Disruption of Nurse-like Cell Differentiation as a Therapeutic Strategy for Chronic Lymphocytic Leukemia Free

This article is featured in Top Reads, p.

Chronic lymphocytic leukemia (CLL) is a disease characterized by the accumulation of mature CD5⁺CD19⁺CD23⁺ cells in the bloodstream. It is the most common type of leukemia in adults in the United States, with an estimated 20,160 new cases and ∼4,410 deaths projected for 2022 (1). There are several therapies available for CLL, but resistance and relapse are still a major concern. In addition, the median patient age is 70 y, raising the likelihood that aggressive therapies will be detrimental (2). Hence, the need for additional novel therapies remains.

CLL cells themselves have been shown to have a short life in culture after isolation (3). Further research found that in proliferation centers, CLL cells rely on different cytokine as well as cell-to-cell contacts to survive and proliferate (4). Among cells that provide survival signals to CLL cells are the nurse-like cells (NLCs). NLCs originate from CD14⁺ monocytes and can be derived in vitro by coculturing monocytes with CLL cells for 14 d (3). Characterized by the expression of CD68, CD163, and CXCL12 (stromal cell–derived factor 1, SDF-1), among other markers, NLCs have been related to tumor-associated macrophages (TAMs) and have been seen to be in close contact with CLL cells in lymphoid tissues of patients (3, 5, 6). NLCs provide different survival signals to CLL cells to promote their survival, such as the secretion of CXCL12 (SDF-1), cell-to-cell contacts through CD38/CD31, BAFF, APRIL (a proliferation-inducing ligand), and LFA-3/CD2 (3, 7–9).

Because of their importance in CLL, NLCs are a potential target for therapeutic approaches. Previously, it has been shown that elimination of macrophages resulted in reduced CD19⁺CD5⁺ burden in tumor models and increased mouse survival (10, 11). Nevertheless, effective NLC elimination in patients with CLL with minimal effect on other endogenous myeloid-lineage cells will require a detailed understanding of how they develop. NLC development seems to be dependent on signaling through the CSF-1 receptor (CSF-1R) or by signaling through the receptor for advanced glycation end products (12, 13). These are not unique to NLC, however, so a search for NLC-specific drivers of differentiation remains. Here, we sought to find the signaling pathways that specifically orchestrate NLC development.

Our studies showed that monocyte-to-NLC differentiation was characterized by loss of methylation in specific transcription factor binding sites, especially AP-1 and musculoaponeurotic fibrosarcoma (MAF). MEK signaling is required for expression of AP-1 transcription factors, and we found that MEK inhibition with PD0325901 or trametinib almost completely abrogated the development of NLC in vitro. In addition, we showed that ERK was constitutively phosphorylated in NLCs and that trametinib treatment did not directly cause CLL cell death. Using the Eμ-TCL1 adoptive transfer mouse model, we extended these findings in vivo, where the MEK inhibitor trametinib decreased WBC counts and extended survival time. Trametinib also reduced the total amount of M2-like MHC class II–positive (MHC-II⁺) and F4/80⁺ cells in the spleen. Finally, ERK phosphorylation was significantly diminished in the spleens of treated mice. These results shed light on the signaling pathways involved with NLC development and suggest that trametinib may be a possible therapeutic agent for CLL.

Blood samples from naive, low-count (<60,000 cells/µl) patients with CLL were obtained according to the Declaration of Helsinki under protocols approved by the institutional review board at The Ohio State University. PBMCs were collected by centrifugation with lymphocyte separation medium (Corning, Corning, NY), washed with incomplete RPMI media, and counted. For NLC development, PBMCs were plated in plastic or collagen I–coated (Corning) dishes at 10 × 10⁶ cells/ml in RPMI 1640 medium (Life Technologies/Thermo Fisher Scientific, Waltham, MA) complemented with 10% FBS (VWR, Radnor, PA), 2 mM l-glutamine (Corning), 56 U/ml penicillin, and 56 µg/ml streptomycin (Invitrogen/Thermo Fisher Scientific, Waltham, MA). Fresh media were added as needed during cell culture. According to previous reports, fully differentiated NLCs are obtained after 2 wk of cell culture (3).

For generating NLCs from healthy donor monocytes, CD14⁺ monocytes were isolated from source leukocytes (American Red Cross, Columbus, OH; Versiti Blood Center of Ohio, Columbus, OH) using positive magnetic isolation as previously described (14). Briefly, PBMCs were resuspended in MACS buffer (0.5% BSA, 2 mM EDTA in PBS) and incubated with CD14⁺ magnetic beads (Miltenyi Biotec, Auburn, CA) for 15 min in ice. Then, cells were washed with MACS buffer and passed through an LS column (Miltenyi Biotec) coupled to a magnet. CD14⁺ cells were recovered from the column by pressure, resuspended in complete RPMI media, and counted. Then, CD14⁺ cells were cocultured with CLL patient B cells, which had been isolated using the Rosette-Sep B cell kit (STEMCELL Technologies, Vancouver, BC, Canada) by the Experimental Hematology Laboratory (The Ohio State University). For healthy donor–derived NLCs, cocultures were incubated for 21 d because their development appeared to have lagged behind that of patient-derived NLCs.

