Development and atomic structure of a new fluorescence-based sensor to probe heme transfer in bacterial pathogens

Abstract

Heme is the most abundant source of iron in the human body and is actively scavenged by bacterial pathogens during infections. Corynebacterium diphtheriae and other species of actinobacteria scavenge heme using cell wall associated and secreted proteins that contain Conserved Region (CR) domains. Here we report the development of a fluorescent sensor to measure heme transfer from the C-terminal CR domain within the HtaA protein (CR2) to other hemoproteins within the heme-uptake system. The sensor contains the CR2 domain inserted into the β2 to β3 turn of the Enhanced Green Fluorescent Protein (EGFP). A 2.45 Å crystal structure reveals the basis of heme binding to the CR2 domain via iron-tyrosyl coordination and shares conserved structural features with CR domains present in Corynebacterium glutamicum. The structure and small angle X-ray scattering experiments are consistent with the sensor adopting a V-shaped structure that exhibits only small fluctuations in inter-domain positioning. We demonstrate heme transfer from the sensor to the CR domains located within the HtaA or HtaB proteins in the heme-uptake system as measured by a ~60% increase in sensor fluorescence and native mass spectrometry.

1. Introduction

Heme is an important protein cofactor that has critical roles in a range of cellular processes, including electron transfer during cellular respiration, photosynthesis, and enzyme catalysis [1, 2]. It is also the most abundant source of iron in the human body, which is exploited by many microbial pathogens as a nutrient [3–9]. Monoderm gram-positive bacterial species are surrounded by a thick peptidoglycan sacculus that can contain a range of hemoproteins that scavenge free heme and promote its transit from the bacterial surface to the underlying plasma membrane [10]. These extracellular heme binding proteins in gram-positive bacteria from the Firmicutes phylum (e.g. Staphylococcus aureus, Bacillus anthracis, Group A streptococci, etc.) have been extensively studied, and use conserved NEAr iron Transporter (NEAT) domains to bind and transfer heme [11, 12]. The NEAT domains transiently associate with one another to transfer heme from one NEAT domain to another in a process that occurs up to ~70,000-times faster than the rate at which each domain spontaneously releases heme to its surroundings [13–16]. Much less is known about how heme is acquired in distantly related bacteria from the Actinobacteria phylum (e.g. Corynebacterium diphtheriae, Corynebacterium glutamicum, Nocardia abscessus, Propionibacterium spp., and Bifidobacterium spp.) whose cell walls are distinguished by additional layers of arabinogalactan and mycolic acids [17]. The extracellular heme binding proteins in these microbes bind heme using distinct Conserved Region (CR) domains, whose molecular mode of action is incompletely understood [18]. An understanding of how these hemoproteins bind and transfer heme is of fundamental importance and could lead to new antibiotics that work by disrupting microbial heme acquisition.

C. diphtheriae is the causative agent of the respiratory disease diphtheria, which can result in effects ranging from sore throat and mild fever, to more severe symptoms including endocarditis and pneumonia [19]. During infections, the bacterium secretes the diphtheria toxin as part of the response to low iron conditions that are sensed by the diphtheria toxin repressor (DtxR) [20, 21]. Several iron and heme acquisition proteins are under the control of DtxR and are important for growth in iron limited conditions [22, 23]. In order to acquire heme, C. diphtheriae employs an array of surface associated and secreted proteins that bind to heme using CR domains [18, 24–26]. The surface-associated proteins are bound to the plasma membrane via a C-terminal transmembrane helix (TMH), and in some instances the proteins are also released from the membrane into the surrounding milieu [18, 27]. Five CR-containing proteins have been identified: HtaA, HtaB, ChtA, ChtB and ChtC (Fig. 1). In the heme acquisition process these proteins bind to free heme and/or to heme that is released from human hemoglobin (Hb) that is bound to the cell surface by the microbial HbpA protein [18, 27]. HbpA can directly bind to Hb and presumably the Hb-haptoglobin (Hp) complex [27–29]. Heme molecules present on the cell surface are then thought to be transferred through the network of CR domain-containing proteins to HmuT, the lipoprotein component of the HmuTUV ABC-type transporter that pumps heme into cytoplasm [22, 30]. Imported heme is then either directly bound by bacterial hemoproteins or degraded by the HmuO heme enzyme to release free iron [31, 32]. The CR-containing hemoproteins play an important role in enabling C. diphtheriae to use Hb as an iron source, as nonpolar deletions of the genes encoding these proteins impair microbial growth in media in which the Hb-Hp complex is the sole source of iron [26, 27]. At present, only structures of the CR domains from C. glutamicum have been determined, revealing a novel fold for a heme-binding protein [33]. The binding affinities and order of heme transfer between the components of the heme uptake machinery are not known.

Figure 1. Heme import pathway in C. diphtheriae.

Hemoglobin (Hb) (and the Hb-haptoglobin complex, not shown) is bound by the HbpA protein. Heme released from Hb or free heme is then bound by a series of membrane-associated hemoproteins: HtaB, ChtB, HtaA, ChtA and ChtC. The HtaB and ChtB proteins, as well as the ChtA and ChtC proteins, share a high level of sequence homology in their CR domains and are presumably paralogs [25]. Heme is then transferred to HmuT, the lipoprotein component of the HmuTUV importer that moves heme into the cytoplasm. Heme is then thought to be degraded by the intracellular HmuO oxidase to release free iron or possibly directly incorporated into microbial hemoproteins. The order of heme transfer via the CR domain-containing proteins is not known.

