Biotechnology Journal

During manufacturing batches, Chinese hamster ovary (CHO) cells encounter critical levels of environmental stressors such as ammonia, lactate, and osmolality accumulation that can significantly reduce cell health and productivity. It is therefore crucial that stress adaptation and resistance be factored into cell line development (CLD). In this study, we employee population-based transcriptomic and differential gene expression analysis on stress-induced CHO cells to identify biomarkers displaying both heritable and stress-responsive properties. Using this workflow, 199 genes displayed transcriptional variability characteristic of a bistable system that formed four network communities of co-fluctuating genes. These communities were enriched in genes related to the regulation of apoptotic processes and gene expression/metabolic pathways. Seven genes were identified as promising biomarkers for engineering a stress-resistant phenotype. Genetic engineering methods may be employed in the future to bias clonal populations for higher stress tolerance to manufacturing stress, therefore increasing cell health and productivity in at-scale bioreactors.

Medium optimization is a crucial step of cell culture for biopharmaceutics and regeneration medicine. It remains challenging, as both media and cells are highly complex systems. To address the issue, we tried active learning to fine-tune the culture medium by combining the high-throughput assay and machine learning. As a pilot study, the cell line HeLa-S3 and the gradient-boosting decision tree algorithm were used. The regular and time-saving approaches were developed, and both successfully fine-tuned 29 components to achieve improved cell culture than the original medium. The fine-tuned media showed a significant decrease in fetal bovine serum and the differentiation in vitamins and amino acids. Unexpectedly, the medium optimization raised the cellular NAD(P)H abundance but not the cell concentration owing to the conventional method used for cell culture assay. Our study demonstrated the efficiency of active learning for medium optimization and provided valuable hints for employing machine learning in cell culture.

Nanobody is one special type of single-domain antibody fragment with multiple advantages over traditional antibody. Our previous work established linear-double-stranded DNA (ldsDNA, or PCR amplicon) as novel biological parts for building AND gate genetic circuits in mammalian cells. During this AND-gate circuit formation process, the co-transfected up- and down-stream ldsDNAs could be linked together to form intact gene expression cassette. Here, we employed this ldsDNA-based AND-gate (LBAG) strategy to construct nanobody library in mammalian cells. The sequence complexity of complementary determining regions (CDRs) was introduced into ldsDNA by PCR amplification. After being co-transfected into mammalian cells, the up- and down-stream ldsDNAs undergo AND gate linkage and form full nanobody coding regions, containing CDR1-3. High throughput sequencing identified 22,173 unique oligonucleotide sequences in total generated by this strategy. Thus, we developed a novel method to construct nanobody library, which is a start point for building high content nanobody library in mammalian cells.

Chinese Hamster Ovary (CHO) cell monoclonal antibody (mAb) production in continuous perfusion has witnessed a renewed interest within the biopharmaceutical industry. Widespread implementation of perfusion biomanufacturing, however, remains hindered by long process development timelines and high costs. Use of predictive scale-down platforms to generate large informative metabolic datasets and guide process development decisions is critical to decreasing a molecules time to market. While scale-down platforms based on the pseudo perfusion concept have been previously reported, they have not been rigorously validated. They are often limited by oxygen transport or insufficient metabolic characterization, reducing their role to a preliminary screening tool. Here, we report the design and validation of a pseudo perfusion platform based on a phenotype-driven approach to ascertain that the process emulates continuous perfusion characteristics and is not oxygen limited. Beyond metabolic and cell size steady state, we show that our pseudo perfusion design enables cell cycle subpopulation and intracellular antibody expression steady state. We also demonstrate that pseudo perfusion robustly predicts amino acid demands in continuous perfusion bioreactors with exceptional linear correlation across a broad range of cell-specific perfusion rates (CSPRs). When coupling the pseudo perfusion platform developed here with a workflow for metabolic characterization, we significantly augment the dimensionality and reliability of data which can be generated at this scale to gain actionable insights towards perfusion process design, ultimately reducing process development timelines and the associated costs. HighlightsResidual lactate is a key proxy for oxygen transport in scale down platform design Novel flow cytometry workflow confirms cell cycle and intracellular steady state Pseudo perfusion robustly predicts metabolic phenotypes in continuous perfusion K-means clustering analysis of nutrient rates provides insight into media design

