Hemophore-like proteins of the HmuY family in the oral and gut microbiome: unraveling the mystery of their evolution

SUMMARY

Heme (iron protoporphyrin IX, FePPIX) is the main source of iron and PPIX for host-associated pathogenic bacteria, including members of the Bacteroidota (formerly Bacteroidetes) phylum. Porphyromonas gingivalis, a keystone oral pathogen, uses a unique heme uptake (Hmu) system, comprising a hemophore-like protein, designated as the first member of the novel HmuY family. Compared to classical, secreted hemophores utilized by Gram-negative bacteria or near-iron transporter domain-based hemophores utilized by Gram-positive bacteria, the HmuY family comprises structurally similar proteins that have undergone diversification during evolution. The best characterized are P. gingivalis HmuY and its homologs from Tannerella forsythia (Tfo), Prevotella intermedia (PinO and PinA), Bacteroides vulgatus (Bvu), and Bacteroides fragilis (BfrA, BfrB, and BfrC). In contrast to the two histidine residues coordinating heme iron in P. gingivalis HmuY, Tfo, PinO, PinA, Bvu, and BfrA preferentially use two methionine residues. Interestingly, BfrB, despite conserved methionine residue, binds the PPIX ring without iron coordination. BfrC binds neither heme nor PPIX in keeping with the lack of conserved histidine or methionine residues used by other members of the HmuY family. HmuY competes for heme binding and heme sequestration from host hemoproteins with other members of the HmuY family to increase P. gingivalis competitiveness. The participation of HmuY in the host immune response confirms its relevance in relation to the survival of P. gingivalis and its ability to induce dysbiosis not only in the oral microbiome but also in the gut microbiome or other host niches, leading to local injuries and involvement in comorbidities.

INTRODUCTION

Oral and gut microbiome in health and disease

A healthy oral microbiome is inhabited mainly by aerobic Gram-positive bacteria, with dominating species of Streptococcus, which form a complex consortium with other bacteria (Fig. 1) (1–8). The development of periodontal diseases is caused by a shift in the composition of the oral microbial consortium, leading to the dominance of facultative anaerobic Gram-negative pathogenic bacteria over aerobic Gram-positive commensal bacteria (1, 9–14). The most abundant bacterial species identified in subgingival sites and associated with clinical features of periodontitis belong to the “red complex,” comprising Porphyromonas gingivalis, Tannerella forsythia, and Treponema denticola (1, 12). Among them, higher numbers of late colonizers, mainly P. gingivalis and T. forsythia, are found in patients with advanced periodontitis (15–17). Other bacteria, mainly Prevotella intermedia, classified in the “orange complex,” serve as early colonizers of dental plaque and act as bridging species with members of the “red complex” (18–21). Although P. gingivalis is an anaerobic bacterium, it can also survive in oral sites exposed to aerobic conditions (7), often co-localizing with very early colonizers, such as Streptococcus gordonii (4–7, 18, 22). The presence of intact P. gingivalis cells, their outer-membrane vesicles (OMVs), or DNA in non-oral sites, including synovial fluid, blood, atherosclerotic plaques, and the brain, suggests that the bacterium or its components may be involved not only in local injuries but also in systemic inflammatory diseases, such as diabetes, atherosclerosis, rheumatoid arthritis, respiratory and cardiovascular diseases, and Alzheimer’s disease (23–36).

Fig 1.

Heme sources for oral and gut microbiome members. In the healthy oral cavity or in the early stage of periodontitis, the main heme source is heme (Hm) complexed to serum albumin (HSA) and hemopexin (Hpx), both hemoproteins provided by gingival crevicular fluid (GCF) (A). In the advanced form of periodontitis, the main heme sources are hemoglobin (Hb) released from erythrocytes and subjected to metHb generation performed by P. gingivalis RgpA and P. intermedia InpA activity or H₂O₂ produced by S. gordonii, as well as HSA and Hpx present in GCF, and heme bound to proteins of cohabitating bacteria (B and C). In a healthy gut, the main heme sources are dietary compounds and bacterial proteins (D). Additional sources of heme, such as Hb, may occur with the development of pathogenic bacteria and inflammation (E). P. gingivalis, as a late colonizer and gut invader, may use metHb and bacterial hemoproteins as a heme source (C and F), which gives this bacterium a competitive advantage over other cohabitating bacteria found in the oral cavity and the gut. Sg, S. gordonii; Pi, P. intermedia; Tf, T. forsythia; Td, Treponema denticola; Aa, Aggregatibacter actinomycetemcomitans; Fn, Fusobacterium nucleatum; Pg, P. gingivalis; Ec, Escherichia coli; Bt, Bacteroides thetaiotaomicron; Bv, B. vulgatus; Bf, B. fragilis; B, bacteria belonging to the Bacillota phylum. GAPDH, glyceraldehyde 3-phosphate dehydrogenase; InpA, interpain A; Kgp, RgpA, and RgpB, gingipains; and PinA, PinO, Tfo, HmuY, Bvu, and BfrA, hemophore-like proteins.

The gut microbiome is composed mainly of the members of Bacillota (formerly Firmicutes) (~75%) and Bacteroidota (formerly Bacteroidetes) (~25%) phyla (Fig. 1) (37–42). The gut Bacteroides species can be both commensal and opportunistic pathogens, with Bacteroides fragilis as the best example (41–46). Also, in the case of the gut microbiome, changes in the composition of the microbial consortium lead to dysbiosis, resulting in the development of inflammation-based diseases (47, 48). Disruption of the surface of the intestinal mucosa, caused by inflammation or surgery, followed by the movement of B. fragilis into the blood and surrounding tissues, causes serious infections, formation of abscesses, septic shock, and often organ failure (38, 47, 48). Several recent reports showed that, as a result of mouth-gut bacterial transmission, periodontopathogens invade the gut, influence host health, and participate in non-oral diseases (49–54).

The high microbial density in the oral cavity and the gut results in nutritional competition among microbiota. One of the important nutrients for host-associated pathogenic bacteria is heme (iron protoporphyrin IX, FePPX), used as the main source of iron and PPIX. In addition, because members of the Bacteroidota phylum cannot synthesize precursors of PPIX, heme serves them as an essential nutrient and key virulence factor (55–59).

Hemophore-based heme acquisition mechanisms

Hemophores utilized by Gram-negative and Gram-positive bacteria

Any heme released into the circulation becomes tightly bound to host heme-scavenging proteins and, therefore, is not easily available for pathogenic bacteria. For this reason, bacteria have evolved a variety of advanced heme acquisition mechanisms. Although they can acquire heme utilizing direct heme transport through the outer membrane carried out by TonB-dependent outer-membrane receptors (TDRs), hemophores facilitate this process [reviewed in references (60–64)]. In Gram-negative bacteria, the best-characterized heme acquisition systems relying on a hemophore are those of Serratia marcescens and Pseudomonas aeruginosa (heme assimilation Has system, HasADEB) [reviewed in references (60–64)]. HasA hemophores are secreted proteins that bind free heme or sequester heme from host hemoproteins directly and then deliver it to TDRs. The typical HasA protein structure is arranged as a unique α/β fold, composed of seven β-strands and four α-helices (Fig. 2) (65–68). The heme-binding pocket is formed by two loops comprising conserved heme iron coordinating His32 and Tyr75 residues, and the supportive His83 residue (65–69). In contrast to HasA from S. marcescens or P. aeruginosa, HasA from Yersinia pestis or Yersinia pseudotuberculosis does not possess the conserved histidine residue, and heme iron coordination occurs with the involvement of the conserved tyrosine residue only (Fig. 2) (70–72). As a result, apo- and holo-HasA do not significantly differ in their tertiary protein structures, which contrasts with a large conformational rearrangement of the region comprising the loop with His32 in other HasA proteins (Fig. 2) (67, 68, 70, 73).

Fig 2.

