Increased IFN responses drive myeloid cell activation in people living with HIV-1

Abstract

People living with HIV and who initiated antiretroviral therapy (PLH) at the chronic stage of the infection are generally exposed to persistent inflammation. We here assessed the impact of the interferon (IFN)/JAK-STAT pathway on myeloid cells from PLH and the potential of a JAK1/2 inhibitor, Baricitinib, to prevent IFN-driven myeloid cell activation. Peripheral blood mononuclear cells (PBMCs) from 16 chronically infected and virologically suppressed PLH were compared to 15 healthy uninfected individuals (UI) before and after exposure to type 1 IFN, type 2 IFN, and Baricitinib. First, we report an increased activation profile on monocytes and type 2 conventional dendritic cells (cDC2s) from PLH compared to UI, associated with a higher expression of PD-L1 and CD11b ex-vivo, and elevated transcription of pd-l1, cxcl-10 and ifnar1. Then, we unveil the role of type 1 and 2 IFN as inducers of PD-L1 expression at the transcriptional and protein level and highlight an increased response to IFN in PLH myeloid cells, associated with the delay before antiretroviral therapy initiation and the level of CD4 T cell depletion. Last, we describe the preventive effects of Baricitinib on IFN-driven PD-L1 expression. Our study shows that PLH harbor signs of myeloid cell activation, associated with increased type 1 and 2 IFN-signaling, which could be reversed by JAK1/2 inhibitors in vitro.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-04613-0.

Introduction

Despite the continuous progress made in developing highly effective and tolerable antiretroviral therapy, the 37 millions of people living with HIV (PLH) still face an elevated risk of mortality when compared to the general population^1,2. This increased morbidity in PLH is linked to unresolved chronic inflammation, stemming from a vicious cycle of chronic antigenic stimulation due to low-grade viremia and the translocation of gut microbial products, which leads to immune activation, exhaustion, and contributes to HIV persistence^3–5. Early initiation of antiretroviral therapy (ART) remains the best option to prevent these events, yet it remains unattainable for many PLH, leaving them exposed to chronic immune activation and related inflammation^6–9. Unfortunately, around 30% of PLH are diagnosed at a late stage, putting them at a higher risk of poor immune reconstitution and increased inflammation-driven morbidity and mortality^10,11. Therefore, understanding the mechanisms underlying persistent chronic inflammation and its consequences is crucial for developing new therapeutic strategies in this population¹².

The chronic phase of HIV infection is characterized by sustained type 1 interferon (IFN) signaling and increased expression of interferon-stimulated genes (ISGs), which promote chronic immune activation, inflammation and its consequences^13–15. Elevated ISG expression persists in PLH, particularly in those with poor CD4 T cell reconstitution^16–18. ISGs expression is induced by type 1 and 2 signaling, through binding of type 1 interferon receptor (IFNAR) and interferon gamma receptor, and induction of the JAK-STAT pathway¹⁹. New strategies targeting the modulation of the IFN/JAK-STAT pathway are being explored, as they could potentially reduce latent viral replication in HIV reservoirs and reverse immune exhaustion^13,20,21.

Myeloid cells, particularly monocytes, have increasingly emerged as key players in HIV-associated chronic inflammation and related comorbidities. These cells exhibit activation markers such as increased expression of the CD11b integrin, heightened production of CXCL-10, and impaired phagocytic function, reminiscent of age-related changes observed in elderly individuals^22–26. The expression of PD-1 ligand, PD-L1, is elevated on myeloid cells from untreated PLH and only partially reduced with ART^27–29. Recent studies have reported elevated levels of soluble PD-L1 in the serum of viremic PLH, reflecting myeloid cell activation^30,31. This increased PD-L1 expression may contribute to the maintenance of HIV reservoirs through engagement with PD-1 on infected CD4 T cells, while also sustaining myeloid cell activation via the recently described PD-L1 back-signaling pathway, which leads to increased expression of class II HLA and inflammatory cytokine production^32–35. Therefore, understanding the mechanisms that drive PD-L1 expression on myeloid cells is critical for developing new therapeutic approaches in PLH in the chronic phase of the HIV infection³⁶.

In this study, we demonstrate that circulating myeloid cells from chronically infected PLH exhibit higher levels of activation markers (PD-L1, CD11b) compared to non-infected individuals. In vitro exposure to type 1 and type 2 IFNs induced PD-L1 and HLA-DR expression on myeloid cells, as well as PDL1 transcription and soluble PD-L1 production, with a greater effect observed in cells from PLH. We identified an increased sensitivity to type 1 IFN in monocytes from PLH with a longer delay in ART initiation and to type 2 IFN in cells from PLH with low CD4 counts. Pharmacological inhibition of JAK-1/2 reduced PD-L1, CD11b, and HLA-DR expression on myeloid cells. These findings support the notion of heightened IFN signaling in myeloid cells from PLH, which can be modulated by JAK-1/2 inhibition.

