Fluorescence correlation spectroscopy (FCS) techniques are well-established tools to investigate molecular dynamics in confocal and super-resolution microscopy. In practice, users often need to handle a variety of sample- or hardware-related artifacts, an example being peak artifacts created by bright, slow-moving clusters. Approaches to address peak artifacts exist, but measurements suffering from severe artifacts are typically nonanalyzable. Here, we trained a one-dimensional U-Net to automatically identify peak artifacts in fluorescence time series and then analyzed the purified, nonartifactual fluctuations by time-series editing. We show that, in samples with peak artifacts, the transit time and particle number distributions can be restored in simulations and validated the approach in two independent biological experiments. We propose that it is adaptable for other FCS artifacts, such as detector dropout, membrane movement, or photobleaching. In conclusion, this simulation-based, automated, open-source pipeline makes measurements analyzable that previously had to be discarded and extends every FCS user's experimental toolbox.

Abstract DNA double-strand breaks (DSBs), such as those produced by radiation and radiomimetics, are amongst the most toxic forms of cellular damage, in part because they involve extensive oxidative modifications at the break termini. Prior to completion of DSB repair, the chemically modified termini must be removed. Various DNA processing enzymes have been implicated in the processing of these dirty ends, but molecular knowledge of this process is limited. Here, we demonstrate a role for the metallo-β-lactamase fold 5′−3′ exonuclease SNM1A in this vital process. Cells disrupted for SNM1A manifest increased sensitivity to radiation and radiomimetic agents and show defects in DSB damage repair. SNM1A is recruited and is retained at the sites of DSB damage via the concerted action of its three highly conserved PBZ, PIP box and UBZ interaction domains, which mediate interactions with poly-ADP-ribose chains, PCNA and the ubiquitinated form of PCNA, respectively. SNM1A can resect DNA containing oxidative lesions induced by radiation damage at break termini. The combined results reveal a crucial role for SNM1A to digest chemically modified DNA during the repair of DSBs and imply that the catalytic domain of SNM1A is an attractive target for potentiation of radiotherapy.

DNA double-strand breaks (DSBs), such as those produced by radiation and radiomimetics, are amongst the most toxic forms of cellular damage, in part because they involve extensive oxidative modifications at the break termini. Prior to completion of DSB repair, the chemically modified termini must be removed. Various DNA processing enzymes have been implicated in the processing of these ‘dirty ends’, but molecular knowledge of this process is limited. Here, we demonstrate a role for the metallo-β-lactamase fold 5′-3′ exonuclease SNM1A in this vital process. Cells disrupted for SNM1A manifest increased sensitivity to radiation and radiomimetic agents and show defects in DSB damage repair. SNM1A is recruited and is retained at the sites of DSB damage via the concerted action of its three highly conserved PBZ, PIP box and UBZ interaction domains, which mediate interactions with poly-ADP-ribose chains, PCNA and the ubiquitinated form of PCNA, respectively. SNM1A can resect DNA containing oxidative lesions induced by radiation damage at break termini. The combined results reveal a crucial role for SNM1A to digest chemically modified DNA during the repair of DSBs and imply that the catalytic domain of SNM1A is an attractive target for potentiation of radiotherapy.

The cell cycle is a complex biological phenomenon, which plays an important role in many cell biological processes and disease states. Machine learning is emerging to be a pivotal technique for the study of the cell cycle, resulting in a number of available tools and models for the analysis of the cell cycle. Most, however, heavily rely on expert annotations, prior knowledge of mechanisms, and imaging with several fluorescent markers to train their models. Many are also limited to processing only the spatial information in the cell images. In this work, we describe a different approach based on representation learning to construct a manifold of the cell life cycle. We trained our model such that the representations are learned without exhaustive annotations nor assumptions. Moreover, our model uses microscopy images derived from a single fluorescence channel and utilizes both the spatial and temporal information in these images. We show that even with fewer channels and self-supervision, information relevant to cell cycle analysis such as staging and estimation of cycle duration can still be extracted, which demonstrates the potential of our approach to aid future cell cycle studies and in discovery cell biology to probe and understand novel dynamic systems.