Comparison of overall structures of representative hemophores produced by Gram-negative (classical hemophores) and Gram-positive [near-iron transporter (NEAT) domain-based hemophores] bacteria. Structures of proteins have been deposited under IDs: S. marcescens apo-HasA (PDB ID: 1YBJ) and HasA-heme (PDB ID: 1DK0), P. aeruginosa apo-HasA (PDB ID: 3MOK) and HasA-heme (PDB ID: 5IQX), Y. pestis apo-HasA (PDB ID: 4JER) and HasA-heme (PDB ID: 4JET), NEAT domain (amino acid residues 341–459) of Staphylococcus aureus apo-IsdB (AlphaFold ID: AF-Q7A656-F1) and IsdB-heme (PDB ID: 3RTL), P. gingivalis apo-HusA (PDB ID: 6CRL), and Bacillus anthracis apo-IsdX1 (PDB ID: 3SZ6) and IsdX1-heme (PDB ID: 3SIK). Heme-binding residues and heme molecules are shown as sticks, and the Fe ion as an orange sphere. The protein structure of apo-IsdB was modeled with AlphaFold (https://alphafold.com) (74, 75). Protein structures were visualized with UCSF Chimera (https://www.cgl.ucsf.edu/chimera/) (76).

In Gram-positive bacteria, the most well-characterized hemophore-dependent heme acquisition system is that of Staphylococcus aureus (iron-regulated surface determinants, IsdABCDEF) [reviewed in references (60–64)]. Components of this system comprise near-iron transporter (NEAT) domains (Fig. 2), of which IsdB is responsible for the binding of free heme or its sequestration from hemoglobin (Hb) (77). Also in Bacillus anthracis, although slightly different in the composition of this system, IsdX1 and IsdX2 hemophores possess one and five NEAT domains, respectively (Fig. 2) (78–80). NEAT domains are formed by eight β-strands, which resemble an immunoglobulin-like fold, and coordinate heme iron using a conserved tyrosine residue (Fig. 2) [reviewed in references (62, 81, 82)].

Hemophore-like proteins utilized by members of the Bacteroidota phylum

The black-pigmented anaerobic Porphyromonas and Prevotella species are able to accumulate Fe(III)PPIX-containing deposits on the bacterial cell surface when they grow on blood agar plates. The main component of the P. gingivalis pigment is the Fe(III) μ-oxo bisheme, which has a green-black color (83), while that of P. intermedia is composed of the monomeric Fe(III)PPIX, which has a brown-black color (84). While these pigments are formed and bound on the bacterial cell surface when Hb is abundant, these species are also capable of binding free μ-oxo bisheme and monomeric ferrihemes from solution. Bound in this manner, the ferrihemes are able to protect the bacterial cells against attack by hydrogen peroxide (H₂O₂) (85) due to their inherent “catalase” activity. However, when bacterial cells become depleted of heme in vivo, free heme is not available due to rapid sequestration by host serum heme-scavenging proteins. This poses a major problem for bacteria with a nutritional reliance upon heme. To overcome this, heme may be derived directly from Hb through proteolytic activity, a capability that has been well documented for both P. gingivalis and P. intermedia, but heme delivery is significantly increased by the engagement of hemophore-like proteins.

To facilitate heme sequestration, anaerobic P. gingivalis and P. intermedia have developed a novel heme acquisition mechanism. In this process, where oxyhemoglobin (oxyHb) is first oxidized to methemoglobin (metHb) (Fig. 1 and 3) (20, 86–88), heme is more easily released. This is possible because the Fe(III)PPIX-binding affinity in the metHb form is lower as compared to the Fe(II)PPIX-binding affinity in the oxyHb form (89). In the case of P. gingivalis, the generation of metHb mainly involves the action of the arginine-specific cysteine protease gingipain A (RgpA) (86), and in the case of P. intermedia, the cysteine protease interpain A (InpA) (20). The bacteria then degrade the more susceptible metHb to release free heme, which is performed mainly by the P. gingivalis lysine-specific cysteine protease gingipain K (Kgp) (87, 88) and P. intermedia InpA (20), or by utilizing proteins with hemophore-like properties to sequester heme directly from metHb (Fig. 1 and 3) (86).

Fig 3.

Schematic presentation of the P. gingivalis heme acquisition (Hmu) system. To release heme, gingipains (Kgp, RgpA, and RgpB) degrade Hb derived from erythrocytes, serum albumin-heme (HSA) and serum hemopexin-heme (Hpx) present in gingival crevicular fluid (GCF) (A). oxyHb is oxidized to metHb mainly by RgpA (B), leading to a decrease in heme binding strength and higher susceptibility to degradation performed by gingipains. Kgp releases HmuY from the bacterial cell surface through limited proteolytic processing (C). HmuY (both soluble and membrane-associated form) binds heme released from host hemoproteins (D) or sequesters heme directly from metHb, Hpx, and HSA (E). HmuY transfers heme to the outer-membrane receptor HmuR (F), which transports heme into the periplasmic space (PS). OM, outer membrane and IM, inner membrane. Protein structures deposited in the RCSB PDB database: HmuY (PDB ID: 3H8T), Hb (PDB ID: 1VWT), HSA (PDB ID: 1N5U), and Hpx (PDB ID: 1QJS). Other protein structures were modeled with AlphaFold (https://alphafold.com) (74, 75): HmuR (UniProt ID: Q7MUG9), catalytic domains of Kgp (UniProt ID: Q51817), RgpB (UniProt ID: P95493), and RgpA (UniProt ID: B2RM93). All structures were visualized with the Swiss-PDB Viewer (https://spdbv.unil.ch) (90).

Compared to other periodontitis-associated anaerobic bacteria, P. gingivalis has the ability to temporarily survive under aerobic conditions. This feature is supported, at least in part, by the observation that P. gingivalis and S. gordonii co-localize in regions of dental plaque exposed to oxygen (7). Co-aggregation between both species and biofilm formation, which supports colonization of P. gingivalis (7, 91–95), employs not only the major (FimA) and minor (Mfa1) fimbrial proteins of P. gingivalis and outer membrane adhesions (SspA and SspB) of S. gordonii (96–98) but also a hemophore-like HmuY protein (22), which we describe in detail below. In addition, H₂O₂ produced by S. gordonii not only causes hemolysis (99) but also converts oxyHb into metHb (100), which increases heme availability as described above. It is worth mentioning here that the formation of the HmuY-heme complex is possible in an environment enriched in H₂O₂ produced by S. gordonii (100) or by neutrophils (101). Increased expression of the hmuY gene during in vitro co-cultures of P. gingivalis with S. gordonii (22) and in vivo in the oral cavity of periodontitis patients (14) suggests the importance of this protein for efficient heme uptake. The heme acquisition HmuY-based mechanism described above is also important in co-infections with other bacteria. P. gingivalis is able to efficiently acquire heme in other niches of the host using similar synergistic mechanisms. One such example is pyocyanin produced by P. aeruginosa, which facilitates heme utilization by Hb oxidation and subsequent heme sequestration by HmuY (32).

Bacteria from the Bacteroidota phylum use a different type of hemophore. The most thoroughly defined heme acquisition (Hmu) system is that utilized by P. gingivalis. The first stage of heme uptake is heme binding by HmuY, the first representative of a novel family, and therefore termed by us a hemophore-like protein. The HmuY protein binds free heme or sequesters heme directly from host hemoproteins, which facilitates heme delivery to a HmuR (Fig. 3) (56, 102), a classical TDR, which transports heme through the outer membrane (102–104). HmuR is composed of 22 anti-parallel β-strands forming a classical β-barrel structure, closed by a plug domain, and of periplasmic and extracellular loops (56). Although the three-dimensional structure of P. gingivalis HmuR has not been solved, the importance of His95 and His434 residues has been experimentally confirmed (105). These heme ligands, located in the plug domain and in one of the extracellular loops, respectively, take part in heme reception from HmuY and subsequent heme transfer through the outer membrane. Other conserved amino acid motifs and glutamic acid residues of HmuR support this process (105, 106).

