SFCi55 and iKLF1.2 hiPSCs were maintained and the differentiation of macrophages was performed as previously described (Yang et al., 2017; Lopez-Yrigoyen et al., 2019; Lopez-Yrigoyen et al., 2020). Briefly hiPSCs were differentiated into embryoid bodies in the presence of 50 ng/mL VEGF, 50 ng/mL BMP4 and 20 ng/mL SCF for 6 days then plated in 25 ng/mL IL3 and 100 ng/mL M-CSF for at least 16 days with media being changed every 3 days. Suspension cells were collected at different time points then macrophages were matured for up to 10 days in the presence of 100 ng/mL m-CSF (macrophage maturation media). KLF1 was activated in iKLF1.2 macrophages with the addition of 200 nM of tamoxifen at days 7 and 9 of maturation.

Umbilical cord blood (UCB)-derived CD34⁺ cells (Stemcell Technologies, 70008.3) were thawed, expanded and differentiated into erythroblasts as described (Douay and Giarratana, 2009; Griffiths et al., 2012). Briefly CD34⁺ cells were first expanded for 6 days in ISHI media [IMDM (Gibco), 3U/mL Heparin (Merck), 10 μg/mL Insulin (Sigma), 200 μg/mL of holo-transferrin (Merck) and 5% Human AB serum (Sigma)) in the presence of cytokine cocktail A (60 ng/mL of SCF (Life Technologies), 5 ng/mL of IL3 (Peprotech) and 3U/mL of EPO (R&D Bio-Techne)]. These expanded cells were cryopreserved in batches of 10⁶ cells/mL in a 1:1 mix of ISHI media and freezing media (60% KnockOut Serum Replacement (Gibco), 20% DMSO (Invitrogen) and 20% ISHI media). These “day 6” cells were thawed and used as a starting point for the erythroid differentiation in macrophage co-culture experiments.

Cells for flow cytometry analysis were resuspended in PBS with 1% BSA (A2153, Sigma-Aldrich) and 5 mM EDTA (15575020, Invitrogen). 10⁵ cells per sample were stained with appropriate antibodies for 15 min at room temperature and with Hoechst dye as previously described (Lopez-Yrigoyen et al., 2019). Data was collected and analysed using the LSR Fortessa (BD Biosciences) with BD FACSDiva and FlowJo 10.8.1 software. Briefly, single and live cells were gated and FMO controls were used to distinguish specifically stained cell populations positive. All antibodies used are listed in Supplementary Table S1.

Immunofluorescence of macrophages was performed by collecting 6 × 10⁴ suspension cells and seeding them on a μClear^® cell culture 96-well plate for 10 days. iKLF1.2 macrophages were treated with tamoxifen twice at days 7 and 9 of maturation. At day 10 mature macrophages were washed twice with DPBS (Gibco) and fixed with 4% PFA for 10 min at room temperature. Cells were then permeabilised with DPBS + Triton-X100, 0.1% (DPBST) twice for 10 min, and blocked for 2 h with blocking solution (DPBST +1% BSA +3% Donkey Serum). Cells were then incubated with goat anti-EKLF/KLF1 antibody (Thermo, cat. PA5-18031) at a dilution of 1:200 in blocking solution overnight at 4°C. Cells were washed with PBST three times for 15 min and incubated with the secondary antibody, donkey anti-goat AF647, (1:1000) and DAPI (1:1000) in blocking solution for 2 h at room temperature in the dark. Cells were washed three times with DPBST for 15 min and DPBS was added before imaging. Images were taken with Zeiss Observer Microscope and Axion Vision software.

