Growth of mesenchymal stem cells at the surface of silicone, mineral and plant-based oils

The culture, scale up of manufacturing and processing of adherent cells requires innovative solutions to enable translation of emerging cell-based technologies in the fields of regenerative medicine and cultivated meat [1, 2]. Although a range of platforms have been designed and proposed to answer these challenges, from hollow fibre reactors to plastic and hydrogel microcarriers for stirred tank reactors [3–6], these remain associated with complex cell processing steps, as well as high costs and risks of contamination with microplastics.

In this context, bioemulsions are attractive alternatives to solid microcarriers for the culture of adherent cells. Oil phases that could be applied are inexpensive and cell products can readily be separated and processed for further applications [7–9]. Recently, the proof-of-concept of long term expansion of mesenchymal stem cells (MSCs) at the surface of bioemulsions was demonstrated [7]. MSCs were cultured from passage 3 to passage 10 on bioemulsions, solid microcarriers and 2D tissue culture plastic in parallel, followed by phenotypic characterisation at multiple time points (cell morphology, flow cytometry, real-time polymerase chain reaction, tri-lineage differentiation assays). The phenotype of MSCs was found to be undistinguishable upon expansion in these three culture systems, at any given round of expansion tested, and instead was found to be sensitive to the passage number and prior rounds of population doubling.

Although the culture of adherent cells may seem surprising, owing to the wealth of reports linking a broad range of cell phenotypes to substrate mechanics [10–14], it was demonstrated that this phenomenon is enabled by the assembly of mechanically strong protein nanosheets at the liquid–liquid interface [15, 16]. Indeed, corresponding interfaces display high interfacial shear moduli and elasticity, enabling resistance to cell-mediated contractile forces. Keese and Giaever had identified that surfactant molecules, combined to poly(L-lysine) (PLL) enabled the culture of fibroblasts at the surface of oil droplets [17] and such phenomenon was recently assigned the formation of protein nanosheets [15]. More systematic studies of the combined impact of surfactant and protein chemistry on the co-assembly of nanosheets at fluorinated oil interfaces identified that covalent coupling of pro-surfactant molecules to associated proteins enabled the formation of quasi-2D (6–10 nm) elastic networks and the storage of strain energy upon cell-mediated contractile forces [18]. The impact of the molecular structure, including molecular weight and charge density, of proteins assembling to such interfaces on the toughness and viscoelasticity of resulting networks was also recently reported [19]. However, although fluorinated oil systems used in these studies have played important roles in discovery, analysis and diagnostics platforms in biotechnologies, such as microdroplet microfluidics [20, 21], they remain unlikely solutions for implantation and cultured meat products. Indeed, it is possible that some fluorinated oil residues remain trapped with the cell products. Other oil alternatives are therefore important to identify.

The present work explores the replacement of fluorinated oils and pro-surfactants with silicone and aliphatic oils, including mineral and rapeseed oils used routinely in cosmetics, consumer healthcare and the food industry. The formation of PLL nanosheets assembled at corresponding oil interfaces in the presence of the pro-surfactants hexadecanoyl chloride (HDC) and sebacoyl chloride (SBC) (hydrolysing into biocompatible palmitic acid and sebacic acid [22, 23], respectively) is explored first via interfacial rheology, enabling the quantification of interfacial shear moduli, the kinetics of nanosheet formation and the investigation of resulting viscoelastic profiles of interfaces. The adhesion and expansion of MSCs cultured at resulting interfaces is then investigated. Finally, the stability of emulsions formed with corresponding protein nanosheets is investigated and the feasibility of MSC culture at corresponding non-fluorinated bioemulsions demonstrated.

2.1. Materials

2.2. Interfacial rheology

Interfacial rheology was carried out on a hybrid rheometer (DHR-3) from TA Instruments fitted with a double wall ring (DWR) geometry and a Delrin trough with a circular channel. The DWR used for this geometry has a radius of 34.5 mm and the thickness of the platinum–iridium wire is 1 mm. The diamond-shaped cross-section of the geometry's ring provides the capability to pin directly onto the interface between two liquids and measure the interface properties without sub-phase correction. The 19 ml of PBS (pH adjusted to 10.5) was placed in the Delrin trough and the ring was lowered, ensuring contact with the surface, via an axial force procedure. The measuring position was set 500 μm lower than the contact point of the ring with the aqueous phase surface. Thereafter, 15 ml of the oil phase containing the pro-surfactant mixture (HDC or SBC at final concentrations of 100 µg ml⁻¹, as specified in corresponding figures) was carefully syringed on top of the aqueous phase.

