transcriptional potential of nascent chromatin - NCBI

The EMBO Journal vol.12 no.10 pp.3855-3864, 1993

Deposition of chromosomal protein HMG-1 7 during replication affects the nucleosomal ladder and transcriptional potential of nascent chromatin Massimo P.Crippa, Lothar Trieschmann, Pedro J.Alfonso, Alan P.Wolffe1 and Michael Bustin2 Laboratory of Molecular Carcinogenesis, NCI and ILaboratory of Molecular Embryology, NICHD, National Institutes of Health, Bethesda, MD 20892, USA 2Corresponding author Communicated by A.Rich

A cell-free system from Xenopus eggs was used to study the role of chromosomal protein HMG-17 in the generation of the chromatin structure of transcriptionally active genes. Addition of HMG-17 protein to the extracts, which do not contain structural homologs of the HMG-14/-17 protein family, indicates the protein is incorporated into the nascent template during replication, prior to completion of chromatin assembly. The protein binds to and stabilizes the structure of the nucleosomal core thereby improving the apparent periodicity of the nucleosomal spacing of nascent chromatin. Assembly of HMG-17 into the nascent chromatin structure significantly increased the transcription potential of the 5S RNA gene and satellite I chromatin. Kinetic studies indicate that the increase in transcriptional potential is observed only when HMG-17 is incorporated into nucleosomes during chromatin assembly. Key words: active chromatin/chromatin assembly/chromosomal proteins/transcription/Xenopus

Introduction Chromatin structure plays an important role in regulating the expression of the genetic information encoded in DNA. The chromatin structure of transcriptionally active genes is different from that of untranscribed genomic regions. However, the nature of these differences and the mechanisms involved in the generation of the chromatin structure of active genes is not understood. The features most commonly associated with active chromatin include increased susceptibility to nuclease digestion (Garel and Axel, 1976; Weintraub and Groudine, 1976), increased content of acetylated histones (Hebbes et al., 1988), reduced levels of lysine rich histones (Kamakaka and Thomas, 1990), increased lability of H2A -H2B dimers (Jackson, 1990; van Holde et al., 1992) and increased content of certain nonhistone chromosomal proteins (Weisbrod and Weintraub, 1979; Albanese and Weintraub, 1980; Weisbrod et al., 1980). Recent results convincingly demonstrate that nucleosome positioning is an integral part of the transcriptional mechanism and that the position of histone -DNA contacts, with respect to initiation of transcription, may facilitate the binding of trans-acting

factors to their target (Grunstein, 1990; Simpson, 1991; Hayes and Wolffe, 1992). Although it is clear that nonhistone chromosomal proteins are associated with the chromatin fiber, the role of these proteins in the various aspects of the transcriptional process remains obscure. Several lines of experimental evidence are consistent with the possibility that nonhistone chromosomal proteins HMG-14 and HMG-17 are part of a process that confers distinct properties to chromatin regions containing transcriptionally active genes (Bustin et al., 1990a). These two proteins are the only known nonhistones with specific affinity for the nucleosome core particle (Albright et al., 1980; Sandeen et al., 1980; Crippa et al., 1992). Each nucleosome has two potential binding sites for either HMG-14 or HMG-17 (Mardian et al., 1980); however, the limited amount of protein in the nucleus confines their presence to a nucleosomal subset. Nuclease digestion experiments suggested that the susceptibility of active genes to DNase I digestion is associated with the presence of these proteins (Weisbrod and Weintraub, 1979; Weisbrod et al., 1980). Although these results remain controversial, data obtained from several laboratories support the involvement of these HMGs in transcriptional processes. Thus, antibodies against HMG-14 preferentially bind to Balbiani rings of polytene chromosomes of Chironomus pallidivittatus (Westermann and Grossbach, 1984) and microinjection of antibodies to HMG-17 into human fibroblast nuclei inhibits transcription (Einck and Bustin, 1983). Affinity chromatography with columns containing anti HMG-17 antibodies revealed that core particles enriched in HMG-17 contained elevated levels of acetylated histones (Malik et al., 1984) and chromatin fractions enriched in HMG-14 and HMG-17 are also enriched in transcribable sequences (Druckmann et al., 1986; Dorbic and Wittig, 1987). Most of the experiments described above attempt to infer the function of the HMGs by analyzing their organization in the assembled chromatin. In the present manuscript we examine whether these abundant and ubiquitous chromosomal proteins are involved in the generation of the chromatin structure of active genes by affecting the process of chromatin assembly. This question is examined in Xenopus laevis egg extracts, which have been previously used to investigate chromatin assembly and transcriptional regulation. In these extracts: (i) chromatin assembly is coupled to DNA replication; (ii) the nucleosomes on the assembled chromatin are physiologically spaced; (iii) transcription factors and histones compete for association with the replicated template and (iv) the transcription of defined genes is dependent on their chromatin structure (Almouzni and Mechali, 1988a,b; Almouzni et al., 1990a,b, 1991). Our findings indicate that this system is also suitable for studies on the role of nonhistone proteins in chromatin structure and function. We show that HMG-17 binds to core particles assembled on replicated DNA and that the

3855

M.P.Crippa et al.

11;'' .:,

..

Ift

Fig. 2. HMG-17 does not affect the rate of the chromatin assembly. Single-stranded DNA and radioactive dCTP were added to extracts in either the presence or absence of Mg2+-ATP and HMG-17 (HMG-17:core molar ratio = 67:1). Radiolabelled DNA was fractionated on an agarose gel. The autoradiographs of the gels depict the mobility of the supercoiled (form I), circular (form III) and linear (form II) DNA.

together with various amounts of purified HMG- 17 protein. The Western blot in Figure IA shows that the antiserum does not react with any protein present in the extract, but

Fig. 1. Egg extracts do not contain HMG-17. 5 yd of egg extract were fractionated on a 15% polyacrylamide SDS gel along with recombinant HMG-17, calf thymus HMG-1, 5% (w/v) PCA-extracted proteins from X. laevis liver and low molecular weight markers (BRL), transferred to nitrocellulose and the membrane probed with either: (A) an antibody against the nucleosome core binding domain of HMG-14/-17 or (B) an antibody against calf thymus HMG-1. The amount of protein markers added to each lane is indicated on top of the lanes. (-) extract; (+) extract supplemented with 300 ng HMG. Liver: 5% perchloric acid extract from X.laevis liver.

incorporation of the protein into nucleosomal particles produces a template with an increased demarcation of nucleosome spacing and an increased potential for transcription by RNA polymerase HI. The studies provide insights into the cellular function of the HMG-14/-17 class of chromosomal proteins and are pertinent to the understanding of processes involved in the generation of active chromatin.

