Abstract
Laser scanning cytometry (LSC) is a powerful tool for qualitative and quantitative analysis of tissue sections in preclinical drug development. LSC combines the strengths of flow cytometry with tissue architecture retention. This technology has been used predominantly with immunofluorescent techniques on cell culture and tissue sections, but recently LSC has shown promise in evaluating chromogenic immunohistochemistry (IHC) and histochemical products in paraffin-embedded and/or frozen tissue sections. Inverted light scatter measurements or a combination of inverted scatter and fluorescence allows automated determination of cell/nuclear counts (e.g., proliferation labeling indices), cell area (e.g., cellular hypertrophy), stromal elements, and labeling intensity (e.g., cytoplasmic/organellar proteins) in chromogen-labeled IHC or histochemical stained sections that correlates well with standard manual quantification methods. Segmentation with autofluorescence or dual immunolabeling facilitates capture of labeling data from specific cell populations. LSC evaluation of HE-stained sections is accomplished using autofluorescence/eosin fluorescence and inverse scatter. A standardized fluorescent approach with archivability, a lack of fluorescence quenching (photobleaching), and amenability to evaluation of multiple markers in a section has been demonstrated using Qdot® nanocrystals. Examples of LSC use in chromogenic IHC, routine histopathology, and Qdot® labeling will be reviewed, and advantages and disadvantages of this technology will be discussed.
Keywords
Introduction
Laser scanning cytometry (LSC) is a recently developed technology (late 1990s) that is capable of quantifying fluorescent and chromatic events in cells and tissues. The technology has the quantitative ability of flow cytometry but allows evaluation of cell populations in situ (i.e., retention of tissue architecture). High-content data are generated, which can include X and Y event positions (i.e., images produced as pixel maps), event area in square microns, event counts, circularity, integrated fluorescence/event (integral), MaxPixel (maximal pixel intensity/event), and peripheral integral and MaxPixel (with peripheral contour use). LSC data can be displayed in scatter plots, histograms, distribution plots, and statistics tables. LSC has traditionally been used with fluorescently labeled cell culture (live or fixed cells) samples and tissue sections. Examples of fluorescence-based LSC applications include: determination of cellular DNA content analysis (Kamiya et al. 1999); spatial resolution of nuclear vs. cytoplasmic fluorescence (Darzynkiewicz et al. 1999); cellular morphometry and cell cycle analysis using maximal pixel intensity (Gorczyca et al. 1996; Luther and Kamentsky 1996); analysis of enzyme kinetics (Bedner et al. 1998); drug uptake (Bedner et al. 1998); ligand binding (Bedner et al. 1998); evaluation of cytoplasmic/ nuclear translocation (Deptala et al. 1998); fluorescence in situ hybridization (FISH) analysis (Kamentsky et al. 1997); cell-to-cell interactions (Darzynkiewicz 1999); and quantification of fluorescent immunohistochemistry (IHC) labeling in tissue sections (Gorczyca et al. 1999; Pruimboom-Brees et al. 2005). Recently LSC has found use with quantification of chromogenically labeled IHC product in tissue sections and with quantitative morphologic analysis of routine histopathology specimens (i.e., hematoxylin and eosin (HE) and histochemical special stain sections) with detection of event counts, event area/volume, MaxPixel measurements, and integral values. Routine histomorphometric analysis and subjective scoring methods have traditionally been used to define, either in a quantitative or qualitative manner, morphologic endpoints in chromogenic immunohistochemistry (IHC) and routine histo-chemically stained sections. Automated high-content quantitative methods such as LSC offer more efficient data collection without interobserver variation. Exploiting the 488 nm and 633 nm laser light scatter/loss detection capability of the LSC (Figure 1), using autofluorescence for segmentation on certain cell types, measurement by use of nuclear (Figure 2) or phantom (Figure 3) contours of the entire section or in randomly placed areas of interest, and the ability to manipulate the laser channels (i.e., compensation) with production of virtual channels have allowed high content and reliable quantification in this medium. The ability for relatively high throughput (i.e., up to 180 slides) and robotic-based automation are also strengths for this technology. The LSC is also able to evaluate an entire tissue section up to 50 μm in thickness with < 20% loss of maximum pixel intensity using a 20X objective lens, therefore resulting in volume measurement that is a factor of section thickness and area in square microns (personal communication, Ed Luther, CompuCyte Inc., Cambridge MA). This paper will discuss LSC techniques that allow quantification of routine histochemical and immunohistochemical products in tissue sections and advantages and disadvantages of these techniques.
