Diagnostic value of miR-155, miR-29a and lncRNAs (NEAT1, GAS5 and COX2) in HIV, tuberculosis, and HIV/TB co-infection

Abstract

Tuberculosis (TB) and human immunodeficiency virus (HIV) co-infection remain leading causes of mortality, especially in source-limited settings where diagnostic challenges impede timely management. Non-coding RNAs (ncRNAs) are emerging as non-invasive biomarkers for infectious diseases due to their role in immune regulation. However, the diagnostic potential of ncRNAs in distinguishing TB monoinfection, HIV monoinfection, and TB/HIV co-infection remains unclear. In this study, we examined the expression of miR-155, miR-29a, and lncRNAs (lncRNA-COX2, lncRNA-NEAT1, lncRNA-GAS5) by qRT-PCR in peripheral blood mononuclear cells (PBMCs) from 95 participants: 25 with HIV monoinfection, 20 with TB monoinfection, 25 with TB/HIV co-infection, and 25 healthy controls. Statistical analyses, including Spearman’s rank correlation and receiver operating characteristic (ROC) curves, were used to evaluate diagnostic performance. miR-155 was significantly downregulated in HIV+ (P = 0.0004) and TB+ (P = 0.02) groups compared to controls, but not in TB/HIV+ (P = 0.3). miR-29a was upregulated in TB+ (P = 0.03) but not significantly altered in other infected groups. lncRNA-COX2 was upregulated in HIV+ (P = 0.03) compared to controls, with non-significant trends in TB + and TB/HIV+. lncRNA-NEAT1 was upregulated in HIV+ (P = 0.0002) and TB+ (P < 0.0001), but not in TB/HIV+ (P = 0.3). lncRNA-GAS5 was downregulated in HIV+ (P < 0.0001), with no significant changes in TB+ (P = 0.4) or TB/HIV+ (P = 0.1). These group-specific patterns are detailed in Table 2 with fold-changes. Both lncRNA-COX2 and lncRNA-NEAT1 were upregulated across all infected groups compared to controls, while lncRNA-GAS5 was increased in TB + and TB/HIV + groups but decreased in HIV + alone. Notably, lncRNA-COX2 exhibited the highest expression levels in TB + and TB/HIV + groups, indicating an inflammatory response related to TB. Similarly, elevated lncRNA-GAS5 levels in TB + and TB/HIV + suggest its role in TB-associated pathology and co-infection effects. lncRNA-GAS5 and lncRNA-NEAT1 demonstrated high diagnostic accuracy for TB (AUC = 0.79 and 0.85, respectively). Selective biomarkers enhanced diagnostic performance, with a combination of miR-29a, lncRNA-NEAT1, and lncRNA-GAS5 achieving an AUC of 0.98 for TB. These findings suggest that multiplex ncRNA profiles provide a powerful diagnostic tool for TB/HIV co-infection, offering a robust, blood-based alternative for early detection in high-burden regions.

Introduction

Mycobacterium tuberculos is (Mtb) is a dangerous pathogen that impacts humans and causes the disease known as tuberculosis (TB)¹; this illness remains a major health concern in many parts of the world, especially in source-limited countries, human immunodeficiency virus (HIV) and TB account for a heavy burden of medical issue. Furthermore, Tuberculosis is the primary cause of death for people living with HIV (PLWH)². The development of active TB is mainly affected by how it interacts with various components of the immune system. Several mechanisms have been recognized as playing a role in the dysregulated immune responses typical of HIV/TB co-infection³. Previous studies have described HIV-related alterations in apoptosis⁴, antigen presentation⁵, immune cell functionality⁶, and cytokine release^7,8 in the response to TB.

It is essential to identify PLWH who are at risk of developing tuberculosis, as they could benefit from enhanced monitoring and clinical assessment. Traditional TB diagnostic methods have limitations in PLWH, with sputum smear microscopy showing negative results in 24–61% of pulmonary TB cases among those co-infected with HIV⁹. The rapid molecular assay Xpert MTB/RIF offers enhanced diagnostic capabilities but, for smear-negative cases, has an estimated sensitivity of only 55% in PLWH, compared to 67% in HIV-negative individuals^10,11. Infants of less than six years of age are unable to produce sputum for microscopy and culture which are currently considered as the confirmatory tests for TB. In this age group, the tuberculin skin test (TST) is used but this test is not specific for MTB, also the TST cannot distinguish between healthy individuals vaccinated with the BCG and those with active TB disease. Interferon gamma release assay (IGRA) testing must be conducted under standard laboratory conditions by trained personnel and its reproducibility is disputed^12–16. Diagnosis by culture is considered as the gold standard for TB but it is time consuming and not very suitable for extra-pulmonary TB. The diagnosis of smear-negative and extra-pulmonary tuberculosis continues to pose substantial clinical challenges with pediatric TB not left out because of the difficulty of sampling sputum^17,18, in this regard there is a demand for more sufficient diagnostic systems with alternative specimens such as blood, feces and urine which can be obtained from any age group. An increasing number of studies have shown that the expression of several specific RNAs has been associated with the occurrence, development, therapy, and prognosis of TB^19–22.

Previous studies have focused on identifying circulating host metabolite or microRNA (miRNAs) or long non-coding RNA (lncRNAs) profiles for TB diagnosis^23–27, however there are limited data on the changes of these analytes in the serum of patients with TB and HIV co-infection. Furthermore, HIV infection alone leads to changes in host serum metabolites, miRNAs and lncRNAs, thus the profile of altered metabolites and miRNAs in TB and HIV co-infection may differ compared to either TB infection or HIV infection^28–34. Serum miRNAs are not readily degraded by enzymes, unaffected by changes in temperature and time, and resistant to acids and alkalis^35,36. Furthermore, serum miRNAs are stable, and evaluating serum miRNA profiles is a feasible diagnostic procedure in clinical laboratories. Because of their high diagnostic potential, serum miRNAs have been evaluated as biomarkers in several pathological conditions including TB, with some studies achieving 82% to 100% accuracy in diagnosing TB by evaluating miRNA^37–42. In addition, high-throughput sequencing has also shown that in addition to mRNA, extracellular RNA contains a variety of non-coding RNA types, including lncRNA and miRNA⁴³. Therefore, lncRNA can be identified in specimens, such as in the serum and sputum, rather than being detected due to the degradation of lncRNA into small fragments⁴⁴.

This study aimed to investigate the power of miRNAs of Hsa-miR-29a, Hsa-miR-155, and lncRNAs of NEAT1, GAS5 and COX2 as a potential biomarker in Mtb and HIV infection.

Results

Participant characteristics

A total of 95 participants were included: 25 HIV+ (mean age 35.2 ± 10.1 years; 60% male), 20 TB+ (mean age 36.4 ± 9.8 years; 55% male), 25 TB/HIV+ (mean age 34.8 ± 11.2 years; 65% male), and 25 healthy controls (mean age 35.0 ± 10.5 years; 52% male). Median CD4 + T-cell counts were 420 cells/µL (range: 150–950) in HIV+, 650 cells/µL (range: 300–1100) in TB/HIV+, and not applicable for TB + or healthy groups. HIV viral loads were < 2000 copies/mL in all HIV + and TB/HIV + participants. No significant differences in age or sex were observed across groups (P > 0.05). All participants were negative for HBV and HCV.

