haptoglobin, a hemoglobin-binding plasma protein, is ... · haptoglobin, a hemoglobin-binding...

6
Haptoglobin, a hemoglobin-binding plasma protein, is present in bony fish and mammals but not in frog and chicken Krzysztof B. Wicher and Erik Fries Department of Medical Biochemistry and Microbiology, Uppsala University, 75123 Uppsala, Sweden Edited by Morris Goodman, Wayne State University School of Medicine, Detroit, MI, and approved January 18, 2006 (received for review October 5, 2005) Hemoglobin (Hb) released from erythrocytes may cause oxidation of lipids and proteins. Haptoglobin (Hp), which occurs in the plasma of all mammals, binds free Hb and inhibits its oxidative activity. It is not known whether this protective protein also exists in lower vertebrates. By analyzing available genomic sequences, we have found that bony fish, but not more primitive animals, have a gene coding for a protein homologous to mammalian Hp. Furthermore, we show that this protein is present in the plasma of Japanese pufferfish (Takifugu rubripes) and that it binds Hb. These results, together with a phylogenetic analysis, suggest that Hp evolved from a complement-associated protein (mannose-binding lectin-associated serine proteinase, MASP), with the emergence of fish. Surprisingly, we found that both chicken (Gallus gallus) and the Western clawed frog (Xenopus tropicalis) lack the Hp gene. In chicken plasma, however, we identified a different type of Hb- binding protein, PIT54, which has been reported to be a potent antioxidant. PIT54 is a soluble member of the family of scavenger receptor cysteine-rich proteins, and we found that its gene exists only in birds. We also show that the plasma of ostrich (Strutio camelus), a primitive bird, contains both PIT54 and Hp. Collectively, our data suggest that PIT54 has successively taken over the function of Hp during the evolution of the avian lineage and has completely replaced the latter protein in chicken. evolution PIT54 complement proteinase H emoglobin (Hb) is an ancient protein used by species from all five kingdoms of organisms to transport and store oxygen. In vertebrates, Hb is confined to erythrocytes, in which it occurs at concentrations as high as 300 mgml (1). Under certain pathological conditions, such as infection, chemical intoxication, or certain genetic disorders, the erythrocyte mem- brane may be destabilized, resulting in leakage of Hb into the circulation (2). This free Hb can oxidize lipids and proteins and may therefore be harmful to the surrounding tissues. The plasma of all mammals studied so far has been shown to contain haptoglobin (Hp), a protein that binds free Hb with high affinity and inhibits its oxidative activity (3). The formation of a complex between Hp and Hb also prevents Hb from passing through the glomeruli. Thus, mice lacking Hp display renal failure (4). Whether lower vertebrates also use Hp as a protective measure against the harmful effects of free Hb is not known. Hb possesses intrinsic peroxidase activity, which has been exploited for the detection of Hb-binding proteins after agarose gel electrophoresis. Based on this indirect method, it has been claimed that the plasma of a number of animals from different classes, such as birds, snakes, turtles, and fish, contain Hp (5). Hb-binding proteins have been isolated from newt and chicken plasma, but their amino acid sequences have not been deter- mined (6, 7). The primary translation product of mammalian Hp mRNA is a polypeptide that dimerizes cotranslationally and is proteolyti- cally cleaved while still in the endoplasmic reticulum (8). This cleavage seems to increase the affinity of the molecule for Hb (9). The mature, tetrameric molecule consists of two short and two long polypeptides: the - and -chains, respectively. The -chain contains a complement control protein (CCP) domain, which is also found in many proteins involved in complement regulation, e.g., complement factors H and C1r, mannose- binding lectin-associated serine proteinases (MASPs), and C1 receptor (10). In humans, the -chain exists in two major allelic forms of 9 kDa (Hp1) and 18 kDa (Hp2), the latter having arisen from partial gene duplication (11). The -chain, with a molecular mass of 35 kDa, is related to chymotrypsin-like serine pro- teinases, but does not possess the essential catalytic amino acid residues (12). Primate Hp has been used as a model in studies of the evolution of multigene families. It has provided the first example of a gene family in which the appearance of a new member has occurred by gene duplication followed by unequal crossing-over (13). Furthermore, other recombinational events, such as gene conversion and intragenic duplication, have been shown to be involved in the formation of a Hp-gene cluster (14). In this article, we present evidence that a Hb-binding protein homologous to the -chain of mammalian Hp appeared early in vertebrate evolution, possibly coinciding with the emergence of bony fish. Results Hp in Invertebrates. To identify proteins similar to mammalian (human) Hp in different species, we analyzed genomic and EST sequences with the BLAST program. The predicted protein sequences with the highest scores were examined for the pres- ence of the Hp hallmark: a serine proteinase (SP) domain lacking the essential serine and histidine residues of the active site and a ‘‘histidine-loop’’ cysteine pair (15). We investigated the ge- nomes of the five invertebrates: Saccharomyces cerevisiae, Cae- norhabditis elegans, Apis mellifera, Drosophila melanogaster, and Ciona intestinalis and found that none contained a gene coding for a protein similar to mammalian Hp. The closest relative found was a SP, MASP, in C. intestinalis (data not shown). Hp Proteins in Bony Fish. The genomes of three bony (teleost) fishes have been fully sequenced to date: zebrafish (Danio rerio), spotted green pufferfish (Tetraodon nigroviridis), and Japanese pufferfish (Takifugu rubripes). In all three genomes, we found a gene encoding a protein with an amino acid sequence 32–34% identical and 49 –52% similar to that of the -chain of human Hp (Fig. 1). The Hp-like protein (HpL) encoded by this gene seems to comprise a single domain with similar structural characteris- tics to Hp described above. In addition, the cysteine residues of this domain are present at the positions corresponding to those Conflict of interest statement: No conflicts declared. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: Hp, haptoglobin; HpL, Hp-like protein; SP, serine proteinase; MASP, mannose-binding lectin-associated SP; CCP, complement control protein; DHODH, dihy- droorotate dehydrogenase. To whom correspondence should be addressed. E-mail: [email protected]. © 2006 by The National Academy of Sciences of the USA 4168 – 4173 PNAS March 14, 2006 vol. 103 no. 11 www.pnas.orgcgidoi10.1073pnas.0508723103 Downloaded by guest on December 10, 2020

