Evidence for Roles of the Escherichia coli Hda Protein Beyond RIDA

Abstract

The ATP-bound form of the Escherichia coli DnaA protein binds ‘DnaA boxes’ present in the origin of replication (oriC) and operator sites of several genes, including dnaA, to coordinate their transcription with initiation of replication. The Hda protein, together with the β sliding clamp, stimulates the ATPase activity of DnaA via a process termed Regulatory Inactivation of DnaA (RIDA), to regulate the activity of DnaA in DNA replication. Here, we used the mutant dnaN159 strain, which expresses the β159 clamp protein, to gain insight into how the actions of Hda are coordinated with replication. Elevated expression of Hda impeded growth of the dnaN159 strain in a Pol II- and Pol IV-dependent manner, suggesting a role for Hda managing the actions of these Pols. In a wild type strain, elevated levels of Hda conferred sensitivity to nitrofurazone, and suppressed the frequency of −1 frameshift mutations characteristic of Pol IV, while loss of hda conferred cold sensitivity. Using the dnaN159 strain, we identified 24 novel hda alleles, four of which supported E. coli viability despite their RIDA defect. Taken together, these findings suggest that although one or more Hda functions are essential for cell viability, RIDA may be dispensable.

Introduction

The Escherichia coli DnaA protein binds to an organized set of sites within the chromosomal origin of replication (oriC) in an ATP-dependent manner to form an initiation complex that ultimately recruits the replication machinery (Kaguni, 2006, Katayama et al., 2010). Establishment of this initiation complex through the aid of the DnaA-binding protein DiaA and specific arrangement of the DnaA-binding sites results in unwinding of the AT-rich region within oriC to form the open complex (Keyamura et al., 2007, Rozgaja et al., 2011). The main replicative helicase, DnaB, is loaded into this open complex in a DnaC- and DnaA-dependent manner (Learn et al., 1997, Seitz et al., 2000, Sutton et al., 1998). During expansion of the unwound region, DnaB recruits DnaG primase, which synthesizes short primers that recruit the DNA polymerase III holoenzyme (Pol III HE), which then catalyzes bidirectional replication of the genome (O’Donnell & Kuriyan, 2006). Pol III HE is comprised of three subassemblies: a catalytic core complex (α, ε, and θ polymerase subunits) responsible for DNA synthesis, the homodimeric β (DnaN) processivity clamp that tethers the core complex to DNA, increasing its processivity, and the DnaX clamp loader complex (δ, δ′, τ₂, γ, χ, and ψ) that loads clamp onto DNA in an ATP-dependent manner, and tethers two core assemblies in Pol III HE to enable simultaneous synthesis of both leading and lagging strands (McHenry, 2011, O’Donnell & Kuriyan, 2006).

Replication of the chromosome is tightly controlled at the stage of initiation in order to prevent premature rounds of replication, which can be toxic (Kellenberger-Gujer et al., 1978, Simmons et al., 2004). In E. coli, at least four systems have been identified that specifically target DnaA or oriC to prevent extra initiation events (Kaguni, 2006). The first of these is SeqA-dependent sequestration of hemi-methylated oriC, which acts to prevent re-association of ATP-DnaA to oriC immediately after duplication of the origin (Lu et al., 1994). The second is achieved through titration of ATP-DnaA by the chromosomal datA locus (71 min), located ~13 min away from oriC (84.5 min) (Kitagawa et al., 1998). Following its duplication, the 2 copies of datA sequester multiple ATP-DnaA, impeding their ability to bind to oriC and promote premature initiation (Morigen et al., 2005). Third, in addition to oriC, ATP-DnaA also binds to DnaA boxes upstream of dnaA to repress transcription, which prevents accumulation of ATP-DnaA to limit extra initiations (Riber & Lobner-Olesen, 2005). Lastly, the Regulatory Inactivation of DnaA (RIDA) promotes hydrolysis of ATP bound to DnaA to yield the ADP-DnaA complex, which is inactive for initiation (Kato & Katayama, 2001). Since ATP-DnaA also helps regulate transcription of several genes whose products act in DNA metabolism, RIDA also affects the ability of DnaA to coordinate transcription of these genes with replication (Riber et al., 2006).

In E. coli, RIDA is catalyzed by a nucleoprotein complex comprised at a minimum of duplex DNA, the DNA-loaded β clamp, and ADP-bound Hda protein (Su’etsugu et al., 2008). Hda resembles the AAA⁺ ATPase module of the DnaA protein (Kato & Katayama, 2001). Hda interacts physically with both the β clamp and ATPDnaA (Kurz et al., 2004, Nakamura & Katayama, 2010, Su’etsugu et al., 2008, Su’etsugu et al., 2005). Its interaction with the clamp involves, at least in part, association of a short hydrophobic sequence common to clamp binding partners, termed the Clamp Binding Motif (CBM), with a hydrophobic cleft located near the C-terminal tail of each clamp protomer ((Dalrymple et al., 2001); see Fig. 1). Mutations targeting either the CBM (Kawakami et al., 2006, Su’etsugu et al., 2005), the conserved AAA⁺ ATPase arginine finger (R153) (Su’etsugu et al., 2005), or residues involved in interaction with DnaA (Nakamura & Katayama, 2010) render Hda deficient for RIDA both in vitro and in vivo. Disruption of RIDA increases the cellular ATP-DnaA level, contributing to extra initiations (Camara et al., 2005, Kato & Katayama, 2001, Riber et al., 2009, Su’etsugu et al., 2001). The finding that hda function was required for cellular proliferation under certain conditions suggests that RIDA is essential for cell viability (Fujimitsu et al., 2008). However, efforts to understand the Δhda phenotype have thus far yielded a collection of non-redundant extragenic suppressor mutations mapping to a broad range of loci with no obvious relationship to each other (Charbon et al., 2011).

Figure 1. Structural model for the Hda-β clamp interaction.

(A) Amino acid alignment of the clamp binding motifs (CBM) of E. coli Pol IV (EcPol IV) and E. coli Hda (EcHda). The Pol IV^LF domain (residues 243-351) bears the CBM at its C-terminus, whereas the Hda CBM resides at its N-terminus. This alignment provides a reference for residue alignment in the structural model, and defines the directionality of the partner proteins relative to the cleft of the β clamp. The E. coli Hda sequence was threaded through the crystal structure of a single molecule of the Shewanella amazonensis Hda (SaHda) protein (PDB ID 3BOS, chain B) (Xu et al., 2009) using the SWISS-MODEL Workspace (Schwede et al., 2003). The resulting model for the EcHda monomer is shown (B), with the CBM colored red. (C) Crystal structure of the Pol IV^LF domain (pink) in complex with the β clamp homodimer (yellow and orange) (PDB ID 1UNN) (Bunting et al., 2003). (D) Overlay of the EcHda monomer structure (blue) on the β clamp (yellow and orange), using the Pol IV^LF (pink) CBM as an alignment guide. Using the pair-fitting function within PyMOL 1.5.0.2 software, α-carbon residues 5-9 (LQLSL) of EcHda were aligned to the α-carbon residues 346-351 of Pol IV^LF (LVLGL), theoretically representing a similar interaction of the CBM for both partner proteins. EcHda clearly extends outward and away from where the Pol IV^LF resides on the β clamp. (E) The modeled structure of EcHda (blue) in complex with the β clamp (yellow and orange) without Pol IV^LF is shown. Structural images were generated using PyMOL v1.5.0.2 software (Schrödinger, 2010).

Current models for RIDA are based largely on results of experiments attempting to reconstitute RIDA in vitro using purified components. Loading of the β clamp onto duplex DNA is necessary: an RNA-DNA duplex cannot substitute (Su’etsugu et al., 2004). The requirement for DNA-loaded β clamp may ensure coupling of RIDA to the replication cycle (Katayama et al., 1998, Su’etsugu et al., 2004). Interestingly, RIDA is stimulated by Pol III HE in the presence of dNTPs (Kato & Katayama, 2001). Taken together, these findings suggest that RIDA may occur within the replisome, using clamps that are either in association with Pol III HE, or persist on lagging strand when the lagging strand Pol III cycles to a new Okazaki fragment (Leonard & Grimwade, 2010, Su’etsugu et al., 2004). An alternative model was proposed in which RIDA takes place at oriC immediately following recruitment of β clamp (Clarey et al., 2006, Mott & Berger, 2007). This model posits that clamp-bound Hda associates with ATP-DnaA filaments bound to oriC to trigger a wave of ATP hydrolysis throughout the filament. Given that only a handful of the more than ~1,000 DnaA molecules in the cell are bound to oriC at initiation (Atlung et al., 1987, Carr & Kaguni, 2001), this mechanism would inactivate only a small subset of the total ATP-DnaA, unless it were catalytic. Although both models are possible, and each provides a useful framework for studying Hda and RIDA, neither address how the actions of Hda are coordinated with those of Pol III HE during the replication process.

