Research Article www.advmattechnol.de A Novel Ratiometric Fluorescent Approach for the Modulation of the Dynamic Range of Lateral Flow Immunoassays Amadeo Sena-Torralba, Helena Torné-Morató, Claudio Parolo,* Saba Ranjbar, Mohammad Amin Farahmand Nejad, Ruslan Álvarez-Diduk, Andrea Idili, Mohammad Reza Hormozi-Nezhad, and Arben Merkoçi* 1. Introduction The majority of lateral flow assays (LFAs) use single-color optical labels to pro- vide a qualitative naked-eye detection, however this detection method displays During the last few decades, lateral flow two important limitations. First, the use of a single-color label makes the LFA assays (LFA) have proven to be one of the most successful point-of-care diagnostic prone to results misinterpretation. Second, it does not allow the precise modula- tests, both in clinical and environmental tion of the sensitivity and dynamic range of the test. To overcome these limita- applications.[1–4] Paper-based biosensors tions, a ratiometric approach is developed. In particular, using anti-HIgG offer several important advantages such as functionalized red-fluorescent quantum dots on the conjugate pad (as target cost-effectiveness, sustainability, washing- dependent labels) and blue-fluorescent nanoparticles fixed on the test line (as free operability, and high tunability.[5,6] target independent reporters), it is possible to generate a wide color palette Additionally, thanks to their ease of use, speed, and simplicity, LFAs are routinely (blue, purple, pink, red) on the test line. It is believed that this strategy will used for applications requiring mass- facilitate the development of LFAs by easily adjusting their analytical properties testing and qualitative evaluation.[2,7,8] For to the needs required by the specific application. example, LFAs are commonly used to diagnose pregnancy at home[9] or, more recently, to quickly identify the presence of COVID-19-specific antibodies and anti- gens in pharmacies and mobile testing sites.[7,10,11] Despite A. Sena-Torralba, H. Torné-Morató, C. Parolo, S. Ranjbar, M. A. Farahmand Nejad, R. Álvarez-Diduk, A. Idili, A. Merkoçi this, their recognized low sensitivity[12] and difficulty to inter- Nanobioelectronics & Biosensors Group pret faint bands[13] are still hampering their use in challenging Institut Català de Nanociència i Nanotecnologia (ICN2) clinical applications, where a quantitative detection of the target CSIC and The Barcelona Institute of Science and Technology (BIST) analyte is required.[14] To overcome this limitation, researchers Campus UAB, Bellaterra, Barcelona 08193, Spain have developed different strategies able to enhance the sen- E-mail: claudio.parolo@isglobal.org; arben.merkoci@icn2.cat sitivity in LFA[12,15–18] and to allow on-site assay quantifica- M. A. Farahmand Nejad, M. R. Hormozi-Nezhad Department of Chemistry tion.[19–21] However, these approaches are still mostly confined Sharif University of Technology to academic laboratories, due to their complexity and poten- Tehran 11155-9516, Iran tially high cost that would affect the affordability and usability M. R. Hormozi-Nezhad of LFAs in real settings.[22] Therefore, there is an urgent need Institute for Nanoscience and Nanotechnology for simple and cost-effective strategies that can overcome the Sharif University of Technology Tehran 11155-9516, Iran aforementioned limitations of LFA allowing their implementa- A. Merkoçi tion in a wide range of clinical scenarios. Catalan Institution for Research and Advanced Studies (ICREA) Currently, the majority of LFAs exploit colorimetric labels Pg. Lluís Companys 23, Barcelona 08010, Spain (e.g., gold nanoparticles and polystyrene beads)[23,24] that allow for a convenient naked-eye or smartphone-based detection. The The ORCID identification number(s) for the author(s) of this article former is still the preferred detection mode of LFAs, since it is can be found under https://doi.org/10.1002/admt202101450. equipment-free and cost-effective, therefore, ideal for resource- © 2022 The Authors. Advanced Materials Technologies published by limited settings.[25] Instead, the latter method is on the rise Wiley-VCH GmbH. This is an open access article under the terms of (thanks to the ubiquity of smartphones) and tends to improve test the Creative Commons Attribution-NonCommercial-NoDerivs License, reproducibility (i.e., removing the subjective part of the naked-eye which permits use and distribution in any medium, provided the original detection).[26–30] In both cases, however, the use of a colorimetric work is properly cited, the use is non-commercial and no modifications or adaptations are made. label limits the readout of an LFA to the identification/measure- ment of a single-color signal. Unfortunately, this may generate DOI: 10.1002/admt.202101450 uncertain situations when the presence of faint bands may not Adv. Mater. Technol. 2022, 7, 2101450 2101450 (1 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de be obvious to the final, untrained user[31] or difficult to detect even markers[43] (see Table S1, Supporting Information for pre- with the smartphone under sub-optimal light conditions.[32] Addi- vious studies). In all these cases, ratiometry has been applied tionally, LFAs based on the generation of a single-color signal dis- either as a sensitivity enhancement strategy[44–46] or as a tool play a poorly tunable dynamic range (the minimum to maximum to reduce background interfering signals.