FogBank: a single cell segmentation across multiple cell lines and image modalities

Background

Many cell lines currently used in medical research, such as cancer cells or stem cells, grow in confluent sheets or colonies. The biology of individual cells provide valuable information, thus the separation of touching cells in these microscopy images is critical for counting, identification and measurement of individual cells. Over-segmentation of single cells continues to be a major problem for methods based on morphological watershed due to the high level of noise in microscopy cell images. There is a need for a new segmentation method that is robust over a wide variety of biological images and can accurately separate individual cells even in challenging datasets such as confluent sheets or colonies.

Results

We present a new automated segmentation method called FogBank that accurately separates cells when confluent and touching each other. This technique is successfully applied to phase contrast, bright field, fluorescence microscopy and binary images. The method is based on morphological watershed principles with two new features to improve accuracy and minimize over-segmentation.First, FogBank uses histogram binning to quantize pixel intensities which minimizes the image noise that causes over-segmentation. Second, FogBank uses a geodesic distance mask derived from raw images to detect the shapes of individual cells, in contrast to the more linear cell edges that other watershed-like algorithms produce.We evaluated the segmentation accuracy against manually segmented datasets using two metrics. FogBank achieved segmentation accuracy on the order of 0.75 (1 being a perfect match). We compared our method with other available segmentation techniques in term of achieved performance over the reference data sets. FogBank outperformed all related algorithms. The accuracy has also been visually verified on data sets with 14 cell lines across 3 imaging modalities leading to 876 segmentation evaluation images.

Conclusions

FogBank produces single cell segmentation from confluent cell sheets with high accuracy. It can be applied to microscopy images of multiple cell lines and a variety of imaging modalities. The code for the segmentation method is available as open-source and includes a Graphical User Interface for user friendly execution.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-014-0431-x) contains supplementary material, which is available to authorized users.     

Otopathogens Detected in Middle Ear Fluid Obtained during Tympanostomy Tube Insertion: Contrasting Purulent and Non-Purulent Effusions

Otitis media is a prominent disease among children. Previous literature indicates that otitis media is a polymicrobial disease, with Haemophilus influenzae, Streptococcus pneumoniae, Alloiococcus otitidis and Moraxella catarrhalis being the most commonly associated bacterial pathogens. Recent literature suggests that introduction of pneumococcal conjugate vaccines has had an effect on the etiology of otitis media. Using a multiplex PCR procedure, we sought to investigate the presence of the aforementioned bacterial pathogens in middle ear fluid collected from children undergoing routine tympanostomy tube placement at Wake Forest Baptist Medical Center during the period between January 2011 and March 2014. In purulent effusions, one or more bacterial organisms were detected in ~90% of samples. Most often the presence of H. influenzae alone was detected in purulent effusions (32%; 10 of 31). In non-purulent effusions, the most prevalent organism detected was A. otitidis (26%; 63 of 245). Half of the non-purulent effusions had none of these otopathogens detected. In purulent and non-purulent effusions, the overall presence of S. pneumoniae was lower (19%; 6 of 31, and 4%; 9 of 245, respectively) than that of the other pathogens being identified. The ratio of the percentage of each otopathogen identified in purulent vs. non-purulent effusions was >1 for the classic otopathogens but not for A. otitidis.     

The pore-lining regions in cytochrome c oxidases: A computational analysis of caveolin,cholesterol and transmembrane helix contributions to proton movement

Cytochrome c oxidase (CcO) is the terminal enzyme in the electron transfer chain. CcO catalyzes a four electron reduction of O₂ to water at a catalytic site formed by high-spin heme (a₃) and copper atoms (Cu_B). While it is recognized that proton movement is coupled to oxygen reduction, the proton channel(s) have not been well defined. Using computational methods developed to study protein topology, membrane channels and 3D packing arrangements within transmembrane (TM) helix arrays, we find that subunit-1 (COX-1), subunit-2 (COX-2) and subunit-3 (COX-3) contribute 139, 46 and 25 residues, respectively, to channel formation between the mitochondrial matrix and intermembrane space. Nine of 12 TM helices in COX-1, both helices in COX-2 and 5 of the 6 TM helices in COX-3 are pore-lining regions (possible channel formers). Heme a₃ and the Cu_B sites (as well as the Cu_A center of COX-2) are located within the channel that includes TM-6, TM-7, TM-10 and TM-11 of COX-1 and are associated with multiple cholesterol and caveolin-binding (CB) motifs. Sequence analysis identifies five CB motifs within COX-1, two within COX-2 and four within COX-3; each caveolin containing a pore-lining helix C-terminal to a TM helix–turn–helix. Channel formation involves interaction between multiple pore-lining regions within protein subunits and/or dimers. PoreWalker analysis lends support to the D-channel model of proton translocation. Under physiological conditions, caveolins may introduce channel formers juxtaposed to those in COX-1, COX-2 and COX-3, which together with cholesterol may form channel(s) essential for proton translocation through the inner mitochondrial membrane.     