RNA extraction was performed using the RNAeasy Mini Kit (74106, QIAGEN) following the manufacturer’s instructions and cDNA was generated as previously described (May et al., 2023). Quantitative real time PCR reactions were performed on the Roche LightCycler^® 480 Instrument. 2 ng of cDNA was amplified with LightCycler^® 480 SYBR Green I Master (4887352001, Roche) according to the manufacturer’s instructions. All reactions were performed with 3 biological and 3 technical replicates. Ct values were normalised to the reference gene GAPDH or the mean of the reference genes GAPDH and β-Actin as indicated. Data was analysed using the 2^−ΔΔCT method and graphs were generated and statistical analysis was performed using GraphPad Prism 8 software. Primers used are listed in Supplementary Table S2.

For protein extraction, hiPSCs-derived macrophages (3 × 10⁶ cells/mL per sample) were washed twice with ice cold DPBS and detached with 200ul of Radioimmunoprecipitation assay (RIPA) buffer (25 mM Tris-HCl pH7.6, 150 nM NaCl, 1% NP-40, 1% Sodium deoxycholate, 0.1% SDS) (Thermo Scientific) and Protease inhibitor cocktail (Sigma), incubated on ice for 30 min and vortexed every 10 min. Samples were sonicated (Diagenode Bioruptor^® Pico) three times for 30 s in 30 s intervals, centrifuged for 15 min at 14000 g at 4°C and protein concentration of the supernatant was quantified using a Pierce Rapid Gold BCA Protein Assay Kit (Thermo Scientific) protein quantification assay. Lysates were snap frozen and stored at −80°C until analysis.

Quantitative proteomic analysis of hiPSC-derived macrophages was performed using Tandem Mass Tagging (TMT) as previously described (Daniels et al., 2021; Ferrer-Vicens et al., 2023). Briefly, 100 mg lysates were labelled with TMT 11-plex reagents (Thermo Scientific). An equal amount of all samples was pooled and used as reference during analysis. All spectra were acquired using Orbitrap Fusion Tribrid mass spectrometer on Xcallibur 2.0 software. Raw data files were processed and quantified using Proteome Discoverer v2.1 software (Thermo Scientific). Proteins were identified by searching against Uniprot Human database and Common Contaminants database, using Sequest^® algorithm. For the ‘Total Proteome’ analysis, each sample analysed was normalised on the ‘Total Peptide Amount’ and filtered using a 5% false discovery rate (FDR). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [1] partner repository with the dataset identifier PXD050357.

Network of biological and pathway analysis of differently expresses proteins were assessed using Web-based gene set analysis toolkit (WebGestalt) (https://www.webgestalt.org/, accessed September 2023) (Liao et al., 2019) and Reactome (https://reactome.org/, accessed September 2023). Using WebGestalt the differently expressed proteins were annotated against the identified proteins (ID mapped) in their database, from which gene ontology (GO) and over-representation analysis (ORA) were performed.

To prepare conditioned media, ISHI media was double filtered with PVDF membrane low protein binding filters with 0.22 μm (E4780-1221, Starlab) and 0.1 μm (SLVV033RS, Millipore) pore size, and was added to macrophages at day 7 of maturation (with 100 ng/ml m-CSF). Where appropriate filtered tamoxifen was added to macrophage conditioned media at a concentration of 200 nM. On day 10 of macrophage maturation, the media in contact with macrophages was collected, passed through a 5 μm strainer (43-50005-13, Pluriselect) then centrifuged at 300 g for 5 min to exclude cell debris. To subsequently deplete EVs, the conditioned media collected from macrophages was again double filtered with 0.22 μm (Starlab) and 0.1 μm low protein attachment syringe filters (Millipore).

For erythroid differentiation, day 6 UCB-derived erythroblasts were thawed and cultured in ISHI media in the presence of cytokine cocktail A until day 7 of differentiation (see above). At day 8 cells were centrifuged and cultured until day 10 in filtered and unfiltered macrophage condition media in the presence of cytokine cocktail B (60 ng/mL of SCF, 3U/mL of EPO, 1uM of Hydrocortisone (Stemcell Technologies) and 300 μg/mL of holo-transferrin (Merck)). From day 11 to day 21 of differentiation cells were cultured in either filtered and unfiltered macrophage conditioned media in the presence of cytokine cocktail C (3U/mL of EPO and 300 μg/mL of holo-transferrin).

