Q

Cytometry 13:404-415 (1992)

1992 Wiley-Liss, Inc.

of Analytical Methods for the Study 1 Liver Cell Proliferation' Per Gerlyng, Trond Stokke, Henrik S. Huitfeldt, Tom Stenersen, Havard E. Danielsen, Tom Grotmol, and Per 0. Seglen2 Departments of Tissue Culture (P.G., T.G., P.O.S.), Biophysics (Tr,S.), and Pathology (To.S., H.E.D.), Institute for Cancer Research, The Norwegian Radium Hospital, and Department of Environmental Health, The National Institute of Public Health (H.S.H.), Oslo, Norway Received for publication June 28, 1991; accepted November 23, 1991

Various cytometric methods for analysis of regenerating rat liver growth (DNA ploidy distributions, binucleation, and DNA synthesis by in vivo BrdUrd incorporation) were evaluated. The overall hepatocellular growth rate (labeling index), the binucleation rate, and separate indices for mononuclear and binuclear cells could be measured simply by microscope counting of collagenase-isolated hepatocytes immunostained for BrdUrd. Flow cytometry of cells stained for BrdUrd and DNA provided labeling indices for the various hepatocellular DNA ploidy classes as well as for nonparenchymal cells (identified by their size-dependent light scatter), but could not distinguish between mononuclear and binuclear hepatocytes. Image cytometry, using fluorescence or Feulgen staining, was inferior to flow cytometry in terms of speed and DNA resolution, but allowed a complete analysis of all hepatocellular DNA ploidy and nuclearity classes. It may

The study of liver cell proliferation represents a particular methodological challenge, since the cellular properties of liver tissue are inordinately complex (25). Rat hepatocytes can be diploid, tetraploid, or octoploid (and a few even hexadecaploid), and they can have one or two nuclei. During proliferation there may therefore be five major cell classes (2N, 2 x 2N, 4N, 2 x 4N, and 8N) which each distributes through the different phases of the cell cycle. After nuclear division a hepatocyte may either divide completely or enter the next G, phase without cytokinesis, forming a binuclear cell. The rate of binucleation vs. mononucleation is an important aspect of the hepatocellular growth pattern: whereas binucleation is prominent in normal, developmental growth, it is strongly diminished in regenera-

therefore be the method of choice, particularly for analysis of liver cell cultures from which single cells are not easily obtained. Fluorescence staining would seem to be preferable to Feulgen staining, since the latter could not be used simultaneously with BrdUrd staining and therefore required a two-step analysis. A non-immunological method, based on the ability of incorporated BrdUrd to quench DNA staining by a Hoechst dye, could only be applied to isolated nuclei, thus giving no information about binucleation. The latter method may be useful for analysis of tumors which are difficult to dissociate to intact whole cells. 0 1992 Wiley-Liss, Inc.

Key terms: Rat, hepatocyte, growth, regeneration, flow cytometry, fluorescence image cytometry, Feulgen image cytometry, image analysis, ploidy, binucleation,

DNA

tive, hyperplastic, and neoplastic growth (24). In addition to the hepatocytes, the liver also contains, by number, about 40% nonparenchymal cells (vascular and connective tissue cells, Kupffer cells, and bile ductular cells) which may further complicate the picture. We have previously attempted to deduce hepatocellular growth patterns from flow cytometric analyses of cellular and nuclear DNA (ploidy) distributions 'This study was generously supported by the Norwegian Cancer Society. 'Address reprint requests to Dr. Per 0. Seglen, Department of Tissue Culture, Institute for Cancer Research, The Norwegian Radium Hospital, Montebello, N-0310 Oslo 3, Norway.

