Lyophilizing SERS biosensors to enable translation into an easy-to-use assay

Abstract

The COVID-19 pandemic has highlighted the importance of point-of-care (POC) pathogen detection. Accurate and accessible diagnostic techniques for virus identification are crucial for controlling the spread of diseases and have profound implications for our communities and global health. Reagentless surface-enhanced Raman scattering (SERS) sensors offer a promising solution for POC testing due to their capability to function without complex processing steps. However, their application in this space is limited by the fact that these solution-based assays are challenging to administer, transport and store. To overcome these limitations, we employed a freeze-drying (lyophilization) process on reagentless SERS sensors and investigated their improved stability and shelf-life. We explored this mechanism using different concentrations of cryoprotectants. Lyophilized sensors were then tested in a mix-and-detect fashion by adding to the dry sensors a drop of the sample, consisting of saliva spiked with target DNA oligonucleotides relative to different SARS-CoV-2 variants. In addition, we further uncovered how lyophilization benefits sensors with a DNA-catalysis mechanism. In summary, our findings indicate that lyophilization substantially enhances the practicality and usability of reagentless SERS sensors, contributing to the translation of this powerful diagnostic tool to POC testing in remote areas with limited resources.

Introduction

The COVID-19 pandemic has demonstrated the importance of point-of-care (POC) pathogen diagnosis and highlighted current issues with the global diagnostic infrastructure. Early identification of pathogens via POC diagnostics is crucial and has helped government agencies control the spread of dangerous viruses not only in the community but also in healthcare settings.^1,2 Current standard diagnostic practices for infectious diseases are reverse transcription polymerase chain reaction (RT-PCR) in healthcare settings and lateral flow assays (LFAs) for POC. RT-PCR is considered to be the gold standard because of its ability to detect specific sequences of virus genomes with high sensitivity.³ PCR has several issues: it requires specialized lab settings, skilled technicians and longer amounts of time to produce final results.^4,5 Conversely, LFAs are simple, allow for faster testing and provide results in just a few minutes. These features make them ideally suited for POC diagnosis; however, LFAs are limited by low sensitivity and can produce a high rate of false negative results.^4,6 An example of this limitation is the known issue of false negative results in SARS-CoV-2 asymptomatic and early infections. Because of the limitations of the current diagnostic infrastructure, we still need sensitive diagnostic technologies to detect infectious diseases at the POC.

Recent years have seen many advances in the development of POC technologies, as described in this review and interesting examples.^7-10 In most cases, these technologies involve a reporter/tag binding a target analyte and different forms of sample processing to remove excess reporter (i.e., immobilization assay). In contrast, reagentless sensors are emerging as a powerful solution for POC testing because they do not require any processing steps.^11,12 Among them, surface-enhanced Raman scattering-based (SERS) biosensors based on functional nucleic acids are a convenient solution for infectious disease detection.^13-15 These sensors operate without any processing steps (i.e., reagentless), have high specificity and are cost-effective to fabricate. Because of these features, these sensors have been used with clinical samples and for in situ diagnostic applications.^16-19 In addition, SERS-based sensors allow for highly multiplexed assays, detecting many targets simultaneously to increase the accuracy and reach (versatility) of the assay. This capability has been displayed in a recent report using a 26-plex with SERS by combining Raman reporters with machine learning.²⁰ Our group has recently developed a reagentless SERS sensor that uses DNA catalysis to improve the sensitivity of the sensor.²¹ This approach allows for an improved limit of detection (LOD) of these SERS sensors, while conserving their reagentless feature, which is key for in situ and POC applications.

Despite their numerous promising advantages, reagentless SERS sensors have limited applications in POC settings. This application gap is partially due to the fact that these sensors are currently used in liquid solutions. Solution-based sensors require trained personnel or streamlined operations to be able to transfer the sensors and mix them with the sample. Additionally, the solution medium makes it harder to transport and store the sensors. Specifically, for reagentless SERS sensors based on functional nucleic acids, we have observed a limited shelf-life, which can be due to loss of colloidal properties, the effect of enzyme degradation and/or an entropy driven process (loss of the functionalized layer).²²

