Differentiating erythroblasts adapt to mechanical stimulation by upregulation of cholesterol biosynthesis via S1P/SREBP-induced HMGCR expression

Abstract

Understanding how mechanical stress affects erythropoiesis is crucial to produce transfusable erythrocytes in fluid-turbulent bioreactors. We investigated the effects of shear-stress on differentiating CD49d⁺CD235a⁺ primary human erythroblasts (EBL) at molecular, cellular, and functional level. Shear-stress, at differentiation onset, enhanced EBL maturation and induced upregulation of genes regulating cholesterol/lipids biosynthesis, causing changes in cell lipid composition. Of note, the osmotic resistance, and the expression of 3-Hydroxy-3-methylglutaryl-CoA reductase (HMGCR), the rate-limiting enzyme of the cholesterol biosynthesis pathway, were higher in dynamic cultures. Inhibition of the S1P-induced proteolytic cleavage, activating SREBPs, led to abrogation of HMCGR expression, and loss of EBL in dynamic cultures, similar to lovastatin administration. This data reveals a role for the S1P-SREBP-HMGCR-axis in the regulation of shear-stress induced adaptation during erythropoiesis, shedding light into mechanisms that will assist the upscaling of erythroid differentiation into bioreactors. Moreover, as shear-stress on hematopoietic cells occurs within the bone-marrow, these results introduce a novel signalling axis in the transduction pathways controlling erythropoiesis.

Fig. 1

Orbital shaker accelerates maturation and enucleation compared to static cultures. Erythroblasts (EBL) were differentiated for up to 10 days in dishes (static) or orbital shakers (dynamic). A) Erythroid cells were harvested at indicated days from dishes and flasks and stained with antiCD49d-PB (BD Biosciences, San Jose, CA, US ) (Alexa Fluor 405, y-axis) and antiCD235a-PE (OriGene Technologies, Rockville, MD, USA) (x-axis), and analysed by flow cytometry. B) Maturation states in static vs. dynamic conditions as explained in (A) were averaged for 5 donors (n = 5) by quantifying the percentage of cells within the respective gates Q1: CD49d⁺/CD235^- immature EBL, Q2: CD49⁺/CD235a⁺ EBL, and Q3: CD49^-/CD235a⁺ late EBL and reticulocytes. C) Enucleation was quantified by staining cells with the cell-permeable DNA-dye DRAQ5 (1 ng/mL, incubation 5 min at RT; BioLegend, San Diego, CA, USA). (APC, y-axis), forward scatter (FSC, x-axis) on the indicated days. Representative scatter plot displaying the gating for reticulocytes, EBL, and pyrenocytes is shown in Supplemental Fig. 1A. Enucleation percentage is defined as the %reticulocyte/ (%reticulocytes + %nucleated cells) using the gates as defined in Supplemental Fig. 1A. D) Enucleation states in static vs. dynamic conditions as explained in (C) were averaged for 5 donors (n = 5) by quantifying the percentage of cells within the respective gates: reticulocytes, erythroblasts, and pyrenocytes for 5 donors (n = 5). E) Cell-counts for reticulocytes and EBL that were differentiated in static or dynamic conditions for 4 donors (n = 4). The distinction between the two populations was done via DRAQ5 staining. B) 2way ANOVA test was performed, with ***p < 0.001, ** p < 0.01 and *p < 0.05. D and E) Paired two-tailed student t-test was performed, with p** p < 0.01 and *p < 0.05. Unless marked, no significance was observed.

Fig. 2

Shear-induced enhanced maturation is progressively lost during erythroblast differentiation. A) Enucleation percentage (calculated as described for Fig. 1C) of samples switched from dynamic to static culture on Day 2, 4, or 6 of differentiation, or not switched, for 3 donors (n = 3). B) Enucleation percentage of samples switched from static culture to dynamic culture on Day 2, 4 or 6 of differentiation, as well as dish (static) and flask (dynamic) controls averaged for 3 donors (n = 3). Paired two-tailed student t-test was performed, with ** p < 0.01 and *p < 0.05. Unless marked, no significance was observed.