METHODS FOR STUDY O F LIVER C E L L PROLIFERATION

(18,19,24). The development of immunocytochemical methods for detection of BrdUrd incorporated into the DNA of proliferating cells (9) now makes it possible to investigate liver cell proliferation more directly. A few studies of hepatocellular growth utilizing BrdUrd incorporation have been published (5,11,21,22,28), showing that the technique is, in principle, applicable t o hepatocytes. However, for a complete analysis of hepatocellular growth i t is necessary to develop analytical procedures capable of resolving the different cell types and the ploidyinuclearity classes of proliferating and nonproliferating hepatocytes. In the present study BrdUrd staining techniques have been optimized for isolated hepatocytes, and various analytical approaches have been investigated in an attempt t o find methods, or combinations of methods, suitable for analysis of hepatocellular growth. MATERIALS AND METHODS Animal Treatment, Isolation of Hepatocytes and Nuclei Male Wistar Kyoto rats (8 weeks old, -150 g) were subjected to a two-thirds partial hepatectomy (PH). 5Bromo-2‘-deoxyuridine (BrdUrd) was administered as a single 50 mg tablet (Boehringer Mannheim, Mannheim, Germany) 15 h after PH in a surgically prcpared subcutaneous pocket to ensure a continuously high level of labeling (1).The liver was isolated and perfused 8 h later (23 h after PH) by the two-step collagenase perfusion technique (20). The resulting cell suspension contained both parenchymal and nonparenchymal cells. The hepatocytes were purified by differential centrifugation as described previously (20). The final cell suspension contained 5-1210 nonparenchyma1 cells (microscopy). Isolated nuclei were prepared from the final cell suspension by the detergent method of Vindelgiv (27); this method produces clean, single nuclei essentially devoid of residual cytoplasm.

405

Burlingame, CA, USA), 0.5% Tween 20, and 0.5% w/v bovine serum albumin and maintained a t 4°C for 30 min. After one centrifugation in PBS (4°C) the cells were resuspended in 50 pl PBS with 1/50 fluorescein isothiocyanate (F1TC)-streptavidin (Amersham, Little Chalfont, UK), 0.5% Tween 20, and 0.5% w/v bovine serum albumin at 4°C for 30 min. The cells were washed once in PBS, resuspended in PBS with 2.5 pg/ ml propidium iodide (PI)(Calbiochem, San Diego, CA, USA) and 100 pg/ml RNAse (type IA from bovine pancreas, Pharmacia LKB, Uppsala, Sweden), and left a t room temperature for 30 min in the dark. The cells were stored at 4°C until analysis by flow cytometry or image cytofluorometry. Cells prepared for Feulgen spectrometry measurements were fixed in 4% formaldehyde, and BrdUrd was exposed by using nuclease digestion of DNA (7) (Amersham Cell Proliferation Kit). Bound BrdUrd-antibody was detected by peroxidase-conjugated antibody against mouse immunoglobulin (Amersham). Diaminobenzidine (DAB; Amersham) gave brown-black staining at sites of BrdUrd incorporation. (For details see procedure from Amersham Cell Proliferation Kit.)

Flow Cytometry Using Hoechst Dye Quenching For analysis of liver cell proliferation by BrdUrdquenching of Hoechst dye binding (16), nuclei from BrdUrd-labeled hepatocytes or liver tumors were prepared by the Vindelgiv method (27) and stained with 2 pg/ml Hoechst 33258 (Riedel de Haen, Hannover, Germany) and 1.5 pg/ml ethidium bromide (EB; Calbiochem). The fluorescences of Hoechst 33258 and EB were measured simultaneously with forward angle light scattering in a n Argus flow cytometer (Skatron, Lier, Norway) equipped with a UV filter block (excitation a t 365 nm; emission at >420 nm) and a “Bl” filter block (excitation a t 395-440 nm; emission a t >470 nm) a s beam splitter. The EB fluorescence was further B r d U r d Immunocytochemistry filtered through a >625 nm longpass filter to exclude For flow cytometry and fluorescence image analysis Hoechst 33528 fluorescence from entering the photothe staining protocol of van Erp et al. (26) with the multiplicator for detection of EB fluorescence. BrdUrd modification proposed by Kirkhus and Clausen (13) incorporation in DNA results in decreased fluorescence was employed. Cells, fixed in 70% ethanol a t -20°C from Hoechst 33258, while the EB fluorescence is unovernight, were washed once in phosphate-buffered sa- affected. Hence, when EB was excited directly, i.e., a t line (PBS) at 4“C, resuspended in 2 ml 0.2% pepsin 546 nm or 488 nm, the fluorescence of this fluorophore (porcine stomach mucosa pepsin, type P7125, from accurately reflected DNA content. Note that EB was Sigma Chemicals, St. Louis, MO, USA) in 2 N HC1, and partially excited by transfer of energy from Hoechst left a t room temperature for 60 min in the dark. After 33258 in the experiments where UV excitation was pepsin treatment, the cells were washed three times employed. This led to a small BrdUrd quenching of the with PBS (4”C), and finally resuspended in 50 pl PBS EB fluorescence a s well. In some experiments UV-exwith antibody to BrdUrd ( 1 : l O ) (Becton Dickinson, cited Hoechst 33258 fluorescence was measured versus Mountain View, CA, USA), 0.5% Tween 20 (Sigma), 488 nm-excited EB fluorescence (and 488 nm forward and 0.5% w/v bovine serum albumin (Sigma type angle light scattering) in a Coulter Epics V flow cyA4378, crystallized and lyophilized). The cells were tometer (Coulter Corporation, Hialeah, FL, USA) maintained in suspension at 4°C for 30 min, then cen- equipped with argon lasers tuned to UV and 488 nm, trifuged three times at 2,000 rpm x 7 min in PBS respectively. As described above, the EB fluorescence (4°C). The pellets were resuspended in PBS with 1/50 was unaffected by incorporated BrdUrd in these cases biotinylated horse antimouse IgG antibody (Vector, and reflected total DNA content. The light scattering

