2.1.

Animal Model and Preparation of Specimens

All the experiments followed procedures approved by the Institutional Animal Ethics Committee of Huazhong University of Science and Technology.

Male Sprague-Dawley rats weighing 120 to 150 g were purchased from the Animal Biosafety Level III Laboratory (Wuhan, China) and housed under specific pathogen-free conditions. The experimental model of chronic pancreatitis was induced by a single intravenous administration of dibutyltin dichloride (DBTC), as described elsewhere.³¹ Briefly, DBTC (Sigma-Aldrich, China) was first dissolved in 100% ethanol (two parts) and mixed with glycerol (three parts), then the DBTC solution was injected into the tail vein at a dose of $8\mathrm{mg}/\mathrm{kg}$ body weight.

At 1, 7, and 28 days after the application of the DBTC, animals were sacrificed with an overdose of isoflurane. The fresh pancreatic specimens were excised and immediately imaged in phosphate buffered saline (PBS) within 1 h after dissection. A part of each pancreas was used for histological analysis [haematoxylin-eosin (H&E) staining, Masson’s trichrome staining, and immunostaining for desmin and $a$-SMA]. The images of the three groups of inflammatory tissues were compared with those of the normal group. Ten rats were used for the normal group and eight rats per group were used for all the pancreatitis groups.

2.2.

Nonlinear Optical Imaging and Spectral Measurements

Nonlinear optical imaging was performed using a modified system based on a commercial microscope (Fluoview 1000, Olympus, Japan), as described previously.²⁹ It is reported that the two-photon absorption of the major intrinsic fluorophores such as NADH and FAD decreases as the excitation wavelength increases in the range of 700 to 900 nm.²¹ We tuned the excitation wavelength in the 730 to 890 nm range and found that 750 nm was the optimal wavelength to give the best signal-to-noise ratio in our microscopic system. A tunable Ti:Sapphire laser (Mai Tai, Spectra-Physics) operating at 750 nm was used for excitation, except where otherwise noted. A water-immersion objective ($60\times /1.2\mathrm{NA}$ water-immersion, Olympus) was employed for focusing the excitation beam into tissue samples and collecting the backscattered intrinsic nonlinear optical signals. The average laser power on the surface of the sample was about 15 mW. The scanning rate was $10\mu \mathrm{s}/\mathrm{pixel}$. The emission signals are detected by three channels in the nondescanned mode. A narrow bandpass filter (FF01-380/14-25, Semrock) was used for the detection of SHG signals ($380\pm 7\mathrm{nm}$; blue). Another two emission filters (FF01-445/45-25 and $\mathrm{FF}01-530/43-25$; Semrock) were used to extract the intrinsic TPEF signals for the differentiation of acini, PSCs, and other nonacinar cells. The TPEF signals were detected with the same system parameters, including the voltage, gain, and offset of the photomultiplier tubes. The nonlinear optical signals were presented by the pseudocolor images. Blue, green, and red color-coded images correspond to SHG signals, TPEF signals between 423 and 467 nm, and TPEF signals between 509 and 551 nm, respectively.

For the spectral measurements, the emitted signals are collected by a fiber bundle with a lens ($\mathrm{focal}\mathrm{length}=50\mathrm{mm}$) after the dichroic mirror ($\mathrm{FF}665\text{-}\mathrm{Di}02\text{-}25\times 36$, Semrock) and sent to a spectrograph (Acton SP2356, Princeton Instruments) with a grating of 300 grooves blazed at 500 nm, followed by a thermoelectrically cooled charge-coupled device camera (PIXIS:256E, Princeton Instruments).

2.3.

Quantitative Analysis of PSC Size and Collagen

To calculate the size of PSCs, the nonlinear optical image among the axial consecutive stack ($z$ step of 1 μm) that shows the largest PSC size in appearance was chosen, and at least five images of different areas were used for each sample. The superimposed images of two TPEF channels were first processed with Adobe Photoshop CS4 (Adobe Systems Inc.) to improve contrast and dissect out the PSCs visually. The size of PSCs was evaluated by the number of pixels that each PSC accounted for and converted to area unit (${\mu \mathrm{m}}^2$) according to the size of the image. The average sizes were calculated for the normal group and pancreatitis groups harvested at different stages.

