EXPERIMENTAL AND MOLECULAR PATHOLOQY 16, 1-15 (1972)

An Ultrastructural Study of Primary Cultures of Adult

Human Liver Tissue

A. GUILLOUZO, P. OUDEA, Y. LE GULLY, M. C. OUDEA, P. LENOIR AND M. BOUREL

Unit& de Recherche HJpatologique U 49 de l’I.N.S.E.R.M., H6pital de Pontchaillou,

Rennes, France and Laboratoire de Recherche H&patologique et Chaire de Clinique Me’dicale B, Facultk de MJdecine, Nantes, France

Received May 3, 1971

Human liver fragments obtained by needle biopsy or surgically were cultured in plastic flasks, using Eagle’s medium, human serum, and 95% Or5% CO* gas mixture.

Electron microscope examination of primary cultures on Day 12 of cultivation showed the presence round certain explants of cells morphologically identical to in vivo hepatocytes. By the same date other cells in the same explants had fewer organelles, cytoplasmic filaments, and sometimes lipid overload. By Day 20 all cells had simpler ultrastructure. Intermediate morphologic patterns suggest that dedifferentiation rather than progressive cell selection may be responsible for such changes.

Other explants produced an outgrowth of poorly differentiated cells the origin of which is difficult to determine.

Several attempts to culture liver cells have yielded cells of various morphology which bore some resemblance to the different types of adult liver cells.

A specific type of epithelial cell was noted in cultures from adult human liver by Iber et al. (1965) and by Taylor et al. (1969). Two cell types-granular and clear-reported by Bourel et al. (1968) and by Le Guilly et al. (1970) in adult human liver cultures are also encountered in most cultures of embryonic or new- born liver (Alexander and Grisham, 1970; Hillis and Bang, 1962; SandstrGm, 1965 and 1966; Watanabe, 1966).

The aim of the study presented here was to define the ultrastructural character- istics of epithelial cells cultured from adult human liver and thus to establish standards of reference for the utilization of such cultures in experimental physio- pathology and pharmacodynamics.

MATERIALS AND METHODS

Liver fragments, obtained either by needle biopsy using a Vim-Silverman needle, or surgically, from 14 patients suffering from various diseases (Table I) were cultured by a technique elaborated at this research center (Bourel et al., 1968; LeGuilly et al., 1970). In two of the patients the hepatic histology was normal, in five it showed the changes of mild, nonspecific reactive hepatitis. Ac- cording to the type of biopsy specimen, 5-200 explants were distributed among l-

1 Copyright Q 1972 by Academic Press, Inc.

2 GUILLOUZO ET AZ,.

15 Falcon polystyrene bottles. The living cultures were observed by phase-con- trast microscopy. Some cultures were stained by the May-Griinwald-Giemsa technique, others with the periodic acid-Schiff reagent after fixation in Bouin’s fluid, others again with Sudan red III Bx after fixation in Baker’s form01 calcium.

Cultures for the ultrastructural study, which was conducted from Day 12 to Day 30 postexplantation, were washed with phosphate-buffered saline and fixed in situ in a 1% osmium tetroxide solution buffered to pH 7.2 with acetate Verona1 and made isotonic by addition of 0.045 g of sucrose per milliliter of fixative. Dehydration was effected by passing the sample through ethanol. Em- bedding was carried out as follows. A thin film of Epon 812 was poured onto the bottom of a polystyrene bottle, the top of which had been cut out. After incuba- tion for 3 hours at 37”C, gelatin capsules were placed over the areas selected for electron microscope examination. The whole was incubated for 24 hours at 60°C then cooled for 10 minutes at -30°C. When the Epon film was detached with a scalpel blade it brought with it the gelatin capsules with the cells at the bottom. The capsules were then filled with Epon and returned to a temperature of 60°C for 2448 hours. A layer of cells at the surface of the blocks, parallel to the plane of frontal section, was thus available. Lateral sections in a plane at right angles to the preceding were prepared from parallelepipeds of Epon, 34 mm thick, obtained by adding Epon to both surfaces of the film containing the cells. Sections cut on an LKB Ultrotome using glass knives were doubly stained with uranyl acetate and lead citrate after Reynolds (1963). The electron microscope used was the Hitachi HU-llE-1F model using an accelerating voltage of 75 KV.