The function of HmuY is quite similar to classical hemophores produced by Gram-negative bacteria. However, unlike secreted HasA proteins, HmuY is initially anchored with the bacterial cell surface and OMVs and is then released as a soluble protein after limited proteolytic cleavage (Fig. 3) (107, 108). Significant differences also exist in their amino acid sequences, tertiary structures, heme/PPIX binding ligands, and heme iron coordination modes (Fig. 2, 4, and 5) (60, 61, 65–69, 72, 73, 102).

Fig 4.

Comparison of overall structures of hemophore-like proteins characterized in periodontopathogens. Structures of proteins determined by crystallography have been deposited under PDB IDs: P. gingivalis apo-HmuY (6EWM), P. gingivalis HmuY-heme (3H8T), T. forsythia apo-Tfo (6EU8), and P. intermedia apo-PinO (6R2H). The protein structure of P. intermedia apo-PinA was predicted with AlphaFold with very high (predicted local distance difference test, pLDDT > 90) confidence (AF-A0A1P8JJ12-F1) (74, 75). Heme iron coordinating residues and heme molecule are shown as sticks, and the Fe ion as an orange sphere. Protein structures were visualized with PyMol 2.5 (The PyMOL Molecular Graphics System, version 2.5 Schrodinger, LLC; http://www.pymol.org/pymol).

Fig 5.

Amino acid sequence comparison of HmuY family members belonging to the Bacteroidota phylum. Amino acids of HmuY engaged in heme iron coordination, as confirmed experimentally by site-directed mutagenesis and crystallography, are shaded in pink, while those engaged in heme iron coordination in HmuY homologs and confirmed by site-directed mutagenesis only are shaded in green. Consensus amino acid sequences are shown as black bars. Blue arrows indicate the regions with the highest amino acid similarity, located in the core of the protein structures. Pg, P. gingivalis; Tf, T. forsythia, Pi, P. intermedia; Bv, B. vulgatus; and Bf, B. fragilis.

Phylogenetic analysis has revealed three groups of HmuY homologs, with the majority of them identified in the Bacteroidota phylum (109, 110). So far, the proteins studied are from P. gingivalis, T. forsythia, P. intermedia, B. vulgatus, and B. fragilis (102, 107–112). Also, distant relatives of the oral and gut species produce proteins belonging to the HmuY family, including those of Elizabethkingia anopheles, a human pathogen connected with life-threatening healthcare-associated infections (113), the fish pathogens Flavobacterium psychrophilum (114) and Flavobacterium columnare (115), and that of Riemerella anatipestifer (116), which is a pathogen of ducks.

HmuY FAMILY OF HEMOPHORE-LIKE PROTEINS

Genetic organization

Typical heme uptake systems utilized by Gram-negative bacteria are composed of TDR, periplasmic binding protein, and ABC transporter, the latter located in the inner membrane (117). The best example is the Hem system (HemRTUV) of Yersinia enterocolitica (118, 119). Members of the Bacteroidota phylum are different in the genetic organization of genes encoding proteins engaged in hemophore-based heme uptake. They form the P. gingivalis hmu operon, hmu-like operons in other periodontopathogens, and hmu-like gene clusters in Bacteroides species, all being potential counterparts of classical hem operons of other Gram-negative bacteria (117, 118) (Fig. 6).

Fig 6.

Organization of genes encoding HmuY protein in P. gingivalis and its homologs in other Bacteroidota members. Genes encoding HmuY family proteins are shown in blue. Genes encoding TonB-dependent outer membrane receptors are shown in black. Genes present in the hmu operon (P. gingivalis), hmu-like operons (T. forsythia and P. intermedia), and hmu-like gene clusters (B. vulgatus and B. fragilis) are shown in white. HmuR, TonB-dependent outer-membrane receptor; HmuS, cobaltochelatase subunit CobN; HmuT, putative ATPase; HmuU, MotA/TolQ/ExbB proton channel family protein; and HmuV, putative protein (DUF2149 domain-containing protein). Additional genes are shown in other colors.

Next to the gene encoding HmuY in P. gingivalis (locus ID in A7436 strain: PGA7_RS02055), a gene encoding TDR (HmuR; locus ID in A7436 strain: PGA7_RS02050) is located. In the T. forsythia hmu-like operon, a duplication results in an additional HmuR homolog (Fig. 6). The first TDR (locus ID in ATCC 43037 strain: Tanf_RS09475) lacks the two histidine residues involved in heme iron coordination in HmuR and in other typical heme TDRs (109, 111). Instead, methionine residues are located in homologous positions. The second gene encoding an HmuR homolog (locus ID in ATCC 43037 strain: Tanf_RS09470) is more like the P. gingivalis HmuR.

Similar to P. gingivalis, T. forsythia expresses one hemophore-like protein, termed Tfo (locus ID in ATCC 43037 strain: Tanf_RS09480) (Fig. 6) (109). Interestingly, the presence of two homologous genes in P. intermedia (locus ID in chromosomes I and II in strain 17: PIN17_RS00035 and PIN17_RS05355), encoding PinO and PinA proteins, respectively (Fig. 6) (110), suggests some functional differentiation in their products. In addition, PinA is encoded on the large chromosome II singly, without a gene encoding TDR or other adjacent genes. Like P. gingivalis, B. vulgatus possesses one HmuY homolog, termed Bvu (locus ID in ATCC8482 strain: BVU_RS11035) (Fig. 6) (112). Surprisingly, B. fragilis expresses three HmuY homologs (locus ID in NCTC 9343 strain: BF9343_RS12925, BF9343_RS04845, and BF9343_RS10155), termed BfrA, BfrB, and BfrC, respectively (Fig. 6) (111), which points to an even more diverse function of their products.

The functions of the other proteins encoded on the P. gingivalis hmu operon are unknown, but they are presumably involved in heme transport in the periplasmic space and into the cytoplasm and/or in heme metabolism (Fig. 6). The third gene in this operon encodes a protein with putative CobN/Mg chelatase/dechelatase function (HmuS) (55–57, 120–122), homologous to B. fragilis BtuS2 protein (55). The next two genes encode proteins with homology to MotA/TolQ/ExbB (HmuT) and permeases (HmuU). The last gene encodes a protein with no homology to any known protein family (HmuV). T. forsythia and P. intermedia possess a similar composition of this locus (Fig. 6).

A slightly more complicated organization can be seen in hmu-like gene clusters in Bacteroides species (Fig. 6). A gene encoding a calycin-like domain-containing protein precedes genes encoding HmuY homologs, i.e., Bvu and BfrA. In other bacteria, this β-barrel outer membrane protein binds small hydrophobic molecules (123, 124). In the hmu-like gene cluster in the B. fragilis 638R strain, an additional gene encoding anti-bacterial toxin (BSAP4) with a membrane attack complex/perforin domain (MACPF) is located. MACPF domain proteins belong to a family of antibacterial toxins of gut Bacteroides, which mediate competition between Bacteroides species (123–125).

Genes encoding BfrB or BfrC are organized as smaller operons, composed of genes encoding TDR, HmuY homolog, and PepSY (peptidase of M4 and Subtilis YpeB protein) domain-containing protein or DUF4903 domain-containing protein, respectively (Fig. 6). The PepSY domain is found in a diverse family of secreted and cell-associated proteins with unknown functions, which are used to regulate protease activity (126). In general, the relationship(s) between those genes, present in many members of the Bacteroidota phylum and typical P. gingivalis hmu operon genes, is an area that needs to be investigated.

Protein expression and localization

P. gingivalis HmuY is expressed constitutively, but its expression significantly increases when bacteria are starved of iron and heme in liquid culture media. Importantly, HmuY production increases when bacteria grow as a homotypic biofilm constituent and when bacteria remain inside host cells (108, 122, 127–129). Interestingly, in patients with periodontitis, HmuY expression is higher and correlates with high expression of P. gingivalis gingipains and P. intermedia InpA (130). This suggests an important synergy between the HmuY family proteins and proteases in heme acquisition (131, 132). The requirement of HmuY for efficient heme acquisition is supported by its significantly higher production when P. gingivalis forms a biofilm together with Candida albicans (133), when P. gingivalis is co-cultured with S. gordonii (22), and when P. gingivalis is a constituent of dental plaque formed in patients with periodontitis (14, 130).