Fig. 3

RNA analysis between static and dynamic erythroblast differentiation confirms accelerated maturation and characterises the involved processes. Erythroblasts (EBL) of four different donors (n = 4) were differentiated for 4 days as indicated in material and methods. A) Heatmap showing a Pearson hierarchical clustering of z-scores of the 505 differentially expressed RNAs between Day1 and Day4. K-means clustering, reveals 3 clusters indicated as K1, K2, K3. Genes involved in the cholesterol biosynthesis pathway are reported next to their specific coordinates on the heatmap. B) The gene identifiers of the upregulated RNAs of K2 from (A) were extracted and their expression dynamics through complete differentiation of EBL to enucleated reticulocytes, as previously published by Heshusius, was datamined and determined. Pearson hierarchical clustering was performed on z-scores over time in days as indicated (x-axis). Kmeans clustering (K = 3) was additionally performed and indicated as K2-a, K2-b and K2-c. C) ENRICHR analysis of K2-a to K2-c from (B) shows the top 10 enriched Gene Ontology (GO) term biological processes that are associated with the RNAs in the indicated Kmeans cluster according to the adjusted p-value. D) STRING analysis of the genes within K2-b was performed and genes within the first 3 biological processes according to the false discovery rate highlighted in red, blue, and green. Lines thickness indicates the strength of data support. E) Representation of cholesterol biosynthesis pathway with related z-score of genes up-regulated on day 4 in dynamic (D) compared to static (S) conditions. F) HMGCR (ab242315, Abcam) expression was assessed through western blot analysis, during EBL differentiation of 3 different donors (n = 3) in static and dynamic conditions. GAPDH (MAB374, Millipore) was used as loading control. G) Quantified expression of HMGCR during cell differentiation in dynamic and static conditions. HMGCR expression was normalised according to GAPDH expression. G) 2way ANOVA test was performed, with ** p < 0.01 and *p < 0.05. Unless marked, no significance was observed.

Fig. 4

Day 4 erythroblasts differentiated in dynamic conditions show higher concentration of cholesterol triglycerides and phosphatidylcholine compared to cells differentiated in static. Day 4 erythroblast (EBL) of three different donors (n = 3) were characterized by global lipidomics to determine their lipid composition. A) Partial Least Squares-Discriminant Analysis (PLS-DA) of samples differentiated in static and dynamic conditions. X-axis represents variations between conditions, y-axis represents variations between donors. B) Heatmap of the top 75 lipids differentially expressed at day 4 of differentiation in static (left- red) and dynamic (right- green) conditions according to the adjusted p-value. Cluster A represents lipids upregulated in static and cluster B lipids upregulated in dynamic. C) Pie-charts representing the lipid composition in percentage of day 4 EBL obtained from static and dynamic cultures. The most expressed lipids in static and dynamic condition are outlined in yellow. D) Comparison of the single lipids detected in reticulocytes obtained from static and dynamic conditions. X-axis represents the abbreviation of lipids name, y-axis the area under the curve. Paired two-tailed student t-test was performed, with ** p < 0.01 and *p < 0.05. Unless marked, no significance was observed.

Fig. 5

Reticulocytes obtained from dynamic cultures show higher cell stability but similar deformability and lipid composition when compared to static. Erythroblasts (EBL) of three different donors (n = 3) were differentiated for 12 days as indicated in material and methods. On day 12 of differentiation, cells were filtered, and the obtained reticulocytes subjected to cell deformability and stability assays. Measure of the reticulocyte’s deformability was detected using the ARCA. Cells were subjected to a shear stress of 3 Pa and 10 Pa. A schematic representation of the ARCA is shown in supplemental Fig. 4D. A) The x-axis represents the ratio between the length and the width of the cell (A/B), the y-axis is the normalised number of cells. A/B ratio directly correlates with reticulocyte deformability. B) Hemolysis percentage (y-axis) measured after incubation with different concentrations of NaCl solution (x-axis) was assessed to determine the stability of reticulocytes obtained from static and dynamic cultures. A and B) Native RBCs were used as control in deformability and stability assays. A) Unpaired two-tailed student t-test of the area under the curve of static condition vs RBC and dynamic condition vs RBC was performed, with *** p < 0.001, ** p < 0.01 and *p < 0.05. B) 2way ANOVA test was performed, with*** p < 0.001, ** p < 0.01 and *p < 0.05. Unless marked, no significance was observed.

