Comparative Study of Gold and Carbon Nanoparticles in Nucleic Acid Lateral Flow Assay

A lateral flow assay (LFA) is a paper-based, point-of-need test designed to detect a specific analyte in complex samples in low-resource settings. Although LFA has been successfully used in different applications, its use is still limited when high sensitivity is required, especially in the diagnosis of an early-stage condition. The limit of detection (LOD) is clearly related to the signal-generating system used to achieve the visual readout, in many cases involving nanoparticles coupled to a biomolecule, which, when combined, provides sensitivity and specificity, respectively. While colloidal gold is currently the most-used label, other detection systems are being developed. Carbon nanoparticles (CNPs) demonstrate outstanding features to improve the sensitivity of this technology by producing an increased contrast in the paper background. Based on the necessity of sensitivity improvement, the aim of this work is a comparative study, in terms of analytical performance, between commercial streptavidin gold nanoparticles (streptAv-AuNPs) and avidin carbon nanoparticles (Av-CNPs) in a nucleic acid lateral flow assay. The visual LOD of the method was calculated by serial dilution of the DNA template, ranging from 0.0 to 7 pg μL−1/1.5 × 104 CFU mL−1). The LFA achieved visual detection of as low as 2.2 × 10−2 pg μL−1 using Av-CNPs and 8.4 × 10−2 pg μL−1 using streptAv-AuNPs. These LODs could be obtained without the assistance of any instrumentation. The results demonstrate that CNPs showed an increased sensitivity, achieving the nanomolar range even by visual inspection. Furthermore, CNPs are the cheapest labels, and the suspensions are very stable and easy to modify.


Introduction
The recent outbreak of SARS-CoV-2 that spread worldwide in a matter of days due to the globalization process highlighted the need for rapid diagnostic tests (RDTs) able to provide a rapid result and to help implement security and prevention measures to interrupt the transmission chain. At present, and more than ever, these RDTs are a major research field of growing interest among researchers around the globe. In 2003, the World Health Organization (WHO/TDR) published a set of criteria for the ideal RDT test under the acronym ASSURED [1]: Affordable, Sensitive, Specific, User-friendly, Rapid and Robust, Equipment-free, and Deliverable to end-users. Although the original features remain relevant, it has been recently revised to include two additional criteria as REASSURED of R (Real-time connectivity) and E (Ease of specimen collection and environmental friendliness), in order to design future devices [2].
Since the presentation of the first US patents by three individual researchers in 1987 (Robert Rosenstein for Becton Dickinson & Co., Keith May for Unilever, and David Charlton for Carter Wallace), lateral flow assays (LFAs) are the most common commercially available RDTs. There has since been an increasing interest in the development of LFAs

Reagents and Equipment
The carbon nanoparticles (CNPs) obtained from carbon black (CB) Spezial Schwartz 4 (Degussa AG, Essen, Germany, Ref For the preparation of carbon black conjugates, an Eppendorf Thermomixer C (Eppendorf AG, DE, Hamburg, Germany), a microtube centrifuge Biocen 22R (Ortoalresa, ES), and a probe sonicator (Sonics Materials, Newtown, CT, USA Ref. VCX130PB220) were used. A SimpliAmp Thermal Cycler (Applied Biosystems, Foster City, CA, USA) was used for the double-tagging PCR amplification. For the immobilization of biomolecules into LFA membranes, an IsoFlow reagent dispensing system (Claremont Bio, Upland, CA, USA) coupled with a syringe bomb KDS LegatoTM 200 (KD Scientific Inc., Holliston, MA, USA) were used at a flow rate of 38 µL min −1 .

Carbon Nanoparticles Modification and Storage
Briefly, 10 mg·mL −1 CNP solution was homogenized in Milli-Q water by sonication at 27 W for 5 min and 5-fold diluted in 5 mmol·L −1 borate buffer, pH 8.8. The immobilization was performed by physical adsorption [20]. To achieve this, 350 µL of avidin (1 mg·mL −1 ) was added to 1 mL CNP solution and incubated for 3 h at RT using a thermomixer at 550 rpm (slight rotation). The suspension containing the avidin-modified carbon nanoparticles (Av-CNPs) was centrifuged for 15 min at 13,600× g, and the supernatant was removed to eliminate the excess avidin. Afterward, Av-CNPs were resuspended in 1 mL of 5 mmol·L −1 borate buffer, pH 8.8 with 1% (w/v) BSA, and centrifuged again under the same conditions. Finally, Av-CNPs were resuspended in 1 mL storage buffer (100 mmol·L −1 borate, pH 8.8 with 1% (w/v) BSA) to a final concentration of 0.2% (w/v) of Av-CNPs. The final conjugate solution was stored at 4 • C until further use.