Newly synthesized peroxisomal proteins are recognized in the cytosol by the cycling receptor PEX5 and directed to a docking complex comprising PEX14 and PEX13 at the peroxisomal membrane. After cargo translocation, the unloaded PEX5 is recycled in an ATP-dependent manner. Receptor docking involves the WxxxF-motifs in the N-terminal domain (NTD) of PEX5 that are recognized by the N-terminal domain of PEX14. Here, we combine biochemical methods and NMR spectroscopy to identify a novel binding interface between human PEX5 and PEX14. The interaction involves the PEX5 C-terminal cargo-binding TPR domain and a conserved IPSWQI peptide motif in the C-terminal intrinsically disordered region of PEX14. The three-dimensional structure of the PEX14 IPSWQI peptide bound the PEX5 TPR domain, shows the PEX14 interaction is non-overlapping with PTS1 binding to the TPR domain. Notably, PEX14 IPSWQI motif binding to a hinge region in the TPR domain shows a more open supercoil of the TPR fold that resembles the apo conformation in the absence of PTS1 peptide. Mutation of binding site residues in PEX5 or PEX14 leads to a partial protein import defect and decrease of the steady-state-concentration of PEX5. This resembles the mutant phenotype of cells affected in receptor recycling, suggesting a role in this process.

Cellular functions rely on proper actions of organelles such as peroxisomes. These organelles rely on the import of proteins from the cytosol. The peroxisomal import receptor PEX5 takes up target proteins in the cytosol and transports them to the peroxisomal matrix. However, its cytosolic molecular interactions have so far not directly been disclosed. Here, we combined advanced optical microscopy and spectroscopy techniques such as fluorescence correlation spectroscopy and stimulated emission depletion microscopy with biochemical tools to present a detailed characterization of the cytosolic diffusion and interaction dynamics of PEX5. Among other features, we highlight a slow diffusion of PEX5, independent of aggregation or target binding, but associated with cytosolic interaction partners via its N-terminal domain. This sheds new light on the functionality of the receptor in the cytosol as well as highlighting the potential of using complementary microscopy tools to decipher molecular interactions in the cytosol by studying their diffusion dynamics.

