HIV infection is associated with loss of anti-inflammatory alveolar macrophages

Abstract

HIV-1 is associated with pulmonary dysfunction which is exacerbated by cigarette smoke. Alveolar macrophages (AM) are the most prominent immune cell in the alveolar space. These cells play an important role in clearing inhaled pathogens and regulating the inflammatory environment; however, how HIV infection impacts AM phenotype and function is not well understood, in part due to their autofluorescence and the absence of well-defined surface markers. The main aim of this study was to evaluate the impact of HIV infection on human AM and to compare the effect of smoking on their phenotype and function. Time of Flight Mass Cytometry (CyTOF) and RNAseq were used to characterize macrophages from human bronchoalveolar lavage (BAL) of HIV-infected and uninfected smokers and nonsmokers. We found that the frequency of CD163⁺ anti-inflammatory was decreased while CD163⁻CCR7⁺ proinflammatory AM were increased in HIV infection. HIV-mediated pro-inflammatory polarization was associated with increased levels of inflammatory cytokines and macrophage activation. Conversely, smoking heightened the inflammatory response evident by change in the expression of CXCR4 and TLR4. Altogether, these findings suggest that HIV infection along with cigarette smoke favors a pro-inflammatory macrophage phenotype associated with enhanced expression of inflammatory molecules. Further, this study highlights CyTOF as a reliable method for immunophenotyping the highly autofluorescent cells present in the BAL of cigarette smokers.

Introduction

Human immunodeficiency virus type 1 (HIV-1) infection has long been associated with non-infectious pulmonary disease (1) in the era of combination antiretroviral therapy. The underlying cause of non-infectious lung disease in people living with HIV (PLWH) stems in part from chronic inflammation mediated by dysregulated alveolar macrophages (AM) (2, 3). AM play a pivotal role in host defense against pathogenic microorganisms, tissue remodeling (4) and immunological homeostasis (5). AM dysregulation is facilitated by direct infection and expression of viral proteins (2), cytokines released by HIV-specific CD8+ T cells (6), high levels of microbial products (7) and cigarette smoke (8). AM from PLWH exhibit increased levels of activation (9) and TNF-α production (10) but are impaired in their ability to clear particular pathogens (11).

Macrophages can be polarized towards a CD163⁻ proinflammatory “M1”, classically activated, or CD163⁺ anti-inflammatory “M2”, alternatively activated state (12, 13). Alternatively activated macrophages can be further defined by the expression of the mannose receptor, CD206 (14–17). CD206⁺ macrophages have been reported to produce large amounts of anti-inflammatory cytokines, such as IL-10, and help with the resolution of inflammation (18, 19). A decreased frequency of CD206⁺ macrophage has been associated with increased inflammation not only during HIV infection (20) but also with other diseases such as acute myeloid leukemia (AML) (21). However, human AM have mixed phenotypes during health and disease, and there are conflicting reports about the identification and frequency of anti-inflammatory AM in humans (22–24).

Additionally, during pulmonary inflammation, monocytes are recruited to the lung and differentiate into monocyte-derived alveolar macrophages (Mo-AM), which are phenotypically distinct from tissue-resident alveolar macrophages (TR-AM) (25, 26) and express similar markers used to identify M1 and M2 AM. For example, TR-AM have similar phenotypic and functional properties to that of CD163⁺ AM(27), which are supposedly contributed by CD14⁺CD206⁺ monocytes(28); however, it remains unclear how these cells differ. Whether HIV infection impacts macrophage polarization and/or induces a shift in the frequency of Mo-AM and TR-AM and how AM control inflammation during HIV infection is not fully understood, which is partly associated with the difficulty of characterizing AM in the human lung using traditional flow cytometry due to their auto-fluorescent nature (29).

