The Evolution & Medicine Review

Iron is a critical metal for essential cellular processes, such as respiration, in both human and microbial cells.  Thus, in the context of infection, iron is a high-value cellular commodity and an evolutionist might reasonably expect a metallic tug-of-war between host and pathogen iron-binding proteins or other iron-binding molecules (siderophores).  This speculation is impressively supported in a paper published this month (Barber and Elde, 2014).  These authors provide strong evidence for positive selection affecting several sites in host (transferrin, Tf) and pathogen (transferrin binding protein A) iron-binding proteins based on a combination of genetic, structural, and functional experimental methods.

Tf is a ferric (Fe3+) iron-binding protein synthesized in the liver and secreted into the blood.  This plasma protein contains two similar lobes, N (N-terminal) and C (C-terminal), each of which can bind one ferric iron ion.  Tf serves to ferry iron absorbed from the intestines to cells, especially in the bone marrow, where erythrocyte precursors require iron for heme, a critical component of hemoglobin and necessary for binding oxygen.  Tf saturated with bound iron (diferric or holo-Tf) is captured by transferrin receptors (Tf-R) on cells, and the Tf-R internalizes the Tf-iron complex via receptor-mediated endocytosis.  The iron is released from Tf at low pH in endocytic compartments, and the Tf-R returns to the plasma membrane and releases the ‘empty’ Tf back into the circulation.  Binding of iron by Tf or by other proteins or smaller molecules (siderophores) is also important to prevent free iron from catalyzing oxygen free radical formation, which can cause substantial cellular damage by denaturing proteins, causing DNA strand breaks, and oxidizing lipids.

A third reason for the host to sequester iron in addition to transport and inhibition of free radical generation is to deny host iron supplies to any microbes that may have invaded host tissues.  Barber and Elde expected that transferrin would be subject to purifying selection that maintains both the capacity to bind iron and the ability to effectively interact with the Tf-R on host cell surfaces.  But they wondered if the potential for iron piracy by pathogen derived iron-binding proteins would lead to positive selection on portions of the transferrin molecule.

To begin to address this interesting question, the authors cloned and sequenced transferrin orthologs from 21 species of hominoids and monkeys, both Old and New World.  They then carried out phylogenetic analysis including determination of ratios of nonsynonymous to synonymous amino acid substitution rates as a means to detect positive selection.  Although this approach is not without limitations, the results strongly supported the notion that some positions in the Tf amino acid sequence were subject to positive selection in this clade.  This result is what is normally expected of immune system receptors involved in co-evolutionary processes between host and pathogen, sometimes referred to as “arms races.”

What was not necessarily expected is that of 18 Tf amino acid positions found to be rapidly evolving, 16 were located in the C lobe and only two were located in the N lobe.  So, while the C lobe has been subjected to positive selection, the N lobe has been subjected primarily to purifying selection.  Since only one of the rapidly evolving sites is implicated in binding to Tf-R, Barber and Elde argue that it is unlikely that co-evolution with Tf-R is responsible for the rapid evolutionary changes in the C lobe.

A study (Cheng et al., 2004) of the structure of the Tf/Tf-R complex suggests that both the N and C lobes of Tf contact Tf-R.  Based on the results recounted above, one might predict that the C lobe uses a different face to interact with Tf-R versus TbpA, given the purifying selection predicted for the former and the positive selection demonstrated for the latter.

Insight into why the C lobe (or more precisely, part of the C lobe), and not the N lobe, is subject to rapid evolution stemming from positive selection of amino acid substitutions is provided by the authors who note that 14 of the 16 quickly evolving sites appear to be contacted by the transferrin binding protein A (TbpA) of Neisseria meningitidis as determined by inspection of a co-crystal structure of human Tf and TbpA.  The TbpA proteins, which are surface receptors, of the Gram-negative human pathogens N. meningitidis, N. gonorrhoeae, and Haemophilus influenzae all extract iron from the C lobe, but not the N lobe, of Tf.

Previous work by others demonstrated that the TbpAs from N. gonorrhoeae and H. influenzae bound recombinant human Tf.  Barber and Elde expressed recombinant forms of these TbpAs, and show that they can bind human and gorilla Tfs but not the Tfs from chimpanzees, orangutans, gibbons, or baboons.

Chimpanzee Tf differs from human Tf by only four amino acids.  One of the positions that differs, 591, exhibits the substitution patterns associated with multiple cycles of positive selection.  Whereas human Tf has glutamic acid at position 591, chimpanzee Tf has lysine.  If glutamic acid is substituted for lysine at position 591 in chimpanzee Tf, TbpA molecules from both N. gonorrhoeae and H. influenzae bind dramatically better.  Conversely, replacing glutamic acid by lysine in human Tf substantially reduces binding by TbpA from both N. gonorrhoeae and H. influenzae.

