Ea22 Proteins from Lambda and Shiga Toxin-Producing Bacteriophages Balance Structural Diversity with Functional Similarity.

Enterohemorrhagic Escherichia coli (EHEC) outbreaks are commonly associated with contaminated food sources. Unlike normal intestinal bacteria, EHEC are lysogens of lambdoid bacteriophages that also carry a gene for Shiga toxin. Oxidative attack by the immune system or other stressors on the bacterial host can activate the lytic pathway of the latent phage genome to produce phage progeny and the release of Shiga toxin into the surrounding tissues. Within the genomes of bacteriophage λ and Shiga toxin-expressing (Stx+) phages such as φ24B and φP27, there is a conserved set of open reading frames that is located between the exo and xis genes that influences the lysogenic-lytic decision. In this report, we have focused on the largest exo-xis region open reading frame termed ea22 that has been shown previously to have prolysogenic properties. Using a variety of biophysical and bioinformatic methods, we demonstrate that λ and φP27 Ea22 proteins are tetrameric in solution and can be considered in terms of an amino-terminal region, a central coiled-coil region, and a carboxy-terminal region. The carboxy-terminal regions of λ and φ24B Ea22, expressed on their own, form dimers with exceptional thermostability. Limited proteolysis of φP27 Ea22 also identified a C-terminal region along the predicted boundaries. While the three Ea22 proteins all appear to have the hallmarks of a domain in their respective C-terminal regions, each sequence is remarkably dissimilar. To reconcile this difference among Ea22 proteins from λ and Stx+ phages alike, we speculate that each Ea22 may achieve the same function by targeting different components of the same regulatory process in the host.

Introduction

Escherichia coli (EHEC) outbreaks due to contaminated meat, vegetable, and water supplies have been recorded since early 1980 with a spectrum of health issues ranging from mild gastrointestinal problems to kidney failure.[1,2] Unlike normal intestinal bacteria, EHEC carry a stably integrated bacteriophage genome that encodes a ribosome-inactivating Shiga toxin (typically Stx2).[3,4] When EHEC are exposed to a variety of environmental stressors such as oxidative attack from the immune system, the endogenous phage genome transitions from a dormant lysogenic state to a lytic state along with the production and release of Shiga toxin.[5,6] Antibiotic treatment, typically the first choice in a bacterial infection, can be counterproductive in EHEC infections because this treatment represents another type of stress that may drive the endogenous phage toward a lytic response. Since Shiga toxin-producing (Stx+) phages belong to the same family as bacteriophage λ,[7] a molecular level comparison may reveal new ways in which Stx+ phages contribute to the pathogenicity of EHEC and potentially new therapeutic leads to combat infection.[8,9]

A relatively unknown region between the exo and xis genes of λ and Stx+ phages is known to affect the host cell-cycle[10,11] and the lysogenic–lytic transition[12,13] yet, paradoxically, is also dispensable for normal viral development. This so-called exo–xis region consists of four open reading frames termed ea22,[14]orf73, orf61,[15] and orf63.[16] While the overall number of exo–xis region genes in Stx+ phages can be larger, these four genes are conserved, suggesting that there may be some evolutionary pressure to maintain them. The hybrid zinc finger/homeodomain fold of Ea8.5 suggests that it may serve as a transcription factor or play another role associated with nucleic acid binding.[17] Less is known about Orf63, except that it has an oligomeric two-helix fold.[16]

Ea22 is the largest protein of the exo–xis region and is expressed early in the lytic cycle by the pL promoter. Unlike other exo–xis genes studied to date that accelerate the development of the lytic state,[15,16,18]ea22 promotes the maintenance of the lysogenic state.[14] Before we embarked on this study, it was not known if ea22 even encoded a protein although the functional data appeared to suggest it was the case. With a set of expressed and purified Ea22 proteins and protein fragments from λ phage and two representative Stx+ phages, φP27[19] and φ24B,[12] we demonstrate that Ea22 is a multidomain protein that can be considered in terms of an amino-terminal region, a central coiled-coil region, and a carboxy-terminal region. In this report, we have used circular dichroism spectroscopy, nuclear magnetic resonance spectroscopy, differential scanning calorimetry, and limited proteolysis to show that the C-terminal region of each Ea22 protein has the qualities of a protein domain, but each C-terminal domain is not necessarily similar. We consider this observation in terms of the functional differences between λ and Stx+ phage.

