Super-resolution Imaging of the Bacterial Division Machinery

Summary

We describe a super-resolution imaging method to probe the structural organization of the bacterial FtsZ-ring, an essential apparatus for cell division. This method is based on quantitative analyses of photoactivated localization microscopy (PALM) images and can be applied to other bacterial cytoskeletal proteins.

Abstract

Bacterial cell division requires the coordinated assembly of more than ten essential proteins at midcell^1,2. Central to this process is the formation of a ring-like suprastructure (Z-ring) by the FtsZ protein at the division plan^3,4. The Z-ring consists of multiple single-stranded FtsZ protofilaments, and understanding the arrangement of the protofilaments inside the Z-ring will provide insight into the mechanism of Z-ring assembly and its function as a force generator^5,6. This information has remained elusive due to current limitations in conventional fluorescence microscopy and electron microscopy. Conventional fluorescence microscopy is unable to provide a high-resolution image of the Z-ring due to the diffraction limit of light (~200 nm). Electron cryotomographic imaging has detected scattered FtsZ protofilaments in small C. crescentus cells⁷, but is difficult to apply to larger cells such as E. coli or B. subtilis. Here we describe the application of a super-resolution fluorescence microscopy method, Photoactivated Localization Microscopy (PALM), to quantitatively characterize the structural organization of the E. coli Z-ring⁸.

PALM imaging offers both high spatial resolution (~35 nm) and specific labeling to enable unambiguous identification of target proteins. We labeled FtsZ with the photoactivatable fluorescent protein mEos2, which switches from green fluorescence (excitation = 488 nm) to red fluorescence (excitation = 561 nm) upon activation at 405 nm⁹. During a PALM experiment, single FtsZ-mEos2 molecules are stochastically activated and the corresponding centroid positions of the single molecules are determined with <20 nm precision. A super-resolution image of the Z-ring is then reconstructed by superimposing the centroid positions of all detected FtsZ-mEos2 molecules.

Using this method, we found that the Z-ring has a fixed width of ~100 nm and is composed of a loose bundle of FtsZ protofilaments that overlap with each other in three dimensions. These data provide a springboard for further investigations of the cell cycle dependent changes of the Z-ring¹⁰ and can be applied to other proteins of interest.

Protocol

1. Sample Preparation

1. Inoculate LB media with a single colony of strain JB281 [BW25113 / pJB042 (P_Lac:FtsZ-mEos2)]. Grow overnight in a shaker at 37 °C.
2. Dilute the culture 1:1,000 into M9⁺ minimal media [M9 Salts, 0.4% Glucose, 2 mM MgSO₄, 0.1 mM CaCl₂, MEM Amino Acids and Vitamins] and grow to mid-log phase (OD₆₀₀= 0.2-0.3) in the presence of chloramphenicol (150 μg/ml) at room temperature (RT).
3. Induce culture with 20 μM IPTG for 2 hr (see Note #1).
4. Pellet culture in microcentrifuge (8,000 rpm, 1 min) and resuspend in an equal volume of M9⁺; repeat. After the second resuspension, continue growth at RT for 2 hr.
5. Fix induced culture with 4% paraformaldehyde in PBS (pH=7.4) at RT for 40 min. Pellet culture and resuspend in an equal volume of PBS; repeat. If imaging live cells, skip fixation, pellet culture, and resuspend with equal volume of M9^- [M9 Salts, 0.4% Glucose, 2 mM MgSO₄, 0.1 mM CaCl₂, MEM Amino Acids]; repeat.
6. Dilute gold beads 1:10 with resuspended culture. Apply sample to prepared imaging chamber (see step 2.4).

[1] Note: The above induction protocol (steps 1.1-1.3) has been optimized for the expression system of strain JB281. Detailed induction conditions may vary for other expression systems or proteins of interest.

2. Assembly of the Imaging Chamber

1. Make a 3% (w/v) agarose solution in M9^-. Melt agarose at 70 °C in a bench-top heat block for 40min. Store melted agarose at 50 °C for up to 5 hr.
2. Place a clean microaqueduct slide in the upper half of the imaging chamber with the electrodes facing down. Align the lower gasket on the microaqueduct slide so as to cover the perfusion channels.
3. Apply ~50 μl of the melted agarose to the center of the glass slide. Immediately top the agarose gel droplet with a clean, dry coverslip. Allow the gel to solidify at RT for 30 min.
   1. To clean coverslips: arrange coverslips in a ceramic holder and sonicate for 20 min in rotating baths of acetone, ethanol and 1M KOH in glass jars. Repeat cycle three times, for a total of 9 sonications, followed by a single sonication in dH₂O. Store the cleaned coverslips in fresh dH₂O. Prior to use, blow dry coverslips with filtered air.
4. Carefully remove coverslip, leaving gel pad on the microaqueduct slide. Immediately add 1 μl of cell culture sample (see step 1.6) to the top of the gel pad. Wait for ~2 min, allowing solution to be absorbed by the gel pad. Top gel pad with a new clean, dry coverslip. Assemble the whole imaging chamber according to manufacturer's instructions.