For CLL patient monocytes, cells were isolated from whole blood from patients with CLL, similar to the method described above. For viability assays, monocytes or PBMCs were plated at 3 × 10⁶ cells/ml or 10 × 10⁶ cells/ml, respectively, in preprepared collagen-coated plates (50 μg/ml of rat collagen type I; Life Technologies/Thermo Fisher).

For inhibition assays, the following inhibitors and final concentrations were used: SP600125 1 µM, SB202190 1 µM, PD0325901 0.5 µM, and trametinib 1–10 µM (Selleck Chemicals, Houston, TX). Inhibitors were added at the stated concentrations to CLL-PBMC culture every 96 h.

After MEK or MAPK inhibition, cells were detached from plastic dishes using EDTA-trypsin 0.25% (Life Technologies) at 37°C for 10 min. The trypsin reaction was stopped using complete RPMI media. Cells were then gently scraped using rubber scrapers (Sarstedt, Newton, NC) to promote detachment, then centrifuged and resuspended in fresh media or staining buffer (5% BSA in PBS). For cultures that were done in collagen-coated plates, cells were detached using collagenase type IV (Worthington Biochemical, Lakewood, NJ) at 0.1% in HBSS (Life Technologies) with 10 µg/ml of Polymyxin B (Life Technologies) for up to 1 h until cells showed detachment. Cells were then washed and resuspended in media or staining buffer (0.5% BSA in PBS).

For cell sorting, PBMCs from culture were washed twice with PBS and blocked with whole human IgG at 10 µg/ml in PBS for 15 min at 4°C. Then, cells were stained with anti-human CD14 FITC (clone M5E2; BD Biosciences, Franklin Lakes, NJ) and anti-human CD19 PE-cyanine 7 (Cy7) (clone HIB19; eBioscience/Thermo Fisher Scientific, Waltham, MA) Abs for 30 min at 4°C in the dark; the appropriate isotype controls were used. After incubation, cells were washed, resuspended in staining buffer, and sorted for CD19⁻CD14⁺ cells using a BD FACSAria or FACSAria III cytometer (BD Biosciences, San Jose, CA) at The Ohio State University Analytical Cytometry shared resource.

For methylation analysis, paired CD14 cells from time 0 and time 14 d (0 versus 21 d for healthy donor–derived NLCs) were used. CD14⁺ cells at time 0 were isolated by positive selection using MACS (Miltenyi Biotec, Cambridge, MA) of PBMCs, whereas cells at day 14 or 21 were obtained by cytometry-based cell sorting as described above. Cells were lysed and DNA was isolated using the Gentra Puregene cell kit (Qiagen, Germantown, MD) according to the manufacturer’s protocol. Analysis of DNA methylation was done at the German Cancer Research Center (Heidelberg, Germany) using Illumina EPIC/850k methylation arrays (Illumina, San Diego, CA). The Infinium methylation assay was performed as described previously (15). Raw 850K data were normalized by a β-mixture quantile normalization method (16) using the RnBeads version 2.0 software (17). Probes present on sex chromosomes, non-CG, and those affected by proximity to a known single-nucleotide polymorphism were removed to generate a final probe set of 804,572 probes. Enrichment of transcription factor (TF) motifs was determined using HOMER (18). For motif identification in hypomethylated regions in Illumina 850K data, appropriate background probe sets were generated to have equal guanine-cytosine and CpG content as well as a similar distribution of methylation levels compared with foreground. A known motif search was performed using ±100 bp from the CpG of interest. TF binding site (TFBS) enrichment was calculated using the Epiannotator R package with a minimum overlap of 1 bp (19). Association of hypomethylated regions with TFBS binding was performed by comparing with TF chromatin immunoprecipitation–sequencing peak intervals available from ENCODE (20) for K562 cells, and adjusted p values were calculated using Fisher’s exact test. For data visualization and other analysis, the Qlucore Omics Explorer software (Qlucore, Lund, Sweden) was used. Genes were identified from the top 5000 differentially hypomethylated CpGs by selecting those with hypomethylation in the promoter regions (5′ untranslated region [UTR], transcriptional start sites, and first exons). The data discussed in this publication have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (21) and are accessible through accession number GSE206842 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE206842). Data were then subjected to further enrichment analysis using ShinyGO (http://bioinformatics.sdstate.edu/go/) (22).

