Background
Chemokines are small proteins and the largest subfamily of cytokines that act on G protein-coupled receptors (GPCRs) with seven transmembrane domains [
1]. Chemokines are classified into four subfamilies according to their structural characteristics: CXC, CC, CX3C, and XC [
2]. To date, 17 CXC-chemokines, 28 CC chemokines, 2 XC chemokines, 1 CX3C chemokine, and approximately 20 chemokine receptors have been identified [
3,
4]. Plasma chemokines are primarily secreted by monocytes, T cells, dendritic cells (DCs), and some epithelia [
3]. Chemokines can attract immunocytes to areas of inflammation, infection, and tissue damage [
5], and are critical in the development and differentiation of immune cells. Additionally, chemokines contribute to the pathogenesis of multiple diseases, including autoimmune disorders, hypersensitive reactions, cancers, and viral infections [
1,
6‐
8]. A chemokine panel was used to screen plasma samples from patients with colorectal cancer, and five chemokines and four cytokines identified as associated with increased total mortality, among which TNF-α and CCL24 were exclusively associated with colorectal cancer-specific mortality [
9]. Further, Weisheng et al. [
10] identified a 14-biomarker panel (chemokines and cytokines) for detecting endometriosis. Hence, chemokine panels can be valuable for disease diagnosis and/or progression prediction.
Primary HIV infection (PHI) refers to the early period of HIV infection (approximately the first 12 weeks after infection) [
11], during which the virus disseminates from the original infection site to different tissues and organs. Subsequently, numerous events occur, including establishment of a viral set-point, which determines subsequent HIV progression [
12‐
16]. Different immune cell subsets are activated in response to HIV and secrete large quantities of cytokines and chemokines potentially associated with disease progression, opportunistic infection, and increased mortality [
17‐
20].
Investigators have focused on the immunological role of chemokines in early HIV infection. During acute simian immunodeficiency virus (SIV) infection, levels of MCP-1, MIP-1α, and MIP-1β can distinguish progressive and non-progressive SIV infection in
Chlorocebus sabaeus [
21]. Further, seven human plasma chemokines were assessed, and fold-change in CXCL10 (HIV
+ vs. HIV
− plasma level) was significantly higher in HIV rapid progressors, with CXCL10 level during PHI negatively correlated with CD4
+ T-cell counts at the 4-month-infection point [
22]. Additionally, 15 cytokines and 1 chemokine (CXCL10) are present at higher levels in rapid relative to slow disease progressors during acute HIV-1 infection [
23]. Furthermore, the combination of IL-12p40, IL-12p70, IFN-γ, IL-7, and IL-15, but not chemokines, could predict HIV disease progression in women with acute HIV-1 infection [
24]. Overall, cytokine levels during HIV infection have been studied; however, only 6–8 chemokines were generally assessed, hence the magnitude of alterations in the majority of chemokine profiles during PHI remain unknown. Further, studies usually compare concentrations of chemokines in samples from HIV-positive and healthy HIV-negative individuals, and high-within-person-variability can result in measurement errors. Changes in chemokine profiles pre- and post-HIV infection during PHI in the same individual may more accurately represent disease conditions.
Here, we used 108 plasma samples collected from 54 patients at two sampling points (pre- and post-PHI) to determine alterations in profiles of 30 chemokines and 10 cytokines between the two sampling points. Furthermore, we analyzed the relationship between chemokine concentrations and disease progression. Finally, we developed the combination of CXCL9, CXCL10, and CXCL11 levels during PHI as a biomarker to predict HIV disease progression.
Methods
Study participants
We set up a prospective open cohort study in the Key Laboratory of AIDS Immunology of the National Health and Family Planning Commission [The First Affiliated Hospital of China Medical University (CMU)], which to date includes > 2000 men who had sex with men (MSM) high-risk study participants, who were HIV-negative when they were enrolled, and all of whom have been screened for HIV infection every 1–3 months. Among newly-diagnosed participants, 54 participants with available clinical information and blood samples from two time points; pre-HIV infection (HIV seronegative and HIV RNA-negative) and post-HIV infection (HIV seropositive or HIV RNA-positive), were selected; the estimated time of HIV infection ranged from 13 to 155 days (Table
1). The estimated time of infection was defined as previously described [
25]. Briefly: (i) by referring to Fiebig stage [
26]; (ii) if the patient could clearly recall the time of high-risk exposure, that time point was the estimated infection time [
25]; (iii) the midpoint between the last time point of HIV antibody negative test and the first HIV antibody positive test was the estimated infection time [
25]. On collection, all plasma samples were immediately stored at − 80 °C until use. After diagnosis with HIV infection, the 54 participants were followed up for an average of 1745 days (range from 7 to 3431 days). All clinical study protocols were approved by the Ethics Review Committee of The First Affiliated Hospital of China Medical University, Shenyang, P. R. China, and the study was conducted according to the principles of the Declaration of Helsinki ([2018] 2015-140-5).
