Addressing Bioreactor hiPSC Aggregate Stability, Maintenance and Scaleup Challenges Using a Design of Experiment Approach

Addressing Bioreactor hiPSC Aggregate Stability, Maintenance and Scaleup Challenges Using a Design of Experiment Approach | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Addressing Bioreactor hiPSC Aggregate Stability, Maintenance and Scaleup Challenges Using a Design of Experiment Approach Haneen Yehya, Sofija Raudins, Jan Jensen, Michael A. Bukys This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3500780/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Background Stem cell-derived therapies hold the potential for treatment of regenerative clinical indications. Static culture has a limited ability to scale up thus restricting its use. Suspension culturing can be used to produce target cells in large quantities, but also presents challenges related to stress and aggregation stability. Methods Utilizing a design of experiments (DoE) approach in vertical wheel bioreactors, we evaluated media additives that have versatile properties. The additives evaluated are Heparin sodium salt (HS), polyethylene glycol (PEG), poly (vinyl alcohol) (PVA), Pluronic F68 and dextran sulfate (DS). Multiple response parameters were chosen to assess cell growth, pluripotency maintenance and aggregate stability in response to the additive inputs, and mathematical models were generated and tuned for maximal predictive power. Results Expansion of iPSCs using 100 ml vertical wheel bioreactor assay for 4 days on 19 different media combinations resulted in models that can optimize pluripotency, stability, and expansion. The expansion optimization resulted in the combination of PA, PVA and PEG with E8. This mixture resulted in an expansion doubling time that was 40% shorter than that of E8 alone. Pluripotency optimizer highlighted the importance of adding 1% PEG to the E8 medium. Aggregate stability optimization that minimizes aggregate fusion in 3D culture indicated that the interaction of both Heparin and PEG can limit aggregation as well as increase the maintenance capacity and expansion of hiPSCs, suggesting that controlling fusion is a critical parameter for expansion and maintenance. Validation of optimized solution on two cell lines in bioreactors with decreased speed of 40 RPM, showed consistency and prolonged control over aggregates that have high frequency of pluripotency markers of OCT4 and SOX2 (> 90%). A doubling time of around 1 to 1.4 days was maintained after passaging as clumps in the optimized medium. Controlling aggregate fusion allowed for a decrease in bioreactor speed and therefore shear stress exerted on the cells in a large-scale expansion. Conclusion This study resulted in a control of aggregate size within suspension cultures, while informing about concomitant state control of the iPSC state. Wider application of this approach can address media optimization complexity and bioreactor scale-up challenges. iPSC human induced pluripotent stem cell pluripotent state Bioreactor DoE culture media suspension culture aggregates aggregate size aggregate stability proliferation Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Background Human pluripotent stem cells (hPSCs) can proliferate continuously and have the capacity to differentiate into any human cell type. Upon differentiation, this unlocks their potential to treat various clinical indications such as diabetes, orthopedic injuries, neurological, cardiovascular diseases, and others by supplying cells needed to restore tissue function. Induced pluripotent stem cells (iPSCs) or reprogrammed cells generated from mature specialized cells [ 1 ], are becoming the focus in the field of regenerative medicine, as they are not restricted by ethical concerns of embryonic stem cells but display a similar differentiation and growth capacity as previously used embryonic stem (ES) cells. While no current FDA-approved iPSC cell therapy exists, many are in clinical trials. Therapeutic applications of iPSCs in regenerative medicine typically depend on the availability of 10 8 -10 10 clinical grade cells per patient, where such material is manufactured using a current Good Manufacturing Practice (cGMP) process and performed within an isolated production environment to assure patient safety [ 2 – 4 ]. Conventional adherent culture flasks are impractical for large scale production as they require frequent manual intervention. Adherent cultures have a limited ability to scale up and result in high batch-to-batch variability and a lack of cost effectiveness. Alternatively, continuously stirred-tank suspension bioreactors provide an hiPSC 3D culture that generates aggregates that are more biologically comparable to an in vivo environment [ 5 ]. In addition, bioreactors can provide a continuous monitoring of environmental factors including temperature, pH, oxygen, and nutrients in a sterile environment required to produce clinical grade cells. Maintaining pluripotent state in large-scale suspension environment will improve reproducibility and cell quality, but it requires control over media composition, aggregation and physicochemical stresses exerted on cells. iPSC expansion in bioreactors presents several challenges including cell clumping, shear stress, complex media composition and cost. Cell clumping or aggregate fusion can cause heterogenous cell populations to arise [ 6 ]. Shear stress leads to aggregate breakage, DNA-breaks, resulting in cell death and karyotypic instability [ 7 ]. An environment that can prevent unwanted cell adhesion and maintain aggregate stability while maintaining pluripotency for manufacturing expansion is in demand. Cell State control is particularly relevant, as heterogeneous populations presents a serious risk, as incomplete differentiation impacts both clinical safety and potency [ 6 ]. While simple elements such as temperature and pH can be controlled and monitored, molecular component interactions, concentration sensitivity, chemical stability of additives, when combined with physical stresses, mechanical forces present all combine into a very challenging production problem; this also presents an opportunity to identify which key critical process parameters truly govern culture performance [ 8 ]. For instance, although recent studies confirm that mechanical stimuli affect iPSCs during their differentiation [ 9 , 10 ], mechanosensing signaling is not well understood especially in terms of stem cell mechanobiological interactions within the physical environment, of a bioreactor [ 11 , 12 ]. A critical challenge to generating well-controlled studies assessing higher-order interactions is characterizing the complex mixture of variable parameters that affect the molecular and functional response of iPSCs [ 12 ]. Over the last 18 years, exhaustive optimization has been done by multiple groups mainly using various adherent hiPSC culture mediums. This has led to the identification of FGF2[ 13 ] and TGFb1 as the two critical growth factors in the Essential 8 (E8) medium formulation. E8 is a further modification identifying the essential components from the original TeSR medium which pioneered early pluripotency studies. Through an intermediary B8 medium which is a cost reduced optimization from the E8 medium [ 14 ]. Currently, there is broad focus on determining defined, stable, cost-effective solutions for iPSC culture maintenance and growth. One of the main improvements in the E8 formulation is the absence of bovine serum albumin (BSA) component, since it is a xeno component that has issues with consistencies. Pairwise dropout testing done in the E8 optimization effort uncovered that β-mercaptoethanol (BME) without BSA caused toxicity, but with the absence of BME, BSA was able to be eliminated from the medium. This finding highlights the complexity of doing optimization experiments without studying the interactions and non-linear effects. A key factor in obtaining well-founded results is understanding interaction effects and reducing the variability of environmental factors in the process [ 14 ]. Limited research evaluated the media needs for suspension culture that can address the shear stress generated from a bioreactor and the absence of substrate attachment in a 3D environment. Here, we demonstrate a systematic approach maintaining iPSC aggregates in a uniform controlled stable size within a bioreactor. Heparin sodium salt (HS), polyethylene glycol (PEG), poly (vinyl alcohol) (PVA), Pluronic F68 (Pluronic F68) and dextran sulfate (DS) are compounds commonly used in biomedical and pharmaceutical applications. They have versatile properties such as reducing shear stress by decreasing surface tension of the media, enhancing extracellular matrix and cell membrane interaction, increasing aggregate stability, and preventing cell fusion [ 15 – 17 ]. Using a design of experiment (DoE) factorial study we develop process understanding of the complex interactions between these components. The DoE approach identifies statistically significant interactions resulting in a media formulation that is reproduceable by limiting variation bias. Models generated using these 5 factors can be optimized for three major cell culture criteria: (1) growth rate or doubling time, (2) pluripotency maintenance, and (3) aggregate stability. These parameters are shown to be independent, and optimization for each attribute is related to different culture conditions. Consequently, attaining ideal culture conditions across the desirability criteria spectrum is an exercise based on compromise. Methods Generating DoE designs and perturbation matrixes The DoE runs were computer generated from MODDE software (Sartorius Stedim Data Analytical Solutions, SSDAS) using D-optimal interaction designs. All components tested and outputs measured were manually inputted into Design Wizard in MODDE. Factors known to impact the aggregate stability in suspension culture were chosen for the design. Reagent concentration ranges were based on previous publications. Design runs were chosen to have 16 reactions conditions with the addition of 3 center point conditions. One of the reactions had no additives, which is the E8 bioreactor. Examples of reaction additives of different bioreactors are shown in (Figure S1 A-B). The highest G-efficiency design was chosen from the DoE designs generated by the software. Media reactions were made manually and filtered before use. Negative log transform was only used to normalize distribution for flow data sets for OCT4 and TRA-1-60 to meet normality assumptions. Assay and media preparation along with detailed methods are described in the supplemental information. qPCR primers, antibodies and reagents used are listed in (Table S1 and S2). Human induced pluripotent stem cell culture Cells were maintained at 37°C and 5% CO2. E8 medium was purchased ready in solution. Other medium components were purchased as powder. Prior to the assay, cells were grown to 70–80% confluence after 4 days of culture on vitronectin coated 6 well plates. Cells were dissociated with TrypLE, for 3 minutes at 37°C and resuspended in E8 medium, transferred to 50 ml conical tubes, and centrifuged at 400 x g for 6 minutes. The pellet was resuspended in E8 and 10 µM Y-27632 ROCK inhibitor. 11 million cells were seeded in a 100ml bioreactor with the corresponding mediums. Cell counting and aggregate size Samples from the bioreactor were taken after mixing using a pipette to avoid sampling bias and gradient formation after settling. 3 ml total was sampled daily for 3 samples of 1ml for cell count. The cells were first dissociated using Accutase and incubated for 10 minutes. Cells were then quenched with E8 then centrifuged at 400 x g for 6 minutes. The pellet was resuspended in the same volume of PBS. Samples were then analyzed for total cell count using Attune flow. Triplicate counts were collected for each bioreactor for each day of culture. Duplicate samples were collected for aggregate imaging of 500ul and put on a 24 well plate. EVOS 7000 was used for bright field images of the aggregates. ImageJ was then used to analyze aggregate size and distribution. A minimum of 30 aggregates were measured for each bioreactor every day of the culture. This data was then analyzed for standard deviation on excel and growth rate using GraphPad Prism9. The cell counts were also analyzed for doubling time using GraphPad Prism. Average data from aggregate size and counts were then inputted into the response tab in MODDE. Using the aggregate size generated from day 1 from all bioreactors and their corresponding cell growth rate, a predicted aggregate size was calculated for day 3 and compared to the actual size measured. The predicted aggregate size was calculated from theoretical aggregate volume. Assumptions made were the size of the aggregate diameter being 10µm based on in house measurements and 64% random aggregate packing density based on empirical observations and simulation experiments [ 18 ]. Aggregate size measured on Day1 was used to calculate aggregate volume using equation: (1) V 1 = (4/3) · π · R 3 The aggregate volume was divided by the individual cell volume defined as (2) V c = (4/3) · π · (10/2) 3 = 523.6 The division was multiplied by the packing density to obtain cell number in an aggregate volume: (3) N 1 = V1/V c . 