Results of MiRNAs and LncRNAs selection

Common genes between HIV and tuberculosis pathways

KEGG pathway analysis confirmed significant overlap between HIV-1 infection and TB pathways (overlap 60/212 and 60/180, respectively; adjusted p < 1E-100). A total of 60 common genes were identified (e.g., AKT1-3, BCL2, CASP3/8/9, CYCS, IFNA1-21/IFNB1, IRAK1/4, MAPK1/3/8–14, MYD88, NFKB1, PPP3CA-CB-CC/R1-R2, RAF1, RELA, TLR2/4, TNF, TNFRSF1A, TRAF6). These genes cluster in immune activation, apoptosis, and inflammation pathways (e.g., lipid/atherosclerosis: 56/215 genes, adjusted p = 4.39E-108), representing critical hubs for co-infection synergy.

microRNAs targeting the common genes

Enrichr/miRTarBase analysis identified multiple miRNAs targeting the 60 shared genes. Top enriched miRNAs included hsa-miR-34a-5p (13 targets: e.g., BCL2, BAX, CASP3/8/9; adjusted p = 1.99E-04), hsa-miR-146a-5p (4 targets: e.g., IRAK1, NFKB1, TRAF6; adjusted p = 0.0015), and hsa-miR-143-3p (7 targets: e.g., AKT1/2, BCL2, MAPK1; adjusted p = 0.0018). Notably, miR-29a-3p (5 targets: e.g., AKT2/3, BCL2, CALM3, CASP8; adjusted p = 0.055) and miR-155-5p (8 targets: e.g., AKT1, CASP3, FADD, MAPK13/14, MYD88, NFKB1; adjusted p = 0.0032) were selected for focus due to their validated roles in immune regulation and pathogen response. These miRNAs suggest post-transcriptional modulation of co-infection outcomes, with miR-29a often downregulated in HIV (promoting latency) and upregulated in TB (diagnostic marker), while miR-155 is upregulated in both (enhancing inflammation).

lncRNAs associated with HIV and tuberculosis

2 GeneCards and co-expression analyses identified multiple lncRNAs linked to HIV/TB. High-relevance hits included GAS5 (associated with apoptosis/T-cell dysfunction) and NEAT1 (involved in latency/paraspeckle formation). PTGS2-AS1 (lncRNA-COX2) was selected for its inflammation-modulating role via PTGS2. These lncRNAs interact with shared genes (e.g., NEAT1 co-expressed with MAPK/TLR) or sponge miR-29a/miR-155, regulating immunity. Expression patterns: GAS5 downregulated in both infections (T-cell exhaustion); NEAT1 upregulated (promoting persistence); COX2 upregulated (inflammation).

Expression patterns of MiRNAs and LncRNAs in PBMCs

The expression levels of miR-155, miR-29a, and the lncRNAs (COX2, NEAT1, and GAS5) were quantified in peripheral blood mononuclear cells (PBMCs) across the four study groups (healthy controls, HIV+, TB+, and TB/HIV+), with results expressed as log2 fold changes relative to the healthy group. MiR-155 exhibited a progressive downregulation across infection groups, with the most pronounced reduction observed in the TB-HIV co-infected cohort (Fig. 1A). Compared to healthy controls, miR-155 levels were significantly decreased in TB-HIV patients (p = 0.004), while no significant difference was noted in HIV+ (p = 0.2) or TB+ (p = 0.2) groups. Inter-group comparisons revealed not significant downregulation in TB + relative to HIV+ (p = 0.25) and in TB-HIV relative to both TB+ (p = 0.3) but a significance in HIV+ (p = 0.04) cohorts. In contrast, miR-29a displayed an upregulation trend, particularly in TB-related groups (Fig. 1B). Significant increases were observed in TB-HIV+ (p = 0.03) and HIV+ (p = 0.03) but not for TB+ (p = 0.8) compared to healthy controls. Notably, miR-29a was significantly elevated in TB-HIV relative to TB+ (p = 0.001) and but not for HIV+ (p = 0.9) groups. Expression of lncRNA-COX2 showed modest downregulation in infected cohorts, with strong co-infection specificity (Fig. 1C). Significant reductions were detected in both TB-HIV (p = 0.03) and TB+ (p = 0.03) but not for HIV+ (p = 0.08) relative to healthy controls. No significant inter-group differences emerged between HIV + and TB+ (p = 0.9), HIV + and TB-HIV (p = 0.9), or TB + and TB-HIV (p = 0.9). lncRNA-NEAT1 was markedly upregulated across all infection groups, with the highest levels in TB + and TB-HIV cohorts (Fig. 1D). Highly significant elevations were noted in HIV+ (p < 0.0001), TB+ (p = 0.0002), and TB-HIV+ (p < 0.0001) compared to healthy controls. Inter-group analyses indicated no significant differences between HIV + and TB+ (p = 0.5), HIV + and TB-HIV (p = 0.5), or TB + and TB-HIV+ (p = 0.3), showing a broad induction potentially linked to inflammatory pathways in both infections. lncRNA-GAS5 demonstrated significant downregulation, most evident in co-infection (Fig. 1E). Levels were reduced in HIV+ (p = 0.0004), TB-HIV (p < 0.0001) and TB+ (p = 0.01) relative to healthy controls. Pairwise comparisons showed no significant differences between HIV + and TB+ (p = 0.4), HIV + and TB-HIV+ (p = 0.3) and TB + and TB-HIV (p = 0.3), underscoring a co-infection-exacerbated suppression that may contribute to immune dysregulation.

Fig. 1.

Relative expression of (A) miR-155, (B) miR-29a, (C) lncRNA-COX2, (D) lncRNA-NEAT1, and (E) lncRNA-GAS5 in PBMCs from healthy controls, HIV+, TB+, and TB/HIV + patients (log2 fold change vs. healthy).

To improve clarity, we have summarized the numerical fold changes in Table 1, showing the median log2 fold change (IQR) for each biomarker in the infected groups relative to healthy controls. The table also includes the corresponding median fold change (2^{median log2FC}) for reference.

Table 1.

Median log2 fold changes (IQR) and fold changes for biomarkers in infected groups compared to healthy controls.

