Peripheral blood monocytes (Mo) originate from hematopoietic progenitor cells in bone marrow and play important roles in innate and adaptive immunity due to their ability to differentiate into macrophages (MΦ) and dendritic cells (DC) 1234567. The heterogeneity and plasticity of MΦ and DC result from their differentiation in specific tissue microenvironments 8910. The expression of CD16 (FcγRIII) distinguishes two Mo subsets in peripheral blood of healthy individuals: a major CD16^- subset (80–95%) and a minor CD16⁺ subset (5–15%) 11. Compared to classical CD16^- Mo, CD16⁺ Mo exhibit a more MΦ-like morphology, produce higher levels of TNF and IL-1 1213, have higher antigen presenting potential 141516, and differentiate into DC upon transendothelial migration in vitro 17. CD16⁺ Mo express CX3CR1 and migrate in response to CX3CL1 1819, a membrane-bound chemokine expressed on inflamed endothelial cells, while CD16^- Mo express CD62L and CCR2 and migrate in response to CCL2 1820, which mediates Mo migration from bone marrow and recruitment to inflammatory sites 221. CD16⁺ Mo produce IL-6, CCL2, and matrix metalloproteinase-9 upon interaction with CX3CL1-expressing endothelial cells 22 and activate resting T-cells for HIV infection by producing CCR3 and CCR4 ligands 23. Together, these findings suggest that CD16⁺ and CD16^- Mo are recruited into different anatomic sites under constitutive or inflammatory conditions, and play distinct functional roles in immunity and disease pathogenesis.

A dramatic increase in circulating CD16^- Mo has been reported in inflammatory pathologies such as sepsis, HIV infection, tuberculosis, and asthma 112425. Studies of patients infected with Mycobacterium leprae demonstrated that CD16⁺ and CD16^- Mo differentiate into DC-SIGN⁺ MΦ and CD1b⁺DC-SIGN^- DC, respectively, and the presence of CD1b⁺DC-SIGN^- DC in M. leprae lesions was associated with healing 26. An increased frequency of CD16⁺ Mo was associated with non-healing Leishmania chagasi lesions 27. Thus, CD16⁺ and CD16^- Mo differentiation into MΦ or DC subpopulations with distinct phenotypes influences host defenses in infectious disease. Consequently, there is interest in developing therapeutic strategies that target specific Mo subpopulations 682829.

Mo heterogeneity is conserved across mammalian species 781930. In mice, Gr1⁺CX3CR1^low Mo (homolog of human CD16^- Mo) are recruited into the peritoneal cavity or draining lymph nodes under inflammatory conditions by mechanisms dependent on CCR2 and CD62L, and subsequently differentiate into DC 19313233. In contrast, Gr1^-CX3CR1^high (homolog of human CD16⁺ Mo) are constitutively recruited into peripheral tissues including spleen, gut, lungs, and brain 19. Gr1^-CX3CR1^high patrol vascular endothelium by mechanisms involving LFA-1 and CX3CR1, and are rapidly recruited into inflamed tissues where they differentiate into MΦ expressing the transcription factors cMaf and MafB and transiently producing TNF-α 33. Studies in CX3CR1/ApoE double knockout mice suggest that CX3CR1^high Mo play a critical role in development of atherosclerotic lesions 3435. CX3CR1⁺ Mo may be precursors for lamina propria DC, which depend on CX3CR1 to form transepithelial dendrites, enabling direct sampling of luminal antigens 36. Furthermore, adoptive transfer studies in rats demonstrated that CCR2^lowCX3CR1^high Mo are constitutively recruited into the gut where they give rise to intestinal lymph DC 37. Studies on the origin of myeloid pulmonary DC demonstrated that Ly-6C^highCCR2^highCX3CR1^low Mo differentiate into CD103⁺ DC 3839), whereas Ly-6C^lowCCR2^lowCX3CR1^high Mo give rise to CD11b^high DC 3940. Thus, CX3CR1^high and CX3CR1^low Mo subsets play distinct functional roles under constitutive and inflammatory conditions.

The developmental relationship between Mo subsets is poorly understood. In mice, Mo recently emigrating from bone marrow exhibit a Ly-6C^high phenotype and gradually downregulate Ly-6C 41. Mo acquire CD16 expression upon exposure to M-CSF 42, TGF-β 1743, or IL-10 44, and upregulate CX3CR1 expression upon CCL2 stimulation via CCR2 45. Mo differentiation is associated with decreased CCR2 expression and increased CCL2 production 46. Engrafted Gr1^highCX3CR1^low Mo in peripheral blood traffic to the bone marrow, differentiate into Gr1^lowCX3CR1^high Mo, and contribute to mucosal, but not splenic, generation of DC 5. These findings suggest that human CD16⁺ Mo and mouse Ly-6C^lowCCR2^lowCX3CR1^high Mo differentiate from CD16^- Mo and Gr1^-Ly-6C^highCCR2^highCX3CR1^low Mo, respectively.

