ABSTRACT
Transcription factor IIIA (TFIIIA) and p43 zinc finger protein form distinct complexes with 5S ribosomal RNA in Xenopus oocytes. Additionally, TFIIIA binds the internal promoter of the 5S RNA gene and supports assembly of a transcription initiation complex. Both proteins have nine tandemly repeated zinc fingers with almost identical linker lengths between corresponding fingers, yet p43 has no detectable affinity for the 5S RNA gene. TFIIIA zinc fingers 1-3 are connected by highly conserved linkers, first identified in the Drosophila protein Krüppel, that are found in many DNA binding zinc finger proteins. To understand the role of these linkers in RNA and DNA binding we exchanged three TFIIIA linker amino acids with the equivalent amino acids from p43. The major effect of linker substitution is a 50-fold reduction in DNA specificity, concomitant with an 8-fold reduction in affinity. N-Terminal zinc fingers from either TFIIIA or p43 bind to multiple specific sites on 5S RNA that are resistant to competition by tRNA or poly(rA). This mode of RNA binding is unaffected by linker substitution. These data suggest that zinc finger linkers significantly facilitate the specificity of DNA binding.
Zinc finger proteins perform diverse biological functions in regulation of cell growth and differentiation through DNA-, RNA- and protein-protein interactions. Zinc fingers of the C2H2 class coordinate a single zinc atom through two invariant cysteines in an antiparallel [beta]-sheet and two histidines in an [alpha]-helix. Molecular determinants for DNA binding have been identified by biochemical experiments and by direct crystallographic observation of zinc finger-DNA complexes (1-3). Zinc finger proteins that mediate binding to RNA have also been identified (4-7). Recently the zinc finger protein WT-1, implicated in Wilm's tumor, has been identified in splicing complexes and is reported to bind RNA in vitro (8,9). The molecular determinants for interaction of individual zinc fingers from these proteins with RNA are not fully understood.
The archetypal C2H2 zinc finger protein is the Xenopus transcription factor TFIIIA (10,11). TFIIIA has nine tandemly repeated zinc fingers and a C-terminal domain required for transcription (12,13). TFIIIA binds both the 5S RNA gene promoter (internal control region, ICR) and, in developing oocytes, is complexed with 5S RNA in a 7S RNP (14). A related 5S rRNA binding protein, p43 (4), found complexed with 5S RNA in a 42S RNP (15), has identical zinc finger length and similar H/C linker length and appears to bind to the same region of 5S RNA (16). However, binding assays have failed to detect p43 binding to the 5S RNA gene (4). Furthermore, p43 has no sequence homology with TFIIIA in the C-terminal domain, suggesting that it is not a transcription factor for 5S RNA genes. Since p43 binds exclusively to 5S RNA, comparison of the effects of mutation in TFIIIA and p43 provides an excellent model system for investigating the differences in the molecular basis for RNA and DNA recognition.
Within TFIIIA, different groups of zinc fingers have differential binding affinities for RNA and DNA (17-19). The three N-terminal zinc fingers have the highest DNA affinity of any isolated group of three fingers from TFIIIA (19-22). These zinc fingers are uniquely connected by H/C linker sequences similar to a highly conserved linker sequence, TGEKPY, first identified in the Drosophila homeotic gene Krüppel and present in many DNA binding zinc finger proteins (23). Conservation of the Krüppel-like H/C linker sequence suggests that linkers play a role in enabling adjacent zinc fingers to bind DNA. Mutation of H/C linkers between N-terminal fingers of TFIIIA has been shown to decrease or abolish DNA affinity. Mutation of a single glycine to serine or lysine to glutamic acid between zinc fingers 1 and 2 in full-length TFIIIA results in a weakened TFIIIA footprint on the 5S RNA gene (24). Similar effects for linker mutagenesis, at most a 3-fold reduction in affinity, have been reported recently in extensive scanning mutagenesis of TFIIIA (25). In the context of three N-terminal zinc fingers, mutations in linkers can cause more severe disruption of DNA binding. Substitution of threonine with serine in linker 1 decreases DNA binding affinity 11-fold by increasing the dissociation rate (26). More extensive amino acid changes that substitute both H/C linkers in TFIIIA zinc fingers 1-3 with p43-derived linkers have been reported to completely abolish DNA binding activity (27,28).