Group	miR-155	miR-29a	lncRNA-COX2	lncRNA-NEAT1	lncRNA-GAS5
HIV+	-0.44 (1.74) Fold: 0.74	-0.28 (1.5) Fold: 0.82	1.83 (2.75) Fold: 3.57	2.77 (2.37) Fold: 6.82	-2.14 (1.82) Fold: 0.23
TB+	-2.11 (2.38) Fold: 0.23	1.34 (4.6) Fold: 2.53	1.55 (2.17) Fold: 2.92	2.37 (3.23) Fold: 5.15	-1.73 (2.06) Fold: 0.30
TB/HIV+	-1.10 (1.87) Fold: 0.46	1.50 (1.47) Fold: 2.82	2.09 (3.06) Fold: 4.27	3.53 (3.70) Fold: 11.66	-2.92 (1.52) Fold: 0.13

Correlations between lncRNAs and MiRNAs

Spearman’s correlation analysis was employed to explore relationships between lncRNAs and miRNAs across the study groups, with results visualized in scatter plots (Fig. 2A–F). A positive correlation was observed between lncRNA-COX2 and miR-155 (r = 0.24, P = 0.01), suggesting a potential cooperative regulatory mechanism, possibly linked to inflammatory pathways activated in infection. Conversely, lncRNA-COX2 exhibited a negative correlation with miR-29a (r = -0.38, P = 0.0001), indicating an antagonistic interaction that may reflect divergent roles in immune regulation, with miR-29a potentially suppressed in the presence of elevated COX2 expression. lncRNA-GAS5 showed negative correlations with both miR-155 (r = -0.41, P = 0.0001) and miR-29a (r = -0.15, P = 0.01), suggesting that GAS5 upregulation might suppress these miRNAs, possibly as part of a feedback loop to modulate immune responses. In contrast, lncRNA-NEAT1 displayed no significant correlation with miR-155 (r = -0.05, P = 0.6) but a negative correlation with miR-29a (r = -0.33, P = 0.0008), indicating a selective regulatory influence on miR-29a. These correlation patterns provide insights into the complex regulatory networks involving lncRNAs and miRNAs, which may underpin the differential immune responses observed in HIV and TB infections.

Fig. 2.

Spearman’s correlation analysis for exploring relationships between lncRNA-COX2, lncRNA-NEAT1, lncRNA-GAS5 and miR-29a and miR-155 across the study groups, (A) Healthy and HIV+, (B) Healthy and TB+, (C) Healthy and HIV/TB+, (D) HIV + and HIV/TB+ (E) TB + and HIV/TB+, and (F) HIV + and TB + patients.

Diagnostic performance of individual biomarkers

Receiver operating characteristic (ROC) curve analysis was conducted to evaluate the diagnostic potential of individual biomarkers in distinguishing between study groups (Fig. 3A–F). For the comparison of healthy controls versus HIV+, the area under the curve (AUC) values were as follows: lncRNA-GAS5 (0.78, P = 0.0005), lncRNA-NEAT1 (0.81, P = 0.0001), lncRNA-COX2 (0.70, P = 0.01), miR-29a (0.76, P = 0.001), and miR-155 (0.71, P = 0.008). These moderate to good AUC values suggest that these biomarkers, particularly NEAT1 and GAS5, have reasonable discriminatory power for detecting HIV infection. For healthy controls versus TB+, the AUCs were higher: lncRNA-GAS5 (0.97, P < 0.0001), lncRNA-NEAT1 (0.84, P < 0.0001), lncRNA-COX2 (0.76, P = 0.001), miR-29a (0.58, P = not significant), and miR-155 (0.66, P = 0.007), indicating excellent performance for GAS5 and NEAT1 in identifying TB, though miR-29a showed limited utility. For healthy controls versus TB/HIV+, all biomarkers demonstrated excellent discriminatory power (AUCs ranging from 0.79 to 1.00, P < 0.0001), reflecting their strong potential as markers of co-infection.

Fig. 3.

Receiver operating characteristic (ROC) curve of study factors (miR-155, miR-29a, COX2, NEAT1 and GAS5) for discrimination among (A) Healthy and HIV+, (B) Healthy and TB+, (C) Healthy and HIV/TB+, (D) HIV + and HIV/TB+, (E) TB + and HIV/TB+, and (F) HIV + and TB+.

In pairwise comparisons among infected groups, HIV + versus TB/HIV + yielded AUCs ranging from 0.51 to 0.91 (P = 0.0007 to not significant), with lncRNA-NEAT1 and miR-155 showing the highest values, suggesting some ability to differentiate co-infection from HIV alone. For TB + versus TB/HIV+, AUCs ranged from 0.63 to 0.72 (P = 0.001 to 0.05), indicating moderate discrimination, with lncRNA-GAS5 and miR-155 being most effective. For HIV + versus TB+, AUCs ranged from 0.56 to 0.84 (P = 0.0001 to not significant), with lncRNA-COX2 and miR-29a showing the best performance. These results underscore the variable diagnostic utility of individual biomarkers depending on the specific infection context, with higher AUCs generally observed in comparisons involving healthy controls versus infected groups.

To detail the mentioned correlations between ncRNAs as shown in Fig. 3, Spearman analysis across all samples (n = 120) revealed moderate positive correlation between lncRNA-COX2 and miR-155 (r = 0.24, P = 0.01), moderate negative correlation between lncRNA-COX2 and miR-29a (r=-0.38, P < 0.0001), strong negative correlation between lncRNA-GAS5 and miR-155 (r=-0.41, P < 0.0001), weak negative correlation between lncRNA-GAS5 and miR-29a (r=-0.15, P = 0.01), no significant correlation between lncRNA-NEAT1 and miR-155 (r=-0.05, P = 0.6), and moderate negative correlation between lncRNA-NEAT1 and miR-29a (r=-0.33, P = 0.0008); no correlations were performed with CD4 count or viral load due to incomplete data (available for only 45/60 HIV-infected samples) and lack of significance (r < 0.2, P > 0.1).

Enhanced diagnostic power of combined biomarkers

To enhance diagnostic accuracy, multivariate logistic regression models were developed to assess the combined performance of miR-155, miR-29a, lncRNA-COX2, lncRNA-NEAT1, and lncRNA-GAS5, with probability scores calculated as logit(P) = ln(P/(1-P)) = b0 + b1ΔCt1 + … + bnΔCtn, where bi are regression coefficients and ΔCti are relative expression levels (Fig. 4A–F). For healthy controls versus HIV+, a combination of miR-155, lncRNA-NEAT1, and lncRNA-GAS5 achieved an AUC of 0.95 (P < 0.0001), significantly outperforming individual biomarkers and indicating a robust approach for detecting HIV infection. For healthy controls versus TB+, a trio including miR-29a, lncRNA-NEAT1, and lncRNA-GAS5 yielded an AUC of 0.98 (P < 0.0001), demonstrating excellent discriminatory power for TB detection. For healthy controls versus TB/HIV+, the same three-biomarkers reached an AUC of 1.00 (P < 0.0001), suggesting near-perfect classification of co-infection cases.

Fig. 4.

Receiver operating characteristic (ROC) curve of (A) Forward stepwise (likelihood ratio) (combination of miR-155, lncRNA-GAS5 and lncRNA-NEAT1), (B) Forward stepwise (Conditional) (combination of miR-29a, lncRNA-GAS5 and lncRNA-NEAT1), (C) backward stepwise (likelihood ratio) (combination of miR-155, lncRNA-GAS5 and lncRNA-NEAT1) for discrimination among healthy vs. HIV + patients, healthy vs. TB + and healthy vs. HIV/TB + patients respectively respectively, (D) (Forward stepwise (conditional) (combination of miR-155, miR-29a, lncRNA-GAS5 and lncRNA-NEAT1) for discrimination among HIV + vs. HIV/TB + patients, (E) backward stepwise (conditional) (combination of miR-155, lncRNA-GAS5 and lncRNA-NEAT1) for discrimination among TB + vs. HIV/TB + patients, (F) backward stepwise (likelihood ratio) (combination of miR-155, miR-29a) for discrimination among HIV + vs. TB + patients.

