Shape-preserved transformation of biological cells into synthetic hydrogel microparticles

Abstract

The synthesis of materials that can mimic the mechanical, and ultimately functional, properties of biological cells can broadly impact the development of biomimetic materials, as well as engineered tissues and therapeutics. Yet, it is challenging to synthesize, for example, microparticles that share both the anisotropic shapes as well as the elastic properties of living cells. Here we describe a cell-directed route to replicate cellular structures into synthetic hydrogels such as polyethylene glycol (PEG). First, the internal and external surfaces of chemically fixed cells are replicated in a conformal layer of silica using a sol-gel process. The template is subsequently removed to render shape-preserved, mesoporous silica replicas. Infiltration and crosslinking of PEG precursors and dissolution of the silica results in a soft hydrogel replica of the cellular template as demonstrated using erythrocytes, Hela, and neuronal cultured cells. We find that the elastic modulus can be tuned over an order of magnitude (~10-100 kPa) though with a high degree of variability. Furthermore, synthesis without removing the biotemplate results in stimuli responsive particles that swell/deswell in response to environmental cues. Overall this work provides a foundation to develop soft particles with nearly limitless architectural complexity derived from dynamic biological templates.

Graphical Abstract

Abilities to synthetically replicate the complex morphology and mechanical properties of single cells is important for materials science and medicine and has thus received considerable attention. Here we describe a route to replicate cellular structures into synthetic hydrogels that retain the internal/external shapes/features of the biotemplate and result in mechanically tunable and stimuli-responsive cellular mimics.

The demand to create precision 3D soft materials is increasingly important for burgeoning applications such as flexible electronics, soft robotics, artificial muscles and tissue engineering. Often materials are needed at scales that enable relatively straightforward manufacturing approaches using extrusion, roll-to-roll, casting/spin coating, molding, and various modes of 3D printing. For complex 3D shapes, the manufacturing challenge increases as dimensions scale down, for example, to the size of micro- and nano-particles. Thus, there has been considerable attention aimed at developing strategies for batch synthesis of complex or otherwise non-spherical soft colloidal particles.^[1] Noteworthy approaches include emulsion^[2] and microfluidic techniques,^[3] electro-jetting,^[4] methods based on flow (photo) lithography,^[5] and particle replication in a non-wetting template (PRINT) method.^[6] However, these methods do not yet approach the structural complexity found in living systems such as single cells and multicellular organisms. In particular, abilities to synthetically replicate the complex morphology and extracellular topography of single cells, using hard and soft materials, would enable numerous areas of materials science and medicine such as the development of biological vectors that can mimic, for example, the shape and mechanical properties of red blood cells.

Recently we have developed a process to generate bio-composites and inorganic replicas of biological cells and soft tissues.^[7] External and internal bio-surfaces of, for example, single cells are replicated in nanometer thick, conformal silica layers using a solution-based, sol gel process. Removal of the biological template via high temperature calcination produces a shape-preserved mesoporous silica cell replica (SCR) which, in the case of RBC-templated silica, results in a high surface area (~600-700 m²/g) particle with an average pore diameter of 5 nm.^[7b] Building upon this approach, we surmised that this shape-preserved porous silica could serve as a suitable scaffold for subsequent templating of soft, synthetic materials. Indeed, Caruso and colleagues^[8] have previously developed approaches to use mesoporous silica particles as sacrificial templates to generate spherical polymer particles of various compositions,^[8-9] and recently demonstrated soft hydrogel micro-particles comprised of polyethylene glycol (PEG) with tunable modulus.^[10] Inspired by this work, here we describe the synthesis of polymer replicas of mammalian cells. This provides a foundation to develop particles with nearly limitless architectural complexity derived from dynamic biological templates, tissues and organisms.^[7a]