Results

Signs of activation and type 1 IFN signaling in monocytes and cDC2s from PLH

We first investigated the activation profile of myeloid cells in PLH. Cells from 16 chronically infected and ART-treated HIV-1 patients (PLH) and 15 uninfected individuals (UI) were analyzed for the expression of PD-L1, CD11b, and HLA-DR using flow cytometry. While no differences were observed in the distribution of monocyte subsets between PLH and UI, PLH showed a decreased proportion of conventional dendritic cells (cDC)1s and cDC2s compared to UIs (Fig. 1A). Monocytes and cDC2s from PLH exhibited increased expression of PD-L1 and CD11b, though not of HLA-DR (Fig. 1B and C). This heightened expression of PD-L1 and CD11b on monocytes was driven by the CD14+/CD16- and CD14+/CD16 + subsets, with no significant differences observed in cDC1s and pDCs regarding PD-L1, CD11b, and HLA-DR expression (data not shown).

Fig. 1.

Increased ex-vivo expression of activation markers and interferon-related genes in PLH myeloid cells. (A) ex-vivo proportion of monocytes and dendritic cells from 14 PLH and 15 UI. See Supplemental Fig. 1A for gating strategy. Statistical analysis performed using Mann-Whitney test. cDC1s: conventional dendritic cells type 1, cDC2s: conventional dendritic cells type 2, pDCs: plasmacytoid dendritic cells. (B) Ex-vivo expression of PD-L1, CD11b and HLA-DR geometric mean fluorescence intensity (MFI) on monocytes from 14 PLH and 15 UI. Statistical analysis performed using Wilcoxon signed-rank test. (C) Ex-vivo expression of PD-L1, CD11b and HLA-DR MFI on cDC2s from 14 PLH and 15 UI. Statistical analysis performed using Wilcoxon signed-rank test. (D)Ex-vivo expression of pd-l1, mx1, cxcl-10, ifnar-1 mRNA on bulk PBMCs. Results from 15 uninfected individuals and 17 PLH for pd-l1, mx1 & cxcl-10, and from 11 uninfected individuals and 13 PLH for ifnar-1. Quantitative PCR results, presented as the fold expression of the mean ex-vivo delta-ct values for each gene, using the delta-delta-CT method, and 18 S as the housekeeping gene. (E)Correlations between the ex-vivo expression of pd-l1 and mx1 in PLH, performed through Spearman correlation test, presented as log¹⁰ relative expression. (F)Correlations between the ex-vivo expression of pd-l1 and ifnar1 in PLH, performed through Spearman correlation test, presented as log¹⁰ relative expression. (G) Correlation between the ex-vivo expression of ifnar-1 and the delay from HIV diagnosis to antiretroviral therapy initiation in 13 PLH.

Next, we assessed the ex vivo expression of a set of IFN-related genes (mx1, cxcl-10), ifnar-1) and pd-l1 in PBMCs by RT-qPCR. The ex vivo expression levels of pd-l1, cxcl-10, and ifnar-1 mRNAs were elevated in PLH compared to UI (Fig. 1D). Notably, pd-l1 expression was positively correlated with mx1 (Spearman’s r = 0.66, p = 0.005) and ifnar-1 (Spearman’s r = 0.62, p = 0.026) in PLH, but not in uninfected individuals (Fig. 1E-F). Additionally, pd-l1 mRNA expression correlated with PD-L1 mean fluorescence intensity (MFI) on monocytes from PLH (Spearman’s r = 0.75, p = 0.025), suggesting that circulating monocytes are the primary contributors to pd-l1 expression in PLH’s PBMCs. Ifnar-1 mRNA expression was also correlated with the time from HIV diagnosis to ART initiation (p = 0.01, r = 0.69) (Fig. 1G). These findings suggest a relationship between myeloid cell activation and IFN signaling in PLH cells, prompting us to further compare the effects of type 1 and type 2 IFNs on cells from PLH and UI. We also assessed the levels of circulating soluble CD14 which did not differ between UI and PLH and did not correlate with ex-vivo expression of myeloid cell activation markers (Supplemental Fig. 2A).

Differential effects of type 1 and 2 interferons on monocytes and cDC2s from PLH and UI

To further investigate the relationship between IFN signaling and myeloid cell activation, we cultured PBMCs from both PLH and UI, exposed them to either type 1 (IFN-α2a) or type 2 (IFN-γ) interferons, and assessed the expression of activation markers after 24 h.