Upload: others

Post on 23-Aug-2020

18 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Haptoglobin, a hemoglobin-binding plasma protein, is ... · Haptoglobin, a hemoglobin-binding plasma protein, is present in bony fish and mammals but not in frog and chicken Krzysztof

Haptoglobin, a hemoglobin-binding plasma protein,is present in bony fish and mammals but notin frog and chickenKrzysztof B. Wicher† and Erik Fries

Department of Medical Biochemistry and Microbiology, Uppsala University, 75123 Uppsala, Sweden

Edited by Morris Goodman, Wayne State University School of Medicine, Detroit, MI, and approved January 18, 2006 (received for review October 5, 2005)

Hemoglobin (Hb) released from erythrocytes may cause oxidationof lipids and proteins. Haptoglobin (Hp), which occurs in theplasma of all mammals, binds free Hb and inhibits its oxidativeactivity. It is not known whether this protective protein also existsin lower vertebrates. By analyzing available genomic sequences,we have found that bony fish, but not more primitive animals, havea gene coding for a protein homologous to mammalian Hp.Furthermore, we show that this protein is present in the plasma ofJapanese pufferfish (Takifugu rubripes) and that it binds Hb. Theseresults, together with a phylogenetic analysis, suggest that Hpevolved from a complement-associated protein (mannose-bindinglectin-associated serine proteinase, MASP), with the emergence offish. Surprisingly, we found that both chicken (Gallus gallus) andthe Western clawed frog (Xenopus tropicalis) lack the Hp gene. Inchicken plasma, however, we identified a different type of Hb-binding protein, PIT54, which has been reported to be a potentantioxidant. PIT54 is a soluble member of the family of scavengerreceptor cysteine-rich proteins, and we found that its gene existsonly in birds. We also show that the plasma of ostrich (Strutiocamelus), a primitive bird, contains both PIT54 and Hp. Collectively,our data suggest that PIT54 has successively taken over thefunction of Hp during the evolution of the avian lineage and hascompletely replaced the latter protein in chicken.

evolution � PIT54 � complement proteinase

Hemoglobin (Hb) is an ancient protein used by species fromall five kingdoms of organisms to transport and store

oxygen. In vertebrates, Hb is confined to erythrocytes, in whichit occurs at concentrations as high as 300 mg�ml (1). Undercertain pathological conditions, such as infection, chemicalintoxication, or certain genetic disorders, the erythrocyte mem-brane may be destabilized, resulting in leakage of Hb into thecirculation (2). This free Hb can oxidize lipids and proteins andmay therefore be harmful to the surrounding tissues. The plasmaof all mammals studied so far has been shown to containhaptoglobin (Hp), a protein that binds free Hb with high affinityand inhibits its oxidative activity (3). The formation of a complexbetween Hp and Hb also prevents Hb from passing through theglomeruli. Thus, mice lacking Hp display renal failure (4).Whether lower vertebrates also use Hp as a protective measureagainst the harmful effects of free Hb is not known.

Hb possesses intrinsic peroxidase activity, which has beenexploited for the detection of Hb-binding proteins after agarosegel electrophoresis. Based on this indirect method, it has beenclaimed that the plasma of a number of animals from differentclasses, such as birds, snakes, turtles, and fish, contain Hp (5).Hb-binding proteins have been isolated from newt and chickenplasma, but their amino acid sequences have not been deter-mined (6, 7).

The primary translation product of mammalian Hp mRNA isa polypeptide that dimerizes cotranslationally and is proteolyti-cally cleaved while still in the endoplasmic reticulum (8). Thiscleavage seems to increase the affinity of the molecule for Hb(9). The mature, tetrameric molecule consists of two short and

two long polypeptides: the �- and �-chains, respectively. The�-chain contains a complement control protein (CCP) domain,which is also found in many proteins involved in complementregulation, e.g., complement factors H and C1r, mannose-binding lectin-associated serine proteinases (MASPs), and C1receptor (10). In humans, the �-chain exists in two major allelicforms of 9 kDa (Hp1) and 18 kDa (Hp2), the latter having arisenfrom partial gene duplication (11). The �-chain, with a molecularmass of �35 kDa, is related to chymotrypsin-like serine pro-teinases, but does not possess the essential catalytic amino acidresidues (12). Primate Hp has been used as a model in studies ofthe evolution of multigene families. It has provided the firstexample of a gene family in which the appearance of a newmember has occurred by gene duplication followed by unequalcrossing-over (13). Furthermore, other recombinational events,such as gene conversion and intragenic duplication, have beenshown to be involved in the formation of a Hp-gene cluster (14).

In this article, we present evidence that a Hb-binding proteinhomologous to the �-chain of mammalian Hp appeared early invertebrate evolution, possibly coinciding with the emergence ofbony fish.