3. Image Acquisition

1. Turn on the microscope, camera and lasers. Open MetaMorph software (Molecular Devices).
2. Place appropriate excitation/emission filters in the imaging path (Figure 1).
3. Set laser powers and acquisition settings accordingly.
   1. 488-nm laser = 15 mW (see Note #2)
   2. 561-nm laser = 75 mW
   3. 405-nm laser = 2 mW ± neutral density filter (see Note #3)
4. Lock imaging chamber into the stage via the complementary stage adaptor.
5. Designate an appropriate imaging region (100x100 pixel) that is homogenously illuminated by all three lasers.
6. Identify a sample area that contains both cells and fiducial markers (gold beads) in close proximity but not overlapping.
7. Focus on the surface of cells closest to the coverslip. Acquire one brightfield image with 50 ms integration time and move the focus 0.5 μm into the sample (Figure 3Ai).
8. Acquire ensemble green fluorescence image with excitation from the 488-nm laser using a 50 ms integration time (Figure 3Aii).
9. Acquire streaming video in red channel with continuous illumination from 405-nm and 561-nm lasers. For fixed cells, a 50 ms frame rate is used for a total of 20,000 frames (~17 min total) with the 405-nm laser power ramped by ~10% every 1,000 frames (see Note #4). For live cells, a 10 ms frame rate is used for a total of 3,000 frames (~30 s total) at a constant 405-nm laser power.
10. Move the focus back to the lower surface of the cells and acquire another brightfield image as in step 3.7.

[2] Note: The integrated intensity of the green fluorescence of a cell is directly proportional to the total number of FtsZ-mEos2 molecules expressed in the cell. The 488-nm excitation power and ensemble fluorescence acquisition settings should be kept constant for all cells so that the relative FtsZ-mEos2 expression levels in different cells can be compared.

[3] Note: Fixed cells are imaged with the Neutral Density (ND) filter (Figure 1, Optical Component #5) in place in order to achieve a slow activation rate so that each individual molecule can be accurately identified. The ND filter is removed when imaging live cells to increase the activation rate, while maintaining a low probability of two molecules being activated simultaneously in a diffraction-limited area. The faster rate applied to live-cell imaging is needed to decrease the total acquisition time, thereby "freezing" the Z-ring in time and limiting the effect of photodamage to the cell.

[4] Note: The numbers of frames acquired were optimized for our system and are dependent on the particular cellular structure, labeling density, activation rate and whether exhausting the entire pool of fluorophores is important for analysis. In live cells, it is important to obtain a sufficient number of localizations in as short a period of time as possible to avoid the blurring of the image due to movement of the cellular structure.

4. PALM Image Construction

1. Determine centroid position of each detected spot.
   1. Load the image stack into Matlab (MathWorks, Inc).
   2. Crop the image stack to a region around one cell that excludes all fiducial beads.
   3. Select an intensity threshold for spot detection that is above the background level, but below the average spot intensity. We account for the variation in background level throughout an experiment by calculating the running average of maximum intensity using a 150 frame window. The intensity threshold for each frame is then interpolated from the running average values.
   4. Subtract the respective threshold from each frame. Prospective spots are identified as three adjacent pixels with intensities above threshold.
   5. Fit the fluorescence intensity of each prospective spot to a two-dimensional Gaussian function [Equation #1] using a nonlinear least squares algorithm (Figure 2). The localization precision of each spot is calculated using the total number of detected photons [Equation #2]. Any poorly-fit spot with a localization precision greater than 20 nm (fixed cells) or 45 nm (live cells) is discarded (see Note #5). Any spot with an intensity greater than two-fold of the mean intensity, possibly due to overlapping emitters, is also discarded.
   6. For fixed cells, remove repeated observations of the same molecule by disregarding any spot whose centroid position is within 45 nm of any other spot that preceded it by 6 frames or less. For live cells, all spots are retained.
2. Calibrate sample drift using fiducial marker movement.
   1. Crop out a 10x10 pixel region around a fiducial marker.
   2. Subtract the respective threshold determined in 4.1.3 from each frame.
   3. Fit the intensity profile in each frame via least-square fitting algorithm assuming a two-dimensional Gaussian shape.
   4. Calculate displacement between centroid positions relative to frame 1.
3. Correct the centroid position of each unique molecule identified in 4.1 with the appropriate sample drift calculated in 4.2. Compare the brightfield images acquired in 3.7 and 3.10. If cells are observed to move relative to the fiducial markers, the data is thrown out.
4. Superimpose all corrected centroid positions on a single image composed of 15x15 nm pixels. Plot each unique molecule as a unit-area, two-dimensional Gaussian profile centered at the centroid position with a standard deviation equal to the localization precision [Equation #2]. This results in a probability density map, where a pixel's intensity is proportional to the likelihood of finding a molecule in that pixel (Figure 3Aiv).
5. Comparing PALM images to ensemble images.
   1. Replot the centroid positions as in 4.4, but on a plane with 150x150 nm pixels and a standard deviation equal to 250 nm. This generates a PALM image that mimics a diffraction-limited image, which can be used to compare to the diffraction-limited, ensemble image acquired in 3.8 (Figure 3Aiii).
   2. To confirm that the detected molecules are representative of the whole population, compare the reconstructed ensemble image (Figure 3Aiii, step 4.5.1) with the experimental ensemble fluorescence image (Figure 3Aii, step 3.8) to confirm that both images show structures of similar shape, orientation and relative intensity.