Time sweeps were performed at a constant frequency of 0.1 Hz and a temperature of 25 °C, with a displacement of 1.0 10⁻³ rad, to follow the formation of the protein layers at the interface. The concentration of PLL used for all rheology experiments were 100 μg ml⁻¹ (with respect to aqueous phase volume). Before and after each time sweep, frequency sweeps (with a constant displacement of 10⁻³ rad) were conducted to examine the frequency-dependant characteristics of the interface. Amplitude sweeps (with constant frequencies of 0.1 Hz) were carried out to ensure that the selected displacement was within the linear viscoelastic region. Considering the low moduli initially measured for pristine liquid–liquid interfaces (in the absence of protein and/or surfactant), viscous drag from both phases were not corrected. We note that although the interfacial shear moduli observed at liquid–liquid interfaces in the absence of protein or surfactant are expected to be considerably lower than those measured in our assay [24, 25], due to lack of viscous drag correction, they should not completely vanish and interfacial shear viscosity should be expected to persist, even with liquid–liquid interfaces that display very limited roughness (at the molecular scale).

Stress relaxation experiments were performed at 1% strain for 120 s. Stress relaxation data from the 10th second onwards was plotted as a function of time and fitted with a double exponential decay function, according to the following equation:

$$\begin{align}\sigma = {\sigma _{\text{e}}} + {\sigma _1}\left( {1 - {{\text{e}}^{ - t/{\tau _1}}}} \right) + {\sigma _2}\left( {1 - {{\text{e}}^{ - t/{\tau _2}}}} \right).\end{align} \tag{ 1 }$$

In this equation, σ is the measured residual stress, σ_e is the elastic stress and σ₁ and σ₂ are viscous relaxation components. τ₁ and τ₂ are relaxation time constants. The degree of stress retention (σ_r) is calculated as:

$$\begin{align}{\sigma _{\text{r}}} = \frac{{{\sigma _{\text{e}}}}}{{{\sigma _{\text{e}}} + {\sigma _1} + {\sigma _2}}} \times 100.\end{align} \tag{ 2 }$$

2.3. Generation of pinned droplets

2.4. Preparation of glass-mounted wells for higher resolution imaging (20X objective)

2.5. Formation of bioemulsions

2.6. MSCs culture and seeding

2.7. Hoechst staining and cell counting

2.8. Cytocompatibility assay

2.9. Immunostaining

2.10. Immuno-fluorescence microscopy and data analysis

2.11. Statistical analysis

In order to promote cell adhesion and expansion at the surface of non-fluorinated liquids, we proposed to adapt the formulation of protein nanosheets previously investigated at fluorinated interfaces [18, 26]. In these systems, the cationic protein PLL was introduced in the aqueous phase (PBS) and allowed to co-assemble with the pro-surfactant pentafluorobenzoyl chloride (PFBC) introduced in the oil phase. We referred to this latter molecule as pro-surfactant as neither PFBC, nor PLL, on their own display sufficient tensio-active properties to stabilise emulsions. However, when combined, these two molecules can couple and form a layer of protein that is physically cross-linked at the interface between the two liquids. Covalent coupling of pro-surfactant to proteins such as PLL, albumin or lysozyme was found to be essential to confer sufficient elasticity to the interface to store energy upon deformation and resist cell-mediated contraction [18].

Therefore, we translated such design to the formation of PLL nanosheets at non-fluorinated oil–water interfaces, combining PLL with aliphatic acyl chlorides. Two pro-surfactants were selected (figure 1): hexadecanoyl chloride (HDC), displaying a long aliphatic tail and hydrolysing into the food-compatible and biocompatible palmitic acid [27, 28], and SBC [22, 23], enabling direct intermolecular covalent crosslinking within nanosheets and also hydrolysing into cytocompatible residues (sebacic acids, anhydrides and esters enter the composition of drug release systems) [18, 22]. As in the case of nanosheet assembly at fluorinated interfaces, the pH of the aqueous phase was adjusted to 10.5, corresponding to the pKa of PLL, to enable sufficient coupling of acyl chloride residues [19, 26].

