A stable cryogenic fluorescence microscope for correlative super-resolution light and electron microscopy

Reviewer #1 (Public review):

Summary:

In the manuscript "A stable cryogenic fluorescence microscope for correlative super-resolution light and electron microscopy," the authors demonstrate a new cryogenic light microscopy design and characterize its temperature and spatial stability. The manuscript does a good job of reviewing the state of the field and highlights the need for improved cryogenic microscope stages. The system avoids challenges associated with vacuum-based designs, particularly vacuum transfer systems that can be difficult to engineer, while also showing minimal ice contamination and drift, which are the primary challenges associated with open cryostat systems.

Strengths:

The key strengths of the manuscript are the simple design and the significant level of detail provided in the description of the cryogenic stage. This represents a valuable step forward for the field by providing a home-built, non-vacuum stage design that others can emulate.

Weaknesses:

There are only minor weaknesses or issues to address, which, if resolved, would strengthen the manuscript overall.

(1) A key element of the design gets little attention, which is the plastic cap for the objective. It is not entirely clear to the reader how this is being used except as something of a thermal break between the cryogen environment and the objective, but there are some questions. Is the objective housing touching the plastic cap? Where is the front of the cap relative to the front objective lens? Is the front objective lens exposed to the cryogenic environment? Could the authors provide some 3D views of that in an SI figure? This would help clarify.

(2) The refilling system is not shown in the diagrams provided in Figure 1 and S1 in sufficient detail. How is the system mechanically coupled to the dewar on the microscope stage? Are there any concerns about coupling vibrations onto the table?

(3) There is a description on page 6 that a rectangular aperture is used to align the excitation with the position and orientation of the sample. I know the authors are using this for excitation of the lamella, but without saying so in this text, it is confusing. I would consider stating that this is for future work involving excitation of lamella and then citing their preprint.

(4) In Figure 2d, the z-drift is shown with the focus lock correction applied. This is highly relevant, but I also think it would be good to plot the z position plus the stage position in an SI figure. This will give a better idea of the mechanical stability of the system. Also, in this figure, I wonder if the authors could comment on the source of the jumps in lateral position. For example, just before 30 minutes. Lastly, I would make the lower plot have a tighter y-axis range. It is hard to see anything, hence the inset.

(5) The ice contamination looks minimal in Figure 3. I think it would benefit the manuscript to have lower magnification images as well, to show the level of ice contamination across a representative square. This would be good, but only if the authors have it in hand.

(6) In Figure 4b, the y-axis is unclear. It looks like it has been normalized. Consider revising.

(7) A fluorescence intensity trace for the data shown in Figures 4c and f would be helpful to show the single-molecule behavior.

Reviewer #2 (Public review):

Summary:

This manuscript reports the development of a cryo super-resolution fluorescence microscopy system. The authors demonstrate that they can achieve a mechanical and thermal stability that is sufficient to perform cryo-SMLM over the course of several hours. Focus instability is compensated for by tracking a fluorescent bead for its movement in the axial direction and adjusting the sample stage accordingly during data acquisition. Lateral instabilities are corrected after data acquisition. An enclosure around the microscope allows to significantly reduce ice contamination during cryo-SMLM imaging and sample transfer. The authors show an example of correlative cryo-SMLM and cryo-ET imaging achieved with their microscope system, which depicts the distribution of FtsZ-rsEGFP2 in E. coli.

Strengths:

The authors have designed a microscopy system for SR-cryo-CLEM, which achieves high stability while reducing complexity and costs substantially when compared to vacuum-insulated systems (e.g., Hoffman et al., 2020). They also provide software for controlling the microscope and data acquisition. This lowers the barrier for other labs to implement SR-cryo-CLEM into existing cryo-ET workflows. Reduction of ice contamination helps to increase throughput, which is currently one of the biggest bottlenecks for SR-cryo-CLEM.

Weaknesses:

To correct for focus drift, the authors track a fluorescent bead in the far-red channel. This is possible for bacterial samples as used in this work, as beads can easily be introduced to surround the cells.

