Sp1 is phosphorylated and its DNA binding activity down-regulated upon terminal differentiation of the liver.

Using nuclear extracts prepared from rat liver it was demonstrated that binding of a transcription factor to site II of the D-site binding protein promoter could be induced by dephosphorylation of these extracts. Competition band shifts and supershift assays reveal this protein to be the general transcription factor Sp1. Phosphorylation of Sp1 appears to occur as a result of terminal differentiation of the liver. Proteins from both 1-day-old rat liver and adult liver undergoing regeneration have less of the phosphorylated form of Sp1 present with consequent increased DNA binding activity. Sp1 is similarly phosphorylated in brain, kidney, and spleen with phosphatase treatment of the extracts significantly increasing the level of DNA binding activity. Dephosphorylation of Sp1 results in a 10-fold increase in the affinity of Sp1 for its cognate site. Two-dimensional gel electrophoresis reveals that approximately 20% of the detectable protein appears to be in the phosphorylated form in adult liver extracts. Another protein with similar characteristics also appears to be present in the liver. Decreasing Sp1 DNA binding activity by phosphorylation may be an important mechanism for regulating gene expression, and possibly bringing about growth arrest during terminal differentiation.

Using nuclear extracts prepared from rat liver it was demonstrated that binding of a transcription factor to site II of the D-site binding protein promoter could be induced by dephosphorylation of these extracts. Competition band shifts and supershift assays reveal this protein to be the general transcription factor Sp1. Phosphorylation of Sp1 appears to occur as a result of terminal differentiation of the liver. Proteins from both 1-day-old rat liver and adult liver undergoing regeneration have less of the phosphorylated form of Sp1 present with consequent increased DNA binding activity. Sp1 is similarly phosphorylated in brain, kidney, and spleen with phosphatase treatment of the extracts significantly increasing the level of DNA binding activity. Dephosphorylation of Sp1 results in a 10-fold increase in the affinity of Sp1 for its cognate site. Two-dimensional gel electrophoresis reveals that approximately 20% of the detectable protein appears to be in the phosphorylated form in adult liver extracts. Another protein with similar characteristics also appears to be present in the liver. Decreasing Sp1 DNA binding activity by phosphorylation may be an important mechanism for regulating gene expression, and possibly bringing about growth arrest during terminal differentiation.
The regulation of terminal differentiation is a complex process involving both the induction of specific genes and growth arrest of the differentiating cells. There are several model systems where the expression of transcription factors involved in the induction of differentiation are regulated by factors directly linked to the control of the cell cycle (1,2). During myogenesis interaction between Myo-DI, and the retinoblastoma protein appears to be critical for bringing about, and maintaining, the terminally differentiated state (3)(4)(5). The plasticity of differentiation in the liver, characterized by its ability to undergo compensatory regenerative growth (6), provides an in vivo model for the linkage of growth and differentiation. Several liver enriched transcription factors including CAAT/enhancer-binding protein, liver activator protein, and DBP 1 are down-regulated in response to the induc-tion of growth and the loss of differentiation which occurs during liver regeneration (7,8). Recently, the regulation of the DBP gene, which encodes a member of the proline-and acidic rich domain subfamily (9) of basic/leucine zipper proteins, has pointed to another link between the cell cycle and differentiation. Factors binding to the retinoblastoma control element (RCE) (10) were found to be induced upon terminal differentiation and to bind to several sites (site I and III) within the DBP proximal promoter (11). In contrast, the binding of proteins to site II of the DBP promoter was found to be altered during differentiation. Site II binds at least two proteins, one of which does not change during differentiation. During the growth phase of regeneration the presence of a slowly migrating complex was noted which is not present in extracts from the normal liver. This pattern was paralleled by the presence of a similar, slowly migrating complex in extracts from the newborn liver. The reduction in DNA binding activity to site II in the adult extracts appears to involve the well studied transcription factor Sp1 and has suggested that Sp1 DNA binding activity is modified in response to differentiation.
