Unraveling the Role of MicroRNAs in Mycobacterium tuberculosis Infection and Disease: Advances and Pitfalls

Tuberculosis (TB) is an infectious disease of extremely high epidemiological burden worldwide that is easily acquired through the inhalation of infected respiratory droplets. The complex pathogenesis of this infection spans from subjects never developing this disease despite intense exposure, to others in which immune containment fails catastrophically and severe or disseminated forms of disease ensue.

ABSTRACT

Tuberculosis (TB) is an infectious disease of extremely high epidemiological burden worldwide that is easily acquired through the inhalation of infected respiratory droplets. The complex pathogenesis of this infection spans from subjects never developing this disease despite intense exposure, to others in which immune containment fails catastrophically and severe or disseminated forms of disease ensue. In recent decades, microRNAs (miRNAs) have gained increasing attention due to their role as gene silencers and because of their altered expression in diverse human diseases, including some infections. Recent research regarding miRNAs and TB has revealed that the expression profile for particular miRNAs clearly changes upon Mycobacterium tuberculosis infection and also varies in the different stages of this disease. However, despite the growing number of studies—some of which have even proposed some miRNAs as potential biomarkers—methodological variations and key differences in relevant factors, such as sex and age, cell type analyzed, M. tuberculosis strain, and antimicrobial therapy status, strongly hinder the comparison of data. In this review, we summarize and discuss the literature and highlight the role of selected miRNAs that have specifically and more consistently been associated with M. tuberculosis infection, together with a discussion of the possible gene and immune regulation pathways involved.

INTRODUCTION

Tuberculosis (TB), an infectious airborne disease caused by Mycobacterium tuberculosis, remains the leading cause of death among infectious diseases from a single agent as well as the ninth leading cause of death worldwide. In 2018, around 10 million people developed TB, and 1.2 million died as a result of this disease, exceeding the combined number of deaths from HIV and malaria during the same period (1). According to the World Health Organization (WHO), about a quarter of the global population is infected with this bacterium and about 5% to 15% will develop the disease in their lifetime, with an increasing risk in subjects with HIV infection or other conditions, such as undernourishment, diabetes, smoking, and heavy alcohol consumption (2).

M. tuberculosis infection is commonly acquired through inhalation of infected respiratory droplets that reach the alveolar space, where mycobacteria undergo phagocytosis by the alveolar macrophages (3–5). Once inside the macrophages, M. tuberculosis displays several mechanisms that avoid or counterfeit the immune response of the host, such as modulating the endosomal/phagosomal maturation pathway, blocking the fusion of the mycobacterial phagosome with the lysosome, and/or avoiding the bactericidal mechanisms of the host cells (3–5).

Currently, the diagnosis of active pulmonary TB (PTB) requires direct M. tuberculosis detection (microbiological or DNA amplification) and, therefore, is highly dependent on the availability of clinical samples, such as sputum, for analysis. Conversely, none of the currently available non-sputum-based tests to diagnose M. tuberculosis infection, such as the tuberculin skin test or immunoassays based on interferon gamma (IFN-γ) release, are able to discriminate between latent TB infection (LTBI) and active TB disease (6). Therefore, in order to improve early diagnosis and control transmission, there has been an intense global effort in the scientific community searching for new biomarkers that might distinguish between the different states of this infection that range from asymptomatic and latent to progressive and destructive active disease.

MicroRNAs (miRNAs) are small, noncoding, single-stranded RNA molecules that bind to specific mRNA to inhibit the translation of multiple genes into proteins (7). Compelling evidence shows that they are involved in a number of normal cellular processes, such as cell cycle control, apoptosis, and several developmental and physiological processes (8–11), and that their altered expression is associated with the pathogenesis of diverse diseases, including cancer and autoimmune conditions (12–16).

In recent years, it has also become evident that miRNAs have a role in the control of several infections (17, 18), mediating antiviral defenses (in plants and animals) (19, 20) and regulating the host cellular environment (21) as well as the immune response (22–24), with a critical role in adaptive immunity (25–28). Particularly, miRNAs can mediate the host response to M. tuberculosis infection mostly through the regulation of proteins involved in the innate and adaptive immune pathways (29).

In this review, recently discovered miRNAs that are related to M. tuberculosis infection are summarized, and we shed light on how these small molecules might be involved in TB pathogenesis and immune response.

miRNAs

miRNAs were first described as small noncoding RNA (ncRNA) with no biological role. However, in 1993, Ambros and Ruvkun discovered quite the opposite, when they found that lin-4 was essential for the normal temporal control of diverse postembryonic developmental events in the nematode Caenorhabditis elegans by negatively regulating the LIN-14 protein level (30, 31). This remarkable finding opened the door to the exploration of a novel cellular regulatory mechanism involving non-protein-coding transcripts and led, a decade later, to the discovery of the let-7 miRNA (32), which was found to be highly conserved across different species ranging from flies to humans (33). Since then, thousands of miRNAs have been widely studied, increasing our understanding of their biogenesis, target recognition, and role in gene regulation (34–38).

It is presently understood that the main biological role of miRNA—negative gene regulation—can be performed in the following three ways: mRNA degradation, translational repression, or miRNA‐mediated mRNA decay. These mechanisms depend on the level of complementarity between the miRNA and the mRNA target. Translation repression and miRNA‐mediated mRNA decay require a limited microRNA target base pairing, while mRNA cleavage requires a more extensive microRNA target base pairing (39).

Currently, the study of miRNA has been accelerated due to new available technologies, such as next-generation sequencing (NGS), specifically, RNA transcriptome sequencing (RNA-seq), allowing researchers to identify and determine the relative expression levels of all miRNA (closely related miRNA and isoforms) within a high number of samples (40). MicroRNA array, conversely, allows the study of multiple miRNA (41), while quantitative real-time PCR (qPCR) and multiplex miRNA profiling allow the validation of several miRNAs across a number of samples with high sensitivity (42, 43). Some of these technologies are considered supplementary to each other in the process of validation of findings, in which using an independent method helps to avoid misleading signals (e.g., RNA-seq can introduce bias through multiple PCR amplification and ligation steps during library preparation) (44).

