Unlocking early detection of Alzheimer’s disease: The emerging role of nanomaterial-based optical sensors

Abstract

Alzheimer’s disease (AD) is a chronic and progressive neurodegenerative disorder that affects millions of individuals worldwide. Researchers have conducted numerous studies to find accurate biomarkers for early AD diagnosis and develop more effective treatments. The main pathological hallmarks of AD are amyloid beta and Tau proteins. Other biomarkers, such as DNA, RNA, and proteins, can also be helpful in early AD diagnosis. To diagnose and treat AD promptly, it is essential to accurately measure the concentration of biomarkers in the cerebrospinal fluid or blood. However, due to the low concentrations of these biomarkers in the body, highly sensitive analytical techniques are required. To date, sensors have become increasingly important due to their high sensitivity, swift detection, and adaptable manipulation features. These qualities make them an excellent substitute for conventional instruments. Nanomaterials are commonly employed in sensors to amplify signals and improve sensitivity. This review paper summarized the integration of nanomaterials in optical sensor systems, including colorimetric, fluorescent, and surface-enhanced Raman scattering sensors for AD biomarkers detection.

1. Alzheimer’s disease

Alzheimer’s disease (AD), known as a chronic progressive neuronal degeneration disease, is the most common cause of dementia and usually affects people over the age of 60. AD progresses through different stages. The first stage, preclinical AD, is when the brain is still able to compensate for the damage to neurons caused by the disease so that no pathogenesis can be observed. When the brain is no longer able to compensate for the damage from AD, patients will progress into the stage of mild cognitive impairment (MCI). A common symptom of MCI is memory loss and difficulty with thinking, which gradually worsens over time. As the progression of AD goes on, patients will sequentially confront mild, moderate, and severe AD. In addition to aggravating memory loss, several psychological disorders, such as depression, agitation, and hallucination, will also appear, and patients will no longer be able to perform daily activities. Eventually, patients are prone to fatality due to the damage to basic physiological regulation [1]. The exact mechanism of AD is not fully understood.

Current research indicates that amyloid plaques and neurofibrillary tangles (NFTs) in the brain are the primary pathological phenomena of AD. These plaques and tangles ultimately break down the brain’s structure, leading to cognitive decline and other symptoms associated with the disease. The cooperation of plaques and tangles increases the oxidative stress and induces the inflammatory reaction, which gradually causes irreversible synaptic dysfunction and neuronal death. Extracellular accumulation of advanced glycation end products (AGEs) by glycoxidation involving free radical production can be found in plaques. In addition, AGEs also facilitate the formation of plaques by cross-linking with amyloid beta (Aβ) aggregates [2]. The formation of plaques and NFTs can also trigger the secretion of proinflammatory cytokines by microglia and astrocytes. Subsequently, these cytokines induce inflammatory responses, which further cause dystrophic neuritis and eventually lead to neuronal degeneration [3]. The pathogeneses of plaques and NFTs start from the hippocampus, which is in charge of forming memory, and gradually spread throughout other parts of the brain [4].

According to the report from Alzheimer’s Disease International, the number of people above 60 years old worldwide has reached nearly 900 million, and the population living with dementia is predicted to be approximately 152 million in 2050 [5,6]. The severity of AD has become not only a public health problem but also a social issue. Leqembi, the first Aβ-targeting antibody, was approved by the Food and Drug Administration to treat adult AD patients in 2023. However, Leqembi should be initiated in patients with MCI or mild dementia stage of AD [7]. Early and accurate diagnosis of AD is crucial for the development of effective AD therapies. Therefore, analyzing AD fluid biomarkers has become a popular research direction in recent years. These approaches can help detect the disease in its early stages and facilitate the development of targeted treatments.

2. Biomarkers of Alzheimer’s disease

The National Association on Aging and Alzheimer’s Disease proposed a research framework to biologically define AD by AT(N) biomarker profiles in 2018. The AT(N) scheme categorizes biomarkers into three groups based on the nature of the pathologic process. “A” refers to Aggregated Aβ evaluated by cerebrospinal fluid (CSF) Aβ₄₂, or Aβ₄₂/Aβ₄₀ ratio or amyloid positron emission tomography (PET). “T” refers to aggregated tau (NFTs) evaluated by CSF p-tau or Tau PET. “(N) refers to neurodegeneration or neuronal injury evaluated by anatomic magneticresonance imaging (MRI) or fluorodeoxyglucose (FDG) PET or CSF total tau [8]. The measured biomarkers for AD include the concentration of Aβ₄₂, Aβ₄₂/Aβ₄₀ ratio, total Tau protein (t-tau), and phosphorylated Tau protein (p-tau) 181 in CSF [9]. PET is valuable for evaluating the process of AD, but the need for radioactive substances limits its clinical use. Similarly, CSF sampling, which requires invasive lumbar punctures, is also impractical for routine use. Recently, a strategy focusing on blood-based biomarkers has gained attention from researchers [6].

Except for Aβ and Tau proteins as AD biomarkers, other biomarkers include DNA, RNA, and proteins originating from neuronal pathogenesis processes such as neuroinflammation, synaptic dysfunction, neuronal injury, and associated genetic variants could be helpful for early diagnosis of AD [10]. Given that many of these biomarkers also indicate other neurodegenerative diseases, numerous studies have explored their correlations with core AD biomarkers to enhance clinical interpretations [11–13]. Hawksworth et al. conducted a recent review focusing on protein biomarkers of AD, emphasizing their associations with amyloid or tau pathology, as well as cognitive deficits in AD [14]. In this context, we introduce Aβ, Tau proteins, and several potential AD biomarkers.

2.1. Amyloid-beta

Aβ peptides are short peptide chains produced when amyloid precursor protein (APP) is broken down through proteolysis. APP is a membrane protein that is commonly found in synapses. Although the exact function of APP is not yet fully understood, it is believed to play a role in supporting the growth of neurons and synapses, as well as acting as a receptor within cells [15,16]. The metabolism of APP is regulated by three types of secretase enzymes - α-secretase, β-secretase, and γ-secretase. Approximately 90% of APPs are metabolized by α-secretase and γ-secretase, which produces soluble peptide residues that are easier to eliminate from the body. It has been observed that around 10% of APPs are metabolized by β-secretase and γ-secretase [17]. When β-secretase hydrolyzes APPs, it generates longer and hydrophobic peptide residues (Aβ) that tend to clump together and are difficult to eliminate from the body. This process is also known as amyloidogenesis. The Aβ family can have different lengths due to the imprecise cleaving site of APP by γ-secretase. These lengths can range from 36 to 43 amino acids. Aβ₄₀ accounts for approximately 80–90% of Aβs produced, while Aβ₄₂ accounts for approximately 5–10% [17,18]. When Aβs fail to be effectively eliminated, they self-aggregate from monomers to plaques via oligomers and fibrils [19]. In AD patients, unknown or genetic mutations cause increased β-secretase activity, resulting in excessive accumulation of Aβ. On the other hand, the oligomeric forms of Aβ (AβO) can lead to various neuronal damages, such as receptor-mediated toxicity, dysregulation of ion channels, proteasome impairment, and mitochondrial dysfunction. These damages eventually induce the pathogenesis of AD [20].

In current clinical scenarios, PET can be used to visualize the distribution of Aβ and the atrophying structure of the brain [1], and the levels of Aβ₄₂ or the ratio of Aβ₄₂ to Aβ₄₀ in CSF or plasma is measured by suitable analytical methods for AD diagnosis [21–24]. These values can help evaluate the progression of AD in the clinical setting and were integrated in Table 1. AβO plays a crucial role in AD development and is considered a potential diagnostic candidate. The AβO in plasma measured using a multimer detection system-oligomeric assay was compared between 52 AD patients treated and 52 individuals with normal cognition. The cut-off value for AβO is ≥ 0.78 ng/mL [25]. A significant 3- to 5-fold increase in AβO has been detected in CSF of AD patients (63) compared to controls (54). The AβO in CSF ranged between 0.1 and 10 pg/mL [26].

Table 1.

Clinical research of biomarkers of AD in CSF and plasma samples.

Biomarker	Sample	Detection outcome		Cut-off value	Reference
		NC a	AD b
Aβ42	CSF	916 ± 182 pg/mL (ApoE ɛ4−)	507 ± 180 pg/mL (ApoE ɛ4−)	≤ 550 pg/mL	[22]
		745 ± 205 pg/mL (ApoE ɛ4+)	460 ± 137 pg/mL (ApoE ɛ4+)
t-tau		276 ± 116 pg/mL (ApoE ɛ4−)	804 ± 404 pg/mL (ApoE ɛ4−)	≥ 375 pg/mL
		300 ± 161 pg/mL (ApoE ɛ4+)	792 ± 484 pg/mL (ApoE ɛ4+)
p-tau		44 ± 14 pg/mL (ApoE ɛ4−)	91 ± 39 pg/mL (ApoE ɛ4−)	≥ 52 pg/mL
		50 ± 23 pg/mL (ApoE ɛ4+)	89 ± 39 pg/mL (ApoE ɛ4+)
Aβ40	CSF	6.7 ± 2.5 ng/mL	5.9 ± 2.3 ng/mL	-	[23]
Aβ42		0.87 ± 0.36 ng/mL	0.67 ± 0.26 ng/mL	≤ 0.64 ng/mL
Aβ42/Aβ40		1.5 ± 1.1	1.3 ± 0.66	≤ 0.95
Aβ40	CSF	1935 pg/mL (cohort 1)	2088 pg/mL (cohort 1)	-	[26]
		1996 pg/mL (cohort 2)	2249 pg/mL (cohort 2)
Aβ42		44 pg/mL (cohort 1)	108 pg/mL (cohort 1)
		114 pg/mL (cohort 2)	216 pg/mL (cohort 2)
AβO (ADDLs)		0.28 (0.18–0.34) pg/mL (cohort 1)	2.6 (0.75–3.3) pg/mL (cohort 1)
		0.52 (0.41–0.66) pg/mL (cohort 2)	1.52 (1.19–1.94) pg/mL (cohort 2)
Aβ42	CSF	503.99 ± 156.67 pg/mL	328.76 ± 110.54 pg/mL	≤ 416.0 pg/mL	[35]
t-tau		86.03 ± 47.47 pg/mL	145.69 ± 80.22 pg/mL	≥ 76.7 pg/mL
p-tau		41.59 ± 21.81 pg/mL	66.72 ± 35.82 pg/mL	≥ 36.1 pg/mL
Aβ42/p-tau		14.38 ± 6.46	7.39 ± 7.39	≤ 9.53
Aβ42/t-tau		7.12 ± 3.48	3.45 ± 4.01	≤ 4.13
Aβ40	Plasma	60.3 ± 7.2 pg/mL	43.9 ± 22.6 pg/mL	-	[24]
Aβ42		15.8 ± 0.7 pg/mL	23.5 ± 18.8 pg/mL	≥ 16.41 pg/mL
Tau		13.3 ± 6.9 pg/mL	47.6 ± 19.2 pg/mL	≥ 24.90 pg/mL
Aβ42/Aβ40		0.267 ± 0.038	0.557 ± 0.231	≥ 0.361
AβO	Plasma	0.45 ± 0.19 ng/mL	1.43 ± 0.30 ng/mL	≥ 0.78 ng/mL	[25]

Abbreviations: ADDLs: amyloid-derived diffusible ligands.

^aNormal control

^bPatient with a dementia due to AD.