Here, we investigate the developmental and functional relationship between CD16⁺ and CD16^- Mo subsets. Whole genome transcriptome analysis suggests that these Mo subsets originate from a common myeloid precursor, with CD16⁺ Mo being at a more advanced stage of differentiation and having a more MΦ – and DC-like transcription program. Upregulation of the transcription factors RARA and KLF2 in CD16⁺ Mo coincided with the absence of cutaneous lymphocyte associated antigen (CLA) expression, indicating potential imprinting for non-skin homing in CD16⁺ Mo. These results define distinct transcriptional profiles of CD16^- and CD16⁺ Mo subsets suggesting different stages of myeloid differentiation, new markers to distinguish these Mo subpopulations, and unique roles in immune responses and inflammatory diseases.

In this study, we define transcriptional profiles of human CD16⁺ and CD16^- monocytes (Mo) and provide new insights into their developmental relationship and biological functions. Despite remarkable transcriptional similarity (approximately 83%), a significant number of transcripts were differentially expressed (n = 2,759), with 228 and 250 >2-fold upregulated and downregulated, respectively, in CD16⁺ compared to CD16^- Mo. Differentially expressed genes related to cell-to-cell adhesion and trafficking, immune responses and inflammation, metabolism and stress response, signaling and signal transduction, cell cycle, proliferation, and differentiation, cytoskeleton, and regulation of transcription. Gene set enrichment analysis (GSEA) demonstrated that CD16⁺ Mo are enriched in genes related to NK-mediated cytotoxicity, inositol phosphate metabolism, actin binding, and oxidative stress, while CD16^- Mo are enriched in genes related to hematopoietic cell lineage, receptor-mediated endocytosis, arginine and proline metabolism, NTHi pathway, and lipid binding. The transcriptional profiles suggest that CD16⁺ and CD16^- Mo subsets originate from a common myeloid precursor, with CD16⁺ Mo being at a more advanced stage of myeloid differentiation and having distinct biological functions in vivo.

Previous studies in mice provide evidence for a developmental relationship between Ly6C^highCCR2^highGr1⁺CX3CR1^low and Ly6C^lowCCR2^lowGr1^-CX3CR1^high Mo (homologs of human CD16^- and CD16⁺ Mo, respectively), with Ly6C^lowCCR2^lowGr1^-CX3CR1^high Mo being more mature and derived from Ly6C^highCCR2^highGr1⁺CX3CR1^low Mo 54185. Likewise, studies on human Mo demonstrated the ability of CD16^-CX3CR1^low Mo to differentiate into CD16⁺CX3CR1^high Mo upon stimulation with TGF-β, IL-10, M-CSF, or CCL2 17434445. Our comparative transcriptome analysis provides further evidence for the idea that CD16^- Mo originate from a common granulocyte-macrophage (GM) precursor and give rise to CD16⁺ Mo, which are more closely related to macrophages (MΦ) and dendritic cells (DC). CD16^- Mo preferentially expressed granulocyte-associated transcripts (i.e., CSF3R, formyl peptide receptor 1 (FPR1), the calgranulins S100A8, S100A9, and S100A12), and myeloid markers (i.e., CD14, MNDA, TREM-1, CD1d, and C1qR1/CD93), together with transcripts suggesting an increased potential for receptor-mediated endocytosis via molecules such as CD14 and FCGR1A/CD64 8586. In contrast, CD16⁺ Mo preferentially expressed MΦ (i.e., CSF1R/CD115, MafB, EGF module-containing mucin-like hormone receptor (EMR)1-3, CD97, and C3aR) 86 and DC markers (i.e., SIGLEC10, CD43, CXCL16, and RARA) 56748788. CD16⁺ Mo expressed higher levels of transcripts encoding the cysteine protease cathepsin L (CTSL), which contributes to phagocytic-endocytic proteolysis in DC for subsequent antigen presentation 64. Upregulation of transcripts encoding dipeptidyl-peptidase I, CTSC 89 may further enhance antigen processing by CD16⁺ Mo or DC derived from these cells. Although some studies classified CD16⁺ Mo as DC based on their increased antigen presenting ability 141516 and transcriptional profile similarities 16, a recent compendium analysis of transcriptional profiles demonstrated that CD16⁺HLA-DR⁺ cells are more closely linked to myeloid CD14⁺ cells than to DC subsets in peripheral blood 90. Our results demonstrate that CD16⁺ Mo share approximately 83% of their transcripts with CD16^- Mo, supporting the idea that these two Mo subsets are developmentally related.