Rapid progress in algal biotechnology has triggered a growing interest in hydrogel-encapsulated microalgal cultivation, especially for the engineering of functional photosynthetic materials and biomass production. An overlooked characteristic of gel-encapsulated cultures is the emergence of cell aggregates, which are the result of the mechanical confinement of the cells. Such aggregates have a dramatic effect on the light management of gel-encapsulated photobioreactors and hence strongly affect the photosynthetic outcome. In order to evaluate such an effect, we experimentally studied the optical response of hydrogels containing algal aggregates and developed optical simulations to study the resultant light intensity profiles. The simulations are validated experimentally via transmittance measurements using an integrating sphere and aggregate volume analysis with confocal microscopy. Specifically, the heterogeneous distribution of cell aggregates in a gel matrix can increase light penetration while alleviating photoinhibition compared to a flat biofilm. Finally, we demonstrate that light harvesting efficiency can be further enhanced with the introduction of scattering particles within the hydrogel matrix, leading to a four-fold increase in biomass growth. Our study, therefore, highlights a new strategy for the design of spatially efficient photosynthetic living materials that have important implications for the engineering of future algal cultivation systems. Significance StatementThe ability to cultivate microalgae at scale efficiently would allow more sustainable production of food and food additives. However, efficient growth of microalgae requires optimised light conditions, which are usually challenging to obtain using biofilm cultivations mode: as the outer layer of cells are necessarily more exposed to incoming light than the inner layer, posing the problem of photoinhibition on the outer cells receiving too much light, and shading the ones below. Here we study both experimentally and numerically, how microalgae aggregates growing in the confinement of a hydrogel can provide an improved light distribution and therefore biomass growth is maximised. This study proposes new strategies on how to engineer future photobioreactors.

Tissue cultures are an important study model for woody plant tissue and can be used to study lignin biosynthesis. The greatest disadvantage of protocols based on extraction of lignin from wood biomass is the almost inevitable alteration of the native structure of lignin. Using a Norway spruce tissue culture with the ability to secrete monolignols into a liquid culture medium, fundamental aspects of lignin have been studied in the past, such as its structure, the enzyme activity related to its polymerization, and its interactions with a secondary cell wall hemicellulose. In this study, parameters that can induce monolignol production and secretion in the tissue culture are investigated via gene expression analysis. The impact of the composition of the solid growth medium, which was in some cases supplemented with xylan, was studied in depth through transcriptomic investigation. We find that the state (i.e. liquid or solid) and the xylan content of the medium can impact gene expression, although microscopic analysis suggests that cellular morphology is consistent. Extracellular lignin was collected from a formulation of liquid medium with the same composition as that used for cellular growth, which was previously presumed to be "non-inducing" of lignin biosynthesis. Chemical analysis of this lignin was performed using nuclear magnetic resonance spectroscopy and size exclusion chromatography, which revealed changes in its structure compared to the polymer produced in the previously developed "inducing" liquid medium. These experiments show that there is still much we do not understand about an oft-used tissue culture system, but show the way to a deeper understanding of the genetic control of lignin biosynthesis.

