ACS Photonics

Light-sheet microscopy (LSM) has revolutionized bioimaging by delivering high-contrast volumetric resolution with minimal photodamage. Spatial wavefront shaping, used to gen{-}erate lattice and Airy light-sheets, has been particularly effective in advancing LSM be{-}yond the Rayleigh limit. Despite its broad adoption, most LSM implementations rely on rigid dual-objective geometries that complicate sample handling and impose a trade-off between imaging field of view (FoV) and axial resolution. Here, we introduce space-time light-sheet microscopy (ST-LSM), a single-objective strategy that exploits space-time (ST) correlations for the first time. ST-LSM goes beyond separate spatial or temporal modulation to jointly modulate the spatiotemporal spectral structure of a pulse. This uniquely enabled light-sheets with wavelength-scale thickness over millimeter-scale dis{-}tances. When compared to state-of-the-art approaches, ST-LSM eliminates the dual-objective constraint, expands the sample-accessible volume by 25x, and increases the FoV by 10x without sacrificing sectioning resolution. We demonstrate the versatility of ST-LSM by using a single setup to image specimens across four orders of magnitude in size, from whole roots and developing embryos, down to mammalian cells with sub-cellular axial resolution. These results position ST-LSM as an accessible and high-performance optical microscopy platform at a variety of biological scales, by translating space-time wave-packet physics into a practical imaging modality.

Quantitative knowledge of nanoparticle properties is desirable in a large number of scientific and technological applications, but measurements with a high degree of precision usually prove to be challenging. Among a range of available methodologies, optical techniques with single particle sensitivity are especially interesting because they can reveal intrinsic hetero-geneities in a fast non-invasive manner. Recently, we presented interferometric nanoparticle tracking analysis (iNTA) as a highly sensitive label-free technique that is capable of determining the size, concentration and index of refraction of different subpopulations in a suspension mixture. Here, we enhance this method with biochemical specificity through multicolor fluorescence detection at the single-molecule sensitivity limit. We benchmark the performance of the combined technique, which we name iNTA-F, by distinguishing populations of fluorescent and non-fluorescent nanoparticles of different material, size, and fluorescence intensity, with an emphasis on the characterization of lipid vesicles and biological extracellular vesicles (EVs).

Tissue-clearing techniques have transformed optical imaging of fixed specimens, yet their application to living systems remains limited by toxicity and removal of key tissue components. We recently demonstrated that absorbing molecules such as tartrazine can reversibly render live mouse skin transparent. Subsequently, it was reported that isotonic protein solutions can achieve ex vivo and in vivo cellular clearing. However, discrepancies remain regarding the optimal refractive index (RI) for live-cell clearing and the impact of elevated osmolality on cell viability. Here, using cultured mammalian cells, we systematically examine the dependence of optical contrast on medium RI and the effects of hyperosmolality. We find that, contrary to the recent report of an optimal RI of 1.36[~]1.37 for suspended cells, densely-packed adherent cells exhibit a monotonic decrease in phase contrast up to an RI of 1.41 with tartrazine. Moreover, even under highly hyperosmotic conditions ([~]1200 mOsm/kg), cultured cells exhibit minimal deformation and negligible loss of viability for up to 30 min in the clearing solution. These results demonstrate that tartrazine enables effective live-cell clearing at RI up to 1.41 while preserving viability under elevated osmolality, and motivate future studies to define optimal conditions for in vivo optical clearing. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=44 SRC="FIGDIR/small/717314v1_ufig1.gif" ALT="Figure 1"> View larger version (17K): org.highwire.dtl.DTLVardef@1c45280org.highwire.dtl.DTLVardef@483a5org.highwire.dtl.DTLVardef@5ed60forg.highwire.dtl.DTLVardef@377714_HPS_FORMAT_FIGEXP M_FIG C_FIG

Three-photon microscopy (3PM) has enabled the optical access of neurons [~]500-1500{micro}m below the brain surface but has been limited to slow imaging frame rates or small imaged area due to the combination of a nonlinear peak power requirement and the need to limit average power below the thermal damage threshold. High sensitivity to laser fluctuation and inherently dim signals introduce additional challenges and add error. Combined with the effects of brain motion in behaving animals, 3P imaging of neuronal activity during animal behavior has remained practically unachievable. Herein, we systematically address these limitations by carefully balancing scanning speed with power requirements, using a deeply cooled silicon photomultiplier detector with Bayesian statistics-based processing to reduce excess noise, and through spatiotemporal shaping of excitation pulses. Our improvements enable rapid (20-30Hz) imaging of calcium activity in the dorsal hippocampal dentate gyrus of behaving mice, allowing the identification of spatially tuned neurons and the recapitulation of established functional properties across different cell types in this brain region. PRED-3P imaging provides a new approach to functional characterization of cells deep in the brain that were previously inaccessible to two-photon imaging.

