Interplay of graphene–DNA interactions: Unveiling sensing potential of graphene materials

Abstract

Graphene-based materials and DNA probes/nanostructures have emerged as building blocks for constructing powerful biosensors. Graphene-based materials possess exceptional properties, including two-dimensional atomically flat basal planes for biomolecule binding. DNA probes serve as excellent selective probes, exhibiting specific recognition capabilities toward diverse target analytes. Meanwhile, DNA nanostructures function as placement scaffolds, enabling the precise organization of molecular species at nanoscale and the positioning of complex biomolecular assays. The interplay of DNA probes/nanostructures and graphene-based materials has fostered the creation of intricate hybrid materials with user-defined architectures. This advancement has resulted in significant progress in developing novel biosensors for detecting DNA, RNA, small molecules, and proteins, as well as for DNA sequencing. Consequently, a profound understanding of the interactions between DNA and graphene-based materials is key to developing these biological devices. In this review, we systematically discussed the current comprehension of the interaction between DNA probes and graphene-based materials, and elucidated the latest advancements in DNA probe–graphene-based biosensors. Additionally, we concisely summarized recent research endeavors involving the deposition of DNA nanostructures on graphene-based materials and explored imminent biosensing applications by seamlessly integrating DNA nanostructures with graphene-based materials. Finally, we delineated the primary challenges and provided prospective insights into this rapidly developing field. We envision that this review will aid researchers in understanding the interactions between DNA and graphene-based materials, gaining deeper insight into the biosensing mechanisms of DNA–graphene-based biosensors, and designing novel biosensors for desired applications.

I. INTRODUCTION

Graphene is an allotrope of carbon consisting of a single layer of atoms arranged in a hexagonal lattice structure.¹ Due to its extraordinary electrical, optical, mechanical, and chemical properties, graphene has become a research hotspot over the past decades.^2–5 It exhibits great potential in diverse applications, including flexible electronics, supercapacitors, optoelectronic devices, and biosensors.^6–10 In the realm of biosensing, graphene, and its derivatives, graphene oxide (GO) and reduced graphene oxide (rGO) have garnered particular interest due to several advantages they offer: The excellent fluorescence quenching ability enables the fabrication of a wide range of optical biosensors.^11–14 Good biocompatibility renders them suitable for in vivo biosensing applications.¹⁵ The large surface area provides ample binding sites for biomolecules, enhancing the sensitivity of biosensors.¹⁶ In recent years, the binding of DNA probes on graphene-based materials has enabled the construction of more complex hybrid materials with user-defined architectures, advancing the selectivity and sensitivity of biosensors for complex biomolecular assays.

DNA probes have been designed to exhibit exceptional sensitivity and specificity toward their target molecules, including DNA, RNA, proteins, and small molecules. In the context of biosensors, DNA probes can specially be engineered as DNA aptamers to recognize and bind to molecules of interest. Upon binding, DNA aptamers undergo conformational changes that enable signal transduction and detection.^17,18 DNA probes possess a range of unique features, including high binding affinity, selectivity, specificity, sequence programmability, excellent chemical stability, good biocompatibility, and well-developed chemistry for further functionalization.¹⁷ Leveraging these features, DNA probes have been successfully employed in the design of numerous biosensors for bioanalysis and disease diagnosis. The distinctive binding interaction between DNA and graphene-based materials, such as graphene and GO, holds particular significance. This interaction allows for the anchoring of DNA probes onto the surface of these materials, resulting in the creation of novel hybrid nanomaterials, leading to the development of highly sensitive and selective biosensing platforms.^16,19–21

In addition, DNA nanostructures^22–24 can incorporate further functionalities on graphene materials as placement scaffolds and chemical adapters. Over the past decades, structural DNA nanotechnology has undergone tremendous growth.^24–26 One prominent technique in this field is DNA origami, which entails folding a long single-stranded DNA (ssDNA) scaffold through hybridization with hundreds of short synthetic “staple strands.” This technique enables the precise programming of complex one-dimensional (1D), two-dimensional (2D), and three-dimensional (3D) shapes with nanoscale features.^22,27,28 These versatile DNA nanostructures offer advances for the precise spatial organization of various DNA-linked nanomaterials, such as target DNA, proteins, quantum dots, and metal nanoparticles (NPs).^29–32 Using DNA nanostructures to precisely pattern species on graphene-based materials significantly expand their structural and functional diversity, allowing for the placement of complex biomolecular assays at predetermined positions on the surface of graphene-based materials.^14,33,34

Despite the potential of utilizing DNA probes and DNA nanostructures in graphene-based biosensing, there remain significant challenges to overcome. Specifically, the complicated interactions between DNA and graphene-based materials are not yet fully comprehended, including issues related to DNA probe adsorption and desorption mechanisms, as well as the structural rearrangement of DNA nanostructures when deposited on graphene-based materials.^35–37 To tackle these challenges and thus unveil the DNA-empowered graphene-based materials in biosensing application, both theoretical and experimental studies have been undertaken to explore the fundamental nature of interactions between DNA and graphene-based materials, and the possibilities of manipulating these interactions for biosensing purposes. In this perspective, we have conducted a comprehensive review of the current knowledge regarding the interactions between DNA probes and graphene-based materials. We also provided an overview of the studies focusing on the interplay of DNA nanostructures and graphene-based materials. Furthermore, we highlighted how these understandings have pushed forward the integration of DNA probes and DNA nanostructures with graphene-based materials for various biosensing applications. Here, DNA probes specifically encompass single-stranded DNA (ssDNA), double-stranded DNA (dsDNA) and DNA aptamers, etc. DNA nanostructures specifically include DNA origami and DNA tetrahedron, etc.

II. GRAPHENE-BASED MATERIALS

Graphite possesses a layered structure where carbon atoms are arranged in a hexagonal pattern.³⁸ Each carbon atom is covalently bonded to three neighboring carbon atoms, forming strong sigma bonds within the layers. The layers are stacked in an AB sequence and are held together by weak van der Walls (vdW) forces, generated by delocalized π-orbitals [Fig. 1(a)]. Graphite exhibits anisotropy,³⁹ with its highest electrical and thermal conductivity occurring within the layers. Conversely, perpendicular to the layers, the weak vdW forces lead to lower conductivity. These properties make graphite widely used in electrodes, biosensing, conductive coatings, heat sinks, and thermal interface materials. The anisotropy of graphite enables the carbon layers to slide over each other with minimal resistance, facilitated by the weak vdW forces between the layers.⁴⁰ This property makes it an excellent solid lubricant in machinery and automotive components. Furthermore, while graphite is chemically stable and resistant to most chemical reactions,⁴¹ the weak vdW forces between adjacent layers enable chemical intercalation reactions by allowing the intercalate (such as atoms, ions, small molecules, and even polymers) to insert between the layers. This process creates modified graphene structures, expanding their applications in energy storage devices, sensors, catalysis, and electronic devices.^42,43

FIG. 1.

(a) The crystal structure of graphite. The carbon atoms are arranged in a honeycomb lattice with a bond length of 1.42 Å, and the distance between adjacent layers is 3.35 Å. There are four atoms per unit cell, namely, A, A’, B, and B’. Reproduced with permission from Chung, J. Mater. Sci. 37, 1475–1489 (2002). Copyright 2002 Springer Nature. (b) Structure of the pristine graphene with sp² hybridized carbon atoms. (c) and (d) Structures of graphene oxide (GO) and reduced graphene oxide (RGO). Reproduced with permission from P. Suvarnaphaet and S. Pechprasarn, Sensors 17(10), 2161 (2017). Copyright 2017 MDPI.

The layers in graphite are referred to as graphene layers, which are 2D carbon allotrope consisting of a single layer of carbon atoms arranged in a hexagonal lattice.¹ Each carbon atom is covalently bonded to three neighboring carbon atoms, forming strong sp²-hybridized bonds [Fig. 1(b)].⁴⁴ Graphene’s unique structure gives it exceptional mechanical strength, flexibility, optical transparency, electrical, and thermal conductivity.⁴⁵ As a result, it has sparked tremendous interest and holds great potential for a wide range of applications.^46–48 Particularly, graphene-based materials are highly promising for biosensing applications. Its ability to quench fluorescence makes it well-suited for optical sensing.²⁰ When target molecules interact with graphene surface, it induces alterations in the optical properties, enabling the detection of their presence and concentration.^49–52 Moreover, graphene’s exceptional electrical conductivity, high surface-to-volume ratio, and biocompatibility make it an ideal material for constructing sensitive electrical biosensors. Graphene has been utilized in a widely adopted design as a conducting channel in a graphene field-effect transistor (GFET).⁵³ As target molecules bind to the graphene surface, there is a consequential change in electrical properties such as conductivity, capacitance, or gate voltage, facilitating the measurement and correlation of these changes with target concentration.^54–56 These biosensors hold immense potential in biomedical diagnostics, environmental monitoring, food safety, and other fields necessitating rapid and precise sensing capabilities.^12,57–59

GO is typically prepared by chemically exfoliating graphite under strong acidic and oxidative conditions.^60,61 Unlike pristine graphene, which composed of sp²-hybridized carbons only, GO possesses a variety of oxygen-containing functional groups on its basal plane and edges, including carboxyls, hydroxyls, esters, and epoxides [Fig. 1(c)].^44,62,63 The presence of oxygen functional groups, especially carboxyl groups, renders GO negatively charged and allows it to be electrostatically stabilized as a colloidal suspension in water and certain organic solvents.⁶⁴ In addition, the oxygen functional groups in GO provide sites for chemical modification and functionalization, allowing the incorporation of specific molecules or functional groups, such as antibodies, enzymes, and DNA probes, for biosensing applications.^44,65 When specific biomolecules bind to GO, changes in the electrical, optical, or electrochemical signals can be induced, enabling the detection of biomarkers, pathogens, or DNA sequences.^16,66,67 rGO is derived from GO through a reduction process that removes a substantial portion of the oxygen functional groups [Fig. 1(d)].^44,68 The reduction restores the sp²-hybridized carbon atoms in the graphene lattice, resulting in improved electrical conductivity and broader utilization in electronic devices such as transistors, electrodes, and supercapacitors.^69,70 Additionally, the large surface area and chemical reactivity of rGO make it well-suited for sensing applications.^52,71–73

III. DNA-BASED MATERIALS

The basic chemical formula of DNA is well established. The backbone of ssDNA consists of alternating phosphate groups and deoxyribose sugar, joined together in regular 3’ 5’ phosphate di-ester linkages. Each sugar is connected to a nucleobase. DNA commonly contains four different kinds of bases: adenine (A) and guanine (G) are purines, consisting of fused five- and six-membered heterocyclic compounds, while cytosine (C) and thymine (T) are pyrimidines, consisting of six-membered rings. In dsDNA, hydrogen bonds form between the bases of the two strands.^74,75 A is typically paired with T through two hydrogen bonds, and G is paired with C through three hydrogen bonds [Fig. 2(a)].

FIG. 2.

(a) DNA helix is unfolded to illustrate the sugar–phosphate backbones and the base-pair rungs of DNA. These backbones exhibit opposing orientations. The 3’ and 5’ ends denote the orientation of the carbon atoms on the sugar rings. Each base pair consists of one pyrimidine base, either T or C, and one purine base, either A or G, linked by hydrogen bonds (represented by dotted lines). (b) The general process of SELEX entails multiple cycles of target incubation with a nucleic acid library, target-binding strand separation, PCR amplification, and ssDNA generation to regenerate a new library for another iteration of this process. Reproduced with permission from Yu et al., Angew. Chem., Int. Ed. 60(31), 16800–16823 (2021). Copyright 2021 John Wiley and Sons. (c) DNA origami shapes. From left to right: Smiley face shape. Scale bar: 50 nm. 12-tooth gear. Scale bar: 20 nm. Regular tetrahedron. Scale bar: 20 nm. Closed-ring oligomer. Ring thickness: 56 nm. Self-limiting hierarchical assembly of closed polyhedra. Scale bar, 50 nm. Reproduced with permissions from P. W. K. Rothemund, Nature 440(7082), 297–302 (2006). Copyright 2006 Springer Nature; Dietz et al., Science 325(5941), 725–730 (2009). Copyright 2009 AAAS; Hyungmin et al., ACS Nano 13, 2083–2093 (2019). Copyright 2019 American Chemical Society; and Wagenbauer et al., Nature 552(7683), 78–83 (2017). Copyright 2017 Springer Nature. (d) DNA tiles assembly. Reproduced with permission from He et al., Nature 452(7184), 198–201 (2008). Copyright 2008 Springer Nature. (e) DNA brick assembly. Reproduced with permission from Shen et al., Nat. Mater. 20(5), 683–690 (2021). Copyright 2021 Springer Nature. (f) DNA lattice assembly. Scale bars: 200 nm. Reproduced with permission from Tang et al., J. Am. Chem. Soc. 145(25), 13858–13868 (2023). Copyright 2023 American Chemical Society.

