Methods for analyzing the role of phospholipase A₂ enzymes in endosome membrane tubule formation

Abstract

Cargo export from mammalian endosomal compartments often involves membrane tubules, into which soluble and membrane-bound cargos are segregated for subsequent intracellular transport. These membrane tubules are highly dynamic and their formation is mediated by a variety of endosome-associated proteins. However, little is known about how these membrane tubules are temporally or spatially regulated, so other tubule-associated proteins are likely to be discovered and analyzed. Therefore, methods to examine the biogenesis and regulation of endosome membrane tubules will prove to be valuable for cell biologists. In this chapter, we describe methods for studying this process using both cell-free, in vitro reconstitution assays, and in vivo image analysis tools.

INTRODUCTION

Endosomes of mammalian cells are pleiotropic organelles containing distinct morphological domains: (1) a spherical vesicular or vacuolar domain with a relatively large luminal space and (2) membrane tubules and budding vesicles with higher surface-to-volume ratios that emanate from the vacuolar domain (Bonifacino & Rojas, 2006; Huotari & Helenius, 2011; Zerial & McBride, 2001). The pioneering studies of Geuze and colleagues demonstrated that endosome tubules, generally 60–80 nm in diameter, function to sort and concentrate membrane receptors away from the vacuolar domain for recycling to other compartments (Geuze, Slot, & Schwartz, 1987; Geuze, Slot, Strous, Lodish, & Schwartz, 1983). Despite 25 years of effort to understand tubule-mediated endocytic trafficking, it is unclear how these endosome tubules export cargoes: do they undergo fission by themselves, and/or serve as platforms for subsequent coated vesicle budding (Huotari & Helenius, 2011) (Figure 1)?

FIGURE 1.

Endocytic organelles and steps utilizing membrane tubules and cytoplasmic phospholipase A₂ (PLA₂) enzymes. The inhibitory symbol (T) indicates membrane tubule-mediated trafficking steps that are inhibited by PLA₂ antagonists. Previous studies have shown that endocytic trafficking of transferrin receptors (TfR), low-density lipoprotein (LDL), and epidermal growth factor receptors (EGFR) via clathrin-dependent endocytosis (CDE) involves PLA₂-mediated membrane tubule formation at several steps. The cytoplasmic PLA₂ PAFAHIb catalyzes the formation of membrane tubules and facilitates export of TfRs from early endosomes (EE) and the endocytic recycling compartment (ERC). Trafficking of EGFRs and LDL from late endosomes/multivesicular bodies (LE/MVB) is inhibited by PLA₂ antagonists. Following clathrin-independent endocytosis (CIE), proteins such as MHC class I, CD55, and CD59 utilize cytoplasmic cPLA₂ε at the CIE recycling compartment (CIE RE) to generatemembrane tubules for recycling back to the plasma membrane. TGN, trans-Golgi network. (See color plate)

Another area of active research has been to elucidate the molecular mechanisms responsible for endosome (and Golgi) membrane tubule formation. For example, studies have characterized the role of the cytoplasmic protein complexes retromer and sorting nexins, which bind to endosome membranes and catalyze the export of membrane cargo by sorting and fission mechanisms (Gallon & Cullen, 2015). Additionally, membrane tubule formation is often facilitated by microtubules and associated motor proteins (Bechler, de Figueiredo, & Brown, 2012). Finally, in the last several years other studies have established a surprising role for cytoplasmic phospholipases in the formation of membrane tubules (Bechler et al., 2012; Ha, Clarke, & Brown, 2012). Work from our laboratory and those of Polishchuk, Luini, and colleagues has identified several phospholipase A₂ (PLA₂) enzymes that catalyze the formation of endosome membrane tubules, which contribute to the sorting and trafficking of various membrane cargoes. These PLA₂ enzymes, and possibly others, function in the trafficking of transferrin (Tf) and epidermal growth factor (EGF) receptors via the clathrin-dependent endocytic pathway (Bechler et al., 2011; Doody, Antosh, & Brown, 2009; de Figueiredo et al., 2001), and major histocompatibility complex class I (MHC-I) proteins via the clathrin-independent endocytic pathway (Capestrano et al., 2014; Figure 1).

Endosome membrane tubules are similar in many ways to those that emanate from the Golgi complex, ER-Golgi-Intermediate Compartment, and trans-Golgi network, which have been shown to mediate several trafficking events (Bechler et al., 2012; Polishchuk, Capestrano, & Polishchuk, 2009). Moreover, these membrane tubules also have a requirement for cytoplasmic PLA₂ enzyme activity (Bechler et al., 2010; de Figueiredo, Drecktrah, Katzenellenbogen, Strang, & Brown, 1998; de Figueiredo, Polizotto, Drecktrah, & Brown, 1999; San Pietro et al., 2009), suggesting a common mechanism for shaping endosome and Golgi membranes into tubules.

In spite of these advances, little is known about how the formation of PLA₂-mediated endosome membrane tubules is spatially or temporally regulated. Therefore, future research of the molecular mechanisms and functional roles of endosome membrane tubules will benefit from multiple experimental approaches. Thus, the goals of this chapter are to: (1) describe both in vitro and in vivo methods for quantitatively examining endosome membrane tubule formation in cultured mammalian cells; and (2) describe methods for investigating the role of PLA₂ enzymes membrane tubule formation. The methods described here are taken from well-established protocols that are routinely used in our lab. We have used these methods to elucidate the roles of cytoplasmic PLA₂ enzymes in the biogenesis and functions of endosome membrane tubules, but most are generally applicable for the study of any protein involved in these processes.

