The role of zinc fingers in transcriptional activation by transcription ...

Proc. Natl. Acad. Sci. USA Vol. 90, pp. 168-172, January 1993 Biochemistry

The role of zinc fingers in transcriptional activation by transcription factor IIIA (5S rRNA/Xenopus/DNA binding/RNA polymerase HI)

SAMUEL DEL RIO AND DAVID R. SETZER Department of Molecular Biology and Microbiology, Case Western Reserve University, School of Medicine, Cleveland, OH 44106

Communicated by Donald D. Brown, October 1, 1992

ABSTRACT We have described elsewhere a number of the properties of a set of mutant forms of Xenopus transcription factor IIIA (TFIIIA) containing single amino acid substitutions that result in the structural disruption of individual zinc finger domains. These "broken finger" proteins have now been analyzed with respect to their ability to support transcription of 5S rRNA genes in vitro. Disruption of any one of the first six zinc fingers of TFIIIA has no discernible effect on the activity of the protein in supporting 5S rRNA synthesis in standard in vitro transcription assays, despite the fact that some of these mutant proteins exhibit large decreases in their binding affinity for 5S rRNA genes in binary complexes. These results indicate that the activity of TFIIIA as a transcription factor can be largely independent of its equilibrium binding constant for the 5S rRNA gene in the absence of other components of the RNA polymerase III transcriptional apparatus. In fact, this finding is consistent with the known pathway and kinetics of assembly of 5S rRNA transcription complexes. In contrast to the results obtained with finger 1-6 mutants, analogous mutations in zinc fingers 7-9 of TFIIIA result in moderate to complete loss of transcriptional activity. We interpret these results to mean that the three C-terminal zinc fingers of TFIIIA are not only involved in binding to the internal control region of 5S rRNA genes but are also required, either directly or indirectly, for higher-order interactions that are important in transcription complex assembly, stability, or activity.

DNA-binding affinity will result in a lower transcriptional activity for the mutant protein. This expectation is reasonable, of course, provided that the occupancy of the target site by the transcription factor in a transcriptional activity assay is dependent on the affinity of the purified factor for the site. There are multiple circumstances, however, under which this condition will not hold; it should therefore not be surprising to discover instances in which there is little or no dependence of transcriptional activity on DNA-binding affinity, at least when the latter is measured in simple binary reaction mixtures containing only the transcription factor and an appropriate DNA fragment. As described in detail below, we believe for kinetic reasons that transcription of 5S rRNA genes under typical in vitro conditions constitutes one such example. We have recently constructed and analyzed a set of nine mutant forms of TFIIIA that we have called "broken finger" proteins (unpublished data). Each of these mutant forms of TFIIIA contains a single amino acid substitution in one of the nine zinc fingers of TFIIIA; specifically, each contains an asparagine substitution for the first of the two conserved histidine residues that serves as a Zn2+ ligand in the canonical C2H2 zinc finger motif (5). We have shown that these mutations lead to structural disruption of the single zinc finger domain containing the mutation with little or no effect on the structural integrity of other portions of the protein. By quantitatively analyzing the DNA-binding properties of these mutant TFIIIA molecules, we have been able to assess the energetic contribution of different fingers to the overall interaction of TFIIIA with the internal control region of 5S rRNA genes. It has also been possible to define sites of interaction of various fingers with short oligonucleotides within the internal control region by using footprinting methods to identify alterations in the pattern of protection afforded by each broken finger variant. Of particular relevance to the experiments of this paper, we have found that structural disruption of individual zinc fingers leads to reductions in binding affinity of ==2.5-fold to nearly 30-fold, depending on the identity of the mutated finger (Table 1). In the current study, we reasoned that these broken finger proteins could be used to assess the importance of DNAbinding affinity in mediating TFIIIA-dependent transcription of 5S rRNA genes. In fact, we have found no correlation of transcriptional activity with the DNA-binding affinity of TFIIIA in simple binary complexes and that the lowestaffinity mutants are indistinguishable from wild-type TFIIIA in standard in vitro transcription assays. This result can be understood in terms of a well-supported model for the pathway and kinetics of 5S rRNA transcription complex assembly (6-8). Interestingly, mutations in the three C-terminal zinc fingers of TFIIIA do lead to a reduction in the ability of TFIIIA to support transcription of 5S rRNA genes, and we argue that this results from a functional role of these

