|Home | About | Journals | Submit | Contact Us | Français|
Two-photon intravital microscopy is a powerful tool that allows the examination of dynamic cellular processes in the live animal with unprecedented resolution. Indeed, it offers the ability to address unique biological questions that may not be solved by other means. While two-photon intravital microscopy has been successfully applied to study many organs, the kidney presents its own unique challenges that need to be overcome in order to optimize and validate imaging data. For kidney imaging, the complexity of renal architecture and salient autofluorescence merit special considerations as these elements directly impact image acquisition and data interpretation. Here, using illustrative cases, we provide practical guides and discuss issues that may arise during two-photon live imaging of the rodent kidney.
The kidney is a highly complex organ that consists of more than 10 specialized epithelial cell types interlaced with numerous immune and interstitial cells as well as an organized vascular system. Intravital multiphoton microscopy studies conducted over the past 15 years have provided unique insights into kidney physiology and pathophysiology. However, there are many subtle issues commonly encountered that can complicate image acquisition when studying various aspects of kidney structure and function in the live animal. We have previously described in detail the basic surgical and microscopy platform we utilize for 2-photon imaging of the kidney [1–4]. We have also reported on the detailed use of various probes and delivery methods including advantages and limitations [5–15]. Therefore, in this review, we will focus on illustrating with examples specific aspects of 2-photon imaging of the kidneys including commonly encountered pitfalls and difficulties.
All animal protocols were approved by Indiana University Institutional Animal Care Committee and conform to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Male C57BL/6J mice, B6129S1-Casp3tm1F1v/J (caspase 3 knockout mice), B6;129S-Gt(ROSA)26Sortm1.1(CAG-COX8A/Dendra2)Dcc/J (photo-activatable mitochondria mice), B6.129P-Cx3cr1tm1Litt/J (CX3CR1-EGFP), and C57BL/6-Tg(CAG-EGFP)1Osb/J (all 8–10 weeks of age) were obtained from The Jackson Laboratory.
In some experiments, mice were subjected to a single dose of 5 mg/kg Alexa Fluor 568-labeled LPS from S. minnesota (Life Technologies and Sigma-Aldrich). Hoechst 33342 (Life Technologies), dissolved in normal saline, was administered intraperitoneally as a 2 mg/kg bolus 1–2 h before imaging. FITC-inulin (Sigma), a marker of tubular flow was administered intravenously (25 ng/kg). Tetramethylrhodamine methyl ester (TMRM, Life Technologies) was administered intravenously as a 10 μg/kg bolus from a DMSO stock diluted in normal saline. Phiphilux G2D2 (Calbiochem) 570 μl per mouse was injected iv through a jugular line and the kidney was imaged for periods up to 2 hours. The pH sensitive liposomes (DOPE/CHEMS) were a kind gift from Dr. Rudolph Juliano . Approximately 80 mM of calcein was entrapped in 24 mM phospholipids. To remove free calcein, the liposomes were centrifuged (3000g) for 10 min immediately before use. The pellet was resuspended with 2 mL of TBS and 200 μl of the resuspended liposomes was injected via tail vein. E. coli were transformed with CFP plasmid (BL21 Star DE3 and pRSET/CFP, Life Technologies) and incubated in LB media until OD600 was achieved. Approximately 0.5x106 CFU E coli were injected via tail vein.
Live animal imaging was performed using an Olympus FV1000-MPE confocal/multiphoton microscope equipped with a Spectra Physics MaiTai Deep See laser and gallium arsenide detectors, with signal digitized to 12 bit resolution. The system is mounted on an Olympus Ix81 inverted microscope stand with an Olympus 20× and 60× NA 1.2 water-immersion objective . For most experiments, the laser was tuned to 800 nm wavelength. Animals were placed on the stage with the exposed intact kidney placed in a coverslip-bottomed cell culture dish (HBSt-5040; Warner Instruments) bathed in isotonic saline. Two ReptiTherm pads (Zoo Med) and a heated water jacket blanket were used to maintain the temperature at 36 °C.
A typical two-photon microscopy setup (non-descanned detector) allows us to image a rodent kidney down to about 100 μm in depth from the renal capsule (Fig. 1A). Because of strong light scattering, deeper live imaging is currently limited as compared to other organs such as the brain, where less scattering species are present [1,18,19]. Improving imaging depth can be achieved with: 1) an increment in laser power, 2) the use of a high numerical aperture of the objective lens with matched refractive indices of the sample and immersion medium, 3) a long free working distance (e.g., Olympus objectives: XLPLanN 25x/1.05W MP, UPlanSApo 30x/1.05 Silicon oil, XLUMPlanFI 20x/0.95 W), 4) longer excitation wavelength as achieved with an optical parametric oscillator [20,21]. These approaches have been used with variable success and will not be discussed further in this review.
