A total of 186 patients with HIV (age, 18-86 years; 61.8% male and 38.2% female) attending the Antiretroviral Therapy Clinic in Nakorn Nayok Hospital, Nakorn Nayok (Thailand) from April 2014 to October 2020 were recruited into the present study. The inclusion criteria were as follows: i) Patients aged >18 years with documented HIV infection; ii) receiving suppressive ART for >6 months and iii) available blood samples and clinical data. Patients who consumed alcohol, herbal medicine and steroidal drugs, in addition to patients with active opportunistic infections, were excluded. The patients were divided into 150 HIV-moninfected patients, aged 18-86 years, 58.7% male and 41.3% female, 19 HBV-coinfected HIV patients, 29-58 years, 73.7% male and 26.3% female, and 17 HCV-coinfected HIV patients, aged 20-62 years, 76.5% male and 23.5% female. In total, 50 uninfected controls, aged 20-60 years, 40% male and 60% female, were recruited from Police General Hospital, Bangkok (Thailand) in November 2022. Inclusion criteria for controls were individuals aged >18 years with available blood samples and clinical data who were seronegative for HIV, HBV and HCV infection. Individuals with notable liver fibrosis, determined by fibrosis-4 (FIB-4) score >1.45 or aspartate aminotransferase (AST)-to-platelet ratio index (APRI) >0.5, were excluded (24-26). The present study protocol was approved by the Human Ethics Committee No. 3, Thammasat University, Pathumthani (Thailand; approval no. 116/2565) and the Certificated Biological Safety Committee of Thammasat University, Pathumthani, (Thailand, approve no. 079/2565).

Clinical and laboratory data of the HIV group were obtained and characterized as described in previous studies (22,23). Briefly, the data collected from the medical records of the patients were age, sex, CD4⁺ cell count, HBV/HCV coinfection, levels of AST and alanine aminotransferases (ALT), platelet count, HIV viral load and ART regimen and duration. Anti-HIV, HBV surface antigen and anti-HCV tests were performed to recruit seronegative HIV, HBV and HCV-uninfected controls. The extent of liver fibrosis in HIV-infected patients and uninfected controls was evaluated by measuring non-invasive markers FIB-4 and APRI, which are commonly used scoring systems recommended for assessing liver fibrosis in the presenceof chronic HCV by the World Health Organization (27). FIB-4 score >1.45 or APRI >0.5 was considered to be significant liver fibrosis (24-26). Additionally, levels of fibrotic markers, namely laminin (LN), procollagen type III N-terminal peptide (PIIINP), hyaluronic acid (HA) and type IV collagen (IVC), in addition to tumor marker α-fetoprotein (AFP), were measured using electrochemiluminescence immunoassay according to the maufacturer's instructions (Mindray CL-900i using LN, PIIINP, HA, IVC and AFP test kits (cat. nos. LV111, PIIINP111, HA111, C IV111 and AFP111, respectively; all Mindray Medical International Co., Ltd.).

EDTA-blood samples left over from routine examination were collected from the clinical laboratory of Nakorn Nayok Hospital. Plasma was separated within 8 h of collection by centrifuging at 2,000 x g for 10 min at room temperature and stored at -80˚C until further use. Plasma samples from 186 patients infected with HIV were classified into the following three groups: HIV-monoinfection (n=150), HIV/HBV coinfection (n=19) and HIV/HCV coinfection (n=17). Blood samples from seronegative HIV, HBV and HCV individuals (n=50) with FIB-4 score ≤1.45 and APRI ≤0.5 were designated as uninfected controls. All samples were then subjected to proteomic analysis (Fig. 1).

Total protein concentration in plasma samples was measured using the Lowry method, with bovine serum albumin applied as the standard (28). Protein samples were subjected to in-solution digestion. The samples, which were dissolved in 10 mM ammonium bicarbonate (AMBIC), underwent disulfide bond reduction using 5 mM dithiothreitol in 10 mM AMBIC at 60˚C for 1 h and alkylation of sulfhydryl groups using 15 mM iodoacetamide in 10 mM AMBIC at room temperature for 45 min in the dark, followed by digestion in porcine trypsin (Promega Corporation) in a 1:20 ratio at 37˚C overnight. Peptides were dried and resuspended with 0.1% formic acid prior to nano-LC-MS/MS analysis. The LC-MS/MS analysis was performed using an Ultimate 3000 Nano/Capillary LC System (Thermo Fisher Scientific, Inc.) coupled to a ZenoTOF 7600 mass spectrometer (SCIEX). Peptide separation was performed using a 75 µm x 15 cm column, packed with Acclaim PepMap RSLC C18 resin (2 µm, 100Å, nanoViper, Thermo Scientific, UK). The C18 column was maintained at 60˚C within a thermostatted column oven. Solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in 80% acetonitrile) were delivered to the analytical column. A gradient elution of 5-55% solvent B was employed to separate peptides at a constant flow rate of 0.30 µl/min over a 30-min period.

