Main

Biomarkers, defined as measurable indicators of disease, are valuable tools for patient screening, prognosis, informing treatment and treatment monitoring. The detection of biomarkers typically relies on functional nanoprobes (for example, labelled antibody or peptide) due to their high sensitivity and specificity1. Ideal functional nanoprobes should have high binding affinity to the target, cost-effective production, scalability and flexible storage to benefit population-based patient screening, which is particularly important for the rapid detection of emerging diseases like COVID-19 (refs. 2,3). The controlled and precise attachment of the target binder (for example, antibody) to material surfaces in optimal densities and orientations to maximize the binding affinity, multivalency and target-specific recognition are also crucial for improved diagnostic outcomes4,5,6. Although various strategies are available for attaching biomolecules to materials, these methods are challenging because of problems with conjugation efficiency, site specificity and inhibited target binding. As such, it is important to develop better approaches to overcome such challenges for effective disease detection.

Here we developed a class of bioinspired multifunctional nanoprobes based on yeast-cell-derived synthetic bionanofragments (SynBioNFs) for the rapid detection of emerging diseases. SynBioNFs were genetically programmed to display multifunctional components including specific target binder (nanobody) and generic surface binder (adhesive peptide) anchored on nanosized yeast cell wall fragments through the endogenous yeast cell wall protein Aga2p. We demonstrated the universal nature of SynBioNFs through their utility in SARS-CoV-2 (SCV2) virion detection on various detection platforms, including surface-enhanced Raman scattering (SERS), fluorescence, electrochemical and colorimetric lateral flow assay (LFA) systems.

SynBioNFs as multifunctional nanoprobes

To allow highly efficient and effective biomarker detection, we genetically engineered SynBioNFs to express multifunctional protein or peptide components on the cell wall via a yeast surface display system, followed by simple fragmentation and separation of cell wall fragments into nanosized particles (Fig. 1a). As a proof-of-concept demonstration of their use in disease detection, SynBioNFs displaying anti-SCV2 receptor-binding-domain (RBD) nanobodies (W25 (ref. 7), Sb68 (ref. 8) and Nb21 (ref. 6)) were applied on different diagnostic platforms for the detection of SCV2 (Fig. 1b), including SERS, fluorescence, electrochemical and LFA systems. Nanobodies, the smallest antigen-binding domain, were selected as the recognition module due to their high binding affinity, stability and suitability for large-scale production9,10. MatterTags11, SpyTags4,9 and StrepTags12 were used for the tethering of SynBioNFs to various surfaces used in the diagnostic platforms. Specifically, MatterTag—a type of adhesion-promoting peptide—is able to attach to various surfaces (for example, gold, silicon and plastic) via both hydrophobic and polar interactions11. SpyTag enables a spontaneous covalent conjugation with SpyCatcher for surface immobilization, which is a rapid self-catalysed assembly process4,9. StrepTag allows the specific interaction with streptavidin for an easy surface coupling with commercially available streptavidin-functionalized particles. The use of a defined biological immobilization strategy instead of non-specific chemistry approach preserved the binding function as well as ensured the precisely controlled orientation of the nanobody.

Fig. 1: Design of SynBioNFs as a reagent for disease detection.
figure 1

a,b, Generation (a) and universal application (b) of SynBioNFs in the detection of SCV2 on diagnostic platforms including SERS, fluorescence, electrochemical and naked-eye lateral flow platforms. The following capture and/or detection SynBioNFs pairings were used: capture SynBioNFs (W25–MatterTag) and detection SynBioNFs (Sb68–SpyTag) on the SERS platform; capture SynBioNFs (Sb68–MatterTag) and detection SynBioNFs (Nb21–StrepTag) on the fluorescent platform; capture SynBioNFs (Sb68–MatterTag) on gold electrodes for electrochemical detection; and SynBioNFs (Nb21–SpyTag) conjugated with SpyCatcher-coated spherical gold particles used as the detection reagents for LFA.

Full size image

Compared with commercially available antibodies in biomarker detection, SynBioNFs featured a higher stability under the conditions of high temperature (Supplementary Fig. 1), low pH and strong detergents (Supplementary Fig. 2a–c), as well as sufficient stability in the presence of high concentrations of salt (Supplementary Fig. 2d), which largely resulted from the stabilization of functional proteins by the abundant hydroxyl groups of polysaccharides in the yeast cell wall13,14. Furthermore, SynBioNFs demonstrated higher avidity towards targets due to the capability of loading multiple copies of functional proteins by the nanosized yeast cell wall fragments (Supplementary Fig. 3). Additionally, SynBioNFs demonstrated a much weaker disassociation than the corresponding nanobody (Supplementary Fig. 4). Thus, SynBioNFs with high stability and high avidity can undoubtedly benefit the widespread and highly sensitive biomarker detection, especially for pandemic disease screening.

Generation and characterization of SynBioNFs

We employed yeast surface display to project the desired nanobodies on the cell wall due to several advantages, including cost-effective growth media, robustness in the fermentor and availability of well-established genetic tools to precisely edit the yeast cell15. A panel of nanobodies targeting distinct epitopes of SCV2 RBD (W25, Sb68 and Nb21) were bioengineered on the N terminal of the yeast surface protein Aga2p, connected via a 15 amino acid flexible GGGGS linker. To enable the attachment of SynBioNFs to material surfaces, surface-binding peptides including either MatterTag, SpyTag or StrepTag were added to the C terminus of the Aga2p protein through a 30 amino acid flexible GGGGS linker. The flexible GGGGS linkers added on either side of the Aga2p anchor protein were used to minimize steric hindrances between the nanobody and functional peptide fusions. Haemagglutinin (HA) and c-Myc epitope tags were also engineered at the N and C termini to allow the detection of the displayed fusion proteins.