Tissue autofluorescence may be problematic because it can overlap with the emission of commonly used probes. However, it can also be harnessed to define various structures in the field under examination. Fig. 1B shows a typical appearance of mouse kidney cortex. Note the presence of abundant proximal tubules (PT) recognized by their thickness and strong autofluorescence. Distal tubules, collecting ducts and thick ascending limbs are also seen as thinner tubules with dimmer autofluorescence. Surface cortical glomeruli are scarce especially in an adult mouse . Importantly, note the presence of two distinct populations of PT segments, one with fine bright green punctate autofluorescence and the other with coarse red-green (appearing as brown color) autofluorescence. These two populations of PT segments represent S1 and S2 tubules, known to differ in their metabolic profiles and hence, their autofluorescence. To distinguish upstream S1 from downstream S2 segments, we followed the sequential appearance of systemically injected fluorescently-labeled inulin (Fig. 1B–D). All experiments confirmed that the tubules with red-green coarse autofluorescence are indeed upstream S1 while the tubules with fine bright green autofluorescence are downstream S2 . Thus, autofluorescence can be useful in defining the various structures in the cortex without the use of exogenous probes.
Prolonged time-lapse imaging offers the ability to capture “slow” dynamic cellular changes that may not be otherwise apparent. Here we present two examples that highlight the power of extended kidney imaging. In Fig. 2A and B we show the dynamic development of acute tubular necrosis observed over 75 min (see also Supplemental Video 1). Note the flattening of the lining epithelium, swelling of the lumen and shedding of epithelial cells. These dynamic changes at subcellular resolution cannot be appreciated with short imaging times. Another example involves the trafficking of immune cells in the renal interstitium. The kidney hosts a breadth of immune cells that participate in the pathophysiology of a variety of renal diseases. The way in which immune cells interact with each other and with other renal structures is poorly understood. Using a transgenic mouse that expresses eGFP in all cell types and treated with endotoxin, we imaged a renal cortical field over a one hour time period. Note the trafficking of immune cells and their interactions with tubules and glomerular structures (Fig. 2C–E; Supplemental Video 2). These are likely slow moving macrophages and their activity can be easily missed with shorter imaging times.
Fluorescent probes have been used extensively to characterize and image cellular processes in vitro. However, the in vivo use of these probes is met with unique challenges primarily related to probe delivery and distribution. Furthermore, many probes require strict microenvironmental physico-chemical parameters for optimal fluorescence. These can be easily set in vitro but much harder to control in vivo. In the kidney, the complex cellular architecture, glomerular filtration barrier and vigorous PT endocytosis further dictate that very careful consideration be given to the selection of fluorescent probes for intravital imaging. Commonly used primary probes for intravital renal imaging are listed in Table 1. A number of secondary probes (endogenous or exogenous molecules conjugated to fluorescent probes) are also used to image various processes in the kidney after systemic injection. Finally, the list of transgenic animals with fluorescent indicators continues to grow. Some of the genetically encoded probes have the advantage of bypassing the technical difficulties associated with in vivo probe use we discuss here [23–25].
We first address the importance of probe size. Because of the size selectivity of the glomerulus, a large molecular probe will not be filtered (typically >40 Kd or >4 nm) [26,27]. Conversely, smaller probes are freely filtered and readily appear in the urinary lumen. Thus, the half-life of a large probe in the circulation is generally much longer than that of a small molecular probe. This size difference can be harnessed to measure parameters such as glomerular filtration rate (GFR), blood and tubular flow. For example, we used high-molecular weight poly (I:C) (not filtered) to label Toll Like Receptor 3 (TLR3) positive immune cells in the circulation and interstitial space  (Fig. 6A). The use of low-molecular weight poly (I:C) could lead to low labeling efficiency due to the probe loss in the urinary lumen as a function of GFR. For filtered probes, the possible presence of PT fluid-phase and/or receptor mediated endocytosis requires additional consideration. For example, internalized probes may get decomposed or quenched in the endosomal/lysosomal compartment resulting in an underestimation of the reabsorption magnitude .
It is possible to distinguish fluid-phase from receptor-mediated endocytosis. This is done traditionally by analyzing the kinetics and co-localization of the probe with a known fluid-phase markers such as low molecular weight dextran . In addition, if a transgenic knockout mouse is available for that particular receptor, it can be used to estimate the residual fluid-phase reabsorption component. We made use of such a model to distinguish the two components of endotoxin uptake in the PT. In Fig. 3A, we show that systemically administered Alexa-labeled endotoxin is internalized primarily by S1 proximal tubules in wild-type mice. In a TLR4 knockout mouse, we documented the presence of residual mild endotoxin uptake (Fig. 3B). This residual uptake likely represents receptor-independent fluid phase endocytosis .