0.64 The predicted cell number of an aggregate after (3) days in culture was calculated from the cell growth rate measured from daily count (K) using the equation below: (4) N 3P = N 1 *exp(K*3) From the cell number, the predicted aggregate volume (V 3P ) was calculated again from the first equation. We defined the difference between the predicted and actual size as “Aggregate % Error”. This is calculated by subtracting the predicted from the actual average size measured and divided by the actual value: (5) Aggregate % Error = (V 3 -V 3P )/V 3 Flow cytometry and qPCR testing A 3 ml sample was taken from each bioreactor on day 4 of the culture for qPCR testing. RNA samples were dissolved in Trizol and extracted according to manufacturer’s protocol. Quantification of RNA was performed on epoch reader. A high-Capacity cDNA RT Kit was used for reverse transcription of RNA samples of each bioreactor. Triplicate samples of cDNA were obtained from triplicate samples of each bioreactor. Data collection was performed using QuantStudio for qPCR testing. The runs were performed per the manufacturer’s protocol and recommendations. The primers used were NANOG, SOX2, POU5F1 and GAPDH (Table S1 ). The data set obtained was then exported to Excel and normalized against corresponding housekeeping gene GAPDH. Final expression levels were expressed as 1/(2 Crt ) x 1000 and inputted into the MODDE software. A 3ml sample was taken from each bioreactor on day 4 of the culture for flow testing. The aggregates were dissociated with Accutase for 10 minutes into single cells. Cells were resuspended in PBS and divided into samples for intracellular staining and cells for extracellular staining. Sample for intracellular staining were fixed with a live/dead stain FVS 780, then permeabilized and stained for OCT4 and SOX2 using Attune™ Flow Cytometry (Table S2 ). Immunofluorescent staining hiPSC aggregates from day 4 were dissociated with Accutase and plated onto a Vitronectin treated 24 well plate and grown for 1 day. Cells were fixed with 5% PFA in DPBS for 15 minutes at room temperature, permeabilized and blocked with a blocking buffer solution (1% BSA and 0.1% Tween-20) for 60 minutes at room temperature and stained with primary antibodies at 1:500 (POU5F1, SSEA4, SOX2, NANOG) (Table S2 ) overnight. Cells were then washed with DPBS and stained with secondary antibodies (488-Donkey anti- Mouse and 594-Donkey anti-Rabbit) for 60 minutes. Cells were washed three times with DPBS (5minutes each) and stained with DAPI. Plates were imagined using EVOS M7000 Imaging System microscope. Statistical analysis Data were analyzed and graphed in Excel, R, GraphPad Prism9 and MODDE software. Modeling of design space was performed using MODDE software. Three replicated center point experiments with all five additives at mid-concentrations were used and split between the batches for normalization and model evaluation purposes. The model responses for each parameter were statistically evaluated for model reproducibility, model validity, fit (R2) and prediction precision (Q2). Comparisons were conducted via ANOVA test with a significant difference defined as P < 0.05. The number of bioreactor reactions was predetermined by the MODDE generator. The experiments were not randomized for blocks before testing. Results Using a DoE Approach to Optimize Pluripotent Aggregate Expansion and Cell State D-optimal DoE interaction designs were used to compress the number of experimental runs compared with a full factorial design. Design compression allows determination of interactions and predictions within the design space explored within the concentration range used (i.e., ‘‘known space’) [ 19 ]. Multiple high-molecular weight polymers have previously been shown to impact pluripotent stem cell cultures, but combinatorial assessments have not been performed to any large extent. Here, we investigated dextran sulfate (DS), Heparin sodium salt (HS), poly (vinyl alcohol) (PVA), Pluronic acid F68 (PA) and polyethylene glycol (PEG) (Fig. 1A). These factors and their concentration range were chosen based on previous publications about their impact on the iPSC aggregates growth and maintenance. Previous work showed that expansion of hPSCs in stirred-type bioreactors was improved by the addition of Pluronic F68 acting as non-ionic shear protectant [ 15 ]. PA has an average molecular weight of 8400 Da and is a triblock copolymer of poly (ethylene oxide)–poly (propylene oxide)–poly (ethylene oxide). PA was also demonstrated to restore cell growth and viability in cell cultures [ 20 ]. PVA is a biocompatible synthetic polymer with a molecular weight that ranges between 31,000–50,000. It is commonly used in the pharmaceutical and food industries [ 21 ]. In a recent study, researchers replaced serum albumin with PVA to develop a culture system that supports long-term expansion of functional mouse hematopoietic stem cells (HSCs) [ 22 ]. PVA was also evaluated in hiPSC culture to promote proliferation. At 1 mg/mL PVA resulted in maximal cell density within the range tested from 0.1 to 10 mg/ml [ 23 ]. PEG is a polymer based on the -CH2CH20- repeat unit that comes in different size ranges [ 24 ]. Similarly, to PVA, PEG is approved for drug and pharmaceutical applications. It functions as an aggregation inhibitor/emulsifying agent that may improve protein stability [ 16 ]. DS is a polysulphate that may increase aggregate stability and cell growth. Heparin sodium salt supports hiPSC growth due to FGF2 stabilization. FGF2 degrades relatively quickly at 37°C [ 17 , 25 ]. Dextran sulphate has been commonly used in the biopharmaceutical industry to prevent cell aggregation. It is a poly-sulphated compound and was recently reported to control aggregate size and shape properties of hiPSCs, without compromising the maintenance of pluripotency [ 5 , 26 ]. We generated D-optimal DoE interaction designs to evaluate these effectors. Cultures were maintained for four days in a suspension environment of a 100ml vertical wheel bioreactor (Fig. 1B). Reactors were initially seeded from 60–70% confluent hiPSCs pre-cultured on coated vitronectin plates [ 24 ] that were dissociated into single cells with TrypLE and seeded at 1.1E5 cells/ml in the presence of Rho kinase inhibitor Y-27632 for the first 24h. Daily aggregate samples were collected to obtain images. On the final day of culture, RNA was extracted and used in cDNA synthesis. Cell samples for flow, counts and plating were digested using Accutase and processed for FLOW analysis using cell state specific antibodies. Pluripotency data was assessed at the end of the experiment through Immunohistochemistry (IHC), flow cytometry and qPCR testing. The experiment was divided into two batches of experiments to be run at different times (Fig. 1C). Three replicated center point experiments with all five additives at mid-concentrations were used and split between the batches for normalization and model evaluation purposes. The optimization criteria were based on multiple attributes related to aggregate stability, cell growth and pluripotency (Fig. 1D). Mathematical models were generated and tuned for maximal predictive power and fit. The model responses for each parameter were statistically evaluated for model reproducibility, model validity, fit (R2) and prediction precision (Q2). All response variables measured had an R2 value above 0.5 indicating a model with high significance. The model had model validity above 0.25 indicating the absence of problems related to outliers and transformation issues (Fig. 1D). Inspection of the response models reveal insights on the degree of single/combinatorial control of key culture parameters. Invariably, coefficient listings of responses revealed that most significant predictors of culture behavior are contributed by factor interactions (Fig. 1E). All bioreactor runs were performed using E8 basal medium for pluripotency maintenance creating a suitable bioreactor growth assay in which the impact of additives upon iPSC cell aggregate stability and growth could be measured daily (Figure S2 A). The assay conditions were based on our previous work evaluating process parameters such as bioreactor seeding density, digestion frequency (Figure S2 B), single cell dissociation (Figure S2 C), passage time (Figure S2 D), plate coating, and bioreactor speed. Using the data obtained from the short-term growth assay (Figure S2 E), prediction models were constructed using computer modeling software. Response parameters included aggregate diameter sizes, cell concentration, growth rates, pluripotency marker expression, cell viability and aggregate variability. Each response model estimates main factor effects along with interactions between factors. Impact of Additives on Suspension Growth Kinetics Various cell culture additives have been used to protect cells from environments that involve agitation-aeration cell damage [ 15 , 20 , 27 – 29 ]. The additives selected in this design are components known to have a positive impact on growth and protection from shear stress. An optimizer setpoint dynamic profile was performed to detect main factor contributions (Fig. 2A) using doubling time data collected over 4 days of culture in bioreactors (Fig. 2B). Desirability criteria were set for the minimization of doubling time (i.e. maximal growth rate, Fig. 2C). Optimizer results detected read out conditions revealed that Pluronic F68, PVA and PEG all contribute to increased proliferation. The process capability index (Cpk) obtained from this optimizer was 0.9 (Fig. 2C) with a probability of failure of 1%. Modeling terms for maximal growth rate control include interaction terms between the additives. While interpreting and confirming a triple interaction can be complex, the obtained results from the model optimizer and coefficients plot confirm a combined effect of these three additives that is not simply additive through independent factor contributions (Fig. 1E) but rather there is a synergistic relationship observed contributing to the desired overall growth in this suspension system. hiPSC culture without daily media changes presented caveats related to impact of increased cell growth eventually plateauing due to nutrients limitation. Also, production of glycolytic waste products such as lactate and ammonia will reduce the medium pH, negatively impacting pluripotency, and continuous growth. For this purpose, the original seeding density of cells was at 1.1E5 cells/ml and the assay was revalidated at day 3 in addition to day 4 for many variables measured. Validation experiments compared the optimized medium to the control E8 medium without additives (Fig. 2D). The difference between the growth rates of the two mediums confirms the advantage of the additives (k = 0.4) over the control (k = 0.35). Optimization of Pluripotency Maintenance Attaining maximal growth conditions of iPSCs is not meaningful unless the iPSC state characteristics are also maintained. To demonstrate the maintenance of pluripotency, it was essential to test for the markers of undifferentiated iPSCs. At the end of the experiment, using flow cytometry we tested for SSEA4, OCT4, SOX2 and TRA-1-60 while a complimentary q-PCR assessment was conducted for NANOG, OCT4 and SOX2. The results of these experiments were used to model the effect of the additives on pluripotency. We optimized for the markers individually as shown in (Fig. 3A), since model results are non-confounding when optimization is performed for a singular response at a time. This is because that until proven that individual factors are not in regulatory conflict (i.e., SOX2, NANOG, or OCT4 being subject to differential control), single-response factor optimization is prudent. As we found, these factors are indeed subject to differential control. Model optimizers for SOX2 and OCT4 had PEG as a major contributor with factor contributions (FC) of 18.72 and 53.88 respectively. The model for SOX2 also contained a significant contribution from Pluronic F68 (FC = 15.28 and 3% probability of failure). Optimization of NANOG had a contribution of PEG and Pluronic F68 with FC of 7.16 and 15.44 respectively, however the addition of HS and DS had the highest contribution for NANOG optimization with FC of 21.12 and 48.04 respectively. Validation using these optimizers are shown in stained images for the pluripotency markers (Fig. 3B). Of the pluripotency markers measured, OCT4 expression was most strongly affected by tested media additives (Fig. 3C). Coefficients for all flow and qPCR markers (Fig. 3D) demonstrated that DS had opposing impact on OCT4 (negative) and NANOG (positive) making it challenging to select additive inputs that would satisfy all pluripotent markers. Upon validation, the common factor from all optimizers, PEG improved flow expression percentage of pluripotency markers compared to control media (Fig. 3E). Co-expression of pluripotency intracellular markers (SOX2 and OCT4) with the extracellular markers (SSEA4 and TRA-1-60) was analyzed using flow cytometry (Figure S3 and S4). Comparative data showed co-expression of intracellular markers (Fig. 4A) and extracellular co-expression (Fig. 4B). Models demonstrated a strong correlation of 89% between OCT4 and TRA-1-60 expression (Fig. 4C). A correlation this strong was not observed between any of the other pluripotency markers. Using model optimizers for the maximization, and minimization, of OCT4 expression confirmed that TRA-1-60 was coregulated with OCT4 expression (Fig. 4D). This finding shows a direct predictive connection between OCT4 and TRA-1-60 suggesting a coregulatory relationship. The model showed additional correlation between these pluripotency markers and the response variable of aggregate size. Further validation of this correlation was evaluated using two software tools, R (Fig. 4E) and MODDE (Figure C-D-F). A positive correlation of 76–85% was observed between increased diameter size and OCT4/TRA-1-60 expression respectively (Fig. 4E-F). Whereas a negative correlation was observed between increased diameter size and SOX2 (-46%) and NANOG (-50%) expression (Fig. 4E-F). Correlation results from both statistical tools showed that there is a significant correlation between aggregate size change and the expression of pluripotency markers with a P-value < 0.05, therefore the observed correlation is unlikely to occur by chance alone. Controlling Aggregate Architecture Using Media Additives Since aggregate size was demonstrated to impact pluripotency and its heterogeneity could cause inconsistencies in nutrients and oxygen intake by the cells, we sought to establish parameters qualifying the aggregate state. Ideally, optimal aggregate growth should occur through the formation of aggregates from single cells upon seeding, whereafter aggregate size growth should occur due to cell proliferation only. Large deviation in aggregate sizes could have two causes 1) unstable aggregates breaking apart due to shear forces and separate to become smaller or 2) aggregate fusion causing larger aggregates to form from the addition of smaller ones (Fig. 5A). Therefore, a parameter can be calculated and set for a desired range in an optimizer (Fig. 5B) using the data collected on cell proliferation and aggregate size throughout culture time. A violin plot of aggregate diameters in bioreactors on day 3 is shown (Fig. 5C) demonstrating that additives modify the distribution of the observed sizes. Aggregate sizes measured have a deviation and therefore modelling on average size could potentially mask that response output. To overcome this limitation of using the average aggregate size as a response variable, the average size was inputted in the model in addition to its corresponding standard deviation. Using the average aggregate size generated from day 1 from all reactions in the design with their corresponding growth rates, a predicted aggregate size was calculated for day 3 and compared to the observed size measured. We define this difference as “Aggregate % Error”. This is calculated by subtracting the predicted from the actual average size measured and divided by the observed. Using this derivative assay parameters, it was then possible to obtain a combination of components that would predict stable size limiting, hereby limiting aggregate fusion (maximization) or aggregate instability (minimization) (Fig. 5B). The aggregate diameter deviation was calculated from aggregate diameter measurements of images