Antha improves yield and robustness of lentiviral vector particle production

Oxford Biomedica used Antha to perform a multifactorial optimisation of their lentiviral vector transfection / transduction system, resulting in: a 3 to 10-fold increase in vector titre upon transduction, 81% reduction in pure error, 83% time saving and 32% resource saving


Oxford Biomedica and Antha

Oxford Biomedica is a world leader and pioneer in lentiviral vector research, manufacture and clinical development. Oxford Biomedica have over 20 years of experience in the field of gene and cell therapy and were the first organisation to treat humans with in vivo lentiviral based vectors. Using their unique LentiVector® platform they have created a valuable portfolio of gene and cell therapy product candidates for oncological, ophthalmological and CNS indications.

Oxford Biomedica started working with Antha in 2018 as they recognised the need for a scalable, flexible and robust software for automating highly complex experiments. They chose Synthace and Antha as they provide rapid and flexible design and in silico simulation of laboratory automation protocols through to physical execution of protocols in a hardware agnostic manner.

Executive Summary

Oxford Biomedica employed Antha in their Research and Development labs where they were looking to improve the efficiency and robustness of their in-house lentiviral vector production through optimisation of their transfection reagent mixes. To comprehensively investigate this complex biological system a multifactorial experimental approach was taken that would have been very challenging via manual execution, due to the number of experimental runs and the complexity of the experimental design.

However with Antha this is achievable and afforded Oxford Biomedica up to 40 hours time saving on experimental planning and execution whilst providing robust and reliable execution via an automation liquid handling platform and an Antha protocol that could be flexibly tuned for subsequent Design of Experiments (DoE) campaigns. The initial optimisation results identified transfection conditions in which between a 3 to 10-fold increase in vector titre was achieved upon transduction, with an 81% reduction in pure error indicative of a more robust process achieving the two key optimisation objectives within two weeks of physical execution.

Gene and Cell Therapy

The gene therapy field first emerged in 1989 with the clinical trial of a genetically modified human tumour-infiltrating lymphocyte (TIL). [1] The first approved gene therapy in the EU, Glybera, for the inherited disease Lipoprotein Lipase Deficiency used an Adeno Associated Virus (AAV) therapy. Since this approval, the field has rapidly expanded with both ex vivo gene therapies (autologous and allogeneic gene modified cell therapies) and in vivo gene therapies [2] all reaching the clinic or market including Kymriah, Yescarta and Luxtarna.

Since 2014, over $36.8 billion of funding has flowed into the gene and cell therapy field, with the broader gene therapy field experiencing over 30% year-on-year growth between 2016 and 2018. On the back of this funding, there are now over 700 clinical trials for gene therapies (ex vivo and in vivo), the vast majority of these being in phase I or II clinical trials, clustered around oncology or monogenic indications.

The public imagination has also been stirred by the growing prominence of chimeric antigen receptor (CAR) technology, and in particular the approvals of Kymriah (Novartis) and Yescarta (Kite, Gilead) for blood cancers. In 2018, Kymriah was approved for use in the NHS, and in January 2019 the first patient in the UK was treated for B-cell acute lymphoblastic leukaemia (ALL) with this therapy. Oxford Biomedica is the sole supplier and manufacturer of the lentiviral vector used to generate Kymriah.

These successes and the attention they have brought have highlighted the challenges in this space. One of these is the high cost to therapy, driven by both market structure and the potential one-time curative nature of these therapies. Another is the global demand for quality lentiviral vector creating a vector crunch in the sector, encouraging companies to optimise product quality, developability and vector titres wherever they can. In response to these challenges, gene and cell therapy companies are turning to technology – automation equipment, informatics software and supply chain systems in order to improve product quality, reduce costs, increase facility throughput and ultimately satisfy market demand.

Lentiviral Vectors

Lentiviral vectors are able to transduce (genetically modify) a broad range of both dividing and non-dividing cell types. This leads to stable and long term expression of the gene or genes of interest. Lentiviral vectors have been found to be safe and well tolerated in multiple clinical trials. They are being used in in vivo gene therapy applications for Central Nervous system (CNS) indications such as Parkinson’s Disease and Amyotrophic Lateral Sclerosis (ALS), and Ophthalmic indications such as Wet Age-Related Macular Degeneration and Retinitis Pigmentosa. They are also used in ex vivo gene therapy or cell therapy applications like the afore mentioned Kymriah for the treatment of B-Cell Acute Lymphoblastic Lymphoma (ALL) and Diffuse Large B-Cell Lymphoma (DLBCL) amongst many other cancer indications.

Third-generation lentiviral vectors have now been developed, with improved safety profiles, by removing the likelihood of making a self-replicating virion by several methods; removal of genes encoding virulence accessory proteins and separation of core viral elements onto separate plasmids [3, 4]. Swapping the natural viral envelope protein to that of the Vesicular stomatitis Indiana virus, VSV-G, conferred tropism to broad target cell types (Figure 1).