Presently, more than 2,600 mature miRNA sequences have been identified within the human genome (http://www.mirbase.org/, v.22.1) (45–47), with more than 15,000 target genes and over 380,000 miRNA-target interactions (MTIs) identified and validated by Western blotting, luciferase assay, pulsed stable isotope labeling by/with amino acids in cell culture (pSILAC), microarray, and/or NGS (miRTarBase, v.7.0) (48).

miRNA EXPRESSION PROFILE UPON MYCOBACTERIUM TUBERCULOSIS INFECTION

Individuals with M. tuberculosis infection are usually classified binarily as having LTBI or active TB. However, a variety of stages exist between initial exposure and the development of active disease. After first exposure, some individuals may successfully eliminate the bacterium, and they would no longer carry viable bacteria but may still have immunological evidence of prior infection (49). When the immune system is unable to clear the infection but can control and contain it, the bacteria persist in a latent state of infection for years, may progress in a slow or fast manner to a primary active disease, or can cycle through incipient and subclinical states before developing into symptomatic disease or eventual disease resolution (49). Based on these facts, the fate of M. tuberculosis infection is highly dependent on individual immune response, and therefore genetic differences in the host’s immune regulatory mechanisms may influence the risk of developing clinical TB (50). Gene silencing via host miRNA has been proposed as one of the mechanisms by which human macrophages battle intracellular pathogens like M. tuberculosis (51). However, bacteria can also manipulate host cell pathways by regulating miRNA expression (52). Knowing how miRNAs modulate gene expression upon M. tuberculosis infection and identifying their target genes are critical in understanding the host response to infection. In this regard, a comparative study of three evolutionarily closely related Mycobacterium genomes (M. tuberculosis H37Rv, M. tuberculosis CDC1551, and Mycobacterium leprae TN), in which only conserved interactions between human miRNAs and putative M. tuberculosis target genes were considered, predicted 26 candidate M. tuberculosis genes that might be targeted by human miRNAs in lung and macrophages (51).

The miRNA expression profile can also be determined by phylogenetic differences in M. tuberculosis strains, as suggested by an in vitro model of THP-1 cells infected with Beijing/W versus the non-Beijing/W M. tuberculosis genotypes. Macrophages infected with the Beijing/W strain (more virulent) exhibited a repression of 13 miRNAs with respect to macrophages infected with the non-Beijing/W strain (53). Furthermore, profound differences in miRNA expression have also been described in relation to drug-resistant M. tuberculosis strains compared to drug-sensitive strains (54), a factor to be considered for study comparisons.

Interestingly, miRNA expression is also modified under the effect of successful antimicrobial therapy. Wang et al. (55) found that serum levels of miR-21-5p, miR-92a-3p, and miR-148b-3p distinguished between untreated, partially treated (for 2 months), and cured TB, in comparison with those with noninfection. Besides, these miRNAs were significantly reduced in patients with cured TB compared to those of patients with untreated TB, and miR-125a-5p levels were upregulated in the partially treated TB samples compared to those of untreated or cured TB. Also, significant increases in miR-21-5p, miR-92a-3p, and miR-148b-3p levels were found in patients with untreated TB compared to those of healthy controls, and miR-21-5p alone discriminated between cured and untreated TB (area under the receiver operating characteristic [ROC] curve [AUC] = 0.851). In a similar study, serum levels of miR-16 and miR-155 exhibited a down- and upregulation, respectively, after the completion of treatment for active TB, while no significant difference was observed when compared to levels in uninfected subjects (56). These studies suggest that miRNAs might be useful as biomarkers to distinguish between the different stages of TB infection or therapy responsiveness; however, none of them have been validated in prospective studies, and patients with other infections should be included.

PITFALLS OF miRNA PROFILING IN MYCOBACTERIUM TUBERCULOSIS INFECTION AND DISEASE

A substantial number of studies have collected an enormous amount of data about miRNA expression profiles in different stages of M. tuberculosis infection and disease. Table 1 summarizes these results, compares different samples from healthy volunteers, subjects with LTBI, and patients with active TB, and shows the number of miRNAs found to be up- or downregulated. Discrepancies among studies due to methodological variations, choice of biological sample, differing stages of TB infection, and differences in RNA extraction processes, number of miRNAs obtained, cut-off selection criteria, and data analysis process (including data normalization) introduce a potential bias that influences these results and must be considered for accurate interpretation. For example, when comparing serum samples from patients with active TB and healthy controls, a study showed 59 upregulated and 33 downregulated miRNAs (57), while another report showed 10 and 25 miRNAs, respectively (58). Regardless of these variations, several miRNAs have been consistently shown to be either upregulated or downregulated across studies (Table 2). Particularly, hsa-let-7g, hsa-miR-30c, and hsa-miR-433 showed an upregulation in serum samples (57, 59, 60), and hsa-miR-1270, hsa-miR-371-3p, hsa-miR-380*, hsa-miR-582-3p, and hsa-miR-618 showed a downregulation in sputum (61) and serum samples (57). However, some other miRNAs presented a variable regulation depending on the study or sample analyzed (Table 3). As an example, hsa-miR-1 showed an upregulation in sputum samples (61), a downregulation in peripheral blood mononuclear cells (PBMCs) (62), and both an up- and downregulation in serum samples (60, 63) of patients with TB when compared to those of samples from healthy controls.

TABLE 1.

