PIE-scope, integrated cryo-correlative light and FIB/SEM microscopy

link管理

链接快照平台

输入网页链接，自动生成快照
标签化管理网页链接

相关文章推荐

豁达的西装 · [BT/网盘下载][巅峰拍档/英国疯狂汽车秀 ...· 2 周前 ·

道上混的斑马 · 沈阳地铁3号线最新进展_腾讯新闻· 1 月前 ·

开朗的牛排 · 「3D图形开发(OpenGL/VTK/OSG ...· 3 月前 ·

忧郁的作业本 · MySQL :: Security in ...· 3 月前 ·

风流的匕首 · 优秀的工具=生产力！15款高效原型设计软件种 ...· 4 月前 ·

Version of Record published

July 4, 2019 (This version)

Accepted Manuscript published

July 1, 2019 (Go to version)

Accepted

June 26, 2019

Received

February 8, 2019

Cryo-electron tomography (cryo-ET) is emerging as a revolutionary method for resolving the structure of macromolecular complexes in situ. However, sample preparation for in situ Cryo-ET is labour-intensive and can require both cryo-lamella preparation through cryo-focused ion beam (FIB) milling and correlative light microscopy to ensure that the event of interest is present in the lamella. Here, we present an integrated cryo-FIB and light microscope setup called the P hoton I on E lectron microscope (PIE-scope) that enables direct and rapid isolation of cellular regions containing protein complexes of interest. Specifically, we demonstrate the versatility of PIE-scope by preparing targeted cryo-lamellae from subcellular compartments of neurons from transgenic Caenorhabditis elegans and Drosophila melanogaster expressing fluorescent proteins. We designed PIE-scope to enable retrofitting of existing microscopes, which will increase the throughput and accuracy on projects requiring correlative microscopy to target protein complexes. This new approach will make cryo-correlative workflow safer and more accessible.

Cryo-electron tomography (cryo-ET) is currently the principal method for investigating the structure of proteins and protein complexes directly in their native environment ( Beck and Baumeister, 2016 ). Here, a cellular sample is vitrified by plunge or high-pressure freezing to instantaneously cease cell activity and prevent crystallisation of water in the sample ( Dubochet et al., 1988 ) ( Lucić et al., 2005 ). Despite the high-resolution potential of cryo-ET, there are two major challenges that limit the applicability of this technique. First, it is not possible to image a cell in regions thicker than 500 nm using a conventional 300 keV transmission electron microscope ( Frank, 1996 ). Second, the cellular environment is extremely heterogeneous and it is impossible to unequivocally identify every visible density ( Förster et al., 2010 ). To overcome the issue of cell thickness, the most common and successful approach is to use a cryo-focused ion beam microscope (FIB) to thin the sample and produce flat electron-transparent lamellas of approximately 300 nm thick ( Marko et al., 2006 ; Marko et al., 2007 ) ( Hsieh et al., 2014 ; Rigort et al., 2010 ). Thus, in a single lamella, only a portion of the cell is retained for cryo-ET but it remains unknown as to whether the protein/complex of interest is present. Considering that every step from FIB milling to cryo-ET imaging is time-consuming, it is crucial to accurately target the FIB milling procedure to the region on interest. The solution to this has been correlative light and electron microscopy (CLEM) ( Arnold et al., 2016 ; Sartori et al., 2007 ), which also offers the opportunity to validate the interpretation of the cryo-ET data by matching the fluorescence data with tomographic densities.

The most complete cryo-CLEM workflow links light microscopy (live cell imaging and cryo-LM), cryo-FIB and cryo-ET ( Figure 1 ). The goal, in this case, is to image a protein of interest within a living specimen at the highest possible/required resolution (this step is optional but it can be crucial when working on time-sensitive cellular processes). Then, when appropriate, the sample is vitrified though plunge freezing (or high-pressure freezing). At this stage, the sample must be transferred under cryogenic conditions into a cryo-LM ( Sartori et al., 2007 ; Schorb and Briggs, 2014 ; van Driel et al., 2009 ) and imaged to obtain a still after cryo-fixation to locate all the regions that will be targeted for cryo-FIB milling. Cryo-LM on the vitrified sample is required (even if fluorescence images have been acquired on the living sample) to ensure that the images recorded during the living specimen experiment can be related to the cryo-LM image as effects due to dynamic cell process, blotting and plunge-freezing could cause the target structure to move position. The sample is then transferred (again under cryogenic conditions) into the cryo-FIB. Once it has been milled, the specimen needs to be either transferred to the cryo-TEM for imaging or back to the cryo-LM to confirm that the FIB milling successfully isolated the region housing the protein of interest ( Arnold et al., 2016 ).

( a ) Live cell imaging. ( b ) Plunge or high-pressure freezing of the sample for cryo-preservation (note, cells can be either cultured on the grid in ( a ) or in case of suspension be added onto the grid at this stage). ( c ) Cryo-light fluorescence microscopy in the PIE-scope to identify regions of interest. ( d ) Simple lateral translation of the sample brings the selected region of interest under the FIB/SEM columns. ( e ) Once the region of interest is under the FIB, a lamella can be thinned down. The angle between the prepared lamella to the grid plane can be controlled by the stage tilt. ( f ) Simple translation and tilt of the prepared lamella inside the system allow verification of successful targeting of the region of interest with light microscopy. ( g,h ) An optional repeat of ( e ) and ( f ) for further thinning of the lamella. Since two cryo-transfers are required, the system allows rapid and multiple intermediate verifications of targeting with optical and scanning electron microscopes. The x,y,z-axes shown in c-h correspond to the stage axes of the FIB/SEM. ( i ) Once the region of interest is successfully targeted and optically verified, the lamella is cryo-transferred into a TEM for high-resolution sample analysis. Distances and dimensions throughout the figure have been adapted for illustrative purposes and are not representative.