First, IFN exposure had significant effects on the myeloid cell populations distribution. Exposure to either type of IFN reduced the proportion of classical CD14+/CD16- monocytes and increased the proportion of intermediate CD14+/CD16 + and non-classical CD14-/CD16 + monocytes. Both IFN also had an effect on the proportion of dendritic cells, and were associated with reduction of the proportion of cDC1s and increase in the proportion of pDCs but did not affect the proportion of cDC2s. (Supplemental Fig. 2B). Neither the proportion of CD4 nor CD8 T cells were modified by IFNs (data not shown). IFN-γ was a strong inducer of PD-L1 expression at the surface of monocytes and cDC2s (Fig. 2A and C) at the mRNA level (Fig. 2E) and as a soluble protein form (Fig. 2F). IFN-γ exposure also led to the expression of HLA-DR on monocytes but not on cDC2s (Fig. 2B and D). IFN-α2a exposure increased PD-L1 expression at the surface of monocytes and cDC2s (Fig. 2A and C), at the mRNA level (Fig. 2E) and as a soluble protein (Fig. 2B and D). These effects were consistent across all three monocyte subsets (Supplemental Fig. 2C). Both IFNs also induced PD-L1 MFI expression on cDC1s and pDCs in both PLH and UI (Supplemental Fig. 2D). Neither IFN significantly increased CD11b expression (data not shown).

Fig. 2.

Increased IFN responses in monocytes and cDC2s from PLH. (A-B) Expression of PD-L1 and HLA-DR on monocytes after IFN-α2a and IFN-γ exposure for 24 H in PBMCs from 13 PLH and 14 UI assessed by flow cytometry. Statistical analysis performed with Friedman’s paired test and Dunn’s correction. (C-D) Expression of PD-L1 and HLA-DR on cDC2s after IFN-α2a and IFN-γ exposure for 24 H in PBMCs from 13 PLH and 14 UI assessed by flow cytometry. Statistical analysis performed with Friedman’s paired test and Dunn’s correction. (E) IFN-α2a and IFN-γ stimulation effects on pd-l1, mx1, cxcl10 and ifnar1 expression in bulk PBMCs. Results from up to 13 uninfected individuals and 14 PLH. Statistical analysis performed with Friedman’s paired test and Dunn’s correction or Kruskal-Wallis test. (F)IFN-α2a and IFN-γ stimulation effects on soluble PD-L1 production in bulk PBMCs supernatants. Results from 15 uninfected individuals and 12 PLH (11 for the IFN-γ condition).

Most of the effects described above were shared between UI and PLH cells, with exceptions. The IFN-α2a induction of PD-L1 and HLA-DR expression at the surface of monocytes was found to be significant only in PLH cells and not in UI’s (Fig. 2A and B) with paired Friedman test, yet levels of PD-L1 expression were similar between UI and PLHs. The IFN-α2a induction of soluble PD-L1 production was also only significant in PLH cells and the levels of soluble PD-L1 produced after IFN-α2a exposure were higher in PLH cells compared to UI’s (Fig. 2E).

As described above (Fig. 1), PD-L1 MFI ex vivo was higher on monocytes and cDC2s from PLH compared to UI. However, although PD-L1 MFI significantly increased in PLH after 24 h of IFN stimulation, the levels equalized between PLH and UI on both monocytes and cDC2s (Supplemental Fig. 2E). To explore this further, we analyzed the effects of type 1 and type 2 IFNs on mRNA expression of ISGs and PD-L1 in PBMCs from both groups. Both IFNs similarly induced pd-l1 and cxcl10 mRNA expression, while IFN-α2a was the primary driver of mx1 expression (Fig. 2E). Interestingly, PLH cells showed higher mRNA expression levels of pd-l1, mx1, and ifnar1 after 24 h of culture compared to UI. This suggests potential post-transcriptional regulation that might reconcile these findings with the MFI results (Supplemental Fig. 2F).

To better understand the differences in IFN response between PLH and UI, we explored correlations between the IFN-induced activation in monocytes and cDC2s and various immunovirological factors in PLH (Table 1). The most significant association observed with IFN-α2a response was the time to ART initiation, which positively correlated with PD-L1 MFI on monocytes and showed a trend toward correlation with PD-L1 expression on cDC2s. For IFN-γ response, low CD4 T cell counts were correlated with greater PD-L1 upregulation on monocytes and cDC2s, increased pd-l1 mRNA in PBMCs, heightened HLA-DR expression on monocytes, and a trend toward correlation with sPD-L1 release in cell supernatants. Overall, these results underscore a differential sensitivity to IFN exposure in PBMCs from PLH compared to UI and highlight IFNs as key drivers of PD-L1 expression in the myeloid cells of PLH. We thus identified signs of an increased response to type 1 IFN in monocytes from PLH with a longer delay in ART initiation and to type 2 IFN in cells from PLH with low CD4 counts (Table 1).