Cell wall embedded hemoproteins in C. diphtheriae presumably function to transfer heme from the cell surface to the membrane. Indeed, co-incubation of heme acceptor and donor CR domains for 30 minutes followed by chromatographic separation was previously used to demonstrate that heme is transferred from HtaA to HtaB [24]. However, more rigorous kinetic measurements of this heme transfer process have not been possible because many of the CR domains exhibit very similar UV/Vis absorbance spectra [18, 24]. To overcome this limitation, we constructed a CR domain-containing heme binding sensor using a previously described strategy that exploits the ability of heme to quench the fluorescence of the green fluorescent protein [34, 35]. This sensor, called CiG, contains the C-terminal CR domain from the HtaA (CR2) protein genetically fused within the enhanced green fluorescent protein (EGFP). We show that the intrinsic fluorescence of the EGFP chromophore in the sensor is quenched by heme bound to the adjacent CR domain, providing a facile method to probe heme binding by CR2, as well as homotypic (same CR type) and heterotypic (different hemoproteins) heme transfer events. Further, by determining the crystal structure and performing small-angle X-ray scattering (SAXS) experiments of the CiG sensor, we gained insight into the determinants through which CR domains bind heme in Actinobacteria.

2. Materials and Methods

2.1. Cloning, expression, and purification of proteins

The nucleotide sequences encoding the HtaA and HtaB CR domains were cloned out of C. diphtheriae (NCTC 13129) genomic DNA by polymerase chain reaction (PCR) using primers flanking the regions encoding HtaA residues 32–214 for its N-terminal CR domain (CR1), residues 344–506 for its C-terminal CR domain (CR2) (Integrated DNA Technologies, Coralville, IA), and HtaB residues 30–250 for its CR domain. The inserts were then cloned into a pET-28b-based plasmid containing N-terminal 6xHis and small ubiquitin-like modifier (SUMO) fusion tags (pSUMO) [36]. The intra-domain fusion of CR2 within EGFP (CiG) was constructed by multi-fragment Gibson assembly with the gene fragment encoding the CR2 domain inserted in the EGFP sequence between amino acids Y39 and G40, and inserted into the pSUMO vector. The EGFP gene used was a variant containing the mutations F64L/S65T/L221K/H231L relative to wild-type GFP, which results in the enhanced fluorescence phenotype (EGFP, F64L/S65T/H231L) while favoring the monomeric form of the protein (L221K) [37–39]. Site-directed mutagenesis was used to generate the Y60F (Y361F in HtaA numbering) variant of CiG. The resulting plasmids were transformed into E. coli BL21 (DE3) competent cells (New England Biolabs, Beverly, MA), and cultures were grown in LB media at 37°C to an OD₆₀₀ of 0.6–0.8. Protein expression was induced with 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) and allowed to continue overnight at 17°C. Cells were harvested by centrifugation and resuspended in a lysis buffer (50 mM sodium phosphate, 300 mM NaCl, pH 7.0) supplemented with 1 mg/mL egg white lysozyme, 2 mM phenylmethylsulfonyl fluoride (PMSF, Sigma), a protease inhibitor cocktail (Sigma-Aldrich), and 0.5 mg Serratia marcescens nuclease [40]. The cells were lysed by sonication and then clarified by centrifugation. Supernatant was applied to HisPur Co²⁺-chelating resin (Thermo Fisher Scientific), washing unbound proteins out with lysis buffer plus 10 mM imidazole. The 6xHis-SUMO tag was removed from the proteins of interest by adding the ULP1 protease (purified in-house) to the resin, followed by elution of the target proteins in low imidazole buffer. An Akta FPLC system (GE Healthcare Life Sciences) was used for further purification using a Superdex S75 size exclusion column. Proteins were concentrated in Amicon Ultra centrifugal filters (EMD Millipore), and final purity was confirmed to be >98% by SDS-PAGE.

2.2. Preparation of apo and heme bound proteins

The CR1, CR2, HtaB and CiG proteins were initially purified as a mixture of heme-free (apo) and heme-bound (holo) forms due to their ability to scavenge heme from the E. coli expression host. The amount of heme-bound protein can be estimated via UV/Vis spectroscopy using the relative intensity of the bound heme Soret peak at ~404 nm compared to the absorbance at 280 nm (A_Soret/A₂₈₀), further described in Supplementary Methods. In order to obtain apo-CR2 and apo-HtaB CR domains, purified protein was subjected to heme extraction using methyl ethyl ketone followed by dialysis back into the lysis buffer. After this procedure the Soret absorbance was negligible indicating successful removal of heme, while ¹H-¹⁵N heteronuclear single quantum coherence (HSQC) nuclear magnetic resonance (NMR) spectra indicated that the CR domains properly refolded (Supplementary Methods, Fig. S1) [41, 42]. Unfortunately, this approach could not be used to produce apo-CiG as the refolding step after heme removal does not result in a fluorescent GFP chromophore. The Y60F-CiG variant was purified in the fully apo form.

