Mitigating the deleterious impact of saliva on cell-free expression sensors for point-of-use diagnostics

Abstract

Point-of-use sensors that detect biomarkers in noninvasive biofluids, such as saliva, could provide individuals with valuable insights into their health and performance status at home or in the field. Cell-free biosensors have the potential to meet these requirements, as these sensors can detect a wide variety of analytes and can be stored at room temperature after freeze-drying. However, saliva inhibits the function of cell-free biosensors, in part due to the presence of interfering proteins and nucleases. Here we show simple strategies to mitigate the inhibitory effects of saliva on the function of cell-free expression systems, including the use of nuclease inhibitors and pretreatment of saliva with centrifugal filtration or heat. We also show successful implementation of our mitigation strategies to detect cortisol spiked into pooled human saliva using a cell-free biosensor. Combined, our results show that simple approaches can be used to enable the function of cell-free biosensors in saliva.

Graphical Abstract

Graphical Abstract.

Introduction

Point-of-use diagnostics provide practical mechanisms for monitoring health and performance, giving individuals actionable insights to adjust medication, improve health, or refine athletic training. Cell-free expression (CFE) systems offer a promising platform for developing point-of-use diagnostics, as they provide open, membrane-free system that can detect a wide variety of analytes and can be stored without a cold chain after freeze-drying [1, 2]. Furthermore, CFE systems can be embedded within various materials for functionalization in complex, real-world samples [3, 4]. CFE biosensors have been developed to detect pathogens [3, 5–7], water quality targets [8, 9], pharmaceutical contaminants [10], and human health and performance biomarkers like indicators of metabolic and cognitive states [11–13]. The cumulative advantages of CFE systems suggest this platform has potential for developing a variety of point-of-use sensors.

Non-invasive sampling, such as saliva collection, is ideal for point-of-use biomarker monitoring. However, many biofluids are known to inhibit CFE assays, likely due to the presence of nucleases, proteases, and other interfering molecules [6, 12, 14–17]. To mitigate the inhibitory effects, previous studies have used various approaches, including adding nuclease inhibitors like murine RNase inhibitor (mRI) and implementing different treatments to the biofluids before use [6, 15, 17–21].

Here, we sought to evaluate the impact of pooled human saliva, either untreated or pretreated with different filtration or heating strategies, and mRI on the performance of a homemade CFE system. Additionally, we adapted a whole-cell biosensor for detecting cortisol to function in our CFE system. We then demonstrated the ability of the freeze-dried biosensor to detect cortisol when rehydrated using pre-filtered saliva and an extract prepared from Escherichia coli overexpressing the mRI protein. The successful detection of cortisol in pretreated saliva using a freeze-dried CFE system opens the door for developing at-home diagnostic tests for stress and other health and wellness applications.

Materials and methods

Plasmids and materials

The plasmid for constitutive sfGFP expression (pJBL7010) was shared by Julius Lucks’ laboratory (Addgene 136 942) [25]. The cortisol sensor plasmid (pLysRsfGFP) was previously generated at the U.S. Air Force Research Laboratory [23] (Supplementary Table S2). The mRI overexpression plasmid (pPLV_C1) was obtained from Addgene (186415). All plasmids were propagated in E. coli DH10B cells and purified using Qiagen Mini Prep kits.

Extract preparation

CFE extracts were prepared from E. coli BL21^* DE3 cells as described previously [12, 25]. Standard cultures were grown overnight in LB media, while mRI-enriched cultures (harboring pPLV_C1) included ampicillin. Overnight cultures were diluted into 2 × YTP media; mRI-enriched cultures were induced with 1 mM IPTG at an OD600 of 0.4–0.6. Cells were grown to an OD600 of 2.5–2.8, harvested, washed twice with cold S30 buffer, and frozen. Thawed pellets were resuspended in S30 buffer (1 ml/g) and lysed via sonication (25% amplitude, 200–600 J). After clarification, lysate underwent an 80-min runoff reaction at 37°C, overnight dialysis (3.5 kDa MWCO) against S30 buffer, and final clarification before storage at −80°C.

CFE reaction setup

A master mix of reagents was prepared with components at concentrations listed in Supplementary Table S3, aliquoted, and stored at −80°C [26]. The 5 μl CFE reactions were mixed as described in Supplementary Table S3. For freeze-dried assays, the reactions were pipetted into tubes or dispensed (Echo 650, Beckman Coulter) into 96-well v-bottom plates (Corning), frozen at −80°C for 30 min, and lyophilized overnight in a Labconco freeze-dryer. The dried reactions were rehydrated with the specified matrix. Reactions were read in a 96-well plate on a BioTek Synergy H1 reader at 30°C (Ex: 485 nm, Em: 528 nm) every 5 min. End-point fluorescence after 1000 min was plotted for all figures, except for Supplementary Fig. S3, which used the first 250 min of data. Fluorescence values were normalized using a fluorescein calibration curve [8].

