SynapCountJ: A Tool for Analyzing Synaptic Densities in Neurons

Gadea Mata

, J

onathan Heras

, Miguel Morales

, Ana Romero

and Julio Rubio

Departamento de Matem

aticas y Computaci

on, Universidad de La Rioja, Logro

no, La Rioja, Spain

Institut de Neurocincies, Universitat Aut

onoma de Barcelona, Barcelona, Spain

Keywords:

Synapses, Synaptic Density, Image Processing, ImageJ.

Abstract:

The quantiﬁcation of synapses is instrumental to measure the evolution of synaptic densities of neurons under

the effect of some physiological conditions, neuronal diseases or even drug treatments. However, the manual

quantiﬁcation of synapses is a tedious, error-prone, time-consuming and subjective task; therefore, tools that

might automate this process are desirable. In this paper, we present SynapCountJ, an ImageJ plugin, that

can measure synaptic density of individual neurons obtained by immunoﬂuorescence techniques, and also

can be applied for batch processing of neurons that have been obtained in the same experiment or using the

same setting. The procedure to quantify synapses implemented in SynapCountJ is based on the colocalization

of three images of the same neuron (the neuron marked with two antibody markers and the structure of the

neuron) and is inspired by methods coming from Computational Algebraic Topology. SynapCountJ provides

a procedure to semi-automatically quantify the number of synapses of neuron cultures; as a result, the time

required for such an analysis is greatly reduced.

1 INTRODUCTION

Synapses are the points of connection between neu-

rons, and they are dynamic structures subject to

a continuous process of formation and elimination.

Pathological conditions, such as the Alzheimer dis-

ease, have been related to synapse loss associated

with memory impairments. Hence, the possibility

of changing the number of synapses may be an im-

portant asset to treat neurological diseases (Selkoe,

2002). To this aim, it is necessary to determine the

evolution of synaptic densities of neurons under the

effect of some physiological conditions, neuronal dis-

eases or even drug treatments.

The procedure to quantify synaptic density of a

neuron is usually based on the colocalization be-

tween the signals generated by two antibodies (Cuesto

et al., 2011). Namely, neuron cultures are permeabi-

lized and treated with two different primary markers

(for instance, bassoon and synapsin). These antibod-

ies recognize speciﬁcally two presynaptic structures.

Then, it is necessary a secondary antibody couple at-

tached to different ﬂuorochromes (for instance red

and green; note, that several other combinations of

color are possible) making these two synaptic proteins

visible under the ﬂuorescence microscope. The two

markers are photographed in two gray-scale images;

that, in turn, are overlapped using respectively the red

and green channels. In the resultant image, the yel-

low points (colocalization of the code channels) are

the candidates to be the synapses.

The ﬁnal step in the above procedure is the selec-

tion of the yellow points that are localized either on

the dendrites of the neuron or adjacent to them. Tools

like MetaMorph (Devices, 2015) or ImageJ (Schnei-

der et al., 2012) — a Java platform for image pro-

cessing that can be easily extended by means of plu-

gins — can be used to manually count the number of

synapses; however, such a manual quantiﬁcation is a

tedious, time-consuming, error-prone, and subjective

task; hence, tools that might automate this process

are desirable. In this paper, we present SynapCountJ,

an ImageJ plugin, that semi-automatically quantiﬁes

synapses and synaptic densities in neuron cultures.

2 METHODOLOGY

SynapCountJ supports two execution modes: individ-

ual treatment of a neuron and batch processing — the

workﬂow of both modes is provided in Figure 1

Mata, G., Heras, J., Morales, M., Romero, A. and Rubio, J.

SynapCountJ: A Tool for Analyzing Synaptic Densities in Neurons.

DOI: 10.5220/0005637700250031

In Proceedings of the 9th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2016) - Volume 2: BIOIMAGING, pages 25-31

ISBN: 978-989-758-170-0

Start

Batch

processing?

NO Configure setting

Channel

Green Image

Channel Red

Image

Structure

(txt-file)

Known

threshold?

Select thresholdNO

Write threshold

YES

Save the

settings?

Image with

analyzed region

Image with

indicated

synapses

Xml-file with

the settings

YES

Show the results

Table with

analysis

YES

Path of

directory with

images

Lif-files?

Xml-file with

settings

YES

Lif-file

End

Calculate number

of synapses and

density

Show table Save results images

Table with

analysis

Image with

analyzed region

Image with

indicated

synapses

Save settings

Calculate number of

synapses and

density

Choose input

files

Input path of

files

Choose

directories

Figure 1: Workﬂow of SynapCountJ.

