In this study, we analyzed the putative functions of stabilin-1 in blood monocytes. Microarray analysis revealed downregulation of several proinflammatory genes in the stabilin-1high monocytes when compared with stabilin-1low monocytes. When cocultured with stabilin-1high monocytes, IFN-γ synthesis by T cells was diminished in Ag-recall assays. Knockdown of stabilin-1 in monocytes increased the synthesis of several proinflammatory molecules, including TNF-α, and supported high IFN-γ and low IL-4 and IL-5 production by T cells in Ag-specific stimulation assays. Anti–stabilin-1 Ab treatment also led to increased IFN-γ synthesis in the recall assays. In clinical settings, the expression of stabilin-1 was diminished on blood monocytes and tissue macrophages under proinflammatory conditions. These data define stabilin-1 as a new immunosuppressive molecule and suggest that stabilin-1high monocytes may dampen proinflammatory reactions in vivo.
Monocytes are bone marrow–derived subsets of circulating WBCs, which can differentiate into macrophages and dendritic cells after migration into different tissues (1). Monocytes play an important role in tissue homeostasis, wound healing, and host defense reactions against microbes, other inflammatory stimuli, and tumor cells. Based on the expression of CD14 and CD16, human monocytes are classified into three subsets, which also have different transcriptomes (1–3). The majority (80–90%) of blood monocytes belong to the CD14+CD16− subset, which, together with the CD14+CD16+ subpopulation, are proinflammatory monocytes analogous to mouse Ly6ChiCCR2hiCX3CR1low cells. The nonclassical minor CD14dimCD16+ human monocyte population, in contrast, resembles mouse Ly6ClowCCR2lowCX3CR1hi cells, which constantly patrol the vessels and are involved in cell maintenance and healing functions (3, 4).
Stabilin-1 (also known as CLEVER-1 or FEEL-1) is a multifunctional type I transmembrane protein most prominently expressed on selected endothelial cells and alternatively activated M2 macrophages (5–9). Stabilin-1 synthesis is induced in macrophages during inflammation (10), and it is highly expressed in a subset of tumor-associated macrophages (11). Stabilin-1 is also expressed in monocytes during specific disease conditions, like hypercholesterolemia, but reported to be absent from normal monocytes (12). In macrophages, stabilin-1 acts as a scavenging receptor during endocytosis of selected proteins such as oxidized and acetylated low-density lipoprotein (Ac-LDL), OVA, SPARC, and placental lactogen (13–16). It has also been reported to be involved in bacterial binding (17), clearance of apoptotic cell bodies (18), and wound healing (10) and in adhesion and transmigration of placental leukocytes through endothelial cells (19). Although stabilin-1 is a useful marker for a subset of type 2 macrophages, nothing is known about its functional role during the polarization of immune responses.
In this study, we unexpectedly found definitive expression of stabilin-1 on the surface of normal blood monocytes when using new sensitive Abs. In functional experiments, the stabilin-1low monocyte population and stabilin-1 knockdown monocytes supported the generation of robust proinflammatory immune responses. Notably, ligation of monocyte stabilin-1 by an anti–stabilin-1 Ab also diverted Ag-specific responses into the Th1 direction. These data thus define stabilin-1 as a novel immune modulator.
Materials and Methods
Blood samples from healthy volunteers, tetanus-vaccinated persons (previously vaccinated individuals who had received a booster injection within past 2 y), and timothy grass–allergic (self-reported) persons were collected using venipuncture. All donors gave their informed consent.
Blood samples and samples of placenta and placental bed were taken at elective caesarean sections performed for defined obstetrical indications from women with normal pregnancies and from those with pre-eclampsia at the Department of Obstetrics in Turku University Central Hospital. These samples were collected with permission of the Ethical Committee of the University of Turku, and a written consent was signed by all patients participating in this study. Apart from the treating clinician (E.E.), the patients remained anonymous to the other investigators.
