How can surface-enhanced Raman spectroscopy improve diagnostics for bacterial infections?

ABSTRACT

Currently, bacterial infection is still a major global health issue. Although antibiotics have been widely used to control and treat bacterial infections, the overuse and misuse of antibiotics have led to widespread antimicrobial resistance among many bacterial pathogens. Therefore, reducing bacterial infections through rapid and accurate diagnostics is crucial for global public health. Traditional microbiological detection methods have limitations such as poor selectivity, high complexity, and excessive time consumption, highlighting the urgent need to develop efficient and sensitive bacterial diagnosis methods. Surface-enhanced Raman spectroscopy (SERS), as an emerging technique in clinical settings, holds a promising future for bacterial identification due to its rapid, nondestructive, and cost-effective nature. This invited special report discusses the application of SERS technology in bacterial diagnosis using pure culture, clinical samples, and single-cell Raman analysis. Current challenges and prospects of the technology are also addressed with in-depth discussion.

Microbial infection is a major global health issue, causing an estimated 13.7 million infection-related deaths in a single year, of which 7.7 million deaths have been associated with bacterial pathogens such as Staphylococcus aureus (S. aureus), Salmonella enterica (S. enterica), and Klebsiella pneumoniae (K. pneumoniae) [1]. Although antibiotics have been widely applied to control and treat bacterial infection, their overuse and misuse have generated extensive antimicrobial resistance (AMR) in many bacterial pathogens [2]. Therefore, reducing bacterial infection through rapid and accurate diagnosis is an urgent health priority for both local and global public health [3].

1. Conventional methods for bacterial infection diagnosis

So far, many methods have been developed to identify bacterial pathogens causing infectious diseases. Bacterial culture, one of the most conventional methods for bacterial infection diagnosis, is still considered the gold standard and is frequently used to compare other microbiological diagnosis methods [4]. It is worth noting that the cultural method suffers from limitations such as the incapacity of unculturable bacterial identification, high turn-around time, low sensitivity, subjectivity, etc [4]. Staining is another traditional method for bacterial infection diagnosis, which is simple, cheap, and easy to operate. The most commonly used bacterial staining method is Gram-staining, which can assess the quality of clinical specimens and provide early diagnosis of bacterial infection [5]. In addition, Gram-staining generally divides bacterial species into two major groups, Gram-positive and Gram-negative, which can help clinicians avoid unnecessary broad-spectrum antibiotics, hence reducing antibiotic overuse. However, inaccurate staining occurs frequently, and some bacteria, including Mycoplasma, Chlamydia, and Rickettsia, are not suitable for Gram staining [6].

Due to their rapidity and simplicity, many biochemical reactions have also been applied to identify bacterial pathogens in infectious diseases through the identification of specific proteins, fats, carbohydrates, or enzymes. For example, the identification of key enzymes in bacterial pathogens can classify bacterial species and understand bacterial pathogenesis, which further guides the eradication of bacterial pathogens [7]. In particular, catalase can decompose hydrogen peroxide, identification of which through chemical reaction can be used to discriminate catalase-positive and catalase-negative bacterial pathogens [8], while urease is a widespread enzyme in bacterial pathogens for several medical conditions, e.g., peptic ulcer caused by Helicobacter pylori [9]. However, biochemical reactions face limitations such as strain-level phenotypic variabilities, difficulties in complex reaction patterns, long turnaround times (TAT), and lack of expertise in result interpretation. These factors make them less reliable and accurate than other molecular-level bacterial identification methods.

2. Molecular methods for bacterial infection diagnosis

In addition to the above-mentioned conventional methods, various immunoassays for the detection of bacterial antigens and antibodies have been developed for the diagnosis of bacterial infections. Bacterial antigen tests (BAT) are easy and cheap to conduct. However, many laboratories limit their use due to the false positives, low sensitivities from cross-reactivity with other bacteria, and their inability to accurately indicate a viable infection [4,10]. The bacterial antibody test faces similar issues, including the inability to detect antibiotic resistance and frequent cross-reactions caused by proteins shared among various bacterial species [4]. In addition, the antibody tests cannot distinguish current infections from past infections, limiting the application of the technique for the prevention and control of bacterial infections [11].

Different from bacterial antibody or antigen tests, mass spectrometry (MS) is a powerful tool for bacterial identification and antibiotic resistance detection, which is now a routine technique in most clinical microbiology laboratories [12]. Although MS is rapid and accurate, if a bacterial species is not listed in the MS database, the bacterium cannot be identified. In addition, for certain very closely related bacterial species in the genera of Shigella and Escherichia, the differentiation is rather difficult [13,14].