Saliva pretreatments

For heat treatments of pooled human saliva (Medix Biochemica), 1 ml aliquots were heated at 95°C for 10 min and cooled to room temperature. For syringe filtration, saliva was passed a 2 μm filter (Thermo Fisher). For centrifugal filtration assays, 1 ml saliva aliquots were centrifuged in Amicon Filter Units (3, 10, 30, or 100 kDa) at 3000 g for 30 min at 4°C. For the cortisol sensor assays, 2 ml saliva aliquots (with or without 0.5 mM hydrocortisone (Sigma) were centrifuged in 3 kDa filter units at 3000 g for 60 min at 4°C. The flow-through from all filtration methods was used to rehydrate the CFE reactions.

Results and discussion

Several approaches to mitigate the inhibitory effects of saliva on CFE reactions were examined. To investigate the use of nuclease inhibitors, E. coli extract was prepared from cells overexpressing mRI following established protocols [17], thereby creating extract pre-enriched with the inhibitor protein. The presence of mRI in the resulting extract was confirmed with an ELISA assay, demonstrating that the endogenous mRI concentration in the cell-free reaction was comparable to that of standard extract supplemented with 1 U/μl commercial mRI (Supplemental Methods, Supplementary Fig. S1). CFE reactions were then prepared with either the mRI-enriched extract or with standard extract supplemented with 0, 1, or 2 U/μl commercial mRI. A plasmid that constitutively expresses super-folder green fluorescent protein (sfGFP) was used as a readout for CFE reaction performance. The reactions were prepared, freeze-dried, and rehydrated with nuclease-free water or pooled human saliva that was either untreated, passed through a 0.2 μm syringe filter, or heated at 95°C for 10 min (Fig. 1A). Reactions with mRI-enriched extract produced as much, or more, sfGFP fluorescence than reactions with standard extract across all conditions (Fig. 1A). These results support the use of mRI-enriched extracts to help mitigate the deleterious effects of saliva on CFE systems. Conversely, the effect of adding 1 or 2 U/μl of commercial mRI varied depending on the rehydration matrix; in comparison to reactions without any mRI, the performance improved with untreated or syringe-filtered saliva, while the performance decreased with nuclease-free water or heat-treated saliva (Fig. 1A). These results were consistent with reports from the literature that commercial mRI can negatively impact CFE reaction performance, possibility due to glycerol or other incompatible components in the mRI buffer [17, 21]. Pretreatment of the saliva by heating or syringe-filtering improved CFE reaction performance compared to untreated saliva (Fig. 1A). However, neither of these treatments fully restored the sfGFP production to the levels achieved by reactions rehydrated with nuclease-free water. Therefore, we investigated the use of centrifugal filters to remove inhibitory factors from the saliva. CFE reactions were prepared with either mRI-enriched extract or with standard extract combined with 0 or 1 U/μl commercial mRI. The reactions were prepared, freeze-dried, and rehydrated with nuclease-free water or saliva that was either untreated, or passed through a 3, 10, 30, or 100 kDa molecular weight cutoff (MWCO) centrifugal filter (Fig. 1B). The combination of using a low MWCO filter (3 or 10 kDa) combined with the use of mRI-enriched extract resulted in highly functional reactions that produced similar amounts of sfGFP signal to those rehydrated in nuclease-free water (Fig. 1B). To confirm that the optimal mRI-enriched extract protocol was being used, three variant extracts were generated, each excluding different preparation steps, and their performances were evaluated in the presence of nuclease-free water and untreated saliva (Supplementary Fig. S2). These results showed that preparing mRI-enriched lysate with both post-lysis steps (run-off and dialysis) was needed to produce the strongest signal in the presence of untreated saliva. Combined, these results highlight the successful mitigation of saliva's inhibitory effects on CFE reactions through a combined approach of enriching cell extract with mRI and employing low MWCO filtration of saliva samples, ultimately yielding robust sfGFP production comparable to reactions rehydrated with nuclease-free water. While low-MWCO centrifugal filters are effective for restoring CFE compatibility, they are not always ideal for point-of-care settings where centrifuges may not be available. Future efforts to evaluate alternative saliva-clarification methods that achieve similar inhibitor removal without requiring benchtop equipment, such as passive dialysis or tangential-flow filtration, could be beneficial.

Figure 1.

Mitigating saliva inhibition in CFE reactions with mRI and pretreatments. CFE reactions were prepared with a sfGFP-expressing template and either standard extract with commercial mRI added or mRI-enriched extract. Freeze-dried reactions were rehydrated with A) water or saliva (untreated, heat-treated, or syringe-filtered), or B) water or saliva pretreated with 3, 10, 30, or 100 kDa centrifugal filters. Data show averages (n = 3–4), standard deviation, and individual replicates.