Materials and Methods

Cloning and Expression

Several full-length genes and gene fragments of λ_Ea22, φP27_Ea22, and φ24B_Ea22 were synthesized and placed in an N-terminal 6× His-tagged, T5 promoter-driven expression vector by ATUM (Newark CA): Clone #134893 λ_Ea22(1–182), #179655 λ _Ea22(93–182), #179656 λ_Ea22(102–182), #179657 λ_Ea22(109–182), #179658 λ_Ea22(119–182), #350149 φP27_Ea22(1–62), #350148 φP27_Ea22(145–315), #358023 φP27_Ea22 (176–315), #350147 φ24B_Ea22(159–301), #351679 φ24B_Ea22(164–269), #393816 φP27_Ea22(235–315). Full-length N-terminal 6× His-tagged φP27_Ea22(1–315) and φ24B_Ea22(1–301) were made by inserting a suitable gene fragment into the I and HI sites of pET28b (Novagen). For milligram quantities of a purified protein, a 1.5–3.0 L culture of E. coli BL21(DE3) containing 50 μg/mL kanamycin was initially grown at 37 °C with shaking to an A600 of 0.5 and then chilled to either 16 °C (λ _Ea22) or 25 °C (φP27_Ea22, φ24B_Ea22) prior to induction with 1 mM isopropylthiogalactopyranoside (IPTG) for 18 h (λ _Ea22) or 6 h (φP27_Ea22, φ24B_Ea22). The cell pellet was harvested by centrifugation (4000g, Beckman JA-10 rotor), resuspended in T300 buffer (10 mM Tris, 300 mM NaCl, 0.05% NaN3, pH 7.7), and lysed by French press and sonication. The soluble fraction was separated from the cell debris by centrifugation (28 000g, Beckman JA-20 rotor) and applied to a 5 mL Nuvia nickel affinity column (BioRad). Next, the affinity column was washed with T300 + 10 mM imidazole and the protein of interest was eluted with T300 + 500 mM imidazole. The eluate was concentrated to 5 mL, treated with 20 mM dithiothreitol (DTT), and applied to a gel filtration column (HiPrep 16/60 S100, GE Life Sciences) that had been pre-equilibrated to 10 mM Tris, 100 mM NaCl, 0.05% NaN3, pH 7.7. Suitable fractions from the column were assessed for purity by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), pooled, and concentrated accordingly. Protein concentrations were estimated from A280. For studies requiring protein concentrations of φP27_Ea22 >2 mg/mL, the buffer NaCl concentration was increased to 0.5 M to improve protein solubility.

Nuclear Magnetic Resonance (NMR) Spectroscopy

Proteins and protein fragments were uniformly 15N-labeled by following the protocol previously described except that M9 minimal medium was used (6 g of Na2HPO4, 3 g of KH2PO4, 1 g of 15NH4Cl, 0.5 g of NaCl, and 10 g of glucose in 1 L of water supplemented with 1 mM CaCl2, 1 mM MgSO4, 50 μg/mL kanamycin, and a trace mineral mix). The samples were concentrated to 0.2 mM and supplemented with 10% D2O. 1H,15N heteronuclear single quantum coherence (HSQC) spectra were acquired at various temperatures on a 700 MHz Bruker Avance-III spectrometer with a nitrogen-cooled triple-resonance probe. Datasets were processed with NMRPipe[20] and interpreted with CCPN analysis v2.4.[21]

Circular Dichroism (CD) Spectroscopy

Protein samples were dialyzed to 10 mM Tris, 50 mM NaCl, 0.05% NaN3, 5 mM DTT, pH 7.7 and standardized to a 20 μM concentration for a 0.1 cm path-length quartz cell. Spectra were acquired from 260 to 200 nm using a Jasco J810 instrument equipped with a six-position temperature-regulated cell holder. Thermostability was assessed by holding the wavelength constant at 222 nm and ramping the temperature from 20 to 90 °C at a rate of 2 °C/min. Signals were converted from mdeg to mean residue ellipticity and plotted with Pro Fit (Quansoft).

