Shr of Group A Streptococcus is a new type of composite NEAT protein involved in sequestering heme from methemoglobin

SUMMARY

A growing body of evidence suggests that surface or secreted proteins with NEAr Transporter (NEAT) domains play a central role in heme acquisition and trafficking across the cell envelope of Gram-positive bacteria. Group A Streptococcus (GAS), a β-hemolytic human pathogen, expresses a NEAT protein, Shr, which binds several hemoproteins and extracellular matrix (ECM) components. Shr is a complex, membrane-anchored protein, with a unique N-terminal domain (NTD) and two NEAT domains separated by a central leucine-rich repeat region. In this study we have carried out an analysis of the functional domains in Shr. We show that Shr obtains heme in solution and furthermore reduces the heme iron; this is the first report of heme reduction by a NEAT protein. More specifically, we demonstrate that both of the constituent NEAT domains of Shr are responsible for binding heme, although they are missing a critical tyrosine residue found in the ligand-binding pocket of other heme-binding NEAT domains. Further investigations show that a previously undescribed region within the Shr NTD interacts with methemoglobin. Shr NEAT domains, however, do not contribute significantly to the binding of methemoglobin but mediate binding to the ECM components fibronectin and laminin. A protein fragment containing the NTD plus the first NEAT domain was found to be sufficient to sequester heme directly from methemoglobin. Correlating these in vitro findings to in vivo biological function, mutants analysis establishes the role of Shr in GAS growth with methemoglobin as a sole source of iron, and indicates that at least one NEAT domain is necessary for the utilization of methemoglobin. We suggest that Shr is the prototype of a new group of NEAT composite proteins involved in heme uptake found in pyogenic streptococci and Clostridium novyi.

INTRODUCTION

Acquisition of iron from host sources is of vital importance to many pathogenic bacteria during the course of infection. Mammalian hosts limit the availability of free extracellular iron to levels around 10^-18 M (Bullen, 1981) by producing iron-chelating proteins such as lactoferrin and transferrin or by storing it in ferritin within the cells. However, nearly 75% of the iron in the human body is found in the form of heme, where it is incorporated into the protoporphyrin ring and serves as a prosthetic group of hemoglobin, myoglobin, and some enzymes (Stojiljkovic & Perkins-Balding, 2002, Tong & Guo, 2009). The process of obtaining heme from the host by Gram-positive pathogens often involves binding of heme or hemoproteins by bacterial receptor proteins which then deliver the heme to a membrane-bound ABC transporter for translocation to the cytoplasm (Tong & Guo, 2009, Wandersman & Delepelaire, 2004). The first Gram-positive heme transporter to be described was the hmu (hemin/hemoglobin utilization) system of Corynebacterium diptheriae (Drazek et al., 2000). The hmuTUV genes share homology with Gram-negative heme-ABC transporters such as found in Yersinia pestis (Hornung et al., 1996, Thompson et al., 1999). HmuT, which is localized to the cell membrane, binds hemin or hemoglobin directly; HmuU and V are the membrane permease and the ATPase, respectively. Two membrane anchored heme-receptors, HtaA and HtaB, encoded by the hmu chromosomal locus were recently described and hypothesized to work in conjunction with the HmuTUV transporter in heme acquisition and transport across the cell envelope (Allen & Schmitt, 2009). Inside the Corynebacterium cytoplasm, a heme-oxygenase enzyme, HmuO, degrades the heme and releases the iron for use by the pathogen (Kunkle & Schmitt, 2007, Wilks & Schmitt, 1998).

The heme uptake system studied in the most detail in Gram-positive bacteria is that from Staphylococcus aureus and is termed the Isd (iron-regulated surface determinants) system. In the Isd system, the IsdA, B and H (HarA) proteins are covalently attached by the SrtA sortase enzyme to the cell wall, where they interact with a variety of ligands including heme, hemoglobin, hemoglobin-haptoglobin complex, fibrinogen, and fibronectin (Clarke et al., 2004, Dryla et al., 2003, Mazmanian et al., 2003, Torres et al., 2006). These proteins obtain heme and deliver it across the cell wall and the cell membrane in a cascade fashion via IsdC and the IsdEF ABC-transporter (Skaar et al., 2004, Skaar & Schneewind, 2004, Wu et al., 2005). Unlike the surface exposed Isd receptors, IsdC is embedded deep in the cell wall by a dedicated sortase, SrtB (Marraffini & Schneewind, 2005, Mazmanian et al., 2002, Mazmanian et al., 2003). Ligand binding by IsdA, B, C, and H (HarA) is mediated by NEAT (Near transporter) domains, which are found in one or more copies in each of the receptor proteins.

The NEAT domain is a protein motif first identified in 2002 that is encoded in variable copy number near ABC iron transporter genes in the chromosomes of several Gram-positive bacterial species (Andrade et al., 2002). It is approximately 125 amino acids long with low primary sequence homology and a predicted secondary structure of mostly β strands (Andrade et al., 2002). Expressed in recombinant form, the isolated NEAT domains of IsdA, IsdC, IsdH (HarA) and the second NEAT domain of IsdB (IsdBN2) have been studied and manifest different ligand preferences. The IsdA NEAT domain has fibrinogen binding ability (Clarke et al., 2004) and binds heme (Grigg et al., 2007). The IsdC NEAT domain binds heme but not proteins such as hemoglobin, fibrinogen or fibronectin (Pluym et al., 2008, Sharp et al., 2007, Villareal et al., 2008). The third NEAT domain of IsdH (IsdHN3) is also exclusively a heme-binding domain (Watanabe et al., 2008, Pilpa et al., 2009) and IsdBN2 binds heme as well (Tiedemann et al., 2008). In contrast, the two N-terminal NEAT domains of IsdH (HarA) have demonstrated binding to hemoglobin and hemoglobin/haptoglobin, respectively (Dryla et al., 2003, Dryla et al., 2007, Pilpa et al., 2006). Secreted or cell wall anchored NEAT proteins which are central to heme acquisition pathways were identified in other Gram-positive pathogens including Listeria monocytogenes (Jin et al., 2006), Bacillus anthracis, (Gat et al., 2008, Maresso et al., 2006), and Bacillus cereus (Daou et al., 2009).

Heme uptake has also been studied in Group A Streptococcus, a β-hemolytic pathogen that uses host heme-containing proteins as an iron source (Eichenbaum et al., 1996). One system involves a 10 gene iron-regulated operon that has been termed Sia (Streptococcal iron acquisition) (Bates et al., 2003). Genes three through five in the cluster (siaABC or htsABC (Lei et al., 2003)) encode an ABC transporter that shares significant homology with heme or siderophore transporters found in other bacterial species (Bates et al., 2003). The second gene in the operon, shp, encodes a surface protein with β-sandwich fold similar to that of NEAT domains and is considered a distal member of the NEAT family (Aranda et al., 2007). Shp protein has been reported to bind heme at the cell surface and transfer it to SiaA (HtsA), the lipoprotein component of the ABC transporter (Lei et al., 2003). The heme coordination of SiaA has recently been elucidated by further biophysical studies and is described as six-coordinate and low-spin, employing methionine and histidine as axial ligands (Sook et al., 2008). The first gene in the sia operon encodes the Shr (streptococcal hemoprotein receptor) protein. Shr is a large (145kDa) hydrophilic protein that does not share significant overall homology with known heme or hemoprotein receptors (Bates et al., 2003). The first studies of Shr revealed that it plays a role in iron acquisition. It was observed to bind hemoglobin and hemoglobin-haptoglobin complex (Bates et al., 2003). The transfer of heme from Shr to the protein Shp has also been described (Zhu et al., 2008). Shr was recently demonstrated to bind the extracellular matrix proteins fibronectin and laminin, suggesting that it also acts as an adhesin (Fisher et al., 2008). A null shr mutant is attenuated for virulence in a zebrafish model for necrotizing fasciitis, underscoring the importance of Shr to the infection process in vivo (Fisher et al., 2008).

