Chicken erythrocyte histone H5 has been suggested repeatedly to be a general suppressor of transcription and replication. Therefore, the biological functions of H5 were investigated and compared with those of Hl (Hla + Hlb) by microinjection of the purified proteins into proliferating L6 rat myoblasts. By pulse-labelling of the injected cells with [3H] uridine and [3H] thymidine it was shown that H5 blocked both transcription and replication substantially, and that the chromatin of the injected cells became densely compacted. Hl also suppressed these functions, but to a much lesser degree. The effects were specific and not caused by change in intracellular pH caused by introduction of the very basic H5, or its non-specific interaction with nucleic acid, since injection of protamine or lysozyme did not affect the cells. The migration and localization of injected H5 was monitored at different times after injection by immunofluorescence, which revealed that H5 was efficiently and stably concentrated in the nucleus.
The results indicate that H5 indeed might function as an inactivator of the erythroid genome in its natural environment, probably by keeping the chromatin in a very condensed state.
Most of the eukaryotic chromatin is compacted into a 30 nm fibre, the structure of which is known in considerable detail. The fibre is made up of a helical array of nucleosomes, probably held together by the linker histone Hl (for reviews, see Felsenfeld & McGhee, 1986; Thömas, 1984). This structure must form a significant barrier to RNA polymerase as it moves along the DNA; therefore, the 30 nm fibre is thought to unravel to some extent in transcriptionally active regions. Indeed, the chromatin structure as a whole and the nucleosome structure of active genes both differ considerably from those of inactive genes (reviewed by Mathis et al. 1980; Weisbrod, 1982).
Since Hl, or some modified form of it, is considered to keep the 30 nm fibre intact (and inactive) (Weintraub, 1984), and since its removal increases the micrococcal nuclease sensitivity of chromatin (e.g. see Noll & Kornberg, 1977), Hl seems to play a crucial role in determining the state of activity of the chromatin. Recent experiments indicate that Hl possibly acts as a ‘crude’ general gene repressor, keeping the chromatin in an inactive ground state (Hannon et al. 1984; Schlissel & Brown, 1984; Weintraub, 1984; reviewed by Weintraub, 1985).
Two striking examples of special linker histone variants associated with a decrease in gene activity are Hl° and H5. Hl° accumulates in contact-inhibited cultured cells (Pehrson & Cole, 1980), and in murine erythroleukaemia cells after induction of erythroid differentiation (Osborne & Chabanas, 1984). H5 accumulates during maturation in the nucleated erythrocytes of birds, fish and amphibia; the cells become transcriptionally inactive and enter G0. In chicken erythrocytes this histone variant accumulates in the chromatin, causing a 70% net increase in linker histone content. During this process the nucleosomal DNA repeat length increases (Moss et al. 1973; Appels & Wells, 1972; Billett & Hindley, 1972; Thömas, 1984; Affolter et al. 1987). H5 and Hl° resemble each other at both the nucleotide and amino acid levels (Doenecke & Tonjes, 1986; Pehrson & Cole, 1981), and bind to the same region of the nucleosome as Hl, i.e. at the positions where the DNA enters the core histone octamer and leaves it after making two turns around it (Allan et al. 1980; Smith & Johns, 1980; Stein & Künzler, 1983).
Since the accumulation of H5 in chicken erythrocytes parallels the inactivation of the nucleus, the protein has been suggested to be a general chromatin repressor (Appels & Wells, 1972; Billett & Hindley, 1972). Also, we have reported earlier that replication in L6 rat myoblasts and quail myoblasts was suppressed when these cells were fused with mature chicken erythrocytes to form heterokaryons. In the process H5 is taken up by the rat and quail nuclei to some extent, indicating that H5 could function as a repressor in this system (Ringertz et al. 1985).
In the present report we compare the effects of H5 and Hl on transcription and replication by microinjection of the purified proteins into active cells.
