The E. coli
nucleoid undergoes major conformational changes as cells progress through a period of rapid growth, exhaust their supply of nutrients and enter stationary phase (Kim et al., 2004
). These changes are driven by fluctuations in the available pool of nucleoid-associated proteins (Ohniwa et al., 2006
). Thus, the major growth phase nucleoid protein Fis is replaced by Dps and CbpA in stationary phase. This is significant because Fis binds to specific DNA target sites while Dps and CbpA bind DNA with little or no sequence specificity. Here we have investigated the dimerization, DNA binding and aggregation properties of CbpA. While this manuscript was in preparation, a structure for residues 200–302 of Klebsiella pneumonia
CbpA was deposited in the Protein Data Bank (code 3I38, Midwest Center for Structural Genomics, unpublished). This region corresponds to amino acids 201–303 of E. coli
CbpA, essentially comprising an isolated CTD II domain. From an alignment of the two homologues (), it can be seen that the Klebsiella
CbpA CTD II structure is an excellent model for the E. coli
protein, with 81% sequence identity between the two molecules in this region. Importantly, none of the variable residues between the Klebsiella
and E. coli
proteins are located at the dimerization interface. The side chains of W287 and L290 protrude from the same side of an α helix (), and form a hydrophobic surface that locates to the twofold symmetry axis of the dimer (). Hydrophilic residues in the α helix (Q288, Q289, D292, Q294, S295 and S296) are not part of monomer : monomer contacts, consistent with our alanine scanning results. The α helix is flanked by a ‘tail’ at the extreme C-terminus of CbpA that appears to play little role in dimerization. Consistent with this, removal of the C-terminal 10 amino acids of CbpA, that form this tail, does not affect dimerization (). Interestingly, many proteins related to CbpA do not contain this C-terminal tail ().
Structural model of the CbpA CTD II dimer. A. The CbpA CTD II monomers are shown in yellow and green respectively. B. Amino acid side chains W287 and L290 are highlighted and form a hydrophobic surface that is buried at the dimerization interface.
The most intriguing aspect of our analysis is the aggregation of CbpA with DNA revealed by our gel shift assays () and AFM analysis (). Aggregation of CbpA with DNA does not appear to be influenced by the topological state of the DNA () and CbpA forms aggregates with both linear and plasmid DNA (, and S4
). Comparison of CbpA–DNA complexes presented here with Dps–DNA complexes previously imaged by Ceci et al. (2004)
reveal remarkable similarities. Both proteins form large aggregates when bound to DNA and, like Dps, CbpA can protect DNA from digestion by DNase I (). These observations are particularly intriguing given that CbpA and Dps are both expressed in a stationary phase-specific manner, when the nucleoid undergoes a process of aggregation and compaction. However, since there is no obvious homology between Dps and CbpA, the two proteins likely interact with the DNA differently to form aggregates. Indeed, we speculate that CbpA may play a more specialized role than Dps in organization of the stationary phase chromosome; CbpA is present at lower levels than Dps and forms foci in the nucleoid whereas Dps is evenly distributed (Azam et al., 2000
). For example, a CbpA-like protein in Plasmodium falciparum
was recently shown to specifically interact with the chromosome replication origin (Kumar et al., 2010
). It is also possible that CbpA acts as a ‘back-up’ for Dps in certain conditions. Consistent with previous observations we found that Fis, a major growth phase nucleoid protein, did not aggregate across the DNA but bound at specific locations (Schneider et al., 2001
; Grainger et al., 2006
). It is clear that these locations almost always coincide with locations at which DNA strands cross each other ().
In summary, CbpA was discovered as a DNA-binding protein that preferentially recognized curved DNA sequences (Yamada et al., 1990
; Ueguchi et al., 1994
). Subsequent studies have shown that CbpA binds to DNA with a similar affinity to other nucleoid proteins (Azam and Ishihama, 1999
) and is associated with the nucleoid of starved cells (Azam et al., 2000
). In addition to binding DNA, CbpA, by virtue of its J-domain, can act as a substitute for DnaJ in DnaK-mediated protein disaggregation. Disruption of the cbpA
gene in a dnaJ
genetic background results in severe growth defects at 37°C and, at permissive temperatures, these mutants have an elongated appearance and multiple nucleoids (Chenoweth et al., 2007
). Binding of CbpA to DNA inhibits the co-chaperone activity of the protein (Chae et al., 2004
). Moreover, association of CbpA with CbpM inhibits both the DNA-binding and co-chaperone functions of CbpA (Chae et al., 2004
; Chenoweth et al., 2007
). Here we show that, upon binding DNA, CbpA forms aggregates. This likely explains why CbpA is unable to function as a co-chaperone when bound to DNA. Dimerization of CbpA is required for DNA binding and we show that CbpA dimerization in vivo
is sensitive to environmental conditions (). Thus, it is also possible that an environmental signal triggers the co-chaperone activity of CbpA. We suggest that, to fully understand the biological role of CbpA, it will be essential to determine how transitions between the co-chaperone, DNA-binding and CbpM-binding activities of CbpA are controlled.