Folate Receptor Beta for Macrophage Imaging in Rheumatoid Arthritis

Non-invasive imaging modalities constitute an increasingly important tool in diagnostic and therapy response monitoring of patients with autoimmune diseases, including rheumatoid arthritis (RA). In particular, macrophage imaging with positron emission tomography (PET) using novel radiotracers based on differential expression of plasma membrane proteins and functioning of cellular processes may be suited for this. Over the past decade, selective expression of folate receptor β (FRβ), a glycosylphosphatidylinositol-anchored plasma membrane protein, on myeloid cells has emerged as an attractive target for macrophage imaging by exploiting the high binding affinity of folate-based PET tracers. This work discusses molecular, biochemical and functional properties of FRβ, describes the preclinical development of a folate-PET tracer and the evaluation of this tracer in a translational model of arthritis for diagnostics and therapy-response monitoring, and finally the first clinical application of the folate-PET tracer in RA patients with active disease. Consequently, folate-based PET tracers hold great promise for macrophage imaging in a variety of (chronic) inflammatory (autoimmune) diseases beyond RA.


Synovial Macrophages as Biomarkers for RA Disease Activity Assessment
Rheumatoid arthritis (RA) is an autoimmune disease of the joints characterized by the infiltration of various immune cells in the synovium amongst which macrophages play an important role (1,2). Macrophages impact on other immune cells and inflammatory processes via the release of proinflammatory cytokines (e.g. TNFa) and chemokines, which may promote activation of T cells and other immune cells, trigger endothelial cell activation and (pathological) angiogenesis, and induce osteoclast activation (3,4). Synovial tissue analysis has pointed out that the (activated) synovial macrophage is a key biomarker for disease activity assessment from the early disease onwards and for monitoring of therapeutic efficacy at later stages of the disease (5,6). The basic synovial joint architecture of the healthy joint is comprised of a double layered structure (synovial lining) which holds tissue resident macrophages, and underneath a vascularized sub-lining layer of connective tissue (7). In the early stages of RA, infiltration of immune cells is observed in combination with activation of resident macrophages present in the synovial lining layer (2). Established RA is marked by progressive macrophage infiltration in the synovium (~10-20 layers) (8). In fact, macrophages represent one of the most prominent cell types present in the synovium during early stage and also established RA (2,9,10), being responsive to treatment (6), and thus underscoring their exploitation as a biomarker for the assessment of RA disease through positron emission tomography (PET) imaging.
The importance of macrophages as key player in the pathogenesis of RA has been explored in both preclinical and clinical studies. It has for example been shown in animal models of arthritis that depletion of macrophages significantly decreases the severity of the disease (11,12). Also, in RA patients, macrophage infiltration in the RA synovium has been found to significantly correlate with disease severity (e.g. with changes in disease activity score (DAS) and composite change index) (5,6,13). In addition, recent in depth cellular and molecular analyses of RA synovial tissues revealed that RA patients could be stratified in 3 pathology groups based on the presence of specific immune cell types (14)(15)(16). These 3 pathotypes were designated diffuse-myeloid (characterized by predominantly myeloid cell infiltration, notably macrophages), lymphomyeloid (predominantly B-cell infiltration), and pauci-immune (low immune cell infiltration) (14). Remarkably, a higher diffusemyeloid gene expression profile, thus characterized by macrophage infiltration, was associated with a higher DAS 28-ESR and a larger DAS 28-ESR reduction after treatment with disease modifying anti-rheumatic drugs (DMARDs) (14), again underscoring the importance of the synovial macrophage as biomarker for RA disease activity.
Although it is well recognized that macrophages play an important role in the pathology of RA, they can exert both pro-and anti-inflammatory roles associated with their polarization (17). The synovial cytokine milieu, in particular granulocyte-macrophage colony-stimulating factor (GM-CSF) and macrophage colony-stimulating-factor (M-CSF), constitutes the driving force in skewing macrophages to the M1-type pro-inflammatory phenotype and the M2-type antiinflammatory macrophage, respectively (9,(18)(19)(20). M1 and M2 represent the extremes of macrophage polarization and have been characterized based on differences in their transcriptome, secretome and proteome profiles (19)(20)(21)(22). Several (membrane) marker proteins are commonly used to classify M1 (e.g. CD80, TNFa, iNOS) and M2 (e.g. CD163, IL-10, Arginase) macrophage subpopulations, and have been associated to an inflammatory or remission state of RA. For example, macrophage subpopulations were found associated with RA disease remission such as synovial tissue macrophages that are MerTK positive (MerTKpos), lymphatic vessel endothelial hyaluronan receptor 1 positive (LYVE1pos) and have a high expression of Folate Receptor (FR) beta (FRb-high) (23). In the context of this review, FRb expression has long been recognized on macrophages that are triggered by inflammatory stimuli and on activated macrophages in inflamed joints of RA patients (24,25). In exvivo M-CSF skewed monocyte-derived macrophages, FRb is differentially expressed on M2-type macrophages (26)(27)(28). However, in inflamed RA synovium, these FRb-expressing M2macrophages can produce pro-inflammatory cytokines when exposed to either pro-inflammatory stimuli (i.e. lipopolysaccharide (LPS) + interferon-g (IFNg) (19) or an RA synovial microenvironment with anti-citrullinated protein antibodies or complex IgGs (29,30). Together, FRb is a bona fide marker on synovial macrophage subpopulations, even though its exact role in function in either pro-and antiinflammatory macrophages needs to be defined in greater detail. FRb expression on (activated) macrophages in RA has initiated research aimed at therapeutic targeting as well disease monitoring with imaging modalities (31,32), which will be discussed in the next sections.