Shr has two NEAT domains, but its overall domain architecture, which includes a unique N-terminal domain and a series of leucine-rich repeats, is different from any of the characterized heme and hemoprotein-binding receptors, including the NEAT-containing Isd proteins of S. aureus. Shr also lacks a cell wall anchoring motif typical of the Isd receptor proteins of S. aureus; at the C terminus, Shr has a hydrophobic segment with a positively charged tail that threads the protein through to the cytoplasmic membrane. It was recently demonstrated that Shr spans the cell wall and is exposed to the extracellular environment (Fisher et al., 2008). Shr is proposed to participate in the acquisition of heme by GAS and its delivery to Shp and/or the SiaABC transporter. The uptake of heme from hemoproteins by Shr or its direct role in iron acquisition has not been shown, however. In this study we establish the function of Shr in hemoglobin use and heme uptake by GAS. We analyze the functional domains of this receptor protein and present evidence that the mechanism of heme uptake by Shr is different from that of the characterized Isd proteins. We suggest that Shr is a prototype of a new group of NEAT proteins involved in heme uptake.

To distinguish between the two NEAT domains in Shr, the closest NEAT domain to the amino group will be referred to as NEAT1 and the second NEAT domain from the amino group will be referred to as NEAT2.

RESULTS

Shr is a composite NEAT protein found in pyogenic streptococci and C. novyi

NEAT domains are key ligand-binding domains used by receptor proteins involved in heme acquisition and translocation in Gram-positive bacteria. We recently reported that the GAS Shr protein has two NEAT domains that are separated by an LRR region (Fisher et al., 2008). Additional sequence examination also identified an EF-hand motif between the first NEAT domain of Shr and the LRR segment (residues 532 and 544) and two copies of a short domain with unknown function, DUF1533 (residues 61-123 and 203-269) in the N- terminal region of Shr (Fig. 1). In silico analysis using the web based SMART tool (Letunic et al., 2009, Schultz et al., 1998) reveals that there are about 160 NEAT domains in 80 proteins encoded by Gram-positive bacteria from the Firmicutes phylum. Most of these NEAT-proteins are Isd-like molecules, which contain one or more copies of NEAT domains, a leader peptide, and in some cases a sortase recognition signal or other type of cell-wall binding region. A few proteins consist of an LRR segment in addition to export signals and NEAT domain(s); these include the heme uptake protein of B. cereus, IlsA (Daou et al., 2009), and the following hypothetical proteins A0PYT7 (C. novyi), O6HLL6 & O6HNR0 (B. thuringensis), and 073BH4 (B. cereus). Shr appears to be the first characterized protein with DUF1533 domains. An examination of the database demonstrated that DUF1533 is found in duplication in putative proteins with unknown function from the Clostridia class and in two species of Paenibacillus. Therefore, the domain architecture of Shr is different and more complex than most of the previously described NEAT receptors or the hypothetical NEAT-proteins found in bacterial genomes. The complex domain arrangement of Shr is intriguing and suggests that it evolved by joining several domains found separately in bacterial proteins of Firmicutes. Shr orthologues, which share identical or nearly identical domain architecture, are found in C. novyi as well as in the pyogenic streptococci S. equi zooepidemicus and S. dysgalactiae (Fig. 1). A shr orthologue is found in the genome of S. equi subsp. equi as well; however, a frame shift mutation results in a truncated protein (Holden et al., 2009). All streptococcal Shr orthologues are closely related in their primary amino acid sequence (58-86% identity) while the C. novyi orthologue shares fewer identical residues (~30%). Intriguingly, the shr gene in all the streptococci is part of a 10-gene cluster which is homologous to the sia operon of GAS. The shr gene in C. novyi on the other hand, is found in a genomic locus that encodes only a putative LRR-NEAT protein (A0PYT7). Together these observations suggest that Shr may represent a new type of composite NEAT protein family.

Figure 1. Shr proteins in pyogenic Streptococci and Clostridium novyi.

LP: Leader Peptide; DUF: Domain of Unknown Function 1533; NEAT: NEAr Transporter domain; EF: EF-hand motif; TM: TransMembrane domain.

Shr obtains heme from solution and reduces ferric heme to ferrous heme

Shr binding to hemoproteins in vitro and its genomic location in the sia operon together with the heme binding protein, Shp, and the SiaABC heme-transporter, suggest that Shr is involved in heme acquisition and transport by GAS (Bates et al., 2003). This hypothesis was recently supported by the observation that purified Shr transfers heme to Shp (Zhu et al., 2008). The sequestering of heme from host proteins by Shr was not previously demonstrated, however, and the mechanism of heme and hemoprotein binding has not been investigated. To characterize heme uptake by Shr, a histidine-tagged recombinant protein (rShr) (Bates et al., 2003) was prepared and analyzed (Fig. 2A). Shr is readily reduced following treatment with dithionite or oxidized by ferricyanide, producing the corresponding absorption spectra with maxima around 410 nm for the bound ferric heme or 430, 540 and 560 nm for the ferrous heme (Zhu et al., 2008). In this study we found that the spectral properties of rShr after ferricyanide treatment were almost identical to those without treatment, with Soret bands at 412 and 414 nm (Fig. 2B green and blue lines, respectively). On the other hand, the addition of D, L-dithiothreitol (DTT) resulted in a shift in the absorption peak to 427 nm (Fig 2B red line) and the production of more resolved peaks at 536 and 566 nm, spectral characteristics of ferrous heme-protein complexes (Makinen, 1983). Therefore, our data indicate that rShr was purified from E. coli as mostly ferric heme complex. In the course of Shr purification, we observed that heme-bound rShr was considerably more stable than the heme-free protein, and that the addition of hemin to the E. coli culture prior to harvesting the cells resulted in a higher production of intact rShr (data not shown).

Figure 2. Hemin binding and reduction by rShr.

(A) SDS-PAGE showing purified rShr. (B) 9 μM Shr (blue line) was treated with 10 mM of DTT (red line) or 30 μM ferricyanide (FCN) (green line). Histag elution buffer containing 10 mM DTT was used as a blank for DTT-treated Shr spectrum and excess ferricyanide was removed from the protein sample by dialysis in phosphate buffer. (C) An increase of heme bound to Shr (3 μM) as increasing concentrations (1 μM, red; 3 μM, green; 5 μM, purple; 10 μM, blue; or 20 μM, orange) of hemin were added to the protein is shown by the sharp peak at 414 nm. Hemin reduction is indicated by the growing absorbance at 427 nm and at ~540 and ~564 nm. The corresponding hemin chloride concentrations in Histag elution buffer served as blanks for the UV-visible scans (see figure S2). The insert magnifies the 500 nm – 700 nm region. (D) UV-visible spectra of rShr following the addition of 20 μM hemin (red line) and treatment with ferricyanide (blue line). Hemin reduction shown by the presence of a Soret peak at 427 nm and by the peaks at ~540 and ~564 nm (red spectrum) is reversed by the addition of ferricyanide (blue spectrum).