Lyophilization via a freeze-drying process has been previously used to improve the long-term stability of metallic nanoparticles for biomedical applications.²³ This process involves two steps: freezing and drying in which the drying process occurs in two ways: primary drying (ice sublimation) and secondary drying (desorption of unfrozen water). During lyophilization, several stresses can occur in the solution such as physical stresses including changing the concentration of nanoparticles during ice formation, mechanical stress due to ice crystal formation, pH and ionic strength changes, and surface dehydration.²⁴ To protect the materials in the solution from stresses and further damage during lyophilization, generally cryoprotectants are used. Sugar molecules such as sucrose, trehalose, sorbitol, maltose, etc. are employed as cryoprotectants, which facilitate the stabilization of the materials (i.e., sensor nanoparticles) by forming a glassy matrix with mobility. For example, previous studies reported that trehalose as a cryoprotectant helped to lyophilize a variety of reagents, ranging from antibodies to polymer microspheres without degradation because it offers several benefits over other sugar molecules.²⁵ For instance, trehalose has a high glass transition temperature, low intramolecular hydrogen bonding, and large hydration number, allowing nanoparticles not to aggregate during lyophilization.²⁶ Due to these properties, trehalose as a cryoprotectant can play an important role during the freeze-drying process. Literature studies showed that various cryoprotectants were used to preserve the photothermal characteristics and photoacoustic imaging signal generation of the nanoparticles after the freeze-drying process.²⁷

This approach was applied to a diagnostic assay to allow the use of molecular assays in remote areas with limited infrastructure.²⁶ In this report, all the reagents involved in an immobilization assay were lyophilized with cryoprotectants and bulking agents, to be pressed into easy-to-handle tablets. These reagents included quantum dots and magnetic microbeads functionalized with bio-receptors. While this approach demonstrated that the elements of bioassays can be tableted to enable POC applications, immobilization assays (as the one shown in the report) require multi-step processes with devices or operations to perform the necessary washing steps. The multi-step issue was partially mitigated by the same group by designing tablets capable of delivering the assay elements sequentially by layering the tablet.²⁸ However, the other issues related to immobilization assays remain. Because reagentless sensors do not suffer from the same operational issues, they could be a powerful expansion in this technological space by offering an easy-to-administer assay for remote areas with limited infrastructure. However, no study has been reported yet on the lyophilization of reagentless sensors to achieve long-term stability and to facilitate their application as POC diagnostic assays.

To fill this technological gap, we have designed and optimized a method to lyophilize SERS-based sensors and demonstrated their use as a one-pot diagnostic assay. Herein, we have employed freeze-drying on reagentless SERS sensors in the presence of various trehalose concentrations and investigated their stability and improved shelf-life. We expanded our study to test the lyophilization of a SERS sensor based on a DNA-catalysis mechanism, showing the benefits of lyophilization for the DNA-catalysis mechanism in reducing unwanted sensor activation. We demonstrated the application of the lyophilized sensors as a one-pot assay by directly adding to lyophilized sensors spiked saliva. Finally, we showed that the lyophilization process can be performed with freezing at −20 °C, showing the accessibility of this method. Overall, we demonstrated the lyophilization of reagentless SERS sensors showing the first steps in enabling the use of this powerful diagnostic tool in remote areas with limited infrastructure.

Experimental

Materials and reagents

D-(+)-Trehalose dihydrate, gold(III) chloride solution (200 mg dL⁻¹), silver nitrate, ammonium hydroxide, ascorbic acid, hydrochloric acid (1.0 N), magnesium chloride (1.0 M), trisodium citrate dihydrate, and 6-mercapto-1-hexanol (MCH) were acquired from Sigma-Aldrich. Tween-20 (10% solution) was purchased from Roche. Thiol–PEG (MW = 5000 Da) was obtained from NANOCS. 10 mM Tris-HCl and phosphate-buffered saline (PBS) buffer solutions were prepared from stock solutions of 1.0 M Tris–HCl and PBS 10× (both molecular biology grade) purchased from Fisher Scientific. Sodium dodecyl sulfate was purchased from Fisher Scientific. All synthetic DNA oligonucleotide strands (modified and unmodified) were purchased from Integrated DNA Technologies received in pH 8.0 IDTE buffer at 100 μM and stored in a −20 °C freezer. DI water mentioned herein refers to 18 MΩ obtained from a Millipore-Sigma Synergy Reverse-Osmosis system filtered through a 0.22 μm filter.