2.2. Tau protein

Tau protein is generated in the brain through the transcription of the microtubule-associated protein tau (MAPT) gene on human chromosome 17. There are six isoforms of water-soluble, non-folded proteins, including Tau352 (0N3R), Tau381 (1N3R), Tau410 (2N3R), Tau383 (0N4R), Tau412 (1N4R) and Tau441 (2N4R) [27]. The function of Tau protein was used to support the structure of microtubes, which dominate the transportation of nutrients throughout neurons in the brain. As Tau proteins are phosphorylated, they collapse from microtubes and twist into tangles to form the NFTs. Subsequently, the structure of microtubes will disintegrate and disrupt the transportation of nutrients. Eventually, the gradual neuronal death in the brain will occur due to the dysfunction of microtubes and the neurotoxicity of tangles [18]. Phosphorylation is the most common post-translational modification of Tau protein. Other modifications include glycosylation, nitration, and proteolysis. Tau441 contains 85 potential phosphorylation sites, with 45 serine (S), 35 threonine (T), and 5 tyrosine (Y) sites. These sites are concentrated in the proline-rich region and microtubule-binding repeat region [28,29]. When the Tau protein appears hyperphosphorylated, it reduces its affinity with microtubules. Consequently, the Tau protein gets released into the cell [30]. This leads to an increase in the concentration of both t-tau and p-tau compared to normal individuals. P-tau inCSF, mainly p-tau181, p-tau217, and p-tau231, is considered an important biomarker for AD and has been included in the diagnostic criteria [31]. Several studies have reported that p-tau217 and p-tau231 in CSF may be better diagnostic biomarkers for AD than p-tau181 [32,33]. In addition, p-tau217 in plasma performs better as an AD diagnostic biomarker than p-tau181 and p-tau231 [34]. Clinically relevant concentrations of Tau protein in CSF and plasma were integrated in Table 1.

2.3. Neurofilament light chain

Neurofilament light chain (NfL) is one of the four subunits of axonal cytoskeletal protein, released upon axonal damage and detectable in both CSF and plasma [36]. Given that various neurological diseases and aging entail this pathological process, NfL levels cannot reliably differentiate AD from other conditions but can aid in early diagnosis [37]. Jung et al. noted that the longitudinal changes in NfL levels can reflect and predict the severity of AD, categorizing NfL as a monitoring biomarker for AD per the definitions of the FDA-NIH Biomarker Working Group in 2016 [38]. Notably, the measurability of NfL in blood underscores its significance for in-depth discussion [39]. Multiple studies have confirmed plasma NfL as a potential AD biomarker, with its dynamics predictive not only of cortical atrophy and neurogenesis in the presymptomatic stage but also of white/grey matter atrophy in symptomatic AD patients [37,40–42]. Moreover, while NfL exists in multiple forms, its complete characterization remains ongoing. Budelier et al. observed significant differences in concentrations of neurofilament light fragments between AD patients and healthy individuals CSF, including NfL165, NfL324, and NfL530 by immunoprecipitation-mass spectrometry assay and the commercially available immunoassay [43].

2.4. Neurogranin

Neurogranin (Ng) is a postsynaptic protein that binds to calmodulin, initiating signaling pathways crucial for synaptic plasticity, synaptic regeneration, and long-term potentiation [44]. Ng is predominantly found in the hippocampus and cerebral cortex, and elevated CSF levels indicate synaptic loss, leading to cognitive decline. A recent postmortem study of human brain tissue revealed reduced Ng levels in cognitive brain regions in AD and healthy aging cohorts, supporting the notion that CSF Ng reflects synaptic loss in the brain and serves as a hallmark of cognitive decline [45]. Tarawneh et al. compared symptomatic AD patients with cognitively normal groups and concluded that CSF Ng is associated with other CSF markers, representing a promising marker for synaptic dysfunction in AD [46]. Wellington et al. observed Ng level differences in typical and atypical AD patients; Ng levels were higher in amnestic AD and lower in AD with posterior cortical atrophy. They also identified distinct relationships between Ng, t-tau, and NfL, suggesting that Ng levels may aid in characterizing various AD phenotypes [47]. However, it is worth noting that the specificity of CSF Ng for AD is controversial, as CSF Ng levels are also elevated in aging and MCI cohorts [48]. Nonetheless, Ng still holds promise as a prognostic biomarker for early AD diagnosis, although further research and evidence are required.

2.5. α-Synuclein and β-synuclein

The synuclein family comprises α-synuclein (αSyn), β-synuclein (βSyn), and γ-synuclein (γSyn), among which αSyn binds to synaptic vesicles, facilitating soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex assembly to regulate neurotransmitter release [49]. Typically, the α-helical form of αSyn exhibits membrane binding, while the unstructured form remains soluble in the cytosol [50]. The transformation between these two forms is modulated by βSyn and γSyn. Pathologically, αSyn aggregates into oligomers or fibrils, leading to neurotoxicity and contributing to synucleinopathy, a hallmark of Parkinson’s disease [49,51]. In AD, CSF levels of αSyn are strongly correlated with the triple core CSF markers (t-tau, p-tau181, and Aβ), with CSF αSyn levels increasing as MCI progresses to symptomatic AD, thus supporting the utility of CSF αSyn for early or differential diagnosis [52,53]. Majbour et al. observed increased total αSyn CSF levels in AD, with no changes in oligomeric or fibrillar forms, suggesting variations arise from the soluble form of αSyn; notably, AD patients did not exhibit synucleinopathy as seen in Parkinson’s disease [54]. Elevated levels of βSyn and γSyn have also been detected in the CSF of AD patients, with βSyn showing correlations with Tau proteins; furthermore, βSyn is detectable in blood, indicating its potential as an alternative AD marker [55].

2.6. Soluble triggering receptor expressed on myeloid cell 2

Neuroinflammation marker proteins are promising biomarkers for early AD diagnosis, including soluble triggering receptor expressed on myeloid cells 2 (sTREM2), glial fibrillary acidic protein (GFAP), YKL-40, and others. Neuroinflammation, characterized by dysregulation of the central nervous system immunity, primarily involves microglial cells and astrocytes [56,57].

In AD, the aggregation of Aβ and deposition of other proteins in brain tissues are neurotoxic, triggering microglial cells to engage in phagocytosis and activating astrocytes to sustain overall neuronal function [56]. Pathologically, the imbalance between pro-inflammatory and anti-inflammatory responses results in the transcriptional activation of downstream genes and the overproduction of toxic factors, ultimately leading to neuronal death [57]. TREM2, expressed on the surface of microglia, plays a crucial role in recognition. Its soluble cleavage product, sTREM2, detectable in CSF, serves as a marker of microglial activation in AD. Knapskog et al. found that elevated CSF levels of sTREM2 are associated with aging and tauopathy but not cognitive status [58]. Various studies have reported decreased sTREM2 levels in the CSF of AD patients, with individual baseline levels varying [59–61]. Nevertheless, sTREM2 remains a potential biomarker for investigating the dual role of microglia in AD pathology [62].

Moreover, TREM2 is among the gene variant expression targets of microglia and has been reported to be regulated by multiple AD risk genes in different studies. TREM2 levels are higher in carriers of the rs7748513 or R47H variants, lower in L2112P carriers, and unaffected in apolipoprotein E gene ɛ4 (ApoE4) carriers [61,63,64]. Kim et al. studied an autopsy-confirmed cohort and found that TREM2 variants were associated with atypical AD clinical phenotypes and accelerated global cognitive decline [65]. Deming et al. discovered an association between variants in the membrane-spanning 4-domains subfamily A (MS4A) gene region and CSF sTREM2 levels, including the single nucleotide polymorphism rs1582763 [66].

2.7. GFAP and YKL-40

GFAP is an intermediate filament protein present in the cytoskeleton of astrocytes, overexpressed in reactive astrocytes and astrogliosis. Its levels in CSF or plasma can reflect astrocyte status, with reduced GFAP indicating decreased astrocyte activity [57]. In AD and other neurodegenerative conditions, GFAP levels are elevated in both plasma and CSF. Given the convenience of blood-based biomarkers for clinical monitoring and research, most studies have focused on the reliability of plasma GFAP for AD diagnosis [67,68]. Yang et al. found that plasma GFAP exhibited stronger potential for screening AD amyloid pathology compared to other plasma biomarkers, such as p-tau181 and NfL, and showed prognostic power in predicting future cognitive decline [69]. Chatterjee et al. reported alterations in GFAP and other plasma markers both cross-sectionally and longitudinally in AD, collectively providing improved predictive capability for Aβ pathological status [70]. Stevenson-Hoare et al. obtained similar results and also noted a close association between plasma GFAP levels and AD risk genes such as ApoE4 [71]. Taken together, these studies suggest that plasma GFAP could serve as a diagnostic and predictive biomarker for AD.

Another astrocyte biomarker under discussion is YKL-40, also known as chitinase-3-like protein 1 (CHI3L1), which can be detected in CSF and plasma [67]. YKL-40 is a secreted glycoprotein of 40 kDa involved in various inflammatory pathways associated with conditions such as rheumatoid arthritis, cancers, and neurological diseases [72]. Pelkmans et al. identified a greater role for YKL-40 in mediating Aβ-induced tau phosphorylation and neuronal injury, while GFAP mediates early Aβ misfolding [68]. In a clinical trial investigating inflammatory plasma markers, preclinical AD patients showed significantly higher plasma levels of both GFAP and YKL-40, indicating that these markers could help identify individuals at risk of developing AD [57,73]. Although YKL-40 is gaining recognition as a potential early diagnostic biomarker for AD, further research is needed to fully understand its role, significance, and correlation with other biomarkers.

2.8. Transactive response DNA binding protein of 43 kDa

Transactive response DNA binding protein of 43 kDa (TDP-43), an intranuclear protein encoded by the TAR DNA Binding Protein (TARDBP) gene, is associated with TDP-43-mediated RNA splicing, trafficking, stabilization, and regulation of gene expression [74]. Mutations in TDP-43-related genes are linked to amyotrophic lateral sclerosis and frontotemporal lobar degeneration, leading to symptoms such as muscle weakness and personality changes [11]. Additionally, TDP-43 pathology has been observed to form deposits in the brains of individuals aged 80 years and older, termed limbic predominant age-related TDP-43 encephalopathy (LATE) [74]. LATE coexisting with AD, known as LATE + AD, is common in AD patients with accelerated cognitive decline, and TDP-43 may also trigger neuroinflammation in the brain [75]. Noteworthy, TDP-43 pathology is also common in other subtypes of AD and is involved in AD progression. TDP-43 directly interacts with Aβ and Tau proteins, cross-seeding and co-localizing with them to form oligomers and aggregates [74]. Genetically, TDP-43 pathology is associated with the ApoE4, and the burden of TDP-43 increases with the number of ApoE4 alleles, collectively accelerating the development of AD-type pathology. Altogether, TDP-43 serves as a significant risk factor of AD and plays a role in the progression for AD-type dementia [74].

2.9. Genetic risk factors

In addition to the extensively studied genetic risk factors such as ApoE and the previously mentioned gene variants, as well as the early-discovered APP, PSEN1, and PSEN2 mutations [76,77], other gene variants, including SORL1, ABCA7, ABCA1, PLCγ2, and ADAM10, are also associated with AD risk. Many of these variants are rare and are linked to factors such as age, sex, or race [78]. The discovery of numerous genes affecting AD provides multiple avenues for a comprehensive and diverse understanding of the disease [76].

A recent meta-analysis, combining data from previous genome-wide association studies with findings from a new large case–control study, identified 42 new risk loci and confirmed 33 known loci for AD [76]. Furthermore, this analysis highlighted the involvement of these loci in pathways related to amyloid, tau, and microglia, further elucidating the molecular mechanisms underlying AD pathogenesis [76].

2.9.1. ApoE

Corder et al. indicated that individuals who possess the genetic variant ɛ4 in the ApoE gene have a higher frequency in AD patients with late-onset Alzheimer’s disease (LOAD) [79]. The ApoE gene, located on chromosome 19, has three alleles: ɛ2, ɛ3, and ɛ4. These alleles are generated by two single-nucleotide polymorphisms (SNPs), rs429358 (C/T) and rs7412 (C/T). The ApoE genotype ɛ3 is the most common in the general population, accounting for 50–90%, followed by ɛ4 at 5–35% and ɛ2 at 1–5%. An antagonistic effect that reduces the risk of AD can be observed in carriers of ApoE-ɛ2, while carriers of ApoE-ɛ3 do not show any influence on AD. However, carriers of ApoE-ɛ4 have a higher risk of suffering from AD. Individuals who are heterozygous for ɛ4 have a 2–3 times higher risk of developing AD, and this risk increases to 8–12 times for those who are homozygous for ɛ4 [80].

