Enhancing the Scalable Manufacturing of Human Mesenchymal Stromal Cells in Dynamic Culture Using Bioactive Heparin/Collagen Layer‐by‐Layer Coated Microcarriers

ABSTRACT

Expansion of human mesenchymal stromal cells (hMSCs) remains a bottleneck during manufacturing, where high cell yields must be achieved without compromising quality. Microcarriers provide elevated surface areas but often rely on synthetic materials. Here we use heparin/collagen Layer‐by‐Layer (LbL) coatings on microcarriers to enhance dynamic hMSC culture. HMSCs were cultured on LbL‐coated and untreated microcarriers in up to 500 mL flasks, and were assessed over time, up to 18 days across three seeding densities (2600–20,000 cells/cm²), while yield was assessed at the end of culture. Metabolic analysis suggests no induction of a generalized stress response, increased glycolytic activity, and a shift toward a more active secretory state for cells grown on coated surfaces, as measured by noninvasive assays (Luminex multiplex and Raman spectroscopy). At medium and high densities, the cost (18.6%) and yield (60% and 30%, respectively) increases (coated: untreated) were moderately favorable. However, at the lower density, the cost increase remained similar, despite a yield increase of 110%. Together, the findings of this study demonstrate that our LbL surface coatings enhance hMSC functional performance and scalability of microcarrier‐based culture, especially at lower initial cell densities, representing a promising approach for scalable cell manufacturing with the potential for GMP adaptation.

Graphical Abstract and Lay Summary

Successful manufacturing of human mesenchymal stromal cells (hMSCs) for cell therapy applications requires large quantities of potent cells per dose. Expansion of hMSCs remains a bottleneck during manufacturing, where high cell yields must be achieved without compromising quality. A current approach to manufacture large amounts of hMSCs relies on the use of spherical microcarriers for suspension‐based dynamic culture. However, many of the available commercial microcarriers rely on synthetic materials dissimilar to the cell's native environment. In this work, we demonstrate how a biomimetic coating composed of stratified layers of heparin and type I collagen can be used to coat microcarriers to enhance hMSC culture. We demonstrate a significant increase in yield, no induction of a generalized stress response, increased glycolytic activity, and a shift toward a more active secretory state for cells grown on coated surfaces compared to uncoated samples.

.

1. Introduction

Human mesenchymal stromal cells (hMSCs) are clinically important tools of regenerative and immunomodulatory therapy due to their multipotent nature and paracrine signaling capacity [1, 2]. Their clinical translation; however, still leaves a lot to be desired, including consistency, functional quality, and desirable, scalable yield.

Given that clinical batches can require hundreds of billions of cells, large‐scale therapeutic manufacturing presents significant challenges. This necessitates the development of systems that allow for minimization of passage number, maximization of cell yield, and maintenance of therapeutic efficacy [3, 4]. Protracted expansion presents both material benefits and logistic challenges, with complex economic consequences. Most notably, protracted expansion can increase risk factors to cell quality and potency, such as cellular senescence, replicative exhaustion, genetic drift, and increased risk for contamination [5, 6]. This fundamental balance between maximizing potency and achieving required cell yields, while minimizing passage counts, remains a major barrier to widespread and cost‐conscious clinical adoption of hMSC therapies [7].

To address these demands, scalable three‐dimensional (3D) suspension culture systems employing microcarriers have become central to hMSC manufacturing processes [8, 9]. Microcarriers are macroscopic beads that have surfaces treated to permit cell attachment and are often used with cells grown in spinner flask bioreactors [10]. They provide large surface area to volume ratios, enabling dense adherent growth in stirred bioreactors while maintaining cell–cell and cell–matrix interactions.

Many commercial carriers are based on inert polymers, such as polystyrene, with limited biological functionality. Despite research highlighting the significance of surface chemistry and topography having a profound influence on cell attachment, metabolism, and secretory behavior, the efforts to engineer more bioactive carrier interfaces have recently become a true focus [11, 12]. Among these up‐and‐coming research areas, layer‐by‐layer (LbL) polyelectrolyte coating assemblies have emerged as a versatile method to tailor the physicochemical and biochemical properties of microcarrier surfaces by altering electrostatic interactions and mimicking aspects of ECM. LbL assembly refers to the sequential deposition of oppositely charged polyelectrolyte layers, including polycations such as poly‐l‐lysine, collagen, and chitosan, and polyanions such as heparin, alginate, or hyaluronic acid, to construct tunable, often extracellular matrix‐mimetic or ECM‐mimetic coatings to enhance adhesion and regulate cellular signaling [7, 13, 14].

In our previous work, it was demonstrated that heparin‐ending, heparin‐collagen (HEP/COL), coatings significantly improved hMSC adhesion and viability under static 2D and microcarrier culture [15]. Additionally, studies have shown that hMSCs and more specifically hMSCs grown on HEP/COL coatings can limit production costs and reliance on serum when grown on microcarrriers [16]. In some cases, low‐ or no‐serum conditions showed increased adhesion and decreased clumping [12]. Building upon these findings, the next critical step is to determine how these coatings perform under dynamic, scaled‐up conditions, where shear stress, nutrient gradients, and constant mixing are added to the equation. Understanding these coating‐dependent effects on hMSC growth and function will allow for the bridging of the gap toward high‐quality manufacturing for clinical translation. Which is why our objective was to determine the contribution of the HEP/COL relative to an unmodified control substrate rather than another variable surface.