Differential Scanning Calorimetry (DSC)

Protein samples were treated with 20 mM DTT, standardized to 1 mg/mL, and then dialyzed to 10 mM Tris, 0.1 M NaCl, 0.05% NaN3, pH 7.7. In cases where the solubility was limited, the NaCl concentration was adjusted to 0.5 M. Before analysis, the samples were centrifuged at 24 000g for 5 min at 4 °C to remove any precipitates. The protein and matched buffer were loaded into the sample and reference cells, respectively, and the temperature was increased from 20 to 100 °C at 1 °C/min at 3 atm pressure. Background subtracted data were fit to the simplest model (single or multistate) that conformed to a 99% confidence level.

Size-Exclusion Chromatography with Multi-Angle Laser Scattering (SEC-MALS)

Protein samples were prepared in the same manner for DSC experiments. The chromatography unit of the instrument was configured with an Infinity-II HPLC (Agilent) and an AdvanceBio SEC 300 Å column (Agilent). The chromatography system was equilibrated in the sample buffer for 18 h before the first injection for optimal stability. The detection section of the instrument was configured with a MiniDAWN TREOS MALS and OptiLab T-rEX refractive index components (Wyatt). Prior to the analysis, a 2 mg/mL bovine serum albumin (BSA) standard was injected to calibrate the peak and retention time characteristics of the flow cells. After calibration, 20 μL of the protein sample was injected. Chromatograms were analyzed and molecular masses determined with ASTRA software (Wyatt).

Bioinformatics

Sequences were submitted to NetSurfP-2.0[22] for secondary structure analysis. Coiled-coil prediction was performed with the programs, DeepCoil, MARCOIL, and PCOILS, all part of the HHpred suite.[23]

Limited Proteolysis

A 20 μM solution of φP27_Ea22(1–315) in a buffer containing 20 mM Tris, 150 mM NaCl, 10 mM DTT, pH 7.7 was digested with 200 nM of proteomic-grade trypsin (Sigma-Aldrich) for 30 min at 37 °C. The digestion was stopped with either 2× SDS sample loading solution for electrophoretic assays or 1% formic acid (v/v) for liquid chromatography coupled with mass spectrometry (LC-MS) assays.

Mass Spectrometry

Electrospray ionization (ESI) mass spectra were acquired on a 6538 UHD model quadrupole time-of-flight mass spectrometer coupled to a 1260 Infinity model high-performance liquid chromatography system (Agilent Technologies). Approximately 1 pmol of protein mixture was separated with a 50 × 2 mm2 Jupiter 5 μm C4 300 Å reverse-phase column using a flow rate of 0.25 mL/min. After a 1 min wash with 0.1% (v/v) formic acid and 5% acetonitrile, a 15 min separating gradient to 60% acetonitrile was applied. MS data were processed and analyzed with MassHunter (Agilent).

Bacterial Survival Assay

A bacterial culture was grown in LB at 30 °C to A600 = 0.2. From a 1 mL aliquot, bacteria were harvested by centrifugation, washed with 0.85% NaCl, and centrifuged again. In preparation for infection, the pellet was resuspended in 1 mL of TCM buffer (10 mM Tris–HCl, 10 mM MgSO4, and 10 mM CaCl2, pH 7.2). After a 30 min incubation at 30 °C, the phage was added to m.o.i. of 1 and maintained for 15 min. Serial dilutions were made in 0.85% NaCl and plated. After an overnight incubation at 37 °C, the percentage of surviving E. coli was calculated relative to a culture in which TCM buffer was added instead of the phage. Bacteria, bacteriophage, and plasmids for this assay are summarized in Table .

