Applications of surface enhanced Raman scattering (SERS) spectroscopy for detection of nucleic acids

5.1 Methods based on the attachment of a Raman reporter to the SERS-active surface

The first developed group of SERS DNA/RNA detection methods utilising selective hybridisation with “capture” complementary ssDNA/RNA strands were based on hybridisation-induced placing of a Raman reporter directly on the SERS-active surface. The first example of such SERS DNA detection was described in 1994 by Vo-Dinh and co-workers [33]. This group attached a Raman reporter (cresyl violet) to the 5′ end of ssDNA strands complementary to the sought-after (target) ssDNA, and then carried out hybridisation with the analysed ssDNA adsorbed on a nitrocellulose filter [33]. After the hybridisation process, the ssDNA which had not hybridised was removed by careful washing. Then, the DNA absorbed on the nitrocellulose (both that which did not hybridise and that which hybridised with DNA complementary to the target DNA with cresyl violet attached) was desorbed from the nitrocellulose in a 0.1 M NaOH solution. After neutralisation, the desorbed DNA was deposited on a SERS-active substrate and the SERS spectrum was measured. The presence in the measured SERS spectrum of strong Raman bands of cresyl violet suggests that the analysed ssDNA contained the sought-after sequence complementary to the sequence of the ssDNA to which the Raman reporter had been attached.

This general principle of DNA detection (deposition of a Raman reporter on the SERS-active surface as a result of DNA hybridisation) has been applied in many slightly different SERS sensors. For example, in 1998, the same group developed a SERS sensor for detecting a DNA fragment from the gag gene region of HIV1 [34]. At first, the analysed DNA was multiplied in a polymerase chain reaction (PCR) in such a way that the obtained (multiplied) DNA fragments were labelled with cresyl violet. Cresyl violet was introduced to the formed DNA by using a cresyl violet-labelled oligonucleotide primer for PCR [34]. In another process, the capture probe ssDNA, which was complementary to an internal sequence of the amplified target ssDNA, was bound to a polystyrene substrate. After hybridisation of the capture probe ssDNA with the cresyl violet-labelled product ssDNA (multiplied and modified analysed ssDNA), the polystyrene wells were carefully washed to remove non-specifically bonded ssDNA. The polystyrene substrate with attached DNA was then coated via thermal evaporation with a 10 nm layer of silver. Finally, when the target ssDNA was present in the analysed sample, the SERS signals from the attached cresyl violet are detected [34].

A significantly simpler SERS DNA sensor which utilises hybridisation-induced placing of a Raman reporter on the SERS-active substrate was achieved by immobilising capture ssDNA (complementary to the target ssDNA) on a SERS-active substrate [35]. Then, after proper modification of the analysed ssDNA (by attaching to it chains of, for example, rhodamine B), hybridisation with the immobilised capture ssDNA was carried out and the SERS signal measured. If the target ssDNA was present in the analysed sample, rhodamine B was attached to the SERS-active substrate and a strong SERS spectrum of this Raman reporter was measured [35]. Using such an approach, Allain et al. detected a fragment of the BRCA1 breast cancer susceptibility gen [35], [36]. Barhoumi and Halas noticed that, because the Raman scattering cross-section of adenine is so large, when the probe DNA sequence was adenine-free, adenine could be used as a Raman reporter [37]. Therefore, in this type of approach, there is no need to use any Raman reporter. To form an adenine-free probe ssDNA sequence general enough to detect any target ssDNA sequence, adenine was replaced in the probe ssDNA by 2–aminopurine [37]. This substitution preserved the same hybridisation characteristics of the substituted sequence [37]. Using such adenine-free probe ssDNA sequences, any adenine-containing DNA target oligonucleotide can be easily detected using SERS, due to the presence of the characteristic Raman band of adenine at 736 cm⁻¹ (see Figure 11) [37].

Figure 11:

Schemes showing DNA hybridisation between the thiolated probe DNA, with 2–aminopurine substituted for adenine, and the unlabelled target DNA; any adenine-containing DNA target oligonucleotide leads to the appearance in the measured SERS spectrum of a characteristic band of adenine at 736 cm⁻¹. Reprinted with permission from ref. [37]. Copyright 2010 American Chemical Society.

The next significant improvement of SERS DNA sensors based on the hybridisation-induced placing of a Raman reporter on a SERS-active surface was achieved by eliminating the need to modify the analysed DNA. In created what is known as a “sandwich-type” SERS DNA sensor, two strands of ssDNA complementary to two different fragments of the target ssDNA are used. One ssDNA, complementary to one fragment of the target DNA, (known as the capture DNA) is immobilised on the SERS-active substrate [38], [39]. Then, hybridizations with the analysed ssDNA and the reporter ssDNA (which is also complementary to a fragment of the target ssDNA and has a Raman reporter attached to it) are carried out. When the analysed sample contains the target ssDNA, the reporter ssDNA with the attached Raman reporter is immobilised on the SERS-active substrate (see Figure 12), which means that the Raman reporter is located in close proximity to the SERS-active substrate, and hence a strong SERS spectrum of the Raman reporter is observed [38], [39]. Such SERS DNA sensors can detect a remarkably low DNA concentration at ∼1 fM [38]. Analysis of how changing various construction details of such a sensor affects the intensity of the generated Raman signal showed that, although in general there is an increase in the intensity of the SERS signal when the distance between the Raman scatterer and the SERS-active surface decreases, for this type of sensor a greater intensity of the measured Raman signal is usually observed when the Raman reporter is farther away from the plasmonic substrate (see Figure 12) [40]. This is probably caused by a significant change in the hybridisation efficiency for the different structures of the sensor analysed, due to some steric hindrances [49].

