While cell culture platforms are widely used for P. aeruginosa infection diagnosis, they exhibit laborious and time-consuming characteristics, necessitate selective enrichment and biochemical screening, and are linked to low sensitivity, selectivity, and contamination risk. Consequently, there is a demand for faster, more precise, and sensitive diagnostic approaches for P. aeruginosa identification. In recent years, researchers have been continuously developing novel detection methods^[1–4]. Pseudomonas exotoxin A (PE) is a protein toxin derived from Pseudomonas aeruginosa, exhibiting strong cytotoxicity and immunogenicity. Consequently, determining the content of PE is of paramount significance for the diagnosis and treatment of related infections. Presently, detection techniques for PE primarily rely on the biological enzyme-linked immunosorbent assay (ELISA), a method characterized by time-consuming procedures, intricate operations, high costs, and the need for labeling^[5,6]. Thus, there is a compelling necessity to develop novel and efficient detection technologies.

In recent decades, the development of optical fiber technology has led to the increasing maturity of optical fiber biosensing technology based on fiber microstructures as sensing elements. Due to the small structural volume, low cost, strong interference resistance, high refractive index sensitivity, and real-time online detection capabilities of optical fibers^[7,8], combined with the strong molecular recognition specificity of biosensing probe molecules, optical fiber biosensing technology has played a complementary and alternative role in traditional biochemical detection techniques. Commonly used optical fiber biosensing elements currently include long-period gratings (LPG), fiber Bragg gratings (FBG), tilted fiber gratings (TFG), and D-type fiber evanescent wave sensor elements^[9–13]. Nevertheless, due to the difficulty of optical fiber biosensors in detecting small molecules or analytes at low concentrations, there is still significant room for improvement in optical fiber biosensing technology^[14,15]. To overcome these limitations, researchers have primarily focused on optimizing fiber microstructures and biosensitive materials to enhance sensor performance. In order to increase the sensitivity of the fiber to refractive index (RI) changes, researchers have developed an S-tapered fiber (STF) and have successfully applied it to the detection of temperature^[16], humidity^[17], RI^[18], strain^[19], and more. Through the analysis of detection principles, it is evident that STF also exhibits considerable potential in antigen–antibody detection, thereby enhancing sensor sensitivity and specificity.

In the realm of fiber optic biosensors, the utilization of biologically sensitive materials is indispensable, exerting a profound impact on the performance of biosensors, particularly in bridging the inorganic-organic interface. Through continuous exploration by researchers in recent years, graphene oxide (GO)^[20–22] and gold nanoparticles (GNP)^[23-25] have emerged as focal points in the realm of biologically sensitive materials due to their outstanding optical properties, biological compatibility, adsorption capacity, and chemical stability. For instance, a long-period grating (LPG) biosensor based on GO has been designed for highly sensitive label-free antibody–antigen immunosensing^[26]. Recently, a novel fiber optic biosensor utilizing the unique optical properties of GNP has been developed for the determination of cadmium ion [Cd(II)] concentration^[27]. Currently, the sensitive materials used in fiber optic biosensing technology are predominantly single materials. Graphene oxide, with its excellent transparency and fiber-coating capability, holds a crucial position in fiber optic sensors. In contrast, GNP exhibits superior bio-adsorption capability and surface area, granting it a superior position in biosensors. Considering the aforementioned factors comprehensively, we propose a scheme to combine these two materials to form a bilayer biologically sensitive film, thereby enhancing the excellent performance of fiber optic biosensors.

Currently, commonly used biological probes for specific recognition and binding analysis of analytes include antibodies^[28], nucleic acids^[29,30], and enzymes^[31]. In fiber optic immunosensors, traditional monoclonal antibodies are frequently employed as biological probes. However, they suffer from drawbacks such as large molecular size, labor-intensive production, long development cycles, high costs, and limited stability. Nevertheless, with advancements in antibody technology, in recent years, researchers have developed superior antibodies known as nanobodies (Nbs), originating from camels and representing the smallest fully functional antigen-binding fragments^[32,33]. Figure 1 illustrates the distinctions between traditional antibodies and Nbs. Due to their small size, strong affinity, high specificity, structural stability, ease of dispersion, and short development cycles, Nbs exhibit characteristics that make them promising alternatives as biological probes in the field of biosensing technologies.