Results The egg extract does not contain HMG-17 Since we wished to examine the effect of HMG-17 on nucleosome assembly it was important to determine whether HMG-17 or HMG-14/-17 analogs are present in the egg extract. Comparative sequence analysis of all members of the HMG-14/-17 protein family revealed that a stretch of -30 amino acids, comprising the nucleosomal binding domain of the protein, is evolutionarily conserved (Bustin et al., 1990a). A peptide comprising this region of the protein binds to core particles (Crippa et al., 1992) therefore this region is a hallmark of the HMG-14/-17 protein family. Indeed, antibodies specific to this peptide recognize HMG-14/-17 protein from a variety of species (Bustin et al., 1990b). This antiserum was used to test whether the egg extract contained HMG-14/-17-like proteins. Egg extracts were fractionated on SDS-containing polyacrylamide gels,

3856

recognizes HMG-17 exogenously added to the extract. The antiserum detects a protein present in a 5 % perchloric acid extract of X. laevis liver, suggesting that HMG-14/-17-like proteins may be present in somatic tissue, but are absent in the egg. As an additional positive control we used an antiserum against calf HMG-1, which has been previously used to identify HMG-A, an HMG-1-like protein present in amphibian oocytes (Kleinschmidt et al., 1983). Indeed, under the conditions used in an attempt to detect HMG-17, the anti HMG-1 serum detects protein HMG-A in the egg extract. We conclude therefore that chromosomal proteins HMG-14/-17 or close structural homologs are not present in the egg extract. However, these studies do not exclude the possibility that distinct embryonic forms of these proteins or functional homologs are present in the egg extract. Assembly of HMG- 17 into nucleosomes In a Xenopus egg extract, supplemented with Mg2+-ATP, single-stranded DNA is replicated and assembled into regularly spaced chromatin in a time-dependent manner (Almouzni and Mechali, 1988a,b). The kinetics of the assembly of the replicated DNA into chromatin can be monitored by the appearance of supercoiled DNA. The rate of chromatin assembly is not significantly affected by the presence of HMG-17 (Figure 2). Supercoiled (form I) is detected -75 min after addition of the single-stranded template. DNA replication and chromatin assembly continue for a longer period; however, the relative amounts of supercoiled (form I), nicked circular (form II) and linear DNA (form IE) remain constant. All the assembly reactions described in this paper were carried out for 120 min. Next we tested whether the HMG-17 protein was incorporated into the assembled nucleosomes. Specific binding of HMG-17 to nucleosome cores can be detected by mobility shift assays. (Albright et al., 1980; Mardian et al., 1980; Sandeen et al., 1980; Crippa et al., 1992). Indeed, sequential binding of one or two molecules of HMG to cores produces two distinct mobility shifts (Figure 3A). To test whether HMG- 17 was incorporated into nucleosomes in the egg extract, minichromosomes which were assembled either in the presence or absence of HMG-17 were purified on Sepharose 4B and digested with micrococcal nuclease. Figure 3B indicates that core particles obtained from

HMG-17 and chromatin assembly A.

: .-

w

c

b

1

.

-

.hi~.~______

_."

-

Miar atic-_----

._

CP 2 CoP-,_ _,,i

1

a

,

I.a

D' A

B. HsFr>- :;

.;.

HMG1 N1G 7 rIl!)(G"-" i

~~-

" _

.I

:.

..z.

aftim

,mp

ww

T-%.-

el

f,

- AWL Ak. am. qw NOW qw"

..

46

-...

Fig. 3. HMG-17 is assembled into nucleosomes. (A) Mobility shift assays of specific interaction of HMG-17 with nucleosome core particles. 32P-end labelled core particles prepared from chicken erythrocytes were mixed with DNA extracted from these particles and the mixture was incubated with recombinant HMG-17 at the molar ratios indicated at the top of each column. Note that the protein binds to core particles and produces two mobility shifts corresponding to the binding of either one or two HMG molecules to the core. (B) Binding of HMG-17 into chromatin assembled on replicated DNA. Minichromosomes assembled in the presence of HMG-17 were fractionated on Sephacryl 300, digested with micrococcal nuclease and the digest fractionated on a 4% (w/v) polyacrylamide gel. The position of the core particle (CP) and the core particle containing either one (CP + 1) or two molecules of HMG-17 (CP + 2) is indicated on the left. The molar ratio indicates the number of moles of HMG-17 per mole of core particle, assuming that all the input DNA was replicated and assembled into minichromosomes. (C) Western analysis of proteins in the assembled chromatin. Chromatin assembled in the extract was purified by sucrose gradient centrifugation and the proteins present in the minichromosomes analyzed by immunoblotting with antibodies specific to histone H2A, H2B and HMG-17.

chromatin assembled, on replicating DNA, in the presence of HMG-17 migrate slower than the core particles obtained from chromatin assembled in the absence of HMG- 17. The shift in mobility was very similar to that obtained with in vitro reconstituted core particles. In a separate experiment minichromosomes assembled either in the presence or absence of HMG- 17 were purified on sucrose gradient and the protein determined by Western analysis with antibodies against histones H2A, H2B and HMG-17. The results presented in Figure 3C clearly indicated that HMG-17 copurified with the minichromosomes. We conclude therefore that HMG-17 is incorporated into the assembled chromatin. HMG- 17 affects the apparent nucleosomal periodicity of the assembled minichromosomes Studies with cell free systems indicate that the assembly of chromatin during replications involves several distinct steps.

Fig. 4. HMG-17 improves the apparent nucleosomal spacing on assembled chromatin. Chromatin assembled on single-stranded M13 DNA under the various conditions described in the figure was digested in the extract with micrococcal nuclease for the times indicated, the DNA purified, fractionated on an agarose gels and exposed for autoradiography, as described in Materials and methods. Comparison of panels A and B illustrates the effect of Mg2+-ATP on chromatin assembly. The presence of HMG-17 during chromatin assembly, in the absence of Mg2+-ATP, results in a clearly defined core particle with little background between the bands. Note, however, that the number of bands is not significantly affected (compare 2' and 5' digestion point in panel B with that in panel C). In the presence of Mg2+-ATP (panel D) the effect of HMG-17 on the background between the bands is less pronounced, however, the number of resolvable bands is somewhat increased (compare four middle lanes in panel D). The position of DNA corresponding to mono-, di-, tri- and tetranucleosomes is indicated on the left. The input molar ratio of HMG-17 to core was 50.