Cytometry by Time of Flight (CyTOF) utilizes pure element conjugated antibodies that produce discrete isotope peaks. This eliminates the need for compensation, increases the number of simultaneously analyzed markers, and eliminates autofluorescence. Because of the aforementioned properties of CyTOF, it is an ideal method for phenotyping autofluorescent cells, such as AM (30, 31). AM autofluorescence is enhanced by cigarette smoking, which is highly prevalent in the HIV population (32), further highlighting the utility of CyTOF for analysis of lung samples. In addition, simultaneous examination of a larger number of markers made possible by CyTOF technology is necessary to clearly define the known and unique AM population in the lung associated with various disease conditions. Here, CyTOF was used to analyze the expression of 33 unique markers on cells isolated from bronchoalveolar lavage fluid (BALF) obtained from untreated HIV-infected and seronegative individuals composed equally of smokers and non-smokers. The primary goal of this study was to better elucidate the role of AM in HIV-associated lung inflammation. A decreased frequency of CD163⁺ AM in HIV-infected subjects which was associated with increased inflammatory cytokines was observed. Consistent with previous findings, cigarette smoking-induced increased expression of CCR2 (33), TLR4 (34) and CXCR4 (35) was also found in our samples. Taken together, these data suggest that HIV induces a shift in the AM population from an anti to pro-inflammatory, thus promoting pulmonary inflammation during HIV infection.

Methods

Study Population

BAL cells and peripheral blood mononuclear cells (PBMCs) were obtained from 10 antiretroviral naïve HIV-infected non-smokers, 9 antiretroviral naïve HIV-infected smokers, 10 HIV seronegative non-smokers and 9 HIV seronegative smokers (Table 1). To improve statistical power, in some cases subjects were grouped either by smoking or HIV status alone. All participants were 26 years or older. The median viral load in HIV-1-infected participants was 72,400 copies of HIV-1 RNA/ml plasma (range: 418 to 2,070,000 HIV-1 RNA/ml plasma), and the median CD4⁺ T cell count was 562 cells/μl (range, 219–1342 cells/μl).

Table 1.

Demographics of the Study Population^A

Cohort	HIV-Infected Smokers	HIV-Infected Nonsmokers	HIV-Uninfected Smokers	HIV-Uninfected Nonsmokers
Number	10	9	9	10
Gender (M/F)	9/1	8/1	9/0	10/0
Race (C/AF/H)B	7/2/1	8/1/0	8/0/1	9/1/0
Age (years)	32.4 (26.7 – 54.6)	44.6 (27.2 – 57.6)	45.1 (28.6 – 58.9)	39.4 (26.8 – 57.3)
Cigarette use (pack years)	9.1 (0.5 – 22.0)	0	22.4 (2.5 – 33.0)	0
Viral load	55601 (736 – 353767)	279516 (49 – 1130000)	N/A	N/A
CD4 T cell count	550 (103 – 948)	562 (220 – 735)	N/A	N/A
BAL cells
WBC count (×106)	43.0 (17.5 – 62.0)*	31.4 (21.2 – 43.4)*	25.8 (3.5 – 57.2)	22.5 (3.5 – 50.0)
Macrophages (%)	88.9 (64 – 100)	80.8 (67 – 92)	94.7 (88 – 100)	86.3 (58 – 96)
Lymphocytes (%)	9.2 (0 – 32)	17.0 (7 – 33)*	3.8 (0 – 14)	12 (2 −36)
Neutrophils (%)	1.0 (0 – 8)	0.3 (0 – 1)	0.5 (0 – 2)	0.2 (0 – 1)

^AData expressed as average (range).

^BC = Caucasian; AF = African-American; H = Hispanic.

^*Statistically significant compared to HIV-Uninfected Nonsmokers

Ethics Statement

The study protocol was approved by the Colorado Multiple Institution Review Board (COMIRB No: 14–1595) and written informed consent was obtained from all participants.

Mass cytometry

Human PBMCs were isolated by Ficol density gradient centrifugation as previously described(36). Cryopreserved BAL cells were thawed in complete RPMI medium supplemented with benzonase (Sigma-Aldrich, St. Louis, MO, USA). PBMCs from a healthy donor were included in each assay as a staining control. For live–dead cell distinction, cells were stained with 2.5 μM Cisplatin (Fluidigm, South San Francisco, CA, USA). Cells were re-suspended in 70 μl barium free FACS buffer (PBS with 0.1% BSA and 2 mM EDTA) and incubated for 30 min at 4°C with a 30 μl cocktail of metal-conjugated antibodies focused on AM for BAL or T cells for PBMC (Supplemental Table 1). All metals prone to oxidation and metals that are present in lighter flint were excluded. Fixed samples were barcoded using the Cell-ID 20-Plex Barcoding Kit (Fluidigm, South San Francisco, CA, USA). Cells were stained with DNA intercalator (0.125 μM Iridium-191/193; Fluidigm, South San Francisco, CA, USA) and acquired using a CyTOF2 mass cytometer (Fluidigm, South San Francisco, CA, USA), CyTOF software v.6.0.626 with noise reduction, a lower convolution threshold of 200, event length limits of 10–150 pushes, a sigma value of 3, and a flow rate of 0.045 ml/min. Runs were concatenated using the FCS file concatenation tool from Cytobank and were de-barcoded following manufacturer’s protocol.

