Folate Receptor β (FRβ) Expression in Tissue-Resident and Tumor-Associated Macrophages Associates with and Depends on the Expression of PU.1

As macrophages exhibit a huge functional plasticity under homeostasis and pathological conditions, they have become a therapeutic target for chronic inflammatory diseases. Hence, the identification of macrophage subset-specific markers is a requisite for the development of macrophage-directed therapeutic interventions. In this regard, the macrophage-specific Folate Receptor β (FRβ, encoded by the FOLR2 gene) has been already validated as a target for molecular delivery in cancer as well as in macrophage-targeting therapeutic strategies for chronic inflammatory pathologies. We now show that the transcriptome of human macrophages from healthy and inflamed tissues (tumor; rheumatoid arthritis, RA) share a significant over-representation of the “anti-inflammatory gene set”, which defines the gene profile of M-CSF-dependent IL-10-producing human macrophages (M-MØ). More specifically, FOLR2 expression has been found to strongly correlate with the expression of M-MØ-specific genes in tissue-resident macrophages, tumor-associated macrophages (TAM) and macrophages from inflamed synovium, and also correlates with the presence of the PU.1 transcription factor. In fact, PU.1-binding elements are found upstream of the first exon of FOLR2 and most M-MØ-specific- and TAM-specific genes. The functional relevance of PU.1 binding was demonstrated through analysis of the proximal regulatory region of the FOLR2 gene, whose activity was dependent on a cluster of PU.1-binding sequences. Further, siRNA-mediated knockdown established the importance of PU.1 for FOLR2 gene expression in myeloid cells. Therefore, we provide evidence that FRβ marks tissue-resident macrophages as well as macrophages within inflamed tissues, and its expression is dependent on PU.1.


Melanoma Xenograft Model
Immunodeficient NOD-scid-IL2Rg null (NSG) mice (The Jackson Laboratory, Bar Harbor, ME 04609, USA) were maintained under specific pathogen-free conditions. Male mice (4-6 weeks of age) were subcutaneously inoculated with 10 6 BLM melanoma cells. When tumors reached approximately 1 cm in width (approximately at day 14th), mice were euthanized and tumors were resected and frozen for histologic analyses. This procedure was approved by the IiSGM animal care/use and Comunidad de Madrid committees (PROEX-084/18).

Confocal Microscopy and Immunohistochemistry
Normal skin samples were obtained from abdominoplasty. Normal colon and muscle samples were localized adjacent to tumor and obtained from colon adenocarcinoma and melanoma patients. Informed consent was obtained, and all the procedures were performed following Medical Ethics Committee (Hospital General Universitario Gregorio Marañón) guidance. Thick sections (4 µm in depth) of cryopreserved tissue were first blocked for 10 min with 1% human immunoglobulins and then incubated for 1 h with either a mouse monoclonal antibody against human FRβ [52], a rat monoclonal antibody against murine FRβ [48], an anti-CD163 monoclonal antibody (clone Ber-Mac3, MBL International Corp., Woburn, MA 01801, USA), an anti-Von Willebrand factor (rabbit polyclonal, Dako, Santa Clara, CA 95051, USA), an anti-F4/80 (clone BM8, labeled with Alexa Fluor 647, Biolegend, San Diego, CA 92121, USA), or isotype-matched control antibodies. All primary antibodies were used at 1-5 µg/mL, followed by incubation with Cy3-labeled anti-mouse and Cy5-labeled anti-rabbit secondary antibodies. Tissues were imaged with the 20X PL-APO NA 0.7 immersion objective of a confocal scanning inverted AOBS/SP2-microscope (Leica Microsystems, Wetzlar, 35578 Germany).

Quantitative Real Time RT-PCR
Oligonucleotides for selected genes were designed according to the Roche software for quantitative real-time PCR, and RNA was amplified using the Universal Human Probe Roche Library (Roche Diagnostics, Indianapolis, IN 46256, USA). Assays were made in triplicate and results normalized according to the expression levels of GAPDH. In all cases, the results were expressed using the ∆∆CT method for quantitation.

siRNA-Mediated Knockdown
THP-1 cells (2 × 10 6 cells) were nucleofected with 3 µg of siRNA for PU.1 (sc-36330 PU.1 siRNA gene silencer; Santa Cruz Biotechnology, Dallas, Texas 75220 USA) or a negative control siRNA (sc-37007 Control siRNA-A; Santa Cruz Biotechnology, Dallas, Texas 75220 USA) using the Cell Line Nucleofector Kit V (Amaxa, Cologne, Germany). After nucleofection, cells were kept in culture for 24 h, and one-fifth of the cells were lysed and subjected to Western Blot for PU.1 detection. Total RNA was isolated from the remaining nucleofected cells and subjected to real time-PCR.