Spatiotemporal control over cell fate and behaviour within bioprinted constructs remains a key challenge in tissue engineering. Optogenetics offers versatile potential for non-invasive regulation of biological processes. Yet, its integration within large-scale, cell-laden bioprinted materials is still limited, especially considering spatial constraints of existing light delivery methods. In this study, we introduce a novel approach that repurposes tomographic volumetric bioprinting to enable post-printing stimulation of photosensitive protein-switches and optogenetic circuits in cells deep within hydrogel constructs. By converging different bioprinting approaches, computer vision, context-aware model generation, and synthetic biology and cell engineering, we demonstrated selective activation of a fluorescent, light-responsive protein probe within multi-material centimeter-scale constructs. Moreover, leveraging a multi-wavelength volumetric bioprinter, we further demonstrate this concept by selectively stimulating cells expressing a near-infrared optogenetic system that triggers gene expression and the induction of pancreas-specific transcription factors. The described methods provide platforms for remote, repeatable, and localized control of biological events in volumetric constructs, opening new possibilities for advanced tissue models, and dynamic tuning of cell-mediated protein production in engineered living systems.

In this study, we investigate the individuality and information content of infrared molecular profiles derived from blood samples in a large, longitudinal health-profiling cohort and compare them to a standard clinical laboratory panel. Using Fourier-transform infrared spectroscopy, we obtained comprehensive molecular fingerprints from 4,704 self-reported healthy individuals over five visits spanning 1.5 years, alongside routine clinical laboratory measurements. We show that infrared profiles are highly individual-specific and remarkably stable over time, with intra-individual variability significantly lower than inter-individual differences--paralleling the characteristics observed in clinical laboratory data. To quantify and compare the information content of these molecular datasets, we employ individual identification as a proxy for Shannon entropy. In this framework, higher identification accuracy reflects a higher amount of information. Infrared profiles outperform the clinical laboratory panel in identifying individuals at scale, suggesting higher intrinsic information content. Furthermore, combining infrared and clinical laboratory data substantially improves identification performance (the identification of less than 3000 individuals by the clinical laboratory panel is boosted to more than 4000 by incorporating the infrared spectroscopic markers), highlighting the value of integrating complementary data modalities. These findings suggest a practical framework, rooted in information theory, for comparing molecular profiling approaches and emphasize the potential of infrared spectroscopy as a complementary tool in personalized medicine.

Kidney organoids degrade in long-term culture and lack joint basement membranes between epithelial and endothelial cells characteristic of renal tissue. Here we show that these limitations can be overcome in static cultures simply by optimizing the microenvironment. Supplementing standard media with tubular-enhancing factors (TEFs) dramatically improves organoid yield and longevity, while vascular-enhancing factors (VEFs) and replating increases endothelial cell yield and invasiveness. A transcriptomic and imaging atlas demonstrates maintenance of nephron structures for six months with increased metabolism, signaling, differentiation, and aging-related pathways. In addition to adherent cultures, these media also enable organoid differentiation and vascularization in suspension cultures and hydrogels. Remarkably, addition of TEFs and VEFs to organoids in suspension induces self-assembly of joint basement membranes between endothelial cells and podocytes or tubules, a major feature of renal tissue. Microenvironment optimization thus enables longitudinal stabilization and higher-order vascularization of kidney organoids, offering a diverse resource for long-term studies and tissue engineering applications.