2.9.2. MicroRNAs

MicroRNAs (miRNAs) are non-coding RNA molecules approximately 22–23 nucleotides in length that play crucial roles in regulating gene expression. miRNAs detected in CSF, blood, and brain tissues have shown stability and are being investigated as candidate biomarkers for AD, although their precise relationships with AD pathology are not yet fully understood [81–83]. Swarbrick et al. identified miRNAs as predominantly involved in AD pathological pathways, including those related to amyloid, tau, and neuroinflammatory signaling pathways. They also reported dysregulated miRNA expression in the peripheral blood of AD patients [84]. In a systematic review, Van den Berg et al. identified 253 circulating miRNAs with various functions, such as upregulation, downregulation, or dual-functionality. They noted differences in expression levels between AD patients and healthy controls [83]. For example, the miR-29 family is believed to downregulate β-site amyloid precursor protein-cleaving enzyme 1 (BACE1) [85]. Additionally, miR-125b-5p has been implicated in inducing tau phosphorylation, potentially contributing to glial proliferation and synapse-related dysfunction [86–88]. Furthermore, miR-146a-5p levels were found to be elevated in both the brains and blood of AD patients, promoting neuroinflammation and neurodegeneration [81,86].

3. Nanomaterial-based optical sensors

The low trace levels of biomarkers in body fluids make it challenging to detect them accurately. Fortunately, optical sensors have emerged as highly sensitive tools that can swiftly detect the slightest changes in biomarker levels. They are an excellent alternative to conventional instruments due to their adaptability and ability to provide reliable results. Nanomaterials are often employed in sensors to amplify the signal and enhance sensitivity. With these benefits, nanomaterials-based optical sensors, including colorimetric, fluorescent, and surface-enhanced Raman scattering (SERS) sensors, have been developed to detect AD biomarkers.

3.1. Colorimetric sensors

Colorimetric sensors transform biochemical reactions occurring during the sensing process into observable color changes, which can be quantified through alterations in absorption spectra or visually observed without the need for specialized equipment [89,90]. Typically, the biochemical events induced by target analytes cause changes in the electronic energy of chromogens or alter the surrounding dielectric environment of metal nanoparticles, resulting in changes in absorbance or shifts in the absorption peak. The absorbance change is directly proportional to the concentration of the target analyte, making colorimetric sensors ideal for quantitative, label-free, real-time monitoring, and easy implementation [89,91,92]. Therefore, the selection of chromogenic substrates with distinct color transitions is crucial for their efficacy. Upon reviewing previously published colorimetric sensors for AD biomarkers, the primary substrates utilized include chromogenic indicators coupled with peroxidase-activity enzymes and plasmonic particles, such as noble metal nanoparticles (shown in Fig. 1).

Fig. 1.

Colorimetric strategies for AD biomarkers based on nanomaterials with different mechanisms. (a) Oxidative enzyme based colorimetric sensor. (b) Surface plasmon resonance. (c) Self-aggregation of nanomaterials based colorimetric sensor. (d) Localized surface plasmon resonance.

3.1.1. Oxidative enzyme based colorimetric sensor

In the context of colorimetric sensors employing chromogens, color transitions may occur through enzymatic catalysis, pH mediation, or other mechanisms [89]. For instance, Khachornsakkul et al. utilized bromocresol purple (BCP) as a sensing element and employed a dry-protein binding process for the detection of urine Aβ [93]. BCP, with a pKa of 6.3, exhibits a yellow color below pH 5.2 and violet above pH 6.8 [94]. In the pH 5 sensing system, BCP and Aβ interact via electrostatic forces, causing BCP to increase its conjugation degree and reduce the π* state energy, resulting in a color change from yellow to blue-purple. Consequently, there is a bathochromic and hyperchromic shift in its absorption spectrum. The proposed sensing system also incorporates BCP onto a distance-based paper analytical device (dPAD). The dPAD uses a self-designed pattern, wax-printed on filter paper, and consists of two functional areas: a sample inlet and a longitudinal detection area coated with BCP. When Aβ flows through the detection channel through capillary action, it binds to BCP, causing a color change. The higher the concentration of target Aβ, the longer the length of the color-changing zone, which facilitates quantitative analysis [93].

Certain chromogenic substrates undergo color transitions catalyzed by oxidative enzymes. One such substrate is 3,3′,5,5′-tetramethylbenzidine (TMB), a well-known peroxidase substrate widely used in colorimetric and enzymatic assays [95]. TMB exhibits various colors depending on its oxidation state, appearing transparent (with absorbance at 285 nm) as diamine (TMB), blue-green (at 370 nm and 652 nm) as a charge–transfer complex (TMB¹⁺), and yellow (at 450 nm) as diimine (TMB²⁺) [96]. This chromatic variability of TMB has also been harnessed in colorimetric sensors for AD diagnosis.

The colorimetric signals of the tau protein aptasensor introduced by Duan et al. were generated through the oxidation of TMB catalyzed by horseradish peroxidase (HRP). To enhance the signal, they designed a double-stranded nanostructure known as polyvalent biotinylated aptamer scaffold (PBAS), as shown in Fig. 2. PBAS consists of numerous biotinylated Tau protein aptamers hybridized with linker DNAs, where the biotin labels can be attached to streptavidin-modified magnetic beads for separation or streptavidin-modified HRP (strep-HRP) for catalysis. This configuration provides multiple effects on surface immobilization, target recognition, and signal amplification. In the absence of Tau proteins, PBAS forms enzymatic clusters that catalyze TMB oxidation, resulting in a color transition from colorless to yellow (450 nm). However, when target Tau proteins bind to the aptamer, they prevent the formation of PBAS, leading to a reduction in strep-HRP levels and decreased catalytic activity on TMB. Furthermore, the authors compared PBAS with monovalent biotinylated aptamers (MBA) and methods using hybridization chain reaction (HCR). The PBAS-based aptasensor showed rapid and efficient signal amplification, producing signals four times greater than the MBA-based method, demonstrating its superior efficiency [97]. Chen et al. also combined signal amplification with TMB in the proposed aptasensor for the detection of AβO in CSF. They adopted a cyclic amplification approach involving an AβO-specific aptamer, four single-stranded DNA (ssDNA) probes, Exonuclease III (Exo III), and nicking endonucleases (Nt.Alw1). Initially, AβO-specific aptamer forms triple-helix with single-strand DNA. In the presence of AβO, the triplexhelix disintegrates, aptamer captures AβO, singlestrand DNA is released, which then initiates a dual-cyclic amplification cascade. This process involves a series of hybridization and cleavage steps, using enzymes like Exo III and Nt.Alw1 to cleave fragments, which are then reused for repeated hybridization cycles. After two cycles of amplification, the final products contain numerous G-quadruplexes. These G-quadruplexes form complexes with heme, acting as DNAzymes with peroxidase properties that catalyze TMB oxidation in the presence of H₂O₂. The absorbance of blue oxidized TMB at 650 nm can be measured to determine AβO levels in CSF with high sensitivity and specificity [98].

Fig. 2.

Schematic illustration of the PBAS-based colorimetric assay of Tau proteins. Reprinted with permission from Polyvalent Biotinylated Aptamer Scaffold for Rapid and Sensitive Detection of Tau Proteins (Ref [97]). Copyright 2020 American Chemical Society.

Another colorimetric sensor based on HRP-TMB, developed by Khan et al., offers a method for detecting Aβ₄₀ without the need for antibodies or other recognition elements. They utilize the specific binding affinity of Aβ₄₀ to gold nanoparticles (AuNPs) and various metal ions. In this assay, citrate-capped AuNPs capture Aβ₄₀, forming complexes. Metal ions are then added, not to induce AuNP aggregation but to attach to Aβ₄₀ through electrostatic attraction, enabling an indirect colorimetric signal output. Nickel (Ni) is used as the source of metal ions, conjugated with HRP (Ni-HRP), which catalyzes the oxidation of TMB, producing a colorimetric signal. Thus, Aβ₄₀ binds to both AuNPs and Ni-HRP, forming a sandwich-like complex that allows the quantification of Aβ₄₀ in human serum. Additionally, to select the optimal metal ion, the minimum aggregatory concentration of AuNPs with different metal ions was investigated, including Nickel (Ni), Copper (Cu), Zinc (Zn), and Cadmium (Cd). Ni was chosen because it provides optimal binding to Aβ₄₀ without causing AuNP aggregation, ensuring that the color changes in the sensing system originate from TMB oxidation [99].

In addition to natural enzymes, nanozymes have emerged as promising tools for biosensor development based on enzyme–substrate reactions. Nanozymes are nanomaterials that mimic enzyme-like functions, including hydrolases, oxidoreductases, and transferases. They offer advantages in synthesis, design, modification, stability, and adjustable catalysis [95,100]. Among nanozymes, metal-based nanomaterials with peroxidase activities have been extensively explored [101].

Zhou et al. synthesized porous crystalline materials known as ZnO–Co₃O₄ nanocages (NCs) and employed them as nanozymes for a colorimetric sensor targeting Aβ [102]. These ZnO–Co₃O₄ NCs, derived from metal–organic frameworks (MOFs) through oxidation, possess unique stability characteristics compared to traditional MOFs [103], as they can prevent disassembly due to potential ligand exchange during the sensing process. The metal nodes of ZnCo₃O₄ NCs exhibit peroxidase activity and can catalyze TMB, leading to a color change from colorless to blue-green (652 nm). Additionally, both Zn²⁺ and Co³⁺ exhibit binding affinity toward Aβ monomers. Consequently, the activity of the nanozyme is inhibited by the binding of Aβ monomers, resulting in a decrease in the absorbance intensity of TMB at 652 nm. Furthermore, Aβ species such as oligomers and fibrils exhibit lower binding affinity to nanozymes. Therefore, monitoring changes in TMB absorbance intensity enables quantitative analysis of Aβ and monitoring of amyloidogenesis [102].

Hu et al. discovered that different Aβ species modulate the peroxidative activity of MnO₂ nanosheets to varying degrees. Building on this finding, they developed an enzymatic colorimetric sensor array for versatile detection of Aβ species (Aβ₄₀, Aβ₄₂, Pro-Aβ₄₂, Aβ₄₂ fibrils, AβE22G, and Pro-AβArc). In this sensing array, colorimetric signals arise from the oxidation of o-phenylenediamine, catalyzed by MnO₂ nanosheets, resulting in a color change from colorless to yellow (absorbance at 450 nm). Additionally, since different Aβ species exhibit distinct binding forces with MnO₂ nanosheets, MnO₂ nanosheets are conjugated with various types of metal ions (Zn²⁺, Fe³⁺, or Cu²⁺) to expand the differences in binding affinities of the six Aβ species for linear discriminant analysis. As a result, the sensor array can distinguish and quantify the six types of Aβ species, allowing for monitoring of the Aβ aggregation process [104].

According to the review presented above (Table 2), oxidative enzyme based colorimetric biosensors for detecting AD biomarkers are diverse, mainly utilizing chromogenic substrates, peroxidase enzymes, metallic and composite nanomaterials, and various sensing strategies such as homogeneous assays, paper-based devices, signal amplification, and optical arrays. While not all methods incorporate recognition elements, these approaches have demonstrated specificity for target markers and have been applied in both simulated and real biological samples. They offer sensing advantages including label-free detection, visual observation, and sensitivity.

Table 2.

Nanomaterial-based colorimetric (absorbance) sensors for AD biomarkers.

Nanomaterials	Biomarkers	Recognition element	Limit of detection	Sample	References
AuNPs	Aβ40	–	0.23 nM	Human serum	[99]
ZnO–Co3O4 nanocages	Aβ40	–	3.5 nM	Rat CSF	[102]
Metal ion conjugated MnO2 nanosheet	Aβ species	–	0.44 pM	Human blood	[104]
AuNPs	AβO	Aptamer	0.56 nM	aCSF	[107]
pA-pT-apt@AuNPs	AβO	Aptamer	3.03 nM	aCSF	[108]
AgNPs, AuNPs	Aβ40, Aβ42	–	–	Human plasma	[109]
Gelsolin modified AgNTs and AgNRs	Aβ40, Aβ42	–	–	Rat CSF and brain tissue	[111]

Abbreviations: aCSF: artificial cerebrospinal fluid, AgNPs: silver nanoparticles, AgNRs: silver nanorods, AgNTs: silver nanotriangles, AuNPs: gold nanoparticles, pA-pT-apt@AuNPs: aptamer-polythymine-polyadenine-gold nanoparticles, PBAS: polyvalent biotinylated aptamer scaffold.