Also, higher expression of proteins homologous to HmuY is observed when T. forsythia, P. intermedia, B. vulgatus, and B. fragilis grow in vitro in an environment characterized by limited availability of iron and heme (107–112). The expression of bfrA and bfrB genes and their adjacent TDRs is significantly downregulated in the B. fragilis metronidazole-resistant 638R strain as compared to the metronidazole-susceptible 638R strain, both lacking the nimA gene and cultured in the medium lacking heme (134, 135). Surprisingly, this is highly visible in the case of the bfrB gene, presumably not involved in heme but rather in PPIX uptake, as we indicate below (111). Therefore, bfr and hmuY genes are good candidates for further studies to resolve mechanisms of metronidazole resistance in anaerobic Bacteroidota in relation to heme acquisition.

OMVs are small, spherical membranous structures formed via vesiculation of the outer membrane of Gram-negative bacteria. They serve as an important vehicle transporting virulence factors and, therefore, contribute to nutrient delivery, bacterial co-aggregation, biofilm formation, host cell invasion, and efficient evasion of host immune response. Importantly, OMVs are enriched in crucial virulence factors, including HmuY family proteins (108–112, 136–145). This way of delivering virulence factors to many niches of the host is more effective as compared to whole bacterial cells and increases the heme acquisition ability of P. gingivalis. In general, the distribution pattern of HmuY homologs between whole bacterial cells, OMVs, and culture medium containing soluble proteins is similar to that for the P. gingivalis HmuY (108–112, 136–145).

Another important virulence feature of P. gingivalis HmuY is its resistance to proteolytic degradation by a variety of enzymes, including P. gingivalis gingipains (108), P. intermedia InpA (20, 109–112), and enzymes released by host immune cells (e.g., neutrophil elastase) (32). In contrast, HmuY homologs produced by other bacteria are sensitive to proteolysis (108–112). Therefore, the extremely high proteolytic activity of P. gingivalis and the simultaneous resistance of HmuY to proteolytic degradation serve as an advantage for P. gingivalis over cohabitating bacteria, allowing the former to better survive in heme-limited host environments.

Regulation of gene expression

For most pathogenic bacteria, iron is required at a concentration that is not available in the host (146). Free heme is also limited in the host, mostly due to the high sequestration activities of host heme-scavenging proteins (147–150). On the other hand, an excess of iron and heme is toxic due to the generation of reactive oxygen species (151). To avoid these consequences, cellular iron and heme homeostasis is tightly regulated in the bacterial cell (64, 152, 153). Although the mechanisms regulating the expression of genes encoding the HmuY family proteins are still poorly understood, it is quite well documented, mainly in P. gingivalis, that their expression is regulated directly or indirectly by iron/heme availability (26, 104, 109–112, 154, 155).

In the majority of bacteria, genes are controlled by proteins belonging to the FUR (ferric uptake regulator) superfamily, which respond to the iron concentration (156, 157). The best-characterized member of this family, Fur protein, usually functions as a global regulator. Using a variety of mechanisms, it controls the expression of genes engaged in iron uptake in many pathogenic bacteria but also regulates the expression of genes whose products participate in stress response and virulence potential (158). Although Fur homologs exist in the members of the Bacteroidota phylum (159), so far only the Fur protein from P. gingivalis (PgFur) has been characterized (Table 1) (155, 160–162).

TABLE 1.

Proteins involved in the regulation of P. gingivalis hmuY gene expression and their homologs in other Bacteroidota members^a

P. gingivalis A7436 (NCBI accession number)	Regulator description and function in the hmu operon regulation	References	%identity in relation to P. gingivalis proteins (NCBI accession number)
			P. intermedia 17	T. forsythia ATCC 43037	B. fragilis 638R	B. vulgatus ATCC 8482
PgFur (QDE53638)	Ferric uptake regulator—regulation of the hmu operon under iron/heme excess conditions	(155, 159, 160)	24.54 (WP_028906176)	24.84 (WP_014224877)	24.22 (WP_005789793)	29.30 (WP_005841116)
CdhR (AKV63980)	Transcription activator protein/response regulator, containing a CheY-like receiver domain and an HTH DNA-binding domain— regulation of the hmu operon in response to growth phase	(163–165)	30.22 (WP_014710295)	30.26 (WP_014225829)	44.16 (WP_005788304)	58.91 (WP_005843420)
PG1236 (AKV63981)	Nitric oxide-induced protein (hemerythrin domain-containing protein)—regulation of the hmuY gene	(165)	41.23 (WP_014710296)	33.62 (WP_014225828)	frameshift	48.03 (WP_005843418)
TrkA (AKV65246)	Potassium transport regulatory protein/NAD-binding component—regulation of cdhR and hmuY genes	(166)	43.60 (WP_014709064)	52.91 (WP_014223971)	49.21 (WP_005785058)	52.36 (WP_005842854)
SigCH (AKV63443)	Extracytoplasmic function sigma factor/sigma-70 family—regulation of the hmu operon	(167)	37.42 (WP_014708340)	38.27 (WP_014224343)	34.94 (WP_005786485)	32.53 (ABR41229)
SigH (AKV64889)	Extracytoplasmic function sigma factor/sigma-70 family—regulation of the hmu operon	(168)	ND	65.87 (WP_041590746)	50.90 (WP_122139738)	50.30 (WP_117853272)
HaeR (AKV64439), HaeS (AKV64440)	Two-component system, comprising the response regulator (HaeR)/response regulator with CheY-like receiver domain and winged-helix DNA-binding domain and sensor histidine kinase (HaeS)—regulation of the hmu operon	(169)	29.49 (WP_014710097) 23.79 (WP_014710098)	61.06 (WP_124752021) 44.47 (WP_124790238)	65.33 (WP_014299062) 45.41 (WP_005788537)	63.11 (WP_178284964) 44.71 (WP_005846805)
PgRsp (AKV63644)	Redox-sensing protein—regulation of the hmu operon in response to the redox state	(170)	ND	ND	ND	ND
PG_0686 (AKV64470)	Oxidative stress-induced protein (diguanylate cyclase)/DUF438 domain, hemerythrin domain—regulation of the hmu operon	(171)	50.20 (WP_014708659)	30.05 (WP_014225769)	49.51 (WP_041161332)	ND

^a ND, not detected.

In the absence of PgFur and in an environment rich in iron and heme, expression of the hmuY gene is increased, demonstrating that PgFur acts as a classical iron-dependent repressor. However, in contrast to other Fur proteins, PgFur does not possess a histidine-rich motif responsible for iron binding, has an atypical positively charged C terminus, and does not regulate expression of genes typically involved in iron acquisition in other Gram-negative bacteria, e.g., the feoB gene (149, 155–158). This suggests that its mechanism of action is different (155).

There is a high probability that PgFur modulates in vitro and in vivo growth of P. gingivalis mainly by affecting iron/heme acquisition mechanisms through a multi-layer regulatory network. Such complex regulation of genes expressing virulence factors in this bacterium, especially in environments limited with respect to both iron and heme, may involve not only direct regulation of some genes by PgFur (so far not precisely identified) but also indirect PgFur regulation by influencing the expression of other regulators.

Another mechanism used to regulate the expression of the hmu operon genes in P. gingivalis involves the two-component HaeSR system (Table 1), comprising a sensor histidine kinase and a response regulator (169). Phosphorylated HaeR binds directly to the hmu promoter as well as to the upstream region of the hmuS gene. It seems that regulation of expression of the hmu operon genes in P. gingivalis involves a two-layer control mechanism: one occurring with the direct involvement of the HaeSR system (169) and the other employing indirect regulation of haeSR genes by PgFur (155).