Table 1
Patient demographic and clinical characteristics
Sociodemographic variables |
N | 54 |
Sex | Male (54/54) |
Age, years [median (IQR)]a | 32.5 (28, 41) |
< 20 | 0 |
20–30 | 19 |
30–40 | 20 |
40–50 | 10 |
> 50 | 5 |
Unknown | 4 |
Marital status [n/total (%)] |
Unmarried | 45 (83%) |
Married | 6 (11%) |
Divorced | 3 (6%) |
MSM | 54/54 |
Chinese | 54/54 |
Ethnicity | Han (45/54) |
Man (7/54) |
Other (2/54) |
Clinical |
HBV Ag | N/A | 4/54 |
HCV Ab | N/A | 1/54 |
TPPA | N/A | 26/54 |
Fiebig staging |
I–II | – | 14 |
III–IV | – | 16 |
V–VI | – | 24 |
Estimated infection day (days) [median (IQR)]a | − 91 (− 141, − 56) | 31 (25, 49) |
Methods for estimating infection day (N) |
i. Fiebig stage | N/A | 6 |
ii. Remembered time of high-risk exposure | N/A | 24 |
iii. Laboratory HIV antibody detection | N/A | 24 |
Follow-up (days) [median (IQR)]a | N/A | 1719 (1142, 2625) |
CD4+T-cell count (cells/μL) [median (IQR)]a |
Chemokine detection sampling point | N/A | 424 (328, 531) |
Set-point | N/A | 488 (390, 634) |
One-year-infection pointb | N/A | 412 (328, 555) |
Viral load (copies/mL) [median (IQR)]a |
Chemokines detection sampling point | N/A | 78,500 (11,370, 753,000) |
Set-point | N/A | 24,900 (8450, 83,050) |
One-year-infection pointb | N/A | 22,887 (8482.5, 36,600) |
Viral subtype [N/total] | – | CRF-01AE (53/54) |
CRF-01AE/BC (1/54) |
Detection of CD4+ T-cell count and HIV viral load (VL)
The FACSCalibur (BD, Franklin Lakes, New Jersey, USA) flow cytometer were used to measured absolute blood CD4+ T-cell counts (cells/μL). The levels of plasma HIV-1 RNA (copies/mL) were detected by the COBAS Ampliprep/COBAS TaqMan 48 Analyzer (Roche Diagnostics, Branchburg, New Jersey, USA).
Plasma chemokine/cytokine detection
Plasma chemokine/cytokine levels were determined using a Bio-Plex Pro™ Human Chemokine Panel 40-Plex (BIO-RAD), including CXCL1/Gro-α, CXCL2/Gro-β, CXCL5/ENA-78, CXCL6/GCP-2, CXCL8/IL-8, CXCL9/MIG, CXCL10/IP-10, CXCL11/I-TAC, CXCL12/SDF-1A+β, CXCL13/BCA-1, CXCL16/SCYB16, CCL1/I-309, CCL2/MCP-1, CCL3/MIP-1α, CCL7/MCP-3, CCL8/MCP-2, CCL11/Eotaxin, CCL13/MCP-4, CCL15/MIP-1δ, CCL17/TARC, CCL19/MIP-3β, CCL20/MIP-3α, CCL21/6Ckine, CCL22/MDC, CCL23/MPIF-1, CCL24/Eotaxin-2, CCL25/TECK, CCL26/Eotaxin-3, CCL27/CTACK, CX3CL1/Fractalkine, GM-CSF, MIF, TNF-α, IFN-γ, IL-1β, IL-2, IL-4, IL-6, IL-10, and IL-16. Chemokine/cytokine standards supplied by the manufacturer were detected in duplicate and run on each plate. All data were acquired using a Bio-Plex 200 System (BIO-RAD). Chemokine/cytokine levels below/above the minimum/maximum thresholds of detection are reported as the minimum/maximum threshold values for each chemokine/cytokine.
