Increasing the sensitivity of the CytoBuoy flow cytometers

03.11.2014  by  George Dubelaar


We are expanding the huge particle size range of the CytoSense towards extremely small and dim fluorescent cells, such as Prochlorococcus marina.  The benefits are clear: to cover the whole size spectrum of phytoplankton and bacteria with a single instrument in the lab or in-situ.  The capability of the CytoSense to analyse filaments, big cells and colonies is a long known fact, its sensitivity for small cells is less well known.

In cooperation with the Mediterranean Institute of Oceanology we are exploring the sensitivity of the CytoSense.  The objective is to optimize the signal to noise ratio of the relevant fluorescence signals.  The extremely large flow cell and fast profiling electronics easily cause high backgrounds and noise levels, a challenging boundary condition. First tests with Mediterranean samples were very promising.  More efficient regeneration of the recirculating sheath fluid (with our new automatic filtering system) decreased background fluorescence significantly.  We also introduced selectable sheath fluid velocity to let particles move slower through the laser focus, reducing noise bandwidth.  Finally the spatial filtering of stray light in the detection unit was improved by a better lens design.  

The pictures show a histogram and a scatterplot of the red fluorescence (620-700nm) versus sideward light scattering for fluorescence intensity calibration beads having 8 levels of fluorescent dye.  The dynamic range in signal heights of the CytoSense proved sufficiently large to analyse the complete range of 8 levels of fluorescence in one run (the dynamic range in integrated signals is virtually unlimited owing to the scanning data format).  Important is to note that even the lowest pigment containing cluster is well separated from the blank cluster without pigment.  The sideward light scatter detection proved also very sensitive, as 200 nm beads were easily detected (not shown).

In the meantime, more approaches are explored to further improve the result shown here, such as de-convolution of the pulse profiles to separate noise components from particle related signal emission.