Differential miRNA expression profiles in different stages of tuberculosis

Samplea	No. of upregulated miRNAs	No. of downregulated miRNAs	No. of validated miRNAs	Participantsa	Reference
PBMC	28	2	1	PTB vs control	Liu et al., 2011 (78)
PBMC	9	1		LTBI vs ATB	Wu et al., 2014 (62)
		1		ATB vs control
	16	6		LTBI vs control
PBMC	15	14	8	ATB vs controlb	Zhou et al., 2016 (86)
CD4+ T cells	82	53	6	ATB vs control	Fu et al., 2014 (117)
	33	46		LTBI vs control
	62	62		LTBI vs ATB
Serum	59	33	3	ATB vs control	Fu et al., 2011 (57)
Serum	90	7	7	PTB vs control	Qi et al., 2012 (59)
Serum	44	47	6	PTB vs control	Zhang et al., 2013 (63)
Serum	62	33	2	ATB vs LTBI	Zhang et al., 2014 (60)
	82	22		ATB vs BCG+
	37	65		ATB vs BCG−
	24	6		ATB vs three groups
Serum	10	25		ATB vs control	Xu et al., 2015 (58)
Differentiated macrophages derived from PBMC	6	3	11	ATB vs control	Zheng et al., 2015 (53)
	8	1		LTBI vs control
	4			LTBI vs ATB
THP-1 cells	1	13	14	Beijing/W vs non-Beijing/W M. tuberculosis strains
Sputum	43	52	3	ATB vs control	Yi et al., 2012 (61)

^aATB, active tuberculosis; PTB, pulmonary tuberculosis; LTBI, latent tuberculosis infection; PBMC, peripheral blood mononuclear cells.

^bSamples obtained from children (2 to 14 years old).

TABLE 2.

miRNAs consistently upregulated and downregulated across studies analyzing miRNA expression profile

miRNAa	Samples	References
Upregulated miRNA
hsa-let-7e	Serum/serum	Qi et al., 2012 (59)/Zhang et al., 2014 (60)
hsa-miR-127-3p
hsa-miR-145
hsa-miR-21
hsa-miR-24
hsa-miR-25
hsa-miR-26b
hsa-miR-340
hsa-miR-432
hsa-miR-20a		Qi et al., 2012 (59)/Xu et al., 2015 (58)
hsa-miR-93		Qi et al., 2012 (59)/Zhang et al., 2013 (63)
hsa-miR-191		Fu et al., 2011 (57)/Zhang et al., 2014 (60)
hsa-miR-23b
hsa-miR-146b-5p		Fu et al., 2011 (57)/Zhang et al., 2013 (63)
hsa-miR-483-5p
hsa-miR-122		Zhang et al., 2013 (63)/Zhang et al., 2014 (60)
hsa-miR-29c
hsa-miR-320c
hsa-miR-378		Zhang et al., 2013 (63)/Xu et al., 2015 (58)
hsa-miR-193a-3p		Wu et al., 2014 (62)/Zhang et al., 2014 (60)
hsa-miR-143	Serum/serum/serum	Fu et al., 2011 (57)/Zhang et al., 2014 (60)/Xu et al., 2015 (58)
hsa-let-7g		Fu et al., 2011 (57)/Qi et al., 2012 (59)/Zhang et al., 2014 (60)
hsa-miR-30c
hsa-miR-433
hsa-miR-22		Fu et al., 2011 (57)/Zhang et al., 2013 (63)/Zhang et al., 2014 (60)
hsa-miR-151-3p	Serum/sputum	Qi et al., 2012 (59)/Yi et al., 2012 (61)
hsa-miR-409-3p
hsa-miR-1204		Fu et al., 2011 (57)/Yi et al., 2012 (61)
hsa-miR-23a		Zhang et al., 2014 (60)/Yi et al., 2012 (61)
hsa-miR-548d-5p		Xu et al., 2015 (58)/Yi et al., 2012 (61)
hsa-miR-376c	Serum/serum/sputum	Qi et al., 2012 (59)/Zhang et al., 2014 (60)/Yi et al., 2012 (61)
hsa-miR-29a	Serum/serum/serum/sputum	Fu et al., 2011 (57)/Qi et al., 2012 (59)/Zhang et al., 2014 (60)/Yi et al., 2012 (61)
hsa-miR-144*	PBMC/serum	Liu et al., 2011 (78)/Qi et al., 2012 (59)
hsa-miR-660
hsa-miR-296-5p		Liu et al., 2011 (78)/Zhang et al., 2014 (60)
hsa-miR-142-3p		Wu et al., 2014 (62)/Qi et al., 2012 (59)
Downregulated miRNA
hsa-miR-1270	Serum/sputum	Fu et al., 2011 (57)/Yi et al., 2012 (61)
hsa-miR-371-3p
hsa-miR-380*
hsa-miR-582-3p
hsa-miR-618

^ahsa, Homo sapiens.

TABLE 3.

miRNAs with variable expression (either upregulation or downregulation) across studies analyzing miRNA expression profile

