Density gradient centrifugation compromises bone marrow mononuclear cell yield

Abstract

Bone marrow mononuclear cells (BMNCs) are widely used in regenerative medicine, but recent data suggests that the isolation of BMNCs by commonly used Ficoll-Paque density gradient centrifugation (DGC) causes significant cell loss and influences graft function. The objective of this study was to determine in an animal study whether and how Ficoll-Paque DGC affects the yield and composition of BMNCs compared to alternative isolation methods such as adjusted Percoll DGC or immunomagnetic separation of polymorphonuclear cells (PMNs). Each isolation procedure was confounded by a significant loss of BMNCs that was maximal after Ficoll-Paque DGC, moderate after adjusted Percoll DGC and least after immunomagnetic PMN depletion (25.6±5.8%, 51.5±2.3 and 72.3±6.7% recovery of total BMNCs in lysed bone marrow). Interestingly, proportions of BMNC subpopulations resembled those of lysed bone marrow indicating symmetric BMNC loss independent from the isolation protocol. Hematopoietic stem cell (HSC) content, determined by colony-forming units for granulocytes-macrophages (CFU-GM), was significantly reduced after Ficoll-Paque DGC compared to Percoll DGC and immunomagnetic PMN depletion. Finally, in a proof-of-concept study, we successfully applied the protocol for BMNC isolation by immunodepletion to fresh human bone marrow aspirates. Our findings indicate that the common method to isolate BMNCs in both preclinical and clinical research can be considerably improved by replacing Ficoll-Paque DGC with adapted Percoll DGC, or particularly by immunodepletion of PMNs.

Figure 1. Representative illustration of the gating strategy in rat samples.

The cells of interest were first categorized into cells belonging to the erythroid lineage (CD45−/Ery+), CD45−/Ery- cells and leukocytes (CD45+/Ery-). The latter were then differentiated into B-cells (CD45R+/CD11b−), CD11b+ cells (CD45+/CD45R−/CD11b+) and other CD45+ cells (CD45+/CD45R−/CD11b−). Next, CD11b+ cells were separated into RP1+ polymorphonuclear cells (PMN) and RP1- monocytes/dendritic cells (Mo/DC).

Figure 2. Influence of ammonium chloride lysis on rat bone marrow composition.

(A) Freshly isolated rat bone marrow consists of 60% CD45+ and 40% CD45- cells. Lysis caused a significant decrease of CD45- (−85%) and CD45+ (−36%) cells. (B) The forward/sideward scatter diagram revealed an almost complete depletion of erythroid cells (arrow) whereas other main bone marrow populations remained largely unchanged. (C) Quantification of the cell fractions revealed that the loss of CD45+ cells due to lysis occurred symmetrically among the main leukocyte subpopulations. (D) Staining against an erythroid marker supported the finding that most cells of the erythroid line (CD45−/Ery+) disappeared following lysis whereas other CD45−/Ery- cells were significantly enriched. This was further potentiated after the application of Ficoll, Percoll or MACS depletion leading to an extended loss of erythroid cells in favor of CD45−/Ery-. Values are means ± SD for 5 samples. *p<0.05 by t-test.

Figure 3. Adjusted density gradient centrifugation by using Percoll.

Percoll of different densities (1.071 to 1.084 g/mL) differentially separated the major subclasses of lysed rat bone marrow. The higher the density, the higher was the BMNC yield (within layer 1) and the lower the BMNC loss (within layer 2), respectively. The CD45- population remained stable in layer 1 at densities from 1.073 g/mL upwards. At a density of 1.084 g/mL the BMNCs were increasingly contaminated with PMNs. Values are means from 2 pooled samples per density.

Figure 4. Rat BMNC yield following different isolation procedures.

(A) Compared to lysed bone marrow, each of the isolation procedures caused an almost complete depletion of PMNs (C; arrows indicate the remaining RP1+ PMNs). As an unwanted side effect, this was accompanied by a significant loss of BMNCs and CD45- cells (A). BMNC loss was maximal after Ficoll DGC followed by Percoll DGC and MACS separation (*p<0.05). Both Percoll and MACS preserved the CD45- population compared to Ficoll (A; ^§p<0.05). Further analysis revealed a symmetric cell loss among the BMNC subpopulations (B; C, representative forward/sideward scatter diagrams). (D) For both Percoll and MACS, primarily RP1+ PMNs but also particular BMNC populations were detected within the waste (Percoll: layer 2; MACS: content of the columns). Values are means ± SD for 5 samples. *^,§p<0.05 by one-way ANOVA.

Figure 5. Recovery of rat hematopoietic progenitors.

(A) Analysis of the CFU-GM frequency revealed a significant enrichment of hematopoietic stem cells (HSC) by Ficoll compared to MACS. However, when relating the CFU-GM number to the absolute BMNC yield, Ficoll DGC resulted in a significant loss of HSCs compared to both Percoll and MACS separation. Values are means ± SD for 4 samples. *p<0.05 by one-way ANOVA.

Figure 6. Isolation of BMNCs from fresh human bone marrow aspirates and determination of progenitor cells.

(A) Fresh whole bone marrow (BM) contained a high proportion of CD45- erythroid cells. The amount of erythroid cells could be significantly decreased by lysis or by immunodepletion (MACS or PluriB) of erythrocytes whereas highest purity of CD45+ leukocytes was attained after immunomagnetic depletion (MACS). (B) Lysis of whole bone marrow caused a significant loss of BMNCs (*p<0.05 versus whole bone marrow) that was further extended by PMN depletion with either MACS or PluriB (^#p<0.05 versus lysed bone marrow). The combined depletion of erythrocytes and PMNs by sequential MACS or by combined PluriB yielded comparable BMNC counts. The amount of remaining PMNs was constant at a low level after each of the isolation procedures. (C) Gating strategy for progenitor characterization. CD34+ hematopoietic stem cells (HSCs; C 1) featured a uniform, lymphoid-like phenotype and were partially CD133+ (C 3). In contrast, CD105+ mesenchymal stromal cells (MSC; C 2) exhibited increased variability of size and granularity. (D) Quantification of progenitors revealed that HSCs and MSCs were lost due to the different BMNC isolation procedures (p<0.05 versus lysed bone marrow). Cell loss was comparable in all experimental approaches except for the combined one-step depletion (PluriB), where the HSC yield was significantly lower compared to the other isolation protocols (^#p<0.05 versus lysis+MACS, lysis+PluriB and MACS+MACS). Values are means ± SD for 3 samples. *^,#p<0.05 by one-way ANOVA.