Free and Protein-conjugated Polyamines in Mouse Epidermal Cells EFFECT OF HIGH CALCIUM AND RETINOIC ACID*

We have investigated polyamine metabolism in primary cultures of mouse epidermal cells. These cells, which grow at low Caa* levels as a monolayer with characteristics of basal cells, terminally differentiate when the extracellular Ca2+ level is raised above 1 mM. The cellular levels of free polyamines were measured, and, after incubation of cell cultures with [‘Hlputres-cine, the distribution of label in both acid-soluble and acid-insoluble cellular components was examined. Free polyamine levels were reduced in cells induced to differentiate. Treatment with retinoic acid, which pre- vents differentiation and causes increased prolifera-tion, resulted in an increase in free putrescine. Upon adjustment of the calcium concentration to a level that induces differentiation, the enzyme transglutaminase was activated, and a concomitant increase in the level of both protein-bound mono- and bis-y-glutamyl deriv- atives of putrescine and spermidine was observed. Iso-lation of a material of apparent molecular weight about 6000 which contains only mono-y-glutamylpolyamines and the finding of both mono- and bis-y-glutamylpo- lyamines in the protein fraction containing cornified cell envelopes provided the basis for speculation on polyamines in envelope formation. Our data suggest that polyamines play a role during epidermal cell differentiation through transglutaminase-mediated post- translational from Nonidet extraction was suspended in 20 mM Tris-HC1 containing 1 mM phenylmethylsulfonyl fluoride, 10 mM dithiothreitol, and 2% SDS,’ boiled for 5 min, and centrifuged at 3,000 X g. This extract (SDS-soluble fraction) was also examined in the fast protein liquid chromatography system. The remaining insoluble pellet containing cross-linked protein of cornified envelopes was retained for digestion and analysis for 7-glutamylpolyamines. Protein was determined as described by Lowry et al. (35) using bovine serum albumin as standard. Radioactivity was measured in Hydrofluor counting fluid (National Diagnostic) with the use of a Beckman liquid scintillation spectrometer.

(1 Supported by a grant from the Washington Hospital Research Center. chondrocytes (6), L6 myoblasts (7), HL60 promyelocytes (8) and 3T3 fibroblasts (9). This evidence of polyamine dependence is strengthened by the finding that, in each case, addition of polyamines to the culture medium allows differentiation to proceed (5)(6)(7)(8)(9). There are other reports, however, that polyamine depletion stimulates differentiation in tumor cell lines, such as neuroblastoma (10) and melanoma (11), and more recent reports claim stimulation in various embryonal carcinoma lines (12)(13)(14). Despite these evident discrepancies, the findings suggest that polyamines are involved in cell differentiation, although no conclusions as to their specific role can be made.
In the search for the function of polyamines in cell differentiation, consideration should be taken of novel aspects of polyamine metabolism. Thus, it has been reported that polyamines are present in mammalian tissues and body fluids, not only per se as highly charged cations but also covalently conjugated to protein (15)(16)(17)(18). The posttranslational modification of proteins involving structural elements of polyamines occurs by two separate metabolic pathways. One results in the formation of hypusine (N"(4-amino-2-hydroxybu-ty1)lysine) (19). Here the butylamine residue derived from spermidine is transferred to the t-amino group of peptidebound lysine and subsequently hydroxylated (20). The other is the covalent attachment of polyamines through amide linkage to the y-carboxyl groups of glutamyl residues of protein. Formation of these protein components through exchange of the polyamine for ammonia at the carboxamide groups of glutaminyl residues is catalyzed by transglutaminases, Ca2+-dependent enzymes that are widely distributed in mammalian cells and in biological fluids (15). These enzymes, which catalyze the formation of t-(y-glutamy1)lysine crosslinks, play an important role in some extracellular reactions such as fibrin clot stabilization and seminal plug formation (for reviews, see Refs. 15 and 21). Although the specific function of the ubiquitous intracellular transglutaminase is unknown, it has been suggested that the enzyme plays a role in cell differentiation, since the differentiation of several cell lines in cultures is paralleled by increased enzymatic activity (22)(23)(24). In addition, higher transglutaminase enzymatic activity is seen in untransformed cells as compared with transformed cell lines (25) and in nonproliferating as compared with proliferating cells (26).