Investigation of Ixodid ticks as vectors of Babesia caballi and Theileria equi (Protozoa: Apicomplexa) in central Italy

Babesia caballi and Theileria equi are widely recognized as causative agents of equine pirolasmosis (EP), an acute, sub‐acute, and chronic disease of equines, with relevant economic impact on horse trade worldwide. Although several studies on EP prevalence from central Italy have been published, data on ticks responsible for its transmission are still lacking. In this study, we identified a potential competent vector, investigating main features of its ecology together with EP infection rates. A two‐year sampling of questing ticks was carried out for the first time in Italy in an area known for high EP prevalence in horse sera, detecting the association between Rhipicephalus bursa and causative agents of EP. Most of the positive pools harbored a single infection (91.1%); mixed infections were also detected (8.9%). The infection rate for T. equi slightly decreased among years; B. caballi showed a lower, but increasing, infection rate. Tick phenology, climate variables, and peaks of EP prevalence indicated late May and second half of June as periods with the highest risk of new infections, especially during warm and dry days.     

Positive and negative regulation of the gamma-secretase activity by nicastrin in a murine model

Nicastrin is a component of the gamma-secretase complex that has been shown to adhere to presenilin-1 (PS1), Notch, and APP. Here we demonstrate that Nicastrin-deficient mice showed a phenotype that is indistinguishable from PS1/PS2 double knock-out mice, whereas heterozygotes were healthy and viable. Fibroblasts derived from Nicastrin-deficient embryos were unable to generate amyloid beta-peptide and failed to release the intracellular domain of APP- or Notch1-Gal4-VP16 fusion proteins. Additionally, C- and N-terminal fragments of PS1 and the C-terminal fragments of PS2 were not detectable in Nicastrin-null fibroblasts, whereas full-length PS1 accumulated in null fibroblasts, indicating that Nicastrin is required for the endoproteolytic processing of presenilins. Interestingly, cells derived from Nicastrin heterozygotes produced relatively higher levels of amyloid beta-peptide whether the source was endogenous mouse or transfected human APP. These data demonstrate that Nicastrin is essential for the gamma-secretase cleavage of APP and Notch in mammalian cells and that Nicastrin has both positive and negative functions in the regulation of gamma-secretase activity.     

The effects of ultraviolet irradiation on P680+ reduction in PS II core complexes measured for individual S-states and during repetitive cycling of the oxygen-evolving complex

Flash-induced absorbance measurements at 830 nm on both nanosecond and microsecond timescales have been used to characterise the effect of ultraviolet light on Photosystem II core particles. A combination of UV-A and UV-B, closely simulating the spectrum of sunlight below 350 nm, was found to have a primary effect on the donor side of P680. Repetitive measurements indicated reductions in the nanosecond components of the absorbance decay with a concomitant appearance and increase in the amplitude of a component with a 10 s time constant attributed to slow reduction of P680⁺ by Tyr_z when the function of the oxygen evolving complex is inhibited. Single-flash measurements show that the nanosecond components have amplitudes which vary with S-state. Increasing UV irradiation inhibited the amplitude of these components without changing their S-state dependence. In addition, UV irradiation resulted in a reduction in the total amplitude, with no change in the proportion of the 10 s contribution.Abbreviations BBY- PS II membrane fragments - P680- primary electron donor of PS II - PS II- Photosystem II - Q_A and Q_B– primary and secondary quinone electron acceptors of PS II - S-state- redox state of the oxygenevolving complex - Tyr_z– tyrosine residue in PS II - UV-A- ultraviolet radiation 320–400 nm - UV-B- ultraviolet radiation 280–320 nm     

CD4+ Th0 cell clones,isolated from a metastatic lymph node of a melanoma patient,possess cytolytic function

In the present study T lymphocytes isolated from a metastatic lymph node (T-LNL) of a melanoma patient have been cloned. In the attempt to verify whether T-LNL may acquire in vitro functional activities in the absence of tumour-associated antigens, they were cloned utilizing allogeneic lymphocytes as feeder cells. Nineteen clones generated from T-LNL proved to be CD4+ and, among these, five were able to kill autologous and allogeneic human melanoma cells in HLA-class-II-restricted way. On the basis of their cytokine production, these CD4+ cytolytic T-LNL clones were shown to belong to the Th0 subset and three of them expersseed the V17 chain of the T cell receptor. These results suggest the presence of melanomaspecific but functionally inactive lymphocytes with T cell receptor oligoclonality in the lymph node environment. These specific T cells may acquire in vitro the capacity to kill autologous and allogeneic tumours without any induction by autologous melanoma cells.     

Analysis of the distribution and evolution of the ATP‐dependent DNA ligases of bacteria delineates a distinct phylogenetic group ‘Lig E’

Prior to the discovery of a minimal ATP‐dependent DNA ligase in Haemophilus influenzae, bacteria were thought to only possess a NAD‐dependent ligase, which was involved in sealing of Okazaki fragments. We now know that a diverse range of bacterial species possess up to six of these accessory bacterial ATP‐dependent DNA ligases (b‐ADLs), which vary in size and enzymatic domain associations. Here we compare the domain structure of different types of b‐ADLs and investigate their distribution among the bacterial domain to describe possible evolutionary trajectories that gave rise to the sequence and structural diversity of these enzymes. Previous biochemical and genetic analyses have delineated three main classes of these enzymes: Lig B, Lig C and Lig D, which appear to have descended from a common ancestor within the bacterial domain. In the present study, we delineate a fourth group of b‐ADLs, Lig E, which possesses a number of unique features at the primary and tertiary structural levels. The biochemical characteristics, domain structure and inferred extracellular location sets this group apart from the other b‐ADLs. The results presented here indicate that the Lig E type ligases were horizontally transferred into bacteria in a separate event from other b‐ADLs possibly from a bacteriophage.