Automated Detection and Analysis of Exocytosis

The Automated Detection and Analysis Of Exocytosis software will allow the user to automatically detect exocytic events and TIRF image sequences of pH-sensitive fluorophores. It will also automatically output features of exocytosis such as the spatial distribution or the frequency, as well as individual properties of exocytic events, such as the half-life or change in fluorescence over the background. In addition, an option is included to classify exocytic events into four modes of exocytosis, previously described in literature.

In order to use the Automated Detection and Analysis of Exocytosis software, first you’ll click the find dataset button, and you’ll navigate to where your data is deposited, and you’ll wanna put in a folder called raw data. Your data files will automatically populate the list here, and you can have any number of data files in this folder. Next, you’ll wanna choose a directory for which your analysis files will be deposited.

Here, I’ve chosen a directory called test. You’ll also wanna fill in the frame rate of your images as well as the pixel size. Here my frame rates are a hundred milliseconds per frame, my pixel size is eight nanometers.

Finally, you will need mask files in order to run the Automated Detection and Analysis of Exocytosis software. You can use the included mask maker button in order to automatically generate mask files from your data files. The run indicator will turn yellow and then back to green when the mask maker is finished running.

Your mask will be deposited in a new folder called mask files in your chosen directory. And note, mask files will automatically populate the list here. You will wanna check that your mask files are appropriately made for your data files, and you can do this by highlighting any of the data files in the list, as well as the corresponding mask file.

The first frame of your data file will show up and the mask file selected will show up as well.Here. We can see that our mask files are appropriate for the analysis at hand. Mask files may also be supplied by the user separately.

When one wishes to make a mask file from a current data file we recommend using Image J.In order to do so, first open up the image in Image J that you wish to make a mask file from. You can then use the polygon selection tool in order to begin creating a mask file around by clicking around the edge of the cell. When you have completed your mask, double-click in order to connect to the entire polygon.

Once this is complete, you will head to edit, selection and create mask. An inverted mask will be created. You will want to save this mask file using the same name as the data file followed by _mask_file.

Now, if you provide your own custom mask files, it is important to let the automated detection software know where those mask files are located. To do so you will click the find mask files button, and navigate to the directory with your mask files in it. The new mask files will then populate the list here.

It’s important that you have a mask for every single data file before the analysis can be run. Now, once you have your dataset loaded, your mask files, your frame rate and pixel size adjusted correctly, and a chosen directory, you can finally decide whether you want to include classification as part of your analysis. If you toggle the classification button, in addition to detecting exocytic events, each exocytic event will be classified into one of four classes.

Once you have decided the way your analysis is to be run, you can then begin the analysis by clicking the analysis button. The run indicator will turn to yellow to indicate that the analysis is underway and will turn back to green when your analysis is finished. Once the analysis is complete, as indicated by your run indicator changing from yellow to green, you will notice that a new data files folder has appeared within your chosen directory.

Within the data files folder, you’ll find analysis files corresponding to each image set in your analysis run. In addition, a cell statistics file containing summary information such as the frequency of exocytosis for each of the image sets is here. For each image set, you have a fluorescent traces file, which contains information on the X position, Y position and frame number for where they exocytic events occur.

In addition, the average fluorescence in a region of interest around each exocytic event is presented both before, during and after exocytosis. In addition, there is also a tracking file, which contains similar information of the X, Y and temporal positions. However, if the classification checkbox is checked, in addition, there will be four extra columns, which indicate the probability of the exocytic event belonging to one of four classes.

Either full vesicle fusion instantaneous, full vesicle fusion delayed, kiss-and-run instantaneous, or kiss-and-run delayed. An exocytic event belongs to one of the four classes if it is greater than 0.5 and is the highest probability within the four classes governed. In this case, the first exocytic event here belongs to the full vesicle fusion instantaneous class, as it is the highest number above the four classes and it is greater than 0.5.

In addition, there are a number of other feature files for each image set, which are used during the classification of exocytosis and may be of interest for further analysis. Finally, if we want to use Ripley’s K analysis in order to detect the spatial-temporal organization of exocytosis, we will first start by splitting our mask file into a neurite mask file and a soma mask file. We’ll do this by first opening up our mask in Image J.We will wanna use the color picker in order to select a background pixel.