DNA aptamers are short, single-stranded DNA oligonucleotides that possess diverse 3D conformations and are capable of binding to specific target molecules with high affinity and specificity.⁷⁶ They can be generated through an in vitro technique, systematic evolution of ligands by exponential enrichment (SELEX) [Fig. 2(b)].^77,78 In the SELEX process, a large library of random DNA sequences undergoes iterative selection and amplification steps to isolate DNA aptamers that have strong binding affinity to target molecules, including nucleic acids, peptides, small molecules, proteins, and even complex structures like cells or tissues.^79,80 As synthetic biorecognition elements, aptamers offer compelling advantages over naturally derived antibodies.⁸¹ They can be easily synthesized and modified with various functional groups or conjugated with other molecules, enabling efficient immobilization and the generation of electrochemical, electronic, or optical signals, respectively.⁸² Additionally, aptamers exhibit superior thermal and chemical stability, allowing for recovering their binding affinity after thermal denaturation, and remaining stable across a wide range of pH and ionic strength.⁸³ Moreover, the in vitro technology SELEX allows for the specific selection of aptamers against toxic and non-immunogenic targets.⁸⁴ Therefore, DNA aptamers have been adopted widely in fundamental and translational research regarding diagnostics and therapeutics, addressing limitations associated with antibodies.^81,84 They have been successfully integrated into electronic and optical sensors as molecular probes for target detection and quantification, as well as for selective drug delivery or therapeutic agent targeting to specific cells or tissues.^16–18

DNA nanostructures involve the assembly of ssDNA into complex but precisely controlled architectures.²² Several methods were commonly employed for synthesizing DNA nanostructures: (1) DNA Origami: A long ssDNA scaffold with hundreds of nucleotides is folded by a set of short “staple” strands. The scaffold strand serves as the backbone, while the staple strands hybridize with specific regions on the scaffold to guide its folding into the desired shape [Fig. 2(c)].^22,85–89 (2) Tile-based Assembly: Short DNA strands called tiles are designed with sticky ends that can hybridize with adjacent tiles, leading to the self-assembly of intricate nanostructures [Fig. 2(d)].^23,27 (3) Single-stranded tiles (SSTs): SSTs have the remarkable ability to self-assemble into 2D or 3D shapes without the need of a scaffold. SSTs interact with each other through complementary domains, analogous to “Lego” bricks, allowing for the construction of intricate and complex architectures [Fig. 2(e)].^28,90,91 (4) DNA Lattice Assembly: DNA strands are designed with specific sequences that can form a lattice structure through base pairing. The lattice can be designed with specific periodicity and symmetry, enabling the assembly of DNA nanostructures with controlled arrangements [Fig. 2(f)].^92–95 Over the past decades, numerous reports have show-cased successful synthesis of intricate 2D and 3D DNA nanostructures with exceptional programmability, nanoscale addressability, biocompatibility, and versatility in terms of shape, size, and functionality.^27,96–98 DNA nanostructures possess a myriad of desirable properties, rendering them incredibly appealing for an extensive array of applications, including biosensing, nanoelectronics, photonics, drug delivery, and tissue engineering.^26,99

As graphene-based materials and DNA both hold promising potentials on biosensing, by interplay of DNA and graphene-based materials, we can unveil DNA-empowered graphene-based materials in biosensing applications. In Secs. IV–VI, we will review the current knowledge regarding the interactions between DNA and graphene-based materials, including issues related to DNA probes adsorption and desorption mechanisms, as well as the structural rearrangement of DNA nanostructures when deposited on graphene-based materials.^35–37 We will compare these interactions by the binding energy of nucleobases (Table I) on the surface of graphene-based materials using different theoretical or experimental methods, as well as single-molecule force spectroscopy (SMFS) measurements of binding affinity between DNA homopolymers/heteropolymers and graphite (Table II).

TABLE I.

Summary of the binding energies of nucleobases on the surface of graphene-based materials using different theoretical or experimental methods.

Methods	A	T	G	C	U	Energy unit	Adsorption substrates	Results	References
constrained local density approximation (cLDA)	0.49	0.49	0.61	0.49	0.44	eV	Graphene	G > A ≈ T ≈ C > U	100
MP2	0.94	0.83	1.07	0.80	0.74	eV	Graphene	G > A ≈ T ≈ C > U
B97-D/TZV(d,p)	−14.5	−15.6	−18.9	−14.1	−13.5	kcal mol−1	C24H12	G >A > T > C > U	101
B97-D/TZV(d,p)	−19.8	−18.4	−24.1	−18.4	−15.8	kcal mol−1	C54H18	G > A > T ≈ C > U
B97-D/TZV(d,p)	−20.7	−19.4	−25.2	−19.2	−16.7	kcal mol−1	C96H24	G > A > T ≥ C > U
B97-D/TZV(d,p)	−20.9	−19.7	−25.7	−19.3	−17.0	kcal mol−1	C150H30	G > A > T ≥ C > U
Additive simulations	−8.51	−7.73	−8.93	−6.86	−6.83	kcal mol−1	Monolayer graphene	G > A > T > C > U	102
Drude simulations	−8.61	−8.27	−9.51	−7.87	−7.00	kcal mol−1	Monolayer graphene	G > A > T > C > U
Additive simulations	−9.40	−7.97	−9.68	−7.03	−7.35	kcal mol−1	4-layer graphite	G > A > T > U > C
Drude simulations	−8.35	−7.66	−9.85	−5.96	−7.03	kcal mol−1	4-layer graphite	G > A > T > C > U
Molecular mechanics (MM) (Amber FF)	−14.80	−14.75	−15.10	−7.93		kcal mol−1	Graphene	G > A ≈ T > C	103
sTS	0.829	0.742	0.959	0.724	0.664	eV	Graphene	G > A > T > C > U	104
TS	0.849	0.763	0.986	0.745	0.682	eV	Graphene	G > A > T > C > U
DFT-D2	0.636	0.583	0.770	0.573	0.515	eV	Graphene	G > A > T > C > U
DFT-D3	0.618	0.570	0.733	0.567	0.512	eV	Graphene	G > A > T ≈ C > U
vdW-DF	0.637	0.607	0.750	0.582	0.543	eV	Graphene	G > A > T > C > U
vdW-DF2	0.594	0.558	0.717	0.546	0.501	eV	Graphene	G > A > T ≈ C > U
LDA	0.55	0.54	0.72	0.56		eV	Graphene	G > C > A > T	105
Perfect Bayesian Equilibrium (PBE)	0.06	0.08	0.14	0.13		eV	Graphene	G > C > T > A
PBE + vdW	1.00	0.95	1.18	0.93		eV	Graphene	G > A > T > C
First-principles DFT and the vdW-DF2	0.59	0.57	0.70	0.52	0.50	eV	Graphene	G > A > T > C > U	106
DFT-D	−20.31	−18.40	−22.49	−18.94	−16.34	kcal mol−1	Graphene	G > A > C > T > U	107
optB86b	19.9	19.2	22.3	18.0		kcal mol−1	Graphene	G > A > T > C	108
PBE + TS	19.8	19.7	21.7	18.4		kcal mol−1	Graphene	G > A > T > C
M05-2X with 6–31+G(d,p) basis set	27.01	27.98	37.62	27.98	22.19	kJ mol−1	Graphene	G > C ≈ T > A > U	109
M05-2X with 6–311++G(d,p) basis set	16.70	19.64	27.23	20.50	13.93	kJ mol−1	Graphene	G > C > T > A > U
M06-2X with 6–31+G(d,p) basis set	52.19	52.93	65.08	51.02	46.36	kJ mol−1	Graphene	G > T > A > C > U
M06-2X with 6–311++G(d,p) basis set	44.23	46.23	57.46	45.10	35.00	kJ mol−1	Graphene	G > T > C > A > U
DFT combined with non-equilibrium Green’s function (NEGF)	0.497	0.468	0.713	0.577		eV	Graphene nanoribbon	G > C > A > T	110
Time-dependent (TD) M06-2X/6-31G(d) level of theory	−11.4	−11.0	−14.6	−11.1		kcal mol−1	Graphene (in gas phase)	G > A ≥ C ≥ T	111
	−14.0	−14.9	−15.7	−11.7		kcal mol−1	Graphene (in ethanol solution)	G > T > A > C
	−25.0	−24.7	−24.7	−21.9		kcal mol−1	Graphene (in ethanol solution and nucleobases attached with a fluorophore)	A > G ≈ T > C
TD-DFT LDA Perdew Wang function (PWC)	−19.7	−16.9	−23.2	−18.0		kcal mol−1	Graphene (in gas phase)	G > A > C > T
Double numerical (DND) level of theory	−16.6	−13.6	−15.7	−11.7		kcal mol−1	Graphene (in ethanol solution)	A > G > T > C
Isothermal titration calorimetry (ITC)	−11.85	−4.72	−13.45	−9.38		kcal mol−1	Graphene (in alkaline solution)	G > A > C > T	103

TABLE II.

Summary of SMFS measurements of binding affinities between DNA homopolymers/heteropolymers and graphite.

Sequence	Peeling force (pN)	Binding energy per nucleotide (kBT)	References
3′-poly(dT50)	85.3 ± 4.7	11.3 ± 0.8	112
3′-poly(dC50)	60.8 ± 5.5	7.5 ± 0.8
5′-poly(dC50)	73.4 ± 5.5	9.4 ± 0.9	113
5′-poly(dT100)	78.5 ± 5.0	10.2 ± 0.8
3′-poly(dT100	76.6 ± 3.0	9.9 ± 0.5
5′-poly(dT100)	66.4 ± 1.4	8.3 ± 0.2
50 -NH2-(CH2) 6-AGT CAG TGT GGA AAA TCT CTA GC-30	176.8 ± 50.7 (in pure water)		114
	156.2 ± 36.5 (in 20 mM NaCl)
	141.3 ± 36.2 (in 50 mM NaCl)
	126.3 ± 37.7 (in 150 mM NaCl)
A30	64 ± 4		115
T30	57 ± 3
G30	74 ± 8
C30	58 ± 7
Adenosine DNA aptamer (5′-NH2-(CH2)6-AGAGAACCTGGGGGAGTATTGCGGAGGAAGGT-3′)	117.8 ± 29.5		116
	164.3 ± 25.4 (in the presence of 1 μM adenosine)

IV. INTERACTIONS OF GRAPHENE-BASED MATERIALS WITH NUCLEOBASES

A number of modeling and simulation methods have been utilized to investigate the interaction of graphene with nucleobases.^109,117 These studies have indicated that the binding of nucleobases to graphene is primarily driven by π-π stacking. The calculated binding energies differ significantly depending on the specific bases adsorbed on graphene.^100,101,118 The binding energies of nucleobases on the surface of graphene-based materials using different methods are summarized in Table I. Gowtham et al.¹⁰⁰ carried out an investigation of the interaction between nucleobases and graphene using density functional theory (DFT) framework, with additional calculations using Hartree–Fock plus second-order Møller–Plesset perturbation theory. The resulting equilibrium configurations of nucleobases on graphene surface were found to be quite similar for A, T, G, C, and U [Fig. 3(a)]. The calculated binding energies were nearly identical for A, T, and C. However, G exhibited a stronger binding, whereas U showed a weaker binding. This suggests the following hierarchy: G > A ≈ T ≈ C > U. The authors found that the calculated trend of binding energies was remarkably correlated with the polarizabilities of nucleobases. This correlation suggested that the polarizability of the bases governed the interaction strength of the nucleobases with graphene. As vdW energy is proportional to polarizabilities, the remarkable correlation indicates that vdW interaction is the dominant interaction source between the nucleobases and graphene. Hemanth and Mallajosyula conducted polarizable molecular dynamics (MD) simulations to investigate the impact of polarization on the interactions between nucleobases and graphene. Their study revealed that upon incorporating polarization, the binding free energies for nucleobases on monolayer graphene exhibited the following trend: G > A > T > C > U.¹⁰²

FIG. 3.

(a) Equilibrium geometries of nucleobases on top of graphene. From top to bottom: G, A, T, C, and U, respectively. Reproduced with permission from Gowtham et al., Phys. Rev. B 76(3), 033401 (2007). Copyright 2007 American Physical Society. (b) The top view of the optimized geometries of nucleobases on CNTs of varying curvatures and graphene. Reproduced with permission from D. Umadevi and G. N. Sastry, J. Phys. Chem. Lett. 2(13), 1572–1576 (2011). Copyright 2011 American Chemical Society. (c) Scan tunneling microscopy (STM) image of guanine lattice on graphite. Scale bar: 10 nm. Reproduced with permission from N. J. Tao and Z. Shi, J. Phys. Chem. 98(5), 1464–1471 (1994). Copyright 1994 American Chemical Society. (d) Co-adsorption of G and C at the 1-octanol/graphite interface. (i) Typical STM image showing three distinct domains, marked I, II, and III. Scale bar, 10 nm. (ii) and (iii) Correlation-averaged zoom-in on the structures of domain I and domain III. Scale bars, 1 nm. Bottom: height profiles along the blue lines in the top panels. Reproduced with permission from Xu et al., Nano Lett. 6(7), 1434–1438 (2006). Copyright 2006 American Chemical Society.

Whereas the calculated trends of binding energies of nucleobases vary by theoretical methodologies, Varghese et al.¹⁰³ reported that the binding energies followed the order of G > A ≈ T > C when employing the Hartree–Fock approximation with the addition of vdW interaction and solvation energies based on the AMBER generalized Born mode. The binding energies were calculated to be in the following order: G > A > T > C > U when using TS (Tkatchenko and Scheffler’s method),¹¹⁹ sTS (simplified version of TS),¹⁰⁴ DFT-D (all-electron density functional theory in generalized gradient approximation combined with an empirical correction for dispersion interactions),¹⁰¹ vdW energy-corrected DFT calculations,¹⁰⁵ and plane-wave pseudopotential approach within the generalized gradient approximation of DFT.¹⁰⁶ Furthermore, the sequence of binding energies was: G > A > T ≈ C > U when using DFT-D3 and vdW-DF2.^104,120,121

These theoretical studies often assumed perfectly planar graphene sheets. However, the ripple formation is an inherent nature of graphene sheets, which provides it with the structural stability.^122–124 The appearance of curvatures due to ripple formation can significantly alter the electronic properties of graphene and affect its interaction with nucleobases. To investigate the effect of curvature on nucleobases binding, researchers have performed theoretical calculations of the binding energies for nucleobases on carbon nanotubes (CNTs) with varying curvatures and on planar graphene using highly reliable first-principle calculations.¹¹⁸ The results showed that the binding energy increased as the curvature decreased and reached the maximum for planar graphene [Fig. 3(b)]. This phenomenon was attributed to more efficient stacking between the carbon surface and the nucleobases as the curvature decreases. However, these studies of curvature effect on nucleobase binding are based on structures of CNT but not nonplanarity of graphene. Another study reported that isolated large graphene exhibited significant curvature, as observed by quantum chemical calculations and atomic force microscopy (AFM) measurements.¹⁰⁷ Thecurvature may give additional stability to nucleobase binding on graphene surface. This seemingly contradictory conclusion compared to Ref. 118 may be ascribed to the differences in the size, curvature, and electronic properties of CNTs (7 benzene rings), planar graphene (7 benzene rings), and curved graphene (60 benzene rings), which can impact non-bonding interactions.^125,126 It has been reported that the interaction energy systematically increases as the size of the π-system expands.¹²⁵ The larger graphene model (60 benzene rings) in Ref. 107 suggests a potentially stronger binding affinity with bases compared to the smaller CNTs (7 benzene rings) or graphene (7 benzene rings) used in Ref. 118. In addition, Ref. 107 has demonstrated that nucleobases can bind to larger graphene with inherent curvature.¹⁰⁷ The calculated binding energies between them followed the order of G > A > C > T > U, which was well-correlated with previous theoretical and experimental observations. G, A, and C had higher stabilization energy than T and U, presumably due to the presence of amino groups that can interact with the π electron cloud of the graphene sheet to form stronger N – H^…π hydrogen bond, while T and U possess only carbonyl oxygen atoms, capable to form weaker lone-pair π interactions. However, Ref. 107 did not provide direct evidence to confirm the contribution of graphene’s inherent curvature to nucleobase binding. Additional research efforts are crucial to substantiate this aspect.

