Unbiased quantification of Golgi scattering and Golgi-centrosome association

Abstract

The vertebrate Golgi complex is a large dynamic organelle which undergoes morphological changes and fragmentation both as a part of normal physiological dynamics and under disease conditions. Golgi is known to have a functionally important relationship with the centrosome. The extent of the spatial association between these two organelles varies in a dynamic and regulated manner. It is essential to have a reliable unbiased approach to evaluate Golgi volume, Golgi extension/scattering in the 3D cell space, and spatial association of the Golgi with the centrosome. It is also important that each of these features is evaluated by a simple metrics, one measurement per cell, so that the variability and deviations in the cell population can be easily assessed. Here, we present an approach to analyze confocal microscopy image stacks to easily measure Golgi volume, scattering, and association with the centrosome. The approach is based on a custom MATLAB script, provided here as a supplement, and also uses widely available software (ImageJ and/or Imaris). The output of the script is a table with the following parameters: Golgi volume in voxels, Golgi volume in um^3, “Golgi-Golgi” distance (averaged distance between all Golgi voxels), Golgi-centrosome distance (averaged distance between each Golgi vowel and the nearest mother centriole), and centrosome-centrosome distance (for cells with duplicated centrosome, the distance between the mother centrioles). The approach can also be applied to analyze distribution of any fluorescently-labeled structure within a cell and its association with the centrosome or any single point within the cell volume.

1. Introduction

The Golgi complex (the Golgi apparatus, or the Golgi) serves as a major hub for intracellular trafficking, allowing for the proper processing of newly synthesized proteins, their sorting, and transport toward proper destinations [1–3]. Due to its vital role in protein trafficking, organization of the Golgi complex has been studied extensively. The main functional unit of the Golgi is formed of several flat membrane cisternae organized in a tight stack. In most vertebrate cells, Golgi stacks are interconnected to form a single organelle (aka Golgi ribbon) in the cell interior [4–6]. Integrity, morphology, and intracellular position of the Golgi complex are considered important factors in its function as a trafficking hub. In addition, Golgi ribbon integrity is critical for the directionality of post-Golgi trafficking and its contribution to cell polarity and migration [7].

The consensus is that the localization and morphology of the Golgi ribbon are, to a large extent, governed by the cytoskeleton. Microtubules play the decisive role in the integrity of the Golgi ribbon: incoming Golgi membrane are collected by minus-end-directed microtubule-dependent molecular motors [8, 9]. As a result, the Golgi is often found in the close proximity of the centrosome, the major microtubule-organizing center (MTOC), where a large number of microtubule minus ends are located. That being said, both microtubule and actin cytoskeletons are needed to maintain functional Golgi organization [10, 11] and to govern highly dynamic changes in the Golgi that are paramount for the proper function of this organelle [12].

An essential Golgi reorganization occurs in proliferating cells, as they progress through the cell cycle. Pioneering work has shown the Golgi complex undergoes a highly regulated process of progressive fragmentation as it nears mitosis [13–15]. Both Golgi fragmentation and its disassociation from the centrosomes are required for the mitotic entry [14–17]. We have recently described additional dramatic morphological changes of the Golgi complex during interphase [12]. The organelle begins as a highly compact structure tightly associated with the centrosomes at G1, to dissociate from the centrosomes and acquire an extended shape along the nuclear equator in S-phase [12]. Recondensation and extension of the Golgi impairs cell polarity and directionality of cell migration [12]. Later, in mitosis, this expended Golgi becomes fully scattered in preparation for the partitioning into the daughter cells [18] where it reassembles into single ribbons. This cell-cycle dependent dynamics is just one example out of a large number of physiological and disease related processes that are associated with or depend on Golgi complex organization.

Thus, evaluation of Golgi morphology in microscopy images and live-cell recordings has an enormous importance for evaluating phenotypes connected with Golgi ribbon integrity, shape, and association with the centrosome. Multiple approaches, including some in our prior work, have been developed to quantitatively describe the Golgi morphology in a cell population. However, it is difficult to find a convenient yet informative metric that provides comprehensive information about this convoluted three-dimensional structure and is easily obtained in an unbiased manner.