[47] range of concentrations that can be detected), which can affect Here, we propose for the first time the implementation of the adaptation of the LFA to different applications requiring the a ratiometric approach in a paper-based sensor for the pre- detection of the same biomarker, but at different concentrations. cise modulation of a LFA’s dynamic range using human IgG In this sense, time consuming sample pre-treatment (concentra- (H-IgG) as a model analyte. Specifically, we demonstrate how tion or dilution) must be performed in order to drop the target’s we can easily modify the color tonality in the LFA’s detection concentration in the quantitative range.[33] line to quantify the desired amount of target molecule even During the millennia the human eye has evolved to dis- with the naked-eye readout. The approach is based on the use tinguish different color tonalities rather than different color of red-fluorescent CdSe@ZnS QDs (655-QDs) that are modi- intensities.[34] For example, we had to distinguish between fied with target-specific antibodies (Figure 1). At the same time, potentially poisonous red mushrooms from nutritious brown target-independent blue fluorescent polystyrene nanobeads ones.[35,36] In this work, we demonstrate how harnessing (450-NBs) are immobilized on the test line (TL) together with this concept it is possible to create more sensitive and pre- target-specific antibodies. The presence of the target and the cise LFAs able to support naked-eye detection. We achieved subsequent immune-sandwich formation induces the accumu- it using a ratiometric approach, which consists in mixing lation of the 655-QDs on the TL. The co-localization of red and two fluorescent labels to generate a wide color variation.[37] blue fluorescent labels on the TL generates a wide color palette Ratiometric readouts are not new in the bioanalytical field, (including red, pink, purple, and blue). Importantly, since the since they have been used for applications ranging from pH amount of 450-NBs is controlled during the fabrication of the sensing[38] to the detection of several molecular targets (e.g., LFA, we can associate a specific color to a specific target con- thrombin,[39] botulinum neurotoxin,[40] and bovine hemo- centration and, therefore, modulate the dynamic range of the globin[41]). In the field of paper-based sensors, ratiometric LFA. We believe that our ratiometric approach will facilitate the approaches have been employed for the detection of diverse development of new LFAs by easily tuning their dynamic range analytes, ranging from heavy metal ions[42] to protein bio- upon the particular requirements of each application. Figure 1. A) Wavelength emission spectra of the 450-NBs and 655-QDs labels, showing a maximum emission peak at 450 and 655 nm, respectively. B) Schematic representation of the ratiometric LFA strip for H-IgG detection. 655-QDs (red nanoparticles) are conjugated to α-HIgG antibodies and placed on the conjugate pad, while unmodified 450-NBs (blue nanoparticles) are preadsorbed together with α-HigG antibodies on the TL. H-IgG anti- bodies are deposited on the CL. The presence of the target analyte in the sample promotes the formation of the immune-complex on the TL, generating a progressive color change from blue to purple, pink and red, which can be quantitatively related to the concentration of the target analyte. C) The combination of different ratios of 450-NBs and 655-QDs leads to the modulation of the assay’s dynamic range. Adv. Mater. Technol. 2022, 7, 2101450 2101450 (2 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de 2. Results and Discussion desired color for a specific target concentration. For example, assuming the fluorescence signals of the two labels equally 2.1. Concept change with changes in their concentration, if we want to shift the dynamic range of the LFA one order of magnitude, we can We propose the application of a ratiometric strategy in LFA simply deposit on the TL tenfold the amount of 450-NBs. In exploiting red-fluorescent 655-QDs (as target-dependent this way, it would be required tenfold the amount of target to signal transducers) and blue fluorescent 450-NBs (as target- accumulate enough 655-QDs to compensate for the increased independent reporters) (Figure 1A). The formers are function- amount of 450-NBs. This precise control of the dynamic range alized with anti-HIgG antibodies and located on the conjugate allows us to quantify the desired amount of target with the pad, while the latter are not modified, and they are placed on naked eye. At the same time, it allows also to easily avoid the the TL together with anti-HIgG antibodies (Figure 1B). The hook effect,[48] which is the loss of signal intensity in the test presence of the target (i.e., H-IgG) promotes the accumula- line upon the presence of excess target analyte in the sample tion of the antibody-conjugated 655-QDs on the TL, leading (the consequent low signal intensity can be falsely interpreted to the co-localization of the two fluorescent labels on the TL. as a low target analyte concentration). This induces a color change on the TL that depends on the concentration of the target analyte in the sample. That is, for increasing concentrations of target the color signal will pro- 2.2. Nanoparticles Characterization gressively shift from blue to pink, purple, and finally red. Con- versely, the absence of the target in the sample will keep the To develop our ratiometric LFA we selected two fluorescent blue fluorescence emission in the TL. In both cases, the con- labels (655-QDs and 450-NBs) which display the same excita- trol line (CL) (based on secondary antibodies specific for those tion wavelength, but different emission spectra (Figure 2A). Of on the surface of 655-QDs) will always