The content of collagen fibers for each specimen was evaluated by the proportion of the number of pixels that collagen fibers account for to the total number of pixels of the image. Because of the fibrous connective tissue capsule distributed in the surface of pancreas,²⁹ the SHG images of the parenchyma at an imaging depth showing the best visualization of acinar morphology were picked out for the analysis of collagen density for all samples. The SHG images were preprocessed using the MATLAB (The MathWorks) function “imadjust” for gray level adjustment to increase the contrast of the image, so that the intensities of the SHG signal acquired for different samples are comparable. The background noise is subtracted by counting the pixels with a value above the threshold of 20% of the maximal value for each image as described elsewhere.³² At least five imaging areas were calculated to get the average content value for each tissue.

Statistical significance was calculated with analysis of variance (ANOVA) linear contrast using SPSS software (SPSS). All $p$-values of 0.05 or less were considered to be statistically significant.

2.4.

Histological Analysis

The pancreatic tissues were fixed with 10% neutral buffered formalin, and paraffin-embedded sections with a thickness of 5 μm were routinely prepared. The pancreatic tissue sections were subsequently de-waxed and rehydrated before H&E staining and Masson’s trichrome staining, which were used for the detection of collagen fibers.

The phenotype of PSCs in the time course of chronic pancreatitis induced by DBTC was characterized using immunohistochemical detection of desmin and $\alpha$-SMA using published techniques.¹³ Antigen retrieval was achieved by microwave heating of the tissue sections for 15 to 20 min. Endogenous peroxidase activity was blocked with hydrogen peroxide (10 min) and rinsed in PBS three times. Sections were then incubated overnight at 4°C with primary antibody [monoclonal mouse anti-desmin antibody $1/100$ (ab8470) or monoclonal mouse anti-$\alpha$-SMA antibody $1/200$ (ab18147); Abcam, Cambridge, Massachusetts). After three washes in PBS for 3 min each, the ultrasensitive streptavidin-peroxidase kit (KIT-9710; Maixin Inc., Fuzhou, China) and diaminobenzidine (DAB) kit (DAB-0031; Maixin Inc., Fuzhou, China) were used for color development according to the manufacturer’s instructions. Counterstaining was performed with Mayer’s hematoxylin.

Histological images were acquired using an optical microscope (IX81, Olympus, Japan) equipped with a $20\times$ air objective and a color digital industrial camera (DFK 41BU02, The Imaging Source Europe GmbH, Germany).

3.1.

Identification of PSCs in a Normal Pancreas

It can be observed from the nonlinear optical images that a small amount of collagen fibers [arrow in Fig. 1(a)] distributed around the acini composed of acinar cells [asterisks in Fig. 1(b)], which has been reported previously.²⁹ The fluorescence intensity of acinar cells detected in the red channel [Fig. 1(c)] is much lower than that in the green channel [Fig. 1(b)], indicating that the intrinsic fluorescence primarily comes from NADH with a fluorescence peak at 460 nm.²¹ The emission spectrum of acinar cells excited at 750 nm [Fig. 1(e)] extends over a wide range of wavelengths similar with NADH and FAD, which are the major cellular sources of intrinsic fluorescence. The spectrum with excitation wavelength of 820 nm, where the absorption of NADH is negligible, displays a prominent peak at about 530 nm [Fig. 1(e)], which is the characteristic of FAD.²¹ These results demonstrate that the major intrinsic fluorophore detected in the range of $445\pm 22\mathrm{nm}$ and $530\pm 21\mathrm{nm}$ in acinar cells is NADH, and FAD has a minor contribution to the autofluorescence when excited at 750 nm. Besides, the emission from collagen fibers peaks at 375 nm, verifying that the collagen fibers can be identified by SHG signals detected in the blue channel [$380\pm 7\mathrm{nm}$, arrow in Fig. 1(a)].

