Single cell sorting is a crucial element of biologics discovery and cell line development that enables researchers to screen and sort cell populations for desired characteristics. However, the process has traditionally been complicated and expensive. While conventional methods such as flow cytometry are high-throughput, they can be damaging to the cells and result in limited cell viability. Other methods include limiting dilution - a manual, time-consuming approach, and single-cell printers - an automated, Inkjet-like printing approach, but neither can incorporate selective screening for production. Additionally, it may be difficult to accurately measure antibody secretion rates due to the rapid diffusion of molecules.

The Cyto-Mine® single cell analysis system uses proprietary picodroplet technology to overcome these limitations. The novel and powerful system encapsulates single cells in picodroplets that provide a uniquely protective environment to assay for secreted target proteins or productively titer. The encapsulated single cells are rapidly sorted based on fluorescence to find the positive “hit” cells, then imaged and dispensed, ready for further downstream analysis.

Cyto-Mine® uses photomultiplier tubes (PMT) to detect the fluorescent signal from the picodroplets, and it is essential to optimize the PMT settings to establish a good signal to noise ratio that distinguishes between positive and negative picodroplets for sorting and dispensing.

In this article, we outline some techniques and tips from our Senior Service Engineer, Darren Matthews, on optimizing PMT settings using Cyto-Mine®.

Detecting Picodroplets

Setting PMT gains

When the screen has moved to sorting, under the Sensors tab, the PMT settings can be established.

Care must be taken to optimize the PMT gains. If you set the gains too low, it will be difficult to detect a signal and distinguish what is a picodroplet. On the other hand, gains that are set too high can saturate and damage the PMTs or increase the noise level to the extent where it is difficult to distinguish between the signal and the noise.

Tips for setting PMT Gains:

As the signal detected on the voltage trace is on a non-linear scale, you can quickly go from a small signal to saturation. If you are working with a sample you have not run before, start with a low PMT gain around 0.2 and gradually increase by 0.1 to avoid saturation. If it remains difficult to differentiate between the noise and the picodroplet, then your signal is too low, and you need to keep increasing the gains. You are looking to set gains that differentiate the peak in fluorescence signal from the baseline so that you can identify your picodroplets of interest for sorting. Typically gains around 1-3 V peak height are ‘good’ for picodroplet detection.

Setting the thresholds

Once the PMT gains are established, two thresholds need to be set before the system will recognize and count each pulse as a picodroplet. These thresholds are necessary to determine the base signal that will trigger the detection of a picodroplet and exclude potential false triggers.

The next step is to specify the signal to be used for sorting. Compare the signal-to-noise-ratios of the green and red PMT and select the PMT with the best signal to noise ratio to validate.

Tips for setting the Threshold:

We have made it very simple to set the threshold with our ‘Auto’ button, but if you want to set your own thresholds, here are some tips.

The lower threshold should be set at a value just above the height of the noise to retain the majority of information from the signals. Then, the upper threshold should be set at a value above the lower threshold, approximately the width of the noise height. We use upper and lower thresholds to exclude any potential false trigger that could otherwise be caused by noise. So you’re just telling the system that anything above this line is a picodroplet and anything that doesn’t touch both lines should be ignored.

In the Scatter Plot and Heat Map, you will see the majority of the data points as a tight population: these are the negative or empty picodroplets. If you do not see picodroplets with fluorescence values that are different from the negative population on the Scatter Plot, check if your gains are too low. A low signal may also indicate that the length of incubation was not long enough for the cells to produce the secreted protein of interest. The good news is that you can re-incubate the picodroplets to allow more secreted proteins to build up and interact with the bioassay reagents to improve the signal.

Selective sorting

Setting the gating

Once thresholds have been set and the signal is appearing on the Scatter Plot, a polygon region can be added to the Scatter Plot or Heat Map tab to select the picodroplets for sorting. At this stage, it is best to keep the keep Sorted Frequency less than 20 Hz.

Whether you decide to use the Scatter Plot or Heat Map comes down to personal preference. Still, the Heat Map provides more in-depth knowledge of the density of picodroplets in a particular region, which is beneficial for identifying where the majority of the picodroplets are, and in particular where the negative population and the positive (containing secreting cells) starts. The data can be plotted as Peak or Average signal.

Tips for setting the Gating:

Peak is the maximum signal value from each picodroplet, whereas Average is the area of the signal divided by its maximum. Typically, I use whatever gives the best differentiation (spread on the Scatter Plot/ Heat Map) to Gate on, as some assays work better at Peak and others work better on Average. Homogeneous assays (where the fluorescence is homogeneously distributed in the picodroplet) like FRET-based single cell secretion, tend to work better using Average, while assays where fluorescence is clustered within the cell membrane or intracellular, tend to work better using Peak.