provide a red fluores- note, as excitation source we choose an ultraviolet (UV) LED cence signal, indicating that the assay has properly worked because it emits light at a wavelength of 365 nm and it is suit- (Figure 1B,C). able for point-of-care applications (i.e., it is cost-effective (0.85€) The ability to control the dynamic range of our LFA depends and portable (5 mm)). Upon excitation of the labels at 365 nm on two parameters: 1) the color tonality of TL which depends on we collected the maximum emission peaks at 450 and 655 nm the red/blue fluorescent ratio, and 2) the amount of 450-NBs on for the 450-NBs and 655-QDs, respectively (Figure 2A). Next, the TL which can be controlled during its deposition. Regarding we characterized the fluorescence color variations when mixing the former, since each color of our palette is associated with a the nanoparticles at different ratios in solution. Specifically, specific red/blue ratio, the two labels must increase or decrease we started identifying the concentrations that provide an equal accordingly to always obtain the desired color. Specifically, fluorescence signal intensity (Figure 2B, purple spectrum). the more 450-NBs on the TL, the more 655-QDs are required This was achieved by using a solution with 0.32% 655-QDs and to produce the same color shift. Regarding the latter, we can 450-NBs at 0.45% (w/v), which was related to a red emission vary the amount of 450-NBs on the TL to precisely obtain the peak of 2.9E+08 a.u. and a blue emission peak of 2.7E+08 a.u. Figure 2. A) Excitation and emission spectra of the 450-NBs (0.45%) and 655-QDs (0.32%). B) Emission spectra resulting from the ratiometric com- binations of the 450-NBs and the 655-QDs. Inset of the emitted fluorescence under the UV lamp. C) TEM images of the i) 450-NBs and ii) 655-QDs. Adv. Mater. Technol. 2022, 7, 2101450 2101450 (3 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de With this combination, we obtained a purple fluorescent signal commonly used as a detection pad of LFAs) and a smartphone (Figure 2B). The next step was to vary these concentrations to camera (the ideal point-of-care optical reader).[50] Generally, there generate color changes. Specifically, to obtain a pink fluores- are two main color models used to quantify images: RGB and cent signal we kept the concentration of 655-QDs at 0.32% and HSB.[51] In RGB, colors are represented by the combination of we diluted to half (0.225%) the concentration of 450-NBs. This the three primary colors Red, Green, and Blue described by a was related to a red emission peak of 2.9E+08 a.u. and a blue numerical scale from 0 to 255. Instead, the HSB model repre- emission peak of 1.4E+08 a.u. Finally, using 0.32% of 655-QDs sents a color by its Hue (H), its Saturation (S), and Brightness and the 450-NBs at 0.056% we have suppressed the blue signal (B). The Hue values are generally based on a 360° color wheel, obtaining the expected red fluorescence. This was related to a where 0, 120°, and 240° are related to red, green, and blue, respec- red emission of 3.0 E+8 a.u. and a blue emission of 3.8 E+7 a.u. tively. This latter system conceives colors more like the human The collected data demonstrate that we can generate four dif- eye does.[52] That is, the HSB model (and in particular the Hue) ferent colors (blue, pink, purple, red) by simply changing the relates a color tonality with a single numerical value, simplifying relative amounts of 450-NBs and 655-QDs, suggesting the fea- the quantification of color variations in LFA. In contrast, the sibility of the ratiometric approach using these two fluorescent RGB model is better suited to analyze color intensity, as it ena- nanoparticle labels. bles an independent quantification of the three primary colors.[53] Besides having a strong measurable optical signal, the ideal For instance, when analyzing the accumulation of 655-QDs on label of a LFA must flow through the nitrocellulose membrane the TL of a non-ratiometric LFA, the parameter of interest is the and be stable enough to be dried on the conjugate pad without R value, since the G and B values remain constant. Therefore, in aggregating. Therefore, we characterized the size and dispersity our work we used both color scales, the HSB model to measure of the nanoparticles using transmission electron microscopy ratiometry-based LFA and the RGB to measure LFA relying on (TEM), which can provide a good indication of the particle size a single optical label. Importantly, while the HSB values range and dispersity. As we can observe in Figure 2C-i 450-NBs dis- from 0 to 360, since we rely on the ImageJ’s “HSB stack” tool play a diameter of 43.3 ± 5.6 nm (in agreement with the one the values change between 0 and 255 (Figure S2, Supporting specified by the company)[49]; however they tend to aggregate Information).