Numerous experimental studies have investigated the structures of nucleobases on graphite, as they were presumed to exhibit similar base stacking interactions as on graphene. It was demonstrated that all four DNA bases (A, T, G, and C) can spontaneously self-assemble into compact 2D lattices on graphite surface using scanning tunneling microscopy (STM). In two independent studies, Allen et al.¹²⁷ and Heckl et al.¹²⁸ investigate the molecular arrangements of nucleobases on graphite in air using STM. They both observed that A, T, and G molecules formed highly organized lattices through the evaporation of their aqueous solutions on freshly cleaved graphite surface. The base molecules registered heteroepitaxial on the graphite with their planar rings lying flat on the surface. Tao and Shi¹²⁹ further examined the nucleation, growth, and molecular packing structures of A and G on graphite in NaCl solution, using both STM and AFM. They discovered that nucleation began from step edges or flat terraces, and line-shaped nuclei tended to form during the nucleation. As the nuclei grew and collided, their growth halted upon contact with adjacent nuclei, preventing any coalescence or coagulation. Consequently, both A and G molecules underwent spontaneous condensation, eventually organizing into ordered lattices on the graphite surface [Fig. 3(c)]. The spontaneous lattice formation and structure determination of nucleobases, modified nucleobases, and co-adsorption of nucleobases on graphite were further proved by the STM in the subsequent studies.^130–132 For instance, Xu et al. demonstrated a well-ordered structure by co-adsorption of the complementary nucleobases G and C on graphite surface in 1-octanol solvent by STM.¹³² The measurements revealed three distinct domains, labeled as I, II, and III in Fig. 3(d). The domains I/II were formed from C molecules alone, while the domain III, featuring a double-row structure, was attributed to aligned, hydrogen-bonded GC dimers. The observed dimer chains were further established by theoretical modeling, providing detailed insight into the intermolecular interactions underlying their formation.

Experimental works have also explored the binding energy between nucleobases and graphite/graphene. The obtained binding orders closely matched those obtained from the theoretical calculations mentioned earlier. Sowerby et al.¹³³ determined the equilibrium adsorption isotherms for the nucleobases on graphite. They found that the adsorption behavior of bases differed, with the binding order being G > A > T > C > U. In another study conducted by Varghese et al.,¹⁰³ isothermal titration calorimetry (ITC) was used to investigate the interaction of nucleobases with graphene. The results showed that the binding energy varied in the order G > A > C > T in aqueous solutions, with the positions of C and T appeared to be interchangeable.

V. INTERACTIONS OF GRAPHENE-BASED MATERIALS WITH SSDNA

π-π stacking interaction has been demonstrated to be a driving force for ssDNA physisorption on graphene.^{16,20,134–136} However, the interaction strength follows a different sequence,^113,137,138 compared to that of individual nucleotides obtained from solution studies, G > A > T > C,^133,139 or of nucleobases and nucleosides determined by ITC, G > A> C > T.¹⁰³ Single-molecule force spectroscopy (SMFS) that involves oligonucleotides covalently linked to AFM tips provides a flexible approach to measure the strength of interactions between DNA and solid substrates directly. SMFS measurements of interaction strength between DNA homopolymers/heteropolymers and graphite are summarized in Table II. In a study by Manohar et al.,¹¹² the force required to peel ssDNA from graphite was measured using SMFS. In their experimental approach, gold-coated AFM tips were covalently functionalized with thiol-modified ssDNA [Fig. 4(a)]. These functionalized tips were brought into contact with the graphite surface and then retracted. During the tip retraction, the force–distance curves displayed characteristic plateaus connected by abrupt force jumps, reflecting a steady-state peeling process interrupted by complete release of one or more ssDNA molecules. The corresponding release force was interpreted as the binding strength of ssDNA with graphite. For pyrimidine oligomers, the measured forces were 85.3 ± 4.7 and 60.8 ± 5.5 pN for polythymine [3’-poly(dT₅₀)) and polycytosine (3’-poly(dC₅₀)], respectively. Based on these measurements, the authors estimated the average binding energy per monomer to be 11.5 ± 0.6 kBT and 8.3 ± 0.7 kBT for thymine and cytosine nucleotides, respectively. In a similar approach, Iliafar et al.¹¹³ observed that the peeling forces of homopolymers were negligibly affected by the DNA chain length. The forces measured for polyadenine [3’-poly(dA₅₀) and polyguanine (5’-poly(dG₁₀₀)] were 76.6 ± 3.0 and 66.4 ± 1.4 pN, respectively. These forces corresponded to the average binding energies per monomer of 9.9 ± 0.5 and 8.3 ± 0.2 kBT, respectively. The ranking of the average binding energies per monomer for all four bases was determined to be T ≥ A > G ≥ C. Interestingly, peeling 5’-poly(dG₁₀₀) from graphite exhibited significant variations in the shape of force–distance curves and peeling forces, compared to the other homopolymers. This variation suggested the possible presence of secondary structures in 5’-poly(dG₁₀₀).

FIG. 4.

(a) Experimental scheme of peeling a ssDNA molecule from graphite surface. Reproduced with permission from Manohar et al., Nano Lett. 8(12), 4365–4372 (2008). Copyright 2008 American Chemical Society. (b) (i) Schematic of the forced migration simulation. The ssDNA is pulled down (+y), up (−y), and parallel to (+x) the step defect. (ii) ssDNA is subjected to the same-magnitude constant force directed up and down the step defect. (iii) The average speed of ssDNA vs the constant force for which the force is directed up, down, and parallel to step defect. Reproduced with permission from M. Shankla and A. Aksimentiev, Nat. Nanotechnol. 14(9), 858–865 (2019). Copyright 2019 Springer Nature. (c) Representative snapshots of ssDNA on pristine graphene, GO 20%, and GO 60%, respectively. Reproduced with permission from Kim et al., Adv. Mater. Interfaces 4(8), 1601168 (2017). Copyright 2017 John Wiley and Sons. (d) (i) Typical snapshots of a ssDNA segment adsorbed on the (1) oxidized and (2) unoxidized region of the GO surface. (ii) Snapshots of the adsorption process at different times. At 0.170 ns, a hydrogen bond is formed between ssDNA’s phosphate backbone and GO. At 0.932 ns, a hydrogen bond is formed between ssDNA’s nucleobase and GO. At 0.986 ns, stable π-π stacking is formed between the aromatic rings of the nucleobases and GO. Reproduced with permission from Xu et al., Chem. - Eur. J. 23(53), 13100–13104 (2017). Copyright 2017 John Wiley and Sons.

Physisorption of ssDNA on graphite can induce the assembly of ssDNA into distinct patterns. Akca et al.¹⁴⁰ discovered that the poly-T and poly-G formed elongated networks, while the poly-A and poly-C formed small spherical particles on graphite surface. This is potentially due to that when DNA adsorbed on the graphite surface, the bases must rotate around the sugar–phosphate backbone to bind to the graphite, breaking the base–base bond and resulting in energy loss. Meanwhile, the DNA–graphite interaction provided an energy gain. The balance between these two energies led to the formation of distinct ssDNA patterns. These specific patterns will potentially serve as templates for assembling other nanomaterials or be utilized in the design of graphene-based biosensors and DNA nanodevices. In addition, Shankla and Aksimentiev¹⁴¹ observed that ssDNA tended to localize and stretch linearly along step defects on mechanically exfoliated graphite as revealed by AFM imaging. They further employed MD simulations to show that these step defects can guide the transport of adsorbed ssDNA. The adsorbed ssDNA moved more rapidly down a step defect than upward, and even faster along the defect edge under external force [Fig. 4(b)]. This direction-dependent phenomenon was harnessed to deliver ssDNA to and away from a graphene nanopore. The defect-guided delivery principle based on interplay of interactions between ssDNA and defected graphene holds potential for nanopore sensing and sequencing applications.¹⁴¹

The interaction between DNA and GO is sophisticated compared to that between DNA and graphene.^135,142,143 π-π stacking of DNA nucleobases with carbon-rich domains, hydrogen bonding, and electrostatic repulsion with the oxygen-rich domains play crucial roles in the interactions between DNA and GO. Among all contributing factors, Park et al.¹⁴⁴ demonstrated that hydrogen bonding was a major contributor to the interaction between ssDNA and GO. They successfully dissociated ssDNA from GO by disrupting the hydrogen bonding using urea. The oxidation rate of graphene surfaces also influences the interactions between ssDNA and GO. Kim et al.¹⁴² conducted simulations of ssDNA adsorption on GO with varying oxygen coverages [Fig. 4(c)]. They found that on pristine graphene or GO with low oxygen coverage, ssDNA unfolded due to vdW interactions, particularly π-π stacking; on GO with moderate oxygen coverage, the folded ssDNA structure was preserved due to a balance between vdW and electrostatic interactions; on GO with high oxygen coverage, the folded ssDNA structure was disrupted because of the formation of strong DNA–surface hydrogen bonds. Additionally, dynamic adsorption of ssDNA onto GO has been explored by studies based on MD simulations.^135,143 Xu et al.¹⁴³ discovered that ssDNA could bind to GO surface in a stable manner through a combination of π-π stacking interactions and hydrogen bonding, with a higher preference for adsorption on the oxidized areas compared to the unoxidized areas. Importantly, the adsorption of ssDNA onto GO occurred in a stepwise fashion, facilitated by the dynamic cooperation of π-π stacking and hydrogen bonding. Initially, ssDNA bound to the oxidized groups by forming hydrogen bonds between its phosphate backbone and GO. Subsequently, the ssDNA was dragged closer to the GO surface, with hydrogen bonds transitioning from the phosphate backbone to the nucleobases. Once the aromatic rings of the ssDNA aligned with graphene carbon rings, π–π stacking interactions rapidly formed [Fig. 4(d)]. This behavior was attributed to the faster formation of hydrogen bonds compared to π-π stacking interactions. The π-π stacking interactions required more relaxation time to adjust the conformation of ssDNA to establish stacking interactions with GO.¹⁴³

VI. INTERACTIONS OF GRAPHENE-BASED MATERIALS WITH DSDNA

ssDNA readily adsorbs onto graphene mainly through π-π stacking interactions.^{16,20,134,135} However, the adsorption behavior of dsDNA on graphene remains inconclusive. Zhao¹⁴⁵ conducted MD simulations to investigate the self-assembly of dsDNA segments on graphene surface. They observed that dsDNA segments can regulate their geometric structure rapidly to form two distinct types of structures on graphene surface [Fig. 5(a)]. In the first type, dsDNA segments rotated and “stood up” on graphene, with their helix axes perpendicular to the surface. In the second type, the segments lay flat on graphene, with their axes parallel to the surface. The ending base pairs of the second type were broken apart upon binding to the surface. The primary driving force for both patterns was attributed to the π-π stacking between the ending base pairs and the carbon rings.¹⁴⁵

FIG. 5.

(a) Snapshots of self-assembly of dsDNA segments on graphene taken at different times. dsDNA segments 1, 3, and 4 are able to rotate and stabilize in a “stand-up” state. Segment 2 is lying on the surface and keeping its original orientation. Reproduced with permission from X. Zhao, J. Phys. Chem. C 115(14), 6181–6189 (2011). Copyright 2011 American Chemical Society. (b) Final binding configurations of dsDNA on idealized graphene (i) and wrinkled graphene (ii). (iii) Snapshots showing the adsorption process of dsDNA on the wrinkled graphene. The bases bound to wrinkled graphene are highlighted in red. Reproduced with permission from Li et al., J. Phys. Chem. C 124(5), 3332–3340 (2020). Copyright 2020 American Chemical Society. (c) Three possible mechanisms of hybridization between a probe DNA adsorbed by GO and its cDNA (target DNA): Langmuir–Hinshelwood mechanism (top panel), Eley–Rideal mechanism (middle panel), and Displacement mechanism (bottom panel). In all the cases, the probe DNA with a fluorophore label is pre-adsorbed on GO and the cDNA is added afterward. Reproduced with permission from Liu et al., Anal. Chem. 85(16), 7987–7993 (2013). Copyright 2013 American Chemical Society.

The introduction of self-assembled monolayer can alter the interaction mechanism and control the structural conformation of dsDNA. Kim et al.¹⁴⁶ explored the influence of 1-octadecylamine (ODA) coatings on graphene surface on the conformation of dsDNA using MD simulations. They found that increasing the concentration of ODA led to enhanced stabilization of internal interactions within the DNA double strands, such as base pairs, stackings, and internal hydrogen bonds. On the other hand, the interactions between DNA and graphene decreased. These findings indicated that the formation of DNA–ODA hydrogen bonds hindered the π-π stacking interactions between nucleobases and graphene, resulting in greater structural stability of dsDNA. Simulations further demonstrated that DNA sequences containing A-T base pairs exhibited significant conformational changes, such as sharp kinks or unwinding, in comparison with sequences with G-C base pairs due to the weaker pairing interactions of A-T base pairs.

Graphene exhibits a wrinkled structure rather than an ideally planar one.^{123,147–149} These wrinkled structures not only impact the interactions with nucleobases, as mentioned earlier, but also influence the binding of dsDNA. Li et al.¹⁵⁰ investigated the interactions between wrinkled graphene and dsDNA using MD simulations. They observed that dsDNA maintained its native structure when bound to perfectly planar graphene. However, when bound to wrinkled graphene, dsDNA underwent significant structural deformation [Fig. 5(b)]. Through analysis of free energy profiles, the researchers found that the binding affinity between the terminal bases of dsDNA and the wrinkled area was stronger than that with the planar area. The energy difference restricted the movement of the bases on the wrinkled area, effectively acting as “anchors.” Consequently, the remaining parts of dsDNA underwent substantial movement and generated a “centripetal stretching” force toward the anchoring bases. This force teared apart the hydrogen bonds between the base pairs, leading to the unfolding of dsDNA in a zipper-like manner. The local unfolding then released more base pairs, enabling their attachment to the wrinkled area through π stacking interactions, which accelerated the unfolding process of dsDNA [Fig. 5(b)].

