Subunit interactions in Ascaris hemoglobin octamer formation.

The oxygen-avid, perienteric hemoglobin of Ascaris is a homooctamer. Each subunit contains two tandem globin domains that are highly homologous with the exception of a charged COOH-terminal extension. In solution, recombinant domain one (D1) exists as a monomer, whereas recombinant domain two with the COOH-terminal tail (D2) is primarily an octamer. To examine the role of the COOH-terminal extension in Ascaris hemoglobin multimer formation, we attached the tail to the monomeric, heme-containing proteins, myoglobin and D1; neither construct was capable of multimer formation. Additionally, we removed the tail from both full-length Ascaris hemoglobin and D2. This substantially decreased, but did not eliminate, multimerization. We further characterized subunit interactions by disrupting full-length hemoglobin multimers with the chaotropic salt, NaSCN, which yielded intermediate oligomers. In solution, D2 demonstrated a greater propensity to dissociate than full-length hemoglobin, indicating that D1 contributes to octamer stability. D1 formed a weak dimer in its crystal; thus, we analyzed interactions along the subunit interface. Hydrogen bonds as well as hydrophobic and electrostatic forces appeared to contribute to dimer formation. Amino acid substitutions along this interface in D2 are predicted to enhance subunit interactions for that domain. Our studies reveal that the COOH-terminal tail is necessary, but not sufficient, for efficient octamer formation. Other regions, possibly within the E- and F-helices and AB loops of both domains, appear to be important for Ascaris hemoglobin octamer formation.

The perienteric fluid of the parasitic nematode Ascaris contains an abundant hemoglobin with exceptional oxygen avidity (1). This globin is an octamer composed of identical 43-kDa polypeptides (2,3). Each subunit consists of two tandem globin domains. A highly charged tail is found at the COOH terminus of the second domain ( Fig. 1A) (4,5). The crystal structure of the first globin domain has been solved, revealing striking similarities to the tertiary structures of globins from diverse species (6). The heme pocket contains a network of hydrogen bonds, including one contributed by a critical tyrosine, which locks the liganded oxygen in place. Mutagenesis studies have provided further support for this (7,8).
Although the molecular basis of the oxygen avidity of Ascaris hemoglobin has now been described in some detail, structural motifs responsible for the formation of this remarkable multimer have not yet been elucidated. However, molecular cloning and expression of the separated domains have provided interesting clues (9). Domain one (D1), which does not have the COOH-terminal extension, exists as a monomer in solution. In contrast, domain two (D2), which contains the tail, is capable of polymerizing to an octamer. Excluding the tail, the primary structures of D1 and D2 are highly homologous with 63% amino acid identity, implicating the tail as a potentially important component in multimerization (4,5). The 23-residue COOH-terminal extension is highly polar, possessing 22 charged residues, including four His-Lys-Glu-Glu (HKEE) direct repeats. It has been suggested that this tail may serve as a polar zipper, forming a ␤-barrel comprised of eight antiparallel ␤-strands stabilized by interstrand salt bridges between side chains (8,10). Electron microscopy and cross-linking studies have indicated that the octamer conforms to a two-layered arrangement of subunits, perhaps with one type of interaction of subunits within a layer and another type between the two layers (3). In this study, we have characterized the interaction of subunits and have assessed the contribution of the highly charged tail in Ascaris hemoglobin octamer formation.

EXPERIMENTAL PROCEDURES
Ascaris Hemoglobin Constructs-Ascaris hemoglobin lacking the COOH-terminal tail (AHϪt) was engineered using the full-length expression plasmid insert (AH) (9) as a template for polymerase chain reaction (PCR). 1 Primers contained NcoI restriction sites for insertion into the expression vector. The forward primer was the same used for amplification of the full-length construct as described previously (9). The AHϪt reverse primer (5Ј gcatccatggtcatttgccgtgccttgtggc 3Ј) was complementary to the region just upstream of the sequence encoding the COOH-terminal tail and also provided a stop codon for translation termination.