Sequentially, these steps involve deposition of the H3 -H4 tetramer, assembly of two H2A -H2B dimers and establishment of proper nucleosome spacing (reviewed by Almouzni and Wolffe, 1993). In Xenopus and mammalian extracts the last two steps are influenced by the Mg2+-ATP in the extract (Gilkin et al., 1984; Dilworth et al., 1987; Almouzni and Mechali, 1988a; Banerjee and Cantor, 1990; Almouzni et al., 1991; Kleinschmidt and Steinkeisser, 1991). Indeed, as previously demonstrated by others and as shown in Figure 4, chromatin assembled in the absence of Mg2+-ATP is digested by micrococcal nuclease significantly faster than chromatin assembled in the presence of Mg2+-ATP (compare Figure 4A and B). However, polyacrylamide gel analysis of minichromosomes, assembled under our reaction conditions using several preparations of egg extracts, did not indicate that the composition of the core histones in the chromosomes is significantly changed on the addition of exogenous Mg2+-ATP (not shown). Thus, in our experiments, it is not clear whether Mg2+-ATP affects the final stages of H2A-H2B deposition and nucleosome assembly (Almouzni and Wolffe, 1993), the establishment of proper nucleosome spacing by various factors (Tremethick 3857

M.P.Crippa et al.

and Frommer, 1992) or chromatin compaction by excess Mg2+ ions (Hansen and Wolffe, 1992). Conceivably, all of these factors might independently, or synergistically, influence the digestion of the chromatin template by micrococcal nuclease. Western analysis (see Figure 3) indicated that HMG-17 protein is associated with purified minichromosomes regardless whether they were assembled in the presence or absence of exogenous Mg2+-ATP (not shown). However the presence of HMG-17 has significant effects only on the structure of chromatin assembled in the absence of Mg2+-ATP. Thus, chromatin assembled in the absence of Mg2+-ATP, but in the presence of HMG-17, displays a well-defined nucleosomal spacing (compare Figure 4B with C). The nucleosomal ladder obtained from chromatin assembled in the presence of HMG-17 is extremely well demarcated and the background beneath and between the bands is significantly diminished. In fact, the nucleosomal spacing pattern is more defined than that generated by digestion of a minichromosome assembled in the presence of Mg2+-ATP. However, the number of identifiable bands is not increased. For example, the nucleosomal ladder obtained after 5 min and 10 min extends to the same position (tetramers and trimers respectively) regardless of the presence of HMG-17 (Figure 4B and C). The effect of HMG-17 on the structure of chromatin assembled in the presence of Mg2+-ATP is less pronounced (Figure 4D). The number of resolvable bands in the ladder is increased, however, the background between the bands does not decrease significantly. The difference in the effect of HMG-17 on the nucleosomal spacing between chromatin assembled in the presence or absence of Mg2+-ATP may be due to the kinetics of assembly of HMG-17 protein into the nascent chromatin (see results below). The effect of HMG-17 on the nucleosomal ladder generated by micrococcal nuclease raises the possibility that the protein functions as a nucleosome spacing factor (Drew, 1993; Tremethick and Drew, 1993). However, as elaborated in the Discussion, it is more probable that the apparent increase in the definition of the nucleosomal spacing is due to the stabilization of the core particle by HMG-17. An increase in the stability of the nucleosome core and chromatin by HMGs has been previously described (reviewed in Einck and Bustin, 1985; Bustin et al., 1990a). This increase in nucleosome core stability will lead to preferential digestion of the linker regions. Replication-dependent assembly of HMG- 17 on nascent chromatin The generation of a well-defined nucleosomal ladder upon

digestion of assembled chromatin with micrococcal nuclease is an indication of properly assembled and spaced nucleosomes. We used this assay to test whether HMG-17 preferentially assembles into chromatin together with the histones during replication or whether the protein can be loaded onto the nucleohistone fiber after the assembly of the nucleosome core. In these experiments chromatin was assembled in the absence of Mg2+-ATP, under conditions where the effect of HMG- 17 on nucleosome spacing and nucleosome core stability are easily observed. HMG-17 was added to the assembly mixture either together with, or 2 h after, the addition of the single-stranded DNA template, i.e. either prior to the initiation or after the completion of 3858

a,0

_

_

*

,

lb

i:~~~~~~~~~~~~~~~~~~~~WI...w.:. *

**

*e

*

Fig. 5. HMG-17 is correctly deposited onto nascent chromatin during replication. Autoradiograms of micrococcal nuclease digests of chromatin assembled in the absence of Mg2+-ATP. Lanes 1, 2 and 3, no HMG-17 in the extract. Note that the digest did not generate a distinct nucleosomal ladder. Lanes 4, 5 and 6, HMG-17 present during replication. Note that the average size of the digested DNA is about the same as in lanes 1-3, however, the nuclease digest generated a distinct nucleosomal ladder because the background between the bands is reduced. Lanes 7, 8 and 9, HMG-17 added 2 h after the addition of the single-stranded DNA. In this case the presence of HMG-17 did not re-establish the nucleosomal boundaries and the digest did not generate a distinct nucleosomal ladder. M, markers. The numbers above the lanes indicate time of digestion with micrococcal nuclease.

replication. Analysis of the micrococcal nuclease digests of the assembled minichromosomes (Figure 5) indicates that the late addition of HMG-17 failed to generate a micrococcal nuclease ladder which was observed when the protein was present during replication (see Figure 4). Thus, once the chromatin template is assembled, the protein does not delineate the nucleosomal boundaries and does not stabilize the core particle structure and therefore, micrococcal nuclease digestion does not produce a distinct nucleosomal ladder. Although these results could be interpreted to mean that HMG-17 can serve as a spacing factor, we favor the interpretation that HMG-17 affects the stability of the core particle rather than the spacing between the cores (see Discussion). We conclude therefore that the protein is assembled onto the template during replication, prior to the completion of chromatin assembly. Deposition of HMG- 17 into chromatin increases the transcriptional potential of class 11l genes Next, we examined whether the presence of HMG-17 in the assembled chromatin affects the transcriptional potential of the template. Two class III genes, the somatic 5S RNA gene of Xenopus borealis (Peterson et al., 1980) and the satellite I DNA of X. laevis (Lam and Carroll, 1983; Wolffe, 1989) were used. Transcription by polymerase III of both genes requires the presence of transcription factors TFIIB and TFHIC, which are present in the egg extract. The 5S gene also requires TFIIIA, which was supplied in excess. In addition, the ability of polymerase III to transcribe these genes depends on the structure of their chromatin (Almouzni et al., 1990a). Chromatin assembled in an extract supplemented with Mg2+-ATP contains fully assembled, properly spaced nucleosome cores and is a poor template for transcription (Almouzni et al., 1991; Wolffe, 1991). The chromatin assembled in the absence of exogenous

HMG-17 and chromatin assembly HMG-17 added: during assenibiy

Mg"--ATP HMG-17 Sai

afte _

-

A

-.:

-

-+-

_~w*1

_

5S

1

-

*

2

3

4

5

6

8

Fig. 6. Replication-dependent deposition of HMG-17 on prenucleosomal particles increases the transcriptional potential of class III genes. Single-stranded DNA carrying either the satellite I gene of X.laevis or the somatic 5S RNA gene of X.borealis was incubated in the extract under the various conditions described in the figure. Extracted RNA was fractionated on sequencing gels and the gels were autoradiographed. As reported previously (Almouzni et al., 1991), the amount of RNA transcribed from minichromosomes assembled in the presence of Mg2+-ATP is significantly lower than that transcribed from minichromosomes assembled in the absence of Mg2+-ATP. HMG-17 increases the transcriptional potential of minichromosomes containing prenucleosomal particles only when assembled onto these minichromosomes during replication. Addition of HMG-17 to preassembled minichromosomes did not affect their transcriptional potential.

