IFN Priming Is Necessary but Not Sufficient To Turn on a Migratory Dendritic Cell Program in Lupus Monocytes

Patients with systemic lupus erythematosus (SLE) often suffer from infections, which represent one of the leading causes of morbidity and mortality (1). Both the underlying immune alterations of SLE and the immunosuppressive therapy that patients receive to control disease activity predispose to infections. Infection symptoms mimic exacerbations of SLE and often raise important diagnostic and therapeutic dilemmas. Infections, in turn, trigger SLE disease exacerbations. However, the mechanisms responsible for this association have not been elucidated.

Myeloid cells, including neutrophils and monocytes, are part of the first line of defense against infectious agents. In addition to their innate immune functions, monocytes can differentiate into dendritic cells (DCs) and therefore link innate and adaptive immunity. Although monocytes exposed in vitro to cytokines and TLR ligands differentiate into DCs (2, 3), their capacity to differentiate in vivo has been formally demonstrated only recently (4–6). In mice, in vivo full differentiation of monocytes into DC-SIGN/CD209a⁺ monocyte-derived DCs (Mo-DCs) can be triggered by microbial components, such as LPS. These Mo-DCs rapidly lose expression of monocyte markers such as Gr-1/Ly6C and CD115/c-fms and upregulate TLR4 and CD14. They also acquire DC morphology and localize to the T cell areas of lymph nodes via l-selectin and CCR7. Through Toll/IL-1R domain–containing adapter inducing IFN-β signaling, they become powerful Ag-capturing cells and APCs (6). The extent to which human monocytes differentiate into DCs in vivo remains to be fully determined, although DCs found in inflammatory diseases, such as rheumatoid arthritis (RA) and tumor inflammatory ascitis, share phenotypic and transcriptional characteristics with in vitro–generated Mo-DCs (7).

The blood monocyte compartment is altered in SLE (8), as monocytes from a subset of patients behave similar to myeloid DCs (mDCs) by inducing allogeneic CD4⁺ T cells to proliferate in vitro. Exposure of normal monocytes to SLE serum results in the generation of DCs, which is dependent on IFN-α (8). Unabated DC maturation could lead to the activation/expansion of autoreactive T cells that have escaped central tolerance, thus explaining many of the features of the disease (9). SLE serum also drives healthy monocytes to become DCs with B cell helper capabilities. Thus, these DCs can efficiently stimulate naive and memory B cells to differentiate into IgG and IgA plasmablasts resembling those found in the blood of SLE patients (10).

In this study, we sought to further characterize the blood monocyte compartment of SLE patients to identify candidate pathways that could explain their potential DC behavior. We also sought to investigate the synergism between SLE serum factors, including type I IFN, and pathogen-associated molecular patterns, such as LPS, in the acquisition of a migratory monocyte/DC phenotype that could explain the link between infections and SLE exacerbations.

All SLE patients were collected from the Pediatric Rheumatology Clinic at Texas Scottish Rite Hospital (Dallas, TX). Samples from 51 SLE patients, including 7 males and 44 females, were studied in this project. The average age of the patients at the day of sample collection was 15 y (range, 8–19 y), and the average duration of SLE was 0.69 y (range, 0–2.06 y). The breakdown of the patient ethnicity was 45% Hispanic, 27% African American, 18% white, 4% Asian, and 2% unspecified. Table I depicts the SLE disease activity index (SLEDAI) and medications for each patient in the study. Thirteen sera from pediatric SLE patients were used for the flow cytometry staining on cultured monocytes, most of them in more than one independent experiment. SLE sera were selected based on a high SLEDAI score, absence of immunosuppressive medication, and absence of high-dose prednisone at the moment of blood draw, in addition to absence of i.v. prednisolone bolus administration in the 2 mo previous to the date of blood draw. Clinical characteristics of the SLE patients whose serum was used in our experiments are summarized in Table II. Patients were all females. The average age was 15.8 y (range, 13–17 y). The breakdown of ethnicity was 61.5% Hispanic, 30.8% African American, and 7.7% white.

The control population consisted of 21 randomly selected healthy children, including 5 males and 16 females (average age, 12 y; range, 6–22 y). The ethnic breakdown of the healthy donors was 42% white, 29% Hispanic, 19% African American, and 10% Asian. Some of the flow cytometry staining on cultured monocytes was done using monocytes from three adult healthy donors (two males and one female with ages ranging from 31 to 56 y).

Whole blood samples were collected from pediatric patients and healthy donors using standard venipuncture techniques. Blood was drawn in either EDTA or sodium citrate vacutainer tubes (BD Biosciences, Franklin Lakes, NJ) for the cellular immunology experiments, and the samples were used within 4 h of collection. All blood collection protocols were reviewed and approved by the Institutional Review Boards at the Baylor Research Institute, the Texas Scottish Rite Hospital, and the University of Texas Southwestern Medical Center (all of them in Dallas, TX).

A standard protocol for lymphocyte separation with modifications was used for isolation of PBMCs (11).

Monocytes were isolated from PBMCs from healthy donors and SLE patients using three methods. In the first method, positive selection with CD14 MicroBeads (Miltenyi Biotec, Auburn, CA) as per the manufacturer’s protocol. Monocytes were washed and counted, and their purity was checked by flow cytometry using the following Ab mixture: CD14-FITC (M5E2), CD3-PE (HIT3a), HLA-DR-PerCP (L243), and CD19-allophycocyanin (SJ25C1). Samples with a purity of >97% were used for further processing. In the second method, a two-step procedure began with an automated leukopheresis and was followed by counterflow elutriation. In the third method, elutriated monocytes were further purified by negative selection using the EasySep human monocyte enrichment kit without CD16 depletion (StemCell Technologies, Vancouver, BC, Canada) following the manufacturer’s instructions.