D2 lacking the COOH-terminal tail (D2Ϫt) was constructed using the D2 expression plasmid (9) as a template in PCR. The D2Ϫt forward primer was the same used for amplification of D2 (9). The D2Ϫt reverse primer was the one used above for the formation of AHϪt.
Addition of the COOH-terminal tail to D1 in order to create D1ϩt was performed utilizing a two-step PCR-based strategy modified from Morrison and Desrosiers (11) (Fig. 1B). In the first PCR step, Reaction I used the D1 expression plasmid as template (9). Primer a, the forward primer, contained an NcoI restriction site and was the same used for amplification of D1 as described previously (9). Primer b, the reverse primer in this reaction (5Ј ttctttatgctcgtggtgttctcgtccatgcttgttgatttc 3Ј), contained 21 nucleotides complementary to the extreme 3Ј end of the coding sequence for D1, plus 21 nucleotides complementary to the extreme 5Ј coding sequence for the tail. In Reaction II, the D2 expression plasmid was used as template. Primer bЈ, the complement to Primer b, was used as the forward primer. Primer c, the reverse primer, contained an NcoI restriction site and was same utilized for the amplification of D2 as described previously (9). Reactions I and II resulted in products with a region of shared overlap with one another. In order to * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
generate full-length D1ϩt, a 10-fold dilution of both products was used as template in Reaction III with only the extreme 5Ј and 3Ј end-specific oligonucleotides, Primers a and c. In the first round of amplification, the single-stranded templates hybridized in the region of shared overlap and self-primed the extension reaction, thus creating the desired fulllength double-stranded insert. In all subsequent rounds, the 5Ј-and 3Ј-specific primers amplified this product. PCR was performed in a TC480 thermal cycler (Perkin-Elmer). Final reaction conditions in all PCR were 1.25 units of AmpliTaq (Perkin-Elmer), 0.1 M oligonucleotide primers, 20 mM Tris-HCl (pH 8.2), 10 mM KCl, 6 mM (NH 4 ) 2 SO 4 , 2 mM MgCl 2 , 0.1% Triton X-100, 10 g/ml nuclease-free bovine serum albumin, 0.2 mM deoxynucleotides, and the relevant template in a final volume of 0.05 ml. Reactions were cycled 30 times at 95°C for 1 min, 50 for 1 min, and 75 for 1 min. After ethanol precipitation, PCR products were digested at 37 for 16 h with 10 units of NcoI (Boehringer Mannheim). The DNA was then purified by adsorption to a glass slurry using the Elu-Quik DNA purification kit (Schleicher & Schuell) and cloned into NcoI-digested, phosphatasetreated Studier expression vector pET-8C. Phosphatase reactions were carried out at for 1 h using 1 unit of calf intestinal phosphatase Expression constructs were transformed into Escherichia coli strain XL1-Blue for plasmid screening. Positive clones were selected on LBampicillin plates and were screened for forward orientation by PCR analysis. Positives were verified for the expected mutations by plasmid DNA sequencing with the Sequenase kit (U. S. Biochemical Corp.), and the products were resolved on 6% polyacrylamide gels (National Diagnostics).
Myoglobin Construct-Addition of the COOH-terminal tail of Ascaris hemoglobin to dog myoglobin in order to create Mgbϩt was accomplished using a similar PCR-based strategy with the following changes. In Reaction I, dog myoglobin cDNA was used as a template. The forward primer (5Ј ccatggggctcagcgacggggaatg 3Ј) contained an NcoI restriction site. The reverse primer (5Ј tcgtggtgttctttgccgtggccctggaagcccagctcct 3Ј) contained 20 nucleotides complementary to the extreme 3Ј end of dog myoglobin, plus 20 nucleotides complementary to the extreme 5Ј end of the Ascaris hemoglobin COOH-terminal tail. The complement of this primer was used as the forward primer in Reaction II with the D2 plasmid as template and reverse primer for D2 as above. A 10-fold dilution of both products was used in Reaction III with the extreme 5Ј and 3Ј oligonucleotides, resulting in Mgbϩt. PCR conditions, cloning, and subsequent analyses were performed essentially as described above for the Ascaris hemoglobin constructs.