BIOIMAGING 2016 - 3rd International Conference on Bioimaging

Figure 2: Neuron with two antibody markers and its struc-

ture. Top. Neuron marked with the bassoon antibody

marker. Center. Neuron marked with the synapsin antibody

marker. Bottom. Structure of the neuron.

2.1 Individual Treatment of a Neuron

The input of SynapCountJ in this execution mode are

two images of a neuron marked with two antibodies

(an image per antibody), see Figure 2. SynapCountJ

is able to read tiff (a standard format for biological im-

ages) and lif ﬁles (obtained from Leica confocal mi-

croscopes) — the latter requires the Bio-Formats plu-

gin (Linkert et al., 2010). The following steps are ap-

plied to quantify the number of synapses in the given

images.

In the ﬁrst step, from one of the two images, the

region of interest (i.e. the dendrites where the quan-

tiﬁcation of synapses will be performed) is speciﬁed

using NeuronJ (Meijering et al., 2004) — an ImageJ

plugin for tracing elongated image structures. In this

way, the background of the image is removed. The

result is a ﬁle containing the traces of each dendrite

of the image.

Subsequently, the user can decide whether she

wants to perform a global analysis of the whole neu-

ron, or a local analysis focused on each dendrite of

Figure 3: SynapCountJ window to conﬁgure the analysis.

Figure 4: SynapCountJ window to modify the threshold of

the red and green channels. Top. Window to ﬁx the thresh-

old of the image. Bottom. Fragment of the neuron image

with the synapses indicated as the red areas on the structure

of the neuron marked in blue. Moving the scrollbars of left

window, the marked areas of the image are changed.

the neuron. In both cases, SynapCountJ requires ad-

ditional information such as the scale and the mean

thickness (that is determined by the size of the subja-

cent dendrite) of the region to analyze (see Figure 3)

— these parameters determine the area of the dendrite

avoiding the background (i.e. all the non-synaptic

marking).

Taking into account the settings provided by the

user, SynapCountJ overlaps the two original images

SynapCountJ: A Tool for Analyzing Synaptic Densities in Neurons

of the neuron and the structure of the neuron pre-

viously deﬁned. From the resultant image, Synap-

CountJ identiﬁes the almost white points (the result

of green, red, and blue combination) as synaptic can-

didates, and it allows the user to modify the values

of the red and green channels in order to modify the

detection threshold (see Figure 4).

Once that the detection threshold has been ﬁxed,

the counting process is started. Such a process is

inspired by techniques coming Computational Alge-

braic Topology. In spite of being an abstract math-

ematical subject, Algebraic Topology has been suc-

cessfully applied in digital image analysis (S

egonne

et al., 2003; Gonz

alez-D

ıaz and Real, 2005) In our

particular case, the white areas are segmented from

the overlapped image, and the colors of the resultant

image are inverted — obtaining as a result a black-

and-white image where the synapses are the black ar-

eas. From such an image, the problem of quantifying

the number of synapses is reduced to compute the ho-

mology group in dimension 0 of the image; this corre-

sponds to the computation of the number of connected

components of the image.

Finally, SynapCountJ returns a table with the ob-

tained data (length of dendrites both in pixels and mi-

cras, number of synapses, and density of synapses per

100 micron) and two images showing, respectively,

the analyzed region and the marked synapses (see Fig-

ure 5).

2.2 Batch Processing

Images obtained from the same biological experiment

usually have similar settings; hence, their processing

in SynapCountJ will use the same conﬁguration pa-

rameters. In order to deal with this situation, Synap-

CountJ can be applied for batch processing of several

images using a conﬁguration ﬁle. It is necessary to

study at least one image from experiment to get the

optimal settings. The parameters are saved and used

to process the set of images from the same experi-

ment.

For batch processing, SynapCountJ reads tiff ﬁles

organized in folders or a lif ﬁle (the kind of ﬁles pro-

duced by Leica confocal microscopes), and using the

conﬁguration ﬁle processes the different images. As

a result, a table with the information related to each

neuron from the batch (the table includes an analy-

sis for both the whole neuron and from each of its

dendrites) is obtained. In addition, in the same direc-

tory where the lif-ﬁle or tiff-ﬁles are stored, the plu-

gin saves all the resultant images for each image from

experiment (one of them shows the marked synapses

and the other one, the region which has been studied).