Heme binding by rShr was investigated further by monitoring the changes in the UV-visible spectrum following the addition of increasing amounts of hemin (heme with ferric iron) to the protein solution. The addition of free hemin resulted in a concomitant increase in rShr-bound hemin, as was indicated by the growing absorbance at 414 nm (Fig. 2C). Surprisingly, the UV-visible spectrum of rShr following the hemin addition also revealed growing absorption peaks at 427, and at 540 and 564 nm as well (Fig. 2C insert). Removing the free heme by dialysis did not lead to changes in the spectrum (data not shown). The growing peaks at 427, 540 and 564 nm indicate a simultaneous increase in rShr-bound ferrous heme. The addition of hemin to rShr was done under aerobic conditions (using an open tube and vigorous mixing) and in the absence of reducing agents. Therefore, the observed rise in rShr-bound heme suggests that Shr reduces the added hemin. The addition of ferricyanide to the protein solution following titration with 20 μM hemin resulted in a shift of the absorption peak from 427 back to 410 nm. The 410/280 absorbance ratio that is indicative of the ferric heme load in Shr (Zhu et al., 2008) was changed from 0.59 to 0.75. Oxidation with ferricyanide also eliminated the peaks at 540 and 564 nm (Fig. 2D). Therefore, ferricyanide was able to oxidize the ferrous iron of the Shr bound heme, confirming that the changes in Shr spectrum seen following the addition of free hemin were due to a reversible reduction of the protein-bound hemin. It was previously reported that Shr could be purified from E. coli as a mixture of ferric and ferrous iron heme (Zhu et al., 2008). Our observations suggest that Shr has an inherent ability to reduce the ferric heme and to provide a stable environment for the produced ferrous complex. The autoreduction activity is a very intriguing characteristic of Shr. To the best of our knowledge, it is the first report of heme reduction by a bacterial heme receptor.

NEAT1 and NEAT2 are both heme-binding domains in Shr

Heme binding by the Isd proteins is imputable to their NEAT domains (Grigg et al., 2010). However some NEAT domains have been reported to not bind heme. For example, IsdH NEAT3 domain binds heme (Pilpa et al., 2009, Tiedemann et al., 2008, Watanabe et al., 2008), whereas IsdH NEAT1 and IsdH NEAT2 domains interact only with hemoglobin and haptoglobin (Pilpa et al., 2009). Isd NEAT domains hold the heme within a hydrophobic pocket through several conserved residues including two invariant tyrosines, one of which coordinates the iron (Tyr 166 in IsdA) and a second residue (Tyr 170 in IsdA) that interacts with both the heme pyrrole ring and the coordinating tyrosine (Grigg et al., 2007). Sequence analysis revealed that not all of the conserved residues in the Isd heme binding sites are found in Shr NEAT domains (Fig. S1); most significantly, both of the NEAT domains in Shr are missing the iron-coordinating residue Tyr 166, and only NEAT1 has the Tyr 170. Interestingly, the second heme binding protein coded by the sia operon, Shp, does not use tyrosine residues to coordinate the heme iron, and instead utilizes two methionines (Aranda et al., 2007).

To investigate heme acquisition by Shr, several recombinant proteins containing one or more of the component domains of Shr were constructed with an N-terminal fusion to the Strep-Tag epitope. These Shr variants include recombinant proteins with the amino terminal domain of Shr (NTD) or the amino terminal domain through NEAT1 (NTD-N1), the NEAT1 domain, and the NEAT2 region (Fig. 3A). The recombinant proteins were overexpressed, purified by FPLC using a Strep-Tactin column, and analyzed. The protein containing only the NEAT1 region turned out to be highly insoluble and was therefore excluded from further investigations. SDS-PAGE analysis confirmed the production and purification of the recombinant Shr protein fragments, revealing protein bands at the expected size for each construct: 61 kD (NTD-N1), 42 kD (NTD), and 23 kD (NEAT2, Fig 3B).

Figure 3. Heme binding by Shr fragments.

(A) Schematic representation of NTD, NTD-N1 and NEAT2. LP: Leader Peptide; ST: Strep-Tag; DUF: Domain of Unknown Function 1533; NEAT: NEAr Transporter domain; (B) SDS-PAGE showing purified recombinant Shr fragments (1) Molecular weight marker, (2) NEAT2, (3) NTD, (4) NTD-N1. The UV-visible spectra of NEAT1 (C), NTD (D), NTD with additions of hemin chloride (1 μM, red; 5 μM, green; or 10 μM, purple) (E) and NEAT2 (F). (G) UV-visible spectra of NEAT2 following titration with 20 μM hemin (blue). The red and green lines, respectively, represent the spectrum 5 min and 24 h, after addition of 6 μM ferricyanide. The insert magnifies the 400 nm – 460 nm region, showing the shifts in the Soret peaks. The Strep-tag wash buffer alone was treated exactly the same way as the protein solution, and used as blank for UV-visible scan.

The Shr protein variants that contained the first NEAT domain (NTD-N1) or the second NEAT domain (NEAT2) both had red color when purified and a UV-visible spectrum consistent with bound heme. The purified NTD-N1 protein showed a significant absorbance at 410 nm indicating that it was co purified with ferric heme (Fig. 3C). In contrast, the optical spectrum of the purified NTD fragment showed no band at the Soret region indicating that it did not contain heme (Fig. 3D). To test the heme binding ability by Shr's N-terminal region, the recombinant NTD protein was incubated with increasing concentrations of hemin. The optical spectrum of the protein after incubation showed no absorption at the Soret region. Instead, similar to the spectrum of free heme (Fig. S2), a broad peak at 390 nm was formed as increasing amounts of hemin were added, indicating an accumulation of unbound heme in the NTD protein solution (Fig. 3E). Therefore, the NEAT1 domain contained within the NTD-N1 protein is responsible for the ferric heme binding observed by this protein fragment.

Unlike the ferric-heme load of the NTD-N1 protein, the optical spectrum of the purified NEAT2 protein suggests that it is purified with a mixture of ferric and ferrous heme. NEAT2 spectra consistently had a significant Soret band at 428 nm along with sharp peaks at 535 nm and 564 nm. However, variations in the intensity of absorbance at 410 and 428 nm (indicating different ratio of the protein bound ferrous and the ferric forms) were observed in the spectrum of NEAT2 from different protein preps (Fig. 3F and the blue line in Fig. S3). To further investigate heme binding by NEAT2, the protein was titrated with increasing concentrations of free hemin and the UV-visible spectrum was monitored (Fig. S3). As observed with the full length Shr, the addition of free hemin resulted in concomitant increase of absorption at around 410 nm as well as 428, 535 and 564 nm. NEAT2 was then treated with ferricyanide and the UV-visible spectrum was taken after 5 and 30 min intervals and after 24 h. Within 5 min after addition of ferricyanide (Fig. 3G red line), the absorbance at ~410 nm increased compared to NEAT2 without ferricyanide (Fig. 3G blue line), and the bands at ~ 428, 535, 564 nm were almost gone, indicating that the NEAT2 protein was mostly in the oxidized form. The spectra did not change significantly at 30 min (data not shown). After 24 h (Fig. 3G green line), a red shoulder on the Soret band was seen, and the absorbance at 428, 535 and 564 nm had increased. Together these observations indicated that the NEAT2 protein autoreduced slowly.