RESULTS

Phase-contrast microscopy

From the seventh culture day onward certain explants, their contours still uneven, became surrounded by closely packed polygonal cells having many gran- ules and sometimes refractive vacuoles in their cytoplasm. These cells had round nuclei with one or two nucleoli, many were binucleate, and they appeared to spread out and enlarge their diameters as they got nearer to the periphery of the culture. Absence of a clear-cut demarcation between this ring of granular cells and the explant suggested that at the perimeter of the latter the cells might be arranged in several layers (Fig. 1). Maximum outgrowth was reached during the third week. After a month, those near the explant still preserved their organiza- tion and their granularity but toward the periphery of the outgrowth the cell layer became progressively thinner and the granules, reduced in number, were congregated around the nuclei.

Other explants assumed smooth rounded contours toward the fifth culture day. A single-cell layer of mono or binucleate epithelial cells, with clear cytoplasm and a larger diameter than the granular cells just described, could then be seen spreading over the explant surface (Fig. 2). A month after explantation these clear cells were fundamentally unchanged in size and appearance.

Sometimes the two cell types, granular and clear, were present together in cul- tures from the same subject, but in this event one or the other type always pre- dominated (Table I). On the other hand, as a general rule, each explant sent out

PRIMARY CULTURES OF HUMAN LIVER TISSUE! 3

FIQ. 1. Granular epithelial cells; normal liver, Day 16 of culture. X470.

one type of epithelial cell only. Certain sheets of epithelial cells had a lin fibroblasts interspersed with occasional macrophages. Some explants wer rounded exclusively by mesenchymal cells; they were excluded from the structural study.

ling of e sur- ultra-

Light microscopy

May-Griinwald-Giemsa staining confirmed the existence of two contr *asting epithelial cells, granular and clear, but yielded no additional informatior 1. The cytoplasm of the granular cells was more intensely periodic acid-Schiff-pc ositive

4 GUILLOUZO ET AL.

FIG. 2. Clear epithelial cells; normal liver, Day 13 of culture. x470.

than that of the clear cells, and in neither was the intensity substantially less after diastase pretreatment. The lipid nature of the refractive vacuoles which phase-contrast microscopy had revealed in many of the granular cells was demon- strated by staining with Sudan red III Bx.

Electron microscopy

Explants with granular epithelial cells. By Day 12 of culture the center of the explant was the site of necrotic foci, aggregations of collagen fibers, and lipid droplets; biliary and vascular structures alone had preserved their morphological

PRIMARY CULTURES OF HUMAN LIVER TISSUE 5

TABLE I

SUMMARY OF PATIENTS FROM WHOM LIVER FRAGMENTS WERE TAKEN FOR TISSUE

CULTURE

Age of patient (wars)

Clinical diagnosis Liver histology

30 59

67

71 6

27

54

56 16

17 22

56

40 52

Pancreatic cyst

Lymphosarcoma

Cholecystitis with cho- lelithiasis

Duodenal ulcer

Atresia of biliary pass-

ages Ruptured intracranial

aneurysm

Cholecystitis with chol- elithiasis

Acute pancreatitis Alcoholic hepatitis

Cholelithiasis Cholecystitis with cho-

lelithiasis Myeloid splenomegaly

Hemochromatosis Alcoholism

Predominant cell type”

M.N. S.R.H.”

M.N. S.R.H.; slight infiltration of portal connective tissue

M.N. S.R.H.

Steatosis Portal sclerosis with dilatation of bile

ducts Normal

M.N. S.R.H.; foci of clarification; inflam- mation of portal spaces

Fibrosis; steatosis; nonspecific hepat,itis Annular cirrhosis ; steatosis ; inflamma-

tory infiltration

M.N. S.R.H. Normal

Myeloid metaplasia

Portal and periportal sclerosis; mild steatosis

a G = granular epit,helial; C = clear epithelial.

b M.N. S.R.H. = mild, nonspecific react,ive hepatitis.

G G

G

G G

G

C

C C

C

C

C

C C

integrity. At the periphery of the explant disintegrating cells lay side by side with apparently healthy cells similar to those in the outgrowth.