Microfluidic vortex shedding (VS) can rapidly deliver mRNA to T cells with high yield. The mechanistic underpinning of VS intracellular delivery remains undefined and VS-Cas9 genome editing requires further studies. Herein, we evaluated a series of VS devices containing splitter plates to attenuate vortex shedding and understand the contribution of computed force and frequency on efficiency and viability. We then selected a VS design to knockout the expression of the endogenous T cell receptor in primary human T cells via delivery of CRISPR-Cas9 ribonucleoprotein (RNP) with and without brief exposure to an electric field (eVS). VS alone resulted in an equivalent yield of genome-edited T cells relative to electroporation with improved cell quality. A 1.8-fold increase in editing efficiency was demonstrated with eVS with negligible impact on cell viability. Cumulatively, these results demonstrate the utility of VS and eVS for genome editing human primary T cells with Cas9 RNPs.

The application of CRISPR-based genome editing tools in plants is often challenged by low editing efficiencies, requiring most plant editing workflows to proceed through the delivery of pre-formed ribonucleoprotein (RNP) complexes to protoplasts. Here, we report increases in protoplast-based RNP delivery and genome editing efficiencies through the addition of anionic polymers to standard protoplast transfection protocols. We test addition of various polymers and peptides for their ability to increase genome editing efficiencies in both Nicotiana benthamiana and Arabidopsis thaliana protoplasts, by adding these components to standard PEG-based protoplast transfection protocols: i) non-covalent addition of charged polymers, ii) non-covalent addition of amphiphilic peptide A5K, and iii) tyrosinase-mediated covalent conjugation of various relevant peptide motifs directly to the RNP. Incorporation of the amphiphilic peptide A5K or covalent attachment of peptides to the RNP had no positive effect on editing efficiencies. However, we found that addition of anionic polymer polyglutamic acid to standard PEG transfection protocols significantly improved editing efficiencies in both Nicotiana benthamiana and Arabidopsis thaliana protoplasts relative to RNPs alone without negatively impacting protoplast viability. Our results suggest anionic polymers stabilize the RNP and increase the colloidal stability of the protoplast transfection workflow. This simple and straightforward method of stabilizing Cas9 RNPs can be easily adopted by others working on direct protein delivery to plant protoplasts to increase genome editing efficiencies. Key messageThe addition of anionic polymer polyglutamic acid to standard protoplast PEG transfection workflows enhance CRISPR-Cas9-mediated gene editing in plant protoplasts. Our results suggest the mechanism of increased transfection efficiency is due to colloidal stabilization of RNPs. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=96 SRC="FIGDIR/small/635105v1_ufig1.gif" ALT="Figure 1"> View larger version (18K): org.highwire.dtl.DTLVardef@1101f0org.highwire.dtl.DTLVardef@964402org.highwire.dtl.DTLVardef@14afe61org.highwire.dtl.DTLVardef@82a982_HPS_FORMAT_FIGEXP M_FIG C_FIG

Stable and productive CHO cell lines are essential for biopharmaceutical manufacturing, yet early expansion steps are often constrained by prolonged period required for suspension adaptation. Single-cell cloning (SCC) ensures monoclonality and regulatory compliance, but cells transitioning from static to suspension culture frequently exhibit variable recovery, which prolongs timelines and increases process variability. To address this challenge, mixing-based microplate culture systems have been developed to improve early expansion efficiency. The C.NEST platform provides controlled pneumatic mixing and environmental monitoring that facilitates earlier adaptation to suspension conditions. At the 96-well and 24-well stages, this system allows cells to establish stable growth under suspension-like environments, thereby shortening the adaptation period following transfer to shaking culture. In this study, we applied C.NEST to the SCC workflow for developing CHO-K1 stable cell lines. Integrating C.NESTs controlled mixing reduced adaptation time, enhanced the consistency of clone expansion, and improved the ability to identify high-yield clones. These findings highlight the potential of C.NEST to streamline cell line development workflows by accelerating early suspension adaptation and improving clone selection reliability. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=165 SRC="FIGDIR/small/693844v1_ufig1.gif" ALT="Figure 1"> View larger version (40K): org.highwire.dtl.DTLVardef@ca76bforg.highwire.dtl.DTLVardef@3a3079org.highwire.dtl.DTLVardef@447e01org.highwire.dtl.DTLVardef@ac8d6f_HPS_FORMAT_FIGEXP M_FIG C_FIG C.NEST mixing shortens suspension adaptation, accelerates clone expansion, and enhances early-stage screening. HighlightO_LIThe C.NEST microplate agitation culture system accelerates early CHO-K1 cell line development. C_LIO_LIControlled pneumatic mixing improved oxygen transfer and medium homogeneity, promoting stable growth during early expansion. C_LIO_LIEarly mixing shortens suspension adaptation by approximately one week. C_LIO_LIMixing cultures enabled more accurate clone performance assessment, revealing high-producing outliers. C_LIO_LIC.NEST provides a scalable and reproducible solution for integrating mixing-based culture into single-cell cloning workflows. C_LI