From PBMCs, CD4 T cells were then enriched using a StemSep human CD4⁺ T cell enrichment kit (StemCell Technologies), following the manufacturer’s instructions. Possible contaminating DCs were removed by using anti–HLA-DR magnetic beads (Miltenyi Biotec). CD4 T cells were counted and frozen at −80°C.

mDCs and monocytes from healthy volunteers were sorted on a FACSAria (BD Biosciences) as lineage⁻CD11c⁺HLA-DR⁺ and CD14⁺HLA-DR⁺ cells, respectively.

We performed multicolor surface staining using a standard protocol. Briefly, washed cells were resuspended in PBS and labeled with specific Abs for 15 min. Then they were washed to eliminate the excess of Ab and resuspended in PBS. 7-Aminoactinomycin D (7-AAD) was added to each tube right before acquiring them in an LSR II (BD Biosciences). FlowJo software (Tree Star) was used for analysis. Our gating strategy consisted on gating on monocytes on the forward scatter versus side scatter plot and then on the CD14⁺ cells. For experiments in which 7-AAD was used to discriminate between alive and necrotic/apoptotic cells, we initially gated on 7-AAD⁻ cells. We then followed the same explained gating strategy.

Purified LPS (Escherichia coli 0111:B4, catalog no. L4391) was obtained from Sigma-Aldrich (St. Louis, MO). Recombinant IFN-α2b (Intron A) was obtained from Schering-Plough (Bloomfield, NJ). Allophycocyanin-conjugated anti-human CCR7, PerCP-eFluor 710 anti-CD11c, and appropriate isotype controls were obtained from eBioscience (San Diego, CA). FITC-conjugated anti-CD2, PE anti-CD16, anti–HLA-DR V450, Alexa Fluor 700 anti-CD56, lineage mixture 1, 7-AAD, and appropriate isotype controls were purchased from BD Biosciences (San Jose, CA). Pacific Orange anti-CD3 and quantum dots 800 anti-CD14 and corresponding isotype controls were purchased from Invitrogen (San Diego, CA). ECD anti-CD16, ECD anti-CD19, and isotype controls were purchased from Beckman Coulter (Fullerton, CA). PerCP-Cy5.5 anti–blood DC Ag (BDCA)-1 (or CD1c), FITC anti-human DC-SIGN, and corresponding controls were obtained from BioLegend (San Diego, CA). FITC anti–BDCA-2 (CD303), PE anti–BDCA-3, PE anti–BDCA-4, corresponding controls, and FcR human blocking reagent were purchased from Miltenyi Biotec.

Frozen CD4 T cells were thawed, washed, and live cells were counted. Ten million CD4 T cells were resuspended in 1 ml PBS (Invitrogen) and stained with 1 μM CFSE (Invitrogen) for 10 min in the dark. Staining was blocked with cold complete RPMI 1640 medium containing 10% human AB serum. Cells were washed twice, resuspended in complete RPMI 1640 containing human AB serum, and counted for viability.

Blood monocytes isolated from healthy volunteers were incubated with 20% autologous serum alone or in the presence of 1000 U/ml IFN-α2b (Schering-Plough) in six-well plates at a concentration of 10⁶ monocytes/well in 3 ml media. After incubation for 1 h at 37°C, cells were harvested and RNA was extracted. Identical experiments were done after the following incubation time points: 6 h, 24 h, 2 d, and 3 d.

Blood monocytes isolated from healthy volunteers were incubated in RPMI 1640 supplemented with l-glutamine, gentamicin, and penicillin in six-well plates (1 million cells/well, 3 ml medium) with either 20% autologous serum or 20% SLE sera from active and untreated patients. Cells were harvested and RNA was extracted after 6 h incubation.

In some experiments, blood monocytes isolated from healthy volunteers were incubated in RPMI 1640 supplemented with l-glutamine, gentamicin, and penicillin in 24-well plates (500,000 cells/well/ml medium) with either 20% autologous serum or 20% SLE sera from active and untreated lupus patients.

Healthy monocytes were incubated with 20% autologous sera or SLE sera for 16 h. Cells were then harvested and an in vitro transwell migration assay was performed. Briefly, a 4-μm-pore size insert with a polycarbonate membrane designed for a 24-well plate was used for this assay (Corning, Corning, NY). The lower chamber contained 600 μl 0.5 μg/ml CCL19 solution in RPMI 1640 supplemented with l-glutamine, gentamicin, and penicillin. A total of 0.5 × 10⁶ monocytes in 200 μl of the same medium was added to the upper chamber. The plate was incubated at 37°C for 3 h, and the number of migrated cells was counted by flow cytometry using CountBright absolute counting beads (Invitrogen).

CFSE-labeled CD4 T cells (1 × 10⁵) were incubated with monocytes (2 × 10⁴) in 96-well plates in complete RPMI 1640 medium. Cells were harvested at 6 h, 2 d, and 5 d. After 6 h of culture, harvested cells were enriched by depleting T cells with CD3 Dynabeads (Invitrogen), and RNA was extracted for use in microarray testing. After 2 d of culture, cells were harvested and monocytes were stained for different cell surface markers. After 5 d of culture, cells were harvested and the CFSE dilution was measured to assess the level of T cell proliferation.