Methemoglobin binding is mediated by the N-terminal domain of Shr, which specifically recognizes the holo form

Following erythrocyte lysis, the α₂β₂ heterodimeric hemoglobin converts to methemoglobin, in which the heme is found in the ferric form. This is largely an αβ heterodimer (Ascenzi et al., 2005, Umbreit, 2007). Methemoglobin is likely to be a physiologically relevant heme source for the hemolytic GAS. To investigate Shr interactions with methemoglobin, we developed and used an ELISA. rShr and the recombinant Shr fragments NTD-N1, NEAT2, and NTD, were used to coat the wells of microtiter plates and allowed to interact with increasing concentrations of methemoglobin (Fig. 4A). Ligand binding by the immobilized proteins was detected using hemoglobin antiserum. Wells coated with BSA and uncoated wells were used as controls for non-specific interactions. rShr bound methemoglobin in a dose dependent and saturable manner, while only low background binding (OD 405 ≤ 0.1) of the hemoglobin antiserum to the control wells was observed. The recombinant Shr fragments NTD and NTD-N1 bound methemoglobin with binding profiles that were similar to the full length Shr, and methemoglobin binding appeared saturated at a concentration of 10 nM (Fig. 4A). The observation that the full length Shr, NTD-N1 and NTD alone equivalently bind methemoglobin indicates a preponderant role for the NTD in hemoglobin binding. In contrast, no methemoglobin binding by the NEAT2 protein was detected. Therefore, unlike IsdA, IsdB, and IsdH proteins, which use NEAT domains to bind hemoglobin, an uncharacterized protein pattern found in the N-terminal domain of Shr interacts with hemoglobin.

Figure 4. (A) Methemoglobin binding by Shr fragments.

ELISA showed methemoglobin binding by rShr (crosses), NTD (black squares), and NTD-N1 (triangles). In contrast, NEAT2 (dots) or BSA (diamonds) did not bind methemoglobin. ELISA testing apohemoglobin binding by NTD (Red squares) showed no binding. The plates were coated with rShr or the Shr fragments and subsequently reacted with increasing concentrations of methemoglobin or apohemoglobin. Protein binding was detected with anti-hemoglobin antibodies as described in Materials and Methods. (B) Direct detection of immobilized holohemoglobin (triangles) and apohemoglobin (squares) with anti-hemoglobin antibodies. Uncoated wells (diamonds) were used as a negative control. Each datum point in sections A and B represents the mean ± SD (represented by the error bars) from data from at least two independent experiments done in triplicates.

We hypothesize that Shr interacts with methemoglobin to acquire heme from the host. We therefore asked whether Shr could differentiate between the apo and the holo forms of hemoglobin. Heme was removed from methemoglobin according to Asakura et al, (Asakura et al., 1964) and the formation of the apoprotein was confirmed by the UV-visible spectrum, which revealed no absorption at the Soret region (data not shown). The binding of NTD to immobilized apohemoglobin was tested by ELISA. In contrast to its interactions with holoprotein, no binding of NTD to apohemoglobin was observed (Fig. 4A, red line). Similar to the NTD, the full length Shr did not bind apohemoglobin (Fig. S4). A control ELISA performed with immobilized holo and apohemoglobin demonstrated that the hemoglobin-specific antibody was able to detect both forms of hemoglobin similarly over the range of hemoglobin concentrations studied (Fig. 4B). Therefore, the absence of binding of apohemoglobin in the experimental ELISA shown in Fig. 4A (red line) was not due to a lack of ability of the hemoglobin antiserum to recognize the apoprotein. In conclusion, these experiments demonstrate that Shr differentiates between the holo and the apo forms of hemoglobin and binds only to heme-loaded protein.

The NEAT 2 domain in Shr mediates most of its binding to ECM

We have recently observed that in addition to its interactions with hemoproteins, Shr also functions as an adhesin and binds fibronectin and laminin (Fisher et al., 2008). To determine the domains involved in the ability of Shr to bind these proteins components of the extracellular matrix (ECM), immobilized rShr, NTD, NTD-N1, and NEAT2 were allowed to react with the ECM components using ELISAs. Ligand binding was detected with antibodies specific for fibronectin or laminin. When fibronectin was added in increasing concentrations to the immobilized proteins, rShr as well as the recombinant fragments NTD-N1 and NEAT2 bound it in a concentration-dependent and saturable manner (Fig. 5A). The NEAT2 protein demonstrated the highest binding to fibronectin, while only low level binding was seen with NTD-N1. No interactions with fibronectin were demonstrated by the immobilized NTD protein. These observations suggest that NEAT regions mediate the observed Shr binding to fibronectin. Similar observations were made with laminin; as shown on Fig. 5B, rShr, and NEAT2 proteins bound laminin, while no significant binding to laminin was observed by the NTD or NTD-N1. Together, these observations indicate that while both Shr NEAT domains are able to interact with some ECM components, the NEAT2 domain plays a more significant role in this activity of Shr.

Figure 5. Binding of extracellular matrix proteins by Shr.

Elisa assay showing fibronectin (A) and laminin (B) binding by rShr (crosses) and NEAT2 (dots). NTD-N1 (triangles) slightly bound fibronectin but did not bind laminin. In contrast, NTD (squares) or BSA (diamonds) did not bind fibronectin or laminin. The plates were coated with rShr or Shr fragments and subsequently reacted with increasing concentrations of fibronectin or laminin as described in Materials and Methods. Each datum point in panels A and B stands for the mean ± SD (shown by the error bars) of three independent experiments done in triplicates.

The NTD-NEAT 1 region of Shr is sufficient for heme acquisition from methemoglobin

We next asked if the NTD-N1 fragment of Shr, which contains the hemoglobin-binding region and the heme-binding NEAT1, is sufficient for heme acquisition from methemoglobin. Heme was removed from the purified NTD-N1 (Asakura et al., 1964), and the formation of apo NTD-N1 was confirmed by UV-visible spectrum analysis (Fig. 6A). The heme transfer assay was done over a Strep-Tactin column with immobilized apoNTD-N1 protein. Methemoglobin in equimolar amounts to the immobilized apoNTD-N1 protein was flowed through the column. The bound hemoglobin was removed by extensive washes with salt containing buffer (see Materials and Methods), and the NTD-N1 protein was then eluted with desthiobiotin. Western blot analysis of the fractions collected during this procedure revealed that the hemoglobin containing fractions also included low amounts of NTD-N1 in addition to methemoglobin (lane 3, Fig. 6B), suggesting that some methemoglobin/NTD-N1 complexes were washed from the column. The NTD-N1 fraction that was eluted with desthiobiotin, however, did not contain a detectable amount of hemoglobin (lane 4, Fig. 6B). The optical absorbance of NTD-N1 after the methemoglobin passage showed a sharp peak at 411 nm, indicating that the apoNTD-N1 acquired heme from methemoglobin (red line, Fig. 6A). To confirm that the observed absorbance at 411 nm resulted from NTD-N1/heme complex and not from trace amounts of methemoglobin, we analyzed the absorbance of 0.7 μM methemoglobin solution, a concentration which is above the detection level of the hemoglobin antibody utilized in the assay (lane 6, Fig. 6B). This analysis revealed that methemoglobin, at the tested concentration, had a significantly lower Soret band in than the NTD-N1 fraction (data not shown).

Figure 6. Heme transfer from methemoglobin to apoNTD-N1.

(A) UV-visible spectra of 10 μM apoNTD-N1 after contact with methemoglobin (red) or hemin chloride (blue) and 10 μM apoNTD-N1 (green). (B) Western blot analysis of the fraction containing the Hb washes and NTD-N1 elution. Proteins (50 ng/well) were detected with anti-Hb (upper panel) or anti-Shr (lower panel) antibodies. (1) MW Marker; (2) purified apoNTD-N1; (3) Hb fraction; (4) NTD-N1 after Hb flow; (5) Empty lane; (6) Hb 50 ng ; (7) Hb 100 ng.