I tend to use the Heat Map to do my Gating as you can see the intensity of picodroplets at different points. To set the polygon region, you can draw a polygon around the picodroplets of interest, close the Gate, then drag and drop the edges to adjust the shape of the polygon.

*Note* If you have a very rare sample or a low concentration of positives you wish to sort, then the picodroplet population needs to be ‘bulked out’ with empty picodroplets to support the flow through the microfluidic channels. We recommend dragging the polygon region over the empty picodroplet section on the Scatter Plot/ Heat map to bulk out the population but avoid the densest area (yellow and red in the Heat Map).

Collecting empty picodroplets

Ideally, you want to collect at the start around 800 to 1,000 empty picodroplets that you can use at the flow stabilization phase before dispensing. During flow stabilization, everything is directed to waste. Therefore, only empty picodroplets should be used. After 800-1,000 empty picodroplets are sorted, you then want to select the positive picodroplets.

It is worth calculating if, and when, you are going to run out of picodroplets to better monitor the sorting process and to make sure you have enough time at the end of sorting, to select and bulk out with empty picodroplets. If you only have a small number of positives, “bulking out” helps to push the picodroplets through the microfluidic chambers.

The length of time it will take to run out of picodroplets depends on the ‘Approximate Number of Generated Picodroplets’ selected at Pre-Experiment Configuration. On average, you will run out of picodroplets approximately between 60 and 120 minutes of sorting.

Dispensing

Once sorting has concluded the picodroplets are collected in a chamber within the Cyto-Cartridge® ready to select for dispensing.

Setting the flow rate

Flow stabilization is essential to get a stable flow rate of picodroplets with good spacing. Picodroplets can be driven to the dispenser at a high flow rate (20 µL/h). The high flow rate is used to get the picodroplets out of the chamber, through the sorting channel and spaced out for dispensing.

However, the flow rate must be decreased as soon as you start to see picodroplets appearing, to reduce the number of picodroplets lost at this stage, as everything is directed to waste during flow stabilization.

Tips for getting the flow right

Picodroplet frequency is calculated in real-time so you can be more responsive. It can take time for the picodroplets to appear at the dispenser, usually around 5-10 minutes. Once the picodroplets have arrived, I tend to reduce the flow rate incrementally every few seconds, from 20 to 15 to 10 to 5 (going straight from 20 to 5 is sometimes too much of a jump). Once the flow rate has been set at 5, load the multi-well microtiter plate for dispensing. 

During flow stabilization, you can also adjust the focus of the dispensing camera to ensure clear images are taken before dispensing. Doing this while the flow rate is still at 20 is usually easier because a higher number of picodroplets can be seen in the channel.

Adjusting the PMT settings

At dispensing, PMT gains, threshold and detection trigger need to be set as described before. Starting slowly, increase the PMT gains until a signal appears. Set the PMT gains with a good signal to noise ratio then automatically select the threshold, choose the detection trigger best suited for the experiment, then validate.

Once the PMTs have been established for dispensing, move onto the Scatter Plot or Heat Map tab, set the polygon gating for dispensing and start reducing the flow rate. You should aim for a picodroplet frequency of 1 Hz (one picodroplet per second) or lower, using the lowest flow rate setting possible. You don’t want to dispense any faster than 1. Otherwise, you will increase the chance of having two picodroplets dispensed in one well. If this happens, it will be flagged in the data analysis software as a red circle.

Tips to set the Gate for dispensing:

If the assay is a FRET-based assay, you are looking for the region with the greatest shift towards the upper left corner. If everything you have sorted is positive, then everything is a ‘hit’ and can be selected for dispensing. If a rare sample is being analyzed, and both the positives and negative picodroplets were selected for sorting, then a split in the Scatter Plot will be seen at dispensing, similar to what observed at sorting.

Monitoring the dispensing process

You can view the progress of dispensing on the multi-well microtiter plate (green wells have been dispensed into, while blue are still empty) and view one image of the last imaged picodroplet. If the focus of the picodroplet is not satisfactory, you can modify it again now, and you will see the result in the next imaged picodroplet. As the image tab is not in real-time and the progress tab can be slightly behind, we recommend that you keep the Scatter Plot or the Heat Map tab open to watch the picodroplet frequency and ensure that it does not go too fast.

Do you need further advice or troubleshooting tips? 

Take a look at our series of How-To-Videos.  Here are two How-To Videos on PMT optimization you might find useful:

How to ensure a good signal from the PMT

How to selectively sort Picodroplets