Sensor design

The sensors utilized in this study were designed and prepared based on our previously published work.²¹ Sensor mechanisms were designed based on the calculated Gibbs free energy (ΔG) for the DNA strands using UNAFold, with a strand concentration of 1 nM, [Na] = 150 mM at a temperature of 20 °C.²⁹ Table S1^† contains a list of all the nucleic acid sequences used herein.

Synthesis of nanostars

A gold seed solution (gold nanoparticles with a 12 nm diameter) was prepared using the Turkevich method.³⁰ Silver-coated gold nanostars (Ag@AuNS) were also prepared by following a published protocol.³¹ The synthesis process involves mixing 10 mL of DI water, 10 μL of 1.0 N HCl, 493 μL of 5.08 mM HAuCl₄ and 100 μL of gold seed solution in a flask while stirring at room temperature for less than 60 seconds. Then, 50 μL of 2 mM AgNO₃ and 50 μL of 0.1 M ascorbic acid were quickly added to the reaction mixture, causing a color change from pale orange to deep blue. Finally, the silver coating process was carried out by rapidly adding 0.1 M AgNO₃ and NH₄OH to the mixture and stirring until the solution changed to a wine-red color. The synthesized Ag@AuNS solutions were stored at 4 °C for future use.

Sensor functionalization

First, 2 μL of 100 μM stem-loop DNA probe solution was added to 898 μL of ≈0.1 nM Ag@AuNS solution, followed by 100 μL of 2.5 mM MgCl₂ solution, and incubated overnight. Thiolated PEG was added to the solution at a final concentration of 10 μM and kept at room temperature for 1 h. The solution was centrifuged at 4000 rcf for 10 min and washed in 10 mM Tris–HCl with 0.01% Tween-20 to remove excess salts and PEG. The washed sensors were incubated in 0.1 mM MCH for 10 minutes at 37 °C, centrifuged, and washed three times. After the third wash, the sensors were redispersed in 1× PBS with 0.01% Tween-20 with 500 nM placeholder DNA (final concentration) and incubated at 37 °C for at least 12 h. The sensors were then centrifuged at 4000 rcf for 10 min and washed in 1× PBS with 0.01% Tween-20 three times to remove the unhybridized placeholder. These sensors use Cy5 as a Raman reporter attached to the probe strand.

Lyophilization of sensors

The prepared sensors were mixed with an aqueous solution of 20%, 10%, and 2% trehalose (cryoprotectants) in a 1 : 1 volume ratio to gain final trehalose concentrations of 10%, 5%, and 1%. For freezing, the samples were placed in a −20 °C or −80 °C freezer for at least 48 h until use. For freeze-drying, dry ice was put in a freeze-drier (LABCONCO Dry Ice Benchtop Freeze Dry Chamber 7522700), the pre-frozen sample was attached to the sample valve and the process was performed for at least 24 h. The quantity of dry ice was checked periodically and replenished as necessary. The resulting lyophilized samples were stored at room temperature until use (for 0–60 days). The frozen and freeze-dried samples were visually inspected for appearance and then reconstituted with 0.5 mL DI water (the same volume as before lyophilization). The control/room temperature (C), freeze-thaw (FT), and freeze-dried (FD) reconstituted sensors were analyzed using UV-vis spectroscopy, Nanoparticle Tracking Analysis (NTA), Transmission Electron Microscopy (TEM), and SERS. For all experiments conducted in this work, triplicate measurements were utilized.

UV-vis absorbance spectroscopy

The extinction spectra of the nanostars and sensor solutions were recorded between 300 and 900 nm, using 100 μL of samples, with a Synergy HTX microplate reader (BioTek) in a flat-bottom polystyrene 96 well plate (Corning).

Nanoparticle tracking analysis

A NanoSight NS300 (Malvern Panalytical) nanoparticle tracking analysis (NTA) system with a 532 nm laser was used to measure the size of nanoparticles and their relative concentration. For the NTA measurements, all sensors control, freeze-thaw, and freeze-dried reconstituted in DI water were diluted 20 to 40 times.

Transmission electron microscopy

The nanostars were characterized using transmission electron microscopy. The sample preparation process for TEM imaging is briefly described here. First, PEG–SH (final concentration: 10 μM) was attached to the nanostars for 30 minutes and then centrifuged at 4000 rcf for 10 minutes at 4 °C and resuspended using DI water. Then 10 μL of the resuspended solution was placed on a 300 mesh TEM Cu grid (Ted Pella) and allowed to air dry at room temperature. Images were taken using a TEM Talos F200i (Thermo Fisher) at 200 kV at the Advanced Material Characterization Center (AMCC) of the University of Cincinnati. Fig. S1^† shows the images of gold nanostars and silver-coated gold nanostars.