Folate Receptors: Function, Structure and Targeting
Human FRs are high affinity binding proteins for folates (folic acid and reduced folate cofactors) which are essential vitamins necessary for single carbon transfer reactions in amino acid biosynthesis (e.g. conversion of homocysteine into methionine) and for de novo purine and thymidylate biosynthesis (33)(34)(35). There are four types of FRs: FR-alpha (FRa), FR-beta (FRb), FRgamma (FRg) and FR-delta (FRd) ( Figure 1A). Of those, FRa, FRb and FRd are glycosylphosphatidylinositol (GPI) membrane anchored, whereas FRg is a secreted form (mainly from hematopoietic cells) because of a lack of an efficient signal for GPI modification (36)(37)(38)(39). FRa is expressed on normal epithelial cells (e.g. kidney, spleen and lung tissue) (40) and tumor tissues like ovarian, breast (41), pancreatic and lung carcinomas (42,43). FRb is selectively expressed on cells of the myeloid lineage (44,45) and is upregulated on activated macrophages in active RA disease (24,25) wherein its expression is regulated by PU.1 transcription factor (46). The FRd gene was originally identified being highly homologous mouse folate binding protein 3 (Folbp3) (47), but does not harbor folate binding capacity (48). FRd (and splice variants thereof) is expressed on regulatory T cells and has a proposed function in immune regulation (49).
Macrophage FRb is a valid target for folate-based imaging ( Figure 1B), along with other FRb targeting approaches ( Figure 1C) by (drug-conjugated) monoclonal antibodies, folate-conjugated drugs, CAR T cells and folate antagonists (25,31,34,50,51). FRb contains a binding pocket for folic acid where binding of folic acid is facilitated by a conformational change in two regions of the receptor, in particular the region  connecting beta strand 1 and 2 and the region of alpha helix 1 (52). Like folic acid, folate antagonist therapeutic drugs such as methotrexate (MTX, the anchor drug in RA treatment) (53,54) and pemetrexed (PMX) (55) share a pterin moiety in their structure which can bind in the hydrophobic region of the binding pocket of FRb (52) ( Figure 1D). FRa and FRb share high affinity binding of folic acid with Kd's in the low nanomolar range (0.1-1 nM) (36). Structure activity testing disclosed antifolate structures with FRa and FRb affinities close to folic acid, 2-3 orders of magnitude lower affinities for MTX than folic acid and intermediate affinities for PMX (25,56). In order to elicit therapeutic activity against macrophages, a folate antagonist should compete with FRb binding of the primary circulating reduced folate in plasma, 5methyltetrahydrofolate (5-CH 3 -THF) (25). Also FRb functions as one of three transport proteins for folates and antifolates, the two others are the reduced folate carrier (RFC, SLC19A1) and proton-coupled folate transporter (PCFT, SLC46A1) (57) ( Figure 1E). The expression of these 3 folate transporters does vary between polarized macrophages; in ex vivo skewed monocytes to M1-type macrophages by GM-CSF and M2-type macrophage by M-CSF, RFC gene expression is differentially higher in M1-type macrophages, whereas FRb and PCFT expression are markedly higher in M2-type macrophages (27,28). Given that FRa/b also retain a high affinity binding profile for folate conjugates, this allowed the design of folate conjugates which could serve as folate based imaging agents like [ 18 F]fluoro-PEG-folate ( Figure 1F) (58-60).