Enhancing targeted delivery of biomedicines improves efficacy and can reduce off-target effects by lowering the effective dose, but achieving targeting is challenging. Extracellular vesicles (EVs) are promising biological nanovesicles which can be targeted by displaying binding proteins and are being developed as therapeutics. Currently, discovering EV targeting constructs is limited by low throughput and resource-intensive EV production and isolation. To accelerate discovery, we developed a screening pipeline to identify EV targeting constructs without requiring EV production. This approach is premised on the hypothesis that cell-cell interactions may predict some cell-EV interactions. Our cell binding assay (CELLISA) quantifies binding of a cell surface-displayed targeting protein to its cognate receptor on a target cell, employing a microscopy-based analysis pipeline. After validating the premise using existing T cell-targeting reagents, we develop CELLISA for either adherent or suspension EV producer cells. Finally, we use CELLISA to evaluate new binders and validate that hits mediate targeting and/or delivery of genetic cargo to natural killer cells and T cells. CELLISA increased throughput > 6-fold and decreased time by 40% compared to standard EV screens, and it identified a T-cell binder conferring efficient gene delivery. CELLISA is easily adaptable to other laboratories and can accelerate EV research.

DNA double-crossover (DX) molecules, comprising two Holliday junctions connected by two duplex arms, are fundamental building blocks of DNA nanostructures, but their mechanical properties remain poorly understood. Here we investigate the elasticity of isolated antiparallel DX motifs with 18 to 22 base pairs between the crossovers. Using mechanical models parameterized by extensive all-atom molecular dynamics simulations, we demonstrate that the bending rigidity of the duplexes within a DX motif is highly anisotropic, and that this anisotropy results from long-range elastic couplings involving all the duplex base pairs between the crossovers. The duplex stretch modulus decreases due to localized defects, while the twist stiffness is close to that of an isolated duplex. The DX core as a whole follows an analytical beam theory in bending but not in torsion. Our results extend beyond local elastic models of DNA nanostructures and pave the way for probing peculiar mechanical properties of other key motifs for DNA and RNA nanotechnology.

Biomolecular condensates are increasingly implicated as intermediates in the formation of pathological amyloid assemblies, yet the mechanisms by which sequence-encoded structural motifs and non-equilibrium molecular transport cooperate at condensate interfaces remain incompletely understood. Here, we introduce Flux-Driven Molecular Dynamics (FD-MD), a simulation framework that combines sequence-encoded {beta}-prone interactions with sustained molecular influx to examine fibril formation at condensate interfaces. Within this framework, we establish three main results. First, a scaling analysis of orientational entropy suggests that condensate interfaces can enhance nucleation relative to the bulk by as much as two orders of magnitude, by reducing the entropic cost of coalignment of rigid {beta}-prone segments. Second, varying segment rigidity and molecular supply rate organizes a non-equilibrium phase diagram with four interfacial growth morphologies, ranging from uniform wetting to fibrillar protrusions and inter-condensate bridging networks. Third, directional fibril elongation displays an inverse relationship with drift velocity, consistent with a mechanism in which higher transport rates to the interface favor planar saturation over directed tip incorporation. Together, these results support a picture in which condensate interfaces can act as kinetically favorable nucleation environments, sequence-encoded rigidity helps determine whether interfaces remain liquid-like or become fibrillar, and molecular flux emerges as an additional control axis in the model for condensate aging trajectories.

Bispecific T cell engagers and related immunotherapies are dosed using equilibrium binding models derived for well-mixed solution, yet therapeutic activity occurs at nanoscale membrane synapses with finite receptor copy numbers. Here we show that membrane confinement introduces geometry-dependent corrections to the landmark Douglass [6] ternary binding model, shifting formation half-points (TF50) by 2-10-fold at clinically relevant antigen densities. We present two complementary formulations--effective concentration and surface density--that preserve the Douglass framework while explicitly accounting for synapse geometry, surface topology, and the accessibility factor () of surface receptors. We further derive stochastic descriptions of trimer formation via the chemical master equation, demonstrating the recovery of the classical Ternary Binding Model equilibrium in appropriate limits. We illustrate the framework using the CD19-targeting BiTE blinatumomab as a case study. Accounting for microvillus-driven patchy close contact during immune surveillance yields a mechanistic explanation for why higher target antigen density can increase the dose required to achieve a fixed level of ternary formation: in the membrane-confined regime, excess target acts as a local antigen sink that sequesters drug and reduces the free fraction available for productive bridging. Rather than fitting to a single shift value, we emphasize the robust scaling and regime structure predicted by the theory (density-proportional behavior in the sink-dominated limit, and collapse toward affinity-limited behavior outside that limit). The generalized framework provides ready-to-use correction formulas and parameter-estimation guidance, establishing a rigorous foundation for antigen-density-aware dosing strategies in T cell engager pharmacology. Statement of SignificanceBispecific T cell engagers are currently interpreted largely through bulk-solution binding models, but at nanometer-scale immune synapses those models can miss a distinct, decision-relevant regime. We generalize the standard ternary binding model to membrane-confined synapses and relate whole-cell receptor counts to effective concentrations in the contact zone. In a blinatumomab case study, the framework explains a paradoxical observation: increasing target density can shift half-maximal ternary-complex formation to higher doses because abundant antigen acts as a local sink that sequesters drug. These results show when bulk affinity alone is insufficient for potency interpretation and support antigen-density-aware dosing and experiments that distinguish geometric from chemical control.