A similar experiment was done with immobilized NTD-N1 with hemin chloride solution (in 4-fold molar excess) instead of methemoglobin in the mobile phase. The optical analysis of NTD-NEAT1 after passage of free hemin also revealed that it acquired heme as indicated by the peak at 412 nm (blue line Fig. 6A). However the peak at the Soret region of NTD-N1 after the passage of methemoglobin was significantly higher than that after contact with the hemin chloride solution. This observation suggests that the NTD-N1 protein acquires more heme from hemoglobin than from the hemin solution, supporting heme transfer from methemoglobin directly to the immobilized NTD-N1. Heme acquisition by apoNTD-N1 was also investigated using an alternative assay in which the NTD-N1 was allowed to interact with methemoglobin in solution at room temperature. The UV-visible spectrum of the NTD-N1 protein after its separation from methemoglobin (using Strep-Tactin column) revealed a sharp peak at 410 nm (Fig. S5), demonstrating heme transport from methemoglobin to the NTD-N1 protein. Interestingly, the absorption intensity at the Soret region following 5 min of co-incubation was about 80% of that seen following 75 min, indicating rapid heme sequestering by the NTD-N1 protein.

Shr is required for GAS growth using hemoglobin as the sole iron source

To determine the importance of the heme binding domains to the function of Shr in vivo, several GAS mutants containing in frame deletions of various regions in shr were constructed. These include a mutant with NEAT1 deletion mutant (ΔNEAT1), a mutant with a deletion that spans the distal part of the LRR and most of the second NEAT domain (ΔNEAT2), and a mutant with a large deletion that includes both NEAT domains and the region in between (ΔNEAT1-2, Fig. 7A). The production of shr alleles in the expected size in the genome of each of the mutants and of the corresponding Shr proteins was confirmed by PCR and Western blot analyses (Fig. 7B and 7C). This analysis also confirmed the production of the wild type Shr when the ΔNEAT1-2 was complemented. Successful complementation of the null shr mutant was previously shown (Fisher et al., 2008). RT-PCR analysis with primers specific for siaA, which is located downstream of shr, verified that the shr mutations were not polar and did not affect the expression of the downstream genes in the sia operon (Fig. 7B).

Figure 7. Growth analysis of shr deletion mutants.

(A) Schematic representation of the in frame deletions created in Shr. WT: wild type; ΔNEAT1: a deletion of the NEAT1 domain; ΔNEAT2: a deletion of the NEAT2 domain; ΔNEAT1-2: a deletion of both NEAT1 and NEAT2 domains. Panels B-E describe the characterization of the constructed GAS mutants (B) The first panel shows PCR analysis of the chromosomal shr gene. Total RNA from each strain is shown in the second panel. The third and fourth panels respectively show RT-PCR analysis of the expression of shr (ZE106/126 primers) and siaA (204A-Fwd/Rev primers) genes. (C) Western blot showing the expression of the corresponding Shr protein variants. WT: wild type GAS (strain NZ131); ΔN1: NEAT1 mutant (strain ZE4925); ΔN2: NEAT2 mutant (strain ZE4926); ΔN1N2: NEAT1-2 mutant (strain ZE4929); Shr::spec: non polar null shr mutant (strain ZE4912); ΔN1N2/pXL14: NEAT1-2 mutant complemented with shr (strain ZE4924). A 1 ml volume of each culture at OD₆₀₀ = 1 was processed and 20 μl of the prepared samples were loaded per well. (D) Growth in CDM in the presence of 20 μM of iron. (E) Growth in CDM with 2 mM dipyridyl and no additional source of iron. (F) Growth in CDM with 2 mM dipyridyl and 20 μM of methemoglobin, as the sole source of iron. Diamonds: wild type GAS (strain NZ131); Squares: ΔNEAT1 mutant (strain ZE4925); Triangles: ΔNEAT2 mutant; Crosses: ΔNEAT1-2 mutant; Dots: null shr mutant. (G) Growth of wild type and Shr complemented strains in CDM with 2 mM dipyridyl and 20 μM of hemoglobin as the sole source of iron. Diamonds: wild type GAS; Crosses: ΔNEAT1-2 mutant complemented with shr; Dots: Null shr mutant (shr::aad9) complemented with shr (strain ZE4924). Cells were grown in a 96 well microplate at 37 °C for 24 h and growth was monitored at OD₆₀₀. Each datum point in all of the panels represents the mean of at least two independent experiments performed in triplicates. For clarity purpose the SD (represented by the error bars) is shown only in panel F (in which significant growth differences are found between the strains).

The ability of wild type, an shr null mutant (Fisher et al., 2008), and the isogenic mutant strains described above to use hemoglobin as a sole source of iron was investigated using a growth assay that is based on iron-depleted chemically defined medium (CDM). The wt strain and all of the shr mutants grew well in complete CDM medium containing 20 μM of free iron (Fig. 7D). On the other hand, CDM that was prepared without iron and contained 2 mM of the ferric chelator 2, 2-dipyridyl did not support significant growth of any of the tested GAS strains (Fig. 7E). The addition of 20 μM of methemoglobin to the iron-depleted CDM restored growth of the wt strain to the level obtained with CDM containing 20 μM iron (Fig. 7F), lower hemoglobin concentration however, did not support growth of any of the strains in the iron depleted medium (Fig. S6 A & B). These observations demonstrate that iron is indeed the limiting factor for GAS growth in the 2,2-dipyridyl CDM and that GAS is able to use hemoglobin to satisfy its iron needs as we previously reported (Eichenbaum et al., 1996). No significant growth differences were observed between the shr mutants containing a deletion of NEAT1 (ΔNEAT1) or of NEAT2 (ΔNEAT2) and the wt strain in the 2, 2-dipyridyl-CDM supplemented with methemoglobin. Therefore, the ΔNEAT1 and the ΔNEAT2 shr mutants are not affected in their ability to use hemoglobin as a source of iron. On the other hand, the growth of the shr null mutant and to a lesser extent that of the mutant that was missing both of the NEAT domains (ΔNEAT1-2) was impaired (Fig. 7F). The growth phenotype demonstrated by the shr mutants was reversed by complementation with the shr gene (Fig. 7G). These findings establish that Shr is required for hemoglobin utilization in vivo and suggest that Shr function requires at least one of the heme binding NEAT domains. The addition of methemoglobin in higher concentration (60 μM) supported better growth of the tested GAS strains (Fig. S6C). Thus, the Shr-dependent pathway for hemoglobin utilization in GAS may be of high affinity. However, additional pathways for acquisition of iron from hemoglobin also exist, as previous findings suggest (Montañez et al., 2005).

DISCUSSION

During the infection process, the β-hemolytic GAS can tap into the intracellular heme reservoir due to the potent hemolysins it produces and satisfy its needs for iron with hemoglobin, hemoglobin-haptoglobin, and other heme-containing proteins (Eichenbaum et al., 1996, Francis et al., 1985). Previous observations implicated the surface-exposed NEAT protein, Shr, in the first step of heme acquisition from host proteins by GAS (Bates et al., 2003, Zhu et al., 2008). In this work we have provided the first direct support for this proposition by demonstrating that Shr can obtain heme from methemoglobin and by establishing that Shr function is important for GAS ability to use methemoglobin as an iron source. GAS use of Shr in heme acquisition is reminiscent of NEAT-containing receptors such as the Isd proteins in S. aureus and related proteins from other Gram-positive bacteria. In this study, however, we demonstrate that GAS Shr structure and function are different from that of previously characterized NEAT proteins and suggest that Shr represents a new type of protein family with a different mode of hemoglobin binding and heme acquisition.

