Microfluidics enabled multi-omics triple-shot mass spectrometry for cell-based therapies Free

In the past decade, cell-based therapies have become transformative for cancer treatment and beyond. In these therapies, cells from a patient (autologous) or donor (allogenic) are used to repair or enhance the biological function of a damaged or diseased system.¹ In 2017, the FDA approved the first cell-based immunotherapy, Kymriah™ (Tisagenlecleucel), a chimeric antigen receptor (CAR) T-cell therapy.² Kymriah and other subsequently approved CAR-T cell therapies have produced remarkably effective and durable clinical responses for treatment of leukemia and other hematological malignancies. On the heels of these pioneering successes, over 20 cell therapies are now FDA approved and thousands of other cell-based therapeutics are currently in development.³ 

Despite their tremendous clinical outcomes and potential, cell-based therapies present unique discovery, characterization, and manufacturing challenges as compared to conventional small-molecule drugs and biopharmaceuticals. Small-molecule drugs have established protocols that ensure the accurate synthesis, purity, and potency of the product.^4,5 In the established biopharmaceutical industry, the final product is a single, medically potent biomolecule produced by a cell, e.g., a monoclonal antibody. However, for advanced cell-based therapies, the final product is a population of living cells, each individually comprised of thousands of diverse biochemicals.⁶ For example, during the CAR-T manufacturing process, patient derived cells undergo multiple complex biochemical processes and genetic engineering steps in vitro before they are delivered to a patient after a multi-week manufacturing process.⁷ The large number of diverse and dynamic biomolecules associated with cell-based therapies renders the process of characterizing and controlling the manufacturing of cell-based therapies significantly more difficult, as it requires the understanding and monitoring of numerous interrelated biomolecules that together impact interactions among millions of cells.

It is essential to identify and screen for critical quality attributes (CQAs) that have a direct correlation with the clinical success of a cell therapy during its manufacturing process.⁸ Once CQAs are identified, they can be used as biomarkers of safety and potency to monitor cell therapies as they are developed. The identification of CQAs has commonly been limited to a few individual biomarkers. These might include intracellular, surface-bound, and/or secreted cell constituents. We believe that to truly understand the significance of these biomarkers, it is essential to also comprehend the spatial and dynamic patterns by which these biomarkers behave collectively as an interacting group.⁹ Additionally, gaining insight into the evolution of these biomarkers at the local (individual or few cell) level will provide crucial and novel insights into the heterogeneity of the therapy during its manufacturing, which will likely affect the therapeutic efficacy. Biomarkers that are representative of the cellular state, with particular emphasis on those associated with cell signaling, are strong contenders for relevant CQAs. Ultimately, we expect that the required combination of target biomarker classes should include secreted signaling molecules (i.e., secretome), metabolites in the cells’ interior (i.e., intracellular metabolome), and cell surface markers (i.e., surfaceome). Additionally, the measurement of concentrations of relevant molecules of all three classes should occur at the “same time and place.”

A cell’s secretome consists of secreted biomolecules crucial for cell signaling, communication, and migration, while its intracellular metabolome reflects modifiable metabolic pathways used to modulate cell signaling, and its surfaceome encompasses surface-bound proteins that are integral for cell–cell/cell–environment communication and signal transduction.^10–12 The primary common factor among these three “omes” is their significant role in cell signaling. Intra- and extra-cellular vesicles could also be included in a broad definition of the cell's “omes” as they carry an important function in cell communications, but they warrant a separate dedicated consideration due to complexity of their structural organization and molecular composition.¹³ Surface-exposed proteins serve as key regulators connecting the intra- and extracellular signaling networks, facilitating the conversion of external signals into internal responses, and vice versa.¹⁴ The extracellular metabolome has been used to make inferences regarding the intracellular metabolome in various studies;^15,16 however, the exact correlation between biomarkers from different “omes” and especially their resulting impact on cells’ therapeutic potency and efficacy is yet to be fully established on a fundamental biochemical basis. Simultaneous investigation of biochemical networks and molecular enrichment using multi-sample, multi-omics data is an important avenue for research and would provide valuable insights and enhance our understanding of cellular response mechanisms.¹⁷ 