The goal of the work discussed in this report was to begin to define how E. coli coordinates RIDA with replication. E. coli has five DNA polymerases: Pol I and Pol III HE act in DNA replication and repair, while Pol II, Pol IV, and Pol V act in translesion DNA synthesis (TLS), which contributes to mutations induced by DNA damage (Ohmori et al., 2001, Prakash et al., 2005, Sutton et al., 2000). Importantly, the actions of these Pols are coordinated in large part through their interaction with the β sliding clamp (Kurz et al., 2004, Lopez de Saro et al., 2006, Lopez de Saro & O’Donnell, 2001, Sutton et al., 1999). We hypothesized that if RIDA were catalyzed during DNA replication, then Hda may impact the function of one or more of the E. coli Pols. E. coli strains bearing the dnaN159 allele, which encodes a temperature-sensitive clamp protein (β159) bearing G66E and G174A substitutions, display altered Pol usage (Maul & Sutton, 2005, Sutton, 2004, Sutton & Duzen, 2006, Sutton et al., 2005). We therefore used the dnaN159 strain as a tool to determine whether Hda might reside within the replisome and impact function of one or more Pol(s) during the replication process. Our results demonstrate that a low copy number plasmid directing expression of Hda from its native promoter was toxic to the dnaN159 strain. Importantly, toxicity was dependent upon Pol II and Pol IV, suggesting a role for Hda in helping to manage the actions of these Pols in vivo. Consistent with such a role, elevated levels of Hda conferred sensitivity to nitrofurazone (NFZ), and suppressed the frequency of −1 frameshift mutations characteristic of Pol IV. While analyzing these phenotypes, we determined that loss of hda conferred a cold sensitive phenotype that is tied to the suppressor-acquiring state previously reported (Charbon et al., 2011, Riber et al., 2006). Finally, we exploited the severe sensitivity of the dnaN159 strain to elevated levels of Hda to identify a collection of novel mutant hda alleles unable to confer toxicity. Characterization of these mutant Hda proteins using genetic and biological approaches identified four novel Hda mutations capable of supporting E. coli growth despite their RIDA defect. Taken together, these findings illustrate important cellular functions for Hda in addition to its established role in RIDA, and suggest that although one or more Hda function(s) are essential for E. coli viability, its ability to catalyze RIDA is not required for cell growth.

Results

A structural model for the E. coli Hda-β clamp complex

Bunting et al. reported a crystal structure of the β clamp in complex with the little finger domain of Pol IV (Pol IV^LF) (Bunting et al., 2003). In this structure, Pol IV^LF was bound to two separate clamp surfaces: the CBM located at the the C-tail of Pol IV^LF bound the hydrophobic cleft of the clamp, while residues V303-P305 of Pol IV^LF reached over the dimer interface to contact the rim of the adjacent clamp protomer (Fig. 1C). Although only the cleft contact is required for Pol IV DNA synthesis, both the rim and cleft contacts were required for Pol IV to switch and perform DNA synthesis with a stalled Pol III HE in vitro and in vivo (Heltzel et al., 2009, Sutton et al., 2010). As a detailed understanding of Hda-β clamp interactions is required for a molecular description of the role of the β clamp in coordinating RIDA with DNA replication, we used the Pol IV^LF-β clamp structure described by Bunting et al. as a starting point to construct a working model for the Hda-β clamp complex (Fig. 1E). We began by threading the E. coli Hda sequence through the x-ray crystal structure of the Shewanella amazonensis Hda (SaHda) homodimer (Xu et al., 2009) to generate a model for the E. coli protein (Fig. 1B; EcHda) using SWISS-MODEL Workspace (Schwede et al., 2003). We then manually aligned the CBM of the modeled EcHda monomer with the CBM of the E. coli Pol IV^LF in the Pol IV^LF-β clamp complex (Bunting et al., 2003) (Fig. 1D). Since the CBM is located at opposite ends in the two proteins (Fig. 1A), EcHda exited the clamp cleft in the opposite orientation as Pol IV^LF (Fig. 1D), positioning the globular domain of the EcHda monomer on the rim of the β clamp (Fig. 1E). Although it remains to be demonstrated whether Hda contacts the clamp rim, our model nonetheless suggests that Hda and Pol IV could co-exist on the same clamp, with each making distinct contacts with separate rim surfaces (Fig. 1D).

Elevated levels of Hda are toxic to the dnaN159 strain

A plasmid directing expression of Pol IV at ~4-fold higher than normal SOS-induced levels impairs growth of the dnaN159 strain (Maul & Sutton, 2005). Based on our structural model for the Hda-β clamp complex (Fig. 1), we hypothesized that elevated levels of Hda could act similarly to Pol IV and impair growth of the dnaN159 strain by interfering with the ability of β159 to act in replication. As a test of this hypothesis, isogenic dnaN⁺ and dnaN159 strains were transformed with either the control plasmid (pWSK29), or hda plasmid (pJCB200), which at a copy number of six to eight, expresses ~5-fold higher than physiological levels by Western blot (data not shown). Although the dnaN⁺ strain (MS100) was efficiently transformed with pJCB200, no transformants were obtained with the isogenic dnaN159 (MS101) partner (Table 1). As controls, we examined plasmids expressing seqA and bearing datA, respectively. While extra copies of datA failed to effect growth of either the dnaN⁺ or dnaN159 strains, elevated levels of SeqA severely impaired growth of both (Table 1). Taken together, these results indicate that Hda is unique among the initiation regulators in terms of its ability to impede growth of the dnaN159 strain when expressed at an elevated level.

TABLE 1.

Elevated levels of Hda are lethal to the dnaN159 strain^a

Recipient strain	Relevant Genotype	Plasmid allele	Transformation Efficiencyb	Relative Efficiencyc
MS100	dnaN+	none (pWSK29)	1.43 (±0.10) × 106	≡1.00
		hda (pJCB200)	1.55 (±0.29) × 106	1.08
		empty (pACYC184)	1.74 (±0.29) × 106	≡1.00
		datA (pMMF84)	1.41 (±0.19) × 106	0.81
		seqA (pMMF57)	< 33	< 1.92 × 10−5
MS101	dnaN159	none (pWSK29)	1.26 (±0.12) × 105	≡1.00
		hda (pJCB200)	< 22	< 1.75 × 10−4
		empty (pACYC184)	2.63 (±0.92) × 105	≡1.00
		datA (pMMF84)	1.21 (±0.16) × 105	0.46
		seqA (pMMF57)	< 33	< 1.27 × 10−4
		hda(Q6A) (pJCBQ6A)	1.94 (±0.30) × 105	1.54
		hda(L9A) (pJCBL9A)	1.05 (±0.08) × 105	0.83
		hda(R153A) (pJCBR153A)	2.45 (±0.62) × 105	1.94
		hda(R153M) (pJCBR153M)	1.45 (±0.01) × 105	1.15
JCB150	dnaN+ ΔdinB	none (pWSK29)	4.01 (±0.26) × 104	≡1.00
		hda (pJCB200)	4.37 (±0.13) × 104	0.96
MS125	dnaN159 ΔdinB	none (pWSK29)	1.74 (±0.09) × 105	≡1.00
		hda (pJCB200)d	3.10 (±0.35) × 104	0.18
JCB151	dnaN+ ΔpolB	none (pWSK29)	7.78 (±1.60) × 103	≡1.00
		hda (pJCB200)	2.65 (±0.74) × 104	3.41
JCB152	dnaN159 ΔpolB	none (pWSK29)	6.19 (±1.45) × 103	≡1.00
		hda (pJCB200)d	1.57 (±0.13) × 104	2.54
MS108	dnaN+ ΔumuDC	none (pWSK29)	4.96 (±1.16) × 103	≡1.00
		hda (pJCB200)	1.71 (±0.22) × 104	3.45
MS109	dnaN159 ΔumuDC	none (pWSK29)	1.45 (±0.42) × 104	≡1.00
		hda (pJCB200)	<3.3	<3.82 × 10−4
MS102	dnaN+ lexA3(Ind−)	none (pWSK29)	1.73 (±0.76) × 104	≡1.00
		hda (pJCB200)	1.61 (±0.30) × 104	0.93
MS103	dnaN159 lexA3(Ind−)	none (pWSK29)	2.67 (±1.15) × 103	≡1.00
		hda (pJCB200)	<100	< 3.75 × 10−2
MS104	dnaN+ lexA51(Def)	none (pWSK29)	1.30 (±0.30) × 104	≡1.00
		hda (pJCB200)	2.60 (±0.69) × 103	0.20
MS105	dnaN159 lexA51(Def)	none (pWSK29)	1.10 (±0.50) × 104	≡1.00
		hda (pJCB200)	<100	< 9.10 × 10−3

^aAn n of 2 was performed for each transformation. All strains are derivatives of MS100.

^bTransformation efficiency is expressed as transformants per μg plasmid DNA.

^cRelative transformation efficiency is expressed relative to pWSK29 (for hda) or pACYC184 (for seqA and datA), that was set equal to 1.00 for each strain.

^dMS125 pJCB200 transformants were markedly healthier (1–2 mm in diameter, like MS125 pWSK29 transformants) than JCB152 pJCB200 transformants (0.5 mm in diameter, smaller than JCB152 pWSK29 transformants) after overnight growth.

To determine whether RIDA activity is required for Hda toxicity, we measured transformation efficiencies for pJCB200-derived plasmids expressing mutant Hda proteins known to be defective for RIDA. These plasmids bear substitutions in either the CBM of Hda known to abrogate its interaction with the β clamp (Q6A and L9A) (Su’etsugu et al., 2005), or the highly conserved arginine finger (R153A and R153M) that is important for catalytic activity of AAA⁺-ATPase proteins (Su’etsugu et al., 2005). As summarized in Table 1, all four of these plasmids transformed the dnaN159 strain with efficiencies similar to that of the pWSK29 control plasmid. Taken together, these results indicate that Hda toxicity relies on both the clamp binding activity of Hda, and its arginine finger.