406

GERLYNG ET AL.

Fluorescence Image Cytometry Cytospin preparations of cell suspensions, stained with PI and anti-BrdUrdiFITC, were examined by comFlow Cytometry Using Anti-BrdUrd-Antibody puterized fluorescent image analysis (11).A Nikon Miand FITC-Labeling crophot microscope equipped with a n epifluorescence Ethanol-fixed cells were BrdUrd-stained as de- attachment (HMX-HBO 100 W lamphouse) and filter scribed above. The cells were analyzed with the Argus blocks G (PI fluorescence) and B2E (FITC fluoresflow cytometer equipped with the “FITC 510” filter cence) was used. Excitation light was adjusted to the and the “FITC-PE” filteribeam splitter, the latter be- sensitivity range of the video camera by grey-filters. ing used to separate the red PI fluorescence from the The microscope was equipped with a high sensitivity yellow FITC fluorescence. The forward light scatter of video camera (JAI 733 SIT, JAI, Copenhagen, Denwhole cells was used to discriminate between hepato- mark) modified to allow manual setting of the gain and cytes and the smaller nonparenchymal cells. black offset. The camera was connected to a Magiscan image analysis system (Joyce-Loebl, Gateshead, UK) Feulgen Image Cytometry with the Genias general image analysis program (8,12) Cell suspensions were fixed in 4% formaldehyde a t extended to measure mean grey intensities of a n object room temperature for 24 h. Monolayers of cells were in there overlaying grey images. The camera was calprepared on polylysine-coated glass slides by cytospin ibrated with a Leitz 621 059 fluorescence standard. centrifugation (Hettich, Tuttlingen, Germany) a t Three grey images (512 x 512 pixels, 64 grey levels) 1,250g. After postfixation in formaldehyde, DNA stain- representing the cellular outline by transmitted light, ing by the Feulgen-Schiff technique (6) was applied, and the fluorochromes PI and FITC, were sequentially using hydrolysis a t room temperature (5 N HC1, 22”C, captured from the same areas. The three images were 60 min) and Schiff (pararosaniline) staining for 1 h. stored in the memory, and the image representing PI The cell monolayers were subsequently analyzed by was used to define the nuclei with a “cut sharpen” optransferring the microscope image from a Zeiss axiop- erator. Confluent nuclei were separated manually with lan microscope equipped with a x40 objective and a a light-pen by the investigator. This image was com546 n m green filter to a SEM-IPS image processing pared to the transmitted light image, and nuclei beunit (Kontron, Munich, Germany) using a Grundig FA longing to the same cell were grouped together. Cells 76 video camera. The nuclear boundaries were defined without cytoplasm, and aggregates of nuclei, were by the operator by establishing a threshold grey level omitted. The procedure allowed measurements of flufor each image, and pixels darker than this threshold orescence intensities and areas of nuclei within indivalue were identified as belonging to a nucleus. A bi- vidual cells. In each cytospin preparation, 500-800 nary image for nucleus identification was thus concells were measured. structed for each microscope image. Image folds were randomly selected in the monolayer, and all nuclei RESULTS AND DISCUSSION with a correctly defined boundary were measured. ViBrdUrd incorporation (DNA synthesis) and DNA sual identification in the grey level image and also in the microscope directly allowed the precise identifica- content in cells from a regenerating liver, labeled with tion of cell types and also classification of nuclei a s BrdUrd for 8 h (15-23 h after PHI, have been measured belonging to mononuclear or binuclear hepatocytes. by five different methods: 1)microscopical counting; 2) The optical density for each nuclear pixel was mea- flow cytometry using Hoechst dye quenching; 3 ) flow sured, and the integrated optical density calculated for cytometry using anti-BrdUrd-antibody and FITCevery whole nucleus. All images were corrected for the labeling; 4) image cytometry using Feulgen-Schiff optical density of the background, ensuring stable and staining, and 5) image cytometry using fluorescence reproducible measurements. Approximately 300 nuclei staining. Two different labeling principles have been were measured for each monolayer. BrdUrd-labeled applied. Method 2 is based on the ability of DNA-innuclei were colored black or brown with peroxidase/ corporated BrdUrd to reduce the fluorescence yield DAB using cobalt and nickel as intensifiers, and could from the DNA staining dye Hoechst 33258 (16,171.The easily be separated from BrdUrd-negative nuclei which other methods utilize a monoclonal antibody which became pink after Feulgen staining. Two sets of mea- binds to BrdUrd moieties in single-stranded DNA (7,9). surements were done: Firstly, nuclear DNA content Microscopical Counting was measured with Feulgen before BrdUrd staining. Secondly, nuclear DNA content was measured with Hepatocytes which have incorporated BrdUrd in Feulgen after BrdUrd staining, but only in BrdUrd- their DNA are clearly recognized in the microscope afnegative cells (BrdUrd-positive nuclei could not be ter staining with DABiFeulgen (Fig. 1A) or FITC (Fig. measured with Feulgen staining). In each ploidy/nucle- 1B). By using a weak counterstain the percentage of arity class the number of BrdUrd-negative cells was BrdUrd-positive hepatocytes (labeling index) could subtracted from the total number of cells to obtain the easily be counted. For hepatocytes labeled with BrdUrd number of BrdUrd-positive cells. from 15 to 23 h after P H the overall labeling index was of nuclei reflected nuclear size and was proportional to DNA content.