All aspects of cell-based therapies are dynamic, including both the evolution of cell cultures (in terms of growth, transduction, differentiation, and activation) throughout the manufacturing process as well as their therapeutic action when administered to patients. These changes occur on a multitude of time scales (from seconds to minutes to hours), and they involve the orchestrated interactions of individual cells within the cell population concurrent with evolution (production, consumption, and conformation) of biomolecules inside, outside, and on the surface of the cell. Researchers do not yet fully understand these dynamic processes, especially those involving shorter time scales, nor their corresponding impact on the therapeutic potency. Marasco et al. put it eloquently that “time remains a largely underappreciated or neglected variable in most comprehensive cellular response experiments, not as much for its perceived lack of value as for the difficulty of precise temporal resolution and control of measurements.”¹⁸ This is further emphasized by Faley et al. as an important consideration for understanding single cell dynamics over time.¹⁹ Thus, an ability to identify relevant biomolecules and track their temporal changes using rapid screening methods are essential for both discovery of new biomarkers and sensitive measurement of known CQAs, which can be applied to near-continuous cell culture state monitoring for possible real-time feedback control during therapy manufacturing.

Despite being widely utilized, continuous measurements are often limited to assessing properties of the bioreactor media (such as pH, temperature, and dissolved gas concentrations) that only indirectly correlate with the cell state.²⁰ The reliance on these indirect methods is not due to confidence in their utility but because there is a notable lack of direct analytical methods capable of untargeted identification (discovery) and targeted monitoring (process control) of molecular biomarkers of cell-therapy cultures in real time. Achieving the necessary high sensitivity detection of CQAs from multi-omes has primarily relied on a combination of several different offline techniques such as microarrays, enzymatic assays (such as ELISA), flow cytometry, nuclear magnetic resonance (NMR) spectroscopy, and liquid chromatography (LC) mass spectrometry (MS). These techniques generate different types of data often with little complementarity and present a challenge for integrated data analysis that could yield a cohesive narrative on the underlying biology. Additionally, these approaches face notable challenges in terms of significant time delays, large size and diversity of sample types, and substantial manual handling with limited throughput and loss of precious analytes. These challenges hinder their effectiveness in understanding the cell culture dynamics on the time scale of relevant biological processes, especially when the fast biochemical changes (on the scale of minutes) are important.