9-11 (a rat IgG2a) and 3-372 (a mouse IgG1) are against human stabilin-1 (8, 19). The other mAbs used were: CD14- FITC (mouse IgG2a; Southern Biotechnology Associates), CD16-PerCP–Cy5.5 (mouse IgG1; BD Pharmingen), HLA-DR–allophycocyanin (mouse IgG2a; BD Pharmingen), and macrophage mannose receptor (CD206, clone 15-2 MRC-1; Lifespan Biosciences). As isotype controls, IgG2a-FITC, IgG1-PerCP–Cy5.5, IgG2a-allophycocyanin (all from BD Pharmingen), MEL-14 (rat IgG2a), and 3G6 (mouse IgG1) were used. The second-stage reagents were PE-conjugated goat anti-mouse IgG (a whole Ig molecule from Southern Biotechnology Associates) and PE-conjugated goat polyclonal F(ab)2 against anti-rat F(ab)2 (Abcam). F(ab)2 fragments of 9-11 and MEL-14 were generated commercially by GenScript.
Immunofluorescence stainings and FACS analyses
Blood samples were collected in EDTA tubes. RBCs were lysed using a commercial lysis buffer (BD Pharm Lyse; BD Biosciences). After preblocking with human Ig (100 μg/ml; KIOVIG from Baxter) the leukocytes were sequentially incubated with anti–stabilin-1 9-11 F(ab)2 or negative control MEL-14 F(ab)2 (20 μg/ml) and PE-conjugated goat anti-rat F(ab)2–F(ab)2 Ab. In monocyte phenotyping experiments, the cells were further incubated with a mixture of anti-human HLA-DR, CD14, and CD16 mAbs (or with the appropriate isotype controls). In the polarization experiments (see below), the cultured leukocytes were harvested and surface stained for stabilin-1 [as above with 9-11 F(ab)2] or for MRC-1 using the anti–MRC-1 mAb (or an isotype control [10 μg/ml]) followed by PE-conjugated secondary goat anti-mouse Ab. Alternatively, aliquots of the cells were permeabilized (15 s in ice-cold acetone), blocked, and stained for total stabilin-1. The cells were fixed using paraformaldehyde and analyzed using FACS Aria II or FACSCalibur (BD Biosciences).
After RBC lysis, the cells were stained for stabilin-1 and CD14 (see above) for sorting by FACSAria (BD Biosciences). Monocytes were identified using scatter profiles, CD14 and stabilin-1 double-positive monocytes were gated using FACS Diva software (BD Biosciences), and the brightest (stabilin-1high) and dimmest (stabilin-1low) 10% of stabilin-1+ monocytes were collected.
Small interfering RNA transfections
Purified monocytes (see below) were transfected with stabilin-1 small interfering RNA (siRNA) and negative control siRNA (5′-UCAAGUCGCUGCCUGCAUA-3′) (ON-TARGETplus Human STAB-1 and Nontargeting siRNA, respectively; GE Healthcare) as previously described (19). The cells were then cultured for 24–96 h under nonpolarizing culture medium. Stabilin-1 expression was determined by FACS from aliquots of cells to verify the silencing efficacy.
Ac-LDL uptake assay
Negative control and stabilin-1 siRNA-transfected monocytes were harvested on day 2. The cells were treated with Alexa 488–labeled Ac-LDL (Invitrogen; 10 μg/ml in RPMI 1640 containing 10% FCS for 4 h), washed, and analyzed by FACS as described (19).
ELISA for quantification of TNF-α
Conditioned culture medium from negative control and stabilin-1 siRNA-transfected monocytes were collected on day 2. TNF-α protein levels were quantified using a sensitive ELISA kit (catalog no. KHC3014; Invitrogen) according to the manufacturer’s instructions.
Ag-specific recall assays
PBMCs from adults recently vaccinated against tetanus and from timothy grass–allergic persons were collected using Ficoll. Monocytes were enriched using MACS negative selection kit (Monocyte isolation Kit II with CD16–MACS; Miltenyi Biotec). Autologous T cells were isolated from the blood samples of the same donors (either on the same day for cocultures with sorted monocytes or on day −1 for cocultures with siRNA-transfected monocytes) using negative selection (T Cell Enrichment Kit, EasySep; Stemcell Technologies).