Sp1 is a member of the C 2 -H 2 zinc finger family (12) and was one of the first transcription factors to be cloned (13). It is involved in the regulation of a wide variety of different genes, including the early promoter of SV40 (14), genes involved in proliferative response (15)(16)(17), extracelluar matrix protein genes (18,19), housekeeping genes (20 -22), and a number of growth factor genes (17,23). Sp1 has also been implicated in the function of the RCE (24,25) and cotransfection of a retinoblastoma gene expression vector is able to modulate the transactivation of responsive genes by Sp1. These findings have linked Sp1 function to that of the retinoblastoma protein, suggesting that Sp1 may be involved in modulating transcription in response to the growth state of the cell.
Two different types of post-translational modification of Sp1 are known to occur. Sp1 contains a number of O-linked N-acetylglucosamine residues clustered in the amino-terminal half of the protein (26). Glycosylation does not appear to affect the ability of this factor to bind DNA but the addition of soluble wheat germ agglutinin (WGA) does interfere with the transactivation of Sp1 responsive genes. Sp1 is also known to be phosphorylated by a DNA-dependent protein kinase which requires the presence of DNA ends for activity (27) and involves a large protein complex which includes the Ku antigen (28). The phosphorylation of Sp1 by DNA-dependent protein kinase does not appear to alter the DNA binding activity of this protein.
Correlation of the changes in occupation of the DBP site II with the differentiation state of the liver and the observation that this site can act as a weak transcriptional activator led us to examine the molecular basis of this phenomenon. In this paper we have been able to demonstrate that Sp1 binds to site II of the DBP proximal promoter. Changes in protein-DNA complexes present on this site during differentiation led to the discovery that Sp1 DNA binding activity is modulated by phosphorylation. In the normal liver Sp1 is phosphorylated result-ing in a decrease in its affinity for DNA. In de-differentiated tissues, such as regenerating or newborn rat liver, Sp1 is less phosphorylated and exhibits increased DNA binding activity. An additional factor, possibly Sp3, which also binds to an Sp1 recognition site but which is not recognized by an Sp1 antibody is also regulated in a similar manner. These results suggest that phosphorylation of Sp1, as well as closely related proteins, may be an important regulatory mechanism for controlling growth and differentiation in vivo.

MATERIALS AND METHODS
Liver Nuclear Extract Preparation-Nuclear extracts were prepared at a concentration of 5-10 g/l, according to Gorski et al. (29), with the modifications described in Maire et al. (30). The liver regeneration extracts were prepared by treating adult male rats with carbon tetrachloride and preparing extracts approximately 48 h later as described by Mueller et al. (7). The development extracts were prepared from both male and female rats 1 day after birth. Extracts from other tissues were prepared from adult rats.
Gel Mobility Shift Assays-Gel mobility shift assays were performed as described in Lichsteiner et al. (31) except that the binding reactions lasted for 15 min on ice before the samples were loaded on a 6% nondenaturing polyacrylamide gel. In the competition assays the unlabeled competitor oligonucleotide was added to the binding reaction at the same time as the labeled oligonucleotide probe for each site. The recombinant human Sp1 was obtained from Promega and 0.25 footprint units were used per binding reaction. The binding reactions for the supershift experiments were allowed to incubate on ice for 10 min before the indicated amount of antibody was added. The reaction was then incubated another 10 min on ice before the samples were loaded on a gel. The anti-Sp1 antibody is a rabbit affinity purified polyclonal antibody raised against residues 520 -538 of the human Sp1 protein, and was obtained from Santa Cruz Biotechnology, Inc. The oligonucleotides for DBP sites I, II, III, and IV and the RCE are described in Leggett and Mueller (11). Other oligonucleotides used in these assays are as follows: Sp-1, 5Ј-CTGCGGGGCGGGGCAGAC-CCCGCCCCGTCTGACG-5Ј; site C, 5Ј-GTAGGAACCAATGAAATGC-GAGGTAAGTATTTGGTTACTTTACGCTCCATTCATACATCC-5Ј.
Nuclear Extract Dephosphorylation-The dephosphorylation of extracts was carried out in 25 mM Hepes, pH 7.5, 34 mM KCl, 50 mM MgCl 2 at 30°C for 5 min, then 15 min on ice with calf intestinal alkaline phosphatase (1 unit/50 g of nuclear extract). The dephosphorylation reaction was stopped by the addition of a mixture of inhibitors to final concentrations of 10 mM NaF, 10 mM sodium vanadate, 10 mM potassium pyrophosphate, and 5 mM sodium phosphate. The inhibitors were added to the extract before the 30°C incubation for the mock dephosphorylation reactions.