This workflow requires handling of samples across multiple microscopes ( Arnold et al., 2016 ; Rigort et al., 2010 ). Each transfer must be performed under cryogenic conditions and in a humidity-free environment in order to avoid the formation or deposition of crystalline ice. Also, handling a cryo-EM grid on which multiple lamellas have been prepared is an extremely delicate process. Even though users do not handle a bare grid as in the formative days of this field ( Marko et al., 2007 ; Rigort et al., 2010 ), but rather a cartridge (which increases the sample stiffness), every manual handling step risks thawing, damaging or contaminating the sample and lowers the overall success of the process ( Hsieh et al., 2014 ). This also means that users are unlikely to confirm whether their lamella preparation was successful via cryo-LM before acquiring the data with a transmission electron microscope (TEM), which may further impact workflow success.

All the risks associated with cryo-transfers described above can be mitigated using an integrated approach, where a light microscope is integrated into the vacuum chamber of the electron microscope. Such an approach has been successfully implemented for CLEM where the introduction of a light microscope objective in a TEM column permitted switching between TEM and LM imaging, known as the iLEM ( Agronskaia et al., 2008 ). The iLEM has demonstrated its value on both room temperature and cryo-samples ( Faas et al., 2013 ), making navigation seamless and accelerating the data collection process by eliminating the need to image areas not representing the region of interest. The major limitations that iLEM suffers from are that high-resolution LM imaging is impossible, because of the tight geometry of the TEM column, and the need to remove the cryo-box usually protecting the sample, which can lead to long-term contamination. Also, for the specific requirement of targeting the lamella preparation, having a light microscope embedded in the cryo-TEM will not be beneficial.

Here, we report the design and functionality of an integrated cryo-LM and cryo-FIB/SEM which we call the P hoton I on E lectron Microscope, or PIE-scope. We demonstrate how the PIE-scope can be used to accurately guide cryo-FIB milling to target cells and tissues using a variety of specimens. We also present an example of software integration to control both the cryo-FIB and the cryo-LM.

To ensure that existing FIB/SEM systems in the field can be retrofitted with the solution proposed here, we chose a design that would be applicable to as many systems as possible regardless of the SEM working distance and column shape. For this reason, the LM does not image the sample at the coincidence point but 49 mm away from it along the X axis ( Figure 2 ). Although we developed it to fit on a standard ThermoFisher FIB/SEM chamber, with minor modifications it can be adapted to other systems.

( a ) Outside view of the PIE-scope. The atmospheric (external) retrofitted component consisting of one excitation and one emission arms. The camera at the end of the emission arm can be replaced. The excitation arm has been designed to fit an FC/APC fibre-end with a 6 µm core and a NA = 0.06. BS represents the position of the dichroic beam-splitter. ( b ) A view of the FIB/SEM chamber containing showing the in-vacuum section of the PIE-scope. An objective is mounted on a high-precision motorized stage for sample focusing (LM focus drive). The light from the external arm (cyan arrow) is delivered through a glass flange and directed into the objective by a mirror. ( c,d ) CAD renderings of the in-vacuum section of the PIE-scope. These renderings are a simplified representation of the chamber and do not include all components visible in ( b ), but the renderings clarify the geometry of the critical components of PIE-scope. The XYZ axes at the bottom left of panels c and d are aligned with the stage axes.

The position of the objective relative to the FIB/SEM offers multiple advantages: (i) switching between the LM and FIB/SEM imaging can be performed with a single axis translation ( Figure 1 ); (ii) no rotation is required - only tilt might be needed in some designs (important for cryo-stages where rotation is limited); (iii) if the geometry of the sample holder on the stage allows, objectives with high numerical aperture can be used; (iv) focussing with the LM does not require the sample to be moved.

When designing LM assembly, we considered that if not in use, the objective should be retracted to prevent collision with the stage. Furthermore, in order to ensure safety during stage movements when the objective is not retracted, the entire in-vacuum assembly is grounded to the FIB/SEM chamber, therefore allowing for sensing stage collisions. To focus, we use a piezoelectric linear positioner with active feedback and a precision better than 10 nm, which provides fine focus control and high repetitivity. Considering the geometry of the vacuum chamber, in order to convey the beam-path normal to the glass window at the vacuum flange, we had to steer the beam at an angle of 17.2 degrees which imposed the use of a 3 × 1’ mirror. Here, in order to prevent mechanical distortions of the mirror, the holder was designed to exert an equal pressure across the rim of the mirror. The design allows for use of high NA objectives such as the Olympus MPLAPON 50x/0.95. We tested multiple objectives for vacuum compatibility and found that degassing will occur for an average of 24 hr from the first pumping cycle (see below for details). Although we have not noticed any change in the optical performance of the objectives as a consequence of their permanence under high-vacuum, we expect that the cement used in the assembly of those complex optical elements will degrade over time and the objective will probably need to be replaced (our current experience describe this as >12 months).

Outside the vacuum chamber, the LM is composed of an aluminium body that is directly attached to the flange and guides the position of the required optical elements. In front of the glass window on the flange, a rectangular mirror steers the beam and ensures a $n (\frac{π}{2}) t i l t (w h e r e n \in Z^{+})$ relative to the orientation of the objective optical axis. The microscope body has two arms, the one devoted to the illumination contains the fibre-end and the illumination tube-lens; while the detection arm hosts the imaging tube-lens and the camera. The body includes a dichroic selector, which in our case allows for four channel fluorescence (DAPI, GFP, RFP, Far-red) and reflected light illumination. The possibility of having a reflected bright-field illumination image can simplify correlation in cases where there are no fiducial beads or where the fluorescence signal cannot be directly matched to the shape of the sample. The current design allows for wide-field imaging, but Structured-Light-Illumination (SIM) ( Gustafsson, 2000 ) could be implemented by modifying the excitation arm to accommodate the polarising optics and either a transmission grating or a programmable reflective ferroelectric liquid-crystal-on-silicon (LCOS) micro-display ( Kner et al., 2009 ; Shao et al., 2011 ). The implementation of SIM would add optical sectioning and super-resolution capabilities.

In order to control all imaging modalities from the same computer and to have an interface designed around the needs of CLEM ( Figure 3 ), we developed a program in Labview and Python. The interface allows for (i) simple image acquisition in all modalities; (ii) seamless navigation between modalities which can bring the same ROI centred relative to each beam and (iii) provides hierarchical data storage to simplify post-processing. This control interface, which we called PIE-scope-commander is freely available from the laboratory resource webpage ( https://www.demarco-lab.com/resources ).

