In situ X-ray-assisted electron microscopy staining for large biological samples
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This study explores the kinetics of heavy metal staining of tissue using time-lapse imaging with X-ray micro computed tomography (CT). It will be of interest to the wide community of scientists preparing biological samples for electron microscopy (EM), in particular large-volume EM. While at present the relation between CT imaging and EM contrast remains to be quantified, this study has the potential to become a reference for the field in establishing a quantitative tool for assessing and developing staining protocols.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Electron microscopy of biological tissue has recently seen an unprecedented increase in imaging throughput moving the ultrastructural analysis of large tissue blocks such as whole brains into the realm of the feasible. However, homogeneous, high-quality electron microscopy staining of large biological samples is still a major challenge. To date, assessing the staining quality in electron microscopy requires running a sample through the entire staining protocol end-to-end, which can take weeks or even months for large samples, rendering protocol optimization for such samples to be inefficient. Here, we present an in situ time-lapsed X-ray-assisted staining procedure that opens the ‘black box’ of electron microscopy staining and allows observation of individual staining steps in real time. Using this novel method, we measured the accumulation of heavy metals in large tissue samples immersed in different staining solutions. We show that the measured accumulation of osmium in fixed tissue obeys empirically a quadratic dependence between the incubation time and sample size. We found that potassium ferrocyanide, a classic reducing agent for osmium tetroxide, clears the tissue after osmium staining and that the tissue expands in osmium tetroxide solution, but shrinks in potassium ferrocyanide reduced osmium solution. X-ray-assisted staining gave access to the in situ staining kinetics and allowed us to develop a diffusion-reaction-advection model that accurately simulates the measured accumulation of osmium in tissue. These are first steps towards in silico staining experiments and simulation-guided optimization of staining protocols for large samples. Hence, X-ray-assisted staining will be a useful tool for the development of reliable staining procedures for large samples such as entire brains of mice, monkeys, or humans.
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
In this work, Ströh et al. characterize the kinetics of osmium tetroxide staining of soft mouse brain tissue samples, the first step in many protocols aimed to prepare samples for electron microscopy imaging. The authors used time-lapsed single-projection X-ray images of the sample immersed in the staining solution to monitor the staining process. They have then been able to not only accurately model osmium tetroxide diffusion in the tissue across time and depth, but also to compare the performance of osmium tetroxide to other commonly used first reagents: osmium reduced in potassium ferrocyanide and the same reduced osmium in formamide. Overall, they provide a clear insight on the kinetics of osmium diffusion in tissue - obeying a long-established quadratic law - while also provide clear …
Author Response
Reviewer #1 (Public Review):
In this work, Ströh et al. characterize the kinetics of osmium tetroxide staining of soft mouse brain tissue samples, the first step in many protocols aimed to prepare samples for electron microscopy imaging. The authors used time-lapsed single-projection X-ray images of the sample immersed in the staining solution to monitor the staining process. They have then been able to not only accurately model osmium tetroxide diffusion in the tissue across time and depth, but also to compare the performance of osmium tetroxide to other commonly used first reagents: osmium reduced in potassium ferrocyanide and the same reduced osmium in formamide. Overall, they provide a clear insight on the kinetics of osmium diffusion in tissue - obeying a long-established quadratic law - while also provide clear insight on how osmium concentration in the sample rises above its concentration in the staining solution. Finally, the authors also manage to put in perspective the effects of osmium reduction on the osmium staining of the tissue. Their results showcase that osmium reduction triggers a washout of the osmium in the sample and not only counteracts an osmium-triggered sample expansion but also manages to reverse its sign, resulting in sample shrinkage and even leading to sample degradation if left for long periods of time (evident after several tens of hours).
One minor weakness of the manuscript is that it does not characterize the presence of osmium in the tissue after the water washes that typically follow osmium staining. That would provide a valuable control for the interpretation of the potassium ferrocyanide-triggered osmium washout. Also, it would provide a valuable insight on the presence of bound osmium in the sample at the moment of starting the next staining step in the protocol, which would facilitate escalating the use their approach to modularly optimize complex heavy metal soft tissue staining protocols consistent of multiple successive steps.
We added a supplementary figure 8 and 9a that show the dynamics of heavy metal washout in double distilled H2O for 22 hours after 22 hrs of incubation in 2% buffered OsO4. First, the sample was quickly flushed with double-distilled H2O four times to completely remove any buffered OsO4 solution. Subsequently, the sample was immersed in double distilled H2O for 22 hrs. The accumulation of heavy metals decreases by about 6% in a depth of 100-1200μm. Our hypothesis is that this can be explained by unbound osmium that diffuses out of the sample and by the slight expansion of the sample (Supplementary Figure 9b). But in contrast to the washout effect in K4[Fe(CN)6] (Supplementary Figure 12c), the reduction in heavy metal density appears to be small.