In comparisons among infected groups, the HIV + versus TB/HIV + distinction was improved with a five-biomarker group (miR-155, miR-29a, lncRNA-COX2, lncRNA-NEAT1, lncRNA-GAS5), achieving an AUC of 0.91 (P = 0.0001), reflecting enhanced sensitivity to co-infection-specific changes. For TB + versus TB/HIV+, a three-biomarker group (miR-155, lncRNA-NEAT1, lncRNA-GAS5) resulted in an AUC of 0.74 (P = 0.005), indicating moderate improvement. For HIV + versus TB+, a combination of miR-29a and lncRNA-COX2 yielded an AUC of 0.84 (P < 0.0001), suggesting a specific utility in differentiating these infections. These combined selective groups consistently outperformed individual biomarkers, highlighting the synergistic diagnostic potential of integrating multiple non-coding RNAs. This approach could be particularly valuable for clinical settings where distinguishing between HIV, TB, and their co-infection is critical for tailored therapeutic strategies (Table 2).

Table 2.

Diagnostic performance for all key comparisons (AUC with 95% CI, sensitivity/specificity at Youden cutoff). These metrics were derived from delta Ct values; for combinations, predicted probabilities from logistic regression.

Comparison	Biomarker/Combination	AUC (95% CI)	Sensitivity	Specificity	Cutoff (delta Ct or prob.)
Healthy vs. HIV+	lncRNA-NEAT1	0.81 (0.70–0.90)	0.85	0.72	2.05
Healthy vs. HIV+	lncRNA-GAS5	0.87 (0.77–0.95)	0.80	0.84	-1.50
Healthy vs. HIV+	Combination (miR-155, lncRNA-NEAT1, lncRNA-GAS5)	0.95 (0.89–0.99)	0.90	0.92	0.65
Healthy vs. TB+	lncRNA-NEAT1	0.85 (0.72–0.94)	0.96	0.72	1.96
Healthy vs. TB+	lncRNA-GAS5	0.79 (0.66–0.90)	0.76	0.76	1.27
Healthy vs. TB+	Combination (miR-29a, lncRNA-NEAT1, lncRNA-GAS5)	0.98 (0.93-1.00)	0.92	0.96	0.77
Healthy vs. TB/HIV+	lncRNA-NEAT1	0.62 (0.48–0.75)	0.68	0.60	2.50
Healthy vs. TB/HIV+	lncRNA-GAS5	0.70 (0.57–0.82)	0.72	0.64	-2.00
Healthy vs. TB/HIV+	Combination (miR-155, lncRNA-NEAT1, lncRNA-GAS5)	1.00 (0.98-1.00)	1.00	0.96	0.82
HIV + vs. TB/HIV+	lncRNA-COX2	0.59 (0.45–0.73)	0.64	0.56	3.10
TB + vs. TB/HIV+	miR-29a	0.64 (0.50–0.77)	0.70	0.60	1.80

To provide a more comprehensive evaluation of diagnostic performance, we calculated 95% confidence intervals for the AUCs, as well as sensitivity, specificity, and optimal cut-off values based on the Youden index (maximizing sensitivity + specificity − 1) for key biomarkers in distinguishing healthy controls from TB mono-infection (Table 3). These metrics confirm the strong discriminatory power of lncRNA-NEAT1 and the biomarker combination, with the latter achieving near-perfect classification.

Table 3.

Diagnostic performance of key biomarkers for distinguishing healthy controls from TB mono-infection.

Biomarker	AUC (95% CI)	Sensitivity	Specificity	Optimal cut-off (Youden index)
lncRNA-GAS5	0.79 (0.66–0.90)	0.76	0.76	1.27 (ΔCt)
lncRNA-NEAT1	0.85 (0.72–0.94)	0.96	0.72	1.96 (ΔCt)
Combination (miR-29a, lncRNA-NEAT1, lncRNA-GAS5)	0.98 (0.93-1.00)	0.92	0.96	0.77 (predicted probability)

Discussion

In this study, we profiled the expression of two miRNAs (miR-155 and miR-29a) and three lncRNAs (COX2 [PTGS2-AS1], NEAT1, and GAS5) in PBMCs from individuals with HIV monoinfection, TB monoinfection, TB/HIV co-infection, and healthy controls. Our bioinformatics-driven selection, leveraging KEGG pathway overlaps and miRTarBase-validated interactions, identified these ncRNAs as regulators of 60 shared immune genes (e.g., NF-κB, MAPK, apoptosis pathways), providing a rational basis for their investigation. The qRT-PCR results corroborated these predictions, revealing downregulation of miR-155 across infections and TB-specific modulation of miR-29a, alongside lncRNA upregulation, which aligned with anticipated roles in pathogen-host crosstalk. Correlation networks and ROC/logistic regression analyses further demonstrated synergistic diagnostic utility, with panels achieving AUCs up to 1.00 for co-infection detection. These findings elucidate ncRNA-mediated immune dysregulation, validate bioinformatics insights experimentally, and position these molecules as candidate biomarkers for disease stratification and therapeutic targets in TB/HIV-endemic settings.

The bioinformatics workflow pinpointed 60 overlapping genes between HIV-1 (hsa05170) and TB (hsa05152) KEGG pathways, enriched in immune activation (e.g., TLR4, TNF, MYD88) and apoptosis (e.g., BCL2, CASP3), with transcription factors like POU1F1 overrepresented (adjusted p = 0.0027). miRTarBase analysis confirmed miR-155 targeting 8 shared genes (e.g., AKT1, NFKB1, MYD88; adjusted p = 0.0032), supporting its role in NF-κB-mediated inflammation, while miR-29a targeted 5 (e.g., AKT2/3, BCL2, CASP8; adjusted p = 0.055), implicating it in IFN-γ and autophagy pathways. lncRNA selection via GeneCards and lncHUB co-expression prioritized NEAT1 (score 20.67, co-expressed with MAPK/TLR), GAS5 (score 37.86, linked to T-cell apoptosis), and COX2 (antisense to PTGS2, inflammation regulator). These predictions were robustly validated by qRT-PCR: miR-155 downregulation (log2 fold change − 1.5 to -2.0 across groups, P < 0.05) mirrored its suppressive targeting of pro-inflammatory hubs, consistent with HIV/TB evasion strategies^17,20,45. Similarly, miR-29a’s TB-specific profile (upregulated vs. HIV+, P = 0.003; downregulated vs. controls, P = 0.03) aligned with its predicted modulation of apoptosis/inflammation genes, attenuated in co-infection as per exosomal miRNA alterations in HIV-TB^7,8,18. lncRNA upregulation—COX2 (log2 + 2.5 in TB/HIV+, P < 0.0001), NEAT1 (+ 1.8 uniformly, P < 0.0001), GAS5 (+ 3.0 in TB+, P < 0.0001)—reflected co-expression with shared pathways, with recent lncRNA profiling in HIV-TB confirming differential expression of inflammation-linked transcripts like COX2^37,38. Discrepancies, such as GAS5’s upregulation contrasting some HIV downregulation reports⁴⁶, may stem from TB dominance in co-infection or PBMC-specific dynamics, underscoring the value of integrated bioinformatics-experimental approaches for context-specific validation^2,37.

