How a flow cytometer reads a cell

A flow cytometer reads a cell by forcing it, usually one at a time, through a focused light beam and recording how that cell scatters the light and whether it emits fluorescence. The decision this page supports is what a dot plot can honestly say, and which controls have to exist before a gate is more than a drawing. It is a research explainer. It is not a clinical protocol, not a panel recipe, and not medical advice. Imaging the same kind of fluorescent label on a microscope is a different measurement, covered in fluorescence microscopy without the myths.

Instrument classes can be compared from the scientific instruments catalogue. The work often sits inside a broader molecular biology pathway. A specification goes to the quote request. Nothing on those pages asserts that a particular cytometer is on a bench.

The question the instrument can answer

Cytometry is good at counting many cells quickly and scoring each one for a handful of optical parameters. It is a poor way to see where inside a cell a protein sits. That spatial question belongs to microscopy. It is also a poor way to prove that a fluorescent antibody is specific. Specificity is a reagent and a control problem. The cytometer will digitise whatever photons arrive.

Typical research samples are blood-cell suspensions, cultured lines, or microbes in a single-cell state, prepared so that clumps are rare. If the particles are tissue fragments, the instrument is not reading cells. It is reading chunks, and the scatter plot will say so if you let it. Culture conditions for the lines you bring to the cytometer are a separate methods problem. ATCC culture guides describe handling expectations for collection material. They do not validate a panel.

One cell, one pulse

The sample is injected into a stream of sheath fluid. Hydrodynamic focusing squeezes that sample into a narrow core so particles pass the laser in single file. If the core is too wide or the concentration is too high, two cells cross together and become one event with the summed fluorescence. That doublet problem is why pulse width, or a related shape parameter, is plotted and gated before anyone believes a double-positive population.

As a particle crosses the beam it generates a pulse in each detector. Height is the peak. Area is the integrated pulse and is often the value used for fluorescence, because it copes better with cells that are not all the same speed. Width helps find aggregates. A sudden change in width across the whole file often means the flow rate or a partial blockage changed mid-tube. Look at time as a parameter. A sample that drifts is not a biological shift until the fluidics are stable.

Scatter and fluorescence

Forward scatter is light deflected by small angles, collected in front of the laser. It is influenced by size and by how the cell bends light. Side scatter is collected around a right angle and is more sensitive to internal complexity: granules, nuclei, membrane folds. Together they separate, for example, many lymphocyte, monocyte and granulocyte clouds in blood, but only as a starting map. The names of those clouds are an interpretation you earn with markers, not a label the axes print.

Fluorescence starts when a fluorophore absorbs in its excitation band and emits at a longer wavelength. Optical filters and dichroic mirrors split that emission toward detectors, historically photomultiplier tubes and now often avalanche photodiodes or similar solid-state detectors. Each detector is a spectral window, not a chemical identity. If two fluorophores both emit into that window, the detector cannot tell them apart by wishing.

The laser line has to match the excitation, not the colour name on the vial. A dye sold as "red" may be poorly excited by the laser you have and brightly excited by one you do not. Panel design is matching dyes to lasers and to filters, then checking that the biology, not the brightest dye, sits on the dimmest antigen. That last habit prevents the common failure in which a dump channel is gorgeous and the question channel is a smear.

Gates, compensation and controls

A gate is a region on a plot. Events inside it are carried forward to the next plot. The order matters. A usual research order removes time drift, debris and doublets, identifies the cells of interest with scatter or a lineage marker, and only then asks a fluorescence question. Gating on the fluorescence you hope to prove, and only afterwards looking at scatter, hides the fact that the "positive" events were debris.

Compensation corrects spectral overlap. You run each fluorophore alone, see how much of its signal lands in the other detectors, and build a matrix that subtracts those portions. The control has to be at least as bright as the sample. A dim compensation control under-corrects a bright sample and manufactures false positives. Compensation cannot repair tandem dyes that have degraded, antibodies that were mixed in one tube before staining, or autofluorescence that differs between samples. Unstained cells show autofluorescence. A fluorescence-minus-one tube, which contains every stain except the one you are gating, shows whether a threshold is sitting on spread from the other dyes.

Isotype controls are sometimes used and often over-trusted. An isotype of the wrong fluorescence-to-protein ratio does not model your antibody. Prefer a biological negative, a fluorescence-minus-one, and a stain you can block or compete if specificity is the claim.

When the control fails

If single stains look dim, pause before you acquire the panel. Voltage and gain should put the negative population out of the noise and the positive population on scale. Digital instruments still saturate. Events piled on the axis maximum are not "very positive". They are unmeasured. Dilute a sample that runs above the recommended event rate. A high rate is not efficiency. It is coincidence.

If compensation requires large negative values or populations dive below zero in a way you cannot explain as spread, a tandem dye may have uncoupled, a control may be mislabelled, or a filter may be wrong. Do not hand-edit the matrix until the plot looks pretty. Find the stain that misbehaves.

Clumps, high dead-cell fractions and viscosity change scatter and uptake of viability dyes. A viability gate that was drawn on a fresh control can be wrong on a sample that sat in a hot room. Record a viability dye when dead cells would otherwise appear positive.

Safety, aerosols and research limits

A cytometer that sorts, and some analysers with a damaged nozzle, can generate aerosols. Whether that matters depends on the sample. The decision is institutional, informed by references such as the CDC BMBL and the WHO Laboratory biosafety manual, not by this page. Do not bring material your laboratory has not assessed because the instrument has a door. Waste fluid is chemically and biologically active. Treat it as the risk assessment says. A research acquisition is not a diagnostic test.

Writers who record the panel, voltages and gate logic in a methods repository such as protocols.io leave a trail a second person can run. A screenshot without the control tubes is not that trail.

Heat, humidity and power

Lasers and fluidics dislike surprise heat and surprise power loss. A building that browns out mid-acquisition can shift voltages or stop sheath flow. Do not splice two halves of a file and call it one sample. In humid weather, condensation on a cold sample tube becomes a drip into the probe; wipe tubes and let them equilibrate. Sheath fluid that was made with poor water will sparkle with particles and raise the debris gate until real cells are a minority. Filter it. Write the sheath lot and the threshold. Two cytometers with the same laser names can still have different filters. Copying a voltage sheet from the other room is how a panel goes off scale.

What to send with an enquiry

State analysis versus sorting, laser and detector needs, plate or tube format, and the biosafety constraints of the sample. Ask for the optical specification in writing. Use the scientific instruments catalogue and the quote request. Ask whether a quotation is possible. Do not read a family name as a configured panel or as a claim that an instrument has been reserved.