LC-MS is a powerful analytical technique that can be used for the detailed discovery of biomarkers in cells and cell therapy products. MS offers extensive coverage of biochemical compounds with high sensitivity and specificity across a wide range of molecular weights, classes, and concentrations.^21,22 The typical LC-MS workflow involves extraction, homogenization, centrifugation, and the separation of complex mixtures based on their physicochemical properties using a stationary phase and a mobile phase.²³ Although LC-MS provides detailed information regarding the composition of individual biomarkers, there are inherent limitations regarding its speed, throughput, and manual handling as well as technological challenges for their implementation in the lab-on-a-chip format. Single-shot electrospray ionization (ESI) MS allows direct analysis of a sample without the need for prior separation, unlike traditional LC-MS. ESI-MS retains the structural integrity of large molecular weight biomolecules, minimizes fragmentation, and, thus, can be used for untargeted biomarker detection.^24,25 However, when no separation is utilized, it is significantly more challenging to identify analytes from an intricate multi-component MS spectra due to overlapping signals and co-elution.²⁶ Additionally, ionization efficiency, charge suppression, interferants, and preferential loses of different ions between electrospray and MS will affect the signal intensity and identification of analyzed biomarkers. There are several powerful solid, liquid, and gas phase separation techniques that can be used prior to sample analysis by MS, but they all add to the total time required for analytical output. In particular, a combination of solid phase extraction (SPE), ultra-performance LC, and ion mobility separation (IMS) could mitigate the concern of co-elution and overlapping signals for improved analyte identification; yet the speed and sample sizes required of this multi-step sample preparation may not be adequate for monitoring rapid changes in biomarkers during cell manufacturing.^27,28 There exists a trade-off between increasing the speed of analysis and reducing the sample size. Furthermore, minimizing the required sample size is important for limiting the amount of valuable therapeutic product (CAR T-cells) that needs to be extracted for analysis during therapy biomanufacturing. For instance, Zhang et al. accomplished rapid (15 s) analysis of metabolites using an automated SPE-IMS-MS workflow with Agilent's RapidFire SPE sample preparation.^29,30 However, these packed-bed SPE cartridges use large volumes (30 μl loaded per sample) and are not suitable for applications that target a small number of cells. Solid phase microextraction has been introduced to address the volume limitations; however, existing systems require time-consuming steps (∼30 min) and offer limited coverage of biomolecules largely limited to metabolites.^30,31 Utilizing microfluidic sample preparation systems in conjunction with mass spectrometry presents a significant opportunity to harness the advantages of single-shot ESI-MS while gaining a portion of the benefit obtained from LC-MS, creating a comprehensive analytical approach for untargeted biomarker discovery and characterization. Microfluidic systems offer precise control over small volumes and enable rapid analyte transport. To sort between different “omes” and gain greater sample control, it becomes essential to employ cell “traps” and barriers within microfluidic systems. Numerous trapping techniques have been documented in scientific literature, serving as a crucial foundation for design innovation.^18,19,32,33 Microfluidic systems can directly integrate with sensing technologies for real-time monitoring of cell behavior. The ability to analyze small populations of cells proves especially advantageous for direct sampling and analysis of the cell state of a bioreactor without disrupting the therapy as a whole. Microfluidic systems also allow for high throughput, multiple point (in time and space) measurements to assess the cell culture's spatial-temporal heterogeneity.

In many cases, the limitations of sample preparation hinder analytical systems when both a broad range of analyte identification and rapidity of analysis are desired. Through the advancement of fast and efficient microfluidic sample processing techniques, microfluidics-enabled MS has the potential to become a unified platform for direct, real-time, and multi-omics discovery and analysis of CQAs. This Perspective article examines the use of microfluidic-enabled MS systems to detect and characterize CQAs resulting from a cell-therapy product's secretome, intracellular metabolome, and surfaceome. In particular, the feasibility of rapidly assessing all these cellular multi-omes at “the same place and the same time,” from a small sample of cells obtained from a bioreactor, each in a single-shot ESI-MS manner, will be evaluated. We envision a fully automated platform that removes all specific analytical expertise from sample preparation. This establishes data continuity and ensures that measurements across all biochemical classifications correspond precisely to the same temporal reference points, thereby enhancing data coherence and facilitating the capture of dynamic processes. Moreover, owing to the speed and constrained sample sizes of microfluidics, analyses could be conducted more frequently, potentially enabling the real-time capture of dynamic changes. Such a system is illustrated in Fig. 1. The ability to detect biomarkers from all three of the target “omes” using a single system with integrated functions targeting different classes of biomarkers does not currently exist. We refer to a vision of such a bioanalytical discovery-monitoring micro-total analysis system (μTAS) as an enabler of “triple-shot” mass spectrometry, in which the secretome, intracellular metabolome, and surfaceome are portioned, extracted, and selectively processed using a microfluidics with control on the scale of individual cells followed by analysis of each “ome” via single-shot ESI-MS. In this Perspective, we will discuss the current state-of-the-art in microfluidic enabled MS for measuring the cell secretome, intracellular metabolome, and surfaceome to assess their readiness for future integration into a powerful “triple-shot” ESI-MS μTAS.