Nucleic acid amplification tests (NAATs) are a group of techniques, e.g., quantitative PCR (qPCR) that have been widely used in the diagnosis of bacterial infections due to their high rapidity, sensitivity, specificity, and multiplexity [15]. Despite these merits, the technique has several limitations. It can be over-sensitive and produce false-positive from trace contamination of the targeted specimens. In addition, sequence variations may cause for certain bacterial infections if specific sequences targeted by a particular NAAT are overlooked [16].

Currently, DNA sequencing techniques, such as 16S rRNA sequencing and whole genome sequencing, are reliable methods for identifying bacteria and are often used in clinical laboratories. However, sequencing also has its limitations for bacterial detection. For example, sequencing cannot distinguish closely related species just based on 16S rRNA, it is difficult to detect low-abundance bacteria, the sequencing cost is comparatively high, and it has the potential for misinterpretation due to sequencing errors [17]. In addition, challenges exist in sample preparation, nucleic acid contamination, data analysis, visualization, and explanation. In particular, when using short-read sequencing methods, the current analytical pipeline cannot accurately assemble repetitive regions of bacterial genomes, leading to genomic gaps and causing incomplete or inaccurate identification [18].

Additionally, vibrational spectroscopy techniques utilize the characteristic vibrations of molecules in biological samples as detectable features to obtain detailed chemical information about the target sample [19]. The Fourier transform infrared (FTIR) spectra of bacteria exhibit highly specific fingerprint patterns, making FTIR a cost-effective alternative detection technique capable of identifying bacteria at the species and subspecies levels [20]. However, the spatial resolution of conventional vibrational spectroscopy is constrained by the optical diffraction limit, restricting its application to the analysis of particles in the size range of 50–500 μm [21]. Raman spectroscopy presents a promising alternative, enabling high spatial resolution analysis of single particles [22]. Nevertheless, Raman signals from samples are typically weak and can only be effectively enhanced when the sample molecules are adsorbed onto a metal surface with nanoscale roughness [23].

3. Surface-enhanced raman spectroscopy (SERS) for bacterial infection diagnosis

SERS is considered a promising technique, benefiting from the localized surface plasmon resonance of plasmonic metal nanoparticles, making it an ideal alternative for rapid, sensitive, and nondestructive bacterial diagnosis [14,24,25]. Currently, there are two main methods for bacterial detection using SERS sensors: label-free and label-based strategies [24].

The label-free method primarily involves mixing noble metal nanoparticles with bacteria or forming nanoparticle films on substrates for bacterial attachment [26]. The prepared SERS substrates can be classified into single-metal substrates and nanocomposite substrates. For example, in-situ synthesis of silver nanoparticles (AgNPs) on bacterial cell membranes has been used to detect bacteria such as K. pneumoniae [27,28], E. coli [29], and S. aureus [30], allowing for simple and rapid collection of SERS signals. Alternatively, double-layer or multilayer metal nanoparticles can be used as SERS substrates to achieve higher signal enhancement. For instance, the self-assembly of transition metal carbides/nitrides MXene with gold nanoparticles (AuNPs) into MXene-Au composites, which possess good hydrophilicity and biocompatibility, has been employed for the detection of E. coli and Bacillus subtilis [31]. However, the label-free SERS method may result in weak target signals and difficulty in obtaining highly reproducible spectral signals due to its non-targeted bacterial recognition.

Label-based bacterial detection utilizes optical nanoprobes, which combine noble metal nanoparticles with specific organic signaling molecules. SERS tags bind to bacteria through specific recognition units, allowing indirect bacterial detection by measuring the Raman signals of the Raman reporter molecules. This method offers higher sensitivity and reproducibility [32]. Conventional SERS probes capture bacteria by directly adsorbing Raman reporter molecules onto the surface of nanoparticles (NPs), but their separation from the SERS substrate is easily triggered by the instability of complex environments [33]. The issue can be effectively alleviated by modifying the surface of the label with a “protective shell” (such as polymers, liposomes, or SiO2nanolayers), although this may weaken the SERS signal. Some studies have chosen Au nano-bridged nanogap particles (Au NNPs), which consist of Au core-Au shell nanoparticles with an internal nanogap, as novel SERS tags. By embedding Raman reporter molecules in the internal nanogap, high and stable SERS signals can be generated [33]. In a reported study, Concanavalin A-conjugated Fe₃O₄@SiO₂ NPs (ConA-Fe₃O₄@SiO₂ NPs)/bacteria/aptamer were used to modify Au NNPs, enabling the specific differentiation of various pathogenic bacteria species and allowing for the ultra-low detection of bacteria in serum samples [34]. Compared to label-free SERS, the label-based strategy provides more unique and sensitive signals, but its preparation and detection processes are more complex. Overall, SERS is an effective bacterial detection technology, capable of directly obtaining bacterial molecular fingerprints and performing qualitative and quantitative analysis of samples.