Collagen, the most abundant protein in the body and a major structural component of cellular ECM, is a crucial component of the coatings. In addition, it serves as the polycation, enhancing adhesion and cytoskeletal organization through integrin binding [17, 18]. Heparin, an integral sulfated glycosaminoglycan of the body, acts as the final layer of our coatings for its observed impressive cytokine sequestering abilities. The polyanionic nature of HEP plays a big role in its potential to sequester and stabilize growth factors such as fibroblast growth factors (FGFs), prolonging and in some cases amplifying bioavailability [14, 18, 19, 20].

Although surface modifications of microcarriers have improved hMSC adhesion under static conditions, the scalability, biochemical consequences, and cost‐effectiveness of these coatings under dynamic and larger volume culture remain poorly defined. Previous work, including our own, established that heparin‐ending LbL (HEP/COL) coatings enhance initial attachment, proliferation, and migration on static microcarrier culture [12, 15]. Still, it must be acknowledged that hydrodynamic shear, mass‐transfer gradients, and carrier–carrier collisions in dynamic culture can alter coating stability, cell physiology, and cell function in a variety of ways [8, 9]. Process metrics, such as total yield or viability, can fail to capture more subtle metrics, including biochemical and immunomodulatory details. Recent literature highlights the benefits of non‐invasive spent media analytics in monitoring cell behavior, including metabolic and secretory profiles, throughout culture [21, 22]. Employing surface‐enhanced Raman spectroscopy (SERS) combined with Multivariate Curve Resolution–Alternating Least Squares (MCR‐ALS) analysis, allows for periodic tracking of biochemical shifts (proteins, nucleic acids, metabolite signatures, etc.) within spent media over extended cultures, without a need for destructive sampling of microcarriers.

To assess how surface chemistry influences function over time, between seeding and harvesting, analytical approaches of Raman spectroscopy and Luminex multiplex assays were tasked with analyzing spent media samples. Allowing the examination of macromolecular composition and metabolic activity, as well as relative quantities of cytokines. Applying these non‐disruptive techniques, from collecting media during exchanges and quantifying cell yield at the end of culture, allows for a simple yet effective approach of qualitative and quantitative analyses with minimal disruptions to culture integrity. This systems‐level understanding is crucial for translating our material product into robust, GMP‐compatible manufacturing processes that maintain cell quality and quantity in large‐scale cultures with minimal disruptions to culture.

In this manuscript, we build on our prior work over the years optimizing these LbL HEP/COL coatings for hMSCs [7, 14, 15, 23, 24, 25]. This study aims to advance that platform toward a dynamic, 3D bioreactor‐relevant system. In our previous work [12, 14, 15], the heparin‐ending coatings demonstrated enhanced surface bioactivity and cell adhesion under static culture conditions, and efficacy in assembling on microcarriers. Still, their behavior under hydrodynamic shear and continuous suspension (in bioreactor‐like systems) across multiple seeding densities and extended culture of over 2 weeks, remained untested. Here, we evaluate the performance of our coatings over extended periods of culture—a crucial step in the path toward therapeutic translation.

Additionally, from a manufacturing perspective, modification strategies must balance biological performance with economic feasibility. The LbL‐coating process is simple and versatile, but the cost of reagents and additional steps required can vary and be taxing. Therefore, the added costs and efforts must be justified by substantial improvements in cell yield, quality, and/or reproducibility.

We hypothesized that the LbL‐coated microcarriers would improve initial attachment, viability, and yield without significantly increasing cost. Additionally, our proposed coated carrier approach may be further considered as a method of increasing manufacturing yield or helping transition to therapeutic scale, while not increasing the cost to patients. Given that our primary objective was to evaluate the performance of the microcarrier coating platform under dynamic culture while minimizing biological variability, we utilized commercially available hMSCs from a single donor that our lab has made use of in a number of studies to isolate coating‐dependent effects. Ultimately, the aim was to ascertain if our coatings support the goal of moving toward high‐volume, reproducible, and efficient manufacturing for cell therapies.

This study, broadly, investigated whether HEP/COL LbL‐coated microcarriers could improve the biological performance and manufacturing efficiency of hMSC expansion under dynamic culture conditions. By evaluating the coatings across different seeding densities and monitoring metabolic and secretory profiles over time, we found that their benefits were density dependent, with the greatest cost‐effectiveness observed at low seeding densities. The findings also suggest that the coatings are associated with increased glycolytic activity and changes in the secretory profile relative to untreated microcarriers, without evidence of a generalized stress response.

2. Materials and Methods

2.1. Solution Preparation

Heparin sodium (HEP) (Celsus Laboratories, Cat. PH3005), lyophilized type 1 collagen sponges (COL) derived from bovine tendon (donated by Integra Lifesciences, Anãsco, PR), and Poly(ethylenimine) (PEI)(Sigma‐Aldrich, Cat. P3143) were weighed and prepared in polymer solutions at a concentration of 1.0 mg/mL in sodium acetate buffer (0.1 ​M sodium acetate anhydrous, 0.1 ​M acetic acid, at pH 5 for HEP and PEI, and pH 4 for COL). In addition, a sodium acetate buffer at pH 5 was prepared for use as a wash solution. The water used for all solutions was Ultrapure water at 18 ​MΩ ​cm from a Millipore‐Sigma Direct‐Q 3 (Cat. ZRQSVP3US).