( a,b ) Optical ( a ) and FIB ( b ) image of the same sample region, containing a series of pre-milled test patterns (crosses). After the stage translation calibration and targeting steps one can select the target feature using the optical microscope ( a ), and simply translate the same field of view under the FIB and the ( b ). ( c ) The precision of direct, unsupervised ROI targeting by simple sample translation is <420 nm for the X-axis (major axis of translation) and <300 nm for the orthogonal Y-axis. ( d, e, f ) The procedure used to quantify the mechanical correlation precision when moving between an optical ( a ) and FIB ( b ) images. Fiducial markers were identified in all modalities and the transformation to perform the correlation was calculated. In ( f ) a scatter plot describing the mean residual for all the measurements (n = 86), showing the precision is ~500 nm (±σ). For every measurement, we selected beads > 100 beads to eliminate/average out eventual fiducial selection errors.

The first step, after the LM has been mounted and aligned, is to measure the exact position of the optical axis relative to the coincidence point of the FIB/SEM. This step requires a fiducial that can be recognised in all modalities. We noticed that, due to mechanical tolerances, every time the system is mounted there is a variability of up to 500 µm in the final position of the light microscope. The calibration is typically performed manually by estimating the shift between the coincident point and the expected position and adjusted by centring the fiducial. The shift values are then inserted in the control software in order to automatically perform direct movements between imaging modalities.

In the software, we assumed three positions (see Figure 1 and Video 1 ), which can be customised: (i) LM imaging, where the sample orientation is normal to the optical axis of the LM; (ii) FIB imaging, where the sample orientation is normal to the optical axis of the FIB; (iii) lamella preparation, where the sample is at 15–22 degrees from the optical axis of the FIB. Once the imaging positions have been defined it is possible to perform pre-computed relative movements (shifts, tilt, and compucentric rotations) in order to image the same ROI under different modalities. The control software interface allows direct control of the FIB/SEM, a visual comparison between the FIB/SEM and LM images and saving the data in organised sub-directories.

In PIE-scope we implemented a basic correlation procedure, which leads the user to identify the location of the region of interest in the FIB or SEM image. Although using the proposed procedure is optional and specific use-cases might benefit from custom designed image processing, we find that the availability of a general method already present and embedded in the software greatly enhances the usability. The PIE-scope correlation is performed through custom made python scripts that allow selecting multiple points on the LM and FIB/SEM images to calculate the appropriate transformation. 2D correlation is performed simply by applying an affine transformation that includes anisotropic scaling to match the pixel spacing resulting from imaging a sample from different tilts. This procedure is best suited for 2D correlation and, according to previous reports ( Kukulski et al., 2011 ; Schorb and Briggs, 2014 ), it can lead to correlation precisions which are better than 100 nm. If 3D correlation is required, this can be performed using the algorithm previously described in Arnold et al. (2016) , which can be downloaded as Python package from 3DCT ( https://github.com/coraxx/3DCT ). 3DCT can be used independently or it can be called from PIE-scope by modifying the call-back function of the correlate button in the LabVIEW GUI to interface with the 3DCT package.

Since we modified the vacuum system of the FIB/SEM we first ensured that the vacuum system performance was unaffected. This was measured using both the vacuum gauge of the FIB/SEM (HVG), and through the use of a residual gas analyser (from Pfeiffer). Once satisfied that both modified flanges did not present any leak we took advantage of the residual gas analyser to identify the average degassing time for the objective. This redefined the minimum pumping cycle length prior to insertion of cryo-samples to >20 hr. After this time, the vacuum in the FIB/SEM chamber is 1e-4 Pa at room temperature and 5e-5 Pa when the system is at cryogenic temperature. The condensation rate on cryo-samples was found to be identical to the rate measured before the installation of the PIE-scope. It is worth to mention that the maximum vacuum level obtainable varies from instrument to instrument and that the pumping time is dependent on the components present in the vacuum chamber. Therefore, the evaluation of the vacuum performance (or loss of performance) should be done on each individual instrument (e.g. Thermo Fisher Helios systems in average have better vacuum than Thermo Fisher Scios), and there should be no variation on the maximum vacuum level that can be reached after the PIE-scope installation.

The addition of an external component to the FIB/SEM leads to the potential introduction of vibrations into the system, which may result in mechanical drift. We measured the drift of the sample at room and cryo temperature with and without the light microscope installed to ensure no significant variation would be detected. With the presented hardware composition (fibre coupled Toptica iChrome-CLE and heated Basler USB3 camera) the drift remained within the specification of the FIB/SEM (<10 nm/min at room temperature and <30 nm/min at cryogenic temperature). Further, when imaging at the maximum framerate achievable with the SEM (50 frames/s) we have not detected changes in the vibrations.

Since the system is designed for correlative studies, the first feature to characterise was the repetitivity of the relocation precision between the LM and the coincidence point. As displayed in Figure 4 , performing direct unsupervised movement shows that the relocation precision is <420 nm along X and <220 nm along Y. The relocation precision is a characteristic of the FIB/SEM stage and could vary from instrument to instrument, but most manufacturer's specifications claim a relocation precision better than 2 µm. Considering that the simplest use-case consists of positioning a 5–10 µm lamella around a target, the above-mentioned stage accuracy is sufficient. The relocation precision was measured by repeatedly moving between the FIB and the LM imaging positions ( Figure 4 ). After each movement, we imaged the sample and calculated the residual error in relocation by calculating the required transformation to match the last image with the previous one (the residuals for each measurement are displayed in Figure 4 ). As previously shown ( Arnold et al., 2016 ; Schorb and Briggs, 2014 ), when using fiducial markers, it is possible to improve the correlation accuracy to <100 nm despite the diffraction limit of the light microscope equipped with a non-immersion objective will be limited to ~320 nm in XY and ~555 nm in Z (assuming a NA of 0.95 and a wavelength of 500 nm). In order to quantify the optical performance of the system, we measured the maximum imaging resolution through the calculation of point-spread-functions. As expected, given the simplicity of the optical design, for all objectives tested we could measure a resolution in line with the theoretical diffraction limit. Through the use of image deconvolution, the LM resolution in all three dimensions could be mildly improved but in order to better resolve the molecular distribution at subcellular level super-resolution imaging would be better suited. The field of cryo-super-resolution fluorescence microscopy is still at its early stages and the most suitable method still needs to be identified but recently multiple approaches have been demonstrated ( Moser et al., 2019 ; Tuijtel et al., 2019 ; Xu et al., 2018 ). The PIE-scope design is per se compatible with any type of epi-fluorescence imaging. Accordingly, if super-resolution imaging is required the implementation will be possible and will only require changes on the atmospheric side of the device. Along these lines, the effectiveness and usability of cryo-super-resolution imaging are predominantly limited by the ice contamination that inevitably builds up during long imaging sessions and by drift present in those systems ( Briegel et al., 2010 ; Wolff et al., 2016 ; Xu et al., 2018 ). The PIE-scope resolves both these problems since the sample is kept in high vacuum during the entire imaging session and the drift rate is determined by the specifications of the cryo-stage in use, which must be compatible with high-resolution SEM imaging and FIB micromachining.