Primary cultures of normal mouse epidermal keratinocytes are an extensively studied model for mammalian cell differentiation (27). In this system at low CaZ+ (0.03 mM) the cells grow as a monolayer with characteristics of basal cells (281, at high Caz+ (1.2 mM) they terminally differentiate, leading to stratification, cornification, and eventual sloughing from the culture dish (28), and in the presence of retinoic acid they are prevented from differentiating (29) and proliferate at an increased rate (30). This system provides suitable features for investigations of polyamine metabolism during cell differentiation. Transglutaminase is generally accepted as an important marker for epidermal cell terminal differentiation (27, 31) and parallel increases of its activity with the onset of keratinization have been detected both in vivo and in vitro (31,32). In this work we present evidence for the role of polyamines as natural substrates for transglutaminase during the first stages of epidermal cell differentiation.
Methods-Epidermal cells were prepared from BALB/c mice as described (28), except using Dispase in place of trypsin. Cells (-7.5 X 106/10-cm culture dish) were plated and cultured for 60 h under low calcium conditions (0.03 mM Ca") in Eagle's minimal essential medium minus calcium, supplemented with 0.1% gentamicin (Gibco) and 8% Chelex 100-treated fetal calf serum (Gibco). After the initial 60-h culture, the cells were either maintained at low calcium conditions, placed in the same medium except under high calcium conditions (1.2 mM Ca"), or kept under low calcium conditions with addition of 5 PM retinoic acid. In each case incubation was continued for 48 h. In the cultures with retinoic acid, fresh medium containing this compound was administered after 24 h and cultures were kept from direct light.
For measurement of transglutaminase activity, cells that had been washed with large volumes of cold phosphate-buffered saline were scraped into an appropriate volume of 50 mM Tris-HC1 buffer, pH 8.3, containing 0.5 mM EDTA and rapidly frozen and thawed five times. After centrifuging at 105,000 X g for 30 min in a Beckman L5-50 centrifuge, the resulting supernatant (soluble fraction) was decanted and the pellet (particulate fraction) was resuspended in a small volume of 50 mM Tris-HC1 buffer, pH 8.3, containing 0.5 mM EDTA. These extractions were carried out at 0-4 'C. Transglutaminase activity was measured on unfractionated cell extracts, and on soluble and particulate fractions by incorporation of [3H]putrescine into N,N'-dimethylcasein essentially as described previously (34).
For determination of free polyamines, washed cells were scraped into phosphate-buffered saline and an equal volume of cold 20% trichloroacetic acid was added. After centrifuging for 15 min at 5,000 X g, the supernatant was collected and the pellet was washed twice with 5% trichloroacetic acid. The supernatant and washes were combined and polyamines were measured on aliquots by means of an ion exchange chromatographic procedure camed out on a Durrum D-400 amino acid analyzer equipped with a 4 X 80-mm column packed with Dionex DC 6A resin. A three-buffer program was employed as described previously (16).
For identification of products of polyamine metabolism, [3H]putrescine was added at the level of 5 pCi/ml after the initial 60-h culture and incubation was continued for the 48-h period under various conditions (see above). Aminoguanidine (20 p~) , an inhibitor of amine oxidases, was also included in the culture to minimize the oxidative loss of labeled putrescine from the medium. Washed cells were collected in a small volume of phosphate-buffered saline and an equal volume of cold 20% trichloroacetic acid was added. After centrifuging for 15 min at 5000 X g, the pellet (acid-insoluble fraction) was washed with 5% trichloroacetic acid containing 1 mM each of unlabeled putrescine, spermidine, and spermine until essentially no radioactivity was detected in the wash. The supernatant and washes were pooled (acid-soluble fraction). Determination of radiolabeled free polyamines and their y-glutamyl derivatives were performed on aliquots of the acid-soluble fraction and on enzymatic digests of the acid-insoluble fraction, respectively, using the automated ion exchange chromatography procedure as described previously (16).