Fig. 6

Lovastatin inhibition of HMGCR prevents cells adaptation to shear stress environment. Erythroblasts (EBL) of three different donors (n = 3) were incubated with different concentrations of lovastatin on day 0 of differentiation and cultured in static and dynamic conditions. Erythroid cells were harvested on day 12 of differentiation and analysed by flow cytometry. A) Representative example of living gate strategy plotting forward scatter (FSC x-axis) vs side scatter (SSC y-axis) in cells differentiated in static and dynamic conditions treated with different concentrations of lovastatin. B) Maturation states of cells differentiated in static and dynamic conditions incubated with different concentrations of lovastatin according to antiCD49d-PB (BD Biosciences, San Jose, CA, US) (Alexa Fluor 405, y-axis) and antiCD235a-PE (OriGene Technologies, Rockville, MD, USA) (x-axis), staining. C) Quantified percentage of cells in the living gate, averaged for 3 donors (n = 3). Gating strategy to identify the living cells gate is described in Supplemental Fig. 4). D) Quantified maturation states of cells differentiate in static and dynamic incubated with different concentrations of lovastatin were measured as described in Fig. 1A. Percentage of cells in gates was averaged for 3 donors (n = 3). ND indicates not-detectable measurement due to the low percentage of cells in the living gate. E) Representative picture of pelleted cells on day 12 of differentiation incubated with different concentration of lovastatin. C and D) 2way Anova test was performed, with *** p < 0.001 ** p < 0.01 and *p < 0.05. Unless marked, no significance was observed.

Fig. 7

S1P inhibition in the first 2 days of differentiation, reduced cell viability and HMGCR expression and led to a loss of EBL viability in dynamic cultures. Erythroblasts (EBL) of three different donors (n = 3) were incubate with the S1P inhibitor PF-429242 on day 0 of differentiation and cultured in static and dynamic conditions for 2 days. The effects of S1P inhibition were evaluated every 24 h. A) Cell-counts for EBL that were cultured in static or dynamic conditions in presence or absence of the inhibitor. B) Representative picture of pelleted cells of 3 donors (n = 3) on day 2 of differentiation in static or dynamic conditions, incubated with PF-429242. C) Representative example of living gate strategy plotting forward scatter (FSC x-axis) vs side scatter (SSC y-axis) in cells differentiated in static and dynamic conditions treated with PF-429242. D) Quantified percentage of cells in the living gate, averaged for 3 donors (n = 3). Gating strategy to identify the living cells gate is described in Supplemental Fig. 4. E) Maturation states in static vs. dynamic conditions as explained in Fig. 1A were averaged for 3 donors (n = 3) by quantifying the percentage of cells within the respective gates Q1: CD49d^-/CD235a^-, Q2: CD49d⁺/CD235^- immature EBL, Q3: CD49^-/CD235a⁺ late EBL and reticulocytes, Q4 CD49⁺/CD235a⁺ EBL. Representative example of gating strategy corresponding to the described conditions are reported below. F) HMGCR expression assessed through western blot analysis, during 2 days of EBL differentiation in static and dynamic conditions treated with PF-429242 and controls. BAND 3 was used as loading control. G) Quantified expression of HMGCR during cell differentiation in dynamic and static conditions. HMGCR expression was normalised according to BAND3 expression. (A, D, and G 2way Anova test was performed, with **** p < 0.0001, *** p < 0.001 **, p < 0.01 and *p < 0.05. Unless marked, no significance was observed.