Recruitment of CD16⁺ and CD16^- Mo into tissues is mediated via distinct molecular mechanisms 78. Our gene expression analysis confirms differential expression of adhesion molecules and chemokine receptors previously reported to be preferentially expressed on CD16⁺ Mo (i.e., LFA-1, PECAM/CD31, CX3CR1) and CD16^- Mo (i.e., CCR1, CCR2, and L-selectin/CD62L) 181920. We also identified new cell surface markers and other molecules that are differentially expressed in these Mo subsets and may influence their trafficking and migration into tissues. The tetraspanins MS4A4A and MS4A7, adhesion molecules SIGLEC10 and ICAM-2, and membrane-bound chemokine CXCL16 56 were preferentially expressed by CD16⁺ Mo, whereas the tetraspanin MS4A6A, adhesion molecules CD99 and junctional adhesion molecule like (JAML or AMICA) 57, and chemokine receptor FPR1 were preferentially expressed by CD16^- Mo. CD31 and CD99 are involved in distinct steps of Mo transendothelial migration 9192. Expression of CXCL16, a chemokine expressed by DC 56, on the surface of CD16⁺ Mo may facilitate interaction with CXCR6⁺ cells (i.e., NKT and activated CD4⁺ and CD8⁺ T-cells 56) and retention of CXCR6⁺ cells in tissues. Similar to mouse and rat CCR2^lowCX3CR1^high Mo 3741, CD16⁺ Mo expressed higher levels of SPN/CD43 (sialophorin, leukosialin, large sialoglycoprotein or gp115), a ligand for ICAM-1 9394, and the macrophage adhesion receptor sialoadhesin (Siglec-1) 95. CD43 has both adhesive and anti-adhesive properties 96, mediates DC maturation 88, and contributes to regulation of immunological synapse formation 97. CD47, a receptor for thrombospondin-1 (TSP-1), is preferentially expressed by CD16⁺ Mo. CD47 ligation selectively inhibits the development of human naive T cells into Th1 effectors by decreasing IL-12 and TNF-α production by Mo-derived DC 9899. Consistent with these findings, CD16^- and CD16⁺ Mo may induce Th1 and Th2-like differentiation, respectively 100. However, CD16⁺ Mo express IL-12RB1, which favors Th1 polarization 58, whereas CD16^- Mo express receptors for the Th2 cytokines IL-6 and IL-13 59 and the anti-Th1 cytokine, IL-27 6061. Accordingly, the influence of CD16⁺ and CD16^- Mo on Th1 versus Th2 polarization of immune responses is likely to be highly dependent on the local microenvironment within tissues.

CD16⁺ Mo expressed high levels of transcripts for RARA, which controls transcription of genes involved in cell trafficking and mucosal homing. RA imprints lymphocytes with non-skin mucosal homing properties by decreasing cutaneous lymphocyte-associated antigen (CLA, an epitope on PSGL-1) expression 83 and increasing expression of CCR9 and integrin beta 7, two mucosal addressins 77. RA also controls reciprocal differentiation of Th17 and regulatory T cells 101, modulates myeloid gene expression and differentiation 75, and regulates survival and antigen presentation by DC 74. Consistent with our hypothesis that the RARA pathway is activated in CD16⁺ Mo, we demonstrated CLA downregulation on these cells, together with upregulation of two RA-induced targets: SLP-76 75 and CXCL16 102. RA induces mucosal-type DC, which produce TGF-β and thereby imprints T-cells for gut homing by inducing CCR9 and integrin beta 7 76. CD16⁺ Mo-derived MΦ and DC constitutively produce TGF-β 23100, but whether they also instruct T-cells for gut homing remains to be determined.

KLF2 mRNA is expressed at very high levels and significantly upregulated in CD16⁺ compared to CD16^- Mo. KLF2 belongs to a family of zinc-finger transcription factors that is induced by PI3K signaling 103 and controls expression of several genes including those coding for CD62L, CCR7, integrin beta7, sphingosine-1-phosphate receptor (S1PR1) 73, and lymphotoxin beta 104. CCR7 and CD62L are essential for migration into lymph nodes, S1P1 regulates T-cell thymic egress and recirculation 105, and integrin beta 7 mediates cell recruitment into Peyer's patches and mesenteric lymph nodes 106. Together, these findings raise the possibility that preferential expression of KLF2 in CD16⁺ Mo may confer an increased potential for trafficking.

