Comparative Analysis of Cytokine mRNA Expressions in Human Tissues With Mycobacterium Tuberculosis Infection


 Background: One of the widely used diagnostic methods for Mycobacterium tuberculosis (MTB) infection is the acid-fast bacilli staining of formalin-fixed paraffin-embedded (FFPE) tissues; however, this method cannot discriminate between MTB and nontuberculous mycobacteria (NTM) species. Moreover, confirming tuberculosis (TB) using FFPE tissue specimens may be difficult owing to their low bacterial load. In addition, interference in molecular diagnostic assays, including polymerase chain reaction (PCR), may occur owing to fragmentation and genomic DNA cross-linkage in FFPE tissues formed during formalin fixation or paraffin-embedding procedures. Therefore, we aimed to investigate whether an automated molecular diagnostic method based on PCR-reverse blot hybridization assay can discriminate between human MTB-positive and -negative FFPE tissues and to compare the relative mRNA expression levels of various host immune markers between MTB-infected and uninfected human tissues using quantitative reverse transcription (qRT) PCR. A total of 52 human FFPE tissue samples from various regions of the body, including the lungs, lymph nodes, tendons, colon, and appendix, were collected and used for the molecular identification of Mycobacterium species and analysis of cytokine mRNA expression. Results: IFN-γ, TNF-α, IP-10, CXCL9, CXCL11, and GM-CSF mRNA expression levels in MTB-infected tissues were significantly higher than those in uninfected samples. Additionally, the differences in the mRNA expression levels of IFN-γ, CXCL9, and GM-CSF between MTB-infected and uninfected tissues were statistically significant were statistically significant (p < 0.05). Correlation curve analysis indicated that the mRNA expression of IFN-γ was inversely proportional to that of IP-10 and that the mRNA expression levels of IFN-γ, TNF-α, CXCL9, CXCL11, GM-CSF, and TNFR were proportional and well-correlated. Furthermore, to establish marker profiles for detecting MTB infection, the statistically significant expression levels of three markers were combined. We confirmed that the combined profile of IFN-γ, CXCL9, and GM-CSF expression levels was statistically significant (P < 0.001). Conclusions: Although the mRNA expression patterns of host immune markers may vary according to MTB infection status, these patterns may be highly correlated and can be simultaneously used as an additional indicator for diagnosing TB in human tissue samples.

Background Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis (MTB) and remains a serious health problem worldwide [1]. According to a World Health Organization report in 2018, the Republic of Korea had the highest prevalence and mortality rate for TB among countries in the Organization for Economic Cooperation and Development [2]. Moreover, cases of multi-drug-resistant and extensively drug-resistant TB are rapidly increasing, leading to increased di culties in e cient TB treatment and control [3].
Globally, one of the most widely used diagnostic methods for detecting MTB infection is the acid-fast bacilli (AFB) staining of formalin-xed para n-embedded (FFPE) tissues; however, this method is affected by some factors: FFPE tissues may contain relatively low bacterial load, and granulomas may be caused by other kinds of microbes, including other bacterial and fungal species such as Cryptococcus species and Listeria monocytogenes. Moreover, AFB stains cannot discriminate between MTB and nontuberculous mycobacteria (NTM) infections [4,5]. Alternative diagnostic tools to detect MTB and simultaneously identify mycobacterial species in tissue samples are therefore needed to improve TB diagnosis and treatment [6].
Molecular diagnostic methods, including polymerase chain reaction (PCR) and real-time PCR, have recently been used to effectively diagnose and identify mycobacterial species in various types of clinical samples [4]; however, interferences in molecular diagnostic assays may occur owing to fragmentation and genomic DNA (gDNA) cross-links in FFPE tissues formed during formalin xation or para nembedding procedures [7].
Several different immune markers such as cytokines, chemokines, immune regulators, and immune stimulating factors have recently been shown to play important roles in hosts infected with MTB [8]. In a study, tumor necrosis factor-alpha (TNF-α), an in ammatory cytokine, was reportedly implicated in not only the proliferation and differentiation of immune cells but also in in ammatory processes, including apoptosis of MTB-infected cells [9]. In another study, interferon-gamma (IFN-γ) and antigen-speci c T cells reportedly controlled MTB infection [10]. In addition, interleukin 2 (IL-2) levels have been found to be signi cantly higher in patients with active TB than in healthy control groups [11], and IL-4 has also been reported to act as an anti-in ammatory cytokine in cases of TB infection [12]. Furthermore, C-X-C motif chemokine 9 (CXCL9; a monokine induced by IFN-γ) levels in the plasma of patients with active pulmonary TB are higher than that in the plasma of healthy individuals [8]. CXCL10 (an IFN-γ-inducible protein of 10, IP-10) levels have also been found to be elevated in patients with TB lymphocytic pleural effusion [13], and CXCL11 (an IFN-γ inducible T cell alpha chemoattractant) levels have been shown to be signi cantly increased by stimulation with an MTB antigen [14]. Importantly, research has also shown that the mRNA levels of these immune markers can be precisely measured in blood immune cells to distinguish among individuals with active TB, those with latent TB infection (LTBI), and healthy controls [15]. Granulocyte-macrophage colony-stimulating factor (GM-CSF) is a biomarker that can be used to differentiate between persons with active TB and QuantiFERON TB Gold In Tube (QFT-GIT) test-positive persons [16]. Higher levels of plasma CCL11 (eotaxin) and TNF receptor (TNFR) have also been associated with loss of CD4 + T cells [17].
Here, we aimed to investigate whether an automated molecular diagnostic method based on PCR-reverse blot hybridization assay (REBA) can discriminate between human MTB-positive FFPE tissues and MTBand NTM-negative FFPE tissues and to compare the mRNA expression levels of various host immune markers in MTB-infected human tissues using quantitative reverse transcription PCR (qRT-PCR) TaqMan probe assays. We also attempted to overcome the limitations of conventional DNA testing by targeting small amounts of fragmented DNA from MTB-infected FFPE tissues, and we present here an evaluation of the usefulness of this diagnostic method.