Materials and Methods
This paper is a review of technical approaches for laser scanning cytometric evaluation of nonfluorescent or “bright field” antibody-based labeling, and also the use of Quantum Dot® conjugated antibody labeling for a permanently mountable fluorescent approach with less “photobleaching.” Actual examples of these approaches are detailed. Where appropriate, the same sections used for LSC or tissue from the same organ were evaluated in parallel with a more traditional quantitative or semiquantitative approach, and LSC data were evaluated by a pathologist.
LSC Quality Control Measures/ Performance Qualification
The iCyte LSC produces data that have the potential to positively or negatively effect the progression of a compound through the development pipeline. For this reason, it is important to have quality control (QC) measures in place to support the integrity of the data. A QC workspace was created to evaluate the alignment and calibration of the instrument by scanning Sphero™ Ultra Rainbow Fluorescent Particle Slides (catalog #FPS-5057-UR5, Spherotech, Libertyville, IL). These slides contain a permanently mounted mixture of 5.0 μm particles in five different fluorescent intensities. The particles were contoured based on blue light loss, and the fluorescent emission was separated into three channels; green, orange, and long red. Histograms were produced showing the log intensity of the five populations separated into five distinct regions in each of the three channels. Once a baseline measurement is established, comparisons can be made to future QC checks to monitor potential fluctuations or drift in system performance. The instrument’s ability to separate populations of beads based on size is a second parameter that is useful for system QC and calibration. This QC check is performed using Sphero™ Rainbow Fluorescent Particle Slides that contain a mixture of six particle sizes: 0.56, 0.96, 3.0, 5.5, 10, and 15.5μm (catalog #FPS-M57-6, Spherotech, Libertyville, IL). The beads are excited with multiple laser lines, and the integral value of the emission is plotted against area in square microns. Individual populations of beads can be gated and viewed in an image gallery. In the field view display, actual measurement of individual beads in gated regions can be confirmed using the “line tool.” Other measures of performance qualification of LSC this laboratory has used include: reproducibility testing, equivalency testing, comparison of LSC results with tissue section (pathologist role), and use of optimally stained/ labeled sections.
Study Animals
Crl:CD(SD) (Charles River Laboratories, Inc., Raleigh, NC) and NTac:SD (Taconic, Germantown, NY) rats were group housed (2 to 3 per cage) in polycarbonate solid-bottom cages with ALPHA-dri™ (Shepherd Specialty Papers, Inc., Kalamazoo, MI) in an environment with a temperature of 64°F–79°F, 30%–70% relative humidity, and a 12-hour light/dark cycle. Rats were fed Prolab™ RMH 3500 Autoclavable Rodent Chow, made by PMI™ Nutrition International, Richmond, IN, ad libitum. ZDF lean/Crl- Lepr +/+ and ZDF obese diabetic/Crl-Lepr fa rats (Charles River Laboratories, Inc., Raleigh, NC) were housed 1 per cage in polycarbonate solid-bottom cages with Bed-O’Cobs™ (The Andersons, Maumee, OH) in an environment with a temperature of 64°F–79°F, 30%–70% relative humidity, and a 12-hour light/dark cycle, and fed Purina LabDiet™ brand Formulab Diet 5008 (pellets), made by PMI™ Nutrition International, Richmond, IN, ad libitum. Municipal water supply with additional treatment by reverse osmosis was available to all rat strains ad libitum from an automatic watering system.
C57BL/6 NTac (Taconic, Germantown, NY) and OB/OB mice on a C57BL/6 genetic background (Jackson Laboratory, Bar Harbor, ME) were housed 1 per cage in polycarbonate solid-bottom cages with Bed-O’Cobs® (The Andersons, Maumee, OH) in an environment with a temperature of 64°F–79°F, 30%–70% relative humidity, and a 12-hour light/dark cycle. The light/dark cycle could be interrupted for study-related activities. Mice were fed LabDiet® brand Certified Rodent Diet 5002 (pellets), made by PMI® Nutrition International, Richmond, IN, ad libitum. Municipal water supply with additional treatment by reverse osmosis was available ad libitum from an automatic watering system.
All animal handling and treatment in these animal studies were performed in accordance with the GlaxoSmithKline Animal Care and Use Committee (ACUC) guidelines.