Wheat Germ Agglutinin Selection-Normal liver nuclear extract was batch selected using wheat germ agglutinin linked to agarose beads (Sigma). The extract and WGA beads were incubated on ice for 10 min and then the WGA beads were washed with 12.5 mM Hepes, pH 7.6, 10% glycerol, 1 mM EDTA, 100 mM KCl, 5 mM MgCl 2 , 1 mM dithiothreitol, 0.1% Trasylol. The bound proteins were eluted by addition of the same buffer plus 0.3 M N-acetylglucosamine. K d Determination-The relative DNA binding affinities of nuclear extracts prepared from livers in different growth states were measured by saturation binding assays. Gel mobility shift assays were performed using a constant amount of nuclear extract and titrating with increasing amounts of Sp1 site probe. The bound and free probe were cut out of the dried gel and quantitated by scintillation counting. The amount of bound probe (fmol) was measured by calculating the ratio of bound/ free probe and multiplying this by the amount of probe added to each reaction. This value is plotted against the concentration of probe (nM) for each reaction to give the saturation curve for each assay. Scatchard plot analysis of this data was then done. The K d was determined from the slope of this graph in which the K d ϭ Ϫ1/slope. The amount of complex formed under saturating conditions (B max ) was calculated as the x intercept of the Scatchard plot. The three different extracts that were assayed were from normal liver, regenerating liver, and a normal liver extract that was dephosphorylated.
Two-dimensional Gel Analysis of Sp1 Modification-Isoelectric focusing of 40 g of rat liver nuclear protein was carried out according to manufacturer's instructions (Hoeffer) using vertical tube gels and an equal concentration of pH 8 -10 and pH 3-10 ampholytes (Bio-Rad). The second dimension consisted of a 7% SDS-polyacrylamide gel electrophoresis gel. Separated proteins were electroblotted onto nitrocellulose and the Sp1 protein detected using the Sp1 antibody, and a peroxidase-linked goat anti-rabbit IgG secondary antibody. The complexes were detected using the Renaissance chemiluminescence system according to manufacturer's instructions (DuPont NEN).

Factors Binding to Site II of the DBP Promoter Are Modified by Phosphorylation-
The binding of rat liver nuclear proteins to site II of the DBP promoter changes significantly depending upon the differentiation state of the liver from which they were extracted. In order to determine if post-translational modification, specifically phosphorylation was responsible for these changes, dephosphorylation of normal liver nuclear extracts using calf intestinal alkaline phosphatase was carried out. Band shift assays were then performed using all four of the DBP proximal promoter binding sites, the RCE (10) and the rat albumin site C (which binds the ubiquitous transcription factor NF-Y (31)). As can be seen in Fig. 1 dephosphorylation of these extracts resulted in the formation of a slowly migrating complex with site II but not with any of the other DBP sites or albumin site C. The inclusion of phosphatase inhibitors in the dephosphorylation reaction efficiently blocked the induction of this activity which indicates that this change was dependent on the action of the phosphatase. Quantitation of the site II complexes revealed that the faster migrating complex present on site II is not altered. It was the induction of a new, slowly migrating DNA binding activity rather than a shift in mobility of the existing factors which accounted for the observed change. These experiments demonstrate that a new complex with site II can be induced by the dephosphorylation of a factor present in normal liver extracts.