The adsorption of dsDNA on GO was much weaker than that of ssDNA.^20,49,151 ssDNA can strongly adsorb on GO through aromatic stacking, hydrogen bonding, and hydrophobic interactions, whereas dsDNA has its DNA bases shielded, exposing only negatively charged phosphate groups.^152,153 Therefore, the adsorption of dsDNA to GO is only possible under certain circumstances. For instance, dsDNA can strongly adsorb on GO in the presence of metal ions, such as Mg²⁺, Ca²⁺, Na⁺, and K⁺.¹⁵² These metal ions could neutralize the electrostatic repulsion between dsDNA and GO, facilitating their π-π interactions to form dsDNA/GO complexes. The dsDNA/GO complex enabled GO to enhance the biostability of dsDNA and protect it from enzymatic cleavage. To gain a deeper understanding of the interaction between dsDNA and GO, Tang et al.¹⁵¹ conducted a series experiments including fluorescence titration, salt-concentration dependency, DNA melting transition analysis, and circular dichroism (CD) spectra analysis. They discovered that dsDNA can strongly bind to GO, although the interaction was weaker than that between ssDNA and GO. This binding may be facilitated by partial deformation of the DNA double-helix structure on GO. In addition, dsDNA on GO exhibited distinct effects on enzymatic degradation: effectively cleaved by DNA enzyme I and restriction endonucleases as EcoR I implied that GO could not fully protect dsDNA from enzymatic cleavage; high resistance to Exo III degradation indicated that the binding of DNA on GO influenced the enzyme activity or resulted in steric hindrance for Exo III. As hybridization of complementary DNA target (cDNA) is often used in DNA probe–GO-based DNA sensing, it is important to understand the DNA desorption from GO. Compared to DNA adsorption on GO, the process of DNA desorption from GO upon the addition of cDNA is more complex, and three possible mechanisms have been proposed [Fig. 5(c)].¹⁵⁴ The first mechanism is the Langmuir–Hinshelwood mechanism, where cDNA is adsorbed on GO, diffuses, and reacts with the adsorbed probe before desorbing. The second mechanism is the Eley–Rideal mechanism, where cDNA in the solution directly reacts with the adsorbed probe on GO. The third mechanism is the nonspecific displacement mechanism, where the adsorbed probe is displaced by cDNA into the solution, and then hybridizes with free cDNA in the solution.

VII. ENGINEEING GRAPHENE-BASED BIOSENSORS BY HARNESSING DNA PROBE DESIGN AND DNA PROBE–GRAPHENE-BASED MATERIAL INTERACTION

The studies on the dynamics and mechanisms of DNA probe adsorption and desorption contribute to the rational design of biosensors:^19,155 (1) Enable the design of stable and reliable biosensors by enhancing the affinity between DNA probes and graphene-based materials and preventing their detachment during sensor operation, through tailoring the surface properties of graphene-based materials. (2) Allow to design biosensors with high selectivity while minimizing false positives from nonspecific interactions. (3) Help to develop effective biofunctionalization strategies for modifying the graphene-based materials with specific functional groups or biomolecules to enhance the biosensor’s performance. (4) Enable the development of biosensors capable of real-time monitoring, particularly in applications such as DNA sequencing, diagnostics, and environmental monitoring, where rapid and accurate results are essential. In summary, the fundamental knowledge of dynamics and mechanisms of DNA probe adsorption and desorption on graphene-based materials serves as a foundation for creating highly efficient and versatile biosensing platforms for a wide range of applications.

In Secs. VII A and VII B, we focus on the optical and electronic sensing, respectively, by integrating DNA probes with graphene-based materials. As per research reports, a considerable portion of the existing literature predominantly revolves around the utilization of DNA probe–GO hybrids for optical sensing, due to the water dispersibility, high selectivity, and sensitivity, as well as the capability for in vitro and in vivo applications. In addition, a significant portion of research centers on the utilization of DNA probe–graphene for electronic sensing, based on the exceptional conductivity of graphene, and the ease of preparing DNA probe–GFET sensors. Based on the existing reports, our primary focus centers on optical sensing by DNA probe–GO in Sec. VII A and electronic sensing by DNA probe–graphene in Sec. VII B.

A. Optical sensing

The interaction between GO and ssDNA stronger than that between GO and dsDNA, coupled with the exceptional fluorescence quenching capability of GO, forms the basis for the construction of ssDNA empowered GO-based optical biosensors.¹⁹ The common mechanism of ssDNA empowered GO-based optical sensing is: GO could effectively bind to dye-labeled ssDNA and quench the fluorescence emitted by the dye.⁴⁹ Upon introducing cDNA, the ssDNA hybridizes with cDNA to form dsDNA. This hybridization disrupts the interaction between GO and the dye-labeled ssDNA, leading to the release of dsDNA and subsequent fluorescence recovery.⁴⁹ Likewise, if the DNA probe is an aptamer, the addition of target molecule can also induce a similar increase in fluorescence by forming an aptamer–target complex.^16,156 Based on this sensing mechanism, the DNA probe–GO architecture offers a universal platform for the detection of multiplex analytes by designing DNA sequences with diverse functional features of biorecognition ability, tertiary structures, and signal amplification capabilities.

The designed DNA sequences include DNA aptamers, molecular beacons (MBs), and DNAzymes. DNA aptamers generated through SELEX⁷⁸ exhibit unique binding characteristics toward their targets, such as high selectivity, sensitivity, specificity, and the capacity to adopt a variety of tertiary structures.^76,157 They are widely employed as molecular probes in biosensors, offering several advantages over antibodies^81–84 (see Sec. III DNA-based materials). These advantages include ease of synthesis, high stability, low immunogenicity, and the ability to be easily modified or conjugated with other molecules. MBs, which consist of loop and stem structures, target complementary sequences, as well as fluorescent and quencher probes, are employed as intelligent molecular valves.^158,159 When a MB encounters its target nucleic acid sequence, it binds to the target through complementary base pairing. This interaction leads to the opening of the stem region in the MB, separating the fluorophore and quencher. This spatial separation allows the fluorophore to emit fluorescence, indicating the presence of the target sequence.¹⁶⁰ DNAzymes, also known as deoxyribozymes, are DNA molecules with catalytic activity. Typically, DNAzymes comprise catalytic domains that contain the necessary sequences for catalysis, and substrate binding domains that interact with target molecules.^161,162 They can catalyze various reactions, such as cleavage, ligation, and even more complex enzymatic activities that can be applied in target molecule detection and quantification.¹⁶³

In addition to the design of DNA probes’ structures, interplay of the non-covalent interaction between DNA probes and graphene-based materials provides opportunities for optimizing DNA and graphene-based biosensors. For instance, electrolytes are often utilized to bring DNA in close proximity to the GO surface for effective binding, which screen and bridge the negatively charged DNA and GO.²⁰ Research has shown that higher ionic strength, particularly with divalent metal ions, is favorable for promoting binding of DNA onto GO.¹⁶⁴ Controlling pH of the solution is another approach to manipulate surface charge of DNA and GO, and thus regulate the interactions between DNA and GO. Lower pH can enhance binding effectiveness. By lowering the pH from 8 to 5, the binding efficiency increased from 30% to 100%.¹⁶⁴ Additionally, functional groups and the C/O ratios of GO can influence their interactions with ssDNA. In terms of the quenching of a fluorescein-labeled DNA probe, the preference order is: GO > GO-COOH > GO-PEG-NH₂.⁵¹ GO with a higher C/O ratio (C/O ratio = 1.9) demonstrates a stronger binding affinity to ssDNA and more efficient fluorescence quenching of organic chromophores than GO with a lower C/O ratio (C/O ratio = 1.1).¹⁶⁵ Furthermore, the length of the ssDNA plays a significant role in the interaction strength. Studies have shown that longer ssDNA tends to have a stronger affinity to GO compared to shorter ssDNA.^166,167 However, there are contradictory findings in different studies regarding the length-dependent binding affinity,¹⁶⁸ which may be attributed to variations in the DNA length regions investigated.

The interaction between DNA probes and graphene-based materials, along with their integration, provides distinct advantages that propel the development of highly sensitive and selective biosensing platforms. Here are several noteworthy examples of recent advancements that highlight the immense potential of optical biosensors based on DNA probe-empowered graphene-based materials in various fields, with particular emphasis on the detection of DNA, RNA, small molecules, and proteins.

1. DNA detection

DNA detection has gained considerable interest in recent years owing to its potential applications in disease diagnosis and treatment, forensic analysis, food safety evaluation, environmental monitoring, etc.^16,66 Though strategic design of DNA probes and leveraging the interactions between GO and DNA, researchers have achieved highly sensitive and selective detection of specific DNA sequences. For instance, a MB probe was developed, featuring stem-loop regions containing random unpaired bases and rigid paired bases [Fig. 6(a)].¹⁶⁹ The hybrids formed by MBs and GO demonstrated improved sensitivity compared to MBs alone, as GO effectively reduced the background fluorescence of the MBs arising from the incompletely quenched fluorophores by quenchers. Additionally, the hybrids exhibited superior thermostability, providing an advantage in maintaining their functionality and fluorescence response even under challenging temperature conditions. The DNA probe–GO biosensor has also exhibited high sequence specificity in DNA detection, showcasing a promising avenue for single-nucleotide polymorphism analysis.¹⁵⁵ Building on this observation, a DNA probe–GO-based biosensor was developed and realized the discrimination between healthy and mutated DNAs associated with lung cancer, allowing for the detection of mutated DNA in cancer patients [Fig. 6(b)].⁶⁶ Moreover, the large planar surface of GO facilitated the simultaneous quenching of multiple DNA probes labeled with different dyes, enabling multicolor sensing for the detection of multiple DNA targets within the same solution [Fig. 6(c)].¹⁵⁵ This breakthrough opens up new possibilities for detecting multiple tumor suppressor genes and identifying early-stage cancers in asymptomatic individuals.

FIG. 6.

(a) Scheme of the MB-GO-based DNA detection. (i) MB directly hybridizes with target DNA, leading to fluorescence recovery. (ii) and (iii) GO first binds and quenches MB’s background fluorescence. Hybridization of MB with target DNA forms dsDNA and induces the release of MB from GO, consequently recovering the fluorescence. Reproduced with permission from Li et al., Nanoscale 2(6), 1021 (2010). Copyright 2010 Royal Society of Chemistry. (b) The DNA probe–GO biosensor responses to healthy DNA and mutated DNA. The introduction of healthy DNA to the biosensor leads to the hybridization of healthy DNA and DNA probe, causing the release of DNA probe from GO surface and initiating fluorescence recovery. Conversely, the addition of mutated DNA fails to induce fluorescence recovery. Reproduced with permission from Kadhim et al., RSC Adv. 13(4), 2487–2500 (2023). Copyright 2023 Royal Society of Chemistry. (c) Scheme for GO-based multicolor DNA analysis. When specific target DNA is introduced into a mixture of dye-labeled DNA probes, only the probe that is complementary to the target forms dsDNA, preventing its adsorption on GO and avoiding fluorescence quenching. In contrast, other probes adsorb on GO and are quenched by GO. Reproduced with permission from He et al., Adv. Funct. Mater. 20(3), 453–459 (2010). Copyright 2010 John Wiley and Sons.

2. RNA detection

DNA probe–GO-based optical biosensors have emerged as versatile platforms for detecting RNA molecules by designing DNA probes. For instance, two distinct DNA probes were developed to target microRNA-21 (miRNA-21) and protein programmed cell death 4 (PDCD4) mRNA, respectively. These DNA probes were adsorbed on GO, forming a sensing platform that successfully achieved simultaneous imaging of miRNA-21 and PDCD4 mRNA in both solutions and MCF-7 human breast cancer cells [Fig. 7(a)].¹⁵⁶ Notably, this approach demonstrated the ability to differentiate the expression levels of the two RNAs in living cells, offering great value for early cancer diagnosis and monitoring cancer development. Another novel ssDNA aptamer has been selected by employing a modified version of GO-SELEX¹⁷⁰ for miRNA-215 detection and achieved a detection limit of 2.6 nM.¹⁷¹

FIG. 7.

(a) Simultaneous detection of miR-21 and protein programmed cell death 4 (PDCD4) mRNA. Two DNA probes targeting miR-21 and PDCD4 mRNA separately can be adsorbed by GO. When DNA probes hybridize with their targets to form duplexes, duplexes deviate from GO and lead to the recovery of fluorescence for detecting miR-21 and PDCD4 mRNA. Reproduced with permission from Pan et al., Nanoscale 10(29), 14264–14271 (2018). Copyright 2018 Royal Society of Chemistry. (b) Principle of the fluorescent microRNA (miRNA) biosensor. (1) Graphdiyne quantum dot (GDQD) modified with DNA probe emits at 405 nm. (2) Adding graphene quantum dot (GQD) can quench the fluorescence of GDQD, and GQD shows a new emission peak at 505 nm. (3) When miRNA-21 is present, it hybridizes with DNA probe and increases the distance between GDQD and GQD, leading to the recovery of GDQD fluorescence and decrease in GQD fluorescence. Reproduced with permission from Bahari et al., J. Lumin. 239, 118371 (2021). Copyright 2021 Elsevier. (c) Simultaneous dual-color imaging of miRNA-451a and miRNA-214-3p in human breast cancer cells. DNA probes targeting miRNA-451a and miRNA-214-3p, respectively, can be adsorbed by rGO. Upon DNA probes hybridize with their targets to form duplexes, the duplexes release from rGO, leading to fluorescence recovery for miRNA detection. Reproduced with permission from Xiong et al., Talanta 225, 121947 (2021). Copyright 2021 Elsevier.

