Cell types share many traits, like specific markers on their surface, or the ability to make a particular molecule. But does that mean every cell of a certain type behaves the same way? Not necessarily. Many experiments are done on large groups of cells – tens or hundreds of thousands of a single type, cultured together in a well. These studies can tell us a lot, but what they can’t reveal is the unique response of each individual cell.
Single cell research has revealed that not every cell of a certain type acts identically; for example, a 2013 study in Nature showed two distinct patterns of mRNA expression and splicing in dendritic cells when they were exposed to bacterial lipopolysaccharide. (Read the paper) Characterizing single cell variations is especially important for understanding diseases like cancer, in which the individual cells in a tumor progressively acquire new mutations and can become, by the later stages of the disease, radically different from their clonal compatriots. These differences can have major consequences when selecting drugs to attack the tumor.
As important as it is to look at individual cells, however, it’s challenging to drill down to that level. The first obstacle is sample quantity; a eukaryotic cell yields only about 5 pg of genomic DNA. How can you sequence DNA when you have so little to work with? Whole-genome amplification is an option, but amplification bias has historically been a stumbling block. Another challenge is that the sampling methods for isolating single cells can lead to mechanical stress and cell death, or nucleic acid fragmentation. Finally, biological factors like cells at different division stages can affect results.
Now for the good news – new technology is finding ways to overcome all of these challenges, helping scientists refine ideas about what each cell is doing during a biological process or a disease state. Do you think single cell analysis is the right direction for your research field? What are some of the key discoveries that you think could be made by looking at individual rather than averaged responses?