The role of H/C linkers in RNA binding is less well understood. The TFIIIA and p43 zinc fingers are generally connected by linkers of similar length. However, TFIIIA zinc fingers 4-7 show the highest discrimination for RNA and these zinc fingers are connected by unusually short H/C linkers (19). In full-length TFIIIA, all zinc fingers contribute to overall 5S RNA binding affinity (17), with mutations in zinc finger 6 that disrupt finger structure having the most profound effects (18,19,29). Deletion analysis of p43 showed that N-terminal zinc fingers 1-4 and full-length p43 have comparable affinity for 5S RNA (17).
To understand the contribution of H/C linkers to specific RNA or DNA binding affinity we have compared the nucleic acid binding properties of N-terminal zinc fingers from TFIIIA and p43 when H/C linkers were exchanged. We find that mutation of H/C linkers has the greatest effect on DNA binding specificity with a smaller effect on affinity. In contrast, 5S RNA binding by three zinc fingers is independent of a unique linker sequence and occurs at multiple specific sites on 5S RNA.
TFIIIA cDNA from plasmid pTA102, kindly provided by Dr David Setzer, was tailed with a double-stranded DNA adapter oligonucleotide, 5'-d(AATTCCCGGGCTAGCTGACTAG) and 5'-d(GATCCTAGTCAGCTAGCCCGGG), and inserted via a BamHI linker into pGEM®5Zf(-) (Promega) to create pRR0. The cDNA was then altered by site-directed mutagenesis to introduce nucleotides encoding amino acids found at the equivalent position in p43 (30). The mutagenic oligonucleotides used were 5'-d(GCAGGCGCATCTGTGCAAACACTCAGAACAGAAACCATTTCCATG), to change TGE to SEQ in the first H/C linker, and 5'-d(CCCGCCACTCACTCACTCATCTTGCCTTTAAAAACTTCACATGTG), to change TGE to LAL in the second H/C linker (nucleotides that differ from wild-type are underlined and the restriction sites introduced are in bold). These oligonucleotides were used separately or in combination with single-stranded pRR0 DNA as a template to introduce linker mutations. The complementary strand was synthesized with T7 DNA polymerase (US Biochemical) (30). Plasmids containing mutated sequence were identified by the presence of either DdeI or DraI restriction sites, introduced by mutagenic oligonucleotides, or the absence of a XmnI site and confirmed by dideoxy sequencing. The full-length wild-type and mutated cDNAs were subcloned into the expression vector pET-15b (Novagen), which introduces an N-terminal polyhistidine tag and thrombin recognition site.
To express the N-terminal zinc fingers from TFIIIA, a DNA fragment encoding zinc fingers 1-3 was generated by PCR amplification of the full-length TFIIIA cDNA using the T7 primer [5'-d(TAATACGACTCACTATAGGG)], 5' of the methionine start in pET-15b, and a 3' primer corresponding to the end of zinc finger 3 [5'-d(GGCGCTAGCTCAGATCTTG-ATGTTATGGAA)]. Fragments were inserted into pET-15b using the 5' NdeI and 3' BamHI sites and sequenced. Similarly, DNA encoding the first three zinc fingers from p43 were amplified from the full-length p43 cDNA (4) by PCR using primers PZN1 [5'-d(CGGGATCCCATGAAAAACGTTGGT)] and PZC3 [5'-d(AGTGGATCCTATACGGCCTCCCCGT)]. The product was digested with BamHI, gel purified and subcloned into the BamHI site of pET15b. The H/C linker between p43 zinc fingers 1 and 2 was introduced into the p43 cDNA using the mutagenic oligonucleotide 5'-d(GGCTGGGCACACCGGTGAGAAG-CCGTGG) to generate pHPZ1. The mutation in the second H/C linker was generated by overlapping PCR. In one PCR a forward primer corresponding to the N-terminal end of finger 1, PZN1, was used with a reverse primer complementary to the H/C linker mutation, 5'-d(GAGTTTCTTCTCGCCGGTGTGTCGCTTC), with pHPZ1 as template. In a second PCR, also with pHPZ1 as template, a forward primer corresponding to the H/C linker mutation, 5'-d(GAAGCGACACACCGGCGAGAAGAAACTC), was used with a reverse primer to the C-terminal end of finger 3, PZC3. The products of these reactions were purified by agarose gel electrophoresis and used as templates in a third PCR with primers corresponding to the N-terminus of zinc finger 1 (PZN1) and C-terminus of zinc finger 3 (PZC3). Final PCR products were subcloned into the pET-15b BamHI site and sequenced.