The 96-well ELISPOT plates (Mabtech) were coated with anti-human IFN-γ (1 μg/ml; Mabtech), IL-4 (15 μg/ml; Mabtech), or IL-5 (15 μg/ml; Mabtech) for 24–48 h at 4°C, washed with PBS, and blocked (RPMI 1640 plus 10% FCS). The purified monocytes and T cells were cocultured in the precoated ELISPOT wells as triplicates at a ratio of 1:10 (10,000 monocytes and 100,000 T cells) in a medium (RPMI 1640, 10% FCS, 2 mmol l-glutamine, 100 μmol 2-ME, and 50 μg/ml gentamicin) containing tetanus toxoid (20 μg/ml; National Public Health Institute, Helsinki, Finland) or timothy extract (100 μg/ml; GREER Laboratories) for 3 d at 37°C in a CO2 incubator. Monocytes and T cells cultured in the same medium without any stimulation served as background controls. On day 3, the wells were washed and sequentially incubated with biotinylated anti–IFN-γ, IL-4, or IL-5 (all at 1 μg/ml) Abs and alkaline phosphatase–conjugated anti-biotin secondary Ab (1:1000), and spots were developed with BCI-NBT solution.
The specific spot numbers for each coculture condition were counted microscopically by subtracting the number of spots recorded in three controls: 1) the cocultures of the corresponding monocytes and T cells without the Ag; 2) the control cultures of the corresponding monocytes with the Ag only (no T cells); and 3) the control cultures of the T cells and the Ag only (no monocytes) from the number of spots recorded in the cocultures in the presence of the monocytes, T cells, and Ag. To normalize the interindividual variation in the absolute numbers of spots, the specific spot number in the stabilin-1high and control siRNA-treated wells was then assigned a value of 100 for each person.
In Ab-blocking experiments, PBMCs (unfractionated) isolated from tetanus- vaccinated persons were incubated with stabilin-1 (3-372) Ab, a nonbinding negative control Ab 3G6, or a binding, irrelevant control Ab CD14 (all at 20 μg/ml) and plated to IFN-γ–coated wells for 3 d.
In vitro polarizations
MACS-purified monocytes (1 × 106 cells/well) were cultured in the growth medium (IMDM containing 10% FCS and 2 mmol l-glutamine to obtain nonpolarized cells. To induce M1 differentiation, TNF-α (50 U/ml) and LPS (10 ng/ml) were added, and to induce M2 differentiation, IL-4 (10 ng/ml) and M-CSF (10 ng/ml); IL-4 (10 ng/ml), M-CSF (10 ng/ml), and dexamethasone (100 nmol); dexamethasone (100 nmol) only; or IL-4 (10 ng/ml) only were added.
RNA was extracted from the sorted stabilin-1high– and stabilin-1low–positive monocytes using the Nucleo-Spin RNAII Total RNA Isolation Kit (Macherey-Nagel). RNA integrity numbers (Agilent 2100 Bioanalyzer; Agilent Technologies) were >8.00 in all samples. The subsequent microarray analysis using the Agilent Sure Print G3 Human Gene Expression Microarray 8 × 60K (Agilent Technologies) was performed at the Finnish Microarray and Sequencing Centre (FMSC). Fluorescence signals were detected using Agilent’s Microarray Scanner System (Agilent Technologies), and the Agilent Feature Extraction Software (Agilent Technologies) was used in further processing of the data.
The resulting intensity data were normalized using quantile normalization. The samples were hierarchically clustered and the correlation values determined with Pearson metrics. R package Limma was used for performing the statistical testing between the groups. The differentially expressed (DE) genes were selected requiring an absolute fold change >2 and p value <0.01. Functional enrichment analysis was conducted against the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes databases for both DE gene lists and unfiltered ranked comparison data using topGO and GOstats packages in R/Bioconductor. The data were further analyzed using Qiagen’s Ingenuity Pathway Analysis (IPA) platform and Gene Set Enrichment Analysis (GSEA) software (http://www.broadinstitute.org/gsea/index.jsp) (20, 21). The data has been deposited into the Gene Expression Omnibus with the accession number GSE63519 (http://www.ncbi.nlm.nih.gov/genbank).
RNA sequencing analysis
RNA was extracted and quality controlled (RNA integrity number >9.00) from stabilin-1 siRNA-transfected and negative control siRNA-transfected human monocytes from three healthy donors as above. The samples were sequenced in FMSC with the HiSeq2500 instrument (Illumina) using single-end sequencing chemistry with 50-bp read length. The reads were aligned against the human reference genome (hg19 assembly, downloaded from University of California, Santa Cruz) using TopHat version 2.0.10. The number of uniquely aligned reads was between 10.2 and 14.2 M per sample. HTSeq (v.0.5.4p3) was used for counting the genewise read counts as defined by RefSeq gene annotations.