Mg2+-ATP is more susceptible to micrococcal nuclease digestion and is more readily transcribed by polymerase m (Camerini-Otero et al., 1976; Almouzni et al., 1990b). Indeed, consistent with earlier observations (Almouzni et al., 1990a, 1991), we find that 5S RNA gene transcription is more efficient in the absence of Mg2+-ATP than in its presence (Figure 6). Satellite I DNA transcription is influenced to a lesser extent. The transcriptional potential of satellite I and 5S RNA chromatin, assembled in the presence of HMG-17 but absence of Mg2+-ATP, is higher than that assembled in the absence of HMG-17 (Figure 6, see also Figures 7 and 9). The autoradiograms in Figure 6 indicate that HMG- 17 increases the transcriptional potential of chromatin only if it is present during chromatin assembly. Thus, if added after the DNA template has been assembled (Figure 6, lanes 5-8), in either the presence (lanes 7 and 8) or absence (lanes 5 and 6) of exogenous Mg2+-ATP, HMG-17 protein has no effect on transcription. These results are consistent with the data presented in Figure 5, i.e. that HMG-17 has an effect only when properly assembled into the nucleosome cores together with the histones, during replication. In addition, our finding that the level of transcription is not affected if HMG is added in the presence of Mg2+-ATP, or after chromatin assembly ended, indicates that the protein affects the template rather than other components of the transcriptional machinery. The increase in transcription of the 5S gene assembled into chromatin in the presence of various molar ratios of HMG-17 is depicted in Figure 7A. An incremental increase in the amount of HMG-17 in the assembly mixture brings about a concomitant increase in the transcription potential of the assembled chromatin. However, large excesses of HMG- 17 are inhibitory. HMG- 17 protein does not affect the transcription of a control gene (satellite I) which was added as 'naked' DNA together with the 32P-labelled CTP

B "..

=

-

E

-

E

1.,

Fig. 7. Dose-dependent specific transcriptional potentiation by HMG-17. (A) HMG-17 affects the transcription of chromatin. Singlestranded M13 (150 ng) containing the 5S RNA gene was replicated and assembled into chromatin in the presence of the amount of HMG-17 indicated on top of each column. After 2 h [32P]CTP and 100 ng double-stranded plasmid carrying the Satellite I gene were added. After 30 min of transcription the products were analyzed as described in Materials and methods. The unidentified band in the extract (marked X) indicated that equal amounts of material were applied to the gels. Note that HMG-17 affects significantly only the transcription of the 5S gene. (B) Addition of MgCl2 does not inhibit transcription. Autoradiograms of RNA transcribed from minichromosomes containing the 5S gene assembled under the various protocols indicated on top of each lane. Amount of HMG-17 added: 1.2 Ag. Amount of either MgCl2 or Mg2+-ATP: 5 mM. These were added to the mixtures either early (E), i.e. at the onset of replication/assembly or late (L), i.e. together with the [32P]CTP used to measure transcription. Note that Mg had an effect only when present during chromatin assembly. The lanes marked as X (possibly resulting from nucleotide exchange with t-RNA) can be used as internal controls.

used to measure transcription. Quantitative analysis with a Molecular Dynamics computing densitometer indicated that addition of HMG-17 increased the transcription of the 5S chromatin 5.3-fold, while the maximum increase obtained with naked DNA was 1.3-fold. This result provides additional evidence to support the conclusion that the HMG-17-dependent increase in transcription is due to an effect on the chromatin template and not to a nonspecific effect on unidentified components of the transcriptional machinery. Quantitative analysis of various autoradiograms (Figure 8) indicates that the presence of HMG-17 increases the transcriptional potential of satellite I and 5S RNA genes, respectively 4- and 5.5-fold. The effect is observed only if 3859

M.P.Crippa et al. 7

7

6

6

5

5

a)

a)

z(4

a)

0

cD L-

3

U- 3

2

2

1

1

0

0

HMG: CORE (molar ratio)

HMG: CORE (molar ratio)

Fig. 8. Quantitation of the transcriptional potential of satellite I (A) and 5S RNA (B) genes. Autoradiograms of RNA transcribed from minichromosomes assembled under the various protocols described in Figure 6 were scanned with a Molecular Dynamics laser scanner and quantitated. (-), assembly in the absence of exogenous Mg2+-ATP; (+), assembly in the presence of exogenous Mg2+-ATP; D, addition of HMG-17 during chromatin assembly; A, addition of HMG-17 after chromatin assembly. Standard deviations (obtained from four different experiments) for the transcription of the 5S gene, in the absence of Mg2+-ATP, are shown.

HMG-17 is present during replication and the assembly reaction does not contain exogenous Mg2+-ATP. Addition of HMG protein after the chromatin has been assembled does not affect the transcriptional potential of the template. High concentrations of HMG are inhibitory. It has been suggested that Mg2 +-ATP facilitates chromatin maturation and inhibits the transcription of the 5S RNA gene by promoting the deposition of the H2A-H2B dimer and the establishment of proper nucleosomal spacing (Almouzni and Wolffe, 1993). However, it is also possible that during assembly the Mg2+-ATP complex dissociates and the Mg ions compact the chromatin and inhibit transcription (Hansen and Wolffe, 1992). The data presented in Figure 7B demonstrate that in the experimental conditions used here, an apparent excess of Mg2+ ions is not inhibitory to transcription. In these experiments singlestranded M13 DNA containing the 5S RNA gene was assembled into chromatin for 2 h and the amount of 5S RNA synthesized from these template measured. To the assembly mixture, HMG-17, 5 mM MgCl2 or 5 mM Mg2+-ATP were added either during assembly (noted as E, i.e. early in Figure 7B) or 2 h after assembly started, together with the 32P-labelled CTP used to measure transcription (noted as L, i.e. late). The amount of RNA synthesized in the absence of any addition can be estimated from lane 1. The presence of HMG-17 during replication/assembly increases the transcription potential (compare lanes 1 and 2). Addition of HMG-17 late, i.e. after assembly is completed has no effect (lanes 1, 2 and 3). Late addition of either MgCl2 or

3860

Mg2+-ATP has no effect either in the absence or presence of HMG-17 (lanes 4-7). However, when MgCl2 or Mg2+-ATP are present during assembly the resulting chromatin is not transcribed (lanes 8-10). We conclude therefore that in this system, chromatin compaction by potentially free Mg2+ ions is not a significant factor in the transcriptional inactivation of 5S chromatin assembled in the presence of Mg2+-ATP. It should be noted that in the experiments of Hansen and Wolffe (1992) an oocyte nuclear extract with a lower protein concentration (5 mg/ml) than the egg extract (50 mg/ml) is used for the transcription experiment. In addition, their protocols involved extensive dilution of the transcription extract. The timing of HMG- 17 protein deposition during chromatin assembly affects the transcriptional potential of nascent chromatin The H3 - H4 histone tetramer initiates nucleosome assembly and positions the nucleosome on the DNA (Camerini-Otero et al., 1976; Dong and van Holde, 1991; Hayes et al., 1991). The results presented above suggest that the transcriptional potential of the 5S RNA gene chromatin is affected by the proteins sequestered on the H3 -H4-containing prenucleosomal assembly complexes. Addition of HMG-17 increases the transcriptional potential while the Mg2+-ATP facilitated formation of a mature chromatin structure, perhaps the final steps involved in the sequestration of H2A -H2B and the establishment of chromatin spacing, decreases the transcriptional potential of the nucleoprotein

hMg2 -ATP

A

HMG i

B asser-it

HMG 17

Mg2- A-p

!:

A

L 0

Westem

e-:

_ =-.-=.

x

replication

Silver stain

-. 20 C 2 -K -

5S

ss5SDNA

HMG-1 7 and chromatin assembly

_

M_

M

M -- ATP late

z

c

0

z

Mg2+ ATP 0 M

+

late

z

6

7

8

iMG-17 H3 H2A"H4A

_

1

-

,

r a- r, s (-

1,

t .:

v ffi I1 .

-.

Fig. 9. The timing of HMG-17 deposition during chromatin assembly is a factor in the generation of active chromatin. (A) Autoradiograms of RNA transcribed from minichromosomes, containing the 5S RNA gene, assembled under the various protocols described above each column. (B) Diagram of the experimental protocol. The lanes marked as X probably derived from nucleotide exchange with t-RNA and can be used as internal controls. The numbers in the HMG-17 lanes refer to the HMG-17:core molar ratios.

template. Furthermore, our results suggest that the processes mediated by Mg2+-ATP are dominant over the effect of the deposition of HMG-17, since in the presence of exogenous Mg2+-ATP HMG-17 has no effect on transcription (Figures 6 and 8). Since the extracts do not contain HMG-17, we assume that the incorporation of the protein into the replicating chromatin is not mediated by specific factors. It is possible therefore that in the absence of exogenous Mg2+-ATP the process of chromatin assembly and maturation is sufficiently slow to allow proper incorporation of HMG-17. In this scenario the effect of HMG-17 can be explained by kinetic consideration, i.e. transcriptional potentiation requires incorporation of HMG prior to chromatin maturation. To clarify these questions an experiment was performed in which HMG-17 was incorporated into chromatin before completion of nucleosome assembly and the establishment of chromatin spacing. The data presented in Figure 9 demonstrate that when the deposition of HMG-17 precedes chromatin maturation transcription is potentiated. In these experiments singlestranded 5S DNA was replicated and assembled into chromatin for 2 h in either the absence or presence of Mg2+-ATP and HMG-17. In some of the samples the Mg2+-ATP was added to the assembly reaction late (Figure 9B presents an outline of the experimental protocol). The transcriptional potential of chromatin assembled in the absence of Mg2+-ATP and HMG-17 (Figure 9A, lane 2) is significantly higher than that assembled in the presence of Mg2+-ATP (lane 1). The presence of HMG-17 in the assembly mixture increases the transcriptional potential (lanes 3 and 4). Assembly in the presence of both HMG-17 and Mg2+-ATP negates the effect of HMG-17 (lanes 5 and 6).

3_4-5

.1

1

3

4

5

Fig. 10. Protein content of minichromosomes assembled under various conditions. The assembled minichromosomes were purified by centrifugation through 10-30% sucrose gradients as described in the methods section. The minichromosomes were pelleted and the protein in the pellet analyzed by SDS electrophoresis in 15% polyacrylamide gels. Duplicate gels were tested for HMG-17 by Western analysis as described in Materials and methods and in the legend to Figure 1. Lanes 1 and 5, histone markers; lanes 2 and 6, minichromosomes assembled in the presence of Mg2+-ATP and HMG-17 (see Figure 9 lane 6); lanes 3 and 7, assembly in the presence of HMG-17 for 2 h at which point Mg2+-ATP was added for additional 90 min prior to isolation of minichromosomes (see Figure 9, lane 10); lanes 4 and 8, extract devoid of DNA was processed as a control.

However, if chromatin is preassembled so as to contain HMG-17, the template remains active even if Mg2+-ATP is added (lanes 7 and 8). To ensure that sufficient time was allocated to complete all the steps involved in chromatin assembly, transcription was also measured 90 min after the addition of the Mg2+-ATP to the extract (lanes 9 and 10). Under these conditions the template remains active. It should be noted that under these conditions the Mg2+-ATP was present in the extract for 90 min and could hydrolyse to produce free Mg2+ ions which could compact chromatin and inhibit transcription. However comparison of lanes 9 and 10 with lanes 7 and 8 (no preincubation with Mg2+ATP) clearly indicates that hydrolysis of Mg2+-ATP is not a significant factor in these experiments. These results are in agreement with the data presented in Figure 7B. The presence of HMG-17 did not affect the final histone composition of the assembled template. Analysis of the proteins present in various sucrose gradient purified

templates by polyacrylamide gel analysis (Figure 10), clearly indicates that both the chromatin assembled in the presence of Mg2+-ATP and HMG-17 (proteins from Figure 9, lane 6) and the chromatin to which Mg2+-ATP was added late (Figure 9, lane 10) contains a full complement of histones and HMG-17 (Figure 10). The presence of the latter was verified by Western analysis (Figure 10). The possibility that the organization of the promoter itself into nucleoprotein complexes is different from that of bulk chromatin still cannot be excluded. The data suggest that transcriptional potentiation of pol HI genes is dependent on the kinetics of chromatin assembly, particularly on the later stages of nucleosome assembly involving the proper deposition of HMG-17 onto the nascent template.

Discussion The data presented in this manuscript suggest that the timing of deposition of HMG-17 protein, during chromatin assembly on replicating DNA, is an important step in the 3861

M.P.Crippa et al.

generation of active chromatin. HMG-17 increases the transcriptional potential of chromatin only when it is deposited on to the chromatin fiber during chromatin assembly. Addition of HMG after the chromatin has been fully assembled does not affect the transcriptional potential of the template. The presence of the HMG-17 protein in chromatin leads to an alteration in the micrococcal nuclease digestion pattern consistent with the possibility that the protein delineates the boundaries of the nucleosomal core and stabilizes its structure. The results support the notion that the structure of chromatin and its transcriptional potential is linked to nucleosome assembly. Replication-dependent deposition of HMG- 17 on nascent chromatin The process of chromatin assembly and maturation involves sequential deposition of the H3-H4 histone tetramer, assembly of two H2A-H2B dimers and establishment of