Table 1.

Correlations between myeloid cell responses to IFN and Immunovirological parameters in PLH.

	Time between HIV diagnosis and ART initiation	CD4 T cell counts
Fold increase in PD-L1 MFI on monocytes after IFN-α2a exposure	r = 0.774 p =0.002
Fold increase in PD-L1 MFI on cDC2s after IFN-α2a exposure	r = 0.538 p = 0.075
Fold increase in PD-L1 MFI on monocytes after IFN-γ exposure		r = - 0.664 p = 0.021
Fold increase in PD-L1 MFI on cDC2s after IFN-γ exposure		r = - 0.672 p = 0.027
Fold increase in pd-l1 mRNA expression from PBMCs after IFN-γ exposure		r = - 0.594 p = 0.045
Fold increase in soluble PD-L1 production after IFN-γ exposure		r = - 0.559 p = 0.062
Fold increase in HLA-DR MFI on monocytes after IFN-γ exposure		r = - 0.678 p = 0.018

Spearman correlation test.

A Pharmacological JAK-1/2 inhibitor prevents IFN-mediated expression of PD-L1, CD11b and HLA-DR in myeloid cells

A potential strategy to modulate the IFN pathway is through the use of JAK inhibitors. We aimed to evaluate whether PLH and uninfected individuals exhibit similar responses to JAK-1/2 inhibition. To this end, we assessed the effect of the JAK-1/2 inhibitor baricitinib on IFN-mediated changes in myeloid cells. Cells were pre-treated with baricitinib at 250 nM for one hour before seeding, followed by IFN exposure and harvesting after 24 h.

Treatment with baricitinib prevented the IFN-induced redistribution of monocyte subsets, specifically inhibiting the differentiation of CD14+/CD16- monocytes into CD14+/CD16 + and CD14-/CD16 + subsets, without reducing the proportion of live cells at the end of the culture (Supplemental Fig. 3A, 3B). Additionally, baricitinib decreased the proportion of pDCs, while not affecting the proportions of cDC1s and cDC2s (Supplemental Fig. 3B).

The expression of pd-l1, mx1, and cxcl-10 mRNA in IFN-exposed PBMCs was significantly reduced (Fig. 3A). Consequently, baricitinib decreased PD-L1 and HLA-DR MFI expression on monocytes but did not affect CD11b MFI expression (Fig. 3B-C, Supplemental Fig. 3C). These effects were observed across all three monocyte subsets, with a slightly lower reduction in PD-L1 MFI expression in PLH monocytes compared to UI (Supplemental Fig. 3D). Baricitinib also reduced PD-L1 expression on dendritic cells (Fig. 3D), decreased HLA-DR and CD11b expression in IFN-exposed cDC2s, and increased CD11b expression in IFN-α2a-exposed cDC1s (Supplemental Fig. 3E-F). Finally, baricitinib reduced IFN-driven production of soluble CXCL-10 in PBMC supernatants in both UI and PLH (Fig. 3E).

Fig. 3.

Baricitinib, a JAK1/2 inhibitor, prevents IFN driven expression of PD-L1, CD11b and HLA-DR on PLH monocytes and cDC2s. (A) JAK1/2 inhibition prevents pd-l1, mx1 and cxcl-10 mRNA expression on bulk PBMCs. Results from 10 UI and 12 PLH. Quantitative PCR results, presented as the fold expression of the mean ex-vivo delta-ct values for each gene, using the delta-delta-CT method, and 18 S as the housekeeping gene. Wilcoxon signed-rank test. (B-C) JAK1/2 inhibition effects on PD-L1 and HLA-DR on total monocytes expressed as geometric mean fluorescence intensity. Results from 13 PLH and 14 UI. Wilcoxon signed-rank test. (D) JAK1/2 inhibition effects on PD-L1 on cDC2s expressed as geometric mean fluorescence intensity. Results from 11 PLH and 14 UI. Wilcoxon signed-rank test. (E) JAK1/2 inhibition effects on soluble CXCL-10 production measured in IFN-exposed PBMC’s supernatants by ELISA. Results from 13 PLH and 12 UI. Wilcoxon signed-rank test and Friedman’s paired test with Dunn’s correction.