Study Design
(1) Quantification of Discrete Nuclear Labeling
Male Sprague-Dawley™ (NTac:SD) (Taconic, Germantown, NY) rats were treated with a single oral dose of a classic biliary toxin at 50 mg/kg in corn oil as the vehicle and were evaluated over a period of 72 hours (2, 6, 24, 48, 72 hours post dose). Rats were euthanized under isoflurane anesthesia. The livers were removed, fixed in 10% neutral buffered formalin (NBF), and processed to paraffin block. Four-micron sections were loaded on the Eridan (Dako, Carpinteria, CA) automated immunohistochemical staining platform and deparaffinized. Antigen retrieval with Target pH6 (Dako) was performed, endogenous peroxidase was quenched, and alkaline phosphatase was inhibited using Dual Endogenous Enzyme Block (Dako). Nonserum protein block was also used (Dako). To identify proliferating cells, liver sections were labeled with a polyclonal rabbit anti-Ki67 clone SP6 (Thermo Scientific/Lab Vision, Fremont, CA) diluted 1:200 in antibody diluent with background reducing components (Dako). The primary antibody reaction was detected using the Envision+ labeled polymer HRP (Dako), followed by incubation in DAB+ (Dako). Next, to identify biliary epithelium, the same tissue section was labeled with a monoclonal mouse anti-AE1/AE3 cytokeratin antibody (Dako) diluted 1:50 in antibody diluent with background reducing components, and detected by a biotinylated horse anti-mouse rat adsorbed secondary antibody (Vector Laboratories, Burlingame, CA) diluted 1:150 in tris-buffered saline followed by an incubation with streptavidin-alkaline phosphatase (KPL Inc., Gaithersburg, MD) that was allowed to react with Liquid Permanent Red (Dako) and counterstained in Mayers Hematoxylin (Dako).
A scanning protocol was created on the iCyte Laser Scanning Cytometer (CompuCyte, Cambridge, MA) using the iNovator (CompuCyte) software. A mosaic scan of the entire tissue section was created with a 40X objective and a 20 μm step size, with tissue contouring by auto-thresholding of absorption produced by the Helium-Neon (HeNe) laser and hematoxylin dye. Once the mosaic scan was completed, a 2.5 × 5 mm rectangular region was randomly placed on each tissue section. Preference was selected in the software to automatically avoid tissue edges. A high-resolution scan was performed for each region. The iNovator was set to field scan using a 40X objective and a 0.5 mm step size. Three channels were configured, and thresholds were optimized to selectively contour primary events. The first channel was set to contour the DAB-stained Ki67 positive cells using inverted scatter (absorption) produced by the 488 argon laser. Compensation was performed to correct for the overlap of the permanent red chromagen by subtracting the long red channel from the absorption of the DAB. This second channel, long red, was set to contour permanent red-stained cytokeratin-positive biliary epithelium using fluorescence produced by argon 488 excitation and the long red (650 to 800 nm) emission filter. The third channel was set to contour hematoxylin-stained nuclei using inverted scatter produced by the HeNe laser.
Peripheral contouring of Ki67 and nuclei was selected to collect signal produced in the long red channel. This step allowed the separation of Ki67 positive cells and nuclei that were present in permanent red-stained biliary structures. The Ki67-positive biliary cell population was selected by gating on cells that were identified on a scatter plot of argon 488 absorption minus long red MaxPixel on the y-axis and long red peripheral max on the x-axis. The Ki67-negative biliary cell population was selected by gating on cells that were identified on a scatter plot of HeNe laser absorption MaxPixel on the y-axis and long red peripheral max on the x-axis. Visual confirmation of desired cell populations was performed by repeating view of image galleries produced from these gated regions. In a similar manner, a count of Ki67-positive hepatocytes was collected by gating on a region of cells produced in a scatter plot of argon 488 absorption minus long red on the y-axis and green peripheral max on the x-axis. The green signal (autofluorescence) represented the cytoplasm of hepatocytes. Ki67 events that also had a high green peripheral max were determined to be hepatocytes. Cells expressing reduced green peripheral max levels were more likely to be Kupffer cells or circulating inflammatory cells within hepatic sinusoidal spaces. Cell size and green peripheral max values were taken together to apply gating to restrict cell counts to the population of interest. Ki-67 labeling indices (LI) were calculated from the following LSC data: Ki-67+ biliary epithelial cell number/total biliary epithelial cells and Ki-67+ hepatocytes/ total hepatocytes, and expressed as a percentage.
(2) Quantification of Cytoplasmic/ Organellar Labeling
Male C57BL/6NTac mice were treated orally with a perox-isome proliferator-activated receptor (PPAR) δ agonist tool compound and WY-14, 643 (PPARα agonist) as a control (0, 100, and 250 mg/kg/day and 50 mg/kg/day, respectively) over 14 days. The vehicle for both treatments was 0.5% hydroxypropyl methylcellulose. Mice were euthanized under isoflurane anesthesia, and livers were removed, fixed in 10% NBF, and processed to paraffin block. Four-micron sections were loaded on the Ventana Discovery XT (Ventana Medical Systems, Tucson, AZ) automated immunohistochemical staining platform, where they were deparaffinized, incubated in cell conditioner 1, avidin/biotin block, and antibody block. To identify peroxisomes, tissue sections were labeled with a polyclonal rabbit anti-peroxisome membrane protein 70 (PMP70) (Affinity BioReagents, Golden, CO) diluted 1:100 in Ventana antibody diluent for 60 min. The primary antibody was detected with a biotinylated goat anti-rabbit secondary antibody (Vector Laboratories, Burlingame, CA) at a concentration of 8 μg/mL in Ventana antibody diluent followed by DAB Map (Ventana Medical Systems), and sections were counterstained with hematoxylin.
