Visualizing liver anatomy, physiology and pharmacology using multiphoton microscopy

MPM

Fluorescence is the emission of light by a substance after absorbing light or other electromagnetic radiation. Usually the absorbed light has a shorter wavelength and higher energy than the emitted light. This is known as single-photon fluorescence that is applied in traditional fluorescence microscopy. In contrast, MPM uses pulsed longer-wavelength light to excite fluorophores in the specimen 5, 10. Two photons must meet the fluorophores simultaneously (within 10^–18 s) in order to excite a fluorophore to emit a fluorescence photon with shorter wavelength and higher energy. A typical optical principle of MPM is shown in Figure 2a. MPM is considered as a revolutionary development in biological imaging because the longer-wavelength and lower-energy excitation light reduces photodamage and increases penetration depth, allowing imaging of living specimens 11. Unlike confocal microscopes, MPM does not contain pinhole apertures, allowing the detector to be placed closer to the objective to optimize collection of scattered light, and further enhancing the imaging depth.

Imaging modality of MPM

As shown in Figure 2b, MPM can excite and detect nonlinear signals including two-photon excited fluorescence (TPEF), second harmonic generation (SHG) and coherent anti-Stokes Raman scattering (CARS) 13. The concept of TPEF is based on the idea mentioned above that two photons of lower energy can excite an electron into a state with higher energy, from which it can decay and emit a shorter-wavelength light. Each photon carries approximately half the energy necessary to excite the fluorophore 5. The most commonly used fluorophores in MPM have excitation spectra in the range of 400–500 nm, whereas the laser used to excite the TPEF lies in the 700∼1000 nm range. As shown in Table 1, a number of endogenous molecules in the liver can generate fluorescence, which is a hindrance of traditional fluorescence microscopy, yet is used as an advantage in TPEF imaging 14. The metabolic coenzymes nicotinamide adenine dinucleotide hydride (NADH) is a respiratory substrate for complex I of the mitochondrial electron transport chain, which is an endogenous fluorophore in the cytoplasm of hepatocytes. As the fluorescence excitation, emission, and lifetime of NADH and NADPH overlap, these two molecules are usually measured together by MPM coupled with FLIM (MPM-FLIM) and referred to as NAD(P)H 15. TPEF images of NAD(P)H in the liver can provide basic information of liver structure. Because NAD⁺, the oxidized form of NAD(P)H, has no fluorescence, changes in the fluorescence intensity of NAD(P)H can provide valuable information regarding cell metabolism. Flavin adenine dinucleotide (FAD) is only fluorescent in the oxidative states, and can provide further information of the cellular redox state 14, 16. The most common optical method for metabolic imaging is the “redox ratio,” which is the ratio of the fluorescence intensity of FAD and NAD(P)H. The redox ratio is sensitive to changes in the cellular metabolic rate and vascular oxygen supply 14, 16. A decrease in the redox ratio or the fluorescence intensity of NAD(P)H is usually observed in cancer cells or injured cells.

In SHG, photons with the same frequency interact with a nonlinear substance and generate new photons with half the wavelength of the initial photons 17. Since SHG is an even-order nonlinear non-absorption process, fluorophore is not required in the specimen. Only highly ordered structures without inversion symmetry are capable of emitting SHG light. Collagen, a sensitive indicator of liver fibrosis, is one of such structures.

CARS is a third-order nonlinear optical process involving multiple types of photons: a pump photon, a Stokes photon and a probe photon. These three photons interact with the sample, and generate a coherent optical signal at the anti-Stokes frequency. This signal will be enhanced when the frequency difference between the pump and the Stokes photons coincides with the frequency of a Raman resonance 18. CARS imaging is intrinsic and specific to molecular vibration, such as aliphatic C-H bonds of lipids. Thus it is ideal for detecting intracellular fat vacuoles in fatty liver tissues.