Holo-CR1, -CR2, and -CiG were generated by incubating them with a 20-fold molar excess of hemin while rocking for 1 hour. The UV/Vis spectrum of these samples showed a broad absorbance band at ~385 nm, indicating the presence of free hemin. Separation of holoprotein from free hemin was achieved by application to DEAE-Sepharose anion exchange resin [13], eliminating the broad free hemin absorbance feature while a relatively sharp protein-bound Soret band was observed. Completeness of heme loading was verified by both NMR experiments and native mass spectrometry (Figs. S1, S4). In the holo forms, the A_Soret/A₂₈₀ ratios were 2.3, 2.8, and 1.7 for CR1, CR2, and CiG, respectively, with these disparities arising from differences in the apo-protein molar extinction coefficients (see Supplementary Methods). Incubation of EGFP and Y60F-CiG was performed as above, but no bound heme was detected in either protein.

2.3. UV/Vis and fluorescence spectroscopy

UV/Vis absorbance spectra were collected on a Shimadzu UV-1800 spectrophotometer over the wavelength range 250–700 nm in a 1 cm path length cuvette. Proteins were prepared in a buffer of 50 mM sodium phosphate, 150 mM NaCl, pH 7.2. Fluorescence excitation and emission spectra of the two CiG constructs were collected on a SpectraMax iD3 plate reader in a 1.0 mm path length SpectraDrop micro-volume plate (Molecular Devices, San Jose, CA). Quantum yield values were determined using the relative determination method with a fluorescein standard (in 0.1 M NaOH) as described in reference [43]. In this procedure spectra were generated with excitation at 460 nm and integrated above 500 nm. The data were there analyzed using the equation:

$${\Phi }_x={\Phi }_{std}\times \left[\frac{{\mathit{\text{Area}}}_x}{{\mathit{\text{Area}}}_{std}}\right]\times \left[\frac{{\eta }_x}{{\eta }_{std}}\right],$$

where Φ is the quantum yield value, Area is the integrated emission intensity, and η is the solvent refractive index, with subscripts ‘x’ for the sample of interest and ‘std’ for the fluorescein standard. The spectral overlap integral (J) between normalized EGFP donor fluorescence emission and CR2 acceptor absorbance spectra was calculated using FluorTools a|e software [44], and was used to calculate the theoretical Förster distance, R₀, using the equation:

$$R_0=0.211\times {\left[\frac{{\kappa }^2\Phi J}{{\eta }^4}\right]}^{\left(1/6\right)},$$

where κ is an orientation factor between donor and acceptor (2/3 for a random mutual orientation of donor and acceptor). The efficiency of fluorescence quenching for a given chromophore pair can then be calculated by the following equation relating the distance between chromophores, R, with the Förster distance R₀:

$$E=\frac{1}{\left[1+\left(R/R_0\right)\right]}$$

Excitation scans were performed by varying excitation wavelength over the range 420–510 nm in 1 nm increments while observing emission intensity at a constant wavelength of 550 nm. The emission scans were conducted by constant wavelength excitation at 440 nm while monitoring emission intensity at a varying wavelength in the range 480–620 nm in 1 nm increments. For the titration of apo-CR domains to holo-CiG, CiG was diluted to a concentration of 0.5 μM with the addition of increasing stoichiometric equivalents of apo-CR2 (homotypic transfer) or apo-HtaB (heterotypic transfer). The resulting intensities at 512 nm were used to calculate the degree of fluorescence recovery.

2.4. Crystallization, data collection, and structure determination

Purified holo-CiG was concentrated to 9.8 mg/mL in 20 mM HEPES, pH 7.5 buffer to screen crystallization conditions. Crystals developed after 3 days at room temperature in a hanging drop, vapor diffusion format against a mother liquor of 1.9 M ammonium sulfate, 0.1 M sodium cacodylate trihydrate, pH 6.4, 0.2 M sodium chloride. The crystals were cryoprotected in a solution of mother liquor containing 20% glycerol for 30 seconds prior to freezing. Diffraction datasets were collected at the Advanced Photon Source (Argonne National Laboratory) on beamline 24-ID-C equipped with an EIGER2 X 16M detector at 100K. Data were collected at a wavelength of 0.97911 Å, a detector distance of 350 mm, and 0.25° oscillations, with crystals diffracting to 2.45 Å resolution. XDS was used to index and integrate reflections in the P2₁ space group [45, 46]. The data were highly anisotropic, so additional ellipsoidal processing and anisotropic scaling was performed using the STARANISO server [47]. The asymmetric unit contains four molecules of CiG, resulting in a Matthews coefficient of 3.27 ų/Da and 62.42% solvent content in the unit cell. Molecular replacement using PHASER with the known structure of EGFP (PDB: 4EUL) and a homologous CR domain structure from Corynebacterium glutamicum HtaA (PDB: 6JSA) led to a single solution [33, 48, 49], which was iteratively improved by manual model building in Coot and automatic refinement in BUSTER [50, 51]. Refinement was carried out with LSSR-type restraints for NCS and a high-resolution EGFP reference target (PDB: 4EUL), and TLS for the individual domains [52, 53]. Statistics for data collection and refinement are presented in Table 2, and final coordinates and structure factors have been deposited in the Protein Data Bank under PDB code 8SMU.

Table 2.