In this study, we propose the development of an optical fiber biosensor designed for highly selective and label-free detection of PE, utilizing a functionalized STF as the transducer. We fabricated a novel GO/GNP bilayer bio-sensitive film, establishing an effective matrix for antibody immobilization with heightened stability and bioactivity. The biosensor’s sensitivity is augmented by this innovative antibody immobilization biointerface, incorporating both GNP and GO. The surface morphology and coated materials were characterized using scanning electron microscopy (SEM). We further enhanced the biosensor to achieve an optical signal boost, enabling the sensitive detection of PE. The biosensor’s selectivity is heightened through the immobilization of anti-PE Nb onto the GNP, facilitating specific interactions with PE. For comparison, an additional STF biosensor was prepared with only GO as the bio-sensitive film. The STF biosensor presents a potentially convenient, low-cost, sensitive, and specific method for antigen detection.

Figure 2(a) depicts the fabrication process of the STF. The S-shaped configuration was achieved by off-axis tapering a single-mode fiber (SMF-28e, Corning) using a commercial fusion splicer (Ericsson FSU-995PM). The STF preparation procedure involved connecting one end of the optical fiber to a light source (Superk Compact, NKT Photonics, Inc.), with the other end linked to the spectrometer (Yokogawa AQ6370D). The process included adjusting the dislocation distance, setting the discharge current and time for each step in the program, and initiating the taper program. The microscopic image of the fabricated STF is presented in Fig. 2(c). The resulting STF possesses a waist diameter, axial offset, and total length of 44.8, 178.5, and 880.6 µm, respectively.

As depicted in Fig. 2(b), the sensing mechanism of the STF hinges upon light interference^[34]. Sensitivity to variations in the external RI is demonstrated by cladding modes which are stimulated at the initial bending section of the STF. The incident light undergoes propagation along distinct trajectories until it reaches the second bending section of the STF. As the light traverses the tapered waist, it induces a particular phase difference, resulting in interference between cladding modes and core modes. The output light intensity can be expressed as I=Ico+Icl+2IcoIclcos(Δφ),(1)where Ico and Icl represent the light intensities of the core mode and cladding mode, respectively. Δφ signifies the phase difference between the core mode and the cladding mode upon traversing the tapered waist, and its expression is given by Δφ=2πΔneffLeffλ,(2)where Δneff denotes the RI difference between the core mode and the cladding mode, Leff represents the effective interference arm length, and λ is the incident wavelength. When Δφ=(2m+1), with m=0,1,2…, the transmission peak wavelength can be expressed by λm=2πΔneffLeff(2m+1).(3)

The interference signal processing method we used was first proposed by Vitrik et al.^[35]. As per Eq. (3), the wavelength shift is attributed to modifications in the RI of the STF surface. Alterations in the thickness of the molecular layer on the fiber surface correspond to changes in the RI of the external environment. Thus, the detection of molecules can be achieved by monitoring the observed wavelength shifts.

Figure 3 illustrates the schematic diagram of the experimental setup for the optical fiber biosensor system based on the STF. The STF sensing area was securely positioned within a square groove, and all samples were dispensed using a pipette. Broadband light from the light source was introduced into the STF, and the real-time monitoring of the transmission spectrum was conducted using a spectrum analyzer within the operational wavelength range of 1000–1700 nm.

Biochemical materials were immobilized onto the STF surface using the layer-by-layer (i-LbL) technique. The i-LbL approach involves the chemical modification of the fiber surface through in situ layer-by-layer self-assembly. Initially, the sensitive region of the STF was securely positioned within a custom square groove and subjected to immersion in an acetone solution for 30 min to eliminate organic contaminants. Subsequently, it underwent rinsing with deionized water and absolute alcohol, respectively. In the next step, the STF was immersed in a 1.0 M NaOH solution for 2 h to introduce hydroxyl groups to the fiber surface, followed by thorough rinsing with deionized water and absolute alcohol. The STF was immersed in a 5% aqueous solution of (3-aminopropyl) triethoxysilane (APTES) for 3 h, during which the hydroxyl groups on the fiber surface underwent a reaction with APTES molecules, forming stable Si-O-Si bonds. After thorough rinsing with absolute alcohol, the sample underwent heating at 95°C for 10 min to enhance APTES fixation. As a result, the positively charged −NH2 groups were able to electrostatically adsorb onto the negatively charged GO. Following this, the GO solution underwent ultrasonic treatment for 20 min to enhance the dispersion of the GO nanosheets. Subsequently, 20 µL of the dispersions were pipetted into the sample cell until the STF was completely immersed in the GO solution. Finally, the sample was incubated in an oven at 42°C until the ethanol solvent completely evaporated, leading to the adsorption of the GO nanosheets onto the STF surface, thereby generating the STF-GO.