MATERIALS AND METHODS
L6J1 rat myoblasts (Ringertz et al. 1978) were grown at 37°C in 5% CO2 in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 5% foetal calf serum (FCS). Two hours before the beginning of an injection experiment, the medium was chjanged to DMEM containing 20 mM-Hepes buffer (pH 7·4) instead of bicarbonate, and the cells were kept at 37°C, without CO2, thereafter.
Chicken erythrocytes (CE) were collected by bleeding 18-day-old embryos. The cells were washed twice in phosphate-buffered saline (PBS) and nuclei were prepared by lysing the cells in reticulocyte standard buffer (RSB; 10mM-Tris-HC1, pH 7·4, 10mM-NaCl, 3mM-MgCl2) with 0·5% NP-40. After washing three to four times in RSB/0·005% NP-40 the pellet was resuspended in 3 pellet vol. of distilled water. A crude linker histone preparation was obtained by sonicating the partially lysed nuclei to a homogeneous chromatin suspension, which was extracted with 3% perchloric acid for 15 min on a magnetic stirrer. The linker histones were extracted from the chromatin for 15 min at +4°C. After centrifugation, the extracted proteins in the supernatant were precipitated by addition of 0·1–0·5 vol. of 100% trichloroacetic acid. To purify H5 and Hl the following steps were performed as described by Spring & Cole (1977). The precipitate was dissolved in a minimum volume of sample buffer (8 M-urea, 1% β-mercapto-ethanol, 0·1 mM-phenylmethylsulphonyl fluoride (PMSF)), typically 70·110mg/2–4ml, and loaded onto a Sephadex G-75 column equilibrated at pH 1·7. Fractions (3 ml) were collected and monitored at 230 nm absorbance, dialysed against distilled water, lyophilized and analysed on SDS-polyacrylamide gels. All steps until the lyophilization were performed at +4°C. One preparation, starting with 1010 nuclei, gave 20–30 mg of pure H5 and 10–15 mg of pure Hl. The Hl fraction was not further fractionated, but used as an Hla4-Hlb mixture. The amount of protein was determined by the Lowry technique (Lowry et al. 1951).
On the day of the experiment, the chicken linker histones and the control proteins were dissolved in sterile, filtered water and centrifuged for 20 min at 27 000 g to remove all particles that could otherwise plug the injection capillaries. The procedure was carried out according to Graessmann et al. (1980), with minor modifications. Grids of 1 mm2 has been etched into glass slides; the cells were grown to about 20% confluence and injected on these slides. For each time point and protein concentration, all cells in one single field had their cytoplasms injected (50–200 cells). The injected volume was calculated by Graessmann et al. (1980) to be approx. 5 ×10 −1 ml per cell. Injection of this volume into the cells gave approx. 0·3 × 106, 1·5 ×106 and 15 × 106 exogenous molecules of H5 or Hl per cell at protein concentrations of 0·2, 1 and 10 mg ml−1, respectively.
As control for possible effects of the actual protein injection, FITC-coupled anti-mouse IgG (0·6mgml−1) and ovalbumin (Sigma; 1 and 10 mg ml−1) were used. To control for effects of injection of basic proteins, protamine (Salmine; Serva) and chicken egg white lysozyme (Merck) were injected at 1 or 10 mg ml−.
After injection the cells were incubated at 37°C for different periods of time and then pulse-labelled with either [3H]uridine or [3H]thymidine.
Pulse-labelling, autoradiography and immune staining
For autoradiography (ARG) the cells were labelled with 20–60 μCi of [3H]uridine (spec, act., 29 Ci mmol−1) per 15 ml culture medium for 1–14 h for the shorter, and 23 h for the longer transcription experiments. The [3H]thymidine incorporation experiments were done with 20—60 μCi of [3H] thymidine/15 ml medium (spec, act., 20Cimmol−1) for 0·5–20·5h in the shorter, and 23 h in the longer experiments.