Statistics of CiG crystal structure data collection and refinement

	CiG
Data collection
Space group	P 1 21 1
Cell dimensions
a, b, c (Å)	63.826, 272.783, 73.518
α, β, γ (°)	90.000, 114.215, 90.000
Resolution (Å)	49.025–2.353 (2.648–2.353)
Wavelength (Å)	0.97911
Total observations	193381 (9600)
Unique reflections	56878 (2846)
Rmerge (%)	10.4 (61.6)
I / σI	7.3 (1.8)
CC1/2	99.4 (68.6)
Completeness	92.2 (74.2)*
Multiplicity	3.4 (3.4)
Wilson B-factor (Å2)	61.9
Refinement
Resolution (Å)	48.32–2.45
No. of reflections	56446
Rwork / Rfree (%)	20.9/23.3
No. atoms	12400
Protein	11858
Ligand/ion	466
Water	76
B-factors (Å2) (all atoms)	62.25
Protein	62.32
Ligand/ion	64.26
Water	39.44
R.m.s. deviations
Bond lengths (Å)	0.007
Bond angles (°)	0.99
Ramachandran favored (%)	97.69
Ramachandran allowed (%)	2.12
Ramachandran outliers (%)	0.19
PDB ID	8SMU

Statistics for the highest resolution shell are shown in parentheses.

^*Values for completeness are reported for the ellipsoidal shells. Spherical completeness values are 60.2 overall, with 98.7 for the lowest resolution shell and 10.1 for the highest resolution shell.

2.5. Small angle X-ray scattering (SAXS) and native mass spectrometry (MS)

SAXS data were acquired at the SIBYLS beamline (Advanced Light Source beamline 12.3.1, Lawrence Berkeley National Laboratory) [54]. Purified holo-CiG at a concentration of 10 mg/mL in 50 mM sodium phosphate, 300 mM NaCl, pH 7.0 buffer was submitted for SEC-SAXS data collection. The sample was injected using an Agilent 1260 series HPLC equipped with a Shodex KW-803 analytical size exclusion column at a flow rate of 0.5 mL/min in sample buffer. SEC-MALS was performed using in-line detection with Wyatt Dawn Heleos multi-angle light scattering, DynaPro Titan quasi-elastic light scattering, and Optilab rEX refractometry detectors, and analysis was carried out using Wyatt Astra 7 software, indicating a monomeric species in solution. Scattering data was acquired on a Dectris PILATUS3 X 2M detector located 1.4 m from the sample, with X-rays at a wavelength of 1.240 Å. Exposures of 2 seconds in length were collected continuously across the chromatographic elution. SEC-SAXS data were processed in ScÅtter IV and analyzed using BioXTAS RAW and the ATSAS Suite [55–57]. C-terminal residues absent from the crystal structure were modeled using Coot prior to performing a minimal ensemble search procedure using the BILBO-MD service to sample possible interdomain orientations [50, 58].

Proteins analyzed by native MS were buffer exchanged into 50 mM ammonium acetate, pH 7.0 and electrosprayed from platinum-coated borosilicate emitter tips pulled on a Sutter P-1000 Micropipette Puller into a Q-Exactive UHMR Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher Scientific, Bremen, Germany). The spray voltage was between 0.6 and 1.8 kV, the capillary temperature was set to 150°C, and transmission and detection ion optics were optimized for low mass-to-charge (m/z) ratio. Data was collected in replicates that were either sampled straight from the mixture or diluted up to 8-fold in 50 mM ammonium acetate to facilitate sample loading.

The relative dissociation constant between CiG and CR2 to heme (K_Relative) was determined from protein-ligand competition experiments using an approach adapted from Krishnaswamy et al. [59]. Briefly, apo-CR2 at concentrations of 1.5, 3.0, and 6.0 μM was incubated with 1.5 μM fully holo-CiG in duplicate for a total of six replicates. Reactions were allowed to equilibrate for 30 minutes at room temperature before native MS analysis to quantify their apo and holo peak intensities. The equation for determining relative K_D for CiG to heme and CR2 to heme (K_Relative) from native MS signal intensities is described below:

$$K_{\mathit{\text{Relative}}}=\frac{\left[CR2\right]\left[CiG\bullet \mathit{\text{heme}}\right]}{\left[CiG\right]\left[CR2\bullet \mathit{\text{heme}}\right]}=\frac{Ab\left(CR2\right)Ab\left(CiG\bullet \mathit{\text{heme}}\right)}{Ab\left(CiG\right)Ab\left(CR2\bullet \mathit{\text{heme}}\right)},$$

where Ab(protein) indicates the sum of all peak intensities corresponding to the indicated protein species. The K_Relative between CiG and CR2 was found to be 1.04 +/− 0.08, indicating nearly identical heme binding affinity.