RNA was extracted using either an RNeasy Mini kit (Qiagen, Valencia, CA) when >5 × 10⁵ cells were recovered or a PicoPure RNA isolation kit (Molecular Devices, Sunnyvale, CA) when <5 × 10⁵ cells were recovered. The quantity of the RNA was determined by NanoDrop ND-1000 UV-Vis spectrophotometer (NanoDrop Technologies, Wilmington, DE) by measuring 1 μl RNA solution against water as a blank. RNA integrity was determined using the Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA) in which the RNA integration number is expressed as a quality factor. Only RNA samples with a RNA integration number of >7.0 were used for final hybridization in the microarray assay.

RNA was labeled using the GeneChip two-cycle target labeling kit (Affymetrix, Santa Clara, CA) following the manufacturer’s recommended procedures. cRNA was fragmented and hybridized to the HG-U133A and HG-U133B Affymetrix GeneChip arrays that contain 45,000 probe sets at 45°C for 16 h. GeneChip arrays were washed, stained, and scanned according to protocols described in the GeneChip expression analysis technical manual (Affymetrix). Scanned GenesChips were inspected visually for abnormalities or irregularities.

Monocytes (300,000 cells/ml) were cultured in RPMI 1640 medium supplemented with penicillin, streptomycin, 2 mM glutamine, and 20% of either autologous serum, 20% of autologous serum plus 1000 IU IFN-α2b, or 20% of SLE serum. Each experiment consisted of a well of monocytes with autologous serum, a well of monocytes with autologous serum plus IFN-α2b, and three wells with different SLE sera. In parallel, we duplicated the same conditions, but monocytes were further stimulated with LPS (10 or 100 ng/μl). After 24 h, adherent monocytes were harvested with cold PBS and 1 mM EDTA. The cells were washed and subsequently resuspended in PBS before staining.

Cell culture supernatants were collected at 24 h. Supernatants were frozen at −80°C until they were analyzed by Luminex assay (BioLegend), as measured on a Luminex 100 system. Multiplex analytes included GM-CSF, IFN-α, IL-1β, IL-4, IL-6, IL-10, IL-15, and TNF.

We screened the supernatants of two experiments in which elutriated monocytes were used and of one experiment in which negatively selected monocytes from elutriated PBMCs were used.

For each Affymetrix U133A and U133B GeneChip, raw intensity data were normalized to the mean intensity of all measurements on that array and scaled to a target intensity value of 500 using Affymetrix Microarray Suite version 5.0 software. Data were then further analyzed using GeneSpring software version 7.3.

To analyze the data from monocytes from untreated and treated SLE patients, five samples in each data set were used for final analysis and compared with five samples from healthy donors. Data were normalized to this set of healthy controls. For each set of experiments, unsupervised clustering of samples was performed using the list of genes present in at least one sample to rule out technical variability. For supervised analysis, an Affymetrix flag call of “present” in three of five samples from each cohort was used to designate the filter for a reliable intensity measurement from each individual gene chip. These two lists combined were used as a quality control measure for class comparison, which was performed using a nonparametric ranking statistical analysis test (Mann–Whitney) as well as a 2-fold difference in the average normalized value of the healthy to test set. Genes from both class comparison methods were combined, and noise in the final data was reduced by filtering genes with a raw value >200 in three of five test samples compared with the average raw signal from the healthy control set.

In the alloreaction experiments, data were normalized to monocytes that did not induce an alloreaction. An Affymetrix flag call of present in three of four samples of each cohort was used to designate the filter for a reliable intensity measurement from each individual gene chip. These two lists combined were used as a quality control measure for class comparison, because there was a 2-fold difference in the average normalized value of the two groups of monocytes. Possible noise in the final data was reduced by filtering in genes with expression raw values >200 in three of four monocyte samples inducing alloreaction compared with the average raw value of monocytes that did not induce an alloreaction.

To identify monocyte genes up- or downregulated after incubation with SLE serum, data were normalized to data from the corresponding control experiment using autologous serum. An Affymetrix flag call of present in three of three samples and a raw value >200 for each condition was used to designate the filter for a reliable intensity measurement from each individual gene chip. These two lists combined were used as a quality control measure for class comparison, because there were 2-fold differences in the average normalized values from the two groups of experiments.

To determine the global pattern of IFN-regulated genes in untreated SLE monocytes compared with healthy controls, a k-mean clustering algorithm was applied to genes expressed in untreated SLE monocytes cultured with IFN and harvested at different time points (1 h, 6 h, 24 h, 2 d, and 3 d).

To analyze whether the association between the stimulus and the percentage of CCR7 surface expression was modified by the addition of LPS to the cultures, a general linear model was used. A Spearman correlation test was used to analyze the association between the percentage of CCR7 surface expression and the level of several cytokines in the supernatant of the 24 h culture.

Blood monocytes from a fraction of pediatric SLE patients function as DCs (8). We surmised that such DC-like monocytes might share phenotypic markers with conventional DC precursors. Thus, we used flow cytometry to characterize 1) the monocyte subpopulation distribution in pediatric SLE patients and healthy children, and 2) the expression of surface molecules that might explain the acquisition of DC function.

