Polyethylene Glycol Nanofiller for Robust Lyophilization of Graphically Encoded Hydrogel Microparticles

Abstract

Graphically encoded hydrogel microparticle‐based biosensing is a promising suspension microarray platform by virtue of multiplexing capability, robust sensitivity, and facilitated downstream analysis. However, the absence of a long‐term and stable storage protocol for the hydrogel microparticle has been a bottleneck for the sensing platform to be adapted to practical fields. In this study, the polyethylene glycol (PEG) nanofiller‐mediated lyophilization strategy of the hydrogel microparticles is presented. To inhibit the lyophilization‐induced deformation of the porous structure and geometries of the particles, PEG is utilized as the filler material occupying the porous region in the hydrogel particles to prevent the interaction between polymer chains and the collapse of the porous structure. Based on the filler effect, the high decoding accuracy (more than 95%) for the lyophilized microparticles after reconstitution can be achieved by outstanding preservation of the particle geometries. Furthermore, the immunoassay performance of the antibody‐functionalized microparticles lyophilized with PEG nanofiller is comparable to that of the non‐lyophilized particles. Finally, the possibility of long‐term storage (more than 6 months) of the lyophilized microparticles is confirmed by thermal aging. This finding is expected to promote the hydrogel microparticle‐based sensing platform to be extended to practical fields via the innovation of the storage protocol.

The polyethylene glycol (PEG) nanofiller exerts a superior stabilization effect on the geometries and porous structures of the graphically encoded hydrogel microparticles in a lyophilization process. The as‐lyophilized microparticles can exert robust immunoassay performances (limit of detection: 6.48 pg mL⁻¹) after thermal aging treatment, which emulates the long‐term storage of the microparticles for 6 months.

1. Introduction

Liquid biopsy is a non‐invasive diagnostic platform in which biomarkers (e.g., protein and nucleic acid) are detected in biofluids such as blood, urine, and saliva.^{[ 1 ]} Based on the high accessibility and safety by minimized invasion for sample acquisition unlike tissue biopsy and endoscopy, the liquid biopsy is a highly potent tool as a next‐generation diagnostic platform. Especially, multiplexed biosensing techniques can contribute to significantly enhanced diagnostic performances of the liquid biopsy enabling simultaneous detection of multiple biomarkers in a single sample.^{[ 2 ]} For instance, multiplexed biomarker detection facilitates low‐cost and rapid diagnosis by high‐throughput profiling of biomarker panels in a limited amount of samples. Furthermore, diagnostic accuracy and reliability can be enhanced by multiplex detection rather than evaluating a single marker.^{[ 3 ]} Precision medicine can also be facilitated by comprehensively analyzing biomarker panels that heterogeneously express between individuals.^{[ 4 ]}

Microarray is a standardized multiplexed biosensing technology that can simultaneously detect up to a few hundred analytes. Microarray is categorized as suspension and planar platforms which typically use the fluorophore and location to designate the codes for each target, respectively. However, each platform still has inherent drawbacks originating from the encoding strategies. For example, the fluorescence‐based suspension microarray (e.g., Luminex technology) has a vulnerability to spectral overlapping between multiple fluorophores which can lead to code misreading. Furthermore, the requirement of multiple fluorophores for the encoding increases the cost and complexity of the entire assay by demanding various types of fluorescent dyes and filters. Even though the planar microarray has high decoding accuracy by spatial encoding process, it has limited detection sensitivity by poor mass transfer kinetics of target analytes to the detection substrates with two‐dimensional structure.^{[ 2 ]} For instance, the sensitivity of the planar microarray‐based microRNA (miRNA) sensing is nanomolar level which is not sufficient to directly detect miRNA in clinical samples (e.g., serum).^{[ 2 ]}

Graphically encoded hydrogel microparticles are highly promising suspension microarray tools targeting diverse analytes such as protein,^{[ 5} , ^{6 ]} miRNA,^{[ 7} , ^{8 ]} and DNA.^{[ 9} , ^{10 ]} The graphical encoding strategy of the microparticles has distinct advantages which can complement the limitations of conventional microarray technologies. Whereas the fluorescence code‐based biosensing technology has an inherent vulnerability for inaccurate decoding, the graphic‐based encoding approaches can accurately and rapidly decode the particular codes by automatic image processing algorithm.^{[ 11 ]} Furthermore, since the hydrogel microparticle has a high surface‐to‐volume ratio and nanoporous structure,^{[ 12} , ^{13 ]} diverse reactants (e.g., target molecule) can be efficiently diffused and captured in the particles, leading to high detection sensitivity.^{[ 14 ]} For example, the hydrogel‐based immunoassay has been reported to have superior sensitivity with more than two orders of magnitudes compared to enzyme‐linked immunosorbent assay (ELISA), the golden‐standard immunoassay platform, even with much higher multiplexing capability.^{[ 5 ]}