A scanning protocol was created on the iCyte Laser Scanning Cytometer (CompuCyte, Cambridge, MA) using the iNovator (CompuCyte) software. A mosaic scan of the entire tissue section was created with a 10X objective and a 20 μm step size, with tissue contouring using signal produced by argon 488 laser light absorption with threshold set at 2500. Once the mosaic scan was completed, four 1 × 1 mm regions were randomly placed, by the software, on each tissue section. Random region placement was manually adjusted as necessary to avoid tissue edges or empty spaces. High-resolution analysis was performed for each region using the field scan module configured to use a 40X objective and a 0.5 mm step size. Phantom contouring used to collect event information was configured in a lattice arrangement, with a radius of 5 μm and a minimum distance between centers of 10 μm. The argon 488 laser was used to scan the high-resolution fields, which generated integral absorption mean values that correlated with the quantity of peroxisomes in the selected tissue regions.
Transmission Electron Microscopy:
Liver samples from each mouse were processed routinely for TEM, semi-thin sections (1 ∝μ section stained with 1% Toluidine blue) were cut to select areas of interest (centrilobular hepatocytes) for thin sectioning of the liver, thin sections (~ 90 nm) were mounted on a 200-mesh copper grid, stained with 5% methanolic uranyl acetate and Reynold’s lead citrate, then examined on a JEOL 1010 (JEOL Inc., Boston, MA) transmission electron microscope. The entire grid was examined, and five digital TEM photomicrographs were taken at the same scope magnification (1000X). For each digital image, the total number of mitochondria and peroxisomes per micrograph were manually counted, and histomorphometric analysis was performed on the same digital images using a computer-assisted software package Image-Pro-Plus™ version 5.1 (Media Cybernetics Inc., Bethesda, MD).
PMP70 and β-actin Immunoblots:
Protein fractions of snap-frozen livers of mice were prepared according to the T-PER™ Tissue Protein Extraction Reagent Protocol (Pierce, Rockford, IL). Approximately 200 mg of liver tissue was homogenized in the extraction reagent containing a protease inhibitor cocktail (Sigma, St. Louis, MO) and centrifuged at 10,000 × g for 5 minutes at 4ºC to pellet cellular debris. The supernatant was filtered through glass wool, analyzed for protein concentration by the BCA protein assay (Pierce, Rockford, IL), and used for PMP70 immunoblot assay. SDS-PAGE was performed using gradient 4% to 12% Tris-Bis precast gels under conditions supplied by the manufacturer (Bio-Rad Laboratories, Inc., Hercules, CA), and separated proteins were electrophoretically transferred to a PVDF membrane (Bio-Rad Laboratories, Inc., Hercules, CA) for 30 minutes at 100 V. The membranes were blocked for one hour; incubated with blocking solution containing both rabbit anti-PMP70 (ABCD3 and PXMP1) antibody (Affinity Bioreagents, Golden, CO) diluted 1:1000 and mouse anti-β-actin antibody (Abcam Inc., Cambridge, MA) diluted 1:4000 for three hours; washed in PBST; incubated for 1 hour in blocking solution containing IRDye™800 conjugated goat anti-rabbit IgG (Rockland, Gilbertsville, PA) at a dilution of 1:1000 and Alexa Fluor™ 680 conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR) diluted 1:5000; washed in PBST; rinsed in PBS; and air dried. Two-color immunoblotting was used to simultaneously detect β-actin and PMP70. Membrane scanning was performed using the 700 and 800 nm channels of an Odyssey™ Infrared Imaging System (LI-COR Biosciences, Lincoln, NE). Quantitation of PMP70 levels was determined by normalizing the signal for PMP70 to the signal for β-actin.
(3) Morphologic Quantification in Hematoxylin and Eosin-Stained Sections
Cellular Hypertrophy Analysis:
OB/OB mice on a C57BL/6 background were treated orally with a classic hypothalamic-pituitary-adrenal (HPA) axis disruptor (Cmpd X) at a dose of 50 mg/kg. Mice were euthanized under isoflurane anesthesia, and adrenal glands were removed, fixed in 10% NBF, processed to paraffin block, and sectioned.