Whole blood samples were obtained after informed consent from 23 pediatric SLE patients (average age, 16.0 y) and 15 healthy children (average age, 12.2 y). The two groups were matched for gender and ethnicity. Pediatric SLE patients and healthy controls displayed similar distribution of CD14^brightCD16⁻ and CD14^dimCD16⁺ cells, whereas the double-positive subpopulation was mildly expanded in SLE patients (p = 0.012) (Fig. 1A). We next analyzed monocyte surface marker expression and found that most of the molecules that become upregulated as monocytes differentiate into DCs, such as maturation (CD83) and costimulatory markers (CD40, CD80, and CD86), were similarly expressed in healthy controls and SLE patients (Table III). Furthermore, a slight downregulation of HLA-DR molecules was observed, especially within the CD14^bright+CD16⁻ monocyte fraction (Fig. 1B, Table III). In SLE patients, CD64 (FcγRI) and the adhesion molecule CD62L were mildly upregulated. Interestingly, CD81, a coreceptor for HCV and HIV that is known to be downregulated by IFN-α (12), was significantly downregulated in CD14^brightCD16⁻ monocytes (Fig. 1C, Table III). Thus, most phenotypic differences found in blood SLE monocytes involve the CD14^brightCD16⁻ population and do not explain DC function.

To better characterize the molecules that could potentially confer Ag-presenting capacity to SLE monocytes, we assessed their transcriptome. All microarray data have been deposited in Gene Expression Omnibus with the common accession number GSE46923 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=udkfguoonzupfyt&acc=GSE46923).

Blood monocytes from five healthy controls and five pediatric SLE patients were isolated using CD14⁺ selection. Because drugs used to treat SLE could induce considerable transcriptional changes, we selected active, newly diagnosed patients who had never received oral or i.v. medications. After positive selection, the purity of the monocyte fraction was >97%. RNA was hybridized to human genome U-133 A and B chips (Affymetrix). An unsupervised clustering algorithm performed on genes present in at least 20% of all samples clustered monocytes from healthy donors and SLE patients into two well-defined groups. SLE expression data were then normalized to those of healthy controls, yielding 1629 differentially regulated transcripts. As shown in Fig. 2A and 2B, 560 of those transcripts could be ascribed to IFN signaling, as they were similarly regulated upon culturing healthy blood monocytes with recombinant type I IFN for various lengths of time (1, 6, 24, 48, 72 h).

Overall, the gene expression profile of SLE monocytes confirmed the flow cytometry data. Thus, DC maturation markers, such as CD83, and costimulatory molecules such as CD40, CD80, and CD86 were not differentially expressed (Fig. 2C). Furthermore, we observed a significant downregulation of MHC class II genes. Other molecules related to Ag presentation such as CD1c and CD1d were downregulated, whereas transcripts encoding surface markers related to monocyte and/or DC lineages were not differentially transcribed. Additional downregulated transcripts included CD44, which encodes an inhibitory receptor for apoptotic particle-induced DC activation (13); CD93, a receptor for C1q that enhances monocyte phagocytosis (14), plays a role in intercellular adhesion (15) and may contribute to the in vivo clearance of dying cells (16); and CD302 (CLEC13A), which mediates Ag uptake (17) (Fig. 2D).

Among cytokine-related transcripts, only IL8, IL6ST, and IL1RN were moderately upregulated, whereas IL13RA1 and TNF were downregulated. Interestingly, ADAM10, one of the shedases that cleaves TNF and E-cadherin (18), was mildly upregulated. The upregulation of ADAM10 could contribute to the previously reported increase of soluble TNF in the blood of SLE patients (18–22) (Fig. 2D).

Additional upregulated transcripts included the activation marker CD69 and CD2AP, a protein involved in dynamic actin remodeling and membrane trafficking that has been linked to glomerular disease (23). IL-4–induced protein 1, a lysosomal l–amino acid oxidase that may play a role in lysosomal Ag processing and presentation, was also mildly upregulated. In mice, the IL-4–induced protein 1 gene maps to a region of SLE susceptibility that includes the Sle3 locus (24). Several signal transducers and transcription activators were upregulated as well, especially STAT1 and STAT2, which are phosphorylated upon type I IFN receptor binding (25), and STAT3, which signals downstream of IFN (26) and IL-6 (27) (Fig. 2D).

As described above (Fig. 2A), an important fraction of the SLE monocyte transcriptome could indeed be ascribed to type I IFN signaling. Thus, IFN-α–inducible protein 27 was the most upregulated (330-fold) transcript in patient monocytes, and SIGLEC1, which acts as an endocytic receptor and is also IFN-inducible, was 88-fold upregulated (Fig. 2D). To explore the full extent of IFN-regulated transcriptional changes, we exposed monocytes from two healthy donors to recombinant type I IFN (IFN-α2b) in vitro. RNA was extracted at different incubation times (1, 6, 24, 48, 72 h) and the expression data were normalized to those of monocytes cultured with medium. These conditions induced a large number of transcriptional changes (n = 4119). Some of these changes were already present at 1 h (n = 226 transcripts), but most were induced upon longer incubation times. Overall, of the 4119 transcripts regulated in healthy monocytes by IFN in vitro, 560 (∼15%) were differentially expressed in ex vivo SLE monocytes, which represents about a third of the SLE monocyte transcriptome (Fig. 2A). Most of these transcripts were previously described in pediatric SLE PBMCs as well (28).