FLIM

The fluorescence lifetime is the mean time a fluorophore remains in the excited state before emitting a photon (fluorescence) and returning to the initial ground energy state. FLIM has been used as an important imaging technique in MPM, which constructs a spatial distribution map of fluorescence lifetimes from a fluorescent specimen 19. The lifetime of the fluorophore signal, rather than its intensity, is used to create the image. Figure 1d shows an example of FLIM image of normal mouse liver. This technique has the advantage of differentiating various endogenous and exogenous fluorophores in the liver tissue based on their unique lifetimes. Table 1 shows the lifetimes of major endogenous fluorophores in the normal liver. Lifetime changes of exogenous fluorophores usually reflect drug-interactions or protein-interactions, while those changes of endogenous fluorophores reflect changes in liver microenvironment, which cannot be revealed by microscopy using fluorescence intensity.

A common application of FLIM in the liver is to measure the fluorescence lifetime of NAD(P)H. NAD(P)H can exist as free or protein-bound molecules, which have similar excitation and emission wavelengths, but can be distinguished by their distinct fluorescence lifetimes. Protein-bound NAD(P)H mainly locates in mitochondrial membrane, and produces adenosine triphosphate (ATP) in aerobic conditions, while free NAD(P)H is located in the cytoplasm, involving in ATP synthesis without oxygen in glycolysis. The relative contribution of protein-bound NAD(P)H decrease in hypoxia, mainly because cellular respiration is shifted to glycolysis, producing ATP in anaerobic conditions 20, 21. The changes of lifetimes of free and protein-bound NAD(P)H also indicate altered metabolic state, as is typically observed in cell differentiation, apoptosis and necrosis 20.

Normal liver

The capsule of the liver, known as Glisson's capsule, is a fibrous connective membrane mainly composed of collagen and elastin fibers. It encompasses the hepatic parenchyma, and contributes significantly to the global mechanical behavior of the liver. MPM has been used to characterize the shape, geometry and microstructure arrangement of the capsule using TPEF and SHG 25, 26. The obtained images revealed that the collagen/elastin ratio increased from the inner to the outer surface of the liver capsule. The microstructure and strain fields of the capsule present heterogeneity among different layers. Reorientation occurs in fiber layers from the inner to the outer surface of the liver capsule.

The cellular structure of liver can be imaged as deep as 250 μm below the liver capsule by MPM. As shown in Figures 1, 3 and 4, hepatocyte cords with bright autofluorescence of NAD(P)H are separated by the blood-filled dark sinusoids in TPEF imaging. The stellate cells are smaller in size but have strong vitamin A-associated autofluorescence and a longer lifetime than that of NAD(P)H 21. Fenestrations are transcellular pores in endothelial cells that facilitate transfer of substrates between blood and the extravascular compartment. Liver sinusoidal endothelial cells could be stained by 6-lauroyl-2-dimethylaminonaphthalene (LAURDAN) to visualize the number and size of fenestrations using MPM 27. Recently, more specific fluorescence-conjugated antibodies (such as anti-PECAM1/CD31 or anti-VCAM-1/CD106) have been developed to stain viable liver sinusoidal endothelium 9, 28.

Liver fibrosis

In liver fibrosis, injury of hepatocytes can be observed as reduced fluorescence of NAD(P)H in TPEF imaging (Figures 3 and 4) 21. Moreover, FLIM shows changes in cellular metabolic pathways as indicated by the decrease of lifetimes of free and protein-bound NAD(P)H, and contribution of protein-bound NAD(P)H (Figure 4). The livers of mice receiving CCl₄ for 5 weeks showed signs of postsinusoidal venules. HSCs can be detected using real-time histology by their distinct vitamin A-associated autofluorescence. Inhomogeneous distribution of stellate cells was found in a mouse model of chronic chemical hepatic injury and advanced liver fibrosis 21. The stellate cells accumulated in centrilobular region around central veins, and reduced in midzonal regions. The regional redistribution of stellate cells formed bridges between postsinusoidal venules in some liver lobules. The morphology and distribution of stellate cells imaged by MPM provide valuable information about cellular microenvironment during development of liver fibrosis, which cannot be obtained from conventional histology.