The outgrowth at this stage contained cells structurally identical to in viva hepatocytes (Figs. 3 and 4) ; they had round or oval nuclei, cytoplasm rich in organelles, well-developed ergastoplasm occurring as parallel flattened cisternae, several Golgi complexes, plentiful free ribosomes, small cytolysomes, pigment granules resembling peribiliary dense bodies, and moderately osmiophilic bodies reminiscent of peroxisomes; in general, they had little glycogen. Their mito- chondria were of various shapes-some round, some elongated, some bent-and their cristae were distributed irregularly like the cristae of hepatocytic mito- chondria. Intercellular contact was mainly rectilinear, interdigitation being un- usual. The cell membranes were linked by desmosomes but diverged from each other in places to form microvilli-filled intercellular spaces comparable to bile canaliculi.

At the same culture age (12 days) other cells differed from normal hepato- cytes. Their cytoplasm was poor in organelles, particularly in endoplasmic re- ticulum; often it contained compact bundles of filaments 60-80 L% in diameter (Fig. 5). Interdigit.ation between adjacent cells was becoming substituted for

6 GUILLOUZO ET AL.

FIG. 3. Granular epithelial cells; normal liver, Day 12 of culture; Lateral view. X8000. The organelles seen resemble those of hepatocytes in viva. m = mitochondrion. + = ergastoplasm. CB. = bile canaliculus. Note microvilli and thick cytoplasmic finger-like processes projecting from the free surface in contact with the culture medium.

rectilinear apposition (Figs. 5 and 8). These changes were more pronounced in the superficial cell strata directly in contact with the medium than in the deeper strata, and at the periphery of the outgrowth than in the explant. Some cells con- tained lipid droplets surrounded by abundant glycogen particles (Fig. 7) and

PRIMARY CULTURES OF HUMAN LIVER TISSUE 7

FIG. 4. Granular epithelial cells; normal liver, Day 12 of culture; lateral view. X18,500. Not.e heterogeneous dense body (black star) resembling a lipofuscin granule, a cytolysome (double nrrows), numerous Golgi complexes (+).

8 GUILLOUZO ETAL.

FIG. 5. Granular epithelial cells, periphery of culture; normal liver, Day 12 of culture;

lateral view. X28,350. The cells in contact with the medium are flat, poor in endoplasmic re- ticulum, and rich in fibrils (F). Note frequent interdigitation (+). Some desmosomes (D) are still in evidence. Inset : a desmosome X51,000.

PRIMARY CULTURES OF HUMAN LIVER TISSUE 9

FIG. 6. Granular epithelid cells; normal liver, Day 15 of culture; frontal view. X10,500. Note numerous cytolysomes-vacuolated (V) and condensed (+ ).

10 GUILLOUZO ET AL.

FIG. 7. Granular epithelial cells; liver of patient with atresia of biliary passages, Day 13

of culture; frontal view. X5800. The cells contain numerous lipid globules (+) (many of them more osmiophilic centrally than peripherally) surrounded by glycogen particles. Inset: detail of glycogen particles. X24,800.

sometimes cytolysomes and vacuoles (Fig. 6), findings which did not correlate with the histology of the fragment from which the culture derived.

That the granular cells, irrespective of their appearance, developed in severa layers, at least in the vicinity of the explant, was evident in lateral view sec- tions. Toward the periphery of the outgrowth the layers became fewer and their constituent cells thinner. One or more strata of elongated flattened cells, some with fibroblastic features, generally lay under the granular cells, an architectural pattern which has already been observed in animals (Alexander and Grisham, 1970; Watanabe, 1966). We saw no basement membrane between this cell stratum and the polystyrene surface.

PRIMARY CULTURES OF HUMAN LIVER TISSUE 11

FIG. 8. Granular epithelial cells, periphery of culture; normal liver, Day 12, lateral view. The smallness of the mitochondria, the close interdigitation of the cell membranes, the scarcity of rndoplasmic reticulum, suggest an analogy between these cells and bile duct epithelium; however. they are not bounded by a basement membrane. Note the extreme rarity of or- ganelles in the superficial cell layer. ~10,400.

During the 2 subsequent weeks necrosis of the center of the explant became total; the peripheral zone alone survived, but could not be distinguished from the proliferating granular cell layers. The changes noted in certain granular cells on Day 12-scanty organelles, cytoplasmic filaments, membrane interdigitation- had progressed and by now affected all the cells of the culture, being especially marked in those situated superficially and peripherally. No cell was similar to in viva hepatocytes. Lateral view sections showed the same arrangement as on Day 12.