Recommendations:

(1) It is not discussed how this can be achieved in other samples than bacterial samples, such as lamellae in mammalian cells. Here, it would be much more difficult to introduce bright point-like markers with far-red fluorescence that would be distributed in the entire cell to capture at least one in the final lamella. Furthermore, it might be important to know for readers whether the far-red channel has to be sacrificed entirely for the focus correction.

(2) The authors show an application of SR-cryo-CLEM imaging of FtsZ-rsEGFP2 in E. coli. In the chosen correlative example (Figure 4d.f), no clear structure can be seen in the fluorescent images. The overview image (Figure 4d) shows no distinct signal in the cell, as it is shown for the non-correlative example in Figure 4a. The cryo-SMLM image (Figure 4f) does not show any ring-like features or accumulations of signals at the constriction site, as would be expected for a projecting along the optical axis. A clearer application example, which would show how increased resolution in cryo fluorescence microscopy enables resolving certain structural details or adds information not accessible in cryo electron tomography, would have strengthened the work. Particularly if taking into consideration that bacteria have a strong auto-fluorescence in the green range (Dahlberg et al., 2020), which could lead to high background or false positive localizations when using green fluorophores as labels.

(3) Access to CAD drawings (particularly for custom-made parts, such as cryostat or humidity enclosure) and a parts list is highly important for other researchers who would like to set up this SR-cryo-CLEM system in their own lab or institution. This is currently missing and, therefore, creating a hurdle for a wider adaptation of the technique.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

In the manuscript "A stable cryogenic fluorescence microscope for correlative super-resolution light and electron microscopy," the authors demonstrate a new cryogenic light microscopy design and characterize its temperature and spatial stability. The manuscript does a good job of reviewing the state of the field and highlights the need for improved cryogenic microscope stages. The system avoids challenges associated with vacuum-based designs, particularly vacuum transfer systems that can be difficult to engineer, while also showing minimal ice contamination and drift, which are the primary challenges associated with open cryostat systems.

Strengths:

The key strengths of the manuscript are the simple design and the significant level of detail provided in the description of the cryogenic stage. This represents a valuable step forward for the field by providing a home-built, non-vacuum stage design that others can emulate.

We thank the reviewer for their positive assessment and strive to address the weaknesses they have constructively raised below.

Weaknesses:

There are only minor weaknesses or issues to address, which, if resolved, would strengthen the manuscript overall.

(1) A key element of the design gets little attention, which is the plastic cap for the objective. It is not entirely clear to the reader how this is being used except as something of a thermal break between the cryogen environment and the objective, but there are some questions. Is the objective housing touching the plastic cap? Where is the front of the cap relative to the front objective lens? Is the front objective lens exposed to the cryogenic environment? Could the authors provide some 3D views of that in an SI figure? This would help clarify.

We thank the reviewer for this helpful suggestion. In the revised manuscript, we will include a new supplementary figure (Fig. S2) providing detailed 3D views of the copper adapter, microscope objective, and plastic cap. The figure will show that the rim surrounding the front lens of the objective is covered by the plastic cap to provide thermal insulation between the objective housing and the cryogenic environment (Fig. S2b). We will also clarify that the front surface of the cap is levelled with the front objective lens to maintain the full working distance of the objective while allowing for axial movement of the z-stage. Finally, we will explicitly state that the front objective lens is exposed to the cryogenic environment (cold nitrogen gas).

(2) The refilling system is not shown in the diagrams provided in Figure 1 and S1 in sufficient detail. How is the system mechanically coupled to the dewar on the microscope stage? Are there any concerns about coupling vibrations onto the table?

To minimise vibrations arising from the nitrogen refilling pumps, the cryostat and liquid nitrogen tubing are mechanically decoupled from the microscope cage system, objective, translation stages, and sample. Specifically, the cryostat and nitrogen tubing are supported independently on a laboratory jack and surround the cage system without rigid mechanical contact. In the revised manuscript, we will update Fig. S1a, b to illustrate the liquid nitrogen tubing and refilling system more clearly. In addition, we will include a new supplementary figure (Fig. S3) to show the detailed cryostat design, refilling tubing, and temperature sensor position.