Protein Expression and Purification-Constructs were transformed into E. coli expression strain BL21 (DE3), and protein expression was carried out as described previously (9), except that cultures were grown for 48 h before induction with isopropyl-1-thio-␤-D-galactopyranoside and then grown an additional 14 -16 h before harvesting bacteria. Protein purification was performed as before (9).
Further purification of hemoglobins for NaSCN dissociation experiments was accomplished by dilution of the samples 1:40 in 20 mM Tris (pH 7.5) and fractionation using Mono Q anion exchange chromatography. A Pharmacia Mono Q HR 5/5 column was equilibrated with 25 mM Tris (pH 7.5), 50 mM NaCl for 15 min, and samples were eluted with a linear gradient from 50 mM to 1000 mM NaCl for 30 min at a flow rate of 1 ml/min. Samples were then concentrated using Centriprep 10 concentrators.
Quaternary Structure Determination-Samples were injected onto a Pharmacia Superose 12 gel filtration column equilibrated with phosphate-buffered saline. Hemoglobin peaks were detected by monitoring absorbance at 410 nm. In order to determine molecular sizes, elution times were calibrated with protein molecular weight standards (Bio-Rad).
Dissociation of Multimers-Ten micrograms of purified hemoglobin were incubated with NaSCN in 20 mM Tris (pH 7.5) at room temperature in a total volume of 24 l. In experiments varying the concentration of NaSCN, incubations were carried out for 20 min. In experiments varying the length of incubation, reactions contained a final concentration of 2 M NaSCN.
Analysis of the D1 Dimer Interface-The crystal structure of D1 (Brookhaven 1ASH) (6) was analyzed on a Silicon Graphics Indigo workstation using the graphics program TURBO to determine the amino acid contacts at the dimer interface.

Analysis of Myoglobin with a Tail-
The purpose of this experiment was to assess the ability of the highly charged Ascaris hemoglobin tail to promote multimer formation of a structur-FIG. 1. Ascaris hemoglobin constructs. A, the primary structures of D1 and D2 are aligned with the charged COOH-terminal tail highlighted in bold (4,9). D1 contains amino acids 1-149 (M r ϳ17,900). D2 contains 151-320 (M r ϳ20,600). B, a two-step PCR strategy was utilized in order to engineer D1 with the tail (D1ϩt). In the first step, complementary oligonucleotide primers (Primers b and bЈ) containing regions complementary to both the extreme 3Ј end of D1 and the extreme 5Ј end of the tail were used in separate reactions with primers complementary to the 5Ј end of D1 (Primer a in Reaction I) or the 3Ј end of the COOHterminal tail (Primer c in Reaction II). The products, containing a region of shared overlap, were utilized as template in the next step with only the extreme 5Ј and 3Ј end-specific primers (Primers a and c) to generate full-length D1ϩt. ally related, monomeric protein. Secondary and tertiary structures of mammalian myoglobins and Ascaris hemoglobin domain one are very similar, despite low amino acid homology (6,12). A two-step PCR strategy as described above was utilized to add the Ascaris hemoglobin tail to the COOH terminus of dog myoglobin, resulting in Mgbϩt. The mutant protein was expressed in E. coli and migrated on SDS-polyacrylamide gel electrophoresis as predicted, based on molecular mass (data not shown). Recombinant Mgbϩt had the visible absorbance spectrum characteristic of a myoglobin, indicating the presence of a properly folded, functional, heme-containing protein. In order to assess quaternary structure, the recombinant protein was analyzed using gel filtration chromatography. Mgbϩt eluted at the position expected for a monomer (21 kDa), just ahead of myoglobin without a tail (17.5 kDa), with no detectable multimeric species (Fig. 2). The minor lower molecular weight peak is likely due to degraded recombinant protein.