Explants with clear epithelial cells. On Day 12 the center of the explant was completely necrotic and its surface comprised only two or three layers of morpho-

12 GUILLOUZO ETAL.

FIG. 9. Clear epithelial cells; normal liver, Day 12 of culture; lateral view. The cells linked by a desmosome contain small mitochondria and numerous fibrils. The surface contiguous with the culture medium (M) has few microvilli. The cells are not subtended by a basement membrane (+-I. ~20,000.

PRIMARY CULTURES OF HUMAN LIVER TISSUE 13

10&ally intact cells. The cells propagating on the polystyrene around the explant had an abundant clear hyaloplasm, with few organelles ; their mitochondria were generally small and elongated, and fascicles of cytoplasmic filaments were al- ready apparent. Lateral views showed these clear cells rapidly forming into a unicellular layer as they grew out from the explant; their borders dovetailed, and intercellular linkage was by desmosomes (Fig. 9). No underlying basement membrane was seen. As culture age advanced the organelles became still fewer and the cytoplasmic filaments more numerous. The interlocking cell edges and the desmosome-like intercellular linkage persisted.

DISCUSSION

Primary cultures of human liver yield two types of epithelial cells: granular and clear. Each explant sends out one type only of epithelial cell, the other being suppressed, seemingly through a process of cell selection.

This study has demonstrated that, on Day 12 of adult human liver cultures, some granular cells growing out from certain explants are ultrastructurally identi- cal to in viwo hepatocytes; the chief points in common are the general appearance of the mitochondria and of the endoplasmic reticulum, the probable presence of peroxisomes, and the presence of lipofuscin granule-like peribiliary dense bodies and of bile canaliculi. Granularity seems to correlate with the large number of mitochondria and dense bodies. Other investigators have made similar observations in liver cultures from chick embryo (Laschi and Rizzoli, 1968a; Westman and Sandstrom, 1966) and from newborn mice and rats (Alexander and Grisham, 1970; Rose et al., 1968 ; Watanabe, 1966).

Likewise on Day 12 other “granular” cells of the same explants mainly at the free surface and in the peripheral areas of the outgrowth have few ergastoplasmic cisternae, no peroxisome, and a large and clear hyaloplasm. These cells are less granular when observed with the phase microscope. These changes progress during the subsequent weeks.

They might. be due to the replacement of the granular cells derived from the hepatocytes by cells from different origin. They might also be due to the simpli- fication of structure customarily termed dedifferentiation. This latter mechanism is strongly suggested by the presence of many cells with intermediate ultra- structural features. Such gradual transformation has also been observed in culture of rat liver (Watanabe, 1966). The literature is not rich in data on pro- gressive loss of specific hepatocyte functions (Alexander and Grisham, 1970; Hillis and Bang, 1962; Rose, et al., 1968; Sandstrom, 1966) and it is too early to attempt correlation with the ultrastructural findings.

Some “simplified” cells bore resemblance to bile duct epithelium although they have no basement membrane and although more than three cells are seldom seen round a space bordered with microvilli. A derivative in vivo relationship between hepatocytes and bile duct epithelium has been postulated (DuBois, 1963) but our researches have neither proved nor disproved it.

Granular cells identical to adult hepatocytes had no ultrastructural lesion re- sembling those which may be observed in liver disease. Other cells which had lost these distinctive histological features contained lipid droplets or cytolysomes,

14 GUILLOUZO ET AL.

which did not correlate with the histological findings in viva. Lipid droplets have been reported by other workers (Laschi and Rizzoli, 1968b; Rose et al., 1968; Watanabe, 1966) in primary cultures from animal livers. One does not know whether they arise from synthesis or lipophanerosis. Cytolysomes might reveal injury of aging cells or transformation into a simpler cell type.

Some explants give way only to ‘Lclear cells.” Their ultrastructural characteris- tics do not facilitate identification of their origin. Their initial image-relatively thick cells with interlocking cell membranes and numerous desmosomes-is more like that of bile duct epithelium than of endothelial cells or even dedifferentiated hepatocytes. Their subsequent spread and further loss of organelles increase the uncertainty. We did not see, between the cells and the polystyrene surface, the basement membrane described by Alexander and Grisham (1970) in their cul- tures on a collagen substrate. Perhaps collagen changes the conditions of detach- ment, or perhaps it provokes the secretion of a basement membrane; yet no base- ment membrane is described by Watanabe (1966) who also used a collagen substrate. Determination of the origin of the clear cells calls for further study aimed at observing exactly how they proliferate inside the explant during the first days of culture.