AM isolation

AM were purified from BAL cells using anti-CD71 magnetic microbeads (Miltenyi Biotec), as previously described (37). Differential cell counts confirmed post-sort purity >90%. AM were lysed in 1 ml TRIzol, and RNA was purified using a RNeasy column (QIAGEN, Germantown, MD). RNA quantitation was performed using a NanoDrop 2000.

Library construction and RNA-seq

RNA-seq was performed on 5 HIV-seronegative and 10 HIV-infected subjects. Library construction and sequencing were performed by the Genomics and Microarray Core Facility at the University of Colorado Anschutz Medical Campus as previously described (38). RNA quality and integrity was first assessed using an Agilent TapeStation 2200. The library was constructed using the Illumina TruSeq mRNA library construction kit. Single-end sequencing was performed using an Illumina HiSeq 4000 for 125 cycles. RNA-seq data have been deposited in the Sequence Read Archive database (http://www.ncbi.nlm.nih.gov/bioproject/496501). Sequencing data were processed using the Illumina BaseSpace online platform and assessed using FASTQC. The FASTQ Toolkit tool was used to filter, adaptor trim, and quality trim reads. Alignment to the hg19 reference genome was performed using the TopHat Alignment tool, with assembly and differential expression analysis performed using the Cufflinks Assembly & Differential Expression tool.

Results

HIV infection is associated with a decrease in CD163⁺ AM

Cells obtained from the BALF were stained with 33 metal-labeled antibodies (Supplemental Table 1), analyzed by CyTOF, and viable AM were defined by DNA content, CD45⁺CD3⁻CD19⁻CD56⁻ and CD11b⁺CD71⁺ expression (Figure 1A) (39, 40). As previously reported by us and others (36), there was an increase in the absolute number of white blood cells in the BALF of HIV-infected subjects compared to uninfected subjects (p<0.0001, Supplemental Figure 1A). However, BALF of HIV-infected nonsmokers contained significantly fewer AM than HIV-seronegative nonsmokers and smokers, as determined by both CyTOF and differentials (Supplemental Figure 1B).

Figure 1. CyTOF reveals that differential expression of CD163, CD206, CCR7 and HLA-DR on AM is mediated by HIV.

BAL from 19 HIV-infected and 19 HIV-seronegative subjects was barcoded and analyzed for 33 markers using CyTOF. (A) Representative dot plots of the gating strategy for alveolar macrophages identified by DNA content, viability, the lack of expression of CD3, CD19 and CD56 and the expression of CD45, CD11b and CD71. (B) Differential analysis between HIV-infected and seronegative subjects of individual macrophage makers on AM. Left frequency, right mean metal intensity (MMI). Statistical significance was calculated by Mann Whitney test. * = p < 0.05, ** = p < 0.005, p = < 0.0005.