Direct sequencing of total cellular RNA enables a better understanding of a broad spectrum of RNA species controlling cellular processes and organismal function. Current nanopore direct RNA sequencing method, however, only captures polyadenylated RNA for sequencing. To address this issue, we developed a unique 3-end RNA tailing method to capture total RNA for nanopore direct RNA sequencing. Due to the distinct electrical signature of the added tail on nanopore, this method allows simultaneous detection of both non-polyadenylated and polyadenylated RNAs. We demonstrated the effectiveness of this method in capturing the dynamics of transcription and polyadenylation of chloroplast RNAs in plant cell. With its high efficiency in retaining total RNA on nanopore, this method has the potential to be broadly applied to RNA metabolism and functional genomics studies.

Scalable, continuous biomanufacturing processes have grown in importance to meet demand for smaller bioreactor sizes, lowered production costs, and improved quality attributes. The Sf9/recombinant baculovirus (rBV) expression system demonstrates promise for virus-like particle (VLP) vaccine and gene therapy production. Here, we present a continuous rBV platform integrating an infection plug flow reactor (PFR) between stirred tank growth (gCSTR) and production (pCSTR) bioreactors. Cell expansion in the gCSTR included a ramp-up stage followed by continuous growth, reaching a steady state of 5x106 cells/mL and >90% viability. Peclet number-fit tracer studies confirmed near-ideal plug flow in the PFR, yielding a 10 h residence time and progressive infection as measured by gp64 signaling. Finally, a pCSTR with a residence time of 48 h exhibited sustained recombinant protein production. An integrated pilot cascade incorporating all reactors ran continuously for 5 days, maintaining stable CSTR cell densities and a measurable increase in infected cell diameter from 14.5 m to 16.1 m. Western blotting and EM of [~]100 nm VLPs in pCSTR effluent demonstrated platform success. Digital twin mechanistic models across four distinct stages of bioreactor operation and Hill-type relationships for rBV infection kinetics predicted cell growth and death for a 7-day run, demonstrating promise for designing continuous systems in silico and building a quantitative framework for scale-up and optimization. Our multi-stage reactor configuration represents a cell host- and product-agnostic production scheme, particularly for processes prone to product heterogeneity, and paves the way towards a true end-to-end continuous platform for myriad modalities in the future.

Intensified perfusion cultures promise higher yields and consistent product quality, yet extended runs frequently stall due to declining Chinese hamster ovary (CHO) cell viability at high cell density. Here, we identify DNA damage accumulation as a central, previously underappreciated driver of this limitation. Using two antibody-expressing CHO cell lines operated at a control and high cell density in perfusion bioreactors, we combined time-resolved transcriptomics with molecular, biophysical and super-resolution imaging analyses. We observed a progressive, global downregulation of DNA damage response (DDR) pathways accompanied by a time-dependent accumulation of DNA lesions. Notably, {gamma}H2AX signalling declined despite DNA damage, indicating impaired damage sensing and repair. Concomitantly, RNA polymerase II protein levels and transcriptional hub organisation were markedly reduced, consistent with widespread transcriptional dysfunction preceding loss of viability. An extended 21-day perfusion run confirmed continued viability decline beyond day 14, supporting a cumulative damage model. Comparison with HEK293 cells revealed substantially more efficient DNA repair, underscoring an intrinsically attenuated DDR in CHO cells. Together, these findings demonstrate that unrepaired DNA damage fundamentally constrains perfusion culture longevity, exposing limits of CHO genomic plasticity and highlighting DDR pathways as promising targets for host cell engineering to enhance perfusion performance.