1. IN VITRO ENDOSOME MEMBRANE TUBULE FORMATION

To characterize the role of PLA₂ enzymes in endosome membrane tubule biogenesis, and to link these biochemical mechanisms to the functions of membrane tubules, we have established a cell-free, in vitro reconstitution assay that faithfully recapitulates the formation of endosome membrane tubules (de Figueiredo et al., 2001). This assay employs negative stain transmission electron microscopy (TEM) to visualize membrane tubule formation and requires the preparation of various reagents as described in the three sections below.

1.1 PREPARATION OF COLLOIDAL GOLD PARTICLES

Endosomes containing 12 nm colloidal gold particles are used in the reconstitution assay. The colloidal gold is delivered to endosomes by fluid-phase endocytosis, and the endosomes are isolated by sucrose density centrifugation. The enriched organelles are visualized by negative stain TEM to unambiguously identify colloidal gold-loaded endosomes. Detailed steps for making 12 nm colloidal gold stabilized with bovine serum albumin (BSA) (gold-BSA particles) have been published elsewhere (de Figueiredo et al., 2001).

1.2 PREPARATION OF COLLOIDAL GOLD-LOADED ENDOSOMES

Many types of cultured mammalian cells can be used to isolate endosomes, but the cells must be adherent and have a reasonably good rate of fluid-phase endocytosis. We have successfully employed rat clone 9 hepatocytes, RAW 264.7 macrophages, and HeLa cells. The following procedure will deliver gold–BSA particles to both early and late endosomes and the endocytic recycling compartment, but modifications to the labeling protocol can be made to label specific compartments, e.g., the amount of time for uptake.

1.2.1 Reagents

• ~ 75 mL of colloidal gold–BSA solution

• Homogenization buffer

  • 250 mM sucrose, 10 mM Tris–HCl, 1 mM EDTA, pH 7.4
• Sucrose step gradient in Beckman SW28.1 centrifuge tubes containing the following concentrations of sucrose at each step (3.4 mL/step): 0.5, 1.0, 1.5, and 2 M sucrose.

  • Each sucrose solution is buffered with 1 mM Tris, 1 mM EDTA, pH 7.4. Additionally, the 0.5 M sucrose step should contain 500 mM KCl, which serves to wash endosomes as they move through the gradient and lower the background level of in vitro tubulation.
  • When preparing the gradient, be sure to account for the volume occupied by the postnuclear supernatant (PNS). For example, if 4 mL of PNS is going to be loaded, then a gradient of ~13 mL total volume is to be prepared (the above instructions would yield a 12.8 mL gradient).
  • Prepare the gradient immediately after the collection of a PNS (see below).

1.2.2 Equipment

• Ultracentrifuge with appropriate rotor and tubes, e.g., Beckman SW28.1 swinging bucket rotor

• Balch/Rothman ball-bearing homogenizer (Balch& Rothman, 1985; de Figueiredo et al., 2001).

1.2.3 Procedure

1. Monolayers of cells are grown to confluence in four to seven 220-mm² plastic tissue culture plates. In all subsequent steps, ensure that the cells do not dry out.

2. To label endosomes with gold–BSA particles by fluid-phase uptake, briefly rinse the cells once with minimum essential medium (MEM) at 37 °C. Remove the MEM.

3. Add to each dish gold–BSA/MEM solution (75 mL/dish) for 15 min (clone 9; HeLa cells) or 10 min (RAW 264.7) at 37 °C.

   1. Note: the gold–BSA particle/MEM solution can be used multiple times in 1 day to reduce the amount of solution needed for an endosome preparation: simply collect the labeling media from one dish and add to the next.

4. Terminate endocytosis by rapidly pouring off the gold–BSA particle/MEM solution and washing the cells three times with ice-cold MEM. Pipetting is not necessary: just pour solutions on and off.

5. Cover cells with ~25 mL of ice-cold MEM and put dishes on ice.

   1. All subsequent steps in the preparation procedure are carried out at 4 °C and/or on ice.

6. Harvest the cells in the MEM with a rubber policeman or cell scraper. Pellet the cells using a tabletop clinical centrifuge (e.g., IEC Centra CL2) at 3000 rpm for 5 min. Aspirate and discard the supernatant. From four to seven 220-mm² dishes with 80% confluent cells, ~1–2 mL of packed cells should be obtained. A minimum of 1 mL packed cells will be needed for subsequent homogenization. The cells should be slightly reddish in color, indicative of endocytosed BSA–gold particles.

7. Resuspend the cells in the homogenization buffer with a Pasteur pipette.

   1. The volume of cells (in the pellet) should be ~20% of the final total volume (cell pellet + homogenization buffer). Too few or too many cells make homogenization problematic. Vigorously, resuspend cells with a Pasteur pipette to break up any clumps.

8. Put the cells in a 15 mL conical centrifuge tube and centrifuge as in step 6.

9. Aspirate and discard the supernatant, and resuspend cells as in step 7.

10. Load the cells into an appropriate sized syringe (3, 5, or 10 mL) for homogenization with a Balch–Rothman homogenizer (Balch & Rothman, 1985; de Figueiredo et al., 2001). Our homogenizer, constructed exactly as described by Balch and Rothman (cited above), used a 0.2501-inch diameter ball bearing. Break open cells with 2–4 passages, and monitor cell breakage by phase contrast microscopy: ~70% of the cells should be broken open.