3. Results

3.1. Development of CiG, a sensor to measure heme binding and transfer to CR domains

Our goal is to learn how components of the C. diphtheriae heme-uptake machinery capture heme and import it into the cell. Typically, heme transfer between proteins is monitored by observing heme-dependent changes in their UV/Vis spectra; the absorbance maxima of heme-binding proteins can be distinct as it reflects the details of heme coordination. If the heme donor and acceptor have sufficiently different wavelengths or molar extinction coefficients for these maxima, changes in their spectra can be used to track transfer. However, monitoring heme transfer between the CR domain components of C. diphtheriae is challenging because many of these proteins exhibit similar UV/Vis spectra when bound to heme [18, 24]. This is evident from the similar spectral properties of the CR domains that are located within the HtaA protein, CR1 and CR2 (Fig. 2b). To explore the mode of binding and heme transfer in HtaA we developed a sensor that enables heme binding to its CR2 domain to be detected by changes in fluorescence. The reporter, called CiG, contains the CR2 domain from HtaA fused inside EGFP, and leverages the established ability of proximally-bound heme molecules to quench EGFP fluorescence via resonance-energy transfer (RET) (Fig. 3a) [34]. Initial modeling studies suggested that a traditional head-to-tail fusion of the CR2 domain with EGFP would position the heme too far from the EGFP chromophore to promote efficient quenching (~70 Å). However, the N- and C-termini in the predicted structure of CR2 are proximally positioned, suggesting that the CR2 domain could be fused to EGFP via an intra-domain insertion, as previously employed by Arpino et al. and Hanna et al. to fuse the E. coli cytochrome b₅₆₂ protein to EGFP [35, 60]. In the CiG reporter the amino acid sequence of CR2 was therefore inserted into the turn in EGFP that connects its β2 and β3 strands (Fig. 2a, bottom). Spectroscopic data are consistent with the component domains within CiG being fully folded and functional, as the UV/Vis spectra of heme bound CiG are similar to CR2 (both proteins exhibit Soret maxima at 405 nm and a charge-transfer band at 624 nm) (Fig. 3b, Table 1) [61]. The spectra of CiG are also consistent with prior studies that have shown that CR2 binds to the oxidized form of heme (called hemin) [62]. As desired, CiG also exhibits absorbance and fluorescence properties similar to EGFP, with a chromophore absorbance maximum centered at 490 nm, and a fluorescence emission peak at 512 nm. Notably, adding hemin to the purified CiG receptor causes its fluorescence emission to steadily decrease, which is consistent with RET between the heme bound to CR2 and the chromophore within the EGFP domain (Fig. 3c). This is further supported by the absence of significant hemin-dependent fluorescence changes in a Y60F receptor variant that is defective in heme binding because its presumed CR2 axial heme-binding tyrosine residue (Y361 in HtaA, Y60 in CiG) is eliminated [24, 62] (Figs. 2a, S2). Based on CiG’s degree of fluorescence quenching at saturating hemin levels and the overlap integral of EGFP fluorescence and CR2 absorption spectra, the distance between the chromophores is expected to be about ~40 Å (see Methods).

Figure 2. The HtaA hemoprotein and CiG sensor.

(a) Schematic of the C. diphtheriae HtaA heme receptor showing the location of its heme binding Conserved Region (CR) domains in the primary sequence. Shown below the schematic are the polypeptides investigated in this study, which include isolated heme binding CR1 and CR2 domains from HtaA. The schematic also shows the fluorescent heme binding reporter CiG, which contains the CR2 domain from HtaA connected via intra-protein fusion to EGFP. The residue numbers of the parent proteins are indicated in parentheses. An asterisk and red bar indicate the position of the substitution for the Y60F-CiG variant, which would correspond to an Y361F substitution in CR2. (b) Overlay of UV/Vis spectra from heme-loaded holo-CR1 (orange-red) and holo-CR2 (blue), highlighting the similarity of their Soret and charge-transfer bands.

Figure 3. Spectral properties of the CiG heme binding sensor.

(a) Schematic showing how heme binding to CiG alters its fluorescence properties. The intra-domain fusion of the CR2 domain within EGFP reports on the heme binding to the CR domain as a reduction in green fluorescence from the EGFP domain, as the bound heme molecule causes quenching. When not bound to heme, CiG fluoresces at 512 nm when excited at 490 nm, which is quenched upon heme binding. (b) Overlay of UV/Vis spectra of heme-loaded (holo) CR2 (blue) and CiG (green). The data show that CiG binds to heme in a similar manner as CR2. (c) Fluorescence emission scans as a function of adding hemin to purified CiG, demonstrating its heme-dependent quenching. Spectra are shown with 0.0, 0.25, 1.0, 4.0, and 20.0 molar equivalents of hemin added to CiG. CiG as purified from E. coli is ~30% bound with heme.

Table 1.

Fluorescence properties of the CiG heme binding reporter

Protein	λSoret (nm)	λex (nm)	λem (nm)	quantum yield
EGFP	n/a	488	511	0.62 ± 0.04
CiG (holo)	404	490	512	0.26 ± 0.02
Y60F-CiG (apo)	n/a	490	512	0.62 ± 0.02
CR2	404	n/a	n/a	n/a

3.2. Crystal structure of the CiG heme binding reporter

In order to gain insight in the molecular basis of heme binding and the relative positioning of the GFP and heme chromophores, the crystal structure of the CiG reporter was determined at 2.45 Å resolution. The reporter crystallized in the P2₁ space group with four protein molecules located in the asymmetric unit (Table 2). The structure was determined using molecular placement methods and is well defined by continuous electron density for the entirety of the protein, except for nine flexible C-terminal amino acids in EGFP (Fig. 4a). In the structure the EGFP domain adopts an expected β-barrel fold that surrounds its chromophore; the coordinates can be overlaid with previous high-resolution structures of isolated EGFP with a backbone root mean square deviation (RMSD) of 0.3 Å (compared to PDB:4EUL) [49]. Within the asymmetric unit, the CiG molecules adopt related structures in which their CR2 and EGFP domains pack against one another to form a V-shaped conformation. However, the positioning of the domains in the molecules differs slightly, such that the angle between them varies between 62.9–79.1° and the positioning of the heme is translated up to 13 Å (Fig. 4d). The domain positioning is maintained by the dual interdomain tethers and a primarily hydrophilic interface that buries 398–482 Ų of surface area. The distance between the heme iron atom within the CR domain and the closest edge of the EGFP chromophore (R) ranges from 41.0–43.9 Å in the four receptor structures that comprise the asymmetric unit (Fig. 4b,c). This inter-chromophore distance agrees well with the observed amount of heme-dependent fluorescence quenching (Table 1), and the calculated R₀ for the CR2-EGFP chromophore pair of 42.3 Å (see Methods). Importantly, the heme binding pocket in the reporter is unobstructed by EGFP suggesting that both the receptor’s affinity and ability to transfer heme will be similar to the native CR2 domain (Fig. 4a).