[54] In the case of the fluorescent labels employed in into clusters of up to 2 µm, limiting their ability to move along this study, we obtained H values of 241° and 235° for 655-QDs in the CNPF200 nitrocellulose membrane (with a pore size of solution and on nitrocellulose, respectively; and H values of 147° 5 µm). Since their inability to move guarantees a stable blue and 148° for 450-NBs in solution and on nitrocellulose, respec- signal for the whole duration of the assay, we used them as tively. As observed, there is no significant difference between the target-independent label printed directly on the TL (Figure S1, H values obtained in solution and in nitrocellulose, which highly Supporting Information). On the contrary, the 655-QDs display supports the applicability of these nanoparticles in ratiometric a diameter of 17.5 ± 2.5 nm and they do not show any indica- strategies on paper-based platforms. tion of being prone to aggregation. Therefore, we used them as We then proceeded to further demonstrate the applicability the target-dependent label since they can easily move along the of the ratiometric approach on a paper-like substrate, using a nitrocellulose membrane (see Figure 2C-ii). smartphone camera as a reader. That is, testing different com- binations of 655-QDs and 450-NBs directly on the nitrocellu- lose. Of note, since 450-NBs tend to create a coffee ring effect 2.3. Ratiometric Concept on Paper Substrate on the paper during the data analysis we only considered the central part of the line (Figure S2, Supporting Information). As To adapt our proposed approach to a LFA, we characterized expected, we achieved the same color palette both on nitrocel- the ratiometric optical signal using nitrocellulose (the material lulose and in solution (Figure 3A). For example, using 0.04% of Figure 3. A) Color combination of different concentrations of 655-QDs (0.64%, 0.32%, 0.16%, 0.08%, 0.04%) and 450-NBs (1%, 0.75%, 0,5%, 0,25%, 0%) on the nitrocellulose (CNPF200). B) Color map representing the hue value of the different combinations of 655-QDs and 450-NBs. Adv. Mater. Technol. 2022, 7, 2101450 2101450 (4 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de the 655-QDs and the 450-NBs at a concentration of 0.25% we on the nitrocellulose membrane different amounts of 450-NBs: could obtain purple (H value of 185). Instead, using 0.16% of 1.0%, 0.1%, 0.01%, and 0%. Then we performed calibration the 655-QDs and the 450-NBs at a concentration of 0.25% we curves using concentrations of the target (H-IgG) between 0 to could obtain pink (H value of 208). Exploiting this data, we built 1000 ng mL−1 (which corresponds to 6.6 × 10−9 m) (Figure 4A). a map (Figure 3B) to better represent and explain the influence The formation of the immune-sandwich complex induces the of the different concentrations of fluorescent labels on the final accumulation of the 655-QDs on the TL producing the desired color. Observing the map, it is clear that the higher the concen- color change. Of note, for this set of experiments we modified the tration of 655-QDs, the more 450-NBs are required to induce conjugate pad with a solution of anti-HIgG-conjugated 655-QDs a color change. Specifically, while using 0.16% of 655-QDs we at 0.16% because it provides a good compromise between cost- can observe all the possible colors (red > pink > purple > blue) effectiveness and a visible optical signal by the naked eye. by increasing the amount of 450-NBs, using 0.64% of 655-QDs Looking at the results obtained in the absence of 450-NBs we inhibited the color shift even using the highest amount of observe that the LFA responded as expected. That is, increasing 450-NBs. concentrations of H-IgG produce a more intense red TL. In this case, since there is no 450-NBs on the TL, there is no color change, but just an increasing color intensity (as for classical 2.4. Design of the Ratiometric LFA LFAs) that starts to be appreciable by naked eye at 63.10 ng mL−1 of H-IgG and it saturates at 398.1 ng mL−1 of H-IgG (Figure 4A). We decided to apply the ratiometric approach to precisely tune Conversely, the results obtained using strips with pre-adsorbed the dynamic range of LFAs. Since the final TL color depends on 450-NBs on the TL confirm that the ratiometric approach can the red/blue ratio, we can control the concentration of analyte indeed be used to tune the LFA’s dynamic range. For instance, that will induce the desired color change based on the amount when fixing 0.01% of 450-NBs on TL, the first color change of 450-NBs we pre-deposit on the TL. To prove this, we printed from blue to purple can be visualized upon the detection of Figure 4. A) Picture of the LFA strips with 0.00%, 0.01%, 0.10%, and 1.00% 450-NBs in TL, after performing the calibration curve for the detection of H-IgG (0–1000 ng mL–1) in PBS buffer. B) Quantification of the red peak intensity variation (left y-axis) of the non-ratiometric LFA strips (0.00% 450- NBs in TL) and the Hue variation (right y-axis) after performing the calibration curve for the detection of H-IgG (0–1000 ng mL–1) using the ratiometric LFA strips with 0.01%, 0.10%, and 1.00% 450-NBs in TL. Data of the curves fitting is shown in Table S2 (Supporting Information). C) Calibration curves showing the normalized red peak intensity variation of the non-ratiometric LFA strips and the normalized Hue variation for the ratiometric strips with 0.01% and 0.10%. Data of the curves fitting is shown in Table S3 (Supporting Information). Adv. Mater. Technol. 2022, 7, 2101450 2101450 (5 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de 31.6 ng mL−1 of H-IgG, the second color change from purple to blue curve) and 0.1% more concentration of 450-NBs on the TL pink occurs when detecting 125.9 ng mL−1, and the third color (Figure 4C, blue curve). Similarly, the Hill coefficient increases change from pink to red is perceived at 1000 ng mL−1 of H-IgG. 1.8-fold (3.99) when depositing tenfold more 450-NBs on the Therefore, with the simple pre-adsorption of 450-NBs on the TL. This is because when fixing a higher concentration of TL, we can shift the linear dynamic range to higher concentra- 450-NBs on the TL, more concentration of target is required to tions of the target analyte. The dynamic range can be further increase the Hue value, which in turn increases faster at higher displaced to the right by raising the amount of 450-NBs on target analyte concentrations.