After labelling, the cells were fixed in absolute ethanol: acetone, 1: 1 (v/v). In the case of [3H]uridine, the slides were hydrated, washed in 5% cold trichloroacetic acid, and then rinsed in water and overlaid with film (Kodak). The film was exposed 4–5 days at +4°C, developed and photographs were taken. The film was then removed and the slides were stained with a rabbit anti-H5 antiserum (a generous gift from T. Graf (Beug et al. 1979)), and then with a FITC-coupled secondary anti-rabbit antibody (Kappel).
In the case of thymidine the slides were fixed, stained with antibody and then covered with film and exposed as above. The cells injected with control protein were treated the same as the histone-injected cells.
Morphology of injected cells and localization of H5 after microinjection
Cells injected with 1 mg ml−1 H5 into the cytoplasm were incubated for 0·10 h, fixed and then stained with anti-H5 antiserum in order to study the migration, localization and visible effects of introduction of a protein specific for inactive cells.
At Oh (fixation immediately after injection) the morphology of the cells appeared normal and not disturbed by the injection (Fig. 1A). All the antigen was spread throughout the whole cytoplasm, with none found in the nucleus (Fig.1B). After 1–5 h all the H5 antigen was concentrated in the nucleus, appearing to accumulate in dots and around the nuclear periphery. Many of the cells had decreased in size, and had vacuolated cytoplasms and compacted nuclei (Fig. 1C,D). By 3–5 h the myoblasts had lost their characteristic shape, and both cells and nuclei were small and compact. H5 was concentrated in the nucleus (Fig. 1E,F). After a 10-h incubation, the volume of the cytoplasm of most of the injected cells was strikingly reduced, and the small nuclei were very compact, resembling chicken erythrocyte nuclei. H5 was highly concentrated in the nuclei (Fig. 1G,H). Cells injected with IgG or ovalbumin (Fig. 8), lysozyme or protamine (not shown) appeared normal.
Effects of microinjection of H5 and HI on transcription in proliferating cells
The experiments presented in Fig. 1 showed that some time was needed for H5 to reach the nucleus, but gave no indication of when effects on transcription and replication would occur. Also, no data of what protein concentrations would give such effects could be obtained from these experiments. We therefore tested various time intervals between injection and pulse-labelling, with a protein concentration of 10 mg ml−1 histone in the solution to be injected. This concentration was reduced during subsequent work; most experiments were done with 1 mg ml−1. Table 1 shows the results of separate injection experiments using different histone concentrations and incubation times. There was an obvious concentration dependence of the inhibition caused by the injected histones, but the effect of varying the time interval between injection and pulse-labelling was not pronounced, being more or less obscured by statistical errors caused by the low number of cells per field (see Materials and methods and Discussion).
Therefore, all data were pooled (as mean values from the different experiments) into two groups, in order to obtain a statistically satisfying number of cells per group. These groups were short and long studies: 1–25 h and 25–42h incubations for the [3H]uridine experiments (incubation including labelling; see Materials and methods). In the transcription experiments, all uninjected cells were labelled with [3H]uridine, as were all cells injected with control protein (see Figs 6, 7 and 8, below), so that the effect of H5 and Hl could be simply calculated as the decrease in the number of labelled cells. Only totally unlabelled (no silver grains at all on the autoradiogram (ARG)) cells were counted as negative, the rest as positive (see Discussion).
The micrographs in Fig. 2A-C show the effects of microinjection of H5 (Imgml−1) on transcription in proliferating rat myoblasts. The cells became compacted and unable to incorporate [3H]uridine (Fig. 2C). Fig. 3 groups the results from the different experiments measuring transcription. Three histone concentrations were used: 10, 1 and 0·2 mg ml−1. For H5 both short and long studies were done, for Hl only the short study. In all the above categories (Fig. 3) H5 inhibited transcription to a greater extent than Hl. Clearly, the effect was concentration-dependent, and the amount of protein needed to obtain inhibition was less for H5 than for Hl. At the longer incubation time, 25–42h, the inhibitory effect of H5 decreased (Fig. 3C).