The sample correlation values (Spearman metrics) were between 0.93 and 0.98 for all sample groups, indicating high reproducibility. The data were normalized using the trimmed mean of M-values normalization method of the edgeR R/Bioconductor package prior to statistical testing with Limma R/Bioconductor package. The DE genes were selected requiring an absolute fold change >2 and p value <0.05. Functional enrichment analysis was conducted against the GO and Kyoto Encyclopedia of Genes and Genomes and Reactome databases for both DE gene lists and unfiltered ranked comparison data using topGO and GOstats Gage Bioconductor packages. The data were further analyzed using Qiagen’s IPA platform and GSEA software (http://www.broadinstitute.org/gsea/index.jsp) (20, 21) and deposited into the Gene Expression Omnibus (accession number GSE63807).
Total RNA was extracted and reverse-transcribed with the iScript cDNA Synthesis Kit (Bio-Rad) or SuperScript VILO cDNA Synthesis Kit (Life Technologies). TaqMan Gene Expression Assays (Applied Biosystems) for ALOX5AP, PADI4, SPP1, ORM2, GNRH2, LHCGR, oncostatin M (OSM), serum amyloid A (SAA) 2, SCGB3A1, TBP, and B2M were used as primer/probe sets, and the PCR reactions were carried out as suggested by the supplier using the Applied Biosystems 7900HT Fast Real-Time PCR System (Applied Biosystems) in the FMSC. All samples were run as triplicates, and the expression values were normalized using human TBP or β2-microglobulin as endogenous controls. The results were analyzed with SDS 2.3 software and DataAssist v3.01. The average mRNA expression of each gene was presented as a relative quantification (RQ) values using the 2−ΔΔCT method, in which RQ = 1 is the value in the control group.
Small pieces of placental bed were embedded in OCT and stored at −70°C. Frozen sections were cut, acetone-fixed, and immunohistochemically stained for stabilin-1 as described (19). The number of stabilin-1+ cells per field (0.268 mm2) was counted with an Olympus BX-60 microscope (Olympus).
Student t test (unpaired, two-sided for the pregnancy samples and paired, two-sided for the rest of the experiments) was used, except for the microarray and RNAseq, in which the R-package Limma was used. The p values <0.05 were considered to be statistically significant.
Normal human monocytes express stabilin-1 on the surface
Stabilin-1 is reported to be present in M2 macrophages (5–8), but not on normal blood monocytes (12). Because monocytes are notoriously difficult to stain due to high expression of FcR, we re-examined this issue by generating a F(ab)2 fragment from our new specific anti–stabilin-1 mAb 9-11 and isotype-matched negative control mAb (19). To further increase the specificity of the stainings, we preblocked the FcRs with human Igs and took advantage of a PE-conjugated F(ab)2 fragment of an anti-F(ab)2 Ab as the second-stage reagent. Using this protocol, we found low, but consistent, stabilin-1 expression on the surface of monocytes (identified based on the typical scatter profiles) isolated from normal healthy donors (Fig. 1A). To further characterize stabilin-1 expression among the heterogeneous monocyte population, we stained PBMCs for HLA-DR, CD14, and CD16 (or the corresponding isotype controls). We then gated for HLA-DR+ cells within the monocyte gate, and identified the three monocyte populations based on CD14 and CD16 expression, as reported (3). Stabilin-1 was clearly expressed on CD14+CD16+ and on CD14+CD16− populations, but not on CD14dimCD16+ cells (Fig. 1B). Stabilin-1 was absent from the lymphocytes and granulocytes (Supplemental Fig. 1). Thus, the two classical CD14+ monocyte subpopulations in human blood express stabilin-1 on their surface under normal conditions.
Stabilin-1high and stabilin-1low monocyte subpopulations have distinct gene expression signatures
We sorted CD14+ monocytes into stabilin-1high– and stabilin-1low–positive cells (Fig. 2A). Reanalyses of the sorted cells verified that the purity of each subpopulation was >95% (Fig. 2A). Heat map and volcano plot analyses of the cDNA microarray data revealed many DE genes between the two populations (Fig. 2B–D, Supplemental Table I). The maximally differently up- and downregulated genes (with fold change >2) in the stabilin-1high monocytes compared with stabilin-1low are shown in Fig. 2D.