Since the landmark experiments of Brent and Ptashne (1), the modular structure of sequence-specific eukaryotic transcription factors has become a central feature of most models of the eukaryotic transcriptional apparatus. This point of view has been repeatedly validated by the ability to structurally and functionally define discrete and independent domains for DNA binding and transcriptional activation in many transcription factors. Transcriptional activation domains are generally presumed to be sites of interaction with other proteins and capable of acting to promote transcription initiation when fused to virtually any DNA-binding domain. While most of these studies have been carried out with sequence-specific class II factors-that is, those that activate transcription by RNA polymerase II-some data suggest that functionally analogous domains exist in the class I factor UBF (2) and in the 5S rRNA gene-specific class III factor transcription factor IIIA (TFIIIA) (3). In the case of TFIIIA, =:z75% of the protein consists of the nine consecutive zinc finger domains (4) shown to contain the DNA-binding activity of the factor, while a shorter sequence near the C terminus of TFIIIA has been demonstrated to be required for transcriptional activation (3). Despite the apparent separation of DNA-binding and transcriptional activation functions in most transcription factors, it is often implicitly assumed that mutations that reduce The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviation: TFIIIA, transcription factor IIIA.

168

Biochemistry: Del Rio and Setzer

Proc. Natl. Acad. Sci. USA 90 (1993)

Table 1. Effects of zinc finger disruptions on DNA-binding affinity of TFIIIA Zinc finger Kd, Relative decrease Mutant nM in affinity repeat Wild type 0.42 1.0 H33N H63N H93N H125N

H155N H183N H210N H241N H272N

1 2 3 4

5 6 7 8 9

1.1 2.9 11.2 4.9 3.6 1.5 2.6 2.9 1.1

2.6 7.0 27 12 8.6 3.6 6.2 7.0 2.7

zinc fingers in higher-order complex formation. Thus, three of the nine zinc fingers of TFIIIA are involved not only in DNA binding but also either directly or indirectly in proteinprotein interactions that are important in the formation of active, stable 5S rRNA transcription complexes.

MATERIALS AND METHODS TFIIIA. Wild-type and mutant forms of TFIIIA were produced in Escherichia coli and purified as described (9). The behavior of the mutant forms of TFIIIA was identical to that of wild type, except on the phenyl-Superose column; on this column, every mutant protein eluted at a unique concentration of (NH4)2SO4, which was in every case lower than that observed for the wild-type protein. We will discuss the structural implications of this observation elsewhere. The concentration of active TFIIIA was determined by Scatchard analysis of DNA-binding data using an electrophoretic mobility-shift assay. Binding reactions were carried out in 20 mM Tris-HCI, pH 7.5/70 mM KCI/7 mM MgCl2/10 pM ZnCI2/1 mM dithiothreitol/bovine serum albumin (100 gg/ ml)/poly[d(IC)] (10 gg/ml)/pepstatin A (1 Ag/ml)/leupeptin (1 Ag/ml)/0.5 mM phenylmethylsulfonyl fluoride/10%o (vol/ vol) glycerol. In Vitro Transcription Assays. The ability of wild-type and mutant forms of TFIIIA to support transcription of a 5S rRNA gene was assessed by using a TFIIIA-dependent in vitro transcription assay as described (9). pC9Xbs201, containing a single copy of the Xenopus borealis somatic-type 5S rRNA gene, was used as the template. Template DNA at a concentration of 2.5 nM was preincubated at 250C for 60 min in a 20-,lI reaction mixture containing 20 mM Tris HCl (pH 7.5), 7 mM MgCI2, 10 ,M ZnC12, 1 mM dithiothreitol, l1o glycerol, 70 mM KCI, bovine serum albumin (100,ug/ml), 20 units of RNasin (Promega), pepstatin A (1 Ag/ml), leupeptin (1 gg/ml), 0.5 mM phenylmethylsulfonyl fluoride, 5 Al of a Dignam nuclear extract (10) prepared from a Xenopus laevis kidney cell line, and variable concentrations of wild-type or mutant forms of TFIIIA. The concentration of TFIIIA was based on its stoichiometric activity in a DNA-binding assay as described above. This allowed us to circumvent any problems that might exist with respect to differential recovery of activity in different protein preparations. The same preparation of nuclear extract was used throughout and exhibits a low level of transcription activity that is independent of added TFIIIA. The amount of 5S rRNA synthesized in each experiment was normalized to this endogenous level of 5S rRNA synthetic activity. After preincubation, 10 gCi of [a-32P]UTP (800 Ci/mmol; 1 Ci = 37 GBq; DuPont/NEN) was added along with ATP, GTP, and CTP to a final concentration of 500 ,uM and UTP was added to a final concentration of 50 AM. Reaction mixtures were incubated for 60 min more at 25°C, and transcription was stopped by the