The question of why certain cell types are differentially susceptible to mutations in broadly expressed genes remains largely unanswered. An important first step in determining the vulnerability of a specific lineage is to identify at what stage its requirement for a given protein arises. Strikingly, many types of congenital anaemia result from mutations in widely, if not ubiquitously, expressed proteins involved in core cellular processes. Examples include: mutations in ribosomal proteins causing Diamond Blackfan Anaemia1; mutations in the secretory vesicle protein SEC23B causing Congential Dyserythropoietic Anaemia (CDA) type II2; in CDA type III, arising from mutations in the kinesin encoded by KIF23 and RACGAP13, 4; and CDA type I (CDA-I) caused by mutations in CDAN1 or CDIN1.5 CDA-I is an example of a rare disease which has the potential to inform us about general cellular processes. CDIN1 and CDAN1 are widely expressed and loss of either protein is incompatible with life; however, only developing erythroblasts appear to be susceptible to biallelic hypomorphic mutations in either gene. We have recently shown that ex vivo differentiated erythroid cells from patients with CDA-I are delayed during terminal erythroid differentiation and this is associated with increased proliferation and widespread changes in chromatin accessibility.6 However, these studies were performed on CD34+ cells, which represent a mixture of haematopoietic stem and progenitor cells (HSPCs). The aim of this study was to investigate the precise cellular stage affected in CDA-I by assaying the functional consequences of CDAN1/CDIN1 mutations (Table 1) on the frequency and erythroid output of myeloid progenitors in patients with CDA-I. We show CDA-I does not alter the myeloid compartment; however, the functional output of CDA-I megakaryocyte and erythroid progenitors (MEPs) is compromised. Both in clonogenic assays and in liquid culture, CDA-I MEPs produce a higher proportion of aberrant colonies and show reduced expression of the erythroid lineage markers CD235 and CD71. c.1104_1106delCTT c.3128A > T p.Phe369del p.Asp1043Val c.2015C > T c.2681_2682delAG Venesection IFN-α2a c.2015C > T c.3338 T > C p.Pro672Leu p.Leu1113Pro c.2044C > T c.2744_2767del Whether the erythroid defect in CDA-I is confined to terminally maturing erythroblasts or also affects their progenitors remains an open question. To address this, we first investigated the expression of both causative genes (CDAN1 and CDIN1) in primary human HSPCs from healthy individuals. We found that similar expression levels of both genes were maintained throughout the haematopoietic hierarchy from HSC through to committed myeloid and erythroid progenitors, showing a decrease during terminal erythroid differentiation from the basophilic erythroblast stage onwards (Figure 1A, B).9 Next, to determine when defects first arise in CDA-I erythropoiesis, we compared the frequency of the progenitor cells upstream of erythroblasts in the haematopoietic hierarchy [Common Myeloid Progenitors (CMP), Megakaryocyte-Erythroid Progenitors (MEPs) and Granulocyte-Macrophage Progenitors (GMPs)] in peripheral blood (PB) of patients with CDA-I and controls (Figure 1C). This showed that the frequency of all three progenitor cell types was the same in CDA-I patients (n = 5) compared to controls (n = 17) (Figure 1C), indicating that biallelic mutations in CDAN1 or CDIN1 have no effect on the numbers of myeloid progenitors with erythroid potential, despite expression of both genes in progenitors (Figure 1B). This finding is in contrast to the situation in Diamond Blackfan Anaemia (DBA), where there are quantitative and qualitative abnormalities in erythroid progenitors.10 To assess whether the function of these myeloid progenitors was affected by biallelic mutations in CDAN1 or CDIN1, we sorted individual MEPs from healthy individuals (n = 10) and CDA-I patients (n = 5) into methylcellulose and cultured them for 14 days. The overall plating efficiency and proportion of MEPs that gave rise to erythroid colonies was similar in patients with CDA-I and controls ~70% (Figure 1D). However, whilst 100% of the erythroid output from control MEPs produced mature BFU-Es, 16.9% ± 3.7 of the erythroid colonies produced by CDA-I MEPs were small, poorly haemoglobinised erythroid clusters that were not seen in the MEP cultures from healthy donors (Figure 1D). These data suggest that, in patients with CDA-I, mutations in CDAN1 or CDIN1 cause perturbation of erythropoiesis at an earlier stage of erythroid differentiation (MEP) than previously reported, with samples from patients with mutations in either gene being similarly affected. Consistent with this, in three CDA-I patients with CDAN1 mutations, treatment with IFN-α2a, the only current therapy for CDA-I,5 restored the normal frequency and pattern of MEP-derived erythroid colonies (Figure 1D) and also normalized Hb levels in patients (Figure 1E). Sorting of individual MEPs into liquid culture, to allow more terminal differentiation, confirmed a qualitative defect in the number of MEP derived colonies grown from CDA-I MEPs (n = 36 colonies) compared to controls (n = 62 colonies) (Figure S1a). Meanwhile, interestingly, MEPs (n = 91) cultured from three CDA-I patients who were on IFN-α2a treatment at the time of sampling, showed normal erythroid maturation in vitro with only 6% aberrant colonies compared to 36% from the untreated CDA-I patients (Figure S1b). In summary, we provide evidence that defects in erythroid development in the rare inherited anaemia CDA-I arise at the MEP stage of erythroid differentiation, much earlier than previously recognized. There is no reported effect on the megakaryocytic lineage and platelet counts in CDA-I patients appear to be normal (Table 1). In common with other cell types, the question of why megakaryocytes are unaffected in CDA-I remains unclear. One explanation may be functional redundancy, as is presumed to be the case in other cell types. Some insight into this may come from analysis of platelet RNA, which suggests that megakaryocytes express a higher level of CDIN1 than CDAN1, the latter of which is below the level of detection.11 Our data also suggest that the erythroid defect can be corrected by IFN-α2a. It remains to be shown whether abnormal MEPs in CDA-I give rise exclusively to abnormal erythroblasts. Current understanding suggests the proteins encoded by CDIN1 and CDAN1 play a role in DNA repair and/or chromatin assembly. Analysing erythroid cells harbouring mutations in both genes would be likely to elucidate this novel pathway, however, generating suitable models will be technically challenging. By refining the stages of erythroid differentiation affected in CDA-I, this study paves the way for further work to identify why erythroid progenitors at the MEP stage develop a specific requirement for the function of the proteins encoded by CDAN1 and CDIN1. To this end, single-cell multiomic approaches to characterize the molecular defect in CDA-I progenitors are ongoing in our laboratory. Such studies would have the potential to inform novel processes in normal and disease erythropoiesis. We thank the CDA-I patients for providing blood samples and the flow cytometry facility at the WIMM for providing cell analysis services and technical expertise. The authors declare no competing financial interests. CS extracted CD34+ cells from CDA-I patient blood and cones; CS and KB performed the flow cytometry analysis and sorting experiments with assistance from SC. CS and KB scored, imaged and flow cytometric analysis of colonies from the Methocult and liquid culture assays. IR, VJB, NR, CB, CS and KB conceived and designed experiments; NR provided conceptual advice and clinical oversight; CB created the figures; CS and CB wrote the paper and all authors reviewed and critically edited the manuscript. RB, QAH, SO, RR, KR, MC and NR are the clinicians responsible for the care of the CDA-I patients. Raw RNAseq data from Gene Expression Omnibus (GEO) GSE74912 and GSE115684 was analysed in Figure 1B. FigS 1 Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

