Label-Free Digital Holotomography Reveals Ibuprofen-Induced Morphological Changes to Red Blood Cells

Understanding the dose-dependent effect of over-the-counter drugs on red blood cells (RBCs) is crucial for hematology and digital pathology. Yet, it is challenging to continuously record the real-time, drug-induced shape changes of RBCs in a label-free manner. Here, we demonstrate digital holotomography (DHTM)-enabled real-time, label-free concentration-dependent and time-dependent monitoring of ibuprofen on RBCs from a healthy donor. The RBCs are segmented based on three-dimensional (3D) and four-dimensional (4D) refractive index tomograms, and their morphological and chemical parameters are retrieved with their shapes classified using machine learning. We directly observed the formation and motion of spicules on the RBC membrane when aqueous solutions of ibuprofen were drop-cast on wet blood, creating rough-membraned echinocyte forms. At low concentrations of 0.25–0.50 mM, the ibuprofen-induced morphological change was transient, but at high concentrations (1–3 mM) the spiculated RBC remained over a period of up to 1.5 h. Molecular simulations confirmed that aggregates of ibuprofen molecules at high concentrations significantly disrupted the RBC membrane structural integrity and lipid order but produced negligible effect at low ibuprofen concentrations. Control experiments on the effect of urea, hydrogen peroxide, and aqueous solutions on RBCs showed zero spicule formation. Our work clarifies the dose-dependent chemical effects on RBCs using label-free microscopes that can be deployed for the rapid detection of overdosage of over-the-counter and prescribed drugs.

Section S1. Molecular dynamics model details S1.1 RBC membrane lipid bilayer model. We modelled the RBC membrane lipid bilayer based on the in silico lipid composition of the model erythrocyte membrane in ref.

S1.2 Preparation of the ibuprofen-lipid systems and molecular dynamics simulations.
Molecular Dynamics (MD) simulations were performed using Gromacs 2018.4 (4) software. The ibuprofen molecules and RBC lipid bilayer were represented by CHARMM General force field (5, 6) (CGenFF) and CHARMM36m (7) force field parameters, respectively. Five different ibuprofen aggregates were studied on top of RBC membrane bilayer -(1) single molecule of ibuprofen (Fig.  S12C), preformed aggregates of (2) 80 ibuprofen molecules, representing very low concentration, and (3) 100 ibuprofen molecules representing low concentration, but higher than 80-molecule aggregate, densely packed box of ibuprofen containing 1903 molecules under (4) isothermalisobaric ensemble (NPT) conditions at constant pressure, where the volume of the system is adjusted during simulation representing high concentration, and (5) canonical ensemble (NVT) at constant volume to model very high concentration of ibuprofen aggregates. System (1) will henceforth be regarded as "single ibu", system (2) as "low ibu conc. I", system (3) as "low ibu conc. II", system (4) as "high ibu conc. I", and system (5) as "high ibu conc. II". The preformed aggregates of ibuprofen (80 and 100 molecules) were modelled by running a MD simulation of randomly dispersed molecules of ibuprofen in water with counterions (Na + and Cl -) (see Fig. S12D). The ibuprofen aggregates formed instantly within 1 ns dynamics (Fig. S12D) due to strong hydrophobic intermolecular forces. The starting configuration of all five systems are shown in Figs. S12A-E. All ibuprofen-membrane complexes were solvated by filling the area above and below the membrane with water molecules represented by the modified TIP3P water model (8), creating a >20-Å thick water layer above the ibuprofen and below the membrane to mimic bulk solvation in the z-plane. Each simulation cell was neutralized by adding the appropriate number of counterions. After 5000 steps of energy minimization, each system was equilibrated over six consecutive steps (100 ps each), with the values of the force constants of position and dihedral restraints of lipids gradually decreased from 1000 to 0 (the unit for position and dihedral restraints are kJ/(mol.nm 2 ) and kJ/(mol.rad 2 ), respectively). During equilibration, the Berendsen thermostat and barostat were applied to maintain the temperature at 310 K and pressure at 1 atm. Semi-isotropic pressure coupling was applied to allow the lipid bilayer to fluctuate in the xy plane independent of the z-axis. For the production run, the Velocity rescaling thermostat and Parrinello-Rahman barostat were applied. Long-range electrostatic interactions were treated using the particle-mesh Ewald (PME) method. The time step used in our MD simulations is 2 fs, and the structures were saved every 100 ps during 0.1 μs of production dynamics.                 Table S3. Description of the extracted morphological parameters from Imaris 9.7.

Morphological
Parameter Description

Diameter
The length of the longest principal axis inside the object (BoundingBoxOO Length C)

Surface Area
The sum of the triangle surfaces Volume Quantification of how much a surface object occupies S/V Ratio Surface area divided by the volume

Thickness
The length of the shortest principal axis (BoundingBoxOO Length A)

Sphericity
The ratio of the surface area of a sphere to the surface area of the particle Mean RI Mean intensity of voxels enclosed within the surface Movie S1. Effect of 0.25 mM ibuprofen on RBCs. RI tomograms were acquired at 2 sec intervals over a period of 20 min. 3D renderings of extracted frames from the video are provided in Fig. 4. Timecode is min:sec.
Movie S2. Effect of 0.5 mM ibuprofen on RBCs. RI tomograms were acquired at 2 sec intervals over a period of 20 min. 3D renderings of extracted frames from the video are provided in Fig. 4. Timecode is min:sec.
Movie S3. Effect of 1 mM ibuprofen on RBCs. RI tomograms were acquired at 2 sec intervals over a period of 20 min. 3D renderings of extracted frames from the video are provided in Fig. 4. Timecode is min:sec.
Movie S4. Effect of 1.5 mM ibuprofen on RBCs. RI tomograms were acquired at 2 sec intervals over a period of 20 min. 3D renderings of extracted frames from the video are provided in Fig. 4. Timecode is min:sec.
Movie S5. Effect of 3 mM ibuprofen on RBCs. RI tomograms were acquired at 2 sec intervals over a period of 20 min. 3D renderings of extracted frames from the video are provided in Fig. 4. Timecode is min:sec.
Movie S6. 3D segmented rendering of a single RBC exposed to low (0.25 mM) ibuprofen concentration and measured with DHTM, showing transient spicule formation, movement and dissolution across the RBC membrane. Timecode is min:sec.
Movie S7. 3D segmented rendering of a single RBC exposed to high (1.5 mM) ibuprofen concentration and measured with DHTM, showing irreversible spicule formation and movement across the RBC membrane. Timecode is min:sec.