Shr represents a family of composite NEAT proteins

Shr is a complex NEAT protein that consists of a unique combination of domains and protein motifs (Fig. 1). The database contains many secreted or surface proteins with NEAT domain(s); a few also carry LRR region(s) including IlsA of B. cereus (Daou et al., 2009). LRR are commonly involved in protein recognition and protein-protein interactions (Kobe & Kajava, 2001). It is possible that the LRR may help facilitate the intra-molecular communications that are likely to take place in Shr or its interactions with the other transport components such as Shp. DUF1533 is a domain of unknown function found in hypothetical proteins that contain secretion or export signals and sometimes other functional regions including LRR. The combination of two DUF1533, a LRR region and two NEAT domains is seen for the first time in Shr, however. Interestingly, Shr seems to be the first characterized NEAT protein that has an EF-hand motif (located between the NEAT1 and the LRR regions, Fig. 1). The EF-hand motif is a calcium-binding domain that is ubiquitous among eukaryotic calcium-binding proteins such as calmodulin, but is also found in bacterial sequences (Michiels et al., 2002, Rigden et al., 2003, Zhou et al., 2006). As with other bacterial proteins with EF-hands, the functional significance of this motif is yet to be determined. Further in silico analysis identified several orthologues of Shr in C. novyi, S. equi sub spp zooepidemicus, S. equi sub spp equi, and S. dysgalactiae. These Shr-like proteins share with Shr significant sequence homology and domain architecture, suggesting that GAS Shr is a representative of a small family of NEAT proteins that evolved by combining domains found in surface or secreted proteins encoded by bacteria from Phylum Firmicutes.

Shr is the only NEAT protein to show ferric heme reduction

We isolated rShr with ferric heme (Fig 2B), indicating that this protein can sequester heme from E. coli. It was previously reported that Shr could be isolated from E. coli apoprotein with a mixture of ferric and ferrous heme (Zhu et al., 2008). When we titrated the protein with free hemin, the increase in the absorption around 414 nm demonstrated that Shr can also acquire ferric heme from solution (Fig. 2C). The concurrent rise in absorption at 427, 540 and 564 nm indicated increasing amounts of rShr-bound ferrous heme and was reversed by oxidation (by ferricyanide, Fig. 2D). This increase in bound ferrous heme following the addition of ferric heme to the purified rShr strongly suggests that Shr autoreduces the heme, even in air. Shr clearly provides a stable environment for the bound ferrous heme. This is the first demonstration that Shr can obtain heme from solution and reduce it to ferrous heme. Heme reduction is a unique property of Shr that to our knowledge has not been reported for any other NEAT protein.

The iron oxidation state in protein-bound heme can have a significant impact on the heme fate and its environment. For example, reduction of the ferric heme in IsdC NEAT domain results in the loss of the heme from the protein (Pluym et al., 2008). On the other hand, the full length IsdA and its single NEAT domain can bind both ferric and ferrous heme. The reduction to ferrous heme, however, changes the iron axial ligand from tyrosine to histidine and renders it accessible to small anionic ligands such as CO. Based on these observations, it is suggested that iron oxidation/reduction lead to subsequent conformational changes with closing/opening of the heme-binding pocket in IsdA (Pluym et al., 2008, Vermeiren et al., 2006). Our in vitro study of rShr shows that the full-length protein is able to bind the heme group in both the ferric and the ferrous forms as well. It seems possible that, as in IsdA however, iron oxidation/reduction may be accompanied by structural changes that influence the heme location within the protein (i.e., binding to first or the second NEAT domain) or the subsequent in vivo steps in the heme trafficking such as the heme transfer to Shp or SiaA.

Functional and sequence analysis show that both NEAT domains in Shr are divergent from one another and from the NEAT domains of the Isd protein family

The optical spectrum of the Shr variants NTD-N1 and NEAT2 demonstrate that they both complex heme, while the NTD protein, in contrast, did not show any heme binding. Sequence alignment demonstrated that only a few of the residues that contact the heme in Isd NEAT domains are conserved in the putative heme biding sites of NEAT1 and NEAT2 in Shr (Grigg et al., 2010), (Fig. S1). Most noticeably, both of the NEAT domains in Shr are missing the potential iron-coordinating residue, Tyr 166, a heme ligand in other NEAT proteins, and only NEAT1 has Tyr 170 (proposed to regulate heme binding and release) (Fig. S1). Therefore, the heme-binding region in both Shr NEAT domains is quite different from that of Isd-like heme-binding NEAT domains and the heme iron may not be coordinated by a tyrosine, at least in the NEAT2 domain of Shr. These observations suggest that Shr NEAT domains have different mechanism for interactions with heme. Likewise, Shp, which can acquire heme from Shr, has a unique iron coordination involving two methionines within the single Shp molecule that function as axial ligands (Aranda et al., 2007). Additional work is needed to determine the heme axial ligands of the NEAT domains in Shr. Future investigations of the evolutionary relation between Shr NEAT domains and Shp, which represents a more remote member of the NEAT family, are warranted.

Both of the NEAT domains in Shr bind heme. The optical spectra of NTD-N1 and NEAT2 proteins as isolated from E. coli indicate that NEAT1 is complexed with ferric heme, and that NEAT2 is bound to both ferric and ferrous heme (Fig. 3C and 3F). Like with the full-length Shr, titration of NEAT2 with increasing concentration of hemin resulted in raising amounts of bound ferric and ferrous heme (Fig. S3). Over time some of the ferric heme in NEAT2 (produced by oxidation with ferricyanide) was reduced to ferrous heme (Fig. 3G). Together these observations suggest that the NEAT2 domain in Shr is capable of autoreduction. Functional differences between NEAT1 and NEAT2 were also found in their interaction with non-heme ligands. While Shr NEAT2 interacts with both fibronectin and laminin molecules, NEAT1 binding to fibronectin is significantly weaker and it does not demonstrate significant binding to laminin (Fig. 5).

A unique site in Shr N-terminal region mediates binding to methemoglobin

We demonstrated in this study, using ELISA with immobilized Shr variants, that the full-length rShr, the NTD, and the NTD-N1 all bound methemoglobin similarly. The NEAT2 protein, on the other hand, did not interact with methemoglobin (Fig. 4A). Therefore, it appears that an unspecified region within Shr NTD mediates its binding to methemoglobin. While this manuscript was in preparation Meehan et al (Meehan et al., 2010) reported that like GAS NTD, the truncated Shr molecule produced by S. equi sub spp equi (consisting mostly of Shr NTD) binds hemoglobin and hemoglobin-haptoglobin complex. The finding that the interaction of Shr with methemoglobin is not mediated by NEAT domains distinguishes Shr from the previously studied Isd receptors. Domain analysis of IsdA and IsdH demonstrated that binding to host hemoproteins is carried out by the single NEAT domain in IsdA and two of the NEAT domains of IsdH (Clarke et al., 2004, Pilpa et al., 2009). Site directed mutagenesis localized the binding to a conserved aromatic motif also found in the first NEAT domain of the IsdB (Pilpa, Robson et al. 2009), which binds hemoglobin and hemoglobin/heptoglobin complex (Torres et al., 2006). Consistent with the results of the ELISA experiments, both of the Shr NEAT domains are missing the hemoglobin binding sites identified in the Isd proteins. Additional work is required to identify the region within Shr NTD that is involved in hemoglobin binding. The sequence of the Shr N-terminal region is unique, and besides the two copies of DUF1533, it shares sequence homology only with Shr orthologues. Therefore, hemoglobin binding is mediated by a new protein motif in Shr.

Hemoglobin binding by bacterial receptors is not fully understood. In this work we show that the Shr N-terminal domain binds only heme-containing hemoglobin (Fig. 4A). The functional significance of this observation is that Shr may release the bound hemoglobin after sequestering all the heme. It is not clear how Shr differentiates between the apo and the holo forms of methemoglobin. Recognition of the heme moiety does not seem to be part of Shr binding to methemoglobin as NTD does not bind heme and the heme is mostly buried within hemoglobin (Genco & Dixon, 2001). We hypothesize, therefore, that Shr recognizes a tertiary structure in the holoprotein that is disrupted when heme is lost, rather than recognizing a linear region within the α or β polypeptides of hemoglobin.

