Tuberculosis (TB), caused by the pathogenic bacteria Mycobacterium tuberculosis (Mtb), continues to be a leading public health threat that affects all countries and age groups (Zumla et al., 2013). In 2020, approximately 10 million people developed TB, and 1.3 million people died from the infection worldwide. Of increasing concern, a total of 157,903 drug-resistant cases were reported, with 132,222 cases of multidrug- or rifampicin-resistant TB and 25,681 cases of pre-extensively drug-resistant TB or extensively drug-resistant TB detected, although this was a large fall (of 22%) from the total of 201,997 people detected with drug-resistant TB in 2019 (WHO, 2021). Tuberculosis control strategies aim to reduce the spread of the infection and cure infectious TB patients, who need rapid and accurate diagnosis to facilitate the administration of prompt anti-TB treatment, thereby helping to reduce the transmission of TB and development of drug resistance.

Mtb is an intracellular pathogen that preferentially infects host macrophages and primarily resides within lung granulomas (Brites and Gagneux, 2015; Elkington et al., 2021), making directly detecting Mtb in human body fluids, such as blood, saliva, or urine, impossible. A current challenge in TB diagnosis is the development of rapid point-of-care tests. Sputum smear microscopy is the most common way of diagnosing TB worldwide (Steingart et al., 2006; Molicotti et al., 2014); however, the sensitivity of sputum smear microscopy ranges from 30 to 60% and is largely dependent on the operator and the abundance of Mtb in the samples (Molicotti et al., 2014). Microbiological culture is considered a diagnostic gold standard, but it requires several weeks and laboratory containment facilities to culture and identify Mtb in samples due to the slow growth rate and biohazard level of the microbe (Lagier et al., 2015). The rapid molecular-based diagnostic test Xpert MTB/RIF has high sensitivity and specificity for the diagnosis of smear-positive sputum TB patients (Steingart et al., 2013); however, Xpert MTB/RIF has a lower sensitivity for smear-negative sputum samples, leaving the diagnosis of a significant proportion of TB-infected cases reliant on diagnostic tests with sub-optimal accuracy.

Recently, many researchers have focused on the discovery of host biomarkers for TB diagnostics. Host immune responses to Mtb infection (Dey and Bishai, 2014) leave traceable signals within the host that may prove valuable for the accurate diagnosis and/or prognosis of TB (Parida and Kaufmann, 2010). Increasingly, investigators are validating the feasibility and accuracy of using proteomic signatures for TB diagnosis and prognosis (De Groote et al., 2017b; Penn-Nicholson et al., 2019). For example, our team previously screened for and validated key proteomic TB biomarkers using an antibody-based array for whole-blood samples that were stimulated with pooled Mtb peptides (ESAT-6 and CFP-10 derived peptides) (Chen et al., 2009) or mitogen, and we successfully identified an eight-protein bio-signature of I-TAC, I-309, MIG, granulysin, FAP, MEP1B, furin, and LYVE-1. The combination of the eight biomarkers allowed us to distinguish TB from healthy control individuals in a test cohort with a specificity and sensitivity of 83% (95% CI, 71–91%) and 76% (95% CI, 56–90%), respectively (Yang et al., 2020). In this review, we summarize the proteomics research approaches followed over the past several years, focusing on the progress of TB biomarker discovery via proteomics, with the hope of eliciting more promising biomarker discoveries for rapid and accurate TB diagnosis.

Proteomics is the systematic study of the proteome with the aim of uncovering the nature of sophisticated protein-interaction networks with respect to protein expression, structure, function, and control of biological processes within an organism (Patterson and Aebersold, 2003). By comparing different patterns within the proteome, proteomics has continued to provide a powerful method for studying the changes in protein diversity that accompany health and disease processes, making the clinical diagnosis, prognosis, and even treatment of different diseases possible (Kavallaris and Marshall, 2005).