The electrospray ionization source was set in positive ion mode with an ion source temperature of 200˚C, spray voltage of 3300 V, curtain gas (N₂) of 35 psi, nebulizer gas of 50 psi, auxiliary heating gas of 55 psi, and declustering potential of 80 V. Top 50 most abundant precursor ions per survey MS1 for MS/MS with an intensity threshold exceeding 150 cps were selected. Precursor ions were dynamically excluded for 8 sec after two incidences of MS/MS sampling (with dynamic collision energy enabled). The MS2 spectra were collected in the range 350-1,800 m/z with a 250-ms accumulation time and Zeno trap enabled. The collision energy parameters included a declustering potential of 80 V, no DP spread, and a CE spread of 0 V. The time bins were summed (with all channels enabled) using a 150,000 cps Zeno trap threshold. The cycle time for the Top 50 DDA method was 3.0 sec. To minimize the effect of experimental variation, three independent LC-MS runs were performed for each sample.

Proteins were quantified using MaxQuant 2.1.4.0 (Max-Planck Institute for Biochemistry) using the Andromeda search engine to match MS/MS spectra to the Uniprot Homo sapiens database (uniprot.org/) (29). Label-free quantification with MaxQuant's standard settings was performed as follows: i) Maximum of two missing cleavages; ii) mass tolerance of 0.6 Da for main search; iii) trypsin as a digesting enzyme; iv) carbamidomethylation of cystein as fixed modification and v) oxidation of methionine and acetylation of the protein N-terminus as variable modifications. Peptides with ≥7 amino acids and ≥1 unique peptide were chosen for protein identification. Only proteins with ≥2 two peptides and ≥1 unique peptide were considered as being identified and subjected to further analysis. The false discovery rate (FDR) was set at 1% and estimated using the reversed search sequences. The maximal number of modifications/peptide was set to 5. The proteins present in the H. sapiens proteome were then downloaded from Uniprot as a search FASTA file. Potential contaminants in the contaminant FASTA files were added to the search space using the MaxQuant software.

MaxQuant ProteinGroups.txt file was loaded into Perseus software version 1.6.6.0(29) and potential contaminants that did not correspond to any Universal Proteomics Standard Set 1 (UPS1) protein were removed. Mass intensities were log₂ transformed before pairwise comparisons using unpaired t tests. Missing values were imputed with a constant value (0) using the Perseus software. Intensity values of the MS/MS spectra were analyzed using partial least squares-discriminant analysis (PLS-DA) model, one-way ANOVA (P-value and FDR <0.05) and Kruskal-Wallis test using web-based tool MetaboAnalyst version 6.0 (metaboanalyst.ca/) (30,31). Heatmap analysis was performed by using the Morpheus online tool (software.broadinstitute.org/morpheus/) (32). Molecular function and subcellular localization of identified proteins were obtained from Gene Ontology (GO; geneontology.org/) and Uniprot databases (uniprot.org/). A protein-protein interaction network was constructed according to the STITCH 5.0 database (stitch.embl.de/).

Descriptive statistics were conducted to analyze characteristics of the study groups. Continuous variables are presented as median (range); categorical variables are reported as n (%). χ² and Fisher's exact tests were used to determine the association between viral infection status and categorical variables, whereas the Kruskal-Wallis and Dunn's post hoc tests were performed to compare the median of continuous variables between study groups. P<0.05 was considered to indicate a statistically significant difference. SPSS for Windows version 18.0 (SPSS, Inc.) and GraphPad Prism 9.1.1 (Dotmatics) were used for the statistical analysis.

Table I demonstrates general and clinical characteristics of each group. All patients infected with HIV received suppressive ART for >6 months and had no detectable HIV viral load (data not shown), whereas no specific medications for HBV and HCV infection were used. All study groups had similar median ages, infected groups showed the higher male-to-female sex ratio than the uninfected control group. Although platelet counts of the three HIV-infected groups and uninfected group were not significantly different, medians of liver enzymes AST and ALT, in addition to the fibrosis markers FIB-4 and APRI, were significantly different. The additional analysis also indicated that the median values of AST, ALT, FIB-4 score and APRI in the three HIV-infected groups compared with the uninfected group differed significantly, excluding the FIB-4 score of HIV-mono-infection group. Differences in immune status and ART between the HIV-infected groups were also tested. Median levels of CD4⁺ cell count and duration of ART in the HIV-monifected, HIV/HBV and HIV/HCV coinfected groups, were compared, yielding no statistical difference. ART drug regimens used in these three infected groups were not found to be significantly different (Table I).