Staining for NLC phenotype and for other leukocytes was done as follows. NLC cultures were obtained as above, then cells were washed and blocked with whole human IgG for 15 min at 4°C. For extracellular staining, cells were incubated with anti-CD14 FITC, anti-CD19 PE (clone HIB19, BD Biosciences), anti-CD163 PE-Cy7 (clone GHI/61; BioLegend, San Diego, CA), anti-CD3 Pacific Blue (clone OKT3, BioLegend), anti-CD56 Super Bright 600 (clone TULY56; eBioscience/Thermo Fisher Scientific), and LIVE/DEAD Blue (Thermo Fisher Scientific). Incubation was done for 30 min on ice, followed by two washes with staining buffer. Then cells were fixed and permeabilized using the Cytofix/Cytoperm Fixation/Permeabilization kit (BD Biosciences) according to the manufacturer’s instructions. Cells were stained intracellularly with anti-CD68 Brilliant Violet 785 (clone Y1/82A, BioLegend) and anti-CXCL12/SDF-1 allophycocyanin (clone 79018; R&D Systems/Biotechne, Minneapolis, MN). After intracellular staining, cells were washed and acquired using the LSRFortessa cytometer (BD Biosciences) at The Ohio State University Analytical Cytometry shared resource. To calculate absolute numbers, precision counting beads (BioLegend) were added. Analysis was done using FlowJo software (BD Biosciences). Isotype controls for all Abs were used for gating.

CLL patient monocytes and CLL patient PBMCs were plated as described above and treated with DMSO or trametinib at 1 μM at the beginning of the culture. For CLL-derived PBMCs, treatments were administered again at day 4. Cells were recovered from the plates, washed and stained for flow cytometry with anti-CD14 Alexa Fluor 647 (clone HCD14, BioLegend) and anti-CD163 PE-Cy7 and for PBMCs with anti-CD19 Brilliant Violent 421 (for PBMCs only). After incubations, cells were washed and stained for annexin V FITC and propidium iodide (PI) according to the manufacturer’s instructions (BD Biosciences). Samples were acquired using the LSRFortessa cytometer. For analysis, isotype controls and precision counting beads were used.

To detect phosphorylation in NLCs during the 14-d culture period, PBMCs from patients with CLL were cultured in plastic dishes containing 12-mm poly-d-lysine–coated coverslips (Neuvitro, Vancouver, WA). Cells were treated with DMSO or trametinib every 96 h for 14 d. After 14 d, nonadherent cells were washed with PBS three times. Then, complete RPMI media were added to the culture, and cells were fixed with paraformaldehyde to the cultures at a final concentration of 2% and incubated for 30 min to allow cell fixation. Cells were washed with PBS and stored at 4°C until staining. For staining, coverslips were blocked with 10% goat serum (Abcam, Cambridge, UK) for 1 h at room temperature. Cells were stained with mouse anti-ERK1 (pT202/pY204) with ERK2 (pT185/pY187) Ab (clone MPK-YT; Abcam) or anti-ERK (clone 7D8; Abcam) and anti-human allophycocyanin CD68 (cloneY1/82A, BioLegend) overnight. Then, samples were washed and incubated with goat anti-mouse Alexa Fluor 488 (Abcam). Orange fluorescent tetramethylrhodamine wheat germ agglutinin (Thermo Fisher Scientific), NucBlue Live Rady Probes Reagent, and Hoechst 33342 (Thermo Fisher Scientific) were used as contrast for membrane and DNA staining. Coverslips were mounted with Fluoroshield solution (Sigma-Aldrich/MilliporeSigma, Burlington, MA). Images were acquired using a Zeiss LSM800 confocal microscope (Zeiss, Pleasanton, CA) and then analyzed using ImageJ (National Institutes of Health, Bethesda, MD). For brightfield images shown in Fig. 3C, acquisition was done using the EVOS XL Core imaging system (Thermo Fisher Scientific) with a 20× objective; scale bars were added by calibration with a Neubauer improved chamber according to a previous publication (23).

For fluorescence immunohistochemistry of mouse spleen samples at 9 wk after transfer, paraffin-embedded tissue slides were treated for Ag retrieval with heat in citrate buffer. Tissue sections were then blocked and incubated with anti-ERK (clone 7D8) or anti-ERK1/2 (pT202/pY204) Ab (clone MILAN8R, eBioscience), anti-mouse F4/80 (clone CI:A3-1, Bio-Rad Laboratories), and Hoechst 33342. After overnight incubation, slides were incubated with goat anti-mouse Alexa Fluor 488 or goat anti-rat Alexa Fluor 594 (Abcam). Slides were rinsed and then mounted with Fluoroshield solution (Sigma-Aldrich) before image acquisition with an Apotome microscope (Zeiss). Images were analyzed using ImageJ software. Mouse spleen tissue section staining for B220 (CD45R, clone RA3-6B2; BD Biosciences, San Jose, CA), CD4 (clone 4SM95, eBioscience/Thermo Fisher Scientific), CD8a (clone 4SM15, eBioscience/Thermo Fisher Scientific), and H&E was performed by the Comparative Pathology and Digital Imaging Shared Resource, Department of Veterinary Biosciences, The Ohio State University. Image acquisition was done with the Olympus BX41 microscope (Olympus, Tokyo, Japan) using the Olympus DP71 camera and cellSens Entry software (Olympus).

Mice (8 wk old) were housed at five per cage in a vivarium at The Ohio State University. All institutional animal care and use committee guidelines were followed, and all experimental procedures were approved in an institutionally reviewed animal protocol. The health of each mouse was monitored by laboratory and vivarium personnel. For the CLL mouse model, we used the Eµ-TCL1 adoptive transfer method, as previously reported (11). Briefly, 10 × 10⁶ splenocytes were transplanted into syngeneic C57BL/6 mice (The Jackson Laboratory, Bar Harbor, ME) intravenously. After 2 wk of the isogeneic transplant, mice were allocated to groups, with comparable WBC counts between groups, and treated i.p. with vehicle or with trametinib at 12.5 µg/mouse in 4% DMSO/corn oil three times per week. For survival, mice were treated for a total of 8 wk (10 wk after transfer); total WBC counts were obtained.