Extracellular vesicles produced by hiPSCs-derived macrophages were analysed using Nanosight LM10 with an sCMOS camera (Malvern, Panalytical, United Kingdom). Collected media from macrophages was vigorously vortexed and injected in the Nanosight chamber 300 μL at a time using a 1 mL syringe and data was recorded for 3 × 30 s videos at 22°C, under media viscosity of 0.953 centipoise (cP). The number, size and concentration of particles in each sample was assessed using NTA software 2.3 (build 0011 RC1).

To assess whether our previous reported KLF1-related transcriptomic remodelling was translated into changes in the proteome, we carried out quantitative proteomic analyses of control and iKLF1.2 hiPSC-derived macrophages in the presence and absence of 200 nM tamoxifen (Figure 2A). Protein profiling was performed using Tandem Mass Tagging (TMT) Quantitative Proteomic analysis as previously described (Daniels et al., 2021; Ferrer-Vicens et al., 2023).

Principle Component Analysis (PCA) of the 7,868 independent proteins that were quantified showed that the different samples and their respective replicates fell into distinct clusters that did not overlap (Figure 2B). Pearson Correlation analyses demonstrated that samples derived from iKLF1.2 hiPSCs in the presence and absence of tamoxifen were more closely related to each other than the equivalent samples derived from control hiPSCs indicating some cell line-specific differences (Figure 2C).

Volcano plots of data generated from control and iKLF1.2 hiPSCs-derived macrophages in the absence of tamoxifen also revealed underlying intrinsic proteomic differences between cell lines (Supplementary Figures S2A–S2C). Volcano plots also revealed proteins that were regulated by tamoxifen alone which were proteins that were differentially expressed in macrophages derived from control hiPSCs in the presence and absence of tamoxifen (Supplementary Figure S2B). As expected, more differentially expressed proteins were identified when comparing macrophages derived from iKLF1.2 hiPSCs in the presence and absence of tamoxifen (Supplementary Figure S2C). We anticipated that this cohort of proteins would include proteins that were differentially expressed due to the activation of KLF1 as well as those that were induced by tamoxifen alone. We therefore considered that the most meaningful and biological relevant proteins would be those that were differentially expressed when comparing tamoxifen-treated macrophages derived from control and iKF1.2 hiPSCs (Figure 2D). This comparison identified 2051 differentially expressed proteins and following the application of a false discovery rate of 0.05, 321 differentially expressed proteins were identified. 44 of these proteins were eliminated as they were also identified as being activated by tamoxifen in macrophages derived from control hiPSCs. A final list of 277 proteins that were differentially expressed upon the activation of KLF1 in hiPSC-derived macrophages were then used in subsequent analyses. Of these, 173 proteins were significantly upregulated and 104 were downregulated (Figure 2E). Bioinformatic analyses were carried out proteins that were both KLF1-upregulated (Figures 3, 4) and KLF1 downregulated proteins (Supplementary Figures S4, S5). However, in this study we focussed our attention on the proteins that were upregulated by KLF1 as we considered these to be more likely to be associated with erythroblastic island macrophages.

Gene ontogeny analyses of the upregulated proteins using the web-based tool WebGestalt showed that 160 out of the 173 upregulated proteins were unambiguously identified (mapped), with 13 proteins unmapped due to their lack of function according to the protein database. For these mapped proteins those associated with metabolism was the top category within biological processes; protein binding was the top category within molecular functions and cell membrane and vesicles was top within the cellular components category (Figures 3A–C). As our key interest was in the communication of the macrophages within the EI niche with differentiating erythroid cells, we carried out further analyses on the upregulated proteins that were associated with membranes and vesicles. We annotated the sub-cellular localisation of these proteins and found that many of these proteins are found in the plasma membrane, suggesting a possible involvement in cell-to-cell communication (Figure 3D). Many of the KLF1 upregulated proteins associated with vesicles are associated with extracellular vesicles, including RAB27B which could be responsible for the transport of specific cargo relevant to cell communication (Figure 3E).