Many studies evaluate Golgi integrity by counting cells with “compact” versus “scattered” Golgi [19] which shows a difference between cell populations but has risk of potentially subjective classification and leaves out variability within the population. Other studies count detectable Golgi particles, which does not provide information of distances between those particles [9, 20, 21]. Furthermore, many approaches analyze two-dimensional image projections and lack a three-dimensional perspective [22, 23]. A task of evaluating the association of the Golgi with the centrosome has as additional challenge, since a metric for the special relationship between a differentially scattered mass (the Golgi) and a very small object (centrioles are detected as resolution-limited spots in light microscopy images) is required. In our recent study [12], we introduced an unbiased method to simultaneously measure Golgi scattering and Golgi-centrosome proximity.

In this tutorial, we will present this method in detail. By applying this quantification approach, we can confidently remove any subjectiveness that comes along with qualitative descriptions. Then, results across several studies will become comparable. In our quantification, we evaluate the spread of Golgi by a parameter called Golgi-Golgi distance (GG) (Figure 1). It is calculated as the mean of pairwise distances for all voxels within segmented volume of the Golgi. We evaluate the association of the Golgi with the centrosome by a parameter called Golgi-centrosome distance (GC). It is calculated as the mean distance from each voxel within segmented volume of the Golgi to the mother centriole of the nearest centrosome (Figure 1). Our analysis also determines such informative metrics as the distance between the centrosomes (in S, G2, M cells), and total Golgi volume per cell. The analysis uses Image (NIH), Imaris (Bitplane), and MATLAB (Mathworks) software. The results are exported as an output table demonstrating GG distance, GC distance, centrosome-centrosome distance, and total Golgi volume. Here, we demonstrate sample preparation, image acquisition, and analysis steps, using confocal microscopy images of the Golgi and centrosomes in RPE1 cells as an example.

Figure 1:

Golgi mode quantitative parameters (modified from Supplementary Figure 1 Frye et al 2020). (a) Table summarizes computation of each Golgi and centrosome positioning parameter used for quantification. For Golgi to centrosome distance (GC), the distance from each voxel within segmented volume of the Golgi to both centrosomes was determined, and the mean distance to the nearest centrosome was calculated to describe Golgi association with the proximal centrosome. The spread of Golgi (GG) was calculated as the mean of pairwise distances for all voxels within segmented volume of the Golgi. Centrosome-to-centrosome distance (CC) was calculated by measuring the distance of centrosome foci.

2. Materials

2.1. Cell Culture

1. Stable cell line: Centrin-1-GFP-RPE1 cells were transfected with RFP-TGN (gifted from Enrique Rodriguez-Boulan) via Amaxa (Lonza, Program I-013), then selected with G418 for at least two weeks before cell sorting.

2. Cells are maintained in DMEM/F12 with 10% fetal bovine serum, 100 μM penicillin, and 0.1 mg/mL streptomycin at 37°C in 5% CO₂.

3. T-25 flasks

2.2. Immunostaining

1. 5 μg/mL fibronectin (EMD Millipore, Burlington, MA, USA)

2. 12mm diameter round coverslips #1.5

3. Square dish for humidity chamber

4. Parafilm M Laboratory Film

5. KimWipes

6. 4% PFA stock in CB buffer

7. 20% Triton-X stock in PBS

8. 1X PBS

9. DHS/BSA

10. Primary antibody: rabbit polyclonal antibody against giantin (Abcam, Cambridge, MA, USA)

11. Secondary antibody: Highly cross absorbed Alexa Fluor 647 anti-rabbit (Molecular Probes, Invitrogen, Eugene, OR, USA)

12. DAPI (Fisher Scientific, Waltham, MA, USA)

13. Mounting media (Vectashield Mounting Medium (Vector Labs, Burlingame, CA, USA))

14. Clear nail polish

2.3. Confocal microscope and analysis

1. Microscope: Laser-scanning confocal microscope Nikon A1r based on a Ti-E inverted microscope with SR Apo TIRF 100× NA1.49 oil lens run by NIS Elements C software (Nikon, Tokyo, Japan) with standard A1r laser launch, GaAsP detector, Sola LED light.

2. Software: NIS-Elements (Nikon, Tokyo, Japan), ImageJ (NIH), MATLAB (Mathworks), Imaris (Bitplane).

3. Methods

3.1. Cell maintenance

1. RPE1 cells expressing centrin-GFP and RFP-TGN are cultured in T25 flasks until they reach approximately 80% confluency.