For macrophage characterization and assessment of CLL load, mice were sacrificed at weeks 7 and 9 after transfer. Spleens were collected, a part of the tissue was saved for paraffin sections, and the rest was disaggregated mechanically using a 40-µm cell strainer. Peritoneal cells were collected by flushing the peritoneal cavity with PBS. Both peritoneal cavity and spleen samples were treated with RBC lysis buffer (1.5 M NH₄Cl, 100 mM NaHCO₃, and 11 mM EDTA in H₂O) for 5 min and washed. CLL load for both peritoneal cavity and spleen, as well as M1 and M2 phenotype staining for spleen, were assessed as described above for extracellular staining.

For intracellular staining, the eBioscience Foxp3 intracellular staining kit (Thermo Fisher Scientific) was used according to the manufacturer’s instructions. Briefly, after extracellular staining, cells were washed twice in staining buffer and resuspended in fixation/permeabilization buffer and incubated for 1 h at room temperature. Then, cells were washed and resuspended in permeabilization buffer; samples were incubated with the intracellular marker Abs for 1 h at room temperature. Finally, samples were washed and resuspended in PBS for acquisition. Samples were analyzed using an LSRFortessa equipment (BD Biosciences). For quantification of total cells per sample for the Eµ-TCL1 adoptive transfer model, Precision Count Beads (BioLegend) were used according to the manufacturer’s recommendations. Populations were characterized using the markers and Abs listed in Table I.

Monocytes isolated from healthy donor blood samples (purchased from American Red Cross or Versiti) were plated in collagen plates and left untreated (M0), treated with GM-CSF at 20 ng/ml (M1) (R&D Systems), or treated with M-CSF at 20 ng/ml (M2) (R&D Systems) at days 0, 3, and 6. On day 6, the GM-CSF–treated cells were also treated with IFN-γ at 10 ng/ml (R&D Systems), and M-CSF–treated cells were also treated with IL-4 at 10 ng/ml (R&D Systems). On day 7, cells were treated with DMSO or trametinib at 1 µM and incubated for 72 h before collection. Cells were detached using collagenase type IV supplemented with Polymyxin B as above; cells were then stained for CD14 Alexa Fluor 647 and viability using the LIVE/DEAD Blue dye for flow cytometry. Cell numbers were obtained using Precision Count Beads. To ensure macrophage polarization, cleared supernatants were analyzed for TNF-α and IL-10 using ELISA Duo Set kits (R&D Systems) according to the manufacturer’s instructions.

Methylation statistical analysis was performed using Qlucore software. Mouse survival was analyzed by the log-rank test. Analysis of the following experiments was done with SAS software (SAS Institute, Cary, NC); for Fig. 3E and Fig. 4, data were analyzed by the mixed effects model; for Fig. 6E and 6F, analysis was performed by ANOVA. For Fig. 3D and Fig. 8, analysis was performed by two-way ANOVA with the Bonferroni multiple comparison using Prism software (GraphPad Software, La Jolla, CA). All other analyses were done using t tests.

Although NLCs have been studied extensively, the mechanisms of their development remain poorly understood. To gain a greater understanding of this, we first investigated their DNA methylation patterns. CD14⁺ cells were isolated from freshly collected blood (from both patients with CLL and healthy donors) using magnetic bead separation, then they were processed for methylation analysis. NLCs were obtained after culturing PBMCs from CLL patient blood for 14 d. CLL cells bind tightly to NLCs, so, to avoid any CLL cell contamination of NLC samples for analysis, NLCs were sorted using CD14⁺CD19⁻ gating. Similarly, NLCs were obtained from healthy donor monocytes by culturing CD14⁺ cells for 21 d with CLL cells isolated from patients. Differentiation of NLC-like cells using healthy donor monocytes has been described previously (24).

DNA methylation was measured using Illumina EPIC/850K methylation arrays (Fig. 1). We observed a substantial loss of DNA methylation after NLC differentiation, with only a small proportion of regions gaining methylation (Fig. 1A). In addition, there appeared to be minor yet noticeable differences between patient-derived and healthy donor–derived NLCs. This is reminiscent of the work of Bhattacharya et al. (25, 26), who found strong similarities regarding marker expression but slight differences in healthy donor–derived versus patient-derived NLC transcriptomes. In a short comparison, we found that patient-derived NLCs expressed CXCL12 (SDF-1) more consistently than did healthy donor–derived NLCs (data not shown). We chose the patient-derived NLCs for all subsequent experiments to remain as relevant as possible with regard to patients with CLL.