As a starting point, porous silica replicas of biological cells were generated as described previously.^{[7b, 7c]} Briefly, fixed cells were incubated in a weakly acidic silicic acid solution for 12-16 hrs at ~ 40 °C and subsequently rinsed in phosphate buffered saline (see Methods for details). Following silicification, the organic (bio)template could be removed via calcination at high temperature (> 500 °C); however, we observed that removal of the organic template via acid digestion resulted in more consistent polymer replication, particularly using red blood cell templates. Here the entire synthesis was carried out in liquid, and thus less deformation of shape or pore size of the silica template was expected versus air drying and high temperature calcination. For this, we incubated the biocomposite particles in 70% nitric acid solution under mild heating (55 °C; 12 hrs) resulting in silica cell replicas (SCRs). Initial attempts to electrostatically adsorb and subsequently polymerize low molecular weight cationic polymers (e.g., polyethylenimine; PEI) to the porous silica particles (zeta potential of −12 mV in phosphate buffered saline; PBS) proved ineffective in our hands, and so we adapted the strategy of Cui et al,^[10] where an 8-arm polyethylene glycol (PEG) amine is incubated and electrostatically adsorbed to a mesoporous silica particle, rinsed and subsequently incubated with an 8-arm ‘crosslinking’ PEG (containing terminal n-hydroxysuccinimide groups; NHS; details in Methods). This procedure is described in Figure 1a. Fluorescent PEG particles were generated by incorporating an NHS-functionalized fluorophore at low concentration (relative to the PEG-NHS) during the crosslinking step. As shown in Figure 1b and 1c, this method was effective for translating SCRs of red blood cells (RBCs) and dissociated HeLa cells into shape-preserved, all-polymer replicas. Remarkably, internal features of HeLa cells, such as nuclear membranes and cytoplasmic bodies, appear replicated in the polymer particles (Figure 1c lower panels). Additionally, spinning disk confocal microscopy (SDM) images confirm replication of HeLa cell heterogeneity, further indicating the possibility to replicate fine surface and subcellular structures—such as membrane ruffles and internal compartments—into synthetic materials (Movie S1; Figure S1).

Figure 1.

Biological cells as starting materials for shape-preserving transformation into synthetic hydrogels. (a) Schematic of the procedure for red blood cells (RBCs). (b) Optical microscopy images corresponding to steps in the procedure. The PEG replicas are labeled with Dylight 550. (c) Disassociated HeLa cells as templates for conversion into PEG hydrogels. The arrows in the “Fixed HeLa” panels point to the edge of a presumed nuclear membrane in fixed cells that is evident in the PEG hydrogels (arrows, bottom left panels). All scale bars are 10 μm.

Comparison of the diameter of hydrated particles indicated minimal size changes throughout the conversion procedure (Figure 2a); resultant PEG-RBCs averaged ~9 μm in diameter (Figure 2a), slightly larger than typical human RBCs (8 μm diameter). Thermogravimetric analysis (TGA) of PEG-RBCs following silica etching showed weight loss is complete at 400 °C indicating volatilization of the particles (Figure 2b). Compared to RBC composites (silica cells prior to removal of the bio-template) that maintain < 50% weight loss up to 1000 °C (the > 50% fraction is presumed to be inorganic silica), this result showed that the silica template is completely dissolved via solution etching to yield an all-organic material. Scanning electron microscopy (SEM) images of dehydrated PEG-RBC- and PEG-HeLa-derived particles showed shrinkage (from ~9 μm to an average of 4-5 μm in diameter after drying for PEG-RBCs; Figure 2c) and conformal deformation against the surface compared to the silica-stabilized particles, further indicating complete transformation into a soft, deformable hydrogel. Of note is that many of the dehydrated PEG-RBCs mirror the concave surface of natural RBCs as compared to what would be expected for surface-only templating (see: [7c]; Figure S6) which indicates PEG infiltration/crosslinking througout the mesoporous silica template. Furthermore, PEG particles dried against a substrate appeared to swell upon rehydration (Figure S2) indicating the potential for dry storage. However, we note that particle shapes also appeared to be stable long-term (> 1 yr) if stored hydrated in aqueous environments. Additionally, we applied this procedure to NG-108 cells (a neuroblastoma glioma cell line) differentiated on glass coverslips to form an interconnected, planar cellular network. Following silica etching, the surface adhered cell bodies were observed to undergo significant contraction while maintaining intercellular connections, resulting in a free-standing film (Figure S3; Movie S2). Thus, the generation of mesoporous silica replicas templated from biological cells—a process that has proven generalizable to tissues and organisms^[7a]—appears the only prerequisite to synthesizing shape-preserved polymer replicas of soft biological materials.

Figure 2.

Physical characteristics of PEG materials templated from cells. a) Optically measured diameters of RBCs and templated particles shows minimal size change throughout the hydrogel synthesis. b) Thermogravimetric analysis of RBC-silica cells (composites) versus PEG replicas. c) SEM images of silica and PEG replicas of RBCs and HeLa cells. All scale bars are 20 μm. d) Geometric mean value and standard deviation of the elastic modulus of PEG-RBCs versus crosslinker concentration (for 0.625 and 1.25 mg/ml respectively. The inset shows geometric mean values and standard deviation of PEG-RBCs particle heights (thickness of the discoid particle) measured with AFM.