Figure 12:

Scheme of sandwich-type SERS DNA sensors utilising capture DNA (complementary to one fragment of the target DNA) immobilised on a SERS-active substrate and reporter ssDNA (which is also complementary to a fragment of the target ssDNA and has a Raman reporter attached). Sensor in configuration A (Raman reporter “outside” the double-stranded DNA) and in configuration B (Raman reporter “inside” the double-stranded DNA). Reprinted from ref. [40]. CC BY license.

It should be emphasized that analogous sandwich-type sensors for DNA detection can also be based on fluorescence spectra [41], [42]. As mentioned above, the main advantage of SERS-based sensors over fluorescence-based sensors is that the recorded SERS spectrum consists of sharp bands, while fluorescence generates much broader bands. Therefore, it is much easier to eliminate band overlap in SERS experiments than in fluorescence experiments and to detect more target DNA/RNA strands in one SERS measurement than in one fluorescence measurement.

An interesting observation was made by Guselnikova et al. who, using principal component analysis, showed that it is possible to distinguish the SERS spectrum of capture ssDNA adsorbed on a SERS-active gold substrate from the SERS spectrum of dsDNA formed on a SERS substrate as a result of the hybridization of complementary capture and target ssDNAs [43]. This means that, in principle, the classical so-called sandwich-type SERS DNA sensors described above can be constructed without the use of a Raman reporter. This group also showed that it is possible to distinguish structures formed by two, not completely complementary ssDNA strands based on the obtained SERS spectra.

A very interesting type of SERS DNA/RNA sensors, whose operation is based on changing the distance between the Raman reporter and the surface of the SERS substrate, is what is known as hairpin sensors. Hairpin sensors are divided into two groups: on-off [44], [45], [46] and off-on [47], [48]. In the on-off hairpin sensors, developed first, the DNA probe sequence consists of a middle section that contains a sequence complementary to the target sequence to be detected, as well as two arms that have complementary sequences and thus form a hairpin loop configuration under normal conditions (see Figure 13) [44], [45], [46]. The hairpin loop configuration is designed such that the Raman reporter is in contact or close proximity (<1 nm) to the SERS-active surface, which makes it possible to induce a strong SERS effect of the Raman reporter (see Figure 13). Hybridisation with a complementary target ssDNA opens the hairpin loop and physically separates the Raman reporter from the SERS-active substrate. Since, as mentioned in Section 2, the SERS enhancement factor depends strongly on the distance, d, between the Raman reporter and the SERS-active surface (as d ⁻¹⁰ [12]), the hybridisation process leads to a strong decrease in the SERS effect and causes a quenching of the measured SERS signal after hybridisation [44], [45], [46]. Using this approach, Vo-Dinh et al. detected the gag gene sequence of the human immunodeficiency virus type 1 (HIV-1) [44] and single-nucleotide polymorphisms in breast cancer BRCA1 gene [46], Sha et al. detected RNA of the Hepatitis C virus (HCV) [45], whereas Kang et al. quantitatively determined let-7a miRNA in MCF-7 cell derived exosomes [49]. As a very active substrate in SERS spectroscopy, this last group used a close-packed and ordered array of gold octahedral nanoparticles; they achieved a broad linear range for this let-7a miRNA sensor (from 10 aM to 10 nM) and a low detection limit (5.3 aM), without using any signal amplification strategy [49].

Figure 13:

Schematic diagram of the concept of a hairpin on-off SERS sensor. (a) A SERS signal is observed when the hairpin is in the closed-state conformation, but diminishes in the open-state conformation. (b, c) SERS spectra from a plasmonics nanoprobe in the absence (b) and in the presence (c) of the target ssDNA.

A few years after developing the hairpin on-off SERS sensor, Vo-Dinh and co-workers developed a hairpin sensor having an off-on structure [47], [48]. In this approach, ssDNA (from which the “stem-loop” is formed) with an attached a Raman reporter at one end and a thiol moiety at the other end is immobilised onto a SERS-active substrate via a metal-sulphur bond. ssDNA serving as a “placeholder” strand is partially hybridised with ssDNA immobilised at the metal surface with an attached Raman reporter, keeping the Raman reporter away from the SERS-active surface (see Figure 14). In this configuration (i.e. in the absence of target ssDNA), the probe ssDNA is “open” and the sensor generates a very low SERS signal (the sensor is “off”). Upon exposure to the target ssDNA, the placeholder ssDNA strand leaves the SERS-active surface; this allows the stem-loop structure to “close” and moves the Raman reporter onto the SERS-active surface, yielding a strong SERS signal (the sensor is “on”). Using hairpin off-on SERS sensors, Vo-Dinh et al. detected the RSAD2 human gene [47] and a fragment of DNA formed from DENV 4 RNA of the dengue virus [48]. In both cases, a cyanine5 moiety was used as the Raman reporter [47], [48].