Flow cytometry was utilized to determine nanobody display on the yeast cell surface and nanobody binding to the SCV2 RBD protein. Specifically, yeast cells induced to display nanobodies were labelled with the fluorescent anti-myc-Dy650 antibody to detect nanobody cell surface expression. The cells were also labelled with fluorescent SCV2 RBD–mNeonGreen to determine the nanobody-binding affinity towards SCV2 RBD. As two representative examples, yeast cells encoding W25–MatterTag and Sb68–SpyTag showed the expression of nanobodies on over 60% of yeast cells (Fig. 2a,b,d,e), suggesting a high level of protein expression in the commonly employed plasmid-based approach16,17,18,19. Furthermore, the Sb68–SpyTag fusion protein on the yeast cells revealed a binding affinity to the SCV2 RBD (kD = 42 nM) similar to the soluble Sb68 nanobody (kD = 37 nM)8 (Supplementary Fig. 5), confirming that the fusion design and surface display retained the complete function of the nanobody.

Fig. 2: Profiling of fusion protein expression on yeast cell surfaces and size characterization of the corresponding SynBioNFs.
figure 2

a,b,d,e, Flow cytometry analysis of yeast cell (a and d) controls and positives with W25–MatterTag (b) and Sb68–SpyTag (e) expressions. c,f, Nanoflow cytometry measurement of the particle size distribution of SynBioNFs (W25–MatterTag) (c) and SynBioNFs (Sb68–SpyTag) (f).

Source data

Full size image

SynBioNFs carrying W25–MatterTag and Sb68–SpyTag were generated through mechanically fragmenting the yeast cells via sonication that had no detrimental effect on the protein activity (Supplementary Fig. 6). Using the filters with 100 nm cut-off to further purify the products, the obtained SynBioNFs possessed a mode size of approximately 70 nm as measured by nanoflow cytometry (Fig. 2c,f).

SynBioNFs as capture nanoprobes

Gold surfaces provide a common interface to accommodate capture probes in biomarker detection20. We demonstrated SynBioNF functionalization on gold surfaces in one-step incubation using MatterTag-assisted immobilization, which allows a biological functionalization on gold surfaces for site-specific anchoring, and potentially enhancing SynBioNF orientation on the surface for optimal target recognition.

We immobilized SynBioNFs (W25–MatterTag) on gold microelectrode surfaces to act as capture nanoprobes (Fig. 3a). Before the immobilization, we utilized oxygen plasma cleaning on the gold microelectrodes to remove potential contamination21, which produced a similar performance to traditional electrochemical cleaning22,23, as demonstrated by differential pulse voltammetry (DPV) and cyclic voltammetry (CV) results (Supplementary Fig. 7). To confirm the MatterTag-assisted immobilization of SynBioNFs, we monitored the redox reaction of [Fe(CN)6]3–/4– on gold microelectrodes through Faradaic electrochemical impedance spectroscopy, DPV and CV. As shown by the Nyquist plot, an increase in the surface impedance is observed following the addition of SynBioNFs on the gold microelectrodes due to the attenuated electron transfer (Fig. 3b). In agreement with electrochemical impedance spectroscopy, DPV also demonstrated a substantial reduction in peak current following the incubation of gold surface with SynBioNFs (Fig. 3c). Furthermore, the corresponding CV data indicated the characteristic redox peak of our electrochemical biosensors (Fig. 3d). Additionally, we evaluated the role of MatterTag in assisting the binding of capture SynBioNFs on gold microelectrodes using our SERS platform, which indicated an enhanced capture of the SCV2 RBD target through MatterTag for a higher SERS signal output (Fig. 3e,f). The SERS signal resulting from the identification of SCV2 RBD by detection SynBioNFs conjugated to plasmonic SERS nanoboxes is explained in the following sections.

Fig. 3: Evaluation of MatterTag-assisted SynBioNF immobilization on gold microelectrodes.
figure 3

a, Schematic of the attachment of SynBioNFs to gold via MatterTag. b–d, Electrochemical study of gold microelectrode versus gold microelectrode with SynBioNFs (W25–MatterTag), showing the Nyquist plots (b), DPV responses (c) and CV characterization (d). e,f, SERS result of the representative Raman spectra (e) and Raman intensities (f) at 1,603 cm–1 of SCV2 RBD detection by coating gold microelectrodes using SynBioNFs infused with MatterTag (red) or without MatterTag (blue) (p = 0.0010, two-tailed t-test). Data are presented as mean values ± standard deviation from three independent experiments.

Source data

Full size image

We also determined the required time of MatterTag-assisted SynBioNF immobilization on gold surfaces by measuring the percentage peak current changes in the DPV responses. Supplementary Fig. 8 reveals the complete binding of SynBioNFs on gold surfaces within a 10 min reaction, which was in line with the MatterTag-assisted soluble protein immobilization on material surfaces11.