Sensitivity of the dnaN159 strain to Hda depends on Pol II and Pol IV

We hypothesized that Hda and Pol IV (and/or additional Pols) can interact simultaneously with the same clamp protein (see Fig. 1D), and that in the case of the dnaN159 strain, these interactions serve to impede clamp function, conferring toxicity. As a test of this hypothesis, we asked whether the ability of Hda to impair growth of the dnaN159 strain could be suppressed by inactivation of E. coli Pol II (polB), Pol IV (dinB), or Pol V (umuDC). As summarized in Table 1, loss of either polB or dinB function rendered the dnaN159 strain tolerant to Hda. Careful examination of the growth phenotypes for these strains suggested that inactivation of dinB was slightly more effective at alleviating Hda toxicity compared to polB (Table 1, footnote c). It is interesting that loss of polB enhances transformation efficiency of pJCB200 (hda⁺) into dnaN159, although it is unclear the significance of this. In contrast to these Pols, inactivation of umuDC failed to alleviate Hda toxicity (Table 1). In order to rule out the possibility that a loss of polB or dinB may reduce the number of available β-clamp molecules to serve as substrates for Hda in RIDA, we re-examined these transformations at increasing temperatures, where the steady-state levels of β159 decrease in the cell (Sutton, 2004). At temperatures up to 36°C, we saw no change in transformation efficiencies (data not shown), suggesting we are not seeing alleviation of the lethality due to a reduction in clamp molecules to function in RIDA. Since overproduction of Hda from the pET17B vector was able to induce the global E. coli SOS response (Banack et al., 2005), we asked whether Hda toxicity from pJCB200 in dnaN159 relied on an ability to induce SOS. As summarized in Table 1, transformation frequencies of isogenic dnaN159 lexA3(Ind⁻) (MS103) or dnaN159 lexA51(Def) (MS105) strains with pJCB200 were indistinguishable from the dnaN159 lexA⁺ (MS101) strain. Taken together, these results indicate that the ability of Hda to impede growth of the dnaN159 strain relies on Pol II and Pol IV, but is independent of SOS induction.

Elevated levels of Hda confer nitrofurazone (NFZ) sensitivity upon E. coli, and suppress the frequency of spontaneous −1 frameshift mutations in vivo

Results discussed above suggest that Hda interacts functionally with Pol II and Pol IV (Table 1). Pol IV (Jarosz et al., 2006, Wagner & Nohmi, 2000), and possibly Pol II (Williams et al., 2010), play important roles in tolerating NFZ-induced DNA damage. In the case of Pol IV, NFZ tolerance results from accurate bypass of N²-dG adducts induced by exposure of cellular DNA to NFZ (Jarosz et al., 2006). Using NFZ-sensitivity as a reporter for Pol II and Pol IV function, we asked whether the level of Hda expressed from pJCB200 sensitized E. coli to the bacteriostatic effects of NFZ. As summarized in Fig. 2A, pJCB200 (hda⁺) sensitized E. coli to NFZ in a fashion that was independent of Pol II and/or Pol IV: although loss of these Pols resulted in a modest sensitivity to NFZ, transformation with pJCB200 further sensitized these strains. Likewise, pJCB200 further sensitized a strain impaired for nucleotide excision repair (Fig. 2B), indicating that Hda confers sensitivity independently of this repair function. In contrast to NFZ, pJCB200 failed to confer sensitivity to ultraviolet (UV) light (Fig. 2C), indicating that NFZ sensitivity was due to a specific effect, rather than a general replication or repair defect.

Figure 2. Overexpression of Hda sensitizes E. coli to NFZ.

Sensitivity of isogenic E. coli strains bearing pWSK29 (control plasmid) or pJCB200 (hda⁺) to the indicated concentrations of NFZ was measured as described in Experimental Procedures. Strains were either impaired for polB (Pol II), dinB (Pol IV), or polB and dinB function (A), or impaired for nucleotide excision repair (uvrB) (B), as indicated. (C) Sensitivity to UV irradiation (50 J•m⁻²) of the polB⁺ dinB⁺ uvrB⁺ parent strain bearing pWSK29 or pJCB200 was measured as described in Experimental Procedures. Representative results for each are shown.

In addition to its role in TLS, Pol IV also catalyzes frameshift mutations in homopolymeric DNA tracks (Kim et al., 1997, Wagner et al., 1999). As part of an effort to obtain further evidence for a possible role for Hda in affecting Pol IV function, we asked whether elevated levels of Hda suppressed the frequency of −1 frameshift mutations in vivo. The frequency of −1 frameshift mutagenesis was measured using strain CC108, which bears a mutant lacZ allele containing a +1 frameshift mutation within a homopolymeric run of dG residues (Cupples et al., 1990). Loss of one dG in this run restores the Lac⁺ phenotype. As summarized in Fig. 3, transformation of CC108 with pJCB200 led to a striking ~10-fold reduction in the frequency of −1 frameshift mutations, consistent with a role for Hda in limiting access of Pol IV to the DNA. These results, taken together with those discussed above, suggest that Hda may help to manage the actions of Pol II and Pol IV in vivo.

Figure 3. Overexpression of Hda suppresses −1 frameshift mutations.

The frequency of lacZ⁻′Lac⁺ reversion of strain CC108 bearing pWSK29 (control plasmid) or pJCB200 (hda⁺) are shown. Results represent the average of 12 independent determinations for each strain. Error bars represent the 95% confidence limits.

Loss of hda function confers a cold sensitive phenotype upon E. coli

In light of the phenotypes discussed above for strains expressing elevated levels of Hda, we next inquired whether loss of hda abrogated one or more dnaN159 phenotype. Although hda is reported to be an essential gene, it has nevertheless been deleted previously (Camara et al., 2003, Riber et al., 2006). Δhda strains were recently demonstrated to acquire suppressors that improve growth and varying degrees of initiation defects (Charbon et al., 2011, Riber et al., 2006). Although we were unable to construct a dnaN159 Δhda strain (data not shown), we reproducibly observed a temperature dependency for the construction of the dnaN⁺ Δhda::cat strain. Although we were able to transduce the Δhda::cat allele into MG1655 bearing either pWSK29 (control plasmid) or pJCB200 (hda⁺) at 37° and 42°C, at 30°C we were only able to transduce the MG1655 strain bearing pJCB200 (Table 2). As controls, we examined transduction frequencies for ΔseqA::tet and ΔdatA::kan. Each of these alleles was transduced efficiently into MG1655, irrespective of whether they harbored the empty or respective complementing plasmid (Table 2). Taken together, these results indicate that the cold-sensitive phenotype conferred by the Δhda::cat allele is not a general consequence of an impaired ability to regulate initiation.

TABLE 2.

The Δhda strain displays a cold sensitive growth phenotype^a

Donor allele	Complement allele	Temperature (°C)	Transduction frequencyb	Relative efficiencyc
Δhda::cat	hda (pJCB200)	30	3.07 × 10−7	≡1.00
	none (pWSK29)	30	<0.06 × 10−8	<1.95 × 10−3
	hda (pJCB200)	37	9.86 × 10−8	≡1.00
	none (pWSK29)	37	8.87 × 10−8	0.90
	hda (pJCB200)	42	2.46 × 10−7	≡1.00
	none (pWSK29)	42	2.05 × 10−7	0.83
ΔdatA::kan	datA (pMMF84)	30	3.35 × 10−6	≡1.00
	none (pACYC184)	30	2.80 × 10−6	0.84
ΔseqA::tet	seqA (pMMF57)	30	3.60 × 10−6	≡1.00
	none (pACYC184)	30	1.55 × 10−6	0.43

^aThe indicated allele was introduced into strain MG1655 using P1vir. Transductants were counted after incubation at the indicated temperatures for 30 hours on LB plates containing chloramphenicol (20 μg•mL⁻¹) for Δhda transduction, kanamycin (40 μg•mL⁻¹) for ΔdatA transduction, or tetracycline (10 μg•mL⁻¹) for ΔseqA transduction.

^bTransduction frequency is expressed as number of transductant colonies per plaque forming unit P1vir bacteriophage.

^cEfficiency is relative to the transduction frequency of the respective complemented transduction.

Since the growth defects associated with the loss of hda function can be masked by suppressor mutations (Riber et al., 2006), we asked whether rapid suppression of the cold sensitivity we observe for the Δhda strain might account for the failure of others to observe this phenotype. Closer examination of the Δhda::cat transductants obtained at 37° and 42°C revealed a mixture of large and small colony phenotypes (Fig. 4; Δhda isolate 1). Further analysis of these two classes indicated that large colonies were viable at 30°C, whereas small colonies were not. When representative small colony variants were struck onto plates and grown overnight at 37°C or 42°C, they consistently yielded a mix of small colonies that were impaired for growth at 30°C (Fig. 4, Δhda isolates 2–4), and large colonies capable of growing at 30°C (Fig. 4, Δhda isolate 5). When representative large colonies were struck, only large, cold-resistant colonies were observed. Taken together, these results indicate that hda function is essential for viability of E. coli at 30°C, and, as a result, growth of Δhda strains at this temperature likely relies on suppressor mutations that compensate for the lack of hda function.

Figure 4. The Δhda E. coli strain displays a cold sensitive growth phenotype.

The Δhda::cat allele was transduced into strain MG1655 bearing either pWSK29 (control plasmid) or pJCB200 (hda⁺) using P1vir bacteriophage. Representative colonies were selected, grown overnight at 37 °C, serially diluted, spotted onto LB plates, and grown overnight at 30°, 37°, or 42 °C, as indicated. Strain JCB100 (Δhda) bearing pWSK29 (control plasmid) was severely impaired for growth at 30 °C, and displayed a mix of large and small colonies at 37° and 42 °C. Three small (isolates 2–4) and one large (isolate 5) CFU for strain JCB100 bearing pWSK29 were selected, grown overnight at 37° C, serially diluted, spotted onto LB plates, and grown overnight at 30° or 42 °C, as indicated. * indicates distinct isolates from a fresh transduction of Δhda::cat into MG1655 pWSK29.