11. Deposit the homogenate in a fresh 15 mL conical centrifuge tube and centrifuge as in step 6.

12. Harvest the PNS with a Pasteur pipette, being careful not to disrupt the nuclear pellet. Put the PNS on ice. Both the pellet and PNS should be slightly reddish due to the gold–BSA particles. Check the PNS by phase contrast microscopy to ensure that nuclei, unbroken cells, etc., have been removed. If not, repeat the centrifugation step and harvest the PNS.

13. To obtain more organelles, vigorously resuspend the nuclear pellet in a small volume of homogenization buffer with a Pasteur pipette (e.g., if the first PNS is about 3–4 mL, suspend nuclear pellet in ~0.5–1 mL). Repeat steps 10 and 11, and combine the two PNSs.

14. Layer the PNS (~4 mL) on top of the sucrose step gradient. Centrifuge the gradient tube at 113,000 × g for 2 h in a Beckman SW28.1 swinging bucket rotor with no brake.

    1. Balance the gradient tubes along with the buckets.

15. Material at the 1.0/1.5 M sucrose interface should contain the highest concentration of gold–BSA particle-loaded endosomes, and will often be visibly pink in color.

16. Collect the material at the 1.0/1.5 M sucrose interface with a Pasteur pipette, and either use immediately or slowly freeze at −20 °C and store at −80 °C. Tubulation-competent endosomes can be stored for ~2 days at −80 °C.

1.3 PREPARATION OF BOVINE BRAIN CYTOSOL

Our early studies on endosome and Golgi membrane tubule formation revealed that bovine brain cytoplasm (BBC) is a convenient source of “tubulation factors,” i.e., cytoplasmic PLA₂ enzymes (Banta, Polizotto, Wood, de Figueiredo, & Brown, 1995; Cluett, Wood, Banta, & Brown, 1993). Thus, BBC serves as a useful positive control and likely source of other proteins that regulate the process of membrane tubule formation. The methods for preparing BBC are described in detail elsewhere (Banta et al., 1995).

1.4 CELL-FREE, IN VITRO FORMATION OF ENDOSOME MEMBRANE TUBULES

Membrane tubules formed from isolated endosomes in vitro are visualized by a whole mount TEM-negative stain assay, similar to that used to document the tubulation of Golgi membranes (Banta et al., 1995; de Figueiredo et al., 1998).

1.4.1 Reagents

• BBC

• 10 mM ATP stock (Store at −20 °C)

  • store in aliquots and thaw only once

• 100 mM MgCl₂ stock (store at −20 °C)

• 10 mM HEPES, pH 7.4, stock (store at 4 °C)

  Just before starting the assay, prepare the following:

• ATP/Mg²⁺ solution and place on ice:

  • 80 µL dialysis buffer, 10 µL ATP, 10 µL MgCl₂
• Tubulation reaction mix

  • 20 µL/sample is needed. Make enough to have excess.
  • For solutions with a high protein concentration, e.g., BBC, use the following proportions for 100 µL (=5 reactions), and add in the order listed:

    • -
      86µL dialysis buffer, 12 µL BBC, 1 µL ATP/Mg stock, 1 µL HEPES stock buffer
    • -
      Equilibrate tubulation reaction mix to 37 °C (~5–10 min).
• 2% phosphotungstic acid (PTA), pH 7.4

  • make with dH₂O, pH with NaOH, store at 4 °C.

1.4.2 Equipment

• Self-closing fine-tipped forceps

• 200 hexagonal mesh formvar- and carbon-coated copper grids (EM Sciences)

  • Formvar-/carbon-coated grids can be homemade if you have a vacuum evaporator, which would substantially reduce the cost/grid.

1.4.3 Getting ready

1. Thaw out all solutions and reagents quickly and put on ice.

2. Place the appropriate number of grids in self-closing fine-tipped forceps in a dust-free area.

   1. We line up all the forceps in a large, square cell culture dish with Whatman filter paper on the bottom (see Figure 2(A)).
   2. Grab grids carefully with forceps, as they easily bend and the coat can tear easily.
   3. Try to only hold by the outer edge of the grids with the forceps.
3. Place forceps down on white paper and line them up with coated side up.

   1. Use the top of a pipette tip box to cover the ends of the grids to prevent dust contamination.

4. Label the appropriate number of 0.5 µL microfuge tubes and place in a rack.

FIGURE 2.

Setup for in vitro endosome membrane tubulation reactions and negative staining with multiple EM grids. (A) Multiple self-closing forceps can be organized to facilitate rapid application of tubulation reactions to EM grids. (B) Side view of a single self-closing forceps with sample on an EM grid. (See color plate)

1.4.4 The reaction

1. For each reaction, place 20 µL of the endosome-enriched fraction (obtained from Section 1.2) in a tube and equilibrate to 37 °C in a water bath (~5 min).

2. Initiate the reaction by slowly adding 20 µL of the tubulation reaction mix to the endosomes.

3. Incubate at 37 °C for 0–30 min.

   Note: Steps 4–9 are done at room temperature.

4. Spot a 10 µL drop of the endosome/reaction mixture onto the EM grid (Figure 2(B)), and let sit for 15 min to allow endosomes to adhere to the grids.

5. Stop the tubulation reaction by carefully adding 10 µL of PTA directly to the drop on the grid, being careful to not disturb the drop. Add two more 10 µL aliquots of PTA. All three aliquots should be added within 5 s.