taken on each culture day using ImageJ software. A known source of aggregate size increase is the adhesion of two or more aggregates into a large one [ 6 ]. It is desired that the formation and growth of aggregates is only a contribution of cell proliferation and not fusion between cell spheres. Processes such as changing the media, taking samples, or anything that removes bioreactors from the base, can create static cultures that lead to an environment promoting clump formation. Our optimizer for the target aggregate % error (Fig. 5B), revealed that the main contributors to improving aggregate stability and decreasing aggregate size deviation in culture were HS (21.71) and PEG (17.48). Inspection of the aggregate diameters and growth slope showed a decrease in this optimization which is desirable as it indicates control over aggregation growth. The cell state control was also evaluated at this setpoint. NANOG expression was close to maximum indicating pluripotency maintenance upon aggregate control. When looking at a comparison of the aggregate % error within the bioreactors, the bioreactor with the minimum error and deviation value was a reaction that contains DS, PVA, PEG and HS (Fig. 5D-E). Whereas the maximum % error was reaction 19 that was a center point containing all additives (Fig. 5E). Validation comparing the optimizer results to the control with the E8 medium with no additives (Fig. 5D) shows that the aggregate size spread was minimized with the addition of Heparin and PEG. Using an optimizer model, we determined the combination of factors that satisfies an overall desired output of growth rate, pluripotency, and aggregate stability (Fig. 6A), not sacrificing one over another. The results suggested that the combination of Heparin sodium salt and PEG can be added to maintain aggregate stability, pluripotency, and growth aspects simultaneously. Validation against an E8 control was done and maintenance of pluripotency (Fig. 6B, C), cell proliferation (Fig. 6D), and aggregate stability (Fig. 6E) were specifically addressed. After successful validation, the optimizer model was tested on an additional cell line RCRP005N as compared to NCRM-1 with a decreased bioreactor speed from 60 to 40 RPM. Both cell lines showed continuous aggregate growth (Figure S5A) even after passaging by reseeding 30 million cells as aggregates on the fourth day of culture. The cells maintained their pluripotency markers OCT4 and SOX2 (91.8% and 97.8% for RCRP5005N and 97.4% and 98.4% for NCRM-1 respectively) (Figure S5B). Cell growth rate was steady after serial passaging without dissociation (Figure S5C) with a doubling time of (1.222–1.434 for RCRP5005N and 1.025–1.293 for NCRM-1 before and after passaging respectively). Results validated the robustness and repeatability of the predicted model. The Aggregate stability parameter can indirectly provide pluripotency maintenance and growth throughout the assay without impacting cell viability (Figure S5D). Therefore, this parameter can be regarded as a critical attribute that has a significant impact on iPSC cultures in a suspension environment. Discussion Many efforts at upscaling hiPSC cultures utilize high-cost commercial medium with proprietary media formulations. While adherent cultures predominantly use E8 medium, optimizing this culture medium to sustain a healthy and consistent aggregate culture is needed. Environmental and process stress such as hydrodynamic stress, single cell seeding, agitation and media changes are fundamentally different from adherent culture conditions. Here we empirically defined combinatorial media formulations that directly influence pluripotency, growth and aggregate stability using a five-dimensional DoE based approach. In this study, the two main components that positively influenced the majority of identified critical attributes were Heparin sodium salt and PEG. Interactions between HS and PEG were well modeled, predicted, and enabled a better understanding of controlling aggregate stability. While together improving iPSC aggregate culturing, PEG and Heparin sodium salt have different effects on cellular aggregates. PEG is known to induce cell aggregation and promote the fusion of larger cell aggregates at higher concentrations (10% and above), while also capable of inhibiting aggregation in other applications at lower concentrations [ 16 ]. Heparin sodium salt is known for its ability to interact with different growth factors such as FGF2 influencing cell behavior and pluripotency. Being able to detect this synergy within compounds that have a different effect on the culture reinforces the strength of the Design of Experiments approached used in this study. The only compounds used throughout this study that did not increase ‘aggregate % error’ were PEG and Heparin sodium salt. Modelling PEG indicated the overall effect was a reduced standard deviation of the aggregate diameter size, though no effect on doubling time was observed. The aggregate deviation and diameter growth rate increased slightly in the presence of PEG. While the addition of Heparin sodium salt did not show a strong impact on aggregate size or growth rate individually it was shown to have strong positive interaction in the presence of PEG. Using DoE-based models provide several advantages over conventional one-factor-at-a-time methods. DoE modelling has embedded insights about model validity, reproducibility, fit (R2), and prediction precision (Q2) for each response variable measured allowing for an unbiased assessment of the overall model quality, all while identifying critical interactions. For example, all response variables measured throughout this series of experiments had an R2 value greater than 0.5, ensuring that models evaluated are highly significant. Models were also scrutinized for model validity and were only being considered significant when > 0.25, indicating an absence of outliers and transformation issues. As large-scale expansion of iPSCs and their derivatives become more common, DoE, as part of a Quality-by-Design strategy, should be used as a tool to evaluate multiple parameters impacting the overall cell quality and process stability. Embedding designs at increasing dimensional scales can be informative for the robust statistical analysis and unbiased optimization strategies needed when entering larger scale. Here, the modeling and design was performed within a low-shear reactor environment that is scalable. As there are multiple parameters involved in large scale manufacturing and transferring adherent culture conditions/protocols to suspension culture, more efficient assays are needed that do not require long testing periods or sequential optimizations strategies lacking the ability to evaluate combinatorial interactions. Specifically, we were able to define unexpected combinations of media additives for controlling aggregate formation within a bioreactor system. We also obtained a rich data response space, where mathematical models included doubling time, aggregate deviation, aggregate growth slope, pluripotency markers and cell count measurements. Using this methodical approach many novel insights into the compounds used in this study were discovered. DS, PVA and Pluronic F68 all had similar effects. Previous research showed that DS prevents cellular aggregation and has the ability to control aggregate size and growth properties [ 5 , 26 ]. We could not replicate these observations. Set point analysis (Figure S6) showed that the largest deviation of ‘aggregate % error’ (Fig. 5A) was obtained using DS. In agreement with this observation modelling for maximal ‘aggregate % error’ resulted in a formulation predominantly containing DS (FC = 32.90) (Fig. 5B). Similarly, previous studies evaluating PVA effect on hiPSCs demonstrated that it promoted proliferation [ 23 ]. In contrast, set point analysis (Figure S6A) showed PVA impacts an increased aggregate diameter and the overall daily aggregate growth which is only desirable if these observed growth parameters occur solely due to cell proliferation. However, looking at the parameter of aggregate deviation (Figure S6A), indicates this PVA mediated increase was due to a heterogeneous effect on aggregate size. Collaborative evidence is provided by the fact that PVA had no effect on either growth rate or doubling time. A scenario repeated with Pluronic F68. Previous reports have shown that the inclusion of Pluronic F68 within a bioreactor increased the overall expansion capacity of hPSCs [ 15 ]. Here we show through a set point analysis that Pluronic F68 increased overall aggregate size and the deviation in the aggregate sizes (Figure S6A) again without effecting the aggregate growth slope, doubling time or cell concentration. Indicating again that this increase was due to fusion events between aggregates and not increased proliferation. An explanation for these discrepancies between previous studies is that others have compared individual reagents against controls lacking the additives, whereas in this study computer modelling allowed for the direct comparison of the compounds being evaluated. It should also be noted that Manstein et al used Pluronic F68 so that RPM could be increased, an increased RPM would have countered the fusion effect. Overall, it is indicated here that fusion has a detrimental effect on cellular proliferation, potentially because as aggregates become larger nutrient availability is increasingly restrictive, actively inhibiting a proliferative state. Essential 8 was used as the basal media for all reactions throughout this series of experiments and is therefore considered responsible for the general maintenance of pluripotency within the bioreactors. However, the compounds evaluated in this study did influence this maintenance. The provision of DS, Pluronic F68 or PVA all had an overall negative influence on pluripotent maintenance promoting the down regulation NANOG and SOX2 expression, though a slightly positive effect on OCT4 expression was shown. In contrast, PEG and Heparin sodium salt had a general positive effect on pluripotent maintenance showing a positive correlation to OCT4 and NANOG expression, while having differential influence on SOX2 expression with Heparin sodium salt decreasing its expression and PEG increasing SOX2 levels (Figure S6A). In general, we could not find strong correlations between the core pluripotency markers OCT4, SOX2, and NANOG (Fig. 4C). A possible explanation is that they have divergent roles functioning as master regulators of lineage-specific differentiation as others have suggested. Previous studies have shown that the over expression of OCT4 leads to endodermal and mesodermal fates [ 30 ], whereas the overexpression of SOX2 generates ectodermal fate [ 31 ]suggesting that driving the maximal expression of either would result in the down regulation of the other. Modelling for minimal or maximal OCT4 expression indicated that the only pluripotent marker assayed within this study that followed OCT4 expression was TRA-1-60 (a positive correlation of 0.89), though notably aggregate diameter correlated with both pluripotency markers. This model was confirmed using R Software (Fig. 4E), which indicated a strong correlation between the increase of the diameter size and the expression of OCT4 and TRA-1-60 markers (0.790, 0.816) with a high significance level (P-value < 000.1). This suggests that aggregate size could be used to influence differentiation potential, specifically here that endoderm differentiation may benefit from a population of larger aggregate. This could also explain why compounds that promoted fusion selectively lose SOX2 and NANOG expression while showing a small increase in OCT4 expression. One of the major efforts during this study was the development of quantification methods for aggregate stability with the overall goal of limiting aggregate fusion. Measurements of aggregate size deviation (range of sizes) could determine the overall homogeneity of aggregate size but could not directly inform on fusion since differential growth rates between aggregates and aggregate fractioning due to shear stress are alternative explanations. So, to better define what was occurring on the individual aggregate level a method for quantifying this aggerate parameter was developed. Resting on a theoretical estimation of the rate aggregate size should change according to growth rates observed, we directly compared this to measured size changes observed within the bioreactors. This was defined as ‘aggregate % error’. We found that by modelling using this parameter we could gain a better control over the size and overall homogeneity of the aggregates within bioreactors. Models setting the ‘aggregate % error’ to the predicted diameter (theoretical size based on doubling time) would define conditions that lacked fusion but in which aggregates were stable enough to resist shear forces. While this model proved useful in measuring overall aggregate stability there are inherent limitations to it. The mathematical model assumes cell size uniformity, a parameter likely to change based on the density of the individual aggregates or differentiation state of the cell culture. There is also an assumption based on empirical observations and simulation experiments that the cells are randomly packed. An alternative method for estimating aggregate stability would have been directly measuring aggregate numbers daily, decreasing aggregate number with an increasing aggregate size could directly indicate fusion. Here we show that models quantifying aggregate behavior based on calculations of theoretical predictions as compared to empirical results are crucial for the identification and understanding of the complexities in a biological system. Having predictive models allows for formulating a hypothesis about the underlying process mechanisms and therefore facilitates optimization by identifying key factors and parameters that bring the model to desired results. Conclusion For each model the relationship between additive/response parameter could be analyzed and optimized towards desirability criteria. In all desirability contexts, process capability (Cpk) as a measurement of process robustness was favorable. Optimal conditions for maximal cell growth required combinatorial additive use. Similarly, optimization of the maintenance of pluripotency, as based on markers OCT4, NANOG, SOX2 was also dependent on combinatorial signaling. Based on measurements of aggregate size and size variance, optimization could be performed to increase aggregate homogeneity. In all cases, underlying models included multiple interaction terms, revealing the criticality of simultaneous testing using combinatorial designs. It was demonstrated that the combinatorial effects of Polyethylene Glycol and Heparin Sodium Salt maintain growth and pluripotency and control aggregate stability by limiting fusion events between aggregates in suspension cultures. We conclude that DoE-based interaction testing performed within a manufacturing-relevant environment allows for process understanding of the biomanufacturing process. The method identifies critical process parameters; their interacting criticality, while returning a deep process understanding. Abbreviations HS: Heparin sodium salt PEG: polyethylene glycol PVA: poly (vinyl alcohol) PA: Pluronic F68 DS: dextran sulfate E8: Essential 8 DoE: Design of experiments 3D: Three-dimensional cGMP: Current Good manufacturing practice hiPSC: Human Induced pluripotent stem cell PBS: Phosphate buffer solution q-PCR: Quantitative PCR RPM: Revolution per minute ROCK: Rho-Kinase Declarations Acknowledgements We would like to thank Trailhead Biosystems for funding this research. The authors also thank Michael Majcher and Katie Sears for their critical reading of the manuscript and Cleveland State University for their support. Author contributions M.B conceived and supervised the project. H.Y. designed, conducted all experiments and wrote the manuscript. S.R. cultured the starting iPSC material for the validation. J.J. provided review feedback. All authors approved the manuscript. Funding The work was supported and funded by Trailhead Biosystems Inc. No external funding was used. Data availability Requests for further information or more detailed protocols should be directed to and will be fulfilled by the corresponding author. This study did not generate new unique reagents. The data that support the findings of this study are available on request. Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing Interests J.J. is founder of and shareholder of Trailhead Biosystems, Inc., Beachwood, OH, USA. M. B. is a shareholder in Trailhead Biosystems, Inc., Beachwood OH. This work has been filed as US Provisional Application No. 63/525,083. References Takahashi K, Yamanaka S. Induction of Pluripotent Stem Cells from Mouse Embryonic and Adult Fibroblast Cultures by Defined Factors. Cell. 2006;126:663–76. Fan Y, Wu J, Ashok P, Hsiung M, Tzanakakis ES. Production of Human Pluripotent Stem Cell Therapeutics under Defined Xeno-free Conditions: Progress and Challenges. Stem Cell Rev Rep. 2015;11:96–109. Deinsberger J, Reisinger D, Weber B. Global trends in clinical trials involving pluripotent stem cells: a systematic multi-database analysis. NPJ Regen Med. 2020;5:15. Colter J, Murari K, Biernaskie J, Kallos MS. Induced pluripotency in the context of stem cell expansion bioprocess development, optimization, and manufacturing: a roadmap to the clinic. NPJ Regen Med. 2021;6:72. Nogueira DES, Rodrigues CAV, Carvalho MS, Miranda CC, Hashimura Y, Jung S, et al. Strategies for the expansion of human induced pluripotent stem cells as aggregates in single-use Vertical-Wheel™ bioreactors. J Biol Eng. 2019;13:74. Otsuji TG, Bin J, Yoshimura A, Tomura M, Tateyama D, Minami I, et al. A 3D Sphere Culture System Containing Functional Polymers for Large-Scale Human Pluripotent Stem Cell Production. Stem Cell Reports. 2014;2:734–45. Abbasalizadeh S, Larijani MR, Samadian A, Baharvand H. Bioprocess Development for Mass Production of Size-Controlled Human Pluripotent Stem Cell Aggregates in Stirred Suspension Bioreactor. Tissue Eng Part C Methods. 2012;18:831–51. Liu M. Engineering stem cell niches in bioreactors. World J Stem Cells. 2013;5:124. Teramura T, Takehara T, Onodera Y, Nakagawa K, Hamanishi C, Fukuda K. Mechanical stimulation of cyclic tensile strain induces reduction of pluripotent related gene expressions via activation of Rho/ROCK and subsequent decreasing of AKT phosphorylation in human induced pluripotent stem cells. Biochem Biophys Res Commun. 2012;417:836–41. Saha S, Ji L, de Pablo JJ, Palecek SP. TGFβ/Activin/Nodal Pathway in Inhibition of Human Embryonic Stem Cell Differentiation by Mechanical Strain. Biophys J. 2008;94:4123–33. Kim YM, Kang YG, Park SH, Han M-K, Kim JH, Shin JW, et al. Effects of mechanical stimulation on the reprogramming of somatic cells into human-induced pluripotent stem cells. Stem Cell Res Ther. 2017;8:139. Goetzke R, Sechi A, De Laporte L, Neuss S, Wagner W. Why the impact of mechanical stimuli on stem cells remains a challenge. Cell Mol Life Sci. 2018;75:3297–312. Xu C, Rosler E, Jiang J, Lebkowski JS, Gold JD, O’Sullivan C, et al. Basic Fibroblast Growth Factor Supports Undifferentiated Human Embryonic Stem Cell Growth Without Conditioned Medium. Stem Cells. 2005;23:315–23. Kuo H-H, Gao X, DeKeyser J-M, Fetterman KA, Pinheiro EA, Weddle CJ, et al. Negligible-Cost and Weekend-Free Chemically Defined Human iPSC Culture. Stem Cell Reports. 2020;14:256–70. Manstein F, Ullmann K, Kropp C, Halloin C, Triebert W, Franke A, et al. High Density Bioprocessing of Human Pluripotent Stem Cells by Metabolic Control and in Silico Modeling. Stem Cells Transl Med. 2021;10:1063–80. Castellanos IJ, Crespo R, Griebenow K. Poly(ethylene glycol) as stabilizer and emulsifying agent: a novel stabilization approach preventing aggregation and inactivation of proteins upon encapsulation in bioerodible polyester microspheres. J controlled release : official J Controlled Release Soc. 2003;135–45. Chen G, Gulbranson DR, Hou Z, Bolin JM, Ruotti V, Probasco MD, et al. Chemically defined conditions for human iPSC derivation and culture. Nat Methods. 2011;8:424–9. Wilken S, Guerra RE, Levine D, Chaikin PM. Random Close Packing as a Dynamical Phase Transition. Phys Rev Lett. 2021;127:038002. Bukys MA, Mihas A, Finney K, Sears K, Trivedi D, Wang Y, et al. High-Dimensional Design-Of-Experiments Extracts Small-Molecule-Only Induction Conditions for Dorsal Pancreatic Endoderm from Pluripotency. iScience. 2020;23:101346. Hu W, Rathman JJ, Chalmers JJ. An investigation of small-molecule surfactants to potentially replace pluronic F-68 for reducing bubble-associated cell damage. Biotechnol Bioeng. 2008;101:119–27. Koski A, Yim K, Shivkumar S. Effect of molecular weight on fibrous PVA produced by electrospinning. Mater Lett. 2004;58:493–7. Wilkinson AC, Ishida R, Kikuchi M, Sudo K, Morita M, Crisostomo RV, et al. Long-term ex vivo haematopoietic-stem-cell expansion allows nonconditioned transplantation. Nature. 2019;571:117–21. Tang X, Wu H, Xie J, Wang N, Chen Q, Zhong Z et al. The combination of dextran sulphate and polyvinyl alcohol prevents excess aggregation and promotes proliferation of pluripotent stem cells in suspension culture. Cell Prolif. 2021;54. Braam SR, Zeinstra L, Litjens S, Ward-van Oostwaard D, van den Brink S, van Laake L, et al. Recombinant Vitronectin Is a Functionally Defined Substrate That Supports Human Embryonic Stem Cell Self-Renewal via αVβ5 Integrin. Stem Cells. 2008;26:2257–65. Furue MK, Na J, Jackson JP, Okamoto T, Jones M, Baker D et al. Heparin promotes the growth of human embryonic stem cells in a defined serum-free medium. Proceedings of the National Academy of Sciences. 2008;105:13409–14. Lipsitz YY, Tonge PD, Zandstra PW. Chemically controlled aggregation of pluripotent stem cells. Biotechnol Bioeng. 2018;115:2061–6. Chisti Y. Animal-cell damage in sparged bioreactors. Trends Biotechnol. 2000;18:420–32. Papoutsakis E. Media additives for protecting freely suspended animal cells against agitation and aeration damage. Trends Biotechnol. 1991;9:316–24. Goldblum S, Bae YK, Hink WF, Chalmers J. Protective effect of methylcellulose and other polymers on insect cells subjected to laminar shear stress. Biotechnol Prog. 1990;6:383–90. Niwa H, Miyazaki J, Smith AG. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat Genet. 2000;24:372–6. Kopp JL, Ormsbee BD, Desler M, Rizzino A. Small Increases in the Level of Sox2 Trigger the Differentiation of Mouse Embryonic Stem Cells. Stem Cells. 2008;26:903–11. Supplementary Files Supplementalmaterials.pdf Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 20 Nov, 2023 Reviewers invited by journal 20 Nov, 2023 Editor assigned by journal 29 Oct, 2023 First submitted to journal 27 Oct, 2023 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3500780","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":251591015,"identity":"0cd1250d-e9d7-4fbc-8e1c-106797933d21","order_by":0,"name":"Haneen Yehya","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA00lEQVRIiWNgGAWjYBACAwYGNiBlA8SMjYdBXGK1pEkAtTSQpOWwBIhzmCiHmUskH3vMU3O+zrz9cMPhggIbBv727gS8WixnpKUb8xy7LSFzJrHh8AyDNAaJM2c34HfYjRwzaR622xISDEAtPAZA70jkEtKS/02a5985CQn+h0RryWGT5m07ICEhQbQtZ56ZSc7tS5acIfEQ7Bcewn45nvxM4s03O34J/vSHjwv+2Mjxt/fi18IgkMDAxIPE58GpEg74DzAw/iCsbBSMglEwCkYyAAD5fkd6+Ln7dAAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0009-0006-6117-7442","institution":"Cleveland State University","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Haneen","middleName":"","lastName":"Yehya","suffix":""},{"id":251591016,"identity":"54792247-4dfc-4864-bef8-d56abccd83d4","order_by":1,"name":"Sofija Raudins","email":"","orcid":"","institution":"Trailhead Biosystems","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Sofija","middleName":"","lastName":"Raudins","suffix":""},{"id":251591017,"identity":"e9ca9fae-e4d0-4ba8-a893-3b183207e630","order_by":2,"name":"Jan Jensen","email":"","orcid":"","institution":"Trailhead Biosystems Inc.","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jan","middleName":"","lastName":"Jensen","suffix":""},{"id":251591018,"identity":"08f7bbd3-a0da-4f1d-bcc4-a5557e7b37e6","order_by":3,"name":"Michael A. Bukys","email":"","orcid":"","institution":"Trailhead Biosystems Inc.","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Michael","middleName":"A.","lastName":"Bukys","suffix":""}],"badges":[],"createdAt":"2023-10-27 21:44:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3500780/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3500780/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":47001091,"identity":"4807956a-e790-4b83-8779-615771f36ead","added_by":"auto","created_at":"2023-11-24 00:05:47","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":327000,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version.\u003c/p\u003e","description":"","filename":"fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/c1756b8931f8ee6e751fe99c.png"},{"id":47001090,"identity":"4bfc2668-e39c-4126-a932-42b37f5a3e0f","added_by":"auto","created_at":"2023-11-24 00:05:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":181541,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version.\u003c/p\u003e","description":"","filename":"fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/9518dc5ac6b2e5d5ab87c6ea.png"},{"id":47000436,"identity":"8ddc4bdd-e83f-4512-b17d-663d89060bfd","added_by":"auto","created_at":"2023-11-23 23:57:47","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1216143,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version.\u003c/p\u003e","description":"","filename":"fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/6b5063ddabdf342468730080.png"},{"id":47000434,"identity":"b0575c9f-bd58-416f-9bce-1129d0fabc18","added_by":"auto","created_at":"2023-11-23 23:57:47","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":177747,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version.\u003c/p\u003e","description":"","filename":"fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/bc833b455484bf0171b35128.png"},{"id":47001487,"identity":"984ec6d1-45c5-4a80-b9ea-23b653dbc3a0","added_by":"auto","created_at":"2023-11-24 00:13:47","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":904838,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version.\u003c/p\u003e","description":"","filename":"fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/50592b049248207457d9ae5a.png"},{"id":47001092,"identity":"f7831889-3428-495e-a6f8-8de447ac875a","added_by":"auto","created_at":"2023-11-24 00:05:47","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1455819,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version.\u003c/p\u003e","description":"","filename":"fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/e979ce43a25bfcc97b50d6af.png"},{"id":47001753,"identity":"2595a733-2029-44bb-9636-5bd5ef902c91","added_by":"auto","created_at":"2023-11-24 00:21:48","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2959482,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/990d208a-0baf-420a-905c-1297d509ffaa.pdf"},{"id":47000439,"identity":"a0ba3045-4bfd-47aa-b249-3310fde0b857","added_by":"auto","created_at":"2023-11-23 23:57:47","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1354811,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementalmaterials.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3500780/v1/e4ecf0ade0799120412b128e.pdf"}],"financialInterests":"","formattedTitle":"Addressing Bioreactor hiPSC Aggregate Stability, Maintenance and Scaleup Challenges Using a Design of Experiment Approach","fulltext":[{"header":"Background","content":"\u003cp\u003eHuman pluripotent stem cells (hPSCs) can proliferate continuously and have the capacity to differentiate into any human cell type. Upon differentiation, this unlocks their potential to treat various clinical indications such as diabetes, orthopedic injuries, neurological, cardiovascular diseases, and others by supplying cells needed to restore tissue function. Induced pluripotent stem cells (iPSCs) or reprogrammed cells generated from mature specialized cells [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], are becoming the focus in the field of regenerative medicine, as they are not restricted by ethical concerns of embryonic stem cells but display a similar differentiation and growth capacity as previously used embryonic stem (ES) cells. While no current FDA-approved iPSC cell therapy exists, many are in clinical trials. Therapeutic applications of iPSCs in regenerative medicine typically depend on the availability of 10\u003csup\u003e8\u003c/sup\u003e-10\u003csup\u003e10\u003c/sup\u003e clinical grade cells per patient, where such material is manufactured using a current Good Manufacturing Practice (cGMP) process and performed within an isolated production environment to assure patient safety [\u003cspan additionalcitationids=\"CR3\" citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eConventional adherent culture flasks are impractical for large scale production as they require frequent manual intervention. Adherent cultures have a limited ability to scale up and result in high batch-to-batch variability and a lack of cost effectiveness. Alternatively, continuously stirred-tank suspension bioreactors provide an hiPSC 3D culture that generates aggregates that are more biologically comparable to an \u003cem\u003ein vivo\u003c/em\u003e environment [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. In addition, bioreactors can provide a continuous monitoring of environmental factors including temperature, pH, oxygen, and nutrients in a sterile environment required to produce clinical grade cells. Maintaining pluripotent state in large-scale suspension environment will improve reproducibility and cell quality, but it requires control over media composition, aggregation and physicochemical stresses exerted on cells.