IFN-inducible upregulated transcripts also included apoptosis-inducing molecules such as TRAIL and FAS; the monocyte/macrophage-specific lectin SLAMFM7 (CRACC), a member of the CD2 family of cell surface receptors implicated in the activation of NK cell–mediated cytotoxicity; ITGB7, a receptor for fibronectin; and MADCAM1, VCAM1, and claudin 23, which plays a major role in tight junction–specific obliteration of the intercellular space (Fig. 2C, 2D). Genes encoding ubiquitination-related molecules, such as culin 1, a component of the SCR (SKP1/CUL1/F-box protein) E3 ubiquitin ligase complex, and ISG15 and its deconjugating protease USP18 were also upregulated. Chemokines and chemokine receptors, such as CXCL10 (IP10), CCR1, CCR5, CCR7, CCL2 (MCP1), and CCL8 (MCPL2), were upregulated as well (Table IV). Additionally, we found upregulation of Fc receptors such as CD64 (FcγRIA), which is also upregulated at the protein level in patient monocytes (29) and represents a marker of Mo-DCs (30).

Among transcripts that we could not formally ascribe to IFN-α (in part due to the limitations of the in vitro experiments), there was a remarkable downregulation of ribosomal protein-encoding transcripts (Fig. 2B, 2D), probably a read-out of cellular stress. Autophagy-related transcripts were not significantly differentially expressed in SLE monocytes, except for DRAM1, a lysosomal modulator of autophagy that plays a central role in p53/TP53-mediated apoptosis (31). This transcript was 2.23-fold upregulated in monocytes from untreated SLE patients (p = 2.32 × 10⁻³). Thus, gene expression profiling of ex vivo SLE monocytes did not provide clues to understand what makes these cells function as DCs based on conventional knowledge.

To directly compare the SLE monocyte transcriptional program with that of blood mDC precursors, we purified lineage-negative HLA-DR^highCD11c^high mDCs and CD14⁺ monocytes from the blood of five healthy donors. Their gene expression profiles were then compared with those of blood SLE monocytes. An unsupervised clustering analysis of transcripts present in >20% of the samples classified healthy monocytes, SLE monocytes, and healthy mDCs into three well-defined groups (Fig. 3A). A supervised analysis was then performed to find genes 1) differentially expressed in healthy mDCs compared with monocytes, and 2) shared by healthy blood mDCs and SLE blood monocytes.

Healthy blood mDCs differentially express 2408 transcripts compared with healthy monocytes (Fig. 3A). Among them, as expected, monocyte and macrophage lineage markers (i.e., CD14 and CD163) were downregulated in mDCs, whereas transcripts encoding molecules involved in Ag presentation and Ag processing, such as HLA class II and asparagine endopeptidase (also known as mammalian legumain), a cysteine protease required for endosomal TLR processing and signaling in DCs (32, 33), were significantly upregulated. Costimulatory molecules such as CD80 and CD86 were not differentially expressed, but transcripts encoding additional molecules associated with immature DC activation, such as the purinergic receptor P2RY14, were highly upregulated. Among Fc receptors, FcγRIIa, which is involved in immune complex internalization, was downregulated, whereas FcεR1A was among the most upregulated transcripts (Fig. 3B). Blood mDCs differentially expressed several members of the TLR family compared with monocytes. Thus, TLR2 and TLR5 were significantly downregulated (Fig. 3C, 3D).

Surprisingly, only 97 of the 2408 transcripts that differentiate healthy blood mDCs from monocytes were similarly expressed in SLE monocytes. Among these, we found genes involved in adhesion (ITGB7, SLAMF7, and AMIGO2), chemotaxis (FPRL2 and CCR5), cytoskeleton remodeling (AKAP2), innate immunity (TLR5), signal transduction (CD2AP and GPR18), and transcription (i.e., ETS1, which is relevant for DC-SIGN promoter activity, ARID5B, and HOXA9) (Fig. 3C). Because the proteins encoded by these genes could not easily explain DC function, we next explored whether contact with T cells might be required for SLE monocytes to acquire an Ag-presenting transcriptional program and/or phenotype.

Blood monocytes from a fraction of SLE patients are able to induce alloreactive CD4⁺ T cell proliferation in vitro (8). Thus, we screened SLE monocytes from 19 SLE patients and selected four that induced CD4⁺ T cell proliferation in vitro and four that did not. CFSE-labeled CD4 T cells (10⁵) were incubated with SLE monocytes (2 × 10⁴). Cells were harvested after 6 h for RNA extraction, at day 2 for flow cytometry, and at day 6 for assessment of T cell proliferation (Fig. 4A). Transcriptional profiles and surface phenotype of MLR⁺- and MLR⁻-inducing monocytes were then compared. Overall, we found 265 transcripts differentially regulated in these two groups (Fig. 4B). All SLE monocytes that induced CD4⁺ T cell proliferation had upregulated MHC class II transcripts after 6 h of culture with T cells at levels similar to those of blood mDCs. Other MLR⁺ monocyte–upregulated transcripts included innate immunity–related molecules such as IL-1β, MARCO, a receptor involved in actin cytoskeleton rearrangement and downregulation of Ag uptake function during DC maturation (34), and CXCL10 (IP-10), a ligand for CXCR3 that it is induced by both type I and type II IFN (35) (Fig. 4C).