Despite the robust multiplexing capability and detection sensitivity, the hydrogel‐based bioassay has normally been used in the laboratory fields instead of being applied to the clinical and industrial ones. One of the main reasons for the limited applicability is the absence of robust and long‐term storage protocol of the hydrogel microparticles functionalized with detection probes such as DNA and antibodies. The hydrogel microparticles for bioassays have normally been stored in an aqueous buffer with surfactants (e.g., Tween 20) under 4 °C for alleviation of heat‐induced denaturation of functionalized probes.^{[ 7} , ^{15 ]} However, this suspension‐based storage of the hydrogel particles has a critical drawback. For the hydrogel particles with functional biomolecules such as antibodies, the biomolecules are unstable and feasible to be denatured under aqueous solvent by unfolding of molecular structure and interaction with solvent and salts.^{[ 16 ]} This functional instability of the probe‐functionalized hydrogel particles in the storage limits the robustness and reproducibility of the assay. Furthermore, the suspension phase‐based storage can be vulnerable to microbial contamination in the samples.^{[ 17} , ^{18 ]} These issues in the storage process have hindered the hydrogel‐based microarray platform from being widely used in diverse biofields including clinical diagnosis.

Lyophilization is a highly efficient preservation strategy for diverse functional molecules such as protein and DNA.^{[ 19} , ^{20 ]} The many protein and DNA reagents in commercialized products such as ELISA kits are stored in a lyophilized form.^{[ 20 ]} After the lyophilization, the dried sample is highly stable since the molecular structure of the reagent is highly preservable under a solvent‐free environment. Also, whereas the evaporation of the solution is highly vulnerable to denaturation of functional molecules in a highly concentrated state right before the complete evaporation, the lyophilization process sublimates the solvent (e.g., water) in samples which is highly beneficial for the maintenance of molecular activities. Furthermore, the lyophilized samples have high resistance to microbial contamination by water‐absence conditions.^{[ 18} , ^{21 ]} However, the hydrogel microparticles are typically not compatible with the lyophilization process due to the low modulus to resist the stress induced by the freezing and drying process. Especially for nanoporous hydrogel microparticles typically used for biosensing applications, the particle has a considerably low modulus to resist the deformation of the particle structure.^{[ 22 ]} This structural deformation of the hydrogel microparticles is fatal in biosensing by 1) incapability of accurately identifying code information of deformed particles and 2) significant reduction of detection signal by damage of the porous structure of the particles which is crucial for in‐gel reaction.

Herein, we present a novel application of PEG nanofillers for lyophilizing encoded hydrogel microparticles using PEG as a filler material in the hydrogel structure to resist stress during lyophilization. PEG is a chemically inert polymer that can easily penetrate nanoporous hydrogel microparticles when the molecular weight of PEG is low (<40 kDa)^{[ 14 ]} We hypothesized that the incorporated PEG would contribute to maintaining gel structure and particle geometries in the lyophilization process by preventing the irreversible interaction between polymer chains and the collapse of porous structure attributed to the volume exclusion effect of inert PEG molecules. This hypothesis was established by noting the excluded volume effect of polymers, where each polymer molecule occupies a specific space and prevents the other molecules from entering that space.^{[ 23} , ^{24 ]} Actually, we demonstrate the enhanced geometrical resilience of the particles by adding the PEG as a nanofiller in the lyophilization process. The molecular weight and concentration of PEG are also optimized for the efficient preservation of particle geometries. Then, it is confirmed that the hydrogel microparticles functionalized with antibodies reconstituted from lyophilized form can exert comparable immunoassay performances to non‐lyophilized particles. Finally, the capability of long‐term storage (more than 6 months) of the antibody‐functionalized particles is verified by a thermal aging test.

2. Results and Discussion

2.1. Validation and Optimization of PEG Nanofiller‐Based Particle Lyophilization

The schematic illustration for the PEG nanofiller‐mediated lyophilization process for hydrogel microparticles is presented in Figure  1A. In the lyophilization process, there are several types of physical stresses during the freezing and dehydration processes. For the freezing process, the frozen solvent induces physical stress in and onto the hydrogel particles by phase transition‐based volumetric variation of the aqueous solvent.^{[ 20 ]} Since the frozen solvent has a high elastic modulus compared to the liquid state, the soft and porous hydrogel material is highly susceptible to deformation. Furthermore, during the dehydration process, the hydrogel particles can also be deformed by the sublimation of the water molecules inside the hydrogel structure.^{[ 25 ]} We hypothesized that the interaction between the polymer chains including entanglement in the hydrogel particles can lead to the considerable and irreversible deformation of particle geometry and porous structure which are fatal for the successful hydrogel‐based multiplex bioassay. Thus, in this study, we determined to utilize inert PEG molecules with nanoscale hydrodynamic diameters as filler materials during the lyophilization process for the preservation of the geometries and porous structure of the particles by inhibiting the undesired interaction between the polymeric chains. Among several material candidates, PEG was determined as the nanofiller considering the high hydrophilicity, antifouling property, and biocompatibility.^{[ 26} , ^{27 ]} After the lyophilization process, chemically inert PEG can be easily removed from the particles after reconstitution by repeating the solvent exchange‐based washing process (Figure 1A).

Figure 1.

A) Schematic overview of the PEG‐mediated lyophilization process of anisotropic hydrogel microparticles with graphical codes. B) Image of lyophilized samples containing hydrogel microparticles. C) Bright‐field images of hydrogel microparticles in aqueous buffer without lyophilization (left), after lyophilization without PEG (middle), and after lyophilization with PEG (right). D) Quantitative comparison of decoding accuracy of particles in deep learning‐based image analysis according to the presence of PEG in the lyophilization process (n = 3).