Figure 14:

Schematic diagram of the concept of a hairpin off-on SERS sensor. Reprinted with permission from ref. [47]. Copyright 2015 Elsevier Inc.

Some increase in the usability of a hairpin SERS DNA sensor is achieved by introducing magnetic properties into it. Lu et al. formed hairpin on-off SERS sensor by immobilizing ssDNA in a folded hairpin structure on core-satellite Fe₃O₄@SiO₂–Au nanostructures (see Figure 15) [50]. The ssDNA fragment used in this sensor contained a 23-base loop sequence to specifically capture the target tetracycline resistance gene (tetA) and a 7-base stem sequence on either side to keep the probe DNA in a hairpin structure at room temperature (see Figure 15) [50]. The detection limit of tetA in environmental samples was calculated to be 25 copies μL⁻¹ [50]. This limit of detection is comparable to that of the conventional, quantitative PCR technique, which was 15.8 copies μL⁻¹ in detecting Mycobacterium tuberculosis-specific DNA [50].

Figure 15:

Schematic illustration of the concept of a hairpin on-off SERS sensor having magnetic properties. DNA hairpin structures were immobilized on core-satellite Fe₃O₄@SiO₂–Au nanostructures. Reprinted with permission from ref. [50]. Copyright 2020 American Chemical Society.

The hairpin type strategy is also used to construct DNA sensors based on the measurement of the fluorescence spectrum [51], [52]. Similar to other SERS and fluorescence sensors with an analogous structure, sensors based on the measurement of SERS spectrum allow for better elimination of band overlap and thus allow for the detection of a larger number of target DNA/RNA strands in a single measurement.

5.2 Methods based on agglomeration of nanoparticles or their selective deposition via the links formed by nucleic acids

Another group of commonly used SERS sensors for the detection of nucleic acids having a given sequence are sensors where the selective hybridisation of the probe (capture) nucleic acid and the target nucleic acids leads to the formation of bonds (via double-stranded chains) that connect the nanoparticles to each other or bind the plasmonic nanoparticles to a macroscopic plasmonic substrate. Bringing plasmonic systems very close each other creates sites, in the narrow slits between the plasmonic structures, that are particularly active in SERS spectroscopy, and when such a phenomenon occurs, we observe a very strong SERS signal.

In the first developed SERS DNA sensor utilising DNA links between plasmonic structures, silver nanoparticles with a diameter of about 15 nm were attached to a smooth silver film (practically inactive in SERS spectroscopy) [53]. Linking DNA chains were seen to form between the silver film and the silver nanoparticles when the probe ssDNA was present in the analysed sample. The deposition of Ag nanoparticles on the Ag film resulted in the formation of a very SERS-active structure, and hence in a very large increase in the intensity of the SERS signal generated by the Raman reporter being used (5-((2-(and-3)-S-(acetylmercapto)succinoyl)amino) fluorescein), which had been deposited previously on the silver substrate (see Figure 16) [53]. A diagram of such a sensor is shown in Figure 16 [53]. A model 24-mer oligonucleotide was used as the target ssDNA [53]. The silver nanoparticles were functionalised with ssDNA complementary to one-half of the target ssDNA (shown in blue in Figure 16A), whereas the smooth SERS-inactive silver film was modified with ssDNA complementary to the remaining part of the target ssDNA sentence (shown in red in Figure 16A) [53]. The Ag-film modified with one capture ssDNA and the Raman reporter was then incubated with the target ssDNA and a sol of silver nanoparticles modified with the second capture ssDNA strand [53]. When both the capture ssDNA strands are complementary to the respective parts of target ssDNA, a sandwich structure is formed through sequence-selective hybridisation, and plasmonic coupling occurs between the silver nanoparticles and the silver film, creating electromagnetic SERS hot spots and a strong SERS Raman reporter signal [53]. When a non-complementary DNA strand was used, featureless spectra were obtained, which shows that the proposed method makes it possible to confirm the presence of DNA of a given sequence in the analysed sample [53].

Figure 16:

Detection of ssDNA based on selective deposition of plasmonic nanoparticles on a plasmonic substrate. (A) Schematic illustration of the SERS detection of ssDNA utilising the creation of DNA links between silver nanoparticles and a smooth silver film (virtually inactive in SERS spectroscopy). A target ssDNA strand (a′b′) is captured by the respective ssDNA fragments attached to the silver nanoparticles (probe-a) and the silver film (probe-b), resulting in the creation of a highly SERS-active structure. Target ssDNA detection is confirmed by recording a strong SERS signal from the surface-bound Raman reporter, F. (B) SERS spectra (vertically offset for clarity) obtained from: (a) an Ag film modified with a Raman reporter, probe-b ssDNA and a′b′ target ssDNA, (b) an Ag film modified with a Raman reporter, probe-b ssDNA, a′b′ target ssDNA and silver nanoparticles with attached probe-a ssDNA, and (c) an Ag film prepared as in (b) except for using a non-complementary target sequence in place of a′b′. (C) Representative AFM images of the surface of the sensor after the attachment of Ag nanoparticles: (a) 5 × 5 μm and (b) 1 × 1 μm. Arrows point to silver nanoparticles on the surface. Reprinted with permission from ref. [53]. Copyright 2007 American Chemical Society.