To obtain more detailed information about the effects of H5 on transcription, a kinetic study was done that included time points soon after injection. The inhibitory effects are plotted against time in Fig. 4. Transcription was inhibited both faster and more efficiently with H5 than with Hl. For H5, a near-maximum effect was already seen 1 h after injection, whereas at the same time point the value for Hl was 1β of its maximum. The maximum effects for H5 for Hl at 4–5 h after injection differed by approx. 20% (50% positive cells for H5, 73% for Hl; Fig. 4). The inhibition caused by H5 was more persistent than that caused by H1; 10 h after injection the Hl-injected cells had almost completely recovered. Despite the differences in kinetics and degree of inhibition, the maximum effects occurred around the same time (4–5 h), and the time curve showed similar patterns for both proteins.
Injection of control proteins (IgG, ovalbumin, lysozyme or protamine) did not to any significant degree affect the ability of the cells to incorporate [3H]uridine. Figs 6 and 7 show the results of injection of the different control proteins, the various concentrations and incubation classes. The data were collected, as in the histone experiments, as mean values of the ARG counts in the different groups. The bars (striped) in Fig. 6 indicate that injection of IgG or ovalbumin, even at high concentrations, did not affect the ability of the myoblasts to incorporate [3H]uridine, regardless of duration of incubation. The morphology of the cells was likewise not affected, and the FITC-labelled IgG did not enter the nucleus (Fig. 8B). These results show that injection of a protein per se does not affect transcription. Since histones are very basic (pl for H5 is approx. 12), the effects seen after injection could be those of a sudden change in intracellular pH or of non-specific interactions with the nucleic acids. To rule out these possibilities lysozyme and protamine were injected. Fig. 7A,B (dark, striped pattern) and 7C,D (black) show that the basic proteins initially disturbed transcription to some degree (96% positive cells for lysozyme; 67% for protamine in the short studies) at 10 mg ml−1. At later time points there was no effect (Fig. 7B,D), and at no time did they affect cell morphology (not shown).
Effects of H5 and Hl on DNA replication
In the [3H]thymidine incorporation experiments, the short study was 0·5–26·5 h (H5) or 1·5—26·5 h (Hl), and the long study was 26·5–46 h (incubation including labelling). The labelling index of the uninjected cells varied from slide to slide, with the degree of myogenic differentiation of the L6 myoblasts and with the length of the labelling period, between 95% and 27% positive cells. Therefore, for each slide, the index obtained from the uninjected cells was set as 100%, and the effects of injection were calculated relative to that value.
The results of the individual injection experiments are shown in Table 2, the morphological effects of H5 are shown in Fig. 2D-F, and the results of the different experimental groups are collected in Fig. 5.
H5 inhibited replication more efficiently than Hl in both short and long experiments, and the effects were again concentration-dependent. Also, replication seemed more sensitive to H5 than transcription, and the suppression was more persistent (compare Figs 3A,C and 5A,C). Hl also inhibited replication of the rat myoblasts, but to a much lesser extent: at 10 mg ml−1 Hl in the short study, 22% of the cells were positive; in the other categories the inhibition was negligible.
As a whole, the replicative mechanisms of the cells appeared more sensitive to microinjection than the transcriptional processes, since injection of the neutral control proteins slightly lowered the percentage of labelled cells (Fig. 6A-C, dotted bars). However, the absolute majority of the cells injected with ovalbumin or IgG were able to incorporate [3H]thymidine, as shown in Fig. 8F. Of the basic control proteins only lysozyme at 10 mg ml−1 affected replication slightly (Fig. 7A, striped bar: 74% positive cells; B, striped bar: 80% positive cells).