2.2. Microcarrier Coating

The coating process, adapted from our previous work [15], and illustrated in Figure 1, began with a brief anchoring using PEI solution for 15 min with occasional tilting for 1–3 g of microcarriers (Corning, Cat. #3772) per 50 mL tube, followed by a 3‐min wash. After this initial anchoring, layers were formed in a three‐step process follows: (step 1): the microcarriers were incubated in HEP solution for 3 min in the 50 mL tube with tilting, centrifuged (VWR, Cat. # 10813–152 for 2D culture and initial optimization and Thermo, Cat. #75004221 for all other experiments) at 100–200 g for 1 min, left to settle for 1 min, followed by disposal of supernatant; (step 2): wash solution was incubated with tilting for 1 min, centrifuged at 100–200 g for 1 min, left to settle for 1 min, followed by the disposal of the supernatant; (step 3): the microcarriers were incubated in COL solution for 3 min in the 50 mL tube with tilting, centrifuged at 100–200 g for 1 min, left to settle for 1 min, followed by disposal of supernatant forming one bilayer. This layering process was repeated for 6.5 bilayers, or 13 layers (HEP‐ending). After coating deposition, the coated microcarriers were washed with Dulbecco's phosphate‐buffered saline (DPBS) 1X without Ca²⁺ and Mg²⁺, and sterilized using ultraviolet (UV) light for at least 10 min. This process is illustrated in Figure 1.

FIGURE 1.

Schematic diagram of the coating and culture processes employed. Centrifuge icon by Servier https://smart.servier.com/ is licensed under CC‐BY 3.0 Unported https://creativecommons.org/licenses/by/3.0/.

2.3. General Culture

Bone marrow‐derived hMSCs purchased from RoosterBIO were used: sourced from a healthy 25‐year‐old male (Lot #310272). The product specification sheet provided by the vendor indicates that these cells were positive for CD90 and CD166 identity markers for hMSCs (as determined by flow cytometry), and negative for CD45 and CD34 (as determined by flow cytometry). Additionally, they demonstrated the ability to differentiate into fat and bone cells. hMSCs were grown in alpha‐minimum essential media MEM Alpha from Gibco with l‐glutamine, and without nucleosides (Cat. #12561‐056) containing 20% fetal bovine serum from Gibco (Cat. #12662029), 1.2% penicillin–streptomycin from Corning (Cat. #30002CI), and an additional 1.2% l‐glutamine from Corning (Cat. #25005CI). Culture in the spinner flasks borrowed and expanded upon the basics in previous literature [12, 15, 26]. All incubations occurred at 37°C and 5% CO_2, and all passages occurred at 80%–90% confluency, unless stated otherwise. All experiments were performed using cells between passages 2–4.

2.4. 2D Culture Experiments

During initial optimization, the protocols focused on replicating the coatings we have used previously on microcarriers and ensuring that the coatings remain able to attach viable cells. Following the testing of the methods of coating that were used in our previous work [15], we made note of the convenience of the centrifugation coating method with higher microcarrier amounts and proceeded with that method, as noted above. Following coating microcarriers, approximately 20 mg of coated and untreated microcarriers were added to wells of Corning ultra‐low attachment microplates (Cat. #3474) with 150 µL of cell culture media and incubated for 1 h to acclimate. After which, varying concentrations of cells were added to the wells of the plates, and they were placed on a plate rocker in an incubator. During the first 2 h, the plates were rotated and gently shaken every 30 min.

2.5. 3D Culture Experiments

Four culture conditions were tested in spinner flasks: 125 mL (Cat #CLS3152) at medium density (secondary optimization), or 500 mL unvented/vented (Cat. #CLS3578/#CLS3153) at low/medium/high density. A vented cap was used in all experiments documented in this study, although similar viability results were observed. In all cases, after the coating process, microcarriers (coated and uncoated) were gently pipetted into spinner flasks and left ajar in a biosafety cabinet in 10 mL of PBS for 15 min under UV light to sterilize. Following sterilization, media was added to the flasks, and they were incubated for at least 1 h. After which, cells were passaged, and experimental procedures were established as depicted in Table 1. The initial experiments in 125 mL flasks used two paths to narrow down an ideal procedure; the continuous condition produced more than two times the number of viable cells compared to the intermittent condition; therefore, it was removed from consideration early.

TABLE 1.

A summary table of the conditions found to be the most ideal for different seeding densities.

Condition	Flask size	Seeding density (cells/cm2) *	Microcarrier mass (g) **	Media volume (mL) **	Rotation	Duration
Small‐scale	125 mL	5000–10,000	1–2 g (total)	Initial: 30 Final (at Day 1): 60	Path 1: 50 rpm constant Path 2: Intermittent (2 s on, 3 s off)	8 Days
Low density	500 mL	2600–2800	Initial (0–24 h): 2 g Final (at 24 h): 5 g	Initial (Day 0): 55 (<0.5% serum) At Day 1: 110 (10% serum) At Day 3: 220 (10% serum)	30 rpm (from Day 1) 40 rpm (from Day 3)	17 Days
Medium density	500 mL	4300–4600	Initial (0–24 h): 3.5 g Final (at 24 h): 5 g	Initial (Day 0): 65 (<0.5% serum) At Day 1: 130 (10% serum) At Day 3: 220 (10% serum)	30 rpm (from Day 1) 40 rpm (from Day 3)	17 Days
High density	500 mL	15,000–20,000	Initial (0–24 h): 1.25 g At 24 h: 2.5 g Final (at 48 h): 5 g	Initial (Day 0): 120 (<0.5% serum) At Day 1: 130 (<0.5% serum) Final (at Day 2): 220 (10% serum)	30 rpm (from Day 1) 40 rpm (from Day 2)	18 Days

^*The surface area of the microcarriers used was 360 cm²/g.

