Introduction Degeneration of the intervertebral disc (IVD) is associated with a loss of notochordal cells (NCs) from the nucleus pulposus (NP) and their replacement by chondrocyte-like cells. NCs have a stimulatory effect upon other cells, such as native IVD chondrocyte-like cells and mesenchymal stem cells, and therefore they are an interesting focus for regenerative tissue engineering strategies. However, optimal conditions, such as osmolarity and other characteristics of the culture media, for long-term culture of NCs are not known. The purpose of this study was to investigate the effects of different culture media and osmolarity on the physiology of NCs in vitro. Materials and Methods NC clusters isolated from canine IVDs were suspended in alginate beads and cultured at 37°C under normoxic conditions for 28 days. Three different culture conditions were investigated; (1) DMEM/F12 (300 mOsm/L), (2) α-MEM (300 mOsm/L), and (3) α-MEM adjusted to 400 mOsm/L to mimic a hyperosmolar environment. NC morphology, expression of genes related to NC markers, matrix production and remodeling, and DNA- and GAG analyses were performed on 1, 7, 14, and 28 days in culture. Data analysis were performed with the aid of a mixed linear model which incorporated fixed effects (i.e., “time in culture” 1, 7, 14, 21, and 28 days, and NP tissue), “culture condition” (DMEM/F12, α-MEM, and α-MEM 400 mOsm/L), and the interaction between these factors; in the random part of the mixed models, the factor “dog” (dogs 1-6) was incorporated to take the correlation within each dog into account. Benjamini-Hochberg correction was applied to correct for multiple comparisons. Results Large, vesicle-containing cells organized in clusters, characterized as NCs, remained present during 28 days for all culture conditions. However, the proportion of the NC clusters decreased over time, whereas the proportion of spindle-shaped cells increased. When comparing the different culture conditions, on day 28 in culture, the cell population cultured in α-MEM 400 mOsm/L appeared to contain more vacuolated cells (NCs) compared with the DMEM/F12 and α-MEM and hence fewer spindle and polygonal cells (Fig.). Gene expression profiling at 7, 14, and 28 days in culture compared with day 1 indicated an initial loss of NC phenotype followed by recovery of brachyury and aggrecan expression after 28 days of culture supporting a potential recovery of NC phenotype. NCs cultured in α-MEM adjusted to 400 mOsm/L showed the highest expression of brachyury, cytokeratin 18, and aggrecan, the highest GAG production, and the lowest collagen 1α1 gene expression. Conclusion In conclusion, NCs cultured in alginate in native cell clusters under normoxic conditions, partially retained their characteristic morphology and recovered their phenotype in long-term culture. In comparison to using DMEM/F12 (300 mOsm/L) culture medium, the use of α-MEM (300 mOsm/L) culture medium for culturing NCs has a beneficial effect on preserving the NC phenotype in culture, which is further facilitated by increasing the culture medium osmolarity to physiological levels (400 mOsm/L). This may be explained by the difference in composition of DMEM/F12 and α-MEM culture media: α-MEM contains a lower concentration of D-glucose (dextrose) than DMEM/F12 (5.5 vs. 17.5 mM, respectively). Since IVD cells are subject to low glucose concentrations in vivo, the relatively low glucose content in α-MEM may facilitate NCs to retain their phenotype. Moreover, hyperglycemia has been positively correlated with IVD degeneration in diabetic rats and recently it has been shown that a high glucose environment stimulates apoptosis and inhibits proliferation of NCs in vitro. Furthermore, preservation of the NC clusters may facilitate the NC physiology and the preserved extracellular matrix may participate in the presentation of growth factors to the cultured NCs. These findings provide additional information concerning the maintenance of NCs in vitro that may set the stage for further mechanistic inquiry into the biology of NCs and their potential role in designing new regenerative strategies. Acknowledgments This work was supported by AO Spine International, through an AOSpine Reserach Network grant (# 104982). Disclosure of Interest None declared