The megakaryocyte/erythroid Transient Myeloproliferative Disorder (TMD) in newborns with Down Syndrome (DS) occurs when N-terminal truncating mutations of the hemopoietic transcription factor GATA1, that produce GATA1short protein (GATA1s), are acquired early in development. Prior work has shown that murine GATA1s, by itself, causes a transient yolk sac myeloproliferative disorder. However, it is unclear where in the hemopoietic cellular hierarchy GATA1s exerts its effects to produce this myeloproliferative state. Here, through a detailed examination of hemopoiesis from murine GATA1s ES cells and GATA1s embryos we define defects in erythroid and megakaryocytic differentiation that occur relatively late in hemopoiesis. GATA1s causes an arrest late in erythroid differentiation in vivo, and even more profoundly in ES-cell derived cultures, with a marked reduction of Ter-119 cells and reduced erythroid gene expression. In megakaryopoiesis, GATA1s causes a differentiation delay at a specific stage, with accumulation of immature, kit-expressing CD41hi megakaryocytic cells. In this specific megakaryocytic compartment, there are increased numbers of GATA1s cells in S-phase of cell cycle and reduced number of apoptotic cells compared to GATA1 cells in the same cell compartment. There is also a delay in maturation of these immature GATA1s megakaryocytic lineage cells compared to GATA1 cells at the same stage of differentiation. Finally, even when GATA1s megakaryocytic cells mature, they mature aberrantly with altered megakaryocyte-specific gene expression and activity of the mature megakaryocyte enzyme, acetylcholinesterase. These studies pinpoint the hemopoietic compartment where GATA1s megakaryocyte myeloproliferation occurs, defining where molecular studies should now be focussed to understand the oncogenic action of GATA1s.

Background: Bioimage analysis is an emerging field within the global research community. It is an interdisciplinary discipline which requires knowledge of biology, image analysis and biophysics. This report represents the analysis and discussion of two questionnaires run by the Image Analysis Focused Interest Group of the Royal Microscopical Society (IAFIG-RMS). The goal of this document, which represents the analysis and interpretation of these questionnaires, is to highlight the current research climate for Bioimage Analysts in the UK and discusses some of the problems and possibilities for this emerging discipline. Methods: Two questionnaires (2016 and 2019) were developed and sent to researchers in the UK using mailing lists and forums specific for microscopy and image analysis. The participants were asked a range of questions spanning different aspects of their work and funding. Respondents were collected and analysed using Jupyter notebooks. Results: The analysis of the responses from these questionnaires highlighted many interesting issues and aspects of this community. It is clear that a major issue for the community is the nature of the funding and the long-term career possibilities available. Furthermore, the issue of independence is discussed with clear evidence that researchers would like to pursue their own research with the option of dedicated time to support the research of others. Conclusions: It is our hope that this study will help catalyse funding opportunities which help support this emerging discipline and help it establish a unique identity for itself within the research community in the UK and beyond.

Seeing the hidden: In expansion microscopy (ExM) subcellular structures are imaged in isotropically expanded fixed samples, consequently allowing enlarged subcellular structures that would otherwise be hidden to standard microscopy to be resolved. Upon comparison of the expansion factors of different cellular compartments in cells within the same gel, we found significant differences in expansion factors of a factor of above 2. More information can be found in the full paper by K. Reglinski et al.