In this study, the GNPs were synthesized following the method proposed by Frens^[36]. The STF-GO was immersed in a benzene solution containing 0.1% (3-mercaptopropyl)trimethoxysilane (MPTES) for 4 h to introduce thiol groups onto the GO surface. These thiol groups formed stable Au-S bonds with the GNP surface, facilitating the immobilization of GNPs. Subsequently, the STF-GO was immersed in a colloidal gold solution for 12 h and cleaned with deionized water, resulting in the successful preparation of the STF-GO/GNP. The specific procedure for assembling Nb onto STF-GO/GNP involved the following steps: the STF-GO/GNP was immersed in a 20 mmol/L 11-mercaptoundecanoic acid (11-MUA) ethanol solution for 30 min. The thiol group within the 11-MUA molecule established a stable Au-S bond with the GNP surface, anchoring the 11-MUA securely onto the GNP and exposing the free -COOH group. Subsequently, the STF-GO/GNP was immersed in a 25 mmol/L 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) solution in a 2-(N-morpholino)ethanesulfonic acid (MES) buffer 0.1 mol/L, pH=5.6, with 0.9% NaCl) for 20 min, followed by cleaning with deionized water. It was then immersed in a 60 mmol/L N-hydroxysuccinimide (NHS) solution [0.1 mol/L phosphate-buffered saline (PBS) buffer, pH=7.4] for an additional 20 min. Finally, STF-GO/GNP was immersed in a Nb solution (0.1 mol/L PBS buffer) for 1 h, followed by cleaning with deionized water. This resulted in the preparation of the STF-GO/GNP/Nb, serving as the sensing probe for the optical fiber biosensor. The morphology and optical properties of the obtained samples were characterized using SEM. The characterization results are presented in Fig. 4.

Before initiating biochemical modifications, sensitivity tests were conducted on the STF to verify its suitability for biochemical detection. A series of NaCl solutions with varying concentrations were applied onto the STF, covering the entire tapered region. The RI of the NaCl solutions was measured using an Abbe refractometer and ranged from 1.4115 to 1.4190. It is evident that the STF spectral line exhibits a noticeable right shift with the increasing RI values. The relationship between the change in the RI and the right shift reveals a linear correlation with a sensitivity of 3050.6 nm/RIU.

SEM (FEI, United States) was employed for the initial characterization of the nano-biochemicals coated on the surface of the STF. SEM characterization of the STF, both before and after modification with various nanomaterials, is illustrated in Figs. 4(a), 4(c), 4(e), 4(g), and 4(i). However, due to the microstructure of the nanomaterials, distinguishing these materials under SEM poses challenges. To address this issue, the samples underwent further characterization using energy-dispersive X ray (EDX), as shown in Figs. 4(b), 4(d), 4(f), 4(h), and 4(j). The image in Fig. 4(b) reveals the presence of oxygen and silicon atoms before the introduction of the GO nanosheets.

Considering that the main component of the optical fiber used in this study is silica, composed of oxygen and silicon atoms, this suggests that the STF remained unmodified. In Fig. 4(d), the appearance of carbon atoms is observed after the introduction of GO nanosheets, confirming the deposition of GO materials on the STF surface. Figure 4(f) illustrates the presence of nitrogen atoms following the introduction of Nb, aligning with the main component of Nb being nitrogen atoms and indicating the successful deposition of Nb on the STF-GO surface. Figure 4(h) shows the presence of gold atoms after the introduction of colloidal gold, aligning with the main component of colloidal gold being gold atoms and indicating the successful deposition of GNP materials on the STF-GO surface. Furthermore, Fig. 4(j) displays the appearance of nitrogen atoms after the introduction of Nb, consistent with the main components of Nb being nitrogen and carbon atoms, signifying the deposition of Nb proteins on the surface of the STF-GO/GNP. These findings collectively affirm the successful coating of the Nb/GNP/GO composite on the STF.