169

addition of 150 ,1l of a solution of 50 mM Tris HCl, pH 8.0/150 mM NaCl/5 mM EDTA, 0.5% SDS/glycogen (50 ,g/ml). After extraction and precipitation, reaction products were resolved on a 10% polyacrylamide (0.5% bisacrylamide)/8.33 M urea/90 mM Tris borate/2 mM EDTA gel. Wet gels were fixed and scanned on an AMBIS Systems (San Diego) radioanalytic imaging system. Ambis version 3.02 software was used for quantification. It may be noted (see Figs. 1 and 2) that transcriptional activity reaches a maximum at substoichiometric concentrations of TFIIIA relative to the DNA template. This may result from the presence of subsaturating amounts of one or more other transcription factors at higher TFIIIA concentrations, from a reduction in the concentration of template DNA available for transcription complex formation because of the presence of inhibitors in the relatively crude nuclear extract, or from a combination of these factors. In either case, the interpretation of our results with respect to TFIIIA activity should be unaffected.

RESULTS We will describe elsewhere the quantitative analysis of DNA-binding properties of mutant forms of TFIIIA containing structural disruptions of individual zinc finger domains. The equilibrium constants for simple binary reactions involving the internal control region of the 5S rRNA gene and each of these mutant proteins or wild-type TFIIIA are provided in Table 1. As we will discuss in detail elsewhere, each of the nine zinc fingers of TFIIIA appears to contribute to the binding energy when the protein interacts with the internal control region of the 5S rRNA gene, but the energetic contribution made by each finger is quite variable. Reductions in binding affinity range from =2.5-fold for disruption of finger 1 or 9 to nearly 30-fold for an equivalent mutation in finger 3. We reasoned that we should be able to use this collection of mutants to assess the importance of DNA-binding affinity, as measured in simple binary reaction mixtures containing only the DNA substrate and TFIIIA, in mediating transcriptional activation of 5S rRNA genes in vitro. Accordingly, we have measured the activity of wild-type and mutant forms of TFIIIA in a TFIIIA-dependent transcription assay in which other required factors are provided in the form of a nuclear extract from a Xenopus cell line. The results of such an analysis for wild-type TFIIIA and variant proteins with mutations in fingers 1-6 are shown in Fig. 1. Despite the substantial decrease in binding affinity for a 5S rRNA gene exhibited by some ofthe mutants [e.g., His-93 to Asn (H93N) and H125N] relative to that observed for wild-type TFIIIA, there is no discernible difference in the ability of these proteins to support 5S rRNA synthesis in a standard in vitro transcription assay. Thus, it would appear that there is no correlation between DNA-binding affinity and the ability of TFIIIA to support transcription of a 5S rRNA gene. In fact, as we argue below, this result is entirely consistent with the known pathway and kinetics of 5S rRNA transcription complex assembly and provides strong support for the conclusion that zinc fingers 1-6 of TFIIIA are required only for sequence-specific recognition of DNA and nucleation of transcription complex formation. These domains of the protein do not play a role in higher-order interactions leading to active transcription complex assembly. In contrast, structural disruption of zinc fingers 7-9 through analogous mutagenic lesions results in significant impairment of the ability of the resulting proteins to support 5S rRNA synthesis (Fig. 2). Interestingly, the three mutant proteins exhibit varying degrees of transcriptional activity, all significantly lower than that of wild-type TFIIIA. Disruption of zinc finger 7 results in a significant but modest reduction in transcriptional activity, while a similar mutation in zinc finger 9 causes a severe loss of activity and zinc finger