In addition to GO, GQD and rGO have also been used for RNA detection by employing the interaction between GQD/rGO and DNA. By utilizing DNA to regulate the distance between GDQDs (graphdiyne quantum dots, donor) and GQDs (graphene quantum dots, acceptor), a DNA/GDQDs/GQD based fluorescence resonance energy transfer (FRET) assay was established for miRNA-21 measurement [Fig. 7(b)].¹⁷² This assay achieved a detection limit of 0.5 pM and successfully detected target miRNA-21 in human serum, biological cell lines, and MDA-MB-231 human breast cancer cells.¹⁷² By harnessing the specific adsorption of ssDNA and dsDNA on rGO and the potent fluorescence quenching effect of rGO,¹⁷³ a DNA probe–rGO nanohybrid platform was developed for the simultaneous visualization of miRNA-451a and miRNA-214-3p in living cells [Fig. 7(c)].⁵² Experiments in solution demonstrated the selectivity and sensitivity of the probe, with a detection limit as low as 1 nM for both miRNA-451a and miRNA-214-3p. Furthermore, intracellular experiments performed on MDA-MB-231 cells realized simultaneous in situ imaging of miRNA-451a and miRNA-214-3;3p.

3. Small-molecule detection

DNA probe–GO-based optical biosensors have been developed to detect and monitor potentially harmful compounds, such as antibiotics, pesticides, and heavy metal ions,^174–176 by designing specific DNA aptamers. To detect residual antibiotics in food products, DNA aptamers were designed to target antibiotics such as sulfadimethoxine, kanamycin, ampicillin, tetracyclines (TCs) and aflatoxin B1 (AFB1).^177–179 [Figs. 8(a)–8(c)] These aptamers adsorbed on GO or GQD, forming fluorescence biosensors with detection limits of 1.997 (sulfadimethoxine), 2.664 (kanamycin), 2.337 (ampicillin) ng/ml, 45 ng/ml (tetracyclines), and 0.25 ng/ml (aflatoxin B1), respectively. These biosensors were effectively validated in real food samples such as corn, milk, and rice, highlighting their reliability and suitability for detecting antibiotics in various food matrices. In addition, DNA aptamers were designed for the selective screening of pesticides. By utilizing L-cysteine capped CdS QDs/aptamer bioconjugates as fluorescent signal, the developed sensor achieved remarkable sensitivity for diazinon, exhibiting a low detection limit of 0.13 nM and an excellent linear dynamic range from 1.05 to 206 nM [Fig. 8(d)].¹⁸⁰ Another strategy utilized AT-rich three-way-junction DNA-stabilized copper nanoparticles as a fluorescent signal. This strategy demonstrated the detection of isocarbophos within a linear range of 10–500 nM, with a detection limit of 3.38 nM.¹⁸¹ Moreover, DNA aptamers were developed to target specific metal ions. The metal ions can trigger a conformational switching of these aptamers on GO surface, inducing the release of aptamers from GO surface and recovery of fluorescence. The proposed DNA aptamer–GO sensors exhibited wide linear detection ranges with ultralow detection limits. For example, they enabled the determination of Pb²⁺ in the range of 5–70 pM and 0.07–20 nM with a detection limit of 0.5 pM [Fig. 8(e)];¹⁸² allowed simultaneous detection of Hg²⁺ and Ag⁺ in the range of 0.05–50 nM, with detection limits of 1.33 and 1.01 pM, respectively;¹⁸³ and facilitated label-free and quantitative detection of Hg²⁺ in a linear detection range of 0.01–0.5 μM with a detection limit of 3.63 nM.¹⁸⁴ Similarly, a DNAzyme branched junction structure was designed for the simultaneous detection of Cu²⁺, Mg²⁺, and Pb²⁺ in both homogeneous solution and human serum samples [Fig. 8(f)].¹⁸⁵ The fluorescence intensity exhibited good linear relationships with the concentration of metal ions, with detection ranges of 2–500 nM for Cu²⁺, 0.5–200 μM for Mg²⁺, and 1–500 nM for Pb²⁺. The calculated detection limits were 1 nM for Cu²⁺, 200 nM for Mg²⁺, and 0.3 nM for Pb²⁺.

FIG. 8.

(a) Schematic illustration of a high-efficient multiplex aptamer/GO-based biosensor for the detection of antibiotics using cyclic enzymatic signal amplification (CESA) method. Reproduced with permission from Youn et al., Sci. Rep. 9(1), 7659 (2019). Copyright 2019 Springer Nature. (b) Experimental design of TCs detection. Reproduced with permission from Ahmed et al., Food Chem. 346, 128893 (2021). Copyright 2021 Elsevier. (c) Label-free TPE-Z/AFB1 aptamer/GO biosensor for AFB1 detection in food samples. Reproduced with permission from Jia et al., Talanta 198, 71–77 (2019). Copyright 2019 Elsevier. (d) The principle of the fluorescence “turn off-on” QDs/aptamer/GO biosensor for diazinon detection. Reproduced with permission from M. Arvand and A. A. Mirroshandel, Food Chem. 280, 115–122 (2019). Copyright 2019 Elsevier. (e) Paper-based aptamer/GO sensor combined with the FRET process for Pb²⁺ detection. Reproduced with permission from Khoshbin et al., Anal. Chim. Acta 1071, 70–77 (2019). Copyright 2019 Elsevier. (f) Simultaneous fluorescent detection of Cu²⁺, Pb²⁺, and Mg²⁺ based on DNAzyme/GO platform. Reproduced with permission from Yun et al., Anal. Chim. Acta 986, 115–121 (2017). Copyright 2017 Elsevier.

4. Protein detection

The DNA probe–GO nanohybrids have emerged as novel bioassay probes for protein sensing, utilizing specific DNA probes, together with novel fluorescent probes, or DNA signal amplification technologies. For instance, a thrombin-binding aptamer linked with a G-quadruplex region was developed for thrombin determination [Fig. 9(a)].¹⁸⁶ The G-quadruplex region can interact with N-methylmesoporphyrin IX (NMM) and emit a significant fluorescent signal, which served as fluorescent probe.^187,188 The fabricated thrombin aptamer/G-quadruplex/NMM/GO biosensor achieved a detection range of 0.37 nM to 50 μM, with a detection limit of 0.37 nM. An H5N1 Influenza A virus hemagglutinin (HA) aptamer/GO optical sensor was also developed for HA detection by integrating HA-specific DNA aptamer and GO [Fig. 9(b)].¹⁸⁹ This biosensor incorporated a novel fluorescent probe, sandwich-structured upconversion nanoparticle (UCNP), and was applicable for quantifying HA in human serum, with a linear range of 0.2–12 ng ml⁻¹ and a detection limit of 114.7 pg ml⁻¹. By combing the general design mechanism of DNA probe–GO optical biosensor and DNA hybridization chain reaction, a novel label-free biosensor was constructed for sensitive detection of mucin 1 (MUC1) [Fig. 9(c)].¹⁹⁰ This biosensor achieved a low detection limit (0.36 fg ml⁻¹) and a wide linear range (1fg ml⁻¹–1 ng ml⁻¹).

FIG. 9.

(a) Detection of thrombin based on DNA probe–GO biosensor. DNA probe has two parts: Sa, the thrombin aptamer sequence, and Se, the G-quadruplex sequence. The probe initially binds to GO and is released from GO in the presence of thrombin. NMM inserts into the G-quadruplex and emits a high fluorescence. In the absence of thrombin, high fluorescence emission cannot be induced. Reproduced with permission from Wei et al., Sensors 19(20), 4424 (2019). Copyright 2019 MDPI. (b) UCNP/HA aptamer/GO biosensor for HA detection. HA aptamer functionalized UCNP can adsorb on GO and enables quenching of its upconversion luminescence. When HA is present, the aptamer undergoes conformational changes, moving away from GO and resulting in the recovery of upconversion fluorescence. Reproduced with permission from Zhao et al., ACS Omega 6(23), 15236–15245 (2021). Copyright 2021 American Chemical Society. (c) Fluorescent biosensing strategy for MUC1 detection. The detection system comprises an M-shaped aptamer probe P1-P2-P3, two hairpin probes AgNCs-H1 and AgNCs-H2, biocatalyst Exo I, and quencher GO. The addition of MUC1 triggers Exo I-assisted target recycling and GO-assisted hybridization chain reaction, resulting in the formation of long linear double-stranded nanowires containing numerous AgNCs. These nanowires cannot be quenched by GO due to their weak binding affinity, thereby retaining a strong fluorescence indicative of the MUC1 concentration. Reproduced with permission from Wu et al., Anal. Chim. Acta 1129, 40–48 (2020). Copyright 2020 Elsevier. (d) DNA probe–GO-based biosensor for MNase detection. fluorescein amidites (FAM)-labeled 20-mer ssDNA initially binds to GO and is quenched by GO. While in the presence of MNase, the ssDNA is cleaved into fragments by MNase, resulting in weak quenching of the FAM fluorescence. Reproduced with permission from He et al., Biosens. Bioelectron. 42, 467–473 (2013). Copyright 2013 Elsevier.

Another approach for protein detection using DNA probe–GO sensors is utilizing competing affinities between longer ssDNA and shorter ssDNA on GO. Previous studies indicated that longer ssDNA exhibits a higher affinity to GO compared to shorter ssDNA.^166,167 Based on this affinity difference, a model enzyme micrococcal nuclease (MNase) was used, which preferentially digested ssDNA at AT-rich regions under slightly basic pH conditions. A FAM-labeled 20-mer ssDNA was used as a substrate for MNase. In the absence of MNase, the FAM-labeled ssDNA was adsorbed on the surface of GO, leading to efficient quenching of FAM fluorescence. However, upon the addition of MNase, the ssDNA underwent cleavage into smaller fragments. The shorter FAM-labeled ssDNA fragments had weak affinity for GO, leading to minimal quenching of the FAM fluorescence [Fig. 9(d)].¹⁹¹ The detection of MNase can be achieved in a range of 8 × 10⁻⁵ to 1.6 × 10⁻³ unit/ml, with a detection limit of 2.7 × 10⁻⁵ unit/ml. Such biosensor model can readily be adapted for the detection of other nucleases that cleave ssDNA, such as S1 nuclease, by modifying the reaction buffer.

B. Electronic sensing

Graphene, renowned for its remarkable electrical properties, is an excellent material for applications in electrical sensing, particularly when utilized in the fabrication of GFETs.^57,192,193 The functionality of GFETs can be harnessed by incorporating probe molecules onto their surfaces. A variety of DNA probes have been successfully employed in the development of DNA probe–GFETs.^7,194–196 In DNA probe–GFETs, ssDNA probes are immobilized on graphene surface and selectively bind to target molecules in solution phase, leading to a conductivity change associated with target-binding events.^8,194 These GFETs offer exceptional properties in the detection of charged biomolecules, with notable attributes, including high selectivity, sensitivity, throughput, and low detection limits in biomolecular detection and analysis, which surpass existing fluorescence methods such as polymerase chain reaction and enzyme-linked immunosorbent assays.

The sensitivity of DNA probe–GFETs relies on the interactions between DNA molecules and graphene surface. Several factors can influence the interactions and consequently the sensitivity of DNA probe–GFETs. The quality of graphene used in DNA probe–GFETs is crucial.¹⁹⁷ High-quality graphene with low defect density and minimal surface contamination provides more available binding sites and higher electrical conductivity, enhancing the sensing capability.^198,199 Decorating graphene surface with NPs, such as platinum or gold NPs, can increase the surface-to-volume ratio to load more DNA probes, which therefore improve the sensitivity or expanding the upper detection limit of the biosensors.^58,200 The length of DNA probes can also impact the sensitivity of DNA probe–GFET by influencing DNA adsorption and binding efficiency to targets.²⁰¹ Among these factors, Debye screening remains a major limitation of the sensitivity of GFET for biosensing in highly ionic physiological solutions.²⁰² This limitation arises from the fact that GFETs primarily respond to the charges carried by biomolecules adhering to the graphene surface. Such responses suffer from the ionic screening by counter ions in solution.²⁰³ This screening effect is characterized by the Debye length that is a measure of the distance over which a charge carrier’s net electrostatic effect persists. Beyond this length, charges are effectively screened. In physiological conditions, the Debye length (typically ≈0.7 nm) is very small compared to the typical sizes of biomolecules (e.g., antibodies ≈ 10–15 nm).^203,204 Consequently, the Debye screening effect substantially hampers the sensitivity of GFETs in detecting macro-biomolecules within highly ionic physiological solutions. Several approaches have been explored to tackle the Debye screening effect: manipulate Debye length by modifying the electrolyte composition and adjusting buffer conditions; use nanoscale structures to influence the local ionic environment surrounding target molecules; functionalize graphene surface with specific molecules or polymers.^8,205 In addition to the methods previously mentioned, a noteworthy strategy for minimizing the Debye screening effect involves using a second gate electrode.²⁰⁴ This secondary gate electrode functions autonomously, allowing precise control over the electrostatic potential at the sensor’s solid–liquid interface. This innovative approach has demonstrated a substantial enhancement in the threshold voltage shift induced by the presence of target molecules, with an increase in nearly two orders of magnitude.²⁰⁴ It is also worth noting the remarkable capabilities of graphene biosensors operating at radio and microwave frequencies.^203,206,207 At these frequencies, the electrolyte ions lag behind the alternating current electric field due to the solution’s viscosity. As a result, the Debye screening is reduced or even eliminated and the buffer solution can be regarded as a pure dielectric.²⁰³ Waves at these frequencies interact with molecules, inducing frequency dependent reorientation of molecular dipoles and translation of electric charges. Different molecules exhibit diverse relaxation processes and interact differently with oscillating electromagnetic fields.²⁰⁸ GFET sensors capitalize on these interactions to identify and distinguish various analytes. Different types of sensors have been developed, including reflectometers, waveguides, resonators, and interferometers. For instance, Zhang et al. reported the operation of GFET near its resonant frequency (i.e., 1.83 GHz) in reflectometry mode, achieving a limit of detection (LOD) of 1 nM for the detection of streptavidin.²⁰⁶ Gubeljak et al. developed an electrochemically gated graphene broadband microwave waveguide, which enabled the unambiguous and reproducible discrimination of single-base mismatch target strands at low DNA concentrations, down to 1 aM.²⁰⁷ These accomplishments underscore the extraordinary potential of ultrahigh frequency operation to unlock the true potential of graphene biosensors, particularly for point-of-care diagnostics under physiological conditions with high salt concentrations.

