Content aware image restoration improves spatiotemporal resolution in luminescence imaging

Luminescent reporters are due to their intrinsically high signal-to-noise ratio a powerful labelling tool for microscopy and macroscopic in vivo imaging in biomedical research. However, luminescence signal detection requires longer exposure times than fluorescence imaging and is consequently less suited for applications requiring high temporal resolution or throughput. Here we demonstrate that content aware image restoration can drastically reduce the exposure time requirements in luminescence imaging, thus overcoming one of the major limitations of the technique.

Content-aware image restoration: pushing the limits of fluorescence microscopy

Fluorescence microscopy is a key driver of discoveries in the life sciences, with observable phenomena being limited by the optics of the microscope, the chemistry of the fluorophores, and the maximum photon exposure tolerated by the sample. These limits necessitate trade-offs between imaging speed, spatial resolution, light exposure, and imaging depth. In this work we show how content-aware image restoration based on deep learning extends the range of biological phenomena observable by microscopy. We demonstrate on eight concrete examples how microscopy images can be restored even if 60-fold fewer photons are used during acquisition, how near isotropic resolution can be achieved with up to tenfold under-sampling along the axial direction, and how tubular and granular structures smaller than the diffraction limit can be resolved at 20-times-higher frame rates compared to state-of-the-art methods. All developed image restoration methods are freely available as open source software in Python, FIJI, and KNIME.

A tunable refractive index matching medium for live imaging cells, tissues and model organisms

In light microscopy, refractive index mismatches between media and sample cause spherical aberrations that often limit penetration depth and resolution. Optical clearing techniques can alleviate these mismatches, but they are so far limited to fixed samples. We present Iodixanol as a non-toxic medium supplement that allows refractive index matching in live specimens and thus substantially improves image quality in live-imaged primary cell cultures, planarians, zebrafish and human cerebral organoids.

DOI:http://dx.doi.org/10.7554/eLife.27240.001

Light microscopy is a key tool in biomedical research. For perfect images, light needs to be able to pass through the sample, the material (or “mounting medium”) that holds the sample in place, and finally the image-detecting equipment in a straight line. However, in practice, light rays often deviate away from this line because they move at different speeds in different materials; how much the speed of light changes is related to a property called the refractive index of the material. This is exactly the effect that causes a stick stuck into water to look bent at the water’s surface. In light microscopy, mismatches in refractive index significantly reduce quality of the images that can be obtained.

Live specimens are particularly challenging to image because different specimens have very different refractive indices compared to the mounting medium, which holds specimens in place but must also keep them alive. Although the addition of chemical compounds can theoretically match the refractive index of the mounting medium to that of the specimen, this approach has so far not been practical because such manipulations tend to kill the specimen. An important challenge has therefore been to identify a compound that can adjust, or “tune”, the refractive index of mounting media over a wide range, yet without harming the specimens.

Now, Boothe et al. have identified a chemical called Iodixanol as an ideal and easy to use supplement for tuning the refractive index of water-based live imaging media. Adding Iodixanol to the mounting media did not appear to have any toxic effects on cell cultures, developing zebrafish embryos or regenerating planarian flatworms. Importantly, Boothe et al. found that Iodixanol significantly improved the quality of the images collected from all of these different specimens.

It is important to stress that Iodixanol does not change the refractive index of the sample or cancel out refractive index differences within the sample – so it cannot render opaque specimens transparent. Nevertheless, Iodixanol supplementation is a simple and affordable technique to improve image quality in any live imaging application without having to resort to more expensive and highly specialized microscopes.

DOI:http://dx.doi.org/10.7554/eLife.27240.002

Inter-domain tagging implicates caveolin-1 in insulin receptor trafficking and Erk signaling bias in pancreatic beta-cells

OBJECTIVE

The role and mechanisms of insulin receptor internalization remain incompletely understood. Previous trafficking studies of insulin receptors involved fluorescent protein tagging at their termini, manipulations that may be expected to result in dysfunctional receptors. Our objective was to determine the trafficking route and molecular mechanisms of functional tagged insulin receptors and endogenous insulin receptors in pancreatic beta-cells.

METHODS

We generated functional insulin receptors tagged with pH-resistant fluorescent proteins between domains. Confocal, TIRF and STED imaging revealed a trafficking pattern of inter-domain tagged insulin receptors and endogenous insulin receptors detected with antibodies.

RESULTS

Surprisingly, interdomain-tagged and endogenous insulin receptors in beta-cells bypassed classical Rab5a- or Rab7-mediated endocytic routes. Instead, we found that removal of insulin receptors from the plasma membrane involved tyrosine-phosphorylated caveolin-1, prior to trafficking within flotillin-1-positive structures to lysosomes. Multiple methods of inhibiting caveolin-1 significantly reduced Erk activation in vitro or in vivo, while leaving Akt signaling mostly intact.

CONCLUSIONS

We conclude that phosphorylated caveolin-1 plays a role in insulin receptor internalization towards lysosomes through flotillin-1-positive structures and that caveolin-1 helps bias physiological beta-cell insulin signaling towards Erk activation.