To mitigate an excess of vacancies, the STF-GO/GNP/Nb was immersed in a 1 mg/mL bovine serum albumin (BSA) solution for 1 h at room temperature, followed by washing with deionized water to eliminate any surplus unbound protein. The sensing system, as depicted in Fig. 3, is utilized for the measurement of optical properties and biochemical sensing of the STF-GO/GNP/Nb. Light for the analysis emanated from a broad-spectrum light source, entered the biosensor system, and ultimately, the spectrum was acquired by the spectrometer and analyzed using Origin software. The specific operational procedures were as follows: PE detection was conducted over a range of concentrations, progressing from low to high concentrations. When transitioning between different PE concentrations, the sensitive area of the STF was delicately rinsed with a phosphate-buffered saline (PBS) solution (0.1 mol/L, pH=7.4).

The STF-GO/GNP/Nb was utilized as an optical biosensor for the detection of PE. A series of PE concentrations, ranging from 0 to 80 ng/mL, was prepared in 0.01 mol/L PBS. Figure 5 depicts the spectral response to various concentrations of PE, revealing a consistent red shift as the concentration increased. This observation is attributed to the correlation between the thickness of the molecular layer and the RI. The change in Δneff in Eq. (3) corresponds to the increasing PE concentration, resulting in spectral drift. Specifically, the molecular layer on the STF surface increases with higher PE concentrations, leading to changes in the environmental RI. Consequently, different absolute transmission values are obtained due to variations in the coupling efficiency.

Figure 6 presents a trendline depicting the average wavelength shift observed over three repeated experiments. It is evident that the wavelength shifted to the right by 20.1 nm as the PE concentration varied from 0 to 80 ng/mL for the sensor with a bilayer and 11.8 nm for the sensor with a monolayer. By defining the concentration sensitivity as the change induced by 1 ng/mL PE, the sensor with the bilayer sensitivity value was determined to be 0.28 nm/(ng/mL) (R2=0.994), and the sensor with the monolayer sensitivity value was determined to be 0.17 nm/(ng/mL) (R2=0.995). Furthermore, the limit of detection (LOD) was calculated as the PE concentration resulting in a 3N signal response, where N is the noise level defined as the standard deviation of five replicate measurements on blank samples. The LOD for the sensor with a bilayer was calculated to be 0.21 ng/mL, and for the sensor with a monolayer was calculated to be 0.53 ng/mL. According to the results, it can be observed that the sensitivity and LOD of the sensor with a bilayer are significantly superior to those of the sensor with a monolayer. The achieved high sensitivity and low LOD demonstrate that the STF sensing system based on the GNP/GO-sensitive film and Nb exhibited excellent label-free biosensing capabilities.

To ascertain whether the observed spectral shifts arise from the interaction between PE and Nb during the evaluation of biosensor specificity, the sensing probe underwent regeneration and exposure in PBS, PE, carcinoembryonic antigen (CEA), BSA, goat serum albumin (GSA), and bisphenol A (BPA) solution environments, respectively. The resulting spectral drifts were observed and recorded individually, as shown in Fig. 7.

In summary, we have developed and assessed the performance of an STF biosensor designed for the rapid, highly sensitive, and label-free detection of PE. The STF, with favorable morphology and spectral characteristics, was effectively prepared in this investigation. The nanomaterials, including GO, GNP, and Nb, were successively assembled around the STF. Our proposed STF biosensor design achieved a sensitivity value of 0.28 nm/(ng/mL) (R2=0.994) and a detection limit of 0.21 ng/mL. For comparison, an additional STF biosensor was prepared with only GO as the bio-sensitive film, achieving a sensitivity value of 0.17 nm/(ng/mL) (R2=0.995) and a detection limit of 0.53 ng/mL. The performance is significantly lower compared to STF biosensors equipped with a bilayer bio-sensitive film. The suggested STF biosensor holds promise as a biophotonic platform for clinical diagnostics, medical applications, and biomedical research.