The consistent downregulation of miR-155 in all infected groups relative to controls aligns with its bioinformatics-predicted role as a negative regulator of antiviral/antibacterial innate immunity, targeting SHIP1/SOCS1 to promote NF-κB activation and IFN-γ production for enhanced macrophage phagocytosis and T-cell responses^47–49. In HIV, miR-155 suppression facilitates viral latency by impairing Th1 polarization and dendritic cell maturation^17,49, while in TB, it attenuates NLRP3 inflammasome activity, reducing IL-1β-driven granuloma formation^20,50,51. While miR-155 is often upregulated in HIV-infected monocytes to promote viral latency, it can be downregulated in CD4 + T cells during active replication or immune exhaustion, highlighting context-dependent roles⁵². Our data, showing no inter-group differences among infections (P > 0.05), suggest a convergent suppressive mechanism via HIV Tat or Mtb ESAT-6-mediated epigenetic silencing^52,53, validating the 8 shared targets identified. Therapeutically, restoring miR-155 via synthetic mimics or AAV-delivered vectors could bolster anti-Mtb immunity in co-infected patients, as preclinical models demonstrate reduced bacterial burden and HIV replication upon overexpression^17,49,54. As a biomarker, miR-155’s moderate standalone AUC (0.71 for HIV vs. controls) improves in panels (e.g., with NEAT1/GAS5, AUC = 0.95), enabling discrimination of HIV + from controls and, to a lesser extent, from TB/HIV+ (AUC = 0.91 in five-ncRNA panel), outperforming single markers in source-limited diagnostics^18,55.

In contrast, miR-29a’s profile exhibited TB-specific downregulation (P = 0.03 vs. controls) with attenuation in co-infection, reflecting its dual bioinformatics-implicated role in IFN-γ-mediated Th1 responses and apoptosis inhibition via targeting BCL2 and DNMT3A^7,8,56. During TB, miR-29a upregulation promotes antimycobacterial autophagy and restricts Mtb survival in macrophages^8,19, but HIV co-infection may blunt this through Vpr-induced suppression, exacerbating immune exhaustion^9,23. This differential expression—higher in TB + vs. HIV+ (P = 0.003)—positions miR-29a as a discriminator between monoinfections, with limited standalone utility for TB (AUC = 0.58) but enhanced in panels (e.g., miR-29a/NEAT1/GAS5, AUC = 0.98 for TB vs. controls; miR-29a/COX2, AUC = 0.84 for HIV + vs. TB+), aligning with meta-analyses of miR-29 as a TB biomarker (sensitivity ~ 80%) even in HIV + cohorts^7–9.

Therapeutically, miR-29a agonists could synergize with ART and ATT to restore IFN-γ signaling and reduce co-infection progression, as evidenced by in vitro studies showing decreased HIV latency and Mtb persistence^14,18.

Among the lncRNAs, COX2 (PTGS2-AS1) displayed marked upregulation in TB + and TB/HIV + groups (P < 0.0001 vs. controls), linked to its antisense regulation of PTGS2 (COX-2), which amplifies PGE2-mediated inflammation and NLRP3 activation^37,38,57,58. In TB/HIV co-infection, this may drive synergistic cytokine storms (e.g., IL-6/TNF-α), impairing granuloma integrity and promoting dissemination^3,37,44. The positive correlation with miR-155 (r = 0.24, P = 0.01) suggests coordinated pro-inflammatory loops, while antagonism with miR-29a (r=-0.38, P = 0.0001) implies COX2-mediated suppression of protective autophagy, consistent with recent differential lncRNA expression in HIV-TB³⁷. As a biomarker, COX2’s AUC = 0.76 for TB vs. controls and utility in HIV + vs. TB + discrimination (part of panels) highlights its TB-specific value^37,38. Therapeutically, COX2 inhibitors (e.g., celecoxib) or siRNA-based knockdown could mitigate hyperinflammation without compromising antiviral responses, as shown in Mtb-infected models where silencing reduced bacterial load^57,58; however, HIV contexts require caution due to potential impacts on latency reservoirs³⁵.

NEAT1’s uniform upregulation across groups (P < 0.0001) underscores its architectural role in paraspeckle formation, sequestering RNA-binding proteins to modulate antiviral ISG expression and apoptosis^26,59,60, co-expressed with predicted MAPK/TLR hubs. In HIV, NEAT1 promotes viral persistence by stabilizing Rev/Rev-responsive element interactions^27,31, while in TB, it enhances macrophage survival and IL-6/IFN-β production^24,30,32,38. The lack of inter-group differences and negative correlation with miR-29a (r=-0.33, P = 0.0008) indicate broad immune activation without specificity, yet its high AUC = 0.84 for TB and inclusion in all high-performing panels (e.g., AUC = 1.00 for co-infection) make it a robust general discriminator from controls^25,28,38. Therapeutically, NEAT1 knockdown via ASOs has shown promise in reducing HIV replication and Mtb-induced inflammation in vitro^27,29, potentially as an adjunct to standard therapies to enhance clearance in co-infected patients²⁶.

GAS5’s upregulation, amplified in TB + and TB/HIV+ (P < 0.0001), acts as a tumor suppressor and immune modulator by sponging miR-21 and glucocorticoid receptor interactions, promoting T-cell apoptosis and restraining hyperinflammation^3,54,61, aligning with its apoptosis-linked co-expression profile. Contrasting some viral downregulation reports⁴⁶, our PBMC data suggest TB-driven induction overrides HIV suppression, possibly via Mtb-induced glucocorticoid resistance^37,39. Negative correlations with miR-155/miR-29a (r=-0.41/-0.15, P = 0.0001/0.01) imply GAS5-mediated feedback to prevent exhaustion. With the highest individual AUC = 0.97 for TB vs. controls and key role in panels (e.g., AUC = 0.95 for HIV; 0.74 for TB + vs. TB/HIV+), GAS5 excels in monoinfection discrimination and co-infection detection^37,41,43. Therapeutically, GAS5 overexpression vectors could restore T-cell function in co-infection, inhibiting HIV via miR-873 sponging and enhancing anti-TB apoptosis^34,35,38; recent analyses affirm its prognostic/therapeutic versatility⁴³. Although lncRNA-GAS5 showed consistent downregulation across all infected groups in our bulk PBMC samples (HIV+: log₂FC = − 2.14, TB+: log₂FC = − 1.73, TB/HIV+: log₂FC = − 2.92; all P < 0.0001), this pattern must be interpreted cautiously due to the known cell-type-specific regulation of GAS5 in HIV and TB. In HIV mono-infection, GAS5 is markedly downregulated in CD4⁺ T lymphocytes, where it acts as a molecular sponge for miR-21, thereby relieving repression of pro-apoptotic and immune activation pathways; persistent GAS5 downregulation is associated with T-cell exhaustion and immune senescence even in virally suppressed individuals⁵⁴. Similarly, in active tuberculosis, GAS5 is downregulated in macrophages and serum of patients, correlating with exaggerated inflammatory responses upon M. tuberculosis infection⁶⁰. These reports align with our observed downregulation in PBMCs. However, opposite upregulation of GAS5 has been documented in certain contexts (e.g., some cancer-associated macrophages or latent TB granulomas), highlighting its dual anti- and pro-inflammatory roles depending on the dominant cell population and disease stage. Because our study used total PBMCs without sorting into T cells, monocytes/macrophages, or other subsets, we cannot empirically resolve whether the observed downregulation predominantly reflects CD4⁺ T-cell loss (in HIV and co-infection) or macrophage responses (in TB). Future studies employing single-cell or sorted-population qRT-PCR/flow-FISH are warranted to dissect cell-type-specific contributions and to determine whether GAS5 signature differs sufficiently between lymphocyte- and myeloid-dominated responses to improve diagnostic specificity in HIV/TB co-infection.