ResultsHp in Invertebrates. To identify proteins similar to mammalian(human) Hp in different species, we analyzed genomic and ESTsequences with the BLAST program. The predicted proteinsequences with the highest scores were examined for the pres-ence of the Hp hallmark: a serine proteinase (SP) domain lackingthe essential serine and histidine residues of the active site anda ‘‘histidine-loop’’ cysteine pair (15). We investigated the ge-nomes of the five invertebrates: Saccharomyces cerevisiae, Cae-norhabditis elegans, Apis mellifera, Drosophila melanogaster, andCiona intestinalis and found that none contained a gene codingfor a protein similar to mammalian Hp. The closest relativefound was a SP, MASP, in C. intestinalis (data not shown).

Hp Proteins in Bony Fish. The genomes of three bony (teleost)fishes have been fully sequenced to date: zebrafish (Danio rerio),spotted green pufferfish (Tetraodon nigroviridis), and Japanesepufferfish (Takifugu rubripes). In all three genomes, we found agene encoding a protein with an amino acid sequence 32–34%identical and 49–52% similar to that of the �-chain of human Hp(Fig. 1). The Hp-like protein (HpL) encoded by this gene seemsto comprise a single domain with similar structural characteris-tics to Hp described above. In addition, the cysteine residues ofthis domain are present at the positions corresponding to those

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: Hp, haptoglobin; HpL, Hp-like protein; SP, serine proteinase; MASP,mannose-binding lectin-associated SP; CCP, complement control protein; DHODH, dihy-droorotate dehydrogenase.

†To whom correspondence should be addressed. E-mail: [email protected].

© 2006 by The National Academy of Sciences of the USA

4168–4173 � PNAS � March 14, 2006 � vol. 103 � no. 11 www.pnas.org�cgi�doi�10.1073�pnas.0508723103

Dow

nloa

ded

by g

uest

on

Dec

embe

r 10

, 202

0

Page 2: Haptoglobin, a hemoglobin-binding plasma protein, is ... · Haptoglobin, a hemoglobin-binding plasma protein, is present in bony fish and mammals but not in frog and chicken Krzysztof

in the SP domain of mammalian Hps (the �-chain), with theexception of the residue that in mammalian Hps joins the �- andthe �-chains. The SP domain of the fish HpLs differs from thatof mammalian Hps in that it is encoded by two, rather than one,exon (data not shown). Each of these exons codes for the sameregion in all three proteins, while the length of the separatingintron, as inferred from the ENSEMBL annotation, varies (97, 122,and 4,383 nt long for T. nigroviridis, T. rubripes, and D. rerio,respectively). The SP domain of these HpLs is preceded by a 16-to 22-residue extension and a putative signal peptide. The ORFsfor the HpL genes of T. nigroviridis and T. rubripes code for anadditional 90- to 100-aa sequence upstream from the predictedsignal peptides. However, analysis of the Kozak sequencesaround the possible translation initiation sites suggested that itis more likely that the second methionine codon is used forinitiating (data not shown). The amino acid sequences of theextension differ considerably between the different species.However, the C termini of all of these extensions contain theconsensus cleavage sequence for subtilisin-like proprotein con-vertases: R-X-(R�K)-R (16) (Fig. 1). Using the sequences ofthese three fish proteins, we identified ESTs coding for HpLfrom other teleost species. From their sequences, we were ableto deduce the amino acid sequence of HpL from seven morebony fishes (Data Set 1, which is published as supportinginformation on the PNAS web site).

To investigate whether fish HpL binds to Hb, we incubatedserum from T. rubripes with gel beads to which Hb from koi carp(Ciprinus carpio) was coupled, eluted bound proteins with urea,and analyzed them by SDS�PAGE (Fig. 2). Two of the majorbands, with apparent molecular masses of �75 and 45 kDa(bands 1 and 2), were identified by MS as IgM heavy chain(�-chain) and HpL of T. rubripes, respectively. The other two

major bands of �30 and 15 kDa, bands 3 and 4, respectively, werefound to be globin, the subunit of Hb.

Amphibian Hb-Binding Proteins. We searched the Xenopus tropica-lis genomic database for a gene coding for a protein homologousto mammalian Hp. The closest related proteins we could findwere orthologs of C1r, C1s, and MASPs. However, the active

Fig. 1. Amino acid sequence alignment of human Hp (gi: 1620396) and HpLs of three bony fishes: T. nigroviridis, T. rubripes, and D. rerio. Amino acid residuescorresponding to the �-chain of human Hps are shaded. Identical residues, conserved substitutions, and semiconserved substitutions are indicated by *, :, and �,respectively. The signal peptide of human Hp and the putative signal sequences of the fish proteins are in bold. (The HpL genes of T. nigroviridis and T. rubripescode for an additional 90–100 aa upstream of the predicted signal peptides. Analysis of the Kozak sequences around the possible initiation codons shows thatthe second methionine codon is more likely to be used for initiating translation.) Consensus cleavage sequences for subtilisin-like proprotein convertases in theHpLs and the corresponding region in human Hp are framed. Conserved cysteine residues are indicated by arrows. Amino acids substituting the essential serineand histidine residues in the active site of homologous SPs are underlined. The bar over the alignment marks the amino acid sequence corresponding to thehistidine loop of SPs.

Fig. 2. Identification of Hb-binding proteins in serum of T. rubripes. Serum(Con) was incubated with Hb-containing gel beads. After centrifugation,unbound proteins were removed (Sup) and bound proteins were eluted with8 M urea (Eluate). All samples were then analyzed by SDS�PAGE underreducing conditions followed by staining. Protein bands marked 1–4 weresubjected to MS. The molecular masses in kDa of reference proteins are shownto the left.