Zoom In Zoom Out Reset image size

The assembly of PLL nanosheets at silicone oil interfaces was first examined, using interfacial rheology (figure 1(A)). Time sweeps demonstrated the rapid increase in interfacial shear moduli associated with the self-assembly of PLL-pro-surfactant nanosheets at silicone–water interfaces. Upon introduction of PLL, both the interfacial storage (figure 1(A)) and loss moduli (supplementary figure S1) rapidly increased, and diverged as the interfacial storage moduli rapidly dominated. Moduli gradually levelled to a plateau, with interfacial storage moduli in the range of 1–2 N m⁻¹. Hence these data confirm the rapid formation of protein networks at interfaces, associated with >3 orders of magnitude increase in interfacial shear moduli. Such phenomenon is comparable, in terms of kinetics and magnitude, to the self-assembly of PLL nanosheets at fluorinated oil interfaces [18, 26]. Indeed, the interfacial storage modulus of fluorinated oil interfaces (Novec 7500) at which PLL was assembled in the presence of the acyl chloride PFBC rapidly raised to moduli in the range of 1–2 N m⁻¹, within 1 h of PLL injection [18].

Although the pro-surfactant formulation had no significant impact on the ultimate modulus of PLL nanosheets (figure 1(D)), it impacted the kinetics of the evolution of interfacial storage moduli (figure 1(A)). As HDC content increased, the interfacial storage moduli increased more slowly as a function of time. In addition, in the case of nanosheets assembled in the presence of HDC alone, a two-step process was observed (supplementary figure S2 for three replicates of this time sweep). After a first plateau seemed to be reached near 2000 s post injection, around 3000 s a second increase was observed. Such two stage process was not observed with PLL nanosheets assembling in the presence of PFBC, on fluorinated oil interfaces [18], and was only observed in the case of albumin nanosheets forming at significantly higher PFBC concentrations (10 mg ml⁻¹), which was proposed to result from multi-layer formation [15]. This could reflect the formation of bilayers or simply the more gradual shift from a soft physically cross-linked nanosheet to a more structured bilayer (see proposed schematic representation of such bilayer in figure 1(B)). Indeed, evidence for such bilayer was obtained in similar systems (with hexadecane as the oil) for PLL-HDC nanosheets, from neutron reflectometry experiments [18]. The longer alkyl chains associated with HDC, compared to PFBC and SBC (able to generate strong physical and covalent crosslinks respectively), may be associated with the initial softer character of the interfaces observed and the requirement for relatively high functionalisation levels to achieve comparable interfacial moduli. However, it should be noted that PLL surface densities and HDC/SBC functionalisation levels were not quantified in the present specific system.

Further analysis of the frequency sweeps revealed comparable viscoelastic profiles (figure 1(C)), in agreement with the establishment of elastic networks at corresponding interfaces. Little frequency dependency was observed over the range probed (0.01–10 Hz). Interfacial storage moduli were consistently well above loss moduli (approximately one order of magnitude), further supporting the elastic nature of nanosheets formed. Overall, all pro-surfactant formulations resulted in nanosheets with comparable moduli and viscoelastic profiles (figures 1(C) and (D)). Interfacial elasticity is an important property that was recently proposed to play an essential role in enabling cell adhesion at liquid–liquid interfaces [18]. Therefore establishing a detailed understanding of the viscoelastic profile of PLL nanosheets at non-fluorinated oil interfaces is particularly important.

To further investigate interfacial viscoelasticity, stress relaxation experiments were carried out at interfaces. In such experiments, the du Noüy ring allowed the establishment of a controlled strain on corresponding nanosheets, and the relaxation of interfacial stress was monitored as a function of time. The relaxation profiles were then fit with a double exponential decay function, in order to extract stress retention σ_r and relaxation rate constants [18]. Relaxation profiles recorded for PLL nanosheets formed in the presence of different pro-surfactant formulations were comparable, almost overlapping (figure 2(A)). In turn, the stress retentions and relaxation constants measured remained within narrow ranges (σ_r between 74% and 80% and τ₁ between 73 and 91 s; figure 2(B) and supplementary figure S3). These values are comparable to those measured for PLL nanosheets assembling at Novec 7500 interfaces, in the presence of PFBC at concentrations of 1.25 µg ml⁻¹ [18]. However, interestingly, at the surfactant concentration used (10–100 µg ml⁻¹), PFBC-PLL nanosheets were found to be less elastic (σ_r near 60%). This may indicate a more brittle character of PLL nanosheets formed at the same concentration of acyl chloride pro-surfactant, or may reflect a reduce level of functionalisation, although the interfacial moduli achieve were comparable at this concentration. These results confirm the high level of elasticity associated with PLL nanosheets formed at silicone–water interfaces, for all pro-surfactant combinations tested. Therefore, PLL nanosheets formed with aliphatic pro-surfactants mediating physical or covalent crosslinks displayed combinations of high interfacial shear moduli and elasticity that are expected to support cell adhesion, spreading and proliferation at liquid interfaces.