SERS measurements

All SERS measurements were performed using a lab-built Raman microscope setup to read from 96 well plates. Samples were measured in a flat-bottom polystyrene 96 well plate (Corning). The sensor response was tested by making a solution of 50 μL of sensors and 50 μL of target of interest. For the blank measurement, the target volume is replaced with 1× PBS. After mixing with target solutions, the sample was incubated for 2 h at room temperature before taking the measurements. Three spectra were collected for each sample. The lab-built microscope uses a 633 nm laser (IPS; power at the sample of 5 mW), a 633 nm Raman filter set manufactured by Semrock and a fiber-coupled spectrometer custom-made by Wasatch, which includes a WP633 connected to a Newton920 CCD (Andor) coupled with a circle-to-line 7-fiber bundle. The integration time for each spectrum was 1 second. The automated stage was controlled through LabView software (National Instruments).

Saliva extraction

A PURE-SAL^™ saliva purification for liquid biopsy kit (Catalog #PRSAL-401) was utilized to collect roughly 1 mL of saliva samples from two healthy individuals (lab volunteers). The saliva was freshly collected from under the tongue, following kit instructions.

Data analysis

Data analysis and visualization were carried out using Matlab R2021b (MathWorks) and Igor Pro V8.04 (WaveMetrics). A lab-built MATLAB script was used for spectral baseline subtraction using a Savitzky–Golay filter. For the catalytic mechanism, peak intensities were normalized by dividing the spectra by the maximum fluorescent signal from the Cy5 Raman reporter sensor. The spectra and peak intensities were plotted using Igor Pro. Our sensors were characterized based on their SERS response, which is characterized by a Cy5 peak at 560 cm⁻¹.

Results and discussion

Proof-of-principle demonstration

To prove the main hypothesis that reagentless SERS sensors can be lyophilized to enable their use at the POC, we first demonstrated and optimized a lyophilization process for the sensors. Fig. 1a shows a schematic of the reagentless sensing mechanism of a SERS sensor. In this mechanism, the sensor is turned on by a genetic target by displacing the placeholder (PH), complementary to the target, and leaving the probe to form a hairpin structure. This process changes the distance of a Raman reporter, bringing it within a few nm of the surface of the metallic nanoparticle. The nanoparticles used in our sensors are silver-coated gold nanostars, characterized in Fig. S1.^† To test the lyophilization process for these sensors, we prepared four sensor solutions with varying trehalose concentrations (Th0%, Th1%, Th5%, and Th10%) at room temperature. Trehalose has been used as a cryoprotectant to protect nanoparticles and assay elements from degrading during the lyophilization process; however, possible interactions with nucleic acid-based sensors are not known.^22,26 These solutions containing the sensors in a PBS buffer underwent three different processes: one was kept at room temperature (control), one was frozen at −80 °C and then thawed (FT), and one was frozen at −80 °C and then dried with a lyophilizer (FD). The FD samples were then resuspended in deionized water to the initial volume. The insets in Fig. 1b show pictures taken from the sensor solutions in different states of the process (initial, frozen and lyophilized, respectively, from left to right). Additionally, calibration curves for the sensors before and after the freeze-drying process are reported in Fig. S3,^† showing no significant changes in the LOD after the process.

Fig. 1.

Lyophilization process for a reagentless SERS sensor. (a) Schematic representation of the reagentless SERS sensing mechanism. (b) SERS sensor response through different processes representative of different steps in the freeze-drying process for samples containing different concentrations (% w/w) of trehalose (Th) as a cryoprotectant. Response measured as the difference between sensor signals for a blank and in the presence of target DNA oligonucleotides ([T] = 100 nM). Error bars represent the standard deviation between independently prepared samples (n = 3). The insets show a sensors solution (Th5%) at their initial stage, after freezing and after lyophilization. Photos of all sensor samples and conditions are reported in Fig. S2.^†