Wicher and Fries PNAS � March 14, 2006 � vol. 103 � no. 11 � 4169

EVO

LUTI

ON

Dow

nloa

ded

by g

uest

on

Dec

embe

r 10

, 202

0

Page 3: Haptoglobin, a hemoglobin-binding plasma protein, is ... · Haptoglobin, a hemoglobin-binding plasma protein, is present in bony fish and mammals but not in frog and chicken Krzysztof

sites of all of these proteins contain the amino acids residuesessential for proteolytic activity. We therefore conclude thatthere is no Hp in X. tropicalis. To determine whether the serumof this type of amphibian contains any Hb-binding protein, weused affinity chromatography and SDS�PAGE, as describedabove; for practical reasons, Xenopus laevis was used for theseexperiments. The final MS analysis of the major electrophoreticband yielded MGC69066, a protein similar to the �-chain of IgM.This protein has not previously been assigned any function andwas not studied further.

Avian Hb-Binding Proteins. We did not find any gene coding for aprotein similar to mammalian Hp in the genome of chicken. Themost similar proteins, as was the case for X. tropicalis, wereorthologs of mammalian C1r, C1s, MASPs, and other relatedSPs. We conclude, therefore, that chicken also does not have Hp.However, the fact that a Hb-binding protein was previouslyisolated from chicken plasma prompted us to identify thisprotein. In agreement with a previous study (6), we found thataffinity chromatography with immobilized chicken Hb followedby SDS�PAGE yielded a 69-kDa band. MS analysis showed thatthis band contained PIT54 (or 18-B), a soluble member of thefamily of scavenger receptor cysteine-rich proteins. Similarly, weisolated a Hb-binding protein of 75 kDa from plasma of domesticgoose (Anser anser). Analysis by liquid chromatography tandemMS showed that is was a PIT54 homolog (data not shown).

We also examined the presence of Hb-binding proteins inostrich, a representative of paleognaths, a separate phylogeneticline of modern birds (reviewed in ref. 17). To this end, plasmafrom common ostrich (Strutio camelus) was incubated with gelbeds coupled with ostrich Hb. After washing, bound proteinswere eluted with solutions of increasing urea concentration (Fig.3A). Analysis by SDS�PAGE under reducing conditions showedthat a 75-kDa polypeptide (band 1) was eluted with 1.5–3.0 Murea, and three polypeptides of 35, 15, and 12 kDa (bands 2, 3,and 4, respectively) were eluted with 3.0–8.0 M urea. Band 4 wasapparently globin because it was also eluted without priorincubation with plasma (data not shown). Under nonreducingconditions, the polypeptide eluted with 1.5–3.0 M urea had anapparent molecular mass of 60 kDa (band 5). Under the sameconditions, bands 2 and 3 were absent, but a single band of �100kDa appeared (band 6), indicating that the polypeptides of theformer bands were linked by disulfide bridges.

N-terminal sequencing of the 35-kDa polypeptide (band 2)yielded the sequence IIGGLLAGKGSFPWQ, which is 66–80%identical with that of the N terminus of the �-chain of mamma-lian Hps. No sequence was obtained when the same techniquewas applied to the 75-kDa polypeptide, probably because the�-amino group of the N-terminal residue was blocked. Thispolypeptide, therefore, was analyzed with liquid chromatogra-phy tandem MS. Five of the seven peptide sequences obtainedwith this technique could be aligned with a high degree ofidentity with the protein sequence of chicken PIT54 (Fig. 3B).Three of these peptides aligned at two to three places, presum-ably reflecting the fact that PIT54 consists of four repeateddomains (18).

Phylogenetic Analysis. Given the fact that fish HpLs differ frommammalian Hps both with respect to domain and exon organi-zation, we wanted to determine whether phylogenetic analysisindicated a common ancestor. To this end, we constructed aphylogenetic tree of the SP domains of fish HpLs, human Hp,human and fish C1r�s, and MASP proteins. High bootstrapvalues were obtained with the maximum-parsimony method(Fig. 4) and the neighbor-joining method (Fig. 7, which ispublished as supporting information on the PNAS web site), 964of 1,000 and 1,000 of 1,000 respectively, for the branch contain-ing human Hp and fish HpLs, indicating that these proteins have

a common ancestor. The two methods did not give a consistentbootstrap support for a hierarchy between the branch containingHp and HpL and the other branches.

Chromosomal Localization of Hp and HpL. To further test thehypothesis that Hp and HpL have a common origin, we analyzedthe chromosomal regions close to the respective genes. We foundthat both Hp in mammals and HpL in fish are located next to thegene for dihydroorotate dehydrogenase (DHODH) (Fig. 5). Thefact that DHODH is a single-copy gene in both groups of animalssuggests that Hp and HpL are orthologs.

The finding that the Hp gene is not present in frog and chickenprompted us to study the corresponding part of the genome ofthese animals. We found that, in both Xenopus and chicken,there is a gene coding for ATP-dependent RNA helicase DHX38(DHX38) located next to DHODH (Fig. 5). In mammals, DHX38

Fig. 3. Identification of Hb-binding proteins in ostrich serum. (A) Ostrichserum (Con) was incubated with Hb-containing gel beads. After centrifuga-tion, unbound proteins were removed (Sup), and bound proteins were elutedwith increasing concentrations of urea (Eluate). All samples were then ana-lyzed by SDS�PAGE under reducing or nonreducing conditions, followed bystaining. Protein bands marked 1–6 are described in the text. The molecularmasses in kDa of reference proteins are shown to the left. (B) Protein fromband 1 was analyzed by liquid chromatography tandem MS yielding thesequences of several peptides. The alignment of these peptides with theknown sequence of chicken PIT54 is shown. Peptides aligning at multipleplaces are underlined. Amino acid residues in locations common to PIT54 andthe peptides are in bold.