A SERS DNA sensor quite similar to the one described above, but based on the formation of aggregates of plasmonic nanoparticles in solution, was described by Qian et al. [54]. In this approach, two types of gold nanoparticles with an attached Raman reporter (such as malachite green) were used. One portion of gold nanoparticles was functionalised with capure-1 ssDNA complementary to one part of the target ssDNA, whereas the second part of the gold nanoparticles was modified with capture-2 ssDNA complementary to the remaining part of the target ssDNA strand [54]. Adding target ssDNA to the mixture of gold nanoparticles modified with capture-1 and capture-2 ssDNA caused an agglomeration of gold nanoparticles (see Figure 17). This led to a coupling of plasmons, the formation of electromagnetic SERS hot spots in the slits between gold nanostructures, and a strong SERS signal of the Raman reporter (in this case, malachite green) [54].

Figure 17:

Schematic illustration of the DNA-induced agglomeration of colloidal gold nanoparticles. The agglomeration permits long-range plasmonic coupling and a large increase in the SERS-activity of the nanostructure formed. Reprinted with permission from ref. [54]. Copyright 2008 American Chemical Society.

When different Raman reporters are used (or the constructed sensor contains well-defined and easily distinguishable places where the generated analytical signal indicates the presence of a specific ssDNA fragment), it is possible to detect various ssDNA fragments using just one sensor. For example, using a SERS DNA sensor of gold nanoparticles on a gold nanowire (a scheme of which is shown in Figure 18), Kang et al. managed to detect multiple target DNAs (DNA from Enterococcus faecium, Staphylococcus aureus, Stenotrophomonas maltophilia, and Vibrio vulnificus) in a single assay [55]. These groups formed gold nanowires on a c-sapphire substrate using a vapour transport method. The formed gold nanowires with a diameter of ca. 150 nm and a length of several dozen micrometers are highly single-crystalline without twins and have atomically smooth surfaces. Therefore, the observed SERS enhancement factor is very uniform along the whole Au nanorod. The surfaces of gold nanowires were modified with the thiolated probe ssDNAs and then were transferred using a nanomanipulator to a silicon substrate. Since the created nanorods are easily visible under an optical microscope, it is relatively easy to determine which nanorod (modified with ssDNA probe complementary to a given fragment of the DNA being searched for) is located and, in this way, using a Raman microscope the analytical signal characteristic of a given fragment of the DNA being sought can be selectively measured. In order to be able to detect four different ssDNA fragments, Kang et al. used gold nanorods modified with two different probe ssDNAs (see Figure 18) and gold nanoparticles modified with two different reporter ssDNAs (see Figure 18), with two different Raman reporters attached (TAMRA and cyanine 5 dyes) [55]. It should be emphasized that the sensor created by Kang et al. cannot be classified as a sensor that could play a useful practical role in its current form. In addition to the relatively complicated procedure of identifying different Au nanowires, the target DNA was amplified by PCR with universal and species-specific primers. PCR amplicons were then purified using a PCR purification kit.

Figure 18:

Detection of ssDNA based selective deposition of gold nanoparticles on gold nanowires. (a) Schematic illustration of a SERS DNA sensor based on what is known as a gold nanoparticle on a gold nanowire system. (b) SERS spectra obtained using such a sensor after DNA hybridisation with the complementary target ssDNA (blue spectrum), and prepared when the non-complementary target ssDNA is used in place of complementary target ssDNA (magenta spectrum). The inset shows UV–vis absorption spectra of the Au nanoparticles (green spectrum), Au nanowires (magenta spectrum), and gold nanoparticle on gold nanowire systems (blue spectrum). (c) SEM image of a typical gold nanoparticle on gold nanowire structure constructed by adding complementary target ssDNA (top) and a clean gold nanowire after the addition of non-complementary target ssDNA (bottom). Reprinted with permission from ref. [55]. Copyright 2010 American Chemical Society.

When the Raman reporters are properly selected (so as to provide non-overlapping spectra), the number of Raman reporters used in one experiment can be increased. For example, Li et al. constructed a SERS DNA sensor in which four Raman reporters (4-aminothiophenol, 4-chlorothiophenol, 4-hydroxythiophenol, and 5,5′-dithiolbis(2-nitrobenzoic acid)) were used simultaneously [56].

Using the “connection mechanism” described above, Li et al. constructed a sensor which permits the simultaneous acquisition of two different analytical signals: an SPR (surface plasmon resonance) signal and a SERS signal [57]. In this sensor, gold nanoparticles functionalised with a Raman reporter and ssDNA complementary to one part of the target micro-RNA were captured onto a smooth gold film – also used as an active part of the SPR sensor – functionalised with ssDNA complementary to another part of the target micro-RNA. This sensor was used to detect three micro-RNA (miRNA) fragments that are typical cervical cancer markers: miRNA21, miRNA124 and miRNA143 [57]. The attached gold nanoparticles greatly improved the SPR response due to the mass increase and larger refractive index changes. Moreover, attaching many Au nanoparticles to the gold surface of the SPR sensor also makes it possible to obtain a strong SERS signal. Three miRNA fragments were detected simultaneously in serum to verify the practicality, selectivity and sensitivity of the sensor. From an analysis of the SPR signals, all three miRNA could be detected simultaneously, with limits of detection (LODs) of 6.3 fM for miRNA21, 5.3 fM for miRNA124, and 4.6 fM for miRNA143, and with a wide dynamic response range of 500 pM-10 nM [57]. Based on the analysis of the SERS signal, higher sensitivity could be achieved with LODs as low as 1 fM for miRNA21, 0.8 fM for miRNA124 and 1.2 fM for miRNA143 [57].