Figure 5: Results provided by SynapCountJ. Top. Table

with the results obtained by SynapCountJ. Center. Image

with the analyzed region of the neuron. Bottom. Image with

the counted synapses indicated by means of blue crosses.

3 EXPERIMENTAL RESULTS

The original aim of SynapCountJ was the automatic

analysis of synaptic density on neurons treated with

SB 415286 — an organic inhibitor of GSK3, a ki-

nase which inhibition was proposed as a therapy in

AD treatment (DaRocha-Souto et al., 2003) — such

a treatment, as it was previously demonstrated, pro-

motes synaptogenesis and spinogenesis in primary

cultures of rodent hippocampal neurons and in Dros-

ophyla neurons (Cuesto et al., 2015; Franco et al.,

2004). In this setting, a comparative study has been

performed in order to evaluate the results that can be

obtained with SynapCountJ.

Primary hippocampal cultures were obtained from

P0 rat pups (Sprague-Dawley, strain, Harlan Labo-

BIOIMAGING 2016 - 3rd International Conference on Bioimaging

Figure 6: Quantiﬁcation of synapses. Left. Manual quantiﬁcation of synapses. Right. Quantiﬁcation of synapses using

SynapCountJ.

ratories Models SL, France). Animals were anes-

thetized by hypothermia in paper-lined towel over

crushed-ice surface during 2-4 minutes and eutha-

nized by decapitation. Animals were handled and

maintained in accordance with the Council Direc-

tive guidelines 2010/63EU of the European Parlia-

ment. Brieﬂy, glass coverslips (12 mm in diame-

ter) were coated with poly-L-lysine and laminin, 100

and 4 µg/ml respectively. Neurons at a 10 × 104

neurons/cm

density were seeded and grown in Neu-

robasal (Invitrogen, USA) culture medium supple-

mented with glutamine 0.5 mM, 50 mg/ml penicillin,

50 units/ml streptomycin, 4% FBS and 4% B27 (In-

vitrogen, CA, USA), as described before in (Cuesto

et al., 2011). At days 4, 7 and 14 in culture a 20%

of culture medium was replace by fresh medium.

Cytosine-D-arabinofuranoside (4 µM) was added to

prevent overgrowth of glial cells (day 4).

Synaptic density on hippocampal cultures was

identiﬁed as previously described in (Cuesto et al.,

2011). In short, cultures were rinsed in phosphate

buffer saline (PBS) and ﬁxed for 30 min in 4%

paraformaldehyde-PBS. Coverslips were incubated

overnight in blocking solution with the following an-

tibodies: anti-Bassoon monoclonal mouse antibody

(ref. VAM-PS003, Stress Gen, USA) and rabbit

polyclonal sera against Synapsin (ref. 2312, Cell

Signaling, USA). Samples were incubated with a

ﬂuorescence-conjugated secondary antibody in PBS

for 30 min. After that, coverslips were washed three

times in PBS and mounted using Mowiol (all sec-

ondary antibodies from Molecular Probes-Invitrogen,

USA). Stack images (pixel size 90 nm with 0.5 µm

Z step) were obtained with a Leica SP5 Confocal mi-

croscope (40x lens, 1. 3 NA). Percentage of synaptic

change is the average of different cultures under the

same experimental conditions. As a control, we used

sister untreated cultures growing in the same 24 well

multi plate.

A total of 13 individual images from three in-

dependent cultures has been analyzed. In Figure 6

we can observe that using a manual method to iden-

tify and count synapses, we obtain a mean of 24.12

synapses in control cultures and 16.74 in treated cul-

tures. The results obtained with SynapCountJ are

similar, there is a mean of 26.03 synapses in con-

trol cultures and 16.50 in the ones which have been

treated.

Notwithstanding the differences in the quantiﬁca-

tion, in both procedures we obtain almost the same in-

hibition percentage, a 30.51% manually and 36.61%

automatically. This shows the suitability of Synap-

CountJ to count synapses, meaning a considerably re-

duction of the time employed in the manual process.

Namely, the manual analysis of an image takes ap-

proximately 5 minutes; of a batch, 1 hour; and, of

a complete study, 4 hours. Using SynapCountJ, the

time to analyze an image is 30 seconds; a batch, 2

minutes; and, a complete study, 6 minutes.