Spearman’s correlations unveiled a regulatory axis—cooperative COX2-miR-155 inflammation, antagonistic COX2/miR-29a/NEAT1-miR-29a suppression, and GAS5-miRNA feedback—mirroring predicted ceRNA networks in infections^5,7,33,37 and supporting bioinformatics-derived interactions (e.g., miRNA-lncRNA sponging of shared targets). These likely underpin co-infection synergy, where HIV attenuates TB-protective miR-29a while lncRNAs amplify persistence³⁷.

ROC and logistic models affirmed biomarker efficacy: individual lncRNAs (GAS5/NEAT1) surpassed miRNAs for vs. controls discrimination, but panels integrated synergies for superior group stratification—e.g., five-ncRNA for HIV+/TB/HIV+ (AUC = 0.91), three-ncRNA for TB+/TB/HIV+ (AUC = 0.74)—aligning with recent lncRNA signatures in HIV-TB cohorts (sensitivity > 90%)^37,41. This multiplex approach outperforms Xpert MTB/RIF (55% sensitivity in HIV-TB)^10,11, offering non-invasive PBMC-based assays for smear-negative/extrapulmonary cases^9,37,52.

Our results advance ncRNA utility in TB/HIV, where co-infection accelerates progression via immune dysregulation^2,3,37. Biomarker panels enable rapid triage, while therapeutic targeting (e.g., miRNA mimics, lncRNA ASOs) could personalize interventions, reducing mortality in high-burden areas^2,4,27,37.

Our study has several limitations that should be considered when interpreting the results. First, while we observed consistent downregulation of lncRNA-GAS5 across infected groups in bulk PBMCs, this may mask cell-type-specific differences, such as downregulation in CD4⁺ T cells during HIV infection (contributing to T-cell apoptosis via miR-21 derepression) versus potential variable expression in macrophages during TB. Without cell-type sorting or single-cell RNA-seq, we could not perform stratified analyses to resolve these context-dependent patterns, limiting our ability to pinpoint the cellular drivers of GAS5 dysregulation. Future studies incorporating flow-sorted populations or single-cell profiling are needed to clarify these mechanisms. Additionally, the lack of a priori power calculation and reliance on post-hoc estimates may limit generalizability for smaller effects; future studies should prospectively power for specific AUC thresholds (e.g., > 0.80). Diagnostic criteria, while standard, may introduce selection bias toward sputum-positive TB cases, potentially missing extrapulmonary forms. Furthermore, sample size (n = 95, co-infection n = 25), potentially limiting power; cross-sectional design omits dynamics; and reliance on bioinformatics without direct functional assays for all interaction; not fully controlled potential effects of other underlying conditions or treatments which could influence RNA expression profiles; region-specific study population warrant validation in broader, multicenter cohorts for generalizability.

This study has addressed novelty since most previous work examined single miRNAs or lncRNAs in isolation, whereas here combine both miRNA and lncRNA signatures and demonstrate synergistic diagnostic performance that outperforms the widely used Xpert MTB/RIF assay in HIV-positive patients. A second innovation is the bioinformatics-driven selection of biomarkers: overlapping genes from HIV-1 and TB KEGG pathways (60 shared genes) were mined for miRNA and lncRNA regulators, leading to the targeted panel of five ncRNAs. Finally, the research provides mechanistic insight through correlation analyses, revealing coordinated networks (e.g., COX2-miR-155 cooperation, GAS5-miRNA feedback) that link ncRNA expression to immune dysregulation in co-infection. Together, these elements constitute a new, non-invasive diagnostic strategy with potential therapeutic implications for high-burden settings.

The ncRNA signatures identified in this study offer a feasible translational route toward clinically deployable assays, particularly in settings where current diagnostics underperform in HIV-associated or smear-negative TB. Given that PBMCs can be obtained through minimally invasive venipuncture and qRT-PCR platforms are already established in most clinical laboratories, these biomarkers could be incorporated as adjunct molecular tests alongside Xpert MTB/RIF to enhance diagnostic resolution. A structured translation pathway would involve sequential steps, including: (i) analytical validation of the ncRNA panel in larger and demographically diverse cohorts; (ii) standardization of assay conditions, such as PBMC processing, RNA extraction, internal reference controls, and threshold cycle (Ct) cut-offs; (iii) development of multiplex qPCR formats to enable simultaneous quantification of miRNA–lncRNA combinations with demonstrated high AUC values; and (iv) clinical evaluation against existing standards of care, particularly in smear-negative, extrapulmonary, and HIV-affected populations where diagnostic uncertainty remains greatest. Integration of such ncRNA-based assays into routine workflows could support earlier detection, stratification of patients, and improved triaging decisions in source-limited settings. Collectively, these steps delineate a realistic pathway from biomarker discovery to diagnostic implementation and underscore the potential of PBMC-derived ncRNAs as clinically meaningful molecular indicators of infection. Future multicenter studies with diverse cohorts should assess longitudinal/prognostic utility^4,25,37. Integrated omics and CRISPR models could dissect mechanisms, paving way for clinical assays^2,5,37,46.

In conclusion, this study elucidates distinct expression profiles of lncRNAs (COX2, NEAT1, GAS5) and miRNAs (miR-155, miR-29a) in PBMCs from individuals with HIV mono-infection, TB mono-infection, TB/HIV co-infection, and healthy controls, as validated through qRT-PCR analyses that corroborated our bioinformatics predictions of their regulatory roles in overlapping immune pathways. Spearman’s correlation analyses further revealed intricate ncRNA-miRNA networks underpinning pathogen-host interactions, while ROC and logistic regression models demonstrated the superior diagnostic accuracy of integrated biomarker panels (e.g., AUCs of 0.95–1.00 for disease detection and differentiation), surpassing individual markers and conventional assays like Xpert MTB/RIF in HIV contexts. These findings not only highlight the mechanistic contributions of ncRNAs to immune dysregulation in TB/HIV co-infection but also underscore their translational potential as non-invasive, multiplex biomarkers for enhanced detection in source-limited settings. Therapeutically, targeting these ncRNAs—such as via miRNA mimics or lncRNA antisense oligonucleotides—could complement antiretroviral and antitubercular therapies to mitigate disease progression, aligning with emerging evidence on dysregulated lncRNAs and miRNAs in co-infected cohorts. Ultimately, ncRNA-based diagnostics and interventions hold promise for revolutionizing precision medicine, reducing the global burden of TB/HIV co-infection, and improving outcomes in high-prevalence regions.