It is the development of technology and informatics that make a new concept achievable, as is the case for proteomics. Figure 1 illustrates the timeline of proteomic technology and protein database development that has promoted proteomics research. However, the proteome of a particular cell type or tissue is a mixture of all the proteins expressed (Wilkins, 2009); thus, protein separation is a prerequisite for single-protein analysis. As early as 1975, two-dimensional polyacrylamide gel electrophoresis (2D-PAGE), which separates proteins based on their size and surface charge, was first employed to separate ribosomal proteins of Escherichia coli (Knopf et al., 1975). Mass spectrometry (MS), which is usually applied to the identification of individual protein spots separated by 2D-PAGE, was first described in 1899. It was not, however, until the development of non-destructive large-biomolecule ionization methods, including matrix-assisted laser desorption/ionization (MALDI) (Karas and Hillenkamp, 1988) and electrospray ionization (Fenn et al., 1989), that the widespread application of MS became feasible. Peptide mass fingerprinting (Henzel et al., 2003) and isotope-coded affinity tag (Gygi et al., 1999) approaches were then combined with different MS strategies for the accurate quantification of concurrent peptides from proteins that are found abundantly or even in trace amounts in complex mixtures.

Benefiting from the rapid development of technology during the 21st century, some classical technologies have helped to further revolutionize proteomics. For example, 2D-PAGE and MS have been developed into techniques such as difference-gel 2D-electrophoresis (Larbi and Jefferies, 2009), surface-enhanced laser desorption/ionization time of flight (SELDI-TOF) MS (Marcos et al., 2013), quantitative chemical cross-linking with MS (Wippel et al., 2021), and immunodepletion techniques for “low-abundance” protein determination (Liu et al., 2021b). Moreover, “protein chips” have been designed and combined with MS approaches for the specific analysis of selected proteins (Zhou et al., 2001). It is important to consider that, coupled with proteomic technologies, proteomics databases, including Swiss-Prot, Entrez, and the Human Proteome Organization, also continue to contribute to the development of proteomics (Kavallaris and Marshall, 2005). Moreover, in addition to other single-cell approaches (Mauger et al., 2021), single-cell proteomics is emerging as a way to identify and quantify the proteins in individual cells (Li et al., 2021). Although there are no high-throughput proteomic “sequencers” available as yet, it is believed that this technology will contribute to the discovery of new aspects of protein and cell biology in the future.

Upon Mtb infection, the host cells’ response is to produce and secrete certain effectors to deal with the invading bacteria (Dey and Bishai, 2014; Brites and Gagneux, 2015; Llibre et al., 2021). Blood is the primary vehicle for the transport of either host effectors or Mtb factors and is usually routinely sampled for clinical testing, making blood samples readily accessible for research purposes and diagnostic tests. The recent advances in proteomics make the simultaneous detection of thousands of proteins possible, facilitating the capture of effectors and thus paving the way for alternative and efficient TB biomarker discovery. In this section, we summarize the recent developments in the identification of proteomic biomarkers that can be used for differentiating TB from healthy or other disease status individuals. This has been achieved by categorizing the origin of the biomarkers from either blood samples, urine samples, or other types of human body fluids.

Intensive investigations into Mtb–host interactions have broadened and deepened our knowledge on the strategies applied by Mtb to infect hosts and the host responses upon infection. The development of technologies, especially MS, has facilitated proteomic biomarker identification with high efficiency, but there are still some technical challenges slowing down the application of proteomics to the diagnosis of active TB. Sampling complexities (e.g., the timing of sample collection, small-molecule interference, and the coexistence of TB with other types of disease), differences in protein abundances, the presence of isoforms, and post-translational modifications are barriers preventing the identification of accurate and universal biomarkers with current technologies and strategies. More powerful techniques with improved sensitivity will be required, especially for low-abundance proteins and for differentiating protein isoforms and modifications. In addition, there is a risk of diagnostic bias when using a single protein biomarker, and a group of biomarkers within a panel could be detected simultaneously using current technologies, but discrepancies among different reports are increasing, providing confusing or even misleading evidence for the optimum TB diagnostic markers. One reason for the unsatisfactory reproducibility of the discovered biomarkers is that the studied cohorts have differed with regard to their genomic backgrounds, immune responses, or other factors. Large cohorts of samples from various populations that include a diversity of TB patient statuses may be needed to overcome this problem. Moreover, a combination of different “omics” techniques may be useful for improving TB diagnostic accuracy and consistency, and single-cell proteomics holds particular promise as a method that will advance active TB diagnosis in the future (Petelski et al., 2021).

JG and YC contributed to the conceptualization and design, reviewed the literature, prepared the figures and tables, and wrote the original draft of the manuscript. XZ contributed to literature review and data preparation. XC and YC supervised the data and wrote, reviewed, and edited the manuscript. JG, XC, and YC contributed to the project administration, funding acquisition, and resources. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.