miRNA	Upregulation sample (reference)	Downregulation sample (reference)
hsa-miR-1284	Serum (Fu et al., 2011 [57])	Serum (Xu et al., 2015 [58])
hsa-miR-139-3p	Serum (Qi et al., 2012 [59])	Serum (Zhang et al., 2014 [60])
hsa-miR-15b
hsa-miR-330-3p
hsa-miR-886-5p
hsa-miR-889
hsa-miR-17	Serum (Qi et al., 2012 [59])	Serum (Xu et al., 2015 [58])
hsa-miR-19b
hsa-miR-28-3p
hsa-miR-345
hsa-miR-206	Serum (Zhang et al., 2014 [60])	Serum (Fu et al., 2011 [57])
hsa-miR-221*
hsa-miR-301a	Serum (Qi et al., 2012 [59])	Serum (Zhang et al., 2013 [63])
hsa-miR-320b	Serum (Zhang et al., 2014 [60])	Serum (Zhang et al., 2013 [63])
hsa-miR-320d	Serum (Zhang et al., 2014 [60])	Serum (Fu et al., 2011 [57])
hsa-let-7c	Serum (Zhang et al., 2014 [60])	Sputum (Yi et al., 2012 [61])
hsa-miR-100
hsa-miR-181a
hsa-miR-1260b	Serum (Fu et al., 2011 [57])	Sputum (Yi et al., 2012 [61])
hsa-miR-10a	Serum (Zhang et al., 2014 [60])	PBMC (Wu et al., 2014 [62])
hsa-miR-31
hsa-miR-181a-2*	Serum (Zhang et al., 2013 [63])	PBMC (Wu et al., 2014 [62])
hsa-miR-199a-5p	Serum (Zhang et al., 2014 [60])	PBMC (Liu et al., 2011 [78])
hsa-miR-335	Serum (Qi et al., 2012 [59])	PBMC (Liu et al., 2011 [78])
hsa-miR-29b	PBMC (Wu et al., 2014 [62])	Serum (Zhang et al., 2014 [60])
hsa-miR-503
hsa-miR-452	PBMC (Liu et al., 2011 [78])	Serum (Zhang et al., 2014 [60])
hsa-miR-532-5p	PBMC (Liu et al., 2011 [78])	Serum (Xu et al., 2015 [58])
hsa-miR-101	Serum (Fu et al., 2011 [57]), serum (Zhang et al., 2014 [60])	Serum (Zhang et al., 2013 [63])
hsa-miR-196b	Serum (Qi et al., 2012 [59]), serum (Zhang et al., 2014 [60])	Serum (Xu et al., 2015 [58])
hsa-miR-199a-3p
hsa-miR-30d
hsa-miR-20b	Serum (Qi et al., 2012 [59]), serum (Zhang et al., 2013 [63])	Serum (Xu et al., 2015 [58])
hsa-miR-26a	Serum (Qi et al., 2012 [59]), serum (Zhang et al., 2014 [60])	Serum (Zhang et al., 2013 [63])
hsa-miR-744	Serum (Fu et al., 2011 [57]), serum (Qi et al., 2012 [59])	Sputum (Yi et al., 2012 [61])
hsa-miR-99b	Serum (Qi et al., 2012 [59]), serum (Zhang et al., 2014 [60])	Sputum (Yi et al., 2012 [61])
hsa-let-7i	Serum (Fu et al., 2011 [57]), serum (Zhang et al., 2014 [60])	PBMC (Zhou et al., 2016 [86])
hsa-miR-125b	Serum (Fu et al., 2011 [57]), serum (Zhang et al., 2014 [60])	PBMC (Wu et al., 2014 [62])
hsa-miR-146a	Serum (Fu et al., 2011 [57]), serum (Qi et al., 2012 [59])	PBMC (Wu et al., 2014 [62])
hsa-miR-103	Serum (Fu et al., 2011 [57]), serum (Qi et al., 2012 [59]), serum (Zhang et al., 2014 [60])	Serum (Xu et al., 2015 [58])
hsa-miR-151-5p	Serum (Zhang et al., 2014 [60]), sputum (Yi et al., 2012 [61])	Serum (Xu et al., 2015 [58])
hsa-miR-1	Serum (Zhang et al., 2014 [60]), sputum (Yi et al., 2012 [61])	PBMC (Wu et al., 2014 [62]), serum (Zhang et al., 2013 [63])
hsa-let-7b	Serum (Zhang et al., 2014 [60])	Serum (Xu et al., 2015 [58]), sputum (Yi et al., 2012 [61])

Another very common problem encountered across miRNA TB biomarker studies including human samples is that the varying stages of TB infection are not always well defined, and recent infection, long-standing LTBI, newly diagnosed active TB, or treated TB are not clearly differentiated, hindering the group comparisons. In addition, very often, there is no differential analysis between males and females, even though this factor has been reported to define a different profile expression for several miRNAs (64).

Importantly, potential pitfalls can also be present in the current platforms used for miRNA expression profiling. The search for miRNAs as possible biomarkers usually involves studies of NGS or microRNA array, due to their advantages in allowing the identification and analysis of multiple targets (40, 41). However, both techniques may have disadvantages. NGS can produce misleading results by introducing bias during the multiple PCR amplification, where the expression levels of some RNAs can be artificially enhanced, diminished, or even undetectable. Also, during the ligation steps, the use of different adapters and barcodes influences cDNA synthesis efficacy, and the variable-specific RNA G+C content can be associated with unequal PCR amplification efficiency (44). Regarding microRNA arrays, this platform permits the study of a large number of miRNAs already identified; however, data comparison is hindered by the divergence in the stringency of the detection call criteria and the poor inter-platform concordance of microRNA expression values (65). Another step commonly involved in miRNA studies involves the validation of NGS or microarray findings by qPCR amplification. This technique allows miRNA profiling and validation with high sensitivity (42). However, qPCR data normalization remains one of the most challenging steps due to the lack of a reliable normalizer for miRNA quantification, hindering the accurate data normalization strategy (66, 67). Currently, small RNAs, in particular RNU6B, are widely used as a normalizer for miRNAs; however, it has been shown that it can experience changes in serum samples in a disease-specific manner. Therefore, it may not be a good normalizer for inflammatory conditions (68). In addition, small RNAs do not share the same biological and biochemical properties of miRNA molecules in terms of their transcription, processing, and tissue-specific expression patterns (68, 69). Therefore, it would be most suitable to use as a qPCR reference other miRNAs exhibiting stable expression under the same experimental conditions (70). Alternatively, another more rigorous strategy for selection of a good qPCR normalizer gene proposes the use of a global mean normalization of a set of reference genes (71, 72). According to this method, the geometric mean of a minimum of three stable housekeeping genes provided a more reliable normalization factor that can control for outliers and differences in abundance between genes (73).