( a ) A scheme of the communication setup. Here the microscope computer controls the FIB/SEM and runs the python control server. On the support computer, we run a python client (ThermoFisher Autoscript) and the custom built PIE-scope commander to control both the FIB/SEM and the LM. ( b ) The PIE-scope commander interface, all the required imaging positions can be calibrated from the main window, basic imaging parameters can be set, both FIB and LM images are visible side-by-side and data location can be defined for each experiment. File naming will have a user-defined root and appendix describing the imaging parameters (e.g. FIB, SEM, LM, laser line, dwell-time, exposure time).

Finally, one aspect to consider if one wishes to use commercial objectives with an NA >0.8 is that the orientation and the position of the sample when mounted on the cryo-stage must allow imaging with sub-mm working distances. Luckily, all available cryo-stages come with inserts that can easily be modified or re-designed according to this specification if required.

In order to evaluate our new system on a real-life sample, we prepared targeted cryo-lamellae where the milling position was defined entirely through the fluorescence signal. Here, we used larvae of the nematode Caenorhabditis elegans , where four head neurons (BAGL/R and ASGL/R) express mCherry, to prepare a lamella centred around a neuronal cell body and one of the main dendrites which extend anteriorly towards the mouth of the animal. In Figure 5 we show stills through the milling procedure, from the identification of the ROI, through the controls while thinning, to the final image after polishing. As shown in Figure 5 , the fluorescence is maintained after milling, which can be used to confirm that the target structure (a dendrite of a BAGL/R neuron) is effectively contained in the lamella.

( a ) Correlation between the SEM (top view) and fluorescence microscopy images. ( b ) Correlation between the FIB and fluorescence microscopy images. The sample is tilted into the lamella preparation position (22° with respect to the grid plane). ( c ) SEM image of the thinned down lamella overlayed with the fluorescence image from the same region. ( d,h ) SEM and optical images, respectively, of the initial state of the sample, while the pairs ( e, i ), ( f, j ) and ( g, k ) show SEM and optical images, respectively, of the intermediate steps in the lamella, thinning. Since the optical microscope is integrated within the system, multiple ROI-verifications are possible throughout the workflow without the need for cryo-transfers.

We further tested the ability of targeted lamella preparation in a bulk cryo-sample. Here a whole brain dissected from a Drosophila melanogaster larva expressing EGFP in interneurons (Engrailed-Gal4, UAS-eGFP) was cryo-fixed and imaged under fluorescence. Based on the fluorescence signal, 1.5 µm thick lamellas were isolated around the neuronal body ( Figure 6 ). This setup is compatible with emerging techniques such as the cryo-lift-out which will allow selective isolation of lamellae from bulk tissue ( Mahamid et al., 2015 ; Parmenter et al., 2016 ; Rubino et al., 2012 ; Zhang et al., 2016 ). The position of the LM does not interfere with the manipulator, making it compatible with the development of future applications involving cryo-lift-out.

Targeted lamella preparation in a bulk cryo-sample (larval D. melanogaster brain expressing EGFP in interneurons).

( a, b, c ) Initial FIB ( a ), SEM ( b ) and LM ( c ) images. ( d ) Milling of straight edges in the sample allows improved precision of correlation between images acquired in different modalities. Here, the overlay between LM and FIB images of the pre-milled sample is shown. ( d, e, f, g ) Pairs of SEM and optical microscopy images of the initial sample state ( d ), two intermediate verification steps ( e and f ), and the final result ( g ). Based on the fluorescence signal, 1.5 µm thick lamellae were isolated around the neuronal body ( g ). Inset in ( g ) shows a magnified image of a lamella ready for a cryo-lift-out.

Considering the sample size, in both the C. elegans larvae and the D. melanogaster brain the lamellas were kept of thicknesses above 1 µm. This thickness still allows high-quality fluorescence imaging because the intensity in the lamella and in the uncut tissue have comparable values. When a lamella has a thickness compatible with cryo-TEM, typically the intensity is 30% of the one measured in a focal plane within the uncut tissue. This value is in line with the difference between the average thickness of a lamella (~250 nm) and the depth of focus of the objective we use (~800 nm). Therefore, in the absence of an optical sectioning method such as confocal microscopy, the limited dynamic range of the detectors make imaging of thin lamellas challenging. Nevertheless, in order to show that this is possible, we prepared cryo-TEM compatible lamellas on Yeast a strain stained with a nuclear stain (Nuclear-ID, EnzoLife Sciences). In Figure 7 it is possible to see that fluorescence is detectable in a lamella, and further that the fluorescence signal can be used to target the cryo-ET data collection. In this case, the fluorescence was diffused through the nucleus and we used Cryo-ET analyses ( Figure 7e and Video 2 ) show that LM imaging does not induce de-vitrification of the sample.

( a–b ) S. cerevisiae cells were labelled with nucleolar staining, lamellas were prepared and then imaged using cryo-fluorescence microscopy. ( c ) Correlated image between the SEM and LM images, the region imaged through cryo-ET is labelled in the yellow square. ( d ) Cryo-TEM image of the lamella showing that the sample is still vitreous.