Hypoxia is a common phenomenon in solid tumours strongly linked to the hallmarks of cancer. Hypoxia promotes local immunosuppression and downregulates type I interferon (IFN) expression and signalling, which contribute to the success of many cancer therapies. Double-stranded RNA (dsRNA), transiently generated during mitochondrial transcription, endogenously activates the type I IFN pathway. We report the effects of hypoxia on the generation of mitochondrial dsRNA (mtdsRNA) in breast cancer. We found a significant decrease in dsRNA production in different cell lines under hypoxia. This effect was HIF1α/2α-independent. mtdsRNA was responsible for induction of type I IFN and significantly decreased after hypoxia. Mitochondrially encoded gene expression was downregulated and mtdsRNA bound by the dsRNA-specific J2 antibody was decreased during hypoxia. These findings reveal a new mechanism of hypoxia-induced immunosuppression that could be targeted by hypoxia-activated therapies.

A major challenge in cell and developmental biology is the automated identification and quantitation of cells in complex multilayered tissues. We developed CytoCensus: an easily deployed implementation of supervised machine learning that extends convenient 2D 'point-and-click' user training to 3D detection of cells in challenging datasets with ill-defined cell boundaries. In tests on such datasets, CytoCensus outperforms other freely available image analysis software in accuracy and speed of cell detection. We used CytoCensus to count stem cells and their progeny, and to quantify individual cell divisions from time-lapse movies of explanted Drosophila larval brains, comparing wild-type and mutant phenotypes. We further illustrate the general utility and future potential of CytoCensus by analysing the 3D organisation of multiple cell classes in Zebrafish retinal organoids and cell distributions in mouse embryos. CytoCensus opens the possibility of straightforward and robust automated analysis of developmental phenotypes in complex tissues.

The antigen-presenting molecule MR1 presents riboflavin-based metabolites to Mucosal-Associated Invariant T (MAIT) cells. While MR1 egress to the cell surface is ligand-dependent, the ability of small-molecule ligands to impact on MR1 cellular trafficking remains unknown. Arising from an in silico screen of the MR1 ligandbinding pocket, we identify one ligand, 3-([2,6-dioxo-1,2,3,6-tetrahydropyrimidin-4-yl]formamido)propanoic acid, DB28, as well as an analog, methyl 3-([2,6-dioxo-1,2,3,6-tetrahydropyrimidin-4-yl]formamido)propanoate, NV18.1, that down-regulate MR1 from the cell surface and retain MR1 molecules in the endoplasmic reticulum (ER) in an immature form. DB28 and NV18.1 compete with the known MR1 ligands, 5-OP-RU and acetyl-6-FP, for MR1 binding and inhibit MR1-dependent MAIT cell activation. Crystal structures of the MAIT T cell receptor (TCR) complexed with MR1-DB28 and MR1-NV18.1, show that these two ligands reside within the A'pocket of MR1. Neither ligand forms a Schiff base with MR1 molecules; both are nevertheless sequestered by a network of hydrophobic and polar contacts. Accordingly, we define a class of compounds that inhibits MR1 cellular trafficking.

Department of Medical Oncology, Molecular Oncology Laboratories, Weatherall Institute of 5 Molecular Medicine, University of Oxford, John Radcliffe Hospital, Oxford, OX3 9DS, UK 6 Current address: Cambridge Institute for Therapeutic Immunology & Infectious Disease. 7 Jeffrey Cheah Biomedical Centre. Puddlecombe Way, Cambridge, CA CB20AW, UK 8 Department of Oncology, Old Road Campus Research Building, University of Oxford, 9 Oxford, OX3 7DQ, UK 10 Medical Research Council Human Immunology Unit, Medical Research Council Weatherall 11 Institute of Molecular Medicine, Radcliffe Department of Medicine, University of Oxford, 12 Oxford OX3 9DS, UK. 13 Current address: Division of Infectious Diseases and Vaccinology, School of Public Health, 14 University of California, Berkeley, USA. 15 Radcliffe Department of Medicine, Weatherall Institute of Molecular Medicine, University of 16 Oxford, John Radcliffe Hospital, Oxford, OX3 9DS, UK 17 18 equal contribution 19

MUSK encodes the muscle-specific receptor tyrosine kinase (MuSK), a key component of the agrin-LRP4-MuSK-DOK7 signaling pathway, which is essential for the formation and maintenance of highly specialized synapses between motor neurons and muscle fibers. We report a patient with severe early-onset congenital myasthenic syndrome and two novel missense mutations in MUSK (p.C317R and p.A617V). Functional studies show that MUSK p.C317R, located at the frizzled-like cysteine-rich domain of MuSK, disrupts an integral part of MuSK architecture resulting in ablated MuSK phosphorylation and acetylcholine receptor (AChR) cluster formation. MUSK p.A617V, located at the kinase domain of MuSK, enhances MuSK phosphorylation resulting in anomalous AChR cluster formation. The identification and evidence for pathogenicity of MUSK mutations supported the initiation of treatment with β2-adrenergic agonists with a dramatic improvement of muscle strength in the patient. This work suggests uncharacterized mechanisms in which control of the precise level of MuSK phosphorylation is crucial in governing synaptic structure.