Figure 4. Structure of CiG heme binding reporter.

(a) Cartoon representation of a single receptor protein in the 2.45 Å CiG crystal structure. The EGFP domain is colored green, while the CR2 domain from HtaA is blue. The β1-β2 portion of EGFP (EGFP residues 1–39) that precedes the insertion point for the CR2 domain is shaded light green, while the β3-β11 portion (EGFP residues 40–238) that follows CR2 is colored bright green. The CR2-bound heme molecule is shown in stick representation and colored yellow, while the EGFP chromophore is shown in green sticks. (b) Top view of the CiG structure with surfaces shown. The image shows the relative locations of the EGFP chromophore and the CR2-bound heme molecule. A dashed line connects the chromophores, which are separated by 41.0–43.9 Å in the four reporter proteins that form the asymmetric unit. (c) The four molecules of CiG in the crystallographic asymmetric unit, colored by chain. (d) Overlay of the four molecules present in the asymmetric unit. The coordinates of the EGFP domain are superimposed, revealing that the relative orientation of the CR2 domains varies slightly. In the four structures the location of the heme pocket differs by as much as 13 Å relative to EGFP. The structure of the CR2 domains in each of the four CiG molecules that comprise the asymmetric unit are similar (backbone coordinate RMSD of ~0.4 Å).

The CR2 domain in the sensor is composed of 11 β-strands that form a β-sandwich constructed from two anti-parallel β-sheets. One of the sheets extends to form the heme binding pocket that also contains two α-helices (Fig. 5a). The binding pocket is formed by a groove that is at the edge of the β-sandwich (Fig. 5b). In the pocket, Tyr361 (numbered based on the sequence of the full-length HtaA protein) located in the α1 helix provides the axial coordination to the iron atom within heme, and forms additional hydrogen bonding interactions with the side chain of His412 located in the β6-β7 loop (Fig. 5c). These side chains are also positioned to form π-π stacking interactions with the pyrrole rings of the heme molecule. Additional hydrogen bonds are present between a propionate group of the bound heme and the side chain hydroxyl groups of Ser357 (helix α1) and Tyr490 (α2-β11 loop), while the other propionate group in the heme is solvent exposed. On the face of the heme molecule opposite the Tyr361-iron axial bond, two residues form capping interactions that appear to limit heme exposure to the solvent: Met443 (β8-β9 loop) and Phe489 (α2-β11 loop). Phe489 likely forms a π-π interaction with a pyrrole group in heme that appears to be a common feature in recognition by CR domains as this residue is highly conserved. The conformation of the CR domain in the sensor is closely related to the recently reported structure of the C-terminal CR domain from the HtaA protein in C. glutamicum (called HtaA-C, PDB: 6JSA) [33]. The backbone coordinates of CR2 and HtaA-C can be superimposed with a RMSD 0.84–0.92 Å, consistent with the proteins sharing 56.4% sequence identity.

Figure 5. Molecular basis of heme binding to the CR2 domain within CiG.

(a) Cartoon representation of the CR2 domain located within the structure of CiG. The secondary structure elements are labeled: α-helices are colored red, β-strands are colored blue, and loops are colored light cyan. The heavy atoms of heme are shown, with carbon, oxygen, and nitrogen atoms colored yellow, red, and blue, respectively. (b) Surface representation of the CR2 domain, showing the location of the bound heme. Residues that form direct interactions with heme (heme pocket) are highlighted with orange shading. (c) Heme-interacting residues are colored orange and annotated with their residue numbering as in the full-length HtaA protein. The axial linkage from Tyr361 to the heme-iron is shown with magenta dashes, and hydrogen bonds are shown as cyan dashed lines. (d) Contacts between the four CR2 domains in the asymmetric unit of the crystal are partially mediated by their heme pockets, with two different relative interprotein orientations: one in which the heme molecules are at an angle relative to each other (chain A:C interface) (top), and one in which the heme molecules are parallel-displaced (chain B:D interface) (bottom).