The image of the lyophilized sample containing hydrogel microparticles and PEG additives is presented in Figure 1B. Since the high concentration (10%) of PEG 8000 was incorporated into the suspension before the lyophilization process, the lyophilized product was in powder form. As shown in Figure 1C, the PEG‐based enhanced conservation of particle geometries can be confirmed by the remarkable collapse of the particle shapes for the particle without using PEG in the lyophilization process. To quantitatively evaluate the geometrical resilience after the lyophilization according to the presence of PEG, we characterized and compared the decoding accuracy of the encoded particles (Figure 1D). For the objective and precise evaluation, the deep learning‐based particle image analysis technique was adopted.^{[ 11 ]} Resultantly, high decoding accuracy (95.51%) of the lyophilized particles mediated with the PEG nanofiller was observed indicating the superb adaptability to the actual bioassay. Whereas the lyophilized particles without PEG filler showed remarkably decreased decoding accuracy (41.77%) by deformation of particle geometry. The decoding accuracy of more than 95% for encoded particles in multiplexed bioassay is an acceptable value considering conventional study even though there still exists the necessity of improving the accuracy in future works.^{[ 28 ]} The detailed procedures and data for microparticle synthesis and deep learning‐based particle analysis can be confirmed in Figures S1, S2 (Supporting Information).

To further validate the uniformity of the geometries of the particles after the PEG‐mediated lyophilization process, we imaged the reconstituted particles after the lyophilization and calculated the square root of the outer area of the particles as the characteristic length. As shown in Figure  2A, the uniformity of the encoded particles after reconstitution can be observed qualitatively validating the geometrical resilience after the PEG‐based lyophilization. Size distribution data of the lyophilized or non‐lyophilized particles in terms of the characteristic length is presented in Figure 2B. Despite the slight reduction of the characteristic length of the particles for the lyophilized case, the uniform size distribution could be confirmed by 5.4% of the coefficient of variation (C.V.).

Figure 2.

A) Bright‐field image of the hydrogel microparticles after reconstitution from the lyophilized state. B) The size‐distribution data of reconstituted hydrogel microparticles before and after lyophilization. The particular size was characterized by calculating the square root of the outer area of the particles. Decoding accuracy optimization according to the molecular weight C) and concentration of PEG D) in the lyophilization process (n = 3).

To optimize the PEG‐based particle lyophilization process, we screened the decoding accuracy of the encoded microparticles after lyophilization according to the molecular weight and concentration of the PEG nanofiller (Figure 2). For molecular weight, 4 kDa, 8 kDa, and 12 kDa of PEG molecular weight were screened at 10% concentration, and the optimal decoding accuracy was confirmed for PEG 8000 (Figure 2C). The decrease in decoding accuracy for PEG 4000 and PEG 12000 can be explained by considerable particle deformation induced by the low size‐exclusion effect of PEG 4000 and the high physical stress of viscous high‐molecular weight PEG 12000 solution, respectively. We further optimized the decoding accuracy by varying the concentration of PEG 8000 in the sample before the lyophilization (Figure 2D). Consequently, the highest decoding accuracy could be observed for 10% of the PEG concentration despite the slight difference compared to 15% of the concentration. This tendency also can be explained by the balance between the size‐exclusion effect and the viscosity‐driven stress depending on the PEG concentration. In further experiments, we determined to use 10% of the PEG 8000 in the lyophilization process instead of 15% concentration condition considering the decoding accuracy, higher processability at low solute concentration, and minimized consumption of the materials.

To demonstrate that the PEG 8000 can easily diffuse into the hydrogel microparticles for inducing filler effect, we fabricated the hydrogel post in the microfluidic channel and characterized the lateral diffusion aspect of the fluorophore (fluorescein isothiocyanate(FITC))‐linked PEG 8000 into the gel post (Figure S3, Supporting Information). Since the hydrogel post faces the solution only in a lateral direction, the penetration of the PEG 8000‐FITC can be analyzed by characterizing the fluorescence intensity of the gel post by top‐view imaging.^{[ 29 ]} Resultantly, the PEG 8000 could penetrate the hydrogel post within 20 seconds, which indicates the ease of the PEG nanofiller penetration into the hydrogel microparticles before the lyophilization process. We also confirmed the facile removal of the PEG nanofiller from the hydrogel microparticles by the washing process using the gel post‐based lateral diffusion assay (Figure S4, Supporting Information). The washing process for about 20s remarkably reduced (∼a factor of 200) the fluorescence signal of the gel post. This indicates that the PEG nanofillers remain in the hydrogel microparticles with inert material, and washing can easily remove the PEG nanofiller from the particle after lyophilization. In addition, the scanning electron microscopy (SEM)‐based morphological analysis of the hydrogel particles lyophilized with or without PEG was conducted (Figure S5, Supporting Information). As a result, the presence of the amorphous PEG after the lyophilization in the porous polymer network could be observed. Furthermore, the pore structures were more uniformly distributed, and the gel structure exhibited less collapse in the PEG‐based lyophilization case. This is consistent with the superior preservation effect of PEG for the hydrogel microparticles in terms of the particle geometry (Figure 1C,D).