An interesting type of SERS DNA sensor that uses both the hairpin-structured DNA sequences described in detail in the previous Section and plasmonic nanoparticles attached via the DNA chain formed is the sensor developed by Si et al. (see Figure 19) [58]. In the first stage of its construction, four thiolated hairpin-structured DNA sequences 1 (hp1-SH) having fragments complementary to four different cancer-associated miRNAs (marked in Figure 19 by: miR-21, miR-221, miR-133a and miR-1246) were attached onto one of the four Au/Ag alloy nanoparticles, which were immobilized on four different locations on a quartz sheet, thus forming a SERS sensor array with four sensing units (sensing unit for: miR-21, miR-221, miR-133a, and miR-1246). As mentioned above, detecting low expression levels of miRNA is extremely challenging, but is very important because miRNAs are biomarkers of various cancers. When the target miRNA is present in the analyzed sample, the hybridization between the related hairpin-structured 1 (hp1) and the target miRNA exposed the hidden toehold in the stem of the hp1, which allowed the corresponding SERS tags to be recognized. The SERS tags were prepared by modifying the Au/Ag alloy nanoparticles with 4−mercaptobenzonitrile (a Raman reporter) and with thiolated hairpin-structured DNA sequences 2 (hp2-SH) complementary to the four hp1 sentences used. These SERS tags with adsorbed hp2-SH displaced the target miRNA from the hp1 to form a stable hp1/hp2 duplex (see Figure 19). The released target miRNA induced the next cycle, which ultimately led to more SERS tags becoming anchored on the sensor array chip; they formed more SERS hot-spots, producing a strong SERS signal [58]. This SERS DNA sensor array offers a versatile and efficient platform for the simultaneous detection of multiple cancer-associated miRNAs with a detection limit of 0.15 pM.

Figure 19:

Scheme of a SERS DNA sensor developed by Si et al. that uses both hairpin-structured DNA sequences and the attachment of plasmonic nanoparticles via the DNA chain formed. Reprinted with permission from ref. [58]. Copyright 2020 American Chemical Society.

A SERS DNA sensor that uses two response amplification mechanisms was designed by Liu et al. [59]. One mechanism for increasing the sensor’s selectivity is nicking endonuclease amplification [59]. The enzyme used for this amplification (Nt.BstNBI) can recognize the ssDNA sequence of 5′-GAGTC-3′ in dsDNA, and cleave only one DNA strand at 4 bases away from the 3′-end of its recognition site; thus, it has very high specificity, since it requires a specific recognition site [59]. The second response amplification mechanism consists of silver nanocubes (such nanoparticles, which have sharp edges and corners, provide high-quality SERS hot spots, making it possible to obtain a stronger SERS signal) [59]. The biosensor developed demonstrated an impressive limit of detection of 3.1 fM for DNA species related to oral cancer (the ORAOV1 gene) [59]. Additionally, the platform exhibited high specificity, even against single-base mismatches, making it promising for early clinical diagnosis of oral cancer [49].

In general, the detection of nucleic acids by SERS measurements using the strategy described above, that is, linking various plasmonic nanostructures to plasmonic or non-plasmonic substrates by creating DNA strands as a result of hybridization, can be modified in many various ways. For example, the substrate to which the plasmonic nanoparticles are attached can be made of: gold nanoparticles on a porous anodic aluminum oxide [60], a film of gold nanoparticles on the surface of flat silica [61], or even TiO₂ nanoparticles [62]. The plasmonic nanoparticles attached by creating DNA strands can consist of, for example: gold nanoparticles modified by 4-mercaptobenzoic acid (MBA) and then covered with silver (Au^MBA@Ag) [60], silver nanoparticles modified by MBA and then covered with gold (Ag^MBA@Au) [61], or silver nanocubes [59], [62]. In order to increase sensitivity, various recycle amplification strategies for the target nucleic acid can also be applied [60], [61], [62]; for example, catalytic hairpin assembly amplification technology by means of rehybridization with thousands of hairpin probes to trigger amplification cycles [61], [62].

By using an RNA-induced aggregation of gold nanoparticles modified with the right capture ssDNA, Tan et al. were able to image miRNA-21 and tumor tissues in vivo in a mouse [63]. The distribution of a given miRNA can even be mapped in situ inside individual cells using a hybridization-induced agglomeration of modified plasmonic nanoparticles introduced to the cells (see Figure 20) [64]. The plasmonic nanoparticles were modified with molecules of the Raman reporter and with capture nucleic acids complementary to the target miRNA. The SERS detection of miRNA in different subtypes of breast cancer cells, SK-BR-3 and MCF-7 showed that the expression level of the target miRNA in the SK-BR-3 cells was five times higher than in the MCF-7 cells, which was confirmed by the result of a gene analysis [64].

