Jul 4, 1994 - other proteins which bend DNA (Goodman and Nash,. 1989 ...... expressing the IntW22 1 amber mutant; E.Davis and M.Brennan for yeast.
The EMBO Journal vol.13 no. 19 pp.4536-4548, 1994
Architectural elements in nucleoprotein complexes: interchangeability of specific and non-specific DNA binding proteins Anca M.Segaill, Steven D.Goodman2 and Howard A.Nash Laboratory of Molecular Biology, NIMH, Bethesda, MD 20892-4034, USA 'Present address: Department of Biology, San Diego State University, San Diego, CA 92182-0057, USA 2Present address: Department of Basic Sciences, University of Southern California, School of Dentistry, Los Angeles, CA 90089-0641, USA
Communicated by M.Hofnung
Integration host factor (IHF) is required in lambda site-specific recombination to deform the DNA substrates into conformations active for recombination. HU, a homolog of IHF, can also deform DNA but binds without any apparent sequence specificity. We demonstrate that HU can replace IHF by cooperating with the recombinase protein, integrase, to generate a stable and specific complex with electrophoretic mobility and biochemical activity very close to the complex formed by IHF and integrase. The eukaryotic HMG1 and HMG2 proteins differ entirely in structure from HU but they also bind DNA non-specifically and induce or stabilize deformed DNA. We show that the eukaryotic HMG1 and HMG2 proteins cooperate with integrase at least as well as does HU to make a defined structure. We also find that the eukaryotic core histone dimer H2A-H2B can replace IHF, suggesting that the histone dimer is functional outside the context of a nucleosome. HU and the HMG proteins not only contribute to the formation of stable complexes, but they can at least partially replace IHF for the integrative and excisive recombination reactions. These results, together with our analysis of nucleoprotein complexes made with damaged recombination sites, lead us to conclude that the cooperation between HU and integrase does not depend on protein-protein contacts. Rather, cooperation is manifested through building of higher order structures and depends on the capacity of the non-specific DNA binding proteins to bend DNA. While all these non-specific binding proteins appear to fulfil the same bending function, they do so with different efficiencies. This probably reflects subtle structural differences between the assembled complexes. Key words: histones/HMG box/HU/IHF/phage lambda site-specific recombination
Introduction Many DNA transactions require that DNA be deformed into specific conformations and this contortion is often
4536
assisted by proteins which act as architectural elements within a larger nucleoprotein array. The integration host factor (IHF) protein of Escherichia coli is one such bending protein. IHF was identified as an accessory factor in lambda site-specific recombination (Miller and Friedman, 1977; Williams et al., 1977) and has subsequently been shown to play roles in gene expression, transposition, replication and phage DNA packaging (reviewed by Friedman, 1988). The IHF protein bends dramatically the DNA to which it binds (reviewed by Nash, 1990). Its suggested role as a structural element (Craig and Nash, 1984; Richet et al., 1986) was confirmed when site-specific recombination reactions were made independent of IHF by replacing its binding site with either stably bent kinetoplast DNA or binding sites for other proteins which bend DNA (Goodman and Nash, 1989; Giese et al., 1992; Goodman et al., 1992; MolinaLopez et al., 1994; S.Austin, personal communication). The IHF protein is homologous to the HU protein of E.coli, and both proteins are ubiquitous among bacteria (reviewed by Drlica and Rouviere-Yaniv, 1987; Schmid, 1990). Like IHF, HU bends DNA, as demonstrated by its ability to facilitate closure of linear molecules too small to circularize in its absence (Hodges-Garcia et al., 1989). However, unlike IHF, which binds DNA in a sequencespecific manner, HU has no known preferred binding sequence. The similarity between IHF and HU at the amino acid level (50-62% identities and very conservative substitutions) suggested that the two proteins may be able to functionally substitute for one another. This prediction has been met to varying degrees in many systems, including replication initiating at the E.coli oriC locus (Skarstad et al., 1990; Hwang and Komberg, 1992), regulation of bacteriophage lambda replication (Mensa-Wilmot et al., 1989; L.Huang and R.McMacken, manuscript in preparation), TnJO transposition (Morisato and Kleckner, 1987), phage Mu transposition (Surette and Chaconas, 1989), phage lambda DNA packaging (Mendelson et al., 1991) and phage P1 DNA packaging (Skorupski et al., 1994). The ability of HU to replace IHF in lambda site-specific recombination was tested by Goodman et al. (1992), who showed that HU can support a substantial amount of excisive recombination but can barely support integrative recombination. In that study, the ability of stably bent A-tract DNA to replace both IHF and HU confirmed that the function shared by these proteins is to bend DNA. The molecular basis for the ability of a non-specific DNA binding protein to substitute for a sequence-specific DNA binding protein has not been established. Here we address this issue in the context of lambda site-specific recombination. Our data show that HU directly replaces IHF in protein-DNA complexes assembled in cooperation with integrase (Int) on recombination substrates. Several tests indicate that the cooperation between HU and Int
Architectural elements in nucleoprotein complexes
does not depend on specific contacts between these two proteins but represents cooperation through effects on higher order structure. This conclusion is strengthened further by our ability to replace IHF or HU with HMG 1 or HMG2 (henceforth referred to as HMG1/2, if both proteins were tested). Although these eukaryotic proteins differ entirely in structure from HU (Read et al., 1993; Weir et al., 1993), they have been shown to replace it in Salmonella Hin-mediated recombination (Paull et al., 1993) and to suppress growth defects of E.coli HU mutants (Megraw and Chae, 1993). We further investigate the degree of similarity between these proteins by using mixtures of IHF and either HU or HMG 1 to address the apparent obligatory requirement for IHF in integrative, as opposed to excisive, recombination.