(3) There is a description on page 6 that a rectangular aperture is used to align the excitation with the position and orientation of the sample. I know the authors are using this for excitation of the lamella, but without saying so in this text, it is confusing. I would consider stating that this is for future work involving excitation of lamella and then citing their preprint.

We agree that the purpose of the rectangular aperture should be made clearer. In line with the suggestion from the reviewer, in the revised manuscript, we will briefly explain that the aperture is intended for selective illumination, such as in applications to cryo-FIB lamellae, and will cite our recent preprint describing this approach.

(4) In Figure 2d, the z-drift is shown with the focus lock correction applied. This is highly relevant, but I also think it would be good to plot the z position plus the stage position in an SI figure. This will give a better idea of the mechanical stability of the system. Also, in this figure, I wonder if the authors could comment on the source of the jumps in lateral position. For example, just before 30 minutes. Lastly, I would make the lower plot have a tighter y-axis range. It is hard to see anything, hence the inset.

We thank the reviewer for this suggestion. In the revised manuscript, we will include an additional supplementary figure (Fig. S4) showing the axial drift measured without focus-lock correction to illustrate the intrinsic mechanical stability of the microscope. We will also clarify that the periodic lateral displacement observed along the x-direction (approximately 300 nm amplitude with a period of ~22 minutes) arises from slight lateral repositioning accompanying z-stage stepping during focus-lock operation, likely due to mechanical coupling between the axes of the translation stage. We will revise the lower panel of Fig. 2d by reducing the y-axis range to improve data visibility.

(5) The ice contamination looks minimal in Figure 3. I think it would benefit the manuscript to have lower magnification images as well, to show the level of ice contamination across a representative square. This would be good, but only if the authors have it in hand.

We agree that this would be useful. In the revised manuscript, we will update Fig. 3 to include two additional low and intermediate-magnification cryo-EM images showing a representative grid square and a zoomed-in region of it, including a few grid holes. These images provide an overview of the ice contamination across a substantially larger field of view.

(6) In Figure 4b, the y-axis is unclear. It looks like it has been normalized. Consider revising.

The y-axis in Fig. 4b represents the localization rate (number of detected localizations per frame) within the selected ROI in Fig.4c and was not normalized. The values were calculated in SMAP by binning the localization frames into 100 temporal bins and dividing the number of localizations in each bin by the corresponding bin width, resulting in units of localizations per frame. Therefore, values close to 1 indicate approximately one localization detected per frame at that time point. To avoid potential confusion regarding the interpretation of this representation, we will replace this plot in the revised manuscript with a more explicit visualization showing the number of detected localizations per defined number of frames as a function of time (frame number) for the specific ROI shown in Fig. 4c.

(7) A fluorescence intensity trace for the data shown in Figures 4c and f would be helpful to show the single-molecule behavior.

In the revised manuscript, we will add fluorescence intensity traces corresponding to the single-molecule events shown in Fig. 4c and Fig. 4f to further demonstrate their single-molecule emission characteristics.

Reviewer #2 (Public review):

Summary:

This manuscript reports the development of a cryo super-resolution fluorescence microscopy system. The authors demonstrate that they can achieve a mechanical and thermal stability that is sufficient to perform cryo-SMLM over the course of several hours. Focus instability is compensated for by tracking a fluorescent bead for its movement in the axial direction and adjusting the sample stage accordingly during data acquisition. Lateral instabilities are corrected after data acquisition. An enclosure around the microscope allows to significantly reduce ice contamination during cryo-SMLM imaging and sample transfer. The authors show an example of correlative cryo-SMLM and cryo-ET imaging achieved with their microscope system, which depicts the distribution of FtsZ-rsEGFP2 in E. coli.

Strengths:

The authors have designed a microscopy system for SR-cryo-CLEM, which achieves high stability while reducing complexity and costs substantially when compared to vacuum-insulated systems (e.g., Hoffman et al., 2020). They also provide software for controlling the microscope and data acquisition. This lowers the barrier for other labs to implement SR-cryo-CLEM into existing cryo-ET workflows. Reduction of ice contamination helps to increase throughput, which is currently one of the biggest bottlenecks for SR-cryo-CLEM.