2.2. Lyophilized Microparticle‐Based Immunoassay

To investigate the assay performance of the encoded hydrogel microparticles lyophilized using a PEG nanofiller, we conducted an immunoassay following the standard hydrogel‐based assay protocol (Figure  3A). The fluorescent particle image and detection signal after detecting the sEng target using the microparticles with or without lyophilization are presented in Figure 3B. The detection signal of the lyophilized case was a factor of 0.86 compared to the non‐lyophilized case indicating robust signal retention even with the lyophilization process. The singleplex assay for the sEng was conducted using the lyophilized and non‐lyophilized particles to compare the detection sensitivity (Figure 3C). As with the non‐lyophilized case, the calibration curve from the lyophilized particle‐based assay showed a linearized regression curve (R² = 0.9980) demonstrating the capability of robust quantification of the target analytes. From the calibration curve, the limit of detection (LoD), the target concentration corresponding to three times the standard deviation of the control signal, was calculated. Consequently, the LoD for lyophilized and non‐lyophilized cases was comparable (4.44 pg/mL and 2.16 pg/mL for each case), indicating that the PEG‐based lyophilization process does not significantly influence detection performances. The detection sensitivities of MUC1 were also comparable between the lyophilized and non‐lyophilized cases (0.991 mU/ml and 0.579 mU/ml for each case) (Figure S6, Supporting Information). We also characterized the reproducibility of the detection signal by assaying the five independent samples containing the sEng target using the microparticles lyophilized with the PEG nanofiller (Figure S7, Supporting Information). As a result, the coefficient of variation (CV) was revealed as 2.98%, presenting the detection process's reproducibility compared to the conventional bioassay techniques (CV from 2.8% to 10%).^{[ 30 ]} It is expected that the CV of the presented assay can be further enhanced by instrument‐based automation of the particle fabrication and bioassay procedures, minimizing the intervention of the users.^{[ 31 ]}

Figure 3.

A) Schematic of the encoded hydrogel microparticle‐based protein detection. B) Fluorescence signal comparison according to the lyophilization pretreatment of the particles after detecting 12500 pg/mL of sEng using the encoded microparticles. The scale bar is 30 µm. C) Calibration curve for singleplex detection of sEng using the encoded microparticles. The left and right graphs indicate the detection results using encoded particles before and after lyophilization. D) Multiplexed detection result for sEng (12500 pg/mL) and MUC1 (2500 mU/mL). For each data point, a total of 5–7 particles were analyzed.

To evaluate the multiplexed detection performance of the lyophilized particles, multiplex detection was also conducted using the two types of targets, sEng and MUC1, which are both biomarkers for prostate cancer (Figure 3D).^{[ 32} , ^{33 ]} Four combinations of the samples were assayed according to the presence (+) or absence (−) of each target. As shown in (−+) and (+−) cases in Figure 3D, there was negligible non‐specific detection signal indicating the robust specificity in the multiplexed assay. Furthermore, the detection signal of target‐presence cases was consistent for each target, which also indicates the robustness and reproducibility of the multiplexed assay. To confirm the consistency of the detection signal between the singleplex and multiplex assay, we characterized the recovery rate, the ratio between the calculated concentration by contrasting the detection signal from the multiplex assay with the calibration curve and the spiked target concentration (Table S1, Supporting Information). The recovery rate for sEng and MUC1 were 93.86% and 85.17%, respectively. Considering that the recovery rate ranging from 70% to 130% is acceptable for the industrial field, the developed particle lyophilization process is expected to be utilized in practical applications including diagnosis.

2.3. Thermal Aging‐Based Investigation of Long‐Term Storage of Microparticles

To investigate the feasibility of long‐term storage of encoded hydrogel microparticles via the PEG nanofiller‐based lyophilization, we conducted the immunoassay and characterized the detection sensitivity after thermal aging of microparticles. The 4 °C and 6 months were selected as goal storage conditions considering that many biomolecules or substrates loaded with biomolecules are typically stored at 4 °C or −20 °C.^{[ 7} , ³⁴ , ^{35 ]} To conduct a thermal aging test, it was required to calculate the storage duration under elevated temperatures corresponding to the virtual storage conditions of 4 °C and 6 months for the lyophilized or non‐lyophilized microparticles, respectively. Thus, we utilized the Q₁₀ method which utilizes the Q₁₀ as the temperature coefficient indicating the variation of biological activity according to the temperature difference.^{[ 36 ]} The accelerated aging test has been utilized as a short‐term experimental alternative to long‐term tests, leveraging the thermally accelerated kinetics of changes in the material and its molecular properties. Even though the direct analysis for a long‐term period may be more reliable, due to the constrained time, resources, and low throughput of the direct analysis, the thermal aging test has been widely utilized for diverse materials and devices, including biosensors.^{[ 37} , ³⁸ , ^{39 ]} In detail, the Q₁₀ value can be expressed by the Equation (1):

$$Q_{10}={\left[\frac{k\left(T_2\right)}{k\left(T_1\right)}\right]}^{10/\left(T_2-T_1\right)}$$ (1)

T represents the storage temperature of the microparticles functionalized with probes. Also, k indicates the rate constant for the variation of the biological activity of the probes immobilized in the microparticles during aging depending on temperatures. The variation of the detection signal of microparticles according to the storage period corresponds to k in this study. The Q₁₀ value can be acquired by conducting assays using microparticles stored under two different temperatures (24 °C and 34 °C) and analyzing the detection signal for each temperature case. By using Q₁₀, the k for 4 °C also can be calculated using the Equation (1). Then, we can finally calculate the required storage period of microparticles under 24 °C which corresponds to the storage conditions of 4 °C and 6 months.