Gene expression pro ling of host immune markers in MTB-infected and uninfected tissue samples
Of the 25 target genes assessed, the gene expression levels of CXCL9, CXCL10, CCL2, CCL3, CCL4, CCL11, IL1β, IL-2R, TNFR, and IL-6R were higher in MTB-infected tissues than in uninfected tissues; however, the mRNA expression levels of CXCL9, CCL2, CCL3, CCL4, IL1β, and IL-6R varied in MTB-infected tissues, even though they exhibited higher mRNA expression levels. The mRNA expression levels of IFN-γ, TNF-α, IL-2, IL-4, and CXCL11 reportedly increased upon infection with MTB based on previous studies; however, the differences in the mRNA expression levels of IL-6 and IL-8 between MTB-infected and uninfected tissues were not signi cant (Table 3). mRNA expression patterns in samples according to MTB infection Shown in Figure 1 are the mRNA expression levels of MTB-positive and MTB-negative tissues. The mRNA expression levels of IFN-γ, TNF-α, CXCL9, CXCL10, CXCL11, and GM-CSF were relatively higher in MTBpositive tissues than in MTB-negative tissues. In particular, IFN-γ, CXCL9, and GM-CSF were signi cantly higher (p < 0.05) in MTB-positive tissues than in MTB-negative tissues. IL-2 and CCL11 mRNA expression levels were lower in MTB-positive tissues than in MTB-negative tissues; however, the mRNA expression levels of IL-2R and TNFR were not different between MTB-positive tissues and MTB-negative tissues. IL-4 mRNA expression levels were not detected owing to low levels. To establish marker pro les for detecting MTB infection, the expression levels of three markers whose differences between MTB-positive and MTBnegative tissues were statistically signi cant were combined and validated. The differences in the combined pro le of IFN-γ, CXCL9, and GM-CSF expression levels discriminated between MTB-positive and -negative tissues and was also statistically signi cant (P < 0.001).

Correlation curve analysis patterns of mycobacterial infection
The correlation curve analysis was based on the expression levels of IFN-γ, which is known to speci cally increase in host tissues with MTB ( Figure 2). The mRNA expression levels of IFN-γ were signi cantly correlated with those of TNF-α, CXCL9, CXCL11, GM-CSF, and TNFR; however, there was no correlation between the mRNA expression levels of IFN-γ and IL-2 or IL-2R. Interestingly, the mRNA expression levels of IFN-γ and CXCL10 were inversely proportional.
Receiver operating characteristic (ROC) curve analysis of the mRNA expression levels of host immune markers ROC curve analysis was performed to ensure that the results were clinically applicable (