NETHODS FOR STUDY O F LIVEK CEIdI, PROLIFERATION

407

FIG. 1. Light micrographs of DAB- and Feulgen-stained (A) or FITC-stained hepatocytes (B) from regenerating liver labeled with BrdUrd 15-23 h after PH. BrdUrd-positive mononuclear and binuclear cells are easily recognized. A, x 290; B, X 460.

Table 1 Proliferation and Binucleation of Hepatocytes From Regenerating and Normal Rat Live?

Regenerating liver Nnrmal liver

Hepatocellular BrdUrd labeling index (70) Mononuclear Binuclear All henatocvtes 46.3 i 2.1 (3) 47.2 3.5 (3) 45.1 1.5 (3) 1.3 -t 0.1 i 5 i 1.0 2 0.0 (5) 1.2 i 0.1 (5)

*

*

Binucleation index (96) 2.0 ? 0.4 (3) 10.0 i 1.5 (3)

aYoung male rats (150 g) were labeled with BrdUrd (S.C.tablets) in vivo 15-23 h after a % partial hepatectomy (regenerating liver) or for an equivalent length of time in control rats (normal liver). Hepatocytes were isolated by collagenase perfusion immediately after the BrdUrd pulse for measurement of labeling index, or after a 12 h (normal liver) or 24 h (regenerating liver) chase for measurement of binucleation index. BrdUrd-labeled cells were identified by DAB staining and counted in the microscope (500 cells counted from each suspension index). Each value is the mean & S.E. of the number of rats given in parentheses.

4695, with similar values for mononuclear and binuclear cells (Table 1).After a chase sufficiently long to allow labeled cells to divide or binucleate (routinely 12 h), a n estimate of the binucleation rate could be obtained by measuring the percentage of binuclear cells among those which were BrdUrd-positive (binucleation index). The binucleation index for the cells labeled 1523 h after PH was found to be 2% (Table 1)as measured 24 h later; this very low index value is typical of regenerative growth (24). In comparison, proliferating hepatocytes from normal (non-regenerating) livers of age-matched rats (with labeling indices around 1%) displayed a binucleation index of 10% (Table 1).Microscopical counting can thus provide precise values for overall growth rates as well as some information about growth patterns (labeling indices of mononuclear vs. binuclear cells; binucleation indices), although the method cannot relate these measurements to cellular ploidy (DNA content).