Reviewer #2 (Public Review):
The investigators studied kinetics of Osmium Tetroxide diffusion in large chemical fixed biological samples. So far it has never been monitored so accurately. The use of micro CT scan images gives good insight in what is happening inside tissue blocks. The technical designed approach and mathematical analysis of the data, result in achieving the goal of opening the black box of staining. Other labs might use this X-ray method to understand their -sometimes very specific- conventional electron microscopy sample preparation protocols.
The data shows accumulation of OsO4 in 4mm brain tissue blocks. Quantification of absorption intensity proves a quadratic dependence of time and sample size. OSO4 shows a homogeneous distribution after 20h in contrast to reduced osmium what resulted in a heterogeneous distribution and a high intensity band at 300-880 µm depth.
Adding formamide to reduced Osmium gives a more homogeneous spreading but a side effect of long incubation with Formamide is 10-15% expansion of the tissue (reduced Osmium alone shrinks 5%, Osmium alone expands 5%).
To overcome heterogenous spreading of reduced osmium the reagents were separated: the 1st osmium step was followed by a 2nd reducing ferrocyanide step. Surprisingly this led to wash-out of Osmium from the sample and therefore not useful.
The authors used equations and simulations to develop a diffusion-reaction-advection model. Four coupled processes of diffusion, binding, unmasking and expansion are described to explain the staining reaction.
The future goal of this paper is to set up an in-silico model which can be used for e.g. precious samples and predicts processes in different type of samples. A lot more work needs to be done to get that far though since many more steps are involved in the sample preparation for electron microscopy to get decent morphology. Variations of tissue, cells, species, protocols, imaging techniques are numerous. To create an "one fits all model" is very ambitious.
Strengths:
The results are very well documented and the use of micro CT to monitor chemical processes will be useful to other laboratories to better understand complex sample preparation steps. It will certainly be used by others to adapt their protocols to specific specimens.
Experiments were done consistently and accurately.
Both the introduction and discussion are supported by thorough literature search, which build a thorough reference for laboratories interested in sample preparation for electron microscopy.
Of specific interest are the reported effect of the commonly used osmium mixes on the overall tissue topology and the unexpected shrinkage/swellings. These undoubtedly will raise awareness in the community and should trigger careful (re)consideration of established protocols.
This work could represent a stepping stone for those laboratories studying the ultrastructure of large specimens, in particular (but not exclusively) the neurobiology community.
Weaknesses:
The study is done on brain tissue which is heterogenous and might make the extrapolations quite unprecise when modeling. It might have been easier to work with a more homogeneous samples like liver.
On the tissue type as well. It would be interesting to have some data from other tissue types in order to extract how different various tissues would behave in terms of osmium penetration. Yet this might be slightly beyond the scope of this article.
The fitted parameters of the diffusion-reaction-advection model (Supplementary Figure 7) capture the measured diffusion-reaction-advection kinetics of osmium staining in brain tissue well (Figure 2c, Supplementary Figure 1; residual standard error SEres=0.026 ± 0.005, mean ± s.d.), indicating that the heterogeneity of the brain tissue is not a major constraint for the accuracy of the presented model. But we do agree with the reviewer that the diffusion/reaction-advection kinetics are likely different in different tissue types. To illustrate this, we measured and modeled the staining kinetics of 2% buffered OsO4 in 4mm punches of liver tissue (Supplementary Figure 13). Interestingly, the effective diffusion coefficient appears to be >4X larger in liver tissue compared to brain tissue.
Whilst the penetration of osmium is very important to achieve a good preservation of tissues and a good and homogenous contrast for EM, this step is also very delicate, as it can lead to tissue damages, especially when the reaction is not controlled. It is known, for a long time, that osmium can cause precipitates, loss of components (e.g. cytoskeleton) or even tissue destruction. One way to mitigate this has been to perform the osmium fixation at low temperatures, e.g. on ice. Yet, the authors don't report the temperature at which they performed their experiment. It is assumed that they worked at room temperature, whatever it could be within the Versa. This should be documented.
All staining and washing steps have indeed been done at room temperature. We added that information accordingly in the sample preparation section of the methods.