4 DISCUSSION

Up to the best of our knowledge, 4 tools have

been developed to quantify synapses and measure

synaptic density: Green and Red Puncta (Shi-

warski et al., 2014), Puncta Analyzer (Wark, 2013),

SynD (Schmitz et al., 2011) and SynPAnal (Daniel-

son and Lee, 2014) — a summary of the general fea-

tures of these tools can be seen in Table 1. The rest of

this section is devoted to compare SynapCountJ with

SynapCountJ: A Tool for Analyzing Synaptic Densities in Neurons

Table 1: General features of the analyzed software.

Software Language Underlying Tech-

nology

Types of Im-

ages

Technique for de-

tection

Green and Red Puncta Java ImageJ tiff Colocalization

Puncta Analyzer Java ImageJ2 tiff Colocalization

SynapCountJ Java ImageJ tiff and lif Colocalization

SynD Matlab Matlab tiff and lsm Brightness

SynPAnal Java tiff Brightness

Table 2: Features to quantify synapses and synaptic density of the analyzed software.

Software Detection of den-

drites

Threshold Batch

Processing

Dendrites

length

Density Export Save

Green and Red

Puncta

Not used X

Puncta Analyzer Manual ROI X X

SynapCountJ Manual X X X X X X

SynD Automatic X X X X X

SynPAnal Manual X X X X

these tools — such a comparison is summarized in

Table 2.

There are two approaches to locate synapses in an

RGB image either based on colocalization or bright-

ness. In the former, synapses are identiﬁed as the

colocalization of bright points in the red and green

channels — this is the approach followed by Green

and Red Puncta, Puncta Analyzer and SynapCountJ

— in the latter, synapses are the bright points of a re-

gion of an image — the approach employed in SynD

and SynPAnal. In both approaches, it is necessary a

threshold that can be manually adjusted to increase

(or decrease) the number of detected synapses; such a

functionality is supported by all the tools.

In the quantiﬁcation of synapses from RGB im-

ages, it is instrumental to determine the region of in-

terest (i.e. the dendrites of the neurons where the

synapses are located); otherwise, the analysis will not

be precise due to noise coming from irrelevant regions

or the background of the image — this happens in

the Green and Red Puncta tool since it considers the

whole image for the analysis. Puncta Analyzer al-

lows the user to ﬁx a rectangle containing the den-

drites of the neuron, but this is not completely pre-

cise since some regions of the rectangle might con-

tain points considered as synapses that do not belong

to the structure of the neuron. SynD is the only soft-

ware that automatically detects the dendrites of a neu-

ron; however, it can only be applied to neurons with

a cell-ﬁll marker, and does not support the analysis

from speciﬁc regions, such as soma or distal den-

drites. SynapCountJ and SynPAnal provide the func-

tionality to manually draw the dendrites of the image;

allowing the user to designate the speciﬁc areas where

quantiﬁcation is restricted.

The main output produced by all the available

tools is the number of synapses of a given image;

additionally, SynapCountJ, SynD and SynPAnal pro-

vides the length of the dendrites; and, SynapCountJ

is the only tool that outputs the synaptic density per

micron. All the tools but Green and Red Punctua can

export the results to an external ﬁle for storage and

further processing.

Finally, as we have explained in Subsection 2.2,

images obtained from the same biological experiment

usually have similar settings; hence, batch process-

ing might be useful. This functionality is featured by

SynapCountJ and SynD, and requires a previous step

of saving the conﬁguration of an individual analysis.

SynPAnal does not support batch processing, but the

conﬁguration of an individual analysis can be saved

to be later applied in other individual analysis.

As a summary, SynapCountJ is more complete

than the rest of available programs. It can use differ-

ent types of synaptic markers and can process batch

images. Furthermore, a differential feature of Synap-

CountJ is that it is based on a topological algorithm

(namely, computing the number of connected com-

ponents in a combinatorial structure), allowing us to

validate the correctness of our approach by means of

formal methods in software engineering.

5 CONCLUSIONS AND FURTHER

WORK

SynapCountJ is an ImageJ plugin that provides a

semi-automatic procedure to quantify synapses and

BIOIMAGING 2016 - 3rd International Conference on Bioimaging

measure synaptic density from immunoﬂuorescence

images obtained from neuron cultures. This plugin

has been tested not only with neurons in development,

but also with the neuromuscular union of Drosophila;

therefore, it can be applied to the study of images that

contain two synaptic markers and a determined struc-

ture. The results obtained with SynapCountJ are con-

sistent with the results obtained manually; and Synap-

CountJ dramatically reduces the time required for the

quantiﬁcation of synapses.