Table 1

Strains and Bacteriophages Used in this Study

source	genotype
E. coli strains
MG1655	F– λ– ilvG rfb-50 rph-1
MG1655 (λ)	MG1655 bearing λ prophage
MG1655 (φ24B)	MG1655 bearing φ24B prophage
MG1655 (φP27)	MG1655 bearing φP27 prophage
bacteriophages
λ	carries a frameshift mutation relative to Ur-lambda
Φ24B	Δstx2::catGFP
P27	Δstx2::catGFP

Results

Sequence Analysis of Ea22 Proteins

Given the maturity of the bacteriophage field, it is remarkable that there have been few explorations of the exo–xis region. One reason for the lack of interest may be the dispensability of the exo–xis region for viral development. As shown in Figure a, the number of open reading frames in the exo–xis region is variable among lambdoid phages; however, a core set of four genes is retained (orf63, orf61, orf73, and ea22), suggesting some functional significance. Before we embarked upon the expression studies of Ea22 from λ and two representative Stx phages, φP27 and φ24B, the secondary structure of each protein was predicted and used to supplement a sequence alignment. Overall, the alignment suggests that Ea22 is partitioned into three regions (Figure b). In contrast to φ24B and φP27 Ea22, the N-terminal region of λ Ea22 appears to be truncated to a degree that a potential domain is lost. The central region of Ea22 is predicted to contain several long α-helical segments, which suggests a coiled-coil structure. The C-terminal regions of the three Ea22 proteins are the most intriguing since they bear little, if any, resemblance to each other even at the secondary structure level.

Figure 1

Gene map of the exo–xis regions from φP27, φ24B, and λ phages. (a) Ea22 is highlighted in magenta and other genes that are typically conserved among lambdoid phages are blue. For clarity, some ORFs of φP27, φ24B, and λ are shown without their prefixes of vBEcoSP27_ and vB_24B_, and lambda. (b) Secondary structure prediction of Ea22 from the λ phage and the Stx phages φP27 and φ24B. Predicted α-helices (red), β-strands (blue), coiled-coil regions (green), and extended regions (gray) are indicated. The sequences are aligned according to the sequence and secondary structure similarity. Ea22 sequences typically terminate with a +G or +GE motif (orange) of unknown significance. The secondary structure similarities suggest a tertiary structural level of organization consisting of an N-terminal domain, a central coiled-coil domain, and a C-terminal domain.

Structural and Functional Features of λ Ea22

When lower induction temperatures were used (20 °C), we obtained milligram quantities of soluble 6× His-tagged λ Ea22 (182 aa) and several C-terminal fragments, thereby providing a path toward a biochemical characterization of this unknown phage protein. The largest C-terminal fragment selected for the study, λ_Ea22(93–182), began just inside the coiled-coil region and was followed by four additional fragments that were designed to successively remove one predicted secondary structure at a time (Figure a). The two shortest C-terminal fragments, λ_Ea22(129–182) and λ_Ea22(119–182), expressed as inclusion bodies and could not be refolded from denaturant, leaving λ_Ea22(109–182) as the minimal soluble fragment with a predicted secondary structure of β1β2α1α2α3. Table S1 in the Supporting Information summarizes all expression conditions and solubilities of protein fragments presented in this study.

Figure 2

Structural characterization of λ Ea22 fragments. (a) Various protein fragments extending from the latter part of the predicted coiled-coil region (green) to the native C-terminus were expressed. λ_Ea22(119–182) and λ_Ea22(129–182) were insoluble and not pursued further. (b) SEC-MALS analysis of λ_Ea22(93–182). One peak with a retention time of 16.9 min coincides with a molecular mass of 22.0 kDa, suggesting that the native form of this protein fragment is a dimer (the calculated monomeric molecular mass from the sequence is 10.7 Da). (c) 1H–15N HSQC NMR spectrum of 15N-labeled full-length λ Ea22. (d) 1H–15N HSQC NMR spectrum of the smallest soluble Ea22 C-terminal fragment, λ_Ea22(109–182). (e) CD thermograms of λ_Ea22(93–182, blue), λ_Ea22(102–182, green), and λ_Ea22(109–182, red) at a wavelength that is characteristic of α-helices. (f) Survival of E. coli MG1655 containing various λ Ea22 expression plasmids after infection with λ. Error bars represent the SD of five replicates. Relative survival is expressed against a control culture in which the buffer was added in the place of phage particles. EV = empty vector control.