Sp1 Is the Phosphoprotein Which Binds to Site II-A weak complex of similar mobility to that observed with site II was also formed with the RCE oligonucleotide only in the dephosphorylated extracts (Fig. 1). In normal liver nuclear extracts, common factors bind to DBP sites I and III and the RCE and represent the majority of the DNA binding activity (11). However, the RCE is also known to bind Sp1 through a sequence which comprises the 5Ј-half of the RCE (24,25) and which is distinct from the elements described above. As the complexes formed with DBP sites I and III did not change in the dephosphorylated extracts it appeared that Sp1 might be the common dephosphorylated factor binding to site II and the RCE. Competition experiments with the two parts of the RCE site indicated that the Sp1 site within this FIG. 1. Dephosphorylation of nuclear extracts induces DNA binding activity for site II and the RCE. Band shift assays were carried out with liver nuclear extracts, extracts treated with alkaline phosphatase, or extracts with phosphatase and various phosphatase inhibitors, as indicated. Oligonucleotides from DBP sites I-IV, the RCE, or the rat albumin C site which binds NF-Y, were used as probes. DNA complexed with proteins are indicated (BOUND) as is the position of the free DNA (FREE). element could weakly compete for binding of the dephosphorylated factor to site II, although it did not compete for the other factor binding to site II (data not shown). Competition with a consensus Sp1 site also eliminated binding of the dephosphorylated factor to site II while having no effect on the binding of the other factor which interacts with site II (Fig. 2A). In order to provide additional evidence that it is Sp1 which is the phosphoprotein binding to site II, the ability of Sp1 to bind to a WGA column (26) was used to partially purify the observed activities. Selection of extracts with WGA enriches for the Sp1 site binding activity but does not select the other factor interacting with site II (Fig. 2B, NORMAL/WGA). The low mobility complex is observed with site II if this WGA selected material is dephosphorylated, or if it is first dephosphorylated and then selected on a WGA column. Furthermore, an identical complex capable of being selected on a WGA column is also observed with the consensus Sp1 oligonucleotide (Fig. 2B, left). These observations strongly suggest that Sp1 is the protein which binds to site II following phosphatase treatment.
Significantly less complex was formed between the dephosphorylated factor and the RCE in comparison to site II (Figs. 1 and 2A) suggesting that these sites might have different affinities for this factor. In order to compare their relative affinities, a band shift assay was carried out using dephosphorylated nuclear extracts selected on a WGA column with site II as the probe. Unlabeled DBP sites II and III, the RCE, and the Sp1 consensus site were then used as competitors for this activity. This revealed that site II and the consensus Sp1 site had comparable affinities with a maximum reduction in binding to 1 and 16%, respectively (Fig. 3A). The RCE was a very poor competitor in this assay with a maximal reduction in binding to only 75% of the starting value, only slightly better than the reduction to 91% with the unrelated site III. A similar experiment was carried out using purified Sp1 produced from a vaccinia virus expression vector, demonstrating that site II is a higher affinity binding site for Sp1 than the consensus Sp1 site (Fig. 3B). The RCE is a less efficient competitor than site II, with a maximal reduction in binding to 44%, however, it is a better competitor for recombinant Sp1 compared to the dephosphorylated WGA selected material suggesting that the WGA material may actually contain two different factors, one of which does not bind well to the RCE.
Two or More Sp1-like Phosphoproteins Bind to Site II-A polyclonal antibody recognizing a region of Sp1 near the zinc fingers (27) was used to supershift both normal and dephosphorylated proteins bound to site II and the Sp1 oligonucleotide. The antibody produced a slight supershift with the normal extracts and both probes reflecting the low level of unphosphorylated protein present in these extracts (Fig. 4, A and B). In contrast, approximately half of the complex formed with the dephosphorylated extracts was shifted by the antibody. The normal and supershifted complexes were essentially identical in mobility to those obtained with the recombinant Sp1 protein.
Despite the addition of increasing amounts of antibody and apparent saturation of the supershifted complex approximately half of the DNA binding activity was not supershifted. It appears that the original complex represents an unresolved doublet of two proteins both of which are activated by dephosphorylation and bind to Sp1 and site II oligonucleotides. This is consistent with the previous observation that the WGA selected material contains a factor which, unlike Sp1, does not bind to the RCE. The existence of at least two other Sp1 like genes (32), one of which can recognize Sp1 sites, supports this hypothesis. Preliminary evidence suggests that at least part of this complex is composed of Sp3. Sp1 DNA Binding Activity Is Modulated during Differentiation by Phosphorylation-The changes in occupation of site II observed during differentiation in the normal or regenerating liver suggest that Sp1, and possibly other related factors, are differentially modified such that they become phosphorylated

FIG. 2. An Sp1 oligonucleotide competes for proteins binding to site II and the RCE and this activity is selected by WGA.