Lentiviral vectors carrying a gene of interest are produced via transfection of a packaging cell line with several plasmids encoding the core structural and packaging proteins required for formation of vector particles. The vectors produced by the packaging cell line are then harvested and used to transduce a target cell line for delivery of the genetic payload to the target cells (Figure 1). Oxford Biomedica are actively engaged in continuous improvement of process and analytics for delivery of higher quantity and improved quality of products with better characterisation. Optimising for the production of functional vector particles carrying the desired genetic payload in the manufacturing process will help address the market need and demand for gene therapies by potentially decreasing research and development timelines, increasing yield, decreasing required dosage and decreasing overall time to market.

The process of vector particle production can lead to ineffective as well as active vector particles. Optimal production of active over ineffective vector particles and their ultimate efficacy is largely determined by the respective ratios of the many components contributing to the transfection process. Using Antha to aid planning and execution of a complex multifactorial design space helps address questions surrounding yield and robustness of vector particle production.

BG #ededed.jpg

Figure 1. Lentiviral vector architecture and production process. Third generation lentiviral vectors are composed of protein and RNA components encoded on four separate plasmids to improve safety profiles (top panel). The target transgene of interest is flanked by self-inactivating long terminal repeats further increasing the safety profile of this vector system. Structural proteins are coded for by gag-pol and env genes whilst rev encodes a protein required to facilitate mRNA transport from the nucleus for translation in the cytoplasm, initiating vector particle production. All four plasmids require transfection into a packaging cell line in order for vector particles to be produced (bottom panel). In this study the DoE reaction optimisation carried out by Antha is that of the transfection reagent mix (1), after transfection (2) the production of vector particles is triggered (3). Vector particles are harvested from cell culture media and manually prepped for transduction into a target cell line (4) where the transgene cargo integrates into the host genome (5).


Design of Experiments

DoE is an approach to process optimisation which aims to minimise time and cost of results by simultaneously varying process conditions (factors) in a statistically optimal way, iteratively refining an underlying statistical process model. [5, 6] Oxford Biomedica shortlisted 10 factors for investigation in the first iteration of the DoE optimisation. This allowed for a more complex analysis of how different factors interact, a process that is not addressed by one factor at a time experiments and rarely addressed in DoE given the inability to investigate more than 4-5 factors when executing the experimental set up manually (customer testimonial).

Given the increase in complexity of experimental design execution afforded by Antha, in parallel to the optimisation targets set out in the experimental campaign, Oxford Biomedica wanted to investigate the benefits and value of deviating from a traditional iterative DoE methodology for a quicker optimisation strategy.

A more traditional DoE experimental campaign follows factor screening, iteration, refinement and optimisation stages (Figure 2), often taking anywhere from 2 - 12 weeks depending on sample work up for analytics and the speed of the system being studied. However, given the capability to investigate a larger number of factors (up to 10) and execute many more experimental runs (90-150) in a single iteration providing more data and statistical power, a shorter campaign following a factor screening and combined refinement and optimisation stage (Figure 2) was investigated, taking only 2 weeks in this instance to physically execute both iterations.

BG #ededed.jpg

Figure 2. DoE Optimisation Strategies. Schematic representations of iterating through experimental design space. A typical DoE experimental optimisation usually follows a factor screen, iterate, refinement and optimisation frame work (top panel). In this study, Oxford Biomedica investigated a rapid iteration DoE strategy with a factor screen followed by joint refinement and optimisation iteration.


Antha and DoE

Using the Antha DoE suite of elements, Oxford Biomedica were able to rapidly and flexibly define an automated liquid handling protocol that simulates all the low-level liquid handling instructions interpreted from a DoE design file in order to then execute that experimental run on a Gilson PIPETMAX® (Figure 3). The initial workflow set up and optimisation took ~30 minutes, providing a 95-98% time saving when compared to trying to hard code an equivalent method using alternative liquid handling platform vendors’ software, which still couldn’t provide the user the same breadth or flexibility that the Antha workflow provides.

Figure 3. Simple, rapid and flexible workflow prototyping in Antha’s Workflow Editor. The graphical user interface of Antha’s Workflow Editor affords a biological scientist the ability to rapidly prototype automated liquid handling workflows through programming at a higher level of abstraction with respect to most hardware vendors’ automated liquid handling software. The DoE workflow used for driving the execution of both DoE iterations in this study is shown here.