^**For experiments with the 500 mL flask where there were multiple deposits of microcarriers or media, the posted amount is the total in the vessel at the time point.

Notably, these studies involved extended‐duration, bioreactor‐scale cultures requiring substantial labor, material, and resources. Therefore, the work was designed as an exploratory investigation, building upon previous work to identify meaningful biological trends rather than establish definitive mechanistic relationships. Additionally, the experiments with 125 mL flasks used cells at passages 5–7, while the experiments with 500 mL flasks used cells at passages 4–6, and unless stated otherwise, the serum % in media was 10%.

2.6. Media Exchanges, Sample Collection, Analysis, and Imaging

After spinner flasks reached total volume, 50% media exchanges occurred every 48 h. Simultaneously, 1 mL aliquots of spent media were both collected and frozen at −20°C and −80°C, and samples of small microcarrier volumes were retrieved and added to well plates for imaging.

2.7. Live/Dead Imaging

At each sampling time point, a representative 0.5 mL aliquot of well‐mixed microcarrier suspension was removed aseptically from each 500 mL spinner flask and transferred to wells of sterile black bottom well plates. Samples were allowed to settle briefly (∼60 s), and the supernatant was aspirated and replaced with sterile calcium/magnesium‐free PBS to remove residual serum. A working staining solution of Calcein‐AM (Cal‐AM, 1 µM) and Ethidium Homodimer‐1 (EthD‐1, 4 µM) in 4 mL of PBS was prepared fresh and protected from light. The Live/Dead kit was from Thermo, Cat. # L3224. Microcarriers were resuspended in 100–0 µL of staining solution and incubated at room temperature for 30 min in the dark with gentle shaking after 10–15 min to improve dye access to 3D bead surfaces.

Fluorescence images were acquired on an Olympus IX‐83 confocal microscope (objective 4×–20×) using FITC/GFP settings for Calcein (ex ∼488 nm, em ∼510–540 nm) and TRITC/Texas Red settings for EthD‐1 (ex ∼528–535 nm, em ∼610–630 nm). For each sample, images were captured from the center of the wells and several randomly selected locations (z‐stacking was used when necessary to capture the 3‐D cell projection). Image analysis was performed in FIJI/ImageJ: images were background‐subtracted, a color threshold was applied per channel, and particle analysis was used to count Calcein‐positive (live) and EthD‐1–positive (dead) objects. Viability was roughly calculated as the percentage of live cells among total cells and reported as a total percentage from all relevant images. The approximate quantification of the live/dead results appears in Figure S1. Stained microcarriers were not returned to production cultures to avoid any potential dye carryover; all steps were performed with minimal light exposure to avoid photobleaching.

2.8. Metabolic Approaches

2.8.1. Lactate and Urease Colorimetric Assay

Lactate Assay Kit (Sigma‐Aldrich, Cat# MAK064) and Urease Activity Assay Kit (Sigma‐Aldrich, Cat# MAK120) were used according to the manufacturer's instructions, and samples were plated in duplicate.

2.8.2. Raman Spectroscopy of Spent media

The spent media samples from hMSCs grown on uncoated and LbL‐coated microcarriers were collected and stored in −80°C. Samples were collected every 48 h for a total of 18 days, beginning at Day 2. Colloidal gold nanoparticles of size 40 nm were used as a surface‐enhanced Raman scattering (SERS) substrate for Raman signal enhancement. Nanoparticles and samples were mixed in a 3:1 ratio, respectively, and a drop was placed on a quartz slide for Raman measurement. Raman spectral acquisition was performed using a confocal Raman microscope (XplorRA Plus, JY Horiba, NJ). Raman spectra were acquired from 10 different points of the drop. Raman spectra were acquired using a 785 nm laser with an incident power of 10% corresponding to approximately 4–5 mW and a grating of 600 lines/mm. The acquisition time was 1 s with 1 accumulation. Spectra were recorded in the fingerprint region (400–1800 cm⁻¹) using a 10× objective lens. Wavenumber calibration of the confocal Raman system was performed automatically with a silicon standard.

Raman spectra were preprocessed as described previously [27]. Autofluorescence was removed by a fifth‐order polynomial fit. Further, the background corrected spectra were subjected to median filtering to remove random spikes appearing due to cosmic rays, which interfere with biological signals. The spectra were further vector‐normalized to cancel out the potential variations in laser power and median filtering. The spectral data were then subjected to multivariate curve resolution‐alternating least squares (MCR‐ALS) to decompose the complex composite spectra. MCR‐ALS is a well‐known method for decoding complex composite spectra into individual pure component spectra without prior knowledge of the constituents [28]. The resulting spectral profiles or loadings, and the corresponding contributions of pure components, called “scores,” are thus recovered. Briefly, the composite spectra were refined through iterative optimization while enforcing non‐negativity constraints on both the spectral and concentration matrices. Additionally, the model enforced equal spectral lengths across all groups to allow direct comparison of the resulting scores. The non‐negativity constraint ensured that the obtained loadings and scores reflected physically meaningful pure spectra and their corresponding abundances. To promote efficient convergence, principal component analysis (PCA) was first applied, and the number of output components was empirically determined. The selected PCA loadings were subsequently used as the initial estimates for the MCR‐ALS algorithm. All data analysis and preprocessing were performed using MATLAB (MathWorks, Natick, MA).