The far-UV circular dichroism (CD) spectrum of full-length λ Ea22(1–182) demonstrated broad troughs at 208/222 nm, suggesting that the protein was folded with hallmarks of an α-helical secondary structure. Differential scanning calorimetry showed that full-length λ Ea22(1–182) denatured with a single transition at 48 °C. A similar thermal denaturation midpoint was observed using circular dichroism spectroscopy. The reader is referred to Table for a summary of the biophysical and biochemical parameters described in this report. The oligomeric state of λ Ea22(1–182) was determined by SEC-MALS, a method that combines analytical size-exclusion chromatography (SEC) with multi-angle laser scattering (MALS). While full-length λ Ea22(1–182) was observed to be tetrameric by SEC-MALS, the C-terminal fragment, λ_Ea22(119–182), was dimeric (Figure b). In the discussion, we present a model of the Ea22 protein that reconciles these two observed oligomeric states. The calorimetry, circular dichroism, and SEC-MALS data for full-length λ Ea22(1–182) are available in the Supporting Information as Figure S1.

Table 2

Oligomeric State and Thermodynamic Parameters from SEC-MALS, CD, and DSC Analyses of Ea22 Proteins and Protein Fragments

protein	region	oligomeric state	Tm (CD, °C)	Tm (DSC, °C)	ΔH (DSC, kJ/mol)
λ Ea22(1–182)	full-length	tetramer	48	48.4 ± 0.1	433 ± 3
λ Ea22(93–182)	C-term domain	dimer	>90	>100
λ Ea22(102–182)	C-term domain	dimer	>90	>100
λ Ea22(109–182)	C-term domain	dimer	82	90
φ24B Ea22(135–261)	C-term domain	dimer	>90	>100
φP27 Ea22(1–315)	full-length	tetramer	55	48.1 ± 0.1	1048 ± 24
				52.5 ± 0.1	743 ± 23

A sample of full-length 15N-labeled λ Ea22 was assessed by the nuclear magnetic resonance (NMR) method. At room temperature (298 K), the majority of 1H–15N resonances was extremely broadened. This observation is consistent with a 22.5 kDa Ea22 monomer that exhibits the solution characteristics of a 90 kDa tetramer. Line shapes improved when the temperature was raised to 310 K as a result of the protein being able to tumble faster in solution. In the 1H–15N HSQC spectrum shown in Figure c, each resonance, or peak, is roughly attributed to one backbone amide 1H–15N pair found in each amino acid except proline. Following this rule, approximately two hundred peaks were expected, yet only one-third of that predicted number was observed. This discrepancy can be attributed to the large apparent molecular weight of the protein and additional ms−μs time scale dynamics from unstructured regions and large domain movements. Considering these factors, we hypothesized that the observable peaks were localized to one domain that was separated enough from the rest of Ea22 for it takes on solution characteristics of a small protein. From the secondary structure predictions, the C-terminus was the most plausible region for such a domain.

Of the five C-terminal protein fragments that were cloned, we prepared 15N-labeled NMR samples of the three that were soluble, being λ_Ea22(93–182), λ_Ea22(102–182), and λ_Ea22(109–182). The 1H–15N HSQC spectrum of the λ_Ea22(109–182) protein fragment was comparable to the spectrum of the full-length λ Ea22 protein (Figure d), thereby confirming our hypothesis that the C-terminal region is tethered far enough from the coiled-coil region to appear as if it is unrestrained.

A CD thermal melt assay of the three soluble C-terminal λ Ea22 protein fragments was performed by monitoring the change in ellipticity at 222 nm. As shown in Figure e, no change was observed up to 90 °C for λ_Ea22(93–182) and λ_Ea22(102–182), suggesting that these dimeric protein fragments were highly thermostable. In contrast, a Tm of 82 °C was observed for λ_Ea22(109–182), suggesting that amino acids between 102 and 108 make a minor contribution to the structure of the C-terminal domain. Differential scanning calorimetry (DSC) at 3 atm pressure was performed to extend the CD thermal assay. From this investigation, we observed that the thermostability of λ_Ea22(93–182) and λ_Ea22(102–182) was greater than 100 °C. Thus, this first biochemical and biophysical survey of λ ea22 suggests that it is more than a nonessential open reading frame and, in fact, encodes a multidomain protein that experiences evolutionary pressure to maintain a stable fold.