Identification of hda mutations unable to impede growth of the dnaN159 strain

The ability of pJCB200 to impede growth of the dnaN159 strain provided a strong selection by which to identify alleles impaired for this function. We postulated that analysis of hda alleles unable to impair growth of the dnaN159 strain would provide insights into both biological roles and structure-function of Hda. To identify hda mutations, we simply selected for pJCB200 derivatives that were able to transform the dnaN159 strain. Using this approach, we identified 40 clones that transformed the dnaN159 strain at a frequency comparable to that of the pWSK29 control plasmid, and expressed a full length Hda protein at the expected level, based on Western blot analysis (see Experimental Procedures). The mutation(s) resulting in each hda allele was identified by automated nucleotide sequence of the plasmid-encoded hda cassette. Of the 40 plasmids sequenced, 24 novel and unique hda alleles were identified (Table 3). Of these 24, the majority (21) resulted from a single nucleotide missense mutation; however, 1 insertion [ΩLDQ(208-210)], 1 deletion [ΔLDQ(208-210)], and 1 frameshift (D198fs) mutation were also identified. Interestingly, the insertion mutation sequence duplicated the nucleotide sequence that was lacking in the deletion mutation. The frameshift mutation (D198fs) substituted residue D198 with R, followed by 20 novel amino acids and a stop codon.

TABLE 3.

Summary of nucleotide sequence analysis of novel hda alleles.

Plasmid	hda allele (deduced amino acid substitutiona)	Nucleotide substitutionb	No. of occurrencesc
pJCB217D15G		43GAC→GGC	1
pJCB212A62E		184GCG→GAG	1
pJCB237S68P		202TCG→CCG	1
pJCB225L98P		292CTG→CCG	2
pJCB251E114G		340GAG→GGG	1
pJCB219Y121H		361TAC→CAC	1
pJCB214L133S		397TTG→TCG	1
pJCB240G136D		406GGC→GAC	4
pJCB216P140R		418CCG→CGG	1
pJCB211R141Q		421CGG→CAG	1
pJCB243D155N		463GAC→AAC	3
pJCB218G157V		469GGG→GTG	8
pJCB239D167N		499GAT→AAT	1
pJCB223A178V		532GCG→GTG	2
pJCB215G182A		544GGT→GAT	1
pJCB249R196Q		586CGG→CAG	2
pJCB203D198fs		Δ592(GACAGAG)	2
pJCB238D209G		625GAT→GGT	1
		490CCA→CCC (silent)
pJCB234D212N		634GAT→AAT	1
pJCB221L222P		664CTG→CCG	1
pJCB227P225A		673CCG→GCG	1
pJCB226L233F		697TTG→TTT	1
pJCB202Δ(LDQ208-210)		Δ622(TTGGATCAG)	1
pJCB213Ω(LDQ208-210)		Ω622(TTGGATCAG)	1

^aDeduced amino acid substitutions, based on the DNA sequence, are indicated. Amino acids are abbreviated using the one letter code (i.e. D15G indicates aspartic acid-15 changed to glycine). D198fs indicates a frameshift in the codon frame beginning with residue 198. ΔLDQ and ΩLDQ indicate deletion and duplication respectively of the three-codon sequences.

^bThe position of nucleotide alterations are indicated relative to nucleotides in hda. Numbers represent the first base in the codon, and are relative to the C in CTG encoding the initiator codon, which is defined as 1.

^cThe number of independent times that each hda allele indicated was identified.

The N-terminal region of Hda contains the CBM that was important for the ability of Hda to impede growth of the dnaN159 strain (Table 1). However, with the exception of the D15G substitution, no other mutations were identified in the first 61 residues of the protein (Fig. 5A). Mutations mapping to the rest of Hda were fairly evenly distributed throughout the sequence. When the position of each mutation was represented on a structural model of Hda (Fig. 5, panels B, C, D, and E), it was evident that most mutations affected portions of Hda predicted to be solvent accessible. Exceptions to this include mutations L98P, L133S, and G136D, which are confined to the β-stacking sheet that occupies the hydrophobic core of the protein crystal (Fig. 5, panels C and E), and the three mutations affecting the C-terminal region of Hda (Fig. 5E). The paired deletion/insertion mutants, ΔLDQ208-210 and ΩLDQ208-210, respectively, likely affect the length of the α-helix they reside in (α7). In contrast, the D198fs mutant likely retains an α-helix similar in nature to that of α7 of the wild type protein, based on a secondary structure prediction (PHYRE 2.0 (Kelley & Sternberg, 2009)), but, due to the frameshift, lacks the terminal α-helix (α8). Finally, several mutations also map nearby residues previously demonstrated to impair discrete biochemical functions of Hda (Nakamura & Katayama, 2010). Possible defects conferred by these mutations are discussed later (see Discussion).

Figure 5. Positions of mutations identified in Hda.

(A) The primary structure of Hda depicting positions of the CBM (green), Walker A and B motifs (blue), Box VI and VII (light blue), sensors 1 and 2 (black line), and the conserved arginine finger (black line) is shown. Residues of Hda that form alpha (α) and 3₁₀ (η) helices or beta strands (β), based on the SaHda structure (PDB ID 3BOS), are indicated. Positions of mutants identified in this study are shown. Mutations are colored according to mutant class: class I is in green, class II is in orange, class III is in red, and class IV is in purple. The new reading frame generated by the D198fs mutant is shown. Positions of each mutation are also represented on the EcHda structural model from Figure 1B. Each panel (B–E) represents a single class of Hda mutations: (B) class I mutations are depicted as green spheres; (C) class II mutations are depicted as orange spheres; (D) class III mutations are depicted as red spheres; and (E) class IV mutations are depicted as purple spheres, while the C-terminal region affected by the D198fs is colored in pink. Residues corresponding to the CBM (blue) or the arginine finger (red) are depicted in stick form for reference. Structural images were generated using PyMOL v1.5.0.2 software (Schrödinger, 2010).

All but 6 of the novel hda alleles support viability of E. coli at 30°C

We exploited the cold sensitive growth phenotype of the Δhda strain in order to screen the various hda alleles for their ability to support viability of E. coli. Strain MG1655 was transformed with the different hda plasmids prior to deleting the chromosomal hda allele using P1vir generalized transduction. As summarized in Table 2, we were unable to transduce the Δhda::cat allele into the MG1655 bearing pWSK29 at 30°C. In contrast, transduction efficiencies of MG1655 bearing pJCB200 were comparable to each other at 30°, 37°, and 42°C (Table 4). The 24 mutant hda alleles were analyzed in a similar fashion, and results of this analysis are summarized in Table 4. All but 6 (L98P, G136D, L233F, D198fs, ΔLDQ208-210, ΩLDQ208-210) of the 24 were able to support growth of strain JB100 (Δhda::cat) at 30°C. Interestingly, 4 (D198fs, L233F, ΔLDQ208-210, and ΩLDQ208-210) of these 6 contained mutations affecting the two carboxy-terminal α-helices of Hda. It was previously suggested that the ability of the hda-185 (K170C) allele to destabilize the C-terminal domain of Hda may result in the cold-sensitive phenotype previously seen (Fujimitsu et al., 2008). Our finding that the two carboxy-terminal α-helices are required for viability of E. coli is consistent with this model. The inability of these hda alleles to support viability of E. coli at 30°C indicates a crucial role for the C-terminus of Hda.

TABLE 4.

Summary of the ability of mutant hda alleles to support cell growth and RIDA in vivo.

Class	hda allele	Transduction Efficiency with Δhda::cata			oriC:terC
		Temperature (°C)	Transduction frequencyb	Relative efficiencyc	Temperature (°C)	oriC:terCd
	hda+	30	3.07 × 10−7	≡1.00	nde	nde
	hda+	37	9.86 × 10−8	0.32	37	4.2 ± 1.1
	hda+	42	2.46 × 10−7	0.80	nde	nde
	Q6A	30	0.06 × 10−8	1.95 × 10−3	nde	nde
	L9A	30	0.06 × 10−8	1.95 × 10−3	nde	nde
	R153A	30	< 0.06 × 10−8	< 1.95 × 10−3	nde	nde
	R153M	30	< 0.06 × 10−8	< 1.95 × 10−3	nde	nde
I	D15G	30	4.30 × 10−8	0.14	37	3.5 ± 1.7
	R141Q	30	9.09 × 10−8	0.30	37	3.5 ± 0.4
	D155N	30	1.25 × 10−7	0.41	37	3.6 ± 0.8
	G182A	30	5.21 × 10−8	0.17	37	3.4 ± 0.2
	D209G	30	1.78 × 10−7	0.58	37	3.4 ± 0.4
	L222P	30	1.91 × 10−7	0.62	37	3.3 ± 0.1
	P225A	30	1.12 × 10−7	0.36	37	3.8 ± 1.3
II	A62E	30	7.64 × 10−8	0.25	37	3.2 ± 0.3
	S68P	30	9.94 × 10−8	0.32	37	4.4 ± 0.6
	E114G	30	8.97 × 10−8	0.29	37	4.0 ± 0.1
	Y121H	30	3.88 × 10−8	0.13	37	3.6 ± 0.1
	L133S	30	8.91 × 10−8	0.29	37	4.3 ± 1.1
	P140R	30	3.82 × 10−8	0.12	37	2.9 ± 0.8
	A178V	30	1.22× 10−7	0.40	37	4.0 ± 1.2
III	G157V	30	4.30 × 10−8	0.14	37	8.2 ± 0.8
	D167N	30	9.82 × 10−8	0.32	37	6.8 ± 0.8
	R196Q	30	8.12 × 10−8	0.26	37	13.0 ± 0.3
	D212N	30	1.62 × 10−7	0.53	37	8.2 ± 0.9
IV	L98P	30	< 0.06 × 10−8	< 1.95 × 10−3	37	36.4 ± 9.2
	G136D	30	< 0.06 × 10−8	< 1.95 × 10−3	37	41.5 ± 15.6
	D198fs	30	< 0.06 × 10−8	< 1.95 × 10−3	37	27.2 ± 10.5
	L233F	30	< 0.06 × 10−8	< 1.95 × 10−3	37	36.1 ± 14.7
	Δ(LDQ208-210)	30	< 0.06 × 10−8	< 1.95 × 10−3	37	33.9 ± 13.0
	Ω(LDQ208-210)	30	< 0.06 × 10−8	< 1.95 × 10−3	37	23.1 ± 2.3

^aThe plasmid in each recipient MG1655 strain used for transduction experiments is indicated. Transductants were counted after incubation at the indicated temperatures for 30 hours on LB plates containing chloramphenicol (20 μg•mL⁻¹) and ampicillin (150 μg•mL⁻¹).