PET Imaging of Macrophages
Over the past decade, non-invasive molecular imaging techniques that assess RA disease activity using macrophage PET imaging have been developed (31,58,(61)(62)(63). Disease activity assessment of RA is currently performed clinically through calculation of the DAS, which takes into account the number of swollen and tender joints, the erythrocyte sedimentation rate (ESR) and the visual analogue score (VAS) (1). However, the DAS score contains subjective elements and is limited to the sensitivity and specificity of clinical assessment of tenderness and swelling of the joints (64). To make further steps in improvement of diagnostics and monitoring of disease activity, objective tools such as macrophage (whole body) PET imaging may offer new opportunities for early diagnosis and early determination of the treatment outcome. Early diagnosis (even before clinical diagnosis) and effective, personalized treatment may ultimately result in prevention of (progression of) joint damage (1).
Initially, imaging studies were performed for FRa-expressing tumors with folate-SPECT tracer [ 99 Tc]EC20 (48). By serendipity, imaging of a cancer patient who had an arthritic comorbidity showed a positive scan of the inflamed knee joint due to infiltration of FRb-positive macrophages (65). This was the start of exploration of arthritis imaging by FRb targeting. The easy accessibility as GPI-linked plasma membrane protein and its myeloid cell specific expression constitutes FRb a suitable target for macrophage imaging. [ 99 Tc]EC20 proved its suitability in visualizing FRb-positive macrophages in inflamed joints of arthritic rats (66) and RA patients (67). These findings encouraged the development of folate-based PET tracer which would provide a higher sensitivity and spatial resolution compared to scintigraphy.
The original synthesis of folate-based PET tracers relied on chemistry linking folic acid with an spacer moiety (polyethylene glycol) as a precursor molecule to which the PET isotope is coupled (68). This approach was adopted in studies by Gent et al. (58) and Kularatne et al. (59) to synthesize [ 18 F]fluoro-PEGfolate ( Figure 1F). Assessment of FR binding affinity for fluoro-PEG-folate in a [ 3 H]folic acid competition assay showed that the unlabeled tracer had a 2-fold lower affinity than for folic acid, but a 2.5-fold higher affinity than for the circulating plasma folate 5CH 3 -THF (58). After fulfilling this important criterium of high affinity binding and outcompeting binding of circulating plasma 5-CH 3 -THF, next [ 18 F]fluoro-PEG-folate was examined further for in vivo PET-based monitoring of disease activity and therapy response in a preclinical model of RA (see section 2).