4. Promising application of the SERS technique in clinical laboratories

Detection and identification of bacterial pathogens from real-world clinical samples, while avoiding the labor-intensive culture process, can effectively save diagnostic time and promptly provide a guideline for the treatment of infectious diseases [35,36].

Blood is a widely analyzed body fluid that contains abundant sample information and closely reflects the physiological state of the human body [26]. For patients with bloodstream infections, rapid identification of pathogens and determination of antibiotic sensitivity are crucial. Traditional diagnostic methods take 2–3 days, which is too long for effective disease management [37]. In a study, a three-dimensional Alternative Current Electrokinetic/SERS (3D-ACEK/SERS) system was designed for concentrating bacteria in the blood. This system can identify bacterial species with antibiotic sensitivity within minutes, with a sensitivity five orders of magnitude lower than the current detection limit [38].

Urine, as another common body fluid, has garnered attention due to its ease of collection and noninvasive nature. Urinary tract infection (UTI) is a prevalent public health issue, and regular and rapid monitoring of the pathogens causing UTI can enhance the treatment effectiveness [39]. Considering the outstanding colorimetric, Raman, and photothermal signal modes of gold nanoflowers in gold-based nanomaterials, researchers have developed a multimodal capture antibody-independent lateral flow immunoassay (MCI-LFIA) method for the rapid diagnosis of bacterial UTI. This method uses p-mercaptophenylboronic acid-modified gold nanoflower (AuNF-PMBA) as multifunctional labels, which exhibit outstanding bacteria capture ability [40], thereby enhancing the flexibility and accuracy of SERS technology in point-of-care testing (POCT) [41]. The AuNF-PMBA-based MCI-LFIA can detect E. coli-positive samples in human urine with high accuracy within 45 minutes [40].

Apart from bodily fluid samples, wound infections caused by bacterial invasion also pose a significant threat to life safety. Some functionalized platforms, such as phenylboronic acid, mannose, and cyclodextrin, have been employed for the capture or release control of pathogens [42,43]. However, the insufficient contact between rigid platforms and wounds leads to low pathogen capture efficiency. It is necessary to design soft materials to construct biological interfaces for pathogen capture. Graphene possesses excellent biocompatibility and an outstanding ability to alter SERS signals [44]. Based on this, a smart tape with “three-in-one” characteristics was designed. Dense gold nanostars were incorporated between two graphene sheets as the SERS substrate and modified with synthetic nitrophenyl derivative molecules to capture pathogens through electrostatic interactions [45]. These detection methods significantly reduce the time required for identification and have the potential to serve as POCT devices for bacterial infection detection, providing effective guidance for clinical diagnosis and treatment.

5. The potential of single-cell Raman monitoring technique

Currently, microbial cell analysis mostly relies on population-average measurements, which obscure the characteristics of subpopulations within the community [46]. Single-cell Raman spectroscopy (SCRS) overcomes the limitations of low cell counts and matrix complexity in biological samples, providing spatially resolved chemical information at the single-cell level for bacteria [47,48]. It eliminates the need for tedious sample pretreatment and microbial culturing processes, provides results within minutes [49], and offers opportunities to monitor biochemical and biophysical changes in whole cells [50]. Combining nanomaterials with single-cell analysis enables highly sensitive detection at the single-cell level [51]. A study utilized the unique optical properties and macromolecule adsorption capabilities of MoS₂, combined with gold nanostars, to prepare MoS2@AuNSs as SERS substrates and photothermal tags for the in-situ diagnosis and eradication of pathogens at the single-cell level in biological samples [52]. This material shows potential applications in SERS sensing and photothermal therapy. Biological nanoparticles can achieve clear structures with high affinity and biocompatibility by binding to bacterial surface components, providing attractive materials for targeted delivery, infection treatment, and biosensing [53].