After the process and resuspension, the sensors were tested with synthetic targets to evaluate their sensor response. These specific sensors were designed in our previous work to target a region of the SARS-CoV-2 genome used for identifying the omicron variant.²¹ Fig. 1b shows the SERS signal from different samples with and without the addition of the target. A control experiment shows no-significant differences between the samples at different trehalose concentrations, indicating that the cryoprotectant does not negatively impact the sensors. The only exception is a slightly reduced signal at the highest concentration of trehalose (Th10%). The FT samples showed a significant role of the cryoprotectant in the process. Specifically, Th0% samples had significantly higher signals under both blank and target conditions. This observation can be an indication of aggregation in the sensors, which pushes the Raman reporters to be in contact with the nanostars. Some effect can also be seen in Th1%, with a slightly increased blank signal. The FD samples showed similar results to FT, with the samples containing enough trehalose (i.e., Th5% and Th10%) showing a good response. The samples without trehalose had a high blank signal, although the overall signal was lower, likely because of irreversible precipitation. Th1% had a higher blank signal. In conclusion, from our experiment, we identified the key role of trehalose in the freeze-drying (lyophilization) process and a likely optimal concentration of 5% in weight.

To investigate our hypothesis of trehalose protecting the sensors from aggregation, we analyzed our sensor by recording the extinction spectra and size distribution of the samples. These results are shown in Fig. 2. Ag@AuNS functionalized with a synthetic DNA have a characteristic localized surface plasmon peak at 508 nm (≈500 nm, before DNA functionalization), as observed in the spectrum of the sensor solutions. For the sensor solution under all conditions, no red- or blue-shift in the UV-vis spectra was observed, suggesting that there was no change in the morphology or functionalized layer of the nanostars during the process. However, variations in the intensity were observed. For example, FT-treated Th0% sensor solutions (Fig. 2b) showed a decrease in the absorbance compared to other solutions indicating that the sensor concentration decreased, which can be attributed to the loss of nanoparticles by aggregation (and precipitation) in the absence of a cryoprotectant. Fig. 2b also shows that almost no change in the intensity was observed for the Th1% to 10% solutions, highlighting that cryoprotectants are protecting the nanoparticles during the freezing process. A drastic drop in the absorbance for FD Th0% was observed, as depicted in Fig. 2c, suggesting that almost all the sensor nanoparticles precipitated in the absence of a cryoprotectant during the lyophilization process. There was a slight increase observed in the absorbance for the rest of the solutions with the increase of the Th% concentration, likely because of the same mechanism.

Fig. 2.

Characterization of sensor solutions through different processes representative of different steps in the freeze-drying process (C = control, FT = freeze-thaw, and FD = freeze-dry). (a–c) The extinction spectra of sensors. (d–f) NTA showing the size and concentration of functionalized sensors (y-offset for clarity). Each figure reports the calculated concentration. All graphs report different concentrations of trehalose as different colors. Specific sizes of the sensors from these samples are reported in Table S2.^†

To further characterize and support the extinction spectra observation, nanoparticle tracking analysis (NTA) was carried out to determine the size distribution and concentration. Fig. 2d-f show the NTA measurements for all sensors in the different processes. The position of the peak represents the size and the height represents the concentration. We observed that all the sensors under different thermal conditions are consistent in size; however, the observed concentration was different. The significant decrease in the particle concentration for FD samples indicates some loss of sensors during the freeze-drying process or inaccuracies in the resuspension step. Also, FD samples seem to show a secondary peak with roughly double the size of the particles, possibly indicating some aggregation residual from resuspending the lyophilized particles. Fig. S2^† shows the images of all the solutions undergoing these processes, showing the aggregation as a color loss, in agreement with the extinction and NTA data. Interestingly, frozen samples at Th1% show almost no color, which is then regained after thawing. In general, these complementary techniques support the conclusions made based on the SERS signal data.

Shelf-life extension via lyophilization

After demonstrating the feasibility of lyophilization for reagentless SERS sensors, we tested the effect of lyophilization on the shelf-life/long-term stability of sensors. In this study, sensors with different concentrations of trehalose (Th0%, Th1%, Th5%, and Th10%) were dried and stored at room temperature. The sensor response was measured by resuspending the sensors in DI water in 1 week intervals. These results were compared with those of sensors stored in solution. Fig. 3 shows the response of sensors as a function of time under the two storage conditions, stored in solution (C) and lyophilized (FD). The SERS sensor response was calculated as shown in Fig. 1b, and blank and target signals for all samples are reported in Fig. S6.^† As can be seen in the figure, the response decreases over time, with almost no response observed after 40 days. However, lyophilized sensors exhibited constant SERS responses retaining roughly 100% of the response after 35 days. As previously observed, the samples undergoing the freeze-drying process without the cryoprotectant (Th0%) did not show any response. These results show that the lyophilization can stop the degradation processes occurring in reagentless SERS sensors and extend the shelf-life of the sensors possibly for years, assuming that the sensors can be kept dry.