In order to study the biological functions and effects of the erythrospecific linker histone H5 in living cells, as compared to those of Hl, we have microinjected these proteins into proliferating rat myoblasts. Since H5 accumulates in the chicken erythrocyte (CE) nucleus during erythroid maturation concomitantly with a compaction of the chromatin, while both transcription and replication are drastically reduced, we have studied these parameters in myoblasts after injection of H5. The results show that H5 blocks both transcription and replication to a higher degree than Hl, and that the myoblast nuclei become compacted. It should be pointed out that many of the H5-injected cells were detached from the dish and lost their shape as they became inactivated after subsequent incubation (e.g. in a uridine experiment approx. 35% of the cells were lost; in a thymidine experiment approx. 25%; not shown). This gave a bias to the results. Also, many H5-injected cells that were counted as positive here showed a reduced label compared to control cells, in both uridine and thymidine experiments. None of these phenomena occurred when control proteins were injected. These facts indicate an even greater effect of H5 than that reported numerically above.
The effects occurred quite rapidly: already after one hour of incubation, transcription was reduced by almost 50% at 1 mg ml−1 of H5 (Fig. 4), and replication in one experiment by 80% at 10 mg ml−1 after half an hour (Table 2). Einck & Bustin (1983) have shown that microinjected antibodies against HMG 17 and histones H2A, H2B and H3 inhibited transcription, when concentrated into the nucleus. This points to the fact that the core histones may also be involved in determining the activity of the chromatin. To investigate this, we tried to inject core histones (our own preparations from CE) and histone H4 (Boehringer) as controls. However, these preparations were extremely toxic to the cells, even at low concentrations (OTmgml−1), causing massive cell detachment and death (not shown). This might be due to contamination by non-histone chromosomal proteins in the preparations. It should be stressed, however, that the aim of this study was to compare the effects of the linker histones H5 and Hl.
Our results indicate that replication is more sensitive to the introduction of H5 than is transcription. This may be due to a disturbance of a variety of mechanisms determining the length and regulation of the cell cycle, compared to a possible blockage of the movement of RNA polymerase or of initiation of transcription. From Fig. 5 it can also be seen that the inhibition by H5 of replication was more persistent than that by Hl. These results on replication also support our earlier notion that leakage of H5 from the chicken erythrocyte nuclei to the active rat nuclei in CE × L6J1 heterokaryons may suppress DNA synthesis in the rat nuclei (Ringertz et al. 1985). The inhibitory effect of H5 was clearly concentration-dependent, in agreement with the situation in early (immature) polychromatic CE, which contain H5 and are genetically active. They are only inactivated later in the differentiation process when H5 has accumulated to large amounts (Billett & Hindley, 1972; Williams, 1972). From the time curve experiment (Fig. 4) it can be seen that H5 had to be present in the nucleus to exert its inhibitory effect, indicating that the effects were not mediated by other cytoplasmic factors induced or activated by H5. The protein became detectable exclusively in the nucleus within 1-5 h after injection (Fig. ID), suggesting a rapid and efficient translocation mechanism. Moreover, once concentrated in the nucleus, H5 remained there and did not seem to be massively degraded, since the immunofluorescence was still intense after 30-40h (not shown). Uptake of proteins into the nucleus has been shown to depend on the size of the molecule, its origin and its charge (De Robertis, 1983). Paine et al. (1975) have shown the functional diameter of the nuclear pore to be approx. 45 A, and proteins below that size can enter the nucleus but do not accumulate unless they are of nuclear origin. However, nuclear proteins of up to 165 000Mr can transverse the nuclear membrane, but non-nuclear proteins such as IgG (160000Mr) cannot enter at all (Bonner, 1975; Fig. 7B of this paper). Histones have earlier been shown to be taken up rapidly by isolated nuclei in suspension (Cox, 1982), and HMG 1 and 2 have been shown to accumulate in the nucleus after microinjection (Wu et al. 1981). The faithful transport and localization of H5 (20 580Mr) suggests that the histone molecules might contain some sort of transport signal, as has been shown for nucleoplasmin (Dingwall et al. 1982), SV40 large T antigen (Kalderon et al. 1984) and yeast histone H2B (Moreland et al. 1987). However, the seven amino acid stretch that determines the nuclear localization of the latter histone is not present in H5.