Proc. Natl. Acad Sci. USA 90 (1993)

Biochemistry: Del Rio and Setzer

170

A 5 0

0)

0

z

4~_ 3 /-

A

A

4

'C

a

A A

CD, ._

cr

2 11 I

0

0

0.2 0.4 0.6 0.8

1.2 1.4 1.6 [protein] (nM)

1

1.8

A

2

2.2

2.4

0

0.2 0.4 0.6 0.8

1.2 1.4 1.6 1.8 [protein] (nM)

1

2

2.2 2.4

FIG. 2. Transcriptional activity of wild-type and broken finger variants of TFIIIA containing mutations in zinc fingers 7-9. Reactions were carried out as described in Materials and Methods and the legend to Fig. 1. Results with wild-type TFIIIA (*), H21ON (4), H241N (m), and H272N (o) are shown.

9 also adversely affect higher-order interactions important in 5S rRNA transcription complex formation.

DISCUSSION

0

0.2 0.4 0.6 0.8

1.2 1.4 1.6 1.8 [protein] (nM)

1

2

2.2 2.4

FIG. 1. Transcriptional activity of wild-type and broken finger variants of TFIIIA containing mutations in zinc fingers 1-6. A constant concentration of a 5S rRNA gene-containing plasmid (pC9Xbs2Ol) was incubated with a TFIIIA-deficient extract and variable concentrations of wild-type or mutant TFII1A for 60 min. Nucleoside triphosphates were then added and the incubation continued for 60 min more. Labeled RNA products were isolated and resolved on a denaturing polyacrylamide gel, after which 5S rRNA synthesis was quantified with an AMBIS Systems radioanalytic imaging system. (A) Results with wild-type TFIIIA (*), H33N (A), H63N (i), and H93N (e) are shown. (B) Results with wild-type T-FIIIA (*), H125N (A), H155N (a), and H183N (e) are shown.

8 disruption yields a protein that is completely inactive in mediating 5S rRNA gene transcription. Loss of or reduction in transcriptional activity in these cases cannot result simply from reduced binding affinity of the proteins for DNA, since each of these mutants displays a greater affinity for 5S rRNA genes than is found for proteins with mutations in fingers 4, 5, or 6 (Table 1), all of which exhibit wild-type activity in a transcription assay. Furthermore, the lower transcriptional activity cannot result from anomalously low recovery of active protein from E. coli in the cases of mutants H210N, H241N, and H272N, since the concentration of protein was determined in each case from Scatchard analysis of DNAbinding data. Thus, these three proteins all exhibit a reduction in their ability to support 5S rRNA synthesis relative to their activity in a simple DNA-binding assay. We conclude that, in addition to their effects on DNA binding (Table 1; unpublished data), mutations that disrupt zinc fingers 7, 8, or

Although perhaps superficially surprising, the finding that the activity of TFIIIA in standard in vitro transcription assays is independent of its DNA-binding affinity as determined in simple binary reactions with the DNA template is actually consistent with the previously characterized pathway and kinetics of 5S rRNA transcription complex assembly (6-8). A highly simplified representation of that pathway is shown in Fig. 3. There are several features of this pathway that are relevant to the current discussion. First, transcription complex formation is initiated by the reversible binding ofTFIIIA to the 5S rRNA gene. [Although TFIIIC can also bind first (6), it does so with relatively low affinity, suggesting that the TFIIIA-first pathway predominates (6, 7).] The equilibrium constant for this step, in the absence of other components of the transcriptional apparatus, is the value that has been previously determined for wild-type and mutant forms of TFIIIA (9) and that is presented in Table 1. Second, one or more subsequent steps in complex assembly, probably including sequential interaction of TFIIIC and TFIIIB with the TFIIIA-DNA complex (6, 7), are essentially irreversible; that is, the complexes formed as a result of these steps are extremely stable and do not measurably dissociate or funcKd

DNA + TFIIIA

[DNA-TFIIIA] TFIIIC

k-2 lk2

(k2k>>k2)

[stable complex 11

TFIIIB

kI3tIk3

(k3>>k43)

[stable complex 11 + RNA Polymerase

(k2,k4,k5

>>k3)

Ill

Ikr SS rRNA

FIG. 3. Simplified kinetic scheme for 5S rRNA transcription complex formation and 5S rRNA synthesis.