ROLE OF VALIDATED miRNA IN DIFFERENT STAGES OF DISEASE

Active TB is believed to reflect a failure of the immune response restraining M. tuberculosis, with uncontrolled replication leading to progressive development of TB pathology and clinical manifestations. In this context, several investigators have looked for changes in molecular elements that might be related to TB infection development. When looking at miRNAs, a variable number of circulating or intracellular miRNAs have been found and described accordingly in the different stages of M. tuberculosis infection. Of these, only a small proportion has been further validated after discovery (Table 4). As an example, among validated miRNA, hsa-miR-196b and hsa-miR-376c exhibited a clear upregulation in serum from patients with active TB compared with that of the control group (LTBI and uninfected controls) (60).

TABLE 4.

miRNAs involved in different disease stages validated^a in different samples and studies

Sample	Validated miRNA	Regulation	Individuals	Reference
PBMCs	hsa-miR-144*	Upregulated	PTB vs control	Liu et al., 2011 (78)
	hsa-miR‑29b	Upregulated		Zhou et al., 2016 (86)b
	hsa-miR‑1	Downregulated
	hsa-miR‑150	Downregulated
	hsa-miR‑155	Downregulated
	hsa-miR‑31	Downregulated
	hsa-miR‑146a	Downregulated
	hsa-miR‑10a	Downregulated
	hsa-miR‑125b	Downregulated
Serum	hsa-miR-93*	Upregulated	PTB vs control	Fu et al., 2011 (57)
	hsa-miR-29a	Upregulated
	hsa-miR-3125	Downregulated
	hsa-miR-361-5p	Upregulated		Qi et al., 2012 (59)
	hsa-miR-889	Upregulated
	hsa-miR-576-3p	Upregulated
	hsa-miR-210	Upregulated
	hsa-miR-26a	Upregulated
	hsa-miR-432	Upregulated
	hsa-miR-134	Upregulated
	hsa-miR-378	Upregulated		Zhang et al., 2013 (63)
	hsa-miR-483-5p	Upregulated
	hsa-miR-22	Upregulated
	hsa-miR-29c	Upregulated
	hsa-miR-320b	Downregulated
	hsa-miR-101	Downregulated
	hsa-miR-196b	Upregulated	ATB vs control	Zhang et al., 2014 (60)
	hsa-miR-376c	Upregulated	ATB vs LTBI/BCG−
	hsa-miR-16	Upregulated	PTB vs control and TB treated	Wagh et al., 2017 (56)
	hsa-miR-155	Downregulated
	hsa-miR-148b-3p	Upregulated	Control vs ATB; ATB vs partially treated TB; ATB vs cured TB	Wang et al., 2018 (55)
	hsa-miR-21-5p	Upregulated
	hsa-miR-92a-3p	Upregulated	Control vs ATB; ATB vs cured TB
	Has-miR-125a-5p	Upregulated	Partially treated TB vs untreated; partially treated TB vs cured TB
	hsa-miR-144	Upregulated	ATB vs control	Lv et al., 2016 (74)
Sputum	hsa-miR-29a	Upregulated	PTB vs control	Fu et al., 2011 (57)
	hsa-miR-19b-2*	Downregulated		Yi et al., 2012 (61)
	hsa-miR-3179	Upregulated
	hsa-miR-147	Upregulated
CD4+ cells	miR-451a	Upregulated	ATB vs control	Fu et al., 2014 (117)
	miR-340-5p	Upregulated
	miR-136-5p	Upregulated
	miR-4292	Downregulated
	miR-29b	Upregulated
THP-1 cells	hsa-miR-16	Upregulated	ATB vs control	Zheng et al., 2015 (53)
	hsa-miR-137	Upregulated
	hsa-miR-140-3p	Upregulated
	hsa-miR-193a-3p	Upregulated
	hsa-miR-501-5p	Upregulated
	hsa-miR-598	Upregulated
	hsa-miR-95	Downregulated

^aExpression of miRNAs was confirmed by real-time PCR (qPCR) after sequencing.

^bSamples obtained from children (2 to 14 years old).

In the following sections, the analysis is focused on six miRNAs that have been specifically and consistently associated with TB disease (hsa-miR-144-3p, hsa-miR-144-5p, hsa-miR-146a, hsa-miR-155, hsa-miR-21, and hsa-miR-29a) according to the miRTarBase database (48), and evidence is shown that backs up this association as well as the possible gene regulation pathways involved.

miR-144-3p.

The expression of miR-144-3p in sputum and serum from patients with active TB was not only higher than that of healthy uninfected controls but was also able to discriminate patients that showed improvement after antibiotic treatment from those that did not (74). Accumulative evidence suggests that autophagy is a key process of innate immune response employed by the host against M. tuberculosis and other intracellular bacteria (75, 76). Autophagy is regulated by numerous distinct autophagy-related (ATG) proteins. Among the ATGs, ATG4 and microtubule-associated protein light chain 3 (LC3) are essential for autophagy induction. Interestingly, Mycobacterium bovis bacillus Calmette-Guérin (BCG) infection of murine macrophages induced increased expression of miR-144-3p, which targets ATG4a to inhibit autophagy activation and antimicrobial responses to BCG (Fig. 1A). Overexpression of miR-144-3p decreased both mRNA and protein levels of ATG4a as well as protein levels of p62 (molecular carrier of cargo to be degraded by autophagy) and LC3I/II, inhibiting maturation of autophagosome in macrophages, thus increasing intracellular survival of BCG. Conversely, when macrophage miR-144-3p expression was inhibited, ATG4a levels were increased, leading to an accelerated autophagic response and a decrease in BCG survival (77).

FIG 1.

miRNA regulation of miR-144-3p and miR-144-5p upon mycobacterial infection. (A) After Mycobacterium bovis (BCG) infection in murine macrophages and dendritic cells, miR-144-3p is upregulated and targets ATG4a mRNA to inhibit autophagy enhancing mycobacterial survival. (B) In human monocytes infected with Mycobacterium tuberculosis, miR-144-5p targets DRAM2 mRNA to inhibit autophagy promoting mycobacterial survival.

miR-144-5p.