A partial separation of the radiolabeled proteins in a bufferinsoluble fraction from cells treated with high calcium and labeled by 24-h exposure to [3H]putrescine in the presence of 20 p~ aminoguanidine was performed by means of exclusion chromatography on a Pharmacia-fast protein liquid chromatography system. Washed cells from 25 dishes were homogenized in cold 20 mM Tris-HC1 buffer, pH 6.0, containing 1 mM EDTA and 1 mM phenylmethylsulfonyl fluoride. The cell homogenate was centrifuged at 3,000 X g, and the resulting were applied to a Superose 6 HR 10/30 column (Pharmacia LKB Biotechnology Inc.) for exclusion chromatography. The insoluble pellet from Nonidet extraction was suspended in 20 mM Tris-HC1 containing 1 mM phenylmethylsulfonyl fluoride, 10 mM dithiothreitol, and 2% SDS,' boiled for 5 min, and centrifuged at 3,000 X g. This extract (SDS-soluble fraction) was also examined in the fast protein liquid chromatography system. The remaining insoluble pellet containing cross-linked protein of cornified envelopes was retained for digestion and analysis for 7-glutamylpolyamines. Protein was determined as described by Lowry et al. (35) using bovine serum albumin as standard. Radioactivity was measured in Hydrofluor counting fluid (National Diagnostic) with the use of a Beckman liquid scintillation spectrometer. (27, 36). The calcium-induced changes in cultured epidermal cells mimic some aspects of in vivo differentiation in epidermis (27). A distinctive aspect of epidermal cell differentiation is the establishment of the cornified envelope beneath the plasma membrane. This envelope is composed of cross-linked proteins (27), and its formation has been related to the progressive increase in particulate epidermal transglutaminase activity (36,37). During differentiation the transglutaminase catalyzes the production of isopeptide bonds between precursor proteins to form the envelope (38, 39). Cells growing in low calcium form very low levels of envelopes (27,40). However, when the Ca2+ level is increased there is a progressive increase of the epidermal transglutaminase activity (Table I)  Retinoic acid treatment also triggers an increase in transglutaminase activity (29), as is apparent in Table 1. However, investigation of the subcellular localization of this activity revealed that the major change is in activity of a soluble enzyme, probably that of so-called tissue transglutaminase which is present in all eukaryotic cells (41). Indeed, it has been reported that during culture of cells with retinoic acid there is some reduction in the activity of the particulate enzyme (42) and this appears to be confirmed by the data given in Table I. In Table I1 is shown the effect of calcium-induced epidermal cell differentiation on the cellular levels of free putrescine, spermidine, and spermine. Forty-eight hours after raising the calcium concentration in the medium there is a pronounced decrease in polyamines with respect to the level detected in cells grown in low calcium medium. When cells in low calcium are treated with retinoic acid, a large increase (&fold) in The abbreviation used is: SDS, sodium dodecyl sulfate.

TABLE I1
Free polyamines in muse epidermal cells in culture: effect of calcium and retinoic acid Cells were cultured for 48 h after the initial 60 h in low Ca2+ culture and polyamines were measured as described under "Experimental Procedures." Values are means of three determinations f S.D.  Table  I with [3H]putrescine added after the initial 60-h incubation as described under "Experimental Procedures." Polyamines were separated chromatographically (16). Values are means of three determinations with a S.D. less than 15%. Given in parentheses are the relative percentages of the total radioactivity for the individual culture conditions. intracellular putrescine is seen with little change in spermidine and spermine. Table I11 shows the distribution of [3H]putrescine and its radiolabeled derivatives in the acid-soluble fractions from these cells. It is evident from the data of Table I11 that when aminoguanidine is added together with radiolabeled putrescine at low Ca2+, a significant reduction occurs in labeled unknown compounds that elute near the front in the ion exchange chromatography. Less than 5% of the total radioactivity of low Ca2+-grown cells was found as spermine, as was also the case with cells treated with high Ca2+ and those grown with retinoic acid.
In differentiated epidermal cells somewhat more putrescine may be converted to spermidine than in the cells cultured in low Ca2+. However, retinoic acid treatment clearly leads to a greater uptake of putrescine.
In Table IV are presented data on the distribution of radioactivity in various components of the protein fraction (acid-insoluble fraction) from epidermal cells treated with [3H]putrescine as above, A fraction from each of the peaks of radioactivity that was separated by ion exchange chromatography from enzymatic digests of the cellular protein, was treated with 6 N HCl for 18 h at 110 "C. The identity and homogeneity of the labeled y-glutamylpolyamine was in each case verified by the finding of a stochiometric amount of the expected labeled polyamine in the acid hydrolysate (data not shown). Further verification was made with the use of the enzyme y-glutamylamine cyclotransferase, the action of which is specific for y-glutamylamines (43). Treatment of the y-glutamyl polyamines after their separation with this enzyme gave rise to the expected free polyamine in each case (data not shown). Hypusine, which is a radioactive polyamine TABLE IV Distribution of radioactivity from PHIputrescine in the protein fraction of mouse epidermal cells in culture: effect of calcium and retinoic acid Cells were cultured as outlined in Table  I with [3H]putrescine added after the initial 60-h incubation as described under "Experimental Procedures." Protein components were identified after exhaustive proteolytic digestion of the acid-insoluble fractions as described (16). The values given are the mean of three determinations with S.D. less than 10%. ND, not detected. *The N'-and N8-(y-glutamyl)spermidines were not separated by derivative found as a protein component in each group of cells, is unchanged by both acid hydrolysis and treatment with the cyclotransferase. This amino acid displays chromatographic properties very similar to those of the N-(y-glutamyl)spermidines, but is distinguishable on the basis of its resistance to these treatments (19). The data given in Table  IV show that the protein fraction from differentiated cells contains much larger amounts of radiolabeled y-glutamylpolyamines than that from cells cultured at the low Ca2+ level, whereas the protein fraction from retinoic acid-treated cells is practically devoid of these y-glutamylpolyamines. Significant amounts of free radiolabeled polyamines were also found associated with the protein fraction of each group of cells (data not shown) probably as a consequence of strong electrostatic interactions. Some raholabeled material was eluted near the front in ion exchange chromatography of each digest (6700, 6000, and 2800 cpm in that of the low Ca2+-, the high Ca2+-, and the retinoic acid-treated cells, respectively). This material was not identified.