Fig. 1

Identification of pancreatic stellate cells (PSCs) in normal pancreatic tissues. The intrinsic nonlinear optical images including (a) second harmonic generation (SHG) image, (b) two-photon excited fluorescence (TPEF) image detected in the range of 423 to 467 nm, (c) TPEF image detected in the range of 509 to 551 nm, and (d) the superimposed image of the above three channels show collagen fibers and PSCs distributed around the exocrine acini. The contrast of the SHG image (a) is enhanced for better visualization using Fluoview FV10-ASW software (Version 1.4a; Olympus, Japan). The solid arrows, asterisks, solid arrowheads, and dotted lines indicate the collagen fibers, acinar cells, the PSCs, and the cellular processes of the PSCs, respectively. (e) The emission spectrum of different pancreatic components excited at 750 or 820 nm can be related with the major endogenous biochemical constituents such as collagen fibers, nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), and retinol. The rapid decrease of the spectra in the range of 665 nm is due to the cut-off of the dichroic mirror before the spectrograph. Scale bar is 60 μm.

Specially, a few cells in the periacinar space with a star-like appearance and cellular processes [dotted lines in Fig. 1(c)] show higher intensity of TPEF signals in the red channel [arrowheads in Fig. 1(c)], leading to an orange color in the superimposed image of three channels [Fig. 1(d)]. The corresponding spectroscopic data exhibit a major peak around 500 nm [Fig. 1(e)], which is close to that of vitamin A.²¹ These cells are identified as PSCs according to their long cytoplasmic processes encircling the base of pancreatic acini and the presence of retinol-containing fat droplets.

Correspondingly, the anatomical structures of the pancreatic acini and collagen fibers are correspondent with H&E and Masson’s trichrome staining [Fig. 2(a) and 2(b)], respectively. The immunostaining images for the expression of the intermediate filament desmin [Fig. 2(c)] show the same characteristics of PSCs with the nonlinear optical images, including the cytoplasmic processes and periacinar distribution. The activation marker $\alpha$-SMA that only expressed around blood vessels and pancreatic ducts [Fig. 2(d)] further validated the quiescent phenotype of PSCs with vitamin A storage in normal pancreas.

Fig. 2

Histological images of normal pancreas. (a) Haematoxylin-eosin (H&E) staining and (b) Masson’s trichrome staining show the same pattern of exocrine pancreas with the nonlinear optical images (Fig. 1). (c) Desmin-positive and (d) $\alpha$-SMA-negative assessed by immunostaining show that the PSCs in normal pancreas exhibit a quiescent phenotype. The morphology of PSCs revealed by immunostaining is identical to that in the nonlinear optical images (Fig. 1). Scale bar is 60 μm.

The above results reveal that intrinsic emission contributions from different biochemical constitutions can be distinguished by the green ($445\pm 22\mathrm{nm}$) and red ($530\pm 21\mathrm{nm}$) channels, and combined with morphological characteristics to recognize PSCs from acinar cells in pancreatic tissues, indicating the potential to depict the other pancreatic components by nonlinear optical microscopy.

3.2.

Visualization of Multiple Pancreatic Components During Chronic Pancreatitis

At day 1 after the induction of chronic pancreatitis, there are no obvious alterations of the acinar structure in the nonlinear optical image [Fig. 3(a)], which is consistent with the H&E-stained image [Fig. 3(a)]. Inflammatory cells without any processes of the cytoplasm appear in the interstitium matrix at day 7 [open arrows in Fig. 3(b)]. The infiltrating inflammatory cells manifest an orange color, with high fluorescence intensity at longer wavelength ($530\pm 21\mathrm{nm}$).¹⁶ Most of these cells show parallels with the characteristics of lymphocytes including a small round or oval shape and a thin rim of cytoplasm surrounding the dark round nucleus,²² as can be found in the H&E-stained images [Fig. 3(b)]. At day 28 after the induction of chronic pancreatitis, a decrease in the number of acinar cells and the presence of tubular complexes consisting of duct-like structures with a large lumen lined by a monolayer of low cuboidal or flattened cells [asterisks in Fig. 3(c)] can be seen, which is verified by H&E staining [Fig. 3(c)]. In addition, macrophage [open arrowhead in Fig. 3(c)] with cytoplasmic inclusions²² can also be observed except for lymphocytes [open arrows in Fig. 3(c)], and they possibly play a role in the removal of necrotic cellular debris in the pancreas.