Characterization of CNPs by Transmission Electron Microscopy
A JEOL 1400 transmission electron microscope (TEM) at an accelerating voltage of 120 kV was used for the characterization of the CNPs. For sample preparation, a microquantity of CNP solution was sonicated at 27 W for 5 min and dropped on a copper plate with a carbon film grid until dried. Images were further processed by the Nucleus counting plugin for ImageJ to calculate their size distribution. Before the analysis, images were transformed to 8-bit, with the contrast and brightness adjusted and binarized to make them processable by the software. Moreover, before the plugin was employed, a watershed filter was applied to separate them correctly. Only well-separated CNPs were counted and measured.
2.4. Bacterial Strains Culture, DNA Extraction, and Double-Tagging PCR E. coli DH5-α strain was cultured in Luria-Bertani (LB) plates for 16 h at 37 • C. After the overnight culturing, the concentration of bacteria was found to be approximately 10 8 CFU·mL −1 . Lysis of bacteria and DNA extraction were performed, incubating 1 mL of bacteria at 99 • C for 1 h at different concentrations and centrifuging at 15,500× g for 2 min. The lysate was washed with Milli-Q water, resuspended with 200 µL Tris-EDTA (10 mmol·L −1 Tris; 1 mmol·L −1 ; pH 8.0) for DNA extraction, and incubated for 10 min at 99 • C followed by 15 min in ice. Genomic DNA was centrifuged for 5 min at 12,000× g and resuspended at different concentrations.

Double-Tagging PCR and Quantification by Gel Electrophoresis
During the double-tagging PCR procedure (Figure 1), bacterial DNA was amplified and labeled with biotin and digoxigenin using modified primers [29,30]. PCR was carried out in 15 µL reactions. The primers for the double-tagging PCR were selected for the specific amplification of the E. coli 16S ribosomal gene (NCBI Reference Sequence MN661169.1, as detailed in Supplementary Materials). The sequences of the primers were as shown in Table 1. transformed to 8-bit, with the contrast and brightness adjusted and binarized to make them processable by the software. Moreover, before the plugin was employed, a watershed filter was applied to separate them correctly. Only well-separated CNPs were counted and measured.

Bacterial Strains Culture, DNA Extraction, and Double-Tagging PCR
E. coli DH5-α strain was cultured in Luria-Bertani (LB) plates for 16 h at 37 °C. After the overnight culturing, the concentration of bacteria was found to be approximately 10 8 CFU•mL −1 . Lysis of bacteria and DNA extraction were performed, incubating 1 mL of bacteria at 99 °C for 1 h at different concentrations and centrifuging at 15,500 × g for 2 min. The lysate was washed with Milli-Q water, resuspended with 200 μL Tris-EDTA (10 mmol•L -1 Tris; 1 mmol•L -1 ; pH 8.0) for DNA extraction, and incubated for 10 min at 99 °C followed by 15 min in ice. Genomic DNA was centrifuged for 5 min at 12,000 × g and resuspended at different concentrations.