Moreover, and in line with the previous comment, it seems very important, if not crucial for this study, to thoroughly document the effect of long term exposure to osmium on the tissue integrity, at the ultrastructural level. The authors should perform the full workflow, i.e. down to the EM analysis, not necessarily on the full time series but at least on key timepoints. Assessment of various key components, e.g. synaptic structure, myelin sheets integrity, visibility of organelles, microtubules etc. would be very important. Indeed, what would be the interest of a 20 hours incubation in osmium if this would lead to a loss of the fine subcellular organization?
We added Supplementary Figure 14 showing the ultrastructural preservation after 20 hours of incubation in OsO4.
Another point that might be interesting to investigate and report on would be the potential damages caused by X-ray irradiation over long time periods. Does this interfere with the stability of the osmium solution? Of the sample itself?
This is a very important point which has to be kept in mind for every staining solution and procedure studied by X-rays. For the staining solutions presented here, we did not find any difference in the qualitative appearance of the samples or the staining solutions with or without X-ray exposure.
I am not able to comment of the modeling part itself, but it seems that the diffusion-reaction-advection model is based on many assumptions e.g. tissue density, isotropic expansion, homogeneous diffusion medium. The validation on experimental brain sample looks convincing, but it would be interesting to check how these could be generalized to a larger spectrum of biological material.
As the reviewer points out, the fitted parameters of the diffusion-reaction-advection model (Supplementary Figure 7) capture the measured diffusion-reaction-advection kinetics of osmium staining in brain tissue well (Figure 2c, Supplementary Figure 1; residual standard error SEres=0.026 ± 0.005, mean ± s.d.), indicating that the heterogeneity of the brain tissue is not a major constraint for the accuracy of the presented model. But we do agree with the reviewer that the diffusion/reaction-advection kinetics are likely different in different tissue types. To illustrate this, we measured and modeled the staining kinetics of 2% buffered OsO4 in 4mm punches of liver tissue (Supplementary Figure 13). Interestingly, the effective diffusion coefficient appears to be >4X larger in liver tissue compared to brain tissue.
Reviewer #3 (Public Review):
Ströh et al use time-lapse X-ray imaging to monitor the diffusion of heavy metal stains into large brain samples. Uniform staining of large (thicker than 1 mm) tissue samples is a prerequisite for future whole-brain 3D EM reconstructions of synaptic connectivity. Until now, staining optimization has essentially been achieved through trial and error. The reported approach allows the rapid measurement of staining gradients and the determination of diffusion rates within tissue specimens. This offers the possibility to modify staining parameters with a more rapid turn-around. The authors develop a diffusion/binding model to describe the occupancy of free and masked osmium binding sites and fit the model parameters to the diffusion of osmium solution. The authors also demonstrate that an approach that separates the osmium staining and reduction steps seems to counterintuitively 'washout' the osmium in the tissue.
While the approach seems promising as a diagnostic tool and offers a principled approach to gaining a better understanding of staining processes, a weakness is the lack of a demonstration that the x-ray imaged staining gradients correlate with what is actually observed under the electron microscope. For example, the figures show that reduced osmium stains tissue with a maximum intensity of ~1.1 (a.u.) compared to osmium alone at ~0.9 (a.u.). Because these intensities are not calibrated against the appearance of the staining in EM sections, their interpretation is limited.
We added Supplementary Figure 14 showing the ultrastructural preservation after 20 hours of incubation in OsO4 and Supplementary Figure 15 that shows the covariation of the EM and X-ray pixel intensities. Note, in the presented study we acquired X-ray projection images that represent the cumulative tissue and heavy metal density along the direction of projection through a 4 mm thick tissue punch. Therefore our X-ray projection pixel intensities cannot directly be compared to the EM pixel intensity of a thin section. As has been shown previously (see for example Figure 4b in Mikula & Denk 2015), in computed X-ray tomograms the intensity scales linearly with EM intensity, if the pixel intensity in an EM section is compared to the corresponding reslice of a computed X-ray tomogram.