2. One day prior to experiment, place glass coverslips into an empty, sterile 35-mm dish and add one drop of 20 μl fibronectin onto each drop. Incubate for 30 mins at 37° or 60 mins at room temperature.

3. After incubation period, add ~1mL of sterile PBS into dish and gently swirl to rinse coverslips.

4. Remove PBS and seed cells into the 35-mm tissue culture dish containing fibronectin coated coverslips. Split the cells at 1:8 ratio from a confluent dish. Allow cells to attach overnight.

3.2. Immunostaining

1. Aspirate culture media, quickly rinse with PBS once, and replace with fixation media (0.25% Triton-X in 4% PFA/CB). Allow to incubate for 15 mins at room temperature.

2. At lab bench, aspirate fixative and rinse cells with 1x PBS for 5 mins. Repeat 3 times.

3. Prepare humidity chambers for antibody incubation by adhering Parafilm to the bottom half of a square plate. Remove paper backing. Label the bottom of the dish on the outside for the desired number of coverslips (data points). Moisten small KimWipe and adhere to the top lid. It is convenient to prepare two chambers.

4. Add one drop of 20 μl blocking solution (DHS/BSA) to exposed parafilm for each coverslip. Place coverslips face down on solution. Incubate for 1hr at room temperature in closed humidity chamber.

5. In the second humidity chamber, place one drop of 20 μl giantin antibody dilution (1:250) for each coverslip. Do not rinse coverslips after blocking.

6. Gently lift coverslips out of blocking solution, wick away excess blocking solution by touching coverslip edge to a filter paper, and place face down in giantin solution. Close the chamber and seal with parafilm. Incubate 1 hour at room temperature or overnight at 4°C.

7. Rinse coverslips with 1x PBS by flipping them face-up into a well-labeled multi-well plate. Replace PBS 3 times for three 5-minute washes.

8. In the humidity chamber, replace Parafilm and place one drop of 20 μl secondary antibody dilution (DAPI 1:1000, Alexa Flour 647 anti-rabbit 1:500) for each coverslip.

9. Wick away excess PBS from coverslips and place face down in secondary solution. Incubate for 1hr at room temperature in a dark drawer.

10. Repeat step 7 (three washes in PBS).

11. To mount coverslips, place one drop of mounting media (Vectashield) onto a clean glass microscope slide. Gently press coverslips to wick away excess mounting media.

12. After a drying period, use clear nail polish to seal all sides of the coverslips.

3.3. Image Acquisition/ Parameters

We use Nikon A1r microscope run by Nikon Elements NIS to create optical configurations with the following settings. Similar parameters should be easily obtained by your specific confocal microscope whether from Nikon or a different manufacturer. Ultimately, parameters will be different but the goal is to obtain high signal to noise ratio at the highest possible optical resolution in three-dimensions.

1. Objective: 100× (oil immersion), 1.49 NA

2. Scan mode: Galvano unidirectional

3. Scan speed: 0.125

4. Frame size: 512 X 512

5. Averaging: Normal

6. Pinhole: 76.63 μm

7. Zoom: 1x

8. Pixel size: 0.12 μm/px

9. Z-step size: 0.125 μm

10. Laser Power

    1. 405: 1.3%

    2. 488: 0.7 %

    3. 561: 1.5%

    4. 647: 1.5%

11. PMT HV (Gain of the detector)

    1. 405: 120

    2. 488: 60

    3. 561: 50

    4. 647: 90

12. Offset (of the detector)

    1. 405: 0

    2. 488: 0

    3. 561: −20

    4. 647: −4

13. Open Nikon Elements NIS.

14. Enable the “XY Position” tab in the ND Acquisition dialog box. This will allow the user to record multiple XYZ positions that will be imaged in sequence. Ensure “Include Z” is selected.

15. Prepare 100× objective by adding correct immersion oil.

16. Fasten microscope slide onto slide holder.

17. Use LED lamp to focus on sample. Enable perfect focus.

18. Search coverslip and record positions in “XY Position.” To avoid bias while choosing cells, only use the DAPI channel to record multiple positions. Imaging fields should have multiple cells, with nuclei that are well spaced.