Biochemistry: Del Rio and Setzer tionally decay even during prolonged periods of incubation involving multiple rounds of transcription (6, 7, 11). Third, the ultimate transcriptional activity of the template is dependent on the concentration of the stable complexes formed as a consequence of TFIIIC and TFIIIB activity. These features of the pathway lead to a clear prediction: so long as transcription complex formation is not kinetically limited-that is, so long as the first of the irreversible steps in complex assembly is allowed to proceed to completion-the ultimate rate of 5S rRNA synthesis will be essentially independent of the binding constant governing the TFIIIA-5S rRNA gene equilibrium. This conclusion follows from the fact that later, irreversible steps in the pathway deplete the pool of TFIIIADNA binary complexes, resulting in the stoichiometric conversion of templates into higher-order complexes independent of the equilibrium constant governing formation of binary TFIIIA-DNA complexes, up to the capacity of the extract being used as a source of other transcription factors. Thus, the concentration of these higher-order complexes, and therefore the measured transcription activity, will simply reflect a stoichiometric measure of the TFIIIA concentration and not its affinity for the 5S rRNA gene in a binary complex. A key question, therefore, is "Is the formation of higherorder stable complexes kinetically litrhited in these in vitro reactions?" In fact, we can be quite confident that it is not. In the experiments of Figs. 1 and 2, reaction mixtures were preincubated for 60 min to allow transcription complex formation to occur before the concentration of such complexes was determined by measuring 5S rRNA synthesis during a subsequent 60-min incubation. The prolonged preincubation period eliminates the lag associated with 5S rRNA synthesis that results from the relatively slow-acting activity of TFIIIB (6, 8). It is most unlikely, however, that decreasing the preincubation period would lead to a kinetic limitation in the formation of stable higher-order complexes. As originally shown by Lassar et al. (7), formation of stable (that is, functionally irreversible) complexes is conferred by a combination of TFIIIA and TFIIIC, both of which act quite rapidly relative to TFIIIB (6). Consequently, not only were the experiments of Figs. 1 and 2 carried out under conditions in which formation of stable higher-order complexes was not kinetically limited, but it would, in fact, be difficult to design an experiment in which such a kinetic limitation existed, at least if the experiment depended on the synthesis of 5S rRNA as an ultimate assay. In any event, the result of Fig. 1-namely, that 5S rRNA synthesis activity is independent of the equilibrium constant governing formation of TFIIIA-DNA binary complexes-is in complete accord with predictions made from the known properties of transcription complexes and their assembly. An additional conclusion becomes apparent from the preceding analysis. Specifically, structural disruptions of zinc fingers 1-6 of TFIIIA affect only the ability of the protein to form complexes with the DNA substrate and have no effect on other aspects of transcription complex assembly or stability that ultimately determine the activity of the template in 5S rRNA synthesis. This conclusion is consistent with the widely held view that the DNA-binding and transcriptional activation domains of transcription factors are functionally independent. A corollary of this reasoning is that any reduction in transcriptional activity associated with a particular mutation would indicate that the mutant protein is altered in interactions beyond those important in binary complex formation. Such a result was obtained for TFIIIA variants with structural disruptions of zinc fingers 7, 8, or 9. We are therefore led to the surprising conclusion that, in addition to their roles in binding to the internal control region of the 5S rRNA gene (Table 1; unpublished data), these domains of TFIIIA must also be involved in higher-order interactions in the transcription complex. We can imagine two mechanisms to account

Proc. Natl. Acad. Sci. USA 90 (1993)