nucleosome spacing (reviewed in Almouzni and Wolffe, 1993). These steps are facilitated by the addition of Mg2+-ATP. Although we were not able to demonstrate that the histone content of chromatin is dependent on the addition of Mg2+-ATP, we did notice that these ions affect the structure and transcription potential of the assembled template. It is not presently clear whether Mg2+-ATP affects the deposition of the histone dimer or of a potential spacing factor (Tremethick and Frommer, 1992; Dimitrov et al., 1993). It is well-documented that HMG-17 can be reconstituted with H2A -H2B depleted particles (Piniero et al., 1992). In fact early reconstitution experiments suggested that the H2A -H2B dimer will inhibit the binding of HMG-14/-17 to nucleosomes (Weisbrod et al., 1980). The binding of HMG-14 and -17 to intact nucleosomal core particles is not random; these proteins have preferred binding sites on the core (Albright et al., 1980; Mardian et al., 1980; Sandeen et al., 1980; Crippa et al., 1992). The protein binds to the nucleosomal DNA at the entry/exit point on the core and at the dyad axis protecting these sites from DNase I digestion (Mardian et al., 1980; Sandeen et al., 1980; Brawley and Martinson, 1992; Crippa et al., 1992). Indeed, our results are consistent with the positioning of the HMG-17 at the ends of the nucleosomal DNA and the overall stabilization of the nucleosome structure (reviewed in Einck and Bustin, 1985; Bustin et al., 1990a) suggesting that the in vitro reconstitution faithfully reproduces the placement of the protein by the replication-assembly system. Micrococcal nuclease digestion of the minichromosomes reveals that the protein helps delineate the core particle boundaries and improves the apparent nucleosomal spacing of chromatin. However, HMG-17 delineates the nucleosomal particle, improves the nucleosomal spacing and increases the transcriptional potential of pol III genes only if it is deposited during chromatin assembly. Addition of HMG-17 after completion of chromatin assembly has no effect on the micrococcal nuclease digestion pattern or on the transcriptional potential of the genes studied. The requirement for replication-dependent deposition of HMG-17 may question the validity of previous experiments in which the protein is added to purified core particles. However, detailed mapping of the association of HMG-14 and HMG- 17 with isolated cores clearly indicates that the proteins bind to distinct sites and that this binding is not random (Mardian et al., 1980; Sandeen et al., 1980; Crippa 3862

et al., 1992). Conceivably the HMG is properly placed on the purified core particle in which the DNA ends of the nucleosomal cores are accessible. These regions may be sterically hindered in the assembled, supercoiled minichromosome. Our results also indicate that HMG-17 does not affect the rate of DNA replication and the assembly into minichromosome. We have also noted that the protein does not affect the linking number of the assembled minichromosome (not shown). Most probably the protein affects the structure of the core particle itself, without significant consequences to the higher order chromatin structure (McGhee et al., 1982). Involvement of HMG in the generation of active chromatin The finding that HMG-17 is assembled into the nascent chromatin during DNA replication is relevant to understanding its cellular function and may provide insights into the mechanism whereby HMG-17 preferentially binds to a nucleosomal subset, independent of the underlying DNA sequence. This situation is consistent with the possibility that the proteins are involved in the generation of active chromatin, which may be an initial step in determining the cell-specific transcriptional potential of certain genes. The levels of HMG-17 mRNA increase sharply at the GI/S interphase (Bustin et al., 1987). Thus, it is possible that the early replicating DNA, which is enriched in transcribable genes (Wolffe, 1991), benefits from a larger pool of HMG-17 than late replicating DNA. Therefore the content and organization of HMG-14/17 in early replicating genes may be different from that in late replicating genes.

The role of HMG in transcribing chromatin Several different mechanisms may account for the transcriptional potentiation of chromatin by HMG-17. The first mechanism involves the possibility that the protein acts as a transcriptional activator. The structure of the HMG-14/-17 proteins is reminiscent of that of certain transcriptional activators. Negative charges are clustered into a restricted region of the protein which has the potential to form ae-helices with negatively charged surfaces. However, we have demonstrated that the HMG proteins do not function as 'classical' transcription activators (Landsman and Bustin, 1991). An alternative possibility is that the binding of HMG- 17 to prenucleosomal particles influences nucleosome positioning and fixes the position of the particles over sequences which favor the interaction of polymerase Ill transcription factors with their binding site, or allows access of these factors to their recognition elements within nucleosomes. In this model the function of HMG would be to increase the number of genes which can be efficiently transcribed by the polymerase. However, nucleosome positioning is not seen in Xenopus egg extracts (Almouzni et al., 1990a, 1991); hence, models involving positioning as a regulatory mechanism are unlikely to operate. A third possibility is that the protein facilitates transitions in chromatin structure which accompany transcription through nucleosomes. We note, however, that in the case of the 5S RNA gene the entire gene is complexed with the factors in the preinitiation complex (Wolffe and Morse, 1990). Therefore, RNA polymerase III never has to negotiate histone-DNA contacts. We favor an additional possibility

HMG-1 7 and chromatin assembly

and suggest that an interplay between HMG- 17 and certain nucleosomal proteins, perhaps the H2A -H2B histone dimer or the histone HI-like B4 (Dimitrov et al., 1993), may affect the competition between transcription factors and histone binding during replication. Likewise, the presence of HMG- 17 could affect the release of the H2A - H2B dimer (van Holde et al., 1992) or even the entire histone octamer (Clark and Felsenfeld, 1992) from association with DNA. Is HMG- 17 a nucleosomal spacing factor? Although the HMG protein affects the micrococcal nuclease pattern of the assembled chromatin our data do not necessarily suggest that the protein functions as a physiological nucleosomal spacing factor in Xenopus eggs. Immunochemical analysis clearly demonstrates that the egg extract is devoid of any proteins which contain the nucleosomal binding region of this class of proteins. Yet, the extract can assemble replicated DNA into physiologically spaced nucleosomes (see also Tremethick and Frommer, 1992; Drew, 1993; Tremethick and Drew, 1993). The extract may contain functional homologs, however, so far these have not been identified. In fact the Western analysis indicates that the extract is devoid of proteins containing the highly conserved nucleosome-binding domain of HMG-14/-17, a hallmark of this protein family (Bustin et al., 1990). We favor the possibility that the effects of HMG-17 on the micrococcal nuclease digestion of the chromatin reflects the fact that the protein stabilizes the nucleosome core (reviewed in Einck and Bustin, 1985; Bustin et al., 1990a) and protects it from nuclease digestion. In addition, the presence of HMG at the nucleosomal core boundaries may provide a kinetic barrier to the exonucleolytic activity of micrococcal nuclease. Thus, the frequency of cuts in the core decreases and the relative rate of digestion of the linker DNA increases. The end result is a well-defined micrococcal nuclease ladder, with little background between the bands. Analysis of the data in Figures 4 and 5 is consistent with this notion. The immature particle, assembled without exogenous addition of Mg2+-ATP is not an effective barrier to the enzyme (Camerini-Otero et al., 1976; Almouzni et al., 1990b). In the presence of HMG-17 this particle is more resistant to digestion and the ends of the core more defined, resulting in a marginally extended ladder of welldefined nucleosome multimers. However, the major effect of HMG-17 addition is the marked diminution of the background between adjacent bands. In fact, Figure 5 indicates that the nucleosomal ladder produced in the presence extends to the same position (up to tetramers) as the smear obtained in the absence of HMG-17 (or the late addition of HMG-17). Thus, the major effect of HMG-17 assembly is an increase in core particle stability rather than the determination of proper spacing between adjacent core particles. Relation to previous work on HMG-14/-17 More than 10 years ago, Weintraub and his collaborators suggested that chromosomal proteins HMG-14 and HMG-17 may modulate the chromatin structure of active genes (Weisbrod and Weintraub, 1979; Weisbrod et al., 1980). In spite of evidence consistent with this possibility, the proposal remained controversial chiefly due to the inability to show significant differences between HMG-free and