[55] Therefore, the combination the TL. For example, the use of 0.1% of 450-NBs promotes a of 0.16% 655-QDs and 0.1% 450-NBs provides a 1.8-fold sensi- delay in the generation of the first and second color changes, tivity enhancement compared to the combination of the same which occur at 398.1 and 1000 ng mL−1 of H-IgG (Figure 4A). concentration of 655-QDs and tenfold lower concentration of Following the same trend, no color change can be observed 450-NBs. within the range of 0–1000 ng mL−1 of H-IgG when using 1% of 450-NBs on TL (Figure 4A). We then proceeded with the quantification of the results. In 2.5. Applicability with Complex Samples the case of the non-ratiometric LFA, for the quantification of the signal, we cannot rely on the H value (since it measures Finally, we tested the applicability of the ratiometric strategy color change) (Figure S3, Supporting Information), but instead, using an undiluted and untreated serum. Serum samples are we used a previously reported method based on the measure- known to interfere with the signal generated on the TL, by ment of the RGB pixels and their conversion to brightness masking the optical signal due to their intrinsic opacity and/ values.[23] Regarding the ratiometric LFA, we used the H value or by hindering the immunorecognition of the target analyte since the output is not the intensity of the TL, but the color due to the presence of competing protein-based molecules.[56] change (Figure S2, Supporting Information). The collected To prove the compatibility of the proposed biosensor with com- data can be fitted to a 4-parameter logistic (sigmoidal) equa- plex samples we spiked H-IgG into fetal bovine serum (FBS). tion. Figure 4B compares the H value variation of the three We performed the calibration curves using the same concen- ratiometric conditions with the red peak intensity variation of trations of fluorescence labels and analyte as in the previous the non-ratiometric LFA. Of note, to better observe the signal experiment, and we observed the same trend. Specifically, variation we subtracted the baseline from the curves obtained the signal achieved on the TL when evaluating a blank serum for 0.01%, 0.1%, and 1% 450-NBs. Instead, Figure 4C shows the sample (without the spiked H-IgG) was significantly similar same data but normalized, in this way it is easier to compare to the signal obtained when analyzing a blank PBS sample. the shift in the dynamic range among the different LFAs tested. Thus, there was no interference between the antibodies used To better explain and estimate the different dynamic ranges, as bioreceptors and the biomolecules of the serum sample. In we have compared the inflection point and Hill coefficient of addition, the signal intensity in TL was not masked by the FBS each calibration curve (see Table 1). The former provides infor- and the color tonalities were not affected for all the evaluated mation regarding the dynamic range of the curve (i.e., the conditions (Figure S4, Supporting Information). Analyzing target concentration corresponding to the half of the maxi­­ the collected data, we achieved a Hill coefficient of 1.52 and an ­mum signal), while the latter enables the quantification of the inflection point of 79.29 ng mL−1, resulting in a dynamic range assay’s sensitivity, as it is related to the steepness of the cali- of 5 to 300 ng mL−1 for 0% 450 NBs (see Table 1). Whereas we bration curve slope. As expected, the application of ratiometry achieved a Hill coefficient of 2.39 and an inflection point of and the subsequent increase of the 450-NBs concentration on 120.14 ng mL−1, giving a dynamic range of 30 to 400 ng mL−1 for the TL promotes the progressive increase of the inflection point 0.1% 450 NBs (Figure 5). Therefore, the same trend is observed in the calibration curve, which in turn results in a shift of the when evaluating a more complex sample matrix such as fetal dynamic range toward higher concentrations of the analyte. For bovine serum, where both a higher inflection point, and Hill instance, almost a twofold and threefold increase in the inflec- coefficient are achieved when using the ratiometric LFA with tion point is achieved when depositing 0.01 (Figure 4C, light 0.16% 655-QDs and 0.1% 450-NBs. Table 1. Comparison of the analytical parameters of the ratiometric and non-ratiometric LFA for H-IgG detection in PBS buffer and FBS. 3. Final Remarks and Conclusions In this study, we have developed a simple and cost-effective % 450-NBs Hill Inflection point Dynamic range coefficient [ng mL−1] [ng mL−1] approach to modulate the dynamic range and sensitivity of LFAs through the application of a ratiometric strategy based Calibration curve in PBS buffer on target-dependent fluorescent QDs (655-QDs) and target- 0.00 2.40 137.55 56–300 independent polystyrene nanobeads (450-NBs). Firstly, we 0.01 2.16 272.89 100–700 have demonstrated the ratiometric capability of these labels 0.10 3.99 401.11 112–700 in solution, where the expected color change from blue to red Calibration curve in FBS can be achieved only by changing the concentration ratios of each nanoparticle. We have then proved the ratiometric strategy 0.00 1.52 79.29 5–300 on a paper-based substrate. More specifically, the same color 0.10 2.39 120.14 30–400 tonalities (obtained in solution) can be perceived when the Adv. Mater. Technol. 