SHG imaging has been performed on liver biopsies taken from patients with liver fibrosis or cirrhosis 21, 29, rodents with CCl₄ induced chronic injury and fibrosis 21, and cholangitis and biliary fibrosis (Mdr2−/− and bile duct ligation) 21, 29. MPM can clearly reveal the increase of the collagen amount during fibrosis progression and offer the possibility of an accurate characterization of fibrosis without specific staining. As shown in Figure 3, collagen deposition was obvious in the centrilobular region and pseudolobular formation was evident in in CCl₄ induced fibrotic liver. While the livers with biliary fibrosis of 20-week-old Mdr2−/− mice showed intercellular fibrosis, especially around the medium sized to large bile ducts of the periportal region. Two quantification systems, Fibro-C-Index [29] and Fibrosis-SHG index (Figure 5) 12 have been developed for the assessment of liver fibrosis using MPM. Both systems have been compared and validated by pathological examination, indicating the high reliability and sensitivity for potential application of MPM in clinical diagnosis of liver fibrosis. Since collagen SHG response is effectively instantaneous 30, the mean lifetimes decreased significantly in fibrotic liver at excitation of 800 nm (containing both NAD(P)H and collagen signals), allowing the discrimination from normal liver (Figure 3). MPM-FLIM would be a powerful tool providing realtime histology which combines morphology and quantitative evaluation of liver fibrosis simultaneously.

Liver cancer

The most common type of primary liver cancer is hepatocellular carcinoma (HCC). The MPM optical diagnostic features for HCC have been established and validated in different studies 21, 31. High-resolution MPM images have clearly demonstrated extensive cell heterogeneity characterized by irregular size and shape, increased nuclear to cytoplasmic ratio, decreased autofluorescence of NAD(P)H, central necrosis and intercellular collagen in HCC (Figure 4). Yan et al. performed a pilot preclinical study investigating 224 surgical specimens including benign and malignant liver lesions such as hemangioma, focal nodular hyperplasia, HCC, cholangiocarcinoma, and colorectal cancer liver metastases. They found that MPM is able to diagnose liver cancer and differentiate benign and malignant liver lesions with high sensitivity, specificity and accuracy 31.

In FLIM images, the cellular lifetimes of free and protein-bound NAD(P)H, and contribution of protein-bound NAD(P)H significantly decreased in HCC, which is consistent with hypoxia and the increased levels of glycolysis in neoplastic cells (Figure 4). In poorly-differentiated HCC sections, an increased cellular lifetime of red fluorescent pigments including porphyrins, biliverdins and bilirubins, and scattered short-lifetime spots can also be detected non-uniformly, which may represent the inflammation-induced infiltrated leukocytes 32. Thus, MPM-FLIM has the potential to become a powerful clinical tool in the future to perform real-time diagnosis of liver cancer and differentiation of benign and malignant liver lesions.

Liver steatosis

In liver with steatosis, black intracellular fat vacuole can be detected as spots with significantly reduced fluorescence of NAD(P)H (darker than the nuclei) in TPEF images (Figure 4) 21. The cytoplasm of hepatocytes has a sponge-like appearance because of numerous small black intracellular fat inclusions in mice after feeding a high fat diet for 14 days (Figure 4). Compared to TPEF imaging, CARS imaging can specifically visualize fat droplets in fresh liver samples and extract the fat content through image analysis 33. Similarly to the way using SHG to quantify the collagen amount in the fibrotic liver, CARS imaging has been used to quantitatively assess the fat in rat livers with steatosis 33. The content of hepatic fat measured by MPM has been correlated well with that determined by biochemical analysis. This staining free technique has the potential for early diagnosis and rapid detection of liver steatosis.