171

for such a role. First, these zinc finger domains may be directly involved in protein-protein interactions [or proteinRNA interactions, given evidence that an RNA component plays a role in polymerase III transcription, at least in Bombyx (12)] that are important in transcription complex assembly, stability, or activity. Second, it is possible that these zinc finger domains are not directly involved in proteinprotein interactions but instead are responsible for precisely positioning the C-terminal, nonfinger domain of TFIIIA so that it can make appropriate contacts with other components of the transcriptional apparatus. This domain has previously been shown to be required for efficient 5S rRNA synthesis (3). Unfortunately, it is not currently possible to rigorously distinguish between these two possibilities. We do note that footprinting analyses with these mutant proteins demonstrate that the protection pattern afforded by H21ON at the very 5' end of the internal control region is essentially equivalent to that seen with wild-type TFIIIA and that alterations in the footprint pattern are found more 3'-ward (unpublished data). This result would suggest that DNA contacts at the very 5' end of the control region, which would presumably be important in correct positioning of the C-terminal, nonfinger domain of TFIIIA, are unaffected in H21ON and that these data would therefore argue for direct protein-protein contacts by finger 7. We do not consider these data conclusive, however, because of the potentially low resolution of DNase I as a footprinting probe and because of the fact that H21ON exhibits only a moderate reduction in transcriptional activity. Unfortunately, the protection pattern at the very 5' end of the control region is altered with H241N and H272N, thereby making the footprinting analysis uninformative with respect to this particular issue. Regardless of the precise mechanism, however, it is clear that the three C-terminal zinc fingers of TFIIIA are either directly or indirectly involved in higherorder interactions in the 5S rRNA transcription complex and that no clear distinction can be made in this region of the protein between functionally defined DNA-binding and transcriptional activation domains. What do these data tell us about broader issues of eukaryotic transcriptional mechanisms? First, it is important to note that the stability of intermediates in the pathway of transcription complex assembly has not been carefully studied for eukaryotic RNA polymerases other than RNA polymerase III. It is therefore possible that the affinity of DNA-protein interactions occurring as upstream steps in transcription complex assembly may be generally unimportant in measurements of the ultimate transcriptional activity of a gene, as we have found to be the case for the TFIIIA-5S rRNA gene interaction. Second, a potential role for the zinc finger structural motif in protein-protein or higher-order interactions beyond those involved in direct DNA recognition should not be excluded at the present time; the three C-terminal zinc fingers of TFIIIA appear to play such a role, although we cannot exclude the possibility that the mechanism of action is an indirect one. Given the widespread occurrence of zinc finger domains in many proteins of undemonstrated function, it is possible that similar results remain to be discovered elsewhere. We thank Pieter deHaseth for comments on the manuscript. S.D.R. was supported in part by a Medical Scientist Program training grant from the National Institutes of Health and by a Predoctoral Minority Fellowship from the American Society for Microbiology. This work was supported by Grant DMB 88-19286 from the National Science Foundation and by Grant GM-48035 from the National Institute of General Medical Sciences. 1. Brent, R. & Ptashne, M. (1985) Cell 43, 729-736. 2. Jantzen, H.-M., Admon, A., Bell, S. P. & Tjian, R. (1990) Nature (London) 344, 830-836.

172

Biochemistry: Del Rio and Setzer

3. Smith, D. R., Jackson, 1. J. & Brown, D. D. (1984) Cell 37, 645-652. 4. Miller, J., McLachlan, A. D. & Klug, A. (1985) EMBO J. 4, 1609-1614. 5. Diakun, G. P., Fairall, L. & Klug, A. (1986) Nature (London) 324, 698-699. 6. Setzer, D. R. & Brown, D. D. (1985) J. Biol. Chem. 260, 2483-2492. 7. Lassar, A. B., Martin, P. L. & Roeder, R. G. (1983) Science 222, 740-748.

Proc. Natl. Acad. Sci. USA 90 (1993) 8. Bieker, J. J., Martin, P. L. & Roeder, R. G. (1985) Cell 40, 119-127. 9. Del Rio, S. & Setzer, D. R. (1991) Nucleic Acids Res. 19, 6197-6203. 10. Dignam, J. D., Martin, P. L., Shastry, B. S. & Roeder, R. G. (1983) Methods Enzymol. 101, 582-598. 11. Bogenhagen, D. F., Wormington, W. M. & Brown, D. D. (1982) Cell 28, 413-421. 12. Young, L. S., Dunstan, H. M., Witte, P. R., Smith, T. P., Ottonello, S. & Sprague, K. U. (1991) Science 252, 542-546.