\u003c/p\u003e \u003cp\u003eiPSC expansion in bioreactors presents several challenges including cell clumping, shear stress, complex media composition and cost. Cell clumping or aggregate fusion can cause heterogenous cell populations to arise [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Shear stress leads to aggregate breakage, DNA-breaks, resulting in cell death and karyotypic instability [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. An environment that can prevent unwanted cell adhesion and maintain aggregate stability while maintaining pluripotency for manufacturing expansion is in demand. Cell State control is particularly relevant, as heterogeneous populations presents a serious risk, as incomplete differentiation impacts both clinical safety and potency [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. While simple elements such as temperature and pH can be controlled and monitored, molecular component interactions, concentration sensitivity, chemical stability of additives, when combined with physical stresses, mechanical forces present all combine into a very challenging production problem; this also presents an opportunity to identify which key critical process parameters truly govern culture performance [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. For instance, although recent studies confirm that mechanical stimuli affect iPSCs during their differentiation [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], mechanosensing signaling is not well understood especially in terms of stem cell mechanobiological interactions within the physical environment, of a bioreactor [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eA critical challenge to generating well-controlled studies assessing higher-order interactions is characterizing the complex mixture of variable parameters that affect the molecular and functional response of iPSCs [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Over the last 18 years, exhaustive optimization has been done by multiple groups mainly using various adherent hiPSC culture mediums. This has led to the identification of FGF2[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] and TGFb1 as the two critical growth factors in the Essential 8 (E8) medium formulation. E8 is a further modification identifying the essential components from the original TeSR medium which pioneered early pluripotency studies. Through an intermediary B8 medium which is a cost reduced optimization from the E8 medium [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Currently, there is broad focus on determining defined, stable, cost-effective solutions for iPSC culture maintenance and growth. One of the main improvements in the E8 formulation is the absence of bovine serum albumin (BSA) component, since it is a xeno component that has issues with consistencies. Pairwise dropout testing done in the E8 optimization effort uncovered that β-mercaptoethanol (BME) without BSA caused toxicity, but with the absence of BME, BSA was able to be eliminated from the medium. This finding highlights the complexity of doing optimization experiments without studying the interactions and non-linear effects. A key factor in obtaining well-founded results is understanding interaction effects and reducing the variability of environmental factors in the process [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Limited research evaluated the media needs for suspension culture that can address the shear stress generated from a bioreactor and the absence of substrate attachment in a 3D environment.\u003c/p\u003e \u003cp\u003eHere, we demonstrate a systematic approach maintaining iPSC aggregates in a uniform controlled stable size within a bioreactor. Heparin sodium salt (HS), polyethylene glycol (PEG), poly (vinyl alcohol) (PVA), Pluronic F68 (Pluronic F68) and dextran sulfate (DS) are compounds commonly used in biomedical and pharmaceutical applications. They have versatile properties such as reducing shear stress by decreasing surface tension of the media, enhancing extracellular matrix and cell membrane interaction, increasing aggregate stability, and preventing cell fusion [\u003cspan additionalcitationids=\"CR16\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Using a design of experiment (DoE) factorial study we develop process understanding of the complex interactions between these components. The DoE approach identifies statistically significant interactions resulting in a media formulation that is reproduceable by limiting variation bias. Models generated using these 5 factors can be optimized for three major cell culture criteria: (1) growth rate or doubling time, (2) pluripotency maintenance, and (3) aggregate stability. These parameters are shown to be independent, and optimization for each attribute is related to different culture conditions. Consequently, attaining ideal culture conditions across the desirability criteria spectrum is an exercise based on compromise.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eGenerating DoE designs and perturbation matrixes\u003c/h2\u003e\n \u003cp\u003eThe DoE runs were computer generated from MODDE software (Sartorius Stedim Data Analytical Solutions, SSDAS) using D-optimal interaction designs. All components tested and outputs measured were manually inputted into Design Wizard in MODDE. Factors known to impact the aggregate stability in suspension culture were chosen for the design. Reagent concentration ranges were based on previous publications. Design runs were chosen to have 16 reactions conditions with the addition of 3 center point conditions. One of the reactions had no additives, which is the E8 bioreactor. Examples of reaction additives of different bioreactors are shown in (Figure \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003eA-B). The highest G-efficiency design was chosen from the DoE designs generated by the software. Media reactions were made manually and filtered before use. Negative log transform was only used to normalize distribution for flow data sets for OCT4 and TRA-1-60 to meet normality assumptions. Assay and media preparation along with detailed methods are described in the supplemental information. qPCR primers, antibodies and reagents used are listed in (Table \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e and S2).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\n \u003ch2\u003eHuman induced pluripotent stem cell culture\u003c/h2\u003e\n \u003cp\u003eCells were maintained at 37\u0026deg;C and 5% CO2. E8 medium was purchased ready in solution. Other medium components were purchased as powder. Prior to the assay, cells were grown to 70\u0026ndash;80% confluence after 4 days of culture on vitronectin coated 6 well plates. Cells were dissociated with TrypLE, for 3 minutes at 37\u0026deg;C and resuspended in E8 medium, transferred to 50 ml conical tubes, and centrifuged at 400 x g for 6 minutes. The pellet was resuspended in E8 and 10 \u0026micro;M Y-27632 ROCK inhibitor. 11\u0026nbsp;million cells were seeded in a 100ml bioreactor with the corresponding mediums.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003eCell counting and aggregate size\u003c/h2\u003e\n \u003cp\u003eSamples from the bioreactor were taken after mixing using a pipette to avoid sampling bias and gradient formation after settling. 3 ml total was sampled daily for 3 samples of 1ml for cell count. The cells were first dissociated using Accutase and incubated for 10 minutes. Cells were then quenched with E8 then centrifuged at 400 x g for 6 minutes. The pellet was resuspended in the same volume of PBS. Samples were then analyzed for total cell count using Attune flow. Triplicate counts were collected for each bioreactor for each day of culture. Duplicate samples were collected for aggregate imaging of 500ul and put on a 24 well plate. EVOS 7000 was used for bright field images of the aggregates. ImageJ was then used to analyze aggregate size and distribution. A minimum of 30 aggregates were measured for each bioreactor every day of the culture. This data was then analyzed for standard deviation on excel and growth rate using GraphPad Prism9. The cell counts were also analyzed for doubling time using GraphPad Prism. Average data from aggregate size and counts were then inputted into the response tab in MODDE. Using the aggregate size generated from day 1 from all bioreactors and their corresponding cell growth rate, a predicted aggregate size was calculated for day 3 and compared to the actual size measured. The predicted aggregate size was calculated from theoretical aggregate volume. Assumptions made were the size of the aggregate diameter being 10\u0026micro;m based on in house measurements and 64% random aggregate packing density based on empirical observations and simulation experiments [\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e\n \u003cp\u003eAggregate size measured on Day1 was used to calculate aggregate volume using equation:\u003c/p\u003e\u003cspan\u003e\n \u003cp\u003e(1) V\u003csub\u003e1\u003c/sub\u003e= (4/3) \u0026middot; \u0026pi; \u0026middot; R\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\n \u003c/span\u003e\n \u003cp\u003eThe aggregate volume was divided by the individual cell volume defined as\u003c/p\u003e\u003cspan\u003e\n \u003cp\u003e(2) V\u003csub\u003ec\u003c/sub\u003e = (4/3) \u0026middot; \u0026pi; \u0026middot; (10/2)\u003csup\u003e3 =\u003c/sup\u003e 523.6\u003c/p\u003e\n \u003c/span\u003e\n \u003cp\u003eThe division was multiplied by the packing density to obtain cell number in an aggregate volume:\u003c/p\u003e\u003cspan\u003e\n \u003cp\u003e(3) N\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;V1/V\u003csub\u003ec\u003c/sub\u003e. 0.64\u003c/p\u003e\n \u003c/span\u003e\n \u003cp\u003eThe predicted cell number of an aggregate after (3) days in culture was calculated from the cell growth rate measured from daily count (K) using the equation below:\u003c/p\u003e\u003cspan\u003e\n \u003cp\u003e(4) N\u003csub\u003e3P\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;N\u003csub\u003e1\u003c/sub\u003e*exp(K*3)\u003c/p\u003e\n \u003c/span\u003e\n \u003cp\u003eFrom the cell number, the predicted aggregate volume (V\u003csub\u003e3P\u003c/sub\u003e) was calculated again from the first equation. We defined the difference between the predicted and actual size as \u0026ldquo;Aggregate % Error\u0026rdquo;. This is calculated by subtracting the predicted from the actual average size measured and divided by the actual value:\u003c/p\u003e\u003cspan\u003e\n \u003cp\u003e(5) Aggregate % Error = (V\u003csub\u003e3\u003c/sub\u003e-V\u003csub\u003e3P\u003c/sub\u003e)/V\u003csub\u003e3\u003c/sub\u003e\u003c/p\u003e\n \u003c/span\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\n \u003ch2\u003eFlow cytometry and qPCR testing\u003c/h2\u003e\n \u003cp\u003eA 3 ml sample was taken from each bioreactor on day 4 of the culture for qPCR testing. RNA samples were dissolved in Trizol and extracted according to manufacturer\u0026rsquo;s protocol. Quantification of RNA was performed on epoch reader. A high-Capacity cDNA RT Kit was used for reverse transcription of RNA samples of each bioreactor. Triplicate samples of cDNA were obtained from triplicate samples of each bioreactor. Data collection was performed using QuantStudio for qPCR testing. The runs were performed per the manufacturer\u0026rsquo;s protocol and recommendations. The primers used were NANOG, SOX2, POU5F1 and GAPDH (Table \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e). The data set obtained was then exported to Excel and normalized against corresponding housekeeping gene GAPDH. Final expression levels were expressed as 1/(2\u003csup\u003eCrt\u003c/sup\u003e) x 1000 and inputted into the MODDE software. A 3ml sample was taken from each bioreactor on day 4 of the culture for flow testing. The aggregates were dissociated with Accutase for 10 minutes into single cells. Cells were resuspended in PBS and divided into samples for intracellular staining and cells for extracellular staining. Sample for intracellular staining were fixed with a live/dead stain FVS 780, then permeabilized and stained for OCT4 and SOX2 using Attune\u0026trade; Flow Cytometry (Table \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n \u003ch2\u003eImmunofluorescent staining\u003c/h2\u003e\n \u003cp\u003ehiPSC aggregates from day 4 were dissociated with Accutase and plated onto a Vitronectin treated 24 well plate and grown for 1 day. Cells were fixed with 5% PFA in DPBS for 15 minutes at room temperature, permeabilized and blocked with a blocking buffer solution (1% BSA and 0.1% Tween-20) for 60 minutes at room temperature and stained with primary antibodies at 1:500 (POU5F1, SSEA4, SOX2, NANOG) (Table \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e) overnight. Cells were then washed with DPBS and stained with secondary antibodies (488-Donkey anti- Mouse and 594-Donkey anti-Rabbit) for 60 minutes. Cells were washed three times with DPBS (5minutes each) and stained with DAPI. Plates were imagined using EVOS M7000 Imaging System microscope.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003eStatistical analysis\u003c/h2\u003e\n \u003cp\u003eData were analyzed and graphed in Excel, R, GraphPad Prism9 and MODDE software. Modeling of design space was performed using MODDE software. Three replicated center point experiments with all five additives at mid-concentrations were used and split between the batches for normalization and model evaluation purposes. The model responses for each parameter were statistically evaluated for model reproducibility, model validity, fit (R2) and prediction precision (Q2). Comparisons were conducted via ANOVA test with a significant difference defined as P\u0026thinsp;\u0026lt;\u0026thinsp;0.05. The number of bioreactor reactions was predetermined by the MODDE generator. The experiments were not randomized for blocks before testing.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eUsing a DoE Approach to Optimize Pluripotent Aggregate Expansion and Cell State\u003c/h2\u003e \u003cp\u003eD-optimal DoE interaction designs were used to compress the number of experimental runs compared with a full factorial design. Design compression allows determination of interactions and predictions within the design space explored within the concentration range used (i.e., \u0026lsquo;\u0026lsquo;known space\u0026rsquo;) [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Multiple high-molecular weight polymers have previously been shown to impact pluripotent stem cell cultures, but combinatorial assessments have not been performed to any large extent. Here, we investigated dextran sulfate (DS), Heparin sodium salt (HS), poly (vinyl alcohol) (PVA), Pluronic acid F68 (PA) and polyethylene glycol (PEG) (Fig.