SynBioNFs as detection nanoprobes

The conjugation of SynBioNFs with plasmonic nanostructures generated detection nanoprobes for reading highly specific and sensitive signals. Anisotropic gold–silver alloy nanoboxes with strong Raman enhancement capability were selected as representative nanostructures to conjugate with SynBioNFs. Exploiting the SpyTag and SpyCatcher systems, SynBioNFs (Sb68–SpyTag) were incubated with SpyCatcher-coated nanoboxes for a simple self-assembly (Fig. 4a). After incubation with SynBioNFs, the nanoboxes with an average edge length of around 80 nm (Fig. 4b) were surrounded by a layer of low-contrast material (Supplementary Fig. 9), which indicated the conjugation of SynBioNFs on nanobox surfaces. Crucially, the use of SpyTag and SpyCatcher as a simple and modular approach for bioconjugation provided a potentially universal and useful strategy for biofunctionalization.

Fig. 4: Demonstration of SpyTag and SpyCatcher systems in conjugating SynBioNFs with gold–silver alloy nanoboxes as detection nanoprobes.
figure 4

a, Schematic of SynBioNFs conjugated with nanoboxes for virus detection by identifying RBD on the spike protein. b, Scanning electron microscopy image of the nanoboxes, showing the successful preparation of nanoparticles with the desired box-like morphology. Data from one independent experiment. c,d, Histogram (c) and mean fluorescence intensities (MFIs) (d) of bioconjugates measured in flow cytometry. Bioconjugates 1–3 were prepared with SynBioNFs (SpyTag–Sb68) and nanoboxes (coated with SpyCatcher), SynBioNFs (Sb68) and nanoboxes (coated with SpyCatcher) and SynBioNFs (SpyTag–Sb68) and nanoboxes, respectively, in which the (+) samples were reacted with RBD–mNeonGreen as positives and the (–) ones were incubated with PBS as controls. Data are presented as mean values from two independent experiments.

Source data

Full size image

To demonstrate the crucial role of SpyTag and SpyCatcher in building the bioconjugates, we prepared different bioconjugates and reacted with RBD–mNeonGreen for fluorescence-based nanoflow cytometry profiling. For bioconjugate 1 prepared in the presence of SynBioNFs (SpyTag–Sb68) and nanoboxes (coated with SpyCatcher), the fluorescence intensities were substantially higher than the corresponding control (Fig. 4c). In contrast, bioconjugates 2 and 3 using nanoboxes without SpyCatcher or SynBioNFs without SpyTag showed remarkably low fluorescence intensities like their controls. The mean fluorescence intensity further quantitatively showed much higher intensities for bioconjugate 1 that had both SpyTag and SpyCatcher present (Fig. 4d). Such a result clearly supported the necessity of the SpyTag and SpyCatcher systems in connecting detection SynBioNFs and nanoboxes. These data also illustrated that the binding components (that is, nanobody) on SynBioNFs were highly functional even after conjugation with nanoboxes.

As demonstrated in previous reports, the SpyTag and SpyCatcher systems allowed a prompt autocatalytic reaction under physiological conditions and rapidly formed a covalent isopeptide bond4,9. In our study, sodium dodecyl sulfate–polyacrylamide gel electrophoresis indeed revealed the formation of a peptide bond in ~1 min (Supplementary Fig. 10) using the purified SpyTag and SpyCatcher soluble proteins. Similarly, the preparation of bioconjugates of SynBioNFs (Sb68–SpyTag) and nanoboxes (coated with SpyCatcher) demonstrated rapid assembly within 10 min, as suggested by the nanoflow cytometry measurement (Supplementary Fig. 11).

SynBioNF-enabled SCV2 detection on SERS platform

Having engineered SynBioNFs that can act as capture and detection nanoprobes, we showcased their application on the SERS platform for SCV2 detection (Fig. 5a–e). First, capture SynBioNFs (W25–MatterTag) were applied on gold microelectrodes by using the W25 nanobody to capture SCV2 by binding the class-2 epitope site of RBD on the spike protein7. Subsequently, detection SynBioNFs (Sb68–SpyTag) conjugated with nanoboxes (coated with SpyCatcher and Raman reporters) were introduced to achieve virus identification and highly sensitive signal readout. Unlike the W25 nanobody, Sb68 nanobody recognizes a class-4 epitope region of RBD and has high affinity towards RBD on the spike protein (kD = 37 nM)8. As the RBD amounts positively correlated with the viral loads, the SCV2 concentration can be inferred from the Raman reporter (that is, 2,7-mercapto-4-methylcoumarin (MMC)) intensity at 1,603 cm−1. To further intensify the Raman readout for a highly sensitive and specific SCV2 detection, we integrated an alternating-current electrohydrodynamic-induced nanoscopic fluid flow (that is, nanomixing)24 to improve the molecular collision of the sample and detection SynBioNFs with the gold microelectrodes (Supplementary Fig. 12).

Fig. 5: Schematic and specific detection of RBD and SCV2 by using SynBioNFs as capture and detection nanoprobes on the SERS platform.
figure 5

a, Schematic of the workflow of SCV2 detection. b, Immobilization of capturing SynBioNFs (W25–MatterTag) on the gold surface and the capture of virus from the sample. c, SynBioNFs using MatterTag to bind gold atoms and W25 nanobody to capture the RBD on the spike protein. d, Identification of the captured virus by detection SynBioNFs (Sb68–SpyTag) conjugated with nanoboxes. e, SynBioNFs using SpyTag to conjugate SpyCatcher-coated gold–silver alloy surface and Sb68 nanobody to detect the RBD on the spike protein. f,g, SERS mapping images (f) and average Raman spectra (g) in the presence of positive samples and negative controls. The red colour in f indicates the presence of RBD/SCV2 on the gold microelectrodes, which clearly differentiated the positive samples from the negative controls.