To acquire the Q₁₀ value, we stored the lyophilized or non‐lyophilized (control case) microparticles immobilized with sEng antibodies for 0, 1, 3, 5, and 7 days. For each time point, the immunoassay was conducted for 12500 pg/mL of sEng, and the variance of a normalized detection signal was tracked as presented in Figure  4A,B. The regression graph was plotted for each time point to acquire the k value. By substituting slope values for each temperature condition in Equation (1), the Q₁₀ values were acquired (non‐lyophilization: 2.8, lyophilization: 4.1). The higher Q₁₀ value for the lyophilization case may be attributed to the high activation energy (E_a) of the stable lyophilized state considering that the higher E_a leads to the higher Q₁₀ according to the Arrhenius equation.^{[ 40} , ⁴¹ , ^{42 ]} The three temperature points‐based Q₁₀ analysis by tracking changes of k values according to the 24 °C, 29 °C, 34 °C was conducted to validate the reliability of the conducted Q₁₀ modeling using two temperature points (Figure S8A, Supporting Information). We confirmed the linearity of the graphs representing the rate constant variation according to the temperature (Figure S8B,C, Supporting Information). The Q₁₀ values were calculated from the regression line for the lyophilized and non‐lyophilized particles, respectively. As a result, the acquired Q₁₀ values were identical to the previous two points‐based approaches for the non‐lyophilized particles (Q₁₀ = 2.8) and lyophilized particles (Q₁₀ = 4.1), indicating the robustness and reliability of the presented modeling. Furthermore, using the acquired Q₁₀ values, the required storage period under 24 °C corresponding to the condition of 4 °C and 6 months could be calculated (non‐lyophilization: 22.8 days, lyophilization: 10.4 days).

Figure 4.

Thermal aging‐based investigation of the relationship between long‐term storage and detection performances of microparticles according to lyophilization. A, B) The normalized fluorescence intensity by detecting the 12500 pg/mL of sEng using the non‐lyophilized (A) or lyophilized (B) microparticles stored under 24 °C and 34 °C. The calibration curve for sEng using the non‐lyophilized C) or lyophilized D) microparticles that have been thermally aged under 24 °C to synchronize with the condition of 4 °C and 6 months. For each data point, a total of 5–7 particles were analyzed.

Finally, we conducted immunoassays using microparticles thermally aged under 24 °C for specific periods corresponding to the conditions of 4 °C and 6 months for non‐lyophilized and lyophilized cases, respectively. The calibration curves for non‐lyophilized and lyophilized cases can be seen in Figure 4C,D. Interestingly, the LoD of the lyophilized particle‐based assay after aging was 6.48 pg/mL, which is 5.2 times superior to that of the non‐lyophilized case (33.56 pg/mL). Furthermore, the LoD was comparable to that of the aging‐free assay using the lyophilized microparticles (4.44 pg/mL). Whereas, the non‐lyophilized particle‐based assay reported 15.5 times inferior LoD to that of the non‐aged case (2.16 pg/mL). The significant decrease of LoD only for the non‐lyophilized case can be explained by the denaturation of antibodies in the liquid phase in which the transport of molecules can be facilitated.^{[ 16 ]} Conclusively, these results demonstrate that the PEG nanofiller‐based lyophilization strategy not only preserves the particle geometry but also maintains the detection performances of the encoded hydrogel microparticles for a long‐term period.

In this study, we newly developed PEG nanofiller‐mediated lyophilization strategy of the graphically encoded hydrogel microparticles. There are two main novelties of the research in aspects of firstly addressing the lyophilization of encoded hydrogel microparticles and the idea of filling the porous hydrogel materials with polymeric molecules.

First, despite the high potential of the graphically encoded hydrogel microparticles to be applied to clinical and biological fields, there has not been a study to develop robust storage techniques for them. In the studies related to the encoded hydrogel microparticle‐based biosensing, the storage of the particles typically has been conducted in the suspension state at 4 °C,^{[ 7} , ^{15 ]} even though the biomolecules in hydrogel particles are susceptible to denaturing in solution phases. Despite the well‐known prominence of the lyophilization approach for stabilizing biomaterials, the suspension‐dependent particle storage protocol could not be avoided due to the deformation of the particle geometry by the lyophilization‐induced physical stress, as clearly shown in Figure 1C. To circumvent this limitation, for the first time, we developed the lyophilization strategy of the graphically encoded hydrogel microparticles with the robust preservation capability of the particle geometry and bioassay performances. Even though there have been several reports on lyophilizing hydrogel microparticles,^{[ 43} , ^{44 ]} those particles were tailored to be used in drug delivery carriers, for which geometrical preservation during the lyophilization was not a critical issue to implement the role of drug carrier. Thus, the fit‐for‐purpose development of the lyophilization technology capable of geometrical preservation of microparticles was a significant need for the encoded hydrogel microparticle‐based biosensing field, and we first showcased it with the PEG nanofiller strategy.