3.3. Small angle X-ray scattering experiments reveal that CiG adopts a semi-rigid structure

In the asymmetric unit of the crystal all four CiG reporters adopt related V-shaped structures, suggesting that their component domains likely undergo only limited inter-domain motions in solution (Fig. 4d). To explore CiG’s mobility in solution we performed size exclusion chromatography coupled to small angle X-ray scattering (SEC-SAXS) experiments (Fig. 6a). Molecular weight estimates from SAXS and multi-angle light scattering (MALS) are consistent with CiG being monomeric in solution. CiG’s scattering profile indicates that it has a radius of gyration (R_g) of 25.3 Å, which agrees well with its crystal structure (R_g values calculated using the crystal structures range from 23.3–24.3 Å) [63]. Additionally, a dimensionless Kratky plot of the scattering data is consistent with CiG adopting a relatively rigid multi-domain structure, as the peak maximum in the plot is near the theoretical coordinate for a folded protein (q*R_g = √3 with a normalized intensity of 1.1), and the scattering data approach the baseline at higher values of q*R_g (Fig. 6b) [64]. The distance distribution function, P(r), also shows the expected profile for an ordered multi-domain protein, with an initial maximum at ~23 Å due primarily to intra-domain interatomic distances and a broad second maximum at ~41 Å due to interdomain distances (Fig. 6c). The maximum dimension of the P(r) function (D_max) is estimated to be 82 Å, which is also in excellent agreement with the crystal structure (81–83 Å). Finally, direct comparison of the experimental SAXS scattering data with profiles predicted from the crystal structure are in reasonable agreement (the lowest χ² value is 2.02 when chain B is considered) (Fig. S3a) [63]. An even better fit to the SAXS data (χ² = 1.20) is obtained when the nine disordered C-terminal residues are accounted for and the positioning of the domains is allowed to move to best fit the SAXS data (Fig. 6a) [58]. This SAXS-derived structure closely resembles the crystal structures, except that it is slightly less compact (Fig. S3b). Thus, collectively the SAXS data indicate that the V-shaped domain arrangement observed in the crystal structure also predominates in solution.

Figure 6. The CiG sensor adopts a semi-rigid, V-shaped conformation in solution.

(a) CiG small angle X-ray scattering (SAXS) data. The solid line shows the fit to the scattering data using the CiG crystal structure optimized by a minimal ensemble search procedure [58]. (b) A dimensionless Kratky representation of CiG’s scattering profile. The peak maximum resides close to the intersection of q*R_g = √3 and a normalized intensity of 1.1. This is expected for a compact, folded protein. (c) The P(r) function derived from the SAXS data shows the bimodal distribution expected for a rigid two-domain protein. The D_max value obtained from this plot is consistent with the crystal structure of CiG.

3.4. The CiG reporter can be used to monitor heme transfer between CR domains

To demonstrate its utility, CiG was initially used to probe homotypic heme transfer between CR2 domains by monitoring the emission spectra of holo-CiG after adding apo-CR2. Consistent with heme transfer to the acceptor, progressive increases in CiG fluorescence emission occur upon adding apo-CR2 (Fig. 7a). A plot of the fluorescence response as function of acceptor added shows the changes saturate at high concentrations of apo-CR2 at acceptor:donor ratios consistent with the CR2 and CiG proteins having similar affinities for heme (Fig. 7b). To substantiate that CiG fluorescence changes were caused by heme transfer to apo-CR2, native MS spectra were recorded before and after mixing the reaction components (Fig. 7c, S4). The native MS data confirms that heme is transferred from holo-CiG to apo-CR2 as their co-incubation results in the formation of holo-CR2. As described in the Methods section, an analysis of mass spectra recorded at varying acceptor:donor ratios confirms that CiG and apo-CR2 have similar binding affinities for heme (K_relative = 1.04 +/− 0.08). In a separate set of experiments holo-CiG was incubated with apo-HtaB and its fluorescence emission was determined to probe for heterotypic transfer as well. Consistent with heme transfer to apo-HtaB, progressive fluorescence increases occur as apo-HtaB is added, and the changes plateau at high acceptor concentrations (Fig. 7b). When the sensor’s fluorescence emission is plotted as function of acceptor:donor ratio the apo-HtaB and apo-CR2 acceptors exhibit similar behavior, suggesting that they have similar heme binding affinities. Collectively, the data are compatible with homotypic and heterotypic heme transfer by CiG.

Figure 7. The CiG sensor can be used to detect heme transfer to other CR proteins.

(a) Fluorescence emission scans as a function of adding apo-CR2 to holo-CiG. Scans are shown after 0.0, 0.5, 1.0, 2.0, and 4.0 molar equivalents of apo-CR2 were added. Progressive increases in fluorescence are observed that are consistent with heme transfer. (b) Changes in the relative fluorescence intensity of the CiG reporter as a result of adding varying amounts of apo-CR2 or apo-HtaB. (c) Native MS spectra demonstrating heme transfer from holo-CiG to apo-CR2. The panels on the left show MS spectra (deconvolved to the mass domain) of the isolated apo-CR2 and holo-CiG proteins. The panel on the right shows the mass spectrum acquired 30 minutes after mixing.

4. Discussion

C. diphtheriae and other microbial pathogens scavenge heme using CR domain-containing hemoproteins [24]. At least five proteins in C. diphtheriae containing either one or two heme-binding CR domains are located in the cell wall, and in some instances, also secreted [24, 25]. How they collectively promote heme uptake is not well understood, but it presumably involves the transfer of heme between the proteins. It has been challenging to easily measure heme transfer between C. diphtheriae’s CR domains because they have similar UV/Vis spectra (Fig. 2b). To overcome this limitation, we developed the CiG heme binding sensor in which the CR2 domain from the HtaA protein is inserted into a surface exposed turn in EGFP (the turn connecting strands β2 to β3) (Fig. 2a). In this report we have shown that the CiG sensor exhibits heme-dependent fluorescence changes (Fig. 3c) that can be used to monitor either homotypic (donation to CR2) or heterotypic (donation to HtaB) heme transfer (Fig. 7). A 2.45 Å crystal structure and SAXS data reveal that CiG adopts a semi-rigid structure in which its CR2 and EGFP domains adopt a V-shaped conformation. The angle defining the positioning of the domains in the four protein molecules within asymmetric unit varies by only ~16°, consistent with SAXS analyses that show that CiG exhibits limited inter-domain mobility in solution. Heme binding to CiG quenches ~60% of its fluorescence as judged by the relative quantum yields of holo-CiG and a Y60F variant that is incapable of binding to heme due to mutation of the CR2’s axial tyrosine residue (Y361 in HtaA) (Table 1). These fluorescence changes are consistent with the atomic structure, since the measured degree of quenching predicts that the chromophores are separated by ~40 Å and in the crystal structure the distance between the heme iron atom within CR2 and the closest edge of the EGFP chromophore ranges from 41.0–43.9 Å in the four receptor molecules that form the asymmetric unit.