Because protein–protein interactions and signaling pathways related to stabilin-1 are relatively poorly understood, we used manual inspection of the top downregulated hits and performed IPA to assign these DE genes into functional categories and pathways. These analyses revealed that several of the downregulated genes in the stabilin-1high population (Supplemental Table I) exhibit immune-related activities. For instance, the three top hits are, in general, known to support proinflammatory reactions (GO: 0006954 [inflammatory response, 16 genes; p value 2.5 × 10−9], GO: 0006952 [defense response, 19 genes; p value 6.3 × 10−9], and GO: 0009611 [response to wounding, 18 genes; p value 6.8 × 10−9] and data not shown). Furthermore, the upstream regulator analysis of IPA predicted that TNF-α, a prototype proinflammatory cytokine, is inhibited in the stabilin-1high population (data not shown). Further GSEA analysis revealed four significantly enriched gene sets (normalized enrichment scores with false discovery rates <25%), including proinflammatory IL-2/STAT5 and TNF/NF-κB gene sets, in the stabilin-1low population (Fig. 2E). Therefore, we then chose three of the proinflammatory genes, ALOX5AP, PADI4, and SPP1 (22–24), which were found to be downregulated in stabilin-1high cells in the microarray, and further verified their expression by quantitative PCR (qPCR). The results showed that the expression of all three were indeed lower in the stabilin-1high population (RQs for ALOX5AP, PADI4, and SPP1 were 0.23, 0.23, and 0.29, respectively) when compared with stabilin-1low population. Collectively, these data imply that a population of CD14+ monocytes characterized by high stabilin-1 expression has a transcriptome suggestive of a reduced proinflammatory potential.
Downregulation of stabilin-1 is linked to upregulation of proinflammatory genes
To study the effects of stabilin-1 knockdown on global gene expression, we aimed at silencing stabilin-1 in monocytes using siRNA. Initial FACS analyses of stabilin-1 expression in permeabilized cells (detecting both surface and intracellular protein) showed that stabilin-1 knockdown led to a 44.8 ± 9.0% (mean ± SEM; n = 3) decrease in stabilin-1 protein expression on day 1 and 64.1 ± 9.8% (mean ± SEM; n = 3) decrease on day 2 (Fig. 2F). The reduction in the stabilin-1 expression after the knockdown also correlated to diminished binding of Alexa 488–labeled Ac-LDL (24.2 ± 7.2% [mean ± SEM; n = 6; p value = 0.0001]), which is one of the known functions of stabilin-1 (13, 17, 19).
After confirming the efficacy and functional consequences of stabilin-1 silencing in monocytes, we then subjected stabilin-1 and negative control siRNA-treated monocytes to RNA sequencing (RNAseq) analyses after 1 and 2 d culture under nonpolarizing conditions. Already on day 1, several genes (such as OSM, CXCL13, and SAA2) involved in the activation of the proinflammatory pathway were upregulated in stabilin-1–silenced cells when compared with negative control siRNA-transfected monocytes (Fig. 2G–I, Supplemental Table II). Our qPCR results verified the altered gene expression of the selected hits in stabilin-1–silenced cells (RQs for CXCL13, OSM, SAA2, and SCGB3A1 were 2.71, 2.22, 1.26, and 0.69, respectively, in stabilin-1 siRNA-treated cells when compared with negative control siRNA-treated cells). GSEA analysis did not reveal any significantly enriched gene sets in stabilin-1–silenced cells. However, further analysis with IPA on DE genes from stabilin-1–silenced monocytes predicted that TNF-α is involved in the regulation of several of these induced genes (Fig. 2J). To verify this prediction, we analyzed TNF-α levels from the conditioned culture medium of control and stabilin-1 siRNA-treated monocytes by ELISA. The results showed that stabilin-1–silenced monocytes produced significantly more TNF-α when compared with control transfected cells (Fig. 2K). These data indicate that stabilin-1 in monocytes directly or indirectly controls the activation of several proinflammatory genes in human monocytes.
