Periodic Reporting for period 3 - IMAGEOMICS (Imaging the proteome at the nanoscale)
Reporting period: 2023-07-01 to 2024-12-31
The key principle of biological imaging is specific labeling. The protein of interest is revealed by tagging with fluorophores, either by genetic encoding or by specific affinity labeling. This procedure has the great disadvantage that each protein needs to be tagged individually. This has so far stopped imaging from becoming an “omics” approach, for large numbers of targets. We proposed to change this concept in our project, by inventing an imageomics approach. We planned here to rely on small affinity probes, nanobodies, that bind amino acid sequences (peptides) that are present in multiple proteins. We applied the nanobodies to biological samples in a combinatorial fashion, and we imaged them at the nanoscale, with a resolution that is sufficient to image single proteins (in the best implementations, better than 1 nm). Proteins were identified by the binding of a unique combination of affinity probes, corresponding to their specific sequences.
In brief, the company NanoTag Biotechnologies (NTB) generated the initial nanobody candidates, which were tested in different fashions, by both the microfluidics laboratory at Bar-Ilan University (BIU) and the biology laboratory at the University Medical Center Göttingen (UMG). In parallel, the optics developers at the KTH Royal Institute of Technology (KTH) generated microscope hardware and software that can be employed for excellent imaging of nanobodies. To obtain optimal resolution, the UMG team generated and optimized expansion microscopy procedures to enable efficient imaging of nanobodies. Most of these technical advancements were published as articles, with several raising substantial interest in the community. Moreover, the UMG team optimized the nanobody treatment within samples, their delivery and removal, together with BIU, enabling KTH and UMG to perform combinatorial imaging of nanobodies.
In brief, the company NanoTag Biotechnologies (NTB) generated the initial nanobody candidates, which were tested in different fashions, by both the microfluidics laboratory at Bar-Ilan University (BIU) and the biology laboratory at the University Medical Center Göttingen (UMG). In parallel, the optics developers at the KTH Royal Institute of Technology (KTH) generated microscope hardware and software that can be employed for excellent imaging of nanobodies. To obtain optimal resolution, the UMG team generated and optimized expansion microscopy procedures to enable efficient imaging of nanobodies. Most of these technical advancements were published as articles, with several raising substantial interest in the community. Moreover, the UMG team optimized the nanobody treatment within samples, their delivery and removal, together with BIU, enabling KTH and UMG to perform combinatorial imaging of nanobodies.
We initially required the generation of nanobodies for numerous peptides that are present in more than one protein, to enable the combinatorial identification of large numbers of proteins. We injected alpacas with a fusion protein comprising 20 different target peptides, along with a high number of off-target and near-target peptides. A total of 46 non-redundant nanobody clones were identified.
We then proceeded to validate the nanobodies in an extensive fashion, by mass spectrometry.
As we suggested in the initial description of our project, the nanobodies should recognize different but overlapping proteins. This is precisely what we observed. The nanobodies have varying levels of overlap, with some recognizing more than 300 common proteins, while others overlap very little.
The protein identities are typically unique, meaning that one protein is observed by only one set of nanobodies, for more than 90% of all proteins. Proteins are typically bound by ~10 nanobodies, although a broad distribution is observed, as predicted in our initial grant application, with some proteins bound by all 42 nanobodies tested. Most erroneous identifications are observed for proteins that are only bound by one nanobody.
Even before the nanobodies are fully available, we needed to ensure that we are able to screen and characterize the nanobodies for the project. The BIU partners have therefore first developed a microfluidic device that can automate the screening and perform a binding assay to characterize the affinity of the different antibodies to their targets and their cross reactivity. The ultimate goal is to quantify the binding of each of the candidate nanobodies to a particular peptide.
As planned, we have tested the optimal conditions for this quantification, and we prepared the screening technology, based on microfluidic devices. We originally planned to use biotinylated peptides, immobilized to the device surface via Neutravidin, which are then detected using fluorescent nanobodies. Such a screen can be multiplexed easily by using small mixes of nanobodies, and later by iterating the screen on the best-binding mixes. We have finished establishing the assay in 2022.
To obtain higher resolution, we turned to expansion microscopy, where our progress has been far more than previously envisioned, as we (UMG team) have obtained resolutions of 1 nm or better. A major gap still exists in the imaging field, between the precise analysis of single proteins by cryo-EM, at Ångstrom resolutions, and the analysis of cellular samples by efficient optical super-resolution (≥10 nm). Optical microscopy should be able to fill this gap, but it is limited by two fundamental problems. First, the achievable structural resolution in biological samples is determined by the labeling density, which is limited by the size of the fluorescent probe (several nanometers) and by the labeling efficiency. Second, fluorophores can interact via energy transfer at sub-10 nm distances, which results in accelerated photoswitching (blinking) and photobleaching, and thus in substantially lower localization probabilities.
The simple solution to this question is to separate the fluorophores spatially, without changing the labeling efficiency. This is best achieved by the physical expansion of the specimen, after embedding it in a swellable gel, in ExM. We implemented this approach, which we termed optimized nanoscale expansion (ONE) microscopy. To improve ONE and ExM resolution, the KTH laboratory also introduced a setup capable of imaging large gels, using sample- and signa-adapted imaging approaches. The details of this work were published in several articles, throughout the project.
We next proceeded to the proof-of-principle demonstration of imaging all of our selected nanobodies. GFP-carrying nanobodies were used, were applied to expanded gels immobilized on coverslips, and resulted in a complex, combined image, as shown in the attached file.