Several studies have shown that expression levels of miR-144-5p are increased in PBMCs from active TB patients compared to those of the same cells from uninfected healthy controls (78–80); however, it is worth noting that miR-144-5p expression might be under a compartmentally distinct type of regulatory process, since pleural fluid mononuclear cells exhibited lower levels of miR-144-5p than PBMCs from patients with pleural TB (80). Evidence from in vitro experiments using purified T cells overexpressing miR-144-5p revealed that this miRNA inhibited the proliferation rate of T cells and the production of crucial Th1 cytokines, such as IFN-γ and tumor necrosis factor alpha (TNF-α), in response to TB infection (78). Additionally, miR-144-5p was capable of downregulating DNA damage-regulated autophagy modulator 2 (DRAM2) and lysosome-associated membrane protein 2 (LAMP2), reducing their mRNA expression in human monocytes. Since DRAM2 and LAMP2 are important proteins in the initiation of autophagy and in the chaperone-mediated autophagy, respectively, these findings suggest that upon infection, miR-144-5p levels contribute to autophagy inhibition, allowing M. tuberculosis to escape the immune system (Fig. 1B) (79).

miR-146a.

Contradictory results have been found regarding miR-146a expression levels. Some studies showed that its expression was upregulated in murine peritoneal fluid after BCG challenge; in human THP-1 cells, murine bone marrow-derived macrophages (BMDMs), and human monocyte-derived macrophages (MDM) upon BCG infection; in lungs and MDM of experimentally M. tuberculosis-infected mice; and in human dendritic cells (DCs) infected with M. tuberculosis (81–85). Conversely, studies performed on a large range of patients with active TB, specifically in white blood cells (children with active TB), PBMCs (pulmonary and extra pulmonary TB), mononuclear cells of pleural fluid, and in plasma samples, showed a downregulation in miR-146a expression (80, 86, 87). Additionally, this downregulation was also observed in alveolar macrophages of bronchoalveolar lavage fluid from patients with active PTB, where miRNA levels in alveolar macrophages from smear-positive patients were lower than in those from smear-negative patients, and higher in healthy uninfected subjects (81). Moreover, miR-146a expression was also downregulated in CD4⁺ T cells and CD19⁺ B lymphocytes obtained from mice challenged with M. bovis BCG (83) as well as in liver tissue from BCG-infected mice (82).

The role of miR-146a in immune regulation involved and required the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway for its induction after BCG infection in vitro, and, in fact, mutation of NF-κB sites and pharmacological inhibition of NF-κB signaling significantly decreased miR-146a basal promoter activity and its expression in mycobacterium-challenged macrophages (81, 82). Interleukin-1 receptor-associated kinase 1 (IRAK1) (critical in initiating an innate immune response against foreign pathogens and important in leading to NF-κB nuclear translocation and activation), TNF receptor-associated factor 6 (TRAF6) (a signal transducer in the NF-κB pathway), and prostaglandin-endoperoxide synthase 2 (PTGS2) (responsible for the production of inflammatory prostaglandins) are potential regulatory targets of miR-146a since it downregulated their mRNA levels in THP-1 cells (Fig. 2A) (81). In addition, miR-146a inhibited NF-κB, c-Jun N-terminal kinase (JNK), and p38 mitogen-activated protein kinase (p38 MAPK) pathways in RAW264.7 cells, impairing BCG-induced iNOS expression and nitric oxide (NO) production in macrophages by reducing the TRAF6 protein level (Fig. 2B) (82).

FIG 2.

Dual role of miR-146a in the survival of mycobacteria. (A) Upon Mycobacterium bovis (BCG) infection in macrophages differentiated from human THP-1 cells, miR-146a increases its expression and downregulates transcription factors to suppress the inflammatory response leading to a reduction in mycobacterial survival. (B and C) After BCG infection in murine macrophages, miR-146a increases its expression and downregulates transcription factors and cytokines to suppress the inflammatory response leading to an increase in mycobacterial survival.

Mycobacterial growth or bacterial fitness might also be regulated by miR-146a, as a lower mycobacterial colony count was found in macrophages differentiated from human THP-1 cells transfected with a miR-146a mimic (81). In contrast, in RAW264.7 cells, miR-146a promoted the survival of M. bovis BCG by suppressing the inflammatory response (Fig. 2C) (88) while miR-146a inhibition enhanced the mycobactericidal activity and, thus, mycobacterial clearance due to an increase in NO production (82). A possible explanation for the differences in mycobacterial survival found between these two studies could be the distinct origin of the cells analyzed, i.e., macrophages differentiated from the human monocytic THP-1 cell line (81) versus the mouse macrophage RAW264.7 cell line (88). Other methodological differences, such as the incubation period for the infection or the assay used to measure mycobacterial load, could also explain the differences found in these results.

Interestingly, a higher risk of developing active PTB has been related to miR-146 single-nucleotide polymorphisms (SNPs) in Tibetan, Kasak, and Han populations but not in others such as the Uygur (90, 91).

miR-155.

A few studies have shown that miR-155 levels were reduced in serum from patients with active TB (56, 92) and increased after TB therapy, although not reaching the same levels as those in uninfected subjects (56). Another study showed that human MDMs exposed to lipomannan from the M. tuberculosis H37Rv strain exhibited a lower miR-155 expression with correspondingly low TNF-α production (Fig. 3A) (93); this reduction in miR-155 levels was also present in human macrophages infected with a virulent M. tuberculosis strain, compared with macrophages infected with the avirulent strain or uninfected macrophages (Fig. 3B) (94). Interestingly, natural killer (NK) cells from TB patients that exhibited higher serum levels of miR-155 displayed lower cytotoxicity and lower TNF-α production in in vitro assays (92). Opposing these findings, a very large number of studies have shown miR-155 upregulation occurs in several types of cells and tissues in response to M. tuberculosis. These cases include RAW264.7 cells after M. tuberculosis infection (95), lung tissue from a PTB murine model at early time points of infection (30 days) (83), murine BMDMs and RAW264.7 cells challenged with M. bovis BCG (96), the J774A.1 mouse macrophage cell line infected with M. tuberculosis H37Rv and to a lesser extent when infected with M. bovis and M. bovis BCG (97), PBMCs from patients with active TB stimulated with purified protein derivatives of tuberculin (PPD) (98), and THP-1 cells infected with BCG (99).

