Decision letter: Label-free imaging of immune cell dynamics in the living retina using adaptive optics [peer_review]

Johnny Tam, Timothy Secomb, Tyson Kim, Anna Akhmanova, Johnny Tam
2020 unpublished
Our recent work characterized the movement of single blood cells within the retinal vasculature (Joseph et al. 2019) using adaptive optics ophthalmoscopy. Here, we apply this technique to the context of acute inflammation and discover both infiltrating and tissue-resident immune cells to be visible without any labeling in the living mouse retina using near-infrared light alone. Intravital imaging of immune cells can be negatively impacted by surgical manipulation, exogenous dyes, transgenic
more » ... yes, transgenic manipulation and phototoxicity. These confounds are now overcome, using phase contrast and time-lapse videography to reveal the dynamic behavior of myeloid cells as they interact, extravasate and survey the mouse retina. Cellular motility and differential vascular responses were measured noninvasively and in vivo across hours to months at the same retinal location, from initiation to the resolution of inflammation. As comparable systems are already available for clinical research, this approach could be readily translated to human application. Joseph, Chu, et al. eLife 2020;9:e60547. DOI: https://doi.org/10.7554/eLife.60547 1 of 14 RESEARCH ADVANCE single-cell blood flow in vessels (Joseph et al., 2019) revealing the dynamic interplay of blood flow and single immune cells in response to inflammation in the living eye. Our approach is in a similar power range to those already employed safely in human AOSLO studies, where phase contrast approaches have also been used to study other retinal cell types Rossi et al., 2017) . This speaks to the feasibility of translating this technique to the clinic. Results To model an immune response, ocular injection of lipopolysaccharide (LPS) was used to provide an acute but self-resolving inflammatory stimulus: endotoxin induced uveitis (EIU) (Chu et al., 2016) . Potential immune cells were observed adjacent to retinal veins only after image registration, frame averaging and time-lapse imaging ( Figure 1A , Video 1). Membrane remodeling, pseudopodia formation and motility (consistent with immune cell structure and function) were visible, distinct from static neurons or macroglia ( Figure 1B and Video 2). Within post-capillary venules, leukocyte rolling, crawling, and trans-endothelial migration behaviors were detectable ( Figure 1C , Videos 3 and 4). Heterogeneity in cell distribution, size and morphology was imaged with multiple cell types in different stages of interaction ( Figure 1D ). We verified these cells comprised neutrophil and monocyte populations by fluorescent marker co-localization. Simultaneous phase contrast and confocal fluorescence AOSLO revealed most leukocytes rolling along venular endothelium were neutrophils using intravenous anti-Ly6G fluorescent antibody labeling, although the proportion may be higher as only 10% of circulating leukocytes may be labeled using this method (Bucher et al., 2015; Figure 1E , Video 5). Conversely, CD68 GFP/+ mice distinguished a population of cells were infiltrating monocytes and macrophages present both in vessels and extravasated into retinal tissue ( Figure 1F ). More cells were visible using phase contrast than by fluorescence labeling, demonstrating its utility for comprehensively detecting diverse and mixed cellular populations. Tissue resident myeloid cells were also visible by AOSLO phase contrast even in healthy eyes without LPS injection. These were confirmed as microglia or hyalocytes by colocalization of Cx3cr1 GFP/+ and CD68 GFP/+ fluorescence ( Figure 1G -I; Lazarus, 1994). Phase contrast even revealed subcellular features, including structures that could represent internal processes such as endosomes ( Figure 1H , Video 6; Uderhardt et al., 2019). As our approach is uniquely non-invasive, repeated imaging at the same tissue location permits longitudinal study throughout the initiation, peak and resolution of an immune response across hours to months within individual eyes ( Figure 1J, Video 7) . To quantify immune cell behaviour in these studies, we had to distinguish immune cells from surrounding tissue by using semi-automated deep learning strategy (Falk et al., 2019) . This correlated well with counts made by masked human observers (R 2 = 0.99, p=0.004, Video 8 (final segment)). Immune cell metrics were quantified in six mice over five timepoints following LPS injection (Figure 2A-C) . Compared to baseline (28.9 ± 34.1 cells/mm 2 , Mean ± SD) a seven-fold influx of cells was detected by 6 hr post-injection (208.3 ± 108.6 cells/mm 2 ) rising to over an 18-fold increase by 24 hr (510.4 ± 441 cells/mm 2 ) before returning toward baseline at 72 hr (59.0 ± 27.8 cells/mm 2 ) and 10 days (69.4 ± 41.7 cells/mm 2 ). AOSLO also allowed cell motility quantification with maximum cell displacement observed at 6 hr (16.1 ± 9.9 mm, n = 12 cells). Despite peak infiltration at 24 hr, motility was greatly reduced (4.6 ± 4.3 mm, n = 58 cells), best appreciable by longitudinal imaging, consistent with Resolvin-mediated suppression of chemotaxis (Schwab et al., 2007) . As retinal tissue is not depressurized by this intravital system, true vascular alterations arising from inflammation can be isolated and correlated to simultaneous immune cell measurements. Adapting our recent work (Joseph et al., 2019) , red blood cell (RBC) velocimetry, vessel dilation and flow-rate changes were quantified in this same cohort of mice ( Figure 2D -H). AOSLO revealed micron-level vascular dilations and heterogeneous changes in blood flow in arterioles and venules in response to LPS. Total blood flow increased in the retinal circulation, yet elevated flow in arterioles and venules was achieved in fundamentally different ways. Venules dilated on average 36% (±8%) at 24 hr post-LPS injection facilitating a total flow increase of 67% (±27%) relative to baseline. Conversely, arterioles also showed a total flow increase, however with minimal dilation and a dramatic elevation in RBC velocity (48% (±31%) increase in arteriole RBC velocity at 24 hr) ( Figure 2D-F) . Joseph, Chu, et al. eLife 2020;9:e60547. DOI: https://doi.org/10.7554/eLife.60547 Colin J Chu Worldwide Universities Network Research Mobility Programme Award Colin J Chu The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
doi:10.7554/elife.60547.sa1 fatcat:lqyj52gvnbff7aif3pstsj7x6y