Materials and methods

Ethical approval

The study was approved by the Institutional Review Board of Iran University of Medical Sciences (IR.IUMS.REC.1401.123, October 2023), and all participants provided written informed consent in accordance with the Declaration of Helsinki.

Patient samples

Participants for this cross-sectional study were recruited and followed at infectious disease clinics providing TB and HIV care affiliated with Iran University of Medical Sciences, Tehran between 2023 and 2025. Four groups of participants were included in this study. In total, 95 participants were recruited in this study, including 25 patients with HIV infection, 20 patients with TB infection, 25 with TB/HIV coinfection and 25 healthy volunteers. Inclusion criteria: adults > 18 years with confirmed diagnoses; exclusion: recent (< 6 months) antimicrobial/ART treatment, comorbidities (e.g., diabetes, cancer), or pregnancy. TB was diagnosed by positive sputum smear/GeneXpert MTB/RIF assay (Cepheid) and confirmed by culture; HIV by positive fourth-generation ELISA (Bio-Rad GS HIV Combo Ag/Ab) confirmed by Western blot and viral load > 1000 copies/mL via quantitative PCR (Roche COBAS AmpliPrep/TaqMan).

MiRNA and LncRNAs selection

The bioinformatics analysis was conducted to identify overlapping genes between HIV and TB pathways, predict regulatory miRNAs, and associate lncRNAs with disease pathogenesis. All analyses were performed using publicly available databases and tools, with data retrieved in 2023. The workflow integrates pathway enrichment, gene overlap assessment, miRNA target prediction, and lncRNA association screening.

Retrieval of disease-associated genes

Gene sets associated with HIV infection and TB were retrieved from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (release 2021). Specifically, genes involved in the “HIV-1 infection” pathway (KEGG ID: hsa05170; 212 genes) and the “Tuberculosis” pathway (KEGG ID: hsa05152; 180 genes) were extracted using Enrichr. Gene set enrichment analysis was performed using Enrichr to identify enriched pathways, lncRNA co-expression networks, and transcription factors associated with the differentially expressed genes (miR-155, miR-29a, lncRNA-COX2, lncRNA-NEAT1, lncRNA-GAS5) and their potential targets. Specific database versions queried included KEGG_2021_Human for pathways, lncHUB_lncRNA_Co-Expression_2021 for lncRNA networks, and TRANSFAC_and_JASPAR_PWMs (integrated as of Enrichr’s 2024 collection) for transcription factors. Terms with adjusted p-values < 0.05 (Benjamini-Hochberg correction) were considered significant. Post-Enrichr output, enriched terms were manually filtered for relevance to HIV/TB immunopathology through literature review (PubMed searches for keyword combinations like “ncRNA + HIV/TB + [pathway term]”), prioritizing those with established roles in immune regulation and infection (e.g., Tuberculosis and Human immunodeficiency virus 1 infection pathways). No additional statistical cross-validation was applied, as the analysis aimed to generate hypotheses for further experimental validation.

These genes represent host factors implicated in viral/bacterial pathogenesis, immune modulation, and disease progression.

Identification of common genes

The two gene lists were compared using the online Venn diagram tool to identify overlapping genes (Fig. 5). The analysis revealed 60 shared genes (e.g., IFNA family, MAPK, TLR4, TNF), presumed to be involved in synergistic molecular mechanisms during HIV/TB co-infection. Unique genes were 121 for TB and 153 for HIV. Transcription factor enrichment for these common genes was performed using Enrichr’s TRANSFAC_and_JASPAR_PWMs dataset, highlighting regulators such as POU1F1 (15/1408 genes, adjusted p = 0.0027) and NKX3-1 (6/264 genes, adjusted p = 0.0167).

Prediction of MiRNAs targeting common genes

The 60 shared genes were submitted to Enrichr, selecting the miRTarBase 2017 dataset (updated to 2023 for validation) to identify experimentally validated miRNAs. Only interactions supported by strong evidence (e.g., reporter assays, Western blotting, qPCR) were considered. Enrichment analysis prioritized miRNAs based on overlap, p-value, and odds ratio (e.g., hsa-miR-34a-5p: 13/735 genes, adjusted p-value = 1.99E-04; hsa-miR-155-5p: 8/456 genes, adjusted p-value = 0.0032; hsa-miR-29a-3p: 5/264 genes, adjusted p-value = 0.055). Focus was placed on miR-29a and miR-155 due to their established roles in immune regulation and pathogen response.

Fig. 5.

KEGG pathway analysis confirmed significant overlap between HIV-1 infection and TB pathways.

Identification of lncRNAs associated with HIV and tuberculosis

Keyword-based searches (“HIV”, “AIDS”, “tuberculosis”, “Mycobacterium tuberculosis”) were conducted in the GeneCards database to compile lncRNAs associated with either or both diseases. Top hits included GAS5 (relevance score 37.86) and NEAT1 (relevance score 20.67). Co-expression analysis with the 60 common genes was performed using Enrichr’s lncHUB_lncRNA_Co-Expression dataset, identifying lncRNAs like LINC01255 (4/100 genes, adjusted p = 0.117) but prioritizing those with known immune/inflammatory roles. PTGS2-AS1 (lncRNA-COX2) was included based on its antisense regulation of PTGS2 (COX2) and literature associations with inflammation. Selected lncRNAs (NEAT1, COX2/PTGS2-AS1, GAS5) are known to modulate immune pathways, interact with shared genes, or sponge miRNAs like miR-29a and miR-155.

Preparation of peripheral blood mononuclear cells (PBMCs)

6 mL of peripheral blood samples were collected in syringes and directly transferred to tubes containing Ethylenediaminetetraacetic Acid (EDTA), and was transported to the processing lab at ambient temperature within 2 h of collection. PBMCs were isolated by Ficoll-Paque PREMIUM (GE Healthcare) density gradient centrifugation (400 × g, 30 min at room temperature), with viability > 95% by trypan blue), technique according to the manufacturer’s instructions, and then the pellet of PBMCs was washed three times with phosphate-buffered saline (PBS) solution (pH: 7.3 ± 0.1), and finally resuspended with 250 µL of RNA maintenance solution (RNA-Later, Ambion, Inc., Austin, TX), and kept at -80 °C until extraction of the total RNA.

Total RNA isolation and complementary DNA (cDNA) synthesis

Total RNA was extracted from PBMC samples using TRIzol reagent (Invitrogen) (Thermo Fisher Scientific, Wilmington, NC, USA), according to the manufacturer’s protocols with minor modifications⁶², followed by DNase I treatment (Thermo Fisher) to remove genomic DNA. RNA integrity was assessed by Agilent 2100 Bioanalyzer (RIN > 7.0) and quantity by NanoDrop (Thermo Fisher Scientific, Wilmington, NC, USA), (A260/280 > 1.8). To mitigate bias, sample collection and RNA extraction were performed by technicians blinded to clinical groups, with samples randomized before processing and were kept at -20 °C until use, and then kept at -80 °C until the cDNA synthesis.