Double-Tagging PCR and Quantification by Gel Electrophoresis
During the double-tagging PCR procedure (Figure 1), bacterial DNA was amplified and labeled with biotin and digoxigenin using modified primers [29,30]. PCR was carried out in 15 μL reactions. The primers for the double-tagging PCR were selected for the specific amplification of the E. coli 16S ribosomal gene (NCBI Reference Sequence MN661169.1, as detailed in Supplementary Materials). The sequences of the primers were as shown in Table 1.    Further experimental details of the double-tagging PCR procedure are also described in the Supplementary Materials.
Afterward, amplimers were analyzed with conventional agarose gel electrophoresis on 2% agarose gel in TAE buffer (40 mmol·L −1 Tris, 40 mmol·L −1 acetate, 1 mmol·L −1 EDTA, containing a DNA stain to check its size distribution. Figure 2 shows the configuration of Av-CNPs ( Figure 2A) and streptavidin-modified AuNPs ( Figure 2B) lateral flow immunoassays. In both instances, the biotin tag of the double-tagged amplicon reacted with the signal-generating systems, while the immobilization of the DNA was achieved by the digoxigenin tag and using anti-DIG antibody on the nitrocellulose strip in the test line. As a control line, a biotinylated macromolecule (such as biotinylated protein or biotinylated dendrimer as previously reported by our research group) [31,32] can be used in order to capture all the remaining Av-CNPs, as shown in Figure 2. It is important to highlight that biotin alone cannot be used, since it is not adsorbed on nitrocellulose strips due to its low MW and, as such, it is washed during the test, as previously reported by our research group [32]. In this instance, a biotinylated protein (a biotin-modified rabbit polyclonal antibody) was used. Afterward, amplimers were analyzed with conventional agarose gel electrophoresis on 2% agarose gel in TAE buffer (40 mmol•L -1 Tris, 40 mmol•L -1 acetate, 1 mmol•L -1 EDTA, containing a DNA stain to check its size distribution. Figure 2 shows the configuration of Av-CNPs ( Figure 2A) and streptavidin-modified AuNPs ( Figure 2B) lateral flow immunoassays. In both instances, the biotin tag of the double-tagged amplicon reacted with the signal-generating systems, while the immobilization of the DNA was achieved by the digoxigenin tag and using anti-DIG antibody on the nitrocellulose strip in the test line. As a control line, a biotinylated macromolecule (such as biotinylated protein or biotinylated dendrimer as previously reported by our research group) [31,32] can be used in order to capture all the remaining Av-CNPs, as shown in Figure 2. It is important to highlight that biotin alone cannot be used, since it is not adsorbed on nitrocellulose strips due to its low MW and, as such, it is washed during the test, as previously reported by our research group [32]. In this instance, a biotinylated protein (a biotin-modified rabbit polyclonal antibody) was used. In the CNP strips, 5 μL of Av-CNPs was added to the sample in a final volume of 50 μL running buffer (100 mmol•L −1 borate buffer, pH 8.8, with 1% (w/v) BSA, 0.05% (v/v) Tween ® 20), and was dropped into the sample pad after 2 min of incubation. A further 200 μL of running buffer was dropped into the sample pad. The readout was achieved after 10 min from sample deposition. For the semiquantitative plots, the samples are the double-tagged amplicons at different concentrations (from 0.00 to 7.09 pg μL −1 /1.5 × 10 4 CFU mL −1 ). In the CNP strips, 5 µL of Av-CNPs was added to the sample in a final volume of 50 µL running buffer (100 mmol·L −1 borate buffer, pH 8.8, with 1% (w/v) BSA, 0.05% (v/v) Tween ® 20), and was dropped into the sample pad after 2 min of incubation. A further 200 µL of running buffer was dropped into the sample pad. The readout was achieved after 10 min from sample deposition. For the semiquantitative plots, the samples are the double-tagged amplicons at different concentrations (from 0.00 to 7.09 pg µL −1 /1.5 × 10 4 CFU mL −1 ).

Lateral Flow Assays Based on Carbon and Gold Nanoparticles as Signal-Generating Systems
Additionally, AuNP-based lateral flow strips were performed as previously described by our research group [31,32], and included a conjugate pad in the configuration as a reservoir of AuNPs. In this instance, a final readout was achieved at 15 min from sample deposition.

Data Interpretation and Analysis
In order to avoid any bias in the images due to luminosity, all the strips for a single experiment were photographed in a single image and the conditions were the same in all cases. The images were taken at the same time for all the strips in a single image, using a portable photographic studio based on LED lights of 1100 lm and a color temperature of 6000-6500 K. The images were taken at a distance of 23 cm with a smartphone. The rear camera was used, with a maximum resolution of 12 megapixels (4032 × 3024 pixels). Autofocus was enabled, and the flashlight was turned off during the data acquisition procedure. Then, the images were converted to an 8-bit grey-scale format using the command Image > Type > 8-bit. The test and control lines were outlined using the rectangular selection tool, and the area under each peak was then numerically integrated using the ImageJ gel analysis toolbox.