FIG 3.

miRNA regulation of miR-155 upon mycobacterial infection and dual role of miR-155 in the survival of mycobacteria. (A and B) Upon lipomannan stimulation in human monocyte-derived macrophages (MDM) or Mycobacterium tuberculosis infection in human macrophages, miR-155 expression decreases. (C) Expression of miR-155 in macrophages stimulated with ESAT-6 increases and triggers a series of events to increase apoptosis and lead to a reduction of mycobacterial survival. (D) After M. tuberculosis infection in mice, miR-155 expression increases in liver and lungs, which is relevant to the immune response, since mice lacking miR-155 exhibit higher susceptibility and lower survival. (E) Upon Mycobacterium bovis (BCG) infection in macrophages, miR-155 expression increases and triggers a series of events to increase autophagy to reduce mycobacterial survival. (F to H) M. tuberculosis infection of human dendritic cells and murine macrophages or BCG infection of human monocytes promotes miR-155 expression that leads to downregulation of transcription factors that consequently favors the survival of the intracellular pathogens. (I) In mouse experiments, miR-155 expression induced by M. tuberculosis infection or ESAT-6 stimulation promotes Toll-like receptor pathway activation, upregulating the expression of certain cytokines that leads to renal injuries.

The upregulation of miR-155 in RAW264.7 cells induced by M. tuberculosis peptide ESAT-6 stimulation depended on the activation of the Toll-like receptor 2 (TLR2)/NF-κB pathway. In turn, the SOCS1 pathway was targeted by miR-155 for the ESAT-6-mediated protective immune response and macrophage apoptosis (Fig. 3C) (100). Moreover, experiments with mice lacking miR-155 (miR-155^−/−) showed that they are more susceptible to M. tuberculosis infection and exhibited lower numbers of CD4⁺ T cells and an impaired M. tuberculosis clearance compared with the wild-type mice (Fig. 3D) (101). Additionally, IFN-γ levels measured in supernatants from cultures of splenocytes from M. tuberculosis-infected mice and stimulated in vitro with M. tuberculosis were lower in the miR-155^−/− mice than in the wild type. These results suggest that miR-155 has an important role in the in vivo immune response against TB. Moreover, in infected RAW264.7 cells, miR-155 decreased the survival of intracellular BCG, M. tuberculosis H37Ra, and H37Rv. This was done targeting and downregulating posttranscriptionally the protein Rheb, consequently promoting the maturation of mycobacterial phagosomes and also activating autophagy (Fig. 3E) (96).

Virulent and live mycobacterium infection of human DCs upregulated miR-155 in opposition to BCG and heat-inactivated M. tuberculosis. The increased expression of miR-155 consequently decreased ATG3 protein levels as well as LC3-II and gamma-amino butyric acid receptor-associated protein (GABARAP) (involved in apoptosis and autophagy), thus inhibiting autophagosome formation (Fig. 3F) (85). In this same regard, human MDMs infected with M. tuberculosis H37Rv or stimulated with ESAT-6 showed an increase in miR-155 levels (102), as well as the human monocytic leukemia cell line (U937 cells) infected with BCG, where the upregulation was further enhanced by stimulating with Rv2346c, a putative ESAT-6-like protein of M. tuberculosis (103).

In contrast, opposing results found in M. tuberculosis-infected RAW264.7 cells have shown that miR-155 was involved in the downregulation of target proteins, such as SH2 domain-containing inositol 5′-phosphatase 1 (SHIP1) (affects cellular behaviors in innate and adaptive immune cells), transcription regulator protein BACH1 (negative regulator of hmox1 that increases its expression upon M. tuberculosis infection), Cox-2/PTGS2, and interleukin-6 (IL-6) (important in the differentiation of B-cells and required for the generation of Th17 cells), benefiting mycobacterial growth (Fig. 3G) (95). Other studies have reported that miR-155 contributed to mycobacterial survival by reducing apoptosis through modulation of Forkhead box O3 (FOXO3) (regulates apoptosis and autophagy) in the THP-1 human monocytic cell line (Fig. 3H) (99). Also, miR-155 increased mRNA expression of TLR4, MyD88, TNF-α, IL-17, and IFN-γ through the TLR4 signaling pathway, promoting the progression of infection, leading to renal injuries in extra PTB disease in mice (Fig. 3I) (104).

Furthermore, studies exploring the diagnostic value of specific miRNAs in TB have described that lower levels of miR-155 could be a potential biomarker for childhood TB (86) and in adults could discriminate between PTB and control groups with diagnostic AUCs consistently found above 0.9 (56, 86, 105).

miR-21.

The role of miR-21 in TB pathogenesis has been less studied; however, a lower expression was found in CD4⁺ T cells and plasma from patients with active TB than from subjects with LTBI (87, 106) but there was no difference between subjects with LTBI and healthy volunteers (106). Conversely, miR-21 expression was upregulated in M. tuberculosis-infected macrophages compared to uninfected macrophages and macrophages infected with M. bovis or M. bovis BCG (97). The same was observed in RAW264.7 and THP-1 cells where miR-21 enhanced M. tuberculosis survival and apoptosis and attenuated the secretion of inflammatory cytokines (IL-1β, IL-6, and TNF-α) by targeting B-cell lymphoma-2 (Bcl-2) and Toll-like receptor 4 (TLR4) (Fig. 4A) (107). Finally, miR-21 levels also increased in PPD-MPT64-stimulated RAW264.7 cells in contrast to PPD-stimulated cells, showing that M. tuberculosis immunogenic protein MPT64 upregulates miR-21 due to increased NF-κB expression, leading to the upregulation of Bcl-2, an antiapoptotic protein, thus decreasing cell death (Fig. 4B) (108). Possible explanations for the discrepancies related to the enhanced or decreased apoptosis might be due to the different stimulus performed (M. tuberculosis versus MPT64), which triggered different mechanisms leading to decreased or increased Bcl-2 levels that in turn correlate with the level of apoptosis.