Crl: CD™(SD) rats were treated orally with a PPARδ agonist tool compound at doses of 5, 30, and 100 mg/kg/day and a PPARα agonist at doses of 50, 300, and 1000 mg/kg/day for 14 days. Rats were euthanized under isoflurane anesthesia, and livers were removed, fixed in 10% NBF, processed to paraffin block, and sectioned.
Four-micron sections of formalin-fixed, paraffin-embedded mouse adrenals and rat liver were stained with HE using a routine method. Completed slides were loaded onto the laser scanning cytometer for semiautomated evaluation. A scanning protocol was created to measure change in cytoplasmic cell area/volume to nuclei area/volume, which was done by first producing a mosaic image of the entire tissue section. To produce this image, the iNovator was configured to produce a low-resolution mosaic scan with a 20X objective and a 5 μm step size of the adrenals. Mosaic scans were performed on the liver sections using a 10X objective and a 20 μm step size. The tissue was contoured using blue light loss (absorption) created by the argon 488 laser, with the threshold set at 4500 for the adrenals and 7500 for the liver. Once the adrenal mosaic scan was completed, five 200 × 200 μm regions were manually placed to limit data acquisition to the adrenal cortex. For the liver, one 3 × 3 mm region was randomly placed by the software, with preference selected to avoid edges. High-resolution analysis was performed for each region using the field scan module configured to use a 20X objective and a 0.5 μm step size for adrenals and a 40X objective and a 0.5 μm step size for the liver. Phantom contouring, used to collect event information, was configured in a lattice arrangement with a radius of 5 μm and a minimum distance between centers at 10 μm. The HeNe laser was configured to evaluate the hematoxylin-stained nuclei by measurement of red light loss. The mean integral values of the red light loss in the scatter 2 channel were used to assess the amount of nuclear material. Argon 488 excitation of the eosin-stained cytoplasm produced a fluorescent emission in the green channel. The mean integral values of the green channel were used to assess the amount of cytoplasm. In addition, the mean integral absorption of the eosin-stained cytoplasm was measured using blue light loss in the scatter 1 channel. This process took advantage of eosin’s dual property of being both a fluorescent and light-scattering dye and allowing comparison of the two techniques on the same sample simultaneously.
(4) Morphologic Component Quantification in Histochemically Special Stained Sections
ZDF lean/Crl- Lepr +/+ and ZDF obese diabetic/Crl-Lepr fa rats were evaluated at 6, 8, 11, and 17 weeks of age. Rats were euthanized under isoflurane anesthesia, and pancreata were collected at each time point, fixed in 10% neutral buffered for-malin, and transferred to 70% ethanol after 24 hours. Entire pancreata were trimmed by a uniform random method, routinely processed to paraffin block, sectioned at 4 μm, and stained with the routine histochemical stain, Masson’s Trichrome.
Sections were loaded on the laser scanning cytometer, and a mosaic scan was performed using a 20X objective and a 20 μm step size. The tissue was contoured using the signal produced by the argon 488 blue light loss with the threshold set at 11000. Once the mosaic scan was completed, a high-resolution field scan of the entire tissue section was performed using a 20X objective and a 0.5 μm step size. Blue-stained collagen was measured using red light loss produced by the HeNe laser, inverted scatter 2 channel. Other tissue components, which were stained red, were measured using blue light loss produced by the argon laser, inverted scatter 1 channel. Compensation was performed to remove spectral overlap of scatter 1 into scatter 2 by creating a virtual channel where scatter 1 is subtracted from scatter 2. The integral mean values for the virtual channel represent the amount of collagen in a given tissue section.
(5) Quantification of Qdot® Nanocrystal Fluorescently Labeled Sections
ZDF lean/Crl- Lepr + + and ZDF obese diabetic/Crl-Lepr fa rats were evaluated at 6, 8, 11, and 17 weeks of age. Rats were euthanized under isoflurane anesthesia, and pancreata were collected at each time point, fixed in 10% NBF, and transferred to 70% ethanol after 24 hours. Entire pancreata were trimmed by a uniform random method, routinely processed to paraffin block, sectioned at 5 μm, and immunohistochemically labeled for insulin, glucagon, and somatostatin by QuantumDot (Qdot®)-conjugated antibodies.