Further analysis of genomic regions exhibiting methylation loss in the patient-derived NLCs revealed significant enrichment of TF binding motifs (Fig. 1B). The most prominent motifs belonged to the basic leucine zipper superfamily and included motifs from the AP-1 and MAF TF families. Regions hypomethylated during NLC differentiation were also enriched for early growth response (EGR) and microphthalmia-associated TF motifs. Hypomethylated regions were compared with known TFBSs in the myeloid cell line K-562 and showed strong enrichment for AP-1 proteins (JUN/FOS) along with MAF and EGR families (Fig. 1C). These results indicate that these TFs play a role in programming DNA methylation patterns during NLC differentiation, which significantly differs from their monocyte counterparts.

We next created a list of genes that were hypomethylated in the 5′ UTR, in the transcriptional start sites or in the first exon, at day 14. This contained 761 entries (Supplemental Table I), which we then submitted to the ShinyGO gene enrichment analysis program (22). We chose the default p value cutoff of 0.05 and examined the top 30 Biological Process entries. Consistent with the TF families found above being stimulated by signaling pathway activation, we found enrichment in processes related to immune response, activation, motility, metabolism, and signaling (Fig. 2A).

We next visualized these with a correlation tree. As shown in (Fig. 2B, the strongest correlations were expectedly within the immune-response processes, depicted by the blue dots. Of note, however, we observed a large cluster of signaling-related entries. These comprised approximately one-third of total entries (9 out of 30). Within the list itself, we saw SRC and GRB2, with six and two hypomethylated regions within their 5′ UTRs, respectively. SRC drives multiple signaling pathways, including RAS, and is linked to multiple cancers (27). GRB2 facilitates RAS activation and is also associated with many types of cancer (28). These results may suggest an involvement of MAPK signaling during NLC development, and in particular the RAS/ERK pathway.

It has been shown that MAPK signaling can drive the expression of AP-1 TFs, along with other identified TFs, such as MAF, EGR, and microphthalmia-associated TF. This is seen especially during inflammation and in response to different stimuli (29). Thus, we hypothesized that inhibition of specific arms of MAPK signaling pathways might lead to disrupted NLC differentiation in vitro. To test this, PBMCs from CLL patient samples were cultured in plastic/collagen dishes for 14 d. We targeted the three major MAPK signaling pathways using four different inhibitors: MEK (PD0325901 or trametinib), p38/MAPK (SB202190), and JNK (SP600125) (Fig. 3).

The blockade of JNK did not affect the number of derived NLCs versus control. However, p38/MAPK inhibition resulted in a higher number of NLCs produced than the DMSO control. Notably, we found that inhibition of MEK significantly reduced the number of NLCs. This effect was seen with two different MEK inhibitors, suggesting that MEK signaling plays a primary role in NLC differentiation (Fig. 3A, 3B). For further experiments, we focused on the effects of the MEK inhibitor trametinib, because it has been shown to have an antitumor effect against melanoma and is approved by the U.S. Food and Drug Administration (30). A representation of cells in culture treated with DMSO or trametinib is shown in (Fig. 3C. To further characterize NLCs, we marked cells with CD14 and CD68 and then measured expression of CD163 and CXCL12 (SDF-1), known markers for NLCs (Fig. 3D). In the DMSO-treated group, a large population was seen in the side versus forward scatterplots, whereas this population seems to be lost in the trametinib-treated culture. In addition, in the DMSO-treated culture, there was a clear population positive for both CD14 and CD68, and these cells also expressed CD163 and CXCL12 (SDF-1) (Fig. 2D, top panel), in accordance with previous reports (3, 31). However, this population was largely lost after MEK inhibition. Instead, there was a population with dimmer expression of CD68 and low to no expression of CD14. This population appeared to be negative for CD163 and showed lower expression of CXCL12 (SDF-1) than NLCs (Fig. 2D, lower panel and left graphs). Taken together, these data suggest that trametinib affects NLC development and phenotype.

It has previously been reported that MEK signaling is important for CLL cell survival (32), and it has also been suggested that MEK inhibition may affect CLL survival in vitro even in the presence of specific survival signals (33, 34). Thus, we tested if trametinib had a direct effect on CLL samples. When CLL cells were cultured alone, trametinib treatment at 1 µM or 10 µM did not cause a significant decrease in the number of viable cells compared with DMSO, with the exception of 10 μM at 7 d (Fig. 3E). These results suggest that MEK inhibition directly affects NLC differentiation but likely not CLL cell survival, at least within our experimental conditions.

To determine whether trametinib reduced NLC numbers by promoting cell death versus blocking differentiation, we first plated CLL patient monocytes onto collagen-coated plates and treated them with DMSO or trametinib (1 μM). Cells were collected at 24, 48, or 72 h and stained for viability (annexin V/PI), along with CD14 and CD163 (differentiation markers). Absolute counts were obtained using counting beads. The results showed that trametinib led to a small reduction in the number of viable monocytes at 72 h, which did not reach significance (Fig. 4A). However, CD163 expression was significantly reduced by trametinib at 48 and 72 h (Fig. 4B), suggesting that although slightly affecting cell viability, trametinib primarily blocked differentiation. In contrast, when we treated CLL patient PBMCs, we found that trametinib significantly reduced monocyte viability at days 3 and 4 (Fig. 4C), suggesting that intercellular signaling may have been affected. In addition, as seen with monocytes cultured alone, we saw a significant reduction of CD163 expression in monocytes within the PBMC cultures (Fig. 4D).