Xenopus laevis rTFIIIA was expressed in Escherichia coli strain BL21(DE3) from plasmid pTA102 and purified by ammonium sulfate precipitation and Biorex70 ion exchange chromatography as described by Del Rio and Setzer (31). Three zinc finger polypeptides derived from TFIIIA or p43 were expressed from pET vectors as polyhistidine tagged molecules in E.coli strain BL21(DE3), which harbors an IPTG-inducible T7 RNA polymerase gene (32). Cells harboring the appropriate plasmid were grown to an OD600 of 0.4-0.6 in Rich broth supplemented with 100 µM zinc sulfate, 40 mM glucose and 200 µg/ml ampicillin. Protein expression was induced by addition of 0.5 mM IPTG for 3-4 h. Cells were harvested, lysed with lysozyme (150 µg/ml) and deoxycholate (0.05%) and disrupted by sonication. Inclusion bodies containing the fusion protein were precipitated by centrifugation and extracted overnight in 15 ml binding buffer (5 mM imidazole, 500 mM NaCl, 50 µM ZnSO4, 100 µM PMSF, 1 mM benzamidine, 20 mM Tris, pH 8.0) containing 6 M urea. The extract was clarified by centrifugation and passed over a 1-2 ml nickel-agarose column (HIS-Bind resin; Novagen) equilibrated with binding buffer containing 6 M urea. The column was washed with 10 column vol binding buffer containing 6 M urea and proteins eluted with 400 mM imidazole, 200 mM NaCl, 6 M urea, 50 µM ZnSO4, 8 mM Tris, pH 8.0 (elution buffer). Protein-containing fractions were analyzed by Tris-tricine gel electrophoresis (33) and peak fractions were pooled. Zinc fingers were renatured by step-wise dialysis from elution buffer containing 6 M urea to 5 mM MgCl2, 2.5 mM CaCl2, 150 mM NaCl, 50 µM ZnSO4, 10 mM DTT, 10% glycerol, 50 mM Tris, pH 8.0, prior to digestion. Proteins were digested for 2 h with 1-2 U thrombin (Calbiochem)/100 µg protein at 25°C and dialyzed at 4°C prior to storage in buffer containing 1 mM benzamidine at -80°C. Protein concentration was determined by BioRad protein assay. BSA was used as the standard and final protein concentrations were determined by multiplying the apparent concentration by 0.62 to correct for dye binding differences between BSA and TFIIIA (34). Amino acid compositions and concentrations of TFIIIA-derived protein fragments were confirmed by amino acid analysis.
The DNA substrate used was a 66 bp AvaI fragment containing the Xenopus borealis somatic 5S RNA gene internal control region (ICR) from bp 39 to 99 (17). DNA concentration was determined as OD260. Gel-purified DNA (2 µg) was end-labeled with [[alpha]-32P]dCTP using the Klenow fragment of DNA polymerase I and purified by spin column chromatography over Sephadex G-25 (Boehringer Mannheim).
