Hiding Inside Cells

 BY DEREK LOWE

Now here's one that you probably didn't see coming. As everyone knows, relapses after various types of cancer therapy are all too common. There are a lot of possible explanations for this, and a ready one is clonal selection. When you start sequencing solid tumors on a cell-by-cell level, you realize that many of them are a mass of competing lineages, formed as they inevitably throw off mutations. The belief (naturally enough) is that some of these are more targetable than others by a given therapy and will be killed off, which unfortunately clears the way for more resistant lines to prosper afterwards. (This is one of the several ways that oncology can resemble infectious disease work).

This situation certainly applies to classic chemotherapy regimes, and to the more recent immunotherapies as well. But there's something odd going on in that case, and this new paper has a rather startling explanation. As the authors note, these immune approaches target "neoantigens", peculiar proteins that mark off tumor cells from the normal background. These are no picnic to find and validate, but there are some known for particular tumors, of course. This work shows that you can use an immune-targeted approach in a mouse model of melanoma to almost completely wipe out the tumors while targeting the melanoma antigen TRP1. But after that happens, you actually get recurrance in many of the animals, producing tumors that are resistant to further treatment. But here's the weird part: these tumors have almost identical neoantigen profiles as compared to the original tumors. This behavior holds up across more than one immunotherapy route and more than one tumor type (breast versus melanoma, e.g.), and it seems to apply to human patient data as well. What's more, when you isolate cells from these second-wave tumors, they seem to be just as vulnerable to T-cell attack. So what's going on?

The evidence suggests that it's not something in the genetic lineage of these cells that is allowing them to survive, but rather some strategy that's used in vivo and might thus relate to something that occurs with multiple cells around rather than being dependent on the characteristics of individual ones. Looking at primary tumor tissue (either isolated ex vivo or via histology of the tumor itself) showed something quite odd. There were cells that appeared to have multiple nuclei, and closer inspection showed multiple membrane structures as well. What seems to be going on is tumor cells hiding inside other tumor cells. Injecting mice with two tumor cell lines, otherwise identical but labeled with two different fluorescent markers, illustrated this: cells formed that showed both of these colors simultaneously, one within the other. And the same behavior could be seen in the ex vivo cells - on standing, they formed these cell-in-cell structures. The cytoplasms don't mix; the two cells are still distinct.

Further experiments showed that it was the presence of T cells that led to this change, especially the CD8+ ones, and that STAT3 and EGR-1 are important pathways in the tumor cells during this behavior. Looking over different tumor types, colon and ovarian cancer cells almost always formed these things to survive T-cell conditions, while pancreatic and B-cell lymphomas never did, with other types on a spectrum between these. Immortalized non-tumor cells were just flat-out killed by T-cell treatment and did not form cell-in-cell structures. The authors emphasize that the great majority of cells are still killed by T-cell attack and that these cell types are transient, but that the survivors are disproportionately from the cell-in-cell population. These results hold up in human tumor tissue from patients with relapsing melanoma and other tumor types. It looks from further studies that it's not that the cell-in-cell structures completely evade T-cell attack, but rather that the "double layering" just buys them more time. T-cell derived lytic granules, which are doing the killing here, don't seem to make it to the inner cells.

This is pretty odd, and nothing quite like it seems to have been described before. Cell-in-cell behavior has been seen under other conditions, but it's generally associated with apoptosis or one cell flat-out consuming the other. This reversible they'll-never-find-me stuff, though, seems new. The authors suggest that immunotherapy be preceded clinically by STAT3 inhibition, and demonstrated that pretreatment with a known tool compound (Stattic) did indeed make the tumor cells notably more sensitive to T-cell attack. Definitely worth knowing about!

https://www.science.org/content/blog-post/hiding-inside-cells