-
Evaluation Summary:
This study explores the kinetics of heavy metal staining of tissue using time-lapse imaging with X-ray micro computed tomography (CT). It will be of interest to the wide community of scientists preparing biological samples for electron microscopy (EM), in particular large-volume EM. While at present the relation between CT imaging and EM contrast remains to be quantified, this study has the potential to become a reference for the field in establishing a quantitative tool for assessing and developing staining protocols.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)
-
Reviewer #1 (Public Review):
In this work, Ströh et al. characterize the kinetics of osmium tetroxide staining of soft mouse brain tissue samples, the first step in many protocols aimed to prepare samples for electron microscopy imaging. The authors used time-lapsed single-projection X-ray images of the sample immersed in the staining solution to monitor the staining process. They have then been able to not only accurately model osmium tetroxide diffusion in the tissue across time and depth, but also to compare the performance of osmium tetroxide to other commonly used first reagents: osmium reduced in potassium ferrocyanide and the same reduced osmium in formamide. Overall, they provide a clear insight on the kinetics of osmium diffusion in tissue - obeying a long-established quadratic law - while also provide clear insight on how …
Reviewer #1 (Public Review):
In this work, Ströh et al. characterize the kinetics of osmium tetroxide staining of soft mouse brain tissue samples, the first step in many protocols aimed to prepare samples for electron microscopy imaging. The authors used time-lapsed single-projection X-ray images of the sample immersed in the staining solution to monitor the staining process. They have then been able to not only accurately model osmium tetroxide diffusion in the tissue across time and depth, but also to compare the performance of osmium tetroxide to other commonly used first reagents: osmium reduced in potassium ferrocyanide and the same reduced osmium in formamide. Overall, they provide a clear insight on the kinetics of osmium diffusion in tissue - obeying a long-established quadratic law - while also provide clear insight on how osmium concentration in the sample rises above its concentration in the staining solution. Finally, the authors also manage to put in perspective the effects of osmium reduction on the osmium staining of the tissue. Their results showcase that osmium reduction triggers a washout of the osmium in the sample and not only counteracts an osmium-triggered sample expansion but also manages to reverse its sign, resulting in sample shrinkage and even leading to sample degradation if left for long periods of time (evident after several tens of hours).
One minor weakness of the manuscript is that it does not characterize the presence of osmium in the tissue after the water washes that typically follow osmium staining. That would provide a valuable control for the interpretation of the potassium ferrocyanide-triggered osmium washout. Also, it would provide a valuable insight on the presence of bound osmium in the sample at the moment of starting the next staining step in the protocol, which would facilitate escalating the use their approach to modularly optimize complex heavy metal soft tissue staining protocols consistent of multiple successive steps.
-
Reviewer #2 (Public Review):
The investigators studied kinetics of Osmium Tetroxide diffusion in large chemical fixed biological samples. So far it has never been monitored so accurately. The use of micro CT scan images gives good insight in what is happening inside tissue blocks. The technical designed approach and mathematical analysis of the data, result in achieving the goal of opening the black box of staining. Other labs might use this X-ray method to understand their -sometimes very specific- conventional electron microscopy sample preparation protocols.
The data shows accumulation of OsO4 in 4mm brain tissue blocks. Quantification of absorption intensity proves a quadratic dependence of time and sample size. OSO4 shows a homogeneous distribution after 20h in contrast to reduced osmium what resulted in a heterogeneous …
Reviewer #2 (Public Review):
The investigators studied kinetics of Osmium Tetroxide diffusion in large chemical fixed biological samples. So far it has never been monitored so accurately. The use of micro CT scan images gives good insight in what is happening inside tissue blocks. The technical designed approach and mathematical analysis of the data, result in achieving the goal of opening the black box of staining. Other labs might use this X-ray method to understand their -sometimes very specific- conventional electron microscopy sample preparation protocols.
The data shows accumulation of OsO4 in 4mm brain tissue blocks. Quantification of absorption intensity proves a quadratic dependence of time and sample size. OSO4 shows a homogeneous distribution after 20h in contrast to reduced osmium what resulted in a heterogeneous distribution and a high intensity band at 300-880 µm depth.
Adding formamide to reduced Osmium gives a more homogeneous spreading but a side effect of long incubation with Formamide is 10-15% expansion of the tissue (reduced Osmium alone shrinks 5%, Osmium alone expands 5%).
To overcome heterogenous spreading of reduced osmium the reagents were separated: the 1st osmium step was followed by a 2nd reducing ferrocyanide step. Surprisingly this led to wash-out of Osmium from the sample and therefore not useful.
The authors used equations and simulations to develop a diffusion-reaction-advection model. Four coupled processes of diffusion, binding, unmasking and expansion are described to explain the staining reaction.
The future goal of this paper is to set up an in-silico model which can be used for e.g. precious samples and predicts processes in different type of samples. A lot more work needs to be done to get that far though since many more steps are involved in the sample preparation for electron microscopy to get decent morphology. Variations of tissue, cells, species, protocols, imaging techniques are numerous. To create an "one fits all model" is very ambitious.