The Tris buffer-insoluble portion of the protein from cells treated with high Ca2+ and [3H]putrescine was examined by exclusion chromatography using a fast protein liquid chromatography apparatus. The results are given in Fig. 1. Most of the radioactivity in the Nonidet-soluble fraction of these proteins was found in a single peak eluting at a position corresponding to a molecular weight of about 6000 (Fig. L4).
Radioactivity in peaks corresponding in positions to a wide range of molecular weights, up to 2.5 X lo6, was found upon examination of the SDS-soluble fraction (Fig. 1B). The identities of the radioactive components of the major Nonidetsoluble material were determined by ion exchange chromatography following proteolytic digestion. These results (Table V) show that this low molecular weight material contains covalently bound putrescine and spermidine in the form of monoy-glutamyl derivatives only. Also shown in Table V is the distribution of covalently bound polyamines in the fraction of differentiated cells that contains cornified envelopes (SDSinsoluble fraction). Clearly this fraction is seen to contain putrescine and spermidine in the form of both mono-and bisy-glutamyl derivatives.

DISCUSSION
In this study we have investigated the metabolism of polyamines in calcium-induced mouse epidermal cell differentiation. The findings strongly support a conclusion that putres-

TABLE V Radioactivity in the Nonidet-soluble low molecular weight material and in the SDS-insoluble fraction from differentiated epidermal cells
The fractions of the Nonidet-soluble low molecular weight material (Fig. lA) were pooled, made 1% in bovine serum albumin and 10% in trichloroacetic acid, and held at 4 "C overnight. The precipitate was dissolved in 0.2 M morpholine acetate buffer, pH 8.0, extracted two times with ether, and subjected to proteolytic digestion as described (16). ND, not detected. cine and spermidine act as physiological substrates for transglutaminase as evidenced by their attachment through yglutamyl linkage to cellular protein. Whereas the growing cells (those at low Ca2+ both with and without retinoic acid) appear to incorporate radioactivity covalently from polyamines into their protein fraction primarily in the form of the amino acid hypusine, y-glutamyl-bound polyamines are by far the main radioactive components in protein of differentiating cells (Table IV). Our observation of a reduced level of hypusine in cells at the high Ca2+ level is in accordance with the reported reduction in hypusine biosynthesis in differentiating Friend erythroleukemia cells (44). It is also as would be expected if, in these cells, hypusine is indeed an entity that is essential for the biological activity of an eukaryotic protein synthesis initiation factor, specifically factor 4D, of which it is a component (45). Treatment of mouse keratinocytes at low Ca2+ with retinoic acid prevents their terminal differentiation (cornified envelope formation) (29) and leads to an increase in their proliferative rate (30). This is accompanied by an increase in putrescine level in the cells and a higher uptake of radiolabeled putrescine (Table 111), which is characteristic of actively dividing cells (1-4). The increase in transglutaminase activity seen upon retinoic acid inhibition of terminal differentiation (Table I) is well documented (41,46). It is interesting that in these retinoic acid-treated cells in which there are ( i ) increased amounts of polyamines, ( i i ) high levels of polyamine radiolabeling, and (iii) enhanced transglutaminase activity, few if any polyamine-protein conjugates are seen (Table IV). It may be significant that in Chinese hamster ovary cells, after polyamine labeling by culturing in the presence of [3H] putrescine, there are essentially no y-glutamylpolyamines found in the protein fraction (47). Even after inducing a significant increase in transglutaminase activity by treatment with sodium butyrate and after raising the specific radioactivity of the polyamines, only barely detectable levels of proteinpolyamine conjugates were observed. Keratinocytes after terminal differentiation and those capable of undergoing this change, i.e. cells at high Ca2+ and those at low Ca2+ without retinoic acid, respectively, produce measurable amounts of yglutamylpolyamines (Table IV). Both contain almost all of their transglutaminase activity in particulate form (Table I).