6. Wick away the liquid from each grid by gently touching the side with a damp Kimwipe.

7. Stain the grids by adding 10 µL aliquots of PTA two more times for 5 s each time, wicking liquid off after each application as in step 6.

8. Let the grids air dry for 30–60 min.

9. Store grids in a grid holder in the dark until observation.

   Other tips: The PTA stain does not adhere or work very well unless there is a minimum amount of protein on each grid. The lowest final protein concentration we have successfully used is ~0.25 mg/mL. If titrating BBC, or examining purified protein preparations, BSA can be used as an inert carrier (Banta et al., 1995).

10. Visualize the endosomes by TEM, an example of which is shown in Figure 3. The extent of endosome tubule formation can be quantified in several ways. An easy first approximation is to score the gold-loaded endosome population for the degree of membrane tubulation, ranging from those that have at least one membrane tubule to endosomes that are completely tubular. A membrane tubule is defined as a membrane extension that is continuous with the lipid bilayer of the endosome, 50–70 nm in diameter, and at least twice as long as it is wide. For each experimental condition, we suggest that at least 150 endosomes are counted from triplicate samples. A second more quantitative method, the Sholl analysis, can be modified to measure tubule number and length, as described below for in vivo endosome membrane tubule formation.

FIGURE 3.

Electron micrographs of endosome membrane tubules. (A–C) Endosomes in vivo labeled by uptake of horseradish peroxidase, and visualized by DAB cytochemistry and thin section electron microscopy (EM). (D–H) Isolated endosomes containing 12 nm colloidal gold particles (arrows) visualized by negative-stain EM. (D, E) Endosomes incubated under control conditions exhibit few membrane tubules. (F–H) Endosomes incubated with bovine brain cytosol (BBC) form varying numbers and lengths of 60–80 nm membrane tubules. Bars = 0.5 µm. This research was originally published in Journal of Biological Chemistry, de Figueiredo et al. (2001) © the American Society for Biochemistry and Molecular Biology.

2. IN VIVO ENDOSOME MEMBRANE TUBULE FORMATION

Mammalian endosome compartments exhibit varying degrees of membrane tubules, which can be enhanced or inhibited in vivo by controlling the expression of key regulatory proteins such as BAR domain-containing sorting nexins (Gallon & Cullen, 2015) or cytoplasmic PLA₂ enzymes like PAFAHIb2 and Ib3 and cPLA2ε (Bechler et al., 2011; Capestrano et al., 2014). Additionally, the formation of endosome membrane tubules can be stimulated with small molecules such as brefeldin A (BFA) or inhibited by enzyme antagonists (Bechler et al., 2011; Doody et al., 2009; de Figueiredo et al., 2001). Indeed, the use of BFA led to the identification of PLA₂ enzymes in membrane tubule formation (de Figueiredo et al., 2001, 1998). The stimulation or inhibition of these membrane tubules can have a profound impact on endosome trafficking, so methods to quantify tubule formation are needed to mechanistically link membrane tubules to trafficking events. In this section, we will describe methods to label and examine endosome membrane tubule formation using the trafficking of Tf as a model (de Figueiredo et al., 2001).

2.1 LABELING TUBULAR ENDOCYTIC COMPARTMENTS BY UPTAKE OF FLUORESCENT TF

2.1.1 Reagents and supplies

• Alexa Fluor 488-Tf

• MEM

• MEM with 10 mM HEPES

• MEM with 10 mM HEPES supplemented with 0.2% (w/v) BSA

• MEM containing 10% fetal bovine serum (FBS)

• BFA

• PBS, pH 7.4

• Microscope coverslips (sterilized by autoclaving)

  • 12 mm diameter; 1 oz (e.g., Fisherbrand 12-545-82 12CIR-1D)

2.1.2 Procedure

2.1.2.1 BFA stimulation of endosome membrane tubules formation

This procedure was developed to visualize tubular endosomal compartments formed in the presence of BFA in HeLa cells. This procedure can be used with other cell types, but specific labeling and treatment conditions might vary.

1. Plate cells onto 35- or 60-mm cell culture dishes containing coverslips and allow to grow for 48–96 h to reach 70–80% confluency.

   1. 35-mm dishes hold 5-, 12-mm coverslips; 60-mm dishes hold 12-, 12 mm coverslips

2. Rinse the cells three times with 10 mM HEPES-MEM and incubate in the same solution for 1 h at 37 °C, with washing every 15 min to ensure depletion of endogenous Tf from the cells.

3. Incubate cells with 20 µg/mL Alexa Fluor 488-Tf in 10 mM HEPES-MEM supplemented with 0.2% (w/v) BSA (pH 7.2–7.4) for 45 min at 37 °C to label early and recycling endocytic compartments.

4. Replace the media with MEM containing 1 µg/mL BFA and 20 µg/mL Alexa Fluor 488-Tf and incubate the cells at 37 °C for up to 20 min, removing coverslips for time points as necessary.

5. Rinse the coverslips quickly in MEM.

6. Fix the cells in 3.7% formalin in PBS for 10–15 min at room temperature.

   1. Note: cells can be processed for immunofluorescence staining after this step.

7. Wash the coverslips three times for 5 min each in PBS at room temperature.

8. Mount the coverslips on microscope slides using an antiphotobleaching media, e.g., Vectashield.

9. Store slides at −20 °C until visualization by fluorescence microscopy.

2.1.2.2 Examining Tf uptake into endocytic compartments

1. Repeat steps 1 and 2 from Section 2.1.2.1 above.

2. Incubate the cells with 20 µg/mL of Alexa Fluor 488-Tfn in 10 mM HEPES-MEM supplemented with 0.2% (w/v) BSA (pH 7.2–7.4) for up to 30 min at 37 °C, removing coverslips for time points as necessary.