There were higher FIB-4 scores and APRI in the three HIV-infected groups compared with the uninfected control group, except for the FIB-4 score in the HIV-monoinfected group (Fig. 2A and B). In addition, significantly higher FIB-4 score and APRI in the HIV/HCV group compared with the HIV-monoinfection group were observed. Key components of the extracellular matrix that typically accumulate during liver fibrosis, namely LN, IVC, PIIINP and HA, coupled with the tumor marker AFP (33), were examined in the plasma samples. HIV groups also had significantly higher levels of LN and IVC compared with the uninfected control group, except for LN levels in the HIV-monoinfection group (Fig. 2C and D). Consistent with FIB-4 score and APRI, LN and IVC levels were found to be significantly higher in the HIV/HCV group compared with HIV monoinfection. Notably, elevated plasma PIIINP, HA and AFP levels in the infected groups was not observed, whereas the HIV/HCV group had significantly higher levels of all of these markers compared with the HIV-monoinfection group (data not shown).

Shotgun proteomics was conducted to explore the plasma protein profiles. A total of 1,074 proteins were differentially expressed. PLS-DA model demonstrated clear separation of the proteome between sample groups, indicating significant differences in the plasma protein profiles (Fig. 3A). Of these, 18 were significantly differentially expressed (Fig. 3C; Table II). PLS-DA model of 18 selected proteins also indicated a clear clustering of proteins from the four sample groups (data not shown). Analysis for localization and function of the 18 proteins was conducted using GO and Uniprot databases (Fig. 3D and E). These proteins were primarily located in the nucleus (32.4%) and cytoplasm (29.4%) and in other intracellular and extracellular compartments (Fig. 3E). The cellular and molecular functions were enriched in molecular processes, including ‘cell cycle, cell proliferation, cell differentiation and cell death’ (15.4%), ‘cell signaling’ (11.5%), ‘RNA processing and transport’ (11.5%), ‘binding protein’ (11.5%), ‘extracellular matrix protein and cytoskeleton’ (11.5%), ‘transcription’ (7.7%), ‘protein transport and degradation’ (7.7%), ‘cell adhesion and cell migration’ (7.7%), ‘inflammation, immune response and anti-viral activity’ (7.7%) ‘DNA replication and repair’ (3.8%) and ‘metabolic process’ (3.8%; Fig. 3D).

Heatmap analysis demonstrated differential expression of the 18 selected proteins (Fig. 3B). A total of 10 proteins was downregulated in the HIV-monoinfected, HIV/HBV and HIV/HCV groups compared with those in the uninfected control group, included haptoglobin (HP), HBV X-transactivated gene 13 protein, forkhead-associated domain-containing protein 1, tenascin-X, RNA-binding protein 34, serine/threonine-protein kinase 31, E3 ubiquitin-protein ligase tripartite motif-containing 56, Mps one binder kinase activator 3C, E3 ubiquitin-protein ligase ring finger protein 220 and BRCA1-A complex subunit RAP80; six proteins were upregulated in the HIV/HBV group compared with the uninfected control group (116 kDa U5 small nuclear ribonucleoprotein component, nuclear cap-binding protein subunit 3, transmembrane channel-like protein 6, DNA replication ATP-dependent helicase/nuclease DNA 2, unconventional myosin-V and calpain-6). In addition, dual specificity calcium/calmodulin-dependent 3',5'-cyclic nucleotide phosphodiesterase was downregulated only in the HIV/HBV group and microtubule-associated protein 2 was upregulated only in the HIV-monoinfection group. The differential expression of these proteins suggests their potential roles in HIV monoinfection and coinfection with HBV and HCV.

To identify the molecular function of plasma protein profiles potentially associated with HIV infection and coinfection with HBV and HCV under long-term ART, the 18 selected proteins were analyzed for their interaction with proteins and chemicals in the STITCH 5.0 database (Fig. 4). Since accumulating evidence has indicated key roles of the profibrogenic TGF-β/SMAD and inflammatory TNF pathways in liver fibrogenesis and inflammation, especially in HIV and hepatotropic HBV and HCV (18,21,34,35), the proteins TGF-β, SMAD, TNF and ART drugs used in the patient groups (zidovudine, lamivudine, nevirapine and tenofovir) were included in the STITCH model. Input proteins serine/threonine kinase 31, chromosome 18 open reading frame 21, ring finger protein 220, myosin VA, tenascin XB, forkhead-associated, phosphopeptide binding domain 1, transmembrane channel-like 6, RNA binding motif protein 34, calpain 6, MOB kinase activator 3C, tripartite motif containing 56, microtubule-associated protein 2, nuclear cap binding protein subunit 1 and nuclear cap binding protein subunit 2 did not form any interaction with the protein networks. However, the model predicted close interactions between ubiquitin interaction motif containing 1 (UIMC1), also named BRCA1-A complex subunit RAP80, and BRCA1 and TGF-β/SMAD pathways. In addition, an interaction between haptoglobin (HP) and the TNF pathway was found. Notably, no protein exhibited interactions with ART drugs. Additionally, box plots indicated that the expression levels of UIMC1 and HP were 1.6-2.4 and 4.0-5.5 times lower in the three HIV-infected groups compared with those in the uninfected control group, who had no significant liver fibrosis, respectively. These results suggested the potential involvement of BRCA1-linked TGF-β/SMAD and TNF pathways during HIV-monoinfection and coinfection with HBV and HCV, in PLWH receiving suppressive ART.