Over-representation analyses (ORA) analysis was also used to identify the most enriched biological pathways associated with differentially expressed genes (Figure 4A). The expression of all proteins associated with these enriched pathways was then assessed and we noted the upregulation of proteins associated with the citric acid cycle, formation of ATP and pyruvate metabolism implying that the KLF1 activation alters the metabolic profile of macrophages (Figure 4B). In keeping with the GO analyses, ORA analyses also revealed a significant enrichment of proteins associated with cell-cell communication and vesicle-mediated transport (Figure 4A). Further analyses of proteins in these categories in our dataset revealed that KLF1 upregulated proteins included those associated with cell junctions, membrane trafficking and the uptake and binding of vesicles (Figures 4C,D). This set of proteins was of particular interest because it is possible that they could be involved in the communication between macrophages and erythroid cells within the EBI niche. Although we had previously predicted that both secreted factors and direct cell contact were involved in communication, we had not considered nor explored the role of extracellular vesicles.

We previously reported that the communication between macrophages and differentiating erythroid cells involved direct cell-cell interaction as well as factors present in the media that could cross the membrane of a trans-well culture system. In that study we demonstrated that the addition of a few selected recombinant proteins (ANGPTL7, IL33 and SERPINB2) had an enhancing effect on erythroid maturation (Lopez-Yrigoyen et al., 2019). However, we had not assessed the effect of the complete conditioned media from control nor KLF1-activated macrophages. The content of that conditioned media would be predicted to include secreted factors as well as extracellular vesicles (EVs). EVs derived from macrophages have been shown to be associated with the inflammatory response and tissue repair (Wang et al., 2020) but to our knowledge there have been no studies on EVs derived from macrophages within the EBI niche. Given that some of the proteins identified in our proteomic screen were associated with extracellular vesicles (EVs) we set out to tested whether this form of intracellular communication may play a role in this system.

We first assessed whether cultured hiPSC-derived macrophages were able to produce EVs and whether their production was altered in response to the activation of KLF1. To ensure the exclusion of any media-derived particles, maturing macrophages from control and iKLF1.2 hiPSCs were switched to pre-filtered media at day 7 of their maturation. Three days later (day 10 of macrophage differentiation), conditioned media was collected and the number and size of EVs were quantified using Nanoparticle Tracking Analysis (NanoSight, Malvern) (Figure 5A). Filtered macrophage-conditioned media was used as a control and as expected, there were virtually no particles detected following filtration (Figure 5B). The concentration of particles in the unfiltered macrophage-conditioned media was highly variable, even between replicate samples (Figure 5B). Although media from macrophages derived from both KLF1.2 and its parental hiPSCs showed a slightly higher concentration of EVs in the presence of tamoxifen, indicating a minor effect of tamoxifen, this trend was not statistically significant (Figure 5B). There was no discernible effect of KLF1 activation on the production of EVs (Figure 5B).

To further characterise the EVs released by hiPSC-derived macrophages, the size distribution of the particles was determined in the conditioned-media derived from each cell line in the presence and absence of tamoxifen (Figure 5C). There was significant variability in the size of detected particles but the most common size of EVs detected was between 150 and 300 nm (Figure 5C). There were no differences in the size distribution of EVs between conditioned media derived from control and KLF1.2 macrophages and the addition of tamoxifen did not have a significant effect (Figure 5D). Taken together these data indicate that the activation of KLF1 had no discernible effect on the number nor the size distribution of EVs produced from hiPSC-derived macrophages.