It should be noted that direct adsorption of DNA probes on graphene channel may not provide sufficient stability in the presence of surfactants or complex solvents.²¹ These unstable DNA probes easily detached from the channel can result in failure to identify and bind targets, which reduces the performance of DNA probe–GFETs. To improve the performance of DNA probe–GFETs, many attempts have been made to enhance the stability and loading density of DNA probes on graphene channel.^54,59,209 These strategies involve the immobilization of pyrene-modified DNA probes on graphene surface, or the use of linker molecules. For instance, 1-pyrenebutanoic acid succinimidyl ester (PASE), which is commonly used as a linker, can be functionalized on graphene surface via π–π stacking between the pyrenyl groups at one end and the graphene plane. The amine-modified aptamer is then covalently attached to the succinimidyl ester group at the other end of the linker using NHS chemistry.⁵⁴ Other molecular linkers, such as 1-pyrenebutyric acid N-hydroxysuccinimide ester (PBASE) and poly-L-lysine (PLL), have also been employed for the immobilization of DNA probes.^210,211

Capitalizing on the benefits of DNA probes and GFETs, the DNA probe-empowered GFET biosensors have opened a wide range of applications and achieved numerous advancements in the detection of various biomolecules. Here are several recent notable examples highlighting the use of DNA probe-empowered GFETs for ultrasensitive and selective detection of DNA, RNA, small molecules, and proteins.

1. DNA detection

DNA probe–GFETs could produce measurable electronic responses upon the adsorption and hybridization of DNA on graphene channels. It was reported that the adsorption and hybridization of DNA probes on GFET channel can trigger either a left or right shift in the charge neutrality point voltages (V_cnp).^{8,198,212,213} Right shift of V_cnp is typically ascribed by the p-doping of graphene. In this scenario, negatively charged DNA molecules induce a negative electric field effect on graphene, thereby facilitating the injection of extra holes into the graphene layer, resulting in the observed right shift of V_cnp.^212,214 Conversely, left shift is generally attributed to the n-doping of graphene caused by the π-π stacking interaction between graphene and electron-rich nucleobases in DNA molecules.^198,215,216 For instance, Xu et al. observed a right shift of V_cnp when introducing DNA probes into GFETs system, which was explained by p-doping effect. The V_cnp shifted toward higher positive gate voltage upon the hybridization with DNA targets, indicating an increased binding of negative charges to the DNA probe-modified GFET [Fig. 10(a)].²¹³ In contrast, Hwang et al. found that the physical contact of DNA probes imposed n-doping effect on graphene, causing a left shift of V_cnp. The V_cnp kept shifting left with the increased concentration of DNA targets, indicating a progressively increased graphene–nucleotide stacking interaction [Fig. 10(a)].⁸ Furthermore, a charge transfer effect induced n-doping of graphene was observed in MoS₂/GFET sensors. In this sensor, MoS₂ (polar molecule) can easily induce polarization of DNA molecule (nonpolar molecule) and reduce the distance between them, which permitted direct charge transfer from DNA to MoS₂/graphene nanostructure, resulting in a left shift of V_cnp.²⁰²

FIG. 10.

(a) (i) Transfer characteristics of GFETs for the bare graphene and after each addition of the following reagents in sequence, PBASE, probe DNA P20, complementary DNA T20 and unrelated control DNA U20. Reproduced with permission from Xu et al., Nat. Commun. 8, 14902 (2017). Copyright 2017 Springer Nature. (ii) Transfer characteristics of GFETs for the DNA hybridization. Reproduced with permission from Hwang et al., Nat. Commun. 11, 1543 (2020). Copyright 2020 Springer Nature. (b) (i) Schematic of a DNA probe–GFET biosensor for DNA detection. (ii) Relative Dirac voltage shift as a function of concentration for DNA targets of different lengths. Reproduced with permission from Ping et al., ACS Nano 10(9), 8700–8704 (2016). Copyright 2016 American Chemical Society. (c) (i) Hairpin-probe DNA bound to a back-gated GFET using a pyrene linker. (ii) The principle of the triggered self-assembly amplification for DNA detection on the GFET. Reproduced with permission from Gao et al., Nano Lett. 18(6), 3509–3515 (2018). Copyright 2018 American Chemical Society. (d) Schematic of a Y-dual probe GFET biosensor. Reproduced with permission from Kong et al., J. Am. Chem. Soc. 143(41), 17004–17014 (2021). Copyright 2021 American Chemical Society. (e) (i) Scheme of the flat (left) and crumpled (right) GFET DNA sensors. (ii) Fabrication process of the flat and crumpled GFETs. Graphene on pre-strained PS substrate was annealed at 110 °C to shrink the substrate and crumple the graphene. Then, source and drain electrodes were applied and solution-top gate was used. In the case of flat GFET, the annealing process was omitted. Reproduced with permission from Hwang et al., Nat. Commun. 11, 1543 (2020). Copyright 2020 Springer Nature.

Discrepancies among these literature types for V_cnp shifts are potentially attributed to differences in device and sample designs, as well as the composition and concentration of electrolytes. In particular, the liquid gate electrolyte plays an essential role in the performance of GFETs, e.g., a higher ionic strength promotes DNA hybridization, but reduces the detection sensitivity; a lower ionic strength increases the detection sensitivity due to an expanded Debye length, but when the ionic concentration falls too low, strand repulsion may destabilize the double-helix conformation.²¹⁷ Chen et al. employed phosphate-buffered saline (PBS) with varying concentrations as buffer solutions (10×, 1×, and 0.1× PBS) for DNA hybridization. They found that V_cnp consistently shifted left in varied PBS concentrations as the concentration of DNA targets increased. A larger shift in V_cnp was observed within lower PBS concentration, implying a higher detection sensitivity as a consequence of the extended Debye length.²¹⁸ In contrast, another study reported the right shifts of V_cnp under different ionic strength and electrolyte type [phosphate buffer (PB, 0.1, 1, and 10 mM), and PBS (0.01×, 0.1×, and 1×)]. These shifts were attributed to the electrostatic gating on GFET by the negatively charged DNA molecules, as previously mentioned. A more pronounced V_cnp shift was also observed within lower ionic strength solutions.²¹⁹ Overall, despite these discrepancies, measurable electronic responses to the adsorption and hybridization of DNA on GFET channels showcase the potential of developing label-free and highly sensitive DNA probe–GFET biosensors.

Many strategies have been employed to enhance the sensitivity of DNA probe–GFETs for DNA sensing, including increase in DNA probe length, design of DNA probe structures, improvement of graphene quality, and modulation of the Debye screening effect. Researchers found that longer DNA probe lengths resulted in a higher surface coverage of DNA probes and a stronger binding affinity between the longer probes and their targets. By increasing the DNA probe length to 60 mer, a remarkable improvement of detection limit of 1 fM was achieved [Fig. 10(b)].²⁰¹ The design of DNA probes can address the length-dependent sensitivity of DNA probe–GFETs. For instance, using a hairpin-probe DNA can trigger a self-assembly pathway for target recycling and a hybridization chain reaction, amplifying the signal and improving the detection limit to sub-fM for 21-mer target DNA [Fig. 10(c)].²²⁰ A Y-shaped DNA dual probe was ingeniously designed to simultaneously target ORF1ab and N genes of severe acute respiratory syndrome (SARS)-CoV-2 nucleic acids [Fig. 10(d)].²²¹ This innovative design achieved a higher recognition ratio for SARS-CoV-2 nucleic acids with a remarkable detection limit as low as 0.03 copy μl⁻¹ (50 zM), attributing to the synergistic effect between the two probe sites. Furthermore, the DNA probe–GFET sensing performance was enhanced by improving graphene quality. A gold transfer method was established to obtain cleaner graphene and achieved a 125% improvement in sensing performance compared to traditional poly (methyl methacrylate) (PMMA) transfer method.²¹⁸ A straightforward room-temperature Ar plasma treatment was employed to remove residues from graphene surface and alter its hydrophilic property. The fabricated DNA probe–GFETs demonstrated a robust response as the concentration of target DNA was increased from 10 aM to 100 fM, with a detection limit of 10 aM.²²² In recent years, several approaches have been employed to enhance the sensitivity of DNA probe–GFETs by modulating Debye screening effect. One of such approaches involves the incorporation of a DNA/MoS₂/graphene hybrid structure into a FET biosensor, to weaken the Debye screening effect.²⁰² DNA probes were immobilized on the MoS₂/graphene surface using PBASE as connectors. This approach exhibited an effective electronic response to DNA target concentrations ranging from 10 aM to 100 pM, with a low detection limit of 10 aM. Crumpled graphene, previously fabricated through macroscopic manipulation of a graphene layer, has also showed the ability to modulate Debye screening [Fig. 10(e)].⁸ DNA probes were anchored on crumpled graphene surface using PASE as linkers. The resulting crumpled DNA probe–GFETs exhibited a detection limit of 20 aM. The performance of the GFET was further enhanced by using peptide nucleic acid probes, resulting in a detection limit of 600 zM. Leveraging the unique properties of crumpled graphene, Ganguli et al.²⁰⁵ fabricated crumpled DNA probe–GFET sensors. They demonstrated superior sensitivity of the sensors in detecting enzymatic DNA amplification, by monitoring the reduction in target-specific primer molecules. An ultralow detection limit of 8 zM was achieved.

2. RNA detection

DNA probe–GFETs have emerged as rapid, highly sensitive, and cost-effective strategies for measuring RNA expression levels by surface functionalization of graphene channel. It was reported that direct immobilization of DNA probes on a GFET channel though π-π stacking enables the sensitive miRNAs detection, achieving a remarkable detection limit of 10 fM [Fig. 11(a)].²²³ To further enhance the performance of GFET biosensors, surface functionalization of GFET channel is crucial. Previous approaches involve the functionalization of GFET biosensors with AuNPs or introducing PBASE on GFET channel. These approaches improve the loading density and stability of DNA probes on the channel, enhancing the biosensor’s performance. Cai et al.⁷¹ employed PASE as a cross-linker to densely anchor DNA probes on rGO surface. The constructed DNA probe–rGO-based FET demonstrated high specificity in distinguishing miRNA-21 with perfect matching, single-base mismatching, and complete mismatching, allowing for detecting miRNA-21 in the plasma of breast cancer patients with an improved detection limit of 1 fM [Fig. 11(b)]. This device was capable of detecting complex matrix samples from clinical settings, without preprocessing or signal amplification steps. This development holds significant promise for practical application in clinical settings. In addition, PLL was used to functionalize graphene channels and increase the density of DNA probes via electrostatic interactions.²¹¹ The constructed biosensor was proved to be highly effective in the detection of breast cancer miRNAs and SARS-CoV-2 viral RNAs [Fig. 11(c)]. It exhibited a low detection limit of 1 fM, provided results within a short detection time of 20 min using only 2 μl of human serum and throat swab samples, and showed a sensitivity enhancement over 113% compared to conventional GFET biosensors. These findings highlight the potential of this approach for rapid cancer diagnosis and virus screening.²¹¹

FIG. 11.

(a) Schematic illustration of miRNA detection by DNA probe–GFET biosensor. The DNA probe is immobilized on graphene channel. The hybridization of complementary miRNA with the probe causes n-doping of the device and the shift in Dirac point, while noncomplementary miRNA does not induce an apparent Dirac point shift. Reproduced with permission from Gao et al., ACS Appl. Electron. Mater. 2(4), 1090–1098 (2020). Copyright 2020 American Chemical Society. (b) The integration of FET with rGO for the detection of miRNA-21 in breast cancer patients. DNA probe is modified on rGO channel via a cross-linker. Hybridization of the probe with target miRNA-21 induces a shift in Dirac point, enabling the miRNA-21 detection. Reproduced with permission from Cai et al., ACS Appl. Nano Mater. 5(8), 12035–12044 (2022). Copyright 2022 American Chemical Society. (c) (i) Schematic of poly-L-lysine-modified GFET (PGFET). (ii) and (iii) Comparison of miRNA detection between GFET and PGFET biosensors. Compared to GFET, PGFET biosensor can give a more obvious Dirac point shift for DNA probe–miRNA hybridization. Reproduced with permission from Gao et al., Anal. Chem. 94(3), 1626–1636 (2022). Copyright 2022 American Chemical Society.

3. Small-molecule detection

DNA probe–GFET biosensors have proven to be powerful tools for the quantification of environmental pollutants, such as antibiotics and toxic heavy metals, by improving graphene quality, functionalizing graphene channels, or rationally designing DNA aptamers. For detecting antibiotics, ultraclean graphene was fabricated by employing a novel camphor-rosin clean transfer (CRCT) approach [Fig. 12(a)].²²⁴ This approach led to a remarkable improvement in the carrier mobility of the fabricated GFET, exceeding a tenfold increase compared to that prepared using conventional PMMA transfer method. DNA aptamers were subsequently immobilized on the graphene channel using PLL molecules as connectors. The resulting biosensor exhibited exceptional performances, including a dynamic detection range spanning five orders of magnitude, a sensitivity of 21.7 mV/decade, and a low detection limit of 100 fM. For metal ion detection, DNA aptamers were designed and immobilized on graphene channels using cross-linkers.²²⁵ For example, Li et al.²²⁶ immobilized arsenite-specific aptamers on the graphene channel using PBASE molecules. Upon target binding, the aptamer underwent conformational changes, altering the proximity of the negatively charged aptamer backbone to the graphene surface and inducing changes in electrostatic charge. This led to highly sensitive detection of arsenite with a wide detection range from 0.05 to 1000 ppb, and a low detection limit of 0.02 ppb [Fig. 12(b)].

FIG. 12.

(a) (i) GFET construction based on camphor-rosin clean transfer (CRCT) and PMMA-assisted transfer of graphene. (ii) Tetracycline detection by the CRCT-GFET. Reproduced with permission from Wang et al., Anal. Chem. 94(42), 14785–14793 (2022). Copyright 2022 American Chemical Society. (b) (i) Schematic illustration of the arsenite detection mechanism in a GFET biosensor array. (ii) Transport characteristics expressed as curves of the biosensor, measured upon the exposure to 0.5–1000 ppb arsenite solutions (left panel). Biosensor response as a function of target ion concentrations (right panel). Reproduced with permission from Li et al., ACS Appl. Nano Mater. 5(9), 12848–12854 (2022). Copyright 2022 American Chemical Society.