4170 � www.pnas.org�cgi�doi�10.1073�pnas.0508723103 Wicher and Fries

Dow

nloa

ded

by g

uest

on

Dec

embe

r 10

, 202

0

Page 4: Haptoglobin, a hemoglobin-binding plasma protein, is ... · Haptoglobin, a hemoglobin-binding plasma protein, is present in bony fish and mammals but not in frog and chicken Krzysztof

is also located close to DHODH, separated only by Hp andTXNL4B. Taken together, these observations indicate that Hphas been deleted from the region between DHODH and DHX38in both X. tropicalis and chicken.

DiscussionIn this study, we have sought to ascertain whether animals lowerthan mammals express the plasma protein Hp. We show that, inbony fish, there is a HpL consisting of only one of the twodomains of mammalian Hp, whereas, in both chicken and theAfrican clawed frog, the protein is absent. On the other hand, inostrich, which is a primitive type of bird, there is Hp withapparently the same domain structure as that of mammalian Hp.

Hp in Bony Fish. In the three bony fishes whose genome has beenfully sequenced, we found a gene coding for a protein with anamino acid sequence similar to that of the �-chain of mammalianHp (Fig. 1). Using the sequences of these three HpLs as well asESTs, we obtained sequences of HpLs from seven more bony fish

(Data Set 1), indicating that this protein is widely expressed inthis class of vertebrates. The predicted translational products ofthe genes coding for HpL consist of a putative signal peptide, a16- to 22-residue extension with the consensus cleavage se-quence for subtilisin-like proprotein convertases at its C-terminal end, and a SP domain (Fig. 1). These characteristicssuggest that HpL is a secretory protein and that it is proteolyti-cally cleaved near the SP domain while passing through the Golgiapparatus. Indeed, our preliminary experiments with HpL fromT. rubripes expressed in COS-7 cells support this notion (ourunpublished observation). In contrast, the primary translationproduct of mammalian Hp mRNA consists of a signal sequence,a CCP, and a SP domain. After the release of the signal sequenceand while still in the endoplasmic reticulum, this polypeptide iscleaved between the two domains by C1r-like protein (19).Whether the difference in cleavage mechanisms between HpLand Hp has any physiologic significance remains to be studied.We found that HpL of at least one bony fish occurs in plasma andbinds Hb (Fig. 2), indicating that this protein has the samefunction as that of mammalian Hp: to inhibit the oxidativereactions of free Hb.

Our finding that fish HpL binds Hb is in agreement withearlier reports, showing that the binding of mammalian Hp to Hbis mediated by the �-chain (the SP domain) (20). The functionof the �-chain of mammalian Hp is unknown. This part of theprotein possibly mediates the binding of Hp to CD11b�CD18(21), an �M-�2 integrin present on monocytes, granulocytes,and natural killer cells; this type of integrin is not present inT. rubripes (22).

Hp in Birds. Among birds, only the genome of chicken (Gallusgallus) has been fully sequenced. In this genome, we could notfind any gene coding for a protein similar to mammalian Hp. Bybiochemical means, however, we identified PIT54 (or 18-B), aprotein homologous to the scavenger receptor cysteine-richfamily of proteins, as the major Hb-binding protein in bothchicken and goose plasma. Whether also the goose genome lacksthe Hp gene remains to be determined. Interestingly, CD163,which has been shown to be the receptor for Hp–Hb complexeson human monocytes, is a member of the same protein family asPIT54 (23). Consistent with a role as a protection against freeHb, PIT54 has been found to be a potent antioxidant (18).Chicken and goose are representatives of two different neognath

Fig. 4. Phylogenetic analysis using the maximum-parsimony method ofhuman Hp, fish HpLs, and related human and fish proteinases. The amino acidsequences of the SP domain of the following proteins were aligned by usingCLUSTALW and then manually corrected: human (HSa), Hp (gi: 1620396), C1r (gi:115204), C1s (gi: 115205), MASP-1 (gi: 19860140), MASP-2 (gi: 7387859), andMASP-3 (gi: 15088517); MASP-1 (gi: 30089299), MASP-2a (gi: 26106026), andMASP-2b (gi: 30089301) from Lethenteron japonicum (LJa); HpL, MASP-2 (gi:49354669), C1r�sA (gi: 11177022), and C1r�sB (gi: 11177024) from Cyprinuscarpio (CCa); MASP (gi: 6407545) from Triakis scyllium (TSc); HpL, C1r�s (gi:40217256) from Oncorhynchus mykiss (OMy); HpL, C1r�sA* (ensembl: SIN-FRUP00000156151), C1r�sB* (ensembl: SINFRUP00000174510), and MASP-2*(ensembl: SINFRUP00000139601) from T. rubripes (TFu); HpL, MASP-2* (en-sembl: GSTENT00021129001), C1r�sA* (ensembl: GSTENT00015325001), andC1r�sB* (ensembl: GSTENT00015326001) from T. nigroviridis (TNi); HpL,MASP2a* (ensembl: ENSDARP00000049759), MASP2b* (ensembl: ENS-DARP00000028217), and C1r�s* (ensembl: ENSDARP00000017015) from D.rerio (DRe). The names of protein sequences indicated by a star were assignedbased on the following rules: C1r�s for the proteinases that were most similarto either mammalian C1r or C1s and MASP-2 for the proteinases most similarto mammalian MASPs and lacking the histidine-loop cysteins (see text). Anunrooted phylogenetic tree was constructed by using the maximum-parsimony method. Numbers at nodes denote bootstrap values with 1,000replicates. The Hp, MASP, and C1r�s families of proteins are circled.