Fig. 3

Simultaneous visualization of PSCs, inflammatory infiltrates, and collagen fibers during chronic pancreatitis. (a) In the nonlinear optical images (the first four columns), the morphology of the PSCs (solid arrowheads) in the pancreatic tissues harvested one day after induction of chronic pancreatitis changed from star-like to triangular shape. (b) At day 7, the PSCs are visually absent due to the decrease of retinol fluorescence in the red channel. Lymphocytes without cellular processes (open arrows) appeared in the pancreatic tissues. Slight fibrosis can be observed. (c) A few macrophages with larger size and visible cellular inclusions (open arrowhead) can be seen in the pancreatic tissues at day 28 except for the lymphocytes (open arrows). Besides, the tubular complexes (asterisks) and intense deposition of collagen fibers indicate the progress of chronic pancreatitis. The corresponding H&E images (right column) are consistent with the above results obtained by nonlinear optical microscopy. Scale bar is 60 μm.

The collagen fibers in pancreatic tissues are visualized simultaneously and the density of collagen fibers during chronic pancreatitis is analyzed. As can be seen from the SHG signals, the inflamed pancreatic tissues harvested at day 7 after the induction of chronic pancreatitis show slight increase of collagen fibers around individual acini [Fig. 3(b)], while those harvested at day 28 after the induction of chronic pancreatitis develop intensive periacinar fibrosis [Fig. 3(c)]. Quantitative analysis shows that the density of collagen fibers on day 28 after the induction of chronic pancreatitis is significantly larger than the day 1 group and the normal group [Fig. 4(a); $p<0.001$, ANOVA linear contrast; $n=8$ for each group], which is further validated by Masson’s trichrome staining [Fig. 4(b)–4(d)]. The gradual accumulation of stromal collagen fibers after the phenotype conversions of PSCs implies the contribution of activated PSCs to fibrosis in chronic pancreatitis.

Fig. 4

Content of collagen fibers during chronic pancreatitis. (a) Quantitative analysis of the SHG images show that the density of collagen fibers in the tissues harvested at 28 days after the induction of pancreatitis is significantly larger than those of the other groups. $*$, $p<0.001$, ANOVA linear contrast. The sample size is 8 for each group (eight rats); $+$, outliers on the box-and-whisker diagram; bars, total extent of the data. The Masson’s trichrome images of pancreatic tissues at (b) day 1, (c) day 7, and (d) day 28 are in agreement with the quantitative results. Scale bar is 60 μm.

Dark duct-like structure with a diameter of 6 to 9 μm is delineated at day 28 (dotted lines in Fig. 5). This structure represents blood vessels since the intrinsic fluorescence can be quenched by the heme moieties of hemoglobin in erythrocytes that exist in blood plasma. The edge of the duct is lined by thin fluorescent dots, which are resulted from a layer of flattened epithelium cells (open arrows in Fig. 5), and the spindle-shaped cell that locates adjacent to the duct (solid arrow in Fig. 5) is identical to the pericyte of a capillary vessel around base membrane of epithelium.³³ Besides, the overall pattern is the same as the vessels revealed by fluorescent agents and transgenic reporters.^34,35 Microvessels are also observed in the other groups of pancreatic tissues (Fig. 6), and they are relatively rare in the pancreatic tissues selected. It suggests angiogenesis in the inflammatory pancreatic tissues as pancreatitis progress, which has been reported by previous studies³⁶ and can be confirmed by high vascular density in the H&amp;E-stained image of pancreatic tissues at day 28 [Fig. 3(c)].

Fig. 5

Magnified images of the blood vessel acquired at different imaging depths corresponding to the area indicated by the white box in Fig. 3(c). In the pancreatic tissues at day 28 after the induction of chronic pancreatitis, blood vessels (dotted lines) are present in the inflamed area with pericytes (solid arrow) around the endothelial cells (open arrows). The imaging depths within the tissues are indicated by the $z$ value in the upper right corner of each image. Scale bar is 30 μm.