As therapeutic cells develop and mature, they secrete various paracrine factors that can be used to assess their condition and therapeutic efficacy.^5,34,35 Secreted proteins actively engage in diverse physiological functions, encompassing immune defense, blood coagulation, and cell signaling.³⁶ Additionally, they play critical roles in pathological processes, such as cancer angiogenesis, differentiation, invasion, and metastasis.³⁷ The concentration of these secreted molecules rapidly decreases with distance from the cell's surface, and detecting these biomolecules is difficult without a localized sampling technique that captures the concentrated secreted biomolecules in the cellular microenvironment before they become diluted in the surrounding bulk media.¹⁹ If, in addition to localized sampling, rapid sampling is possible, then transient changes of the cells’ secretome can be tracked. Use of ESI-MS for direct analysis of secreted biomolecules in cell culture is complicated by the need to remove salts that result in adduct and cluster signatures that convolute the resulting MS spectra.³⁸ 

Xiang et al. developed a microfabricated device that provides online dual microdialysis and electrospray ionization (ESI) MS of biological samples.³⁹ By utilizing both high-molecular-weight and low-molecular-weight cut-off membranes, the researchers were able to target biomolecules of interest and remove interfering molecules from their ESI-MS signal. Although this system effectively filtered the sample, the large dead volume (∼80 μl) in the system led to inadequate time resolution of the analysis (∼20 min) and required a large sample size (∼150 μl). Moreover, the hand assembled device is not suitable for practical applications requiring production scale-up.

Progress has been made to increase the speed of microdialysis sample preparation for ESI MS. Olivero et al. demonstrated a continuous, locally sampled in vitro biochemical detection system that utilizes inline microdialysis to remove interfering salts as well as ESI-MS to analyze the sample directly from a pressurized cell culture.⁴⁰ The employed cellulose dialysis membrane resulted in substantial mass transfer resistance for separation, which led to a moderate improvement in analysis speed (∼1 min from sampling in a bioreactor to MS spectra acquisition). Tibavinsky et al. presented a new microdialysis method with a monolithically integrated ultra-thin nanoporous alumina membrane, which substantially reduced the mass transfer resistance of trans-membrane separation and sped up the online sample treatment further (∼1 s).⁴¹ 

Chilmonczyk et al. advanced microfluidic desalination technology coupled with ESI-MS for secretome characterization even further with the dynamic sampling platform (DSP) (Fig. 2).⁴² The DSP is a microfabricated mass exchanger with a nanoporous alumina membrane and a sample conditioning channel. The DSP system can be used to provide direct-from-culture monitoring of secreted large molecules and includes a localized sampling interface, an optimized “cross flow” sample treatment mass exchanger, and an inline ESI emitter for online MS analysis. In addition to effective desalting of the sample and retention of high molecular weight proteins, the use of active sample conditioning provided an opportunity to introduce MS signal enhancing compounds into the sample prior to the ESI-MS analysis. In total, over 99% of salts were filtered from the sample enabling increases in MS signal intensity and signal-to-noise ratio and improving the sensitivity for low-concentration biomolecules. The use of active sample conditioning through introduction of 3-nitrobenzyl alcohol (m-NBA) lead to two to three times improvement in sensitivity of protein detection at sub-nanomolar protein concentrations.⁴³ 

The DSP presented by Chilmonczyk et al. provides a pathway to rapid, localized, and direct analysis of the secretome locally from the immediate surrounding of individual cells.⁴³ This approach is immediately applicable to adherent cell cultures but would require further development of the sampling interface to be able to monitor cells in suspension. The effectiveness of this technology could benefit from additional separation methods beyond size exclusion, including multi-stage analysis with add-on separation units in order to prevent the loss of smaller proteins and other low molecular weight biomolecules. Additionally, the DSP could be adapted to analyze extracellular vesicles to fully characterize the secretome. With these advancements, this technology could be well equipped to provide discovery and monitoring of CQAs and has the potential to be utilized with other platforms for multi-omics analysis.