Fig. 5. Synteny map of human Hp, fish HpL of T. nigroviridis, T. rubripes,and D. rerio loci, and the corresponding regions of amphibian X. tropicalisand chicken (G. gallus) genome, according to the ENSEMBL program (www.ensembl.org). Hpr, Hp-related; DHX38, ATP-dependent RNA helicase;TXNL4B, thioredoxin-like protein 4B; ST5A1, sulfotransferase 5A1; DPEP1,dehydropeptidase 1; PKD1L3, polycystic kidney disease 1-like 3; ?, genes ofunknown function.

Wicher and Fries PNAS � March 14, 2006 � vol. 103 � no. 11 � 4171

EVO

LUTI

ON

Dow

nloa

ded

by g

uest

on

Dec

embe

r 10

, 202

0

Page 5: Haptoglobin, a hemoglobin-binding plasma protein, is ... · Haptoglobin, a hemoglobin-binding plasma protein, is present in bony fish and mammals but not in frog and chicken Krzysztof

orders. The absence of Hp in their plasma suggests that the Hpgene was lost at least by the common ancestor of these orders.

Ostriches represent a line of modern birds (the paleognaths)that seems to have separated 80 million to 90 million years agofrom the neognathous birds (24). Using immobilized Hb, weisolated three polypeptides from ostrich plasma with the appar-ent molecular masses of 15, 36, and 75 kDa (Fig. 3). The 36-kDapolypeptide, which was found to be homologous to the �-chainof Hp, formed a disulfide-linked complex of 100 kDa with the15-kDa polypeptide. This finding suggests that the complex is aheterotetramer as is the case for mammalian Hp (type 1). Fromthe 75-kDa polypeptide, we obtained a partial amino acidsequence similar to that of chicken PIT54. A PIT54-like proteinhas also been found in emu, another paleognathous bird (25).

Based on the currently available genomic data, PIT54 seemsto exist only in birds. It is possible that this protein first appearedin an ancestor of paleognathous and neognathous birds, whichalso possessed Hp. The paleognaths then retained both proteinswhereas at least one of the neognaths lost Hp, making PIT54 theonly Hb-binding protein in these animals. We are not aware ofany other case where the function of one gene has been takenover by another, completely unrelated gene.

Evolution of Hp. On the basis of sequence similarity, mammalianHps and fish HpLs seem to be related to the SPs that activate thecomplement system: C1r, C1s, MASP-1, MASP-2, and MASP-3.All of these proteinases consist of six domains: two C1r�C1s�Uegf�bone morphogenic protein 1 domains, one epidermalgrowth factor-like protein domain, two CCP domains, and oneSP domain. C1r, C1s, and MASP-2 are encoded by separategenes, whereas MASP-1 and MASP-3 result from alternativesplicing of the MASP-1�3 gene, which has two different regions,each coding for a SP domain. The SP domain of MASP-3 appearsto have arisen by retrotransposition of a fragment of MASP1mRNA. MASP-1 can be traced back to the urochordates and hasbeen suggested to be the prototype of the family. This proteindiffers from Hp, C1r, C1s, MASP-2, and MASP-3 in that its SPdomain is encoded by multiple exons and has a histidine-loopcysteine bridge. Based on these differences, it is likely that Hp,C1r, C1s, and MASP-2 have a common ancestor, whose geneappeared through a partial duplication of an ancestral MASP-1�3 (26). Indeed, our phylogenetic analysis of the human pro-teins placed Hp, MASP-2, MASP-3, C1r, and C1s on the samebranch (Fig. 8, which is published as supporting information onthe PNAS web site).

Despite the differences at both the DNA and protein level, fishHpLs and mammalian Hps seem to be orthologs. This notion issupported by two observations. First, our phylogenetic analysisindicated that their SP domain originates from a commonancestor (Figs. 4 and 7). Second, we found that, in all analyzedfish and mammals, the gene coding for the respective protein islocated close to the DHODH gene (Fig. 5), indicating conservedsynteny.

The evolution of fish HpL and mammalian Hp could beexplained by the following sequence of events (Fig. 6). Throughpartial duplication (step I) of the gene coding for the commonancestor of Hp, C1r, C1s, and MASP-2 [anc(MASP-2�C1r�C1s�Hp)], the Hp prototype appeared. It had a domain organizationsimilar to that of mammalian Hp (consisting of both a CPP anda SP domain) but was still an active proteinase. Subsequently, theessential serine (S) and histidine (H) residues in the active siteof its SP domain were substituted with alanine (A) and eitherlysine (K) or arginine (R) residues (step II), resulting in aproteolytically inactive protein. All of these events occurred ina common ancestor of bony fish and mammals. In the fishlineage, the CCP domain of Hp was then deleted (step III), andan intron was inserted in the exon coding for the SP domain (stepIV), giving rise to the present HpL gene.

In mammalian genomes, the Hp gene is f lanked by DHODHand DHX38 (Fig. 5). In chicken and X. tropicalis, which lack Hp,the latter two genes are located closely together, indicating thatthe Hp gene has been deleted. Given the fact that amphibiansemerged before the divergence of reptiles�birds and mammals,this deletion must have occurred independently in the amphibianand neognathan lineages. Alternatively, Hp has been lost beforethe emergence of amphibians and has then reappeared inde-pendently in mammals and paleognaths. This possibility is lesslikely in view of the fact that we found a Hb-binding proteinsimilar to the �-chain of mammalian Hp in serum of Chinesesoft-shell turtle (data not shown); turtles seem to represent aseparate reptilian class (27).