Flow Cytometry Using Hoechst Dye Quenching Figure 2A shows a flow cytometric diagram of Hoechst-stained hepatocyte nuclei from the regenerating liver of a r a t not exposed to BrdUrd (control animal). Three main peaks (2N, 4N, and 8N) were clearly separated. Many hepatocytes were synthesizing DNA (S-phase) as indicated by the elevated fluorescence baseline between the major nuclear DNA peaks. Figure 2B shows a corresponding diagram of hepatocyte nuclei from a r a t exposed to BrdUrd for 8 h during the regeneration period. All BrdUrd-labeled nuclei subsequently exhibited quenched fluorescence and thus shifted their position to lower apparent DNA values. This caused the appearance of a new peak in the hypotetraploid (hyperdiploid) position, corresponding to labelled tetraploid nuclei (of diploid origin, now in late S and G,) with quenched fluorescence. Another peak appeared in the hypo-octoploid (hypertetraploid)

GEBLYNG ET AL.

nuclei displayed quenched fluorescence would indicate that they represented a nearly pure G, population, i.e., there would seem to be practically no octoploid G, nuclei present in livers from rats of this age. There was also a small peak in the hypodiploid position, probably representing diploid nuclei which had been exposed to BrdUrd in late S-phase and which had divided during the 8 h pulse period. LN Figure 2C shows a flow cytometry diagram of nuclei Hoechst from the same preparation as in Figure 2B but stained with ethidium bromide (EB) and excited at 488 nm. In 8N contrast to the UV-excited Hoechst fluorescence, the EB fluorescence was not affected by the incorporation of BrdUrd into DNA, and therefore accurately reBrdU pulse flected the DNA content. This was confirmed in a n 115-23h) experiment where fluorescence from the two dyes was measured both with UV excitation and 488 nm excitation, using dual wavelength excitation flow cytometry (not shown). If BrdUrd-labeled cells are allowed time to complete mitosis, the nuclear DNA content will shift from the high SIG, value to the low G, value. With Hoechst staining, the BrdUrd-labeled, quenched nuclei will assume positions below those of unlabeled nuclei, i.e., in Hoechst the hypodiploid, hypotetraploid, and hypo-octoploid reB gions of the DNA histogram (Fig. 3). d The Hoechst quenching method gives excellent sepBrdU pulse aration between BrdUrd-positive and BrdUrd-negative (15-23 h ) nuclei and also between the different ploidy classes. The procedure is quick and easy to perform since the fluorochromes are mixed directly in the nuclear suspension after preparation. There are no centrifugation or filtration steps involved; the possibility of selective cell loss is therefore small. The method has been reported to require high amounts of incorporated BrdUrd (141,but the sensitivity can apparently be increased by applying bivariate analysis (4). In our experience labeling of 150 g animals with 50 mg BrdUrd (s.c. tablet) for 3 h or more provides sufficient sensitivity for detection of BrdUrd incorporation. The limitation of the Hoechst method as applied to FLUORESCENCE INTENSITY liver cells lies is its inability to distinguish between BrdUrd-positive and BrdUrd-negative whole hepatoFIG.2 . Quenching of Hoechst dye binding by DNA-incorporated cytes. This is mainly due to the extremely high backBrdUrd. The figure shows flow cytometric DNA distribution histoground fluorescence from hepatocytic cytoplasm. Algrams of hepatocyte nuclei isolated from regenerating rat liver 23 h after PH. ZN, 4N, and 8N denote nuclei with diploid, tetraploid, and though the background fluorescence could be strongly octoploid DNA content, respectively. A Hoechst staining, control an- reduced by acidlpepsin treatment (see below), such imals (no BrdUrd exposure). Many hepatocytes are in S-phase as indicated by the fact that the fluorescence between the DNA peaks is treatment reduced the Hoechst fluorescence yield (probably due to formation of single-stranded DNA) by elevated well above the baseline. B: Hoechst staining, animals exposed to BrdUrd 8 h before death. DNA-synthesizing nuclei are evi75% and gave histograms of poor quality. Crossdent a s one quenched tetraploid peak (in the hyperdiploid position) transfer of energy from excited Hoechst to EB or PI and one quenched octoploid peak (in the hypertetraploid position) under these conditions furthermore compromised mulrepresenting labeled nuclei of tetraploid and octoploid origin, respectively. C: BrdUrd-labeled nuclei stained with ethidium bromide, tiparameter analyses. Accordingly, the cytoplasm had to be removed, e.g., by using Vindelov’s method a s dewhich accurately reflects DNA content (no quenching). scribed above. With other, mononuclear cell types this will not cause any problem, but with hepatocytes analposition, corresponding to labelled octoploid nuclei (of ysis of nuclei does not give a representative picture of tetraploid origin, now in late S and G,) with quenched cellular events since 20-40% of all hepatocytes are bifluorescence. The fact that virtually all of the octoploid nucleated, belonging either to the tetraploid (two dip2N

