**1. Introduction**

Two-dimensional layered (2D) nanomaterial (e.g., graphene and its derivatives, transition metal dichalcogenides (TMDs) and other layered nanosheets)-based electrochemical signal amplifications have grea<sup>t</sup> potential for improving both the sensitivity and selectivity of electrochemical biosensors because of their unique physical, chemical, and electrical properties [1–21]. Graphene is a single layer of densely packed carbon atoms with a benzene-ring structure, and is the first known 2D layered nanomaterial [22]. The unique properties of graphene, including its exceptional mechanical strength [23], extremely large surface area (2630 m<sup>2</sup>/g) [24], very high thermal conductivity in the range of ∼3080–5150 W mK−<sup>1</sup> [25], high conductivity [26], good charge carrier mobility [27], and wide potential window [28], endow it with grea<sup>t</sup> applicability in the development of biosensors, and, in particular, electrochemical biosensors [1–5,12,22–24,29]. In addition, based on the molar ratio of carbon to oxygen (C/O), graphene can be roughly divided into two categories, graphene oxide (GO) or reduced graphene oxide (rGO). It is demonstrated that rGO has better electrical conductivity than GO. Because pure graphene lacks an intrinsic band gap and is limited by chemical modification, there is an increasing interest in synthesizing graphene derivatives/nanocomposites and graphene-like 2D nanomaterials. Among the graphene-like 2D nanomaterials, TMDs (e.g., molybdenum disulfide (MoS2) and molybdenum selenide (MoSe2)) show excellent physicochemical properties and remarkable biocompatibility, and also have significant attraction for the fabrication of electrochemical (bio)sensors [6,7,9,14,21,29]. Driven by their unprecedented properties, massive synthetic methods/protocols have been developed for preparing 2D nanomaterials and 2D nanomaterial composites, which involves both physical strategies and chemical approaches, such as dry mechanical exfoliation (e.g., Scotch tape), chemical (e.g., solution-based exfoliation, graphite oxide exfoliation/reduction) and/or electrochemical (oxidation/reduction and exfoliation) processes, chemical vapor deposition (CVD), chemical synthesis, thermal decomposition of SiC wafers and unzipping carbon nanotubes [26,30–43]. In this review, we will not describe the detailed synthetic methods/protocols mentioned above for synthesis of 2D nanomaterials, however, we sugges<sup>t</sup> reading several recently published comprehensive review articles [40–45]. These methods of 2D nanomaterial preparation produce di fferent forms of nanomaterials with a diversity of properties including mechanical, optical, electrical, chemical and biological properties. These diverse properties make 2D nanomaterials suitable for an extensive range of applications, such as drug delivery, in vitro and in vivo imaging, tissue engineering, biosensor construction, and energy conversion and storage [39,43,46]. For biosensor applications, 2D nanomaterials should be extensively characterized because their properties strongly dependent on their characteristics such as thickness or number of layers, morphology, chemical structure and surface functional groups.

Microbial toxins are the general term for a class of substances covering a broad range from small molecules to biomacromolecules (e.g., peptides and proteins), which are produced by living organisms including bacteria, fungus and algae [47–53]. They are widespread throughout the whole world, threatening the health and/or life of humans and livestock, and a ffecting domestic and international trade. For instance, aflatoxin B1 (AFB1, a kind of mycotoxin produced by fungi) has been defined as a group I carcinogen by the World Health Organization (WHO) [53]. Some microbial toxins can generate acute poisonous e ffects even at very low doses, and the co-occurrence of microbial toxins in nature may cause significantly additive and/or synergistic toxicity. In order to e fficiently avoid potential hazards on public health and safety, it is important to precisely and reliably determine the toxins in practical samples from di fferent sources. Liquid chromatography-based methods including high-performance liquid chromatography (HPLC) and high-performance liquid chromatography-tandem mass spectrometry (HPLC/MS/MS) are the gold standards for accurate analysis of toxins [54–59]. Although the HPLC-based methods have high reliability and accuracy, they typically require expensive laboratory facilities and instruments, complex pre-treatment processing of the sample and well-trained operators. These drawbacks strongly limit the application of HPLC-based methods in on-site detections of toxin. Various sensing systems such as surface plasmon resonance (SPR) biosensors, electrochemical biosensors, fluorescence biosensors, colorimetric assays, competitive enzyme-linked immunosorbent assay (ELISAs) and microfluidic immunoassay have been developed for analysis of toxins from di fferent sources including clinical samples, foods, water and feeds [60–66]. Among these biosensing systems, electrochemical biosensors and biotransducers are more attractive because they offer several advantages such as high sensitivity, operational simplicity, relatively low cost, easily miniaturization and suitable on-site analysis [8,11–19,67–69]. These advantages make electrochemical biosensors/transducers of microbial toxins powerful tools in many areas including food, environmental and medical monitoring, disease diagnosis and anti-terrorism security. Owing to the large surface areas and excellent conductivities, the integration of 2D nanomaterials (e.g., graphene and TMDs) and their nanocomposites with electrochemical transducers has grea<sup>t</sup> potential to enhance the analytical performance of electrochemical biosensors for detection of toxins [8,11–19]. For example, since its birth, multiple research initiatives on graphene applied to electroanalytical chemistry have been launched worldwide, and analysts have been developing a plethora of di fferent graphene-based electrochemical sensing platforms for detection of various targets including microbial toxins. Typically, these electrochemical biosensors comprise a graphene and/or a graphene derivative/nanocomposite-modified electrode as an electrochemical signal transduction element, and a biological recognition element (e.g., antibodies, aptamer and microbial cells). The signal from the biological recognition event is converted to a quantifiable electrical signal because the biological target is normally in close contact with the electrochemical signal transduction element through physical or chemical interactions (e.g., electrostatic interactions, π-π interactions and covalent bonds). Because of their unique properties (e.g., large surface area and good conductivity), the detection performance of an electrochemical biosensor can be significantly improved by using the graphene and/or a graphene derivative/nanocomposite. Therefore, the scope of application of 2D nanomaterial-based electrochemical biosensors has been constantly expanding in the field of toxin detection. Some of these studies have been reviewed elsewhere with a focus on the fabrication and toxin detection of graphene-based electrochemical biosensors or as subclassifications in more generalized overviews of the nanomaterial-based electrochemical biosensors [8,11–19]. In this review, we will focus on the recent development of GO/rGO and/or MoS2/MoSe2-based electrochemical biosensors for the determination of various microbial toxins, such as bacterial toxins, fungal toxins and algal toxins, highlighting some of their current achievements, technical challenges/limitations and the future directions by means of a set of selected recent publications.