The activation of gene circuits can impose a significant burden on cells, leading to heterogeneous expression and reduced productivity. In this work, we focused on the T7 production system in E. coli BL21, a prime example of a burdensome gene circuit, to investigate the main cause for this gene expression heterogeneity and methods to mitigate it. Based on continuous cultivation analyzed and control by automated flow cytometry, we quantified the trade-off between cellular growth and gene expression and tracked the cell-to-cell heterogeneity in gene expression (measured as entropy). We concluded that the growth reduction associated to the activation of the burdensome gene circuit, i.e., the switching cost, is at the origin of the population heterogeneity. The loss of growth rate imposed by the burdensome activation of the gene is compensated at the population level by the overgrowth of less induced cells that safeguard the population by generating entropy. We tried to homogenize the population by pulsing the inducer with increasing frequency but found that the population escapes control through promoter mutation, leading to a genotype exhibiting reduced gene expression, but also, reduced entropy. To engineer a more homogeneous population without sacrificing gene expression, we decreased the switching cost associated to the induction by lowering the quality of the main carbon source. This strategy successfully led to a more homogeneous and productive population. Our approach allows for a precise quantification of the trade-off between growth and gene expression in cell population cultivated under dynamic conditions and highlights the importance of the switching cost for designing efficient approaches of cell population control.

The growth of cultures and formation of mucilage blooms in reaction to salt stress of cyanobacterial cultures are investigated with a focus on the influence of pH. In non-buffered medium, cultures show their pH increasing from 6.5 just after inoculation, up to 11 during the exponential phase. We record the time-evolution of concentration and pH, with different initial OD0. In a second set of experiments, we extract the doubling time of the unbuffered cultures in comparison with those inoculated in pH-buffered BG11 media at four different pH from 6.3 to 10.5 : in the most acid media, all cultures die or grow very slowly. At pH = 10.5, we obtain the fastest growth for all four strains, allowing to qualify these cyanobacteria as being alkaliphiles, though for all strains with comparable initial OD0, the doubling time is shorter for unbuffered cultures. Following a previous study [31]), we finally investigate the influence of pH on mucilage formation and biomass uplift induced by salt stress, involving EPS floculation by cations. Our results show that operating in buffered media significantly influences the mucilage formation, though the observed regimes cannot be simply correlated to the pH value.

PurposeExtracellular vesicles (EVs) are increasingly explored as natural vehicles for drug delivery and gene therapy approaches. However, reproducible yield and scalability of EV production still pose major challenges in the clinical translation of EV-based therapies. In this study, we sought to quantify and characterize EVs released by suspension-cultured HEK293 cells (Expi293F cells) grown in shaker flasks or small-scale bioreactors, to investigate how the culturing environment affects EV production yield. MethodsExpi293F cells were cultivated (N=3) in either shaker flasks or a bioreactor system, and total cell density, viability, and size were monitored. Supernatants were drawn daily post-cell seeding and were analyzed for EV quantity, size, morphology, and CD63 expression. ResultsNo significant differences were observed in terms of total cell density, viability, and cell size between both cultivation settings. However, cultivation of Expi293F cells in the bioreactor environment significantly increased EV yield by 3-fold compared to shaker flask cultivation (p < 0.01). Other parameters such as average nanoparticle size, EV morphology, and CD63 expression remained comparable between both cultivation methods. ConclusionThese results demonstrate that Expi293F-derived EV yield can be increased by culturing cells in a scalable bioreactor system. These findings pave the way towards the production of therapeutic-based EVs in a scalable and reproducible manner suitable for future (pre-)clinical applications.