Monocyte stabilin-1 regulates cytokine production during immune responses
To study the functional difference between stabilin-1high and stabilin-1low monocytes, we evaluated cytokine production in Ag-restimulation experiments. We isolated blood monocytes from tetanus-vaccinated persons, separated stabilin-1high and stabilin-1low monocytes by sorting and cocultured them with T cells isolated from the same persons in the presence of tetanus toxoid (Fig. 3A). In these experiments, T cells cocultured with stabilin-1high monocytes produced lower numbers of IFN-γ spot-forming cells (SFC) compared with those cultured with stabilin-1low–positive monocytes (Fig. 3B). These data show that when the Ag presentation takes place via the stabilin-1high monocyte population, the outcome is an impaired Th1 cytokine response.
To study whether stabilin-1 itself has an impact on cytokine production, we silenced stabilin-1 in the total pool of monocytes using siRNA during the coculture experiments (Fig. 3A). Stabilin-1 knockdown specifically led to a 48.5 ± 9.4% (mean ± SEM; n = 3) decrease in stabilin-1 protein expression in monocytes at the beginning of the coculture and to a 65.4 ± 4.9% (mean ± SEM; n = 3) decrease at the end of the coculture. The ELISPOT assays revealed that the numbers of IFN-γ SFC produced in the presence of stabilin-1 siRNA-transfected monocytes were higher than those produced in the presence of negative control siRNA-transfected monocytes (Fig. 3C). The controls verified the specificity of the assays, because no spots were detectable in the coculture in the absence of the specific Ag or in the absence of T cells (Fig. 3D, Supplemental Fig. 2). These results thus show that stabilin-1 itself, directly or indirectly, is involved in the regulation of the immune responsiveness and that low levels of stabilin-1 in monocytes during Ag presentation favor strong IFN-γ production by T cells.
Tetanus toxoid stimulation did not lead to the generation of quantifiable numbers of type 2 cytokine-forming spots (data not shown). Therefore, we chose timothy grass extract as a potential allergen inducing Th2-type cytokine production (Fig. 3A). To enhance the responsiveness, we isolated both monocytes and the autologous T cells for the coculture experiments from persons with known allergy against timothy. In these ELISPOT assays, we found that T cells in the presence of stabilin-1 siRNA-transfected monocytes produced fewer IL-4 and IL-5 spots compared with those cultured with negative control siRNA-transfected monocytes (Fig. 3E). Collectively, these data show that stabilin-1 in monocytes favors the generation of Th2/immunosuppressive responses and impairs the formation of Th1/proinflammatory immune responsiveness in humans.
Anti–stabilin-1 Abs modulate Ag responses to a proinflammatory direction
To study whether stabilin-1 would represent a new target for therapeutic manipulation of immune responses, we took advantage of anti–stabilin-1 Abs. We treated unfractionated PBMCs (only monocytes express stabilin-1 in this cell population; Supplemental Fig. 1), isolated from tetanus-vaccinated persons with anti–stabilin-1 and control Abs. We found that PBMCs blocked with anti–stabilin-1 Ab produced high numbers of IFN-γ SFC compared with PBMCs treated with irrelevant nonbinding (3G6) or binding (anti-CD14) control Abs (Fig. 3F). Anti–stabilin-1 Ab also specifically increased the number of IFN-γ SFC when a lower dose (1 and 10 μg/ml) of the Ag was used (data not shown). Thus, ligation of stabilin-1 on monocytes with a function blocking Ab tilts the Ag-specific recall response to a proinflammatory Th1 direction.
Stabilin-1 is downregulated in proinflammatory monocytes/macrophages
To verify the induction of stabilin-1 during in vitro differentiation of monocytes into M2 macrophages (7), we cultured the purified monocytes isolated from human blood under nonpolarizing, M1-polarizing (LPS and TNF-α), or M2-polarizing (M-CSF and IL-4; M-CSF, IL-4, and dexamethasone; IL-4 only; or dexamethasone only) conditions. Stainings after 2 d showed clear stabilin-1 positivity with our new anti–stabilin-1 Ab on the surface of monocytes cultured under nonpolarizing and all M2-inducing conditions (Fig. 4A, 4B). In contrast, stabilin-1 expression was strongly downregulated under M1-polarizing (LPS and TNF-α) conditions. Macrophage mannose receptor, a prototype marker for M2 cells, was also induced, as expected, under all M2-inducing conditions, but not under nonpolarizing or M1 (LPS and TNF-α) (Fig. 4C). Notably, M2 stimulation not only prevented the loss of stabilin-1 expression on the cell surface when compared with nonpolarizing conditions, but also induced intracellular stabilin-1 expression (Fig. 4D). Collectively, these data show that human monocytes retain stabilin-1 surface expression when the cells are polarizing to M2 direction, whereas they lose it upon M1 induction, which is consistent with the functional immunosuppressive role of stabilin-1.