2022, 7, 2101450 2101450 (6 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de Figure 5. Red peak intensity and Hue variation along the calibration curve for H-IgG detection (0–1000 ng mL−1) in FBS samples using non-ratiometric and ratiometric LFA strips with 0.1% of 450-NBs on the TL. Inset of the LFA strips with 0.0%, 0.1% of 450-NBs after the detection of 0.00, 10.00, 63.10, 125.90, and 398.10 ng mL–1. Data of the curves fitting is shown in Table S4 (Supporting Information). fluorescent labels are located on the nitrocellulose membrane membranes) to receive approval from regulatory agencies. Or, resulting in the expected wide color palette (blue, purple, pink, in case of an already existing LFA that needs to be adapted for and red). To better understand this, we have created a color map a new application, its sensitivity and dynamic range can also based on the hue values and we demonstrated that the variation be adapted with minor modification on the fabrication process in the color tonality can be modulated upon the combination of (again without the need to select new bioreceptors or mem- the labels at different ratios. Finally, by applying this concept in branes). We believe that the ratiometric approach proposed in a LFA, we have shown that a gradual shift of the dynamic range this manuscript has the potential to dramatically affect the way can be achieved upon the pre-adsorption of higher concentra- LFAs are designed and developed saving precious resources tions of 450-NBs on the TL. We have also observed that the and providing the public with better and more readily available application of ratiometry provides a sensitivity enhancement point-of-care sensors. when evaluating the assay qualitatively with the naked eye. Compared to the already reported analytical methods for H-IgG detection, our approach provides a 78-fold higher analytical sen- 4. Experimental Section sitivity, an 18-fold faster assay time (compared to ELISA) and Materials and Reagents: Human IgG antibody, goat anti-human the possibility to progressively shift the dynamic range, which IgG, bovine se-rum albumin (BSA), phosphate buffer saline (PBS) is something inconceivable in conventional LFAs (Table S5, tablets, sucrose, and tween 20 were purchased from Sigma Aldrich Supporting Information). Looking ahead, we envisage that our (Madrid, Spain). Nitrocellulose membrane CNPF200 was purchased ratiometric approach could benefit from solutions addressing from mdi (Ambala Cantt, India). Cellulose membrane (CFSP001700) issues common to all fluorescence based LFAs. For example, was purchased from Merck Millipore (Billerica, MA, USA), glass fiber by employing better cut-off filters, more defined excitation light Standard 14 was purchased from GE Healthcare (Chicago, IL, USA) and the supporting adhesive cards were purchased from Kenosha sources (with narrower emission peak) and up-converting flu- (Amstelveen, The Netherlands). Fluoro-Max Dyed Blue Aqueous orescent labels (with NIR excitation), it should be possible to Fluorescent Particles (450-NBs), Qdot 655 and Qdot 655 goat F(ab’)2 decrease the background noise and therefore improve the sen- anti-human IgG conjugate (H+L) (655-QDs) were purchased from sitivity, specificity and LoD of the assay. Thermo Fisher Scientific (Eugene, OR, USA). Fetal Bovine Serum (FBS) The results described in this work demonstrate how the pro- was purchased from Linus (Madrid, Spain). posed ratiometric approach can be used as a new tool to pre- Instruments: The IsoFlow Bioreagent dispenser and the lateral flow strips cutter used for the fabrication of the strips were purchased from Imagene cisely control the assay’s dynamic range and to improve the Technology (Hanover, Germany) and Shanghai Kinbio Tech (Pudong sensitivity in LFA (see Table S1, Supporting Information). For New District, Shanghai, China), respectively. The spectrophotometer example, during the development of a LFA its analytical proper- SpectraMax ID3 and transmission electron microscope TEM Technai F20 ties can be easily tuned (without selecting novel bioreceptors or used for the characterization of the nanoparticles were purchased from Adv. Mater. Technol. 2022, 7, 2101450 2101450 (7 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de Molecular devices (San Jose, CA, USA) and (Hillsboro, OR, USA). The Acknowledgements Smartphone Xiaomi Redmi Note 8 used for the assay evaluation was purchased from Xiaomi (Beijing, China). A.S.-T. and H.T.-M. contributed equally to this work. The authors Nanoparticles Characterization: The excitation and emission spectra of acknowledge financial support from NACANCELL project PCIN-2016-066 the Fluoro-Max dyed blue aqueous fluorescent particles (450-NBs) and the (program Euronanomed 2). This work was also funded by the CERCA CdSe@ZnS quantum dots (655-QDs) were evaluated with the SpectraMax Program/Generalitat de Catalunya. The ICN2 was funded by the ID3 spectrophotometer, while the size and shape distribution of the CERCA program/Generalitat de Catalunya. ICN2 acknowledges the nanoparticles were characterized using the Technai F20 TEM. support of the Spanish MINECO for the Project MAT2017-87202-P and LFA Strips Fabrication: The functionalization of the detection pad through the Severo Ochoa Centres of Excellence Program under Grant was always conducted with the nitrocellulose membrane immobilized SEV2201320295. A.I. was supported by PROBIST postdoctoral fellowship on a supporting laminated card. Firstly, the nitrocellulose CNPF200 funded by European Research Council (Marie Sklodowska Curie) membrane was functionalized with a solution containing 450-NBs grant agreement No. 754510). C.P. acknowledges support from the (depending on the experimental condition the concentration changed Spanish Ministry of Science and Innovation and State Research Agency between 0% and 0.1% in PBS buffer 10 × 10−3 m pH 