Discussion
TB remains a serious global health problem [18]. NTM infections make the accurate diagnosis and treatment of TB challenging [19], and effective methods are needed to accurately identify different mycobacterial species. Therefore, we determined, using an automated molecular mycobacterial identi cation system, and analyzed the mRNA expression patterns of various host immune markers in tissues according to MTB infection status. Recent studies have shown that many cytokines play an important role in cell-mediated immune responses to MTB infections [20]. The gold standard methods for analyzing host immune markers in FFPE tissues are immunohistochemistry (IHC) staining and ow cytometry analysis in which target-speci c antibodies and homogenized tissues are used; however, these methods require a large amount of samples to simultaneously detect multiple markers. The simultaneous detection of multiple host markers in a small amount of samples and analysis of mRNA expression levels using molecular diagnostic assays with higher analytical sensitivity and speci city than IHC and ow cytometry methods may improve the diagnosis of TB. Therefore, we aimed to pro le the gene expression patterns of 11 host immune markers in MTB-positive and -negative FFPE tissues.
Gene expression patterns were pro led using QuantiGene ® Plex assay (Thermo Fisher Scienti c), and out of a total of 25 immune cytokine genes, the mRNA expression levels of IFN-γ, TNF-α, IL-2, IL-2R, IL-4, CXCL9, CXCL10, CXCL11, and GM-CSF, which are known to increase upon infection with MTB, were higher than those in uninfected tissues. Speci cally, TNFR and CCL11 expression levels were signi cantly increased. Although the expression levels of CCL2, CCL3, CCL4, IL1β, and IL-6R were also signi cant, the differences among sample results were borderline signi cant and therefore were excluded. Therefore, to de nitively con rm the gene expression patterns of 11 host immune markers, we performed qRT-PCR analyses using TaqMan probes. In a previous study, IFN-γ and CXCL9 mRNA levels were found to be signi cantly higher in individuals with active pulmonary TB than in healthy controls [14]. In another study wherein the GM-CSF levels between IGRA-positive and -negative cases were compared, the cytokine levels in IGRA-positive cases were noted to be signi cantly higher than those in negative cases [16]; our results are consistent with this nding in tissue samples with and without MTB infection. Furthermore, in our ROC curve analysis, the AUCs of IFN-γ, CXCL9, and GM-CSF were approximately 0.7. To establish marker pro les that can discriminate between MTB-positive and -negative tissues, two groups, each consisting of combined and validated statistically signi cant mRNA levels of three immune markers, were used. A combined pro le of IFN-γ, CXCL9, and GM-CSF mRNA expression levels can discriminate between MTB-positive and -negative groups and was statistically signi cant (P < 0.001). With this, even when clinical molecular diagnostic assays such as PCR-REBA and nested PCR assays for detecting MTB indicate negative results, accurately diagnosing a clinical case of TB is possible by analyzing the gene expression patterns of select host immune markers. Additionally, even if DNA fragmentation occurs, the mRNA levels of cytokines remain unaffected.
Our study has some limitations. First, a small sample size was used; further research with a larger sample would provide improved signi cance to succeeding results. Second, the method used in this study is not easily applicable in clinical settings owing to the use of tissue samples from patients. Our analysis was conducted as a preliminary study of the gene expression patterns in peripheral whole-blood samples for the detection of TB. Further research on cytokine analysis using peripheral whole-blood samples in patients with extrapulmonary TB would be required.

Conclusion
The mRNA expression levels of the host immune markers IFN-γ, TNF-α, CXCL9, CXCL10, CXCL11, IL-2R, and GM-CSF were higher in MTB-positive tissues than in MTB-negative tissues. Speci cally, increased mRNA expression levels of IFN-γ, CXCL9, and GM-CSF were statistically signi cant (p < 0.05), and ROC curve data indicated that the AUCs of the combination of these three target mRNA expression levels exceeded 0.7. These results suggest that when TB is suspected, gene expression analysis may be used to accurately diagnose the presence of TB.

Molecular identi cation of Mycobacterium species
To con rm MTB-positive and -negative FFPE tissues, we used REBA Myco-ID ® Kit (YD Diagnostics, Yongin, Republic of Korea), for the molecular identi cation of Mycobacterium species, according to the manufacturer's instructions. A single PCR cycle was used for initial pre-denaturation at 95°C for 5 min, followed by 30 cycles at 95°C for 30 s, 30 cycles at 65°C for 30 s, and one cycle at 72°C for 10 min. The ampli ed PCR products were then loaded into an HybREAD 480 ® System (YD Diagnostics), and all reactions were automatically performed by the instrument. After the processes were complete, the data were interpreted using an HybREAD 480 ® Scanner (YD Diagnostics).

Statistical analysis
All statistical analyses were performed using GraphPad Prism v. 5.00 (GraphPad Software, San Diego, CA, USA). Signi cant differences in the mRNA expression levels of MTB-positive and MTB-negative tissues, together with 90% con dence intervals (CIs), were calculated and compared using unpaired ttests. Associations between the mRNA expression levels of IFN-γ and other cytokines were measured using Spearman's rank correlation coe cient (r). An ROC curve analysis was performed to con rm clinical applicability. A p-value of less than 0.05 was considered statistically signi cant.

Declarations
Ethics approval and consent to participate This study was approved by the Institutional Review Board of the Kosin University Gospel Hospital (KUGH 2017-11-042). FFPE tissue samples were collected after obtaining informed consent for research with the approval of the Biobank of Kosin University Gospel Hospital and all experiments were performed in accordance with relevant guidelines and regulations.

Consent for publication
Not applicable.

Availability of data and materials
The data analyzed during this study are included in this published article.