^bTransduction frequency is defined as the number of Cm^R colonies per plaque forming unit P1vir bacteriophage.

^cRelative efficiency is relative to the transduction frequency of the hda-complemented transduction performed at 30 °C, which was set equal to 1.00.

^dCycle threshold [C(t)] was calculated for each primer pair (n = 4) according to the iCycler software. oriC:terC ratio was calculated as the base-2 antilog of C(t)_oriC lesser the C(t)_terC.

^eNo data (nd) was collected for oriC:terC ratios.

Viability of E. coli is independent of the ability of Hda to catalyze RIDA

We next asked whether any of the mutant Hda proteins were capable of catalyzing RIDA. For this, we exploited the fact that most were able to complement viability of the Δhda strain at 30°C to probe their ability to catalyze RIDA in vivo (Table 4). Briefly, the ability of each Hda mutant to regulate initiation while replicating its genome in vivo was measured indirectly using a combination of flow cytometry and quantitative PCR (qPCR). Flow cytometry was used to analyze genome content, and is a convenient means by which to measure replication synchrony. However, since asynchrony in replication could result from either an initiation defect, or an inability to complete replication, we also examined oriC:terC ratios using qPCR as an indirect measure of RIDA. Strains impaired for RIDA should display both an asynchronous phenotype by flow cytometry, and an increased oriC:terC ratio, indicative of excessive initiation events. In contrast, a strain that is proficient for RIDA but unable to complete replication should display an asynchronous phenotype in flow cytometry (odd numbers of chromosomes) and wild type oriC:terC ratio. Finally, strains proficient for RIDA, but impaired for replication, would display an inability to complete chromosomes (no distinct numbers of chromsomes) by flow cytometry, but maintain wildtype oriC:terC ratios by qPCR.

We began by comparing hda⁺ and Δhda strains in MG1655. Briefly, strains were grown to exponential phase at 37°C, at which time an aliquot of cells was harvested for qPCR analysis, while the remainder was treated with rifampin and cephalexin for analysis by flow cytometry. As summarized in Fig. 6A, the hda⁺ strain exhibited the expected genomic profile by flow cytometry, displaying obvious peaks corresponding to 2, 4, and 8 genome equivalents. In striking contrast to hda⁺, the Δhda strain displayed a very different flow cytometry profile (Fig. 6A), consistent with an asynchronously replicating E. coli strain (Boye et al., 2001, Lu et al., 1994, Morigen et al., 2005). Based on qPCR, Δhda strains displayed an oriC:terC ratio of ~19.4 to ~41.2 (Table 5), which is ~5- to ~10-fold higher than the value of ~4.2 observed for the isogenic wild type strain (Table 5), as expected for a strain displaying a pronounced RIDA defect. A Δhda strain able to grow at 30°C (i.e. suppressed) displayed a similar flow cytometry profile (data not shown), but exhibited a reduced oriC:terC ratio of ~10.6 to ~12.4 (Table 5), indicating that suppression improves both growth and regulation of initiation, consistent with previous reports (Charbon et al., 2011, Riber et al., 2006). We also examined isogenic ΔseqA and ΔdatA strains. Whereas the ΔseqA strain resembled the Δhda strain by flow cytometry, the ΔdatA strain exhibited only a modest reduction in synchrony (Fig. 6A), consistent with previous reports (Kitagawa et al., 1998, Morigen et al., 2005). Based on qPCR, the ΔseqA and ΔdatA strains were comparable to the wildtype strain with respect to their abilities to regulate initiation, displaying oriC:terC ratios of 4.9 (±0.6) and 3.7 (±1.3), respectively (Table 5).

Figure 6. Cell cycle analysis of strains expressing mutant hda alleles.

Flow cytometry data was collected with 50,000 cells of the indicated strain labeled with PicoGreen. Fluorescence intensity (abscissa) is presented in logarithmic scale. (A) Chromosome equivalents were determined using strain MG1655. Isogenic strains bearing Δhda, Δdat, ΔseqA, and dnaN (JCB103) or dnaN159 (JCB104) alleles were also analyzed. Representative results for mutants of class I (B), class II (C), class III (D), and class IV (E) are shown.

TABLE 5.

oriC:terC ratios in hda-, datA-, and seqA-deficient strains.

Strain	Relevant genotype	Plasmid allele	oriC:terC
			Temperature (°C)	oriC:terCa
JCB103	dnaN+	none	30	3.4 ± 0.7
			37	3.3 ± 0.4
JCB104	dnaN159	none	30	2.2 ± 0.3
			37	3.0 ± 0.1
MG1655	hda+	empty	37	2.5 ± 0.2
MG1655	hda+	hda	37	4.2 ± 1.1
MG1655	Δhda (isolate 6)b	empty (30 °C-viable)c	37	12.4 ± 1.8
	Δhda (isolate 7)b	empty (30 °C-viable)c	37	10.6 ± 0.6
	Δhda (isolate 8)b	empty (30 °C-inviable)c	37	32.6 ± 0.5
	Δhda (isolate 9)b	empty (30 °C-inviable)c	37	41.2 ± 1.8
	Δhda (isolate 10)b	empty (30 °C-inviable)c	37	19.4 ± 0.6
	Δhda (isolate 11)b	empty (30 °C-inviable)c	37	37.6 ± 6.1
	Δhda (isolate 12)b	empty (30 °C-inviable)c	37	29.6 ± 0.6
	Δhda	hda	37	2.2 ± 0.4
JCB101	ΔdatA	empty	37	3.7 ± 1.3
JCB102	ΔseqA	empty	37	4.9 ± 0.6

^aCycle threshold [C(t)] was calculated for each primer pair (n = 4) according to the iCycler software. oriC:terC ratio was calculated as the base-2 antilog of C(t)_oriC lesser the C(t)_terC.

^bDistinct isolates from a fresh transduction of Δhda::cat into MG1655.

^cBoth 30 °C-viable, and inviable constructs of JCB100 pWSK29 that were constructed at 37 °C, grown overnight for qPCR analysis, and verified for cold-sensitivity at the time of culture harvest, were examined.

As part of this analysis, we also analyzed the dnaN159 strain by qPCR. This strain is proficient for RIDA (Katayama et al., 1998), but is impaired for replication at 37°C (Burgers et al., 1981, Sutton, 2004). The oriC:terC ratio for the dnaN159 strain was comparable to that of the wild type strain (Table 5), regardless of the growth temperature, indicating that regulation of initiation appears to be unaffected, despite its temperature-dependent replication defect. These results help to confirm that the combination of flow cytometry and qPCR distinguishes strains that are wild type for replication and RIDA from those impaired for these functions.

We next analyzed each of the mutant hda alleles. For this, the six (L98P, G136D, D198fs, ΔLDQ208-210, ΩLDQ208-210, L233F) 30°C-inviable mutant complements were constructed using P1vir transduction at the permissive temperature (37°C) and were screened for growth at 30°C during analysis to confirm lack of suppression. The remaining hda mutations, previously constructed at 30°C, were able to support viability of E. coli, and therefore should not bear suppressor mutations. Based on flow cytometry, the 24 hda mutations displayed one of two distinct phenotypes: seven were replication proficient (Fig. 6B), while seventeen were replication-deficient (Figs. 6C, D & E). Replication proficient mutants (D15G, R141Q, D155N, G182A, D209G, L222P, P225A) were able to generate at least two predominant 4N and 8N chromosomal products. Although four of these seven were similar to wildtype, one (P225A) had a minor 6N product, and two (R141Q and D155N) produced what appear to be asynchronous products (~6N and ~12N). The seventeen replication-deficient mutants could be divided into two groups: those that were able to generate a range of chromosomal products (Fig. 6C; A62E, S68P, E114G, Y121H, L133S, P140R, and A178V), and those that were unable to generate discrete products (Fig 6D and E; L98P, G136D, G157V, D167N, R196Q, D198fs, ΔLDQ208-210, ΩLDQ208-210, D212N, L233F).

Based on results of qPCR experiments, the replication-proficient hda mutations (D15G, R141Q, D155N, G182A, D209G, L222P, P225A) had an oriC:terC ratio of ~3.3 to ~3.8, which was comparable to the hda⁺ control strain (Table 4). Thus, this group, despite the asynchronous chromosomal content of R141Q, D155N, and P225A, appears proficient for regulating initiation (i.e. RIDA). The replication-deficient mutants able to generate discrete products (A62E, S68P, E114G, Y121H, L133S, P140R, A178V) had oriC:terC ratios similar to that of the wildtype strain and replication-proficient hda mutants, ranging from ~2.9 to ~4.4, suggesting they were also competent for RIDA (Table 4). The replication-deficient phenotype (Fig. 6C) is likely due to the inability of these mutants to complete replication, as was the case for the dnaN159 strain when grown at 37°C (Fig. 6A and Table 5). hda mutations unable to generate discrete products (L98P, G136D, G157V, D167N, R196Q, D198fs, ΔLDQ208-210, ΩLDQ208-210, D212N, L233F) displayed oriC:terC ratios ranging from ~6.8 to ~42.1 (Table 4), characteristic of cells with defects in RIDA (Charbon et al., 2011, Fujimitsu et al., 2008). A subset of these mutations (G157V, D167N, R196Q, D212N) are particularly interesting in that they are impaired for RIDA, yet nevertheless able to support growth of E. coli (Table 4), suggesting that RIDA, per se, may be dispensable for cell viability.