19. Switch to confocal mode and create optical configurations based on settings listed above.

20. Enable the “Z Series” tab in the ND Acquisition dialog box. This will allow the user to define Z step-size and range to be imaged.

21. With the image window live mode, use the Golgi channel to set appropriate Z settings using the “Z Series” tab. Designate step size in microns. Set the z- range around center using the symmetric or asymmetric setting. Absolute Z-range is not supported if using PFS.

22. Set the home position. Click “Relative” for use with PFS.

23. Use fine focus to survey the Golgi volume. Make note of the distance in microns above and below the home position which is required to image the entire Golgi volume. Input these distances accordingly.

24. After all settings are finalized, acquire multi-positional multi-dimensional image files by clicking “Run now”. Save to file.

25. See Figure 2 for representative Images

Figure 2:

Representative images of Golgi and centrosome positioning in two distinct Golgi configuration modes. (a-b) Perinuclear region of RPE1 cell. Channels: Centrin1-GFP (yellow), giantin (immunostaining, magenta), Hoechst (cyan). Boxed regions (centrosomes) are enlarged in insets. (a’-b’) Golgi immunostaining is used as input to create masks for mapping each Golgi voxel. (a”, b”) Output of manual thresholding at 5% intensity above the background. Maximum intensity projections of laser scanning confocal stacks. Scale 5μm. (c) Results of calcGGGC.m script for cells depicted in Fig. 1a and 1b.

3.4. Preparing images for analysis

1. Create a folder which will contain all Golgi and centrosome data for a single experimental date.

   1. Naming scheme should follow: expDDMM_X. Where DDMM reflects experimental date and X reflects experimental replicate in a single digit.

   2. This folder will contain .nd2 file created in section 3.3.

   3. This folder will contain all macros and scripts required to obtain final output data.

      1. ImageJ macro: readbinary.imj

      2. Matlab scripts: ImportGolgitxt.m, CENTRfromCSV.m, calcGGGC.m (covered in detail in section 3.7)

2. Create eight subfolders that will contain various elements of data. The folder names should be written exactly as below.

   1. full stack

   2. golgi stacks

   3. golgi masks

   4. golgi_coord

   5. centr coord_csv

   6. centrosome stack

   7. centrosome image sequence (if using Imaris for centrosome detection)

   8. imaris centrosome data (if using Imaris for centrosome detection)

3. Open the .nd2 multidimensional image file acquired in ImageJ.

4. Create a file for each cell in the field.

   1. Use the rectangle tool to draw a box around a single cell (Fig. 1a, 1b).

   2. Select Image→Duplicate.

   3. In the Duplicate dialog box, create title with the naming scheme cellXX.tif. Where XX indicates a two-digit number to indicate cell identity. Ensure “Duplicate hyperstack” is enabled.

5. Save each cell crop as a single, multidimensional image stack in the “full stack” folder.

6. Split channels and save separately into the “golgi stack” and “centrosome stack” subfolders, respectively, continuing to follow the naming scheme cellXX.tif.

3.5. Obtaining Golgi coordinates using ImageJ

1. Open a single Golgi stack created in section 3.4 (Fig 1a’, 1b’).

2. Create binary image by applying thresholding.

   • a.Image→ Adjust→ Threshold
   • b.Within the Thresholding dialog box, enable “Dark background,” “Don’t reset range” options. We utilize the Default threshold method.
   • a.Navigate the z- plane near the middle of the Golgi stack and adjust the min/max sliders until all Golgi signal is included in thresholding area. We set the threshold at ~5% intensity above the background (Fig. 1a”, 1b”) (see Note 2).
   • c.After all positive Golgi is within the threshold, apply settings by selecting “Apply.”
   • d.In the “Convert Stack to Binary” dialog box, ensure the method utilized to create the threshold is selected in the Method drop down box. Select dark background in the Background drop down box.
   • e.Enable “Calculate threshold for each image,” “Black background (of binary masks).”
   • f.Save resulting image stack in the “Golgi masks” folder. Create a subfolder for each mask is made. Each subfolder will only contain the mask for a single cell.

3. Open and run the “readbinary.imj” macro.

   1. A file explorer dialog box will open. Navigate to the Golgi masks folder and select the subfolder of the cell of interest.