Statistical analysis
Statistical analyses were performed using GraphPad Prism version 6.0 and SPSS Statistics 20.0. All tests were two-tailed and
p values < 0.05 were considered significant. The nonparametric Wilcoxon matched pairs test was used to evaluate differences in chemokine/cytokine levels from the same individual at different time points (i.e., pre-and post-HIV infection). The nonparametric Mann–Whitney U test was used to compare between-group distributions. Spearman correlation analysis was used to estimate correlation. Kaplan–Meier curves were used to estimate the time to an outcome (CD4
+ T-cell counts ≤ 500 or VL > 10
4 copies/mL). The predictive value of the combination of CXCL9, CXCL10, and CXCL11 levels for HIV disease progression was estimated using receiver operating characteristic (ROC) curves. Predictive values were expressed as the area under the curve (AUC). For ROC analysis of CXCL9, CXCL10, and CXCL11 combinations,
p (probability of a patient sample) was calculated for inclusion in the ROC analysis using the formula:
$$X = logit\left( p \right) = \ln \left( {\frac{p}{1 - p}} \right) = b_{0} + b_{1} \Delta CT_{1} + b_{2} \Delta CT_{2} + b_{3} \Delta CT_{3} \ldots + b_{n} \Delta CT_{n}$$
where b
i terms were the ith regression coefficients by binary logistic regression, and ∆CT
i terms were the relative expression levels of each chemokine [
24]. In this study, a CXCL9, CXCL10, and CXCL11-chemokine combination panel (X = − 1.724 + 0.007 × CXCL9 + 0.004 × CXCL10 − 0.033 × CXCL11, and
p = e
X/(1 + e
X)) was used predict disease progression.
Discussion
Both determining the exact HIV infection time and obtaining plasma samples from patients with PHI are challenging. In this study, we identified 54 HIV-infected patients with well-documented dates of infection and report comparisons of chemokine and cytokine levels between samples obtained from the same individuals pre-HIV infection and during PHI. During PHI, 16 of 30 chemokines exhibited significant changes post-HIV infection in the same individuals; 12 up-regulated and four down-regulated. We observed that expression levels of the CXC-chemokines, CXCL9, CXCL10, and CXCL11, were dramatically enhanced in plasma, with substantial percentage changes in their levels during PHI. These selective proinflammatory chemokines elicit their biological functions by interacting with a common receptor, CXCR3, a seven-transmembrane GPCR highly expressed by activated lymphocytes, such as CD4
+ T helper (Th) cells, CD8
+ cytotoxic T lymphocytes, natural killer (NK) cells, and DCs [
27‐
29]. We also demonstrate that CXCL9, CXCL10, and CXCL11 have potential as novel biomarkers associated with HIV disease progression. Based on our observations, we propose that the combination of CXCL9, CXCL10, and CXCL11 during PHI is a useful biomarker for prediction of HIV disease progression.
Chemokines are crucial players in regulating lymphocyte functions during inflammatory processes [
30]. CXCL9, is also induced by IFN-γ in macrophages, implicated in cancer inflammation and viral infections, and participates in T-cell trafficking, chemotaxis, and activation [
31‐
38]. In the present study, the percentage change in plasma CXCL9 was the largest among the chemokines screened. Although levels of CXCL9 are also elevated in oral mucosal, intestinal mucosa, semen, and decidual tissue in vivo and in vitro during HIV infection [
39‐
42], the mechanism by which CXCL9 becomes elevated is unclear. HSV induces CXCL9 expression on human epithelial cells by activating the p38-CCAAT/Enhancer-Binding Protein-β pathway [
38], which is a possible mechanism underlying the change in CXCL9 levels during HIV infection. Further, our results show that CXCL9 levels are positively correlated with VL at the sampling point and that patients with high VL (lgVL ≥ 4.5) at the sampling and set-points had higher CXCL9 levels than those with IgVL > 4.5. Elevated CXCL9 is also associated with SIV disease progression and decreased phagocytic activity of mucosal macrophages, which prevents the elimination of bacterial antigens in the small intestine, and induces immune activation [
39,
42,
43]. Moreover, blocking CXCL9 can decrease HIV replication in cervical tissues [
42,
43]. Additionally, Lajoie et al. [
44] found that significantly lower expression of CXCL9 in the genital mucosa was associated with strong protection against HIV infection in HIV-exposed seronegative sex workers. Thus, CXCL9 may be a particularly important factor in relation to VL after HIV infection. Further, during the HIV infection entry process, cortical actin is a physiological barrier to HIV, and HIV uses gp120-CXCR4 signaling to active cortical actin and overcome this restriction [
45]. Similarly, CXCL9-CXCR3 signaling may also activate actin to promote HIV entry and post-entry processes, impacting the pace of disease progression. We speculate that CXCL9-induced actin-related signaling may explain its negative function in HIV infection. Moreover, our results are the first to show that levels of CXCL9 correlate negatively with CD4
+ T-cell count at the 1-year-infection point and we speculate that this may be because, as CXCL9 levels are positively correlated with VL, and VL is negatively correlated with CD4
+ T-cell count [
46], CXCL9 may negatively influence CD4
+ T-cell count. Furthermore, a study of adults with HIV in Mozambique found that, although levels of some factors were up-regulated during the first month of HIV infection and rapidly decreased in the subsequent months, CXCL9 levels increased and remained high [
47]. Thus, the alteration of CXCL9 in the plasma milieu may be a long-term determinant of CD4
+ T-cell count.