7.4) on the test through the “Centro de Excelencia Severo Ochoa 2019-2023” Program line (TL). Once fixed-dried at 37 °C for 2 h, a solution of anti-HIgG (CEX2018-000806-S), and support from the Generalitat de Catalunya (1 mg mL−1) and H-IgG (1 mg mL−1) in PBS buffer 10 × 10−3 m pH 7.4 through the CERCA Program. was deposited on the TL and control line (CL) and dried at 37 °C for 2 h. Note: The Acknowledgements section was updated on August 9, 2022, Both the 450-NBs and the antibody solutions were dispensed using the after initial publication online. bioreagent dispenser at a flow rate of 0.5 µL cm−1 and using a speed of 50 mm s−1. The conjugate solution was prepared by diluting the 655- QDs modified with goat F(ab’)2 anti-human IgG conjugate solution to 0.16% (w/v) using PBS buffer (10 × 10−3 m pH 7.4, 1% BSA, 0.5% Tween Conflict of Interest 20 and 10% sucrose), then the solution was drop-casted on the glass The authors declare no conflict of interest. fiber membrane and dried in a vacuum chamber at room temperature for 1.5 h. Cellulose membrane was used as sample and absorbent pad. For the sample pad the cellulose membrane was first soaked into Data Availability Statement sample pad buffer (PBS buffer containing 0.5% (wt/vol) BSA and 0.05% (vol/vol) Tween-20) and then dried at 37 °C for 2 h, before mounting it The data that support the findings of this study are available from the on the laminated card. The laminated card was cut into 3 mm width corresponding author upon reasonable request. strips using an automatic cutter. H-IgG Detection and Assay Evaluation: The calibration curve was performed by drop-casting 100 µL of known concentrations of H-IgG Keywords (from 0 to 1 µg mL−1 in PBS buffer 10 × 10−3 m pH 7.4 and spiked FBS) on the sample pad. The strips were incubated at room temperature fluorescence, lateral flow assay, quantum dots, ratiometry, sensitivity for 10 min on a table covered with a black matte cardboard. This was used to prevent light reflection during the assay evaluation. The results Received: November 2, 2021 were recorded by taking a picture with the smartphone (48 megapixels), Revised: December 21, 2021 10 min after the sample addition, by keeping a 9 cm distance from Published online: March 14, 2022 the strips and under constant illumination conditions. The camera parameters of the smartphone were manually set to ISO speed (100) and shutter speed (1/4 s). UV LEDs were used to excite the fluorescent nanoparticles at a wavelength of 365 nm. [1] K. J. Land, D. I. Boeras, X. S. Chen, A. R. Ramsay, R. W. Peeling, The pictures were analyzed using the open-source Image J software,[57] Nat. Microbiol. 2019, 4, 46. which enables the quantification of both the color intensity and the color [2] P. Brangel, A. Sobarzo, C. Parolo, B. S. Miller, P. D. Howes, variation. Specifically, the measurement of the color intensity was done S. Gelkop, J. J. Lutwama, J. M. Dye, R. A. McKendry, L. Lobel, following the recently published protocol.[25] Briefly, after splitting the image M. M. Stevens, ACS Nano 2018, 12, 63. into the three different RGB channels, a plot profile (covering the total [3] D. Quesada-González, G. A. Jairo, R. C. Blake, D. A. Blake, width of the nitrocellulose membrane) for each LFA was measured just for A. Merkoçi, Sci. Rep. 2018, 8, 16157. the red channel. The peak height was then measured and used to analyze [4] A. Sena-Torralba, Y. Pallás-Tamarita, S. Morais, Á. Maquieira, TrAC, the sensor performance. In the case of measuring the color variation, Trends Anal. Chem. 2020, 132, 116050. each image was transformed using the HSB stack feature. Then the plot [5] C. Parolo, A. Merkoçi, Chem. Soc. Rev. 2013, 42, 450. profile for the Hue (H) feature was analyzed for each strip. The peak value [6] X. Huang, Y. Liu, B. Yung, Y. Xiong, X. Chen, ACS Nano 2017, 11, 5238. was recorded and used for the analysis. Note that while for the color [7] B. D. Grant, C. E. Anderson, J. R. Williford, L. F. Alonzo, V. A. Glukhova, intensity the peak height (i.e., peak value – baseline) was measured, for D. S. Boyle, B. H. Weigl, K. P. Nichols, Anal. Chem. 2020, 92, 11305. the color variation just the peak value was measured, since it corresponds to the color seen by the naked eye. In order to facilitate the comparison [8] X. Huang, Z. P. Aguilar, H. Xu, W. Lai, Y. Xiong, Biosens. Bioelectron. between different LFAs and conditions, in the manuscript normalized data 2015, 75, 166. (min–max normalization) was also presented.[58] The limit of detection [9] “Pregnancy Tests,” can be found under https://uk.clearblue.com/ (LoD) was calculated as Optical intensity (LoD) = blank + 3 σ blank pregnancy-tests, 2020. (i.e., the corresponding value of blank sample plus 3 times its standard [10] D. Wang, S. He, X. Wang, Y. Yan, J. Liu, S. Wu, S. Liu, Y. Lei, deviation). The limit of quantification (LoQ) was calculated as Optical M. Chen, L. Li, J. Zhang, L. Zhang, X. Hu, X. Zheng, J. Bai, Y. Zhang, intensity (LoQ) = blank + 10 σ blank.[59] Y. Zhang, M. Song, Y. Tang, Nat. Biomed. Eng. 2020, 4, 1150. [11] T. Nicol, C. Lefeuvre, O. Serri, A. Pivert, F. Joubaud, V. Dubée, A. Kouatchet, A. Ducancelle, F. Lunel-Fabiani, H. Le Guillou- Guillemette, J. Clin. Virol. 2020, 129, 104511. Supporting Information [12] A. Sena-Torralba, D. B. Ngo, C. Parolo, L. Hu, R. Álvarez-Diduk, Supporting Information is available from the Wiley Online Library or J. F. Bergua, G. Rosati, W. Surareungchai, A. Merkoçi, Biosens. from the author. Bioelectron. 2020, 168, 112559. Adv. Mater. Technol. 