The Antigen-Induced Arthritis Model
In the preclinical assessment of FRb as a macrophage target for PET imaging of arthritis, Chandrupatla et al. established a preclinical rat model of arthritis with sustained macrophage infiltration in the joints (69). In this model arthritis is induced through immunization with methylated bovine serum albumin (mBSA) in complete Freund's adjuvant (CFA) and custom Bordetella pertussis antigen (CBP) following one or repeated intra-articular (i.a.) injections with mBSA in one knee of the rat (leaving the contralateral knee as control), after which synovial inflammation accompanied by an increase in knee thickness occurs over 3 days (Figures 2A, B) (69). This antigen-induced arthritis model was selected for PET imaging of molecular markers in arthritis and therapeutic evaluations due to several reasons; (i) the model is mono-articular, and hence the contralateral and other joints can serve as control, (ii) the model is relatively mild and is not accompanied by severe bone destruction or polyarticular involvement, (iii) the model is also reminiscent of human RA featuring macrophage infiltration (71, 72) ( Figure 2C) and moderate systemic inflammation manifested by modest macrophage infiltration in the liver and spleen (70), (iv) macrophage infiltration can be sustained by repeated mBSA injections, which is a good condition for therapeutic monitoring (69), and (v) the larger size of the rat allows injecting more radioactivity and the larger knee size of the rat (compared to a mouse) is advantageous given the spatial resolution of most PET scanners. This model therefore allows for studying if [ 18 F]fluoro-PEG-folate is a suitable tracer for objective macrophage imaging in arthritis affected joints and for longitudinally studying the effect of anti-rheumatic drugs on joint macrophage infiltration (58,62,70). knee, liver and spleen of non-treated and PMX-treated AIA rats. Data were corrected for blood %ID/g. Statistics for images E-F were performed in Graph-Pad Prism version 9, # p < 0.01, Mann-Whitney U test for non-parametric divided data, *p < 0.05, ***p < 0.001, unpaired T-TEST for parametric divided data, N = 4 rats/group. All results described in Figure 2 were derived (and reanalyzed were indicated) from own research (58,62,69,70 (58). Treatment of arthritic rats with different dosages of the folate antagonist MTX significantly reduced (~2-4 fold) tracer uptake in the arthritic knee compared to the arthritic knee of untreated rats (62, 70) ( Figure 2E). Consistently, histological analysis demonstrated that macrophage infiltration was also~2-4 fold reduced in the arthritic knee as well as in the liver and spleen following MTX therapy (70). Involvement of liver and spleen point to systemic inflammation in the arthritic rat model, which is suppressed by MTX treatment (70). This is in line with the systemic character of RA in patients (73). Similar results were  obtained after treatment of arthritic rats with the secondgeneration folate antagonist pemetrexed (PMX)/Alimta (74).
[ 18 F]fluoro-PEG-folate uptake in the arthritic knee, liver and spleen of PMX treated rats was reduced~2.2; 3.2 and 1.6-fold, respectively compared to untreated rats ( Figure 2F). PMX was originally developed to overcome MTX resistance in cancer chemotherapy (55) by harboring more efficient transport properties via RFC, PCFT and FR than MTX ( Figure 1C). PMX has shown anti-arthritic effects by suppressing cytokine production in activated T cells of RA patients (75), experimental arthritis models (74) and polarized macrophages in vitro (28)

First Clinical Study With [ 18 F]fluoro-PEG-Folate in RA Patients
Immuno-histochemical and immuno-fluorescent analysis of FRb expression in RA synovial biopsies showed marked expression of double CD68 and FRb-positive macrophages in both lining and sublining of RA synovial tissue (representative example shown in Figure 3A) (

Future Perspectives
Given the fact that macrophage PET imaging with FRb has shown clinical feasibility in RA patients, future challenges will be to use folate PET tracers for detection of disease activity in early stage RA and monitoring/prediction of the therapy response of targeted synthetic or biological DMARDs in joints and other sides affected by systemic inflammation. Furthermore, many other inflammatory diseases with macrophage involvement, e.g. idiopathic pulmonary fibrosis, systemic lupus erythematosus, scleroderma, psoriasis, ulcerative colitis, Crohn's disease (78,79), giant cell arteritis (80), cardiovascular diseases (81,82) and tumor associated macrophages in oncology (43) may benefit from folate PET imaging to detect and monitor disease activity.
Recently, also folate PET imaging of lung macrophages in COVID-19 was advocated to identify patients at risk of a severe or even lethal disease (83). Steps forward in the folate-linker chemistry also allow for more rapid synthesis of alternative folate tracers from precursor molecules, e.g.