Rat pancreata were routinely fixed in 10% NBF and processed to paraffin blocks. Four-micron sections were cut on a rotary microtome and then loaded on the Eridan (Dako, Carpinteria, CA) automated immunohistochemical staining platform. The following steps were programmed in the Eridan software: deparaffinization, antigen retrieval in Target pH6 (Dako), avidin/biotin blocking kit (Dako), nonserum protein block (Dako), and primary and secondary antibody incubations. A cocktail of pancreatic hormones was prepared by combining guinea pig anti-insulin at 1:400 (Dako), rabbit anti-somatostatin 1:100 (Dako), and mouse anti-glucagon (Abcam; Cambridge, MA) at 1:200 in antibody diluent with background reducing components (Dako). This cocktail was applied to the tissue, and the tissue was incubated for 30 min. A second cocktail of antibodies was prepared that contained biotinylated goat anti-guinea pig at 1:150, goat anti-rabbit conjugated quantum dot 585 at 1:100 (Invitrogen, Carlsbad, CA), and goat anti-mouse quantum dot 655 1:100 (Invitrogen) in tris buffered saline. This mixture was applied and incubated for 30 min, followed by final incubation in streptavidin conjugated quantum dot 525 at 1:200. Sections were dehydrated through increasing concentrations of ethanol to xylene and permanently mounted in cytoseal (Richard Allen Scientific, Kalamazoo, MI).
Slides were loaded on the iCyte for fully automated quantification of protein expression. A 20X objective with 20 μm step size was selected to produce a mosaic scan of the entire tissue section using argon excitation of the autofluorescence as the primary signal for contouring at threshold 1394. Following the mosaic scan, a high-resolution field scan was performed using the 20X objective and 2 μm step size. A 405 violet diode laser was used as the single excitation source for all quantum dots. Three quantum dots were selected, one for each of the three emission cubes on the iCyte. Quantum dot® 525 was selected for the green, Qdot® 585 for the orange, and Qdot® 655 for the long red channel. To compensate for overlapping spectra, virtual channels were created as follows: insulin Gr-Or (orange channel subtracted from green), glucagon LR-Gr (green channel subtracted from long red), and somatostatin Or-Gr (green channel subtracted from orange). The signal in each of these channels was optimized by repeated test scans, with adjustments made to photomultiplier tubes (PMTs) and thresholds prior to running the study. Fluorescence in each of the channels was contoured based on threshold settings, and the integral mean value for each virtual channel was collected. These values correlated with the level of protein expression for each of the labeled hormones.
Subjective Histopathologic and Immunohistochemistry Scores and Histomorphometry:
Subjective scoring of histopathologic change or immunoreactivity was based on a six-point scoring system where 0 = no, 1 = minimal, 2 = mild, 3 = moderate, 4 = marked, and 5 = very marked histopathologic change/ immunoreactivity. Data were expressed as group mean scores.
Mean area of insulin immunoreactivity and positive nuclear counts were determined using the Image-Pro® Plus version 5.1 (Media Cybernetics Inc., Bethesda, MD) histomorphometry software package. Segmentation on insulin immunore-activity within sections of ZDF rat pancreas with area measurement, and area measurement of the entire pancreas were performed. Data were expressed as percentage of insulin area as a function of entire pancreas area. Manual tagging of positive and negative nuclei was performed and expressed as a labeling index.
Statistical Analyses
Results are expressed as group means ± SD. Statistically significant differences in results parameters were determined by variance analysis (ANOVA) and/or Student’s t-test. Statistical differences were considered significant with a p value < .05. Pearson’s correlation coefficients (R 2) of data generated by LSC and other semiquantitative or quantitative methods were determined.
Results
Quantification of Discrete Nuclear Labeling
Rats treated with a “classic” biliary toxin had a time-dependent increase in the Ki-67 labeling index (LI) of both biliary epithelial cells (up to 0.53 ± 0.08 at 72 hours) and hepatocytes (up to 0.95 ± 0.02 at 72 hours), which were significantly increased over vehicle controls at the same time point (p < .05, Figures 4 and 5), as determined by LSC analysis (Figure 6). There was a strong correlation between the LSC data and subjective scoring of Ki-67 labeling in both cell types (R 2 = 0.968, p < .05 and R 2 = 0.989, p < .05 in biliary epithelium and hepatocytes, respectively).
Quantification of Cytoplasmic/Organellar Labeling
Mice treated with a PPARδ agonist tool compound and WY-14, 643 had increases in hepatocellular PMP 70 labeling (increased mean signal intensity expressed as integral mean) when compared with vehicle control mice. The increases were greater in mice treated with the PPARδ agonist tool compound when compared with WY-14, 643 treatment (Figures 7 and 8). The LSC data correlated well with the subjective scoring of PMP 70 immunoreactivity, which was performed on the same sections as LSC (R 2 = 0.955, p < .05) (Figure 7), transmission electron microscopy (where only five hepatocytes were evaluated; R 2 = 0.667, p <.05) (Figure 7), and PMP 70 immunoblotting (from frozen liver tissue adjacent to sections collected for histopathology; R 2 = 0.875, p < .05; Figure 7).