Mitochondrial physiology

The mitochondrion is a double membrane-bound organelle found in most eukaryotic cells. The mitochondrial permeability transition (MPT) is defined as an increase in the permeability of the mitochondrial membranes to molecules less than 1500 Dalton 34. MPT results from the opening of MPT pores located in the inner membrane of the mitochondria under certain pathological conditions such as hepatic I/R injury, leading to mitochondrial depolarization and swelling, and cell death through apoptosis or necrosis. MPM has been applied to visualize the mitochondrial physiology using different fluorescent probes. Propidium iodide (PI) labels the nuclei of nonviable cells, and can be used to visualize the overall cell viability 35. The opening of MPT pores could be assessed using calcein acetoxymethyl ester in TPEF imaging. Calcein is a fluorescent molecule that cannot be taken up by mitochondria after intravenous injection, resulting in the mitochondria appearing as dark voids in the hepatocytes in the normal liver 36. In mice with hepatic I/R injury or bile duct ligation, the opening of MPT pores in hepatocytes can be revealed by the appearance of calcein fluorescence in the mitochondria matrix space, which are dark voids in normal hepatocytes 37, 38. Rhodamine 123 (Rh123) is another commonly used fluorescent dye in TPEF imaging to visualize polarized mitochondria 35. After intravenous injection, Rh123 selectively accumulates in polarized mitochondria of the healthy liver, resulting in bright punctate fluorescence of Rh123 in hepatocytes. While the liver with hepatic I/R injury 38, 39 or obstructive cholestasis 37 shows dim diffuse of Rh123 fluorescence, denoting that many mitochondria did not take up Rh123 because of depolarization. Thus, MPM can be employed to investigate the drug efficacy. As shown in Figure 6, the effect of NIM811, an MPT inhibitor, was evaluated using fluorescent probes of PI and Rh123 to visualize the attenuation of mitochondrial depolarization and MPT onset in APAP induced liver injury 40.

Functions of hepatic enzymes and transporters

Hepatic enzymes and transporters play a significant role in drug metabolism and biliary excretion. In most studies, functions of hepatic transporters are evaluated in cultured cell system or isolated perfused liver 41. However, both methods have obvious disadvantages such as limited biological significance in cell lines and indirect analysis of transport processes in perfused liver models 42. Using specific fluorescent probes, MPM allows direct measurement of enzyme and transporter functions. Recently, a two-photon excited ratiometric fluorescent probe has been developed for sensitive and selective detection of CYP1A, one of the most important phase I drug-metabolizing enzymes 34. TPEF imaging can directly assess the functions of hepatic transporters by observing the in vivo distribution kinetics of specific fluorescent substrates in the liver 42-44. For example, the functions of biliary canalicular transporter MDR1 (P-glycoprotein, P-gp) and basolateral transporter organic anion transporting polypeptide (Oatp) have been evaluated using their specific substrates Rh123 45 and sodium fluorescein 46. In hepatic I/R injury, the uptake and clearance of fluorescein were found to be delayed, probably due to the impaired hepatic microcirculation and the dysfunction of efflux transporters 47. And the impaired P-gp function can be revealed by the increased fluorescence intensity of Rh123 in the hepatocytes with I/R injury compared to sham 7, 44.

Hepatobiliary excretory function

Bile excretion is part of hepatobiliary function. The liver is responsible for the uptake, processing, and excretion of many exogenous or endogenous substances into bile by hepatocytes 48. Carboxyfluorescein diacetate (CFDA) has been used to define the hepatobiliary excretory function of healthy mice, mice with APAP-induced liver injury and with obstructive cholestasis 48-50. CFDA is a nonfluorogenic substance and can be hydrolyzed by esterase into fluorogenic carboxyfluorescein (CF) after taken up by hepatocytes. The kinetics of CF can be determined in hepatocytes and sinusoids separately based on its fluorescence intensity. The delayed clearance of CF from hepatocytes into bile canaliculi was observed in mice with bile duct ligation. An active machinery operating backflow of bile containing CF from hepatocytes into sinusoids was found in mice with obstructive cholestasis and APAP-induced liver injury 48, 50, 51.