Zoom In Zoom Out Reset image size

Indeed, a broader screen of different protein and pro-surfactant formulations had indicated that the elasticity of protein nanosheets assembled at liquid–liquid interfaces was essential to sustain cell-mediated spreading and expansion, in the context of primary keratinocytes, MSCs and the epidermal cell line HaCaT [18]. Therefore the viscoelastic properties of PLL nanosheets assembled at silicone interfaces in the presence of HDC and SBC pro-surfactants suggest that these interfaces should enable the adhesion, spreading and expansion of MSCs at corresponding non-fluorinated oil–water interfaces. It is proposed that such combination of high interfacial shear modulus and capacity to store strain energy upon cell-mediate deformation will resist cell mediated contractility and sustain cell spreading.

MSCs were seeded at PLL nanosheet-stabilised interfaces formed in the presence of SBC/HDC mixtures, further coated with fibronectin, and allowed to spread for 24–48 h. The spreading of resulting cells was evaluated by immunostaining and microscopy (figure 3). At both time points, MSCs spreading on PLL nanosheets at silicone oil interfaces could be found to display spread morphologies with total surface areas comparable to those of cells spreading on glass coverslip coated with PLL and fibronectin (2000–2700 µm²), or previously reported for MSCs growing at fluorinated oil interfaces stabilised by PLL/PFBC [19]. In all cases, cells displayed lamellipodia associated with fan-shaped morphologies and numerous focal adhesions, identified from vinculin stainings. MSCs also displayed well developed actin stress fibres connecting to focal adhesions. Therefore, our data indicate that MSCs spread at PLL nanosheet-stabilised silicone oil interfaces via the classic integrin-actomyosin machinery, in good agreement with observations of primary keratinocytes and MSCs spreading on fluorinated oil interfaces stabilised by PLL-PFBC assemblies [15, 18].

Zoom In Zoom Out Reset image size

The proliferation of MSCs at silicone oil interfaces was next examined. The cytocompatibility of these liquid–liquid culture systems was investigated first at two different pro-surfactant concentrations (figure 4). At concentrations of 10 µg ml⁻¹, cell viabilities remained comparable on all liquid interfaces tested, as well as the tissue culture polystyrene (TPS) control, both at days 1 and 3 (figure 4(A)). Cell densities were also comparable on all interfaces, at early and late time points (days 3 and 5), indicating that initial cell adhesion and proliferation remained comparable on PLL nanosheet-stabilised interfaces and rigid tissue culture plastic substrates. At higher pro-surfactant concentrations (100 µg ml⁻¹), although cell densities and viabilities were found to be lower on pure SBC/PLL and HDC/PLL nanosheets, mixed pro-surfactant/PLL assemblies displayed comparable cytocompatibility and capacity to sustain expansion to the TPS controls (figure 4(B)). These results are in good agreement with previous observations that the pro-surfactant PFBC induced some cytotoxicity at concentrations above 100 µg ml⁻¹ [15], although this effect was more pronounced in the case of PFBC. Therefore, although changes in local pH associated with the hydrolysis of residual acyl chloride is likely to play a role in such phenomenon, the inherent toxicity of perfluoroaromatics, compared to palmitic acid and sebacic residues, may also contribute to such effects.

Zoom In Zoom Out Reset image size

To further demonstrate the translatability of aliphatic pro-surfactant/PLL nanosheets for the culture of MSCs at the surface of a broader range of oils, the ability of MSCs to expand on two aliphatic oil substrates was next examined (figure 5). In particular, we selected a mineral oil used in the formulation of consumer health products and rapeseed oil, typically used as cooking oil, as aliphatic oil case studies, owing to their inherent cytocompatibility and potential for translation in cell manufacturing platforms. The assembly of PLL nanosheets at those oil interfaces was first confirmed by interfacial rheology (supplementary figure S4). Although resulting nanosheets formed displayed weaker mechanical properties (100–200 mN m⁻¹) and assembled more slowly (in particular in the case of rapeseed oil), they displayed comparable viscoelastic properties and relatively high residual elastic stress, in stress relaxation experiments (supplementary figures S4(E) and (F)). The differences in kinetics and mechanics involved may reflect differences in water or hydroxylated species content, which potentially cross-react with the acyl chloride co-surfactants used and may reduce their effective concentration for coupling to PLL. It is also possible that the different oils used, with associated differences in chemistry, polarity and surface tension, impact on the co-assembly process of acyl chlorides and PLL.