To determine the expression pattern of lncRNAs (COX-2, NEAT-1 and GAS-5), cDNA was synthesized using cDNA synthesis kit (Biotechrabbit, Berlin, Germany) according to manufacturer’s protocol.

Expression analysis of genes using real-time PCR

The expression patterns of lncRNAs (COX2, NEAT-1 and GAS5), and also GAPDH as the reference housekeeping gene were determined by real-time polymerase chain reaction (PCR) using a Rotorgene Q thermal cycler (Qiagen, Hilden, Germany) instrument. The assays were done on 15µL reaction mixture including: 1µL of 10 pmol concentration of each primer (COX2, NEAT-1, GAS5 and GAPDH) (Table 1) 4.5µL nuclease free distilled water, 7.5µL 2 × SYBR^® Premix Ex Taq (Tli Plus) Master Mix (TaKaRa Bio Inc. Shiga, Japan) and 1µL of cDNA as template.

Primers were designed with Primer3 software and validated for specificity (sequences: miR-155 forward 5’-GCGGTTACTGCTAATCGTGATA-3’, miR-29a forward 5’-GAGCCTACCACCATCTGAA-3’, universal reverse 5’-CGAGGAAGAAGACGGAAGAAT-3’; lncRNA-COX2 forward 5’-CTCCACGGGTCACCAATATAAA-3’, reverse 5’- ACGCATCAGGGAGAGAAATG-3’; lncRNA-NEAT1 forward 5’- TGGCTAGCTCAGGGCTTCAG-3’, reverse 5’-TCTCCTTGCCAAGCTTCCTTC-3’; lncRNA-GAS5 forward 5’-AGTTGTGTCCCCAAGGAAGG-3’, reverse 5’-CGTTACCAGGAGCAGAACCAT-3’; GAPDH forward 5’- CGACCACTTTGTCAAGCTCA − 3’, reverse 5’- CCCTGTTGCTGTAGCCAAAT − 3’) (Table 4).

Table 4.

.

Sequence (5’ → 3’)	Name	Direction	Real-time PCR based on SYBR-green I fluorescence
CTCCACGGGTCACCAATATAAA	COX21-F	Forward primer	Real time PCR for COX2
ACGCATCAGGGAGAGAAATG	COX2-R	Reverse primer
AGTTGTGTCCCCAAGGAAGG	GAS52-F	Forward primer	Real time PCR for GAS5
CGTTACCAGGAGCAGAACCAT	GAS5-R	Reverse primer
TGGCTAGCTCAGGGCTTCAG	NEAT3-F	Forward primer	Real time PCR for NEAT-1
TCTCCTTGCCAAGCTTCCTTC	NEAT-R	Reverse primer
CGACCACTTTGTCAAGCTCA	GAPDH4-F	Forward primer	Real time PCR for GAPDH
CCCTGTTGCTGTAGCCAAAT	GAPDH-R	Reverse primer
CGAGGAAGAAGACGGAAGAAT	Universal reverse	Reverse	Real time PCR for miR-155 and miR-29a
GCGGTTACTGCTAATCGTGATA	miR-1555F	Forward	Real time PCR for miR-155
GAGCCTACCACCATCTGAA	miR-296F	Forward	Real time PCR for miR-29a
GCTTCGGCAGCACATATACTAAAAT	U67-F	Forward	Real time PCR for U6
CGCTTCACGAATTTGCGTGTCAT	U6-R	Reverse

¹COX2, cyclooxygenase2.

²GAS5, growth arrest specific 5.

³NEAT-1, Nuclear Paraspeckle Assembly Transcript 1.

⁴GAPDH, Glyceraldehyde 3-Phosphate Dehydrogenase.

⁵MicroRNA-155.

⁶MicroRNA-29a.

⁷U6 snRNA.

The thermocycling conditions for real-time PCR for COX2 is defined as follows: initial denaturing at 95 °C for 10 min, and 40 cycles, including 15 s at 95 °C, 30 s at 58° C, and 20 s at 72 °C, for the NEAT1 and GAS5 were defined as follows: initial denaturing at 95 °C for 10 min, and 40 cycles, including 15 s at 95 °C, 30 s at 60° C, and 20 s at 72 °C; and for GAPDH initial denaturing at 95 °C for 15 min, and 40 cycles, including 15 s at 95 °C, 30 s at 60° C, and 20 s at 72 °C. Data acquisition from all above was sat at the extension step. All the specimens were tested in duplicate reactions.

Expression was normalized to GAPDH for lncRNAs and to U6 for miRNAs (selected for stability across groups via geNorm analysis, M < 0.5). Relative quantification was performed using the ΔΔCt method, with healthy controls as the reference. Ct values > 35 were excluded; all samples had Ct < 32 for the targets. No-template and no-RT controls were negative.

For statistical analysis group comparisons used Mann-Whitney U tests (GraphPad Prism 8.2). No a priori power calculation was conducted, but post-hoc analysis (G*Power 3.1.9.7) for detected effects (e.g., lncRNA-GAS5 in HIV + vs. healthy, effect size d ≈ 1.2 from median difference/SD) at α = 0.05, n = 25 per group, yields power > 0.95; for minimal detectable AUC > 0.80 (binormal approximation), power = 0.85 assuming equal n = 25 and prevalence 0.5. P values are reported uncorrected in figures; after FDR correction (Benjamini-Hochberg) for 15 tests (5 analytes × 3 comparisons), all significant findings (original P < 0.01) remain at FDR < 0.05. Multiple analysts reviewed data independently to reduce interpretation bias.

miRNA-155 and miR-29a expression analysis

Total RNA was extracted (as described in the previous section) from PBMC samples. Synthesis of cDNA was performed as previous section as manufacturer’s protocol with the difference on using specific RT stem-loop primers of each miRNA (Table 5).

Table 5.

.

Sequence (5’ → 3’)	Name
GAAAGAAGGCGAGGAGCAGATCGAGGAAGAAGACGGAAGAATGTGCGTCTCGCCTTCTTTCACCCCAA	miR-155RT
GAAAGAAGGCGAGGAGCAGATCGAGGAAGAAGACGGAAGAATGTGCGTCTCGCCTTCTTTCAGCCAAT	miR-29aRT

The real time PCR assay was carried out in final 15µL volume, including 1µL of specific forward primer, 1µL of universal reverse primer, 7.5µL of SYBR Green PCR Master Mix (TaKaRa, Kusatsu, Japan), 4.5µL of nuclease-free water, and 1µL of cDNA as template. The thermal profile of this assay (three steps with melt) was set at 95 °C for 10 min as hold time, followed by 40 cycles of denaturation at 95 °C for 15 s, annealing at 58 °C for 30 s, and extension at 72 °C for 20 s. This assay was performed using the Rotorgene Q thermal cycler (Qiagen, Hilden, Germany) instrument. The expression levels of microRNAs were normalized to U6 as reference RNA. It should be noted that all reactions were done in duplicate (Table 1).

Author contributions

All authors reveiwed the manuscript and took parts in writing the main manuscript.

Funding

This research was funded by the Vice-Chancellor for Research, Iran University of Medical Sciences, Tehran, Iran (grant number: IR.IUMS.REC.1402.630).

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.