3. Quickly rinse the cells with ice-cold 10 mM HEPES-MEM on ice to halt endocytosis.

4. Remove surface-bound, non-endocytosed Tfn by washing cells with ice-cold 0.2 N acetic acid-PBS (pH 3.5) for 30 s.

5. Rapidly rinse the cells with ice-cold PBS (pH 7.4) to neutralize the pH.

6. Fix and process cells as in steps 6–9 in Section 2.1.2.1 above.

2.1.2.3 Examining Tf recycling through endocytic compartments

1. Repeat steps 1 and 2 from Section 2.1.2.1 above.

2. Incubate the cells with 40 µg/mL Alexa Fluor 488-Tfn in MEM for 45 min at 37 °C to saturate the Tf receptor population.

3. Quickly rinse the cells with ice-cold MEM to halt endocytosis and remove unbound Tf ligand.

4. Rewarm the cells by replacing ice-cold MEM with 37 °C MEM-10% FBS for up to 60 min, removing coverslips for time points as necessary.

5. Fix and process cells as in steps 6–9 from Section 2.1.2.1 above.

6. Total cellular fluorescence can be determined using standard image analysis tools.

2.2 CONFOCAL MICROSCOPY AND QUANTITATIVE IMAGE ANALYSIS

Visualization and analysis of endosome membrane tubules can only be rigorously achieved via confocal microscopy. The Sholl analysis protocol described below was applied to images obtained by a Perkin Elmer UltraVIEW live cell imaging confocal scanner unit attached to a Nikon Eclipse TE2000-U inverted microscope with a Nikon Plan-Apo 60× or 100× numerical aperture 1.4 oil immersion objective (Nikon Corporation, Japan), and a Hamamatsu 1394 ORCA-ER chargecoupled device camera (Hamamatsu Photonics, Japan). The number of confocal Zslices that should be obtained for later analysis depends on several factors, e.g., type of label, thickness of cells, etc. To visualize endosome membrane tubules, we typically obtain 10–12 confocal slices of 0.4–0.5 µm depth each. For any one set of experiments, exactly the same number of Z-slices should be used for each image.

Several automated image analysis algorithms have been developed to quantify changes in organelle morphology. However, quantifying highly tubular organelles is particularly challenging (see Figure 4) and requires either proprietary or special software packages. To overcome these limitations, and to provide a general software tool for quantifying the extent of endosome tubule formation, we have adapted the automated Sholl analysis, which is freely available in NIH ImageJ or Fiji, to provide a semi-quantitative measure of changes in membrane tubule number and length (Abramoff, Magalhaes, & Ram, 2004; Ferreira et al., 2014; Schindelin et al., 2012; Sholl, 1953). Borrowed from the neuroscience field, this technique has been successfully adapted by our lab for analysis of BFA-stimulated Golgi membrane tubulation (Schmidt & Brown, 2009). Images were processed in Fiji using a custom multistep macro script (see below) that enhanced and filtered the images for tubular features as a means to “trace” the membrane tubules.

FIGURE 4.

Semi-automated endosome membrane tubule segmentation and subsequent Sholl analysis. Confocal z-projections were processed by a custom macro script that segmented the membrane tubules and yielded binary images ready for Sholl analysis. (A) An example original confocal z-projection micrograph. Both images are the same, the right image was contrast enhanced to visualize the membrane tubules. (B) The original unenhanced image is processed by the “Tubeness” plugin in Fiji to enhance the tubular features of the image. The image is then segmented using the Yen thresholding algorithm, which essentially traces all elements in the image. (C) The objects from the resulting binary image in (B) are processed by the “Analyze Particles” feature in Fiji. Tubular objects with a circularity between 0 and 0.4 are transferred to a newbinary image resulting in a binary reproduction of themembrane tubules seen in the original image. (D) and (E) The filtered binary membrane tubule image can then processed with the Sholl analysis plugin. The Sholl plugin counts the number of intersections at specified step intervals (indicated by the different colors) along the distance from the center to the edge of the cell that is denoted by the white line. Given that endosomal tubules do not originate froma single point as the Golgi membrane tubules do, the Sholl analysis indicates the number of tubules found from the center to the edge of the cell. (See color plate)

2.2.1 Procedure

The following section outlines a two-step procedure for processing and analyzing confocal micrographs of endosome membrane tubules. A visual of the workflow is shown in Figure 4. The first step is to segment the membrane tubules in the micrograph, which is challenging, as tubules are often amorphous in shape. One method to achieve reliable membrane tubule segmentation is to painstakingly hand trace the membrane tubules as our lab has done previously (Schmidt & Brown, 2009). However, this is time-consuming and not suitable for high-throughput image analysis. To address this issue, we have devised a membrane tubule segmentation protocol that reliably segments tubular elements in a micrograph utilizing macro scripts that are built into Fiji (Schindelin et al., 2012). To outline briefly, the tubular elements in the image are first enhanced using the “Tubeness” plugin (Sato et al., 1998; Schindelin et al., 2012), then segmented using the Yen thresholding algorithm (Yen, Chang, & Chang, 1995; Sezgin & Sankur, 2004), and finally filtered to eliminate spherical endosomes. The resulting binary image can then be analyzed using the Sholl analysis plugin built into Fiji (Ferreira et al., 2014). Section 2.2.1.1 provides a step-by-step guide, while Section 2.2.1.2 provides a custom multistep macro that allows semi-automation of the membrane tubule segmentation process. Note that it is important to preserve the original bit space of the image (8 or 16 bit) to avoid problems during segmentation. Saving micrographs as tif files is highly recommended as it allows the bit space and original pixel gray level information to be preserved. This procedure assumes the individual has familiarity with Fiji.