Figure 20:

Schematic diagram for the: (a) hybridisation-induced conjugation of SERS-active gold nanoparticles modified with a Raman reporter and with ssDNA complementary to the target miRNA, and (b) cellular uptaking of miRANs, the formation of nanoclusters inside the cells, and the detection of miRNA expression level based on SERS mapping.

Also Wang et al. developed a SERS methodology for detecting miRNA inside cells [65]. In their approach, modified gold nanoparticles were self-assembled into trimers under the presence of target miRNA (miR-21), which generated a stable, enhanced electric field due to the coupling of the plasmons. The enhancement of the intensity of the electric field led to the formation of SERS hot spots, and increased the efficiency of the fluorescence, enabling the sensor to use two different analytical signals (SERS and fluorescence). Wang et al. used adenine as a reporter molecule, which simplified the labelling process [65]. This strategy successfully achieved accurate tracking and quantification of miR-21 in cancer cells, and showed good stability within the cells. The limit of detection values of the dual spectra were 3.58 pM (for the SERS signal) and 11.8 pM (for the fluorescence signal), with a relative standard deviation of less than 2.7 %.

It is also worth mentioning that the above-described aggregation of plasmonic nanoparticles induced by the created DNA strands can be used to detect more than nucleic acids. It can be also used, for example, to detect the caspase-3 protein [66], certain environmentally relevant biomolecules [67], or even molecules of individual proteins [68].

Techniques using DNA hybridization-induced formation of plasmonic nanoparticles aggregates are based on the creation of SERS hot spots in narrow gaps between plasmonic nanoobjects after aggregation and, consequently, the appearance of a strong SERS spectrum after DNA hybridization. As far as we know, the longest fragments of nucleic acids analysed by this SERS technique were from a dozen to several dozen nucleotides in length. Studying much longer nucleic acid fragments using these SERS techniques would generate problems with generated SERS enhancement factors. The optimal gap distance for the highest SERS enhancement is approximately 1 nm [69], [70], [71]. Gaps of a few nanometers are also very efficient in enhancing Raman scattering [69], [70], [71]. However, a further increase in the width of the gap leads to a significant reduction in the generated SERS signal [69], [70], [71] – therefore, in this technique, there is a limit to the thickness of the nucleic acid layer creating a “connection” between the plasmonic nanoparticles.

In addition to detection methods based on the agglomeration of plasmonic nanoparticles only, the agglomeration of plasmonic and magnetic nanoparticles can be also used to construct SERS DNA sensors. In these methods, SERS hot spots are created by bringing plasmonic nanoparticles very close each other (achieving plasmonic coupling); moreover, the magnetic-plasmonic aggregates that form can be concentrated by applying a permanent magnet to the sample being analysed. For example, Zhang et al. detected DNA derived from the West Nile virus by carrying out an agglomeration of silica shell-coated magnetic nanoparticles conjugated with capture oligonucleotides complementary to one part of the target DNA and gold nanoparticles conjugated with the Raman reporter (5,5′-dithiobis(succinimidy-2-nitrobenzoate)), as well as capture oligonucleotides (marked in Figure 21 as reporter oligonucleotides) complementary to another part of the target DNA (see Figure 21) [72]. The DNA hybridisation was conducted at pH = 7.4 by adding the target and control DNA sequences in a phosphate-buffered saline to a mixture of modified gold (plasmonic) and iron oxide (magnetic) nanoparticles [72]. After hybridisation, the agglomerates formed were precipitated from the solution using a small external magnet, and the SERS spectra of the precipitate were then measured. Zhang et al. found that the intensity of the strongest SERS band of the Raman reporter at 1,330 cm⁻¹ was proportional to the logarithm of the concentration of the target DNA sequence in the analysed sample [72].

Figure 21:

Schematic illustration of West Nile virus DNA capture and detection using silica shell-coated iron oxide magnetic nanoparticles (MNPs) conjugated with capture oligonucleotides complementary to one part of the target West Nile virus DNA and gold nanoparticles (GNPs) conjugated with a Raman label (5,5′-dithiobis(succinimidy-2-nitrobenzoate)) and reporter oligonucleotides complementary to another part of the target West Nile virus DNA. Reprinted with permission from ref. [72]. Copyright 2010 American Chemical Society.

When using Raman reporters that generate clearly different SERS spectra, a method based on an agglomeration of magnetic and plasmonic nanoparticles due to the creation of linking DNA chains can be also used to simultaneously detect many different fragments of nucleic acids (see Figure 22). For example, Wu et al. constructed a magnetically-assisted, SERS-based biosensor for the ultrasensitive, multiplex detection of three hepatocellular carcinoma-related miRNA biomarkers: miRNA-122, miRNA-223, and miRNA-21 [73]. The biosensor contained two types of nanoparticles: DNA-engineered fractal gold nanoparticles and Ag-coated Fe₃O₄ nanoparticles (Fe₃O₄@Ag) – see Figure 22. Wu et al. used three different DNA-engineered fractal gold nanoparticles containing three different Raman reporters: rhodamine 6G, crystal violet, and 4-aminothiophenol. Each type of gold nanoparticle containing a different type of a Raman reporter was modified with a different type of capture ssDNA, complementary to a part of the chain of three target miRNA (miRNA-122, miRNA-223, and miRNA-21). The Fe₃O₄@Ag nanoparticles were modified with three different ssDNAs complementary to three other fragments of the target miRNA chains – in this way, in one the multiplexed capture Fe₃O₄@Ag nanoparticle, three different capture ssDNA fragments were formed – see Figure 22 [73].