Source data

Full size image

Highly accurate identification of targets among a variety of biomolecules, which may be present in the samples, is crucial for a reliable diagnosis20. To demonstrate the specificity of the engineered SynBioNFs in SCV2 detection, we challenged the designed SERS platform by analysing a range of samples, including recombinant SCV2 RBD protein, gamma-irradiated SCV2 virions, phosphate-buffered saline (PBS), cell culture medium, SCV2 S2 protein, recombinant influenza HA antigen, recombinant dengue virus 2 NS1 antigen and recombinant SARS-CoV-1 RBD protein. As shown in Fig. 5f,g, the positive samples displayed clear Raman signals due to the correct identification of the captured targets, whereas the negative controls showed negligible Raman signals without successful recognition. Therefore, this result illustrated the high specificity of SynBioNFs in reacting with RBD on SCV2.

The ability to detect a low concentration of SCV is critical for the timely control of disease spreading. We, therefore, examined the sensitivity of SynBioNF-enabled SCV2 detection by measuring the Raman signals from a titration of SCV2 gamma-irradiated virus culture (that is, 0, 10, 100, 1,000 and 10,000 virions µl–1). In all the experiments, virus measurement refers to the number of copies of virus RNA to enable a comparison with reverse-transcription quantitative polymerase chain reaction (RT-qPCR)—the gold standard for SCV2 detection. As shown in Supplementary Fig. 13, an increase in the Raman signal was observed with the use of a higher viral concentration and consequently resulted in a good linear relationship curve (R2 = 0.99) for quantification. The limit of detection of SCV2 detection was calculated to be 5.4 virions µl–1 based on three times the standard deviation of blank signals, which was comparable with the RT-qPCR result (~1 copies µl–1)25 and avoided the potential bias arising from non-specific amplification events. In particular, the SynBioNF-enabled SERS platform allowed the differentiation of simulated patients and healthy controls by titrating a low concentration of gamma-irradiated SCV2 into human saliva (that is, 0, 10 and 100 virions µl–1) (Supplementary Fig. 14), suggesting the potential use in clinical settings26.

For clinical relevance, we utilized the SynBioNF-enabled SERS platform to screen a panel of 134 nasopharyngeal swab samples from patients (84 SCV2 positives and 50 negatives, as confirmed by RT-qPCR). The representative Raman images and the acquired Raman intensities at 1,603 cm−1 of all the clinical samples clearly showed identifiable signals for positive samples but negligible signals for negative ones (Fig. 6a,b). Based on the cut-off value of 94 to classify patients and correlate with RT-qPCR diagnosis, SynBioNFs enabled the identification of 81 positives and 50 negatives (Fig. 6c), which correspond to the sensitivity, specificity and accuracy of 96.43%, 100.00% and 97.76%, respectively. Furthermore, this detection achieved a good receiver operating characteristic graph with an area under the curve of 0.9984 (Fig. 6d). Comparing the normalized Raman intensities with cycle threshold (Ct) values of positive samples, a Bland–Altman analysis suggested no significant difference between SynBioNF-enabled and RT-qPCR detections as the majority of the data fit in the 95% limits of agreement (Fig. 6e). In a selected number of samples near the limit of detection of the SynBioNF-enabled SERS platform, we noticed a variation in Raman intensities of positive samples with RT-qPCR tests (Supplementary Fig. 15), which was probably caused by the different working principles of these two approaches and was also reported in a previous finding9. Additionally, SynBioNFs enabled the detection of different SCV2 variants (that is, Delta and Omicron) (Fig. 6f), which was beneficial for accurate SCV2 screening.

Fig. 6: SynBioNF-enabled SERS screening of clinical samples for COVID-19 patient screening.
figure 6

a, Representative Raman images of identified three positive (P1–P3) and three negative (N1–N3) samples. The red colour indicates the presence of SCV2 on the gold microelectrodes, which clearly differentiates the positive samples from the negative controls. Data from three independent experiments. b, Raman intensities of all the clinical samples, with the horizontal lines showing the mean values (n = 134 biologically independent samples) (p < 0.0001, two-tailed t-test). c, Classification result of SynBioNF-based SCV2 detection with the use of the cut-off value (that is, 94) and correlation with RT-qPCR diagnosis. d, Receiver operating characteristic curve of patient versus healthy detection. AUC, area under the curve. e, Bland–Altman plot of SynBioNF-enabled and RT-qPCR detection. f, Raman intensities for the detection of SCV2 Delta variant, Omicron variant and healthy clinical sample controls, with the horizontal lines showing the mean values (n = 21 biologically independent samples) (p < 0.0001, two-tailed t-test). g,h, Detection of healthy samples spiked with different concentrations of SCV2 virions showing Raman intensities with Ct values labelled on the top of the columns (g) and the corresponding commercially available LFA photographs (h), in which LFA failed to generate visible signals on the use of SCV2 concentration lower than 2 × 104 virions ml–1.

Source data

Full size image

Furthermore, we compared the SynBioNF-enabled SERS platform and commercially available LFA by testing several clinical samples with known Ct values (Supplementary Fig. 16), in which the former successfully detected all the positive samples, whereas the latter failed to find three positive samples. Similarly, both SynBioNF-enabled SERS platform and LFA did not generate very consistent signals with the Ct values in RT-qPCR based on the different working principles. Importantly, the SynBioNF-enabled SERS platform generated concentration-dependent results in the measurement of healthy samples spiked with different concentrations of SCV2 virions (Fig. 6g). The LFA also produced a similar concentration-dependent trend (Fig. 6h), but the detection failed on the use of SCV2 concentration lower than 2 × 104 virions ml–1. Taken together, these comparisons illustrated the capability of SynBioNFs in the highly sensitive screening of COVID-19 patients.