Characterization of CNPs by Transmission Electron Microscopy
The structure of the CNPs in CB is composed of a system of condensed carbon aromatic rings that are deposited forming sheets of different sizes and alignment. These sheets are placed randomly through an axis and are attached to each other by π−π stacking, overlapping one another to form the primary unit of the structure, the primary particle or nodule [13][14][15][16][17][18]. Besides the nodules, CNPs in CB tend to form aggregates (85-500 nm) and agglomerates (1-100 µm) [13][14][15][16][17][18]. After the sonication procedure necessary to disperse the CNPs in water, the remaining superstructures are primarily the aggregates, as shown in Figure 3A. The agglomerates are attached by weaker forces that can be broken by the sonication. The mean diameter value was found to be 13 nm with a standard deviation of 6 ( Figure 3B). The TEM image shows the presence of aggregates without any regular structure that were, in most instances, larger than 200 nm. Additionally, AuNP-based lateral flow strips were performed as previously described by our research group [31,32], and included a conjugate pad in the configuration as a reservoir of AuNPs. In this instance, a final readout was achieved at 15 min from sample deposition.

Data Interpretation and Analysis
In order to avoid any bias in the images due to luminosity, all the strips for a single experiment were photographed in a single image and the conditions were the same in all cases. The images were taken at the same time for all the strips in a single image, using a portable photographic studio based on LED lights of 1100 lm and a color temperature of 6000-6500 K. The images were taken at a distance of 23 cm with a smartphone. The rear camera was used, with a maximum resolution of 12 megapixels (4032 × 3024 pixels). Autofocus was enabled, and the flashlight was turned off during the data acquisition procedure. Then, the images were converted to an 8-bit grey-scale format using the command Image > Type > 8-bit. The test and control lines were outlined using the rectangular selection tool, and the area under each peak was then numerically integrated using the ImageJ gel analysis toolbox.

Characterization of CNPs by Transmission Electron Microscopy
The structure of the CNPs in CB is composed of a system of condensed carbon aromatic rings that are deposited forming sheets of different sizes and alignment. These sheets are placed randomly through an axis and are attached to each other by  stacking, overlapping one another to form the primary unit of the structure, the primary particle or nodule [13][14][15][16][17][18]. Besides the nodules, CNPs in CB tend to form aggregates (85-500 nm) and agglomerates (1-100 μm) [13][14][15][16][17][18]. After the sonication procedure necessary to disperse the CNPs in water, the remaining superstructures are primarily the aggregates, as shown in Figure 3A. The agglomerates are attached by weaker forces that can be broken by the sonication. The mean diameter value was found to be 13 nm with a standard deviation of 6 ( Figure 3B). The TEM image shows the presence of aggregates without any regular structure that were, in most instances, larger than 200 nm.  Figure S1, Supplementary Materials). n = 700. Figure 4 shows the results of the end-point double-tagging PCR analyzed by gel electrophoresis. As can be observed, a unique band at the expected MW (527) was obtained. Further detail of the E. coli 16S ribosomal RNA gene (partial sequence, region from 1 to 530 nt) is shown in Supplementary Materials, including the position of the forward and reverse primer sets framing the region to be amplified (527 bp). As shown in Figure 4, the visual limit of detection can be defined at line 8, corresponding to 3.5 pg µL −1 .   Figure S1, Supplementary Materials). n = 700. Figure 4 shows the results of the end-point double-tagging PCR analyzed by gel electrophoresis. As can be observed, a unique band at the expected MW (527) was obtained. Further detail of the E. coli 16S ribosomal RNA gene (partial sequence, region from 1 to 530 nt) is shown in Supplementary Materials, including the position of the forward and reverse primer sets framing the region to be amplified (527 bp). As shown in Figure 4, the visual limit of detection can be defined at line 8, corresponding to 3.5 pg μL −1 .