We then proceeded to validate the nanobodies in an extensive fashion, by mass spectrometry.
As we suggested in the initial description of our project, the nanobodies should recognize different but overlapping proteins. This is precisely what we observed. The nanobodies have varying levels of overlap, with some recognizing more than 300 common proteins, while others overlap very little.
The protein identities are typically unique, meaning that one protein is observed by only one set of nanobodies, for more than 90% of all proteins. Proteins are typically bound by ~10 nanobodies, although a broad distribution is observed, as predicted in our initial grant application, with some proteins bound by all 42 nanobodies tested. Most erroneous identifications are observed for proteins that are only bound by one nanobody.
Even before the nanobodies are fully available, we needed to ensure that we are able to screen and characterize the nanobodies for the project. The BIU partners have therefore first developed a microfluidic device that can automate the screening and perform a binding assay to characterize the affinity of the different antibodies to their targets and their cross reactivity. The ultimate goal is to quantify the binding of each of the candidate nanobodies to a particular peptide.
As planned, we have tested the optimal conditions for this quantification, and we prepared the screening technology, based on microfluidic devices. We originally planned to use biotinylated peptides, immobilized to the device surface via Neutravidin, which are then detected using fluorescent nanobodies. Such a screen can be multiplexed easily by using small mixes of nanobodies, and later by iterating the screen on the best-binding mixes. We have finished establishing the assay in 2022.
To obtain higher resolution, we turned to expansion microscopy, where our progress has been far more than previously envisioned, as we (UMG team) have obtained resolutions of 1 nm or better. A major gap still exists in the imaging field, between the precise analysis of single proteins by cryo-EM, at Ångstrom resolutions, and the analysis of cellular samples by efficient optical super-resolution (≥10 nm). Optical microscopy should be able to fill this gap, but it is limited by two fundamental problems. First, the achievable structural resolution in biological samples is determined by the labeling density, which is limited by the size of the fluorescent probe (several nanometers) and by the labeling efficiency. Second, fluorophores can interact via energy transfer at sub-10 nm distances, which results in accelerated photoswitching (blinking) and photobleaching, and thus in substantially lower localization probabilities.
The simple solution to this question is to separate the fluorophores spatially, without changing the labeling efficiency. This is best achieved by the physical expansion of the specimen, after embedding it in a swellable gel, in ExM. We implemented this approach, which we termed optimized nanoscale expansion (ONE) microscopy. To improve ONE and ExM resolution, the KTH laboratory also introduced a setup capable of imaging large gels, using sample- and signa-adapted imaging approaches. The details of this work were published in several articles, throughout the project.
We next proceeded to the proof-of-principle demonstration of imaging all of our selected nanobodies. GFP-carrying nanobodies were used, were applied to expanded gels immobilized on coverslips, and resulted in a complex, combined image, as shown in the attached file.
Our work demonstrated the applicability of the IMAGEOMICS approach, implying that a new branch has been opened, within the biological imaging field.
Importantly, we also introduced a technology that enables simple and effective super-resolution imaging, at very high resolutions, in the form of one-nanometer expansion (ONE) microscopy. We combined the 10-fold axial expansion of the specimen (1000-fold by volume) with a fluorescence fluctuation analysis to achieve resolutions down to 1 nm or better. We have successfully applied ONE microscopy to image cultured cells, tissues, viral particles, molecular complexes and single proteins. At the cellular level, using immunostaining, our technology revealed detailed nanoscale arrangements of synaptic proteins, including a quasi-regular organisation of PSD95 clusters. At the single molecule level, upon main chain fluorescent labelling, we could visualise the shape of individual membrane and soluble proteins. Moreover, conformational changes undergone by the ~17 kDa protein calmodulin upon Ca2+ binding were readily observable. We could also image and classify molecular aggregates in cerebrospinal fluid samples from Parkinson’s Disease (PD) patients, which represents a promising new development towards an improved PD diagnosis.
ONE microscopy is compatible with conventional microscopes and can be performed with the software we provided (on Github) as a free, open-source package. This technology bridges the gap between high-resolution structural biology techniques and light microscopy, and provides a new avenue for discoveries in biology and medicine.
Importantly, we also introduced a technology that enables simple and effective super-resolution imaging, at very high resolutions, in the form of one-nanometer expansion (ONE) microscopy. We combined the 10-fold axial expansion of the specimen (1000-fold by volume) with a fluorescence fluctuation analysis to achieve resolutions down to 1 nm or better. We have successfully applied ONE microscopy to image cultured cells, tissues, viral particles, molecular complexes and single proteins. At the cellular level, using immunostaining, our technology revealed detailed nanoscale arrangements of synaptic proteins, including a quasi-regular organisation of PSD95 clusters. At the single molecule level, upon main chain fluorescent labelling, we could visualise the shape of individual membrane and soluble proteins. Moreover, conformational changes undergone by the ~17 kDa protein calmodulin upon Ca2+ binding were readily observable. We could also image and classify molecular aggregates in cerebrospinal fluid samples from Parkinson’s Disease (PD) patients, which represents a promising new development towards an improved PD diagnosis.
ONE microscopy is compatible with conventional microscopes and can be performed with the software we provided (on Github) as a free, open-source package. This technology bridges the gap between high-resolution structural biology techniques and light microscopy, and provides a new avenue for discoveries in biology and medicine.