Discussion

Understanding the mechanisms driving myeloid cell activation in chronic HIV infection is crucial for developing strategies to address persistent inflammation in people living with HIV (PLH). Our study reveals that, despite long-term antiretroviral therapy (ART) and effective viral control, myeloid cells from PLH exhibit persistent activation signs and heightened type 1 and 2 interferon (IFN) signaling ex vivo. IFN exposure in vitro upregulated PD-L1 expression on these cells, which was countered by JAK-1/2 inhibition. PLH monocytes and cDC2s showed increased sensitivity to IFNs for PD-L1 expression, correlating with longer duration before ART initiation and lower CD4 counts. These findings underscore an amplified IFN sensitivity in PLH myeloid cells, potentially contributing to HIV-associated chronic immune activation and exhaustion.

Our results align with previous studies indicating that myeloid cell activation markers remain elevated despite ART^37,38. While age differences between PLH and uninfected individuals (UI) could explain some variations—such as increased CD11b expression with age - the observed changes in dendritic cell proportions and PD-L1 expression are likely HIV-specific rather than age-related³⁹. Additionally, monocytes responses to stimuli tend to be reduced with age^40–42, while they appeared increased in PLH in our study.

We confirm that both type 1 and type 2 IFNs induce PD-L1 expression on monocytes and cDC2s from PLH and identify a link between IFN-stimulated gene expression and PD-L1 upregulation. Specifically, IFN-1 responses were heightened in PLH with delayed ART initiation, with increased IFNAR1 expression and PD-L1 induction on monocytes. Conversely, IFN-2 responses were stronger in PLH with low CD4 counts, leading to greater PD-L1 expression and transcription. These observations suggest differential cell-intrinsic IFN signaling, potentially impacting responses to anti-PD-1/PD-L1 therapies.

Increased IFN-1 responses in PLH may be attributed to higher IFNAR-1 expression, particularly in those with delayed ART initiation. Previous studies indicated downregulation of IFNAR in viremic PLH, possibly due to IFN desensitization^43–45. These results likely account for viremic PLH exposed to high levels of circulating IFN-1 which are undetectable in ART-treated PLH^46,47. ART-treated PLH, having been exposed to high IFN-1 levels previously, might experience long-term epigenetic reprogramming affecting IFN responses⁴⁸. This is supported by inflammatory transcriptomic signatures in PLH monocytes, including ISG, and heightened responses to pattern recognition receptor agonists^49–52. It should be noted that our results describe cell sensitivity to IFN-α2a and different patterns of sensitivity may be observed with other IFN-1 subtypes.

In our study, PLH also exhibited heightened IFN-2 responsiveness. Interestingly, IFN-γ increased mx1 levels, and ifnar-1 expression in those with low CD4/CD8 ratio. Increased plasmatic concentrations of IFN-γ and increased type 1 ISG signaling have been reported and associated with incomplete immune reconstitution in PLH^16,53,54. This suggests a complex interplay between type 1 and type 2 IFN signaling, with type 1 signaling potentially augmented following IFN-2 stimulation in myeloid cells from PLH with poor immune reconstitution. IFN-2 could induce higher proportions of phosphorylated STAT-1 and improve type 1 IFN signaling¹⁹. Further research is needed to explore this hypothesis.

The elevated PD-L1 expression on PLH monocytes and cDC2s corroborates earlier findings^27,29,55,56. PD-L1 expression on myeloid cells has been associated with the expression of IFNARx-1, and IFN-α stimulation has been shown to induce PD-L1 expression on dendritic cells through a STAT3/p38 pathway, resulting in reduced T cell activation^57,58. PD-L1 on antigen-presenting cells is known to contribute to T cell exhaustion in chronic infections, and targeting the PD-1/PD-L1 pathway has shown promise in improving T cell functionality in chronic viral infections models^59,60. In humanized mice models of HIV-1 infection, type 1 IFN signaling blockade leads to decreased immune activation and exhaustion, reduces PD-L1 expression on immune cells and improves T cell functionality^14,15. However, anti-PD-1/PD-L1 antibody treatments in PLH with cancer have shown transient increases in HIV viral load and specific CD8 T cell responses without reducing the HIV reservoir^61,62.