2.9. Luminex Immunoassay

As with the Raman spectroscopy, spent media samples from the same experiment were collected and stored in −80°C. Samples were collected every 48 h for 18 days, beginning on Day 2. Magnetic Luminex Multiplex Immunoassay was performed on a MAGPIX Luminex instrument to measure the levels of various cytokines with ProcartaPlex Human Growth Factor Panel, 11plex (Thermo, Cat. # EPX110‐12170‐901), including BDNF (brain‐derived neurotrophic factor), EGF (epidermal growth factor), FGF‐2 (fibroblast growth factor 2), HGF (hepatocyte growth factor), LIF (leukemia inhibitory factor), NGF beta (nerve growth factor beta), PDGF‐BB (platelet‐derived growth factor BB), PlGF‐1 (placental growth factor 1), SCF (stem cell factor), VEGF‐A (vascular endothelial growth factor A), and VEGF‐D (vascular endothelial growth factor D). All reagents and standards were prepared according to the manufacturer's protocol. Results were normalized by relative cell counts; unnormalized results are in Figure S2.

2.10. Yield and Cost Analysis

Harvesting occurred on the final day of culture, with a tailored protocol adapted from commercial suppliers [26, 29]. Most of the cell media was removed via aspiration, and flasks were washed with DPBS twice. After which, 50 mL of 0.25% trypsin was added to each 500 mL spinner flask, and the flasks were incubated with slow rotation (25 rpm) for 25–30 min. At around 20 min, cell ribbons should be visible; incubation with agitation continued for about 5 min after these ribbons were seen. Alternatively, if no ribbons are seen, small aliquots can be collected in a well plate to view via inverted microscopy if cells on individual microcarriers seem to be rounding/detaching. After which, the trypsin was neutralized with 20–25 mL of 20% serum media, swirled, repeatedly pipetted, and transferred to labeled 50 mL centrifuge tubes through a 70 µm strainer. Centrifugation was performed at 150 g for 10 min. Cells were then counted via a hematocytometer for comparison. Yield increase was calculated by comparing the final harvested cell number of the LbL‐coated microcarrier condition to the final harvested cell number of the uncoated microcarrier control. Cost comparison occurred by adding up the associated recurring/disposable equipment costs for each condition, comparing the coated to the untreated control, as depicted in Figure S3, including microcarriers, media, spinner flask, and coating reagents.

3. Results

3.1. Confocal Live/Dead Images Confirming Attachment and Viability

To confirm hMSC attachment and qualitatively assess cell viability over the culture period, microcarriers were sampled from the spinner flask at 48‐h intervals after agitation began, stained, and imaged. Figure 2 displays representative fluorescence images of hMSCs on both untreated and LbL‐coated microcarriers on Days 6 and 16 of culture. On Day 6, both untreated and LbL‐coated microcarriers show clear cell attachment, with much of the microcarrier surface covered by viable, Calcein‐AM‐positive (green) cells. Minimal EthD‐1‐positive (red) cells, indicating high initial viability after the initial cell attachment period. Still, there are microcarriers without cells attached. Especially so for the untreated condition.

FIGURE 2.

Representative live/dead projected images of hMSCs on untreated and LbL‐coated microcarriers over time (high density condition), selected from representative images taken. Scale bars are 400 µm and are representative of all images. Cal‐AM (Calcein‐AM, live cells, green). EthD‐1 (Ethidium Homodimer‐1, dead cells, red).

On Day 16, the differences between conditions are loosely noticeable. The control conditions show an increase in dead cells (EthD‐1‐positive), indicating an increase in cell death over the long‐term culture period. In contrast, the LbL condition shows less significant expressions of EthD‐1 but more diffuse expression, which is hypothesized to be because of the binding capabilities of the negatively charged LbL surface.

3.2. Metabolic Trends Differ Between Conditions

Multivariate curve resolution (MCR) analysis of the SERS spectra resulted in eight distinct chemical components from the spent spinner culture media. Two stood out as the most significant, time‐dependent variations between the uncoated and LbL‐coated microcarriers. Based on spectral peak assignments (Figure 3A,B) Components 1 and 8 were identified as a mixed signature of proteins and nucleic acids, with key peaks corresponding to those presented in the accompanying charts shown in Tables S1 and S2.

FIGURE 3.

Metabolic, enzymatic activity, and cytokine expression of cells cultured on uncoated and LbL‐coated microcarriers over time, from technical replicates. (A) Normalized high density Raman spectroscopy (RS) showing trends of MCR scores for Component 1 for uncoated and LbL‐coated high‐density microcarrier culture over 18 days. (B) Normalized high density Raman spectroscopy (RS) showing trends of MCR scores for Component 8 for uncoated and LbL‐coated for high‐density microcarrier culture over 18 days. The standard deviation in A and B are represented by error bars. (C) Average lactate activity from medium cell density conditions, and (D) average urease activity from medium cell density conditions. Standard deviation in D and E are represented as error bars. (E) Heatmap of cytokine expression, normalized by cell count, representing fold differences (from 0–1: red and 1–2: green, as described in the legend) of LbL‐coated microcarriers compared to uncoated microcarriers as measured by Luminex multiplex assay over 18 days of cell culture in spent media; the color scale appears on the right side. Unnormalized results are in Figure S2.