Upon infection with λ phage, a greater percentage of bacteria survive when they express full-length λ Ea22 from a plasmid (Figure f). The observed percentage, in this case, was in excess of 100% due to the ability of lysogens to resist superinfection and continue dividing.[12] When this experiment was performed with the minimal thermostable C-terminal fragment, λ_Ea22(102–182), the proportion of bacteria surviving infection was similar to the empty plasmid control, suggesting that there was no effect.

Structural and Functional Features of φ24B Ea22

φ24B is a model Stx phage for functional studies due to its rapid development in bacteria and ability to maintain titer during storage.[15,16] Compared to the λ phage genome, φ24B not only contains more exo–xis open reading frames, but also an Ea22 protein that differs in length and composition, particularly in its C-terminal domain. These combined differences may contribute to the fitness of φ24B and its bacterial host within the human intestine.

We began our biochemical survey by expressing full-length φ24B Ea22; however, the protein was insoluble and could not be refolded either by slow or fast removal of the denaturant. While a direct comparison between full-length φ24B Ea22 and λ Ea22 could not be performed, we were successful at expressing a C-terminal fragment, φ24B_Ea22(135–261), extending from the coiled-coil region to the native C-terminus (Figure a). A 1H–15N HSQC NMR spectrum of isotopically labeled φ24B_Ea22(135–261) suggested the presence of a small, folded domain within the fragment (Figure b). While there was no obvious similarity in the sequence or secondary structure with λ Ea22, the φ24B Ea22 C-terminal fragment was also dimeric by SEC-MALS (Figure c) and demonstrated exceptional thermostability as no structural transitions were observed by either CD or DSC methods performed at temperatures exceeding 100 °C. The φ24B Ea22 C-terminal domain has the predicted secondary structure β1β2α1α2β2β3β4β5.

Figure 3

Structural and functional characterization of φ24B Ea22 fragments. (a) Protein fragment φ24B_Ea22(135–261) from the predicted coiled-coil region (green) to the native C-terminus expressed was soluble. A shorter fragment φs24B_Ea22(144–238) constrained by the predicted secondary structures was insoluble. (b) 1H–15N HSQC NMR spectrum of 15N-labeled φ24B_Ea22(135–261). (c) SEC-MALS assay φ24B_Ea22(135–261). Three peaks were observed. Peak #1 with a retention time of 16.2 min is the most prominent and coincides with a molecular mass of 37.2 kDa, suggesting that the native form of this protein fragment is a dimer (the calculated monomeric molecular mass from the sequence is 17.9 Da). Peak #2 with a retention time of 14.1 min coincides with a molecular mass of 73.1 kDa, suggesting a possibly oxidized tetrameric species occurring as a dimer of dimers. Peak #3 did not produce a light scattering signal and is likely a buffer component. (d) Survival of E. coli MG1655 containing various φ24B Ea22 expression plasmids after infection with φ24B. Error bars represent the SD of five replicates. Relative survival is expressed against a control culture in which the buffer was added in the place of phage particles. EV = empty vector control.

The same assay that tested the ability of λ Ea22 to affect the survival rate of infected E. coli was performed for full-length φ24B_Ea22(1–261) and the soluble C-terminal fragment φ24B_Ea22(135–261). Like the λ assay, survival was improved when φ24B_Ea22(1–261) was overexpressed (Figure d). However, unlike the λ assay, survival was also improved for the C-terminal fragment. It is possible that structural differences between the λ and φ24B C-terminal fragments may have contributed to a different functional outcome during infection.

Oligomeric State and Domain Organization of φP27 Ea22

The exo–xis region of the Stx+ phage φP27 has eleven additional genes that follow ea22, in contrast to four in φ24B and two in λ. The two genes in λ include the putative transcription factor ea85 and orf55, a gene with no known function.

