Novel microfluidic rheology assay for in vitro-differentiated red blood cells: Enhancing quality control and therapeutic evaluation in sickle cell disease gene therapies

Abstract Introduction Gene therapies for sickle cell disease (SCD) edit hematopoietic stem cells (HSCs) to generate healthy red blood cells (RBCs), yet incomplete editing often results in heterogeneous cell populations with variable therapeutic efficacy. In vitro erythroid differentiation of HSCs offers a controlled platform to evaluate next-generation therapies, enabling precise assessment of gene editing efficiency, hemoglobin modifications, and rheological improvements. However, these methods yield limited sample volumes with high variability in cell morphology, size, rheology, and hemoglobin content, complicating biophysical and functional evaluations using traditional tools. No validated assays currently exist for these in vitro-derived RBCs. To overcome this, we pioneered a microfluidic rheology assay tailored for small-volume, heterogeneous RBC populations, providing sensitive, high-throughput insights into cellular function. Methods We evaluated rheological properties of in vitro-differentiated RBCs before and after nucleated cell depletion. Sample S1 comprised a mixed population of nucleated and enucleated cells, while Sample S2 was enriched for enucleated cells; both samples were prepared to comprise 2 million RBCs in 300 µL PBS, and they were tested in triplicate. For benchmarking, we analyzed peripheral blood RBCs from healthy donors (HbAA; n=20) and SCD patients (HbSS; n=34). Samples were perfused through a microfluidic network of 4 × 12 µm channels, with RBC flow captured via high-speed video (94 fps). A custom computer vision and machine learning tracking algorithm was developed to quantify individual RBC velocities, which were then normalized to the fastest flowing cell in each sample. The “slow RBC fraction” was defined as the percentage of cells with normalized velocities below 0.711, which was determined as a cutoff based on the mean normalized velocity. To simulate post-therapy heterogeneity, we tested mixtures of healthy and SCD RBCs (HbAA%:HbSS%, 100%:0%, 75%:25%, 50%:50%, 25%:75%, and 0%:100%). Results Peripheral HbAA samples showed a slow RBC fraction of 19.0% ± 4.9%, versus 42.1% ± 12.0% in HbSS, with unimodal velocity distributions skewed toward higher speeds. Sample S1 displayed a bimodal profile (peaks at 0.78 and 0.24 normalized velocity), reflecting rigid nucleated RBCs, yielding a high slow fraction (67.2% ± 6.4%; p=0.001 vs. HbSS). In contrast, enucleated-enriched Sample S2 exhibited a unimodal profile similar to peripheral RBCs, with a slow fraction of 32.9% ± 0.96% (p=0.01 vs. HbAA; p=0.03 vs. HbSS). The assay sensitively detected small abnormal RBC fractions in mixed samples. Discussion and Conclusion This innovative microfluidic assay represents a breakthrough in assessing in vitro-differentiated RBCs, enabling detection of subtle rheological abnormalities from minimal samples (e.g., <2 million RBCs). By enriching for enucleated cells, velocity profiles approached those of healthy peripheral RBCs, underscoring its utility for optimizing differentiation protocols. Clinically, it serves as a quality control tool in biomanufacturing, a potential biomarker for SCD severity and therapeutic response, and a platform to evaluate gene therapy efficacy in heterogeneous populations. Future applications could extend to real-time monitoring of editing outcomes and in vitro monitoring of personalized SCD treatments.