In addition, fluorescein has also been used to reflect the hepatobiliary excretory function. The uptake and clearance of fluorescein has also been found to be delayed in mice with liver steatosis and hepatic I/R injury using TPEF imaging, and confirmed by FLIM 47, 52. Conventional methods for studying the hepatobiliary excretory function usually involve biochemical analyses of contents in the bile, liver, blood or urine, while MPM enables intravital imaging of fluorescent probes in the liver at the subcellular level. This imaging technique can provide further quantitative measures of metabolic and hepatobiliary excretory function, and hence extend the qualitative liver function tests now available.

Cell physiology and migration

MPM has been applied to study the cell migration and function in the liver in a variety of disease states including HCC, colorectal cancer liver metastases, hepatic I/R injury, parasitic infection and liver regeneration. Kupffer cells can be labeled in vivo by intravenous injection of red fluorescent carboxylated latex particles 53, 70-kDa Texas red-dextran 54 and PD nanobeads (545 marked) 55 due to their macrophage phagocytic ability. After labelling, TPEF imaging of the liver reveals that Kupffer cells are the main cellular scavenger for circulating particle-associated antigens in homeostasis 53. These cells are the only detectable population of mononuclear phagocytes within granulomas induced by Leishmania donovani infection 55. Kupffer cells do not migrate to interact with vessels while infiltrating monocytes interact directly with sinusoids after partial hepatectomy 54.

Leukocytes are the cells of the immune system that are involved in protecting the body against both disease and foreign invaders. These cells can be further divided into five main types: neutrophils, eosinophils, basophils, lymphocytes, and monocytes 56. Green fluorescent protein (GFP)-expressing leukocytes were used for MPM imaging in almost all the studies, because most leukocytes lack the phagocytic ability and are difficult to be labelled using fluorescent dyes 53, 57-59. The GFP-expressing tumor-infiltrating leukocyte adhesion and migration in HCC have been visualized in vivo at the single-cell level 57. Their migration was found in a random manner with frequently changed infiltrating directions. In hepatic I/R injury, MPM successfully detected the numbers, velocity, and morphology of recruited neutrophils in vivo, and has potential for a wide range of applications to investigate the mechanism of I/R injury 58.

Recently, MPM has been used to detect the colorectal cancer liver metastases in living mice taking its advantages of imaging target organs for a long period at a high magnification and in the deep depths from the surface 8, 60. An animal model of colorectal cancer liver metastases was developed by inoculating red fluorescent protein (RFP) expressing colorectal cancer cells into the spleens of GFP transgenic mice. As shown in Figure 7, the multi-stage tumor metastatic processes have been visualized in this mouse model, including tumor cell arrest, tumor cell-platelet interaction, tumor cell-leukocyte interaction, and metastatic colonization in the liver 8, 61. These studies are good examples of the applications of MPM as a tool for basic investigation.

Molecule imaging

The pharmacokinetics of fluorescein and RH123 has been investigated by MPM in rat liver 43, 44, 62. Since MPM allows simultaneous visualization of the organ autofluorescence and molecule fluorescence, it is able to determine the levels of fluorescent molecules in the sinusoids, hepatocytes and the bile respectively (Figure 8a), which enables dividing the liver into subcompartments for pharmacokinetic modelling, as seen in Figure 8b. A physiologically based pharmacokinetic (PBPK) model has been developed to characterize the kinetics of fluorescein at the single-cell resolution in health and diseased liver in vivo 52. Using the same intravital imaging technique, half-life of Rh123 is calculated by fitting the decay of fluorescence intensity in hepatocytes vs. time profile into the exponential equation 44.