\u0026nbsp;1A). These factors and their concentration range were chosen based on previous publications about their impact on the iPSC aggregates growth and maintenance. Previous work showed that expansion of hPSCs in stirred-type bioreactors was improved by the addition of Pluronic F68 acting as non-ionic shear protectant [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. PA has an average molecular weight of 8400 Da and is a triblock copolymer of poly (ethylene oxide)\u0026ndash;poly (propylene oxide)\u0026ndash;poly (ethylene oxide). PA was also demonstrated to restore cell growth and viability in cell cultures [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. PVA is a biocompatible synthetic polymer with a molecular weight that ranges between 31,000\u0026ndash;50,000. It is commonly used in the pharmaceutical and food industries [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. In a recent study, researchers replaced serum albumin with PVA to develop a culture system that supports long-term expansion of functional mouse hematopoietic stem cells (HSCs) [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. PVA was also evaluated in hiPSC culture to promote proliferation. At 1 mg/mL PVA resulted in maximal cell density within the range tested from 0.1 to 10 mg/ml [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. PEG is a polymer based on the -CH2CH20- repeat unit that comes in different size ranges [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Similarly, to PVA, PEG is approved for drug and pharmaceutical applications. It functions as an aggregation inhibitor/emulsifying agent that may improve protein stability [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. DS is a polysulphate that may increase aggregate stability and cell growth. Heparin sodium salt supports hiPSC growth due to FGF2 stabilization. FGF2 degrades relatively quickly at 37\u0026deg;C [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Dextran sulphate has been commonly used in the biopharmaceutical industry to prevent cell aggregation. It is a poly-sulphated compound and was recently reported to control aggregate size and shape properties of hiPSCs, without compromising the maintenance of pluripotency [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWe generated D-optimal DoE interaction designs to evaluate these effectors. Cultures were maintained for four days in a suspension environment of a 100ml vertical wheel bioreactor (Fig.\u0026nbsp;1B). Reactors were initially seeded from 60\u0026ndash;70% confluent hiPSCs pre-cultured on coated vitronectin plates [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] that were dissociated into single cells with TrypLE and seeded at 1.1E5 cells/ml in the presence of Rho kinase inhibitor Y-27632 for the first 24h. Daily aggregate samples were collected to obtain images. On the final day of culture, RNA was extracted and used in cDNA synthesis. Cell samples for flow, counts and plating were digested using Accutase and processed for FLOW analysis using cell state specific antibodies.\u003c/p\u003e \u003cp\u003ePluripotency data was assessed at the end of the experiment through Immunohistochemistry (IHC), flow cytometry and qPCR testing. The experiment was divided into two batches of experiments to be run at different times (Fig.\u0026nbsp;1C). Three replicated center point experiments with all five additives at mid-concentrations were used and split between the batches for normalization and model evaluation purposes. The optimization criteria were based on multiple attributes related to aggregate stability, cell growth and pluripotency (Fig.\u0026nbsp;1D). Mathematical models were generated and tuned for maximal predictive power and fit. The model responses for each parameter were statistically evaluated for model reproducibility, model validity, fit (R2) and prediction precision (Q2). All response variables measured had an R2 value above 0.5 indicating a model with high significance. The model had model validity above 0.25 indicating the absence of problems related to outliers and transformation issues (Fig.\u0026nbsp;1D).\u003c/p\u003e \u003cp\u003eInspection of the response models reveal insights on the degree of single/combinatorial control of key culture parameters. Invariably, coefficient listings of responses revealed that most significant predictors of culture behavior are contributed by factor interactions (Fig.\u0026nbsp;1E).\u003c/p\u003e \u003cp\u003eAll bioreactor runs were performed using E8 basal medium for pluripotency maintenance creating a suitable bioreactor growth assay in which the impact of additives upon iPSC cell aggregate stability and growth could be measured daily (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eA). The assay conditions were based on our previous work evaluating process parameters such as bioreactor seeding density, digestion frequency (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eB), single cell dissociation (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eC), passage time (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eD), plate coating, and bioreactor speed. Using the data obtained from the short-term growth assay (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003eE), prediction models were constructed using computer modeling software. Response parameters included aggregate diameter sizes, cell concentration, growth rates, pluripotency marker expression, cell viability and aggregate variability. Each response model estimates main factor effects along with interactions between factors.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eImpact of Additives on Suspension Growth Kinetics\u003c/h2\u003e \u003cp\u003eVarious cell culture additives have been used to protect cells from environments that involve agitation-aeration cell damage [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan additionalcitationids=\"CR28\" citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. The additives selected in this design are components known to have a positive impact on growth and protection from shear stress. An optimizer setpoint dynamic profile was performed to detect main factor contributions (Fig.\u0026nbsp;2A) using doubling time data collected over 4 days of culture in bioreactors (Fig.\u0026nbsp;2B). Desirability criteria were set for the minimization of doubling time (i.e. maximal growth rate, Fig.\u0026nbsp;2C). Optimizer results detected read out conditions revealed that Pluronic F68, PVA and PEG all contribute to increased proliferation. The process capability index (Cpk) obtained from this optimizer was 0.9 (Fig.\u0026nbsp;2C) with a probability of failure of 1%. Modeling terms for maximal growth rate control include interaction terms between the additives.\u003c/p\u003e \u003cp\u003eWhile interpreting and confirming a triple interaction can be complex, the obtained results from the model optimizer and coefficients plot confirm a combined effect of these three additives that is not simply additive through independent factor contributions (Fig.\u0026nbsp;1E) but rather there is a synergistic relationship observed contributing to the desired overall growth in this suspension system.\u003c/p\u003e \u003cp\u003ehiPSC culture without daily media changes presented caveats related to impact of increased cell growth eventually plateauing due to nutrients limitation. Also, production of glycolytic waste products such as lactate and ammonia will reduce the medium pH, negatively impacting pluripotency, and continuous growth. For this purpose, the original seeding density of cells was at 1.1E5 cells/ml and the assay was revalidated at day 3 in addition to day 4 for many variables measured. Validation experiments compared the optimized medium to the control E8 medium without additives (Fig.\u0026nbsp;2D). The difference between the growth rates of the two mediums confirms the advantage of the additives (k\u0026thinsp;=\u0026thinsp;0.4) over the control (k\u0026thinsp;=\u0026thinsp;0.35).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eOptimization of Pluripotency Maintenance\u003c/h2\u003e \u003cp\u003eAttaining maximal growth conditions of iPSCs is not meaningful unless the iPSC state characteristics are also maintained. To demonstrate the maintenance of pluripotency, it was essential to test for the markers of undifferentiated iPSCs. At the end of the experiment, using flow cytometry we tested for SSEA4, OCT4, SOX2 and TRA-1-60 while a complimentary q-PCR assessment was conducted for NANOG, OCT4 and SOX2. The results of these experiments were used to model the effect of the additives on pluripotency. We optimized for the markers individually as shown in (Fig.\u0026nbsp;3A), since model results are non-confounding when optimization is performed for a singular response at a time. This is because that until proven that individual factors are not in regulatory conflict (i.e., SOX2, NANOG, or OCT4 being subject to differential control), single-response factor optimization is prudent. As we found, these factors are indeed subject to differential control.\u003c/p\u003e \u003cp\u003eModel optimizers for SOX2 and OCT4 had PEG as a major contributor with factor contributions (FC) of 18.72 and 53.88 respectively. The model for SOX2 also contained a significant contribution from Pluronic F68 (FC\u0026thinsp;=\u0026thinsp;15.28 and 3% probability of failure). Optimization of NANOG had a contribution of PEG and Pluronic F68 with FC of 7.16 and 15.44 respectively, however the addition of HS and DS had the highest contribution for NANOG optimization with FC of 21.12 and 48.04 respectively. Validation using these optimizers are shown in stained images for the pluripotency markers (Fig.\u0026nbsp;3B). Of the pluripotency markers measured, OCT4 expression was most strongly affected by tested media additives (Fig.\u0026nbsp;3C). Coefficients for all flow and qPCR markers (Fig.\u0026nbsp;3D) demonstrated that DS had opposing impact on OCT4 (negative) and NANOG (positive) making it challenging to select additive inputs that would satisfy all pluripotent markers. Upon validation, the common factor from all optimizers, PEG improved flow expression percentage of pluripotency markers compared to control media (Fig.\u0026nbsp;3E).\u003c/p\u003e \u003cp\u003eCo-expression of pluripotency intracellular markers (SOX2 and OCT4) with the extracellular markers (SSEA4 and TRA-1-60) was analyzed using flow cytometry (Figure S3 and S4). Comparative data showed co-expression of intracellular markers (Fig.\u0026nbsp;4A) and extracellular co-expression (Fig.\u0026nbsp;4B). Models demonstrated a strong correlation of 89% between OCT4 and TRA-1-60 expression (Fig.\u0026nbsp;4C). A correlation this strong was not observed between any of the other pluripotency markers. Using model optimizers for the maximization, and minimization, of OCT4 expression confirmed that TRA-1-60 was coregulated with OCT4 expression (Fig.\u0026nbsp;4D). This finding shows a direct predictive connection between OCT4 and TRA-1-60 suggesting a coregulatory relationship. The model showed additional correlation between these pluripotency markers and the response variable of aggregate size. Further validation of this correlation was evaluated using two software tools, R (Fig.\u0026nbsp;4E) and MODDE (Figure C-D-F). A positive correlation of 76\u0026ndash;85% was observed between increased diameter size and OCT4/TRA-1-60 expression respectively (Fig.\u0026nbsp;4E-F). Whereas a negative correlation was observed between increased diameter size and SOX2 (-46%) and NANOG (-50%) expression (Fig.\u0026nbsp;4E-F). Correlation results from both statistical tools showed that there is a significant correlation between aggregate size change and the expression of pluripotency markers with a P-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05, therefore the observed correlation is unlikely to occur by chance alone.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eControlling Aggregate Architecture Using Media Additives\u003c/h2\u003e \u003cp\u003eSince aggregate size was demonstrated to impact pluripotency and its heterogeneity could cause inconsistencies in nutrients and oxygen intake by the cells, we sought to establish parameters qualifying the aggregate state. Ideally, optimal aggregate growth should occur through the formation of aggregates from single cells upon seeding, whereafter aggregate size growth should occur due to cell proliferation only. Large deviation in aggregate sizes could have two causes 1) unstable aggregates breaking apart due to shear forces and separate to become smaller or 2) aggregate fusion causing larger aggregates to form from the addition of smaller ones (Fig.\u0026nbsp;5A). Therefore, a parameter can be calculated and set for a desired range in an optimizer (Fig.\u0026nbsp;5B) using the data collected on cell proliferation and aggregate size throughout culture time. A violin plot of aggregate diameters in bioreactors on day 3 is shown (Fig.\u0026nbsp;5C) demonstrating that additives modify the distribution of the observed sizes. Aggregate sizes measured have a deviation and therefore modelling on average size could potentially mask that response output. To overcome this limitation of using the average aggregate size as a response variable, the average size was inputted in the model in addition to its corresponding standard deviation. Using the average aggregate size generated from day 1 from all reactions in the design with their corresponding growth rates, a predicted aggregate size was calculated for day 3 and compared to the observed size measured. We define this difference as \u0026ldquo;Aggregate % Error\u0026rdquo;. This is calculated by subtracting the predicted from the actual average size measured and divided by the observed. Using this derivative assay parameters, it was then possible to obtain a combination of components that would predict stable size limiting, hereby limiting aggregate fusion (maximization) or aggregate instability (minimization) (Fig.\u0026nbsp;5B). The aggregate diameter deviation was calculated from aggregate diameter measurements of images taken on each culture day using ImageJ software. A known source of aggregate size increase is the adhesion of two or more aggregates into a large one [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. It is desired that the formation and growth of aggregates is only a contribution of cell proliferation and not fusion between cell spheres. Processes such as changing the media, taking samples, or anything that removes bioreactors from the base, can create static cultures that lead to an environment promoting clump formation. Our optimizer for the target aggregate % error (Fig.