4. Protein detection

In the context of protein detection, DNA probe–GFET biosensors have been developed by functionalizing DNA aptamers on the graphene channel to capture target proteins. As described before, DNA aptamers can be directly immobilized on GFETs.²²⁷ To enhance the stability and anchoring density of DNA aptamers, cross-linker molecules were utilized as bridges to assist the indirect immobilization of aptamers on graphene channel. In a study by Hugo et al.,²²⁸ matrix metalloproteinase 9 (MMP-9)-specific aptamers were integrated on the GFET using pyrene-maleimide linkers [Fig. 13(a)]. The biosensor achieved a detection limit of 97 pM, providing a sensitive tool for MMP-9 detection. Ban et al.²¹⁰ immobilized SARS-CoV-2 coronavirus spike (S)- and nucleocapsid protein (N)-specific aptamers on GFETs using PBASE linkers [Fig. 13(b)]. Results showed that both the S and N aptamers were capable of detecting receptor-binding domain proteins and nucleocapsid proteins of SARS-CoV-2 in concentrations ranging from fM to nM levels. Chen et al.⁵⁶ further anchored DNA strands on graphene channel using PASE as a linker. The introduction of target VEGF165 can trigger a surface hybridization chain reaction [Fig. 13(c)], which greatly enhanced the negative shift in the Dirac voltage, enabling highly sensitive detection of VEGF165 with a low detection limit of 3.24 pg/ml.

FIG. 13.

(a) Matrix metalloproteinase-9 (MMP-9) sensing in swab samples from patients with diabetes and open wounds using a liquid-gated GFET biosensor. Reproduced with permission from Adrien Hugo et al., Biosens. Bioelectron. X 13, 100305 (2023). Copyright 2023 Elsevier. (b) DNA probe–GFET biosensor diagnostic system for ultrasensitive detection of SARS-CoV-2 using patient samples. Reproduced with permission from Ban et al., Proc. Natl. Acad. Sci. U.S.A. 119(28), 2206521119 (2022). Copyright 2022 National Academy of Science. (c) Sensing strategy of VEGF165 detection based on the designed GFET biosensor and DNA signal amplification. Reproduced with permission from Chen et al., Sens. Actuators B Chem. 351, 130964 (2022). Copyright 2022 Elsevier. (d) Protocol to develop a covalent DNA probe–GFET biosensor and the molecular antenna effect: the covalently bound pATP linker acts as a molecular antenna, enhancing the local polarization of graphene due to the proximity of the aptamer/protein system, consequently influencing the position of the Dirac cone. Reproduced with permission from Palacio et al., Biosens. Bioelectron. 222, 115006 (2023). Copyright 2023 Elsevier.

Since the functionalization process significantly impacts the performance of DNA probe–GFET biosensors, Palacio et al.²²⁹ proposed a two-step functionalization process to improve the performance of biosensors. In the first step, low-energy Ar⁺ ions were used to create single-atom vacancies in graphene lattice. In the second step, pATP was utilized as linkers by covalently coupling to graphene surface. Subsequently, thiol-modified ssDNA aptamers were added to bind to the linkers via covalent disulfide bonds [Fig. 13(d)]. Theoretical calculations conducted in this work revealed that this two-step functionalization induced polarization at the graphene surface, which modulated the charge balance at the graphene/aptamer interface. This effect resulted in a net shift of the Dirac cone, enabling the detection of the hepatitis C virus (HCV) core protein in human blood plasma with attomolar sensitivity.²²⁹ The high sensitivity opens up possibilities for early diagnosis of HCV infection and holds potential for clinical applications.

5. DNA sequencing

In addition to the DNA probe–GFET sensors mentioned above for detecting DNA, RNA, and proteins, various concepts have been proposed to use graphene for DNA sequencing. These concepts encompass: (1) ion current detection as DNA translocates through a graphene nanopore, (2) in-plane current detection as DNA translocates through a graphene nanoribbon, (3) tunneling current detection as DNA passes across a graphene nanogap, and (4) current detection based on DNA adsorption on a graphene nanostructure [Fig. 14(a)].^19,230

FIG. 14.

(a) Four concepts for DNA sequencing using graphene nanostructures. (i) Graphene nanopore. (ii) Graphene nanoribbon. (iii) Graphene nanogap. iv. Physisorption of DNA bases onto graphene nanostructure. Reproduced with permission from S. J. Heerema and C. Dekker, Nat. Nanotechnol. 11(2), 127–136 (2016). Copyright 2016 Springer Nature. (b) Schematic diagram of graphene nanopore. Reproduced with permission from Garaj et al., Nature 467(7312), 190–193 (2010). Copyright 2010 Springer Nature. (c) Simulation box showing the graphene/hexagonal boron nitride (hBN) nanopore with translocating DNA. Reproduced with permission from Balasubramanian et al., ACS Appl. Bio Mater. 4(1), 451–461 (2020). Copyright 2020 American Chemical Society. (d) (i) Modeling ssDNA stretching on the graphene/h-BN/graphene heterostructure. (1) The initial linear conformation of ssDNA on the graphene domain. (2) and (3) Conformations of ssDNA at 181 and 311 ns, respectively. (ii) Electrically driven stretching of ssDNA on the heterostructure at different times. Reproduced with permission from B. Luan and R. Zhou, Nat. Commun. 10, 4610 (2019). Copyright 2019 Springer Nature.

Graphene has proven to be an exceptional material for nanopore DNA sequencing, due to several key advantages: (1) the intrinsic feature of ion impermeability, (2) the nearly identical thickness to the nucleotide interval of ssDNA (~0.6 nm), and (3) high sensitivity to various diameters of translocating polymers.^231,232 In a standard graphene nanopore setup, the nanopore is suspended on a prepatterned silicon nitride film, forming a barrier between two chambers filled with an electrolytic solution. By applying a voltage across the two electrodes, ionic current is compelled to flow through the nanopore. As DNA molecules translocate through the nanopore, they instantaneously reduce or block the ionic current, yielding a signal that reflect their structural conformations and polymer diameters [Fig. 14(b)].²³² Since each DNA nucleotide exhibits distinct molecular size, shape, and conformation, it will generate different blockade signals, thereby offering the potential for DNA sequencing.²³³

Nonetheless, achieving single-base resolution poses substantial challenges, primarily stemming from the fast translocation of DNA, conformational fluctuations, stochastic translocation, nonspecific interactions between DNA and graphene, stochastic pore clogging, and high ionic current noise.²³⁰ Multiple methods have been proposed to address these challenges. For instance, the current methods on slowing down the speed of DNA translocation involve using stick-slip interactions, reducing the nanopore size, increasing the solution’s viscosity, or using hydrophobic interactions between graphene and DNA.^234,235 One example employed multilayered graphene-dielectric nanopores (stacked graphene–Al₂O₃ structures) to slow down ssDNA translocation, which is based on the hydrophobic interactions between ssDNA and graphene.²³⁶ Methods aimed at reducing noise rely on cleaning graphene surface, fabricating multilayered graphene-dielectric architecture or coating graphene using atomic layer deposition.^237,238 It was reported that coating graphene surface with a thin TiO₂ layer can substantially reduce ion current noise.^237,238 Despite the progress and dedicated efforts discussed above, the challenges mentioned above remain in graphene nanopore-based sequencing.

Alternatively, methods utilizing the intrinsic conductivity of graphene have been proposed. In graphene nanoribbon-based sequencing, the basic principle is to monitor the current passing through a graphene nanoribbon that contains a nanopore. As DNA bases pass through the nanopore, there are base-specific modulations in the electronic current, allowing for the measurement of DNA sequence.²³⁹ Graphene nanogap-based sequencing is to measure the tunneling currents between two graphene electrodes separated by a gap. Variations in current reflect the translocation details of a DNA molecule.²⁴⁰ Additionally, when a DNA strand passes through a nanochannel, non-covalent stacking interactions between DNA bases and graphene can also induce base-dependent current modulations.^19,230 These methods provide exciting prospects for DNA sequencing with graphene nanodevices, whereas several challenges persist, such as the precise fabrication of graphene nanostructures with well-defined width and edges, as well as the control of DNA molecule motion during translocation through or alongside the graphene nanostructures.

Recently, heterostructures of 2D materials have shown great potentials for controlling DNA motions during sequencing. Particularly, graphene/hexagonal boron nitride (hBN) heterostructures have been used to control DNA motions and distinguish individual bases.^241–243 Two distinct configurations of graphene/hBN heterostructures have been developed: the vertical and in-plane lateral heterostructures. Balasubramanian et al. investigated the electric-field-driven translocation of dsDNA through a nanopore embedded in a graphene/hBN vertical heterostructure. Their findings revealed that the presence of hBN on the trans side led to dsDNA translocating almost parallel to the applied field, resulting in eliminated DNA adsorption and enhanced base-pair specificity of the heterostructure [Fig. 14(c)].²⁴² In comparison, lateral heterostructures could align the biomolecule and guide the molecular motion within a single plane.^243,244 In the patterned lateral graphene/hBN heterostructure [Fig. 14(d)], DNA molecules exhibited a preference for adsorption on the hBN domain over graphene. In addition, the hBN stripe within the heterostructure was found to stretch the DNA into a single-file conformation during the transport driven by external electric fields.²⁴³ The hBN stripe was further designed for DNA sequencing, showcasing the potential for distinguishing different bases.^244,245 Recently, Huang et al. introduced a nanopore sequencing by a nanoslit sensor based on the lateral graphene/hBN heterostructure. This sequencing method comprises two detection modes: cross-slit and trans-slit, which are based on two pathways of DNA: cross and translocate the nanoslit under applied voltages. The structural design and signals obtained from the two detection modes ensure the control of DNA motions and well identification of four DNA bases.²⁴⁶ These studies shed light on the sensing mechanism based on heterostructures and provide theoretical guidance on devising devices that control molecular transport in nanopore sequencing.

VIII. INTERACTIONS OF GRAPHENE-BASED MATERIALS WITH DNA NANOSTRUCTURES

DNA nanostructures have typically been characterized on atomically flat substrate surfaces using methods such as electrostatic attraction,²⁴⁷ dielectrophoresis,²⁴⁸ or by incorporating surface-binding functional groups on the DNA nanostructures.²⁴⁹ Chemically modified graphene sheets, such as GO, nitrogen-doped reduced GO,²⁵⁰ and amino-functionalized GO,²⁵¹ offer atomically smooth and chemically diverse surfaces. These substrates have demonstrated excellent capabilities for adsorption and spatial patterning of DNA nanostructures without compromising their structural integrity. However, DNA nano-structures experienced severely expansion, melting, and distortion, due to π-π stacking interactions when they were deposited on certain graphene-based substrates, such as graphite,^35,36 monolayer graphene supported by mica,³⁷ and free-standing graphene.²⁵² For example, Norton’s group observed that cross-shaped DNA origami underwent rapid reconstruction upon binding to graphite [Fig. 15(a)].³⁵ Such reconstruction effect was reflected in a significant loss of sharply defined structural features, expansion of surface area, and decrease in height. The reconstruction was attributed to strong π-π stacking interactions between ssDNA and graphite surface, which competed with hydrogen bond interactions in dsDNA. Despite the structural changes, the reconstructed origami could still organize proteins on graphite surface. In addition, the reconstructed structures can maintain their morphologies for at least a week, and promote site-selective chemical vapor deposition of SiO₂.³⁶ In an attempt to prevent this reconstruction, Norton’s group further deposited cross-shaped DNA origami on monolayer graphene supported by mica instead of graphite. Graphene was known to be partially “transparent” to the properties of its underlying substrate, which was expected to reduce the destabilizing electronic effects of graphite and increase the contribution of local ions to origami adsorption, thus preventing structural expansion of origami. Surprisingly, DNA origami was found to be completely disrupted when deposited on mica-supported monolayer graphene [Fig. 15(b)], in contrast to the expansion and partial re-organization on graphite.³⁷ This finding suggested that DNA nanostructures were sensitive to the modified properties of monolayer graphene influenced by its substrate. Furthermore, Kabiri et al.²⁵² characterized the conformational polymorphism of DNA origami on free-standing graphene [Fig. 15(c)]. They observed that origami maintained their well-formed structures on amorphous carbon supports, but experienced severe distortion upon deposition onto graphene. This distortion was attributed to the hydrophobic interaction between DNA bases and graphene.

FIG. 15.

(a) Schematized streptavidin-labeled DNA origami (left). AFM images of streptavidin-labeled DNA origami on mica (middle) showcase their well-defined structure, and on highly oriented pyrolytic graphite (HOPG) (right), where they underwent rapid reconstruction. Scale bars: 500 nm. Reproduced with permission from Rahman et al., Nanomaterials 6(11), 196 (2016). Copyright 2016 MDPI. (b) AFM height and phase images show complete disruption of DNA origami upon deposition on mica-supported monolayer graphene. Scale bars, 100 nm. Reproduced with permission from Green et al., Mater. Res. Express 5(4), 045035 (2018). Copyright 2018 IOP Publishing. (c) (i) Design schematic of DNA origami. (ii) AFM image of DNA origami on mica. (iii) STEM image of uranyl acetate-stained origami on amorphous carbon. Uranyl acetate-stained origami (iv) and unstained origami (v) on suspended graphene. DNA origami maintained their well-formed structures on mica and amorphous carbon, but experienced severe distortion upon deposition on graphene. Scale bars, 100 nm. Reproduced with permission from Kabiri et al., Small 13(31), 1700876 (2017). Copyright 2017 John Wiley and Sons.