A, band shift assays were carried out with the indicated probes and either normal or dephosphorylated nuclear extracts. Increasing amounts of unlabeled Sp1 oligonucleotide were added to the reactions as indicated. B, rat liver nuclear proteins were selected using WGA columns (Normal/WGA), followed by dephosphorylation (Dep/WGA), or proteins were first dephosphorylated and then selected with WGA (Dep/WGA). These proteins were used to carry out band shift assays with the Sp1 or site II probes as indicated. upon terminal differentiation and as a consequence bind less well to DNA. The Sp1 binding activity of normal adult liver extract was compared to that of extracts from regenerating and newborn rat livers (Fig. 5). The activity of the extracts was normalized using the DNA binding activity to an NF-Y site (albumin site C) which remains relatively constant during differentiation. Sp1 binding activity is significantly elevated in the de-differentiated extracts. Treatment with phosphatase reveals similar levels of Sp1 in all of the extracts but with differing levels of phosphorylation. Addition of Sp1 antibodies indicates that, as seen with the dephosphorylated normal extracts, approximately half of the activity is contributed by Sp1.
Sp1 from Rat Liver Exists in Several Different Forms-In order to determine if all of the Sp1 present in the liver is in a phosphorylated form two-dimensional electrophoresis was carried out using nuclear extracts from adult and 1-day-old rats, in addition to the dephosphorylated forms of each sample (Fig.  6). The first dimension was optimized to separate proteins with pI values in the pH range 7 to 10 and Sp1 was detected by Western blotting after separation by size. The full sized Sp1 protein exhibits relatively heterogeneous charge in the normal liver extracts, and a proportion of breakdown products of smaller size are visible. Upon dephosphorylation approximately 20% of the total Sp1 protein in adult extracts shifts to a more basic charge (RNLE verses RNLE deP). Dephosphorylation of extracts from 1-day-old liver results in the shift of 10% of the total protein (1 day verses 1 day deP). This indicates that in the de-differentiated liver only half of the phosphorylatable protein is phosphorylated. This is consistent with the increased binding in extracts from de-differentiated liver as well as phosphatase treatment of these extracts resulting in the additional induction of DNA binding activity (Fig. 5). Sp1 produced from a vaccinia vector migrates as a more discrete complex. These results indicate that approximately 20% of the Sp1 in the adult liver is phosphorylated, possibly at multiple sites.
Phosphorylation Affects the K d of Sp1 for Its Site-The ability of dephosphorylation to increase the binding of Sp1 to its cognate site indicates that this post-translational modification likely affects the dissociation constant (K d ) of Sp1. Band shift assays titrating the amount of Sp1 oligonucleotide against a constant amount of protein were used to derive saturation binding curves for Sp1 in its fully phosphorylated (RNLE), partially dephosphorylated (REGN), or completely dephosphorylated (RNLE deP) states (Fig. 7). Scatchard plots were then used to derive the K d and B max values for each of these conditions (Fig. 7, table). In normal extracts the K d of Sp1 for its consensus site is approximately 35 nM, which is in keeping with the low level of binding observed. Phosphatase treatment dramatically lowers the K d to 3.6 nM, close to 10-fold higher affinity. These differences represent the extreme case where all of the phosphate groups appear to have been removed. During de-differentiation the level of phosphorylation appears to represent an intermediate situation as demonstrated in the previous section and by the observation that phosphatase treatment can still increase the binding activity in regenerating or newborn extracts. The K d value for the regenerating extracts is 8.5 nM, intermediate between the values of the fully phosphorylated and dephosphorylated proteins. The B max values change from 12.3 fmol/g of nuclear protein in the normal extract to 23.7 and 30 fmol/g for the regenerating and dephosphorylated extracts, respectively. These results are consistent with the induction of 10 fmol/g of high affinity binding sites during the transition from adult liver to regenerating liver and from the regenerating to fully dephosphorylated state. The regenerating extract likely represents a mixture of both high and low affinity forms which are not resolved on the Scatchard plot. Overall it appears that in the adult liver approximately 20 fmol/g of Sp1 exists in a phosphorylated form with dephosphorylation resulting in a 10-fold increase in its affinity for its cognate site.