Figure 22:

Schematic illustration of a magnetically assisted SERS-based biosensor for multiplex detection of three miRNA biomarkers: miRNA-122, miRNA-223, and miRNA-21. (a) Schematic processes of synthesizing rhodamine 6G, crystal violet and 4-aminothiophenol encoded fractal Au nanoparticles, used as SERS tags. (b) Design and synthesis of the capture magnetic Fe₃O₄@Ag substrate. (c) Procedure for detecting the multiple miRNAs, based on linking magnetic and plasmonic nanoparticles. Reprinted with permission from ref. [73]. Copyright 2021 American Chemical Society.

The miRNA detection using the sensor developed by Wu et al. was performed in the following steps – see Figure 22 [73]. Firstly, Fe₃O₄@Ag nanoparticles functionalised with three different capture ssDNAs were cultured with a complex sample that contained the target miRNAs, irrelevant miRNAs, and other interferences. During this stage, the target miRNAs were specifically bound to capture ssDNA-modified Fe₃O₄@Ag nanoparticles due to the hybridization combination. Subsequently, the Fe₃O₄@Ag nanoparticles with attached target miRNA were magnetically separated from the complex sample. Next, three different fractal gold nanoparticles incorporating three different Raman reporters corresponding to different target miRNAs were added, and after subsequent hybridization, the modified Au nanoparticles were attached to the modified Fe₃O₄@Ag nanoparticles. Finally, after another magnetic separation, the SERS signal of the resulting agglomerates was measured. The limits of detection of these three miRNAs in human serum were 349 aM for miRNA-122, 374 aM for miRNA-223, and 311 aM for miRNA-21 [73].

Magnetic separation has been also applied in many other SERS sensors designed to detect nucleic acids. To detect, for example: (i) miRNA-122, with a wide linear range of from 10 aM to 10 pM and a low detection limit of 6.82 aM [74]; (ii) microRNA-155 (which regulates a variety of biological processes, including promoting the miRNA inhibition of breast cancer invasion and metastasis), with a detection limit of 36.7 fM [75] or even 1.45 fM [76] and a wide linear range (100 fM–5 nM) [75], [76]; and (iii) circulating miRNAs (such as miRNA-499), with a detection limit of 0.37 fM and a linear range of between 1 fM and 10 nM [77].

A similar SERS DNA sensor for detecting the ssDNA associated with the BRAF V600E mutation was constructed from gold-coated magnetic nanoparticles with attached 6-mercaptopyridine-3-carboxylic acid (MPCA) as an internal reference and silver nanoparticles with an attached Raman reporter (4-mercaptobenzonic acid, 4-MBA) [78]. The gold-coated magnetic nanoparticles were modified with ssDNA complementary to one part of the target ssDNA, whereas the silver nanoparticles were functionalised with ssDNA complementary to another part of the target ssDNA. The main difference between this sensor and the standard one described above was the addition of a ligase enzyme (DNA ligase can join together two fragments of nucleic acids by forming phosphodiester bonds) (see Figure 23) [78]. In the presence of the proper BRAF mutation DNA, the ligase enzyme induced the formation of a respective chemical bond, and these two nanoparticles are covalently linked, which brought the silver nanoparticle close to the surface of the magnetic particle (see Figure 23). Due to the aggregation of nanoparticles, the intensity ratio of the 4-MBA/MPCA increased linearly in the 1–100 fmol range of the matched DNA (BRAF gene with V600E mutation). Different ratios of matched DNA in the background of a large number of the single-base mismatched DNA (BRAF normal) were used to mimic real samples, and the intensity ratios of the 4-MBA/MPCA were linear in the 0.02–1 % range of the matched DNA/mismatched DNA [78].

Figure 23:

Schematic graph showing the process of hybridisation of DNA in a SERS DNA sensor with two different Raman reporters. The ligase enzyme leads to the formation of a chemical bond between the BRAF probes 1 and 2 chains in the presence of the target DNA. Reprinted with permission from ref. [78]. Copyright 2020 American Chemical Society.