In addition to CiG, several other heme binding sensors have been developed which use RET quenching of GFP fluorescence as a primary readout [65]. To the best of our knowledge, the first of these sensors was a head-to-tail fusion of EGFP with the E. coli cyt b₅₆₂ protein, which showed an intermediary degree of RET quenching presumably because of its long interchromophore distance of 41 Å and interdomain mobility [34]. This construct was iterated upon by Arpino et al. who inserted cyt b₅₆₂ into EGFP’s β2 to β3 turn, resulting in a much shorter interchromophore distance of 17 Å, a defined interdomain orientation, and complete quenching [35]. This chromophore pair has been further exploited by adding a second fluorescent protein as an internal control for studies in cells [60], as well as to show that disulfide chemistry and other biorthogonal approaches could be used to report on heme binding [66, 67]. Other systems have been developed that rely on FRET between a pair of fluorescent proteins with fused heme-binding domains [68]. The CiG design exploits the intra-domain fusion approach developed by Arpino et al. and also yielded a similar V-shaped conformation to that documented in their work (Fig. 8). Thus, connection to this site within EGFP appears to be ideally suited for creating semi-rigid fusion structures that may be useful for constructing other types of sensors.

Figure 8. Comparison of CiG to the CG6 intra-domain fusion proteins.

The V-shaped interdomain arrangement observed in the structure of CiG (left) was also observed previously in the structure of CG6 that contains the E. coli cyt b₅₆₂ fused to EGFP (right). In both designed proteins a hemoprotein is fused to the β2 to β3 turn in EGFP. In CG6 the cyt b₅₆₂ heme pocket is positioned close to the interdomain interface resulting in complete fluorescence quenching upon heme binding.

The crystal structure of the CiG reporter is the first experimentally determined structure of a CR domain from C. diphtheriae (Fig. 5a). However, it shares significant structural similarity with the recently reported structure of the HtaA-C CR domain from C. glutamicum, consistent with the proteins sharing 56% sequence identity [33]. A comparison of their structures reveals that they both bind heme using an axial tyrosine residue located in their α1 helix, whose negative charge is presumably stabilized by accepting a hydrogen bond from a nearby histidine side chain. In both proteins, additional hydrogen bonding and aromatic π-mediated interactions are present between the protein and heme ligand which shield it from the solvent. However, many of the heme-interacting residues present in both CR2 and its C. glutamicum homolog are not conserved in other CR domains. Notably, the CR domains from C. diphtheriae’s ChtA and ChtC proteins do not contain a histidine residue that can donate a hydrogen bond to the axial tyrosine, but rather employ another tyrosine. This raises intriguing possibilities about their relative affinities for heme and potential to receive and donate heme. Interestingly, transfer measurements using the reporter suggest that heme can be transferred between CR domains via a direct mechanism in which they associate with one another. This is because ~5 minutes after mixing the spectral changes associated with transfer are complete, which is much faster than the rate at which CiG is expected to release heme into the solvent (Fig. S5). Heme binding NEAT domains found in S. aureus and related bacteria rapidly transfer heme by associating with one another via a handclasp mechanism in which the heme pockets of the donor and acceptor are juxtaposed and related by pseudo two-fold symmetry [12, 69]. Even though CR and NEAT domains are structurally unrelated, an inspection of the packing interactions present within the CiG asymmetric unit suggest that CR2 domains can also associate with their heme-binding pockets facing each other via an interface that possesses two-fold symmetry (Fig. 5d). This packing arrangement was also observed in the previously reported structures of the HtaA-N and HtaA-C proteins from C. glutamicum, suggesting that these CR domains could also transfer heme via a handclasp-like complex [33]. However, alternative models of the transfer mechanism have also been proposed that require the β1-α1-β2 regions in the donor and acceptor proteins to switch positions in order to affect transfer. Future studies will seek to distinguish between these mechanisms by using the newly developed CiG reporter to measure the kinetics of heme transfer between components of the heme-uptake network in C. diphtheriae.

In conclusion, we have developed a new fluorescence reporter that enables facile monitoring of heme binding to the C-terminal CR domain within the HtaA protein from C. diphtheriae (CR2). The CiG reporter leverages the intra-domain fusion approach pioneered by Arpino et al., which coupled with detailed structural studies confirm that the β2-β3 turn in EGFP is generally amenable to insertion for the creation of semi-rigid, V-shaped fusions. Using the reporter, we show that that CR2 domain from HtaA can rapidly transfer heme to other CR2 domains (homotypic) or to proteins containing other types CR domains (heterotypic). Future studies will exploit the reporter’s robust optical properties to explore heme uptake by pathogenic C. diphtheriae and other species of bacteria. An understanding of the molecular details of heme acquisition may lead to therapeutics to treat bacterial infections that work by depriving pathogens of this essential nutrient.