Expression of 18 AM specific cell surface markers were compared either by frequency, if the populations were bimodal, or mean metal intensity (MMI) if they were not (Figure 1B). For these analysis smokers and non-smokers were grouped based on their HIV status. The frequencies of CD163⁺ and CD206⁺ cells were significantly decreased in HIV-infected compared to HIV-seronegative subjects (p<0.0001 and p=0.04, respectively) while the frequency of CCR7⁺ cells and the MMI of HLA-DR was increased with HIV infection (p=0.011 and p<0.0001 respectively). CD163 and CD206 have been used independently and together to identify anti-inflammatory macrophages (41). CCR7 expression has been proposed to be a marker of inflammatory macrophages (42), migrating myeloid cells (43) or potentially a monocyte derived population (Mo-AM). Representative dot plots depicting the frequency of CD163⁺CD206⁺ or CD163⁻CCR7⁺ AM from BAL of an HIV-seronegative and infected subject are shown in Figure 2A and 2B. Examining CD163⁺CD206⁺ AM revealed significantly decreased frequencies and absolute counts in HIV-infected compared to HIV-uninfected subjects (Figure 2C, p<0.0001, p=0.01, respectively). In contrast, CD163⁻CCR7⁺ expressing cells were increased with HIV infection both in frequency and absolute counts (Figure 2D, p=0.001, p=0.0004, respectively). The frequency of cells expressing an intermediate amount of CD163⁺ were also compared between HIV-infected and seronegative subjects, but no statistical differences were identified (data not shown). When examining individual surface marker expression of CD163⁻CD206⁻ AM between HIV-seronegative and HIV-infected subjects, we found the MMI of HLA-DR (p=0.002) and the frequency of CCR7 (p=0.003) were increased in HIV-infected subjects compared to HIV-negative controls (data not shown). No statistical differences were observed when CD163⁻ and CD163⁺ AM frequencies were evaluated in regard to smoking status (data not shown). We also found increased MMI of HLA-DR of CD163⁻CD206⁺ and CD163⁺CD206⁺ AM in those with HIV infection compared to those without however the difference did not reach statistical significance (data not shown). HLA-DR expression was inversely correlated with the frequency and absolute number of CD163⁺CD206⁺ AM (Figure 2E, p=0.0007, r=−0.54, p=0.018, r=−0.42) and was positively correlated the frequency and absolute number of CD163⁻CCR7⁺ (Figure 2F, p=0.05, r=0.33, p=0.017, r=0.42), highlighting that the shift in AM towards a CD163⁻ phenotype is related to their activation state. The above analysis was also performed on AM defined without CD71, specifically CD45⁺CD3⁻CD19⁻CD56⁻CD11b⁺, and similar trends were found (data not shown).

Figure 2. HIV infection is associated with a decrease in CD163⁺CD206⁺ and an increase in CD163⁻CCR7⁺ AM.

Frequency of CD163⁺CD206⁺ anti-inflammatory cells and Mo-AM identified as CD163⁻CCR7⁺ we evaluated in HIV-infected vs. seronegative subjects and correlated with HLA-DR expression on AM (A) Representative dot plots and cumulative data showing frequency and absolute number of CD163⁺CD206⁺ AM cells in uninfected and HIV-infected subjects. (B) Representative dot plots and cumulative data of frequency and absolute number of CD163⁻CCR7⁺ AM cells in uninfected and HIV-infected subjects. (C) Correlation of CD163⁺CD206⁺ frequency and absolute number with mean metal intensity (MMI) of HLA-DR on AM. (D) Correlation of CD163⁻CCR7⁺ frequency and absolute number with MMI of HLA-DR on AM. Statistical significance was calculated by Mann Whitney test and Pearson correlation analysis.

Since CyTOF is a relatively new method for characterization of autofluorescent cells, RNAseq and traditional flow cytometry were employed to examine BAL from 6 non-smokers and 8 smokers that were analyzed by CyTOF to validate our findings. Cells were stained with fluorochrome-labeled antibodies specific for CD3, CD11b, HLA-DR, CD163 and CD206. When using traditional flow cytometry smoking caused high levels of autofluorescence, impacting the characterization of the CD163⁺CD206⁺ AM subset (Supplemental Figure 2A). In contrast, CyTOF analysis of the same sample showed more defined populations of mononuclear phagocytes. Upon examining CD163 gene expression (ENSG0000017757575) of CD71⁺ purified AM determined by RNAseq using CyTOF (p=0.02, r=0.58) and traditional flow cytometry (Supplemental Figure 2B, p=0.87, r=−0.05), it was found that population of AM identified using CyTOF, but not traditional flow cytometry, correlated with gene expression, validating this technique and highlighting the confounding effects of traditional flow cytometry, where the analysis of AM population is hampered by autofluorescence.

Gene expression confirms alteration in CD163⁺ Alveolar Macrophage phenotype

Conflicting reports about the use of CD163 as a marker of polarization or cell origination (tissue resident vs infiltrating cells) has led us to compare gene expression levels of several other anti-inflammatory macrophage markers (e.g., CD204, CD206, CD209 and ARG-1). Gene expression of CD204 (p=0.02, r=0.61) and CD206 (p=0.004, r=0.71) was significantly correlated with expression of CD163⁺CD206⁺ on AM (Figure 3A). We also compared the gene expression of CD204 and CD206 with the absolute count of CD163⁺CD206⁺ AM and found a similar trend with positive correlations however they did not reach statistical significance (data not shown). CD209 and Arg-1 had a positive correlation, but neither reached statistical significance (data not shown). Frequencies of cells expressing other proposed markers of tissue-resident AM (TR-AM) and monocyte-derived AM (Mo-AM) including CCR2, CD14, CD36, and CD64 (44, 45) were also examined individually but there were no statistically significant differences between HIV-infected and seronegative subjects (Supplemental Figure 3).