#### **2. Detection of Bacterial Toxins**

## *2.1. Botulinum Neurotoxins*

The *Botulinum* neurotoxins (BoNTs), which are produced by *Clostridium botulinum*, an anaerobic bacterium, are among the most toxic of all naturally occurring substances [70–72]. Based on their molecular structures, BoNTs are categorized into seven serotypes (from A to G). They inhibit acetylcholine release from presynaptic nerve terminals at the neuro-muscular junction in both the central and peripheral nervous systems through cleavage of soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs), resulting in flaccid muscle paralysis. BoNTs can cause the deadly disease, botulism, with a median lethal dose (LD50) of 1 ng per kg bodyweight. Foods are easily contaminated by *Clostridium botulinum* during processing. Various (impedimetric, voltammetric and amperometric) electrochemical biosensors have been fabricated for BoNT detection [73–76]. In particular, electrochemical biosensors can achieve detection of this toxin in a fast and meticulous way, and they also provide a robust and cost-e ffective approach for real-time monitoring of BoNTs. Recently, 2D nanomaterial-based electrochemical biosensors have been applied to sensitively detect BoNTs in various samples including foods. For instance, Narayanan et al. constructed an electrochemical immunosensor of the BoNT serotype E (BoNT/E) by using graphene nanosheets–aryldiazonium salts as transducers [74]. The as-proposed immunosensor shows a low limit of detection (LOD, 5 pg mL−1) and can be employed for rapid detection of BoNT/E with a total analysis time of 65 min. Chan et al. fabricated an electrochemical biosensor for ultrasensitive detection of BoNT serotype A light chain (BoNT-LcA) through immobilization of the SNAP-25-GFP (synaptosomal associated protein 25-green fluorescent protein) peptide substrate on the rGO modified gold electrode via a pyrenebutyric acid (PA) linker (as shown in Figure 1) [75]. In this case, PA was immobilized on the rGO surface through π-π stacking. Subsequently, SNAP-25-GFP peptide reacted with PA via N-(3-dimethylaminopropyl)-N'-ethylcarbodiimide hydrochloride/N-hydroxysulfosuccinimide (EDC/Sulfo-NHS) activation. After specific cleavage of SNAP-25-GFP by BoNT-LcA, the steric hindrance and electrostatic repulsion of SNAP-25-GFP decreased, resulting in an increase in the electrochemical signal. The amount of BoNT-LcA can be detected through the change of peak current of the electrochemical redox probe (ferricyanide, [Fe(CN)6] <sup>3</sup>−/4−(1:1)) by the di fferential pulse voltammetry (DPV) measurement. The as-fabricated electrochemical biosensor provides a relatively wide linear range (1 pg mL−<sup>1</sup> to 1 ng mL−1) and a relatively low LOD (8.6 pg mL−1) for detection of BoNT-LcA because the rGO modified Au (rGO/Au) electrode provides a robust and biocompatible platform with improved electron transfer capability and a large surface area for peptide immobilization. The feasibility of the as-fabricated biosensor is demonstrated by detection of BoNT-LcA in spiked milk samples. Afkhami et al. developed a gold nanoparticle-graphene-chitosan (Au NPs-Gr-Cs) nanocomposite-based impedimetric immunosensor for the detection of BoNT serotype A (BoNT/A) [76]. The Au NPs-Gr-Cs nanocomposite was used for the amplification of the electrochemical signal, and monoclonal anti-BoNT/A antibodies were conjugated on the Au NPs-Gr-Cs nanocomposite modified glassy carbon electrode (GCE). In the presence of BoNT/A, the immunocomplex formed on the as-prepared electrode surface, which acts as the inert electron and mass transfer blocking layer. Therefore, the di ffusion of [Fe(CN)6] 3−/4− is hindered, resulting in a decrease of the peak current. The Au NPs-Gr-Cs nanocomposite-based impedimetric immunosensor has an excellent linear range (from 0.27 to 268 pg mL−1) with a LOD of 0.11 pg mL−1, and is very suitable for routine analysis of BoNT/A in di fferent matrices, such as serum and milk.