And this way, when we fill in the mask file, it is the correct value. Next we’ll use the polygon selection tool and outline the somatic region. Now this requires a bit of subjective, manual decision-making.

we suggest a rough ellipsoid. Once you’ve completed that you will then go to edit, selection and create mask. Finally, you will come back over to our original mask file and use edit and fill in order to fill in the soma, and now we have a separate neurite and soma mask file, which you will then save.

Once you have saved your separate neurite and soma mask file, here, I have it as mask file underscore neur for neurite and underscores soma, we’ll come over to MATLAB and open up the neurite 2D network MATLAB file. Here, we will navigate the current folder to our directory where we deposited all of our analysis data. Once we have done that, we will then have changed the mask name path to our new mask file that is the neurite.

So in this case, I have my neurite mask file under the mask files folder. We will then change the CSV file name to where our fluorescent traces file is located. In this case, it is still in the data files folder, and so the data files slash and the name of the fluorescent traces CSV file.

Once that has been completed, you can then hit run. This will then create a skeletonized version of the neurite mask file and deposit it as a CSV file under the mask files folder, which we can see here. Next we’ll generate a CSV file for the soma as well.

To do so, open the CSV mask creator file. You’ll want to put in the path for your soma mask and a name for the CSV file to be created. Here I just went ahead and used the same exact file name, just with dot CSV appended.

Hit run, and you’ll see that a new soma CSV file has been created alongside the neurite. Once we have created the CSV files for both the neurite mask and the soma mask, we can run Ripley’s K analysis. To do so we’ll navigate over to R Studio and open the Ripley’s K analysis R file.

There are two main variables here to pay attention to, neuron mask and neuron data points. Neuron mask will point to whichever mask files you wish to run. In this instance, I’m first running the soma mask files.

You will want to run all of your soma mask files separately from all of your neurite mask files. Here, I have two neurons, which I will be using for this analysis. However, you can use as many as you want for the Ripley’s K analysis, you’ll just want to copy and paste this code for neuron mask and change the variable to three, and onward.

The second variable is neuron data points. Here you want to point it to the file that was generated by your features all extracted R file. Now mine was named fusion stats, and so that is what it is reading in here.

As mentioned, I have a second soma mask file and neuron, which is being analyzed alongside so we can aggregate the Ripley’s K together. Once you have changed these paths to the correct pathing, you will then use code, run region, and run all. After the run is complete, several plots will be generated, including the grouped Ripley’s K values, as well as the density plots.

These can be saved by going to export, save image as and choosing the appropriate image format, the directory, file name, and finally hitting save. Here we see representative results from 12 murine cortical neurons expressing vent to fluorine imaged at two days in vitro using TIRF microscopy. In A, we see the frequency of exocytosis divided by class.

Here we can see that full vesicle fusion instantaneous occurs more frequently than the other classes. In B, we can see the mode distribution, confirming that full vesicle fusion instantaneous makes up more than half of all the events. In C we determine the spatial distribution of exocytosis as a heat map.

We can see that the majority of exocytic events are clustered in a hotspot near the soma, as well as at the distal ends of the neurites. In D we can determine that exocytic events are statistically significantly clustered, and that the size of these clusters range from half a micron to one micron in size. The use of an automated analysis program to correctly identify and analyze exocytic events in an unbiased manner increases analysis efficiency, and improves reproducibility and rigor.

To ensure accuracy of detection, it is important to maintain a high signal to noise during imaging. Capturing exocytic events or other pH-sensitive transient events requires an imaging frequency fast enough to capture all events and improve estimations such as the half-life or peak change in fluorescence. We have demonstrated that not only does this program work for accurately capturing pH-sensitive fluorescence in developing neurons, but other cell types as well.

However, if using another cell type, it is important to check for differences in accuracy, due to the distinct behaviors of transient events in other cell types. This classification has only been used in developing neurons to date. And indeed we do not know that these processes exist in other cell types or at later developmental time points in neurons.