Bioinformatic Analysis
The genes selectively expressed by monocytes and macrophages from human gut were obtained from [55] and used to identify those genes contained within the "Pro-inflammatory gene set" and "Anti-inflammatory gene set" previously defined [19,29]. A list of genes specifically expressed by macrophages within melanoma [56] and head and neck squamous carcinoma [57] was derived using Cibersortx [58], and their expression in breast carcinoma determined using the METABRIC (Molecular Taxonomy of Breast Cancer International Consortium) study cohort [59,60] on the cBioPortal for Cancer Genomics [61] and using the TIMER resource [62-64] on data generated by The Cancer Genome Atlas Program Research Network [65]. Identification of genes co-expressed with FOLR2 in various tissues was done using Genevestigator® [66]. Gene ontology analysis of the defined gene sets was performed using the online tool ENRICHR [67,68]. Chip-seq data were derived from the Cistrome data browser [69] and processed using the WashU Epigenome Browser [70].

Statistical Analysis
Statistical analysis was performed using a paired Student´s t-test and a p value < 0.05 was considered significant.

Folate Receptor Beta (FRβ) is Co-Expressed with other Genes of the "Anti-Inflammatory Gene Set" and Marks Human Tissue-Resident Macrophages
We have previously reported the existence of transcriptional overlaps between the variety of macrophage subsets that make up Tumor-Associated Macrophages (TAM) and M-CSF dependent macrophages, as both are enriched in the expression of the 170-gene "anti-inflammatory gene set" [19,29], which includes the FRβ-encoding FOLR2 gene [45]. Assessment of the genes significantly co-expressed with FOLR2 in 431 anatomical locations identified various genes of the "anti-inflammatory gene set", including CD209, C1QC, CD163, LILRB5, F13A1, STAB1, RNASE1 and IGF1, with Pearson´s correlation coefficients ranging from 0.77 to 0.65 ( Figure 1A). In line with these findings, analysis of monocytes or macrophages from human colon, whose transcriptomes have been extensively analyzed [55], also revealed an enrichment of the "anti-inflammatory gene set" (Figure 1B), also including the expression of CD209, C1QC, CD163, LILRB5, F13A1 and IGF1 from the "anti-inflammatory gene set" ( Figure 1C). Given these results, FOLR2-encoded FRβ expression was evaluated in colon and other tissue macrophages under homeostatic conditions. In agreement with the transcriptional data, and using tissue arrays, an FRβ-specific antiserum [54] stained numerous cells in the lamina propria of the colon, where CD68+ macrophages were also detected ( Figure 1D). Besides, FRβ+ cells were found in the paracortical area of the tonsil and, to a lesser extent, in skeletal muscle ( Figure 1D). Further, multicolor immunofluorescence revealed that FRβ is co-expressed with the hemoglobin/haptoglobin scavenger receptor CD163 in lamina propria macrophages, as well as in tonsil and in skeletal muscle, thus indicating that FRβ marks tissue-resident macrophages ( Figure 1E). In addition, FRβ was co-expressed with CD163 in the dermis ( Figure 1F), where most FRβ+/CD163+ macrophages exhibited a perivascular distribution ( Figure 1F), as well as in placenta [71] ( Figure 1G). Therefore, FRβ is broadly expressed in vivo by tissue-resident macrophages, where its expression positively correlates with the macrophage marker CD163 and other genes of the "anti-inflammatory gene set". Cells 2020, 9,1445 5 of 17 been extensively analyzed [55], also revealed an enrichment of the "anti-inflammatory gene set" ( Figure 1B), also including the expression of CD209, C1QC, CD163, LILRB5, F13A1 and IGF1 from the "anti-inflammatory gene set" ( Figure 1C). Given these results, FOLR2-encoded FRβ expression was evaluated in colon and other tissue macrophages under homeostatic conditions. In agreement with the transcriptional data, and using tissue arrays, an FRβ-specific antiserum [54] stained numerous cells in the lamina propria of the colon, where CD68+ macrophages were also detected ( Figure 1D). Besides, FRβ+ cells were found in the paracortical area of the tonsil and, to a lesser extent, in skeletal muscle ( Figure 1D). Further, multicolor immunofluorescence revealed that FRβ is co-expressed with the hemoglobin/haptoglobin scavenger receptor CD163 in lamina propria macrophages, as well as in tonsil and in skeletal muscle, thus indicating that FRβ marks tissue-resident macrophages ( Figure  1E). In addition, FRβ was co-expressed with CD163 in the dermis ( Figure 1F), where most FRβ+/CD163+ macrophages exhibited a perivascular distribution ( Figure 1F), as well as in placenta [71] ( Figure 1G). Therefore, FRβ is broadly expressed in vivo by tissue-resident macrophages, where its expression positively correlates with the macrophage marker CD163 and other genes of the "anti-inflammatory gene set".

