We previously demonstrated that human osteosarcoma cells (SAOS‐2) induce contact‐dependent apoptosis in endothelium, and expected similar apoptosis in human gingival fibroblasts (h‐GF) using SAOS‐2 alkaline phosphatase (AP) to identify cells. However, h‐GF apoptosis did not occur, despite reduction in AP‐negative h‐GF number (p < 0.01) and enhancement of this by h‐GF TNFα pretreatment (p < 0.01). We suggest that TNFα‐enhanced transfer of membrane AP from SAOS‐2 to h‐GF would explain these data. This idea was investigated using fluorescence prelabelled cells and confocal laser scanning microscopy. Co‐cultures of membrane‐labelled h‐GF (marker‐DiO) and SAOS‐2 (marker‐DiD) generated dual‐labelled cells, primarily at the expense of single labelled h‐GF (p < 0.001), suggesting predominant membrane transfer from SAOS‐2 to h‐GF. However, opposite directional transfer predominated when membrane labels were reversed; SAOS‐2 further expressed green fluorescent protein (GFP) in cytoplasm and nuclei, and h‐GF additionally bore nuclear label (Syto59) (p < 0.001). Cytoplasmic exchange was investigated using h‐GF prelabelled with cytoplasmic DDAO‐SE and nuclear Syto59, co‐cultured with SAOS‐2 expressing GFP in cytoplasm and nuclei, and predominant cytoplasmic marker transferred from h‐GF to SAOS‐2 (p < 0.05). Pretreating h‐GF with TNFα increased exchange of membrane markers (p < 0.04) but did not affect either cell surface area profile or circularity. Dual‐labelled cells had a morphological phenotype differing from SAOS‐2 and h‐GF (p < 0.001). Time‐lapse microscopy revealed extensive migration of SAOS‐2 and cell process contact with h‐GF, with the appearance of SAOS‐2 indulging in ‘cellular sipping’ from h‐GF. Similar exchange of membrane was seen between h‐GF and with other cell lines (melanoma MeIRMu, NM39, WMM175, MM200‐B12; osteosarcoma U20S; ovarian carcinoma cells PE01, PE04 and COLO316), while cytoplasmic sharing was also seen in all cell lines other than U20S. We suggest that in some neoplasms, cellular sipping may contribute to phenotypic change and the generation of diverse tumour cell populations independent of genetic change, raising the possibility of a role in tumour progression. Copyright © 2012 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

No abstract

We recently reported exchange of membrane and cytoplasmic markers between SAOS-2 osteosarcoma cells and human gingival fibroblasts (h-GF) without comparable exchange of nuclear markers, while similar h-GF exchange was seen for melanoma and ovarian carcinoma cells. This process of “cellular sipping” changes phenotype such that cells sharing markers of both SAOS-2 and h-GF have morphology intermediate to that of either cell population cultured alone, evidencing increased tumour cell diversity without genetic change. TNF-α increases cellular sipping between h-GF and SAOS-2, and we here study binding of SAOS-2 to TNF-α treated h-GF to determine if increased cellular sipping can be accounted for by cytokine stimulated SAOS-2 binding. More SAOS-2 bound h-GF pe-seeded wells than culture plastic alone (p<0.001), and this was increased by h-GF pre-treatment with TNF-α (p<0.001). TNF-α stimulated binding was dose dependent and maximal at 1.16nM (p<0.05) with no activity below 0.006 nM. SAOS-2 binding to h-GF was independent of serum, while the lipopolysaccharide antagonist Polymyxin B did not affect results, and TNF-α activity was lost on boiling. h-GF binding of SAOS-2 started to increase after 30min TNF-α stimulation and was maximal by 1.5hr pre-treatment (p<0.001). h-GF retained maximal binding up to 6hrs after TNF-α stimulation, but this was lost by 18hrs (p<0.001). FACS analysis demonstrated increased ICAM-1 consistent with the time course of SAOS-2 binding, while antibody against ICAM-1 inhibited SAOS-2 adhesion (p<0.04). Pre-treating SAOS-2 with TNF-α reduced h-GF binding to background levels (p<0.003), and this opposite effect to h-GF cytokine stimulation suggests that the history of cytokine exposure of malignant cells migrating across different microenvironments can influence subsequent interactions with fibroblasts. Since cytokine stimulated binding was comparable in magnitude to earlier reported TNF-α stimulated cellular sipping, we conclude that TNF-α stimulated cellular sipping likely reflects increased SAOS-2 binding as opposed to enhanced exchange mechanisms.