Fig. 3.

SERS response of solutions (C) and freeze-dried (FD) sensors at different time intervals of storage at room temperature. Bars of different colors represent samples with different trehalose amounts mixed with the sensors (Thx%). Dashed control line represents the response of a sensor solution without trehalose as synthesized. Error bars represent the standard deviation (n = 3).

To understand the processes involved in the degradation and the role of lyophilization, we studied SERS spectra, extinction spectra and size distribution for these samples at different time intervals. Fig. 4 shows the background-subtracted spectra for the samples (C and FD) at the beginning and end of the time range, both for the blank and after target addition. As can be observed, the signal for C significantly decreases at 41 days. The signal for FD stays constant for 5% and 10% trehalose samples. For Th0%, the sensors aggregate and cannot be resuspended due to the irreversible precipitation. Interestingly, for Th1%, the sensors seem to have aggregated and increased the overall blank signal, as discussed in Fig. 1 for Th0%. Unlike FD, the C samples seem to show a reduced response rather than an aggregation induced loss or a gain in the signal. This behavior could be due to degradation of the functionalization layer (DNA probe and placeholder). Results observed via images, extinction spectra and size distribution confirm the absence of aggregation. Fig. S2^† shows the photos of samples across different time ranges. As can be seen, the solution-stored samples retain the full color from the suspended sensors. In contrast, FD samples after 35 days with 0 and 1% trehalose lost all color. Fig. S4 and S5^† show the extinction spectra and size distribution of all samples across the studied range of time. The same behavior is observed in the extinction spectra, although more variability is observed in these samples. Th5% and Th10% samples stored in solution retain the initial OD after 40 days, while we observed that the SERS signal is significantly reduced. The size distribution generally verifies these results, although showing a lower concentration of particles for FD samples and signs of aggregation for solution-stored (control) samples after 40 days. In conclusion, the loss of the signal over time in solution from the sensors is at least in part due to degradation of the functionalized DNA layer on the surface of the particles, which we show can be prevented by lyophilization.

Fig. 4.

SERS spectra from sensor samples stored under different conditions at the beginning and end of the time range studied (C # = control/solution-stored for # days and FD # = lyophilized/freeze-dried stored for # days). Light lines represent the blank spectra (_B) and heavy lines represent spectra in the presence of target DNA oligonucleotides (_T). Spectra of different colors represent sensors mixed with different concentrations of trehalose as a cryoprotectant (Th#%, # = % w/w of trehalose). Spectra are offset by 5k counts on the Y axis for the different Th% and offset of 200 counts between the blank and target. Both offsets were done to improve visualization. A yellow highlight is used to indicate the peak at 560 cm⁻¹ used to measure sensor response.

Lyophilized reagentless sensors in mix-and-detect applications

To demonstrate the ease of using lyophilized reagentless SERS sensors, we tested the lyophilized sensors for the detection of samples spiked with DNA oligonucleotides from the SARS-CoV-2 genome sequence in a mix-and-detect fashion (i.e., by simple addition of a liquid sample to the sensors). In this experiment, we used sensors detecting a region of the SARS-CoV-2 genome used to identify the omicron variant. We then added solutions of two DNA oligonucleotide targets (one for the B.1.1.529 – omicron – variant and one for the B.1.617.2 – delta – variant, differing by a single base deletion) in DI water and in saliva extracted from two individuals (diluted by 50%). The sensors were lyophilized with 5% w/w trehalose, based on the optimization shown in the previous sections. The dried sensors for this test were mixed with solid SDS required for a final concentration of 5% w/w after resuspension. This step was added because the dried sensors will require combination with a lysing and denaturing agent (e.g., SDS) in their final mix-and-detect application. Fig. 5 shows the SERS response for the detection of SARS-CoV-2 variants by lyophilized sensors. The results show that the sensors are only activated in the presence of the omicron target. No significant response is observed for the delta target (single-base difference). The same response is observed in buffer and 50% saliva, showing that the sensors can perform in this medium without losing specificity and response. Importantly, these results were obtained with the mix-and-detect method, which involves simply adding a liquid sample to a tube containing the lyophilized sensors. This method is consistent with POC applications because it does not require any processing steps, other than the addition of the sample to the tube. While at the present stage, the method was used on free-DNA-oligonucleotide samples, we expect the lyophilization process to be easily combined with other reagents (e.g., lysing agents as SDS) to be able to work in a one-pot assay for encapsulated RNA as well (i.e., viral particles contained in saliva samples). In summary, these results show how lyophilization will enable POC applications of reagentless SERS sensors.