The effectiveness of using mRI and pretreatment strategies to preserve the functionality of a CFE sensor in the presence of human saliva was investigated. For this, a sensor that detects cortisol was chosen, since cortisol is an important stress-related biomarker associated with the metabolic, cognitive, and physiological status of an individual [22]. The selected sensor uses a LysR-type transcriptional regulator (LTTR) from Clostridium scindens to detect cortisol [23]. The sensor plasmid, pLysRsfGFP, which has the LTTR controlling the expression of sfGFP, was originally designed for whole cell-based assays but was used directly without modification in CFE reactions at a 10 nM concentration (Fig. 2A). When fresh CFE reactions were implemented in nuclease-free water, an ~20 times improvement in sensitivity was observed (EC₅₀ = 0.39 μM) (Fig. 2B), compared to whole cell-based studies performed previously using the same plasmid sequence and an E. coli Nissle 1917 TolC knock out strain (EC₅₀ = 8.9 μM) [23] (Supplementary Table S1). Improvements in sensitivity after transferring sensors from whole cells to CFE systems have previously been observed in other instances [2, 12] and could be due to the lack of a cell membrane in CFE reactions. The function of the sensor in the presence of saliva pretreated with centrifugal filtration was next investigated. CFE reactions with the cortisol sensor were prepared using mRI-enriched extract, freeze-dried, and then rehydrated with either nuclease-free water or saliva pretreated with centrifugal filtration (3 kDa MWCO) (Fig. 2C). Cortisol was spiked into the saliva either prior to filtration or after filtration to determine if this biomarker was preserved within the sample during the filtration process. The process of lyophilization decreased the overall performance of the sensor in terms of total fluorescence output and sensitivity, with the EC₅₀ ranging from about 0.8–1.6 μM for the lyophilized conditions (Supplementary Table S1). Previous studies have shown that freeze-drying CFE reactions can lead to a drop in protein production and sensor performance, with some studies identifying strategies to improve the function of freeze-dried CFE reactions using lyoprotectants or optimization of the rehydration volumes [4, 24]. Salivary cortisol typically ranges from ~ 1–100 nM [27–30], substantially lower than the sensor’s EC₅₀ values (0.39–1.6 μM), underscoring that this assay demonstrates saliva compatibility rather than clinical sensitivity and would require further engineering for diagnostic use. The CFE cortisol sensor could potentially benefit from additional efforts to mitigate the impact of freeze-drying on its performance. While freeze-drying negatively affected the CFE cortisol sensor, the use of filtered saliva as the rehydration matrix did not lead to a clear decrease in performance when compared to using nuclease-free water for rehydration (Fig. 2C and Supplementary Fig. S3). These results indicate that the combination of preparing mRI-enriched extract and pretreating saliva using centrifugal filtration with a low MWCO can be a successful strategy to enable the function of CFE-based sensors, and detection of target analytes, within complex biofluids. In practical diagnostics, water cannot substitute for saliva because the sensor must be rehydrated with the patient’s sample to measure endogenous cortisol. Filtration is therefore essential to remove inhibitory components, establishing performance parity with water as the benchmark for saliva-based CFE assays. While more work is needed to determine how broadly applicable these mitigation strategies are for different CFE sensors and their target analytes, these results provide key insights into strategies for developing more reliable CFE-based diagnostics that use saliva as a non-invasive sample matrix.

Figure 2.

Performance of a CFE-based cortisol sensor in the presence of saliva. A) Sensor circuit design illustrates C. scindens LTTR regulating sfGFP expression in response to cortisol. B) CFE reactions were prepared fresh with 10 nM cortisol sensor plasmid and mRI-enriched extract. C) Reactions were freeze-dried and rehydrated with water or 3 kDa centrifugal-filtered saliva. Cortisol was added to saliva pre/post-filtration at varying concentrations. Data show averages (n = 3), standard deviation, and three-parameter logistic trendlines.

Conflict of interest

No potential conflict of interest was reported by the authors.

Funding

This work was funded by the Defense Advanced Research Projects Agency’s (DARPA’s) Smart Non-invasive Assays of Physiology (SNAP) program and the U.S. Air Force Research Laboratory (AFRL) 711th Human Performance Wing’s Human Effectiveness Directorate (RH).

Data availability

Data available as a supplementary file at SYNBIO online.

Material availability statement

All materials are subject to a Materials Transfer Agreement.

Disclaimers

The views expressed are those of the authors and do not reflect the official guidance or position of the United States Government, the Department of Defense, the United States Air Force, or the United States Space Force. Unless otherwise noted, imagery in this document is property of the U.S. Air Force. Distribution Statement A: Approved for public release; distribution is unlimited. Cleared on 18 Jul 2025, Case Number: AFRL-2025-3 529.