Furthermore, FLIM imaging adds the capability of differentiating fluorescent molecules from biological tissues based on their fluorescent lifetimes. For example, fluorescein and its metabolite fluorescein mono-glucuronide (FG) have an overlapping excitation and emission spectra 62, which make them to be hardly distinguished using TPEF imaging. However, the fluorescence lifetime of fluorescein and metabolite FG is reported to be 3.8 ns to 4.1 ns and 2.3 ns, respectively 62, which can be differentiated by FLIM. The in vivo distribution and metabolism of fluorescein has been studied based on the lifetime change in different zonation of rat liver using MPM-FLIM, showing the mean lifetimes of fluorescein and FG decease over time after injection 62. Therefore, the combination of MPM and FLIM provides a novel technique for studying real-time distribution and metabolism of fluorescent molecules in the liver at a high resolution.

MPM has also been used to real-time evaluate the chemotherapy response on the colorectal cancer liver metastases 64. After administration of 5-fluorouracil or irinotecan, tumor cell fragmentation, condensation, swelling and intracellular vacuoles were observed under time-series intravital TPEF imaging. A pharmacokinetic/pharmacodynamic model could be developed based on these observations.

Nanoparticle imaging

Nanoparticles (NPs) are defined as spherical (or quasispherical) particles with a diameter less than 100 nm, but often refer to the range up to 300–500 nm 65. Since NPs have unique physiochemical properties, and have been intensively applied to drug and gene delivery, imaging and diagnosis 66, it is important to investigate their pharmacokinetics for clinical applications. Although many studies have reported the organ-level distribution of NPs, very few have addressed their disposition in organs at the cellular level 63. The uptake, distribution and excretion of fluorescent or fluorescent labelled NPs can be investigated using TPEF imaging at the single cell resolution 67-70. Polymeric NPs with high bond repetition rate are highly suitable for CARS imaging because the CARS signal scales quadratically with the bond concentration 70.

The in vivo spatiotemporal disposition of quantum dots (QDs), a typical example of long-circulating NPs, has been intensively explored in the liver at a single cell level using MPM 63, 71. The concentration profile of QDs determined from the QD fluorescence intensity in sinusoids correlates well with that in the plasma measured by inductively coupled plasma mass spectrometry (ICP-MS) 63. This suggests that MPM may be used as a semi-quantitative method to investigate the distribution of fluorescent NPs in living organisms, particularly in the initial distribution phase (0.5 min to 5 min), when collecting blood samples is difficult. Furthermore, the enriched quantitative data obtained from MPM imaging, especially at early time, allows more sophisticated pharmacokinetic modelling to reveal more detailed information of NP disposition, while blood sampling and ICP-MS analysis can only provide limited data for modelling 63. As shown in Figure 9, the QD sub-organ distribution visualized by TPEF imaging was further confirmed by FLIM imaging based on the change of average fluorescence lifetime. Since the fluorescence lifetime of QDs is much longer than that of liver cells, the increased fluorescence lifetime in the sinusoid but not in hepatocytes after QDs injection confirmed that these negatively charged QDs predominantly distributed in the sinusoids 63. Thus, even if the excitation and emission spectra of a fluorophore overlap with those of the background, it can still be distinguished based on different fluorescence lifetimes.

Therapeutic cell imaging

Cell therapy has emerged as an evolutionary therapeutic force especially for diseases not curable by traditional therapeutics. Mesenchymal stem cell (MSC) is one of the most promising and widely used therapeutic cells for many debilitating diseases including liver cirrhosis, diabetes, spinal cord injury and myocardial infarction. Recently, a whole body PBPK model has been developed based on the organ disposition and cell-tissue interactions of GFP-expressing MSCs visualized by MPM 72. This imaging technique opens up a new window for more in-depth investigation about the disposition of molecules, nanoparticles and biological agents in the liver and other organs to understand their in vivo pharmacokinetics and pharmacodynamics.