We thank the reviewer for their critical assessment, and for their suggestions below which we have used to improve the manuscript.

Weaknesses:

To correct for focus drift, the authors track a fluorescent bead in the far-red channel. This is possible for bacterial samples as used in this work, as beads can easily be introduced to surround the cells.

Recommendations:

(1) It is not discussed how this can be achieved in other samples than bacterial samples, such as lamellae in mammalian cells. Here, it would be much more difficult to introduce bright point-like markers with far-red fluorescence that would be distributed in the entire cell to capture at least one in the final lamella. Furthermore, it might be important to know for readers whether the far-red channel has to be sacrificed entirely for the focus correction.

We thank the reviewer for highlighting this point. We agree that focus stabilization strategies for cryo-FIB lamellae are likely to differ from those used for the individual bacterial cell samples. For lateral drift correction, the presence of a single continuously detectable bright feature within the field of view is sufficient. Importantly, this feature does not need to be a fluorescent bead; any stable signal that can be continuously detected by the camera can serve as a suitable reference for drift correction. We will expand the Discussion to describe potential strategies for stable cryo-SMLM imaging, including the use of intrinsic sample or lamella features for autofocus, minimal fiducial-based approaches, and the practical implications of dedicating the far-red channel to focus stabilization.

Furthermore, in the revised manuscript, we will include a new supplementary figure (Fig. S4) demonstrating the intrinsic axial stability of the microscope in the absence of active focus-lock correction. These measurements show that the system remains within the objective's depth of focus for a relatively long time, providing adequate stability for experiments in which far-red fluorescent fiducial beads are unavailable, such as cryo-FIB lamella imaging.

(2) The authors show an application of SR-cryo-CLEM imaging of FtsZ-rsEGFP2 in E. coli. In the chosen correlative example (Figure 4d.f), no clear structure can be seen in the fluorescent images. The overview image (Figure 4d) shows no distinct signal in the cell, as it is shown for the non-correlative example in Figure 4a. The cryo-SMLM image (Figure 4f) does not show any ring-like features or accumulations of signals at the constriction site, as would be expected for a projecting along the optical axis. A clearer application example, which would show how increased resolution in cryo fluorescence microscopy enables resolving certain structural details or adds information not accessible in cryo electron tomography, would have strengthened the work. Particularly if taking into consideration that bacteria have a strong auto-fluorescence in the green range (Dahlberg et al., 2020), which could lead to high background or false positive localizations when using green fluorophores as labels.

We thank the reviewer for this thoughtful comment. We agree that a correlative example displaying more pronounced structural features would further illustrate the capabilities of cryo-SMLM. However, the primary aim of the present work is the development and characterization of a robust cryogenic super-resolution microscope for reliable cryo-SMLM and correlative cryo-CLEM, rather than the demonstration of new biological applications. The utility of correlative cryo-SMLM/cryo-ET for resolving cellular structures has already been established in previous studies, including those employing rsEGFP2-labelled targets.

The correlative dataset presented here is intended to demonstrate the compatibility of the microscope with cryo-CLEM workflows rather than to provide detailed biological insight. Moreover, the use of intact E. coli cells imposes inherent limitations on the ultrastructural information accessible by cryo-electron tomography; overcoming these limitations would typically require specimen thinning, for example, by cryo-focused ion beam (cryo-FIB) milling, which is beyond the scope of the present work.

Regarding the concern about auto-fluorescence, elevated background fluorescence is not unique to bacterial samples or green fluorescent proteins but is a general consideration in cryo-SMLM that depends on the specimen and imaging conditions. While auto-fluorescence may reduce image contrast, it does not affect the conclusions of this work, which focuses on the design and performance of the microscope.

(3) Access to CAD drawings (particularly for custom-made parts, such as cryostat or humidity enclosure) and a parts list is highly important for other researchers who would like to set up this SR-cryo-CLEM system in their own lab or institution. This is currently missing and, therefore, creating a hurdle for a wider adaptation of the technique.

Thank you for this useful suggestion. In the revised manuscript, we will make available the complete SolidWorks CAD files for all custom-designed components, together with a comprehensive parts list and the full assembly corresponding to Fig. S1 as supplementary materials.