mRNA therapeutics have demonstrated transformative potential in infectious disease, oncology and genetic disorders. Parallel to the rapid clinical advancement of mRNA therapies, research into in vitro transcription (IVT), the pivotal process of mRNA manufacturing, has intensified significantly. However, the majority of these studies omitted cap analog, an essential substrate for co-transcriptional capped IVT (co-IVT), likely to simplify experimental systems. Nevertheless, co-IVT remains the dominant industrial standard for the production of mRNA therapeutics. Here, we report an optimized co-IVT process that achieves a yield of more than 26 g/L while maintaining capping efficiency of over 99.5%, and achieving a 30-fold reduction in double-stranded RNA (dsRNA), the most critical process-related impurity. These performance gains substantially improve mRNA production efficiency and enhance mRNA quality, thereby accelerating the transition to commercial-scale manufacturing of mRNA therapeutics. Moreover, the performance was attained using either of two mechanistically distinct engineered T7 RNA polymerase (RNAP) mutants. Both mutants enabled robust mRNA synthesis under process conditions that have not been previously reported. These observations provide mechanistic insights into IVT and advance our fundamental understanding of this critical bio-manufacturing step.

As bioprocess intensification has increased over the last 30 years, yields from mammalian cell processes have increased from 10s of milligrams to over 10s of grams per liter. Most of these gains in productivity have been due to increasing cell numbers in the bioreactors, and with those increases in cell numbers, strategies have been developed to minimize metabolite waste accumulation, such as lactate and ammonia. Unfortunately, cell growth cannot occur without some waste metabolite accumulation, as central metabolism is required to produce the biopharmaceutical. Inevitably, metabolic waste accumulation leads to decline and termination of the culture. While it is understood that the accumulation of these unwanted compounds imparts a less than optimal culture environment, little is known about the genotoxic properties and the influence of these compounds on global genome instability. In this study, we examined the effects on Chinese hamster ovary (CHO) cells genome sequences and physiology due to exposure to elevated ammonia levels. We identified genome-wide de novo mutations, in addition to variants in functional regions of certain genes involved in the mismatch repair (MMR) pathway, such as DNA2, BRCA1 and RAD52, which led to loss-of-function and eventual genome instability. Additionally, we characterized the presence of microsatellites against the most recent Chinese Hamster genome assembly and discovered certain loci are not replicated faithfully in the presence of elevated ammonia, which represents microsatellite instability (MSI). Furthermore, we found 124 candidate loci that may be suitable biomarkers to gauge genome stability in CHO cultures.