The persistent inhibition of transcription and replication, and the change in nuclear and cellular morphology detected after injection of H5 were not caused by secondary effects of a sudden change in pH or by nonspecific base-acid interactions between nucleic acid and basic proteins. This was shown by injection of lysozyme and protamine, two very basic proteins. They disturbed cell functions only slightly and reversibly, when injected at high concentrations (Fig. 7A-D). In fact, the initial transcriptional inhibition of protamine seen in Fig. 7C was detected 3 h after injection, and completely overcome 23 h later (Fig. 7D). An even higher degree of inhibition could have been expected for protamine, since this small, very basic peptide binds very strongly to DNA of spermatocytes of higher animals. In a process much resembling avian erythrocyte maturation and inactivation, protamines replace histones concomitantly with an inactivation of the genome (Olson & Busch, 1974). Despite the strong analogy with H5, protamine probably requires the milieu of a more mature and spermatocytelike cell type to exert its function. The weak effect of 10 mg lysozyme ml−1 on replication (Fig. 7A,B) could be explained by non-specific disturbance of cytoplasmic functions (translation, nucleotide synthesis pathways), since this protein should not enter the nucleus. Taken together, the basic proteins caused some inhibition only when injected at the highest concentrations.
The results of the histone injections show a clear difference between the inhibitory effects of H5 and Hl. However, Hl did initially cause a significant decrease in transcription and replication. Chicken erythrocyte Hl (the same Hla + Hlb fraction as used here) has previously been shown to reduce the binding of RNA polymerase to isolated chromatin to about the same extent as low concentrations of H5. However, at high concentrations giving about two molecules per nucleosome (comparable to the situation in mature CE) H5 was shown to block transcription totally (Hannon et al. 1984). Also, Schlissel & Brown (1984) have shown elegantly that the binding of Hl to 5 S RNA gene chromatin inXenopus inhibits formation of a transcription complex. In our experiments, however, the cells recovered from the effects of Hl with time (Figs 3, 4, 5). Differences in transcriptional and replicational blocking and ability to accumulate on chromatin could be explained by differences between H5 and Hl in affinity for DNA. Kumar & Walker (1980) have determined the standard free energy of dissociation of Hl to be 0·5 kcal mol−1, and 3·5 kcal mol−1 for H5, in 1 M-NaCl. We have found the binding constant of H5 to be approximately double that of Hl by a fluorescence-quenching technique (M. Bergman & F. Watanabe, unpublished).
The exact mechanism of inhibition of transcription and replication is unknown. The simplest model would presume a steric blocking of RNA and DNA polymerases by the interaction of H5 with DNA. In conflict with this model, however, it has been shown that so-called active chromatin of chicken erythrocytes contains both H5 and Hl (Weintraub, 1984), and that H5 is scattered all over the genome, intermixed with Hl (Mazen et al. 1982; Torrez-Martinez & Riuz-Carrillo, 1982). A more complicated mode of this blocking, then, would be that H5 has a greater preference than Hl to form stable, higher-order structures of chromatin, and thus condense it to a higher degree than chromatin containing only Hl (Allan et al. 1981; Thömas, 1984; Thömas et al. 1985).
In summary, in a comparison of chicken linker histones H5 and Hl, we have shown that H5 efficiently inhibits both transcription and replication when introduced into proliferating rat myoblasts by microinjection. Hl shows a clearly smaller degree of inhibition of these functions.
That in turn, indicates that H5 might be an important factor in the later stages of inactivation of the erythrocyte genome during avian erythropoesis.
We thank Dr Nils R. Ringertz for supplying facilities and financial support (Swedish Medical Research Council (13U-5951), and funds from Karolinska Institute). We thank Mrs E. Mellqvist for excellent photographic work and assistance with the ARG and immune staining, and Mrs Jennifer Abrahamsson for assistance with the figures. The stay of E. W. at Karolinska Institutet was sponsored by an EMBO short-term fellowship (ASTF 4943).