5.3 Methods based on hybridisation-induced rearrangement of linking and protecting alkanethiol chains attached to SERS-active surfaces

In addition to the above approaches to DNA/RNA detection using SERS spectroscopy, whose principals were developed over a decade ago and are currently only being further developed, in 2019 Kowalczyk et al. proposed a completely new approach to DNA detection based on SERS measurements [79]. When determining the structure of various layers formed on SERS-active metals from capture ssDNA and 6-mercaptohexan-1-ol (such modified SERS substrates are often used in standard SERS DNA sensors), this group observed that hybridisation with target ssDNA complementary to the immobilised capture ssDNA induces a significant change in the conformation of chains of chemisorbed ω-substituted alkanetiols (6-mercaptohexan-1-ol and the alkanethiol linking moiety through which the capture ssDNA is attached to the metal surface) (see Figure 24) [79]. These structural changes can be easily determined from the respective SERS spectra, especially when there are no adenine moieties in a close proximity to the metal surface (see Figure 25) [79], and therefore, Kowalczyk et al. decided to attempt to detect circulating tumor DNA fragments characteristic for the BRAF mutation (c.1799T>A) using this type of SERS DNA sensor [79]. The detection was based on the conformation change of the chemisorbed ω-substituted alkanetiols (gauche → trans transformation) during the DNA hybridisation [79]. This SERS DNA sensor has been tested on clinical samples (from patients with melanoma or thyroid cancer who have the BRAF mutation, and from patients without the BRAF mutation). It was shown that this SERS DNA sensor is characterised by a low detection limit, at a level of pg μL⁻¹, a wide analytical range from 6.75 pg μL⁻¹ to 67.5 ng μL⁻¹, and high selectivity [79]. It should be emphasized, however, that these analyses were not carried out using directly tissues obtained from oncological patients during standard therapies, but DNA isolated from such tissues using respective DNA purification kits. Further experiments (using only DNA samples obtained from commercial suppliers) showed that SERS-active silver substrates are significantly more promising for the SERS observation of such hybridisation-induced rearrangements than the gold substrates originally used [80].

Figure 24:

Scheme of the rearrangement of chemisorbed 6-mercaptohexan-1-ol and an alkanethiolate linking moiety via which the capture ssDNA is attached to the SERS-active surface due to DNA hybridisation. Reprinted from ref. [80]. CC BY-NC-ND license.

Figure 25:

Detection of ssDNA based on DNA hybridisation-induced rearrangement of linking and protecting alkanethiol chains. (A) Scheme of a SERS sensor based on spectral changes of the linking layer and 6-mercaptohexan-1-ol (marked as MCH in the drawing) induced by DNA hybridisation. (B) Possible structures of Au–S–C–C chain: trans (T) conformation and gauche (G) conformation. Reprinted with permission from ref. [79]. Copyright 2019 Elsevier.

The DNA identification based on DNA hybridization-induced rearrangement described above and some other SERS measurements described in the previous parts of this work were performed on nanostructured plasmonic surfaces (either nanostructured surfaces of plasmonic metals or nanostructured surfaces of other materials coated with a layer of a plasmonic metal) – example SEM image of a surface of such SERS substrate is shown in Figure 26. As already shown for this type of SERS substrates, the dominant part of the SERS signal comes from molecules adsorbed on SERS hot spots [81], [82]. However, if very large molecules are adsorbed, they are not able to occupy all SERS hot spots – the larger the molecule is adsorbed, the smaller the ratio of SERS hot spots is occupied [83]. This effect will undoubtedly make the study of very long chains of nucleic acid very difficult.

Figure 26:

SEM image of a hetero-epitaxial GaN layer after photoetching and subsequent sputtering of silver – an example SERS substrate used for DNA hybridisation-induced rearrangement SERS measurements. Reprinted from ref. [80]. CC BY-NC-ND license.

5.4 Methods utilising DNA hydrogels

Interesting materials that significantly change their properties in the presence of certain nucleic acids are DNA hydrogels. After the introduction of the proper target nucleic acid, the DNA hydrogel disintegrates, and then, for example, certain nanostructures are able to easily penetrate through it. Using this approach, Si et al. developed arrays of miRNA-responsive DNA hydrogel-based SERS sensors, for nine miRNA markers of various cancers: miR-21 (breast, pancreatic, lung, ovarian and liver cancers, and leukemia), miR-221 (pancreatic, liver and thyroid cancers, and glioblastoma), miR-224 (liver cancer), miR-205 (lung cancer), miR-155 (breast and pancreatic cancers, and leukemia), miR-141 (ovarian cancer), miR-25 (colon and ovarian cancers), miR-18 (liver cancer), and miR-183 (colon cancer) [84]. At first, in the layer of the plasmonic AuAg alloy nanoparticles immobilised at the surface of the sensor, streptavidin was attached. The layer of the AuAg nanoparticles with attached streptavidin was then protected with a thin layer of DNA hydrogel (see Figure 27). Next, the plasmonic nanoparticles (and in this case the AuAg alloy nanoparticles as well) modified with biotin and a Raman reporter (4-mercaptobenzonitrile) were deposited on the DNA hydrogel [84]. The plasmonic nanoparticles could not pass through the hydrogel and bound with the streptavidin-modified sensor surface, so at this stage only a very weak SERS signals was observed. After the introduction of the target miRNA, the DNA hydrogel was disintegrated so that the plasmonic nanoparticles modified with the Raman reporter were able to pass through the hydrogel and to be captured on the streptavidin-modified detection surface. This resulted in strong SERS signals and the detection of the target miRNA [84].

Figure 27:

Schematic illustration of preparing and applying a target miRNA-responsive DNA hydrogel-based SERS sensor. The array formed makes it possible to measure multiple target miRNAs in one sample. Reprinted with permission from ref. [84]. Copyright 2020 American Chemical Society.