PD-L1 also mediates an inflammatory signature in myeloid cells through reverse signaling. It contributes to regulating IL-6 production by dendritic cells, potentially fueling monocyte inflammation in PLH^63,64. We report an increased production of soluble PD-L1 after IFN stimulation from PLH cells. Soluble PD-L1, produced predominantly by activated myeloid cells, inhibits T cell proliferation and cytokine production⁶⁵. Higher levels of soluble PD-L1 were previously described in the plasma of ART-treated PLH compared to UI. They were correlated with markers of microbial translocation, and produced in vitro by LPS-stimulated monocyte-derived dendritic cells from viremic PLH^30,31. Also, chronic microbial exposure may activate Th1 lymphocytes, leading to an enhanced IFN-2 environment. In line with this hypothesis, we observed a trend toward a correlation between IFN-2 stimulation and the magnitude of sPD-L1 production.

CD11b expression, increased in PLH’s monocytes, aligns with previous reports⁶⁶. CD11b is crucial for monocyte function and inflammation^67,68 but showed minimal induction by IFNs in our study. We did not find any difference in the ex-vivo expression of HLA-DR on myeloid cells between PLH and UI, in line with previous reports of HLA-DR expression normalization with ART⁶⁹. Our observation of an HLA-DR induction on myeloid cells by IFNs is well documented, and baricitinib was previously shown to prevent HIV induced HLA-DR upregulation on primary human macrophages^70,71.

Our study provides a basis for JAK-inhibition in PLH. Several therapeutic approaches have been considered to modulate the IFN pathway in chronic HIV infection⁷². In a humanized mice model, IFNAR blockade in conjunction with ART led to a decrease in the expression of T cell exhaustion markers, as well as reduction in HIV reservoir size^14,15. However, such a strategy may be challenged by the complexity of IFNAR trafficking and membrane recycling, which mirrors the difficulties encountered in assessing the efficacy of anti-IFNAR antibodies in type 1 IFN-driven diseases^73,74. Targeting JAK-1 and JAK-2 reduces both type 1 and type 2 IFN signaling and provides additional anti-cytokine effects, particularly against IL-6⁷⁵. A recent open-label clinical trial evaluated Ruxolitinib, another JAK1/2 inhibitor, in ART-treated and virologically suppressed PLH, reporting an acceptable safety profile, as well as a reduction in circulating IL-6 levels and activated T cells. In our study, baricitinib prevented IFN-mediated induction of PD-L1, CD11b, and HLA-DR on monocytes and myeloid dendritic cells, and also reduced cxcl-10 mRNA expression and protein production in PBMCs⁷⁶. These findings highlight JAK inhibitors as potential therapeutic agents for modulating aberrant IFN signaling and restoring immune homeostasis in ART-treated PLH.

A potential limitation of our results is the choice to study bulk mRNA expression in total PBMCs rather than in sorted cell populations. Of note, we assessed the effect of IFN-α2a and IFN-γ on T cells from the same samples via flow cytometry and found no expression of PD-L1 or signs of activation (data not shown). This suggests that myeloid cells are likely responsible for the observed pd-l1, mx1, and cxcl-10 mRNA expression in bulk PBMCs. Additionally, we did not include markers of ASDCs in our gating strategy and were thus not able to identify the contribution of this new myeloid DC subset to the IFN response⁷⁷. Another limitation is the lack of analysis of the expression of other markers of myeloid cell activation (CD80, CD83 and CD86) which were not included in our cytometry panel. Future research should further explore the implications of IFN-driven PD-L1 expression on monocytes and cDC2s for T cell activation and immune responses.

Conclusion

Our study reveals a dysregulated phenotype in monocytes from chronically infected and virologically suppressed PLH, characterized by heightened sensitivity to interferons and reversible with pharmacological JAK-STAT inhibition.

Materials and methods

Patient recruitment and ethics

PLH were recruited from the Internal Medicine and Clinical Immunology department of Bicêtre Hospital, France, between February and October 2021. Healthy uninfected individuals (UI) were recruited from l’Établissement Français du Sang (La Pitié-Salpêtrière Hospital blood bank, Paris). Inclusion criteria were: age over 18 years, plasma viral load below 50 copies/mL for at least one year, and antiretroviral therapy for at least two years. Exclusion criteria included pregnancy, acute or evolving pathology, absence of consent, inability to understand the consent form, and patients under legal guardianship. UI were significantly younger than PLH (median age: 35 years [IQR 27–47.5] vs. 55 years [IQR 47–60.5], p = 0.001) (Table 2). There were no differences in sex distribution between UI and PLH (sex ratio: 0.875 women/men in UI and 1 women/men in PLH). The median time between HIV diagnosis and ART initiation was 730 days (IQR [26–4440]). PLH had a median CD4 T-cell count of 730 cells/mm³ (IQR [570–827.5]) before sampling (referred to as “Last CD4 T-cell count”) and a median CD4/CD8 ratio of 1.05 (IQR [0.65–1.37]). The median duration of viral control in PLH was 68 months (IQR [33–130]). All subjects provided written informed consent. The study was approved by an independent investigational review board and ethics committee (CPP Sud Est VI, reference AU1597, n° IDRCB 2019-A03120-57) and conducted in accordance with the principles of the Helsinki Declaration. The total number of PLH included in the rt-qPCR ex-vivo is 17, and 15 UI. The total number of PLH included in the ex-vivo flow cytometry analysis is 14 for 15 UI. The total number of PLH included in the soluble PD-L1 concentration measurement was 12 PLH and 15 UI. Variations in samples number during the experiment are due to technical issues.