Expansion microscopy (ExM) has been successfully used to improve the spatial resolution when imaging tissues by optical microscopy. In ExM, proteins of a fixed sample are crosslinked to a swellable acrylamide gel, which expands when incubated in water. Therefore, ExM allows enlarged subcellular structures to be resolved that would otherwise be hidden to standard confocal microscopy. Herein, we aim to validate ExM for the study of peroxisomes, mitochondria, nuclei and the plasma membrane. Upon comparison of the expansion factors of these cellular compartments in HEK293 cells within the same gel, we found significant differences, of a factor of above 2, in expansion factors. For peroxisomes, the expansion factor differed even between peroxisomal membrane and matrix marker; this underlines the need for a thorough validation of expansion factors of this powerful technique. We further give an overview of possible quantification methods for the determination of expansion factors of intracellular organelles, and we highlight some potentials and challenges.

Introduction: Kinase oxidation is a critical signaling mechanism through which changes in the intracellular redox state alter cardiac function. In the myocardium, the regulatory Iα subunit of Protein Kinase A (PKARIα) can be reversibly oxidised, forming interprotein disulfide bonds within the holoenzyme complex. However, the impact of disulfide formation on kinase function, and its influence on PKA signaling in the context of heart disease remains unknown. Methods & Results: Myocardial ischemia-reperfusion (I/R) was found to be a potent inducer of PKARIα disulfide formation in vivo , both in mice and in humans. Using imaging modalities with high spatial and temporal resolution, we found that this conformation did not increase intrinsic PKA catalytic activity, but rather facilitated enhanced AKAP-dependent compartmentation of PKARIα in the adult mouse left ventricular (LV) myocyte, with preferential localization to the lysosome under oxidized conditions (n=38-41 myocytes, N=3 animals, p<0.01). Investigations in isolated LV myocytes revealed disulfide-modified PKARIα to be a significant regulator of lysosomal two pore channel (TPC)-dependent calcium-induced calcium release, with myocytes from ‘redox dead’ PKARIα mice (Cys17Ser) displaying spontaneous sarcoplasmic reticulum calcium release events and pronounced intracellular calcium oscillations. These events were prevented by ryanodine receptor blockade (1 mM tetracaine; n=14, p<0.01), acute depletion of lysosomal calcium stores (100 nM bafilomycin; n=7; p<0.01), or TPC inhibition (5 μM Ned-19; n=9; p<0.05). Absence of I/R-induced disulfide formation in “redox dead” PKARIα mouse hearts resulted in larger infarcts (2-fold increase, p<0.001) and a concomitant reduction in LV contractile recovery (1.6-fold, p<0.001), which could be fully prevented by administering the TPC inhibitor, Ned-19, at the time of reperfusion. Conclusions: Oxidised PKARIα acts as a potent inhibitor of intracellular calcium release in the heart through its redox-dependent interaction with the lysosome. In the setting of I/R, where PKA oxidation is induced, this regulatory mechanism is critical for protecting the heart from injury and offers a novel target for the design of cardioprotective therapeutics.

Object detection networks are high-performance algorithms famously applied to the task of identifying and localizing objects in photography images. We demonstrate their application for the classification and localization of cells in fluorescence microscopy by benchmarking four leading object detection algorithms across multiple challenging 2D microscopy datasets. Furthermore we develop and demonstrate an algorithm that can localize and image cells in 3D, in close to real time, at the microscope using widely available and inexpensive hardware. Furthermore, we exploit the fast processing of these networks and develop a simple and effective augmented reality (AR) system for fluorescence microscopy systems using a display screen and back-projection onto the eyepiece. We show that it is possible to achieve very high classification accuracy using datasets with as few as 26 images present. Using our approach, it is possible for relatively nonskilled users to automate detection of cell classes with a variety of appearances and enable new avenues for automation of fluorescence microscopy acquisition pipelines.