We have presented a novel system to perform cryo-CLEM and targeted cryo-lamella preparation, which simplifies sample handling and reduces the risk of sample contamination by limiting the number of cryo-transfers. Furthermore, having a light microscope within the vacuum chamber provides an opportunity to verify that the target has not been lost while milling. Future developments will likely involve the introduction of optical sectioning and super-resolution imaging. Further, having the light microscope on the same stage as the FIB resolves the current status where each microscope comes with its own image format and no stage information can be directly imported from the LM to the FIB/SEM. This configuration provides the foundation required for automated CLEM to mill targeted cryo-lamellae, which will enable structural studies on rare cellular events.

Considering future software requirements, it is reasonable to expect setups which include automated segmentation of correlated images to define or guide the definition of the milling parameters with limited user intervention. When using the conventional suite of separated microscopes it is challenging to achieve such a target because each microscope has its own image format and no stage information can be directly imported from the LM to the FIB/SEM. In the PIE-scope, all images are acquired using the same stage and software ( Figure 3b ), and fluorescence images can be acquired in between milling steps (therefore providing additional information useful for a more precise correlation, such as the milled edges). Through the PIE-scope, it will be therefore possible to implement CLEM automation for targeted lamella preparation.

The illumination source is a four wavelength combiner with a single multimode fibre output (Toptica iCLE-50 405, 488, 561, 640 nm). The choice between a multi-band-pass dichroic (Chroma 89402 m) and a Thorlabs 10/90 (R/T) beam splitter allows for fluorescence or reflected light imaging. The maximum laser power used during imaging corresponded to 0.5 mW out of the fibre, which was further reduced to 0.2 mW at the back-focal plane of the objective, although there might be occasions where more laser power is required for imaging, we noted that these conditions have provided sufficient illumination on all tested samples. Currently, the camera used is a Basler acA1920-155um with USB3.0 interface. This camera is a heated CMOS detector with a quantum efficiency in the visible range of ~70%, it represents a cost-effective solution which comes with a dark noise of 6.6 e ^- . The major advantage of this component, beside its price, is that it does not require active cooling, making the system less susceptible to vibrations. If single-molecule imaging is required, or if the fluorescence signal is extremely dim, then cooled CMOS cameras such as the Hamamatsu Flash4.0 V3 should be used. In this case, in order to ensure that no vibration is introduced in the system, the camera cooling lines should be connected to the FIB/SEM water chiller. The feedthrough from the atmosphere to the high-vacuum is realised through a glass window (Thorlabs VPCH42-A). The optical path is steered using three axes mirror holders embedded in the custom build microscope body/flange visible in Figure 2a . The focus drive is a high-vacuum-compatible piezoelectric linear positioner (Smaract SLC-2445HV). The objectives tested were Olympus MPLFNL 10x/0.3, MPLFNL 20x/0.45, UPFLN 20x/0.8 and MPLAPON 50x/0.95.

The microscope body and the adapters were machined at the Monash University Mechanical and Aerospace Engineering Workshop. All atmosphere parts were made of Aluminium alloy 7075. All vacuum parts were machined from Stainless Steel 316 l. Mounting of the microscope body on the vacuum chamber of the FIB/SEM occurs directly on the middle front GIS port (GIS2 on ThermoFisher instruments). The only strict requirement of this design is that neither a gas injection system (GIS) nor detector can be mounted on the front GIS ports at the FIB side of the chamber, we resolved this issue by repositioning our GISs from the flanges GIS1 and GIS2 to the rear flanges (GIS4 and GIS5). Although this can be a limitation for highly equipped material science microscopes, on systems where biological imaging and lamella cryo-preparation is the major focus this will not result in a problem. Also, with the advent of multi-line-GISs offered by most instrument manufacturers, it is extremely unlikely that all GIS flanges will be required.

All custom components were designed using Autodesk Inventor Pro 2016 and Siemens NX. Parts were fabricated in the Mechanical and Aerospace engineering mechanical workshop at Monash University. Ray tracing was performed using Zemax OpticsStudio. The control software has been developed using National Instruments LabVIEW 2015, Python three with Numpy 1.15, Scipy 1.1.0 and ThermoFisher AutoScript 4.0. All drawings, latest Bill of Material (matched to the stable software released), a stable software release, list of dependencies and instructions are available at https://www.demarco-lab.com/resources .

All Caenorhabditis elegans strain RJP3088 (Pets-5::mCherry +Pelt-2::GFP) ( Juozaityte et al., 2017 ) was maintained at 20°C on NGM plates seeded with Escherichia coli OP50 bacteria (Porta-de-la-Riva, Fontrodona et al.). Isolation of L1 larvae was performed as previously described ( Porta-de-la-Riva et al., 2012 ). Larvae were deposited on a holey carbon EM-grid and plunge-frozen in liquid ethane. C. elegans samples were kept in M9 medium, concentrated to a density of 3–4 worms / µL. 5 µL of solution were deposited on a Cu-300-mesh Quantifoil R-2/2 grid which was glow discharged for 30 s using a Pelco EasyGlow. The excess of medium was manually blotted using Whatman filter paper grade 595 (Merck) and vitrified by plunge freezing in liquid ethane.

Drosophila melanogaster used were a cross between Engrailed-Gal4 (BL30564) and UAS-eGFP (BL6874) from the Bloomington Drosophila Stock Center (Indiana). Flies were raised on standard media at 25°C. Drosophila brains at the third instar larval stage were dissected in cold phosphate buffered saline before being mounted in 20% (w/v) dextran solution prior vitrification. Samples were high-pressure frozen using a Leica EM-PACT2 on 200 µm deep planchettes. In order to expose the sample to the surface, therefore reducing the milling time required, we performed manual freeze-fracture by scraping the surface with a scalpel under liquid nitrogen. The difference in density between the dextran solution and the sample itself ensures an easy separation.

Saccharomyces cerevisiae cells were cultured at overnight 30°C in 2 YPD+. Cell growth was monitored through the measurement of the optical density. Cells were harvested when the optical density reached 0.6, washed in PBS and stained using Nuclear-ID (Enzo Life Sciences). 5 µL were deposited on a Cu-300-mesh Quantifoil R-2/2 grid which was previously glow discharged for 30 s in a Pelco EasyGlow. Grids were manually blotted for 5 s using Whatman 595 blotting pads and plunge-frozen in liquid ethane.