Stabilin-1 is downregulated in a proinflammatory condition in vivo in humans
To study whether stabilin-1 expression would be different in proinflammatory and immunosuppressive conditions in vivo, we studied blood monocytes, placental macrophages, and placental bed macrophages during pregnancy. Strong immunosuppression and Th2 deviation is characteristic to normal pregnancy (25), whereas an abnormal proinflammatory reaction takes place in the placenta in pre-eclampsia (25, 26). Interestingly, we found that in patients with pre-eclampsia, stabilin-1 expression was significantly lower on CD14+ blood monocytes compared with normal pregnant women (Fig. 5A). Monocyte-derived macrophages also infiltrate the placenta in large numbers during the pregnancy (27). When the leukocytes were isolated from placentas and stained for FACS, the level of stabilin-1 expression was significantly lower on CD14+ macrophages in patients with pre-eclampsia than in patients with normal uncomplicated pregnancy (Fig. 5B). Significantly fewer stabilin-1+ macrophages were found in the placental bed samples in patients with pre-eclampsia than in normal pregnancy (Fig. 5C). These data show that there is a link between the numbers of stabilin-1+ blood monocytes and macrophages and the level of immunosuppression in vivo.
We report in this study that human stabilin-1 is expressed on CD14+CD16− and CD14+CD16+ monocytes in healthy individuals. Silencing of stabilin-1 in monocytes results in activation of several proinflammatory genes. When compared with stabilin-1low monocytes, stabilin-1high monocytes supported more Th2-type and less Th1-type cytokine production in Ag-specific T cell recall assays. Also under pathophysiological conditions in vivo, downregulation of stabilin-1 expression on monocytes correlated with an enhanced proinflammatory response in pre-eclampsia. Importantly, ligation of stabilin-1 on monocytes by anti–stabilin-1 Abs shifted the Ag-specific recall responses to the proinflammatory direction. Collectively, our findings identify stabilin-1 as a new immunosuppressive molecule on monocytes.
Stabilin-1 has been reported to be absent from normal monocytes, even though it is found on monocytes in hypercholesterolemic patients (12). Our phenotypical analysis with refined reagents and protocols unambiguously showed that both CD14+CD16− and CD14+CD16+ monocytes do express stabilin-1 on their surface, whereas CD14dimCD16+ cells do not. Collectively, these data imply that stabilin-1 may serve previously unnoticed functions in normal human monocytes. Indeed, our microarray data showed that stabilin-1high and stabilin-1low monocytes have different transcriptomes, and RNAseq analyses further demonstrated that stabilin-1 is, directly or indirectly, involved in the upregulation of several proinflammatory genes such as OSM, CXCL13, and SAA2. Interestingly, among other functions these molecules have been reported to control cytokine activity, including pathways regulating IFN-γ, and may therefore be linked to the stabilin-1–dependent shift of polarization. OSM treatment of dendritic cells induces IFN-γ secretion by T cells (28). Similarly, SAA–derived peptides, stimulate IFN-γ secretion by CD4 T lymphocytes in human synovial fluid (29) and boost the T cell–stimulating capacity of APCs (30). Moreover, IPA predicted that TNF could serve as a common upstream regulator of several genes, for which expression is regulated by stabilin-1. Notably, analyses of TNF protein concentration in conditioned monocyte medium indeed showed increased TNF production in stabilin-1–silenced cells.
In addition to the correlative data, we demonstrate an immunosuppressive function for stabilin-1 in functional assays. In ELISPOT assays, T cells cocultured with stabilin-1low or stabilin-1–silenced monocytes in the presence of tetanus toxoid produced more IFN-γ spots compared with those cultured with stabilin-1high monocytes or negative control siRNA-treated monocytes, respectively. Notably, also blocking of stabilin-1 with a mAb led to generation of increased numbers of IFN-γ spots. In line with these observations, ELISPOT assays revealed reduced numbers of IL-4– and IL-5–secreting T cells in the presence of stabilin-1 siRNA-treated monocytes. These data not only show that the stabilin-1high monocyte population is functionally different from stabilin-1low monocytes, but also that the stabilin-1 molecule itself, directly or indirectly, regulates the capacity of monocytes to polarize T cells into Th1 versus Th2 direction upon Ag stimulation. Moreover, therapeutic ligation of stabilin-1 with Abs can be used to change the direction of T cell polarization during Ag activation.