\u0026nbsp;5B), revealed that the main contributors to improving aggregate stability and decreasing aggregate size deviation in culture were HS (21.71) and PEG (17.48). Inspection of the aggregate diameters and growth slope showed a decrease in this optimization which is desirable as it indicates control over aggregation growth. The cell state control was also evaluated at this setpoint. NANOG expression was close to maximum indicating pluripotency maintenance upon aggregate control. When looking at a comparison of the aggregate % error within the bioreactors, the bioreactor with the minimum error and deviation value was a reaction that contains DS, PVA, PEG and HS (Fig.\u0026nbsp;5D-E). Whereas the maximum % error was reaction 19 that was a center point containing all additives (Fig.\u0026nbsp;5E). Validation comparing the optimizer results to the control with the E8 medium with no additives (Fig.\u0026nbsp;5D) shows that the aggregate size spread was minimized with the addition of Heparin and PEG.\u003c/p\u003e \u003cp\u003eUsing an optimizer model, we determined the combination of factors that satisfies an overall desired output of growth rate, pluripotency, and aggregate stability (Fig.\u0026nbsp;6A), not sacrificing one over another. The results suggested that the combination of Heparin sodium salt and PEG can be added to maintain aggregate stability, pluripotency, and growth aspects simultaneously. Validation against an E8 control was done and maintenance of pluripotency (Fig.\u0026nbsp;6B, C), cell proliferation (Fig.\u0026nbsp;6D), and aggregate stability (Fig.\u0026nbsp;6E) were specifically addressed. After successful validation, the optimizer model was tested on an additional cell line RCRP005N as compared to NCRM-1 with a decreased bioreactor speed from 60 to 40 RPM. Both cell lines showed continuous aggregate growth (Figure S5A) even after passaging by reseeding 30\u0026nbsp;million cells as aggregates on the fourth day of culture. The cells maintained their pluripotency markers OCT4 and SOX2 (91.8% and 97.8% for RCRP5005N and 97.4% and 98.4% for NCRM-1 respectively) (Figure S5B). Cell growth rate was steady after serial passaging without dissociation (Figure S5C) with a doubling time of (1.222\u0026ndash;1.434 for RCRP5005N and 1.025\u0026ndash;1.293 for NCRM-1 before and after passaging respectively). Results validated the robustness and repeatability of the predicted model. The Aggregate stability parameter can indirectly provide pluripotency maintenance and growth throughout the assay without impacting cell viability (Figure S5D). Therefore, this parameter can be regarded as a critical attribute that has a significant impact on iPSC cultures in a suspension environment.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eMany efforts at upscaling hiPSC cultures utilize high-cost commercial medium with proprietary media formulations. While adherent cultures predominantly use E8 medium, optimizing this culture medium to sustain a healthy and consistent aggregate culture is needed. Environmental and process stress such as hydrodynamic stress, single cell seeding, agitation and media changes are fundamentally different from adherent culture conditions. Here we empirically defined combinatorial media formulations that directly influence pluripotency, growth and aggregate stability using a five-dimensional DoE based approach. In this study, the two main components that positively influenced the majority of identified critical attributes were Heparin sodium salt and PEG. Interactions between HS and PEG were well modeled, predicted, and enabled a better understanding of controlling aggregate stability. While together improving iPSC aggregate culturing, PEG and Heparin sodium salt have different effects on cellular aggregates. PEG is known to induce cell aggregation and promote the fusion of larger cell aggregates at higher concentrations (10% and above), while also capable of inhibiting aggregation in other applications at lower concentrations [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Heparin sodium salt is known for its ability to interact with different growth factors such as FGF2 influencing cell behavior and pluripotency. Being able to detect this synergy within compounds that have a different effect on the culture reinforces the strength of the Design of Experiments approached used in this study. The only compounds used throughout this study that did not increase \u0026lsquo;aggregate % error\u0026rsquo; were PEG and Heparin sodium salt. Modelling PEG indicated the overall effect was a reduced standard deviation of the aggregate diameter size, though no effect on doubling time was observed. The aggregate deviation and diameter growth rate increased slightly in the presence of PEG. While the addition of Heparin sodium salt did not show a strong impact on aggregate size or growth rate individually it was shown to have strong positive interaction in the presence of PEG.\u003c/p\u003e \u003cp\u003eUsing DoE-based models provide several advantages over conventional one-factor-at-a-time methods. DoE modelling has embedded insights about model validity, reproducibility, fit (R2), and prediction precision (Q2) for each response variable measured allowing for an unbiased assessment of the overall model quality, all while identifying critical interactions. For example, all response variables measured throughout this series of experiments had an R2 value greater than 0.5, ensuring that models evaluated are highly significant. Models were also scrutinized for model validity and were only being considered significant when \u0026gt;\u0026thinsp;0.25, indicating an absence of outliers and transformation issues. As large-scale expansion of iPSCs and their derivatives become more common, DoE, as part of a Quality-by-Design strategy, should be used as a tool to evaluate multiple parameters impacting the overall cell quality and process stability. Embedding designs at increasing dimensional scales can be informative for the robust statistical analysis and unbiased optimization strategies needed when entering larger scale. Here, the modeling and design was performed within a low-shear reactor environment that is scalable. As there are multiple parameters involved in large scale manufacturing and transferring adherent culture conditions/protocols to suspension culture, more efficient assays are needed that do not require long testing periods or sequential optimizations strategies lacking the ability to evaluate combinatorial interactions. Specifically, we were able to define unexpected combinations of media additives for controlling aggregate formation within a bioreactor system. We also obtained a rich data response space, where mathematical models included doubling time, aggregate deviation, aggregate growth slope, pluripotency markers and cell count measurements.\u003c/p\u003e \u003cp\u003eUsing this methodical approach many novel insights into the compounds used in this study were discovered. DS, PVA and Pluronic F68 all had similar effects. Previous research showed that DS prevents cellular aggregation and has the ability to control aggregate size and growth properties [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. We could not replicate these observations. Set point analysis (Figure S6) showed that the largest deviation of \u0026lsquo;aggregate % error\u0026rsquo; (Fig.\u0026nbsp;5A) was obtained using DS. In agreement with this observation modelling for maximal \u0026lsquo;aggregate % error\u0026rsquo; resulted in a formulation predominantly containing DS (FC\u0026thinsp;=\u0026thinsp;32.90) (Fig.\u0026nbsp;5B). Similarly, previous studies evaluating PVA effect on hiPSCs demonstrated that it promoted proliferation [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. In contrast, set point analysis (Figure S6A) showed PVA impacts an increased aggregate diameter and the overall daily aggregate growth which is only desirable if these observed growth parameters occur solely due to cell proliferation. However, looking at the parameter of aggregate deviation (Figure S6A), indicates this PVA mediated increase was due to a heterogeneous effect on aggregate size. Collaborative evidence is provided by the fact that PVA had no effect on either growth rate or doubling time. A scenario repeated with Pluronic F68. Previous reports have shown that the inclusion of Pluronic F68 within a bioreactor increased the overall expansion capacity of hPSCs [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Here we show through a set point analysis that Pluronic F68 increased overall aggregate size and the deviation in the aggregate sizes (Figure S6A) again without effecting the aggregate growth slope, doubling time or cell concentration. Indicating again that this increase was due to fusion events between aggregates and not increased proliferation. An explanation for these discrepancies between previous studies is that others have compared individual reagents against controls lacking the additives, whereas in this study computer modelling allowed for the direct comparison of the compounds being evaluated. It should also be noted that Manstein et al used Pluronic F68 so that RPM could be increased, an increased RPM would have countered the fusion effect. Overall, it is indicated here that fusion has a detrimental effect on cellular proliferation, potentially because as aggregates become larger nutrient availability is increasingly restrictive, actively inhibiting a proliferative state.\u003c/p\u003e \u003cp\u003eEssential 8 was used as the basal media for all reactions throughout this series of experiments and is therefore considered responsible for the general maintenance of pluripotency within the bioreactors. However, the compounds evaluated in this study did influence this maintenance. The provision of DS, Pluronic F68 or PVA all had an overall negative influence on pluripotent maintenance promoting the down regulation NANOG and SOX2 expression, though a slightly positive effect on OCT4 expression was shown. In contrast, PEG and Heparin sodium salt had a general positive effect on pluripotent maintenance showing a positive correlation to OCT4 and NANOG expression, while having differential influence on SOX2 expression with Heparin sodium salt decreasing its expression and PEG increasing SOX2 levels (Figure S6A). In general, we could not find strong correlations between the core pluripotency markers OCT4, SOX2, and NANOG (Fig.\u0026nbsp;4C). A possible explanation is that they have divergent roles functioning as master regulators of lineage-specific differentiation as others have suggested. Previous studies have shown that the over expression of OCT4 leads to endodermal and mesodermal fates [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], whereas the overexpression of SOX2 generates ectodermal fate [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]suggesting that driving the maximal expression of either would result in the down regulation of the other. Modelling for minimal or maximal OCT4 expression indicated that the only pluripotent marker assayed within this study that followed OCT4 expression was TRA-1-60 (a positive correlation of 0.89), though notably aggregate diameter correlated with both pluripotency markers. This model was confirmed using R Software (Fig.\u0026nbsp;4E), which indicated a strong correlation between the increase of the diameter size and the expression of OCT4 and TRA-1-60 markers (0.790, 0.816) with a high significance level (P-value\u0026thinsp;\u0026lt;\u0026thinsp;000.1). This suggests that aggregate size could be used to influence differentiation potential, specifically here that endoderm differentiation may benefit from a population of larger aggregate. This could also explain why compounds that promoted fusion selectively lose SOX2 and NANOG expression while showing a small increase in OCT4 expression.\u003c/p\u003e \u003cp\u003eOne of the major efforts during this study was the development of quantification methods for aggregate stability with the overall goal of limiting aggregate fusion. Measurements of aggregate size deviation (range of sizes) could determine the overall homogeneity of aggregate size but could not directly inform on fusion since differential growth rates between aggregates and aggregate fractioning due to shear stress are alternative explanations. So, to better define what was occurring on the individual aggregate level a method for quantifying this aggerate parameter was developed. Resting on a theoretical estimation of the rate aggregate size should change according to growth rates observed, we directly compared this to measured size changes observed within the bioreactors. This was defined as \u0026lsquo;aggregate % error\u0026rsquo;. We found that by modelling using this parameter we could gain a better control over the size and overall homogeneity of the aggregates within bioreactors. Models setting the \u0026lsquo;aggregate % error\u0026rsquo; to the predicted diameter (theoretical size based on doubling time) would define conditions that lacked fusion but in which aggregates were stable enough to resist shear forces. While this model proved useful in measuring overall aggregate stability there are inherent limitations to it. The mathematical model assumes cell size uniformity, a parameter likely to change based on the density of the individual aggregates or differentiation state of the cell culture. There is also an assumption based on empirical observations and simulation experiments that the cells are randomly packed. An alternative method for estimating aggregate stability would have been directly measuring aggregate numbers daily, decreasing aggregate number with an increasing aggregate size could directly indicate fusion. Here we show that models quantifying aggregate behavior based on calculations of theoretical predictions as compared to empirical results are crucial for the identification and understanding of the complexities in a biological system. Having predictive models allows for formulating a hypothesis about the underlying process mechanisms and therefore facilitates optimization by identifying key factors and parameters that bring the model to desired results.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eFor each model the relationship between additive/response parameter could be analyzed and optimized towards desirability criteria. In all desirability contexts, process capability (Cpk) as a measurement of process robustness was favorable. Optimal conditions for maximal cell growth required combinatorial additive use. Similarly, optimization of the maintenance of pluripotency, as based on markers OCT4, NANOG, SOX2 was also dependent on combinatorial signaling. Based on measurements of aggregate size and size variance, optimization could be performed to increase aggregate homogeneity. In all cases, underlying models included multiple interaction terms, revealing the criticality of simultaneous testing using combinatorial designs. It was demonstrated that the combinatorial effects of Polyethylene Glycol and Heparin Sodium Salt maintain growth and pluripotency and control aggregate stability by limiting fusion events between aggregates in suspension cultures. We conclude that DoE-based interaction testing performed within a manufacturing-relevant environment allows for process understanding of the biomanufacturing process. The method identifies critical process parameters; their interacting criticality, while returning a deep process understanding.