammalian expression systems. Initial thin layer chromatography (TLC) analysis indicated that phosphatidylethanolamine (PE) and phosphatidylcholine (PC) were the major lipid components in all cell lines with lower amounts of sphingomyelin (SM) in SP2/0 compared to CHO and HEK, which was subsequently confirmed and expanded upon following mass spectrometry (MS) analysis. HEK contained 4-10-fold higher amounts of lyso phosphatidylethanolamine (LPE) and 2-4-fold higher amounts of lyso phosphatidylcholine (LPC) compared to SP2/0 and CHO cell lines. C18:1 followed by C16:1 were the main contributors to the difference in both LPE and LPC levels. Alternatively, the SP2/0 cell line exhibited 30-65-fold lower amounts of SM principally in the amount of 16:0. By mapping the transcriptomics data to KEGG pathways, we found expression levels of secretory phospholipase A2 (sPLA2), lysophospholipid acyltransferase (LPEAT), lysophosphatidylcholine acyltransferase (LPCAT), and lysophospholipase (LYPLA) can contribute to the differences in LPE and LPC. Sphingomyelin synthases (SMS) and sphingomyelin phosphodiesterase (SMase) enzymes may play roles in SM differences across the three cell lines. The results of this study provide insights that will aid the understanding of the physiological and secretory differences across recombinant protein production systems.