P arm
P' arm
core
attL
attR -160
-1 3
-1 14
0
+12
+84
+66
Fig. 1. Schema of lambda attachment sites used in excision. Triangles represent the Int core-type binding sites. Arrows represent Int arm-type binding sites. Diamonds filled with 'H' represent IHF binding sites. The tall 'F' represents the Fis binding site, while the elongated octagon represents the Xis binding site. The coordinates indicated at the bottom are landmarks for some of the att site variants we refer to in the text.
Results Excisive recombination occurs between two attachment sites, known as attL and attR, and is mediated by the phage-encoded Int protein in conjunction with the accessory factors IHF, Xis and Fis. Excision, unlike integration, can take place to an appreciable extent in vivo, and in vitro without IHF if the host HU protein is present (Goodman et al., 1992). We investigated the molecular interactions which permit HU to substitute for IHF by using electrophoretic mobility shift assays (Fried and Crothers, 1981; Gamer and Revzin, 1981).
Interactions of HU with attL We first studied HU interactions with the simpler recombination partner, attL, the structure of which is shown in Figure 1. A single IHF binding site (H') is flanked by two types of binding sites for the Int recombinase: three tandem high-affinity 'arm' sites known as P'1, P'2 and P'3, and two low-affinity 'core' sites known as B and C' (Ross and Landy, 1982, 1983; Richet et al., 1988). Reactions contained a PCR-generated, radiolabeled DNA fragment with the attL site and a large (>1000-fold) excess of non-specific competitor DNA, as well as the Int and IHF or HU proteins. Figure 2 summarizes our initial observations. Int and IHF added individually to attL each generated characteristic complexes with the attL site (Figure 2, lanes 2 and 3; Segall and Nash, 1993). Int and IHF added together to attL generated a doublet (Figure 2, lane 4). As inferred by changing the concentration of Int protein in the reaction, the two complexes in the doublet reflect different numbers of Int protomers bound to the attL site (data not shown). Under the conditions of our experiment HU did not shift attL by itself (Figure 2, lane 5), but when Int and HU were added together to attL DNA they formed complexes with similar mobility to the complexes formed with Int and IHF (Figure 2, lane 6). We will henceforth refer to the complexes made from both Int and IHF or Int and HU as attL intasomes (Better et al., 1983). Is HU present in the attL intasomes and, if so, where? Three possibilities exist: (i) HU occupies the same space as IHF; (ii) HU binds in a different location to IHF; and (iii) HU does not remain in the complex once it allows arm-bound Int to make stable contacts in the core region. We investigated the presence of HU in intasomes by using
INT
+
-
+
-
IHF
-
+
+
-
*..,I
compiexes
a4tL in:asores
-
+
HU
B,molecu ar
+
__
*l
Free attL
-+_
Ii
Fig. 2. HU and IHF make similar complexes on attL in the presence of Int. Labeled 186 bp DNA containing the attL site is present in all lanes. In addition, pure Int (40 nM), IHF (34 nM) and HU (37 nM) proteins were added as indicated. Each reaction also contained 1.2 tg salmon sperm DNA (60 tg/ml). See Materials and methods for details, and Segall and Nash (1993) for a description of bimolecular complexes.
3H-labeled HU protein to measure its stoichiometry with respect to the DNA. The data show that HU is present in an -1:1 ratio to the attL DNA (Table I), even in the presence of excess non-specific competitor DNA and after PAGE. To infer where HU is located within the complex, we used an attL-containing 102 bp PCR fragment which excludes all sequences external to specific Int binding sites (attL coordinates - 11 to + 84; see Figure 1). Int and HU make intasome complexes efficiently with this minimal attL molecule (data not shown). Assuming that Int is occupying all its binding sites on the minimal attL fragment, the only naked DNA available for HU binding is 4537
A.M.Segall, S.D.Goodman and H.A.Nash Table I. Stoichiometry of HU in attL intasomes
Experimenta
HU (pmol)
DNA (pmol)
HU/DNA
1 2 3 3 3
0.089 0.074 0.470