FOLR2/FRβ Expression Marks Human Tumor-Associated Macrophages and Correlates with the Expression of CD163 and Regulators of Macrophage Differentiation
To determine whether the macrophage-restricted expression of FRβ also applies to pathological settings, we next assessed FOLR2 expression in TAM from various tumor types. To this end, we initially searched for macrophage-specific gene expression in melanoma [57] and head and neck squamous cell carcinoma (HNSC) [58] (Figure 2A) and identified a set of 21 M-MØ-specific genes whose expression is also restricted to melanoma and HNSC TAM ( Figure 2B). Interestingly, FOLR2, CD163, LILRB5, CD209 and C1QC were identified as genes of the "anti-inflammatory gene set", whose expression is also seen in tissue-resident macrophages and tumor-associated macrophages ( Figure 2B). Analysis of breast cancer transcriptomes (METABRIC cohort) also evidenced a very good correlation between the expression of FOLR2 and those of genes of the "anti-inflammatory gene set", which reached statistical significance in most cases ( Figure 2C), and was highly significant for CD163 (Pearson: 0.72; p = 1.71 × 10 −308 ) and even CD68 (Pearson: 0.59; p = 1.25 × 10 −176 ), another widely used marker for macrophage identification ( Figure 2C). Further, a significant correlation was found between the expression of FOLR2 and CD163 using the TCGA cohorts for breast carcinoma, melanoma and HNSC ( Figure 2D and not shown), whereas no correlation was seen between FOLR2 and the epithelial-specific EPCAM gene in any of the analyzed tumors (data not shown). In fact, and at the protein level, FRβ expression was observed in CD163+ cells in melanoma ( Figure 2E) and in areas enriched in CD68+ cells in colon adenocarcinoma ( Figure 2F). Interestingly, and since FRβ+ macrophages are prominent in the tumor-invasive front of pancreatic cancer and associate with poor prognosis [72], it is worth noting that FRβ+ macrophages were mostly detected in the peritumoral area both in human melanoma and in a melanoma xenograft mouse model ( Figure 2E). Altogether, this set of results indicates that FOLR2 expression correlates with the expression of CD163 and other macrophage-specific genes in TAM, and that FRβ expression in TAM overlaps with the expression of the commonly used macrophage markers CD163 and CD68. Cells 2020, 9, 1445 6 of 17