All significant data are included in this manuscript. The mechanical drawings for all of the components, the bill of material, and the control software have been made available through the laboratory webpage ( https://www.demarco-lab.com/resources ) and through Zenodo (10.5281/zenodo.3260173).

The following data sets were generated

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "PIE-scope: integrated cryo-correlative light and FIB/SEM microscopy" for consideration by eLife . Your article has been reviewed John Kuriyan as the Senior Editor, Sriram Subramaniam as Reviewing Editor, and three reviewers. The following individual involved in review of your submission have agreed to reveal their identity: Jürgen M Plitzko (Reviewer #1).

The reviewers have discussed the reviews with one another. You will see that all three reviewers find your work interesting but raise a number of concerns. Since each reviewer's concerns have a somewhat different flavor, we have chosen to send the individual reviews to you below without merging them. Your revised submission must satisfactorily address the major comments and concerns of the reviewers in order for the manuscript to be considered for publication in eLife .

Reviewer #1:

The manuscript "PIE-scope: integrated cryo-correlative light and FIB/SEM microscopy" corresponded by Alex de Marco describes a "triple-beam" instrument that includes photons, ions and electrons, referred to as a "PIE-scope" for the correlation and fabrication of cryo-FIB lamellae from frozen hydrated and high-pressure frozen samples for subsequent analysis by cryo-electron tomography (cryo-ET). The integration eliminates the multiple sample transfers that typically take place in the current cryo-ET workflow, which greatly simplifies its application. As a result, and through its retrofittable design, it can provide a way to make this workflow more accessible to other laboratories while increasing the overall throughput. The authors plausibly document their developments and achievements, and I regard this as important advance in this field. However, while the manuscript is direct and to the point, its writing appears somewhat rushed and misses details. In some places the writing is too technical (or jargon) and therefore it could be definitely improved to make it more accessible.

It is somewhat disappointing that there is no final TEM analysis showing that after correlation and sample preparation, cryo-ET can be successfully performed on the produced lamella from either of the two samples. The question is: Could this still be performed?

The paper by Gorelick et al., presents a new design for an integrated fluorescent and dual-beam microscope. This technological advance is a great addition to the correlative tools available for cellular cryo-electron tomography. The implementation described is simple and elegant and seems to work very well. In particular, the ability to target with a precision of ~500 nm is very encouraging.

I find the technology worthy of publishing, but I think that the manuscript requires significant revisions in order to target a wide readership, but more importantly more results need to be included to make sure it will be a useful tool for people interested in this technology. Besides needing many more details in the text, a manuscript like this should contain detailed supplementary data for the expert reader.

Reviewer #2:

1) Information necessary to reproduce the setup.a) The authors state that the software is constantly changing and can be made available from the authors. In order for anyone to implement this technology, they have to be sure that the work described here can be implemented outside of the authors' set up. Thus, a stable release of the software, with a detailed description of all the requirements, should be a necessary condition for publication.b) Along these lines, the drawings and all needed equipment for reproducing the set up should be available (can be upon request but needs to be explicitly stated).

2) The imaging is only done in 2D. The authors describe three positions in software: LM imaging, FIB imaging, and lamella preparation.a) How are the transformation of coordinates done for 2-D images between these imaging modes? Specifically, how is the fluorescence signal superimposed onto the tilted (FIB milling) position? Is the rotation simply done by a compression in the direction of the tilt?b) What is the SEM imaging position with respect to the others? A table, figure, or paragraph describing the relative distances and angles should be provided.c) When checking for the fluorescent signal at different stages of milling, is the sample tilted back so that it is orthogonal to the FIB beam, or is the tilted image used? Do the authors routinely use FIB imaging during milling, or is only the SEM used for this task?

3) In describing the sample preparation, the authors are not very explicit. A notable example is the lack of details in the brain sample. Was it high-pressure frozen? If it was plunge frozen, there is no chance that this sample will be vitrified. While the authors may argue that for the purposes of this manuscript this is not relevant, it should then be clearly stated.

4) The biggest gap in the manuscript is the characterization of the quality of the lamellae for cryo-ET. Each of the points are essential for the work to be useful:a) While the authors did not alter the cryo-stage, the authors should test and report on whether there are any alterations to the stability of the system in any way. Are there any new vibration or other issues that were introduced?b) Are there any added contamination rates from the introduction of the fluorescence set up? Without subsequent cryo-TEM images, it is hard to gauge the quality of the finished lamella.c) In fact, the finished lamella seems thick and cracked, and the thickness is not reported. Even if one would consider this work not to be about lamellae production, the detectability of the fluorescent signal for very thin lamellae should be demonstrated.d) A major issue with fluorescence imaging at cryogenic temperatures is the potential for devitrification. If the authors do not think this is a concern due to the low power of the illumination used, they should provide clear arguments. otherwise, they should show the vitrification of the lamella in the TEM after preparation in the PIE scope.

5) In their letter, the authors claim that this is a non-expensive set up ($50K). I think that as part of supplementary data they should provide a list of materials required and their approximate cost, which is customary when describing a new set up. For instance, what is the cost of ThermoFishder Autoscirpt?

Reviewer #3:

General assessment:

Very well written. Clearly described what has been reached and how the system provides a significant addition to the set of tools available for cell biological studies that require morphological/structural information of molecular structures in their cellular context.

The capability to perform the preparation of cryo-lamella using a focused ion beam SEM with an integrated light microscopy is very exciting as it can be used to ensure that the feature of interest is present in the lamella. This approach will improve possibilities to perform cell biological experiments in which targeted cryo electron tomography of fluorescently tagged structures are an essential ingredient.

It is very exciting that a proof-of-principle is provided on larvae of C. elegans on which the technological solution is evaluated. It is also exciting that the instrumentation is principle retrofittable and the software to make all this work for CLEM operation is open source.

Major concerns:

A major concern I have with the manuscript is that though it describes overall features and possibilities of the instrumentation with sufficient detail to illustrate the potential of its capabilities with the proof-of-principle, but that it lacks a discussion on some of the remaining technological bottlenecks of 3D CLEM on frozen hydrated specimen. In my view the steps that still need to be taken to meet the requirements to make the system applicable for a more wider audience are not mentioned with sufficient detail.