3.1.2. Localized surface plasmon resonance

Localized surface plasmon resonance (LSPR) refers to the collective oscillation of free electrons on the surface of nanoparticles, typically metal nanoparticles such as gold (Au) or silver (Ag), when illuminated with light of a specific wavelength. This phenomenon leads to enhanced absorption and scattering of light at certain wavelengths, known as the plasmon resonance wavelength, which is highly sensitive to the size, shape, and composition of the nanoparticles [91,105]. Consequently, the optical properties of AuNPs induced by LSPR find wide applications in the development of biosensors for AD biomarkers [90,92]. AuNPs-based colorimetric sensors leverage the absorbance of plasmonic AuNPs and monitor absorption changes for analysis. One commonly utilized strategy is aggregation-mediated color-tuning. The absorption band of dispersed AuNPs typically occurs at 520 nm, appearing as a red color, while aggregation leads to a color shift towards purple or blue hues [106].

Zhu et al. employed citrate-capped AuNPs to develop a label-free colorimetric aptasensor for detecting AβO. Upon the addition of Na⁺ and aptamer, the negatively charged surface of AuNPs is neutralized, leading to aggregation and causing a shift in absorption spectra from 520 nm (red) to 650 nm (purple). In the presence of AβO, the aptamer forms complexes with AβO, stabilizing AuNPs against salt-induced aggregation. As the amount of AβO increases, the ratio of absorbance at 650 nm–520 nm (A₆₅₀/A₅₂₀) decreases, resulting in a series of color changes (from purple to ruby red). Therefore, the A₆₅₀/A₅₂₀ ratio reflects the degree of aggregation and can be used to quantify the amount of AβO [21,107]. Another AβO aptasensor, devised by Tu et al., utilized elongated aptamers, AuNPs, and a salt-induced aggregation strategy. The aptamer is extended with polythymine (pT) and polyadenine (pA), where the terminal pA preferentially anchors the aptamer on AuNPs to prevent the aptamer from interacting directly or randomly with the AuNPs surface; pT serves as a spacer to provide space between adjacent aptamers (pA-pTapt@ AuNPs). With the assistance of the pA-pT sequence, the aptamer on the other end is allowed to form an upright conformation on AuNPs, meaning the aptamer is stretched and stands up from the surface of the AuNPs, resulting in a significant increase in the size of AuNPs. This design preserves space for target recognition and improves sensing specificity and efficiency. In the presence of AβO, the aptamer undergoes adaptive folding, stabilizing and preventing pA-pT-apt@AuNPs from MgCl₂-induced aggregation. Consequently, the initially aggregated AuNPs transition into a dispersed state, changing the color from blue to red [108].

Ghasemi et al. took a different approach by utilizing two types of plasmonic nanoparticles, silver nanoparticles (AgNPs) and AuNPs, to detect and differentiate between Aβ₄₀ and Aβ₄₂, as well as human serum albumin in plasma [109]. Aβ₄₀ and Aβ₄₂ have an isoelectric point of 5.3 [110], rendering them positively charged and showing varying binding affinities towards citrate-capped nanoparticles, thereby aiding in the dispersion of nanoparticles. Copper ions are introduced to induce nanoparticle aggregation by bridging with Aβ₄₀ and Aβ₄₂, resulting in distinct spectral and color changes for the two plasmonic nanoparticles. The original and emerging absorption peaks are at 520 nm and 620 nm for AuNPs, and at 390 nm and 500 nm for AgNPs. By combining these colorimetric responses of the two metal nanoparticles and recording the absorbance at three wavelengths (420 nm, 530 nm, 620 nm), the color patterns and data matrix of Aβ₄₀ and Aβ₄₂ are established for linear discriminant analysis of Aβ₄₀/Aβ₄₂ quantification [109]. Similarly, Liu et al. presented a colorimetric sensor for CSF Aβ₄₀ and Aβ₄₂ without the use of recognition elements [111]. They designed rod- and triangle-shaped AgNPs, named AgNRs and AgNTs, respectively, and modified their surfaces with gelsolin, which has affinity for Aβ₄₀ and Aβ₄₂. The gelsolin-modified AgNPs serve as the sensing elements, with characteristic SPR peaks at 514 nm (AgNRs) and 640 nm (AgNTs). In the presence of Aβ₄₀ or Aβ₄₂, the gelsolin-modified AgNRs or AgNTs aggregate upon binding with the target peptides, causing shifts in the absorption bands and various color transitions. Specifically, AgNRs exhibit new absorption bands around 700 nm and change color from red to purple (with Aβ₄₀) or blue (with Aβ₄₂). AgNTs show new absorption bands around 900 nm and change color from dark blue to light blue (with Aβ₄₀) or light gray (with Aβ₄₂). By evaluating these color changes in parallel, the sensor produces specific color response patterns that can simultaneously identify and distinguish Aβ₄₀ and Aβ₄₂ [111].

As mentioned, LSPR is closely tied to the structure, composition, particle–particle distance, and dielectric environment of metallic nanoparticles, all of which influence the effect and optical signatures of LSPR. Various nanostructures of AuNPs have been utilized to modulate color for biosensing applications [91,105]. Nair et al. introduced Au nanourchins, Au nanomaterials with a spiky structure, for detecting the fibrillation process of Aβ₄₂ starting from trace amounts of Aβ₄₂ monomers. The urchin-like nanostructure offers a larger surface area and enhanced field enhancement at the tips, enabling them to detect very low concentrations of Aβ₄₂. By simply mixing the Au nano-urchin solution with the Aβ₄₂ monomer solution in a cuvette, the fibrillation process is monitored over time using an optical fiber and lens assembly. When fibrils begin to form and attach to the Au nano-urchins, the plasmonic response of the Au nano-urchins is impeded, leading to a more pronounced reduction in LSPR peak intensity (650 nm) compared to spherical AuNPs. These absorbance changes can be observed in as little as 5 min, indicating the potential for sensing Aβ fibrils [112]. Kim et al. developed a multiplex LSPR sensor employing antibodies and three different shapes of AuNPs for the simultaneous detection of Aβ₄₀, Aβ₄₂, and tau protein. AuNPs are synthesized in spherical shapes (54 nm), short rods (aspect ratio 1.6), or long rods (aspect ratio 3.6), and then PEG-functionalized and conjugated with antibodies for AD biomarkers recognition. Each type of immune-AuNPs undergoes individual detection events, resulting in an LSPR shift at a distinct extinction peak (λ_max), enabling simultaneous analysis without the need for additional separation and identification efforts. As the measurement procedure was like bar-code scanning, the authors described it as a shape-coded biosensor [113].

Song et al. devised DNA-assembled advanced plasmonic architecture (DAPA) and established a DAPA-based plasmonic biosensor for detecting exosomal miRNAs associated with AD, such as miRNA-125b, miRNA-15a, and miRNA-361. The DAPA nanoarchitecture comprises seed particles of three spherical AuNPs functionalized with ssDNA. These AuNPs are interconnected via hybridization of ssDNA strands, creating two nanogaps (2 nm) between each pair of particles. Consequently, a rod-shaped structure is formed, leveraging nanogaps to focus incident light and generate hot-spots, thereby amplifying the overall electromagnetic field energy and plasmonic coupling effect. Detection of miRNAs is accomplished on the DAPA surface using capture DNA and locked nucleic acid probes to form complexes with target miRNAs. This interaction induces a red shift in the LSPR peak, with its amplitude enhanced by the presence of nanogaps in the DAPA structure, enabling attomole-level sensitivity [114].

Chang et al. integrated the strategy of fiber optic nanogold-linked sorbent assay (FONLISA) and DNA ligation specificity, developing a fiber optic particle plasmon resonance (FOPPR) biosensor for the assessing of ApoE genotyping in DNA extracts from human blood [115]. This methodology uses two probes that are complementary to target ssDNA. One probe is biotinylated, while the other is immobilized on dual-functional gold-iron oxide core-satellite hybrid nanoparticles (HNPs). When the target ssDNA is present, these probes form pairs through ligation, which can be detected with the assistance of HNPs. The HNPs serve as both plasmonic signal reporters and magnetic carriers for purifying ligated samples. The ligated products are subsequently captured by streptavidin-functionalized optical fibers, forming a sandwich-like complex on the fiber surface. This results in a change in the transmitted light intensities. The magnitude of this difference allows for the discrimination of six ApoE genotypes, with excellent performance in terms of minimal sample consumption (20 μL of blood), sensitivity, and specificity [116].

3.1.3. Surface plasmon resonance

Surface plasmon resonance (SPR), akin to LSPR, involves the oscillation of surface plasmons induced by incident light at specific angles. However, unlike LSPR, this phenomenon is not confined to nanostructures but distributed across the metal surface. In SPR-based biosensors, biomolecular interactions on the sensor surface lead to changes in reflectivity, angle, or wavelength over time. The intensity of these changes is proportional to the target concentration and is measured for analysis [90,92].

Zheng et al. identified two aptamers with specificity for both oligomeric and fibrillar forms of Aβ₄₀ aggregates, showing distinct recognition patterns via single-molecule force spectroscopy [117]. Accordingly, they developed an SPR-based dual-aptasensor to detect and differentiate between the two types of Aβ₄₀ aggregates in CSF. This sensor is created by immobilizing one aptamer on an Au film surface and the other on AuNPs. The utilization of two aptamers helps to avoid steric hindrance during recognition and enhances the selectivity and sensitivity of biosensing. The presence of Aβ₄₀ oligomers or fibrils results in changes in SPR resonance angle shift, reflecting the quantity of Aβ₄₀ aggregates [118]. Špringer et al. employed two types of antibodies and functionalized AuNPs for detecting tau-amyloid β complexes (tau-Aβ complex). In a sandwich-designed assay, when the target tau-Aβ complex passes over the SPR chip surface immobilized with tau antibodies, it is captured by its tau part and subsequently binds to the biotinylated Aβ antibody by its Aβ part. Finally, the biotinylated end of the Aβ antibody binds to streptavidin-functionalized AuNPs (S-AuNPs), enhancing the SPR signal response rather than providing a colorimetric signal. This increases the mass on the metal chip surface, causing complex changes in the refractive index of the sensor chip, which are amplified by the binding of S-AuNPs. Consequently, under incident light, the binding events of tau-amyloid β complexes on the sensor chip cause a wavelength shift at the SPR dip, which is used as the sensor response for quantifying the target protein complexes [119].

To enhance the sensitivity of SPR biosensors, a stronger electric field on the metal surface is required, which can be achieved by incorporating non-plasmonic graphene-based materials. For instance, graphene and graphene oxide (GO) are two-dimensional materials known for their charge transfer capability, high surface area for biomolecule binding, and ease of fabrication. These materials can be combined with metallic plasmonic materials to augment the electric field and improve sensitivity to changes in refractive index [120]. Building on this concept, Nangare et al. developed a GO-SPR biosensor for detecting Aβ₄₂. In addition to enhancing specificity and sensitivity, they employed the layer-by-layer (LbL) technique and assembled nanomaterials to immobilize multiple antibodies for target recognition. The LbL assembly consisted of AgNPs coated with layers of chitosan (CS) and polystyrene sulphonate (PSS) applied sequentially, with antibodies affixed to the outer layer (AgNPs–CS–PSS-CS@anti-Aβ). This assembly was then deposited onto the activated GO surface of the SPR biosensor. Upon capturing Aβ₄₂ by the antibodies, the binding events induced changes in the SPR signal, which were amplified by the coupling of GO and LbL assemblies, facilitating the detection of Aβ₄₂ [121].