The NTD and NEAT1 in Shr are sufficient for heme acquisition from methemoglobin in vitro

It was previously demonstrated that purified Shr transfers heme directly to apoShp in vitro (Zhu et al., 2008), while the direct movement of heme from methemoglobin to Shp was not observed. Therefore it was hypothesized that Shr is involved in the first step of heme sequestering from hemoglobin. In this study we used the column-immobilized apoNTD-N1 to ask if it could obtain heme following transient interactions with methemoglobin or free heme (Fig. 6). Spectral analysis done with NTD-N1 following the passage of methemoglobin showed that it obtained heme. In a separate assay, co-incubation of apoNTD-N1 with methemoglobin in solution demonstrated that the heme transfer from methemoglobin is fast. NTD-N1 was also able to receive heme from solution, but to a lesser extent. These experiments imply that heme is transfered directly from methemoglobin to the NTD-N1 protein. Further investigations are required to determine if NEAT2 can obtain heme from NEAT1 and/or from methemoglobin, and to find out which of the Shr domains are required for the subsequent step in heme trafficking.

Shr is needed for heme uptake from methemoglobin in vivo

The growth of the shr^- and ΔNEAT1-2 mutants in iron-depleted medium supplemented with hemoglobin was impaired in comparison to that of the wild type GAS strain (Fig 7F). The mutant growth phenotypes were reversed by complementation with the shr gene, establishing the role of Shr in hemoglobin utilization in vivo. The residual growth of the ΔNEAT1-2 in low hemoglobin concentration and the full growth of both mutants observed when higher amounts of hemoglobin were added (Fig. S6C) are consistent with a previous report suggesting that additional hemoglobin utilization pathways are found in GAS (Montañez et al., 2005). Since the shr mutants required a high concentration of hemoglobin to restore growth than the wild type strain, we suggest that Shr mediates a high affinity pathway. It is noteworthy, however, that the deletion of a single NEAT domain (as in the ΔNEAT 1 and ΔNEAT 2 mutants) did not have significant growth defect. These findings indicate that either one of the NEAT domains is sufficient for heme uptake from hemoglobin in vivo, suggesting that despite the differences found between Shr NEAT domains, they have some functional redundancy.

In conclusion, this study establishes the role of Shr in heme acquisition from hemoglobin, and demonstrates that the streptococcal receptor is a representative of a structurally and functionally distinct NEAT protein family found in C. novyi and pyogenic streptococci. We have begun to elucidate the functional domains of Shr, although additional investigations are required to fully understand the mechanism of heme uptake mediated by this intriguing protein.

MATERIALS AND METHODS

Strains, media, and growth conditions

Escherichia coli (E. coli) DH5α and XL1 blue were used for cloning and gene expression. The GAS strain used in this study was NZ131, an M type 49 (Simon & Ferretti, 1991); ZE4912 an isogenic strain with a non polar, null mutation in shr (shr::aad9) (Fisher et al., 2008); ZE4924, a merodiploid strain, which contains both the shr::aad9 and the wild type alleles of shr in the chromosome (Fisher et al., 2008). E. coli cells were grown aerobically in Luria Bertani (LB) medium at 37 °C. GAS cells were grown statically at 37 °C in Todd-Hewitt broth with 0.2% w/v yeast extract (THY, Difco Laboratories) or Chemically Defined Medium (CDM; SAFC Biosciences) as described in Montañez et al 2005 (Montañez et al., 2005). When necessary, 100 μg/ml ampicillin, 100 μg/ml spectinomycin, 70 or 300 μg/ml kanamycin (for E. coli and GAS, respectively) was added to the medium.

DNA manipulations

Chromosomal and plasmid DNA extraction and DNA manipulations, including restriction digest, cloning, and DNA transformation into E. coli or GAS, were done according to the manufacturer's recommendations and with standard protocols as previously described (Eichenbaum et al., 1998, Sambrook et al., 1989). PCR for cloning was performed using the High Fidelity AccuTaq LA DNA Polymerase (Sigma). PCR products were purified with the QIAquick PCR Purification Kit (Qiagen). DNA ligation was done with using Fastlink ligation kit (Epicentre). For RNA extraction and analysis, GAS cells were harvested at the logarithmic growth phase and total RNA was prepared using the RiboPure-Bacteria Kit (Ambion). RNA was quantified spectrophotometrically, and its integrity was examined by agarose gel electrophoresis. For RT-PCR cDNA was produced by Superscript III reverse transcriptase (Invitrogen) according to the manufacturer's specification. The oligonucleotide primers used in this study are listed in Table S1. Table S2 lists and describes the construction of the plasmids used in this work.

Strain construction

The following isogenic mutant series was constructed in NZ131 background: ZE4925 (in frame deletion of NEAT1 in shr, ΔNEAT1), ZE4926 (in frame deletion of the LRR 3’ and most of the NEAT2 in shr, ΔNEAT2), and ZE4929 (in frame deletion of the region between NEAT1 up to and including NEAT2 in shr, ΔNEAT1-2). Alleles with unmarked and in frame deletions in the shr gene with about 1 Kb of flanking sequence were cloned into the temperature sensitive shuttle vector pJRS700, as described in Table S2. The mutations were then introduced into GAS chromosome by transforming NZ131 cells with each of the recombinant vectors and selecting for Kanamycin resistance at 30 °C. The transformants were then passed in antibiotic free medium at 37 °C. Mutants in shr gene, generated by allelic replacement via double homologous recombination were identified by screening for plasmid loss (kanamycin sensitivity). The formation of each mutation was confirmed by PCR and Western blot analysis (Fig. S5). The GAS strains ZE4925, ZE4926, and ZE4929 were engineered using plasmids pXL2, pXL13, and pXL3 respectively. The strain ZE4935 is a merodiploid containing both the wt and the ΔNEAT1-2 shr alleles in the chromosome. For ZE4935 construction, the temperature sensitive plasmid pXL14 was introduced into ZE4929 cells and vector integration into the chromosome (via homologous recombination) was selected on Kanamycin at 37 °C; strain construction was confirmed by PCR and Western blot analysis (Fig. S5 and data not shown).

Overexpression and Purification of Recombinant Shr, NTD, and NTD-N1 and NEAT2

The expression of Strep-tag Shr (pEB2), Strep-tag Shr NTD (pEB10), Strep-tag NTD-N1 (pEB11), or Strep-tag NEAT2 (pHSL2) was induced with 200 ng/ml anhydrotetracycline, overnight at 27 °C. Cells were harvested, resuspended in lysis buffer (100 mM Tris/HCl pH 8, 500 mM sucrose, 1 mM EDTA) with the addition of 0.5 mg/ml lysozyme, β-D glucopyranoside final concentration 0.5% and Complete, mini-EDTA-free protease inhibitor cocktail tablets (Roche) then lysed by sonication. The cells pellet was centrifuged and the cleared lysate was then applied to a Strep-Tactin Superflow column (IBA) with a 5 ml bed volume and purified using FPLC. A step gradient program was used and Strep-tag proteins were eluted with 5 column volumes of 100 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 2.5 mM desthiobiotin. A cation exchange column (4 ml bed volume Hi-trap SP HP, GE Healthcare) was used for further purification of Strep-tag Shr by FPLC. The Strep-tag Shr directly after elution from the Strep-Tactin Superflow column was diluted 1:5 in 50 mM acetic acid pH 4.8, applied to the Hi-trap SP HP column and eluted with 50 mM acetic acid pH 4.8 plus 1M NaCl₂.