In terms of the engineering aspect, during the lyophilization and stabilization of diverse targets, including proteins, nanoparticles, and bioparticles (e.g., cells), several types of lyoprotectants have been used, including disaccharides, surfactants, and amino acids.^{[ 45 ]} The representative lyoprotectant is the disaccharide, especially trehalose and sucrose. These disaccharides can protect the complex molecular structure of the biomolecules, especially for proteins, by stabilizing the biomolecules with hydrogen bonds and vitrification of the phase.^{[ 46} , ^{47 ]} However, although those molecules have been commonly used as lyoprotectants, our experimental results demonstrated that these additives alone were insufficient to preserve the geometry of the encoded hydrogel microparticles, making them unsuitable for utilization in multiplexed assays (Figure 1). The surfactants (e.g., Tween 20) and amino acids (e.g., arginine) have also been utilized with the disaccharides in the lyophilization process for stabilizing the lyophilizate, inhibiting aggregation of the proteins and nanoparticles.^{[ 48} , ⁴⁹ , ^{50 ]} However, these surfactant‐ and amino acid‐based stabilization strategies mainly focus on the thermodynamic stabilization of the solution and colloids (e.g., dispersion), which can not be an intrinsic alternative to alleviate the physical stress on the low‐modulus and nanoporous hydrogel microparticles. As analogous to this study, the polymers such as PEG and dextran have often been used in the lyophilization process for biomaterials such as proteins and bacteriophages.^{[ 51} , ^{52 ]} The PEG has been utilized as a co‐solvent or bulking agent in the lyophilization process,^{[ 53 ]} which can induce the viscosity of the sample, leading to reduced denaturation and aggregation.^{[ 54 ]} However, the sole use of the PEG has been considered insufficient for robust lyophilization of biomolecules, and it has been used as an auxiliary reagent co‐utilized with the disaccharide‐based lyophilization process.^{[ 55 ]} Above all, to the best of our knowledge, there has been a lack of studies using polymers such as PEG for the lyophilization of the hydrogel‐based materials, especially for the micron‐scale hydrogel. This may be due to the hydrogel itself is a polymeric material, and the necessity of using additional polymer lyoprotectants can be overlooked.

In this study, we newly took note of the critical potential of the PEG as filler material inside the hydrogel structure. The use of PEG for the microgel lyophilization may be new work. However, the genuine conceptual novelty of the presented work is to fill the porous microgel network with PEG moieties. The filling of the PEG lyoprotectants into the hydrogel structure was validated by the gel post‐based lateral diffusion assay and SEM analysis (Figures S3, S5, Supporting Information). The dependency of the preservation of the particle geometry on the PEG nanofiller was qualitatively and quantitatively presented (Figure 1). It was further validated that the assay performance of the particles lyophilized with PEG nanofiller was comparable to the non‐lyophilized microparticles (Figure 3), and the performance can be maintained for 6 months (Figure 4). Conclusively, the developed lyophilization process of encoded microgel with PEG nanofiller is a conceptually and technically novel approach. Also, it has distinct advantages, the fit‐for‐purpose application to lyophilizing the graphically encoded microparticles with robust preservation efficacy of particle geometry and assay performances.

In the aspect of stabilization and lyophilization for hydrogel material, our PEG nanofiller‐based approach has differentiation and advantages.

Even though Elle et al. conducted lyophilization of the cellulose‐based hydrogel using polysaccharides (trehalose and sucrose),^{[ 56 ]} most of the previous hydrogel lyophilization processes have normally been conducted without specific strategies for the preservation of the hydrogel. Rather, the previous lyophilization processes were tailored for specific purposes such as drug encapsulation and formation of the macroporous intrastructure in the hydrogel. Song et al. presented the lyophilization‐based drug loading techniques into hydrogel microparticles, but the specific lyoprotectants such as trehalose were not used.^{[ 43} , ^{44 ]} Other studies have also conducted the lyophilization of the bulk‐scale hydrogel, and the purpose of the lyophilization was the formation of micron‐scale pores in the hydrogel.^{[ 57} , ^{58 ]} However, these lyophilization processes with or without disaccharides may not be suitable for the micron‐scale hydrogel with low modulus, as confirmed in this study.

For stabilization in the freezing process, several methods for freezing‐based or frozen‐state hydrogel synthesis (cryo‐annealing approach) can be considered. For example, the repeated freezing and thawing process of polyvinyl alcohol (PVA) solution has been presented for the synthesis of PVA hydrogel.^{[ 59 ]} However, the corresponding cryo‐annealed hydrogel has been reported to have low mechanical strengths even for the bulk‐scale formulation.^{[ 60 ]} Even though the photopolymerization of frozen‐state precursor can be exploited to synthesize a hydrogel, the as‐synthesized hydrogel has low mechanical properties due to the ice crystal‐induced large‐scale pores.^{[ 61 ]} Above all, these freezing‐based or frozen‐state hydrogel synthesis techniques have normally been subjected to generating bulk‐scale hydrogel.