The temporal contribution of Component 1 (protein/nucleic acid) followed a more dynamic, two‐phase pattern, as shown in Figure 3A. The LbL group initially showed a higher contribution than the uncoated group, peaking at Day 6. This was followed by a decline and a secondary rise after Day 12, which appeared to correlate with the rise in Component 8. The uncoated group showed a different, less pronounced peak later in the culture (Day 14). ANOVA analysis confirmed that these complex, time‐dependent differences between the groups were statistically significant at numerous time points, indicating a continuous and significant divergence in the secretome profiles of the two groups.

The contributions of Component 8 (nucleic acids), presented in Figure 3B, showed a clear distinction. The uncoated microcarrier group exhibited a minimal and relatively stable contribution across the entire 18‐day culture. In stark contrast, the LbL‐coated microcarrier group, after a lag phase of approximately 10 days, exhibited a dramatic and significant increase in Component 8 contribution, peaking at Day 18. This visual difference was statistically significant, with a Tukey's multiple comparisons test confirming that the LbL group at Day 18 was significantly different from the uncoated control at Day 18 as well as from all other untreated time points. The summary tables for Tukey's multiple comparison tests for Components 1 and 8 are in Figures S6 and S7.

To correlate the secretome data with the cell population, end‐point cell yield was quantified for this high‐density condition. The LbL microcarriers resulted in approximately 30% greater cell yield compared to the untreated control, the smallest increase in yield among the varying densities, as reflected in Figure 4. These findings, from the condition showing the least difference between groups, establishes a moderate effect of the LbL coating on overall cell proliferation and/or viability.

FIGURE 4.

Comparison figures, showing the cost to produce–yield outcomes at differing conditions (n = 2) (A) hMSC yield chart summarizing and comparing the costs paid and yield percentages found when cultured in a 500 mL spinner flask for differing cell and microcarrier seeding densities (B) bar graph showing the comparison of cost and yield included in the right side of (A).

Further assessing the metabolic profiles of hMSCs on the different microcarriers, lactate and urease concentrations in spent media were tracked over 17 days for medium densities (Figure 3C,D). For lactate, Figure 3C, the untreated group's average lactate levels remained consistent, beginning to rise slightly at about Day 11. The LbL group, on the other hand, is far less consistent, but generally higher compared to the control until Day 13, where concentrations are nearly equal. Urease, seen in Figure 3D, follows a similar but somewhat opposite pattern. The LbL group's average urease levels remained consistent throughout culture. While the untreated group remains consistently just about the LbL group throughout, with a peak around Day 9, and then, again, concentrations ended at an approximately equal level at the end of culture.

Figure 3E shows the normalized fold change heatmap from the Luminex multiplex assay, depicting the difference between LbL‐coated and uncoated for the spent media, from the same sample set in the previous parts of Figure 3. Many of the cytokines showed fold change below 1, with HGF (p < 0.05), FGF‐2, VEGF‐A, LIF, and PDGF‐BB deviating from that pattern.

3.3. Cost and Yield Comparisons

The financial and production implications of implementing a novel microcarrier coating were evaluated and displayed in Figures 4 and S3. The latter provides a detailed cost breakdown for hMSC expansion, comparing uncoated and coated microcarriers in representative batch sizes. For a 5 g batch, the coating reagents account for $84.31, or a 18.6% increase in cost, further illustrated in the pie charts of Figure S3. In addition, the pie charts do well at visually confirming that the largest cost portions, in any case, are media, followed by the spinner flask itself. It must be noted that it is difficult to account for the cost increase associated with the added 4–6 h of labor. Figure 4 details the effect of the increase compared to the final hMSC yield, and a clear density‐dependent trend was observed. At low density, an 18.6% increase in cost (relative to the uncoated control) produced a 110% increase in cell yield. At medium density, a 18.6% cost increase resulted in a 60% yield increase. At high density, the same 18.6% cost increase yielded a 30% increase in cell yield. These results demonstrate that while the coating consistently improves cell yield across all densities, the relative cost–yield benefit diminishes as the initial cell seeding density increases.

4. Discussion

4.1. Imaging Reveals Consistent Viability

The results from the live/dead staining and imaging in Figure 2 provide evidence aligning with the hypothesis that the LbL coating supports robust hMSC attachment, early, and long‐term viability comparable to the untreated microcarriers. The consistent green fluorescence over 18 days (denoting viability), combined with the non‐specific red fluorescence (suggesting consistency with HEP's sequestering effect and a maintenance of charge, reacting with the oppositely charged ethidium), suggests that the multilayered surfaces provide a continued altered microenvironment that can preserve metabolic function, in agreement with the steady metabolic profiles observed in Figure 3. The modest differences in cell distribution and compaction on LbL microcarriers may reflect enhanced matrix interactions or altered mechanotransduction cues imparted by the polyelectrolyte layers, as we have seen in previous studies with static culture, where the coatings have shown not to have an adverse effect on viability [7, 25]. Although for the 2D culture experiments, we see a great deal of clumping, as seen in the z‐stack in Figure S4. Collectively, the images and findings support the hypothesis that LbL coatings help maintain hMSC health and function, even during extended culture, and provide inspiration to look deeper into the metabolic effects of the coatings.