Numerous studies have shown that hydrogel elasticity—for both particles and bulk gels—increases via crosslinking density, providing a means to tune material elasticity to match living cells and tissues^[10-11]. Using atomic force microscopy (AFM), we measured the height and elastic moduli of PEG-RBC particles in physiological conditions (PBS) synthesized using various concentrations of crosslinker. As shown in Figure 2d, doubling the crosslinker concentration increased material stiffness while the average height of the particles did not change significantly across the experiment. Beyond 1.25 mg/ml of crosslinker we observed minimal increase in elastic modulus and noted substantial variability (Table S1). This mimics the significant variation (>100%) in elastic properties observed previously amongst, for example, sub-populations (same age/genetics) of adipose-derived stem cells (ASCs)^[12]. Indeed, the large variation in elastic moduli of the RBC template (Table S1) appears mirrored in the resultant PEG particles. The overall stiffness of the template RBCs (73 kPa) is a consequence of the chemical fixation while the most compliant particles (lowest concentration of crosslinker) better represent the physiological mechanical properties of RBCs (~5-30 kPa)^[13]. ]. Thus, to precisely tune or match desired mechanical properties to a specific cell type will likely require cell-specific optimization for this approach.

The use of living cells to generate mechanically tunable hydrogel replicas offers the intriguing possibility of preserving cellular functions in the synthetic particle. For example, retention of innate hemoglobin function in PEG-RBCs could provide an alternative path to develop a universal blood substitute. However, the path forward is formidable. Though crosslinked hemoglobin has shown promise as a blood substitute;^[14] extensive crosslinking using common aldehydes (required here for shape retention during silicification^[7b]) imparts irreversible damage to oxygen binding function^[15]. As a first step, we investigated the ability to derive a PEG-RBC replica without removing the cell template. Zeta potential measurements indicated the feasibility of electrostatic adsorption of the PEG-amine to the silica-cell composite particle (−10 mV in PBS). Following crosslinking the silica was etched and extensively rinsed in di-H₂O (18.2 MΩ•cm at 25 °C) via centrifugation, upon which we observed that the pellet containing the RBC particles had swelled substantially. Under the microscope, individual particles were markedly larger while retaining the overall RBC shape. This observation is diagramed in Figure 3a. Figure 3b shows optical images of the swelled PEG-RBC particles. We observed particle diameters expanded to a range of ~13-20 μm (average swelled size ~16 ± 2 μm; see SI Fig). Here, the particle was comprised of a crosslinked network of cellular proteins (primarily hemoglobin), materials that have been shown to display stimuli-responsive properties^{[11b, 16]}. Salt-free water is known to destabilize/solubilize protein structures,^[17] thus we suspected that the addition of ions would impact swelling. Indeed, upon addition of 0.15 M NaCl, particles rapidly shrank back to near their initial sizes as shown in Figure 3b, right panel. Figure 3c shows images from a time-lapse of two substrate-bound PEG-RBC particles after the addition of NaCl to the surrounding medium. Here particles shrank within seconds as the salt diffused through the bath (see Movie S3). We rinsed swelled PEG-RBC particles in PBS and again observed shrinkage back to nearly their initial size (Figure S4). As a control, PEG-RBC particles without the bio-template (acid digested as described previously) rinsed and pelleted 5 times in di-H₂O showed only minor swelling—indicating the crosslinked protein network to be the dominant stimuli responsive component of the particle (Figure S3). The ‘swelled’ particles were the softest measured in this study, with a mean elastic modulus of 2.6 kPa that increased following rinsing in PBS to a mean value of 8.3 kPa (Figure 3d). This ability to dynamically modulate both the shape and mechanical properties of PEG-replicas may prove useful for developing smart in vivo absorbents and delivery vehicles—applications that can be further enabled by incorporating responsiveness to additional triggers (e.g., temperature, pH).

Figure 3.

Swelling of PEG-RBC hybrid particles induced at low ionic strength. a) Schematic describing synthesis and swelling/deswelling of PEG-RBCs. b) Swelled particles before (left panel) and after (right panel) introduction of 150 mM NaCl. Arrows correspond to particles viewed edge-on to the substrate before and after deswelling. c) Time lapse images from Movie S2 showing two particles shrinking following the introduction of NaCl to the surrounding media. Scale bars in panels b and c are 10 μm. d) The elastic modulus of swelled versus deswelled particles (geometric mean ± standard deviation).

In conclusion, we have developed a general route to synthesize hydrogel materials using biological templates. The resultant hydrogel particles retain the shape of the biological template—potentially simplifying synthesis of otherwise complex particles and tissue architectures. Incorporation of additional functionalities (e.g., O₂ binding capabilities) can be achieved using a wide variety of known chemistries to functionalize PEG gels^[18] including copolymerization strategies, bioorthogonal chemistries, as well as simply adding dilute reactive moieties during the crosslinking step (e.g., fluorophore labeling, see Experimental Methods). An advantage of this templating approach over other methods is the unprecedented resolution in replicating cellular morphologies and surface topographies from silica templates.^[7] We observed that the average mechanical properties of cells can be potentially replicated in these cellular mimics by varying the degree of crosslinking of polymers infiltrated throughout the silica template. Using this approach, we synthesized hybrid particles that are physically and mechanically responsive to environmental cues due to the properties of the crosslinked biotemplate, suggesting further opportunities to develop functionalities based on synthetic/biological material combinations and enable applications from tailored therapeutics to artificial cells.