Fig. 6

Microvessel in a normal pancreas. (a) and (b) The morphology of microvessel is delineated by the dotted lines in the images acquired at various depths within the tissues as indicated by the $z$ value in the upper right corner of each image. (c) and (d) The endothelium and the thin lumen of the microvessel can be clearly seen in the magnified images corresponding to the areas indicated by the white box in (a) and (b). Scale bar is 60 μm.

Based on intrinsic emission and morphology, other pancreatic components including the lymphocytes, macrophages, collagen fibers, and blood vessels can also be visualized simultaneously with the acinar cells using nonlinear optical microscopy. In particular, their corresponding alterations including inflammatory cell invasion, destruction of acinar cells, tubular complex formation, and fibrosis can be characterized simultaneously and related with different stages of chronic pancreatitis.

3.3.

Phenotype Conversions of PSCs Within 3-D Tissue Structure

The changes of PSCs during chronic pancreatitis are investigated by the pancreatic tissues harvested at different stages. At day 1 after the induction of chronic pancreatitis, the PSCs exhibit shorter cellular processes and take on a triangle-shaped morphology [solid arrowheads in Fig. 3(a)]. This is in agreement with previously reported morphological transformation of PSCs on day 3 in a culture model.³⁷ In addition, the PSCs increase in size [Fig. 3(a)] compared with those in normal pancreas [Fig. 1(d)]. It is difficult to quantitatively assess the size of the fibroblast-like PSCs using conventional histological methods, due to the disconnections between tissue sections generally with a thickness of 4 to 7 μm. In comparison, the inherent optical sectioning ability and high spatial resolution of nonlinear optical microscopy allows accurate quantitative analysis of cell size within 3-D tissue structure. Our result shows significant enlargement of the PSCs compared with the normal group (Fig. 7; $p<0.001$, two-tailed $t$ test on the size of the PSCs; $n=8$ for each group), and the area of PSCs increases by about 1.75 times from normal group to the day 1 group.

Fig. 7

Quantitative analysis of the size of the PSCs. The PSCs in the tissues harvested at day 1 after the induction of pancreatitis are significantly larger than those in the normal pancreas. $*$, $p<0.001$, two-tailed $t$ test on the size of the PSCs. The sample size is 8 for each group (eight rats). Bars, total extent of the data.

After 7 days of chronic pancreatitis, the PSCs can hardly be visualized in the inflamed area [Fig. 3(b) and 3(c)], largely due to the loss of retinol-containing fat droplets, which is related to the activation of PSCs.¹ We further verified the activation of PSCs by coexpression of desmin and $\alpha$-SMA. In the corresponding immunostained images, the PSCs are seen to be quiescent on day 1 after the induction of pancreatitis [Fig. 8(a) and 8(b)], indicating the enlargement of endoplasmic reticulum network as an early event before they become activated.¹ As chronic pancreatitis progresses, the PSCs gradually convert to an activated state [Fig. 8(c)–8(f)]. Besides, the number of $\alpha$-SMA-positive cells increased significantly from day 1 to 28 [Fig. 8(d) and 8(f)], which is possibly from a bone-marrow origin.³⁸

Fig. 8

Immunostaining for the assessment of PSC phenotype. The PSCs in pancreatic tissues harvested at 1 day [(a) and (b)], 7 days [(c) and (d)], and 28 days [(e) and (f)] after the induction of pancreatitis were immunostained by desmin [(a), (c), and (e)] and $\alpha$-SMA [(b), (d), and (f)] expressions, which shows that the PSCs are quiescent at day 1 and they gradually transformed to an activated phenotype as chronic pancreatitis progressed. Scale bar is 60 μm.

It can be seen from the above results that the fluorescence from vitamin A enables morphological characterization of PSCs within 3-D tissue structure before they become activated. After the loss of vitamin A in the cell, the TPEF intensity of PSCs in the red channel decreased. Because the cells are densely arranged in the pancreatic tissues, the PSCs are hard to be identified from the acinar cells according to the cell morphology revealed by NADH and FAD fluorescence. Nonetheless, loss of vitamin A fluorescence can be used as an indicator for the phenotype conversions of PSCs during chronic pancreatitis.