Some genes of the hmu operon are also regulated by a CdhR transcription factor (LuxR family member), which activates their expression, although the direct binding to the hmu promoter is unclear (Table 1) (163–165). Deactivation of the cdhR gene results in decreased expression of hmu operon genes, especially in the early growth phase, suggesting a correlation with the growth phase and cell density, the latter in the context of quorum-sensing mechanisms (167, 172). In addition, a protein (PG1236) encoded by the gene adjacent to the cdhR gene regulates the expression of the hmuY gene since the mutant strain exhibits significantly lower expression of this gene, especially under nitric oxide stress (165). CdhR may not only directly regulate the expression of hmu operon genes through binding to their promoter (162) but also indirectly by the PG1236 protein (165) or potassium uptake protein (TrkA) (166). This shows that nitric oxide stress or signal transduction using potassium ions can be an important part of the regulation of the hmu operon genes. Such mechanisms could be crucial for P. gingivalis entering the bloodstream or host cells.

Other regulators of the P. gingivalis comprise PgRsp, a heme-binding protein that regulates the expression of the hmuY gene depending on the redox conditions (170), and extracytoplasmic function sigma factors (SigH and SigCH), which regulate expression the of hmuY, hmuR, and cdhR genes (Table 1) (167, 168). Additionally, the deletion of a gene encoding an oxidative stress-induced protein (diguanylate cyclase, PG_0686) has an impact on the expression of the hmu operon genes (171).

Altogether, it seems that the expression of genes encoded within the P. gingivalis hmu operon varies depending on the external environment conditions and bacterial cell density. Such an assumption might be confirmed by the production of mRNAs differing in length, encompassing different regions of the hmu operon, with mRNA encoding HmuY protein at the highest abundance as compared to the hmuY-hmuR and polycistronic transcripts (122, 128). Differential gene expression could also result from the activity of the stem-loop structures present within the B₁₂-box (working based on the riboswitch mechanism), located in the hmu operon in P. gingivalis and in the hmu-like gene cluster in B. fragilis (120, 128, 173), or from the production of antisense RNAs corresponding to the P. gingivalis hmu operon genes (174, 175).

Similar to P. gingivalis HmuY, T. forsythia, P. intermedia, B. vulgatus, and B. fragilis produce proteins homologous to HmuY in response to the low availability of iron and heme (104, 109–112). These bacteria encode Fur proteins that are more similar to classic Fur proteins (Table 1), including histidine-rich motifs utilized to bind iron; nevertheless, their function remains to be elucidated. Moreover, not all homologous proteins involved in the regulation of the hmu operon genes in P. gingivalis are present in those bacteria or their degree of homology is low (Table 1). Therefore, one may assume that different bacteria have developed different strategies to regulate the production of the HmuY family proteins, allowing them to efficiently adapt to the host niches they occupy and to changing environmental conditions different in health and disease.

HEME-BINDING PROPERTIES OF HmuY FAMILY PROTEINS—STRUCTURE-FUNCTION RELATIONSHIP

Heme iron coordination

The two unique histidine residues coordinating heme iron in P. gingivalis HmuY are poorly conserved in the HmuY homologs characterized so far, the latter preferring methionine residues to coordinate heme iron (Fig. 4 and 5) (107, 109–112, 176–178). Another important difference is the finding that HmuY binds heme under both aerobic and reducing conditions (Fig. 7; Table 2) (109–112, 178). HmuY coordinates heme iron using His134 and His166 (to unify positions of amino acid residues coordinating heme iron, numbering in the full-length proteins is applied in this review) (Tables 2 and 3; Fig. 4 and 5). In contrast to HmuY, Met162 and Met191 in PinA and Met143 and Met169 in Tfo could be directly involved in heme binding (Table 2; Fig. 4 and 5) (109, 110). In PinO, however, when the heme iron is coordinated by Met150, both Met176 and Met76 could interchangeably serve as heme ligands (Table 2; Fig. 4 and 5) (110). While the heme binding to PinO and Tfo has not been confirmed by crystallography, analysis of the dynamic properties of both proteins suggests a possible means of binding and accommodation of the heme during the opening and closing of the loops that form the entrance to the heme-binding pocket (109, 110).

Fig 7.

UV-visible absorbance spectra of the P. gingivalis HmuY protein in complex with heme analyzed under aerobic (blue line) and reducing (red line) conditions. The HmuY-heme absorbance spectra were visualized with GraphPad Prism 8.0 (San Diego, CA, USA).

TABLE 2.

Summary of characteristics of heme binding to the best-characterized HmuY family members

HmuY family protein	Species	Heme iron coordinating ligandsa	Heme dissociation constant (Kd; M)		References
			Aerobic conditions	Reducing conditions
HmuY	P. gingivalis	His134, His166	<10−9	1.1 ± 0.2 × 10−8	(107, 109, 178)
Tfo	T. forsythia	Met143, Met169	4.9 ± 0.8 × 10−6	8.2 ± 2.7 × 10−9	(109, 112)
PinO	P. intermedia	Met150, Met176/Met76	9.2 ± 3.2 × 10−8	3.4 ± 1.6 × 10−9	(110)
PinA	P. intermedia	Met162, Met191	5.7 ± 0.3 × 10−6	2.0 ± 0.5 × 10−8	(110)
Bvu	B. vulgatus	Met145, Met172	ND	1.6 ± 0.6 × 10−9	(112)
BfrA	B. fragilis	Met146, Met175	1.8 ± 1.4 × 10−7	6.4 ± 0.8 × 10−8	(111)
BfrB	B. fragilis	None	1.4 ± 0.3 × 10−8	ND	(111)
BfrC	B. fragilis	None	ND	ND	(111)

^a Numbering in the full-length proteins is applied. ND, K_d not determined.

TABLE 3.

Heme-binding features of the HmuY family members revealed by UV-visible absorbance spectroscopy^a

HmuY family protein in complex with heme	Soret region (nm)		β band (nm)		α band (nm)		CT band (nm)		Additional bands (nm)		References
	Ox	Red	Ox	Red	Ox	Red	Ox	Red	Ox	Red
HmuY	411	424	529	528	559	558	ND	ND	ND	ND	(107, 109, 178)
Tfo	398	426	529	527	565	558	607	ND	ND	ND	(109, 112)
PinO	406	428	530	529	ND	560	607	ND	499	ND	(110)
PinA	406	427	530	529	ND	559	607	ND	495	ND	(110)
Bvu	401	427	532	528	ND	559	610	ND	499	ND	(112)
BfrA	401	426	529	528	ND	558	ND	ND	ND	ND	(111)
BfrB	391	414	512	538	544	571	645	ND	ND	433,446	(111)
BfrC	ND	ND	ND	ND	ND	ND	ND	ND	ND	ND	(111)

^a ND, heme-binding feature not detected. CT, charge transfer transition; Ox, aerobic conditions; Red, reducing conditions formed by the addition of sodium dithionite.

Different heme-binding modes can also be observed in HmuY homologs produced by Bacteroides species. While Bvu could bind heme using Met145 and Met172, it seems that BfrA could use Met146 and Met175 as heme ligands (Table 2; Fig. 5 and 8) (111, 112). Although this was not examined experimentally, a HmuY homolog from Bacteroides thetaiotaomicron (Bth) may use Met141 and Met168 since both residues are located in homologous positions compared to Bvu and BfrA (Fig. 5) (111).

Fig 8.

Comparison of overall structures of hemophore-like proteins characterized in Bacteroides species. Structures of proteins determined by crystallography have been deposited under PDB IDs: B. vulgatus (3U22), B. fragilis apo-BfrA (4GBS), B. fragilis apo-BfrB (8B6A), and B. fragilis apo-BfrC (8B61). Heme iron coordinating residues are shown as sticks. Protein structures were visualized with PyMol 2.5 (The PyMOL Molecular Graphics System, version 2.5 Schrodinger, LLC; http://www.pymol.org/pymol).