Antha and DoE

Antha’s DoE elements provide features that allow you to rapidly prototype and optimise DoE execution in silico whilst minimising the estimated execution time, number of liquid handling transfers and therefore labware used during the run. The Antha workflow prepares intermediate mixtures of user specified factors from the design file before further distributing and mixing these master mixes in such a way as to achieve the final set point concentrations of all factors in a single run (Figure 3). For these DoE designs, Antha provided up to a 22% saving in the number of liquid handling actions required for execution, equating to an estimated 2 hours saving in run times. The saving on execution run time afforded by smart liquid handling steps translates to an estimated 32% saving in pipette tip usage (373 tips or 4 x 96 tip racks). All planning for the execution of the DoE is taken care of by Antha, guiding the user on required reagents, volumes and labware, detailed schematics of how to set up input plates and how the liquid handling platform should be set up (Figure 4).


Figure 4. Simple user set up. A video of Antha’s preview page directing the users as to how to set up their automated liquid handling platform and to run an in silico simulation of all liquid handling steps to validate the experimental workflow prior to physical execution. All low-level decisions are taken care of by Antha so the user isn’t required to determine deck layouts before conducting a physical run in the lab. This decreases risk of human error or the need for repetitive dry run physical testing in the lab before being able to carry out execution with samples.


BG #ededed.jpg


Improving Robustness with DoE

For each DoE iteration carried out by Oxford Biomedica, a plasmid encoding a fluorescent transgene was used providing a quantitative measure for both transfection efficiency (expression of transgene in packaging cell line) and transduction (expression of transgene in target cell line) via flow cytometry.

Models generated against the transduction and transfection data for both DoE iterations agreed well, identifying key factors and factor interactions contributing to the efficiency of the processes (Figure 5 A & D). In particular the models were able to highlight intuitive factor interactions (Figure 5 B & E) and some novel interactions (Figure 5 C & F) that would not have been identified without a DoE optimisation framework.

Data from the factor screening iteration identified two factors that were mutually negatively correlated directing the exclusion of one of these factors from the refine and optimisation iteration. A decision tree analysis of a subset of reactions from the screening iteration also led to identifying a specific level combination of three factors resulting in no transduction, helping guide the setting of new factor levels in the refine and optimisation iteration to avoid this detrimental design space.

Significant Factors and Modelled Factor Interactions

Figure 5. Significant Factors and Modelled Factor Interaction response surfaces. Half normal plots highlighting factors of influence on the screening model (A) and response surface model (D) for transduction titre, Blue lines represent normal distribution. Modelled factor interaction response surfaces observed in the factor screening iteration (B and C) or the refine and optimise iteration (E and F). Data analysis and graphing performed in JMP 14.1.0.


A key objective for this DoE optimisation was to improve the robustness of the transduction process for scaled down reactions in microtiter plates. There are several manual liquid handling steps between the transfection and transduction process that are thought to introduce variability in downstream data and therefore compromise its validity. Replicated centre points from the two DoE iterations demonstrate that the region of design space found after only one iteration is considerably less variable with improved transduction titre (Figure 6).

Estimates of pure error from the design confirm this showing a reduction in mean square pure error from 0.44 to 0.08 between factor screening and refine and optimisation iterations respectively, accomplishing the core optimisation objectives.

BG #ededed.jpg

Comparison of Replicated Values Within the Design


Figure 6. Comparison of replicated values within the design. Five centre points from the factor screening iteration (left) and six from the refine and optimisation iteration (right) are compared for logged transduction response. The results from the refine and optimisation iteration show both an increased transduction titre and substantially less variability. Data analysis and graphing performed in JMP 14.1.0.



Using Antha, Oxford Biomedica have been able to design, plan and execute a reduced time frame two iteration DoE optimisation and robustness study of their transfection and transduction process (Figure 2). The designs in both iterations contained more factors for investigation and more experimental runs than could have been executed manually (Figure 7 and Table 1), expanding the power and potential to explore and understand the complex biological system to a much higher degree.

Antha provided up to an 83% time saving from combined design, planning, and physical execution and a 32% resource saving allowing Oxford Biomedica scientists to focus on the analysis of their results and experimental planning, driving their research and development forwards, rather than focusing on repetitive and error prone manual execution.

Comparison of time savings and design complexity gains provided by Antha over manual execution or alternative liquid handling platform software highlights the rapidity and flexibility that Antha affords, increasing on design complexity (up to 150 runs versus up to 40 runs) whilst reducing overall time to execution (8 hours versus up to 49 hours) (Figure 7).

BG #ededed.jpg

Figure 7. Time saving and design complexity gains provided by Antha. Antha provides up to a 98% time saving on design and planning for DoEs when compared to other automated liquid handling platform software (left) whilst providing the ability to execute complex designs beyond the capability of tractable manual execution (see customer testimonial, Table 1). With Antha, programming automated liquid handling platforms for high complexity experimentation reduces the burden of alternative solutions without compromising on complexity (right).

BG #ededed.jpg

User experience comparisons

Table 1. User experience comparisons for automated or manual execution solutions for DoE.


Ready to Revolutionise your Workflows?

Book a Demo

BG #ededed.jpg