**Figure 1.** Schematic representation of the detection principle of the rGO based electrochemical biosensors (adapted from Chan et al. 2015 [75], Copyright 2015 Elsevier B.V. and reproduced with permission).

#### *2.2. Clostridium di*ffi*cile Toxin B*

*Clostridium di*ffi*cile* toxin A (Tcd A, 308 kDa) and toxin B (Tcd B, 270 kDa) are co-produced by *Clostridium di*ffi*cile (C. di*ffi*cile)*. Tcd A is an enterotoxin responsible for tissue damage, while Tcd B is referred to as a potent cytotoxin [77–81]. In particular, the rapid and sensitive detection of Tcd B is very helpful for early diagnosis and e fficient therapy because Tcd B is critical for virulence and is found in all clinically isolated pathogenic strains [79–85]. Using the advantages of GO, including the large surface area and good conductivity, Fang et al. developed a simple sandwich-assay type electrochemical immunosensor for improving the Tcd B detection sensitivity by using GO as a sca ffold for the enhanced loading of horseradish peroxidase (HRP) and HRP-labeled secondary Tcd B antibody (as shown in Figure 2) [84]. The LOD (0.7 pg mL−1) of the sandwich-assay type electrochemical immunosensor is much lower than those of other current techniques including ELISA. In addition, the as-prepared electrochemical immunosensor was successfully employed to detect Tcd B in practical samples (e.g., real human stool), demonstrating that the immunosensor has promising potential in clinical applications.

**Figure 2.** Schematic representation of the immunosensor array preparation and detection strategy by sandwich-type immunoassay of Tcd B. Here, Tcd B means *C. di*ffi*cile* toxin B, BSA means bovine serum albumin, anti-Tcd B means anti-Tcd B antibody, HRP means horseradish peroxidase, HRP-Ab2 means HRP-labeled second anti-Tcd B antibody, GA means glutaraldehyde, CS means chitosan, PB means Prussian blue, MWCNTs means multi-walled carbon nanotube, GO means graphene oxide, and GCE means glassy carbon electrode (adapted from Fang et al. 2014 [84], Copyright 2013 Elsevier B.V. and reproduced with permission).

#### *2.3. Staphylococcal Enterotoxin B*

Among the toxins secreted by *Staphylococcus aureus*, the staphylococcal enterotoxin B (SEB) shows superantigenic properties in nature. SEB exposure can result in immunosuppression and serious food poisoning [86,87]. Therefore, it is important to develop a cost-effective, easy-to-use, rapid and sensitive method for real-time monitoring of a low concentration (less than 20 ng kg−<sup>1</sup> (i.e., LD50 value)) of SEB in foods. Several graphene-based electrochemical biosensors have been developed for real-time detection of SEB in foods with a high sensitivity [88–91]. For instance, Sharma et al. reported on an electrochemical biosensor based on a rGO-chitosan-AuNPs-capturing antibody (rGR-Ch-AuNPs-CAb)-modified GCE for detecting SEB [88]. The rGR-Ch-AuNPs-CAb modified GCE shows remarkable detecting performance because it has a flat two-dimensional configuration and large surface area with plenty of active sites (i.e., functional groups). Using the as-proposed rGR-Ch-AuNPs-CAb-based electrochemical biosensor, 5 ng mL−<sup>1</sup> SEB can be easily detected within 35 min, which is much lower than the LD50 value of SEB. Very recently, Nodoushan et al. fabricated an electrochemical aptasensor for SEB detection by using a rGO and gold nano-urchins (AuNUs)-modified screen printed carbon electrode (SPCE) (as shown in Figure 3) [91]. The aptamer of SEB was attached on the electrode surface through hybridization with the immobilized single-stranded DNA probe on the surface of the AuNUs. Hematoxylin was used as the electrochemical signal generator. In the presence of SEB, the aptamer released from the electrode surface, resulting in an increase in the peak current of hematoxylin. Benefiting from the high conductivity of rGO and high surface area of AuNUs, a wide linear range from 5.0 to 500.0 fmol L−<sup>1</sup> was achieved and the LOD was calculated as 0.21 fmol L−1. There is no significant difference between the results given by the commercial ELISA kit and the electrochemical aptasensor. In particular, the aptasensor shows better recovery rates and lower standard deviation than those of the commercial ELISA kit, which could be employed as a point-of-care (POC) device for assessing food samples.

**Figure 3.** Schematic representation of the fabrication process of the SEB aptasensor by using rGO and AuNU-modified screen printed carbon electrodes (SPCEs) (adapted from Nodoushan et al. 2019 [91], Copyright 2018 Elsevier B.V. and reproduced with permission).

#### **3. Detection of Fungal Toxins**