2.2.1.1 Step-by-step endosome membrane tubule “tracing” and Sholl analysis

2.2.1.1.1 Endosome membrane tubule “tracing”

1. Open the image of interest in Fiji.

2. Set the image scale through the Analyze menu: Analyze > Set Scale… The conversion factor is specific for every microscope and objective setup and can be obtained from the manufacturer or measured with the aid of a calibration slide. Selection of the “global calibration” option will automatically apply this scale conversion to every image opened in that particular session of Fiji.

3. Set the foreground color to white and the background color to black through the Edit menu: Edit > Options > Colors…

4. Run theTubeness plugin fromthePluginmenu: Plugins > Analyze > Tubeness… Set the desired sigma factor and select the “use calibration information” option in the dialog box. We found for our studies that the default sigma value was most appropriate. However, this value should be empirically determined for every experiment.

5. Threshold the new “tubed” image through the Image menu: Image > Adjust > Threshold… Select the “Yen” thresholding algorithm, and the “dark background” and “stack histogram” options. The upper and lower limits of thresholding can be set with the sliders in the dialog box or can be typed in to the ten-thousandths place. Note that these values should be applied to every image and need to be empirically determined for each unique experiment. Another thresholding algorithm can be used if necessary, but we found of the ones provided in Fiji, that the Yen algorithm worked best.

6. Convert the thresholded image to an 8-bit segmented binary image through the Process menu: Process > Binary > Make Binary… or Process > Binary > Make Mask… It is strongly recommended that the image be saved at this point.

7. On the segmented image, filter the image objects for linear or tubular elements using Analyze Particles through the Analyze menu: Analyze > Analyze Particles… Enter the appropriate circularity cutoff value and select the “add to manager” option. Circularity is a measure of an object’s linearity, with 0 representing a straight line and 1 representing a perfect circle. This value should be empirically determined and used for every image. For our purposes, we found a cutoff of 0.4 worked well. The image objects that meet the circularity requirements can be viewed in the ROI manager window.

8. Create a new 8-bit image with a black background that matches the pixel dimensions of the original micrograph through the File menu: File > New > Image…

9. Transfer the filtered tubular elements in the ROI manager to the new image. Select the new blank image. Go to the ROI manager window and click on the “show all” option to make the objects appear on the blank image in outline form. In the ROI manager window click on the “more” button, and select the “fill” command to fill in the object outlines with white foreground color.

10. Save the new filtered 8-bit binary image of the tubules and tubular endosomes and proceed with the Sholl analysis outlined below.

11. Note that if the membrane tubules in the micrograph make a closed loop, the loop will be filled in at step 9. One must decide if the resulting filled in space is problematic or not. If so, it can be fixed on a per image basis.

    1. Open the binary image obtained at the end of step 6.
    2. Carefully examine the image or cells of interest for closed loops. It may help to compare the binary image to the problematic filtered tubule image to quickly find the filled spaces.
    3. Set foreground color to black (see step 3) and use the pencil tool with a 1–3 pixel width setting to draw a line through the loop. This will break the loop and prevent the filled spaces from appearing at the end.
    4. Save the image and proceed with the rest of the image processing/analysis.

2.2.1.1.2 Sholl analysis

1. Open the binary image obtained from step 10 in the previous section. Ensure that the image scale is set correctly. Using the straight-line tool, draw a line on the binary image from the center of the cell to just past the outermost object in that cell.

2. Run the Sholl analysis plugin through the Analyze menu: Analyze > Sholl Analysis…In the dialog box, set the starting radius and desired radius step size. The ending radius should already be set, as it is based on the length of the line drawn in step 1. The other options can be set based upon experimental needs. All options should be empirically determined. For our purposes, we utilized a linear Sholl method with a start radius of 0 and a step size of 0.5 µm without curve fitting or normalizations.

3. Of the several new dialog boxes that open, one is a table with the radius step size and the number of intersections measured at each radius. Transfer these data to spreadsheets and save for subsequent analysis. To make comparisons between treatment groups, it may be helpful to compare areas under the curve.

4. If there are multiple cells within the image that are too close together (such that the line drawn in step 1 overlaps another cell), each cell needs to be singled out. We have found the following to be the easiest way to achieve this:

   1. Outline each cell in the image using either the free hand or polygon selection tool, and save the cell outlines to the ROI manager.
   2. Select one cell from the ROI manager list to cause the cell outline to appear on the image. Delete the rest of the image through the Editmenu: Edit > Clear Outside.
   3. Run the Sholl analysis on that cell and transfer the data to a spreadsheet.
   4. Without closing the image, revert the image back to its original state through the File menu: File > Revert.
   5. Repeat steps 4b–4d until all cells in the image have been measured.

Please note that when interpreting the resulting Sholl data, the nature of the membrane tubules being analyzed has to be taken into account. Endosome membrane tubules are often short and originate from not only the cell center, but also from the cell periphery. Therefore, one cannot interpret the Sholl analysis results in the same manner as an analysis performed on Golgi membrane tubules, as an increase in object intersections in the cell periphery does not correlate with continuous, longer endosome membrane tubules. Instead, a Sholl analysis performed on endosome membrane tubules would indicate a change in the quantity of membrane tubules at a given distance from the cell center as demonstrated in Figure 4.

