The frequency of cell-cell fusion events induced by PEG using standard conditions is usually low, and this quantitative limitation can impede subsequent investigations. Therefore, we wished to develop an improved method for PEG-mediated cell fusion that yields a higher number of cell fusion products.
PEG damages cells. Consequently, the success of fusion depends at least to some extent on minute details such as the size and shape of the cell pellet, the intensity of shaking, and other factors that are difficult to standardize. PEG's effect on cells is mainly secondary to dehydration forces bringing the cells together. However, at least at a lipid interface, PEG changes the orientation and configuration of molecular dipoles and the organization of water molecules associated with the membrane. Thus, PEG probably induces changes in cell membrane permeability and in the properties of its cytoskeletal components, which may explain the cell rounding observed during fusion. PEG overtreatment results in excessive cell death and the formation of sincytia containing several nuclei, while undertreatment results in insufficient fusion.
Our strategy to improve PEG-induced cell fusion was suggested by the two-step electrofusion technique, which first accomplishes cell membrane breakdown by a high-intensity direct current pulse, followed by a second pulse that induces cell fusion. Similarly, we adopted a two-step approach to PEG-induced cell fusion. Using two under treatments with 35% PEG for 3 min instead of the standard single 50% PEG treatment for 5 min, we were able to obtain a high yield of cell fusion products. Importantly, most of these fusion products were bi- or trinucleated cells.
Once we had devised a method to improve PEG-mediated cell fusion, we addressed the problem of isolating cell fusion products. Genetic selection techniques such as the HAT (hypoxanthine, aminopterin, thymidine) system have been employed to isolate cell fusion products , although there are several practical and theoretical problems associated with their use. For example, only special cell lines that lack the required enzymes can be used. A second limitation is that substantial bias is introduced by the fact that most genetic selection systems depend on the ability of the fusion product to divide in selective medium. The inability to utilize genetic selection techniques for normal cells represents a limitation for many studies, such as the production of immunovaccines for tumour immunotherapy by dendritic-tumour cell fusion [[34–36]].
Flow cytometry has been offered as an alternative approach to quantifying and isolating hybrid cells after fusion. Dot plot-type fluorescence-activated cell sorting shows a quadrant in which a fluorescently coloured probe monitors one cell type, and a second coloured probe tracks a second cell type. A dually staining "fusion" quadrant is considered demonstrative of fusion. However, this evidence is not conclusive because aggregated cells and single cells coated with cell debris can appear in this "fusion" quadrant. Indeed, confocal microscopy confirmed these difficulties and proved to be superior to flow cytometry in detecting cell hybrids .
On the basis of these theoretical considerations, it seemed useful to isolate polykaryocytes, after PEG fusion, by procedures independent from genetic selection or fluorescent tracking. Cell separation by physical methods is an alternative approach to isolating these polykaryocytes.
The theory underlying cell sedimentation holds that the rate of sedimentation is largely a function of cell size, suggesting velocity sedimentation as an appropriate method for purifying cell fusion products . Accordingly, we found that separation of bi- and trinucleated cells was excellent due to their larger size than parent cells.
During this investigation, we made the accidental observation of some similar morphological features between in-vitro-fused cells and polykaryons we previously observed in HD  and erythroleukemia . Intercellular bridges were frequently observed after the first step of PEG undertreatment (Figure 1a-b) that closely resembles analogous structures present in a bone marrow preparation from an erythroleukemia patient (Figure 1d). Furthermore, large multinucleated cells with nuclei arranged in a circular fashion along the cell periphery can be observed both in vivo (in erythroleukemia and HD) and in vitro under different experimental cell fusion conditions (Figure 4). Jin has described similar cells in experiments when human kidney epithelial cells were exposed to chronic activation of protein kinase B  and by McShane while investigating EBV-induced cell fusion .
Hodgkin and Reed-Sternberg cells (HRS cells) are believed to originate from B cells, but their phenotypic pattern is very heterogeneous. B lineage markers such as CD20, B-cell receptor, and CD79a are found only rarely on these cells, and the lack of surface immunoglobulin expression sharply distinguishes them from their healthy counterparts. Whether this results from a) mutations within the rearranged immunoglobulin genes or non-functional rearrangements that prevent expression of these genes , or b) the extinction phenomenon that takes place when two cells fuse [40, 41]., remains an unanswered question.
Overall, the peculiar morphology and immunophenotype of HRS cells in infectious mononucleosis, HD, and other diseases are still a mystery. If cell fusion plays a role in giant cell formation in HD is presently only a tempting hypothesis that is difficult to prove . In infectious mononucleosis and in several cases of HD, Reed-Sternberg cells can be linked to infection with Epstein-Barr virus, a fusogenic agent. Stepwise transformation of virus-infected cells, possibly triggered by inherent genetic instability, might lead to aneuploidy and cancer . The correlation between aneuploidy and cancer has been known for decades; however, the central question of whether aneuploidy results from cell fusion or abortive mitosis is still unanswered.
Altered expression pattern for the I/i antigens have often been observed during oncogenetic processes as well as in benign haemopoietic disorders  Poly-N-acetyllactosaminyl, I (branched structure) and i (linear structure) structures, are important carbohydrate antigens and are presumed to have essential roles in the process of cellular recognition, differentiation , malignant transformation and cancer metastasis, they are also expressed in carcinoma cells in several tissues and organs . The i/I transition in cell differentiation is determined by the transcription factor CCAAT/enhancer binding protein alpha (C/EBPalpha), which enhances transcription of the IGnTC gene, consequently leading to formation of the I antigen. The loss of function of C/EBPalpha has leukemogenic potential by remodelling the transcription network of B cells through a series of parallel and sequential changes that require endogenous PU.1 .
With this in mind we investigated the i/I antigen expression in cell fusion products of lymphocytes isolated from peripheral blood and kept in vitro culture for prolonged time. The presence of i-antigen positive colonies suggest that cell fusion interferes with normal differentiation and gene expression. Microarray technology is a powerful tool that can quantify the expression of thousands of genes in a single analysis. It has the potential to monitor chromosome gains and losses, accomplish DNA resequencing, and detect mutations, allowing more intense probing of the mechanisms of tumour development. A "census of human cancer genes" was compiled that lists selected cancer genes having a causal link to mutation and oncogenesis . Using a microarray approach to compare Reed-Sternberg cells obtained in vivo and similar cells obtained in vitro might contribute to unravelling the enigma if cell fusion plays a role in Reed-Sternberg cell cytogenesis.
In conclusion, we have introduced some technical improvements in PEG-mediated cell fusion and in the isolation of cell fusion products that might prove useful in the study of human malignancy, in the field of gene mapping, and we have offered preliminary evidence that cell fusion might be involved in giant cell formation in HD.