SynBioNF-enabled SCV2 detection on other platforms

To further demonstrate the diverse utility of this nanoprobe for disease detection, we explored its application of SCV2 detection on three widely used platforms, including fluorescence, electrochemical and visual colorimetric LFA systems.

To enable a convenient fluorescence readout, we fused a StrepTag on SynBioNFs to conjugate with commercially available streptavidin fluorescent beads for detection. Specifically, following the capture of SCV2 by SynBioNFs (Sb68–MatterTag) on gold surfaces, SynBioNFs (Nb21–StrepTag) bioconjugated with fluorescent beads detected RBD through fluorescence imaging. As shown in Extended Data Fig. 1a, the presence of SCV2 generated clear fluorescence signals compared with the negligible signals from the medium control, which suggested the possibility of SynBioNFs on fluorescence diagnostic platforms.

For portable electrochemical sensors, SynBioNFs enabled a simple and rapid label-free DPV target detection. To illustrate this, SynBioNFs (Sb68–MatterTag) on disposable screen-printed electrodes were used to capture SCV2, which resulted in increased surface impedance and reduced peak current signals for quantification. In Extended Data Fig. 1b,c, the current decreased, and therefore, a higher percentage of peak current change was observed for SCV2 but not the medium control. The shifted and broadened DPV peak after SCV2 capture was probably triggered by the formation of a biomolecule layer that decreased the effective current density on the electrodes27,28,29. The corresponding CV result further indicated the successful detection of SCV2 (Extended Data Fig. 1d) with the appearance of characteristic redox peaks.

For the LFA, SynBioNFs can be conjugated with gold nanoparticles to visually identify target binding through the specific colour. In this regard, we linked SynBioNFs (Nb21–SpyTag) and gold nanoparticles (coated with SpyCatcher) as bioconjugates through the SpyTag/SpyCatcher system. After capturing SCV2 on the paper-based analytical device, the bioconjugates can recognize the SCV2 RBD for red colour readout. The apparent red colour generated by SCV2 and no colour from the medium negative control (Extended Data Fig. 1e) in the test lines indicated the feasibility of SynBioNFs on LFA devices. Furthermore, SynBioNF-enabled LFA showed clear differences between one SCV2 Omicron sample and one healthy clinical sample (Supplementary Fig. 17).

Conclusions

SynBioNFs represent the next generation of nanoprobes, which are cost-effective and simple-to-manufacture reagents featuring many key advantages for biomarker detection compared with traditional probes on the market. These include (1) the genetically ‘plug and play’ programmable display of an array of potent nanobodies on nanosized yeast cell walls for potential high-avidity ‘molecular Velcro-like’ target binding; and (2) the integration of functional tags for simple, controlled and site-specific nanobody immobilization on flat or micro-/nanoparticle surfaces. These advantages enabled SynBioNFs as convenient nano-reagents to apply on different platforms for biomarker detection. We have demonstrated streamlined and effective nanoprobe functionalization on the surfaces of flat (that is, gold microelectrodes) and micro-/nanoparticles (that is, gold–silver alloy nanoboxes, fluorescent beads and spherical gold nanoparticles). All these properties of SynBioNFs synergistically benefit highly effective biomarker detection. More detailed discussions on the utility and advantages of SynBioNFs are provided in Supplementary Information.

Methods

Chemicals and materials

Hydrogen tetrachloroaurate trihydrate, silver nitrate, dithiobis (succinimidyl propionate), 11-mercaptoundecanoic acid (MUA) and MMC were purchased from Sigma-Aldrich. Analytical-grade ascorbic acid was obtained from MP Biomedicals. 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride and N-hydroxysulfosuccinimide were obtained from Thermo Fisher Scientific. A Milli-Q water system was used to generate ultrapure water (18.2 MΩ cm) to synthesize the nanoparticles. Two commercially available LFA products purchased from Abbott (ARTG #345192) and SD Biosensor (ARTG #345219) were used to test the clinical samples by following the instructions from the providers.

Production of SpyCatcher, SpyTag and SCV2 RBD–mNeonGreen fusion proteins

To produce SpyCatcher and SpyTag, SpyCatcher–MatterTag and EGFP-SpyTag fusions were designed via a rigid linker (AEAAAKEAAAKEAAAKA) and flexible linker (GGGS), respectively. The chimeric genes were commercially synthesized and cloned into the pCDNA3.1 vector. The fusion proteins were expressed in ExpiCHO-S cells according to the manufacturer’s instructions (Thermo Fisher Scientific). For purification, the filtered supernatant was loaded on a Strep-Tactin 4Flow column (IBA Lifesciences) equilibrated with a purification buffer (100 mM Tris–HCl (pH 8.0), 150 mM NaCl). The column was then washed with 50 column volumes of purification buffer. The fusion protein was eluted by a purification buffer containing 50 mM biotin (IBA Lifesciences). The purities of the purified proteins were analysed in sodium dodecyl sulfate–polyacrylamide gel electrophoresis. The expression and purification of SCV2–mNeonGreen fusion protein is described elsewhere2.