Experimental Section

Synthesis of hydrogel replicas of biological cells:

Cell culture and fixation:

HeLa and NG-108 cells were maintained in medium containing 10% fetal bovine serum (FBS) at 37 °C and 5% CO₂. HeLa cells were dissociated from cell culture flasks using a trypsin/EDTA solution and subsequently rinsed in phosphate buffered saline (PBS) 2-3 times via centrifugation at 500 RCF for 5 min. NG-108 cells were differentiated onto poly-L-lysine-coated coverslips in medium containing 1% FBS and cultured for 1-2 days. RBCs were purified from whole blood using the Ficoll density gradient centrifugation procedure, rinsed and pelleted in PBS (3x in 15 ml centrifugation tubes). Following final rinsing, dissociated HeLa cells, substrate adhered NG-108s and RBCs were fixed at roughly a 10:1 (v/v; solution/cells) ratio of 4% formaldehyde in 1.5 mls of PBS at room temperature for 30 mins (HeLa and NG-108 cells) or 16-24 hrs in a volume of 15 mls (RBCs). Cells were rinsed in PBS followed by rinsing in 0.154 M NaCl dissolved in deionized water (0.9% saline).

Silicification of cells:

Detailed methods and mechanistic discussion for this procedure is described in [7]. Following rinsing and pelleting, the supernatent was removed and the pellet of fixed cells was redispered in a solution containing 0.1 M TMOS hydrolyzed in 0.15 M NaCl containing 1.0 mM HCl (pH 3) and incubated in closed containers for 16–24 hrs at 37 °C to generate cell/silica composite particles. Importantly for this step, the pellet should be appropriately diluted in the silicic acid solution. Specifically, we recommend at minimum a volume of 20:1 (v/v; silica solution/cell pellet) to maintain the solution pH and avoid solution gelation (the acidic solution restricts silica deposition to cell surfaces over this timeframe as described in [7c]). To acheive this dilution for the example of RBCs, the pellet can be resuspended in the remnant liquid (agitation of the pellet following removal of the supernatant to yield for example ~0.5 mL of highly concentrated cells at ~10⁷ cells/μL) in a 15 mL conical tube. The TMOS solution is then added to fill the rest of the tube (~ 30:1 silica solution/cell pellet) and the tube is placed in an incubating mini-shaker with mild aggitation (e.g., 250 rpm) overnight at 37 °C. Dilution in excess of this ratio (> 30:1) can be used with no observed upper limit. HeLa and NG-108 cells were dried by rinsing in dH₂0, followed by 1:1 dH₂O:methanol and 100% methanol (2X) for 10 min in each solution and allowed to dry in air.

Generation of PEG replicas.

Digested particles were incubated in a 1 ml solution of PBS containing 5 mg/ml of 8-arm PEG-amine (PSB-812; Creative PEGWorks; MW 20k) for a minimum of 2 hours. Following rinsing in PBS (3X), particles were incubated in a PBS solution containing 0.5-5 mg/ml of 8-arm PEG-succinimidyl NHS ester (PSB-8061; Creative PEGWorks; MW 10k) for 1-2 hrs; fluorescence labeling was achieved by incorporating an amine reactive dye (e.g., Dylight-488, −550) at a concentration of 0.02 mg/ml during this step. Following rinsing in dH₂O, particles were immersed in a solution of Timetch (Transene) or a 1:1 solution of buffered oxide etch (BOE:H₂O) for 1-5 mins, rinsed and used for experiments.

AFM mechanical testing.

All mechanical testing was carried out on hydrated particles that had been fully equilibrated in either salt-free or 0.15 M NaCl solutions, as described in the main text. The elastic moduli of individual PEG-RBC mimics and fixed RBCs (n = 17-35) were characterized using single indentation experiments with a spherically tipped cantilever (k ~ 0.03 N/m, 5 μm borosilicate bead) using an MFP-3D-BIO AFM (Asylum Research, Santa Barbara, CA) according to previously described methods^[19]. Samples settled in solution and adhered to plasma-treated glass coverslips for 30 minutes before dishes were flooded with sample-matched fluid. Particle heights were measured based on differences in the z-position for initial contact point over the particle and the neighboring glass, while the elastic properties were extracted from force vs. indentation curves using a modified, thin-layer Hertz model^[20].