Zoom In Zoom Out Reset image size

Cell densities measured at early stages (day 3), on both mineral and rapeseed oils, were comparable to those achieved on TPS (figure 5). After 6 d of culture, cell densities on PLL nanosheets stabilised with SBC/HDC 1/1 were slightly below those measured on TPS, although these differences were not statistically significant in the experimental conditions tested. Note that cell densities on the control used (TPS) differ in (figures 5(A) and (B)) as the passage numbers used in corresponding experiments differed. Overall, our data indicate that PLL nanosheets stabilised by mixtures of palmitoyl chloride and SBC enable MSC adhesion and expansion on a broad range of non-fluorinated oil interfaces.

The ability to expand MSCs at the surface of non-fluorinated oil emulsions was explored next. The stability of emulsions stabilised by different pro-surfactant combinations was first examined (figure 6(A) and supplementary figure S5). Whereas PLL-based emulsions stabilised by HDC-rich pro-surfactants were stable just after formation and washing (removal of excess PLL), and after 24 h, those formed with SBC-rich nanosheets rapidly destabilised, in particular those based purely on SBC. Hence, the greater conformational freedom associated with longer aliphatic chains of palmitoyl residues, and their reduced degree of tethering with PLL chains enabled the improved stabilisation of non-fluorinated oil emulsions. Although longer term stability of these emulsions was not quantified, we did not observe any apparent change in the stability of mixed HDC/SBC-PLL/fibronectin stabilised emulsions over the course of 7 d (in MSC growth medium, in the incubator).

Zoom In Zoom Out Reset image size

Finally, MSCs were seeded at the surface of mineral oil and silicone emulsions stabilised with PLL/fibronectin nanosheets, and allowed to expand for 7 d. The resulting cultures were then stained with calcein to identify viable cells. MSCs could be seen covering the surface of corresponding microdroplets, with many cells clearly spreading. Therefore, these results are in good agreement with the cell expansion observed at corresponding 2D protein nanosheet stabilised liquid–liquid interfaces. Together, these results indicate that aliphatic oil emulsions could constitute attractive adherent cell culture platforms for the expansion of stem cells in bioreactors for cell manufacturing.

Cells sense the biochemical and physical properties of their environment. Mechanical engagement of integrins with surrounding protein networks, strengthening of focal adhesions and assembly of the actin cytoskeleton requires a balance of forces, supported by macromolecular networks, sensed at the nanoscale [29]. The dimensions of focal adhesions and actin stress fibres, in the range of a few tens of nanometres, and their mechanical properties, inherent of self-assembled soft matter systems, are comparable with those of other protein nanostructures, including protein nanosheets at liquid–liquid interfaces. Therefore, strong PLL networks displaying interfacial moduli in the range of 1–2 N m⁻¹ and thicknesses in the range of 10–20 nm, even in the case of non-fluorinated oil interfaces and pro-surfactants, are well within the range of dimensions and mechanical properties typically associated with the integrin-talin-vinculin-acto-myosin machinery. Nanoscale sensing of mechanical properties therefore further appears as a key element for the design of new generations of biomaterials. The results presented demonstrate that such concepts and design principles are not restricted to fluorinated oil interfaces, and can be more broadly implemented to other interfaces, including silicone and aliphatic oils. The combined role of pro-surfactants and PLL chemistry on the self-assembly of nanosheets displaying strong mechanical properties and high elasticity remains central to the ability to support cell adhesions and expansion. Specifically, MSCs seeded on PLL nanosheet-stabilised silicone interfaces developed focal adhesions and assembled mature actin cytoskeletons. MSCs were found to proliferate to comparable levels on such interfaces to cells cultured on TPS. In turn, aliphatic oil interfaces, reinforced using equivalent assemblies, enabled the expansion of MSCs too and supported their culture on emulsions. This proof-of-concept, although calling for systematic characterisation and quantification of cell phenotype at non-fluorinated bioemulsion interfaces, demonstrates the feasibility of stem cell cultures at the surface of mineral and plant-based oils that are relevant to consumer healthcare and cultured meat industry. Recovery of cell products from bioemulsions and downstream processing will be facilitated as cultures can simply be centrifuged or filtered (as was demonstrated for cells cultured on fluorinated oils [7]). It is also possible to avoid cell detachment altogether if metabolisable oils are used.

Funding for this work from the European Research Council (ProLiCell, 772462; ProBioFac, 966740), the China Scholarship Council (studentship, 201708060335) and the Leverhulme Trust Foundation for financial support (RPG-2017-229, Grant 69241) is gratefully acknowledged.

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

The authors declare no conflict of interest.