The structural disruption of DNA nanostructures on graphene poses a significant challenge for their integration into graphene-based devices. Several methods have been explored to address this issue, including the modifications of substrates^252–254 and DNA nanostructures.^13,14,33 One approach involves the passivation of the graphene/graphite surface using ssDNA through strong π-π interactions.³⁵ By depositing DNA nanostructures onto graphite, excess ssDNA staples in the solution can intervene at the interface and modify the interactions, thereby preventing structural damage to the DNA nanostructures. Other studies have explored the passivation of graphene surfaces using PLL and pyrene derivatives, such as 1-pyrenemethylamine and 1-pyrenecarboxylic acid.^252–254 In particular, these pyrene molecules have aromatic rings that can easily stack on graphene’s hydrophobic basal plane through π-π interactions. Additionally, their functional groups, such as –OH and –COOH, can impart hydrophilicity to graphene.²⁵⁵ Coating the graphene surface with these molecules helps shield the partial stacking interactions between DNA bases and graphene, thereby improving the deposition of DNA nanostructures onto graphene. Another approach to immobilize DNA nanostructures on graphene involves incorporating pyrene molecules into the DNA nanostructures themselves.^13,14,33 By modifying ssDNA with pyrene groups, the modified ssDNA can hybridize with the DNA nanostructures while also binding to graphene through π-π stacking interactions. This method enables the stable immobilization of DNA nanostructures with different geometries on graphene.³³ In a recent study, a new strategy was reported for immobilizing DNA tetrahedron on graphene. First, non-covalent functionalization of monolayer graphene was performed using PASE linkers, which interacted with graphene through stacking interactions. Then, the DNA tetrahedron was modified with amino groups at each vertex located at the bottom. Finally, the amino groups of the DNA tetrahedron were covalently linked to the PASE molecules, resulting in the immobilization of the DNA tetrahedra on graphene.³⁴

IX. ADVANCING GRAPHENE-BASED BIOSENSORS BY INTERACTIONS OF GRAPHENE-BASED MATERIALS AND DNA NANOSTRUCTURES

In this section, we mainly focus on the use of graphene and its integration with DNA nanostructures, due to the limited recent research on the optical and electrical sensing by GO integrated with of DNA nanostructures. For the same reason, we divided this section into optical and electronic sensing as follows.

A. Optical sensing

Benefit from the tunable interaction at the interface, DNA nanostructures can maintain their structural integrity upon adsorption on the graphene surface. This feature facilitates the seamless integration of DNA nanostructures with graphene and unlocks new possibilities for fluorescent sensing.^13,14,33 dsDNA was previously utilized as a rigid spacer to systematically control the distance between fluorophore and GO.⁵⁰ The study demonstrated that graphene exhibited a long-range quenching capability with a quenching efficiency following a d⁻⁴ dependency, where d represents the fluorophore-to-graphene distance. 50% quenching occurred at a separation distance of 7.5 nm, whereas different orientations of DNA or DNA dynamics may affect the results. Compared to dsDNA, DNA nanostructures with customized architectures allow the placement of arbitrary objects and complex biomolecular assays at predetermined positions. Tinnefeld’s group introduced DNA nanostructures as nanopositioners to revisit the distance dependence of energy transfer to graphene.³³ Despite the well-understood distance dependence, reported values of d₀ (the distance at which 50% quenching efficiency occurs) varied widely due to different fluorophore emitters and distance control methods. To tackle this, the authors positioned single, freely rotating fluorophores at specific locations within DNA nanostructures, allowing precise control over the distance between fluorophores and graphene. They further immobilized the DNA nanostructures onto graphene surface using pyrene-modified ssDNA as mentioned above [Fig. 16(a)]. Through the measurements of fluorescence intensity and lifetime of individual emitters at distances ranging from 3 to 58 nm, they confirmed the d⁻⁴ dependence of excitation energy transfer to graphene, with a precise value of d₀ determined as 17.7 nm.³³ Tinnefeld’s group utilized the same immobilization method to anchor DNA nanostructures on a graphene surface and perform a series of assays that demonstrated several innovative capabilities [Fig. 16(b)]: (1) detecting dynamics with graphene electron transfer (GET) (2) determining the orientation of a FRET pair with respect to the substrate, (3) biosensing using graphene as a quencher, (4) GET tracking of a dye-labeled DNA pointer that can transiently hybridize to three protruding strands on the DNA nanostructure, and (5) GET-DNA PAINT super-resolution imaging with <3 nm z-resolution.¹⁴

FIG. 16.

(a) (i). Schematics of DNA origami structures on graphene. A single ATTO542 fluorophore (green sphere) is positioned at different heights in DNA origami. Blue frame: zoom-in of pyrene-modified (orange) DNA strand protruding from DNA origami and interacting with graphene via π-π interactions. (ii) Normalized fluorescence intensity decays of ATTO542 at difference distances to graphene. (iii) Relative fluorescence intensity as a function of fluorescence lifetime. Reproduced with permission from Kaminska et al., Nano Lett. 19(7), 4257–4262 (2019). Copyright 2019 American Chemical Society. (b) (i) Schematic of a L-shaped DNA origami with a flexible pointer with fluorescent dye, and upper and lower binding strands (left panel). Representative transients for 8 (gray), 7 (blue), 6 (green), and 5 (lilac) nt binding (right panel). (ii) Illustration of the pillar-shaped DNA origami with three different acceptor–donor orientations: horizontal, diagonal, and vertical. (iii) Sketch of graphene biosensing. (iv) Graphene electron transfer (GET) tracking of a dye-labeled DNA pointer that can transiently hybridize to three single-stranded protrusions on the DNA origami. (v) GET-super-resolution on a cubic DNA origami with DNA PAINT. Scale bar: 500 nm. Reproduced with permission from Kaminska et al., Adv. Mater. 33(24), 2101099 (2021). Copyright 2021 John Wiley and Sons. (c) Scheme of DNA origami on graphene for characterizing the quality of graphene-on-glass. Reproduced with permission from Krause et al., ACS Nano 15(4), 6430–6438 (2021). Copyright 2021 American Chemical Society.

Using DNA nanostructures on graphene surface further showcases the potential for assessing graphene quality through fluorescence quenching tests. Despite the outstanding fluorescence quenching properties of graphene, the widespread utilization of large-scale CVD graphene as a 2D quenching material for single-molecule imaging poses challenges. The transfer and cleaning methods currently employed inevitably introduce impurities and defects onto the graphene surface,^256,257 leading to inadequate quenching or nonspecific binding of DNA nanostructures. To address this issue, Tinnefeld’s group employed a DNA nanopositioner to evaluate the quality of graphene through fluorescence quenching experiments [Fig. 16(c)].¹³ The design of the nanopositioner allowed the fluorophore to be positioned close to the height where 50% quenching occurs. This design enabled the fluorophore to be most sensitive to small changes in energy transfer, which could reflect variations in height or defects in the graphene. By combining fluorescence lifetime imaging with other conventional characterization techniques, the authors successfully achieved an efficient evaluation and screening of graphene quality for fluorescence-based applications.

B. Electronic sensing

The integration of DNA nanostructures into graphene-based materials has facilitated the design of sensing interfaces and manipulation of systems at the molecular level, to advance the detection sensitivity and specificity of proteins, small molecules, ions and nucleic acids in complex, and high-ionic-strength biofluids. Recently, Wang et al.³⁴ introduced a molecular electro-mechanical system (MolEMS) and immobilized on the channel of a liquid-gated GFET for rapid and ultrasensitive detection [Fig. 17(a)]. This MolEMS consisted of an aptamer probe attached to a ssDNA cantilever, which was linked to a stiff DNA tetrahedron structure. The immobilization process involved the non-covalent functionalization of PASE on graphene, as well as the covalent linking of PASE to amino group-functionalized DNA tetrahedron. The constructed device exhibited excellent performance characteristics, including rapid detection, easy operation, high sensitivity, specificity, and portability. This was due to the dense accommodation of MolEMS on graphene surface with a highly ordered upright orientation and controllable distance, the ability of MolEMS to overcome Debye screening, along with the high electron mobility and structural flexibility of graphene. It enabled specific detection of small molecules, ions, proteins, and nucleic acids, down to 5 × 10⁻²⁰ M in biofluids or high-ionic-strength buffers. Moreover, it allowed for the direct detection of unamplified SARS-CoV-2 RNA, with concentrations as low as ~0.02 copies per μl RNA in nasopharyngeal swab samples. In their continued research, a GFET biosensor was developed that was modified with triple-probe tetrahedral DNA framework (TDF) dimers for SARS-CoV-2 RNA testing [Fig. 17(b)].²⁵⁸ The triple-probe design synergistically enhanced the binding affinity and enabled the biosensors to directly detect SARS-CoV-2 RNA in artificial saliva at concentrations ranging from 4 × 10⁻²⁰ to 8 × 10⁻²⁰mol/l without the need for amplification. In addition, unlike ssDNA probes, the structure of TDF dimer can effectively prevent the probe DNA intertwining with each other and avoid nonspecific biomolecule adsorption on the graphene surface. The synergy effect of triple probes as well as the special structure of the TDF dimer, combined with efficient signal transduction of GFET, gave rise to the high sensitivity and accuracy of the biosensor. This biosensor enabled 10-in-1 nucleic acid pooled testing, holding great potential for the screening of COVID-19 and other epidemic diseases.

FIG. 17.

(a) The device configuration of molecular electro-mechanical system (MolEMS) GFET (i) and the electrostatic actuation of MolEMS (ii). MolEMS comprises an aptamer probe attached to a ssDNA cantilever, which is linked to a stiff DNA tetrahedron structure. When applying a positive or negative gate voltage by the liquid electrolyte (V_lg), a local electrical field lifts up or pushes down the cantilever, causing controlled cantilever movements in the upper or lower regions of the MolEMS. The aptamer probe conjugated at the cantilever tip enables specific biorecognition at controlled regions. Reproduced with permission from Wang et al., Nat. Biomed. Eng. 6(3), 276–285 (2022). Copyright 2022 Springer Nature. (b) The triple-probe tetrahedral DNA framework (TDF) dimer GFET biosensor for SARS-CoV-2 RNA testing. The triple-probe TDF dimer is immobilized on the GFET device through molecule linkers. The hybridization of the three probes with target RNA induces a pronounced Dirac voltage shift, enabling ultrasensitive RNA detection. Reproduced with permission from Wu et al., Nano Lett. 22(8), 3307–3316 (2022). Copyright 2022 American Chemical Society.

X. CONCLUSIONS AND FUTURE DEVELOPMENTS

The integration of DNA probes/DNA nanostructures with graphene-based materials offers significant potential for creating sophisticated hybrid nanomaterials with exceptional biosensing capabilities. This integration benefits from the specific recognition ability of ssDNA, the programmability and addressability of DNA nanostructures, and the optical/electronic/chemical sensitivity of graphene-based materials. To achieve successful integration of these components, it is essential to understand the interactions between DNA and graphene-based materials. This understanding enables precise control and manipulation of their interactions, leading to the development of advanced molecular detection platforms with enhanced accuracy, sensitivity, and specificity. Considerable progress has been achieved in the study and design of biosensors based on DNA and graphene-based materials to detect a wide range of targets. However, there are several important considerations and challenges that require attention and resolution in the future research:

1. There is still much to learn about the interaction between DNA and graphene-based materials, as well as the structural changes of ssDNA upon binding to graphene-based materials, and the conformational transitions of ssDNA upon target recognition. It is essential to continue research efforts that aim to unravel the details of these interactions and processes to drive advancements in the design and optimization of DNA–graphene hybrid systems.

2. The stabilization of DNA nanostructures when deposited on the surface of graphene-based materials remains a challenge. It is crucial to explore novel methods that can enhance the stability of DNA nanostructures without compromising the electronic properties of graphene-based materials.

3. The size, shape, morphology, thickness, and oxidation degree of graphene-based materials undeniably affect their interactions with DNA. To gain a comprehensive understanding, future studies should emphasize on more quantitative analysis to investigate these factors thoroughly.

4. There are several challenges associated with DNA probe–GO-based optical biosensors: Firstly, these biosensors typically rely on labeling, which is associated with limited labeling efficiency, complex multistep analysis, and the potential for sample contamination. Secondly, ensuring the homogeneity of GO, in terms of size, shape, thickness, and the number of surface functional groups, is challenging. The lack of homogeneity significantly affects the performance evaluation and reproducibility of optical biosensors constructed for accurate biomolecule detection. Thirdly, owing to the water dispersibility, high selectivity and sensitivity, and suitability for both in vitro and in vivo applications, DNA probe–GO hybrids have become the more prevalent choice for optical biosensing when compared to pristine hydrophobic graphene. However, in the context of in vivo applications, it is essential to take into account the safety aspects of these biosensors, including long-term toxicity, cellular uptake mechanisms, intracellular and in vivo metabolic pathways, etc. Finally, one critical challenge is the stability and degradation of DNA probes, as well as the potential for nonspecific probe displacement and the generation of false-positive signals caused by non-target biomolecules in in vivo applications.

5. There are also several challenges associated with DNA probe–GFET-based electrical biosensors. Firstly, a major hurdle in achieving higher sensitivity and detection limits for liquid-gated GFET biosensors is Debye screening. In typical physiological fluids, the Debye length is less than 1 nm, requiring the charged molecules to be within 1 nm of the channel surface for detectability. Possible routes to circumvent the Debye screening effect include the functionalization of graphene channel and the modification of graphene channel’s architecture. These approaches, however, have not entirely resolved the screening effect. Recent progress on operating GFETs at high frequencies suggests that Debye screening can be overcome in physiological conditions, which offers a tantalizing prospect to unblock the true potential of GFET biosensors. Secondly, GFETs based on CVD graphene generally contain surface contaminations and disruptions resulting from the transfer process from metallic to dielectric substrates, which negatively impact the devices’ performance. To mitigate these issues, powerful transferring and cleaning methodologies should be developed to obtain clean, high-quality, and nondestructive graphene. Thirdly, nonspecific interactions of biomolecules with graphene channels may occur in the presence of complex biological sample matrixes, which become another crucial concern for biosensor sensitivity.

6. Expanding the research scope to other 2D nanomaterials, such as transition metal dichalcogenides like MoS₂ and WS₂, for the adsorption of DNA probes/DNA nanostructures, opens up new possibilities. By exploring a broader range of 2D nanomaterials, researchers can uncover unique properties and characteristics that may offer advantages over graphene-based materials or complement their capabilities. Investigating the interactions between these alternative nanomaterials and DNA will provide valuable insights into their potential as platforms for biosensing applications, and contribute to the diversification and advancement of sensing technologies empowered by DNA probes and DNA nanostructures.

DATA AVAILABILITY

Data sharing is not applicable to this article as no new data were created or analyzed in this study.