CD16⁺ compared to CD16^- Mo express very high levels of transcripts for cyclin-dependent kinase inhibitor 1C (CDKN1C or p57/KIP2) (18.4-fold increase) and metastasis suppressor 1 MTSS1 (5.7-fold increase). CDKN1C is a potent inhibitor of several G1 cyclin-dependent kinase (cdk) complexes, and negative regulator of G1/S cell cycle transition and cell proliferation 66. CDKN1C 66 and MTSS1 68 are candidate tumor suppressor genes, and their high expression is consistent with the inability of Mo to proliferate 7. CDKN1C is induced by TGF-β 67, a cytokine known to induce CD16⁺ Mo differentiation 1743. Thus, our results are consistent with a potential link between TGF-β pathway activation and CD16⁺ Mo differentiation in vivo.

Several transcripts related to cell activation were upregulated in CD16⁺ Mo including LTB, TNFRSF8, leukocyte specific transcript 1 (LST1), IFITM1-3, HMOX1, superoxide dismutase-1 (SOD-1), tryptophanyl tRNA synthetase (WARS), and monoglyceride lipase (MGLL), indicating increased activation of CD16⁺ compared with CD16^- Mo. LST1 107, HMOX1 108, SOD-1, and WARS are induced by stimulation with lipopolysaccharide 109. The role of LST1 in immune regulation remains elusive. HMOX1 modulates Mo inflammatory responsiveness by decreasing LPS-induced TNF and IL-1β expression 108. WARS and indoleamine 2,3-dioxygenase (IDO) are responsible for tryptophan use in protein synthesis and degradation, respectively 63. WARS was identified as a molecular marker for Mo differentiation into MΦ 110 and DC 111. These results suggest increased activation of CD16⁺ compared to CD16^- Mo in vivo.

The CD16⁺ Mo subset includes two subsets with distinct levels of CD14 expression: CD14^highCD16⁺ and CD14^lowCD16⁺ 11112. CD14^highCD16⁺ Mo exhibit a phenotype intermediate between that of CD14^highCD16^neg and CD14^lowCD16⁺ Mo in terms of adhesion molecule (e.g., CL62L) and chemokine receptor expression (e.g., CCR2, CXCR2, and CX3CR1) 18. Both CD14^highCD16⁺ and CD14^lowCD16⁺ Mo contributed to the transcriptional profile of CD16⁺ Mo in this study. The expression of some genes we identified as markers for CD16⁺ Mo may be distinct on CD14^highCD16⁺ and CD14^lowCD16⁺ Mo. Consistent with this prediction, we demonstrated intermediate expression of CD115 and CD114 on CD14^highCD16⁺ Mo compared to CD14^highCD16^neg and CD14^lowCD16⁺ Mo, and high expression of CD93 and C3aR1, similar to that on CD14^highCD16^- and CD14^lowCD16⁺ Mo, respectively (Additional file 3). These findings suggest a developmental relationship between these Mo subsets in which CD14^highCD16^neg Mo, CD14^highCD16⁺ Mo, and CD14^lowCD16⁺ Mo represent sequential stages of monocyte differentiation 41.

Comparative transcriptome analysis of CD16⁺ and CD16^- Mo indicates that CD16⁺ Mo represent a more advanced stage of myeloid differentiation with a more MΦ – and DC-like transcription program, whereas CD16^- Mo are more closely related to a common myeloid precursor. Given the ability of CD16⁺ and CD16^- Mo to be recruited into specific tissues via distinct mechanisms, these Mo subsets are likely to give rise to DC and MΦ subpopulations with distinct phenotypes and roles in immunity and disease pathogenesis. Further studies to characterize phenotypic differences between CD16⁺ and CD16^- Mo-derived DC and MΦ are relevant for development of DC-based vaccines, and will also provide a better understanding of their functional roles in immune responses, inflammation, and disease pathogenesis.

The authors declare that they have no competing interests.

PA designed and performed experiments, analyzed and interpreted data, prepared graphics, and wrote the manuscript. VM generated heat maps, performed statistical analysis for differentially expressed genes, and drafted the Methods for Figure 5. KYL and XZ performed statistical analysis of microarray data and drafted the Methods and Results. KYL classified genes based on biological functions and performed GSEA. VSW and AG carried out experiments in Figures 4 and 6 and drafted the Methods and Results. DG conceived the study, designed experiments, analyzed and interpreted data, and wrote the manuscript. All authors revised and gave final approval for publication of the manuscript.