2.2.1.2 A Fiji macro script for semiautomated endososme membrane tubule segmentation

The following macro script was written to semiautomate the membrane tubule segmentation process, rendering images ready for subsequent Sholl analysis. All images to be processed must be placed into one folder. Parameters used for the scripted plugins were empirically determined and will need to be customized for each experiment. The following script can be copied into a plain text file and placed into the macros folder of Fiji. To run the script, install it from the Plugins > Macros menu. Note that prior to running the script, the x–y scale has to be set.

// "Tubeness Processor"
//
// This macro batch processes all images in a folder with the tubeness
// plugin, resulting in a binary image that can be used for the Sholl
// Analysis.

d = getDirectory("Choose a Directory ");

list = getFileList(d);
setBatchMode(true) ;
for (i=0; i<list.length; i++) {
  fullName = d + list[i];
  if (endsWith(fullName, ".tif")) {
     open(fullName);
     t = getTitle();
     run("Tubeness", "sigma=0.0409 use");
     selectWindow("tubeness of "+t);
     setAutoThreshold("Yen dark stack");
     setThreshold(0.0080, 0.3000);
     setOption("BlackBackground", true);
     run("Convert to Mask");
     dotIndex = lengthOf(fullName) - 4;
     n = substring(fullName,0,dotIndex) + ".binary.tiff";
     saveAs("Tiff", n);
     close();
     close();
  }
}
list = getFileList(d);
for (i=0; i<list.length; i++) {
   fullName = d + list[i];
           if (endsWith(fullName, ".binary.tiff")) {
              open(fullName);
              t = getTitle();
              run("Analyze Particles.", "circularity=0.00–0.40 add");
              newImage("Untitled", "8-bit black", 1344, 1024, 1);
              roiManager("Fill");
              roiManager("Show None");
              dotIndex = lengthOf(fullName) - 4;
              n = substring(fullName,0,dotIndex) + ".filtered.tiff";
              saveAs("Tiff", n);
              roiManager("Deselect");
              roiManager("Delete");
              close();
              close();
          }
       }

3. PLA₂ ENZYMES IN ENDOSOME MEMBRANE TUBULE FORMATION

The requirement for cytoplasmic PLA₂ enzymes in endosome and Golgi membrane tubule formation has been demonstrated by modulating protein expression and through the use of enzyme inhibitors. Application of RNAi methods has resulted in the identification of specific PLA₂ enzymes involved in organelle membrane tubule formation (Figure 1). These enzymes include platelet-activating factor (PAF) acetylhydrolase 1B2 (α1) and 1B3 (α2 subunit) (Bechler et al., 2010, 2011), cPLA₂α (San Pietro et al., 2009), cPLA₂ε (Capestrano et al., 2014), and PLA2G6 (Ben-Tekaya, Kahn, & Hauri, 2010). RNAi and genome editing are essential tools in cell biology, but they are not without drawbacks, e.g., the length of time needed to reduce protein levels and to observe phenotypic consequences, compensatory upregulation of other PLA₂ enzymes, secondary or off-target effects. An alternative approach to overcome these limitations involves the use of membrane permeable enzyme inhibitors. Inhibitors are fast-acting thus decreasing the chances of long-term secondary effects. Additionally, they can be used to screen any membrane-trafficking step for a possible role of PLA₂ enzymes, both in vivo and in vitro, and reversible inhibitors can be used to rapidly increase PLA₂ activity, i.e., during washout of treatment. Therefore, this section will describe the use of PLA₂ inhibitors in vivo and in vitro.

PLA₂ enzymes constitute a large and diverse class of proteins, which have been classified into 15 groups (Burke & Dennis, 2009). They are commonly clustered into four categories: secretory PLA₂, cytoplasmic PLA₂, calcium-independent PLA₂, and PAF acetyl hydrolase/oxidized lipid lipoprotein associated (Lp)PLA₂s. Several of these enzymes are important mediators of inflammation, thus many PLA₂ inhibitors have been developed, providing an extensive tool kit for cell biologists. The relevant PLA₂ enzymes for endocytic trafficking are cytoplasmic, so those will be the focus of this section.

3.1 PLA₂ ANTAGONISTS

Table 1 lists a selection of PLA₂ antagonists (a vast literature exists on this subject) (Murakami et al., 2011). The antagonists listed here have been shown to inhibit the formation of endosome and/or Golgi membrane tubules in vivo and in vitro, displaying very similar IC₅₀s (Doody et al., 2009; de Figueiredo et al., 2001, 1998, 1999). Although not specific for any particular enzyme, these inhibitors can be used as diagnostic tools to narrow down specific targets. For example, bromoenol lactone (BEL) is ~1000-fold more specific for Ca²⁺-independent PLA₂ versus Ca²⁺-dependent enzymes (Ackermann, Conde-Frieboes, & Dennis, 1995; Hazen, Zupan, Weiss, Getman, & Gross, 1991). Also, there are both irreversible suicide substrates (BEL) and rapidly reversible competitive inhibitors (ONO-RS-082). A final word of caution: be wary of claims that any inhibitor is highly specific. To paraphrase the late Efraim Racker, “Publish quickly with any new inhibitor because its specificity decreases rapidly with time.”

Table 1.