HMG-bound core particles and because the DNase I sensitivity of transcribed genes was not always associated with the presence of these proteins. Our findings indicating that transcriptional potentiation of pol HI genes is dependent on the kinetics of chromatin assembly rather than the composition of the assembled chromatin, explain the inability to show significant differences between HMG-17-bound and HMG-17-free core particles. As discussed above, HMGs will be correctly placed in purified chromatin subunits; however, the specific effect of HMG-17 on transcription is realized only when the protein is assembled prior to chromatin maturation. Furthermore, the HMG proteins rearrange and migrate among nucleosomes at relatively low ionic strengths (Landsman et al., 1986); therefore, minor differences in the preparation of the chromatin may alter the composition of the template. These alterations may account for discrepancies among the results obtained from various laboratories. It is also possible that transcriptional potentiation at the chromatin level occurs by more than one mechanism and the studies described here are pertinent to only a subset of the cellular genes. The present studies attest to suitability of the Xenopus egg extract assembly system for various studies on the structure and function of chromatin. The HMG-17 protein is properly placed onto the assembled chromatin indicating that the system is suitable for studies on both histones and nonhistone chromosomal proteins. Our results suggest that it is suitable for providing insights into the mechanisms of generation of active chromatin, one of the least understood aspects of the transcriptional process.

Materials and methods Xenopus laevis egg extract X. laevis unfertilized egg extract was prepared as described previously by Almouzni and Mechali (1988a,b).

Western blot analysis Proteins fractionated on a 15% (w/v) polyacrylamide (acrylamide:bisacrylamide = 30:0.8) SDS-containing gel (Laemmli, 1970) were transferred to nitrocellulose in a semi-dry blotting apparatus (Jannsen), according to the recommendations of the manufacturer, for 3 h at room temperature. The membrane was washed in phosphate buffered saline for 10 min three times. Quenching with 2 % (w/v) milk in the above buffer was for 60 min at room temperature, followed by washing as above. Probing with the antibody against the nucleosome-binding domain of HMG-17 (Bustin et al., 1991) was done in quenching solution for 1 h at room temperature, followed by overnight incubation at 4°C and washing as above. The blots were incubated with alkaline phosphatase-labelled goat anti-rabbit antibodies for 1.5 h at room temperature and the location of the antibody visualized by the color reaction with alkaline phosphatase. DNA replication, chromatin assembly and transcription All the reactions were carried out as described by Almouzni et al. (1990a,b, 1991) with single-stranded M13 DNA, prepared by buoyant density centrifugation (Sambrook et al., 1989), carrying either the satellite I gene of X. laevis (Lam and Carroll, 1983; Wolffe, 1989) or the somatic 5S RNA gene of X.borealis (Peterson et al., 1980). Routinely, the reactions contained 10 ng DNA per al of extract. When the X. borealis 5S RNA gene was used in transcription reactions, 1 Al of an extract enriched in class I transcription factors TFIIIA and TFIIIC [prepared according to Smith et al. (1984) and Wolffe (1988)] at -0.3 jig/ul was added to the extract prior to the addition of the DNA. DNA replication and chromatin assembly reaction mixtures were considered complete after 2 h incubation at 22°C. Transcription reactions were monitored by addition of radiolabelled CTP, ribonucleotides and RNasin and routinely were carried out for 30 min (Almouzni et al., 1990a, 1991). DNA and RNA purifications and analysis on agarose or sequencing gels were done as described by Almouzni et al. (1990a, 1991).

3863

M.P.Crippa et al. Purification of assembled minichromosomes and analysis Assembled minichromosomes were purified either by SephacrylS-300 chromatography or sedimentation through sucrose gradients. For Sephacryl S-300 chromatography, minichromosomes assembled in 20 /d extract were applied to an 8001I column equilibrated and eluted with 20 mM HEPES pH 7.5 and 70 mM KCl. For sucrose gradient purification, radiolabelled minichromosomes assembled in the extract were fractionated as described (Wolffe and Schild, 1991) on an SW40 rotor for 2.5 h at 40 000 r.p.m., using a 10-30% (w/v) linear sucrose gradient. Minichromosomes recovered from the gradient were pelleted through a 30% (w/w) sucrose cushion in the same buffer (Wolffe and Schild, 1991). The samples were dissolved in loading buffer and analyzed on a 15% (w/v) SDS gel (Laemmli, 1970) and by Western blotting.

Micrococcal nuclease digestion and analysis of digestion products Micrococcal nuclease (Worthington) digestion of either purified minichromosomes or of minichromosomes in the extract, was initiated by adding CaCl2 to a final concentration of1-2 mM and digestion with the enzyme was at the concentration, temperature and for the times indicated in the figures. The digestion was stopped with 20 mM Tris-HCl pH 8, 20 mM EDTA pH 8 and 0.5% (w/v) lithium dodecyl sulphate, followed by proteinase K digestion (500 jig/ml at 37°C for1 h), extraction and fractionation of the DNA on agarose gel (see above). For analysis of the core particles in the assembled minichromosomes, the digestion performed after minichromosome purification was stopped with 10 mM EDTA pH 8 and the digestion mixture loaded on a 4% (w/v) polyacrylamide (acrylamide:bis-acrylamide = 20:0.69) and 1 x TBE (Tris-borateEDTA; Sambrook et al., 1989) gel.

HMG- 17, histone H2A - H2B dimer and Xenopus laevis liver proteins Recombinant HMG-17 (Bustin et al., 1991) was dissolved in 1% (w/v) bovine serum albumin (BSA) and added to the extract at the concentrations yielding the molar ratios indicated in the figures. For the calculation of the molar ratios the following molecular weights were assumed: HMG-17, 8.9 kDa; nucleosome core, 205 kDa. Replicated M13 plasmids, containing the satellite I gene or the 5S RNA gene were assumed to contain one nucleosome core for every 200 base pairs. Samples incubated without HMG-17 were supplemented with an equal volume of 1 % BSA. Proteins from X. laevis liver were extracted with 5% (w/v) perchloric acid (Nicholas and Goodwin, 1982).

Acknowledgements WethankGenevieve Aimouzni, David J.Clark, Dan Lee and Caroline Schild for assistance in the preparation of experimental materials and for their advice and criticism.