Optical fiber-based SPR sensors have been developed for the detection of AD biomarkers [122–125]. These sensors utilize total internal reflection within the optical fiber to generate an evanescent field, while the fiber surface is coated with plasmonic metallic materials, enabling the induction of SPR with enhanced sensitivity. Furthermore, this type of SPR sensor is compact, adaptable to various materials and mechanical configurations, and suitable for remote sensing, making it a potential candidate for point-of-care testing [126,127]. Nu et al. demonstrated an SPR fiber sensor for serum Tau detection. They replaced the fiber cladding with Cr/Au coating and utilized a light source at 632.8 nm to achieve optimal SPR conditions. Binding events cause changes in fiber output power for multiplexed detection of t-tau and p-tau. The signal change increases with increasing Tau concentration, resulting in a nonlinear fitting with two exponential functions. Additionally, as the metal layer is coated with an elliptical cross–sectional profile, the evanescent wave attenuates in different directions, allowing more penetration into the metal layers to excite a greater number of surface plasmons. This enhances sensitivity compared to a uniform metal layer [124,128]. Khatri et al. designed a U-shaped SPR optic fiber biosensor to distinguish between monomeric and aggregated fibril forms of αSyn. The U-shaped optic fiber was manufactured by exposing the decladded region to a butane flame, applying sufficient force on both sides of the fiber, and bending it into a U-shape with a radius of 1.5 mm. The outer side of the uncladded U-bent section was then immobilized with AuNPs and further coated with CS, serving as the probe for αSyn fibrils. CS selectively binds to αSyn fibrils but not monomers, leading to absorbance changes at 530 nm that can be utilized to detect different forms of αSyn in mixtures [122]. The U-shaped configuration alters the light propagation along the fiber compared to a straight fiber, enhancing sensing sensitivity. The bending mechanical force causes the refractive index distribution in the fiber core to be uneven, with more significant refractive index changes occurring as the bend radius decreases [128,129]. Additionally, at the outer side of the U-bend, higher-order modes attenuate, and the penetration depth of the evanescent field increases, which in turn stimulates more localized surface plasmons of the AuNPs, resulting in stronger signals [127].

A dual-channel SPR sensor chip based on enzymatic cleavage reactions was established for ApoE genotyping without DNA amplification. Developed by Yi et al., this method utilizes a gold-filmed chip with immobilized complementary DNA probes, one end of which is modified with biotin, enabling recognition by streptavidin to amplify SPR signals. Depending on the SNP of the ApoE gene, the target gene fragments can be either single-base mismatch (GTGC) or complementary (GCGC) to form a duplex with the anchored probes. The latter duplex pairs are then cleaved by the restriction enzyme HhaI, resulting in a reduction of biotin-streptavidin recognition, followed by differential SPR signals for ApoE genotyping in humans [130].

Biolayer interferometry (BLI) is a robust optical biosensor capable of detecting various types of analytes. Functionally, BLI takes the form of a disposable optical fiber-based biosensor and operates in a Dip-and-read format. The biolayer interface at the end of the fiber is dipped into the sample well to interact with target molecules [131]. As the incident white light travels down the optic fiber to the end, the internal and external surfaces of the biolayer reflect the light back to spectrophotometers asynchronously due to distance differences, establishing an optical interference pattern [131,132]. The pattern is plotted with the reflected wavelength on the x-axis and relative intensity on the y-axis. In the presence of analytes, binding events change the thickness of the biolayer and increase the propagation length of light, resulting in a wavelength shift in the interference pattern. Since the shift is associated with the size, affinity, and amount of bound analytes, BLI can detect the number of analytes in real-time or monitor dissociative events over time by measuring this shift [131–133]. Ziu et al. developed a BLI-based aptasensor for the Tau protein (Tau441) by immobilizing a biotinylated aptamer on the streptavidin interface as a capture probe. This aptasensor marks the first BLI application in AD biomarkers and demonstrates selectivity over Aβ and αSynwhen applied to fetal bovine serum [134].

SPR imaging (SPRi) is an extension technique of SPR, which combines the properties of SPR with microarray imaging, enabling high-throughput and simultaneous analysis of multiple reactions [135]. Gao et al. utilized Aβ₄₂-specific binding material, AD peptoid 3 (ADP3), along with SPRi technique to develop an SPR nanosheet sensor for Aβ₄₂ detection in plasma and serum. The nanosheet sensor was composed of amphiphilic self-assembling peptoids immobilized on a functionalized gold chip surface, with ADP3 inserted and exposed as a loop on the surface to mimic Aβ₄₂ antibodies. The presence of Aβ₄₂ would be recognized by ADP3 loops, resulting in changes in the RI on the chip surface, which are recorded to reflect the binding events on the surface [136].

Plasmonic interferometry, a nano-plasmonic biosensor introduced by Feng et al., employs a nanoscale groove-slit-groove design on a metal surface, as shown in Fig. 3. This innovative technique combines the properties of surface plasmon polaritons in gold with light interference to detect both SPPs excitation and RI changes caused by compositions near the surface [137]. Unlike typical SPR or LSPR, plasmonic interferometry overcomes limitations such as the need for specific frequencies for resonant scattering and extinction production. With its broadband operations, allowing simultaneous detection of RI changes at multiple wavelengths, wide-angle excitation feasibility, and miniaturized sampling volume and sensing area, plasmonic interferometry emerges as a high-throughput device for biosensing [137]. Li et al. employed this technique with antibody-surface functionalization and successfully detected TREM2 in CSF, highlighting the potential of this technical approach for developing other immunoassays in the future [138].

Fig. 3.

Graphical illustration of a plasmonic interferometer with (a) scanning electron microscopy micrograph of a groove-slit-groove design on the metal surface, and (b) schematic working principle. Reprinted with permission from Nanoscale Plasmonic Interferometers for Multispectral, High-Throughput Biochemical Sensing (Ref [137]). Copyright 2012 American Chemical Society.

In summary, SPR and LSPR-based biosensors offer label-free detection, real-time response, and exceptional sensitivity in the picomolar/femtomolar range. As outlined in Table 3, these methods are capable of detecting and distinguishing various AD biomarkers, including but not limited to Aβ species, tau proteins, ApoE genotypes, and miRNAs. Many of these biosensors incorporate recognition elements to enhance selectivity and mitigate non-specific binding. Additionally, they utilize diverse nanostructures, non-plasmonic materials, optical fiber configurations, two-dimensional arrays, and interferometric principles to achieve miniaturization, simplification, high-throughput capabilities, and multiplex analysis. These advancements serve as inspiration for the development of biosensors targeting other analytes.

Table 3.

Nanomaterial-based SPR/LSPR sensors for AD biomarkers.

Nanomaterials	Biomarkers	Recognition element	Limit of detection	Sample	References
Au nano-urchins	Aβ42	–	–	–	[112]
AuNPs, AuNRs	Aβ40, Aβ42, Tau protein	Antibody	34.9 fM for Aβ40	Mimicked blood	[113]
			26 fM for Aβ42
			23.6 fM for Tau protein
DADP, gold nanomaterials	miRNA-125b, miRNA-15a, and miRNA-361	ssDNA	10.54 aM for miR-125b, 13.53 aM for miR-15a, 11.10 aM for miR-361	Human blood	[114]
AuNPs	Aβ40 aggregates	Aptamer	0.2 pM for Aβ40O, 0.05 pM for Aβ40 fibrils	Human CSF	[118]
AuNPs	Tau–Aβ complex	Antibody	1 pM	Human CSF	[119]
AgNPs–CS–PSS-CS@anti-Aβ, GO	Aβ42	Antibody	1.21 fg/mL	Rat blood and saliva	[121]
Au/Cr coated fiber	t-tau, p-tau	Antibody	2.4 pg/mL for t -tau, 1.6 pg/mL for p-tau	Human serum	[124]
AuNPs	α-synuclein		–	–	[122]
Gold-iron oxide core-satellite HNPs	ApoE gene	ssDNA	16–38 fM	Blood DNA extracts	[116]
Gold film	ApoE gene	ssDNA	10–50 fM	Blood DNA extracts	[130]
–	Tau441	Aptamer	6.7 nM	Fetal bovine serum	[134]
Peptoid nanosheet	Aβ42	ADP3 loop	1 pM	Human blood	[136]
Gold layer	TREM2	Antibody	–	–	[138]

Abbreviations: Ach: acetylcholine, AchE: acetylcholinesterase, ADP3: AD peptoid 3, AgNPs: silver nanoparticles, anti-Aβ: Aβ antibody, AuNPs: gold nanoparticles, AuNRs: gold nanorods, CS: chitosan, DADP: DNA-assembled advanced plasmonic, GO: graphene oxide, HNPs: hybrid nanoparticles, PSS: polystyrene sulphonate.

Based on the preceding discussion, colorimetric sensors are essential techniques widely employed for detecting AD biomarkers. These sensors exploit the principle of color change and are typically constructed using various components such as chromogenic substrates, peroxidative enzymes, metallic, and composite nanomaterials. Offering advantages of label-free detection, naked-eye visibility, and high sensitivity in biological samples, these technologies have demonstrated specificity for target biomarkers and applicability in simulated or real biological specimens. Thus, colorimetric sensors hold significant promise for the diagnosis and monitoring of AD and are poised to continue playing a pivotal role in future research and applications.

3.2. Nanomaterial-based fluorescent sensors

Nanomaterials can be applied as a fluorescent indicator, quencher, and carrier in the fluorometry detection strategy (shown in Fig. 4). Compared to traditional fluorophores such as fluorescein amidites (FAM), cyanine-5 (Cy5) and so on, fluorescent nanomaterials, such as quantum dots (QDs) [139], upconversion nanoparticles (UCNPs) [140], and MOF [141], possess unique optical properties like large Stokes shift, ease of modification and controllable emitting wavelength. These properties provide more flexibility to establish diverse fluorometric methods. On the other hand, the application of nanomaterials as a quencher in a fluorometric method is also useful to enhance the performance due to the high extinction coefficient or high interaction surface, which facilitates the electron transfer, especially noble metal nanoparticles and GO [142,143]. Other nanomaterials, such as Prussian blue nanoparticles (PBNPs), polydopamine nanoparticles (PDANS) and transition metal nanosheets [144–146], were also established as a quencher for the development of biosensing methods. Furthermore, nanomaterials, such as magnetic nanoparticles (MNPs), can serve as carriers, capable of functionalizing recognition elements and providing diverse detection mechanisms by magnetic properties [147]. Table 4 summarizes the nanomaterial based fluorescent sensors used for AD biomarkers detection. It includes nanomaterials, recognition elements, limit of detection, and sample matrix.

Fig. 4.

Fluorometric strategies for AD biomarkers based on nanomaterials which were constructed as fluorophore or quencher. (a) Fluorescent sensor based on FRET-induced quenching. (b) Fluorescent sensor based on magnetic separation-induced signal decline.

Table 4.

Nanomaterial-based fluorescent sensors for AD biomarkers.