Sp1 Is Phosphorylated and Its DNA Binding Activity Reduced in Other Tissues-In order to determine if the observed changes in Sp1 binding are specific to the liver, band shift assays were also carried out with nuclear extracts from brain, kidney, and spleen. Low levels of Sp1 binding activity are present in all of the extracts, although slightly higher levels are seen in spleen extracts (Fig. 8). Phosphatase treatment in all FIG. 5. Sp1 phosphorylation is dependent on the differentiation state of the liver. Band shift assays were carried out with nuclear extracts from adult (NORMAL), regenerating (REGEN), or newborn (1 DAY) rat livers. The probe was the rat albumin C site (C) which recognizes NF-Y, or the Sp1 site (all others). Antibody to Sp1 was added in some reactions (Ab) and in others the extracts had first been dephosphorylated by the addition of alkaline phosphatase (DeP). The supershifted complex is indicated SUPERSHIFT.
FIG. 6. Characterization of Sp1 phosphorylation by two-dimensional electrophoresis. Samples of normal adult rat liver nuclear extracts (RNLE), extracts from 1-day-old animals (1 DAY) and recombinant protein (rSp1) and the dephosphorylated forms of each (deP) were separated by isoelectric focusing in the first dimension and by SDS-polyacrylamide gel electrophoresis in the second. Proteins were then transferred to nitrocellulose filters and Sp1 protein detected using the Sp1 antibody and chemiluminescence.
cases results in a significant increase in DNA binding activity. These results indicate that Sp1 is phosphorylated, with concomitant reduction in its DNA binding activity, in all tissues examined. The somewhat higher levels of Sp1 DNA binding activity associated with the spleen are consistent with the increased proportion of actively replicating cells present in that organ compared to the other tissues examined.

DISCUSSION
Proteins binding to site II of the DBP promoter are subject to differential regulation during terminal differentiation. In the normal liver a ubiquitous factor binds to site II, the binding activity of which does not appear to be altered during differentiation. Competition experiments with various oligonucleotides indicate that this factor is present in approximately equal quantities in the normal or regenerating liver (results not shown). In the de-differentiated liver Sp1 DNA binding activity is induced and binds to site II with an affinity comparable to a consensus Sp1 site and significantly higher than to the RCE. This is despite site II having more mismatches to the derived Sp1 consensus sequence than either the Sp1 site or the RCE. Additional experiments have indicated that Sp1 and the ubiquitous factor have overlapping but distinct recognition sites and that binding of these proteins is mutually exclusive. 2 Site II in normal liver nuclear extracts acts as a weak activator (11). De-differentiation results in the down-regulation of DBP tran-scription apparently through changes in occupation of the proximal promoter binding sites. This would suggest that Sp1 acts as a negative regulator of site II activity. The binding of Sp1 to site II may result in the displacement of the other factor and down-regulation of the transcriptional activity of this site. Similar interactions are seen in the human ␥-globin gene (33) where Sp1 competes with a factor called stage selector protein for occupation of several sites. The preferential binding of Sp1 is correlated with the inactivation of this promoter.
There appear to be at least two proteins or Sp1 isoforms present in normal nuclear extracts which can be dephosphorylated and which bind to Sp1 sites. The supershift experiments failed to shift a significant portion of the activated proteins present. The Sp1 antibody used was a polyclonal antiserum directed against a peptide corresponding to amino acids 520 to 538 of human Sp1 (14). Because it recognizes a restricted epitope it is possible that post-translational modification of Sp1 blocks the ability of this antibody to recognize some of the Sp1 present. Alternatively, two new genes related to Sp1 have been isolated, one of which also recognizes Sp1 sites (32). These genes, Sp2 and Sp3, have a low level of identity with the region of Sp1 used to produce the antibody (11/19 and 12/19 residues, respectively) and are likely not recognized in the supershift assay. In addition, the competition experiments with recombinant Sp1 and WGA selected proteins demonstrated that this other factor does not bind well to the RCE in comparison to Sp1. This further suggests that the binding activity observed corresponds to a distinct protein. Thus, another factor, which preliminary experiments suggest may be Sp3, is present in the liver and is phosphorylated upon terminal differentiation and while it recognizes Sp1 sites it appears to have some differences in its sequence specific recognition. Sp3 has recently been identified as a negative regulator of transcription (34) so that competition and differential regulation of Sp1 and Sp3 may be important for control of gene expression.