Xenopus laevis oocyte-type 5S RNA was made by in vitro run-off transcription (T7 Megascript kit; Ambion) from plasmid pT7OD, containing a T7 promoter 5' of the RNA coding region and a DraI site at the 3'-end. The resultant RNA is 121 nt long and has one additional uridine at the 3'-end not found in native 5S RNA. Helix IV/V RNA was similarly generated by run-off transcription of plasmid pMC2, encoding 5S RNA nt 64-112, linearized with BsrGI. RNA was radiolabeled by inclusion of 100 µCi [[alpha]-32P]CTP (456 µM) in a 20 µl transcription reaction. 5S RNA was purified by denaturing electrophoresis through 12% polyacrylamide, 8 M urea, 40 mM Tris-acetate, pH 7.2. After elution from polyacrylamide gel slices using 40 mM Tris-acetate, 1 mM EDTA, pH 7.4, in DEPC-treated H2O, RNA was renatured by heating to 65°C for 15 min in 10 mM MgCl2, followed by slow cooling over 30 min to 25°C. RNA was equilibrated with DEPC H2O by spin column chromatography on Sephadex G-50 (Boehringer Mannheim).
Apparent equilibrium dissociation constants were determined by Scatchard analysis of bound and free nucleic acid (35). A fixed protein concentration (10-200 nM) was incubated with a range of nucleic acid concentrations (1-200 nM DNA, 0.6-20 nM RNA) in binding buffer (70 mM KCl, 1 mM MgCl2, 10 µM ZnSO4, 0.1% NP-40, 5% glycerol, 10 mM DTT, 10 µg/ml poly(dI·dC), 2 µg/ml poly(rA), 20 mM Tris, pH 8.0) in a total volume of 10 µl. Free and protein-bound DNA or 5S RNA were separated by non-denaturing gel electrophoresis through 5% polyacrylamide, 1% glycerol, 20 mM HEPES, pH 8.3, at 4°C, with the upper buffer chamber containing 2 mM DTT. Gels were pre-electrophoresed for 1 h at 23 V/cm and allowed to re-equilibrate to 4°C prior to loading. Samples were loaded with the gel running at 4.5 V/cm and switched to 25 V/cm for the duration of the run (45-90 s). Appropriate protein concentration ranges were determined empirically in preliminary experiments to give a broad range of data points in the Scatchard plot of bound/free nucleic acid against amount of nucleic acid bound. The quantity of radiolabeled substrate in free and bound fractions was determined by laser densitometric scanning of autoradiograms and/or by phosphorimage analysis of dried gels. Apparent Kd values for each protein were calculated by fitting data to a modified form of the Scatchard equation
Approximate ratios of affinity constants for the first (K1) and second (K2) shifted species were derived from the equation
To compare the influence of linker amino acids on RNA and DNA affinities of zinc fingers we substituted linker amino acids between zinc fingers from Xenopus TFIIIA, which bind DNA and RNA, with the corresponding amino acids from a related zinc finger protein, Xenopus p43, which binds exclusively to 5S RNA. Ideally, the influence of an individual linker is best studied in a two zinc finger fragment that is connected by a single linker. However, while two N-terminal zinc fingers are sufficient for 5S RNA binding, three are required for DNA binding (17). Therefore, to determine the effects of linker changes on RNA and DNA affinity of zinc fingers we substituted amino acids linking zinc fingers in the minimal three finger N-terminal fragment of TFIIIA that binds both DNA and RNA.
Three substitution mutants of TFIIIA zinc fingers 1-3 (tz1-3) were made by oligonucleotide-mediated site-directed mutagenesis of the cDNA. Proteins tz1-3.pl1, tz1-3.pl2 and tz1-3.pl1/2 have three linker amino acids substituted between zinc fingers 1 and 2, zinc fingers 2 and 3 and zinc fingers 1, 2 and 3 respectively with corresponding amino acids from p43. Similarly, H/C linkers between the three N-terminal zinc fingers of p43 were substituted with the TGE sequence from TFIIIA. Only three amino acids in the linker were changed, because the other amino acids in the KrYppel-like linker are either the same in TFIIIA and p43 (KP in linker 1 or K in linker 2) or else they are predicted to be part of the zinc finger [beta]-sheet from the Zif268 crystal structure and not part of the linker (2). Amino acid substitutions and protein nomenclature are described in Figure 1A.