Statistical approaches and software for clustering islet cell functional heterogeneity

Worldwide efforts are underway to replace or repair lost or dysfunctional pancreatic β-cells to cure diabetes. However, it is unclear what the final product of these efforts should be, as β-cells are thought to be heterogeneous. To enable the analysis of β-cell heterogeneity in an unbiased and quantitative way, we developed model-free and model-based statistical clustering approaches, and created new software called TraceCluster. Using an example data set, we illustrate the utility of these approaches by clustering dynamic intracellular Ca(2+) responses to high glucose in ∼300 simultaneously imaged single islet cells. Using feature extraction from the Ca(2+) traces on this reference data set, we identified 2 distinct populations of cells with β-like responses to glucose. To the best of our knowledge, this report represents the first unbiased cluster-based analysis of human β-cell functional heterogeneity of simultaneous recordings. We hope that the approaches and tools described here will be helpful for those studying heterogeneity in primary islet cells, as well as excitable cells derived from embryonic stem cells or induced pluripotent cells.

Metabolic asymmetries in asymptomatic HIV-1 seropositive subjects: relationship to disease onset and MRI findings

Fifteen male homosexual subjects (mean age 31.6 +/- 7.2 yr) who were asymptomatic, but HIV-1 seropositive (HIV+) were compared to 15 male age-matched HIV-1 seronegative (HIV-) subjects using resting PET/FDG studies and MR scans. Mean cerebral metabolic rates for glucose (mg/100 g/min) in the HIV+ and HIV- subjects were 7.7 +/- 1.7 and 7.0 +/- 2.1, (p = 0.44), respectively. An index of regional metabolic asymmetry for the whole brain was 5.8% +/- 3.2% in the HIV+ and 2.7% +/- 2.3% in the HIV- (p = 0.002), and the difference was most prominent in the prefrontal area. Significant asymmetries were found in 10/15 HIV+ subjects, primarily in prefrontal (7/15) and premotor (4/15) regions. MRI scans showed no abnormalities on clinical or quantitative evaluation in HIV+ subjects. Upon follow-up of HIV+ subjects over 18-40 mo, seven became symptomatic, of which two died. There was no relationship between the presence of PET scan abnormalities and earlier onset of symptomatic disease.

Positron emission tomographic studies during serial word-reading by normal and dyslexic adults

Positron-emission tomography (PET) was used to study regional cerebral metabolic activity during oral reading in right-handed adult males with, and without a childhood and family history of developmental dyslexia. Significant group differences in normalized regional metabolic values were revealed in prefrontal cortex and in the lingual (inferior) region of the occipital lobe. Lingual values were bilaterally higher for dyslexic than normal readers. In contrast to the asymmetry observed in prefrontal and lingual regions in nondyslexic subjects during reading, the dyslexic pattern was more symmetric. These results demonstrate that individuals who suffered from familial developmental dyslexia as children, activate different brain regions during reading as adults, as compared to individuals without such childhood history.

Sensitivity of cerebral glucose metabolism to age, gender, brain volume, brain atrophy, and cerebrovascular risk factors

In 76 normal volunteers studied by positron emission tomography, with [18F]fluorodeoxyglucose, CMRglu was significantly lower in the elderly as compared with young subjects and significantly higher in females relative to males. However, in 58 of these subjects who also had magnetic resonance imaging scans, age and gender were found to be unrelated to CMRglu, when the effects of brain volume and brain atrophy on CMRglu were partialed out using covariate analyses. Individually, brain volume was found to have a significant effect on CMRglu, explaining approximately 17% of the variability in CMRglu measures and brain atrophy explaining approximately 8% of the variance in CMRglu. Together these two measures accounted for approximately 21% of the variance. Cerebrovascular risk factors in normal subjects were not found to affect mean CMRglu or the variability of CMRglu measures. In this study almost 80% of the variance in CMRglu could not be explained by any of the factors that had been considered. This implies a lack of sensitivity of absolute values of global CMRglu to the mild effects of brain dysfunction. Although some of the unexplained variance is probably methodological in origin, physiological factors that are difficult to quantify, such as the state of arousal, are likely to be contributory as well.

Behavioral activation and the variability of cerebral glucose metabolic measurements

Variability in cerebral glucose metabolism was examined between and within subjects when paired studies were performed in the resting state or in a behaviorally activated state. Both normal and demented subjects were studied twice each, from 1 to 6 weeks apart, under near-identical conditions, using positron emission tomography (PET) and [18F]fluorodeoxyglucose. Resting state studies were repeated in nine normal and four demented subjects. A picture-viewing test, used for activation during PET, was used repeatedly in seven normal and five demented subjects. Within-subject variability, as assessed by the percent difference in metabolic rates in paired studies, was reduced by 60-70% for activation state compared to resting state studies in normals. It is concluded that PET studies of brain metabolism, which are designed to study the active brain, should indeed be performed in functionally activated states, as in addition to demonstrating metabolism during a defined functional state, activation studies show reduced variability of cerebral metabolic measures.