Figure 3. AM gene expressions confirms anti- and pro-inflammatory phenotype correlate with decreased lung function.

RNAseq was performed on CD71⁺ purified AM from a subset of HIV-infected (10) and uninfected subjects (5). (A) Targeted genes of AM, including CD163, CD204, and CD206, were correlated with the frequency of CD163⁺CD206⁺ AM identified by CyTOF. (B) Cathespin-L gene expression correlated with the frequency of CD163⁺CD206⁺ AM and CD163⁻CCR7⁺ AM. (C) Matrix metalloproteinase protein 9 and 24 gene expression correlated with the frequency of CD163⁺CD206⁺ AM in HIV infected subjects. (D) Forced expiratory flow at 25–27% (FEF 25–75%) correlated with the frequency of CD163+CD206+ AM. Statistical significance was calculated by Pearson correlation analysis

Cathepsin proteins play important roles in macrophage function (46), and it has been shown that their mRNA expression and activity increase significantly in the alternatively activated, anti-inflammatory phenotype (47). Supportively, Cathepsin-L and the frequency of CD163⁺CD206⁺ AM had a statistically significant positive relationship (Figure 3B, p=0.04, r=0.55) but Cathepsin-B and -S did not reach statistical significance (p=0.07, r=0.53; p=0.06, r=−0.49, respectively, data not shown). An inverse relationship of Cathepsin-L with CD163⁻CCR7⁺ AM (Figure 3B, p=0.04, r=0.52) was also found. We also found a statistically significant positive correlation when comparing Cathepsin-L with the absolute count of CD163⁺CD206⁺ AM (p=0.03, r=0.56) and a statistically significant negative correlation while with absolute number of CD163⁻CCR7⁺ AM (p=0.04, r=−0.52, data not shown). To further understand the underlying mechanisms of HIV-mediated loss of CD163⁺CD206⁺ AM, gene expression levels of 26 matrix metalloproteinases (MMP) were compared with the frequency of CD163⁺CD206⁺ AM in HIV-infected subjects. Two statistically significant associations were identified: matrix metalloproteinases-9 (MMP-9) (ENSG0000100985) and MMP-24 (ENSG00000125966) (Figure 3C, p=0.02, r=−0.71, p=0.04, r=−0.66, respectively). However, when comparing these gene expressions with the absolute count of CD163⁺CD206⁺ AM no statistically significant correlations were found (data not shown). MMP-9 is induced by oxidative stress, is known to cleave CD163, and its gene expression has been previously associated with HIV infection (48). Supportive of our findings it has also been shown that M1 macrophages (CD163⁻CCR7⁺) are associated with increased expression of MMP24 (previously known as MMP25) while M2 (CD163⁺CD206⁺) macrophages are not (49).

Alteration in anti-inflammatory macrophage frequency is associated with decreased lung function

In order to study the association of change in anti-inflammatory macrophage frequency with lung function, pulmonary function tests (PFT) were performed on 7 HIV-infected individuals and was compared to the frequency and absolute number of CD163⁺CD206⁺ AM. There was a significant association with the frequency and absolute number of CD163⁺CD206⁺ AM and forced expiratory flow at 25–27% (FEF 25–75%) (p=0.008, r=0.78, Figure 3D, p=0.04, r=0.76, data not shown) but not with any of the other PFT variables. Since FEF 25–75% reached statistical significance, these data suggest a link between the loss of CD163⁺CD206⁺ AM and the loss of lung function among HIV-infected subjects.

Loss of CD163⁺ AM is associated with inflammatory cytokine production

To examine the correlation between a change in AM polarization and functional activity, gene expression levels of inflammatory cytokines were examined by RNA-seq performed on CD71⁺ purified AM from a subset of subjects. Differential expression of inflammatory cytokines TNF-α and IL-8 were observed between HIV-infected and uninfected-subjects (Figure 4A, p=0.005, p=0.02, respectively). For these analysis smokers and non-smokers were grouped based on their HIV status. TNF-α was inversely correlated to the frequency of CD163⁺CD206⁺ AM (p=0.02, r=−0.60) and positively correlated with CD163⁻CCR7⁺ cells (p=0.03, r=0.55) and HLA-DR expression (Figure 4B, p=0.04, r=0.53). HLA-DR expression also correlated with IL-8 levels (Figure 4C, p=0.02, r=0.58). These data indicate that HIV-mediated loss of CD163⁺CD206⁺ AM and AM activation are associated with increased levels of inflammatory cytokines.