The phosphorylation of Sp1 during differentiation and the consequent decrease in its DNA binding activity suggests that this process is a significant regulatory mechanism during development. Sp1 has previously been shown to be phosphorylated by a DNA-dependent protein kinase (27,28), however, this modification does not effect the binding of Sp1 to a high affinity site. Differences in the DNA binding activity of Sp1 as assayed 2 K. A. Newcombe, V. Rigg, and C. R. Mueller, unpublished observations. using Southwestern blot analysis have been previously reported (35). Differences in DNA binding activity of a protein which has undergone denaturation and renaturation must be attributable to covalent modification rather than association of an inhibitory protein. Recently inhibitory proteins interacting with the amino-terminal part of Sp1 have been characterized (36,37). It is possible that the phosphorylation of Sp1 is linked to the binding of an inhibitory factor such as Sp1-I. The observation that significant amounts of non-phosphorylated Sp1 which does not appear to bind to DNA is present in the adult liver suggests that an additional level of control is also present.
The wide variety of genes which are regulated by Sp1 make the significance of its differentiation-induced phosphorylation and reduction in affinity particularly important. Many of the genes regulated by Sp1 are required for cell growth and as such the down-regulation of Sp1 binding activity upon terminal differentiation would reduce the expression of these genes and possibly help to bring about growth arrest. We have been able to demonstrate that the serine/threonine kinase, casein kinase II, is capable of phosphorylating the DNA binding domain of Sp1 and that this phosphorylation results in a reduction in Sp1 DNA binding activity. 3 It is possible that this kinase is responsible for the observed phosphorylation of Sp1 during development. Casein kinase II has been predicted to play a central role in the regulation of cellular processes (38,39). Its ability to down-regulate the activity of the AP-1 complex by phosphorylation of c-Jun, and the increase of c-Jun phosphorylation in quiescent cells (40) suggests that this kinase is critical for modulating cell growth. The concomitant need for growth arrest and differentiation suggests that increased phosphorylation of both c-Jun and Sp1 by casein kinase II is a critical step in bringing about terminal differentiation. The demonstration that a transgenic mouse carrying an extra copy of the casein kinase II ␣ subunit (which appears to disrupt casein kinase II function in some way) can produce a block to differentiation resulting in a form of lymphoma (41) gives further evidence that casein kinase II-mediated phosphorylation may play an important role in bringing about growth arrest and differentiation.
The retinoblastoma gene is known to be regulated by Sp1 (42) and its expression is down-regulated during differentiation (43). However, other Sp1 regulated genes carry out housekeeping functions and as such are required to be active in terminally differentiated tissues. The general regulation in Sp1 DNA binding activity would seem to preclude the expression of all Sp1 dependent genes in terminally differentiated tissues. Other transcription factors are induced upon terminal differentiation and could functionally compensate for the loss of Sp1 activity. Alternatively, promoters which remain active after terminal differentiation may have higher affinity binding sites for Sp1. This might be an intrinsic property of the site, as appears to be the case for the DBP site II, or protein-protein interactions could stabilize Sp1 binding. Given that the change in K d of Sp1 is approximately 10-fold upon phosphorylation, significant binding to promoters with high affinity sites would still be possible. Phosphorylation of c-Myb by casein kinase II has been shown to affect its binding to weak sites to a much greater degree than to strong sites (44). This could provide for differential regulation of genes where promoters with low, mid, and high affinity Sp1 sites are all active in replicating, undifferentiated cells. Differentiation would then result in the selective deactivation of promoters with lower affinity sites which, if these genes were required for growth would also bring about growth arrest. The high affinity promoters would remain active under these circumstances and continue to produce necessary proteins. This differentiation-induced phosphorylation of Sp1 does not appear to be unique to the liver as Sp1 activity can be induced by phosphatase treatment in all adult tissues examined. In addition, there does appear to be a correlation between the proportion of actively replicating cells in a given tissue and the amount of Sp1 activity present. These observations suggest that alterations in the phosphorylation state of Sp1 may be part of a ubiquitous mechanism linking the regulation of growth and differentiation.