Engineering yeast cells and production of multifunctional SynBioNFs

EBY100 yeast cells were employed to display eight different functional fusion proteins on the cell walls, including (1) W25–MatterTag, (2) Sb68–SpyTag, (3) Sb68–MatterTag, (4) Nb21–StrepTag, (5) Nb21–SpyTag, (6) Nb21–MatterTag, (7) DD7–MatterTag and (8) DD5–SpyTag. These fusion gene sequences were commercially synthesized and cloned (Gene Universal; Supplementary Data 1 and 2) into the pCTCON2 expression plasmid for yeast surface display by following previous work2. In our design, the target-binding nanobodies (Sb68, W25 and Nb21) and functional peptide tags (MatterTag, SpyTag and StrepTag) were fused into the N and C termini of Aga2p, respectively. A flexible linker (GGGGS) with 15 and 30 amino acids was used at the N and C termini of Aga2p to avoid steric hindrances between the fused proteins/peptides. Furthermore, HA and c-Myc peptide tags that allow the quantification of the displayed fusion proteins were included at the N and C termini of each gene construct, respectively. The fusion proteins carrying Aga2p were able to immobilize on the yeast cell walls by interacting with the Aga1p anchor protein.

To achieve the fusion protein display, EBY100 yeast cells were incubated with recombinant DNA (10 µl, 1,000 ng) under the stimulation of a square wave using electroporation. The produced yeast cells were cultured in the SDCAA medium and monitored until the optical density at 600 nm (OD600) reached 5–10. The yeast cells were then transferred into galactose containing the SGCAA medium and diluted to OD600 of 1.0 to induce fusion protein expression. Following the culture for 48 h, the yeast cells were collected and confirmed the fusion protein expression by performing a flow cytometry analysis, as described below.

For the preparation of SynBioNFs, the yeast cells with the display of fusion proteins were collected from the SGCAA medium (50 ml, OD600 of 6–10) and washed with PBS through centrifugation (2,000×g, 10 min), followed by resuspending into PBS supplemented with a protease inhibitor cocktail with EDTA (10 ml per tablet). The mechanical fragmentation of yeast cells was conducted using a sonicator (Sonics ultrasonic processor VC-505) with a 3 mm tip diameter and 171 mm length at an ultrahigh intensity by repeating the following conditions for five times: 40% amplitude; 1 s ON and 1 s OFF pulse for 2 min. Ultimately, SynBioNFs were obtained using centrifugation (2,500×g, 15 min) to collect the supernatant products and purified through a filter unit (100 nm, Millipore).

Flow cytometry profiling of fusion proteins

The yeast cells with fusion protein expression (107 cells ml–1) were collected and washed using PBS containing 0.1% bovine serum albumin (BSA) (500 µl) through centrifugation (1,500×g, 4 min) at 4 °C. To enable the labelling of the yeast surface protein, the yeast cells were incubated with anti-Myc antibody labelled with DyLight 650 (1:100 dilution) and RBD–mNeonGreen, or anti-dengue NS1 protein followed by anti-His antibody labelled with PE (1:100 dilution) in PBS containing 0.1% BSA (100 µl) with rotation and away from light at 4 °C for 1 h. The labelled yeast cells were centrifuged (1,500×g, 4 min) and washed with PBS containing 0.1% BSA (500 µl), followed by resuspending them into PBS containing 0.1% BSA (500 µl) for testing. The yeast cell controls without the use of anti-Myc or anti-His, anti-dengue NS1 antibody and RBD–mNeonGreen were prepared with the same protocol. The obtained yeast cells were subject to flow cytometry profiling (CytoFLEX, Beckman Coulter) using two lasers (488 and 633 nm) and two band-pass filters (525/40 and 660/20 nm). The data were acquired using CytExpert (2.4.0.28) and analysed with FlowJo software (10.8.1).

SCV2 culturing using cell line

SCV2 was cultured in Vero E6 cells. The Vero E6 cells were first cultured in Dulbecco’s modified Eagle’s medium supplemented with 2% heat-inactivated foetal bovine serum. When the cells were 70–90% confluent, the viral inoculum was inoculated into the Vero E6 cells and incubated at 37 °C (5% CO2); the cytopathic effect was observed. SCV2 was harvested in the supernatant via centrifugation at 4,500×g for 10 min. Virus was gamma irradiated at a dose of 50 kGy to inactivate it.

RT-qPCR quantification of cultured SCV2

To quantify the cultured SCV2 stock, RT-qPCR was performed. MagMAX-96 viral RNA isolation kit was used to extract the SCV2 RNA. The gBlock synthetic E gene standards were utilized to establish the copy-number-related calibration curve. The test employed the AgPath-ID One-Step RT-PCR master mix with the following primers: CoV-E-fwd (5’-AGT ACG AAC TTA TGT ACT CAT TCG TT-3’), CoV-E-R2 (5’-ATA TTG CAG CAG TAC GCA CAC A-3’) and TaqMan probe (CoV E probe 5’-6-FAM-ACA CTA GCC ATC CTT ACT GCG CTT CG-MGB-3’). The detection of SCV2 was conducted in duplicates by using the mean for calibration on an Applied Biosystems instrument. The cycling conditions were 45 °C for 10 min and 95 °C for 10 min, followed by 45 cycles of 95 °C for 15 s and 60 °C for 45 s.

Conjugating detection SynBioNFs with gold–silver alloy nanoboxes

The conjugation of SynBioNFs with gold–silver alloy nanoboxes was performed via the SpyCatcher-/SpyTag-mediated self-assembly. SpyCatcher-coated gold–silver alloy nanoboxes were first prepared as follows: gold–silver alloy nanoboxes were synthesized following our previous work30. One millilitre of nanoboxes were centrifuged at 800×g for 15 min. Then, 10 µl Raman reporter (MMC) and 2 µl linker molecule (MUA) were incubated with the above nanoboxes for 5 h. After removing the free MMC and MUA by centrifuging at 800×g for 15 min, 10 µl of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (10 mM) and 20 µl of N-hydroxysulfosuccinimide (10 mM) were added to activate the carboxyl group on MUA. Then, 0.5 µg SpyCatcher was incubated with the nanoboxes for 30 min at room temperature. The SpyCatcher-coated nanoboxes were purified with centrifuging at 800×g for 15 min and resuspended into 200 µl of 0.1% BSA.