CXCL10, also known as interferon γ-induced protein 10 (IP-10), similar to CXCL9, can be secreted by various cells, including monocytes, leukocytes, endothelia, and epithelia [
48], and is up-regulated in numerous diseases, including hepatitis B, tuberculosis, cancer, diabetes, and autoimmune disorders [
49‐
53]. Here, we demonstrated that CXCL10 was elevated during HIV infection and correlated with VL at the sampling point. Additionally, we found that CXCL10 could predict VL at set-point. Some studies have reported that systemic levels of CXCL10 during PHI are positively associated with VL and negatively associated with CD4
+ T-cell count [
22,
54,
55]; however, there have been no previous reports that CXCL10 levels in plasma during PHI can predict VL at set-point. CXCL10 is up-regulated during HIV infection, and may suppress IFN-γ secretion and T and NK cell cytotoxicity [
55,
56], potentially partially explaining immune system dysfunction during HIV infection. Cecchinato et al. [
57] found that CXCR3
+ Th cell migration in response to CXCL10 was impaired after HIV infection, and could be rescued by modulating actin polymerization. We speculate that high levels of CXCL10 cause impaired immune cell function, leading to high VL.
CXCL11, also referred to as interferon-inducible T-cell alpha chemoattractant (I-TAC) and interferon-gamma-inducible protein 9 (IP-9), is expressed at high levels in peripheral blood leukocytes, pancreas, and liver [
58]. CXCL11 exhibits the highest affinity for CXCR3 among its three selective ligands, followed by CXCL10 and CXCL9 [
58]. Our data demonstrate that CXCL11 is significantly up-regulated during PHI, positively correlating with CXCL9 and CXCL10, and negatively associated with CD4
+ T-cell count at 1-year-infection point. Furthermore, CXCL11 levels can predict CD4
+ T-cell count, with increased levels detected in mixed cryoglobulinemia, Graves’ disease, and some autoimmune diseases [
59‐
61]. Plasma CXCL11 levels have not been reported in HIV, although higher
CXCL11 mRNA levels were observed in monocyte derived macrophages and dendritic cells which infected with HIV in vitro [
62]. Overall, studies of CXCL11 in HIV are limited and the exact mechanism underlying our findings requires further investigation.
As they have a common receptor (CXCR3), CXCL9, CXCL10, and CXCL11 have generally been studied together, and studies of combinations of CXCL9, CXCL10, and CXCL11 in infection, injury, and immunoinflammatory responses have been conducted. Levels of CXCL9 and CXCL10 in lymph nodes are positively associated with disease progression during SIV infection [
63]. CXCL9, CXCL10, and CXCL11 can regulate the balance between CD4
+ effector T-cell subsets and fork-head box P3 (FOXp3)-negative regulatory T cells [
64]. Pineda-Tenor et al. [
65] found that, in HIV/HCV-co-infected patients, genetic polymorphisms in
CXCL9,
CXCL10, and
CXCL11 are associated with sustained virologic response. Recently, the HIV-1 Nef-induced lncRNA, AK006025, was shown to regulate expression of the
CXCL9/10/11 gene cluster in mouse astrocytes [
66]; however, plasma levels of CXCL9, CXCL10, and CXCL11 have not been studied simultaneously in HIV-infected individuals. Our results demonstrate that a combined panel of CXCL9, CXCL10, and CXCL11 has predictive value for HIV disease progression, when both CD4
+ T-cell count and VL are considered. Testing for plasma chemokine levels is convenient and simple; therefore, our model has potential value for clinical application to predict HIV disease progression.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.