Heme binding to HmuY family members causes changes in their tertiary structures upon heme binding, clearly visible in the region encompassing the entrance to the heme-binding pocket (Fig. 4) (107, 109–112, 178). This is a feature very similar but only to some HasA hemophores (65–69), confirming the possibility of evolutionary adjustment of individual bacteria.

Compared to the native HmuY, its variants with His134 and His166 substituted singly by a methionine residue bind Fe(III)PPIX with a lower ability, analogously to B. vulgatus Bvu and T. forsythia Tfo (109, 111, 179). A mixed His-Met pair in HmuY allows the binding of Fe(II)PPIX (179), which is typical for T. forsythia Tfo and P. intermedia PinO and PinA (109, 110). Double substitution of histidine residues by methionine residues, however, abolishes heme binding (179). It is, therefore, possible that the unique properties of HmuY evolved through the acquisition of a unique heme-binding pocket, which allowed the acquisition of an advantage over other hemophore-like proteins in heme binding in environments undergoing changes in redox conditions. This points to a superior virulence potential of P. gingivalis over cohabitating bacteria.

Binding of non-heme porphyrins

Modified hemes, namely mesoheme (FeMPIX) and deuteroheme (FeDPIX), are bound by His134 and His166 of HmuY, similarly to the protoheme (180). Both modified hemes also support P. gingivalis growth, both in planktonic and biofilm forms, as well as its interaction with epithelial cells in a similar fashion to protoheme (57). The affinity of HmuY for the three iron porphyrins is similar (180), suggesting that the protein may also bind non-natural heme derivatives equally to heme. Therefore, it is not surprising that HmuY also binds non-iron metalloporphyrins, although in a different manner to Fe(II/III)PPIX (181): the metal ions in Ga(III)PPIX and Zn(II)PPIX bind only His166, the metal ions in Ni(II)PPIX and Co(II)PPIX bind only His134, whereas the metal ion in Mn(III)PPIX binds either His134 or His166. Although the metal ion in Co(III)PPIX binds His134 and His166, some differences are observed compared to iron binding in heme. What is important from a therapeutic perspective is the discovery that some of these metalloporphyrins, mainly Ga(III)PPIX, Co(III)PPIX, and Cu(II)PPIX, exhibit potent anti-bacterial activity against P. gingivalis (58). Non-iron metalloporphyrins can be used as a tool in an anti-bacterial “Trojan horse” alternative strategy to antibiotic treatment because they exploit heme uptake systems of Gram-negative bacteria as an entry into the cell (182). Among them, gallium seems to be the most suitable choice for this approach since it has an atomic radius almost identical to that of iron (0.62 and 0.64 Å, respectively) but has only one valence state. Therefore, the insertion of Ga(III)PPIX instead of Fe(II/III)PPIX causes the inactivation of pathways utilizing heme-containing proteins. In addition, owing to the strong ability of metalloporphyrins to absorb light in both free and protein-bound forms, they have become a target of diagnostic and therapeutic photodynamic methods (183).

HmuY also forms a complex with PPIX but with very low ability (180). Lower PPIX binding to HmuY is supported by the replacement of PPIX by heme and the subsequent formation of a HmuY-heme complex. In contrast, B. fragilis BfrB can efficiently form a complex with PPIX (K_d ~ 10⁻⁷ M) and coproporphyrin III (CPIII) (K_d ~ 10⁻⁶ M) (111). We hypothesize that BfrB could bind similar iron-free porphyrins as P. gingivalis HusA, a protein exhibiting hemophore-like properties but not belonging either to the HmuY family or to HasA hemophores (Fig. 2) (184, 185).

Three-dimensional protein structures

Structural comparison of proteins forming the HmuY family shows a novel asymmetric all-β-fold structure (Fig. 4 and 8), with a highly similar core found in all proteins (186). This structure is unique (Fig. 4 and 8) as compared to HasA hemophores, NEAT domain-based hemophores, or other less characterized hemophores (e.g., P. gingivalis HusA) (Fig. 2) (65–67, 70, 77, 78, 184, 185, 187).

Crystallographic analysis of P. gingivalis A7436 apo-HmuY (PDB ID: 6EWM, determined at 1.4 Å resolution) and HmuY-heme complex (PDB ID: 3H8T, determined at 1.8 Å resolution) demonstrated that the protein structure is composed mainly of β-strands, with both N and C termini exposed on the protein surface (Fig. 4) (107, 109). The most relevant contacts are the metallo-organic bonds formed between the heme iron and nitrogen atoms of His134 and His166. Complex interactions between amino acid residues forming the heme-binding pocket and a single Fe(III)PPIX involve not only iron coordination with two histidine ligands but also interactions between other amino acid residues and the PPIX ring. Compared to the HmuY-heme complex, apo-HmuY is characterized by the open heme-binding pocket (Fig. 4). Upon heme binding, movement of the loop containing His166 results in the closing of the entrance of the heme-binding pocket.

The three-dimensional structure of T. forsythia ATCC 43037 apo-Tfo (PDB ID: 6EU8, determined at 1.47 Å resolution) is also composed mainly of β-strands. As compared to apo-HmuY, the main difference can be seen in the heme-binding pocket. In Tfo, four shorter β-strands, instead of two long β-strands, form one side of the structure, and Tfo lacks two histidine residues coordinating heme iron in HmuY, which are replaced by methionine residues (Fig. 4) (109). Comparison with apo-HmuY suggests that the loop containing Met169 in apo-Tfo could move significantly, allowing heme entrance into the heme-binding pocket.

The overall three-dimensional structure of P. intermedia 17 apo-PinO (PDB ID: 6R2H, determined at 2.46 Å resolution) is more similar to T. forsythia apo-Tfo than to P. gingivalis apo-HmuY (Fig. 4) (110). As in apo-HmuY, the main part of apo-PinO is formed by two β-sheets. The heme-binding pocket, possessing three hairpins in PinO, is different from that of apo-HmuY, the latter containing two hairpins. Also, in the case of PinO, instead of two histidine residues, two methionine residues engaged in heme binding are present. The differences between apo-PinO and apo-Tfo are much smaller and visible mainly in the heme-binding pocket. So far, the three-dimensional structure of P. intermedia PinA is not known.

The three-dimensional structures of B. vulgatus ATCC 8482 apo-Bvu (PDB ID: 3U22, determined at 2.12 Å resolution) and B. fragilis NCTC 9343 apo-BfrA (PDB ID: 4GBS, determined at 2.75 Å resolution) (Fig. 8) were deposited in the RCSB PDB database by others in 2012 (Joint Center for Structural Genomics, JCSG). The structures of both proteins are most similar to apo-Tfo and apo-PinO as compared to apo-HmuY (Fig. 4 and 8).

The three-dimensional structures of B. fragilis NCTC 9343 apo-BfrB (PDB ID: 8B6A, determined at 1.77 Å resolution) and B. fragilis NCTC 9343 apo-BfrC (PDB ID: 8B61, determined at 1.81 Å resolution) have been solved recently. In the case of the BfrB, the protein structure is arranged by two β-sheets and three small α-helices (Fig. 8) (111). The heme-binding pocket does not contain histidine or methionine residues, but amino acid residues identical to those found in HmuY and engaged in the binding of the PPIX ring are present, confirming the ability of iron-free porphyrins binding. Similar to BfrB, the BfrC structure is formed by two β-sheets (Fig. 8) (111). While overall protein structures of HmuY, Bvu, BfrA, and BfrB are similar, the structure of BfrC is quite different, mainly due to the presence of two longer arms protruding from the core, one of them being flexible and possessing a longer helical structure. This feature could prevent heme entrance into the heme-binding pocket. In addition, only a few residues present in the heme-binding pocket of HmuY are present in BfrC, which could be another reason for the inability of heme binding. Although BfrC binds neither heme nor PPIX, we classified this protein to the HmuY family because of its high structural similarity, mainly residing in the core, typical of all proteins assigned to this family (186) (Fig. 4 and 8).