The large fibril bundles seen in both granular and clear cells, particularly after Day 20 of culture, correspond to the cytoplasmic filaments reported by Biberfeld et al. (1965) in an in vitro-propagated cell line of rat liver origin, and by Rose et al. (1968) in newborn mouse liver explants cultivated on mica coverslips. In our cultures on polystyrene these filaments were strikingly well developed, in some cases traversing the entire cell. It is probable that their role is a mechanical one, and that they play a part in cell stabilization and/or cell movement.

ACKNOWLEDGMENTS

The authors thank Miss M. Boisnard and Miss Strullu for their efficient technical help.

REFERENCES

ALEXANDER, R. W., and GRISHAM, J. W. (1970). Explant culture of rat liver. I. Method, morphology, and cytogenesis. Lab. Invest. 22,50-62.

BIBERFELD, P., ERICSSON, J. L. E., PERLMANN, P., and RAFTELL, M. (1965). Increased occur- rence of cytoplasmic filaments in in vitro propagated rat liver epithelial cells. Exp. Cell Res. 39,301305.

BOUREL, M., LE GUILLY, Y., LENOIR, P., FERRAND, B., and FEBVRE, H. (1968). Culture de foie humain adulte. C.R. Sot. Biol. 162, 979-983.

Dn BOIS, A. M. (1963). “The Liver” (C. Rouiller, ed.), Vol. 1, pp. l-39. Academic Press, New York.

HILLIS, W. D., and BANG, F. B. (1962). The cultivation of human embryonic liver cells. Exp. Cell Res. 26,9-36.

IBER. F., BANG, F. B., and WARWICK, A. (1965). “Cells and Tissues in Culture” (E. N. Willmer, ed.), Unpublished data cited in Vol. 2, p. 624. Academic Press, London.

LASCHI, R., and RIZZOLI, C. (1968a). Etude ultrastructurale de cultures primaires de foie embryonnaire. Mise au point des techniques et premieres observations. J. Microsc. Paris 7, 533-538.

LASCHI. R., and RIZZOLI, C. (1968b). Etude au microscope electronique de cultures primaires de foie embryonnaire. Fourth European Regional Conference on Electron Microscopy, Rome, vol. II, 481-482.

PRIMARY CULTURES OF HUMAN LIVER TISSUE 15

LE GUILLY, Y., LENOIR, P., BOUREL, M., POUPON, R., and GUILLOUZO, A. (1970). La culture prolong&e du tissu hbpatique humain ad&e. Pathol. Biol. 18, 733-741.

REYNOLDS, E. S. (1963). The use of lead citrate at high pH as an electron opaque stain in electron microscopy. J. Cell Biol. 17,208-212.

ROSE, G. G., KUMEGAWA, M., and CATTONI, M. (1968). The circumfusion system for multi- purpose culture chambers. II. The protracted maintenance of differentiation of foetal and newborn mouse liver in vitro. J. Cell BioZ. 39,430-450.

SANDSTR~M, B. (1965). Studies on cells from liver tissue cultivated in vitro. I. Influence of the culture method on cell morphology and growth pattern. Exp. Cell Res. 37, 552-568.

SANDSTR~M, B. (1966). Liver parenchymal cells in tissue culture. A morphological study on foetal rat and chicken liver cells. Acta Sot. Med. Upsal. 71,21-34.

SANDSTR~M, B. (1966). Cytochemistry of foetal chicken liver cultivated in vitro. Acta Sot. Med. Vpsal. 71,65-82.

TAYLOR, P. E., ZUCKERMAN, A. J., and FARROW, L. J. (1969). Culture of needle biopsies of the liver from patients with suspected hepatitis. J. Clin. Pathol. 22, 701-703.

WATANABE, H. (1966). A fine structural study of liver culture. Ezp. Cell. Res. 42,685-699. WESTMAN, J., and SANDSTR~M, B. (1966). Electron microscopy of organic and cultivated

chicken embryonic liver. 2. Zellforsch. 71,271-282.