It is probably only inconsequential that in cells in which insignificant amounts of protein-polyamine conjugates are found, i.e. keratinocytes treated with retinoic acid and Chinese hamster ovary cells, most of the total transglutaminase activity is in a soluble form. It seems clear from the results given in Table I that, despite the high percentage of soluble enzyme activity in retinoic acid-treated keratinocytes, their level of particulate transglutaminase activity is not pronouncedly different from that in cells which were cultured without retinoic acid. Therefore, if particulate enzyme is responsible for yglutamylpolyamine formation, as it has been indicated to be for t-(y-glutamy1)lysine cross-linking to form cell envelopes (40,42), the failure of the cells to form protein-polyamine conjugates upon their treatment with retinoic acid cannot be due solely to depletion of particulate epidermal enzyme. Rather, retinoic acid must prevent formation of these conjugates, at least in part, by another mechanism, perhaps by influencing biosynthesis of conjugate precursor proteins. A similar influence of retinoic acid on the synthesis of precursor proteins has been suggested as a partial basis for the prevention of cell envelope formation by this agent (48).
The similarity in molecular size of a M , -6,000 polyaminecontaining component of differentiated mouse keratinocytes (Fig. L 4 ) to that of the subunits of so-called keratolinin from either bovine snout or human epidermis (49) provides a basis for some speculation. Keratolinin as isolated displays a molecular weight of 36,000 and contains no t-(y-glutamy1)lysine bonds (50). It can be very efficiently cross-linked by epidermal transglutaminase into high molecular weight polymers, some of which are solubilized only by conditions that cleave peptide bonds (50). The subunit size of keratolinin is between 6,000 and 6,200 (50); the subunits are dissociated by urea/SDS and thus appear to be associated in a noncovalent fashion (50). The amino acid composition of keratolinin is known and this material, which is believed to be one of the precursors of cornified cell envelopes (49,50), has been reported to contain the unusual amino acid citrulline (51). We speculate that the low molecular weight polyamine-containing component described in this paper is the mouse keratinocyte equivalent of keratolinin, that it is partially modified during cell differentiation by transglutaminase-catalyzed polyamine incorpora-tion, and that this modification influences its solubility, its subunit composition, and its organization for further enzymecatalyzed change. In accordance with these suggestions are the findings of higher molecular weight polyamine-containing materials in the SDS-soluble fraction from differentiated mouse keratinocytes (Fig. 1B) and of y-glutamylpolyamines as components of the cornified envelope fraction of these cells (Table V). The limited amount of Nonidet-soluble material available to date has not allowed determination of the amino acid composition for comparison with that reported for keratolinin. Of particular interest is whether citrulline is a component of this material from mouse keratinocytes. The finding that polyamines are covalently attached to this mouse low molecular weight fraction in mono-y-glutamyl linkage only, whereas substantial levels of bis-y-glutamylpolyamines, in addition to mono derivatives, are found in the cornified envelope fraction (Table V), is consistent with suggestions that the low molecular weight material participates in envelope formation and that partial conversion of mono-to bis-yglutamyl polyamines occurs during its polymerization into envelope. It is possible that the bis cross-linked forms of yglutamylpolyamines contribute, along with c-(y-glu-tamy1)lysine bonds, to stabilization of envelope structure. Finally, an indication that the small polyamine-polypeptide conjugate may have a special function in the organization of precursor molecules for envelope formation arises from a report that preincubation of purified porcine brain tubulin with putrescine and transglutaminase stimulates both the rate and extent of microtubule assembly (52). Based on this observation, it was suggested that enzyme-catalyzed covalent attachment of putrescine to tubulin produces a modified form of this protein that is more readily interactive with other subunits in an assembly system, and thereby promotes this process (52). Consistent with this suggestion is a report that omission of polyamines from the growth medium of polyamine-auxotrophic Chinese hamster ovary cells causes bundles of actin filaments and microtubules to disappear (53).
A fuller understanding of the role of polyamines in differentiating epidermal cells requires that one consider the pronounced reduction in the levels of free polyamines during this process (Table 11). Free polyamines, if present at high levels during those stages of differentiation in which proper e ( y -glutamy1)lysine cross-link formation is imperative, could cause improper stabilization of the cornified envelopes. Thus, the degree of cross-link production could be reduced as a consequence of competition between polyamines and +amino groups of lysine residues for y-glutamyl bond formation. A critical reduction in free polyamines may, therefore, be an essential feature at the terminal stages of epidermal cell differentiation. It may be significant that in such pathological conditions as psoriasis, which involve skin terminal differentiation, large increases in free polyamines are found (54) associated with altered transglutaminase expression (55).