We also determined whether trametinib affected other cell types during the 14-d PBMC culturing period. We plated CLL patient PBMCs and treated them with DMSO or trametinib as above. After this, we measured numbers of B, NK, and T cells, along with NLCs. The results showed that, as expected, NLC numbers were substantially reduced with trametinib (Fig. 4E). B and NK cells were reduced from day 0 to day 14, but there were no significant differences between trametinib and control at day 14 (Fig. 4F, 4G). However, T cells were significantly reduced by trametinib (Fig. 4H). These results suggest that within the CLL patient PBMC cultures, trametinib preferentially affects NLCs, with T cells also reduced to a significant extent.

Next, we tested whether there was any detectable ERK phosphorylation, and thus MEK signaling pathway activation, in NLCs or CLL cells in culture. For this, we treated cells with DMSO or 1 μM trametinib for 14 d. Samples were stained for CD68, phospho-ERK, and total ERK, then they were analyzed by confocal microscopy. As expected, NLCs showed basal phosphorylation of ERK in addition to expression of the CD68 marker (Fig. 5A, upper panel, and Supplemental Fig. 1A). In contrast, trametinib-treated samples did not contain large CD68⁺ cells, and staining for ERK phosphorylation was not detectable (Fig. 5A, lower panel, and Supplemental Fig. 1A, lower panel). Indeed, trametinib-treated samples appeared to contain only the smaller CLL cells (Fig. 5A, lower panel, and Supplemental Fig. 1A, lower panel). Total ERK was observed in DMSO-treated-samples, but only speckles of total ERK were found in trametinibtreated samples (Fig. 5B, upper and lower panels, respectively, and Supplemental Fig. 1B). This further suggests a loss of the ERK-expressing CD68⁺ population, leaving only the smaller CLL cells. We next compared ERK phosphorylation between CLL patient monocytes (day 0) and their autologous NLC counterparts (day 14). The results showed strong basal phosphorylation within the NLCs (Supplemental Fig. 1C). These results suggest that MEK/ERK activity is basally high in NLCs and that MEK inhibition blocks the development of NLCs.

Next, we tested whether trametinib had any effect on CLL development in vivo (Fig. 6A, 6B). For this, we used the Eµ-TCL1 adoptive transfer model as previously described (10). Mice were treated with vehicle or trametinib three times per week for up to 10 wk, starting 2 wk after adoptive cell transfer. WBC counts were taken, and the results showed a significant reduction in the trametinib-treated group compared with vehicle control (Fig. 6A). There was also a significant increase in survival time in the trametinib group, with a median survival of 164.5 d compared with 89 d in the control group (Fig. 6B).

To further explore the effect of trametinib on CLL disease progression in vivo, we measured the CD45⁺CD5⁺CD19⁺ CLL burden in the spleen and peritoneal cavity (Fig. 6C–6F). Here, engrafted mice were treated with vehicle or trametinib starting at 2 wk after engraftment. Following this, mice were sacrificed at 7 or 9 wk, and the spleen and peritoneal cavity were analyzed. As shown in (Fig. 6C, there was decreased splenomegaly in the trametinib versus vehicle groups after 9 wk, which suggests a difference in disease burden. At 7 wk, flow cytometry of splenic cells stained for CD45, CD19, and CD5 showed a reduction in the number of CD19⁺/CD5⁺ CLL cells (Fig. 6D, 6E). After 9 wk, however, this difference was lessened and only trended toward significance (Fig. 6E). In contrast, there was not a clear trend in the peritoneal cavity at week 7 or 9 (Fig. 6F), so we focused our subsequent attention on the spleen.

Because MEK inhibition strongly reduced NLC development, we sought to define the effects of trametinib on monocyte/macrophage polarization, using inducible NO synthase (iNOS) and EGR2 to mark M1- versus M2-like cells (35). We used the Eµ-TCL1 engraftment model as above, treating with vehicle or trametinib for 5 or 7 wk (7 and 9 wk after engraftment). Spleens were collected, and cells were dissociated and then split into groups to stain for M1- or M2-related markers in monocytes and macrophages using the markers described in Table I (Methods). Expression of iNOS (M1) and EGR2 (M2) was analyzed in CD45⁺CD11b⁺Ly6G⁻ cells expressing either MHC-II or F4/80 or both.

The results showed that there was a significant reduction in the number of EGR2⁺ cells in MHC-II⁺/F4/80⁺ cells from trametinib-treated mice at both weeks 7 and 9 (Fig. 7B, 7F, respectively). Reductions in total MHC-II⁺/F4/80⁺ mirrored the decreases seen with EGR2⁺ cells (Fig. 7B, 7F, respectively), although there was also a significant reduction in MHC-II⁺ cells at week 7 (Fig. 7A). Hence, trametinib reduces the numbers of both total and M2-related monocytes/macrophages.