ConclusionsOur results show that the Hp gene appeared early in vertebrateevolution, between the emergence of urochordates and bonyfish. During this period, blood cells with high concentrations ofHb also seem to have emerged (28), supporting the idea that thefunction of early Hp, like its modern counterpart, was to protectagainst the oxidative damage caused by released Hb. Lamprey,a jawless fish, is the most primitive extant vertebrate witherythrocytes containing a high Hb concentration. When itsgenome sequence becomes available, it will be interesting todetermine whether it has a gene coding for a HpL. Our resultsalso show that in two classes of vertebrates (amphibians andneognathous birds) at least one species has lost the Hp gene. Theconditions that induced this change remain to be elucidated.

Fig. 6. Possible course of HpL and Hp evolution. MASP proteins and C1r andC1s consist of two C1r�C1s�Uegf�bone morphogenic protein 1 (CUB) domains,one epidermal growth factor (EGF) domain, two CCP domains, and one SPdomain (details in text). Through partial duplication (I) of a gene coding for aHp, C1r, C1s, and MASP-2 ancestor [anc(MASP-2�C1r�C1s�Hp)], a gene codingfor a Hp prototype (prtHp) arose. Subsequently, mutations of the essential His(H) and Ser (S) residues in the active site of the SP domain (II) gave rise to aproteolytically inactive protein (ancHp). In fish, the CCP domain was thendeleted (III), and an intron was inserted into the exon coding for the SP domain(IV), resulting in the appearance of present-day HpL. Mammalian Hp retainedthe CCP and the SP domain. Solid and dashed lines under the SP domainrepresent exons and introns, respectively, of the gene fragment coding for thisdomain.

4172 � www.pnas.org�cgi�doi�10.1073�pnas.0508723103 Wicher and Fries

Dow

nloa

ded

by g

uest

on

Dec

embe

r 10

, 202

0

Page 6: Haptoglobin, a hemoglobin-binding plasma protein, is ... · Haptoglobin, a hemoglobin-binding plasma protein, is present in bony fish and mammals but not in frog and chicken Krzysztof

Materials and MethodsBlood and Plasma Samples. Chicken (Gallus domesticus) blood andkoi carp (C. carpio) blood was from the National VeterinaryInstitute (Uppsala, Sweden). Common ostrich (Struthio camelus)blood was from a local ostrich farm, and African clawed frog (X.laevis) blood was a gift from O. Wrange (Karolinska Institute,Stockholm). Plasma from domesticated goose (Anser anser) andChinese soft-shell turtle (Trionyx sinensis) were from Sigma.Plasma of Japanese puffer fish (T. rubripes) was a gift from T.Miyadai (Fukui Prefectural University, Fukui, Japan).

Genes and Genomes. The genome sequences of baker’s yeast (S.cerevisiae), a nematode (C. elegans), honey bee (A. mellifera), fruitfly (D. melanogaster), sea squirt (C. intestinalis), spotted greenpufferfish (T. nigroviridis), Japanese pufferfish (T. rubripes),zebrafish (D. rerio), chicken (red jungle fowl, G. gallus), and humanwere analyzed with tools provided by the ENSEMBL project of theEuropean Molecular Biology Laboratory–European Bioinformat-ics Institute and the Wellcome Trust Sanger Institute (www.ensembl.org). The genome of the African clawed frog (X. tropicalis)was studied by using tools of the Department of Energy JointGenome Institute portal (http:��genome.jgi-psf.org). Peptide andDNA databases were searched by using BLASTP and TBLASTN tools,respectively, with standard settings and the protein sequence ofhuman Hp1 (gi:184316) as the query. The complete Hp genes ofD. rerio and T. rubripes were predicted from their genomic se-quences by using the WISE2 program (www.ebi.ac.uk�Wise2). Pro-tein sequences of previously known mammalian Hps were obtainedfrom the UniProt database (Universal Protein Resource, www.uniprot.org). Additional full-length fish Hp sequences were in-ferred from nucleotide sequences obtained from GenBank (www.ncbi.nih.gov) or the ENSEMBL project (Data Set 1).

Protein Sequence Analysis and Phylogeny. Signal peptides werepredicted by using SIGNALP 3.0 (www.cbs.dtu.dk�services�SignalP) (29). Alignment of amino acid sequences was per-formed by using the CLUSTALW program (30) and then man-ually corrected. Unrooted phylogenetic trees were constructedwith the distance neighbor-joining method and the maximum-parsimony method, using the NEIGHBOR and PROTPARSprograms from the PHYLIP 3.6 package (http:��evolution.genetics.washington.edu�phylip.html), respectively. Bootstrapanalysis was performed with 1,000 random replicates. TheCONSENS program was used to compute the consensus trees by

using the majority rule, and these were visualized by using theDRAWTREE program from the same package.

Isolation of Hb-Binding Proteins. Freshly isolated erythrocytesobtained from various species, as indicated, were washed fivetimes with PBS and suspended in 5 vol of ice-cold 10 mMTris�HCl, pH 8.0. The lysate was then centrifuged for 20 min at40,000 � g at 4°C. The supernatant was collected, and Hb wasisolated by gel filtration on a Superose 12 column. Fractionscontaining the protein were pooled and stored at �20°C.Purified Hb was coupled to CNBr-Sepharose (Amersham Phar-macia Biosciences) according to the manufacturer’s protocol.For the isolation of Hb-binding proteins, serum from variousspecies, as indicated, was diluted three times with PBS andallowed to flow through the Hb-containing gel twice. The gel wasthen washed with 25 vol of PBS, and bound proteins were elutedwith urea.

Identification of Proteins. Proteins were separated by SDS�PAGEin 12.5% gels and stained with colloidal Coomassie brilliant blue,and selected bands were excised. After trypsin treatment, pep-tide masses were determined with a MALDI-TOF mass spec-trometer and analyzed with the Mascot Search Engine of MatrixScience, London (www.matrixscience.com). MALDI-TOF anal-yses were performed at the Expression Proteomics Laboratoryin Uppsala. Liquid chromatography tandem MS analyses wereperformed at the Protein Analysis Center (Karolinska Institute,Stockholm). N-terminal protein sequence analysis was per-formed by BioCentrum (Krakow, Poland).