In addition to assessing the secretome of a cell therapy product, the comprehensive analysis of intracellular content is crucial for developing and evaluating the therapy’s effectiveness. In particular, the intracellular metabolome has been a focal point for CQA discovery and monitoring. The metabolome primarily comprises molecules that participate in cellular metabolism, including amino acids, nucleotides, and lipids.⁴⁴ It serves as a predictive indicator of cellular phenotype and function, and it offers a momentary snapshot of a cell’s reaction to genetic or environmental influences.^45,46 The direct examination of the primary intracellular metabolome, as well as other species that impact cell metabolism in a statistically significant manner, offers valuable information about the cellular condition and provides essential insights into the safety, effectiveness, and potency of the resulting cell therapy product.

The detection of these intracellular biomarkers is challenging due to their diversity in chemical and physical properties and wide concentration range (over nine orders of magnitude, pM to mM).⁴⁷ Conventional workflows for intracellular MS necessitate multiple sample preparation steps, involving the isolation, extraction, and separation of intracellular contents, resulting in a time-consuming analysis process that requires a large numbers of cells. This sample preparation sequence involves manual handling of cell batches and is not suitable for continuous flow analysis. As a result, standard methods are incapable of capturing the dynamic and heterogeneous internal metabolic processes with sufficient rapidity to enable real-time process monitoring and discovery.⁴⁸ 

Hofstadler et al. and Kawai et al. presented single-cell MS techniques utilizing capillary electrophoresis to provide flow separation of charged species.^49–51 Although these techniques can be used to effectively attain metabolomic information, they require complex cell micro-manipulation procedures to ensure selective sampling of the intracellular environment. This limitation restricts their applicability for real-time, continuous monitoring of the cell population state within a biomanufacturing workflow.

Culberson et al. presented a platform to isolate cells, provide buffer/media exchange, and extract intracellular content in a single microfluidic device (Fig. 3).⁵² This platform achieved near-continuous monitoring by directly extracting small cell samples from culture. In this multi-functional system, the samples enter a microfluidic cell processing device where cells are immobilized by microfabricated pillars and media are rinsed. Intracellular components are released by electrical pulses applied across lysis electrodes incorporated in the device, and the flow is then injected to an ESI capillary for direct infusion ESI-MS. This system achieved comprehensive detection of a diverse range of relevant intracellular metabolites (up to 80% of all relevant amino acids) without requiring specific targeting. Additionally, the system could be used to accurately assess whether cells were activated, and it was able to detect 18 upregulated metabolic pathways for the activated T-cells as well as four upregulated metabolic pathways for the non-activated T-cells. However, it is unclear whether all intracellular metabolites are extracted from the cell, and there are likely losses within the cell processing device due to the nature of cell immobilization and lysis resulting in aggregation of the cell debris that may impede metabolite extraction.

The platform presented by Culberson et al. demonstrated rapid (∼10 min per cell sample) and direct from the culture analysis of the intracellular metabolome from a small (about 1000 or fewer) number of suspension cells that could be sampled locally from a bioreactor.⁴⁴ Improvement of the microfluidic technology capable of sampling adherent cell types, such as mesenchymal stem cells (MSCs), in addition to cells in suspension is an important next step. Overall, this platform can be used not only for dedicated metabolome characterization, but also has the potential to be combined with other sample preparation methods (e.g., DSP for secretome) for the comprehensive discovery and monitoring of CQAs across multiple “omes.”

The final component of cell-therapy product screening should involve the extraction, isolation, and analysis of biomarkers on the cell's surface, referred to as the surfaceome. The surfaceome acts as a crucial interface for cellular communication with the extracellular environment and often exhibits early indications of cellular distress and disease.^53,54 Cells use surface markers for intercellular recognition and communication.⁵⁵ These surface markers are also commonly used by researchers to identify and classify cells for their therapeutic potential. Differentiation markers and inhibitory markers are two important classes of surface markers for chimeric antigen receptor (CAR) T-cell therapy characterization. CAR-T cell products with less differentiated T-cells have a stronger potential for expansion, persistence, and tumor eradication.⁵⁶ Inhibitory markers, also referred to as exhaustion markers, have been reported in higher quantities in patients whose CAR-T cells persist for shorter periods of time and patients who fail to respond to treatment.⁵⁷ 