   2. A results window will open and automatically begin to populate XYZ coordinates for each positive pixel detected in the image stack. When the macro is finished, save as a.txt file in the “golgi_coord” folder. The naming scheme should follow: cellXXbin.txt. Where XX indicates a two-digit number to indicate cell identity.

3.6. Obtaining Centrosome coordinates using Imaris

• 1Prepare centrosome files for importing into Imaris.
  • a.Open a centrosome channel file created in section 3.4.
  • b.File→Save As→Image Sequence
  • c.In the “Save Image Sequence” dialog box select TIFF file format, keep consistent cellXX naming scheme, and ensure “Use slice labels as file names” is disabled.
  • b.Save into “imaris centrosome data” subfolder. Create a subfolder for each image sequence. Each subfolder will only contain sequence files for a single cell.
• 2Open Imaris software. Import centrosome image sequences to start a new analysis session.

  1. File→Open

  2. Navigate to the folder containing centrosome image sequence created in the previous step.

• 3Add a new surface.
• 4Set correct XYZ settings.

  1. Edit→ Image Properties

• 5Specify which statistics will be exported using the Preferences menu.

  1. Edit→Preferences→Statistics

  2. In the Surfaces dropdown menu, only select Center of Homogenous Mass X, Y, and Z.

• 6Using the Create wizard, adjust thresholding accordingly to define only the mother centriole (see Note 1) (meaning there will only be either one or two detection spots per file) in the image stack. After centrosomes are detected, execute all creation steps.

  1. Results are exported into a Statistics folder containing .csv files for _Center of Homogenous Mass and _Overall. Save the Center of Homogenous Mass.csv into the “centr coord_csv” folder (see Note 3.). Naming scheme should follow cellXXcentr.csv.

3.7. Calculating Golgi-Golgi distance and Golgi-centrosome distance using Matlab

• 7Ensure all required Matlab scripts are saved into the folder which contains all Golgi and centrosome data for a single experimental date, as created in section 3.4.

  1. Required scripts: ImportGolgiTxt.m, CENTRfromCSV.m, calcGGGC.m

• 8Enter ImportGolgiTxt into the command window and execute. This will prompt the user to navigate to the “golgi_coord” folder.

  1. The result will appear in Workspace as a structure. Rename this structure to expDDMM_Xgolgi.

• 9Enter CENTRfromCSV into the command window and execute. This will prompt the user to navigate to the “centrosome coord_csv” folder.

  1. The result will appear in Workspace as a table. Rename this table to expDDMM_Xcentr.

• 10Double click the calcGGGC.m file within in the Current Folder Pane to make adjustments.

  1. Enter the correct experiment name at line 15: expName = ‘expDDMM_X’;

  2. Ensure pixel size (in microns) of XY (line 16) and Z (line 17) is correct to your system.

• 11Enter calcGGGC into the command window and execute. This will compute values based on the items in the workspace.

  1. The final results will appear in the Workspace as a table named “T.”

  2. This table contains distances of Golgi-Golgi distance, Golgi-centrosome distance, Centrosome-Centrosome distance, total Golgi volume, and total number of Golgi voxels (Fig. 2c) (see Note 4).

4. Notes

1. Centrin-1-GFP will detect both centrioles, however the mother centriole will be significantly brighter. Thus, there will be either one bright and one dim foci in the cell (G1) or two bright and two dim foci in the cell (S/G2).

2. If signal to noise is poor or autofluorescence is apparent, adjust thresholding accordingly and/or manually remove artifacts in the cytosolic area.

3. Alternatively, centrosome coordinates can be obtained using ImageJ. In either case, in order to properly import centrosome coordinate data into Matlab at the next step, centrosome coordinate(s) for each cell must be saved in an .csv file where cell A5= X coordinate of centrosome 1, B5= Y coordinate of centrosome 1, C5= Z coordinate of centrosome 1 and A6= X coordinate of centrosome 2, B6= Y coordinate of centrosome 2, C6= Z coordinate of centrosome 2. If the .csv is not configured properly, an indexing error such as “Index exceeds number of array elements,” will likely return.

4. For cells in S/G2 where there are two centrosomes, the Golgi-Centrosome distance is computed as the distance between each Golgi voxel and the nearer centrosome.