These data are consistent with the earlier reported correlations of stabilin-1+ macrophages with immunosuppressive conditions. In macrophages, stabilin-1 is known to be a good marker for immunosuppressive M2 cells both in vitro and in vivo (13, 16, 18, 19, 31–34). Our current results show that M2-polarizing conditions (31) (M [M-CSF and IL-4], M [M-CSF, IL-4, and dexamethasone], M [dexamethasone] and M [IL-4]) augmented stabilin-1 expression on monocyte-derived macrophages, whereas M1-polarizing (M [LPS and TNF-α]) induction led to the loss of stabilin-1 from the macrophages. Moreover, appearance of stabilin-1–positive macrophages has been reported to be associated with immunosuppressive conditions, such as pregnancy and tumors, in vivo (19, 32). Notably, however, no link between stabilin-1 and the immunosuppressive mechanisms of M2 macrophages has been found earlier. Thereby, our current findings for the first time, to our knowledge, indicate that emergence of stabilin-1 in immunosuppressive cells is not merely a good phenotypic marker, but that the molecule actually is causally involved in the immunosuppressive function in those cells.
Placental macrophages use stabilin-1 as a scavenging and adhesive molecule (19), and it is involved in uptake of placental lactogen (16). We observed that stabilin-1 expression is lower both in blood monocytes and placental macrophages in pre-eclampsia in comparison with normal pregnancy. In the complex pathogenesis of pre-eclampsia, an abnormal placental development leads to a symptomatic second-stage systemic inflammatory response that includes leukocyte and endothelial cell activation. In placental bed, macrophages, among other cell types, have been implicated in formation of proper spiral arteries, and their numbers are reduced in pre-eclampsia (35). These data suggest that stabilin-1 may be involved in maintaining the immunosuppressive state in normal pregnancy in humans.
It is of note that the two gene expression analyses in our study served two different purposes. The comparison of sorted stabilin-1high and stabilin-1low monocyte subpopulations aimed at analyzing whether the monocyte subpopulation expressing the highest level of stabilin-1 displays diminished proinflammatory properties in general (stabilin-1 was used as a phenotypic surface marker for subpopulation isolation). The analyses of DE genes in control and stabilin-1 siRNA-treated monocytes, in contrast, aimed at identifying target genes in the total monocyte pool, for which expression could be directly or indirectly affected by the stabilin-1 molecule itself. Therefore, as expected, when comparing the list of the DE genes between stabilin-1high and stabilin-1low monocytes to that between control and siRNA-silenced monocytes we did not find practically any overlap (only TPST1 and HTRA1 were common). For similar reasons, it was not surprising to see that certain genes (ORM2 and SCGB3A1) DE in the two arrays showed corresponding differences in qPCR analyses of the total pool of blood monocytes or placental macrophages in pre-eclampsia, whereas others (GNHRH2, LHCGR) did not (data not shown).
In conclusion, we define stabilin-1 as a novel immunosuppressive molecule. The stabilin-1low monocyte population, stabilin-1–silenced monocytes, and anti–stabilin-1 Ab-treated monocytes all support enhanced generation of Th1-dominant immune responses. Thus, monocyte stabilin-1 is a novel immune modulator that augments formation of Th2-type Ag-specific T cell responses. Interestingly, anti–stabilin-1 Abs can be used to promote Th1-dependent inflammatory responses, which may be useful, for instance, for counteracting tumor-induced immunosuppression.
We thank Maritta Pohjansalo, Etta-Liisa Väänänen, and Sari Mäki for technical help; the obstetricians in the Turku University Hospital for the blood samples, placentas, and placental beds; and Anne Sovikoski-Georgieva for secretarial help.
The sequences presented in this article have been submitted to the Gene Expression Omnibus under accession numbers GSE63519 and GSE63807.
The online version of this article contains supplemental material.
Abbreviations used in this article:
acetylated low-density lipoprotein
Finnish Microarray and Sequencing Centre
Gene Set Enrichment Analysis
Ingenuity Pathway Analysis
serum amyloid A
small interfering RNA.
The authors have no financial conflicts of interest.