The C-terminal domain of the φP27 Ea22 protein is 54 aa larger than φ24B and 133 aa larger than λ. When full-length φP27_Ea22(1–315) was overexpressed, it partitioned between the soluble and insoluble fractions of the bacterial lysate. Lower induction temperatures improved the yield of soluble protein. Owing to the extensive coiled-coil domain that P27 Ea22 shares with λ and φ24B, the far-UV CD spectrum was characteristically α-helical. A CD-based thermal denaturation assay revealed a broad transition centered at 55 °C. When the thermostability of φP27_Ea22(1–315) was reassessed by differential scanning calorimetry, two transitions were observed around 50 °C, suggesting that φP27 Ea22 is organized into two domains. Despite the sequence and predicted secondary structural differences between P27 and λ Ea22, an SEC-MALS analysis revealed that φP27_Ea22(1–315) is also a tetramer. Experimental data from the CD, calorimetry, and SEC-MALS investigations are shown in the Supporting Information as Figure S2.

To delimit the domain boundaries of φP27 Ea22, a sample was digested with trypsin for varying periods. Over several trials and digestion periods, we observed that the protein was initially cleaved into a 27 kDa fragment and a C-terminal 9 kDa fragment. A 6 kDa fragment was also observed in some trials, albeit to a lesser extent (Figure a). Using the LC-MS method, the 9 and 6 kDa fragments were mapped to 237–315 and 255–315, respectively, in the φP27 Ea22 protein (Figure b). Given that the 9 kDa fragment φP27_Ea22(237–315) is the most stable, the predicted secondary structure of the φP27 Ea22 C-terminal domain is α1β2α2β2α3.

Figure 4

Limited proteolysis of φP27 Ea22. (a) SDS-PAGE of purified full-length φP27 Ea22 protein before and after treatment with trypsin. (b) Mass spectrum and deconvoluted spectrum of a peak obtained by reverse-phase chromatography at a retention time of 10.62–11.03 min. The mass spectrum is consistent with the band labeled φP27-2. (c) Boundary of the best fit deconvoluted mass to the sequence of φP27-2 (observed: 9340.91 Da, expected: 9340.76 Da). In some longer protein digests, an additional C-terminal fragment was observed (observed: 6952.90 Da, expected: 6953.03 Da).

The N-terminal region of φP27 Ea22 is predicted to have the secondary structure α1β2β2 β3α2. A fragment encompassing this region, φP27_Ea22(1–62), was expressed along with two C-terminal fragments, φP27_Ea22(145–315) and φP27_Ea22(176–315). These C-terminal protein fragments were insoluble and could not be refolded from the denaturant. A third, minimal C-terminal protein φP27_Ea22(235–315) identified by mass spectrometry did not express.

Model of Ea22

A model of Ea22 is shown in Figure that reconciles our biochemical investigations of Ea22 from phage λ and Stx+ phages, φ24B and φP27. If the tetrameric coiled-coil region is placed centrally in an antiparallel orientation, the C-terminal sequences are then juxtaposed at each end and are free to dimerize. A long enough tether may permit the C-terminal domain to tumble freely, thereby explaining why 1H–15N HSQC NMR spectra of the C-terminal domain and full-length λ Ea22 were similar. Since the N-terminal region present in φ24B and φP27 could not be expressed, its propensity to dimerize like the C-terminal domain remains unknown.

Figure 5

Possible architecture of Ea22. The tetrameric coiled-coil region (green) is shown as an antiparallel four-helix bundle leaving the C-terminal domain (red) to dimerize on each end of the bundle. For clarity, only one C-terminal domain is shown. The structure and oligomeric state of the N-terminal domain is unknown.

Overall, the proposed architecture implies an adapter-like function for Ea22. While Ea22 was not among the host–phage or phage–phage protein partnerships that were identified in yeast two-hybrid (Y2H) assays,[24,25] Ea22 may still serve in a multiprotein complex that would be undetectable by a Y2H assay or alternatively bind nucleic acids. The proposed architecture permits the N-terminal and C-terminal regions to evolve independently, ultimately leading to different partnerships possibly within the same host regulatory pathway.