FOLR2/FRβ Expression Marks Human Tumor-Associated Macrophages and Correlates with the Expression of CD163 and Regulators of Macrophage Differentiation
To determine whether the macrophage-restricted expression of FRβ also applies to pathological settings, we next assessed FOLR2 expression in TAM from various tumor types. To this end, we initially searched for macrophage-specific gene expression in melanoma [57] and head and neck squamous cell carcinoma (HNSC) [58] (Figure 2A) and identified a set of 21 M-MØ-specific genes whose expression is also restricted to melanoma and HNSC TAM ( Figure 2B). Interestingly, FOLR2, CD163, LILRB5, CD209 and C1QC were identified as genes of the "anti-inflammatory gene set", whose expression is also seen in tissue-resident macrophages and tumor-associated macrophages ( Figure 2B). Analysis of breast cancer transcriptomes (METABRIC cohort) also evidenced a very good correlation between the expression of FOLR2 and those of genes of the "anti-inflammatory gene set", which reached statistical significance in most cases ( Figure 2C), and was highly significant for CD163 (Pearson: 0.72; p = 1.71 × 10 −308 ) and even CD68 (Pearson: 0.59; p = 1.25 × 10 −176 ), another widely used marker for macrophage identification ( Figure 2C). Further, a significant correlation was found between the expression of FOLR2 and CD163 using the TCGA cohorts for breast carcinoma, melanoma and HNSC ( Figure 2D and not shown), whereas no correlation was seen between FOLR2 and the epithelial-specific EPCAM gene in any of the analyzed tumors (data not shown). In fact, and at the protein level, FRβ expression was observed in CD163+ cells in melanoma ( Figure 2E) and in areas enriched in CD68+ cells in colon adenocarcinoma ( Figure 2F). Interestingly, and since FRβ+ macrophages are prominent in the tumor-invasive front of pancreatic cancer and associate with poor prognosis [72], it is worth noting that FRβ+ macrophages were mostly detected in the peritumoral area both in human melanoma and in a melanoma xenograft mouse model ( Figure 2E). Altogether, this set of results indicates that FOLR2 expression correlates with the expression of CD163 and other macrophage-specific genes in TAM, and that FRβ expression in TAM overlaps with the expression of the commonly used macrophage markers CD163 and CD68.  [57] or head and neck squamous carcinoma (HNSC) samples [58]. (B) Identification of M-MØ-specific genes specific within the lists of macrophage-specific genes in human melanoma [57] and head and neck squamous carcinoma samples [58].   [57] or head and neck squamous carcinoma (HNSC) samples [58]. (B) Identification of M-MØ-specific genes specific within the lists of macrophage-specific genes in human melanoma [57] and head and neck squamous carcinoma samples [58]. To identify potential regulators of FOLR2 gene expression, gene ontology analysis was done using Enrichr [64], and results revealed a positive enrichment of genes regulated by MAF, SPI1 (PU.1) and NR1H3 (LXRα) in the TAM-specific genes of the "anti-inflammatory gene set" ( Figure 2G). Indeed, FOLR2 expression was found to correlate with the expression of genes coding for transcription factors that determine macrophage differentiation and specification (SPI1 and MAF) in breast carcinoma (METABRIC cohort, Figure 2H), thus suggesting their involvement in expression of the FRβ-encoding FOLR2 gene. Further analysis of a large variety of tumor types using TIMER2.0 revealed that the positive correlation between FOLR2 and SPI1 expression was highest in colon adenocarcinoma, HNSC and sarcoma ( Figure 2I), and that the FOLR2-SPI1 correlation was more significant than the FOLR2-MAF correlation in almost every tumor type ( Figure 2J). Conversely, no significant correlation was found between FOLR2 or SPI1 expression and the epithelial-specific EPCAM gene expression (data not shown). Altogether, these results established a link between the FOLR2 gene and the expression of the PU.1-encoding SPI1 gene in tumor-associated macrophages.

FOLR2/FRβ Expression also Marks Human Synovial Macrophages
Next, we address the macrophage-restricted expression of FOLR2 in a pathology where macrophages preferentially exhibit a pro-inflammatory polarization, rheumatoid arthritis (RA). Initial assessment of FOLR2 expression in RA indicated an extremely close correlation with the expression of the "anti-inflammatory gene set" ( Figure 3A). In fact, 13 genes of the "anti-inflammatory gene set" were found within the 50 genes more closely correlating with FOLR2 expression in RA ( Figure 3A). Single-cell RNA sequencing (scRNA-seq) on samples from patients with rheumatoid arthritis (RA) or osteoarthritis (OA) has identified 18 unique cell populations in synovial tissue, including four transcriptionally different monocyte subsets [73]: IL1B+ pro-inflammatory monocytes, IFN-activated SPP1+ monocytes, NUPR1+ monocytes and C1QA+ monocytes ( Figure 3B), with the latter two subsets under-represented in RA and thought to exert homeostatic functions [73]. Analysis of the four subsets revealed the presence of genes of the "anti-inflammatory gene set" in IL1B+, NUPR1+ and C1QA+ monocytes, and that the expression of FOLR2 is a specific marker for the NUPR1+ monocyte subset ( Figure 3B) [73]. Indeed, although the expression of FOLR2 diminishes in macrophages from synovial membranes [74] ( Figure 3C) and from synovial fluid [75] ( Figure 3D) in RA, FRβ is still detectable in the lining layer of the synovial membrane of RA patients ( Figure 3E).