Taken together, these findings argue strongly that the ability of Hda to impair growth of the dnaN159 strain is independent of its ability to catalyze RIDA. Although hda mutants impaired for interaction with β clamp (Q6A and L9A) or bearing substitutions of the arginine finger (R153A and R153M) supported growth of the dnaN159 strain (Table 1), several hda alleles proficient for regulating initiation were selected based on their inability to impede growth of the dnaN159 strain.

Mutant Hda proteins are unable to sensitize E. coli to NFZ

Results discussed above revealed a remarkable range in phenotypes for the different hda mutations, making it difficult to identify a common defect underlying their ability to impede growth of the dnaN159 strain when expressed at an elevated level. Since the strain expressing elevated levels of Hda was sensitive to NFZ, we examined the ability of each hda mutation to confer NFZ sensitivity. We first examined mutations bearing substitutions in either the CBM (Q6A) (Su’etsugu et al., 2005), or the arginine finger (R153A) of Hda. As summarized in Fig. 7, both of these mutants were impaired in their ability to confer NFZ sensitivity. We next examined the entire collection of 24 hda mutations deficient for impeding growth of the dnaN159 strain. Remarkably, all 24 were impaired for conferring NFZ sensitivity (Fig. 7). These results suggest that the ability of elevated levels of Hda to impede growth of the dnaN159 strain is related to the ability of Hda to sensitize E. coli to NFZ. As a test of this hypothesis, we asked whether the dnaN159 strain was sensitive to NFZ. As summarized in Fig. 8, the dnaN159 allele sensitized E. coli to NFZ.

Figure 7. Mutant hda alleles are unable to confer NFZ sensitivity.

E. coli strain MS100 bearing the indicated mutant allele of hda were serially diluted in 0.8% NaCl and spotted on LB plates containing NFZ as described in Experimental Procedures.

Figure 8. hda lethality and NFZ sensitivity of dnaN mutants.

Positions P20, G66, and G174 of the β clamp, which bear substitutions in the dnaN159 (G66E, G174A), dnaN780 (G66E), dnaN781 (G174A), dnaN782 (G66A, G174A), or dnaN783 (P20L, G66E, G174A) mutations, are depicted in red on the structural model for the EcHda-clamp complex. Colors are as in Figure 1. This structural representation was generated using PyMOL v1.5.0.2 software. (B) Cultures of E. coli strain MS100 and derivatives bearing the indicated dnaN allele were serially diluted in 0.8% NaCl and spotted on LB plates containing the indicated concentrations of NFZ. Plates were incubated at 30 °C for 16 hours and chilled at 4 °C for 1 hour prior to imaging. (C) Transformation efficiencies of strains bearing the indicated dnaN allele using pWSK29 (control plasmid) or pJCB200 (hda⁺) are shown.

We previously identified 4 intragenic suppressors of dnaN159 that alleviated temperature sensitive growth, named dnaN780 (G66E), dnaN781 (G174A), dnaN782 (G66A, G174A), and dnaN783 (P20L, G66E, G174A) (Maul et al., 2007). Since each of these alleles shares one or both mutations with dnaN159, we reasoned we could use isogenic strains bearing these alleles to determine whether NFZ sensitivity required the G66E and/or the G174A substitution. Although dnaN780 and dnaN781 displayed a near wild type-like resistance to NFZ, the dnaN782 and dnaN783 strains were similar to dnaN159, indicating that NFZ sensitivity relied on the presence of both a G66 and G174 mutation. We next asked whether a correlation existed between NFZ sensitivity and the ability of elevated levels of Hda to impede growth. As summarized in Fig. 8, elevated levels of Hda failed to impede growth of strains bearing the dnaN780 or dnaN781 alleles. In contrast, those bearing dnaN782 or dnaN783 were as sensitive to Hda as was the dnaN159 strain. Taken together, these findings reveal a remarkable correlation between NFZ sensitivity and the ability of elevated levels of Hda to impede growth of E. coli.

Discussion

With the goal of gaining insight into how the E. coli β clamp coordinates the actions of Hda in RIDA with those of other proteins involved in DNA replication, we analyzed phenotypes of strains bearing a low copy number plasmid expressing ~5-fold higher than normal levels of Hda. Although this plasmid failed to affect growth of a wild type E. coli strain, it nevertheless conferred severe sensitivity to NFZ, suggesting that Hda might influence functions of Pol II and Pol IV in vivo. Consistent with this conclusion, elevated levels of Hda reduced the frequency of −1 frameshift mutations characteristic of Pol IV. Moreover, the same Hda plasmid impaired growth of an otherwise isogenic dnaN159 strain in a Pol II- and Pol IV-dependent manner. Since strains bearing the dnaN159 allele are impaired for managing the actions of their Pols, we took advantage of this phenotype to identify 24 novel hda mutations that were unable to impede growth of the dnaN159 strain. Genetic characterization of these hda mutations revealed four phenotypic classes (Table 6). One of these classes (mutants G157N, D167N, R196Q, D212N) was severely impaired for RIDA function in vivo, but nevertheless supported viability of the Δhda::cat E. coli strain at 30°C. Taken together, these findings illustrate important cellular functions for Hda in addition to its established role in RIDA, and suggest that although one or more Hda functions are essential for cell viability, the ability to catalyze RIDA may be dispensable for cell growth.

TABLE 6.

Summary of mutant hda phenotypes.

Phenotypic Class	hda alleles	Viabilitya	RIDAb	Replicationc
Class I	D15G, R141Q, D155N, G182A, D209G, L222P, P225A	+	+	+
Class II	A62E, S68P, E114G, Y121H, L133S, P140R, A178V	+	+	−
Class III	G157V, D167N, R196Q, D212N	+	−	−
Class IV	L98P, G136D, L233F, D198fs, Δ(LDQ208-210), Ω(LDQ208-210)	−	−	−

^aViability was defined as the ability (+) or inability (−) of the plasmid-expressed allele to complement growth of the Δhda::cat strain at 30 °C in the absence of suppression.

^bRIDA proficiency (+) was defined as an oriC:terC ratio < 6, and RIDA deficiency (−) was > 6.

^cReplication proficiency was determined by flow cytometry demonstrating completion of 2–3 distinct and predominant chromosomal populations (+), or greater than three (or lack of) chromosomal populations (−).

Novel hda mutations cluster into four distinct classes

Utilizing a genetic selection that exploits a synthetic lethality between dnaN159 and elevated expression of hda, 24 novel hda mutations were identified. Mutants were characterized through genetic complementation for their ability to (1) support growth at 30°C, (2) complete replication through cell cycle analysis, and (3) regulate initiation through analysis of oriC:terC ratios (Table 6). Based on these criteria, the 24 hda alleles clustered into four distinct classes. Class I supported viability at 30°C, and were proficient for both replication and regulating initiation. With the exception of R141Q, D155N, and P225A, which were asynchronous, mutations in this class were indistinguishable from hda⁺. The second class of mutants (class II) was proficient for both growth at 30°C and regulating initiation, but exhibited a replication defect as depicted by multiple (greater than 3) predominant chromosomal populations seen by flow cytometry. The third class (class III) supported growth at 30°C, but displayed obvious defects in regulation of initiation and replication. These mutations genetically separate roles of Hda in catalyzing RIDA and regulating initiation from those required for viability of E. coli. The final class of mutants (class IV) were indistinguishable from the Δhda::cat allele in that they failed to support growth of E. coli and displayed severe defects in regulation of initiation, and were completely asynchronous.

When we represented the positions of mutations in each class onto a model for E. coli Hda, some correlations between positions and phenotypes were apparent. Although many mutations were distributed throughout the protein, class II mutants reside predominantly in the N-terminal globular region of the protein (Fig. 5C, right half of structures). Class IV mutants clustered primarily in the last two α-helices (α7 and α8). Most of the mutations in this class alter these two helices (D198fs, ΔLDQ208-210, and ΩLDQ208-210), illustrating the importance of these secondary structures for viability in E. coli. Interestingly, L233F, although not in any striking position to dramatically alter secondary structure in the C-terminus, is also important for viability in E. coli.

Several mutations in Hda have been described to date (Feeney et al., 2012, Fujimitsu et al., 2008, Kato & Katayama, 2001, Nakamura & Katayama, 2010, Su’etsugu et al., 2008). Kato et al. (Kato & Katayama, 2001) described mutations residing in the CBM or the arginine finger of Hda that impaired RIDA in vitro: here we demonstrated that strains bearing these hda alleles were inviable at 30°C (Table 4). Surprisingly, our genetic selection yielded no novel mutations within the CBM of Hda. Multiple mutations may have been necessary to disrupt this interaction. Nakamura et al. designed mutations targeting Box VI and VII motifs characteristic of AAA⁺ proteins, demonstrating that mutations of conserved residues in these motifs (F118A, N122A, S152A) broke interaction with DnaA and consequently were broken for RIDA as well (Nakamura & Katayama, 2010). G157V is located just outside of the Box VII motif and is exposed to the same face of Hda believed to be important for DnaA interaction (Keyamura & Katayama, 2011) (Fig. 5), likely lending to its asynchronous phenotype. Similarly, Y121H resides in Box VI, likely impairing interaction with DnaA through Box VI.