2022, 7, 2101450 2101450 (8 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH 2365709x, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/admt.202101450 by Readcube (Labtiva Inc.), Wiley Online Library on [09/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmattechnol.de [13] T. Shaw, C. Tellapragada, K. E. Vandana, D. P. AuCoin, [35] P. Wu, X. Hou, J. J. Xu, H. Y. Chen, Nanoscale 2016, 8, 8427. C. Mukhopadhyay, PLoS One 2018, 13, e0194595. [36] Y. Zhou, X. Huang, C. Liu, R. Zhang, X. Gu, G. Guan, C. Jiang, [14] A. Scohy, A. Anantharajah, M. Bodéus, B. Kabamba-Mukadi, L. Zhang, S. Du, B. Liu, M. Y. Han, Z. Zhang, Anal. Chem. 2016, 88, A. Verroken, H. Rodriguez-Villalobos, J. Clin. Virol. 2020, 129, 6105. 104455. [37] X. Huang, Y. Xiong, J. Song, B. C. Yung, X. Chen, X. Huang, Chem. [15] C. Wang, C. Wang, X. Wang, K. Wang, Y. Zhu, Z. Rong, W. Wang, Soc. Rev. 2018, 47, 2873. R. Xiao, S. Wang, ACS Appl. Mater. Interfaces 2019, 11, 19495. [38] P. T. Snee, R. C. Somers, G. Nair, J. P. Zimmer, M. G. Bawendi, [16] B. S. Miller, L. Bezinge, H. D. Gliddon, D. Huang, G. Dold, D. G. Nocera, J. Am. Chem. Soc. 2006, 128, 13320. E. R. Gray, J. Heaney, P. J. Dobson, E. Nastouli, J. J. L. Morton, [39] C. M. Tyrakowski, P. T. Snee, Anal. Chem. 2014, 86, 2380. R. A. McKendry, Nature 2020, 587, 588. [40] J. Lee, M. B. Brennan, R. Wilton, C. E. Rowland, E. A. Rozhkova, [17] A. Sena-Torralba, R. Álvarez-Diduk, P. Claudio, T.-M. Helena, S. Forrester, D. C. Hannah, J. Carlson, E. V. Shevchenko, A. Müller, A. Merkoçi, Anal. Chem. 2021, 93, 3112. D. S. Schabacker, R. D. Schaller, Nano Lett. 2015, 15, 7161. [18] L. Rivas, M. Medina-Sánchez, A. de la Escosura-Muñiz, A. Merkoçi, [41] X. Wang, S. Yu, W. Liu, L. Fu, Y. Wang, J. Li, L. Chen, ACS Sens. 2018, Lab Chip 2014, 14, 4406. 3, 378. [19] Z. Qu, K. Wang, G. Alfranca, J. M. de la Fuente, D. Cui, Nanoscale [42] K. He, X. Zhan, L. Liu, X. Ruan, Y. Wu, Photochem. Photobiol. 2020, Res. Lett. 2020, 15, 10. 96, 1154. [20] B. S. Miller, C. Parolo, V. Turbé, C. E. Keane, E. R. Gray, [43] Z. Luo, T. Lv, K. Zhu, Y. Li, L. Wang, J. J. Gooding, G. Liu, B. Liu, R. A. McKendry, Chem. - Eur. J. 2018, 24, 9783. Angew. Chem., Int. Ed. 2020, 59, 3131. [21] N. Amin, A. Sena-Torralba, R. Álvarez-Diduk, A. Afkhami, [44] C. Liu, D. Ning, C. Zhang, Z. Liu, R. Zhang, J. Zhao, T. Zhao, B. Liu, A. Merkoçi, Anal. Chem. 2020, 92, 4209. Z. Zhang, ACS Appl. Mater. Interfaces 2017, 9, 18897. [22] Y. Liu, L. Zhan, Z. Qin, J. Sackrison, J. C. Bischof, ACS Nano 2021, [45] J. Wang, C. Jiang, J. Jin, L. Huang, W. Yu, B. Su, J. Hu, Angew. 15, 3593. Chem., Int. Ed. 2021, 100193, 13042. [23] J. F. Bergua, L. Hu, C. Fuentes-Chust, R. Álvarez-Diduk, [46] K. G. Shah, V. Singh, P. C. Kauffman, K. Abe, P. Yager, Anal. Chem. A. H. A. Hassan, C. Parolo, A. Merkoçi, Lab Chip 2021, 21, 2417. 2018, 90, 6967. [24] D. Quesada-González, A. Sena-Torralba, W. P. Wicaksono, [47] L. Li, Y. Zhang, S. Ge, L. Zhang, K. Cui, P. Zhao, M. Yan, J. Yu, Anal. A. de la Escosura-Muñiz, T. A. Ivandini, A. Merkoçi, Biosens. Chem. 2019, 91, 10273. Bioelectron. 2019, 132, 132. [48] G. M. S. Ross, D. Filippini, M. W. F. Nielen, G. I. J. Salentijn, Anal. [25] C. Parolo, A. Sena-Torralba, J. F. Bergua, E. Calucho, C. Fuentes-Chust, Chem. 2020, 92, 15587. L. Hu, L. Rivas, R. Álvarez-Diduk, E. P. Nguyen, S. Cinti, D. Quesada- [49] ThermoFischer Scientific, “Fluoro-Max B0100,” can be found under González, A. Merkoçi, Nat. Protoc. 2020, 15, 3788. https://www.thermofisher.com/order/catalog/product/B0100#/ [26] W. Xiao, C. Huang, F. Xu, J. Yan, H. Bian, Q. Fu, K. Xie, L. Wang, B0100, 2021. Y. Tang, Sens. Actuators, B 2018, 266, 63. [50] J. Liu, Z. Geng, Z. Fan, J. Liu, H. Chen, Biosens. Bioelectron. 2019, [27] C. Ruppert, N. Phogat, S. Laufer, M. Kohl, H. P. Deigner, Microchim. 132, 17. Acta 2019, 186, 119. [51] J. L. D. Nelis, L. Bura, Y. Zhao, K. M. Burkin, K. Rafferty, C. T. Elliott, [28] H. Wang, L. Yang, S. Chu, B. Liu, Q. Zhang, L. Zou, S. Yu, K. Campbell, Sensors 2019, 19, 5104. Changlong Jiang, Anal. Chem. 2019, 91, 9292. [52] L. I. Labrecque, Psychol. Mark. 2020, 37, 855. [29] S. Chu, H. Wang, X. Ling, S. Yu, L. Yang, Changlong Jiang, ACS [53] K. Cantrell, M. M. Erenas, I. de Orbe-Payá, L. F. Capitán-Vallvey, Appl. Mater. Interfaces 2020, 12, 12962. Anal. Chem. 2010, 82, 531. [30] S. Chu, H. Wang, Y. Du, F. Yang, L. Yang, Changlong Jiang, ACS [54] G. Saravanan, G. Yamuna, S. Nandhini, in Int. Conf. Commun. Sustainable Chem. Eng. 2020, 8, 8175. Signal Process., Melmaruvathur, India 2016. [31] D. A. Mendels, L. Dortet, C. Emeraud, S. Oueslati, D. Girlich, [55] F. Ricci, A. Vallée-Bélisle, K. W. Plaxco, PLoS Comput. Biol. 2011, 7, J. B. Ronat, S. Bernabeu, S. Bahi, G. J. H. Atkinson, T. Naas, Proc. e1002171. Natl. Acad. Sci. USA 2021, 118, e2019893118. [56] Y. Rosenberg-Hasson, L. Hansmann, M. Liedtke, I. Herschmann, [32] A. Carrio, C. Sampedro, J. L. Sanchez-Lopez, M. Pimienta, H. T. Maecker, Immunol. Res. 2014, 58, 224. P. Campoy, Sensors 2015, 15, 29569. [57] W. S. Rasband, “ImageJ,” can be found under https://imagej.nih. [33] D. Gasperino, T. Baughman, H. V. Hsieh, D. Bell, B. H. Weigl, Annu. gov/ij/docs/faqs.html, 2018. Rev. Anal. Chem. 2018, 11, 219. [58] O. A. Akanbi, I. S. Amiri, E. Fazeldehkordi, in A Machine-Learning [34] A. Bigdeli, F. Ghasemi, S. Abbasi-Moayed, M. Shahrajabian, Approach to Phishing Detection and Defence, Syngress, Rockland, N. Fahimi-Kashani, S. Jafarinejad, M. A. Farahmand Nejad, USA 2015, p. 45. M. R. Hormozi-Nezhad, Anal. Chim. Acta 2019, 1079, 30. [59] D. A. Armbruster, T. Pry, Clin. Biochem. Rev. 2008, 29, S49. Adv. Mater. Technol. 2022, 7, 2101450 2101450 (9 of 9) © 2022 The Authors. Advanced Materials Technologies published by Wiley-VCH GmbH