Intrinsically Disordered Proteins and protein regions (IDPs) are abundant in many viral proteomes and play diverse roles in the viral infectious cycles. The adenovirus Early Protein 1A (E1A) is one such viral IDP. E1A acts as a molecular hub that regulates viral infection by mediating interactions between viral and multiple host proteins. Like other IDPs, E1A exists in a flexible ensemble of conformations. Despite a demonstrated link between ensemble structure and function in E1A, no real-time measurement of its ensemble has been performed. Here, we use live cell FRET microscopy to measure the local ensemble structure of E1A in human cells, both in healthy cells and in cells infected with adenovirus. We found specific disordered regions undergo significant changes to their ensemble in infected cells. Furthermore, infection also alters the propensity of these regions to partition between the cytoplasm and nucleus, a hallmark of E1A function during infection. Our results showcase that the structural ensembles of viral IDPs are responsive to infection, and suggest that these may play a role in regulating infection progression. SignificanceIntegral to many viral proteomes are intrinsically disordered proteins and protein regions (IDPs), which target and rewire cellular pathways to ensure infection progression. During this process, the physical and chemical composition of the cell changes dramatically: metabolism is rewired, viral proteins are produced en masse, and as a result, the chemical composition of the host proteome is significantly altered. IDPs are known to be structurally sensitive to even mild changes in their environment, and their structural changes can result in a change to function. Here, using live cell FRET microscopy, we show that the structure and spatial localization of an adenovirus IDP, E1A, is altered in cells infected by the virus. Beyond a possible functional role for structural sensitivity in viral IDP function, our findings suggest that host IDPs may also be structurally altered by infection, with downstream functional consequences.

Fibrils formed by 40- and 42-residue amyloid-{beta} peptides (A{beta}40 and A{beta}42) are polymorphic, containing molecular structures that vary with growth conditions in ways that are not fully understood. Here we use cryogenic electron microscopy to characterize the structure of rapidly twisting A{beta}40 fibrils, for which the distance between apparent width minima in electron microscope images ("cross-over distances") is approximately 25 nm. From samples grown under a single set of growth conditions, we obtain high-resolution structures for three different rapidly twisting polymorphs. Although their cross-over distances are similar, the three rapidly twisting polymorphs differ in twist handedness, symmetry, molecular conformations, and intermolecular contacts. Two of the rapidly twisting polymorphs resemble slowly twisting A{beta}40 polymorphs that have been described previously, including polymorphs extracted from brain tissue of Alzheimers disease patients or created by seeded growth from amyloid in brain tissue, but with shorter conformationally ordered segments and other specific conformational differences. These results contribute to our understanding of amyloid polymorphism, connections between morphology and molecular structure, and relationships between brain-derived and in vitro-grown fibrils.

Optogenetics is emerging as a powerful approach for partial vision restoration, with at least three ongoing clinical trials in humans testing novel light-sensitive proteins (opsins) in patients with inherited retinal degenerative disorders. These therapies aim to restore light responsiveness by introducing opsins into surviving retinal cells, such as bipolar or ganglion cells, enabling them to generate neural activity in response to visual stimuli. One ongoing difficulty in selecting promising opsins for clinical development is that there is no way to predict patient perceptual outcomes from optogenetically evoked neural activity as measured ex vivo. Here, we introduce a virtual patient framework that quantitatively links the sensitivity and speed of opsin-mediated retinal responses to predicted patient outcomes, and show how this framework can predict temporal contrast sensitivity functions - a well-established measure of perceptual performance - from microbial opsin photocurrent responses. Our simulations demonstrate that opsin sensitivity and kinetics jointly determine perceptual outcomes, and that enhancing sensitivity at the expense of temporal resolution can degrade the perception of fast-moving stimuli. This computational platform provides a generalizable tool for comparing and selecting the most effective opsins for clinical translation, thereby guiding the design and optimization of next-generation sight restoration strategies.