The ability of elevated levels of Hda to impede growth of the dnaN159 strain relies on Pol II and Pol IV, and correlates with NFZ-sensitivity

Strains bearing the dnaN159 allele are hypersensitive to elevated levels of Hda. Interestingly, disruption of either the polB or dinB genes, encoding Pol II and Pol IV, respectively, relieved this hypersensitivity. Although loss of Pol II or Pol IV could potentially reduce β159 clamps on DNA available to participate in RIDA, we found that at elevated temperatures, which reduce the protein levels of β159 in vivo, this hypersensitivity to elevated levels of Hda is unchanged, suggesting that the hypersensitivity is likely not RIDA-dependent. These results suggest that elevated levels of Hda interfere with the ability of the β159 clamp protein to manage the actions of these Pols, as well as possibly other clamp partners. Our finding that neither the lexA3(Ind⁻) nor lexA51(Def) alleles were able to rescue Hda sensitivity indicates that basal levels of Pol II or Pol IV are sufficient for lethality. Moreover, our finding that the ability of Hda to confer lethality is independent of SOS induction is consistent with our finding that disruption of the umuDC-encoded Pol V fails to contribute to this phenotype. Pol V is undetectable in SOS repressed cells, and accumulates to detectable levels only after SOS induction (Elledge & Walker, 1983, Friedberg et al., 1995, Gonzalez et al., 1998). Finally, overexpression of Hda was previously reported to induce the SOS response (Banack et al., 2005). Based on flow cytometry, we failed to detect cell filamentation of strains expressing wild type or mutant Hda proteins: strains induced for SOS display SulA-dependent cell filamentation, which is apparent based on side and forward scatter in flow cytometry ((Horvath et al., 2011, Wickens et al., 2000) and data not shown). The failure of the RIDA-deficient hda mutants to induced SOS suggests that either replication forks are not collapsing in these strains, or mechanisms exist to deal with the extra initiation events.

Pol IV, and possibly Pol II, contribute to the ability of E. coli to tolerate NFZ (Jarosz et al., 2006, Williams et al., 2010). Pol IV plays a well established role in bypassing N²-dG adducts that are created by exposure to NFZ (Jarosz et al., 2006, Sanders et al., 2006, Yuan et al., 2008). Although NFZ sensitivity conferred by elevated levels of Hda does not appear to be epistatic with polB or dinB (Fig. 2), we cannot rule out a role for Hda in managing the actions of these Pols during TLS in vivo. Indeed, our finding that elevated levels of Hda suppressed −1 frameshift mutations characteristic of Pol IV ~10-fold supports a role for Hda in managing the actions of this Pol in vivo.

Our finding that each of the 24 Hda mutations identified failed to confer NFZ sensitivity, taken together with our observation that the dnaN159 allele sensitized E. coli to NFZ, strongly implies that the mechanism underlying this phenotype serves as the basis for the growth defect conferred by elevated levels of Hda. This correlation is further supported by phenotypes of other dnaN alleles. The dnaN780 (G66E), dnaN781 (G174A), dnaN782 (G66A, G174A), and dnaN783 (P20L, G66E, G174A) alleles were originally identified as intragenic suppressors of the temperature sensitive growth phenotype of dnaN159 (Maul et al., 2007). Phenotypes of isogenic strains bearing these dnaN alleles revealed a remarkable correlation between NFZ sensitivity and the ability of elevated levels of Hda to impede growth (Fig. 8): dnaN780 and dnaN781 were resistant to both NFZ and Hda, while dnaN782 and dnaN783 were sensitive to both. Taken together, these findings indicate that NFZ sensitivity of the dnaN159 strain is independent of temperature sensitivity, and strongly suggests that the mechanism by which dnaN159 sensitizes E. coli to NFZ is the same, or at least shared with, the mechanism by which elevated levels of Hda impedes growth.

Evidence supporting a role for Hda beyond RIDA

It was previously suggested that a capacity to catalyze RIDA is essential for cell viability (Katayama, 1994, Katayama & Crooke, 1995, Kellenberger-Gujer et al., 1978). Kato and Katayama suggested that loss of RIDA was lethal due to an inability to regulate initiation of replication (Kato & Katayama, 2001). More recently, Riber et al. (Riber et al., 2006) suggested that loss of RIDA may be lethal due to an inability to properly express genes whose transcription is regulated in part by ATP-DnaA. Phenotypes of several hda alleles identified in this work provide evidence that hda is required for viability due to a function other than RIDA. Class II mutants appear to be competent for RIDA, but show an obvious defect in replication (Fig. 6b). Each of the hda mutations belonging to class III support viability of the Δhda::cat E. coli strain, despite their inability to properly regulate initiation of replication, as measured by both flow cytometry and oriC:terC ratios (Table 4). R196Q distinguished itself from other mutants in this class in that it mirrors the phenotype for the suppressed Δhda::cat strain in every respect, yet is nevertheless able to support viability of E. coli at 30°C in the absence of suppression. Although we cannot eliminate the possibility that these mutant Hda proteins retain an ability to catalyze sufficient levels of RIDA to support viability, their phenotypes nevertheless argue that it is possible to genetically distinguish roles of Hda in RIDA from those required for viability of E. coli. In contrast to the class III hda alleles discussed above, the class IV mutations are impaired for both RIDA and supporting cell viability. Likewise, mutation of the Hda CBM or arginine finger are also lethal. Thus, certain Hda functions appear to be common to both RIDA and viability.

The use of processivity clamps to regulate initiation of replication is not unique to E. coli. Homologous to E. coli Hda, Caulobacter crescentus HdaA catalyzes RIDA using an analogous mechanism to regulate initiation through inactivation of DnaA (Collier & Shapiro, 2009). By a non-homologous mechanism, YabA, a negative regulator of replication initiation in Bacillus subtilis, interacts with the initiator DnaA and DnaN processivity clamp to limit the amount of DnaA present at oriC in B. subtilis (Goranov et al., 2009). Several results from our study suggest an interaction of Hda with the replication machinery, similar to what has been reported for C. crescentus and B. subtilis. These include: (1) the dependence on Pol II or Pol IV function for elevated levels of Hda to impede growth of the dnaN159 strain; (2) the ability of elevated levels of Hda to inhibit −1 frameshift mutations, and confer NFZ sensitivity; and (3) the finding that several mutant Hda proteins proficient for RIDA in vivo (based on a wild type oriC:terC ratio) displayed a replication defect as measured by flow cytometry. These findings, taken together with other results discussed above, suggest that the requirement of Hda for viability may involve its association with the replisome in a manner that is similar to that depicted in our computational model for the Hda-β clamp complex (Fig. 1). Although we cannot rule out the possibility that some of the mutants described here retain a limited ability to catalyze RIDA despite their initiation defect, we suggest that the essential function of Hda at 30°C may be to stabilize Pol III at the replication fork, and possibly regulate access of Pol II and Pol IV during replication. Although not reported, it is possible that Hda interacts with additional components of the replisome besides the clamp. Alternatively, the mere presence of Hda on the rim of the clamp may serve to limit access of other proteins, such as Pol II and Pol IV, which could impede replication, particularly if they were able to gain frequent access to the fork. Importantly, interactions of Hda with the replisome may not only underly its essential function, but are also presumably required for RIDA. This would explain why some mutations, such as those targeting the CBM, the arginine finger, or the C-terminus of Hda disrupt both RIDA and cell viability. Finally, this type of a role for Hda is consistent with several of the phenotypes observed for hda mutations discussed above, particularly the cold sensitivity phenotype of the Δhda::cat strain, which is common for mutations affecting stability of multi-protein complexes (Boye et al., 2001, Gines-Candelaria et al., 1995). We are currently testing critical aspects of this model.

Experimental Procedures

Bacterial strains and plasmid DNAs

All E. coli strains, plasmid DNAs, and oligonucleotide primers used in this study are described in Table 7. E. coli strains used in this work were derived from either MS100 or MG1655 using generalized P1vir transduction. Unless otherwise stated, strains were grown in Luria-Bertani (LB; 10 g•l⁻¹ tryptone, 5 g•l⁻¹ yeast extract, 10 g•l⁻¹ NaCl) medium (Miller, 1992). When necessary, the following antibiotics were included at the indicated concentrations: ampicillin (Ap), 150 μg•mL⁻¹; chloramphenicol (Cm), 20 μg•mL⁻¹; kanamycin (Km), 40 μg•mL⁻¹; tetracycline (Tc), 10 μg•mL⁻¹. The E. coli hda gene and 225 base pairs of upstream sequence containing the promoter were PCR amplified from genomic DNA using the Hda.forwardN2 and Hda.backwrd primers. The amplified DNA was digested with EcoRI/HindIII and ligated into pWSK29 that had been digested similarly. Ligation reactions were transformed into chemically competent E. coli DH5α. Representative ampicillin resistant transformants were verified by PCR amplification of the gene with primers pWSK.For and pWSK.Rev, and the correct sequence of the cloned fragment was verified by automated nucleotide sequence analysis (Roswell Park Cancer Institute Biopolymer Facility, Buffalo, NY).

TABLE 7.