Table 2.

Characteristics of the 16 ART-PLH and 15 uninfected individuals included in the study.

	Donors (n=15)	HIV infected patients included in the analysis (N=17)
Age (years old, (median, IQR))**	35 [27-47.5]	55.31 [47-60.5]
Gender (W/M)	0,875	1
Active tobacco smoking		N = 1
Delay between seropositivity and initiation of ART (days, median, IQR)		730 [26-4440]
Duration of viral load undetectability (months, median, IQR)		68 [33-130]
Last CD4 counts (/mm3)		730 [570-827.5]
Median CD4 over the last 4 years (/mm3)		606.5 [496.5-803.5]
Median CD4 nadir		207.5 [178.5-252]
Last CD4/CD8 ratio (median, IQR)		1.05 [0.65-1.37]
Antiretroviral therapy regimen
INI + NRTI		N = 7
INI + NNRTI		N = 2
NRTI + NNRTI		N = 4
NRTI + PI		N = 3

**: p<0,01 for the difference between uninfected individuals and ART-PLH patients (whole population, Wilcoxon non-parametric test).

INI: integrase inhibitor, NRTI: nucleosidic reverse transcriptase inhibitor, NNRTI: non-nucleosidic reverse transcriptase inhibitor, PI: protease inhibitors.

PBMC extraction and culture

Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood collected in EDTA tubes using Ficoll-Hypaque centrifugation, followed by two washes with 1% PBS. Cells were counted and seeded into FACS tubes at a concentration of 2 × 10⁶ cells/mL in 500 µL of cell culture medium (RPMI 1640 supplemented with penicillin-streptomycin (2,000 units/mL), 0.02 M HEPES buffer, non-essential amino acids (2X), and sodium pyruvate (2nM)). Cells were pre-treated with baricitinib (LY3009104, INC B028050 Cat n° S2851, Selleckchem) at 250 nM or 500 nM, or with a control (treated with 0.5% DMSO), and incubated for one hour. They were then stimulated with IFN-α2a (PBL Interferon Source, Tebu-Bio, ref 09311100-1) at 500 UI/mL or IFN-γ (PBL Assay Science, ref 11500-1) at 10 ng/mL, or with control PBS 1X, and re-incubated for 23 h. All assays were performed with fresh cells. The choice of the alpha-2 subtype of type 1 IFN relied on previous studies highlighting its increased concentration in untreated PLH and its place among the most induced subtypes during the chronic phase^46,78. Both IFN concentrations were based on previously published ranges of bioactivity and preliminary experiments^79,80. Baricitinib concentration was determined from a previously published dose-activity study and preliminary experiments⁸¹.

Cell staining and flow cytometry

Cells were washed with 2mL PBS 1x and stained with Live/Dead antibody (Clone: eFluor 780, APC-Cy7) for 20 min in the dark at room temperature, then washed again and stained in the dark for 20 min in 200uL of solution composed by the myeloid panel mix, 5uL of Brilliant Buffer Plus (BD Biosciences), 5uL of FC Block (BD Biosciences), and Wash Buffer (500mL PBS 1X + 5 g of Bovine Serum Albumin). The myeloid panel mix included the following fluorophore-conjugated antibodies : CD11b-PE-Vio770 (REA713, Miltenyi), CD14-BV650 (M5E2, Ozyme), CD16-PerCpCy5.5 (3G8, Ozyme), CD123-PE (7G3, BD Biosciences), CD11c-BV421 (3.9, Ozyme), CD141-BV510 (1A4, BD Biosciences), CD1c-BV711 (L161, Ozyme), CD15-BV605 (W6D3, BioLegend), CD33-BV785 (WM53, BioLegend), CD3-FITC (SK7, Ozyme), CD19-FITC (HIB19, Ozyme), CD56-FITC (NCAM16.2, BD Biosciences), CD66b-FITC (G10F5, BioLegend), HLA-DR-APC-R700 (G46.6, BD Biosciences), PD-L1-Pe-Dazzle594 (29E2A3, BioLegend), LILRB2-APC (42D1, BioLegend). Cells were then washed with 2mL PBS 1X and fixed in 1% paraformaldehyde. Data was acquired on an LSR Fortessa flow cytometer (BD Biosciences) and analyzed using FlowJo Software (version 10.6.2, FlowJo SSC, Ashland, OR, USA). To quantify the cells membrane expression of checkpoint inhibitors and activation markers, we either used proportions of positive cells or the geometric mean of the mean fluorescence intensity depending on the antibody intake and percentage of positive cells.