Subsection “Characterizing the system” summarizes the specifications of the system very nicely. Bottlenecks are the accurate 2D positioning of the lamella around a target (~200-400 nm, though is claimed that it could be ~~100 nm), the reduction in imaging quality due to drift (optical astigmatism due to drift, how large is the drift and can limitation be solved?) as well as the outlook for 3 D correlation to achieve an accuracy that would be sufficient for targeted FIB-milling (~500 nm, the need for deconvolution to improve the z-resolution can this indeed be improved, and to what accuracy?)

It would strengthen the paper considerably, if additional discussion/argumentation was giving detailing the possibility of a more accurate performance on next-generation systems.

Reviewer #1:

We re-worded multiple parts of the manuscript in order to make it more accessible. Please refer to the text changes in the re-submitted document.

The analysis could not be performed on the samples presented in the first version of the manuscript because neither sample are available. As pointed out by the reviewer in the minor comments the size can be considered on the extreme side. We therefore decided to add a figure where we used yeast cells with live cell imaging compatible nuclear staining. Here we show that the sample is still vitreous after imaging with the light microscope and that successful targeted cryo-ET can be performed (Figure 7 and Video 2).

Reviewer #2:

We added supplementary files including a link to the lab resource webpage (https://www.demarco-lab.com/resources) with the stable release of the control software with the installation requirements, a bill of material comprising the required and optional (configuration specific) components and the drawings of the custom components. The drawing provided include only the custom components because the FIB-SEM chamber is proprietary intellectual property of ThermoFisher. Upon written request of the model to ThermoFisher and upon written confirmation from ThermoFisher that disclosure is acceptable the authors will be able to disclose the full assembly should this be required.

We agree that although the focus on the paper is on the hardware rather than the image correlation, a more detailed description of the correlation procedure is beneficial to help the readers. Accordingly, we added a paragraph to detail the correlation procedure, further in the Video 1 the correlation procedure is shown.

In short, we provide a tool to perform the 2D correlation which is embedded in the PIE-scope control software, which is based on calculation of the rigid-body transformation between the imaging modalities. If 3D correlation is required users can call external CLEM software such as 3DCT (We decided against including this package in PIE-scope is due to the impossibility for us to support it). The Video 1 shows the correlation procedure.

b) What is the SEM imaging position with respect to the others? A table, figure, or paragraph describing the relative distances and angles should be provided.

We modified Figure 1 to include the information about the relative position and orientation between imaging all modalities.

c) When checking for the fluorescent signal at different stages of milling, is the sample tilted back so that it is orthogonal to the FIB beam, or is the tilted image used? Do the authors routinely use FIB imaging during milling, or is only the SEM used for this task?

For fluorescence imaging the sample must be orthogonal to the FIB in order to access the sample with the short working distance of the LM objective. The FIB is used for imaging only the milling pattern must be placed. In both Figure 5 and Figure 6 all images were acquired using the SEM.

We extended the description of the sample preparation and included detail on the vitrification.

4) The biggest gap in the manuscript is the characterization of the quality of the lamellae for cryo-ET. Each of the points are essential for the work to be useful:a. While the authors did not alter the cryo-stage, the authors should test and report on whether there are any alterations to the stability of the system in any way. Are there any new vibration or other issues that were introduced?

The introduction of the light microscope in its current configuration has no effects on vibrations because the laser combiner is fibre coupled to the instrument and the lasers are solid state diodes which can be passively cooled. The camera is a heated detector, therefore no vibration due to cooling is introduced. The cryo-stage we have on the FIB/SEM has a drift speed limit of 60 nm / min. When first installed it demonstrated to drift 20-30 nm / min depending on the environmental conditions (room temperature fluctuations, multiple people around the instrument, etc.). Since the installation of PIE-scope we have never experienced a drift rate greater than 30 nm / min. Accordingly, a discussion has been added.

b) Are there any added contamination rates from the introduction of the fluorescence set up? Without subsequent cryo-TEM images, it is hard to gauge the quality of the finished lamella.

We have not measured any change in the contamination rate due to the installation of the light microscope. The reason is linked to the fact that the vacuum parameters do not change as a consequence of the installation. All feedthroughs used are high vacuum compatible and their performance is measured in respect to the best obtainable vacuum in the FIB/SEM before and after installation of PIE-scope. Since the LM objective is not designed for operation under high vacuum the degassing is slower than the other components. We added a paragraph to discuss this point, explaining that cryo-experiments should not be conducted if the vacuum chamber has not been pumped for at least 20 hours. The contamination rate is increased as consequence of slow degassing. Once the ideal vacuum conditions have been reached (which can be easily measured through the vacuum gauge on the FIB/SEM or through a residual gas analyser (if available) the experiments can start.

c) In fact, the finished lamella seems thick and cracked, and the thickness is not reported. Even if one would consider this work not to be about lamellae production, the detectability of the fluorescent signal for very thin lamellae should be demonstrated.

In order to address this point, we added a figure (Figure 7) which shows the fluorescence detection on a thin lamella. The thickness is demonstrated also through cryo-TEM. We also added a small discussion about the challenges presented when imaging a thinned lamella. We also added a short discussion relative to this.

d) A major issue with fluorescence imaging at cryogenic temperatures is the potential for devitrification. If the authors do not think this is a concern due to the low power of the illumination used, they should provide clear arguments. otherwise, they should show the vitrification of the lamella in the TEM after preparation in the PIE scope.

We agree with the reviewer that this is a major point to consider. We have provided additional data that include the cryoTEM after preparation using the PIE-scope. We further added the measurement of the total laser power used during imaging in the Materials and methods section.

The cost of the hardware, including the lasers (which represent the vast majority if the cost) is below 50K USD (as described in the supplementary bill of material). It is worth mentioning that the cost can be reduced by ~20K USD if one choses to use a LED combiner or a mercury lamp with filters instead of the lasers (e.g. we identified the LED4D254 led combiner from Thorlabs as a good replacement, which including the driver module and light-guide will cost 5700 USD). Similarly, the cost can increase by 30K USD by integrating a more flexible laser combiner (such as the Coherent Galaxy) and a more sensitive camera such as the Hamamatsu Flash 4.0 V3. The integration using AutoScript is not mandatory, although it provides better control of the instrument, making the process smoother and the correlation faster. The average price of AutoScript is ~ 20K USD.