[5] Note: The minimum required localization precision was determined empirically for each imaging method by plotting the precisions of all molecules for a given sample and selecting an appropriate cutoff.

5. PALM Image Analysis

1. Measuring Z-Ring Width and Diameter.
   1. Rotate the PALM image so as to vertically orient the long axis of the cell (Figure 4A).
   2. Crop out the cellular region containing the Z-ring.
   3. Calculate the cumulative intensity by projecting an intensity profile along the cell's long axis.
   4. Fit the intensity profile to a Gaussian distribution. The ring width is defined as the full width half maximum (FWHM) of the fitted Gaussian distribution (Figure 4B).
   5. Project the cumulative intensity profile of 5.1.2 along the cell's short axis. The ring diameter is defined as the full length of the intensity profile (Figure 4C).
2. Plotting FtsZ Density (see Note #6).
   1. Replot the centroid positions as in (4.4), but without the Gaussian profile such that each unique molecule is given a value of 1 and assigned to a single pixel corresponding to its centroid position. In this way, the intensity of each pixel represents the total number of molecules detected in that pixel. The Density PALM image can also be visualized as a contour plot (Figure 5E).
3. Determining Spatial Resolution.
   1. Theoretically-achievable spatial resolution: create a representative PSF for the entire sample using the average determined localization precision of all detected molecules and calculate the FWHM (Equation #3).
   2. Experimentally-achieved spatial resolution: calculate the average displacement between repeat observations of the same molecule.

[6] Note: We used total internal reflection PALM (TIR-PALM, Figure 1 and 5) to restrict the activation and excitation to a thin layer (~200 nm) above the cell/glass interface. so that only FtsZ molecules associated with the membrane closest to the coverslip will be detected.

Results

Illustrated in Figure 3Aiv is a two-dimensional, super-resolution rendering of the Z-ring generated from the PALM imaging method described above. Below, we summarize qualitative and quantitative information that can be obtained from them.

Qualitatively, we observed that the Z-ring is an irregular structure that adopts multiple configurations (single band or helical arc) that are not distinguishable in conventional fluorescence images (compare Figure 3A-Dii

Discussion

PALM images contain information about molecule counts and positions within a cell, allowing detailed analysis of the distribution and arrangement of target protein molecules that is difficult to achieve by other means. Below we outline precautions that should be taken to extract accurate quantitative information while maintaining the biological relevance of PALM images. We also explore the information that can be best obtained using live vs. fixed cells. Finally, we suggest avenues for obtaining additional super-...

Materials

Name	Company	Catalog Number	Comments
50 x MEM Amino Acids	Sigma	M5550
100 x MEM Vitamins	Sigma	M6895
IPTG	Mediatech	46-102-RF
16% Paraformaldehyde	Electron Micrsocopy Sciences	15710-S
SeaPlaque GTG Agarose	Lonzo	50111
50 nm Gold Beads	Microspheres-Nanospheres	790113-010
FCS2 Imaging Chamber	Bioptechs
Stage Adaptor	ASI	I-3017
Inverted Microscope	Olympus	IX71
1.45 NA, 60x Objective	Olympus
IXON EMCCD Camera	Andor Technology	DU897E
488-nm Sapphire Laser	Coherent
561-nm Sapphire Laser	Coherent
405-nm CUBE Laser	Coherent