Expression of FRβ in Myeloid Cells is Dependent on the PU.1 Transcription Factor
To obtain support for the potential involvement of PU.1 in FOLR2 gene expression, we revised the available ChIP-Seq information on the genes of the "anti-inflammatory gene set", which had been found to be significantly expressed in tissue-resident macrophages ( Figure 1) and TAM ( Figure  2), and identified validated PU.1-binding sites immediately upstream of the first exon of most genes ( Figure 4A and not shown) [76], including FOLR2 as well as PU.1-binding sites within most genes of the "anti-inflammatory gene set" ( Figure 4B) [77]. The presence of 3-4 major peaks in ChIP-seq data for PU.1 binding within the FOLR2 gene [74,76,77], with one of them localized within exon 1, close to a potential FOS-binding site [78] (Figure 4C,D) and overlapping a sequence containing four evolutionary conserved potential Ets-binding sequences (5′-GGAAGGGAAGGAAGAGAGGAA-3′) [79,80] (Figure 4D,E), led us to address the control of FOLR2 expression by PU.1. To analyze the functional significance of this cluster of PU.1-binding elements, we initially evaluated its contribution to the transcriptional activity of the FOLR2 proximal promoter. In HeLa cells devoid of PU.1 [81,82], transfected with the pFOLR2-200Luc construct, which contains the fragment -214 to -34 and includes the PU.1-binding elements, overexpression of PU.1 resulted in 10-fold enhancement of the activity of the promoter ( Figure 4F). Further, mutation of the two distal Ets elements (pFOLR2-200PUmut2pXP2, Figure 4F) reduced PU.1-dependent transactivation to 50% (p = 0.008), while mutation of the four Ets-sequences (pFOLR2-200PUmut4pXP2, Figure 4F) reduced PU.1 transactivation by 86% (p = 0.009), thus implying that PU.1 exerts a positive regulatory action on the FOLR2 proximal regulatory region through interaction with a cluster of Ets-cognate sequences within exon 1 of the FOLR2 gene.

Expression of FRβ in Myeloid Cells is Dependent on the PU.1 Transcription Factor
To obtain support for the potential involvement of PU.1 in FOLR2 gene expression, we revised the available ChIP-Seq information on the genes of the "anti-inflammatory gene set", which had been found to be significantly expressed in tissue-resident macrophages ( Figure 1) and TAM (Figure 2), and identified validated PU.1-binding sites immediately upstream of the first exon of most genes ( Figure 4A and not shown) [76], including FOLR2 as well as PU.1-binding sites within most genes of the "anti-inflammatory gene set" ( Figure 4B) [77]. The presence of 3-4 major peaks in ChIP-seq data for PU.1 binding within the FOLR2 gene [74,76,77], with one of them localized within exon 1, close to a potential FOS-binding site [78] (Figure 4C,D) and overlapping a sequence containing four evolutionary conserved potential Ets-binding sequences (5 -GGAAGGGAAGGAAGAGAGGAA-3 ) [79,80] (Figure 4D,E), led us to address the control of FOLR2 expression by PU.1. To analyze the functional significance of this cluster of PU.1-binding elements, we initially evaluated its contribution to the transcriptional activity of the FOLR2 proximal promoter. In HeLa cells devoid of PU.1 [81,82], transfected with the pFOLR2-200Luc construct, which contains the fragment -214 to -34 and includes the PU.1-binding elements, overexpression of PU.1 resulted in 10-fold enhancement of the activity of the promoter ( Figure 4F). Further, mutation of the two distal Ets elements (pFOLR2-200PUmut2pXP2, Figure 4F) reduced PU.1-dependent transactivation to 50% (p = 0.008), while mutation of the four Ets-sequences (pFOLR2-200PUmut4pXP2, Figure 4F) reduced PU.1 transactivation by 86% (p = 0.009), thus implying that PU.1 exerts a positive regulatory action on the FOLR2 proximal regulatory region through interaction with a cluster of Ets-cognate sequences within exon 1 of the FOLR2 gene. Having demonstrated a direct effect of PU.1 on the FOLR2 proximal regulatory region, we then assayed the role of PU.1 on the activity of the FOLR2 promoter in the FRβ-expressing human THP-1 myeloid cell line, where the receptor is expressed in a functional state ( Figure 5A,B). As shown in Figure 5C, the pFOLR2-200PUmut4pXP2 construct exhibited significantly lower activity than the wild-type pFOLR2-200pXP2 construct (p = 0.008) in THP-1 cells, thus indicating that the activity of the FOLR2 gene regulatory region in myeloid cells is partly dependent on the integrity of the cluster of PU-1-binding elements located within exon 1. To definitively prove the direct involvement of PU.1 on FRβ expression, FOLR2 mRNA expression level was assessed after knocking down PU.1 expression in FRβ+ THP-1 cells. Nucleofection of a PU.1-specific siRNA in THP-1 cells reduced the expression of PU.1 by more than 50% ( Figure 5D). More importantly, siRNA-mediated knockdown of PU.1 led to a significant down-modulation of FOLR2 mRNA levels (p = 0.02 for experiment #1 and p = 0.009 for experiment #2) without affecting the expression of the functionally related PCFT gene ( Figure 5E). Therefore, PU.1 regulates FOLR2 gene expression in THP-1 cells, further confirming PU.1 dependence on the myeloid-restricted expression of the FOLR2 gene. Having demonstrated a direct effect of PU.1 on the FOLR2 proximal regulatory region, we then assayed the role of PU.1 on the activity of the FOLR2 promoter in the FRβ-expressing human THP-1 myeloid cell line, where the receptor is expressed in a functional state ( Figure 5A,B). As shown in Figure 5C, the pFOLR2-200PUmut4pXP2 construct exhibited significantly lower activity than the wild-type pFOLR2-200pXP2 construct (p = 0.008) in THP-1 cells, thus indicating that the activity of the FOLR2 gene regulatory region in myeloid cells is partly dependent on the integrity of the cluster of PU-1-binding elements located within exon 1. To definitively prove the direct involvement of PU.1 on FRβ expression, FOLR2 mRNA expression level was assessed after knocking down PU.1 expression in FRβ+ THP-1 cells. Nucleofection of a PU.1-specific siRNA in THP-1 cells reduced the expression of PU.1 by more than 50% ( Figure 5D). More importantly, siRNA-mediated knockdown of PU.1 led to a significant down-modulation of FOLR2 mRNA levels (p = 0.02 for experiment #1 and p = 0.009 for experiment #2) without affecting the expression of the functionally related PCFT gene ( Figure 5E). Therefore, PU.1 regulates FOLR2 gene expression in THP-1 cells, further confirming PU.1 dependence on the myeloid-restricted expression of the FOLR2 gene.