Cell Cultures and Transfections. Human embryonic kidney cells(HEK293) and green monkey kidney endothelial cells (COS-7)were grown in complete DMEM supplemented with 10% FBS,glutamine, and antibiotics. Cells were transfected with 2 �g DNAby using Lipofectine (Invitrogen).

We thank T. Miyadai, J. Hardig (National Veterinary Institute, Uppsala,Sweden), O. Wrange, and P. Stenius (Stigtomta Struts, Stigtomta, Sweden)for providing plasma and blood samples and H. Schiot, O. Zetterqvist, andH. Ronne for critical comments on the manuscript. This work was supportedby the Swedish Research Council and the E. O. and Edla JohanssonScientific Foundation. MALDI-TOF analyses performed at the ExpressionProteomics Laboratory in Uppsala were supported by Wallenberg Consor-tium North Grant KAW 2000.0013 from the Knut and Alice WallenbergFoundation.

1. Hawkey, C. M., Bennett, P. M., Gascoyne, S. C., Hart, M. G. & Kirkwood, J. K.(1991) Br. J. Haematol. 77, 392–397.

2. Dhaliwal, G., Cornett, P. A. & Tierney, L. M., Jr. (2004) Am. Fam. Physician69, 2599–2606.

3. Gutteridge, J. M. (1987) Biochim. Biophys. Acta 917, 219–223.4. Lim, Y. K., Jenner, A., Ali, A. B., Wang, Y., Hsu, S. I., Chong, S. M., Baumman,

H., Halliwell, B. & Lim, S. K. (2000) Kidney Int. 58, 1033–1044.5. Javid, J. (1978) Curr. Top. Hematol. 1, 151–192.6. Delers, F., Musquera, S., Domingo, M. & Lombart, C. (1983) Comp. Biochem.

Physiol. A 74, 245–248.7. Francis, R. T., Jr. & Becker, R. R. (1984) Comp. Biochem. Physiol. B 77, 341–347.8. Hanley, J. M., Haugen, T. H. & Heath, E. C. (1983) J. Biol. Chem. 258,

7858–7869.9. Heinderyckx, M., Jacobs, P. & Bollen, A. (1988) Mol. Biol. Rep. 13, 225–232.

10. Lehtinen, M. J., Meri, S. & Jokiranta, T. S. (2004) J. Mol. Biol. 344, 1385–1396.11. Bowman, B. H. & Kurosky, A. (1982) Adv. Hum. Genet. 12, 189–261.12. Kurosky, A., Barnett, D. R., Lee, T. H., Touchstone, B., Hay, R. E., Arnott,

M. S., Bowman, B. H. & Fitch, W. M. (1980) Proc. Natl. Acad. Sci. USA 77,3388–3392.

13. Maeda, N., Yang, F., Barnett, D. R., Bowman, B. H. & Smithies, O. (1984)Nature 309, 131–135.

14. Maeda, N. & Smithies, O. (1986) Annu. Rev. Genet. 20, 81–108.15. Tosi, M., Duponchel, C., Meo, T. & Couture-Tosi, E. (1989) J. Mol. Biol. 208,

709–714.

16. Seidah, N. G. & Chretien, M. (1999) Brain Res. 848, 45–62.17. Bennu, D. A. (2004) Lab. Anim. 33, 42–48.18. Iwasaki, K., Morimatsu, M., Inanami, O., Uchida, E., Syuto, B., Kuwabara, M.

& Niiyama, M. (2001) J. Biol. Chem. 276, 9400–9405.19. Wicher, K. B. & Fries, E. (2004) Proc. Natl. Acad. Sci. USA 101, 14390–

14395.20. Lustbader, J. W., Arcoleo, J. P., Birken, S. & Greer, J. (1983) J. Biol. Chem.

258, 1227–1234.21. El Ghmati, S. M., Van Hoeyveld, E. M., Van Strijp, J. G., Ceuppens, J. L. &

Stevens, E. A. (1996) J. Immunol. 156, 2542–2552.22. Huhtala, M., Heino, J., Casciari, D., de Luise, A. & Johnson, M. S. (2005)

Matrix Biol. 24, 83–95.23. Kristiansen, M., Graversen, J. H., Jacobsen, C., Sonne, O., Hoffman, H. J., Law,

S. K. & Moestrup, S. K. (2001) Nature 409, 198–201.24. Harlid, A., Janke, A. & Arnason, U. (1997) Mol. Biol. Evol. 14, 754–761.25. Mann, K. (2004) Br. Poult. Sci. 45, 483–490.26. Fujita, T., Matsushita, M. & Endo, Y. (2004) Immunol. Rev. 198, 185–202.27. Zardoya, R. & Meyer, A. (2001) Naturwissenschaften 88, 193–200.28. Glomski, C. A., Tamburlin, J. & Chainani, M. (1992) Histol. Histopathol. 7,

501–528.29. Bendtsen, J. D., Nielsen, H., von Heijne, G. & Brunak, S. (2004) J. Mol. Biol.

340, 783–795.30. Higgins, D. G., Thompson, J. D. & Gibson, T. J. (1996) Methods Enzymol. 266,

383–402.

Wicher and Fries PNAS � March 14, 2006 � vol. 103 � no. 11 � 4173

EVO

LUTI

ON

Dow

nloa

ded

by g

uest

on

Dec

embe

r 10

, 202

0