These microarray data were further confirmed at two levels. First, cell surface staining of SLE monocytes revealed that upregulation of expression of HLA class II surface protein at 48 h correlated with T cell proliferation at day 6 (Fig. 4D). Second, we found that only alloreaction-inducing SLE monocytes (three of four) secreted increased levels of IL-1β and IL-6 in cocultures with T cells (Fig. 4E). Levels of CXCL10 (IP-10) were also elevated in three of four MLR⁺ cultures, but they did not reach statistical significance when compared with those that did not induce T cell proliferation.

Thus, these studies suggest that blood SLE monocytes require signals derived from T cells to acquire DC function. Additionally, they point toward an important role for innate immunity receptors and/or cytokines, that is, MARCO and IL-1β, in conferring Ag-presenting capacity to monocytes from SLE patients. It has recently been reported that IL-1 has a marked enhancing effect on Ag-specific CD8 T cell expansion, differentiation, migration to the periphery, and memory (36). Whether the type I IFN–rich environment of SLE blood predisposes monocytes to respond to T cell signals and/or to produce the cytokines that we describe above was our next question.

We speculated that exposure of monocytes to SLE serum factors activates some of these cells to subsequently migrate toward the T cell areas of lymphoid organs where they could act as DCs. Thus, these Mo-DC precursors would not circulate in SLE blood but could be identified by exposing healthy monocytes to SLE serum in vitro.

To explore this possibility, monocytes from three healthy donors were cultured for 6 h in the presence of 20% serum from three newly diagnosed, untreated SLE patients. Microarray analysis was then performed upon normalizing the gene expression levels of samples incubated with SLE sera to those incubated with autologous serum. This analysis yielded 2968 transcripts differentially regulated by SLE serum. Functional annotation of these transcripts revealed a significant upregulation of genes involved in the control of apoptosis (i.e., TRADD, FAS, TRAIL, DR6, BIRC3, BNIPL3), adhesion (i.e., JUP, CD47, SLAMF7), pattern recognition (NOD2, TLR7, TLR8), cytokine networks (IL6, IL15, IL4R, IFN-γR2, BLyS/BAFF, APRIL), Fc receptors (FcγRIa, FcγRIIIa), signal transduction and transcription (i.e., STAT1, STAT2, JAK2, IRAK3, SOCS3, IRF5, IRF7, ID2, RUNX3), transport (i.e., TAP1, TAP2, Tapasin), ubiquitination, chemotaxis (CXCL10, CXCR4, CCR7, FPRL2) (Table V), and cell cycle. Many genes within each functional group were found downregulated as well. RNA-processing and protein translation–encoding transcripts were among the most abundant in this category. To determine which genes could be attributed to the effect of type I IFN, we compared the SLE serum-induced transcripts with those resulting from in vitro exposure of healthy monocytes to IFN-α for the same incubation time (6 h). This comparison disclosed 625 common transcripts (∼20% of all SLE serum-induced genes). Of these, about a third (194 transcripts) were equally regulated in freshly isolated blood SLE monocytes.

Of the SLE sera-induced transcripts that would contribute to the migration of DC precursors to lymphoid organs, we focused on the chemokine receptor CCR7. Our data show that 6 h incubation with either IFN-α or SLE serum upregulates the CCR7 transcript in healthy monocytes. To ascertain whether this translates into CCR7 protein expression, CD14⁺ selected healthy monocytes were cultured for 24 h with autologous serum with or without addition of recombinant IFN-α, or with sera from SLE patients with various degrees of clinical activity. Upregulation of CCR7 protein expression under either of these conditions nevertheless was minimal (Fig. 5A). Because TLR4 signaling through the Toll/IL-1R domain–containing adapter inducing IFN-β pathway triggers the expression of CCR7 on murine CD14⁺ monocytes (6), we tested whether this would also happen in human monocytes.

Indeed, we observed that addition of LPS to healthy human serum induces mild CCR7 expression in <24 h on a subset of monocytes, and this effect is further potentiated by IFN-α and SLE sera (Fig. 5A). Furthermore, CCR7 expression resulted in the acquisition of migratory capacity toward its ligands in vitro (Supplemental Fig. 1). These results were reproducible when blood monocytes were obtained through negative rather than positive selection (Fig. 5B), suggesting that triggering through the TLR4 coreceptor CD14 does not play an important role in CCR7 upregulation. Finally, we tested the upregulation of CCR7 on elutriated blood monocytes (fraction V), which contain 5–10% mDCs and plasmacytoid DCs. When cultured under the above-described conditions, CCR7 expression reached the highest levels (Fig. 5C), suggesting that additional factors derived from conventional DCs help to promote the upregulation of this chemokine receptor on monocytes.

Whereas IFN in conjunction with LPS primarily induces CCR7 expression on CD14^bright cells (p < 0.0001), SLE serum does it on both CD14^bright (p = 0.001) and CD14^dim (p = 0.002) cells (Fig. 5D). This most likely reflects a faster monocyte differentiation effect of SLE serum compared with IFN alone. These observations confirm that, as in mice (6), a small population of human blood bona fide monocytes quickly acquires CCR7 expression upon engagement of TLR4. This effect is potentiated by IFN-α and SLE serum, which are sufficient to induce CCR7 transcription but not translation.