Next, 30 µl of the SpyCatcher-coated nanoboxes were incubated with 4 µl of SynBioNFs (Sb68–SpyTag) (8 µg µl–1) in 60 µl of 10 mM PB, 1 mM PBS and 1% BSA buffer for 10 min. The final products were collected by centrifuging at 600×g for 10 min and washing with 0.1% BSA.

NanoFCM characterization of SynBioNFs and nanobox bioconjugates

NanoFCM measurements were performed on a nanoFCM flow NanoAnalyser (NanoFCM). The NanoAnalyser was first calibrated for concentration and size using the standard nanoparticles provided by the company. The size distribution of SynBioNFs was obtained by comparing with the cocktail size standard (that is, premixed silica nanoparticles with different diameters). To profile the fluorescence profiling of gold–silver-alloy-nanobox-conjugated SynBioNFs against RBD–mNeonGreen, 30.0 µl of the conjugates were incubated with 0.5 µl of RBD–mNeonGreen (350 µM) at room temperature for 30 min and the products were washed three times with 0.1% BSA via centrifugation at 600×g for 10 min, followed by recording the events for 1 min. The same amount of gold–silver alloy nanobox and SynBioNFs without reacting with RBD–mNeonGreen were used as negative controls to set the threshold.

SynBioNF-enabled SERS detection of RBD, SCV2, simulated patient samples and clinical COVID-19 samples

The gold microelectrodes were prepared in-house by a photolithography approach with 4-inch borosilicate glass wafers and following a previously established protocol31. After photolithography, the wafer consisted of an array of 28 circular gold microelectrodes with inner working electrodes (1.00 mm in diameter) and outer counter electrode (0.12 mm in diameter). The working and counter electrodes were separated by 1 mm. To contain the sample on the gold microelectrodes, a well structure made of polydimethylsiloxane was attached to the wafer. Before functionalization, the gold microelectrodes were washed with 1× PBS. Subsequently, SynBioNFs (Sb68–MatterTag) were pipetted on the gold microelectrodes and incubated for 30 min at room temperature. Excess SynBioNFs (Sb68–MatterTag) were removed by washing three times with 1× PBS. Finally, the gold microelectrodes were blocked with 5% BSA in 1× PBS for 1 h at room temperature. Before use, the gold microelectrodes were washed with 1× PBS. A 30 µl mixture of the patient sample (20 µl sample + 10 µl PBS/1% Tween-80) was then pipetted on the gold microelectrode and incubated for 45 min under stimulation of nanomixing by alternating-current electrohydrodynamics (frequency, 500 Hz; amplitude, 800 mV). In particular, the inclusion of PBS/1% Tween-80 buffer in the patient samples was aimed to inactivate the virus. For clinical samples that are not contagious after treatment (for example, gamma irradiation), the samples can be directly applied on the platform without the use of PBS/1% Tween-80 buffer. Subsequently, after washing the gold microelectrodes with 1× PBS, the bioconjugates of SynBioNFs (Sb68–MatterTag) and nanoboxes were incubated for 20 min under the same nanomixing conditions as above. Finally, the excess bioconjugates were removed by washing with 1× PBS. The gold microelectrodes were then subject to confocal Raman mapping (WITec alpha300 R spectrometer) and collecting/analysing the data using WITec Suite FIVE software. Specifically, a He–Ne laser with an excitation wavelength of 632.8 nm, ×20 objective, electron-multiplying charge-coupled device camera, 0.05 s integration time and 1 µm step size was used for scanning the images with a size of 60 µm × 60 µm.

Clinical sample details

SCV2-positive clinical patient samples were supplied by the Molecular Diagnostics Unit at Pathology Queensland, and consisted of nasopharyngeal swabs resuspended in PBS. These samples were tested using in vitro diagnostics RT-qPCR at the Molecular Diagnostics Unit very early in the pandemic, whereas diagnostic assays were still being completely validated. Negative and positive samples were provided by the Infectious Diseases Laboratory, Microbiology Prevention Division, Pathology Queensland, and tested using the validated BGI platform. Patient samples were collected under the following ethics approval: HREC ref. no. HREC/2020/QRBW/70461; project title, optimizing clinical diagnostics for SCV2. A waiver of consent was approved by this ethics committee and compensation was not applicable for this study.

Fluorescent platform detection of SCV2

Sixty microlitres of SynBioNFs (Sb68–MatterTag) were incubated on the gold surface at room temperature for 2 h, followed by washing three times with PBS to remove free SynBioNFs. Then, 60 µl of the sample solution (that is, SCV2 or medium control) was loaded and incubated on the sensing area at room temperature for 1 h. The gold chips were washed with PBS three times to remove the uncaptured targets. Next, 50 µl of the bioconjugates were applied on the chips and incubated for 1 h. To prepare the bioconjugates, 500 µl SynBioNFs (Nb21–StrepTag) and 10 µl fluorescence beads (coated with streptavidin) were incubated in an Eppendorf tube at room temperature for 1 h and purified through centrifugation. After getting rid of the free bioconjugates, the gold chips were imaged under a fluorescence microscope. The acquired images were then analysed with ImageJ software (1.53).