For example, polyaniline-containing galactose particles can bind with AuNPs to form Au@PGlyco NPs, which not only effectively enhance SERS signals but also respond to galactose-related enzymes and cell receptors [54]. Researchers found that by incubating Au@PGlyco NPs with E. coli, β-galactoside released by E. coli can react with Au@PGlyco NPs to enhance the change rate of the SERS signal. This study leverages galactose-responsive SERS to monitor single-cell detection insights, demonstrating that the SERS sensing platform holds promise for exploring intracellular mechanisms of exogenous biomolecule interactions with biological reactions [54]. Furthermore, the sensitive and specific detection of individual cells in complex liquid samples and the in-depth understanding of these cells is an important challenge for SERS in advancing precision medicine. SCRS has demonstrated its potential in specifically detecting pathogens in complex liquid samples. A bioprinter capable of converting samples into millions of droplets has been used for high-throughput bacterial identification in blood. When combined with gold nanorods, it enables the separation, counting, and identification of various cell lines [55]. The results indicate that this technology can accurately distinguish between bacteria and blood cells in droplets containing only a few Staphylococcus epidermidis, E. coli, and red blood cells within 15 seconds, effectively identifying the cell type of each spectrum and predicting the cell composition of droplets [55].

6. Current challenges and future perspectives

SERS is rapidly evolving from current biological analysis to promising clinical applications. Compared to traditional detection methods, SERS offers advantages such as low cost, speed, efficiency, and non-destructiveness. However, there are still challenges in achieving clinical translation of the SERS technique.

Firstly, interference from non-target analytes or signals from similar molecules, such as creatinine, uric acid, and bilirubin in serum, or strong background signals from other fluids, can complicate the analysis. An ideal SERS substrate should have the capability to interact with or adsorb bacteria, enabling the enrichment or concentration of target bacteria from low-abundance fluids. Therefore, the preparation of highly active, homogeneous SERS substrates with low background signal intensity is a major direction for future development. Secondly, bacteria undergo different growth stages and metabolic states during their growth process, making it challenging to select a bacterium for analysis without being affected by its growth stage. SERS substrates with ultra-low detection limits or specific capture capabilities can effectively avoid false positive results caused by slight metabolic changes in individual bacteria. Additionally, although SERS has been used for clinical sample analysis, it has not yet been widely adopted in POCT settings. This is primarily due to the lack of portable equipment and stable detection methods. In recent years, some portable Raman spectrometers have emerged, but their laser power and tunability are still inferior to large benchtop devices. Moreover, qualified clinical bacterial detection equipment such as MALDI-TOF or nucleic acid amplification devices have standardized reference libraries for bacterial identification. However, most current studies on SERS signal band identification and assignment still rely on tentative attributions based on previous research, with the specific measured substances remaining unclear. Providing users in clinical laboratories with comprehensive bacterial SERS characteristics and distributions necessitates the construction of a large standardized SERS database. Establishing a comprehensive Raman bacterial library with different substrates, strains, and stages will promote the significant role of SERS in microbiology.

Besides the aforementioned points, there are two additional aspects to consider for the transformation and application of SERS in clinical laboratories. First of all, the biotoxicity and biocompatibility of SERS tags used for wound dressing detection or internal injection should be considered. For example, gold and silica approved by Federal Drug Administrations are good methods for bacterial identification and disease diagnosis [56]. Secondly, the large amount of data generated by SERS typically relies on manual identification, which can only provide semi-quantitative results and is inefficient. Standardized machine learning (ML) based spectral preprocessing and recognition methods shall be developed, as well as to collect and establish a global public database for SERS, ensuring support for the development of robust ML models. Enhancing the interpretability and transparency of ML algorithms will enable healthcare professionals to understand and trust these tools [57].

7. Conclusion

Due to the high sensitivity and in-situ remote sensing capabilities, surface-enhanced Raman spectroscopy has become a promising method for both label-based and label-free diagnosis of bacterial pathogens in clinical laboratories in recent years. With the accumulation of substantial fundamental research and technological innovations, the gap between the knowledge gained from biomedical research and the development of devices that improve patient health has been shortened. Continuous research and development efforts aim to overcome existing challenges, advancing this technology to tackle unmet medical needs, including bacterial pathogen diagnosis and real-time monitoring of body fluids for infectious diseases. This will facilitate the creation of novel commercial SERS POCT systems, providing faster and better diagnostic technologies for individual and public health security.

Funding Statement

The work was supported by the Research Foundation for Advanced Talents of Guangdong Provincial People’s Hospital [KY012023293]; Guangdong Basic and Applied Basic Research Foundation [2022A1515220023].

Article highlights

• SERS is a low-cost and nondestructive alternative to bacterial identification.

• Label-based SERS can achieve specific detection of bacterial pathogens.

• A standardized bacterial SERS database is needed to advance its applications.

• Integration of machine learning with SERS is essential for bacterial identification.

Author contributions

JWT has written the original draft; XRW has done review & editing; YWL has conceptualized and reviewed & edited; and LW has conceptualized, supervised, and reviewed & edited.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.