Lateral Flow Assays Based on Carbon and Gold Nanoparticles as Signal-Generating Systems
The comparative study for the detection of E. coli by double-tagging PCR followed by an LFA based on different signal-generating systems was performed, as schematically shown in Figure 2. The total assay time was less than 15 min. The results of the tests can either be estimated with the naked eye (as shown in Figure 5, the upper panel for Av-CNP and the lower panel for streptAv-AuNP) or with the ImageJ software by processing the images. Figure 5 also shows the relative areas for each type of signal-generating system after processing the images. These areas were used for Figure 6. The visual LOD of the method was calculated by serial dilution of the DNA template, ranging from 0.0 to 7 pg μL −1 . The LFA achieved visual detection as low as 2.2 × 10 −2 pg μL −1 using CNPs and 8.4 × 10 −2 pg μL −1 using AuNPs. These LODs (calculated as the concentration corresponding to the last positive line visible to the naked eye, indicated by the eye icon in Figure 5) can be obtained without the assistance of any instrumentation but is prone to subjectivity.

Lateral Flow Assays Based on Carbon and Gold Nanoparticles as Signal-Generating Systems
The comparative study for the detection of E. coli by double-tagging PCR followed by an LFA based on different signal-generating systems was performed, as schematically shown in Figure 2. The total assay time was less than 15 min. The results of the tests can either be estimated with the naked eye (as shown in Figure 5, the upper panel for Av-CNP and the lower panel for streptAv-AuNP) or with the ImageJ software by processing the images. Figure 5 also shows the relative areas for each type of signal-generating system after processing the images. These areas were used for Figure 6. The visual LOD of the method was calculated by serial dilution of the DNA template, ranging from 0.0 to 7 pg µL −1 . The LFA achieved visual detection as low as 2.2 × 10 −2 pg µL −1 using CNPs and 8.4 × 10 −2 pg µL −1 using AuNPs. These LODs (calculated as the concentration corresponding to the last positive line visible to the naked eye, indicated by the eye icon in Figure 5) can be obtained without the assistance of any instrumentation but is prone to subjectivity.
To further analyze the strips and to avoid subjectivity, the ImageJ analysis tool was used to extract the line intensities from the images of the strips at different concentrations by using a smartphone. The area under the peak (both the intensities of the control and the test line for each system are shown in Figure 5) was then numerically integrated using the ImageJ analysis toolbox, and the intensity of the test line for each strip was fitted using non-linear regression (Sigmoidal dose-response variable slope/Sigmoidal 4 PL) (R 2 = 0.992 and 0.995 for CNPs and AuNPs, respectively); the results are shown in Figure 6A.
The semiquantitative LODs were determined by processing the relative areas for each type of signal-generating system after processing the images, as shown in Figure 6. The LODs are calculated as the last concentration, in which a positive area value is obtained, as shown in Figure 5, indicated by a photo icon.   To further analyze the strips and to avoid subjectivity, the ImageJ analysis tool was used to extract the line intensities from the images of the strips at different concentrations by using a smartphone. The area under the peak (both the intensities of the control and the test line for each system are shown in Figure 5) was then numerically integrated using the ImageJ analysis toolbox, and the intensity of the test line for each strip was fitted using non-linear regression (Sigmoidal dose-response variable slope/Sigmoidal 4 PL) (R 2 = 0.992 and 0.995 for CNPs and AuNPs, respectively); the results are shown in Figure 6A.
The semiquantitative LODs were determined by processing the relative areas for each type of signal-generating system after processing the images, as shown in Figure 6. The LODs are calculated as the last concentration, in which a positive area value is obtained, as shown in Figure 5, indicated by a photo icon.
The LODs obtained showed improved values compared to the visual LOD. In the case of CNPs, the LOD was found to be 6.0 10 −3 pg μL −1 (which corresponds to 10 CFU  To further analyze the strips and to avoid subjectivity, the ImageJ analysis tool was used to extract the line intensities from the images of the strips at different concentrations by using a smartphone. The area under the peak (both the intensities of the control and the test line for each system are shown in Figure 5) was then numerically integrated using the ImageJ analysis toolbox, and the intensity of the test line for each strip was fitted using non-linear regression (Sigmoidal dose-response variable slope/Sigmoidal 4 PL) (R 2 = 0.992 and 0.995 for CNPs and AuNPs, respectively); the results are shown in Figure 6A.
The semiquantitative LODs were determined by processing the relative areas for each type of signal-generating system after processing the images, as shown in Figure 6. The LODs are calculated as the last concentration, in which a positive area value is obtained, as shown in Figure 5, indicated by a photo icon.
The LODs obtained showed improved values compared to the visual LOD. In the case of CNPs, the LOD was found to be 6.0 10 −3 pg μL −1 (which corresponds to 10 CFU The LODs obtained showed improved values compared to the visual LOD. In the case of CNPs, the LOD was found to be 6.0 10 −3 pg µL −1 (which corresponds to 10 CFU mL −1 ), much better than the visual LOD of 2.2 10 −2 pg µL −1 (as shown in Figure 6B). Similarly, the semiquantitative LOD was 2.2 10 −2 for the AuNPs, while the visual LOD was found to be 8.4 10 −2 pg µL −1 .
Besides the improved LODs obtained with the CNPs, the valid analysis range was also found to be wider (from 0 to at least 7.09 pg µL −1 ), since valid tests (with a positive control line) were obtained in all instances, at least up to this concentration. As shown in Figure 5, a clear control line was obtained with CNPs at this concentration. Whereas, at 7.09 pg µL −1 , a non-valid test was observed by the AuNPs system since the control line was negative.
The stability and repeatability study of Av-CNPs is shown in Figure 7. This study was performed with the strips prepared on day 0 and stored at room temperature. Moreover, a batch of Av-CNPs was also prepared on day 0 and stored at 4 • C. This study was conducted for 35 days ( Figure 7C). The results show a mean relative intensity value of 4179, a standard deviation of 389, and a relative standard deviation (RSD) of 9.3% (as shown in Figure 7D). This error shows the repeatability study, including the construction of the strips, the deposition of the line, and the stability of the Av-CNPs. As shown in Figure 7, the Av-CNPs were able to produce a clear control line until day 35, demonstrating outstanding stability at least a month after preparation. Figure 5, a clear control line was obtained with CNPs at this concentration. Whereas, at 7.09 pg μL −1 , a non-valid test was observed by the AuNPs system since the control line was negative.
The stability and repeatability study of Av-CNPs is shown in Figure 7. This study was performed with the strips prepared on day 0 and stored at room temperature. Moreover, a batch of Av-CNPs was also prepared on day 0 and stored at 4 °C. This study was conducted for 35 days ( Figure 7C). The results show a mean relative intensity value of 4179, a standard deviation of 389, and a relative standard deviation (RSD) of 9.3% (as shown in Figure 7D). This error shows the repeatability study, including the construction of the strips, the deposition of the line, and the stability of the Av-CNPs. As shown in Figure 7, the Av-CNPs were able to produce a clear control line until day 35, demonstrating outstanding stability at least a month after preparation.