I

Control ( 2 3 h a f t e r PH 1

L

409

METHODS FOR STUDY OF LIVER CELL PROLIFERATION

HEPATOCELLULAR CARCl NOMA 2N

I

4N

FLUORESCENCE I N T E N S I T Y FI G . 4. DNA histogram (ploidy distribution) of Hoechst-stained, BrdUrd-labeled nuclei from hepatocellular rat carcinoma. The tumor, induced by treatment with 2-acetylaminofluorene (2-AAF)1-5 weeks after PH, was harvested by collagenase perfusion (23) 19 months after start of the carcinogenic treatment. The animal was given a 60 h RrdUrd pulse and a 12 h chase prior to death. Tumor cell nuclei were prepared by the method of VindeLv (27). BrdUrd has been incorporated mainly in the DNA diploid nuclei, as indicated by the prominent subdiploid, quenched peak.

____)

Nuclear ploidy FIG.3. Schematic relationship between nuclear and cellular ploidy of Hoechst-stained, BrdUrd-labeled hepatocyte nuclei (pulse-chase experiment). The upper panel shows the DNA histogram (ploidy distribution) of nuclei from a regenerating liver pulse-labeled with BrdUrd 15-23 h after PH. A given nuclear ploidy class contains nuclei derived both from mononuclear cells of the same ploidy and from binuclear cells with twice as high cellular ploidy. B indicates BrdUrd-labeled nuclei with quenched Hoechst fluorescence. +

loid nuclei) or octoploid (two tetraploid nuclei) cell class (2,3,18,24). The problem is outlined in Figure 3 which indicates that nuclei of a given ploidy may derive either from mononuclear cells of the same ploidy or from binuclear cells belonging to a higher cellular ploidy class. Another potential problem is that in contrast to whole cells, hepatocytic and nonparenchymal nuclei cannot be distinguished by flow cytometric analysis. One application where flow cytometry based on Hoechst quenching may be the preferable alternative is in the study of hepatocellular tumors. Such tumors have been found to be predominantly DNA diploid in most cases; usually with less than 10% binuclear cells in which case the difference between nuclear and cellular ploidy becomes insignificant (23). Figure 4 shows a flow cytometric diagram of nuclei from a hepatocellular carcinoma, demonstrating a prominent quenched peak in the hypodiploid region after a BrdUrd pulse-

chase regimen. Proliferation would therefore seem to occur preferentially in the DNA diploid cell subpopulation.

Flow Cytometry Using FITC-Labeled Anti-BrdUrd-Antibody The method for BrdUrd staining referred to in the methods section proved to be particularly suitable for hepatocytes. The simultaneous denaturation of DNA by acid and proteolytic digestion of cytoplasm by the enzyme pepsin dramatically reduced the background fluorescence from hepatocyte cytoplasm (autofluorescence as well as nonspecific antibody binding) as assessed by microscopy (Fig. 1B). This treatment also diminished the problem of cell aggregation. When 1,000 cells from each of 10 different animals were counted microscopically, the percentage of aggregated cells was found to be 5.7 5 0.6 (mean S.E.M.) in the control cell suspensions but only 1.6 ? 0.4 in the pepsin-treated suspensions. Compared with Hoechst quenching, the use of FITClabeled anti-BrdUrd antibody offers several advantages. Firstly, because the latter method is much more sensitive, BrdUrd pulses of short duration (

Analytical methods for the study of liver cell proliferation.

Various cytometric methods for analysis of regenerating rat liver growth (DNA ploidy distributions, binucleation, and DNA synthesis by in vivo BrdUrd ...
1MB Sizes 0 Downloads 0 Views