Nanomaterials	Biomarkers	Recognition element	Limit of detection	Sample	References
(1) GSH-capped CdTe QDs	AβO	PrP (95–110) peptide	0.2 nM	–	[149]
(2) AuNPs
(1) BaYF5:Yb,Er UCNPs	AβO	Aptamer	36 pM	aCSF	[151]
(2) MNPs
PDANS	AβO	Aptamer	12.5 nM	–	[176]
GO	AβO	Aptamer	20 pM	Exosome	[181]
MNPs	Aβ42	Antibodies	38 fg/mL for Aβ42	Human CSF	[31]
	Tau441	12F4 for Aβ42	197 fg/mL for Tau441
	p-tau 181	BT2 for Tau441	165 fg/mL for p-tau181
		AT270 for p-tau181
CuInS2/ZnS QDs	Tau protein	Aptamer	10 pM	Human serum	[156]
		Antibodies TYR-anti-Tau
CDs@Eu/GMP ICP nanoparticles	Aβ monomer	Cu2+	0.17 nM	AD mice brain tissue	[166]
AuNPs	Aβ40	SEA-SC2 (fluorescent molecule)	–	aCSF	[180]
	Aβ42
PBNPs	AβO	Aptamer	1 nM	aCSF	[175]
				Human CSF
MoS2 NSs	AβO	Aptamer	3.1 nM	AD mice brain tissue	[173]
L-MOF	AβO	Aptamer	0.3 pM	Human serum	[150]
AuNPs
L-MOF-armored-anti-DNA antibody	AβO	Aptamer	0.4 pg/mL	Human serum	[169]
m	Presenilin 1	Spilt SSODN-P1 & P2	0.517 pM for presenilin 1	aCSF	[170]
	Aβ protein	–	0.142 nM for Aβ protein
TAMRA-labeled substrate strand-AuNPs	AβO	Aptamer	22.3 pM	Living cell	[177]
				Mice brain
FAM-dsDNA-GO	AβO	Aptamer	0.1 nM	Human CSF	[179]
UCNPs	AβO	H-USM (Zn2+)	28.4 pM	–	[171]
MOF
CdSe/CdS/ZnS QDs	AβO	AβO aptamer-QDs625	50 pM for AβO	Human serum	[157]
AuNRs@PDA	Tau protein	Tau aptamer-QDs665	20 pM for Tau protein
NCDs	p-tau231	Aptamer	3.64 ng/mL	–	[148]
NCDs	AβO	Antibodies	83 pg/mL	Artificial serum	[165]
BP nanosheets

Abbreviation: Aβ P: aptamer for Aβ42, Ach: acetylcholine, AchE: acetylcholinesterase, aCSF: artificial cerebrospinal fluid, AuNRs@PDA: polydopamine-capped gold nanorods, BP: black phosphorus, CdSe/CdS/ZnS QDs: CdSe/CdS/ZnS core/shell/shell quantum dots, Er-MOF: metal–organic framework [Er(L) (DMF)_1.27]_n, GMP: 5′-guanosine monophosphate disodium, GSH: glutathione, H: hairpin probe, H-USM: highly doped upconversion nanoparticles-SiO₂@MOF microspheres, ICP: infinite coordination polymer, L-MOF: luminescent metal–organic framework, MNPs: Fe₃O₄ magnetic nanoparticles, MoS₂ NSs: molybdenum disulfide nanosheets, NCDs: nitrogen-doped carbon dots PBNPs: prussian blue nanoparticles, SSODN: part of the presenilin 1 gene, Spilt SSODN-P1 & P2: DNA complementary pair sequence with SSODN 1 & 2, Tau P: aptamer for Tau441 and p-tau181, PDANS: polydopamine nanospheres, ThT: thioflavin T, TYR: tyrosinase.

3.2.1. Nanomaterials as a fluorescent indicator

The application of nanomaterials such as QDs [148,149], MOF [150], and UCNPs [151] as fluorescent indicators to develop sensors for detecting biomarkers of AD is summarized here. QDs are the most commonly used nanomaterials-based fluorescent indicator. The QDs can be divided into two categories: semiconductor QDs and carbon-based QDs (CDs) [152,153]. The emitted wavelength of QDs is tunable by manipulating the size of QDs according to the quantum confinement effect [154,155]. The changing size of QDs can influence the discretization of electronic energy states, which induces the confinement of electrons and holes by the conduction and valence bands. With the property of QDs, the detected fluorescent signal can be flexibly designed, depending on the requirement of developed analytical methods. For semiconductor QDs, they were widely applied in the development of optical biosensing methods with diverse fabrication strategies, such as CdTe QDs [149], CuInS₂/ZnS QDs [156], CdSe/CdS/ZnS QDs [157]. A critical drawback of QDs is their poor solubility in water, which can make them incompatible with biosensor applications. Modifying the surface of QDs with hydrophilic functional groups is an ideal solution to overcome this problem. Xia et al. developed a glutathione-capped CdTe QDs sensor based on the inner filter effect (IFE) from AuNPs for the detection of AβO [149]. The non-irradiation energy conversion of IFE provided a flexible quenching of QDs by AuNPs upon the overlapping of emission and absorbance of QDs and AuNPs, respectively [158]. The modification of glutathione onto the surface of QDs as a stabilizer can enhance water-solubility and biocompatibility [159]. In addition to ligand functionalization, a surface-coating of QDs with a layer, such as ZnS, is also feasible to coordinate the compatibility between QDs and aqueous environment. Notably, the surface passivation by ZnS can also enhance the photoluminescence of QDs induced by band-edge recombination due to the suppression of deep trap emission [160]. The deep trap, the band gap in the middle of exciton absorption/emission band, will trap exciton and cause the quantum yield to decrease. With a ZnS shell on the surface of QDs, the energy band gap can be broadened, suppressing deep trap emission and allowing more electrons to return to the ground state [161]. It allows semiconductor QDs to possess outstanding optical characteristics that can enhance their photoluminescence under extra passivation by a surface-coating process. In addition, ligand functionalization or exchange can be achieved in the semiconductor QDs system, allowing them to be applied in various detection strategies [162]. Core/shell QDs, with their optical properties and ligand exchange, were used as fluorescent indicators for biosensing. Lu et al. developed an aptasensor for the simultaneous detection of AβO and tau protein by two kinds of CdSe/CdS/ZnS QDs with different emission wavelengths by an identical excitation and applied polydopamine-capped gold nanorods as the quenching medium [157]. Additionally, a strategy was devised for the functionalization of quantum dots with redox-active biomolecules. Chen et al. developed a sandwich immune-aptasensor for the detection of tau protein. A dopamine-functionalized CuInS₂/ZnS QDs were used as the signal indicator, and the tau-antibody was functionalized with tyrosinase. The dopamine on the QDs can transform to be dopamine quinone due to the catalytic oxidation by tyrosinase. The fluorescence of QDs will be quenched by the dopamine quinone, which serves as an electron acceptor [156].

The optical properties of CDs are similar to those of semiconductor QDs. The surface passivation of CDs induced the energy trap and emitted a photoluminescence [163]. Yet, the synthesis of QDs can be achieved by the carbonization of natural organic materials and can be easily manipulated in the aqueous environment, such as hydrothermal synthesis [164]. Thus, fabricated CDs are more environmentally friendly than metal QDs and more appealing to be applied in biosensing strategy without additional hydrophilic modification. For instance, Phan et al. used citric acid monohydrate as a carbon source and PEI1800 as a nitrogen source to synthesize nitrogen-dopped CDs (NCDs), which possess positive-charged surface and pyridine-type fluorophore intermediates on the surface [148]. The positively charged surface of NCDs allowed nucleic acid, such as aptamers, to attach onto their surface by electrostatic interaction, which can also induce fluorescence resonance energy transfer (FRET), causing fluorescence quenching from NCDs. They applied this mechanism to develop an aptasensor for the detection of p-tau231 upon the detachment of aptamer out of NCDs. When p-tau231 was introduced, it bound to the aptamer, causing the NCDs to be released and resulting in the recovery of fluorescence. Besides, the NCDs can also be implemented a surface functionalization with antibodies so as to construct an immune-based biosensor [165]. On the other hand, CDs can also serve as an antenna ligand to assist the self-assembly of europium (Eu³⁺) and 5′-guanosine monophosphate disodium (GMP) to form CDs@Eu/GMP infinite coordination polymer nanoparticles which encapsulated CDs inside the Eu/GMP network. This can be attributed to the abundant functional groups of CDs (-COOH and –NH₂), which allow for the interaction of coordination. The as-formed nanoparticles can be utilized to detect Aβ monomer in a detection environment with cupric chloride to provide Cu²⁺, which can coordinate with Aβ monomer. The competitive mechanism of Cu²⁺ against Eu²⁺ for the coordination to CDs disrupted the antenna effect of CDs, leading to the decline of fluorescence, and the stronger interaction of Aβ monomer against Cu²⁺ can recover the fluorescence [166]. Compared to semiconductor QDs, CDs have superior biocompatibility and lower toxicity than semiconductor QDs. The manufacture of carbon dots is flexible, allowing for diverse configurations from carbon and ligand sources. However, their actual composition and surface characteristics are challenging to identify.

Other fluorescent nanomaterials, such as MOFs and UCNPs, also possess appealing properties applicable to sensor development. Unlike small QDs or CDs less than 10 nm, MOFs and UCNPs have larger dimensions, providing more capacity to implement surface modification. MOFs are geometrically structural porous crystalline composed of metal clusters, organic ligands (linkers), and guest molecules. MOFs possess a luminescent property attributed to the linker excitation, which performs a charge transfer with metal clusters. In addition, the π-π stacking interaction between adjacent aromatic structures in MOFs provides a large Stokes shift in emission [167]. UCNPs were mainly composed of rare earth elements, including the lanthanide (Ln) family and yttrium (Y) with abundant energy level which can implement intra-configurational transitions. These traits allow UCNPs to emit luminescence under near-infrared excitation [168]. These nanomaterials with luminescent properties can be feasibly utilized in the development of sensors by modification of recognition elements [151] or direct surface interaction upon intermolecular force and electrostatic attraction with recognition elements [150,169]. In the practical application, the optical property of MOF allowed researchers to develop various detection mechanisms [150,169,170]. For instance, Wang et al. construct a multifunction thioflavin T (THT)-embedded Er-MOF sensor for the detection of various biomarkers for AD, including Aβ, acetylcholine, and the gene of presenilin 1. The conjugated structures in MOF provide hydrophobic and π-π stacking interactions against protein or DNA sequences. These surface interactions can further affect the emissive ability of MOF. Additionally, the embedded THT provides selectivity of Aβ, making MOF a suitable candidate for developing diverse biosensing methods [170]. On the other hand, the MOF technique can also be combined with the UCNP technique. Fang et al. developed optically trapped highly doped upconversion nanoparticles, SiO₂@MOF microspheres (H-USM), which wrapped BHQ-1 inside the structure of MOF. The wrapped BHQ-1 can quench the luminescence of UCNPs via luminescent resonance energy transfer. The method was used to detect AβO by the interaction between AβO and the Zn²⁺-fabricating MOF, attributed to the high affinity of histidine (His), glutamate (Glu), and aspartic acid (Asp) of Aβ peptide against Zn²⁺. AβO can induce the disassembly of MOF, and subsequently, BHQ-1 molecules will be released out of microspheres. Eventually, the luminescence of UCNPs can be recovered, and the changing signal can be utilized to quantify the levels of AβO [171].

3.2.2. Nanomaterials as a quencher

AuNPs are commonly used as a quencher due to a high extinction coefficient and a broad absorption spectrum, which enables them to effectively quench fluorescence by overlapping with the emission of common fluorophores [142,172]. Another common quencher, GO, is a planar nanomaterial with abundant aromatic structures composed mostly of sp²-hybridized carbon atoms. GO provides the property of π-π stacking interaction, which can attract substances with aromatic structures and quench fluorescent indicators through electron transfer [143]. Similar to GO, the planar structural transition metal nanosheets, such as black phosphorus nanosheets and MoS₂ nanosheets [165,173], were also regarded as appealing quenchers in the development of analytical methods. These planar nanomaterials also possess electron transfer properties that facilitate fluorescence quench. Attributed to their charged property, the adsorption of recognition elements can easily be achieved by the van der Waals force and intermolecular force [174].

Other nanomaterials, such as PBNPs and PDANS, also possessed the property to quench fluorescence. Fe³⁺ and Fe²⁺ ions can quench fluorescent indicators on PBNPs by attracting negatively charged substances to the surface, facilitating energy transduction [175]. On the other hand, PDANS possessed the structure of aromatic polydopamines which provided intermolecular force, such as π-π stacking interaction and hydrogen binding against positive charged or aromatic substances. When fluorescent indicators are close to PBNPs, the fluorescence can be quenched due to the broad absorption spectrum [176].

In the application of these nanomaterials with quenching properties, researchers can functionalize recognition elements onto the surface of AuNPs [150,177] or GO [178,179], such as nucleic acids (aptamer) and antibody. In addition to surface functionalization of recognition elements onto nanomaterials, several detection mechanisms were designed to directly adsorb recognition elements onto the surface of nanomaterials with quenching properties by intermolecular force or electrostatic attraction, including small molecules [180], peptides [149], nucleic acids [157,173,175,176,181] and antibody [165]. Relying on altering the distance between the fluorescent indicator and nanomaterials with quenching properties, in the presence or absence of a target, the changing fluorescent signal can be utilized to quantify the level of a target.