Fig. 5.

Mix-and-detect analysis of a solution containing SARS-CoV-2 DNA oligonucleotide targets with lyophilized sensors (blank = DI water or 50% saliva; omicron and delta = 10 nM target solutions in DI water or 50% saliva; all samples also contain 5% SDS). Delta and omicron targets differ by a single base deletion. Error bars represent the standard deviation (n = 3). Significant difference from the blank is shown on top of the bars (*** for p < 0.001, no notation indicates no statistical significance).

Freeze-drying of sensors based on the DNA-catalysis mechanism

In a recent report, we demonstrated the use of DNA catalysis to improve the analytical performance of SERS sensors.²¹ The developed catalytic sensing mechanism is shown in Fig. 6a. As can be seen, the forced complementarity of the strands used in the mechanism causes the fuel strand used to interact with the off state of the sensor, which causes the blank signal to increase in the presence of the fuel. We called this effect fuel interference, which is calculated as the difference between signals in the presence and absence of the fuel. While the blank signal increases after the addition of fuel, the significant effect of the DNA catalysis in the presence of the target makes up for the increased blank signal (the signal-to-blank ratio in the presence of the fuel is greater than in its absence). However, the interference effect increases over time, because of the increased probability of the fuel turning on a probe. Fig. 6b shows the interference effect as a function of time. As can be observed, the interference keeps increasing as a function of time over 3 days (72 h). This increasing interference becomes a significant issue if the sensors are stored over a long period of time, which is expected for sensors that must reach remote locations for POC applications as well as for most real-world applications.

Fig. 6.

Role of lyophilization in a sensing mechanism based on DNA catalysis. (a) Schematics of the mechanisms for reagentless SERS sensing amplified by a fuel strand using DNA catalysis. (b) Normalized signal of sensors stored with and without the fuel strand at room temperature for 3 days, showing the fuel interference effect over time. (c) Sensors response as a function of target concentration ([T]) with and without fuel (+fuel) for sensors stored under different conditions (solution or lyophilized (FD)) and for different times (0 and 5 or 7 days for solution and FD, respectively). Fuel concentration ([F]) used is 2 μM. Red dashed line reports the benchmark interference for catalytic sensing. (d) Interference for the different storing conditions and times calculated as fuel – blank. All error bars represent the standard deviation (n = 3). Data without normalization are available in Fig. S7.^†

To remove the issue of increasing interference in long-term storage, we tested the optimized lyophilization process on these sensors. The normal procedure for the use of catalytic sensors is to add the fuel strand to the sensing mix at the same time as the target, to minimize fuel interference. For the lyophilization process, we mixed the sensors with the fuel strand right before starting the lyophilization process. This way, we limited the time for the fuel to interfere with the sensor while also creating a lyophilized sensor with all the required elements in it (sensors + fuel), such that they can still work with by the mix-and-detect method. In this study, we compared the signal as a function of target concentration for normal sensors and catalytic sensors (i.e., with fuel addition), with the catalytic sensors tested under different storage conditions (solution and freeze-dried). Fig. 6c shows the sensor response for these samples. First, we observed the amplification effect of the catalytic sensing mechanism between solution and solution + fuel. As expected, the addition of fuel allows the detection of lower concentrations of the target (0.5 nM vs. 5 nM detectable without fuel). This result is in agreement with our previous report.²¹ The solution + fuel sample represents the traditional way the catalytic sensors are used (i.e., simultaneous addition of the fuel and target) and is used in the figure as the benchmark interference level. When the sensors are used after storage in solution for 5 days (solution + fuel (5 d)), we can easily see an increase of the interference effect, due to the fuel continuous interaction with the sensors. The data labelled as FD + fuel show the sample where the fuel is lyophilized with the sensors, followed by addition of the target and DI water to the lyophilized sensors. As can be observed, these samples are statistically identical to the solution results. Interestingly, when these samples are stored in a lyophilized state, the interference increase is greatly reduced from a 50% increase after 5 days to a 15% increase after 7 days. This difference is easily observable in Fig. 6d, where the interference for the various samples is reported. We believe that any increase must be due to incomplete lyophilization and expect that the increase should be virtually zero for a fully lyophilized sample. It is noteworthy that the results shown in this figure are normalized based on the background fluorescence signal from the reporter used on the sensors (Cy5), which is different than other data in this report. This difference is due to the slight decrease in the sensor signal observed after lyophilization in previous studies. This decrease would have resulted in decreasing interference as a function of time for the lyophilized sample, which is not possible based on the catalytic mechanism. For completeness, the results analyzed without normalization are shown in Fig. S6.^† All samples in this study were lyophilized with 5% trehalose. In summary, these results show that the lyophilization enables the storage of sensors based on DNA catalysis by blocking any off-target effects of the catalytic elements in the sensing mechanism. Importantly, the lyophilization also allows pre-loading of all elements involved in the mechanism in a single tube that only requires addition of the sample solution under analysis. This strategy enables the use of a more complex sensing mechanism for POC applications by eliminating processing steps.