The mechanical stability of the hydrogel in the lyophilization process can be enhanced by increasing the composition of the crosslinking agents (e.g., PEGDA) in the precursor. The higher crosslinking agent concentration in the precursor can facilitate the production of a densely structured hydrogel, enhancing the mechanical integrity.^{[ 62 ]} However, despite the simplicity and adaptability to micron‐scale hydrogel, it should be noted that the low porosity induced by a high amount of crosslinker can exert negative effects on the hydrogel‐based biosensing performances due to the constrained diffusion of the reactants into the hydrogel.^{[ 63 ]}

The direct drying of hydrogel without the freezing process has often been conducted for the functional preservation of the hydrogel.^{[ 64} , ^{65 ]} This direct drying process is highly simple and requires small and portable equipment compared to using a freeze dryer. However, the drying‐induced non‐desired interaction between biomolecules loaded in the hydrogel by the removal of water components can lead to denaturation of the bioactivities of those molecules.^{[ 66 ]} Furthermore, the removal of the water components inside the hydrogel during the direct drying process can lead to undesirable interaction and entanglement between polymer chains.

Encapsulating or filling additives in hydrogel materials can also be a strategy for the stabilization of hydrogel in the lyophilization process. To the best of our knowledge, the presented lyophilization strategy using PEG nanofillers is the premier work in terms of introducing sacrificial matrices in the hydrogel during lyophilization. Compared to other strategies, such as direct lyophilization, cryo‐annealing, addition of cross‐linkers, and direct drying, our approach is unique but powerful for lyophilizing delicate micron‐scale hydrogels by robustly preserving structural integrity, geometrical features, and assay performances. The detailed comparison of the presented study with previously reported stabilization and/or lyophilization processes of the hydrogel can be confirmed in Table S2 (Supporting Information).

3. Conclusion

In this study, we developed the PEG nanofiller‐mediated lyophilization process of graphically encoded hydrogel microparticles. By penetrating PEG molecules into a porous hydrogel structure, the successful preservation of geometries of encoded microparticles can be achieved via enhanced resistance to stress in lyophilization. Using deep learning‐based analysis, the high decoding accuracy (more than 95%) of PEG‐based lyophilized microparticles can be observed compared to that of the control case (41.77%), indicating the PEG nanofiller‐based superior preservation effect. Further optimization of the detailed factors such as sample volume and process time in lyophilization, and refinement of deep learning‐based analysis are expected to advance the decoding accuracy of the lyophilized particles. Noteworthy is that the PEG nanofiller‐based lyophilized microparticles exert comparable immunoassay performances compared to non‐lyophilized microparticles as demonstrated by the calibration‐based analysis. Especially, it is inspiring that the detection performance is still robust after thermal aging treatment which emulates the long‐term storage of the microparticles for more than 6 months under refrigerated temperature. The capability of long‐term particle storage mediated with PEG nanofiller can broaden the utilization of hydrogel‐based bioassays by avoiding redundant particle fabrication steps while maintaining the detection performances of the particles during storage. Furthermore, it can activate the commercialization of the assay technology by facilitating the distribution of the functional hydrogel particles from the manufacturer. In conclusion, the developed lyophilization strategy of the hydrogel microparticles is expected to accelerate the application of the hydrogel microparticle‐based sensing platform in diverse practical fields via the new finding of the PEG‐based filler effect.

4. Experimental Section

Materials

PDMS (Polydimethylsiloxane elastomer, Sylgard 184) was purchased from Dow Corning (USA). PEG‐DA (polyethylene glycol diacrylate, M.W. of 700 Da), PEG (Polyethylene glycol, M.W. of 600 Da and 8000 Da), Darocur 1173 (2‐hydroxy‐2‐methylpropiophenone), Tween 20, D‐(+)‐Trehalose dihydrate (T9531) were purchased from Sigma‐Aldrich (U.K.). BSA (bovine serum albumin 10×) solution, sEng/CD105 Quantikine ELISA Kit (DY1097) were purchased from R&D Systems (USA). Monoclonal Antibody (GP1.4) and ELISA Kit for MUC1, TCEP (Tris(2‐carboxyethyl)phosphine hydrochloride) solution, and SA‐PE (Streptavidin, R‐Phycoerythrin Conjugate) were purchased from ThermoFisher Scientific (USA). The Multi‐Screen HTS‐BV Plate (MSBVN1250) was purchased from Merck Millipore (USA). Phosphate Buffered Saline (PBS) tablet was purchased from Takara Korea Biomedical Inc. (Korea).

Production of Encoded Hydrogel Microparticles

PDMS micromolds were made by mixing the curing agent and PDMS base in a 1:10 ratio. Si wafer master mold with positive patterns of encoded particles was fabricated using photolithography with SU‐8. After pouring mixed PDMS on the wafer, it was baked for at least 4 hours in a 70 °C oven. The cured PDMS was carefully peeled off and sliced in an appropriate size with a razor blade. The discontinuous dewetting with degassed micromolding lithography technique was utilized to synthesize encoded microparticles.^{[ 67} , ^{68 ]} The precursor required for hydrogel particle synthesis was composed of 75% (v/v) PEG 600, 20% (v/v) PEG‐DA 700, and 5% (v/v) Darocur 1173. The PDMS micromold was degassed in a vacuum chamber at 0.1 Pa for 30 min. The precursor was uniformly loaded into micromolds and the cover glass was placed onto the loaded precursor. Then, by horizontally moving the cover glass carefully, the residual precursor was removed. The PDMS mold was exposed to UV light with an intensity of 200 mW/cm² for 135 ms for photopolymerization. The PBST solution (PBS buffer solution with 0.05% (v/v) Tween 20) was loaded onto the mold for particle recovery by capillary rise, followed by the particle rinsing in the microtube with PBST solution five times. After rinsing, the microparticles were collected and stored in 10 µL of PBST.