Zinc finger proteins were expressed in E.coli BL21(DE3) cells with an N-terminal polyhistidine tag and purified to homogeneity by affinity chromatography over nickel-agarose. The N-terminal fusion was proteolytically cleaved from the zinc fingers with thrombin prior to use (Fig. 1B). All proteins contain three vector-derived amino acids at the N-terminus (Gly-Ser-His) after removal of the polyhistidine tag. The mobility of p43 fingers after cleavage with thrombin is significantly slower than TFIIIA fingers in this gel system (33). However, prior to cleavage all recombinant protein molecular weights were as predicted from conceptual translation.
Estimates for Kd were obtained from linear regression of Scatchard plots, since the determination of Kd from the slope of the line is independent of any accurate knowledge of protein concentration (35). This is particularly important for these experiments, because we know the fraction of active protein to be variable between mutant proteins and between preparations of the same protein. A linear Scatchard plot assumes equal binding affinity of all sites on a protein for ligand. Since this is unlikely in the cases where we detect multiple shifted species, we used high RNA concentrations that produce a low fraction of bound material to estimate Kd for the first stoichiometric binding constant (38).
Wild-type TFIIIA zinc fingers 1-3, tz1-3, bind DNA with a dissociation constant of ~3 nM (Fig. 2 and Table 1). Substitution of amino acids TGE (Thr-Gly-Glu) in the conserved Krüppel-like H/C linker, TGEKPF between zinc fingers 1 and 2, with amino acids SEQ (Ser-Glu-Gln) from p43 (tz1-3.pl1), reduced DNA affinity 5-fold. A similar reduction in affinity (3-fold) was observed when the second H/C link, TGE, was substituted with LAL (tz1-3.pl2). The DNA binding affinity of tz1-3.pl1/2, in which both linkers were substituted, was reduced 8-fold (Kd = 24 ± 4 nM) compared with the native TFIIIA fragment (tz1-3).
Substitution of the linker sequences caused a striking reduction in DNA binding specificity that is readily apparent from protein binding isotherms (Fig. 3). The DNA retardation assay using a wild-type TFIIIA fragment (tz1-3) generated a single discrete band representing the binary DNA-protein complex when as much as 90% of the RNA was bound (Fig. 3A). In contrast, the TFIIIA mutants with either one or both H/C linker substitutions generated multiple bands when 50% of the RNA was bound, suggestive of protein binding to multiple sites on the DNA fragment (Fig. 3B). To determine the relative affinity of the multiple species formed with the double linker mutant, complexes were titrated with a non-specific competitor DNA, poly(dI·dC) (Fig. 4). The difference in affinity of the higher order complexes is readily apparent from their titration at low, 10-fold, excess of non-specific competitor. The binary tz1-3-DNA complex is reduced by 50% with a 2000-fold base pair excess of poly(dI·dC), whereas the binary complex formed with the double linker substitution mutant is eliminated with a 500-fold base pair excess. The affinity of the binary complex is estimated to be 200-fold greater than the ternary complex for the wild-type zinc fingers, but only 4-fold for the mutant in which both H/C linkers had been changed to p43 sequences (Table 2; 37).
The three N-terminal zinc fingers from TFIIIA have a comparable affinity (within 5-fold) for both 5S RNA and the ICR DNA fragment under our assay conditions, in contrast to an earlier report of a 20-fold difference (19), but in agreement with Setzer et al. (39). One striking feature of RNA binding for all TFIIIA three zinc finger fragments is formation of multiple shifted species in an RNA mobility shift assay, suggesting that RNA binding by these zinc fingers is inherently less specific than DNA binding (Fig. 3). We measured dissociation constants at low protein excess (10-40% shifted substrate) to give an approximate affinity for the first 5S RNA binding site. 5S RNA affinity of wild-type TFIIIA and the three linker substitution mutants was comparable (Fig. 2 and Table 1), demonstrating that the p43 linker amino acids support RNA binding in the context of TFIIIA zinc fingers. Rates for association and dissociation of the TFIIIA fragments from 5S RNA, like DNA, were also faster than we could measure by gel retardation (data not shown).