Membrane proteins constitute approximately 30% of the proteins encoded by the human genome; however, they are vastly underrepresented in the majority of proteome profiles.⁵⁸ Researchers have uncovered the structures of only about 50 membrane proteins out of the roughly 8000 total that have been identified.⁵⁹ This lack of profiling of the surfaceome can be attributed to the heterogeneity of the membrane proteins, hydrophobicity of the protein's subunits embedded in the plasma membrane, and the overall low abundance of the surface-bound proteins compared to the intracellular proteome.^60,61 The use of machine learning approaches such as AlphaFold-II has promise to increase the number of putative structure annotations for membrane proteins.⁶² 

Microfluidics-enabled flow cytometry is the primary approach utilized for the detection of cell surface markers. The benefits of flow cytometry include the ease of use, fast analysis, and the ability to obtain multi-parameter data on individual cells.⁶³ However, flow cytometry is a targeted technique, and it requires researchers to have a priori knowledge of which surface markers should be analyzed. The targeted nature of conventional flow cytometry also necessitates lengthy sample preparation (specifically incubation with antibody conjugates) that increases the time and cost necessary to use flow cytometry.⁶⁴ 

Mass spectrometry offers the attractive possibility for discovery rather than only targeted monitoring. The most effective strategy for characterizing the surface proteome using MS involves the isolation and enrichment of cell surface proteins. Prior to MS analysis, it is essential to separate the cell membrane from its intracellular components and employ solubilization methods to extract membrane proteins. Standard solubilization techniques include the use of detergents, nanodiscs, and the most recent method, known as Sonication of Lipid Vesicles for Mass Spectrometry (SoLVe-MS), which creates enriched lipid vesicles.^65,66 An illustration of surface marker isolation, enrichment, and detection is shown in Fig. 4.

The most common approaches to isolate the cell membrane involve ultracentrifugation and density-gradient centrifugation. These techniques are prone to contaminants from organelles with densities similar to that of the cell membrane, such as the mitochondria and endoplasmic reticulum.⁶⁷ Additionally, these methods require a large amount of starting material and extended time to extract the desired cell surface fraction.⁶¹ Surface protein enrichment has also been attempted by surface property-based isolation methods such as adhesion to polylysine-coated glass beads or plates,^68,69 free-flow electrophoresis,⁷⁰ and two-phase partitioning.⁷¹ However, these techniques typically result in a low yield, and a time-intensive conditioning process also prone to contaminants.

Chemical covalent coupling to free side groups of surface-exposed proteins or carbohydrates has also been used as a technique to enrich cell surface proteins for MS analysis. Cell surface biotinylation is a method used to selectively label surface-exposed primary amines of lysine, using a biotinylation reagent.⁷² Smolders et al. implemented this technique to evaluate small-tissue samples and identify approximately 1000 surface proteins.⁷³ 

However, to mitigate the presence of contaminants, additional enrichment steps are necessary, such as ultracentrifugation and targeted labeling. As mentioned earlier, these enrichment methods have their own limitations, including large starting samples and extended processing times. Additionally, cell surface biotinylation is only useful for proteins with surface exposed lysines, rendering it ineffective for proteins lacking this characteristic. Similar methods have been identified to selectively label surface-exposed carboxyl groups⁷⁴ and glycan chains.⁷⁵ These methods involve more processing steps and reagents, which further increase the time and cost required, making them unsuitable for dynamic surfaceome discovery and monitoring.