PLA₂ Inhibitors^*

AACOCF3 Arachidonyltrifluoromethyl ketone
Trifluoromethyl ketone derivative of arachidonic acid Substrate analog Reversible and selective for cytoplasmic PLA2 (IC50 = 15 µM) IC50 = 10 µM (in vitro tubulation) Stock solution in DMSO or ETOH
ACA: N-(p-amylcinnamoyl) anthranilic acid
Structural analog of ONO-RS-082 but less potent Also inhibits cPLA2 A moderate leukotriene agonist inhibitor IC50 = 8 µM (in vitro tubulation) Stock solution 40 mM in ETOH
ARA: Aristolochic acid (8-methoxy-6-nitrophenanthro-[3,4-d]-dioxole-5- carboxylic acid)
Inhibits both secretory and cytoplasmic PLA2 (IC50 = 40 µM) IC50 = 8 µM (in vitro tubulation) Stock solution in water
BEL: Bromoenol Lactone (E-6-(bromoethylene)tetrahydro-3-naphthalenyl)-2H- pyran-2-one
Irreversible suicide substrate (IC50 ~1 µM) 1000-fold more specific for cytoplasmic iPLA2 versus cPLA2 IC50 = 5 µM (in vitro tubulation) Stock solution in DMSO
BPB: Bromophenacyl bromide
Nonspecific, irreversible active site modifier of cPLA2 and sPLA2 IC50 = 26 µM (in vitro tubulation)
DEDA: 7,7-Dimethlyeicosadienoic acid
Substrate analog that inhibits both cPLA2 and sPLA2 (IC50 ~15 µM) Also weak 5′-lipoxygenase inhibitor IC50 = 25 µM (in vitro tubulation) Stock solution in ETOH
ITD: Isotetrandrine (bisbenzyltetrahydroisoquinoline)
Inhibits G-protein activation of cytoplasmic PLA2 activity (not a direct enzyme inhibitor) In vitro inhibitory properties unknown IC50 ~10 µM (BFA stimulated tubulation in vivo) Stock solution 25 mM in DMSO
ONO: ONO-RS-082 (2-(p-Amylcinnamoyl)amino-4-chlorobenzoic acid)
Reversible cPLA2 inhibitor (IC50 ~10 µM) IC50 = 6 µM (in vitro tubulation) Soluble in DMSO (up to 25 mg/mL) or ETOH (10 mg/mL)
PACOCF3 Palmityltrifluoromethyl ketone
Substrate analog; trifluoromethyl ketone derivative of palmitic acid Reversible and selective for iPLA2 (IC50 ~4 µM) IC50 = 8 µM (in vitro tubulation) Stock solution in DMSO or ETOH (warm)

^*Where indicated, the IC₅₀ measurements were determined from in vitro and in vivo membrane tubule formation assays of endosomes and/or Golgi complexes (de Figueiredo et al., 1999, 2001). The other IC₅₀ measurements were of enzymatic activities from Balsinde, Balboa, Insel, and Dennis (1999), Ackermann and Dennis (1995), Mukherjee, Miele, and Pattabiraman (1994).

3.1.1 Treatment of living cells

All of the inhibitors in Table 1 are membrane permeable and can be used on live cells in the membrane tubule and trafficking assays described in Section 2.1. Note that many are toxic at high concentrations or after long treatment periods. Also, the effective concentration of most inhibitors is greatly reduced in the presence of serum, i.e., FBS, likely due to tight hydrophobic interactions with albumin. Therefore, all treatments should be done in MEM alone.

1. Prepare ~1000 × concentrated stock solutions in the appropriate organic solvent. Store in aliquots at −20 to 80 °C.

2. Just before use, dilute the stock inhibitor into MEM (warmed to 37 °C if appropriate).

3. Quickly rinse the cells three times with MEM.

4. Add MEM + inhibitor or MEM + solvent control and incubate at 37 °C in a CO₂ incubator.

   1. Many of the antagonists are very fast-acting and achieve a high level of inhibition within 5–10 min.

5. Antagonists must generally be present in any subsequent treatments and incubations.

3.1.2 Use of reversible inhibitors

Reversible inhibitors can be used to great advantage in “washout” experiments. For example, during treatment of cells with the reversible antagonist ONO-RS-082, the Golgi complex fragments into smaller “mini-stacks.” Upon washout, Golgi membranes rapidly form extensive membrane tubules that mediate the coalescence of the mini-stacks into an intact Golgi ribbon (de Figueiredo et al., 1999; Judson & Brown, 2009). Inhibition of endosome membrane tubule formation is similarly reversible with ONO-RS-082 (de Figueiredo et al., 2001).

1. At the end of the treatment period, rapidly rinse the cells 3–5 times with MEM containing 10% FBS (MEM-FBS). Continue the experiment in MEM-FBS.

2. Alternatively, rinsing can be done with MEM-FBS at 4 °C to inhibit membrane trafficking, then switching to MEM-FBS at 37 °C to initiate tubule formation. Care needs to be taken if using 4 °C washing steps because prolonged incubation in the cold will disassemble microtubules, which often facilitate membrane tubule-mediated transport (Lippincott-Schwartz, 1998).

3.1.3 Use of inhibitors in in vitro assays

Many of the antagonists listed in Table 1 have been used in the cell-free endosome membrane tubule reconstitution assay described in Section 1.4. In these experiments, the antagonist target(s) are generally in the BBC fraction, so concentrated stocks can added directly to the BBC. However, it is possible that isolated endosomes may have residual amounts of the cytosolic PLA₂ enzymes bound to their membranes, so it may be necessary to also treat the endosome preparation.