Figure 4. Correlations of CD163⁻CCR7⁺ and CD163⁺CD206⁺ AM frequencies with TNFα and IL-8 gene expression levels.

The gene expression levels of inflammatory cytokines, TNF-α (ENSG0000232810) and IL-8 (ENSG0000169429), from CD71⁺ purified AM were measured by RNAseq. (A) Differential comparisons of TNF-α and IL-8 gene transcription levels in HIV-uninfected and infected subjects. (B) Correlation of TNF-α gene transcription levels with the frequency of CD163⁺CD206⁺ AM, CD163⁻CCR7⁺ AM and HLA-DR mean metal intensity (MMI) on AM. (C) Correlation IL-8 gene transcriptions levels with the MMI of HLA-DR of AM. Statistical significance was calculated by Mann Whitney test and Pearson correlation analysis.

Smoking increases CCR2, TLR4, PD-L1 and CXCR4 expression on AM

To better understand the dual effect of cigarette smoke and HIV-infection on AM, we examined the frequency and mean metal intensity of surface markers on AM from smokers with and without HIV and non-smokers with and without HIV (Figure 5A). Interestingly, regardless of HIV infection, cigarette smoking increases AM expression of toll-like receptor 4 (TLR4), chemokine co-receptor 2 (CCR2), and program death ligand 1 (PD-L1) (Figure 5A left panel), indicating an increased state of activation. The frequency of CXCR4⁺ AM was also statistically significantly elevated, but the frequency for other makers was not (Figure 5A right panel). Representative dot plots are shown for CCR2 and TLR4 from each cohort (Figure 5B). When further broken out by smoking and HIV status a similar tread was seen with increased expression of MMI for TLR4, CCR2, PD-L1 and CXCR4 in smokers with and without HIV (Figure 5C), however, the increase in TLR4 and CXCR4 only reached statistical significance in HIV-infected subjects who smoke compared to uninfected non-smokers (p=0.004) and infected non-smokers (p=0.03), indicating a combined effect of HIV and smoking on AM.

Figure 5. Smoking increase CCR2, TLR4, PD-L1 and CXCR4 expression on AM.

(A) HIV-infected smokers, HIV-infected non-smokers, HIV-uninfected smokers and HIV-uninfected nonsmokers were compared for the expression of various surface receptors on AM determined by the mean metal intensity (MMI) and frequency. (B) Representative dot plots of C-C chemokine receptor 2 (CCR2) and Toll like receptor 4 (TLR4) for each cohort. (C) Differential comparison of CCR2, TLR4, Programed-death receptor ligand 1 (PD-L1), and C-X-C chemokine receptor type 4 (CXCR4). Statistical significance was calculated by one-way Anova with multiple comparisons.

Hallmarks of HIV disease in the blood are associated with a loss of CD163⁺CD206⁺ AM

Having paired CyTOF data from the BAL and blood provided a unique opportunity to evaluate if HIV-mediated changes in the blood were associated to changes in the AM population. In this regard, it was first determined if CyTOF analysis of blood recapitulated known effects of HIV infection. Many hallmarks of HIV mediated immunological effects were observed, including depleted CD4 T cells, increased T cell activation and increased expression of inhibitory pathway markers (Figure 6A). Compared to the frequency of CD163⁺CD206⁺ AM in the lung, blood CD4⁺ T cell frequency was positively correlated (p=0.001, r=0.57) while CD8⁺ T cells and CD4+PD-1+ T cells frequency was inversely correlated (p=0.001, r=−0.55; p=0.004, r=−0.48 respectively) (Figure 6B). We next compared the frequency of CD163⁺ and CD163⁺CD206⁺ AM with plasma viral load in HIV-infected subjects but found no statistically significant correlations (p=0.09, r=−0.43; p=0.47, r=−0.19, data not shown).

Figure 6. CyTOF analysis identifies common HIV-associated phenotypes in the blood which were associated with the frequency of CD163⁺CD206⁺ AM.