Currently, there is a notable lack of microfluidics-enabled surface marker isolation and enrichment techniques to facilitate rapid and untargeted analysis of the proteins on the cell's surface. Although several chemical methodologies have been developed, these techniques are still targeted and require large sample sizes and processing times for efficient labeling.⁶¹ As such, they are incompatible for integration with small-sample microfluidic systems and rapid analytics. Although ultracentrifugation also requires large sample sizes and processing times, it is currently the best method for untargeted membrane isolation for surfaceome discovery and analysis. Thus, filling the unmet need for a microfluidic cell surface marker isolation method might best be inspired by consideration of novel and creative approaches based on the same underlying principle as centrifugation, i.e., separation based on size and density. We consider this the key bottleneck that must be overcome in order to achieve the vital goal of developing the capability for rapid microfluidics-enabled analysis of small (approaching single-cell) samples for characterization of all three “omes.”

We envision an integrated system capable of rapidly characterizing CQAs from a cell-therapy product's secretome, intracellular metabolome, and surfaceome, which we refer to as “triple-shot” ESI-MS μTAS. The key aspect of “triple-shot” ESI-MS μTAS as a tool for dynamic monitoring is the ability to perform all analytical steps rapidly. State of the art proteomics techniques provide unprecedented capability to analyze the complete protein content (inside the cell, outside the cell, and on the surface of the cell); however, it comes with a HPLC-MS time penalty, and there is no clear pathway to address this critical concern. “Triple-shot” ESI-MS μTAS aims to achieve rapidity of the analysis while managing proteome, secretome, and sufaceome coverage through the use of advances in data analytics.

This integrated system could be operated in two modes: parallel, where samples for each “ome” are concurrently analyzed in three separate microfluidic pathways, or in series, where the same sample is sequentially analyzed for each “ome” within a single microfluidic pathway. A parallel and series schematic is shown in Fig. 5. The parallel system’s advantages lie in its potential for faster performance and easier integration in comparison to a series system. This potential for analysis time reduction is due to the capability to process and analyze all three “omes” simultaneously. This advantage comes with the cost of needing multiple mass spectrometers if the MS analyses time is longer than the sample preparation times (MS data acquisition requires ∼1 min for each “ome”); a single MS can be used with automated setting changes if MS analysis is substantially shorter than each sample processing step. Another potential integration benefit of parallel operation results from eliminating the need to preserve various sample aspects for subsequent “omes” screening steps. The series system’s advantage lies in its ability to analyze the secretome, intracellular metabolome, and surfaceome from the very same sample of cells, and it requires a smaller total number of cells to be sampled. The workflow includes three steps: (1) trap cells and extract media (replacing with salt-free isotonic buffer) to capture the secretome through in situ desalting, (2) lyse cells and purge their contents with an aqueous plug to measure intracellular metabolites, and (3) extract membrane constituents from the remaining cell debris (captured within the cell trap after lysis) via enzymatic digestion/detergents to release and isolate surface markers for MS identification. There is potential to broaden this system by integrating extra separation stages (e.g., within steps 1 and 2) aimed at isolating both extra- and intra-cellular vesicles.

For a parallel system, the total analysis time would be approximately that of the longest single sample preparation/analysis step, while for the series system, the total time would be the sum of all of the sample preparation/analysis steps. Therefore, the potential time savings of parallel operation is only truly available if no single step is extremely long compared to the others. The current benchmark secretome processing and MS analysis platform of Chilmonczyk et al. require approximately 1 min from sample to analytical output for secreted protein analysis.⁶ In contrast, the exemplar intracellular metabolome platform of Culberson et al. requires approximately 10 min from sample to analytical output for intracellular metabolite analysis.⁴⁴ If the sample preparation and analysis of the intracellular metabolome consume the majority of the total time, one might explore intriguing alternatives such as the extracellular metabolome. Secreted metabolites can serve as reliable indicators of internal cellular processes. They can be readily assessed without disrupting the overall system and exhibit rapid dynamics, changing within seconds to minutes.¹⁸ Marasco et al. developed a microfluidic system paired with online desalting and mass spectrometry to rapidly assess the exometabolome or secreted metabolites.¹⁸ By utilizing solid phase extraction, the researchers desalted the cellular effluent through a process of several sample loops and LC columns. Practical use of exometabolome monitoring for assessing the cell's internal metabolism requires a comprehensive understanding of the correlation between intracellular and extracellular metabolites, which is currently lacking.