Extensive surface phenotyping of the monocytes cultured under the conditions described above showed that both CD14^dim and CD14^bright CCR7⁺ monocytes express similar levels of HLA-DR and lose CD16 expression. As opposed to mouse CCR7⁺ monocytes, they do not express surface DC-SIGN (data not shown). As previously described for cultured human monocytes, CD304 (BDCA-4) was expressed by most (∼80%) of these cells regardless of the level of CD14 expression (37). CD14^bright and CD14^dimCCR7⁺ monocytes differ greatly, however, in the expression of CD1c (BDCA-1), as 50% of CD14^dim cells express this marker whereas only 3% of CD14^bright cells do. This further supports that, in synergism with LPS, SLE serum is more efficient at differentiating monocytes into Mo-DCs with phenotypic characteristics of mDCs. We observed that the CD14^bright population could be subdivided in two additional subpopulations, described as CD14^bright-int and CD14^bright+. In synergism with LPS, IFN induces CCR7 preferentially in the CD14^bright+ population, whereas SLE serum CCR7 induction predominates in the CD14^bright-int subpopulation. In both scenarios, the percentage of CD1c (BDCA-1)-expressing cells is higher in the CD14^bright-int population as compared with CD14^bright+ cells (Supplemental Fig. 2). Overall, this suggests that SLE serum induces a faster and/or greater degree of differentiation of monocytes into Mo-DCs as evidenced by the higher level of CD1c expression, which is a well-recognized marker of mDCs (38).

To ascertain factors that could contribute to the upregulation of CCR7 in blood monocytes, we measured cytokine levels in the supernatant of the 24-h cultures. We found that levels of IFN-α, GM-CSF, IL-1, IL-6, TNF, and IL-10 (determined by Luminex) correlated with CCR7 surface protein expression. Indeed, proinflammatory cytokine levels correlated better than IFN-α with the induction of this receptor. In the absence of LPS, IL-1 showed the highest correlation with CCR7 expression on CD14^dim monocytes (r = 0.57, p = 0.0271), whereas TNF showed the highest correlation with CCR7 on CD14^bright monocytes (r = 0.56, p = 0.0313) (Supplemental Table I).

SLE monocytes are exposed in vivo to a variety of activation factors, including type I IFN, a cytokine family that significantly reprograms their transcriptional profile and promotes DC differentiation in vitro (8). Our studies show, however, that SLE blood monocytes expressing high IFN-inducible signatures do not share transcriptional and/or surface markers with conventional blood mDC precursors. Moreover, contact with T cells seems to be required for these patient monocytes to acquire DC phenotype and function.

Blood monocytes differentially express chemokine receptors that drive them out of the bone marrow (CCR2) and permit them to recirculate (CCR5) and patrol the vasculature (CX₃CR1) (39, 40). Collectively, these cells lack expression of CCR7, which is the main driver to T cell areas of secondary lymphoid organs. CX₃CR1^hi “patrolling” monocytes correspond to CD14^dimCD16⁺ cells. CX₃CR1^int CD14⁺CD16⁺ and CX₃CR1⁻ CD14⁺CD16⁻ monocytes traffic to sites of infection and inflammation (41). Murine CD14^hi monocytes exposed to LPS have been shown to give rise to a distinct population of TNF and inducible NO synthase–producing inflammatory Mo-DCs in vivo that expresses CCR7 (6, 42). In a murine model of colitis, inflammatory monocytes that infiltrate the lamina propria can also give rise to a phenotypically and functionally distinct CX₃CR1^int population displaying migratory DC hallmarks, including the uptake and processing of orally acquired Ags and priming of naive CD4⁺ T cells. Furthermore, these cells acquire CCR7 expression that endows them with the capacity to migrate out of the colonic lamina propia into the tissue-draining lymph nodes (43). In humans, a unique DC population present in inflammatory environments, including RA synovial fluid, shares gene signatures with in vitro Mo-DCs and secretes Th17 cell–polarizing cytokines that might contribute to disease pathogenesis (7). Whether these cells can acquire CCR7 expression is not known.

In this study, we show that exposure of healthy blood monocytes in vitro to either type I IFN or sera from SLE patients induces the fast (6 h) transcriptional upregulation of CCR7. Additional signals are required, however, to induce monocyte CCR7 protein expression. Thus, addition of LPS to IFN-α or SLE serum induces the expression of functional CCR7 on a subset of monocytes in <24 h. SLE serum further downregulates CD14 and upregulates expression of BDCA-1, an mDC marker. Levels of proinflammatory cytokines (IL-1β and TNF) in the supernatant of these cultures correlate with the induction of CCR7, suggesting that they might be required, in addition to type I IFN, for Mo-DCs to acquire a migratory phenotype.

Monocytes from active adult SLE patients have been reported to display unique phenotypic changes, including lower levels of CD14 and HLA-DR, upregulation of CD71, and higher production of TNF compared with monocytes from healthy controls. Additionally, their exposure to GM-CSF and IL-4 leads to defective Mo-DC differentiation (44). Conversely, our present studies show than the phenotype of blood monocytes from pediatric SLE patients extensively overlaps with that of healthy controls. We observed nonetheless a mild downregulation of HLA-DR and CD81 and upregulation of CD64 and CD62L expression, especially on CD14^brightCD16⁻ cells. We also detected a mild expansion of CD14^brightCD16⁺ monocytes, a subpopulation previously linked to inflammatory diseases (45–47), including RA (48).