3.2.3. Nanomaterials as a carrier

MNPs were the most used nanomaterials as carriers that could load recognition elements. Different from the quenching property, such as FRET or dipole–dipole coupling, the changing fluorescent intensity can be attributed to eliminating magnetic carriers in the detection environment by an external magnetic field. Take the research from Jiang et al. as an example, they established an aptamer complementary DNA-functionalized BaYF5:Yb,Er UCNPs coupled with a magnetic separation technique through aptamer-functionalized MNPs for the detection of AβO [151]. AβO can resist the hybridization between sequences on UCNPs and MNPs when AβO interacts with aptamers on the surface of MNPs. A distinct fluorescent signal can be obtained in the presence of AβO. In contrast, a decline signal can be observed in the absence of AβO due to the elimination of UCNPs from the detection system by MNPs.

In addition to the magnet-assisted separation, MNPs can also serve as carriers with diverse shapes, which can be used to distinguish different analytes when these carriers are functionalized with recognition elements for specific targets. For example, Chan et al. developed two magnetic nanocomposites to detect Aβ₄₂, Tau441, and p-tau181 in visualization. For Tau441 and p-tau181, sphere-shaped magnetic nanocomposites were fabricated and conjugated with Tau441 or Tau181-specific capture antibody; for Aβ₄₂, magnetic Janus rods were constructed and functionalized with Aβ₄₂-specific capture antibody. To enhance the sensitivity, Chan et al. utilized aptamers of Tau181 and Aβ₄₂ coupled with the technique of HCR [182]. By using a tailor-made fluorophore, called SPOH, which can bind to double-strand DNA generated from the HCR, a fluorescent image exhibiting the dispersion of different morphological magnetic carriers with Aβ₄₂, Tau441, and p-tau181 can be obtained and further can be quantified the levels of analytes [31].

Recently, the utilization of fluorescent or quenching indicators has gradually turned from conventional molecules to optical nanomaterials. Optical nanomaterials can be further implemented through surface modifications or functionalization. The modification of nanomaterials allows researchers to refine the sensing performance, which can’t be effectively manipulated in the context of a traditional fluorometric method by optical molecules, such as reduction of nonspecific adsorption, recognition of targets, providing better reaction space, combination with signal amplified techniques and so on. Optical nanomaterials which not only serve as the source of detection signal but also provide other functions, diverse detection mechanisms of fluorometry can be achieved. The high flexibility of constructing a fluorescent sensor allowed researcher to design an appropriate method according to their research strategy. However, some functional fluorescent indicators are hard to fabricate or have a complicated manufacturing process, including synthetic and functionalization procedures. Even if they possessed outstanding optical properties or multiple functions, the drawback of manufacturing inevitably limits the practical application of these fluorescent nanomaterials. The reliability and applicability will be greatly improved if the process of manufacturing these nanomaterials can be simplified. Such an improvement also benefits the quality control of fluorescent nanomaterials from batch to batch.

3.3. Surface-enhanced Raman scattering

Raman scattering is the inelastic scattering of radiation ascribed to molecules’ vibration, namely the Raman effect. However, a weak Raman effect sometimes occurs and leads to low sensitivity. This drawback limits the utilization in the development of analytical methods for trace amounts of analytes. Fleischmann group first proposed an improved strategy for Raman scattering, also known as SERS. While conducting experiments, they discovered that adsorbing pyridine on rough silver surfaces led to an unexpectedly large Raman signal [183]. According to their observations, the increased intensity can be obtained due to the enhanced electromagnetic effect and chemical effect, induced by localized surface plasmons from metals on a nano-scale and the adsorption of molecules onto the surface of metals, respectively [184]. The region of magnified electromagnetic field by the effect of LSPR was also known as “hot spots” [185]. Many researchers have investigated the shape and size of plasmonic nanomaterials to obtain an optimized amplified intensity of SERS. When the distance of immobilized nanomaterials gets into close proximity but are not in touch with each other, the electromagnetic field of “hot spots” can be acquired, which provides the highest enhancement of Raman scattering. The maximum enhancement factor of the Raman signal is approximately ten orders of magnitude [186]. The breakthrough of SERS successfully became prevalent in the development of sensitive analytical methods. SERS has been successfully applied in assays requiring high sensitivity for AD biomarkers. The SERS techniques developed to detect AD biomarkers have been summarized in Table 5, and the schematic diagram is shown in Fig. 5.

Table 5.

Nanomaterial-based SERS sensors for AD biomarkers.

Nanomaterials	Biomarkers	Recognition element	Limit of detection	Sample	References
Non-antibody or aptamer
Au nanoshell	AβO	Sialic acid-modified Au nanoshell	1 pM	–	[201]
Graphene–gold hybrid plasmonic platform	Aβ42	–	1 aM	Human plasma	[190]
AuNPs	Aβ42	RB-AuNPs	–	Mouse brain slice	[202]
Chiral Pt@Au TNRs	Aβ42 monomers	D-Pt@Au TNRs	45 fM for Aβ42 monomer	CSF	[188]
	Aβ42 fibrils		4 fM for Aβ42 fibrils
P3/AgNPs SERS sensor	Aβ42 human insulin	N9-(pyridin-2-ylmethyl)-N6-methoxyadenine-functionalized P3/AgNPs	76 aM for Aβ42	aCSF	[200]
			26 aM for human insulin
AgNP/CS sensor	ZnAβ42	–	1.5 nM for ZnAβ42 by	Plasma	[196]
AgNP/LID sensor	CuAβ42		AgNP/CS sensor 15 pM for CuAβ42 by AgNP/LID sensor
Antibody
AuNPs-coated ITO substrate	Aβ40	β-amyloid antibody-modified Au-dot/ITO	100 fg/mL	–	[193]
Fe3O4@SiO2	Tau	monoclonal anti-Tau-modified Fe3O4@SiO2	25 fM	–	[207]
AuNPs		DTNB-coded polyclonal anti-Tau-modified AuNPs
AgNGS	Aβ40	AgNGS[4-FBT]- Aβ40	0.25 pg/mL for Aβ40	Human serum	[206]
	Aβ42	AgNGS[4-BBT]- Aβ42	0.33 pg/mL for Aβ42
		CAb-MB
Fe3O4@GOs	Aβ42M	Anti-Aβ-amyloid antibody-modified Fe3O4@-GOs	1.62 fg/mL for Aβ42M	Serum	[194]
Au@Ag probe	p-tau181		5.74 fg/mL for p-tau181
		Anti-Aβ-amyloid antibody-modified Au@Ag probe
		Anti-Tau-amyloid antibody-modified Fe3O4@GOs
		Anti-Tau-amyloid antibody-modified Au@Ag probe
Au nanopillars	Tau441	Half tau-antibody fragment-modified head-flocked Au nanopillars	3.21 fM	Plasma	[195]
AuNPs
		Half Cy3-conjugated tau polyclonal antibody-modified AuNPs
Au@SiO2	Aβ40	Aβ40/Aβ42/Tau/NfL monoclonal antibodies-modified nitrocellulose membrane	191.2 fg/mL for Aβ40	–	[210]
	Aβ42		138.1 fg/mL for Aβ42
	Tau441	Au4–MBA@SiO2 with Aβ42/Tau monoclonal antibodies	257.1 fg/mL for Tau441
	NfL		309.1 fg/mL for NfL
		AuDNTB@SiO2 with Aβ40/NfL monoclonal antibodies
SIONs	p-tau396,	Antibody 3G5-SIONs	1.5 pg/mL	whole blood	[208]
AuNPs	p-tau404	4-MBA@AuNP-HRP-R1
Au-plasmonic shell attached polystyrene microsphere	Aβ42	Aβ-Ab1	100 fg/mL	–	[211]
	p-tau181	NBA/AuNPs/Aβ-Ab2
		Tau-Ab1
		4-ATP/AuNPs/Tau-Ab2
Aptamer
AuNPs	AβO	4-AATP-coded PAapt-AuNPs	37 pM for AβO	aCSF	[203]
	Tau381	DTNB-coded PAapt-AuNPs	0.42 fM for Tau381

Abbreviation: 4-AATP: 4-acetamidothiophenol, 4-ATP: 4-aminothiophenol, 4-BBT: 4-bromobenzenethiol, 4-FBT: 4-fluorobenzenethiol, 4-MBA: 4-mercaptobenzoic acid, Aβ-Ab:antibody-Aβ₄₂, AgNGS: Ag nanogap shell, AgNGS_[4-BBT]-Aβ₄₂: anti-Aβ₄₂ antibody-conjugated AgNGS entrapping 4-BBT, AgNGS_[4-FBT]-Aβ_-40: anti-Aβ₄₀ antibody-conjugated AgNGS entrapping 4-FBT, Au@Ag probe: gold-silver alloy nanoprobe, AuNPs: gold nanoparticles, CAb-MB: capture Ab-conjugated magnetic bead, Chiral Pt@Au TNRs: chiral triangular Au nanorings fabricated with a platinum framework, CS: chitosan, CuAβ₄₂: Cu²⁺ assisted aggregated Aβ₄₂, DTNB: 5,5′-dithio-bis (2-nitrobenzoic acid), Fe₃O₄@GOs: magnetic graphene oxide, HRP: horseradish peroxidase, ITO: indium tin oxide, LID: laser induced deposition, NBA: Nile rule A, NfL: neurofilament light chain protein, P₃: ligand (100 mg)-impregnated polyacrylonitrile nanofibers mat, PAapt-AuNPs: polyA aptamer–AuNPs conjugates, RB: rose bengal, SION: superparamagnetic iron oxide nanoparticles, Tau-Ab: antibody-Tau, ZnAβ₄₂: Zn²⁺ assisted aggregated Aβ₄₂.

Fig. 5.

SERS strategies for AD biomarkers based on nanomaterials with different mechanisms.

Attributed to the SERS spectra exhibition, the changing formation of Aβ during self-aggregation can be observed by the changing of dominant wavelength on α-helix and β-sheet structure [187]. In addition to the qualification of Aβ conformation, the SERS technique is also eligible to quantify biomarkers of AD. In the construction of SERS-active plasmonic nanomaterials, researchers have developed various morphologies of nanomaterials to enhance electromagnetic fields. For instance, Wang et al. established chiral triangular nanorings (chiral Pt@Au TNRs) for the detection of Aβ₄₂ monomers and fibrils [188]. The Raman scattering of _D-Pt@Au TNRs after interacting with analytes was significantly enhanced, attributed to surface-enhanced Raman scattering-chiral anisotropy (SERS-ChA) [189]. Additionally, the conformation of Aβ₄₂ monomer and fibrils can also be discriminated and quantified from the vibrational information of Raman spectra by the transformation of Raman peaks from the α-helical structure to the β-sheet structure [188]. Moreover, Yu et al. developed a graphene-gold hybrid plasmonic platform with a pyramid structure which concentrated hot spots and facilitated the SERS enhancement. This platform achieved a high detection efficiency with a detection limit of 1 aM [190].