Hypertrophy Analysis with the Laser Scanning Cytometer
Adrenal Gland
OB/OB mice treated with a classic HPA axis disruptor at doses of 50 mg/kg/day for 14 days had significantly increased cytoplasm-to-nuclei ratios based on light scatter data from HE slides collected by LSC when compared with lean (p = .0021) and OB/OB (p = .0455) vehicle control mice (Figure 9). Overall there was a slight decrease in nuclear area in treated animals, combined with an increase in adrenal cortical cell cytoplasm area. This result correlated with the microscopic findings of adrenal cortical hypertrophy in treated OB/OB mice.
Liver
Rats treated with a PPARα agonist and a PPARδ agonist tool compound for 14 days had microscopic evidence of a dose-dependent centrilobular hepatocellular hypertrophy and cytoplasmic eosinophilic granularity interpreted as peroxi-some proliferation. LSC analysis of HE-stained livers from the study showed a dose-dependent increase in cytoplasm in treated animals compared with vehicle controls, whereas nuclei remained consistent between vehicle controls and treated rats (Figure 10). The LSC changes correlated with histopathology results (R 2 = 0.791556, p < .05) expressed as subjective severity scores showing centrilobular hepatocellular hypertrophy.
Morphologic Component Quantification in Histochemically Special Stained Sections
During diabetogenesis in the ZDF obese diabetic/Crl-Lepr fa rat, pancreatic islets become disorganized, with islet cell loss and replacement with fibrous connective tissue (collagen). Masson’s Trichrome is a routine histochemical stain that allows determination of collagen and that was used to allow both qualitative and quantitative (with LSC) assessment of pancreatic fibrosis. LSC data showed a time-dependent increase in the amount of collagen present in pancreata of aging ZDF obese diabetic/Crl-Lepr fa rats over ZDF lean/Crl- Lepr +/+ controls (Figure 11), which was correlated with subjective scoring of the same sections for the presence of collagen (R 2 = 0.7768, p < .05).
Quantification of Qdot® Fluorescently Labeled Sections
The sums of the intensity and area (square microns) of insulin, glucagon, and somatostatin labeling in ZDF lean/Crl- Lepr +/+ and ZDF obese diabetic/Crl-Lepr fa rats at 6, 8, 11, and 17 weeks of age were measured by LSC. Insulin data only will be described in this paper. Qdots® were conjugated to primary or secondary antibodies and used to label these markers (Figures 12 and 13). Area of insulin labeling in square microns showed a steady time-dependent increase in lean rats, which reached a plateau at 17 weeks of age, and a slow time-dependent decrease in ZDF obese diabetic/Crl-Lepr fa rats, which was significantly decreased compared with the six-week time point by 17 weeks of age (p < .05) (Figure 14). Percentage pancreatic beta cell area, as determined by histomorphometry, correlated well with the LSC-generated area data (R 2 = 0.9365, p < .05). Insulin labeling intensity (expressed in pixel intensity units) as the integral sum of the fluorescence was measured using 525 nm Qdots® by LSC. In ZDF lean/Crl- Lepr +/+ rats, insulin labeling intensity/unit area was variable but present at a higher level than in the ZDF obese diabetic/Crl-Lepr fa rats. In the ZDF obese diabetic/Crl-Lepr fa rats, insulin labeling intensity/unit area was relatively low but was increased at 17 weeks of age (Figure 15).
Discussion
In this manuscript, several approaches to LSC quantitative evaluation of morphologic endpoints in chromogenic immuno-histochemical and routine histochemically stained sections have been detailed by example. LSC has been shown to be an efficient, relatively high-throughput, and reliable automated technology that generates high-content data to quantify morphologic endpoints in nonfluorescent tissue samples with an open/flexible platform. Comparison of LSC data with other standard measures, such as histopathologic subjective scoring, histomorphometry, transmission electron microscopy, and immunoblotting, have shown strong correlations between the individual quantification techniques. In the examples described in this manuscript, a wide range of nonfluorescent applications of LSC and use of Qdot®-conjugated antibodies are detailed.
Discrete nuclear labeling signals in two separate cell types in the same section were quantified simultaneously using segmentation on a specific cytoplasmic label (i.e., cytokeratin AE1/AE3) in biliary epithelium and on autofluorescence in hepatocytes. The number of total and Ki-67-labeled biliary epithelial cells and hepatocytes were enumerated, which allowed calculation of Ki-67 labeling indices for both cell types. These data showed a prominent time-dependent increase in both biliary epithelial and hepatocellular labeling indices after treatment with a classic biliary toxin, which was interpreted as a compensatory regenerative response and was strongly correlated with subjective scoring methods.