Extracellular vesicles (EVs) are mediators of intercellular communication, yet their impact on Chinese Hamster Ovary (CHO) cell physiology and bioprocess performance remains poorly understood. Here, we investigated whether small EVs (sEVs) and large EVs (LgEVs) that accumulate during fed-batch and perfusion cultures modulate CHO cell growth, metabolism, apoptosis, and monoclonal antibody (mAb) production. EVs isolated from early- and late-stage cultures were added to fresh CHO cultures grown with or without glutamine supplementation. Only LgEVs had a significant impact. Late-stage LgEVs markedly altered CHO-cell behavior, reducing cell proliferation, increasing apoptosis under glutamine-limited conditions, and substantially enhancing mAb productivity in a dose-dependent manner. Glutamine supplementation largely alleviated the growth-inhibitory and pro-apoptotic effects of LgEVs while preserving their positive impact on productivity, suggesting that glutamine decouples EV-mediated stress from productivity enhancement. Metabolic analyses revealed increased glucose consumption, a glutamine-dependent shift between glycine and alanine overflow metabolism, and remodeling of amino-acid utilization. Metabolic flux analysis further demonstrated enhanced glycolytic overflow and increased reliance on amino acid-supported anaplerosis. Conversely, selective removal of LgEVs from perfusion medium significantly improved cell expansion without reducing antibody production, supporting an inhibitory role for late-stage LgEVs. These LgEVs were enriched in let-7 family miRNAs and miR-21, consistent with RNAseq analyses demonstrating stress-associated enrichment of these miRNAs in CHO EVs and with functional studies showing that let-7a and miR-21reduce CHO-cell growth. Together, these observations suggest that selective miRNA loading contributes to the growth, metabolic, and productivity phenotypes elicited by late-stage LgEVs. Our findings identify LgEVs as endogenous regulators of CHO-cell physiology and potential targets for optimizing high-density fed-batch and perfusion biomanufacturing processes. HighlightsO_LIEndogenous late-stage Large Extracellular Vesicles (LgEVs) reduce CHO cell growth but boost specific mAb productivity. C_LIO_LIGlutamine supplementation rescues LgEV-mediated growth inhibition and apoptosis. C_LIO_LIMetabolic Flux Analysis (MFA) based on the dynamic behavior of amino acid and other metabolite and substrate concentrations reveals the pyruvate node as a metabolic bottleneck and the associated lactate overflow metabolism as resulting from LgEV exposure. C_LIO_LIStress-associated let-7 and miR-21 microRNAs are highly enriched on a per-EV basis in late-stage LgEVs. C_LIO_LISelective removal of LgEVs improves perfusion cell growth without impacting antibody titer. C_LI

Transfusion-ready red blood cells can be cultured ex vivo from hematopoietic progenitors. Despite its promising outlook, a cultured transfusion unit cannot be produced at competitive costs. Large media volumes are required to maintain a maximum erythroblast cell density of 1-2.106 cells/mL during the erythroblast proliferation stage. To identify the origin of the cell density limitation, we investigated the cellular support and metabolomic phenotype using different media formulations and feeding regimens. Media that were exposed to an increasing density of erythroblasts (termed spent media) displayed a proportional decrease in erythroblast proliferation support. A 1:1 combination of spent media with fresh media (not previously exposed to the cells) restored growth for all tested conditions. Filtering both fresh and spent media with a 3 kDa cut-off filter, and subsequent recombination of the two fractions, indicated that exhaustion of the small molecular weight fraction (<3 kDa) was primarily responsible for growth limitation. We performed targeted and untargeted metabolomics analysis, for both the intra- and extracellular compartments, following seeding in fresh medium (12, 24, 36 h). We observed degradation of nucleosides, depletion of amino acids, and a decrease in intermediates of the glutathione-ascorbate, {gamma}-glutamyl and cysteine-methionine cycles. The latter compounds suggested an increase in oxidative stress in high density erythroblast cultures. Elimination of nucleosides from the medium led to a lower accumulation of purine salvage intermediates, and a 30% increase in cell productivity. In conclusion, we demonstrate that high-density erythroid cultures are subject to metabolic stress, defining critical constraints for scalable culture expansion.

Cell-free transcription and translation systems promise to accelerate and simplify the engineering of proteins, biological circuits and metabolic pathways. Their encapsulation on microfluidic platforms can generate millions of cell-free reactions in picoliter volume droplets. However, current methods struggle to create DNA diversity between droplets while also reaching sufficient protein expression levels. In particular, efficient multi-gene expression has remained elusive. We here demonstrate that co-encapsulation of DNA-coated beads with a defined cell-free system allows high protein expression while also supporting genetic diversity between individual droplets. We optimize DNA loading on commercially available microbeads through direct binding as well as through the sequential coupling of up to three genes via a solid-phase Golden Gate assembly or BxB1 integrase-based recombineering. Encapsulation with an off-the-shelf microfluidics device allows for single or multiple protein expression from a single DNA-coated bead per 14 pL droplet. We envision that this approach will help to scale up and parallelize the rapid prototyping of more complex biological systems.