PBMC from 19 HIV-infected and 19 HIV-seronegative subjects was barcoded and analyzed for 35 markers using CyTOF. (A) Analysis between HIV-infected and seronegative subjects of individual makers on CD4+ T cells. Left frequency, right mean metal intensity (MMI). Statistical significance was calculated by Mann Whitney test. * = p < 0.05, ** = p < 0.005, p = < 0.0005. (B) Correlation of the frequency of CD4⁺ T cells, CD4⁺ T cells or CD8⁺ T cells in the blood with CD163⁺CD206⁺ AM in the lung. Statistical significance was calculated by Pearson correlation analysis.

Discussion

Pulmonary inflammation associated with HIV infection can lead to increased susceptibility to comorbidities such as chronic obstructive pulmonary disease (COPD) and pulmonary hypertension (PH) (2). HIV infection causes changes in the cellular profile of the alveolar space, including an influx of virus-specific T cells (6, 36, 50) and activation of AM (9, 51), both of which contribute to lung inflammation. Smoking, which is highly prevalent in PLWH, also impacts AM function. However, how AM dysregulation contributes to inflammation during HIV infection is not fully understood, in part due the auto-fluorescent nature of AM (29) and smoking, which has limited the usefulness of conventional flow cytometric analysis. Because AM regulate inflammation (52), we used CyTOF to phenotypically characterize AM, defined by CD45⁺CD3⁻CD19⁻CD56⁻CD11b⁺CD71⁺, during HIV infection and found that anti-inflammatory CD163⁺CD206⁺ AM by both frequency and absolute count were decreased, which confirms previous findings using traditional flow cytometry (28), while CCR7⁺CD163⁻ AM were increased. This was also true when AM were identified without CD71 positivity, which has recently been reported to identify a subset of AM (53). When examining levels of inflammatory cytokines using RNAseq, TNF-α and IL-8 were directly related to HLA-DR expression and inversely correlated with the frequency of CD163⁺ AM. Collectively, these findings suggest that HIV-mediated loss of anti-inflammatory CD163⁺ AM is associated with pulmonary inflammation.

AM typically possess anti-inflammatory M2-like characteristics (54, 55); however, M1/M2 polarization can vary during disease. When considering the extremes of M1 (CD163⁻CD206⁻) and M2 (CD163⁺CD206⁺) phenotypes, HIV-mediated polarization has been reported both in vitro and in vivo. For example, in vitro HIV infection drives human monocyte–derived macrophages (MDM) toward an M1-like phenotype, with downregulation of M2-associated markers including CD163, and CD206 (9, 56). In vivo, simian immunodeficiency virus (SIV) infected macaques, a non-human primate (NHP) model of HIV infection, that developed pulmonary arterial hypertension exhibited increased frequencies of M1-like macrophages and decreased frequencies of anti-inflammatory M2-like macrophages (57). As suggested by in vitro assays and NHP models, we found that AM derived from HIV-infected subjects are skewed away from anti-inflammatory M2-like toward an inflammatory M1-like phenotype. However, it is debated whether macrophages can shift between M1 to M2 activation states as a consequence of disease or whether such a switch requires the differentiation of newly recruited blood monocytes.

Recent findings suggest that infiltrating Mo-AM offset the balance of inflammatory and anti-inflammatory macrophages in the lung as opposed to polarization (58). Mo-AM migrate to the lung during lung stress and express more inflammatory cytokines compared to their tissue derived relatives. The identification of these macrophage subsets has been proposed (25, 26, 45); however, no consensus yet exists. Using a 33 monoclonal antibody CyTOF panel and RNAseq from the same subjects gave us an opportunity to determine if our findings support a change in the polarization state or shift in the frequency of Mo-AM/TR-AM. We found that several markers of anti-inflammatory macrophages, including CD204 and Cathepsin L gene expression, were positively correlated with the frequency of CD163⁺CD206⁺ AM. In contrast, we compared several makers with CCR7⁺CD163⁻ cells but found less evidence that an influx of this cell population is responsible for the overall decreased frequency of CD163⁺ cells. These findings suggest that the change in CD163 frequency is due to a loss of CD163⁺ AM opposed to an influx of CD163⁻ infiltrating Mo-AM. However, further clarification of the M1 and M2 AM phenotype compared to Mo-AM and TR-AM, along with more coherent nomenclature for human AM would bring much needed insight to this important question.