His-tagged Shr was expressed (pCB1) and purified as described previously (Fisher et al., 2008) with the following exceptions: when necessary, hemin in dimethyl sulfoxide was added to give a final concentration of 1 μM hemin in the cell culture one hour before cell disruption by sonication and sonication was increased to include ten cycles.

All proteins were prepared in Laemmli sample buffer and separated by sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis (SDS-10%). Western Blot analysis was performed with polyclonal antibodies against Shr raised in rabbit as described previously (Bates et al., 2003). Total protein concentration was measured using a Lowry assay (Pierce Biotechnology, Inc). Each elution fraction was stored in 15% glycerol with 200 μl protease inhibitor cocktail (Complete, mini-EDTA-free, Roche). Fractions used for further study underwent buffer exchange to 20 mM Tris-HCl, 15% Glycerol, pH 8.0 and were stored at -20 °C.

Enzyme-linked Immunoabsorbent Assays (ELISA)

An enzyme-linked immunoabsorbent assay (ELISA) was used to analyze the ability of Strep-tagged Shr, NTD, NTD-N1 and NEAT2 to bind to various ligands. ELISA plate wells (Costar, Corning, inc.) were coated with a 50 μl solution containing the desired concentrations of bait proteins. Wells coated with BSA and uncoated wells were used as controls for non-specific interactions. The bait proteins were diluted in PBS buffer (10 mM phosphate-buffered saline, 100 nM NaCl, pH 7.4) and included rShr, NTD, NTD-N1, NEAT2 and BSA. After the bait proteins were incubated overnight at 4 °C, the wells were washed with PBS-Tween (0.05%) buffer and blocked with 200 μl 5% soy infant formula (Nestle)-PBS-Tween for 1 hour at 37 °C then washed again to remove blocking solution. For apohemoglobin preparation, heme was removed from methemoglobin according to Asakura et al. (Asakura et al., 1964). The desired concentrations of human apo/holomethemoglobin (Sigma), human fibronectin (BD) or mouse laminin (BD) in 5% soy/PBS-Tween were then added to each well (50 μl / well). The wells were then washed PBS-Tween/well to remove unbound protein. Fifty μl of a 1:15,000 dilution of polyclonal rabbit anti-hemoglobin (sigma), rabbit anti-fibronectin (abcam) or rabbit anti-laminin (abcam) antibodies in blocking buffer were subsequently added to each well and incubated at 37 °C for one hour, and the wells were then washed. Fifty μl of goat anti-rabbit IgG conjugated to alkaline phosphatase (Sigma) at 1:6000 dilution in blocking buffer was added to each well and incubated at 37 °C for 1h. Interactions between recombinant Shr, NTD, NTD-N1 or NEAT2 and the ligands were then detected by adding pNPP substrate and developing the chromogenic reaction (KPN). Plates were read at 405 nm on an automated ELISA reader for intervals up to one hour after development. An assay to assess the ability of a variety of blocking buffers to diminish nonspecific binding was also performed as described above except the ELISA wells were uncoated.

rShr, NTD, NTD-N1 and N2 UV-visible spectra

The spectrophotometric analysis of samples from 250 to 700 nm was carried out using a Varian Cary 50 Bio spectrophotometer. Absorption spectra of the purified proteins were measured on the spectrophotometer in a quartz cell with an optical path length of 10 mm. All absorption spectra shown in this study are representative of multiple experiments done with at least three biological replicas.

Heme titration assays

A stock solution of hemin chloride in DMSO was prepared. The absorbance of a 1:1000 dilution of the stock solution at 404 nm was recorded and the concentration of hemin chloride in the stock solution was calculated using Beer's law (A= εbc where hemin in DMSO ε₄₀₄=188,000 m^-1cm^-1 (Brown & Lantzke, 1969, Collier et al., 1979). Protein samples were diluted in Strep-tag elution buffer to 3 μM. Absorbance from 250-700 nm was recorded before addition of hemin chloride. Hemin chloride was added to 1 ml aliquots of 3 μM protein to a final hemin chloride concentration of 1 μM, incubated with stirring at 4° C for 1 h and the absorbance from 250-700 nm was scanned and recorded. This was repeated for hemin chloride concentrations of 3 μM, 5 μM, 10 μM and 20 μM. Strep-tag elution buffer alone was similarly incubated with 1 μM, 3 μM, 5 μM, 10 μM and 20 μM of hemin chloride. These heme-containing buffer solutions were scanned as blanks for the UV-visible spectra of the protein solution containing corresponding concentrations of hemin chloride. The total volume of DMSO added to the protein solutions or the blank solutions ranged from 0.15 μl to 3 μl. Thus, the final DMSO concentration in the sample was 1.5×10^-4 to 3×10^-3 v/v. Ferricyanide and DTT treatment were done in sealed tubes with incubation at room temperature. Ferricyanide was then removed by dialysis.

Heme transfer from methemoglobin

Heme transfer from methemoglobin to Shr fragment NTD-N1 was done using FPLC. ApoNTD-NEAT1 was prepared according to the method described by Asakura et al (Asakura et al., 1964) and 100 nmoles of the protein in 2 ml Strep-tag wash buffer were attached to a Strep-Tactin Superflow column. Equivalent moles of methemoglobin (100 nmoles) were flowed through the immobilized NTD-N1. The bound methemoglobin was removed by washing several times (10 column volumes) with a wash buffer (100 mM Tris-HCl pH 8.0, 250 mM NaCl, 1 mM EDTA). Subsequently, the immobilized NTD-N1 was eluted with Strep-tag elution buffer (100 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 2.5 mM desthiobiotin). Methemoglobin flow-through and NTD-N1 elution samples were collected and analyzed by Western blot using anti-Shr and anti-hemoglobin antibodies. Protein concentration in the different fractions was determined by Modified Lowry assay, and Spectroscopic analysis (250 - 700 nm) of 10 μM apoNTD-N1 before and after passage of methemoglobin was carried out using a Varian Cary 50 Bio spectrophotometer.

Culture in microplate

All GAS strains were grown in CDM supplemented with 3 mM of L-cysteine, 15 mM of sodium bicarbonate, 2.5 mM of magnesium sulfate, 44 μM of calcium chloride, 15 μM of zinc chloride, 20 μM of manganese and either 20 μM of metal iron or a range of concentrations of human hemoglobin. In the latter case, 2 mM of 2, 2-dipyridyl was added to the supplemented CDM to completely chelate residual metal iron prior to the addition of Hb. The prepared media were inoculated to final OD₆₀₀ = 0.005. The inocula consisted of cells grown overnight at 37 °C on blood agar and suspended in iron-free CDM. Bacterial suspensions at OD₆₀₀ = 0.5 were diluted 1:100 into the corresponding medium, which was then dispensed in 200 μl triplicates in a 96 well microplate (Costar 3595, Corning Inc.) at 37 °C for 24 h. The experiments were performed in triplicates and were done at least twice for each strain. Kanamycin (150 μg/ml) was added to the medium for the growth of the complemented strain ZE4924 and ZE4935.

In silico analysis

The following accession numbers were utilized for the NEAT domain-containing proteins examined in this study: S. aureus IsdA ABX29083, S. aureus IsdC ABX29084, S. aureus IsdB YP_001332074, S. aureus IsdH Q6G8J7, S. equi subs. zooepidemicus YP_002122760, C. novyi NT YP_877540, S. dysgalactiae subs. equisimilis YP_002997560, S. pyogenes Shr MGAS5005 ABW80932. Identification of all protein domains was conducted by SMART analysis except the EF-hand domain, which was identified by PROSITE. Multiple sequence alignment the NEAT domains were executed using the ClustalW program.