In an attempt to identify unconventional molecules and/or pathways that might explain the APC function of SLE blood monocytes, we turned to transcriptome analyses. This approach revealed that blood monocytes from newly diagnosed and untreated SLE patients express a unique transcriptional profile, a fraction of which (∼35%) could be explained by exposure to type I IFN. Among IFN-inducible programs, upregulation of transcripts encoding apoptosis-inducing molecules such as TRAIL and FAS might explain the accelerated apoptosis rate described for SLE monocytes in vitro, as well as the proapoptotic effect of SLE serum (49–51). Chemokine receptors such as CXCL10 (IP10) and CCL2 (MCP1) were also found significantly upregulated. IP-10 is chemotactic for activated Th1 cells and together with IL-6 induces plasma cell differentiation (52). Furthermore, levels of serum IP-10 correlate with disease activity (53, 54) and are found increased in the cerebrospinal fluid of neuropsychiatric SLE patients (55). CCL2 (MCP-1) attracts monocytes, T cells, NK cells, and basophils. Increased levels of serum and cerebrospinal fluid CCL2 (56) have also been reported in patients with neuropsychiatric SLE (57).

Our studies suggest that, in addition to type I IFN, blood SLE monocytes require signals generated upon contact with T cells to acquire DC function. Among them, we found innate immunity receptors and cytokines such as IL-1β and IL-6. The capacity of activated T cells to induce IL-1 production in monocytes has been reported in the context of Ag-independent responses. Indeed, previous studies suggest that a cell contact mechanism leads to IL-1 production (58). Type I IFNs, on the contrary, inhibit IL-1 production through two distinct mechanisms. First, these cytokines induce a STAT1-dependent inhibition of NLRP3 and NLRP1 inflammasome activity, thereby suppressing caspase-1–dependent IL-1β processing. Second, type I IFNs enhance the production of IL-10, which by signaling through STAT3 in turn decreases the levels of pro–IL-1α and pro–IL-1β (59). How type I IFN and proinflammatory cytokine networks interact in the context of SLE to promote Mo-DC differentiation remains to be elucidated, but myeloid SLE DC-intrinsic defects could contribute to this phenotype. In this regard, SLE risk-associated polymorphisms in Blipm1 predispose to Mo-DC activation through the upregulation of microRNA let-7c, which in turn downregulates SOCS-1 and permits a robust proinflammatory cytokine response upon LPS triggering (60).

In agreement with our observation that SLE sera drive Mo-DC differentiation, an exuberant transcriptional program, which includes the downregulation of CD14 and the upregulation of CD83, is induced when healthy monocytes are incubated with patient sera. These include both known type I IFN–inducible as well as type I IFN–independent transcriptional pathways. Thus, SLE sera upregulate genes involved in apoptosis, adhesion, chemotaxis, pattern recognition receptors, Fc receptors, RNA and protein modification, signaling, and transcription. Upregulation of innate immunity receptors, such as TLR7 and TLR8, might cause SLE monocytes to become hyperresponsive upon exposure to pathogens and/or to endogenous agonists such as immune complexes containing nucleic acids (61, 62). Genes involved in signal transduction upstream of NF-κB, such as the SLE candidate susceptibility gene IRF5, are also upregulated (63, 64) together with genes encoding molecules involved in nucleosome and chromatin modification, RNA processing, and ubiquitination. Importantly, among the most differentially upregulated transcripts, we found those encoding chemokine receptors, including IP10, CXCR4, and CCR7.

CCR7 regulates the homing of DCs and T cells to lymphoid organs (65), and several CCR7-dependent mechanisms drive the induction and perpetuation of rheumatic diseases such as RA (66). In the present study, we found that monocyte CCR7 transcription is upregulated by both SLE serum and IFN-α, even though the CCR7 promoter does not encode IFN response elements. These conditions, however, do not upregulate CCR7 protein expression to a significant extent. The TLR4 agonist LPS synergizes with IFN-α, and especially with factors in patient sera, to induce CCR7 protein expression on a subset of monocytes in <24 h. Microbial stimulation and LPS have been reported to induce the full differentiation of Mo-DCs in mice in the same time period. These Mo-DCs acquire DC morphology and localize to T cell areas via l-selectin and CCR7, and DC-SIGN/CD209a is a marker of these cells (6). A distinct human Mo-DC subset described in RA synovial fluid and thought to derive from monocytes characteristically lacks DC-SIGN expression (67), which is a feature of the CCR7⁺ monocytes that we observe upon culture with SLE sera.

Infections can trigger SLE flares (68). Although this phenomenon is poorly understood, it is conceivable that infectious organisms induce autoimmune responses by mechanisms such as molecular mimicry in genetically predisposed individuals (68, 69). Bacterial endotoxin/LPS (70) is one of the environmental factors that may contribute to the development and/or exacerbation of autoimmune diseases, as it has been shown in spontaneous murine lupus models (71). Our results suggest that one of the plausible explanations for this phenomenon might be the capacity of LPS to synergize with type I IFN, and possibly additional SLE-related factors, to upregulate chemokine receptors such as CCR7 on monocytes. This would drive these cells to interact with T cells and acquire APC functions that would induce and/or perpetuate pathogenic autoimmune loops.

We thank Michelle Edens and Estrella Thomas for transferring clinical samples and obtaining clinical data. We thank Florentina Marches for sharing technical expertise. We thank Elizabeth Trahan, Sebastien Coquery, and colleagues in the Flow Cytometry Core, Sandra Zurawski in the Luminex Core, Lynnette Walters in the Cell Processing Core, and Shannon Lunt in the Sample Core at the Baylor Institute for Immunology Research for assistance.

The authors have no financial conflicts of interest.