Our findings demonstrate a difference in the contribution of H/C linkers to RNA and DNA binding affinity by two related zinc finger proteins. Although a reduction in DNA binding affinity is observed, qualitatively similar to previous reports, the major effect of linker mutation is a reduction in binding site specificity. RNA binding by the same N-terminal zinc fingers occurs at multiple specific sites. The affinity for these sites is unchanged by linker mutation.
Crystallographic and NMR studies of DNA-zinc finger protein complexes show that linkers are an extended and mobile connection between adjacent zinc fingers (2,40,41). It has been suggested that since no direct contacts with the DNA are made by the linker, the relative orientation of zinc fingers is determined by DNA contacts made with the [alpha]-helix (40). However, a potentially crucial role for linkers may exist in the light of the strong evolutionary conservation of H/C linkers in zinc fingers that bind DNA. N-Terminal TFIIIA fingers have a rod-like structure with extended linkers when free in solution (42). If TFIIIA binds DNA along the major groove, bending around the DNA axis, the necessary change in relative finger orientation may be permitted only by particular linker amino acids. The linker glycine residue has been shown to make an important contribution to TFIIIA DNA binding affinity, perhaps by providing sufficient flexibility to wrap in the DNA major groove, regardless of its position within the linker (24). TFIIIA zinc fingers 1-3, which provide much of the DNA binding affinity, are the only zinc fingers connected with a Krüppel-like linker. If N-terminal p43 zinc fingers interact with helical regions of 5S RNA they may not require the same degree of linker flexibility, since the Krüppel-like linker is absent.
We would like to thank Keith Joho for providing us with plasmid pMC2, David Setzer for plasmid pTA102 and Ken Lanclos and Qin Ryan for constructive criticism of the manuscript. This material is based upon work supported by the National Science Foundation under grant no. MCB-9206873.
Nucleic Acids Research
Pages
Introduction
Materials And Methods
Construction of TFIIIA and p43 mutants
Purification of zinc finger proteins
DNA and RNA substrates
Determination of dissociation constants
Results
Construction of H/C linker mutants
Mutation of the conserved H/C linker sequence reduces DNA affinity
Mutation of the H/C linker does not affect RNA binding
Discussion
Acknowledgements
References
by linear regression using Cricket Graph (Computer Associates) software (22,31,36). Standard errors of the mean were determined by combining values for individual experiments with correlation coefficients >0.7 derived from eight or more data points using SYSTAT (Systat Inc.) software.
[nucleic acidbound]/[nucleic acidfree]
= {-(1/Kd) × [nucleic acidbound]} + {[proteintotal]/Kd}
I0, I1 and I2 are the intensities of radioactive bands representing free nucleic acid, nucleic acid bound in the first complex and nucleic acid bound in the second complex respectively. The error inherent in this equation is (1 + K2/K1)2. The error is small when K1/K2 is large (e.g. when K1/K2 = 20 the error is 1%), but becomes much larger when the ratio of affinity constants is small (when K1/K2 = 2 the error = 56%).
I0I2/(I1)2 [approx] K2/K1, when K1 >> K2 (37).
REFERENCES
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 23 Jan 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. M. Babu, L. M. Iyer, S. Balaji, and L. Aravind The natural history of the WRKY-GCM1 zinc fingers and the relationship between transcription factors and transposons Nucleic Acids Res., December 2, 2006; 34(22): 6505 - 6520. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Jantz and J. M. Berg Reduction in DNA-binding affinity of Cys2His2 zinc finger proteins by linker phosphorylation PNAS, May 18, 2004; 101(20): 7589 - 7593. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||