References Albanese,I. and Weintraub,H. (1980) Nucleic Acids Res., 8, 2787-2805. Albright,S.C., Wiseman,J.M., Lange,R.A. and Garrard,W.T. (1980) J. Biol. Chem., 255, 3673-3684. Almouzni,G. and M6chali,M. (1988a) EMBO J., 7, 665-672. Almouzni,G. and M6chali,M. (1988b) EMBO J., 7, 4355-4365. Almouzni,G. and Wolffe,A.P. (1993) Exp. Cell Res., 205, 1-15. Almouzni,G., Mechali,M. and Wolffe,A.P. (1990a) EMBO J., 9, 573-582. Almouzni,G., Clark,D., Mechali,M. and Wolffe,A.P. (1990b) NucleicAcids Res., 18, 5767-5774.

Almouzni,G., Mdchali,M. and Wolffe,A.P. (1991) Mol. Cell. Biol., 11, 655-665. Banerjee,S. and Cantor,C.R. (1990) Mol. Cell. Biol., 10, 2863-2873. Brawley,J.V. and Martinsin,H.G. (1992) Biochemistry, 31, 264-370. Bustin,M., Soares,N., Landsman,D., Srikanta,T. and Collins,J.M. (1987) Nucleic Acids Res., 15, 3549-3561. Bustin,M., Lehn,D.A. and Landsman,D. (1990a) Biochim. Biophys. Acta, 1049, 231-243. Bustin,M., Crippa,M.P. and Pash,J.M. (1990b) J. Biol. Chem., 265, 20077-20080. Bustin,M., Becerra,P., Crippa,M.P., Lehn,D.A., Pash,J.M. and Schiloach,Y. (1991) Nucleic Acids Res., 19, 3115-3121. Camerini-Otero,R.D., Sollner-Webb,B. and Felsenfeld,G. (1976) Cell, 8, 333 -347.

Clark,D.J. and Felsenfeld,G. (1992) Cell, 71, 11-22.

3864

Crippa,M.P., Alfonso,P.J.

and Bustin,M. (1992) J.

Mol. Biol.,

228,

442-449.

Dilworth,S.M., Black,S.J. and Laskey,R.A. (1987) Cell, 51, 1009-1018. Dimitrov,S., Almouzni,G., Dasso,M. and Wolffe,A.P. (1993) Dev. Biol., in press.

Dong,F.

and van 10596-10600.

Holde,K.E. (1991) Proc. Natl Acad. Sci. USA, 88,

Dorbic,T. and Wittig,H. (1987) EMBO J., 6, 22393-22399. Drew,H.R. (1993)J. Mol. Biol., 230, 824-836. Druckmann,S., Mendelson,E., Landsman,D. and Bustin,M. (1986) Exp. Cell Res., 166, 486-496.

Einck,L. and Bustin,M. (1983) Proc. NatlAcad. Sci. USA, 80, 6735 -6739. Einck,L. and Bustin,M. (1985) Exp. Cell Res., 156, 295-310. Garel,A. and Axel,R. (1976) Proc. NatlAcad. Sci. USA, 73, 3966-3970. Gilkin,G.C., Ruberti,I. and Worcel,A. (1984) Cell, 37, 33-41. Grunstein,M. (1990) Trends Genet., 6, 395-400. 31, 7977-7988. Hansen,J.C. and Wolffe,A.P. (1992) Biochemistry,597-603. Hayes,J.J. and Wolffe,A. (1992) BioEssays, 14, Hayes,J.J., Clark,D.J. and Wolffe,A.P. (1991) Proc. NatlAcad. Sci. USA, 88, 6829-6833.

Hebbes,T.R., Thome,A.W. and Crane-Robinson,C. (1988) EMBO J., 7, 1395-1402.

Jackson,V. (1990) Biochemistry, 29, 719-731. Kamakaka,R.T. and Thomas,J.O. (1990) EMBO J., 9, 3997-4006. Kleinschmidt,J.A. and Steinkeisser,H. (1991) EMBO J., 10, 3043-3050. Kleinschmidt,J.A., Scheer,U., Dabauvalle,M.C., Bustin,M. and Franke,W.W. (1983) J. Cell Biol., 97, 838-848. Laemmli,U.K. (1970) Nature, 227, 680-685. Lam,B.S. and Carroll,D. (1983) J. Mol. Biol., 165, 567-585. Landsman,D. and Bustin,M. (1991) Mol. Cell. Biol., 11, 4483-4489. Landsman,D., Medelson,E., Druckmann,S. and Bustin,M. (1986) Exp. Cell Res., 163, 95-102.

Malik,N., Smulson,M. and Bustin,M. (1984) J. Biol. Chem., 259, 699-702. McGhee,J.D., Rau,D.C. and Felsenfeld,G. (1982) NucleicAcids Res., 10, 2007-2016.

Mardian,J.K.W., Paton,A.E., Bunick,G.L. and Olins,D.E. (1980) Science, 209, 1534-1536. Nicholas,R.H. and Goodwin,G.H. (1982) In Johns,E.W. (ed.), 7he HMG Chromosomal Proteins. Academic Press, pp. 41-68.

Peterson,R.C., Doening,J.L. and Brown,D.D. (1980) Cell, 20, 131-141. Pineiro,M., Gonzalez,P.J., Palacian,E. and Hernandez,F. (1992) Arch. Biochem. Biophys., 295, 115-119. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning, A Laboratory Manual, Second edition. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sandeen,G., Wood,W.I. and Felsenfeld,G. (1980) Nucleic Acids Res., 8, 3757-3778. Simpson,R.T. (1991) Progr. NucleicAcidRes. Mol. Biol., 40, 143-184. Smith,D.R., Jackson,I.J. and Brown,D.D. (1984) Cell, 37, 645-652. Tremethick,D.J. and Drew,H.R. (1993) J. Biol. Chem., 268, 11389-11393. Tremethick,D.J. and Frommer,M. (1992) J. Biol. Chem., 267, 15041 -15048. van Holde,K.E., Lohr,D.E. and Robert,C. (1992) J. Biol. Chem., 267, 2837-2840. Weintraub,H. and Groudine,M. (1976) Science, 193, 848-856. Weisbrod,S. and Weintraub,H. (1979) Proc. Natl Acad. Sci. USA, 76, 630-634. Weisbrod,S., Groudine,M. and Weintraub,H. (1980) Cell, 19, 289-301. Westermann,R. and Grossbach,U. (1984) Cell, 90, 355-365. Wolffe,A.P. (1988) EMBO J., 7, 1071-1079. Wolffe,A.P. (1989) EMBO J., 8, 527-537. Wolffe,A.P. (1991) J. Cell Sci., 99, 201-206. Wolffe,A.P. and Morse,R.H. (1990) J. Biol. Chem., 265, 4592-4599. Wolffe,A.P. and Schild,C. (1991) Methods Cell Biol., 36, 541-549. Received on April 27, 1993; revised on July 5, 1993