HEME SEQUESTRATION PERFORMED BY HmuY FAMILY PROTEINS

Among the Bacteroidota members, the most efficient hemophore function is exhibited by P. gingivalis HmuY, which efficiently sequesters heme from metHb (86). This is possible because in the metHb form, the affinity for Fe(III)PPIX is lower compared to the affinity for Fe(II)PPIX (89). HmuY is also able to sequester heme bound to albumin (20, 86) and hemopexin (109). In the periodontal pocket and inflamed gingival crevice, in the absence of bleeding, the main heme source is serum albumin, which is present in gingival crevicular fluid (Fig. 1 and 3) (188). Importantly, heme complexed to albumin is an important iron and PPIX source for pathogens, when they spread into host niches other than the oral cavity, where free heme is rapidly sequestered by this abundant host heme-scavenging protein.

While this is the case for P. gingivalis, the HmuY homologs produced by other periodontopathogens are not able to carry out heme sequestration when it is part of the metHb form or when complexed to hemopexin and albumin when heme is present in the two latter proteins in the Fe(III)PPIX form (109–112). HmuY homologs are only able to capture the Fe(II)PPIX from the albumin-heme complex, an observation explained by the fact that albumin has a lower affinity for Fe(II)PPIX than for Fe(III)PPIX (189, 190). In addition, the environmental reducing conditions (prevailing in deep anaerobic periodontal pockets) influence iron coordination by methionine residues (where methionyl sulfur atoms serve as ligands) more effectively than by histidine residues (where nitrogen atoms serve as ligands) leading to the destabilization of bis-Met ligand binding under aerobic conditions (191). This is caused by the fact that the coordination through sulfur atoms, being better electron acceptors, results in the rise of the redox potential (192). This is an explanation of efficient Fe(II)PPIX binding to HmuY homologs and heme sequestration from the albumin-Fe(II)PPIX complex by these proteins. Thanks to its unique structure and properties, P. gingivalis HmuY is stable and binds heme also under aerobic conditions (32, 100). It is worth noting here that an oxygen gradient exists not only in different sites of the oral cavity but also in different parts of the human gastrointestinal tract and in other niches of the host (193–197). However, being more versatile in heme binding compared to its homologs, HmuY can enable P. gingivalis to grow better and have a higher tendency to cause dysbiosis.

Many bacterial species employ “moonlighting” proteins to aid colonization and induce pathogenic processes (198–200), and among them is glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (201). Prokaryotic GAPDH is found not only in the cytosolic fraction, where it plays the canonical function, but also in the form released upon bacterial cell lysis or in a bacterial cell wall-associated form (202). Also, S. gordonii GAPDH (SgGAPDH) exhibits similar features. More importantly, while SgGAPDH is able to bind heme (22), HmuY, PinO, and Tfo are capable of sequestering Fe(III)PPIX and Fe(II)PPIX which has been complexed to SgGAPDH (22).

It is possible that both secreted and surface-associated forms of the SgGAPDH-heme complex might serve as a heme reservoir for periodontopathogens. In the early stages of periodontitis, any heme that becomes bound to SgGAPDH could be transferred to albumin, and further accessed by early colonizers, mainly by P. intermedia, and be subsequently exploited by late colonizers, mainly T. forsythia and P. gingivalis (Fig. 1). Therefore, as the disease progresses, any SgGAPDH-heme complexes could be targeted by HmuY and Tfo, which could aid colonization of these later anaerobic colonizers. Importantly, HmuY can efficiently sequester heme, which has been complexed to the other HmuY homologs under aerobic and aerobic conditions. However, under reducing conditions, PinO and Tfo could capture heme bound to HmuY, which raises the possibility of two-way heme transfer between them (109, 110). The different heme-binding properties of the hemophore-like proteins can allow the formation of a mobile reservoir of heme, common to pathogens, but used differentially by other species depending on changing environmental conditions.

HmuY IN RELATION TO INFLAMMATORY BASIS OF PERIODONTITIS AND SYSTEMIC DISEASES

HmuY and related proteins in host immune response

P. gingivalis escapes the host immune response using a variety of mechanisms (33, 127, 159, 203). The bacterium can inhabit host cells and survive in cellular compartments, including endosomes and autophagosomes, using them as a means of escaping the host’s immune response (136, 204–207). These properties enable P. gingivalis to survive in hostile environments and avoid challenge from the host, thus increasing the tendency of this opportunistic pathogen to promote inflammation, which is typical of periodontal diseases. To increase its virulence potential, P. gingivalis releases OMVs, whose production is increased in the late exponential growth phase, in restricted growth conditions, and under environmental stress, including iron and heme starvation (208). OMVs with their cargo can be released into the oral biofilm and gingival tissues and can be internalized into epithelial, endothelial, and immune cells (127, 136, 203–207). It is noteworthy that among the favored cargo of OMVs are HmuY protein and gingipains (137, 138, 209). Taking into consideration the involvement of P. gingivalis in the initiation and progression of comorbidities, OMVs can be delivered more easily than whole bacterial cells to other host niches (203, 206, 207, 209–211).

Initially, it was reported that a 24-kDa protein purified from OMVs of the P. gingivalis W50 strain could stimulate human gingival fibroblasts (212, 213). The protein, designated fibroblast activating factor, exhibited similarity to human growth factors and weak phosphatase activity, the latter potentially contributing to bone resorption (214). Later, the protein was identified by proteomic approach as HmuY (122, 215, 216). Several experiments have demonstrated that HmuY is an important virulence factor in the infection of host cells since inactivation of the hmuY gene results in diminished survival of P. gingivalis and its invasion efficiency of macrophages (127), epithelial cells (57), and gingival keratinocytes (132), both abilities being partially restored by complementation of the mutant strains with the HmuY protein (127). Since the role played by P. gingivalis HmuY in the immunopathogenesis of periodontitis has been described in the previous review in detail (217), here we present only a short overview.

Bacterial antigens, which become accessible to the host immune system after lysis of bacteria, via shedding of OMVs, or after secretion and release of soluble proteins, are identified and processed by host antigen-presenting cells, which results in stimulation of the adaptive immune response and the production of antibodies. This process is also true for P. gingivalis since significantly increased levels of IgG antibodies in sera of patients with periodontitis, compared with healthy control individuals, reacting specifically not only with P. gingivalis total antigens but also with HmuY protein have been found (186, 218). Therefore, we believe that this finding could be employed to develop biological markers of periodontitis used to assess the severity of the disease and progress of the therapeutic treatment. Although similar homologs of HmuY have been identified within the Bacteroidota members co-participating with P. gingivalis in periodontitis, the amino acid sequence of HmuY is unique enough to be used as a molecular marker (109–112, 176, 177). Indeed, in this respect, antibodies raised in rabbits against the HmuY protein are highly specific for this protein and do not recognize homologous proteins of T. forsythia and P. intermedia (177). Another promising candidate for the development of a biological marker is the P. intermedia PinA protein, which also reacts more strongly with serum IgG antibodies in patients with periodontitis (186).

Another member of the HmuY family, the homologous protein produced by E. anopheles, can also be recognized by the host immune system. This multidrug-resistant bacterium, which is responsible for several pathogenic processes (e.g., meningitis, urinary and respiratory tract infections, and septicemia), in the presence of antibiotics increases the production of OMVs, which are enriched in virulence factors, including an HmuY homolog (113). HmuY homologs from F. psychrophilum (HfpY) and F. columnare are also required to ensure host colonization and maintain virulence in rainbow trout (114, 115).

The importance of HmuY family proteins in virulence is similar as compared to that found for other hemophores. For example, a mouse model of P. aeruginosa infection revealed that the hasA gene is among the most upregulated genes during disease progression (219), and IsdB is important for adhesion to host cells, which increases the pathogenicity of S. aureus (220–222). Therefore, it is not surprising that the implication of NEAT-containing proteins of B. anthracis and S. aureus in the pathogenic process is currently being used for research, which could lead to the development of a vaccine (81), attempts similar to those being carried out for P. gingivalis HmuY (186).