Phages are the most abundant biological entities on Earth, yet RNA phages are strikingly scarce compared to their DNA counterparts--a long-standing mystery in phage biology. Here, we use dsRNA phage phiYY as a model to demonstrate that while RNA phages efficiently infect growing bacteria, they are gradually eliminated inside dormant bacteria through weak and time-dependent ROS-mediated damage. The RNA phages decline over days rather than through immediate clearance inside dormant bacteria. Accordingly, scavenging ROS with mannitol or overexpressing ROS degradation enzymes AhpB/TrxB2 rescues RNA phages from elimination. Crucially, because the underlying ROS-mediated RNA damage is minimal, RNA phage survival hinges on genomic redundancy. In single-phage infections, the lone RNA genome is highly vulnerable to cumulative damage and is eventually inactivated. In contrast, during co-infection by multiple RNA phages, the presence of multiple genome copies provides functional redundancy, thereby allowing a fraction of RNA phages to survive inside dormant bacteria. Given that most environmental bacteria are dormant and subject to heterogeneous phage infection, this copy number-dependent vulnerability offers a possible explanation for the scarcity, but not perish, of RNA phages in nature.

Autosomal dominant polycystic liver disease (ADPLD) commonly results from a pathogenic variant in one of 6 genes (GANAB, ALG8, LRP5, PRKCSH, SEC61B, SEC63). Pathogenic variants in these genes are also associated with kidney cysts, which rarely cause kidney failure, but the genes are included in cystic kidney panels. This study determined the population frequency of predicted pathogenic variants in the ADPLD genes in the general population. Variants for each gene were downloaded from gnomAD and annotated with ANNOVAR. The population frequencies were calculated from the number of people with "predicted pathogenic" variants in gnomAD v.2.1.1:loss-of-function structural and copy number; null; and rare, computationally-damaging missense changes that affected a conserved residue. Frequencies were also estimated from the number of gnomADv.4.1 variants assessed as Pathogenic or Likely pathogenic in ClinVar. Predicted pathogenic variants affected one in 95 people using our strategy and gnomAD v.2.1.1, and one in 151 with ClinVar assessments of gnomAD v.4.1 variants. LRP5 and ALG8 which are associated with a milder clinical phenotype, were the commonest affected genes with both strategies. Predicted pathogenic variants in ADPLD appear more frequent in admixed American (one in 100), Finnish (one in 107) and African/African American (one in 130) people (p all <0.0001 compared with Europeans (one in 197).Predicted pathogenic variants for ADPLD may be even more common because of additional unidentified causative genes. However not all ADPLD variants result in liver cysts, nor indeed cystic kidneys, because of incomplete penetrance and variable expressivity.

Saccadic eye movements are established biomarkers in neuroscience and clinical neurology, where video-oculography (VOG) remains the gold standard. However, VOG's high cost, bulky equipment, and poor portability restrict its clinical utility. Electrooculography (EOG) offers a promising alternative by detecting cornea-retinal potential changes during eye movements. To enable quantitative saccadic analysis using EOG as a VOG alternative, this study develops and validates a mathematical transformation model converting EOG data into VOG-equivalent values. A prospective observational study was conducted on 4 healthy adults without neurological or sleep disorders. Horizontal saccades were recorded simultaneously using EOG and VOG during controlled gaze shifts. EOG peak saccadic velocity was derived from voltage change rate, whereas VOG was calculated from angular displacement over time. A derivation dataset of fixed horizontal saccades ({+/-}20{degrees}) formulated the transformation model, achieving a strong correlation coefficient (r = 0.95 rightward, r = 0.93 leftward, p < 0.0001). Multiple filter settings were evaluated, and 0.3 Hz high-pass and 35 Hz low-pass filtering were identified as optimal. The fixed horizontal saccades derived model was applied to a validation dataset of random horizontal saccades, confirming robustness across saccades without significant differences from VOG measurements. These findings establish EOG's feasibility for quantitative analysis of horizontal saccades and provide a validated transformation model. By systematically optimizing filtering parameters, this approach enables EOG as a cost-effective VOG alternative while maintaining high-precision measurement accuracy.