Finally, there is a lack of microfluidic-based methods capable of rapid surfaceome processing for MS analysis. The most common method of surfaceome protein enrichment, ultracentrifugation, requires a lengthy centrifugation step to maintain purity of the integral membrane fraction lasting ∼10 h.⁷⁶ Therefore, the goal for the development of a microfluidic rapid surfaceome technology should be to dramatically decrease the processing time to about that of the metabolome process, i.e., ∼10 min, preserving the possibility of exploiting parallel operation if desired. This Perspective aims to generate awareness of this important analytical challenge and motivate the biomicrofluidics community to invest efforts in developing innovative approaches for rapid cell surfaceome extraction, preparation, and analysis.

This Perspective highlights the need and opportunities for fast and comprehensive analytical methods to discover biomarkers of cell therapies. Invention and development of integrated microfluidic platforms for rapid, locally sampled biochemical analytics are motivated by the complex and dynamic nature of cells as a therapy and its evolution during ex vivo manufacturing and in vivo therapeutic application. Enabling on-demand or real-time analysis of all three “omes” will allow for better discovery of new therapeutic modalities, development of new therapies, and low-cost manufacturing workflows for advanced biotech processes at scale. Although there have been recent advances in the development of promising techniques for the rapid and paired detection of the secretome and intracellular metabolome, there is a notable lack of viable approaches for surfaceome monitoring that could be paired with other multi-omic screening methods. Addressing this gap and developing integrated multi-ome approaches for comprehensive monitoring of the cell state “at the same time and the same place” would greatly enhance our understanding of cell therapies, including culture heterogeneities, and improve efficiency of manufacturing through better process control. Mass spectrometry, the gold standard in molecular analysis, presents a powerful biosensing tool that is sensitive and specific enough to detect even trace amounts of biochemicals. By coupling microfluidic platforms for sample preparation to MS, it is possible to expand the utility of MS for cell-therapy development and manufacturing workflows. Further advancements in microfluidic technology coupled to MS, especially the application of single-shot ESI-MS and its enhancement through integrated secretome, metabolome, and surfacome profiling, such as triple-shot ESI-MS μTAS proposed here, are necessary to enable the rapid and continuous analysis of cell therapy products across multiple dimensions, thereby increasing our understanding of these therapies, reducing developmental timelines, and improving patient accessibility to these life-changing treatments.

The work described herein was supported by the National Science Foundation (NSF) Center for Cell Manufacturing Technologies (CMaT) Award No. 1648035, the Marcus Center for Therapeutic Cell Characterization and Manufacturing Collaboration Grant in Cell Manufacturing (Georgia Tech Foundation), and the Georgia Research Alliance. Partial support was also provided by Grant No. R01 GM138802 from the National Institute of General Medical Science (NIGMS), a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NSF, NIGMS, or NIH. Device fabrication was performed at the Georgia Tech Institute for Electronics and Nanotechnology, a member of the National Nanotechnology Coordinated Infrastructure (NNCI), which is supported by the NSF Grant No. ECCS-2025462. Figures 1, 4, 5, and the summary figure were created using BioRender.com.

A.G.F., M.A.C., and A.L.C. are pursuing commercialization of the DSP technology discussed in this article. The terms of this arrangement have been reviewed and approved by Georgia Tech in accordance with its conflict-of-interest policies. The remaining authors declare no competing interests.

Gianna A. Slusher: Conceptualization (equal); Writing – original draft (lead). Peter A. Kottke: Conceptualization (equal); Writing – review & editing (supporting). Austin L. Culberson: Conceptualization (equal); Writing – review & editing (supporting). Mason A. Chilmonczyk: Conceptualization (equal); Writing – review & editing (supporting). Andrei G. Fedorov: Conceptualization (equal); Writing – review & editing (lead).

Data sharing is not applicable to this article as no new data were created or analyzed in this study.