We also measured changes in iNOS⁺ cells in the three populations (MHC-II⁺, MHC-II⁺F4/80⁺, and F4/80⁺). Total cells were reduced in MHC-II⁺ cells at week 9 (Supplemental Fig. 2D). In addition, iNOS⁺ cell numbers were significantly lower in MHC-II⁺ and MHC-II⁺F4/80⁺ cells, but only at week 9 (Supplemental Fig. 2D, 2E, respectively). This suggests that trametinib reduces M2- as well as M1-related cells, although the effect is earlier and stronger against M2 cells.

To investigate the effect of trametinib on M1-like cells in greater detail, we isolated healthy donor monocytes and treated them with GM-CSF/IFN-γ (M1), M-CSF/IL-4 (M2), or no treatment (M0), analogous to previous reports (36–38). After 7 d, we added DMSO or trametinib (1 μM) and incubated cells for another 72 h. Numbers of macrophages were then compared (Fig. 8). M0 numbers were extremely low by default, because they were primary monocytes in culture for 10 d with no additional growth support. However, trametinib did not substantially affect these M0-like macrophages (Fig. 8A). Similarly, trametinib did not significantly affect the M1-like macrophages (Fig. 8A). In contrast, after trametinib treatment, the M2-like macrophages were significantly decreased (Fig. 8A). To verify phenotype differences, we measured the release of TNF-α and IL-10 by vehicle-treated macrophages (Fig. 8B, 8C).

It has been reported previously that CLL cells from the Eµ-TCL1 model can have a higher infiltration in the spleen, with the tissue losing its normal architecture (39). In addition, CLL cells express lower levels of B220 (CD45R) than normal B cells (40). To test whether trametinib had an effect on B and/or CLL cells, we stained spleen sections from mice 9 wk after engraftment for B220. As shown in (Fig. 9A, two of the four mice treated with trametinib retained considerable levels of B220^high cells; the third showed some B220^high, and the fourth showed widespread B220^dim CLL cell staining throughout the spleen. These first two trametinib-treated mice also showed a morphology similar to our normal, non–tumor-engrafted control. Along with strong B-cell staining in contrast to diffuse CLL cell staining, the red and white pulp was also less diffuse (Fig. 9B). Because we found reductions in T cells when examining CLL patient PBMCs, we also stained for CD4 and CD8 T cells. The results showed that trametinib led to T-cell distributions in the spleen that resembled those of unengrafted spleens (Supplemental Fig. 3). These results, combined with B-cell counts (Figure 6D), further suggest that trametinib treatment may attenuate CLL development in vivo.

Similar to our studies in vitro, we sought to determine the source of MEK activity in the Eµ-TCL1 adoptive transfer mouse model. For this, we stained 9-wk post-transfer spleen samples for either phosphorylated or total ERK and F4/80. As shown in (Fig. 10, there was a marked convergence between ERK expression and F4/80, denoted by the yellow color in the merged image. In addition, trametinib treatment significantly reduced staining of phospho-ERK and total ERK (Fig. 10B, 10C), with corresponding decreases in F4/80. This shows that most of the ERK activity is within the macrophage population and that blocking ERK impacts this population. As expected on the basis of our observations of MEK in the NLC cultures in vitro, the images suggest a greater amount of MEK activity in the engrafted mice than in nonengrafted controls (Fig. 10A, top panel).

As the most common type of leukemia in the United States, CLL has been heavily investigated to design better therapies and improve patient outcomes. To date, Ab therapy and small-molecule inhibitors are successfully being used, although relapse and resistance remain problematic even with relatively new treatments such as ibrutinib (41, 42). In recent years, the importance of TAMs and leukemia-associated macrophages, including NLCs in CLL, has become evident with the help of different animal models (10, 11). Our laboratory has studied NLCs with this purpose in mind, having recently shown that these cells are sensitive to IFN-γ and can increase the elimination of CLL cells through Ab-mediated phagocytosis (43). This suggests that the targeting of NLCs may be an underused, effective means of treatment for CLL.

Although the role of NLCs in supporting CLL cells has been broadly studied, many questions remain about how these cells are generated. In the present work, we aimed to identify the major pathways responsible for NLC differentiation as a novel tool to treat CLL. DNA methylation loss in NLCs was found and shown to be related to TF binding motifs, including AP-1 and MAF. This pattern is similar to a previous report describing macrophage differentiation, where there is substantial loss of methylation, including at the AP-1 TFBSs (44). Interestingly, a report suggests that c-MAF is a driver of the TAMs in lung cancer, where it promotes a tumor-supportive M2-like state (45). Thus, our findings in NLC DNA methylation hallmarks are similar to what has been reported in the solid tumor setting.

We also found slight but noticeable differences between the methylation patterns of healthy donor–derived and patient-derived NLCs. These differences have been seen previously, where patient-derived NLCs express higher quantities of CD68 and CD163, along with a higher capacity to promote CLL survival in vitro (31). This suggests that the use of healthy donor cells as an NLC model, although possibly easier, should be considered carefully.