Discussion

The exo–xis region of λ phage is not required for viral development, yet at least four genes are conserved among λ and Shiga toxin-producing (Stx+) phage found in clinically relevant E. coli. Only recently has the exo–xis region been studied at a molecular level of detail. In the case of ea22, the largest of the conserved exo–xis region genes, it was not even known if it encoded a protein until this study. The exo–xis region also presents a rare new opportunity to examine the vast genomic landscape of phages[26,27] and to examine the protein sequence landscape that determines a given three-dimensional structure. For example, the exo–xis protein Ea8.5,[17] the capsid–tail interface structural protein gpU,[28] and the major tail protein gpV[29] bear little sequence similarity to any other proteins except from a closely related phage, yet the high-resolution structures reveal variations of a common fold. The exo–xis region also presents an opportunity to explore new phage–host relationships and the protein–protein and protein–nucleic acid interactions that underlie them.

A previous study showed that ea22 deletion mutants created a bias toward a sustained lytic developmental outcome by reducing the efficiency of lysogenization and increasing the number of phage progeny.[14] We extended these results with an assay in which bacteria expressing Ea22 were infected with either λ or p24B phage, and then, the number of bacteria surviving were scored relative to an empty expression vector. In the case of full-length Ea22, the relative survival was over 100% because bacteria persisted as lysogens long enough to make it through additional rounds of cell division. While the C-terminal domain of λ Ea22 had no effect on survival, the C-terminal domain from the Stx+ phage φ24B did provide similar prosurvival benefits reinforcing our biochemical and biophysical data. suggesting that the C-terminal domains are not only structurally but also functionally different. Following this idea, we speculate that the λ or Stx+ C-terminal domain from Ea22 may interact with a set of host and phage proteins supporting lysogenization in different ways.

Early phage genes are expressed approximately 15–30 min post infection and include the well-characterized N, cI, and cro gene products along with many exo–xis region members. In Shiga toxin-producing phages such as φ24B and φP27, ea22 expression reached levels up to an order of magnitude higher than other early genes.[30] In λ phage, high levels of ea22 expression were also observed but not to the extent of its Shiga toxin-producing counterparts.[14] Thus, the amount of Ea22 produced during the earliest stage of infection could also have a direct outcome on host survival and suppression of lytic development.

The specific structural features that have been described for λ and Stx+ phage Ea22 proteins may also be considered in terms of the EHEC host. Cattle, for example, do not have a suitable receptor for Shiga toxin and, therefore, serve as a reservoir for EHEC. Within the intestinal tract of a cow, Ea22 may attenuate the response to stress and favor lysogenization in order for the bacterial population to establish itself. Likewise, in the human intestinal tract where bacterial survival is more challenging, Ea22 may slow down the lytic development cycle, so bacteria are not simultaneously attacked from outside and within. Bacterial lysogens overexpressing λ ea22 from a plasmid exhibit a delay in phage development when induced with hydrogen peroxide, suggesting that oxidative stress pathways are involved.[30] When this experiment was repeated in Stx+ phage, the effect was more pronounced, suggesting that Ea22 may be different in Stx+ phage or Ea22-mediated regulation of the response to oxidative stress is more complex.

Recently, a 20 bp small RNA (sRNA) was identified in Stx phage φ24B with many of the same functional features as Ea22.[31] Phage mutants lacking the region for the 80 bp precursor RNA did not lysogenize effectively, responded faster to SOS oxidative stress responses, and were more efficient at producing new progeny. One of the two possible targets of RNA was an antirepressor, d_ant, which is found in other Stx phages and but notably absent in the λ genome. Since the molecular partners of Ea22 are unknown, it also remains to be determined if Ea22 interacts with this antirepressor, or genes, or gene products affected by the antirepressor.

Conclusions

We have demonstrated that the exo–xis region protein Ea22 is a tetrameric functional protein that affects host survival. Future proteomics and high-resolution structural studies are necessary to precisely establish the role of Ea22 and its suitability as a lead for combating clinically important E. coli infections.