Immunohistochemical labeling of peroxisomes for PMP70 in livers of mice treated with PPAR agonists was quantified by LSC. PPAR-agonist treatment caused peroxisome proliferation with an associated increase in the level of PMP70 expression. There was a dose-dependent increase in PMP70 labeling in mice treated with the PPARδ agonist tool compound, which was more severe than in mice treated with WY-14, 643. These LSC data were strongly correlated with subjective scoring of PMP70 immunoreactivity, peroxisome numbers determined by transmission electron microscopy, and hepatic PMP70 protein levels determined by immunoblot. The evaluation of inverse scatter from the PMP70-associated DAB product was also compared with Alexa Fluor 488 conjugated PMP70 antibody fluorescent labeling, where there was a signal saturation effect of fluorescence labeling resulting from the high level of hepatocellular PMP70 expression. This saturation effect did not affect the scatter detection measurements, suggesting that in this situation, LSC quantification was more reliable in chromogenically labeled tissue sections compared with fluorescent labeling. LSC allowed reliable quantification of organellar/ cytoplasmic protein chromogenic labeling, which was validated by several other approaches. This laboratory has also found that LSC quantification of integral intensity of chromogenically labeled nonstructural cytoplasmic proteins can be problematic because of the rather narrow dynamic range of commercially available chromogens (e.g., diaminobenzidine) (Rimm 2006). If quantification of a nonstructural cytoplasmic product is required, it is suggested to use either routine immunofluorescence or Qdot®-conjugated antibody labeling to allow for a broader dynamic range for intergroup/interanimal comparison.
Cellular hypertrophy was measured in HE-stained sections of mouse adrenal gland and rat liver in animals treated with compounds that are known to elicit cellular hypertrophy in the respective tissues. LSC analysis with phantom contours to collect event data generated by light loss/scatter was used to measure change in cytoplasmic area/volume and nuclei area/ volume. A classic HPA axis disruptor elicited adrenal cortical cell hypertrophy, whereas the PPAR agonists elicited peroxisome proliferation, which was associated with hepatocellular hypertrophy. LSC analysis of cellular hypertrophy has been shown to be an accurate measure, and possibly more sensitive than histopathology for the detection of low-grade cellular hypertrophy. LSC data evaluated both nuclear and cytoplasmic changes, which can be used together or separately. The ability of the LSC to evaluate the entire thickness of a tissue section (i.e., 4–5 μm) adds a measure of volumetric analysis to this application compared with more surface-focused technologies.
Pancreatic collagen content (area and integral) in aging ZDF obese diabetic/Crl-Lepr fa rats was measured by evaluation of collagen staining by the routine histochemical stain, Masson’s Trichrome. LSC was consistently able to detect fine fibrillar arrangements of collagen, which were frequently difficult to see by light microscopy. The LSC was not only able to quantify the abundance of collagen, it also reliably detected minimal collagen deposits that were not readily detected by routine histopathologic examination, allowing a more complete evaluation of fibrosis in this experiment.
Qdot®-conjugated antibody-based immunofluorescence has been shown to be a good alternative to routine immuno-fluorescence with the added benefits of permanent mountability and decreased “photobleaching” with laser light. Qdot®-conjugated antibodies for insulin, glucagon, and somatostatin in ZDF lean/Crl- Lepr +/+ and ZDF obese diabetic/Crl-Lepr fa rats were readily detectable and quantified by the LSC. Insulin immunoreactivity as a measure of beta cell mass was determined by LSC measurement of insulin area/volume across representative uniform random samples of the entire pancreas. In ZDF obese diabetic/Crl-Lepr fa rats, there was a significant decrease in insulin area/volume by 17 weeks of age, whereas there was a steady increase in insulin area/volume over time interpreted as normal pancreatic islet growth in the nondiabetic strain.
LSC has been shown in this paper to be a powerful quantitative technique for chromogenic IHC and routine histochemical stains. Advantages of this technology include automation, relatively high throughput capacity (up to 180 slides/run), high correlation with alternative semiquantitative/quantitative methods, and the generation of biologically meaningful morphologic endpoint data. Disadvantages of this technology are few, but quantification of intensity integral of nonstructural cytoplasmic antigens labeled with a chromogen is complicated by the rather narrow dynamic detection range; therefore, fluorescent labeling applications are more suitable for this approach. Overall this paper shows that LSC is comparable to, and in some applications superior to, alternative semiquantitative/quantitative methods with the added benefits of automation and high-throughput capacity.
Footnotes
Acknowledgments
We would like to thank Connie Cummings, DVM, PhD, for ultrastructural pathology and histomorphometry support, and Greg Falls, PhD, for his immunoblot support.