As further work, it remains the tasks of improv-

ing the usability of the plugin and including post-

processing tools to manually edit the obtained results.

Additionally, and since the ﬁnal aim of our project

is the complete automation of the whole process, it

is necessary a procedure to automatically detect the

neuron morphology.

6 AVAILABILITY AND

SOFTWARE REQUIREMENTS

SynapCountJ is an ImageJ plugin that can be

downloaded, together with its documentation, from

http://imagejdocu.tudor.lu/doku.php?id=plugin:utiliti

es:synapsescountj:start. SynapCountJ is open source

and available for use under the GNU General Public

License. This plugin runs within both ImageJ and

Fiji (Schindelin et al., 2012) and has been tested on

Windows, Macintosh and Linux machines.

REFERENCES

Cuesto, G., Enriquez-Barreto, L., Caram

es, C., et al. (2011).

Phosphoinositide-3-kinase activation controls synap-

togenesis and spinogenesis in hippocampal neurons.

Journal of Neuroscience, 31(8):2721–2733.

Cuesto, G., Jord

an-

Alvarez, S., Enriquez-Barreto, L., et al.

(2015). GSK3β inhibition Promotes Synaptogenesis

in Drosophila and Mammalian Neurons. PlosOne,

10(3). doi=10.1371/journal.pone.0118475.

Danielson, E. and Lee, S. H. (2014). SynPAnal: Soft-

ware for Rapid Quantiﬁcation of the Density and

Intensity of Protein Puncta from Fluorescence Mi-

croscopy Images of Neurons. PLoS ONE, 9(12).

doi=10.1371/journal.pone.0115298.

DaRocha-Souto, B., Scotton, T. C., Coma, M., et al. (2003).

Brain oligomeric β-amyloid but not total amyloid

plaque burden correlates with neuronal loss and astro-

cyte inﬂammatory response in amyloid precursor pro-

tein/tau transgenic mice. Journal of Neuropathology

& Experimental Neurology, 70(5):360–376.

Devices, M. (2015). Metamorph research imag-

ing. http://www.moleculardevices.com/systems/meta

morph-research-imaging.

Franco, B., Bogdanik, L., Bobinnec, Y., et al. (2004).

Shaggy, the homolog of glycogen synthase ki-

nase 3, controls neuromuscular junction growth in

Drosophila. Journal of Neuroscience, 24(29):6573–

6577.

Gonz

alez-D

ıaz, R. and Real, P. (2005). On the Cohomology

of 3D Digital Images. Discrete Applied Mathematics,

147(2–3):245–263.

Linkert, M., Rueden, C. T., Allan, C., et al. (2010). Meta-

data matters: access to image data in the real world.

The Journal of Cell Biology, 189(5):777–782.

Meijering, E., Jacob, M., Sarria, J. C. F., et al. (2004). De-

sign and Validation of a Tool for Neurite Tracing and

Analysis in Fluorescence Microscopy Images. Cytom-

etry Part A, 58(2):167–176.

Schindelin, J., Argand-Carreras, I., Frise, E., et al. (2012).

Fiji: an open-source platform for biological-image

analysis. Nature methods, 9(7):676–682.

Schmitz, S. K., Hjorth, J. J. J., Joemail, R. M. S., et al.

(2011). Automated analysis of neuronal morphology,

synapse number and synaptic recruitment. Journal of

Neuroscience Methods, 195(2):185–193.

Schneider, C., Rasband, W., and Eliceiri, K. (2012). NIH

Image to ImageJ. Nature Methods, 9:671–675.

egonne, F., Grimson, E., and Fischl, B. (2003). Topolog-

ical Correction of Subcortical Segmentation. In Pro-

ceedings of the 6th International conference on Med-

ical Image Computing and Computer Assisted Inter-

vention (MICCAI’03), volume 2879 of Lecture Notes

in Computer Science, pages 695–702.

Selkoe, D. J. (2002). Alzheimer’s diseases is a synaptic

failure. Science, 298(5594):789–791.

Shiwarski, D. J., Dagda, R. D., and Chu, C. T.

(2014). Green and red puncta colocalization.

http://imagejdocu.tudor.lu/doku.php?id=plugin:analy

sis:colocalization analysis macro for red and green

puncta:start.

Wark, B. (2013). Puncta analyzer v2.0. https://github.com/

physion/puncta-analyzer.

SynapCountJ: A Tool for Analyzing Synaptic Densities in Neurons