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eHS: Heparin sodium salt\u003c/p\u003e\n\u003cp\u003ePEG: polyethylene glycol\u003c/p\u003e\n\u003cp\u003ePVA: poly (vinyl alcohol)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePA: Pluronic F68\u003c/p\u003e\n\u003cp\u003eDS: dextran sulfate\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eE8: Essential 8\u003c/p\u003e\n\u003cp\u003eDoE: Design of experiments\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e3D: Three-dimensional\u003c/p\u003e\n\u003cp\u003ecGMP: Current Good manufacturing practice\u003c/p\u003e\n\u003cp\u003ehiPSC: Human Induced pluripotent stem cell\u003c/p\u003e\n\u003cp\u003ePBS: Phosphate buffer solution\u003c/p\u003e\n\u003cp\u003eq-PCR: Quantitative PCR\u003c/p\u003e\n\u003cp\u003eRPM: Revolution per minute\u003c/p\u003e\n\u003cp\u003eROCK: Rho-Kinase\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to thank Trailhead Biosystems for funding this research. The authors also thank Michael Majcher and Katie Sears for their critical reading of the manuscript and Cleveland State University for their support.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eM.B conceived and supervised the project. H.Y. designed, conducted all experiments and wrote the manuscript. S.R. cultured the starting iPSC material for the validation. J.J. provided review feedback. All authors approved the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe work was supported and funded by Trailhead Biosystems Inc. No external funding was used.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRequests for further information or more detailed protocols should be directed to and will be fulfilled by the corresponding author. This study did not generate new unique reagents. The data that support the findings of this study are available on request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.J. is founder of and shareholder of Trailhead Biosystems, Inc., Beachwood, OH, USA. M. B. is a shareholder in Trailhead Biosystems, Inc., Beachwood OH. This work has been filed as US Provisional Application No. 63/525,083.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eTakahashi K, Yamanaka S. Induction of Pluripotent Stem Cells from Mouse Embryonic and Adult Fibroblast Cultures by Defined Factors. Cell. 2006;126:663\u0026ndash;76.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFan Y, Wu J, Ashok P, Hsiung M, Tzanakakis ES. Production of Human Pluripotent Stem Cell Therapeutics under Defined Xeno-free Conditions: Progress and Challenges. Stem Cell Rev Rep. 2015;11:96\u0026ndash;109.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeinsberger J, Reisinger D, Weber B. Global trends in clinical trials involving pluripotent stem cells: a systematic multi-database analysis. NPJ Regen Med. 2020;5:15.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eColter J, Murari K, Biernaskie J, Kallos MS. Induced pluripotency in the context of stem cell expansion bioprocess development, optimization, and manufacturing: a roadmap to the clinic. NPJ Regen Med. 2021;6:72.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNogueira DES, Rodrigues CAV, Carvalho MS, Miranda CC, Hashimura Y, Jung S, et al. Strategies for the expansion of human induced pluripotent stem cells as aggregates in single-use Vertical-Wheel\u0026trade; bioreactors. J Biol Eng. 2019;13:74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOtsuji TG, Bin J, Yoshimura A, Tomura M, Tateyama D, Minami I, et al. A 3D Sphere Culture System Containing Functional Polymers for Large-Scale Human Pluripotent Stem Cell Production. Stem Cell Reports. 2014;2:734\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbbasalizadeh S, Larijani MR, Samadian A, Baharvand H. Bioprocess Development for Mass Production of Size-Controlled Human Pluripotent Stem Cell Aggregates in Stirred Suspension Bioreactor. Tissue Eng Part C Methods. 2012;18:831\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu M. Engineering stem cell niches in bioreactors. World J Stem Cells. 2013;5:124.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeramura T, Takehara T, Onodera Y, Nakagawa K, Hamanishi C, Fukuda K. Mechanical stimulation of cyclic tensile strain induces reduction of pluripotent related gene expressions via activation of Rho/ROCK and subsequent decreasing of AKT phosphorylation in human induced pluripotent stem cells. Biochem Biophys Res Commun. 2012;417:836\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaha S, Ji L, de Pablo JJ, Palecek SP. TGFβ/Activin/Nodal Pathway in Inhibition of Human Embryonic Stem Cell Differentiation by Mechanical Strain. Biophys J. 2008;94:4123\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim YM, Kang YG, Park SH, Han M-K, Kim JH, Shin JW, et al. Effects of mechanical stimulation on the reprogramming of somatic cells into human-induced pluripotent stem cells. Stem Cell Res Ther. 2017;8:139.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoetzke R, Sechi A, De Laporte L, Neuss S, Wagner W. Why the impact of mechanical stimuli on stem cells remains a challenge. Cell Mol Life Sci. 2018;75:3297\u0026ndash;312.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu C, Rosler E, Jiang J, Lebkowski JS, Gold JD, O\u0026rsquo;Sullivan C, et al. Basic Fibroblast Growth Factor Supports Undifferentiated Human Embryonic Stem Cell Growth Without Conditioned Medium. Stem Cells. 2005;23:315\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuo H-H, Gao X, DeKeyser J-M, Fetterman KA, Pinheiro EA, Weddle CJ, et al. Negligible-Cost and Weekend-Free Chemically Defined Human iPSC Culture. Stem Cell Reports. 2020;14:256\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eManstein F, Ullmann K, Kropp C, Halloin C, Triebert W, Franke A, et al. High Density Bioprocessing of Human Pluripotent Stem Cells by Metabolic Control and in Silico Modeling. Stem Cells Transl Med. 2021;10:1063\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCastellanos IJ, Crespo R, Griebenow K. Poly(ethylene glycol) as stabilizer and emulsifying agent: a novel stabilization approach preventing aggregation and inactivation of proteins upon encapsulation in bioerodible polyester microspheres. J controlled release : official J Controlled Release Soc. 2003;135\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen G, Gulbranson DR, Hou Z, Bolin JM, Ruotti V, Probasco MD, et al. Chemically defined conditions for human iPSC derivation and culture. Nat Methods. 2011;8:424\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilken S, Guerra RE, Levine D, Chaikin PM. Random Close Packing as a Dynamical Phase Transition. Phys Rev Lett. 2021;127:038002.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBukys MA, Mihas A, Finney K, Sears K, Trivedi D, Wang Y, et al. High-Dimensional Design-Of-Experiments Extracts Small-Molecule-Only Induction Conditions for Dorsal Pancreatic Endoderm from Pluripotency. iScience. 2020;23:101346.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu W, Rathman JJ, Chalmers JJ. An investigation of small-molecule surfactants to potentially replace pluronic F-68 for reducing bubble-associated cell damage. Biotechnol Bioeng. 2008;101:119\u0026ndash;27.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoski A, Yim K, Shivkumar S. Effect of molecular weight on fibrous PVA produced by electrospinning. Mater Lett. 2004;58:493\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilkinson AC, Ishida R, Kikuchi M, Sudo K, Morita M, Crisostomo RV, et al. Long-term ex vivo haematopoietic-stem-cell expansion allows nonconditioned transplantation. Nature. 2019;571:117\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTang X, Wu H, Xie J, Wang N, Chen Q, Zhong Z et al. The combination of dextran sulphate and polyvinyl alcohol prevents excess aggregation and promotes proliferation of pluripotent stem cells in suspension culture. Cell Prolif. 2021;54.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraam SR, Zeinstra L, Litjens S, Ward-van Oostwaard D, van den Brink S, van Laake L, et al. Recombinant Vitronectin Is a Functionally Defined Substrate That Supports Human Embryonic Stem Cell Self-Renewal via αVβ5 Integrin. Stem Cells. 2008;26:2257\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFurue MK, Na J, Jackson JP, Okamoto T, Jones M, Baker D et al. Heparin promotes the growth of human embryonic stem cells in a defined serum-free medium. Proceedings of the National Academy of Sciences. 2008;105:13409\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLipsitz YY, Tonge PD, Zandstra PW. Chemically controlled aggregation of pluripotent stem cells. Biotechnol Bioeng. 2018;115:2061\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChisti Y. Animal-cell damage in sparged bioreactors. Trends Biotechnol. 2000;18:420\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePapoutsakis E. Media additives for protecting freely suspended animal cells against agitation and aeration damage. Trends Biotechnol. 1991;9:316\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoldblum S, Bae YK, Hink WF, Chalmers J. Protective effect of methylcellulose and other polymers on insect cells subjected to laminar shear stress. Biotechnol Prog. 1990;6:383\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNiwa H, Miyazaki J, Smith AG. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat Genet. 2000;24:372\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKopp JL, Ormsbee BD, Desler M, Rizzino A. Small Increases in the Level of Sox2 Trigger the Differentiation of Mouse Embryonic Stem Cells. Stem Cells. 2008;26:903\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"stem-cell-research-and-therapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scrt","sideBox":"Learn more about [Stem Cell Research \u0026 Therapy](http://stemcellres.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/scrt/default.aspx","title":"Stem Cell Research \u0026 Therapy","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"iPSC, human induced pluripotent stem cell, pluripotent state, Bioreactor, DoE, culture media, suspension culture, aggregates, aggregate size, aggregate stability, proliferation","lastPublishedDoi":"10.21203/rs.3.rs-3500780/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3500780/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eStem cell-derived therapies hold the potential for treatment of regenerative clinical indications. Static culture has a limited ability to scale up thus restricting its use. Suspension culturing can be used to produce target cells in large quantities, but also presents challenges related to stress and aggregation stability.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eUtilizing a design of experiments (DoE) approach in vertical wheel bioreactors, we evaluated media additives that have versatile properties. The additives evaluated are Heparin sodium salt (HS), polyethylene glycol (PEG), poly (vinyl alcohol) (PVA), Pluronic F68 and dextran sulfate (DS). Multiple response parameters were chosen to assess cell growth, pluripotency maintenance and aggregate stability in response to the additive inputs, and mathematical models were generated and tuned for maximal predictive power.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eExpansion of iPSCs using 100 ml vertical wheel bioreactor assay for 4 days on 19 different media combinations resulted in models that can optimize pluripotency, stability, and expansion. The expansion optimization resulted in the combination of PA, PVA and PEG with E8. This mixture resulted in an expansion doubling time that was 40% shorter than that of E8 alone. Pluripotency optimizer highlighted the importance of adding 1% PEG to the E8 medium. Aggregate stability optimization that minimizes aggregate fusion in 3D culture indicated that the interaction of both Heparin and PEG can limit aggregation as well as increase the maintenance capacity and expansion of hiPSCs, suggesting that controlling fusion is a critical parameter for expansion and maintenance. Validation of optimized solution on two cell lines in bioreactors with decreased speed of 40 RPM, showed consistency and prolonged control over aggregates that have high frequency of pluripotency markers of OCT4 and SOX2 (\u0026gt;\u0026thinsp;90%). A doubling time of around 1 to 1.4 days was maintained after passaging as clumps in the optimized medium. Controlling aggregate fusion allowed for a decrease in bioreactor speed and therefore shear stress exerted on the cells in a large-scale expansion.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThis study resulted in a control of aggregate size within suspension cultures, while informing about concomitant state control of the iPSC state. Wider application of this approach can address media optimization complexity and bioreactor scale-up challenges.\u003c/p\u003e","manuscriptTitle":"Addressing Bioreactor hiPSC Aggregate Stability, Maintenance and Scaleup Challenges Using a Design of Experiment Approach","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-11-23 23:57:42","doi":"10.21203/rs.3.rs-3500780/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2023-11-20T10:53:17+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2023-11-20T09:58:54+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2023-10-30T00:50:12+00:00","index":"","fulltext":""},{"type":"submitted","content":"Stem Cell Research \u0026 Therapy","date":"2023-10-27T22:23:52+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"stem-cell-research-and-therapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scrt","sideBox":"Learn more about [Stem Cell Research \u0026 Therapy](http://stemcellres.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/scrt/default.aspx","title":"Stem Cell Research \u0026 Therapy","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"15c7e54e-9746-40ad-9879-fb7b926b5982","owner":[],"postedDate":"November 23rd, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2024-06-16T19:04:41+00:00","versionOfRecord":[],"versionCreatedAt":"2023-11-23 23:57:42","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3500780","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3500780","identity":"rs-3500780","version":["v1"]},"buildId":"7rjqhiLT3MXkJMwkYKINL","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}