4.2. Metabolic Differences Between Coated and Untreated Conditions Persist Over Time

The central finding of the Raman spectroscopy analyses, depicted in Figure 3A,B, shows that LbL‐coated microcarriers are not passive surfaces but bioactive modulators of hMSC behavior. More specific molecules were not able to be determined as initially intended, but component analysis was made use of. The late‐stage accumulation of nucleic acids (Component 8) in the LbL group's spent media (significant according to ANOVA testing, p < 0.05). The low, stable profile of the untreated group tells a different story, one of a consistent baseline cellular state. This increase in Component 8's signature nucleic acids likely originate from one of two biological processes, or a combination of both. One is a significantly enhanced rate of cell proliferation, leading to larger cell populations. A higher cell density would correspond to a higher rate of apoptosis, which would passively release large quantities of DNA and RNA into the media. On the other hand, component 8's signature may represent active, disproportionate secretory processes. HMSCs are known for their therapeutic paracrine signaling [30, 31, 32], which is correlated heavily with the secretion of extracellular vesicles, which are rich in nucleic acids [30, 31, 33]. This would be consistent with our previous studies that suggest that HEP/COL LbL coatings can be used to prime cells for therapeutic applications [7, 25]. Although the absence of declining viability throughout culture argues against widespread cell death as the sole contributor to the observed nucleic acid accumulation, further studies would be needed to confirm whether either or both hypotheses should be rejected.

The less direct story of Component 1 supports a more complex narrative. The initial spike from Days 2–8 in the LbL group, not seen in the control, could correspond to a similar adaptation phase seen in Component 8. During this, cells are secreting proteins (collagen, for example) to build their own extracellular matrix, attaching to the microcarrier surface. The rise in Component 1 after Day 12 correlates with the Component 8 spike. This strongly suggests that the late‐stage culture of Component 1 is less dominated by protein secretion but is instead overshadowed by nucleic acid components (adenine, phosphates, etc.) as the release of nucleic acids becomes a dominant event [34, 35].

Figure 3C,D depict a straightforward narrative, suggesting that the LbL coating subtly modulates the energy metabolism of hMSCs during long‐term culture. The more variable lactate levels observed on the coated microcarriers may reflect altered and, in some cases, elevated glycolysis (the Warburg effect), where cells rapidly consume glucose, producing lactate as waste. Elevated lactate can also correlate with paracrine effects and is reported to promote immunoregulatory signaling by enhancing IDO expression, aligning with previously observed trends in LbL and collagen‐coated hMSC systems [7, 36, 37]. The relatively stable urease activity across both groups suggests that nitrogen metabolism remains largely balanced and homeostasis remains constant [38].

In summary, the SERS analysis has captured a complex and time‐dependent cellular response. The LbL microcarriers are validated as a powerful tool for enhancing the secretory potency of hMSCs and validating the SERS method itself as a viable method to distinguish changes in cell number from changes in cell function. The lactate and urease assays suggest that the LbL coating may promote the adoption of glycolytic activity and potential immunomodulatory secretome shifts without compromising metabolic stability.

The findings of the Luminex Multiplex assay, as seen in Figure 3E, further depict a story of significant alteration of the cell behavior. The coincident increased fold change of VEGF‐A and HGF (p < 0.05) is consistent with an enhanced trophic secretory phenotype and serves to corroborate the alternative explanation for the increased nucleic acid signals observed in component 8 of the Raman signaling [39, 40]. Although extracellular vesicles are known carriers of nucleic acids and bioactive proteins, direct characterization has yet to be performed, and therefore, interpretation cannot exceed speculation. LIF, which remained elevated throughout much of the culture, although non‐significantly, adds to the story, further corroborating the suggestion that the coatings are not just increasing cell count but modifying and modulating cell behavior and secretory phenotype while suggesting maintenance of stemness [41]. The findings and the early spike in Component 1, conversely, are hypothesized to correspond to initial ECM production and attachment and are consistent with the modest increases in FGF‐2 and PDGF‐BB and the early rise in HGF. FGF‐2 is a known and widely recognized potent stimulant of MSC proliferation [42, 43]. PDGF‐BB is reported to regulate MSC migration and stromal remodeling [44]. While HGF acts as the biochemical link between early matrix organization and late‐stage secretory activation, inducing ECM turnover and migration [39, 40]. Paired with its ability to alter trophic signaling, HGF may represent the bridge between what was seen in the early Component 1 rise (matrix organization) and mid‐to‐late Component 8 rise (enhanced secretory activity. Additionally, no broad increases across all measured cytokines, including BDNF, NGF‐beta, and SCF, suggest that the HEP/COL coatings do not induce a generalized stress response, as well as suggest retention of phenotypic characteristics consistent with MSC behaviors throughout culture. Markers, such as VEGF‐D and EGF, appearing under‐expressed in the coated populations are interesting, accompanying the enhanced proliferation on coated microcarriers to further substantiate hypotheses centered around the expansion being supported by alternative pathways allowed via FGF2 and HGF [43].