E. coli strains, plasmid DNAs, and oligonucleotides used in this study.

E. coli strains
Strain	Relevant genotype	Source or construction
BL21(DE3)	B F− dcmompThsds (rB-mB-) gal λ (DE3)	Novagen
ALO1917	Δhda::cat	(Riber et al., 2006)
JC126	ΔdatA::kan	(Camara et al., 2005)
JC326	ΔseqA::tet	(Camara et al., 2005)
CC108	ara-600 Δ(gpt-lac)5 Δ(lac-proB) [F lacI373 lacZ(6G→7G) proB+] λ− relA1 spoT1 thi-1	(Cupples et al., 1990)
KM52	ΔmutL460::cat	(Loh et al., 2001)
KM55	ΔmutH461::cat	(Loh et al., 2001)
KM75	ΔmutS465::cat	(Loh et al., 2001)
MS100	thr-1 araD139 Δ(gpt-proA)62 lacY1 tsx-33 supE44 galK2 hisG4(Oc) rpsL31 xyl-5 mtl-1 argE3(Oc) thi-1 sulA211tnaA300::Tn10	(Sutton, 2004)
MS101	MS100: dnaN159	(Sutton, 2004)
MS102	MS100: lexA3(Ind−)	(Sutton, 2004)
MS103	MS101: lexA3(Ind−)	(Sutton, 2004)
MS104	MS100: lexA51(Def)	(Sutton, 2004)
MS105	MS101: lexA51(Def)	(Sutton, 2004)
FC1237	ara Δ(lac-proB) Δ(dinB-yafN)::kan	(Strauss et al., 2000)
JCB150	MS100: Δ(dinB-yafN)::kan	P1(FC1237)×MS100
MS125	MS101: Δ(dinB-yafN)::kan	(Sutton & Duzen, 2006)
STL1336	thr-1 ara-14 leuB6 Δ(gpt-proA)62 lacY1 tsx-33 supE44 galK2 hisG4(Oc) rfbD1 mgl-51 rpsI31 kdgK51 xyl-5 mtl-1 argE3(Oc) thi-1 Δ(araD-polB)::Ω	M. Goodman via J. Courcelle (Rangarajan et al., 1999)
JCB151	MS100: Δ(araD-polB)::Ω	P1(STL1336)×MS100
JCB152	MS101: Δ(araD-polB)::Ω	P1(STL1336)×MS101
JCB153	MS100: Δ(araD-polB)::Ω Δ(dinB-yafN)::kan	P1(STL1336)×JCB150
MS106	MS100: ΔuvrB::cat	(Sutton, 2004)
MS108	MS100: Δ(umuDC)595::cat	(Sutton, 2004)
MS109	MS101: Δ(umuDC)595::cat	(Sutton, 2004)
MG1655	rph-1	CGSC
JCB100	MG1655: Δhda::cat	P1(ALO1917)×MG1655
JCB101	MG1655: ΔdatA::kan	P1(JC126)×MG1655
JCB102	MG1655: ΔseqA::tet	P1(JC326)×MG1655
JCB103	MG1655: dnaN tnaA300::Tn10	P1(MS100)×MG1655
JCB104	MG1655: dnaN159 tnaA300::Tn10	P1(MS101)×MG1655

Plasmid DNAs
Plasmid	Relevant features	Source or construction
pWSK29	ApR; pSC101 origin, general cloning vector	(Wang & Kushner, 1991)
pJCB200	ApR; pWSK29 derivative that bears hda on a HindIII-BamHI fragment downstream of Plac	This work
pJCBQ6A	ApR; pJCB200 derivative that bears hda containing a Q6A mutation	This work
pJCBL9A	ApR; pJCB200 derivative that bears hda containing a L9A mutation	This work
pJCBR153A	ApR; pJCB200 derivative that bears hda containing a R153A mutation	This work
pJCBR153M	ApR; pJCB200 derivative that bears hda containing a R153M mutation	This work
pET16b-Hda	ApR; pET16b derivative that bears hda fused to a 10xHis tag downstream of PT7	This work
pACYC184	CmR; p15A origin, general cloning vector	(Chang & Cohen, 1978)
pMMF57	CmR; pACYC184 derivative bearing seqA	(Felczak & Kaguni, 2009)
pMMF84	CmR; pACYC184 derivative bearing datA	(Felczak & Kaguni, 2009)

Oligonucleotides
Primer	Sequence (5′-3′)
oriC_1	CTGTGAATGATCGGTGATCC
oriC_2	AGCTCAAACGCATCTTCCAG
terC_1	CAGAGCGATATATCACAGCG
terC_2	TATCTTCCTGCTCAACGGTC
Had.forwardN2	GCGAATTCCCGGGTATGGCGCTGGCGACC
Hda.backwrd	CACGGCTCCCCTTATCTACAACTTCAG
pWSK.For	AGAAAGGAAGGGAAGAAAGCGAAAGGAGC
pWSK.Rev	TTAGGCACCCCAGGCTTTACACTTTATGC
Hda.R153A.T	CCGGATCTCGCGTCGGCACTCGACTGGGGGCAG
Hda.R153A.B	CTGCCCCCAGTCGAGTGCCGACGCGAGATCCGG
Hda.R153M.T	CCGGATCTCGCGTCGATGCTCGACTGGGGGCAG
Hda.R153M.B	CTGCCCCCAGTCGAGCATCGACGCGAGATCCGG
Hda.Q6A.T	CCGGCACAGCTCTCTGCGCCACTTTATCTTCCT
Hda.Q6A.B	AGGAAGATAAAGTGGCGCAGAGAGCTGTGCCGG
Hda.L9A.T	CTGAACACACCGGCAGCGCTCTCTTTGCCACTT
Hda.L9A.B	AAGTGGCAAAGAGAGCGCTGCCGGTGTGTTCAG

Nitrofurazone sensitivity and spontaneous frameshift mutation frequency assays

Sensitivity to nitrofurazone (NFZ) was performed as previously described (Sanders et al., 2006). Briefly, strains were grown overnight in LB media to saturation. Cultures were serially diluted 10-fold and spotted onto LB plates with the indicated amounts of NFZ. Plates were incubated for 16 hours at the appropriate temperature (either 30° or 37°C) and moved to 4°C for at least 60 minutes to halt growth before imaging.

Transduction and spotting assays for cold-sensitivity

Strain construction by P1 transduction was performed as described (Miller, 1992). Transduction mixtures were scaled up in volume, spread onto LB supplemented with Cm and 5 mM sodium citrate, and subsequently incubated for 30 hours at 30°, 37°, or 42°C. Fresh transductants were picked from plates incubated overnight at 37°C, inoculated into LB media with Cm and 5mM sodium citrate, and incubated at 37°C for 16 hours. Cultures were serially-diluted 10-fold, and 10 μL spots of each dilution were spotted onto LB plates followed by incubation for 16 hours at 30°, 37°, or 42°C.

Generation of rabbit polyclonal anti-Hda antibody

The Hda protein N-terminally tagged with 10xHis was overproduced using pET16b-Hda in the BL21(DE3) E. coli strain. The tagged protein was purified by nickel affinity chromatography (GE Healthcare) under denaturing conditions with 6M guanidine-HCl and used to immunize rabbits for the production of polyclonal antibodies (Sigma-Aldrich).

Genetic assay for the selection of novel hda alleles

In order to identify mutations in the hda gene that permit transformation of strain MS101 (dnaN159), pJCB200 was passaged through strains lacking mutL (KM52), mutH (KM55), or mutS (KM75) function (Loh et al., 2001). MS101 competent cells were transformed with either pJCB200, or a preparation that was passaged through mismatch repair deficient strains to increase likelihood of mutations, plated onto LB medium for selection of transformants able to grow at 30°C. Twenty-seven of 27 transformants using unpassaged pJCB200, and 113 mutH-passaged, 212 mutL-passaged and 242 mutS-passaged pJCB200 transformants were selected. One hundred twenty transformants were screened by Western Blot analysis to ensure Hda protein expression was comparable to that of the isogenic MS100 (dnaN⁺) strain bearing pJCB200 (hda⁺). Plasmid DNAs were extracted from forty transformants expressing proper levels of Hda. Plasmids were transformed into MS101 (pWSK29, ~10⁶ cells•μg⁻¹ DNA) to confirm hypersensitivity suppression. The hda cassette was PCR amplified from these 40 plasmids using primers pWSK.For and pWSK.Rev, and subjected to automated nucleotide sequence analysis (Roswell Park Cancer Institute Biopolymer Facility, Buffalo, NY) using the same pWSK.For and pWSK.Rev primers. Resultant sequencing traces were analyzed and compared to the hda⁺ reference sequence using MacVector 10.0 software. Twenty-four unique mutations were identified from the 40 clones that were sequenced.

Flow cytometry and quantitative PCR analysis of mutant hda strains

DNA content per cell was determined essentially as described (Ferullo & Lovett, 2008). Briefly, overnight cultures were subcultured 1:1,000 in LB media and incubated at 37°C or 30°C as noted, with aeration, to an optical density at A₆₀₀ ≈0.2. A portion of the cultures was harvested into tubes and kept on ice, cells were collected by centrifugation, resuspended in Tris-EDTA-glycine buffer (TEG) with 50 μg•mL⁻¹ RNAse A, after which genomic DNA was extracted using phenol:chloroform:isoamyl alcohol (pH 8.0), and backwashed with chloroform. Genomic DNA was precipitated with sodium acetate and ethanol at −80°C, resuspended in 50 mM Tris (pH 8.0)-0.1 mM EDTA, for use in quantitative PCR analysis. The remaining culture was treated with 500 μg•mL⁻¹ rifampin and 25 μg•mL⁻¹ cephalexin to arrest protein synthesis and cell division. Cultures were incubated for an additional 2 hours at 37°C with aeration. 1 mL of culture was fixed in 10 mL 70% ethanol and stored at 4°C for >16 hours. Fixed cells were resuspended in 500 μL phosphate-buffered saline (PBS; pH 7.4). 100 μL of resuspended cells were incubated with 20 μL 1% PicoGreen dye (Invitrogen) in 25% DMSO for 1 hour. Samples were analyzed using a Becton Dickinson FACSCalibur flow cytometer equipped with BD CellQuest™ v3.3 software and subsequently analyzed with FCS Express v3.00. Quantitative PCR was performed according to manufacturer’s instructions (Bio-Rad). Briefly, PCR reactions contained 12.5 μL 2× iQ™ SYBR® Green Supermix, 0.5 μL (1:10 diluted) genomic DNA, and 100 nM oriC_1/oriC_2 primer pair or terC_1/terC_2 primer pair in a 25 μL volume. PCR reactions were heated to 95°C for 3 min, followed by 40 cycles with steps of 95°C, 56°C and 72°C for 30 s each. The generation of specific PCR products was confirmed using melting curve analysis.