Exogenous probes are administered for the purpose of reporting on or detecting certain events in the organ or tissue of interest. However, for a variety of poorly understood reasons, non-specific probe activation can occur in the living organism. Therefore, it is important to always try and validate the signal when possible. As an example, we report on the use of the caspase 3 sensor Phiphilux G2D2. This probe is widely used in vitro to detect caspase 3 activity in apoptotic cells. Upon cleavage by caspase 3, this probe emits bright red fluorescence. When administered in vivo to uninjured wildtype mice, we noted the presence of diffuse red fluorescence at the apical border of S1 tubules (Fig. 3C). While this was unlikely to represent true apoptosis in these uninjured tubules, it raised the possibility of apoptosis-independent caspase 3 activity in the highly dynamic brush border. However, when injected into a caspase 3 KO mouse, Phiphilux G2D2 exhibited a similar strong S1 apical fluorescence (Fig. 3D). This indicated non-specific, caspase-3-independent activation of the probe . The origin of this activation remains unknown.
The kidney is a highly metabolic organ and its metabolism is supported by an abundant mitochondrial network. Tetramethylrhodamine (TMRM) is a commonly used probe to monitor mitochondrial membrane potential. We and others have successfully used TMRM and other mitochondrial probes in live animals (Fig. 4A) [5,13]. However, in the case of TMRM, a heterogeneous distribution of the probe among various renal tubules is frequently observed. In fact, differential fluorescence can be even seen within the same tubule. It is unclear if this represents true heterogeneity in mitochondrial potential (related to different metabolic states) among various tubular subsegments or rather, a consequence of heterogeneous probe delivery due to focal microcirculatory failure (Fig. 4B) [29,30]. Note also that TMRM is completely absent from interstitial cells thus hindering the study of their mitochondrial potential (Fig. 4C, D).
Renal resident macrophages play important roles in health and disease. A variety of probes can be directed to the macrophage population by harnessing their phagocytic activity. For example, liposomal clodronate is a popular means to target clodronate to the macrophage thus inducing apoptosis . Similarly, liposomes loaded with fluorescent probes can be used as a vehicle to deliver the probes to macrophages (Fig. 5A, B; Calcein-liposomes). It is important to remove leaked free probe with dialysis or centrifugation before injecting the liposomes into the animal. Fig. 5C–E show the potential confounding effect of free dye. Calcein-liposomes were injected into the animal without strict removal of free dye. Note the markedly different distribution pattern compared to that in Fig. 5A–B. Free calcein was secreted into the urinary lumen primarily by S2 proximal tubules where Multi Drug Resistance 1 transporter is expressed. Calcein is a substrate for the transporter .
Prolonged time-lapse imaging requires close monitoring of the animal under anesthesia. We use isoflurane rather than injectable anesthetics so that the anesthesia dose can be finely titrated and motion artifacts minimized. However, a drift of the imaging field and motion artifacts can still occur and make data interpretation rather problematic. In such cases, acquired imaging data can be salvaged using one of several freely available software to correct for motion artifacts. Fig. 6A, B and Supplemental Video 3 show an example in which motion artifact is reduced sequentially using Intravital Microscopy Artifact Reduction Tool software (IMART; rigid registration, non-rigid registration) . In addition, there are multiple tools available for tracking moving cells in a 2D space, enabling quantitative analysis of cell trafficking. In Fig. 6C, D and Supplemental Video 4, we show such an example using an ImageJ plug-in we developed . This plug-in defines moving objects based on size and highlights the actual path of the moving object with a time-graded color tracing. Finally, commercially available software such as IMARIS can provide advanced volumetric rendering as well as 3D time series analysis (Fig. 6E) .
Here we presented multiple examples to illustrate the power of two-photon intravital microscopy. We also demonstrated select examples to underscore issues that are commonly encountered during intravital imaging of the kidney. The majority of commercially available fluorescent probes have not been tested for intravital microscopy. Understanding renal physiology and using proper controls are crucial to establish a reproducible experimental system. We did not discuss the use of plasmids or viruses as a means to deliver probes to the kidney because of the lack of established methods to date [10,35–40]. A simple injection of plasmids (via tail vein or jugular vein) is not effective due to trapping by other organs. Commonly used viruses such as lentivirus or adeno-associated viruses do not have tropism to the kidney . In the era of genome-editing with the CRISPR system in vivo [42,43], it has become even more important to optimize and refine renal targeting methods. Hopefully, this bottleneck problem will be overcome in the near future with the emergence of exciting next-generation intracellular delivery technologies [44,45].
This work was supported by NIH grant R01-DK107623, VA Merit (1I01BX002901), NIH O’Brien Center grant P30-DK079312 to P.C.D., Paul Teschan Research grant (Dialysis Clinics Inc.), Indiana Clinical and Translational Sciences Institute grant (KL2TR001106) to T.H. IMART was obtained from the Indiana O’Brien Center for Advanced Microscopic Analysis, which is supported by funding from the National Institutes of Health NIH/NIDDK P50 DK61594.