4.3. Cost and Yield Analysis Reveals Commercial Potential

The results of the cost and yield analysis highlight an interesting trade‐off between process cost and cell yield, indicating the value of the LbL coating is context‐dependent. The impressive increase in cell yields with only a modest increase in cost, seen in Figure 4, for the low densities, suggests that cell attachment and proliferation on the microcarrier surface are the primary limiting factors. The coating may enhance the initial attachment efficiency, similar to previous findings [45, 46]. Indicating that the coating allows for greater initial attachment and rapid and efficient attachment, even with a smaller starting cell number. While the higher cell density increases cell yield by ∼30%, it is barely an offset to the 18.6% cost increase. This suggests that at higher densities, the bottleneck for expansion shifts from the microcarriers’ surface properties to system factors such as nutrient depletion or accumulation of metabolic waste. Likewise, looking at the fold increase of the yielded product from the coated microcarriers (1.6, 3.0, and 7.2, respectively) corroborates a difference between the conditions while also highlighting the reality of the inherent constraints of extended culture, a protocol focused on the multi‐week sustainability of the coatings. Furthering the discussion into an economic perspective (Figure S3), the increase in cost for the coating reagents is only part of the picture, as labor is an unspecified unknown, varying greatly depending on the coating procedure and the associated labor costs. Collectively, the cost per 1 million cells was 43.5% lower, 25.9% lower, and 8.8% lower for the coated conditions of the low‐, medium‐, and high‐density cells on average, respectively. Additionally, recovery viability measured by trypan blue (10%) was counted via hemacytometer to be >95%. Thus far, we have shown potential via laboratory‐scale economic modeling.

Costs likely could be further minimized in a standardized, large‐scale, GMP (good manufacturing practice) setting by utilizing larger, optimized batches. Additionally, looking toward scaling up and optimizing for therapeutics, it has been shown that collagen‐coated microcarriers can support advanced expansion of hMSCs, even when gene‐modified [47]. Overall, the results of these analyses indicate that the decision to implement this LbL coating must be strategic and density‐conscious. Additionally, quality control, labor costs, facility costs, and so forth, must be considered on a case‐by‐case basis.

4.4. Limitations and Future Work

First, the spinner flask model is only a simplified version of higher‐level bioreactor dynamics and cannot fully capture the qualities of scalable systems, such as shear, aeration, and nutrient exchange. Therefore, future work must evaluate these coatings in more complex and diverse bioreactor environments, while considering the pathways involved, such as mechanosensitive ones. Second, cell yield quantification was limited, as it only focused on attached cells, and harvesting methods can and should be optimized further. Likewise, attachment, viability, growth rate, and differentiation potential need to be assessed in the future. Third, aggregation was observed at a higher rate in medium and higher density cultures, potentially altering local cell profiles and reducing reproducibility. Optimization and anticlumping strategies should be employed to improve uniformity. Fourth, the Luminex and Raman spectroscopy assays provided useful direction and insight into metabolic trends, but they need to be further specified to establish mechanistic causality. In addition, transcriptomics, proteomics, and so forth, would be beneficial for this same purpose. Fifth, evaluation across multiple donors represents an important next step toward assessing donor‐to‐donor robustness of findings. Lastly, the cost and yield analyses presented are based on laboratory‐scale assumptions and do not include a comparison of other microcarriers or carrier coatings. As established by previous foundational work, the systematic screening of diverse microcarrier chemistries and products is a critical requirement for determining optimal procedures and substrates [48]. Economic and yield outcomes will depend heavily on automatability, reagent sourcing, process volume, and labor situation. Addressing these constraints will help move our LbL‐coated microcarriers to a validated and fully controlled GMP‐compatible manufacturing system that can compete with other options on the market.

5. Conclusions

HEP/COL layer‐by‐layer coatings were shown to support long‐term hMSC expansion on microcarriers while maintaining viability and altering cell behavior separate from changes in cell number. Raman spectroscopy revealed distinct time‐sensitive differences in extracellular biomolecular signatures, including elevated nucleic acid‐associated signals and altered metabolic profiles. Luminex analysis demonstrated elevated expression of trophic and regenerative factors, including fibroblast growth factor‐2 (FGF‐2), hepatocyte growth factor (HGF), and vascular endothelial growth factor‐A (VEGF‐A). Together, suggesting that coated microcarriers promote a unique secretory and metabolic experience for cells during dynamic culture, even for time periods as extensive as 18 days. Additionally, this study serves as a limited example for non‐invasive analysis/comparison of cellular state and behavior. Successful incorporation of methods such as these, possibly even during extended culture, could have countless analytical and maintenance‐oriented benefits

Beyond biological effects, the coatings improved cell yield and reduced cost‐per‐million cells across multiple seeding densities, highlighting their potential value for scalable hMSC manufacturing over untreated surfaces. Collectively, these economic and biological findings demonstrate that HEP/COL‐coated microcarriers function as bioactive expansion substrates capable of modulating hMSC phenotype while improving process/manufacturing performance/potential. Still, the limitations and future work discussed in the discussion need to be addressed to build upon the foundation established for this platform.

Author Contributions

Justin Putman: formal analysis, investigation, methodology, validation, visualization, writing – original draft preparation, writing – review and editing. Varsha Karunakaran: formal analysis, investigation, writing – original draft preparation. Roaa Hadi: visualization, writing – review and editing. Luis Pinzon‐Herrera: formal analysis, investigation. Hallapuranam Krishnaswamy Suresh Kumar: resources. Younghye Song: resources. Jorge Almodovar: conceptualization, funding acquisition, methodology, project administration, resources, supervision, writing – review and editing.

Funding

This work was financially supported in part by the National Science Foundation under grant no. 2051582 and in part by the National Institutes of Health under grant no. P20GM13976.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: biot70308‐sup‐0001‐SuppMat.docx.

Supporting File 2: biot70308‐sup‐0002‐FigureS4.mp4.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.