The persistence of infiltrating inflammatory cells in affected sites in patients with periodontitis is responsible for the exaggerated host immune response and chronic features of periodontitis. In general, an overlapping pattern of regulated genes in model macrophages, exhibiting a generalized inflammatory response to P. gingivalis and its antigens, including HmuY, is observed (223). This leads to the activation of NF-κB-regulated genes, including stimulation not only via the Toll-like receptor 2 (TLR2) but also via the TLR7 and TLR8 pathways (223). It is likely that P. gingivalis may induce the expression of many genes from those pathways that cause macrophages to respond by maintaining infection instead of pathogen elimination. Several reports not only support the finding that P. gingivalis uses host autophagy pathways to hide and survive within host cells (204, 205) but also point to other mechanisms responsible for the evasion of the host immune response.

It is well known that P. gingivalis antigens, including HmuY, significantly stimulate the expression of pro-inflammatory cytokines and chemokines in immune cells (223, 224), allowing for increased intracellular survival of this bacterium and improved survival and chemotaxis of immune cells, resulting in chronic inflammatory conditions and a dysbiotic process in affected individuals.

HmuY has also been reported to participate in cell proliferation and cell death due to its influence on genes encoding proteins responsible for apoptosis (225–227), thus participating in delayed or decreased apoptosis and induction of inflammatory cells to remain in the diseased sites and producing increased levels of pro-inflammatory cytokines. This, in turn, leads to the formation and deterioration of inflammation and contributes to the destruction of tooth-supporting tissues. In this process, P. gingivalis HmuY participates in the induction of a controlled pro-inflammatory response but at the same time participates in processes allowing for insufficient bacterial clearance, bacterial proliferation, and intracellular survival, finally leading to protracted tissue destruction, which is typically seen in advanced periodontitis.

HmuY and its relationship to systemic diseases

The pathogenic processes in periodontitis and diabetes are intercorrelated and, in general, characterized by increased inflammation, which leads to impaired insulin signaling, insulin resistance, and aggravation of diabetes, and therefore diabetic patients are more susceptible to severe forms of periodontitis (30, 211, 228–235). From the other perspective, the non-enzymatic addition of glucose to circulating and structural proteins caused by prolonged periods of hyperglycemia results in glycation of two of the most abundant proteins from the periodontitis perspective, i.e., Hb and collagen, which could contribute to the increased pathogenic potential of P. gingivalis in vivo. Indeed, glycation increases heme sequestration from metHb by HmuY, facilitates homotypic biofilm formation by P. gingivalis on glycated collagen-coated abiotic surfaces, and increases its invasion toward gingival keratinocytes (33).

A correlation between periodontitis and respiratory diseases, mainly in the lungs infected with P. aeruginosa, also exists (31). For example, in individuals with cystic fibrosis, higher co-colonization of P. aeruginosa with P. gingivalis and P. intermedia occurs (236–238), and Hb available from micro-bleeds is the most likely source of heme for oral anaerobic pathogens. Heme acquisition in this environment is facilitated by pyocyanin produced by P. aeruginosa, which efficiently oxidizes oxyHb to metHb, the latter being more susceptible to proteolysis carried out by P. gingivalis Kgp and neutrophil elastase, as well as to direct heme sequestration by P. gingivalis HmuY (32).

Several reports have demonstrated that infectious agents found in the brains of Alzheimer’s patients may be involved in the initiation and progression of this disease (239). Among them, P. gingivalis has recently been demonstrated as an important risk factor involved in developing dementia and Alzheimer’s disease (240–242). Such an assumption has been supported by the presence of P. gingivalis lipopolysaccharide in human brains (243) and P. gingivalis DNA (including that encoding HmuY and gingipains) in human brains and cerebrospinal fluid of individuals with Alzheimer’s disease (24).

CONCLUSIONS

In conclusion, although pathogenic bacteria can take up heme directly, the production and employment of hemophores or hemophore-like proteins significantly enhance this process and contribute to enhanced bacterial survival in heme-limited host environments. Our central hypothesis presented in this review is that the HmuY-based heme acquisition mechanism is used not only to fulfill nutritional requirements but also to increase the virulence potential of opportunistically pathogenic host-associated members of the Bacteroidota phylum and thereby their tendency to cause dysbiosis. Importantly, competition in heme acquisition occurs not only between P. gingivalis and other periodontopathogens but also between P. gingivalis and cohabitating bacteria in host niches other than the oral cavity. Such a phenomenon is, at least in part, possible because HmuY can sequester heme from metHb directly, a process which is facilitated by bacterial proteases or other virulence factors produced by bacteria, and can also compete with albumin and hemopexin, which ordinarily reduce heme availability in the circulation. This advantage could be explained by the independent evolution of heme-binding properties, visible in different heme iron coordination modes in P. gingivalis HmuY versus its homologs. In addition, gene diversity in P. intermedia and B. fragilis points to the significance of the evolutionary steps leading to such differentiation.

Biographies

Teresa Olczak is a professor in the Faculty of Biotechnology at the University of Wrocław (Poland). She earned a Master’s degree in medical analytics from the Medical Academy of Wrocław (Poland), a Ph.D. degree in biochemistry from the University of Wrocław (Poland), and postdoctoral training in the Department of Microbiology, Section of Infectious Diseases at the Boston University School of Medicine (MA, USA). During the first years of the research, she was interested in glycosylation, mainly in connection with human disorders caused by defects in the glycosylation process. For the last 25 years, she has been studying the bases of diseases caused by dysbiosis in the human oral and gut microbiome. Her main research interests comprise the heme uptake mechanisms used by anaerobic bacteria belonging to the Bacteroidota phylum. Her most noteworthy research achievement to date has been the identification and characterization of a novel group of bacterial heme-binding proteins exhibiting hemophore-like functions.

Michał Śmiga is a researcher in the Faculty of Biotechnology at the University of Wrocław (Poland). He earned his Bachelor’s and Master’s degrees in biotechnology and a Ph.D. degree in biochemistry from the University of Wrocław (Poland). His primary research focus is on unraveling the virulence mechanisms of pathogenic bacteria, particularly in relation to the development of periodontal diseases in humans. During his doctoral studies, he analyzed the processes regulating the virulence of Porphyromonas gingivalis. Currently, his area of interest revolves around assessing the significance of heme uptake systems, heme homeostasis, and their regulation in the virulence of Porphyromonas gingivalis, as well as other anaerobic pathogenic bacteria, such as Prevotella intermedia, Tannerella forsythia, and Porphyromonas endodontalis.

Svetlana Antonyuk is a reader in the Faculty of Health and Sciences at the University of Liverpool (UK). She earned her undergraduate degree at Moscow State University (USSR), a Ph.D. degree in X-ray protein crystallography from the Institute of Crystallography (Russia), and postdoctoral training in the Molecular Biophysics Group at Daresbury Laboratory, Cheshire (UK). She applies structural biology tools (X-ray protein crystallography and cryo-EM) to understand the molecular details of enzyme catalysis and determine the structures of proteins involved in MND, liver disease, heart disease, and infectious diseases caused by bacteria, viruses, and parasites. She also helps medical chemists visualize drug-like molecules binding to druggable targets.

John Smalley studied at the University of Liverpool (UK), where he gained a Bachelor’s degree in medical bacteriology and a Ph.D. degree in dental science. He gained post-doctoral connective tissue research experience in the Department of Medical Biochemistry at the University of Manchester (UK). He returned to the University of Liverpool to take up a lectureship in the Department of Dental Science, where he has researched various aspects of the roles played by oral anaerobes in the breakdown and synthesis of periodontal connective tissue components. His most notable work has involved the chemical characterization of the heme-containing pigments of Porphyromonas and Prevotella species. He has extended these studies to elucidate the mechanisms of protease- and hemophore-mediated heme acquisition from hemoglobin and other heme-containing host proteins. His recent work has focused on the influence of non-enzymatically glycated host proteins on the pathogenicity of Porphyromonas gingivalis.