Discussion
In the present manuscript, we show that the FOLR2 gene is expressed by human CD163+ tissue-resident and tumor-associated macrophages (TAM) from various sources, and that its restricted cellular distribution is shared by a limited number of genes, including the commonly used macrophage-specific marker CD163. CD163 is a bona fide macrophage-specific marker [78] that, however, is expressed at higher levels in macrophages polarized toward the anti-inflammatory and reparative side [18,19], a property also shared by FRβ [45]. Indeed, FOLR2 expression parallels that of CD163 in tissue-resident macrophages, TAM from various tumor types and inflamed synovium.

Discussion
In the present manuscript, we show that the FOLR2 gene is expressed by human CD163+ tissue-resident and tumor-associated macrophages (TAM) from various sources, and that its restricted cellular distribution is shared by a limited number of genes, including the commonly used macrophage-specific marker CD163. CD163 is a bona fide macrophage-specific marker [78] that, however, is expressed at higher levels in macrophages polarized toward the anti-inflammatory and reparative side [18,19], a property also shared by FRβ [45]. Indeed, FOLR2 expression parallels that of CD163 in tissue-resident macrophages, TAM from various tumor types and inflamed synovium. Therefore, FRβ can be considered a macrophage-specific marker, in line with previous reports on its expression in distinct macrophage subsets in human and mouse tissues [44,45,48,72,83,84]. Further stressing its cell-restricted expression, the expression of CD163 and FOLR2 in TAM significantly correlates with the presence of the PU.1 transcription factor. Thus, we have found that the PU.1 transcription factor, which is preferentially expressed in myeloid cells, enhances the transcriptional activity of the proximal regulatory region of the FOLR2 gene and directly influences FOLR2 gene expression. The demonstration of the PU.1-dependent expression of FOLR2 is the first evidence of a transcription factor directly controlling FOLR2 expression and suggests that PU.1 contributes to the myeloid-specific expression of FRβ.
Macrophage reprogramming now appears a feasible therapeutic strategy for chronic inflammatory diseases [46]. Accordingly, the identification of macrophage subset-specific markers is a requisite for the development of macrophage-directed therapeutic interventions for human pathologies. The identification of FRβ as a macrophage-specific marker in homeostatic and pathological states has relevant translational implications, because FRβ has already been used as a target for imaging and delivery of therapeutic agents in inflammation-related diseases like rheumatoid arthritis [83][84][85][86]. Therefore, it is tempting to hypothesize that FRβ might also be a useful tool for delivery of agents with ability to shift the macrophage polarization state. Such an approach would benefit from the constitutive FRβ recycling ability [87][88][89] as well as by its huge capacity to transfer ligands towards the macrophage endocytic machinery [89,90]. This strategy would be particularly well suited in the case of tumors, as FRβ is highly expressed in TAM (this report and [45]). TAM promotes malignancy by stimulating angiogenesis, tumor-cell migration and invasion, and TAM accrual correlates with a worse prognosis in numerous tumors (85). Thus, FRβ constitutes an ideal target for delivery of macrophage-repolarizing agents into TAM, and, in line with the results here presented, the characterization of the factors that regulate FRβ expression constitutes relevant information for the development of FRβ-based macrophage targeting strategies.
Besides the involvement of PU.1 in the myeloid expression of FRβ, bioinformatics analysis also indicates that FOLR2 gene expression closely correlates with the expression of the MAF transcription factor in TAM from various sources. In fact, FOLR2 exhibits the highest level of correlation with MAF expression in breast carcinoma, a finding that agrees with the considerable decrease in FOLR2 expression that is seen upon MAF knockdown in human macrophages [74]. However, ChIP-Seq has not provided evidence for any interaction of MAF with the FOLR2 gene. Considering the ability of MAF to heterodimerize with members of the JUN/FOS family of transcription factors [91], and given the existence of FOS-binding sites in exon 1 [78] and additional AP-1-binding elements nearby [79], it is conceivable that MAF might indirectly affect FOLR2 expression by altering the levels of available JUN/FOS family proteins.
The comparison of tissue-resident macrophage-specific genes and TAM-specific genes has resulted in the identification of a group of six genes (MS4A6A, LILRB5, CD209, CD163, FOLR2, C1QC) which are also preferentially/exclusively expressed by macrophages with an anti-inflammatory/reparative polarization (included within the "anti-inflammatory" gene set). The proteins encoded by these six genes participate in either pathogen recognition (CD209, CD163, C1QC) or in modulation of inflammatory responses (MS4A6A, LILRB5). By contrast, and apart from its folate-binding ability, FRβ does not appear to fit within any of these two classes, although it modulates macrophage adhesion to collagen through association to the CD11b/CD18 integrin [92]. As a glycosyl phosphatidylinositol (GPI)-anchored protein, FRβ's potential to exert immunoregulatory actions would be indirect. By contrast, the cellular distribution, structure and recycling behavior of FRβ [85] somewhat resembles that of CD14, a crucial regulator of TLR4 ligand binding, endocytosis and TLR4-initiated signaling from endosomes [93,94]. We speculate that FRβ might exhibit a function similar to CD14, which acts both as a pattern-recognition receptor that binds directly to LPS and a co-receptor for several TLRs [95]. FRβ has a very high affinity for folic acid and folates (Kd~0.1-1nM), but mammals do not synthesize folate and are dependent on other sources. Diet or dietary supplements are not the only sources of folate, as several bacteria in the gastrointestinal tract can synthesize B vitamins, including folates (e.g., Lactococcus lactis, Bifidobacterium adolescentis) [96,97]. The macrophage-specific expression of FRβ described in this report, and the fact that folic acid is produced by numerous bacterial species [96,97], have led us to hypothesize that FRβ acts as a receptor or co-receptor for recognition of bacterial microbiota. If so, gut macrophages could detect high concentrations of folate through FRβ as a mechanism to control bacterial overgrowth through signaling or by phagocytosis, which would allow the microbiota homeostasis to be restored/maintained by a folate-dependent quorum sensing-like mechanism. Whether this mechanism contributes to the interplay between TAM and human microbiota in cancer [98] deserves further investigation. In any event, the hypothesis that FRβ is a sensor for adjusting macrophage effector functions to extracellular folic acid levels is fully compatible with the findings reported in the present manuscript, namely, that FRβ marks tissue-resident macrophages and macrophages within inflamed tissues, and that its expression correlates and is dependent on the expression of the PU.1 transcription factor.