While CD163 is a primary marker for the identification of anti-inflammatory AM, it also has distinct roles in inflammation, and its soluble form in plasma has been extensively correlated with chronic immune activation in HIV infection (59). CD163 is a hemoglobin scavenger receptor and when stimulated leads to the production of anti-inflammatory heme metabolites (60). The role of CD163 in inflammation has been further highlighted by studies in which cross-linking of CD163 led to increased cytokine production (61). We have recently reported that the expression of CD163 on AM was inversely associated with CYP1B1 expression, indicating a relationship between the loss of CD163 and genes associated with COPD (38). MMP-9 is an enzyme that can cleave CD163 (61) and has been found to be increased in BALF and in AM from HIV-infected subjects with emphysema (48, 62). Similarly, we found expression of MMP-9 was increased with HIV infection and inversely correlated with the frequency of anti-inflammatory AM, hinting at a role for MMP-9 in loss of anti-inflammatory AM. However, this mechanism does not account for the increase in inflammatory AM, suggesting a multifaceted effect. When comparing lung function tests, there was a statistically significant relationship between the loss of CD163⁺ AM both by frequency and by absolute count and a reduction of FEF25–75%. Reduced FEF25–75% has been previously shown to associate with lung inflammation, highlighting the importance of CD163⁺ AM role in lung inflammation during HIV infection.

While our primary goal was to evaluate the impact of HIV infection, by bypassing auto-fluoresce, our study design also allowed us to examine the effect of smoking on AM. Historically, smoking has been shown to decrease AM phagocytic and microbicidal activity (8) and induce a distinct activation state (33). Here, we found that cigarette smoking did not induce the loss of CD163⁺ AM, but it did effect AM phenotype by increasing TLR4 and CCR2. CCR2 is involved in immune cells trafficking while TLR4 binds lipopolysaccharide leading to the activation of the NF-KB signaling pathway. Smoking induced upregulation of CCR2 is consistent with previous findings (33), and TLR4 has been shown to be essential for smoking induced pulmonary inflammation (34). We also found that CXCR4, the co-receptor for X4 tropic HIV strains, is increased with smoking. This is of particular importance as upregulation of CXCR4 on AM in other settings, such as Mycobacterium tuberculosis infection, led to increased viral entry of X4 tropic HIV infection in vitro and increased viral evolution in vivo (63). These findings illustrate the utility for CyTOF in evaluating lung samples from smokers as well as highlighting the need for interventions regarding smoking cessation particularly with the high level of smoking in PLWH.

CyTOF analysis on matched blood and lung samples provides us with a unique opportunity to evaluate effects across two distinct immune compartments, particularly associations of immune cell populations in the blood with the loss of anti-inflammatory AM in the lung. The strongest findings were relationships of lung anti-inflammatory AM frequencies with CD4⁺ T cell frequency and the percentage of PD-1 expressing CD4⁺ T cells in the blood, which are each markers of HIV disease progression and T cell exhaustion respectively (64, 65). When we compared the HIV plasma viral load, which has been shown to correlate with the viral load in BAL fluid (66), with the frequency of AM phenotypes no associations were identified suggesting that plasma viral load, and by association BAL viral replication, is not the direct cause of the altered AM phenotypes found in HIV-infected subjects.

Here, we identified HIV-associated polarization of AM from an anti-inflammatory phenotype toward an inflammatory phenotype that was associated with increased immune cell activation, inflammatory cytokine production, and loss of lung function. Phenotypic profiling of immune cell populations in the blood and lung revealed that T cell depletion, activation and exhaustion in the blood is associated with a decreased frequency of anti-inflammatory AM during HIV infection. Smoking increased expression of several surface receptors important for cell signaling on AM, indicating that they are primed for activation. While the aim of characterizing AM during HIV infection and smoking was our primary goal, this study also demonstrated that CyTOF allows the detection of differences in AM surface receptor expression, especially in those who smoke, where conventional flow cytometry staining would otherwise have been masked by auto-fluorescence. Overall, these findings indicate that the loss of anti-inflammatory macrophages plays a role in HIV-induced lung inflammation; however, further evaluation is required to determine the causal relationship.

Key Points:

HIV infection is associated with the loss of CD163+CD206+ AM.

Smoking increases surfaces expression of several activation markers on AM.

CyTOF is a valuable tool for evaluating the phenotypic profile of BAL from smokers.