Conjugation of Capture Antibody to Synthesized Microparticles

The particles were conjugated with capture antibody using the linker‐free antibody conjugation technique.^{[ 69 ]} 5 µL of 1x PBST solution containing capture antibodies (0.15 µg/µL for sEng, 0.25 µg/µL for MUC1) was mixed with 5 µL of 0.2 mM TCEP solution and reduced for 1 hr at 25 °C. The 10 µL of the mixture was added to 10 µL of particle solution and incubated on a thermo‐shaker for 48 hr at 25 °C, followed by the particle rinsing in a microtube with PBST solution seven times.

Lyophilization of Hydrogel Microparticles

Preservatives of 10% (w/v) PEG 8000 and 0.3 M trehalose were added to 1x PBS buffer solution. Then, the solution phase in suspension was exchanged with the preservative solution by repeating three times 1) addition of 450 µL of the preservative solution to 50 µL of particle solution, 2) centrifugation at 6000 rpm for 30 s, and 3) elimination of 450 µL of supernatant. After the microtube was partially sealed with parafilm, the tube was stored in a ‐70 °C deep freezer for 4 hr. Then, the frozen sample was dried under 5 mTorr for 18 hr in the tabletop freeze dryer (TFD8501, IlshinBioBase). Following the completion of lyophilization, the microtube was placed into a vial and sealed with a rubber stopper in a glovebox containing argon gas. Then, the tube was tightly sealed with an aluminum tear‐off cap under air pressure. Before use, the dried sample was reconstituted in DI water and rinsed with PBST solution three times to eliminate preservatives.

Hydrogel‐Based Immunoassay

The sandwich immunoassay was performed in a MultiScreen 96‐wells filter plate on a thermo‐shaker at 25 °C and 1500 rpm.^{[ 5 ]} For each well, the mixture composed of 80 µL of PBST containing 50 microparticles and target proteins was reacted for 2 h. After repeating the rinsing step three times, 40 µL of PBST was refilled. Then the particle solution was mixed with 10 µL of detection antibody solution (12.5 ng/µL for sEng, 3.2‐fold dilution from the detection antibody concentrate of the ELISA kit for MUC1) and reacted on a thermo‐shaker for 1 hr. After repeating the rinsing step three times, 40 µL 1x PBST was refilled. For fluorophore labeling, the particle solution was incubated with 10 µL of SA‐PE solution (20 µg/mL) for 30 min. When the reaction was over, the particle solution was rinsed five times. For the duplex assay, the number of particles per target and concentrations of detection antibodies were maintained identically to the singleplex one. Each rinsing step was conducted using a negative pressure‐based rinsing device controlled by an Arduino microcontroller. All the buffers in the well were removed through a filter except for the particles and refilled with PBST solution.

Fluorescence Imaging and Analysis

Particle images were taken on an inverted fluorescent microscope (Axiovert 200, Zeiss) connected to a CMOS (scientific complementary metal–oxide–semiconductor, Prime, Photometrics, AZ) camera and fluorescence light source (Illuminator HXP 120V, Zeiss). To detect the fluorescence of SA‐PE, a red fluorescence filter set (λex/λem = 546/590 nm) was used. Images saved in TIFF format were analyzed in ImageJ software (National Institute of Health) to measure particle length and fluorescent brightness.

Deep Learning‐Based Decoding Accuracy Analysis

The appearance of undeformed particles before lyophilization was utilized as training data and the particle locations were labeled using the auto‐annotation method (Figure S2A, Supporting Information).^{[ 11 ]} 30 images were used for each code. The Mask R‐CNN model was employed to detect and decode hydrogel particles, using ResNet101 as the backbone for feature extraction from input images. The model was trained using transfer learning, leveraging a pre‐trained model to achieve faster convergence. The pre‐trained model was provided by MMDetection and trained on the COCO datasets. The learning rate scheduler was incorporated to control the learning rate during the training period. Starting with an initial value of 10⁻², the learning rate was progressively reduced by a factor of 10 at 20, 40, 60, and 80 epochs, reaching a final value of 10⁻⁶. The checkpoint minimizing validation loss was utilized for measuring decoding accuracy (Figure S2B, Supporting Information). The decoding accuracy was defined as the ratio of the number of correctly identified particles to the total number of analyzed particles. It was obtained by analyzing 50 particles for three repeated experiments and averaging these three values. This procedure was conducted in the case of lyophilization with PEG or without PEG, respectively (Figure S2C, Supporting Information).

Statistical Analysis

The quantification results for the decoding accuracy and fluorescence signal of the particles were presented with average ± standard deviation.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Jang W., Byun J. W., Choi J. H., Kim B., Bong K. W., Polyethylene Glycol Nanofiller for Robust Lyophilization of Graphically Encoded Hydrogel Microparticles. Small 2025, 21, 2503007. 10.1002/smll.202503007

Data Availability Statement

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