FIG 4.

Dual role of miR-21 upon mycobacterial infection. (A) Expression of miR-21 is increased in human and murine macrophages infected with Mycobacterium tuberculosis, driving downregulation of Bcl-2 and TLR-4, suppression of inflammatory cytokines, and enhanced apoptosis and mycobacterial survival. (B) After MPT64 stimulation in murine macrophages, miR-21 is regulated by NF-κB to increase Bcl-2 expression, reducing apoptosis.

miR-29a.

A study shows that miR-29a expression depended on sex, where PBMCs from females with active TB exhibited a lower expression than PBMCs of male patients before beginning TB therapy and after up to 2 months of treatment. After completing 6 months of treatment, miR-29a levels increased in females, while showing a trend toward reduced levels in males (64). Discrepancies in miR-29a expression arose upon antimycobacterial therapy; while one study showed no significant difference in serum samples (56), other studies showed increased levels of miR-29a in CD4⁺ T cells (64, 109) as well as in plasma samples, where it was reduced to levels equivalent to healthy controls (87). Contradictory results in miR-29a expression were also found regarding the status of M. tuberculosis infection. In PBMCs from children with tuberculosis meningitis and in plasma from patients with active TB, miR-29a expression was found to be higher than that in healthy individuals (87, 110, 111). Additionally, human macrophages infected with M. tuberculosis exhibited an enhancement in miR-29a expression that led to the downregulation of caspase-7 (CASP7) (a key regulator of apoptosis), preventing cell death and allowing mycobacterial intracellular growth (Fig. 5A) (112). Conversely, in CD4⁺ T cells from patients with active TB and in whole blood from children with TB, miR-29a expression was low compared to that of their LTBI counterpart, and while no difference was exhibited between LTBI and healthy adults, miR-29a expression was found to be low in healthy children compared to that in children with LTBI (106). The expression of miR-29a was also reduced in CD4⁺ and CD8⁺ T cells from mice infected with M. bovis BCG; interestingly, both types of T cells showed an increase in IFN-γ concentration, suggesting an inverse correlation between miR-29a and IFN-γ expression during the infection (Fig. 5B) (113). Concordantly, in mice transfected with microRNA sponges—transcripts containing multiple miRNA-binding sites that sequester miRNAs inhibiting their function (114, 115)—and infected with M. bovis, the inverse correlation between miR-29a and IFN-γ was also observed. In addition, the challenged mice with M. tuberculosis PPD in the footpads exhibited more swelling in the infection site, more leukocyte aggregation, increased levels of IFN-γ, lower bacterial burden in the lungs, and a longer life expectancy than the nonchallenged control mice. Together, these results showed that miR-29a regulates the innate and adaptive immunity suppressing IFN-γ through its interaction with IFN-γ mRNA, protein Argonaute 2 (Ago2) (required for RNA-mediated gene silencing), and the RNA-induced silencing complex (RISC) (113). However, evidence also suggested that miR-29a might not alter IFN-γ production in CD4⁺ T cells (106, 109). Furthermore, miR-29a might also regulate other cytokines and chemokines since its inhibition in M. tuberculosis-infected DCs showed an enhanced response of 12 cytokines and chemokines, including IL-12B, IL-2RA, and CXCL10, predicted targets of miR-29a, compared to that of infected but noninhibited cells. Finally, M. tuberculosis-infected DCs transfected with a miR-29a mimic exhibited a reduced CXCL10 expression, suggesting a direct regulation of miR-29a on this chemokine in M. tuberculosis-infected DCs (116).

FIG 5.

Regulation of miR-29a upon mycobacterial infection. (A) Human macrophages infected with Mycobacterium tuberculosis show enhanced miR-29a expression that leads to downregulation of caspase-7, prevention of cell death, and mycobacterial survival. (B) CD4⁺ and CD8⁺ T cells isolated from mice infected with Mycobacterium bovis (BCG) exhibit a downregulation of miR-29a but increased levels of IFN-γ, showing an inverse dynamic between these two molecules.

CONCLUSIONS

Distinctive miRNAs, differentially expressed in the diverse stages of TB infection, might shed light on the nature of the immune response to this elusive pathogen. However, great variability in miRNA expression and its biological effects are dependent on the experimental model conditions, increasing difficulties to conclude any clear or unique effect on the immune response to TB, if any. Furthermore, although some miRNAs, such as miR-21-5p and miR-155, are more consistent as potential diagnostic biomarkers for TB in humans, it is clear that further validation is needed. Nevertheless, the following considerations can be drawn. (i) Different miRNA signatures can emerge from infection with different M. tuberculosis strains and degrees of virulence. (ii) The time and stage of M. tuberculosis infection play a key role in miRNA expression profile, with striking differences between LTBI and active TB. (iii) Ex vivo and in vitro experiments may provide different and even contradicting results on miRNA expression. (iv) Organ- or tissue-specific TB pathogenesis introduces variations into miRNA functions since pulmonary and extrapulmonary TB do not always provide the same miRNA expression profile. (v) Each specific miRNA may function in different cellular signal pathways and exerts diverse effects in macrophages during mycobacterial infection. (vi) Antimycobacterial therapy modifies the miRNA expression profile. (vii) The miRNA expression profile seems to be sex, age, species, and M. tuberculosis strain specific.