Strengths:
The results are very well documented and the use of micro CT to monitor chemical processes will be useful to other laboratories to better understand complex sample preparation steps. It will certainly be used by others to adapt their protocols to specific specimens.
Experiments were done consistently and accurately.
Both the introduction and discussion are supported by thorough literature search, which build a thorough reference for laboratories interested in sample preparation for electron microscopy.
Of specific interest are the reported effect of the commonly used osmium mixes on the overall tissue topology and the unexpected shrinkage/swellings. These undoubtedly will raise awareness in the community and should trigger careful (re)consideration of established protocols.
This work could represent a stepping stone for those laboratories studying the ultrastructure of large specimens, in particular (but not exclusively) the neurobiology community.Weaknesses:
The study is done on brain tissue which is heterogenous and might make the extrapolations quite unprecise when modeling. It might have been easier to work with a more homogeneous samples like liver.
On the tissue type as well. It would be interesting to have some data from other tissue types in order to extract how different various tissues would behave in terms of osmium penetration. Yet this might be slightly beyond the scope of this article.Whilst the penetration of osmium is very important to achieve a good preservation of tissues and a good and homogenous contrast for EM, this step is also very delicate, as it can lead to tissue damages, especially when the reaction is not controlled. It is known, for a long time, that osmium can cause precipitates, loss of components (e.g. cytoskeleton) or even tissue destruction. One way to mitigate this has been to perform the osmium fixation at low temperatures, e.g. on ice. Yet, the authors don't report the temperature at which they performed their experiment. It is assumed that they worked at room temperature, whatever it could be within the Versa. This should be documented.
Moreover, and in line with the previous comment, it seems very important, if not crucial for this study, to thoroughly document the effect of long term exposure to osmium on the tissue integrity, at the ultrastructural level. The authors should perform the full workflow, i.e. down to the EM analysis, not necessarily on the full time series but at least on key timepoints. Assessment of various key components, e.g. synaptic structure, myelin sheets integrity, visibility of organelles, microtubules etc. would be very important. Indeed, what would be the interest of a 20 hours incubation in osmium if this would lead to a loss of the fine subcellular organization?
Another point that might be interesting to investigate and report on would be the potential damages caused by X-ray irradiation over long time periods. Does this interfere with the stability of the osmium solution? Of the sample itself?
I am not able to comment of the modeling part itself, but it seems that the diffusion-reaction-advection model is based on many assumptions e.g. tissue density, isotropic expansion, homogeneous diffusion medium. The validation on experimental brain sample looks convincing, but it would be interesting to check how these could be generalized to a larger spectrum of biological material.
-
Reviewer #3 (Public Review):
Ströh et al use time-lapse X-ray imaging to monitor the diffusion of heavy metal stains into large brain samples. Uniform staining of large (thicker than 1 mm) tissue samples is a prerequisite for future whole-brain 3D EM reconstructions of synaptic connectivity. Until now, staining optimization has essentially been achieved through trial and error. The reported approach allows the rapid measurement of staining gradients and the determination of diffusion rates within tissue specimens. This offers the possibility to modify staining parameters with a more rapid turn-around. The authors develop a diffusion/binding model to describe the occupancy of free and masked osmium binding sites and fit the model parameters to the diffusion of osmium solution. The authors also demonstrate that an approach that separates …
Reviewer #3 (Public Review):
Ströh et al use time-lapse X-ray imaging to monitor the diffusion of heavy metal stains into large brain samples. Uniform staining of large (thicker than 1 mm) tissue samples is a prerequisite for future whole-brain 3D EM reconstructions of synaptic connectivity. Until now, staining optimization has essentially been achieved through trial and error. The reported approach allows the rapid measurement of staining gradients and the determination of diffusion rates within tissue specimens. This offers the possibility to modify staining parameters with a more rapid turn-around. The authors develop a diffusion/binding model to describe the occupancy of free and masked osmium binding sites and fit the model parameters to the diffusion of osmium solution. The authors also demonstrate that an approach that separates the osmium staining and reduction steps seems to counterintuitively 'washout' the osmium in the tissue.
While the approach seems promising as a diagnostic tool and offers a principled approach to gaining a better understanding of staining processes, a weakness is the lack of a demonstration that the x-ray imaged staining gradients correlate with what is actually observed under the electron microscope. For example, the figures show that reduced osmium stains tissue with a maximum intensity of ~1.1 (a.u.) compared to osmium alone at ~0.9 (a.u.). Because these intensities are not calibrated against the appearance of the staining in EM sections, their interpretation is limited.
-
