CenFind: a deep-learning pipeline for efficient centriole detection in microscopy datasets.

BACKGROUND: High-throughput and selective detection of organelles in immunofluorescence images is an important but demanding task in cell biology. The centriole organelle is critical for fundamental cellular processes, and its accurate detection is key for analysing centriole function in health and disease. Centriole detection in human tissue culture cells has been achieved typically by manual determination of organelle number per cell. However, manual cell scoring of centrioles has a low throughput and is not reproducible. Published semi-automated methods tally the centrosome surrounding centrioles and not centrioles themselves. Furthermore, such methods rely on hard-coded parameters or require a multichannel input for cross-correlation. Therefore, there is a need for developing an efficient and versatile pipeline for the automatic detection of centrioles in single channel immunofluorescence datasets. RESULTS: We developed a deep-learning pipeline termed CenFind that automatically scores cells for centriole numbers in immunofluorescence images of human cells. CenFind relies on the multi-scale convolution neural network SpotNet, which allows the accurate detection of sparse and minute foci in high resolution images. We built a dataset using different experimental settings and used it to train the model and evaluate existing detection methods. The resulting average F1-score achieved by CenFind is > 90% across the test set, demonstrating the robustness of the pipeline. Moreover, using the StarDist-based nucleus detector, we link the centrioles and procentrioles detected with CenFind to the cell containing them, overall enabling automatic scoring of centriole numbers per cell. CONCLUSIONS: Efficient, accurate, channel-intrinsic and reproducible detection of centrioles is an important unmet need in the field. Existing methods are either not discriminative enough or focus on a fixed multi-channel input. To fill this methodological gap, we developed CenFind, a command line interface pipeline that automates cell scoring of centrioles, thereby enabling channel-intrinsic, accurate and reproducible detection across experimental modalities. Moreover, the modular nature of CenFind enables its integration in other pipelines. Overall, we anticipate CenFind to prove critical for accelerating discoveries in the field.

Atypical and distinct microtubule radial symmetries in the centriole and the axoneme of Lecudina tuzetae.

The centriole is a minute cylindrical organelle present in a wide range of eukaryotic species. Most centrioles have a signature ninefold radial symmetry of microtubules that is imparted to the axonemes of the cilia and flagella they template, with nine centriolar microtubule doublets growing into nine axonemal microtubule doublets. There are exceptions to the ninefold symmetrical arrangement of axonemal microtubules in some species, with lower or higher fold symmetries. In the few cases where this has been examined, such alterations in axonemal symmetries are grounded in similar alterations in centriolar symmetries. Here, we examine the question of microtubule number continuity between centriole and axoneme in flagellated gametes of the gregarine Lecudina tuzetae, which have been reported to exhibit a sixfold radial symmetry of axonemal microtubules. We used time-lapse differential interference microscopy to identify the stage at which flagellated gametes are present. Thereafter, using electron microscopy and ultrastructure-expansion microscopy coupled to stimulated emission depletion superresolution imaging, we uncover that a six- or fivefold radial symmetry in the axoneme is accompanied by an eightfold radial symmetry in the centriole. We conclude that the transition between centriolar and axonemal microtubules can be characterized by unexpected plasticity.

Coalescence and directed anisotropic growth of starch granule initials in subdomains of Arabidopsis thaliana chloroplasts.

Living cells orchestrate enzyme activities to produce myriads of biopolymers but cell-biological understanding of such processes is scarce. Starch, a plant biopolymer forming discrete, semi-crystalline granules within plastids, plays a central role in glucose storage, which is fundamental to life. Combining complementary imaging techniques and Arabidopsis genetics we reveal that, in chloroplasts, multiple starch granules initiate in stromal pockets between thylakoid membranes. These initials coalesce, then grow anisotropically to form lenticular granules. The major starch polymer, amylopectin, is synthesized at the granule surface, while the minor amylose component is deposited internally. The non-enzymatic domain of STARCH SYNTHASE 4, which controls the protein's localization, is required for anisotropic growth. These results present us with a conceptual framework for understanding the biosynthesis of this key nutrient.

STARCH SYNTHASE5, a Noncanonical Starch Synthase-Like Protein, Promotes Starch Granule Initiation in Arabidopsis.

What determines the number of starch granules in plastids is an enigmatic aspect of starch metabolism. Several structurally and functionally diverse proteins have been implicated in the granule initiation process in Arabidopsis (Arabidopsis thaliana), with each protein exerting a varying degree of influence. Here, we show that a conserved starch synthase-like protein, STARCH SYNTHASE5 (SS5), regulates the number of starch granules that form in Arabidopsis chloroplasts. Among the starch synthases, SS5 is most closely related to SS4, a major determinant of granule initiation and morphology. However, unlike SS4 and the other starch synthases, SS5 is a noncanonical isoform that lacks catalytic glycosyltransferase activity. Nevertheless, loss of SS5 reduces starch granule numbers that form per chloroplast in Arabidopsis, and ss5 mutant starch granules are larger than wild-type granules. Like SS4, SS5 has a conserved putative surface binding site for glucans and also interacts with MYOSIN-RESEMBLING CHLOROPLAST PROTEIN, a proposed structural protein influential in starch granule initiation. Phenotypic analysis of a suite of double mutants lacking both SS5 and other proteins implicated in starch granule initiation allows us to propose how SS5 may act in this process.

Two Plastidial Coiled-Coil Proteins Are Essential for Normal Starch Granule Initiation in Arabidopsis.

The mechanism of starch granule initiation in chloroplasts is not fully understood. Here, we aimed to build on our recent discovery that PROTEIN TARGETING TO STARCH (PTST) family members, PTST2 and PTST3, are key players in starch granule initiation, by identifying and characterizing additional proteins involved in the process in Arabidopsis thaliana chloroplasts. Using immunoprecipitation and mass spectrometry, we demonstrate that PTST2 interacts with two plastidial coiled-coil proteins. Surprisingly, one of the proteins is the thylakoid-associated MAR BINDING FILAMENT-LIKE PROTEIN1 (MFP1), which was proposed to bind plastid nucleoids. The other protein, MYOSIN-RESEMBLING CHLOROPLAST PROTEIN (MRC), contains long coiled coils and no known domains. Whereas wild-type chloroplasts contained multiple starch granules, only one large granule was observed in most chloroplasts of the mfp1 and mrc mutants. The mfp1 mrc double mutant had a higher proportion of chloroplasts containing no visible granule than either single mutant and accumulated ADP-glucose, the substrate for starch synthesis. PTST2 was partially associated with the thylakoid membranes in wild-type plants, and fluorescently tagged PTST2 was located in numerous discrete patches within the chloroplast in which MFP1 was also located. In the mfp1 mutant, PTST2 was not associated with the thylakoids and formed discrete puncta, suggesting that MFP1 is necessary for normal PTST2 localization. Overall, we reveal that proper granule initiation requires the presence of MFP1 and MRC, and the correct location of PTST2.