Freeze-drying procedure with a conventional freezer

To extend the applicability of the developed strategy, we demonstrated that the lyophilization process can be performed using conventional freezer temperature (−20 °C) during the freezing step. We lyophilized sensors with and without fuel using a −80 and a −20 °C freezer for the freezing step. Both methods used 5% trehalose concentration. We used the sensors based on DNA catalysis for this demonstration, which are stored for 1 week in the lyophilized state. The results for the two freezing methods are shown in Fig. 7. As can be observed, both temperatures yield working sensors with the DNA-catalytic mechanism. Importantly, no significant difference was observed between the −20 and −80 °C methods. Both exhibited similar trends as a function of concentration and in the normalized signal magnitude, with the exclusion of the highest concentration samples that showed lower signals. This difference may be due to some deactivation of probes during the −20 °C freezing step possibly due to degradation before reaching or during the frozen state. Importantly, these data show how the lyophilization process can be performed with a −20 °C freezer with almost no impact in analytical performance. Being able to perform this step in a conventional freezer can increase the use of this strategy in labs with fewer resources, as −80 °C freezers are very expensive to maintain and are rarely found in low-resource settings.

Fig. 7.

Response of sensors as a function of target concentration ([T]) with and without fuel (+fuel) for sensors stored in the lyophilized state for one week. Freezing step in the freeze-drying process performed with an −80 °C freezer or a conventional −20 °C freezer. All error bars represent the standard deviation (n = 3).

Conclusion

In summary, we optimized and demonstrated a lyophilization process for reagentless SERS sensors. This strategy enables the extension of the shelf-life of these sensors that tend to lose responsiveness over time, due to loss of colloidal stability and degradation of the functionalized layer. Having the sensors in a lyophilized/powder form also allows them to perform as a very simple one-pot assay that only requires the addition of target solution to a tube, namely mix-and-detect. This simple assay in combination with the extended shelf life can potentially enable the use of this type of sensor in remote settings for POC applications, thanks to the demonstrated strategy. We further extended the use of lyophilization to sensors that use a DNA-catalysis mechanism, showing the benefits of storing this type of sensor in the lyophilized state, removing interference effects between assay components. In addition, we also show that the lyophilization process can be performed with a conventional freezer with identical performance.

In conclusion, reagentless SERS sensors are a powerful tool offering unique features (i.e., highly multiplexed + one-pot assays). However, they have found limited use in POC applications and are currently hard to implement in clinical or low-resource settings. Herein, we demonstrated a strategy that can bridge this translational gap in two ways. (1) Long-term storage allows the sensors to be shipped and used on-demand, facili-tating use at the POC. (2) The mix-and-detect method enables the direct addition of the sample directly to a tube containing sensors in powder form. This solution enables the performance of assays without any processing steps, such as washing or even mixing specific quantities of sample and sensing elements. We believe that the advancements from this work in combination with current and future advances (e.g., use of miniaturized Raman systems for POC applications¹⁰ and combining additional assay elements in the lyophilized powder) will enable the use of reagentless SERS sensors at the POC, offering a new powerful tool to medical professionals in remote settings.

Data availability

The raw data (SERS spectra, UV-vis spectra and nanoparticle tracking analysis) supporting this article have been included as part of the ESI.^†