Electrochemical detection of SCV2

Sixty microlitres of SynBioNFs (Sb68–MatterTag) were incubated on the screen-printed electrodes at room temperature for 2 h. After washing away the free SynBioNFs (Sb68–MatterTag), 60 µl of the sample solution (that is, SCV2 or medium control) was applied on the inner circular working electrodes for an incubation of 1 h and subsequently washed three times with PBS for electrochemical detection. For DPV measurement, 40 µl of 2.5 mM [Fe(CN)6]3−/[Fe(CN)6]4− redox couple in 1× PBS (pH 7.4) containing 0.1 M KCl was added onto the screen-printed electrodes to record the current. The DPV scan was conducted on an electrochemical analyser CHI 650D (CH Instruments) using a scan voltage from –0.2 to 0.4 V, pulse amplitude of 50 mV, pulse width of 50 ms, potential step of 5 mV and pulse period of 10 ms. The CV measurements were performed in 10 mM PBS in the presence of the [Fe(CN)6]3−/4− redox system (pH 7.4, 2.5 mM [Fe(CN)6]3−/4−). The data were recorded between –0.6 and 0.6 V at a scan rate of 100 mV s–1.

LFA for SCV2 detection

Lateral flow test strips (width, 7 mm; length, 80 mm) were prepared using nitrocellulose HP-80 FF strips with laminate backing (Cytiva) and medium-sized absorbent pads attached at the top of the strips. Multiple strips were prepared using a programmable high-speed strip cutter (KinBio). Each strip was spotted with 0.2 µl (200 ng) of capture CR3022 monoclonal antibodies and dried in a 37 °C incubator for 15 min. The CR3022 antibodies were prepared in-house using Chinese hamster ovary cell culture.

Samples for the lateral flow strips were set up in 0.2 ml thin-walled tubes and incubated at 37 °C for 10 min. Next, 10 µl of bioconjugates of SynBioNFs (Nb21–SpyTag) and spherical gold nanoparticles (coated with SpyCatcher) in PBS was mixed with 10 µl SCV2 or medium control. Then, 1% BSA and 1% Tween-80 were included in each reaction. Each reaction was incubated at 37 °C for 15 min. The whole reaction (20 µl) was aliquoted into wells of a 96-well plate, and the tests strips were dipped into the wells to allow the samples to run vertically up the strips towards the absorbent pad for 1–2 min. Then, 50 µl PBST (1× PBS + 0.05% Tween-20) was added to the wells and incubated for further 5 min to move all the bioconjugates up the strip. Visual colorimetric reactions at the capture line were imaged using a digital camera.

ELISA-based assay for stability and avidity test

For the ELISA assay, 10 µg ml–1 of recombinant SCV2 RBD protein diluted in 1× TBS (20 mM Tris (pH 8.0), 300 mM NaCl) was coated on MaxiSorp ELISA plate wells for 1 h at room temperature. The wells were then incubated with the blocking buffer (3.00% BSA in TBS with 0.05% Tween-20) for 1 h at room temperature. Then, 100 µl of anti-SCV2 RBD SynBioNFs (Nb21–SpyTag) (1:5) or rabbit polyclonal antibody (1:10,000) were added to each well and incubated for 1 h at room temperature. For a thermal stability analysis, SynBioNFs or rabbit polyclonal antibody aliquots were incubated at the indicated temperature for 1 h and subsequently added into the respective wells. For avidity assay, urea, guanidine hydrochloride, Triton X-100, sodium dodecyl sulfate or NaCl were added at the indicated concentrations for 1 h at room temperature. To test the acidic pH conditions, a mix of citric acid and sodium phosphate buffers (pH, 2.6–7.6) were used. For alkaline pH conditions, a mix of sodium carbonate and sodium bicarbonate buffers (pH, 9.2–10.8) were used. After washing the wells with TBST five times, HRP-conjugated anti-Myc tag antibody (1:5,000 dilution in 3% BSA–TBST) or HRP-conjugated goat anti-rabbit IgG secondary antibody (1:10,000) was added to the wells for 1 h. The wells were washed with TBST and finally 100 µl of TMB substrate was added to each well, and the reaction was stopped by 100 µl of 1 M sulfuric acid.

Octet assay for measurement of protein interaction

Streptavidin sensors (ForteBio) were pretreated in 200 µl of 10 mM PBS for 10 min. Each well was loaded with 200 µl of the solution. The assay was performed by setting a program: the sensors were dipped in PBS for 120 s in the initial baseline step, loaded with 75 µg of strep RBD per well in the loading sample, dipped in PBS for 120 s in the second baseline step, interacted with SynBioNFs (Nb21–SpyTag) or soluble Nb21 nanobody for 300 s in the association step and ended with disassociation in PBS for 600 s.

Statistical analysis

The diagnostic sensitivity, specificity and accuracy of the SynBioNF-based screening of clinical samples on the SERS platform were determined based on the confusion matrix (Fig. 6c) with the following formulas:

Sensitivity = Number of true positive assessments / Number of all positive assessments = 81/(81 + 3) = 96.43%;

Specificity = Number of true negative assessments / Number of all negative assessments = 50/(0 + 50) = 100 %;

Accuracy = Number of correct assessments / Number of all assessments = (81 + 50)/(81 + 3 + 0 + 50) = 97.76%.

Two-tailed t-tests, receiver operating characteristic curve and Bland–Altman analysis were performed in GraphPad Prism (v. 9.2).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.