Although activation of KLF1 did not alter the number and size distribution of EVs produced from hiPSC-derived macrophages we next assessed whether the EVs generated by KLF1-activated macrophages had any functional difference with respect to their ability to support erythroid cell production and maturation compared to those produced from control macrophages. We considered that the content of EVs might be altered by KLF1 activation and that this might account for some of the enhanced support by KLF1-activated macrophages that we previous reported (Lopez-Yrigoyen et al., 2019).

Umbilical cord blood derived CD34⁺ haematopoietic progenitor cells were differentiated in the presence of conditioned-media from hiPSC-derived macrophages ± tamoxifen and cell number, viability and differentiation status were assessed by flow cytometry (Figure 6A) (Supplementary Figure S3A). The number and phenotype of erythroid cells that were generated in the presence of EV-depleted conditioned media was compared to unfiltered conditioned media where macrophage-derived EVs were present. The proliferation of cells between days 8 and 11 of the differentiation protocol was lower when cells were cultured in filtered compared to unfiltered media conditioned by iKLF1.2-derived macrophages although activation of KLF1 by tamoxifen had no discernible effect (Figure 6B). The viability of cells at day 11, following culture in unfiltered and filtered macrophage-conditioned media, was comparable indicating that the reduction in cell number at this stage was the result of a reduced rate of proliferation rather than an increase in cell death (Figure 6C). However, as differentiation progressed, the viability of cells was reduced significantly to around 30% when the conditioned media was depleted of EVs by filtration, as compared to unfiltered media. Interestingly this difference was apparent and comparable in erythroid cells that were cultured in conditioned-media from macrophages that were generated from iKLF1.2 hiPSCs irrespectively of tamoxifen addition (Figure 6C) and comparable data were generated when cells were cultured in the presence of the conditioned media from macrophages derived from control hiPSCs (Supplementary Figures S3B, S3C).

As differentiation progressed, there was a significant increase in the proportion of cells expressing the erythroid marker CD235a (Glycophorin A) in all culture conditions (Figure 6D). However, the proportion of cells expressing CD235a was significantly lower when cells were cultured in filtered medium, compared to unfiltered conditioned-media derived from macrophages derived from both iKLF1.2 hiPSC (Figure 6D) and control hiPSCs macrophages (Supplementary Figures S3B–S3D). This suggests that the filtration process excluded components that were associated with the commitment and/or the survival of erythroid cells.

The progression in the differentiation of erythroid cells marked by the expression of CD235a is characterised by the reduction in the expression of CD71 and nuclear extrusion that is marked by the lack of DNA staining (Supplementary Figure S3A). At day 11 of differentiation, all CD235a-expressing cells also expressed CD71 and were nucleated and can therefore be defined as immature erythroid cells (Figure 5E). As differentiation progressed the proportion of nucleated CD235a+ cells that co-expressed CD71 increased significantly, reflecting the progressive production of enucleated erythroid cells. By day 21 of the differentiation process, around 25% of the cells cultured in unfiltered conditioned media were negative for both CD71 and Hoechst, whereas when conditioned media was filtered this proportion was less than 10% (Figure 6E). Morphological analyses of cells at day 21 showed the presence of robust nucleated and enucleated erythroid cells when grown in the presence of unfiltered media compared to the poor quality and fragile cells present following culture in the filtered media (Figure 6F). Comparable data were generated from control hiPSCs (Supplementary Figures S3B–S3D).

It was interesting to note that activation of KLF1 using tamoxifen did not alter the supportive effects of conditioned media in any of the parameters tested (Figures 6B–F), which appears to be in contrast with our findings in co-culture and trans-well assays (Lopez-Yrigoyen et al., 2019). However, we predict that this likely reflects either the stability or the concentration of secreted factors within collected and stored conditioned media compared to the media content in the presence of live macrophages. This question could be addressed in the future by the addition of exogenous growth factors such as ANGPTL7, IL33 and SERPINB2 to the conditioned media.

Taken together, these data indicate that components that do not pass through the 100 nM filter presumably including EVs, support the production, maturation and/or survival of differentiating erythroid cells.