In addition to the investigation of the morphology of plasmonic nanomaterials, the substrate underneath them can also influence Raman enhancement performance by providing charge transfer [191,192]. The substrate structure for SERS can be divided into rigid and flexible substrates. Generally, a rigid substrate is the most commonly applied substrate for the development of SERS-based sensors, such as indium tin oxide (ITO) [193], graphene [190], GO [194], gold pellet [195], laser-induced deposition (LID) [196]. For detecting Aβ₄₂, Yu et al. used a single layer of graphene loaded with a pyramid-shaped gold structure [190]. Graphene enhances the Raman signal by transferring charge through π-π stacking between analytes and graphene [197]. Furthermore, graphene also possesses an intrinsic Raman signal, which can serve as an internal standard for the sensor [190]. Yu et al. also utilized GO to immobilize Fe₃O₄ nanoparticles, avoiding self-gathering and enhancing biocompatibility and dispersion [194]. Eremina et al. conducted an LID experiment by immersing the coverslip in a solution of silver benzoate in methanol and illuminating it with a laser source. The laser-induced chemical reaction can reduce metal ions to metals, and metals can further precipitate on the substrate surface with high adhesion [198]. An AgNPs-immobilized LID (AgNPs/LID) was fabricated and utilized to construct a SERS-based sensor for the detection of Cu²⁺ assisted aggregated Aβ₄₂ (CuAβ₄₂) which can acquire a limit of detection down to 1.5 pM by a 633 nm laser [196]. In addition, researchers have investigated the use of a flexible SERS substrate due to its advantageous properties of being easily prepared and scalable [191,199]. Jaiswal et al. fabricated a polyacrylonitrile nanofibrous mat with a purine-based ligand via electrospinning, then functionalized it with AgNPs (P₃/AgNPs) as a SERS sensor for detecting Aβ₄₂ and human insulin. The purine-based ligand possessed non-covalent interaction with peptide analytes and could bind AgNPs due to nitrogen centers in its heterocyclic structure. With the cooperation of surface interaction of AgNPs and non-covalent interaction between analytes and purine-based ligand, the fabricated P₃/AgNPs SERS sensor acquired enhancement factors of 3.1 × 10¹² and 6.6 × 10¹² for Aβ₄₂ and human insulin with detection limits of 76 aM for Aβ₄₂ and 26 aM for human insulin [200]. Likewise, Eremina et al. utilized a uniform distribution of AgNPs on the CS substrate, achieved through a colloidal suspension, to form AgNPs/CS substrate in developing a SERS sensor. This SERS sensor was applied to differentiate between amyloid aggregates and monomers. Mean-while, the sensor can detect Zn²⁺-assisted aggregated Aβ₄₂ (ZnAβ₄₂) with a limit of detection of 1.5 nMfor Aβ₄₂ aggregates using a 532 nmlaser [196].

SERS-based methods possess the property of identifying targets with specific Raman fingerprint information attributed to intrinsic molecular vibration. However, these methods may yield false-positive signals, especially when detecting analytes in complex matrices. Additionally, the specificity of the developed methods may be a dilemma, especially for those targets with compositions similar to their analogs. To overcome this problem, some researchers utilized recognition ligands such as small molecules, antibodies, or aptamers on the surface of nanomaterials in the development of SERS-based sensors [193,201–203], so as to refine the selectivity, avoiding false quantification of targets due to the interference of analogs. SERS tags (dye)-coded nanomaterials coupled with aptamers or antibodies as the recognition element greatly enhance the Raman intensity relied on these molecules with intrinsically strong Raman signals. Take aptamer-based strategy, for example, Zhang et al. developed an aptamer-based SERS sensing platform by designing two Raman dye-coded AuNPs conjugates for a multiplexed detection of Aβ₄₂ oligomers and Tau proteins. In this study, the aptamer sequences were extended with poly consecutive adenine (polyA) sequences (PApt), which can anchor onto the surface of AuNPs by high affinity without modification and prevent adjacent AuNPs from aggregation [204]. In the presence of targets, PApt can bind with corresponding targets and detach from AuNPs. Subsequently, AuNPs will aggregate by the plasmonic coupling effect, and a distinct SERS signal can exhibit due to the enhancement of the electrostatic field from hot spots [203]. Yang et al. have developed sandwich antibody-based SERS sensors for detecting Tau441. This design aims to position Tau441 close to the hot spot during antigen–antibody interaction. The half antibody fragment is placed on the head-flocked gold nanopillar substrate, while the Cy3-half antibody fragment is attached to AuNPs, serving as SERS nanotags [195]. The researchers reduced the size of antibodies by treating tris(2-carboxyethyl) phosphine (TCEP) to break apart the disulfide bonds and divide the antibody into two fragments. This process resulted in half antibody fragments that could be placed close to hot spots on the adjacent point between gold nanopillars on the substrate [205]. The electromagnetic field greatly enhanced and generated a higher Raman signal after the interaction of analytes and antibodies adjacent to hot spots. The sensor achieved a 135-fold increase in sensitivity compared to the antibody-based sensor. Yang et al. constructed two SERS signal probes of Aβ₄₀ and Aβ₄₂ antibodies-conjugated Ag nanogap shells (AgNGS), which respectively entrapped with Raman dyes, namely 4-fluorobenzenethiol (4-FBT) and 4-bromobenzenethiol (4-BBT), inside gaps of AgNPs synthesized on the surface of thio-functionalized SiNPs. A SERS-based sandwich immunosensor was developed for multiplex detection of AD biomarkers by the antibody-conjugated AgNGS entrapping Raman dyes and captured antibody-conjugated magnetic bead which can recognize the 1–16 sequence of Aβ₄₀ and Aβ₄₂. Attributed to different Raman signals exhibited from the 4-FBT and the 4-BBT, the simultaneous detection of Aβ₄₀ and Aβ₄₂ can be achieved, and enhanced Raman intensity was acquired to be 10⁷ for the limit of detection for amounts as low as 0.25 pg/mL [206].

Other studies have also used sandwich immunoassay detection strategies with different platforms to detect AD biomarkers. Zengin et al. constructed a SERS-based sandwich immunoassay for the detection of Tau coupled with a magnetic separation technique by 5,5′-dithio bis-(2-nitrobenzoic acid) (DNTB)-coded AuNPs modified with antibody and antibody-functionalized magnetic silica particles (Fe₃O₄@SiO₂) [207]. Zhang et al. established a magnetic dual-mode immunosensor for the detection of p-tau396 and p-tau404 in whole blood samples. They designed a signal antibody composed of HRP-labeled antibody-functionalized and 4-mercaptobenzoic acid (4-MBA) capped AuNPs (4-MBA@AuNPs-HRP-R1 probe), which can be simultaneously detected targets in Raman and colorimetric modes. This immunosensor can detect p-tau396 and p-tau404 in blood with as low as 1.5 pg/mL in SERS mode and 24 pg/mL in colorimetric mode [208]. Yu et al. fabricated an immunoassay on a magnetic graphene oxide (Fe₃O₄@GO) substrate for the detection of Aβ₄₂ and p-tau181 with the antibody-functionalized and 4-MBA coded tannin-capped AgNPs as a Raman enhanced signal probe [209]; Zhan et al. constructed an immunolateral flow assay, using Au@SiO₂ coded with 4-MBA and DNTB as SERS nanotags, for accomplishing the simultaneous detection of Aβ₄₀, Aβ₄₂, Tau, and NfL with two test lines [210]; Sun et al. established an immune-sensor in a microfluidic chip for the dual detection of Aβ and Tau with Nile rule A and 4-aminothiophenol-coded AuNPs as signal indicators, respectively [211].

According to the clinical research relevant to AD biomarkers, diagnostic values of these substances range from nano to pico orders of magnitude [24,25]. This review introduced many SERS-based sensors for the detection of AD biomarkers, some of which possess very high sensitivity and even reach femto or atto orders of magnitude. To obtain a higher sensitivity, it is crucial to find out the hot spot of nanomaterials against targets. It is also very important to manipulate the location of targets that interact with the SERS system. However, most researchers used antibodies to enhance the selectivity of SERS-based sensors. Antibodies possess a rigid and big-sized structure, which may limit the performance of SERS due to the large distance between SERS signal indicator (target or nanotag) and SERS-induced substrate (nanomaterials). To improve this drawback, half antibody-fragment was utilized to reduce the distance [195]. However, few studies used aptamer in such a scenario. In addition to specifically binding with targets, aptamer also provided properties of stability and ease of modification. Thus, the smaller size of aptamer exhibits less steric hindrance compared with antibodies [212]. The flexible structure of the aptamer may facilitate the manipulation of targets that reach the hot spot. Therefore, the aptamer technique may also be an ideal candidate for developing SERS-based sensors.

4. Conclusion

In the timeline of diagnosing AD biomarkers before 2014, most of the methods for detecting AD biomarkers were strategized in a CSF sample. Although the strategy of detecting AD biomarkers in a blood-based sample was also carried out in some studies, those outcomes showed no apparent evidence to demonstrate AD biomarkers in blood-based samples associated with the pathogenesis of AD [213]. There are several challenges associated with detecting AD biomarkers in blood-based samples. For example, levels of biomarkers in blood-based samples were lower than those in CSF samples due to the blood–brain barrier, which limits the influx of AD biomarkers into the blood circulatory system; proteolytic degradation or nonspecific interaction AD biomarkers with protein interferences in blood-based samples also influence the accurate quantification, especially for the antibody-based methods due to the hindrance of binding sites against targets [214]. However, even if confronted with many challenges during the research, the investigation of AD biomarkers in blood-based samples was still continuously implemented, which is attributed to the major drawback of collecting CSF samples. The collection of a CSF sample is performed by a lumbar puncture, which is an invasive treatment that patients are reluctant to undergo. Besides, such a sample collection method also discourages researchers from promoting the strategy of the early diagnosis of AD. Compared to the intrusive sample collection of CSF samples, the blood test is still an appealing and expected way to diagnose AD [215]. Notably, from 2014 to 2022, research on AD biomarkers in blood-based samples dramatically progressed due to sophisticated analytical techniques with high sensitivity, allowing researchers to effectively distinguish the different levels of AD biomarkers between healthy individuals and AD patients. In addition, it was crucial to establish standardized guidelines for collecting and storing blood-based samples to ensure accurate research outcomes [216]. According to the preanalytic processing guideline established by O’Bryant and co-workers [217], they summarized cohorts undergoing the clinical detection of AD biomarkers in blood-based samples from the collection of blood samples to the storage of plasma or serum samples. Briefly, the storage temperature of plasma samples is the most important point that prevents samples from enzymatic activity to diminish the influence of the proteolytic degradation of the protein. The long-term storage of plasma samples should be at −80 °C or in liquid nitrogen. In addition to the sample collection and storage, the selection of subjects with various conditions should also consider demographic characteristics, genomic type of ApoE, non-AD comorbidities, diet, medications, and so on. These variables can also affect the detection outcome and should be appropriately incorporated into the statistical analysis.

For the development of analytical techniques for the detection of AD biomarkers in blood-based samples, optical biosensing techniques possess great potential to detect AD biomarkers, which were discussed in this review. Nanomaterials have extremely small sizes, high specific surface areas, and unique optical properties. Therefore, combining them with optical sensors can amplify the detection signal and improve the sensitivity. It also has the advantage of improving its selectivity, stability, and system reproducibility. In nanomaterials-based optical sensors, the colorimetric method for detecting biomarkers has the advantages of easy operation and naked-eye observation. However, its sensitivity is only nM, making it challenging to apply to the detection of biomarkers in the body. In the future, combining with isothermal signal amplification technology to improve sensitivity will be a potential strategy to be developed into a point-of-care testing reagent. The systems of SPR and LSPR possess relatively high sensitivity, which can reach a femto or atto-scale in the detection of protein or microRNA, respectively. The LSPR-based technique provided an outstanding ability with a very high sensitivity for genotype detection. The detection limit for AD biomarkers can reach the pico-scale level in fluorometry. However, the most common sample treatment method for biomarker detection involves dilution. Therefore, employing isothermal signal amplification techniques based on nucleic acid branch migration in combination with nanomaterial-based fluorometry can enhance sensitivity, particularly when utilizing an aptamer-based approach. For SERS-based technique, a higher sensitivity can also be obtained down to the level of femto and atto-scale. However, the detection signal of SERS is acquired from the molecular vibration of the target. It is inevitable to obtain other signals from the background, especially when detecting targets in a complex matrix. Therefore, using antibodies and aptamers is an effective way to improve the selectivity and reduce the interference from sample matrices.

Based on the stability, reproducibility and unique optical properties of fabricated nanomaterials, their integration into optical sensors significantly enhances detection performance, particularly for low-concentration AD biomarkers in pico to femto scales. On the other hand, multiple assessments of AD biomarkers can better reflect the correlation between biomarkers and AD progression than relying on a single biomarker. As a result, the development of a multiplex biomarker detection platform can also broaden the utilization of analytical technology in clinical research and testing.