After defining the methylation epigenotype in NLCs, we found that their differentiation was highly dependent upon MEK. When MEK inhibitors were tested on CLL patient PBMC cultures, we observed a marked reduction in NLC numbers. Importantly, this could not be explained solely by the promotion of myeloid cell death. In line with our present findings, a previous report has shown that the use of pacritinib, a JAK2 and FLT3 inhibitor, affects CSF-1R–mediated signaling, including ERK phosphorylation (13). Nevertheless, pacritinib seems to affect both CLL cells and NLCs. Here, we found that the MEK inhibitor trametinib at 1 or 10 µM did not directly affect CLL cell survival in culture, suggesting that the effect of MEK inhibition may be more related to NLC development and thus may affect CLL survival indirectly. This was further observed using annexin V/PI, where a significant decrease in CLL cell viability was shown after 14 d of culture with trametinib treatment (data not shown), which may also explain the nuclear condensation seen in (Fig. 5. A previous report using MEK inhibitor I found that CLL cells were directly sensitive to the compound and that this effect remained in coculture conditions with fibroblasts (34). Nevertheless, the group found that the inhibitor blocked not only this pathway but also AKT, which may explain the difference between their findings and those presented here.

The results from the present study support our hypothesis that trametinib impacts survival by affecting the NLC/macrophage population. Because TAMs are defined as M2-like cells, we used previously published markers for M1 (iNOS) or M2 (EGR2) phenotypes (35) to measure the effects of MEK inhibition in vivo. Of particular interest, we found that EGR2 TFBSs were significantly less methylated in NLCs (Fig. 1) and that the EGR2 transcript was higher (data not shown). The specific roles of EGR2 within the context of NLCs is a topic of further investigation.

Previously, it has been suggested that monocytes from blood infiltrate the tissues in mice and can increase the expression of MHC-II; also, we know that NLCs originate from CD14⁺ monocytes (24, 46). In addition, increased expression of MHC-II in the spleen of Eµ-TCL-1 in comparison with nontransferred wild-type mice has been reported, whereas NLCs have been shown to express MHC-II (10, 24). Interestingly, we may have seen a clear trend, because there was a significant decrease in MHC-II⁺ and MHC-II⁺F4/80⁺ cells in the spleen at week 7 after trametinib treatment, although only MHC-II⁺/F4/80⁺ cells showed a reduction by week 9. Notably, MHC-II⁺/F4/80⁺ cells showed significant reductions in the number of EGR2⁺ cells, suggesting that much of the effect of trametinib was upon this M2-related population. This may be closely related to the enhanced survival time of trametinib-treated mice, because it is in agreement with the findings of Galletti et al., who found a survival advantage after macrophage depletion (11). In addition, we observed some significant changes in the expression of iNOS in both MHC-II⁺ and MHC-II⁺ at week 9. Nevertheless, the number of M1-like cells is a small proportion of the total populations. In addition, when testing M0, M1, and M2 macrophages derived in vitro from healthy donors, trametinib showed no effect on viability within the M1-like population. Hence, the overall impact of trametinib on M2-like macrophages appears greater than that for M1-like cells.

In addition to the flow cytometric analysis, we assessed the histological changes in the spleens of mice treated with vehicle versus trametinib 9 wk after transfer. Previously, it had been reported that B220 expression was lower on the CLL cells of Eµ-TCL1 mice (40), which is readily visible when comparing engrafted, vehicle-treated mice against unengrafted control animals (Fig. 9). Interestingly, two of the trametinib-treated mice showed an almost normal B220 staining, whereas a third showed signs of a milder disease. This observation may be related to the results of our survival experiment, where some trametinib-treated mice died earlier of the disease, and to our observations with flow cytometry (Fig. 6D). Interestingly, results of CD4 and CD8 T-cell staining showed an apparently higher number of positive cells in the trametinib-treated group than in the vehicle group. This is comparable to a previous finding with the CT26 murine carcinoma model, where increased numbers of T cells were found, especially CD4⁺ cells (47). Finally, we confirmed that trametinib had an impact on MEK signaling in vivo, staining spleen sections against phosphorylated and total ERK, along with F4/80 staining. As expected from our experiments in vitro, most MEK activity in vivo appeared to be within the F4/80 macrophages, because staining for phosphorylated ERK and F4/80 substantially overlapped. In addition, trametinib-treated mice showed a reduction in F4/80 cells, similar to the decrease observed with total ERK. As such, it appears as though macrophages rather than the CLL cells are the major population showing ERK activity and are also at least one of the major targets of trametinib in vivo.

Collectively, these results would support the testing of trametinib as a potential treatment for CLL. This drug has been approved by the U.S. Food and Drug Administration for use in melanoma, with specific known side effects (30). An increase in side effects from trametinib in geriatric patients was also not statistically significant when compared with other cohorts (48). This may prove advantageous, given the relatively late onset of CLL.

In conclusion, the present work shows that NLC differentiation involves substantial DNA methylation changes and is a process dependent on the MEK pathway, because treatment with a MEK inhibitor reduces NLC number in vitro. In addition, MEK inhibition may result in fewer splenic monocytes/macrophages, preferentially affecting the M2-like population. Further research is focused on examining the differentiation process of NLCs in greater depth because this may uncover additional, more selective therapeutic targets.

The authors have no financial conflicts of interest.