A license of LabVIEW is advisable (most universities have a site license) as it allows for further customization and development of the software, but it is not required for basic use. We provide both the source code and the compiled version of the stable release which only requires downloading of the LabVIEW runtime (which is free).

Reviewer #3:

General assessment:

It is very exciting that a proof-of-principle is provided on larvae of C. elegans on which the technological solution is evaluated. It is also exciting that the instrumentation is principle retrofittable and the software to make all this work for CLEM operation is open source.

Major concerns:

It would strengthen the paper considerably, if additional discussion/argumentation was giving detailing the possibility of a more accurate performance on next-generation systems.

The effect of astigmatism due to thermal drift, as described by the reviewer can become a major limitation and can result in a loss of resolution up to 9% relative to the expected diffraction limit of the optical setup. Accordingly, we improved the mechanical design of the microscope (the drawings provided in the supplementary material are the latest and include the improvements) and therefore increased stability of the atmospheric section of the light microscope. The changes in the astigmatism over time detected previously are not measurable anymore.

We modified the Discussion section to match the current latest design, added a description of the best alignment performance of the optical path. We added the discussion of the improvements that could lead to better 3D correlation.

ARC Centre of Excellence in Advanced Molecular Imaging, Monash University, Clayton, Australia

Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia

Contribution

Data curation, Software, Formal analysis, Validation, Investigation, Writing—review and editing, Collected and analysed the data, coded the control software, conceived and built the setup

Competing interests

No competing interests declared

ARC Centre of Excellence in Advanced Molecular Imaging, Monash University, Clayton, Australia

Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia

Contribution

Software, Visualization, Methodology, Writing—review and editing, Coded the control software

Competing interests

No competing interests declared Department of Anatomy and Developmental Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia

Contribution

Investigation, Writing—review and editing, Prepared the samples

Competing interests

No competing interests declared

ARC Centre of Excellence in Advanced Molecular Imaging, Monash University, Clayton, Australia

Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia

Contribution

Investigation, Prepared the samples

Competing interests

No competing interests declared

ARC Centre of Excellence in Advanced Molecular Imaging, Monash University, Clayton, Australia

Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia

University of Warwick, Coventry, United Kingdom

EMBL Australia, Clayton, Australia

Contribution

Funding acquisition, Writing—review and editing

Competing interests

No competing interests declared Department of Anatomy and Developmental Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia

Contribution

Formal analysis, Investigation, Writing—review and editing

Competing interests

No competing interests declared

ARC Centre of Excellence in Advanced Molecular Imaging, Monash University, Clayton, Australia

Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Clayton, Australia

University of Warwick, Coventry, United Kingdom

Contribution

Conceptualization, Supervision, Funding acquisition, Writing—original draft, Project administration, Collected and analysed the data, conceived and built the setup

For correspondence

[email protected]

Competing interests

No competing interests declared

Received: February 8, 2019

Accepted: June 26, 2019

Accepted Manuscript published: July 1, 2019 (version 1)

Version of Record published: July 4, 2019 (version 2)

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

The transcriptional regulator SsrB acts as a switch between virulent and biofilm lifestyles of non-typhoidal Salmonella enterica serovar Typhimurium. During infection, phosphorylated SsrB activates genes on Salmonella Pathogenicity Island-2 (SPI-2) essential for survival and replication within the macrophage. Low pH inside the vacuole is a key inducer of expression and SsrB activation. Previous studies demonstrated an increase in SsrB protein levels and DNA-binding affinity at low pH; the molecular basis was unknown (Liew et al., 2019). This study elucidates its underlying mechanism and in vivo significance. Employing single-molecule and transcriptional assays, we report that the SsrB DNA binding domain alone (SsrBc) is insufficient to induce acid pH-sensitivity. Instead, His12, a conserved residue in the receiver domain, confers pH sensitivity to SsrB allosterically. Acid-dependent DNA binding was highly cooperative, suggesting a new configuration of SsrB oligomers at SPI-2-dependent promoters. His12 also plays a role in SsrB phosphorylation; substituting His12 reduced phosphorylation at neutral pH and abolished pH-dependent differences. Failure to flip the switch in SsrB renders Salmonella avirulent and represents a potential means of controlling virulence.

Ca ²⁺ /calmodulin-dependent protein kinase II (CaMKII) is essential for long-term potentiation (LTP) of excitatory synapses that is linked to learning and memory. In this study, we focused on understanding how interactions between CaMKIIα and the actin-crosslinking protein α-actinin-2 underlie long-lasting changes in dendritic spine architecture. We found that association of the two proteins was unexpectedly elevated within 2 minutes of NMDA receptor stimulation that triggers structural LTP in primary hippocampal neurons. Furthermore, disruption of interactions between the two proteins prevented the accumulation of enlarged mushroom-type dendritic spines following NMDA receptor activation. α-Actinin-2 binds to the regulatory segment of CaMKII. Calorimetry experiments, and a crystal structure of α-actinin-2 EF hands 3 and 4 in complex with the CaMKII regulatory segment, indicate that the regulatory segment of autoinhibited CaMKII is not fully accessible to α-actinin-2. Pull-down experiments show that occupation of the CaMKII substrate-binding groove by GluN2B markedly increases α-actinin-2 access to the CaMKII regulatory segment. Furthermore, in situ labelling experiments are consistent with the notion that recruitment of CaMKII to NMDA receptors contributes to elevated interactions between the kinase and α-actinin-2 during structural LTP. Overall, our study provides new mechanistic insight into the molecular basis of structural LTP and reveals an added layer of sophistication to the function of CaMKII.

eLife is a non-profit organisation inspired by research funders and led by scientists. Our mission is to help scientists accelerate discovery by operating a platform for research communication that encourages and recognises the most responsible behaviours in science. eLife Sciences Publications, Ltd is a limited liability non-profit non-stock corporation incorporated in the State of Delaware, USA, with company number 5030732, and is registered in the UK with company number FC030576 and branch number BR015634 at the address: eLife Sciences Publications, Ltd
Westbrook Centre, Milton Road
Cambridge CB4 1YG