Given the dramatic impact of a single point mutation on Tf-TbpA interaction, the authors then asked whether a standard human Tf polymorphism, that happens to occur at position 589 (only two positions from 591), could similarly influence interaction with TbpA molecules.  The two human Tf alleles, C1 and C2 with C1 being the more prevalent, differ at position 589 due to a single C to T transversion resulting in the replacement of proline (C1) by serine (C2).  This single site of variation causes C2 to bind much less well than C1 to the H. influenzae TbpA being used.  The TbpAs from N. meningitidis and N. gonorrhoeae bind to C1 and C2 Tf molecules more comparably.  Based on these results, the authors suggest that for pathogens there may be a trade-off between binding more strongly to C2 Tf versus broader recognition of Tf molecules.

Barber and Elde then compared TbpA gene sequences from a collection of Neisseria and H. influenzae strains isolated from humans.  Through their phylogenetic analysis, the authors concluded that 10 sites in the Neisseria TbpAs and 9 sites in the H. influenzae TbpAs were subject to strong positive selection.  Almost all of these sites, only one of which overlaps between Neisseria and H. influenzae TbpAs, are in the portions of the TbpA structure expected to be extracellular loops that directly interact with Tf.  While TbpA molecules from related pathogen strains vary extensively in these regions, they exhibit a much higher degree of conservation in the transmembrane domains, which do not directly interact with Tf.

The authors mutagenized three of these evolving sites in N. gonorrhoeae TbpA such that the corresponding amino acids from H. influenzae TbpA are substituted.  One of these replacements decreased binding to human Tf.  Barber and Elde conclude that amino acid substitutions at sites identified as subject to positive selection in TbpA, like Tf, can influence the Tf-TbpA interaction.

The grand conclusion from this study is that structural variation in a protein, such as Tf, that performs critical everyday metabolic tasks can nevertheless be consequential for host defense every bit as much as structural variation in proteins dedicated to recognizing pathogen-derived molecules, i.e. more typical immune system receptors.

The centrality of iron in host-pathogen interaction is also supported by another study (Liu, 2014) from June of this year.  Devireddy and colleagues address the battle for iron between mammalian hosts and pathogens that secrete small iron-binding molecules, called siderophores.

One example of a bacterial siderophore is 2,3-dihydroxy benzoic acid (2,3-DHBA), used by Escherichia coli.  Mammalian cells synthesize a similar siderophore, 2,5-DHBA, and a protein 24p3, that can bind either the host or pathogen siderophores.  Similarly, as the authors demonstrate the bacteria can acquire iron from either the host or their own siderophores to support growth.

Liu et al. show that in response to bacterial infection, hosts decrease production of the host siderophore, 2,5-DHBA and increase production of 24p3.  The reduced concentration of host siderophore decreases the amount of iron the bacteria can acquire from 2,5-DHBA, and the increased amount of 24p3 leads to sequestration of the bacterial siderophore, thereby preventing the bacterial cells from capturing as much iron as would otherwise be the case through that pathway.  Mice lacking a gene encoding an enzyme essential for synthesis of 2,5-DHBA were demonstrated to be more resistant to infection with two different E. coli strains.  Supplementation of these enzyme-deficient mice with 2,5-DHBA substantially increased susceptibility to E. coli infection.

In conclusion, there are different host and pathogen strategies in play in different host-pathogen encounters.  The common theme is that the battle for a scarce resource, i.e. iron, generates selective pressures on host and pathogen molecules potentially as consequential as those brought to bear on canonical immune system receptors and their pathogen-derived ligands.  Evolution is unrestrained by our conceptions of the immune system.


Barber MF, Elde NC. Nutritional immunity. Escape from bacterial iron piracy through rapid evolution of transferrin. Science. 2014 Dec 12;346(6215):1362-6. doi: 10.1126/science.1259329. PubMed PMID: 25504720.

Cheng Y, Zak O, Aisen P, Harrison SC, Walz T. Structure of the human transferrin receptor-transferrin complex. Cell. 2004 Feb 20;116(4):565-76. PubMed PMID: 14980223.

Liu Z, Reba S, Chen WD, Porwal SK, Boom WH, Petersen RB, Rojas R, Viswanathan R, Devireddy L. Regulation of mammalian siderophore 2,5-DHBA in the innate immune response to infection. J Exp Med. 2014 Jun 2;211(6):1197-213. doi: 10.1084/jem.20132629. Epub 2014 May 26. PubMed PMID: 24863067; PubMed Central PMCID: PMC4042634.