Conclusions
In this paper, the conjugation of avidin to carbon black nanoparticles and the design of LFAs for the detection of PCR products are presented in order to compare the performance of Av-CNP to the well-known commercial gold nanoparticles in optimal conditions for both cases.

Conclusions
In this paper, the conjugation of avidin to carbon black nanoparticles and the design of LFAs for the detection of PCR products are presented in order to compare the performance of Av-CNP to the well-known commercial gold nanoparticles in optimal conditions for both cases.
The results of the TEM characterization of the CNPs revealed the existence of aggregates and some primary particles following sonication. The sonication is necessary in order to dissolve the particles.
The conjugated CNPs showed good sensitivity and specificity for biotinylated amplicons, as confirmed also by the control line using an adsorbed biotinylated protein. Moreover, the conjugated CNPs showed non-detectable non-specific adsorption, since no signal was observed in the negative test, as in the case of the commercial streptAv-AuNP. The performance of Av-CNPs in this LFA format reveals promising features for use in other applications, enabling a rapid visual readout at an impressive LOD.
The comparative results of Av-CNPs and the commercial streptAv-AuNPs indicated improved LODs for Av-CNPs under the experimental conditions of this work, which can be attributed to the higher signal background contrasts, considering the intensity of the black color. All these features make the CNPs an attractive option to work with and to improve the sensitivity, cost, and simplicity of lateral flow tests.