Gating strategy

Myeloid cells were selected using the SSC-A/FSC-A gating (Supplemental Fig. 1A). After exclusion of doublets and of dead cells, non-polymorphonuclear myeloid cells were selected as CD3-CD19-CD56-CD66b-HLA-DR+. Monocytes subpopulations were then separated using a CD14/CD16 gating: classical monocytes were defined as CD14 + CD16-, intermediate monocytes as CD14 + CD16 + and non-classical monocytes as CD14-CD16+. CD14-CD16- non-polymorphonuclear myeloid cells were then gated on with CD123 and CD11c to define plasmacytoid dendritic cells (CD123 + CD141-) and conventional dendritic cells (cDCs, CD123-CD141+), which were further defined as cDC1s (CD141 + CD1c-) or cDC2s (CD141-CD1c+, confirmed as CD33+). Representative histograms of the data shown in Fig. 1B and C are presented as Supplemental Fig. 1B).

Quantitative PCR

Total RNA was extracted with the GeneJET kit (Thermo Fisher Scientific) according to manufacturer specifications. cDNA synthesis involved use of the Enhanced Avian HS RT-PCR kit (Sigma-217 Aldrich, Saint Quentin Fallavier, France). The quantification of mRNA expression was determined by TaqMan real-time PCR according to the manufacturer’s instructions. The following primers were used: pd-l1 (CD274, assay ID Hs00204257_m1, Thermo Fisher Scientific), mx1 (assay ID Hs00895608_m1, Thermo Fisher Scientific), cxcl-10 (assay ID Hs00171042_m1, Thermo Fisher Scientific). The amounts of mRNA were determined using the delta-delta-CT method and were normalized to the endogenous 18 S (assay ID Hs03003631_g1, Thermo Fisher Scientific). Results were expressed as ratio to the considered gene mean ex- vivo CT.

Cytokine and soluble factors measurements

Ex-vivo plasma and cell-free supernatants after 24 h of cell culture were collected and assayed for CXCL-10 production by a commercially available ELISA kit (BioTechne) and for soluble PD-L1 (Meso-Scale Discovery, K151Z7K-1) used according to the manufacturer’s instruction. CXCL-10 concentrations were quantified by immunofluorescence assay using a Bio-Plex platform (Bio-Rad). Acquisitions and analyses for soluble PD-L1 were performed on a MESO™ QuickPlex SQ120 reader and the MSD’s Discovery Workbench 4.0.

Statistical analyses

All statistical analyses were performed using GraphPad Prism software (version 8, GraphPad.Software Inc., San Diego, CA, USA). When multiple comparisons were involved, groups were compared using a non-parametric, paired Friedmann test; if the groups differed significantly, pairs were compared using Dunn’s correction. When pairing was not available for all datas (i.e. missing point due to technical issue), groups were compared using a non-parametric Kruskal-Wallis test. A non-parametric, paired Wilcoxon test was used for paired data. Spearman’s coefficient was used to assess correlations. The threshold for statistical significance was set to p < 0.05 (* p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001). Continuous variables were described as the median [IQR].

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: LP, NN, CJ, OLMethodology: LP, CJ, NN, CB, OL, VD, AP, MB, CM, DD, DBInvestigation: LP, NNVisualization: LP, NNFunding acquisition: LP, NNProject administration: LP, NN, KBSupervision: NN, CB, OL, BJ, MMT, NC.Writing – original draft: LP, NN, OL, CB, BJ, MMTWriting – review & editing: LP, NN, OL, MMT.

Funding

The study was funded by the Agence Nationale pour la Recherche sur le SIDA et les hépatites virales (ANRS), grant number ECTZ 105732. This work was supported by the Fondation pour la Recherche Médicale, grant number FDM202006011286, to Léo Plaçais.

Data availability

Anonymized data will be available upon reasonable request to LP and acceptance of the investigational review board and ethic committee (CPP Sud Est VI, reference AU1597, n° IDRCB 2019-A03120-57).

Declarations

Competing interests

The authors declare no competing interests.