Introduction: Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease with limited treatment options. Masitinib, a tyrosine kinase inhibitor targeting microglial and mast cell activity in ALS pathogenesis, offers potential neuroprotection. This study presents a post-hoc analysis of long-term survivors treated with masitinib at 4.5 mg/kg/day in study AB10015, comparing observed survival to predicted and historical benchmarks. Methods: Study AB10015 was a randomized, double-blind, placebo-controlled trial assessing masitinib with riluzole in ALS patients. Overall survival (OS) was measured from symptom onset to death, encompassing the double-blind period and post-study follow-up, including an optional open-label program. The ENCALS model predicted survival of long-term survivors ([&ge;]5 years). A delay in the need for mechanical assistance, such as permanent ventilation, gastrostomy, tracheostomy, or wheelchair dependence, was used as a surrogate measure for quality of life (QoL). Results: Among 130 patients receiving masitinib 4.5 mg/kg/day, the 5-year survival rate from onset was 42.3%, increasing to 50.0% in patients with an ALSFRS-R progression rate from disease onset of <1.1 points/month (AB10015 primary efficacy population), and 52.9% in a subgroup of patients without complete loss of functionality at baseline. Half of the long-term survivors had satisfactory QoL, defined as no mechanical assistance. The median OS for long-term survivors (n=55) was 121 months versus the ENCALS-predicted 42 months, yielding a 79-month residual median survival gain. Long-term survivors were prevalent across ALS baseline prognostic factors, including slow or moderate disease progression rate ({Delta}FS), severe or moderate functional severity, bulbar or spinal site of onset, respiratory function, and age. Long-term survival was less likely in patients with complete loss of function at baseline or fast progressing disease ({Delta}FS [&ge;]1.1 points/month) at baseline. Conclusions: Masitinib treatment in ALS patients showed substantial survival benefit. Long-term survivors were largely independent of ALS prognostic factors, suggesting a subpopulation driven by microglial/mast cell activity. A recently identified biomarker detecting masitinib effect on pro-inflammatory microglia may help identify responsive patients.

Background: Large language models (LLMs) are increasingly used in mental health contexts, yet their detection of suicidal ideation is inconsistent, raising patient safety concerns. Objective: To evaluate whether an independent safety monitoring system improves detection of suicide risk compared with native LLM safeguards. Methods: We conducted a cross-sectional evaluation using 224 paired suicide-related clinical vignettes presented in a single-turn format under two conditions (with and without structured clinical information). Native LLM safeguard responses were compared with an independent supervisory safety architecture with asynchronous monitoring. The primary outcome was detection of suicide risk requiring intervention. Results: The supervisory system detected suicide risk in 205 of 224 evaluations (91.5%) versus 41 of 224 (18.3%) for native LLM safeguards. Among 168 discordant evaluations, 166 favored the supervisory system and 2 favored the LLM (matched odds ratio {approx}83.0). Both systems detected risk in 39 evaluations, and neither in 17. Detection was highest in scenarios with explicit suicidal ideation and lower in more ambiguous presentations. Conclusions: Native LLM safeguards frequently failed to detect suicide risk in this structured evaluation. An independent monitoring approach substantially improved detection, supporting the role of external safety systems in high-risk mental health applications of LLMs.

BackgroundProspective studies of pregnant adolescents are essencial to effectively address this global health priority. They help answer vital questions about their health, but such studies are uncommon due to the difficulty in retaining adolescents. This paper describes the successes and challenges of the research strategies used to ensure sufficient recruitment and retention of pregnant adolescents in a longitudinal study about adolescent childbearing in an under-resourced setting. MethodsThe Adolescence and Motherhood Research project was conducted in a rural region of Northeast Brazil in 2017-2019 and assessed 50 primigravids between 13-18 years (adolescents) and 50 primigravids between 23-28 years (young adults) during the first 16 weeks of pregnancy with two follow-ups (third trimester of pregnancy, and 4-6 weeks postpartum). Recruitment strategies involved engagement of health sector and community, as well as referrals from health care professionals and dissemination of the project in different locations. Retention strategies included maintaining contact with the participants between assessments and providing transportation for them to attend the follow-up procedures. ResultsRecruitment took 10 months to complete. A total of 78% of the participants were recruited from the primary health care units, mainly after referral from a health care provider. Retention reached 95% of the sample throughout the study (90%: adolescents; 98%: adults). ConclusionA combination of approaches is necessary to successfully recruit and retain youth in longitudinal studies and engaging local stakeholders may help to increase community-perceived legitimacy of the research. Working closely with front-line staff is essential when conducting research in rural low-income communities.