QSAR Accelerated Discovery of Potent Ice Recrystallization Inhibitors

Abstract

Ice recrystallization is the main contributor to cell damage and death during the cryopreservation of cells and tissues. Over the past five years, many small carbohydrate-based molecules were identified as ice recrystallization inhibitors and several were shown to reduce cryoinjury during the cryopreservation of red blood cells (RBCs) and hematopoietic stems cells (HSCs). Unfortunately, clear structure-activity relationships have not been identified impeding the rational design of future compounds possessing ice recrystallization inhibition (IRI) activity. A set of 124 previously synthesized compounds with known IRI activities were used to calibrate 3D-QSAR classification models using GRid INdependent Descriptors (GRIND) derived from DFT level quantum mechanical calculations. Partial least squares (PLS) model was calibrated with 70% of the data set which successfully identified 80% of the IRI active compounds with a precision of 0.8. This model exhibited good performance in screening the remaining 30% of the data set with 70% of active additives successfully recovered with a precision of ~0.7 and specificity of 0.8. The model was further applied to screen a new library of aryl-alditol molecules which were then experimentally synthesized and tested with a success rate of 82%. Presented is the first computer-aided high-throughput experimental screening for novel IRI active compounds.

Figure 1. Novel small molecule IRIs.

Figure 2. The effects of structural changes on IRI activity.

IRI activity values, represented as a percent mean grain size (MGS), are relative to a solution of phosphate buffered saline (positive control for ice recrystallization).

Figure 3. Select diverse structures from the 124 compounds used to develop the 3D-QSAR model.

A full list of all 124 compounds can be found in the Supporting Info.

Figure 4. List of 29 proposed aryl-alditol structures.

^aThis column indicates whether the QSAR model predicted the compound to be IRI active (A) or inactive (I).

Figure 5. IRI activity of compounds predicted to be active (125–135) at 22 mM (except those marked with asterisks which were measured at 11 mM).

The dark grey bar represents the PBS control. The dotted line represents the cutoff for activity where a MGS >70% is considered inactive and a MGS <70% is considered active.

Figure 6. Most relevant features and their specific correlations for compound 6 (left), possessing potent IRI activity and compound 18 (right) which does not possess IRI activity.

A prominent ESP-ESP interaction occurs at ~10 Å between the aryl substituent and the farthest hydroxyl group for both active and inactive compounds as shown by the green line. The most prominent ESP-ESP interactions (~6 Å) and curvature-ESP interactions (~3 Å) occur between atoms in the aromatic ring in IRI active compound 6, whereas these prominent interactions occur within the carbohydrate region in inactive compound 18.