Analysis of Ascaris Hemoglobin Constructs-To determine the role of the COOH-terminal tail in Ascaris hemoglobin multimerization, constructs of full-length hemoglobin and the separated domains, D1 and D2, were created using PCR (9). D1 plus the COOH-terminal tail (D1ϩt), D2 lacking its tail (D2Ϫt), and full-length hemoglobin lacking the tail (AHϪt) were also constructed ( Fig. 1). Mutant proteins were expressed in E. coli, purified, and analyzed by gel filtration chromatography. As expected, D1 was predominantly monomeric (Fig. 3A). Fulllength hemoglobin (AH) eluted at the position expected for an octamer (Fig. 3B). At the same concentration, D2 was predominantly octameric, but also formed detectable amounts of dimer and an additional multimeric species, most likely tetramer (Fig. 3C). AHϪt was drastically reduced in its ability to form an octamer (Fig. 3B). Elution times indicated the presence of a variety of species, including monomer, dimer, and tetramer with little octamer. D2Ϫt was similarly reduced in its ability to multimerize (Fig. 3C). Major peaks were detected at elution times representing primarily dimer, and to a lesser extent, monomer. Minor peaks at elution times corresponding to tetramer and octamer were also detected. In both tail-less constructs and D2, a substantial amount of high molecular weight aggregate was observed. D1ϩt remained monomeric with an elution profile almost indistinguishable from that of D1 (Fig. 3A).
Dissociation of Full-length Ascaris Hemoglobin-In order to gain insight into the nature of the interaction(s) involved in octamer formation, NaSCN, a chaotropic salt, was utilized to dissociate Ascaris hemoglobin multimers. For these experiments, full-length recombinant hemoglobin was incubated in the presence of varying concentrations of NaSCN. After 20 min, samples were subjected to gel filtration chromatography to assess quaternary structure (Fig. 4). Hemoglobin peaks were detected by monitoring absorbance at 410 nm in order to selectively detect only properly folded molecules. As expected, Ascaris hemoglobin with no added NaSCN eluted at a position consistent with an octamer. In the presence of 1 M NaSCN, analysis revealed shoulders at tetramer and dimer positions. However, a further increase to 1.5 M NaSCN resulted in a substantial decrease in the elution peak corresponding to octamer with a concomitant rise in peaks representing tetramer and dimer. A small peak corresponding to monomer (M r ϳ43,000) was also detected. Dissociation was accentuated in the presence of 2 M NaSCN; elution times suggested a further decrease in octamer along with a marked rise in dimer.
In the next part of this experiment, purified full-length hemoglobin was incubated in the presence of 2 M NaSCN for increasing amounts of time to induce further dissociation of multimers. Quaternary structure was again assessed by gel filtration chromatography (Fig. 5). At 20 min, as in Fig. 4D, substantial dissociation into dimer and tetramer was observed. By 60 min, there was additional loss of octamer and tetramer populations with a striking increase in dimer. Upon incubation FIG. 2. Quaternary structure of myoglobin with the tail. Mgbϩt (monomer mass, 21 kDa) was injected onto a Superose 12 gel filtration column as described under "Experimental Procedures." Elution times (in minutes) were compared with those of molecular weight standards. The 17.5-kDa standard corresponds to myoglobin without a tail. Myoglobin elution from the column was monitored by absorbance at 410 nm. for 90 min, more substantial dissociation into monomer was observed. Similar results were obtained upon incubation for 135 min (data not shown). After 180 min, no peaks were observed, due to subunit denaturation with subsequent loss of the heme moiety.
D1 Dimer Interface Analysis-Although D1 exists as a monomer in solution (9), it forms a weak dimer in the crystalline lattice. The largest intermolecular contact, approximately 1100 Å 2 in area, occurs across a crystallographic dyad and is three times larger than any other contacts. The interface of the D1 dimer is at the E-and F-helices of each subunit with additional contacts from residues in the A-helix and AB-loop to the E-and F-helices (Fig. 6). Significant amino acid interactions between D1 subunits are summarized in Table I. The arginine residue at position 3 (Arg 3 ) in the A-helix forms a strong ionic interaction with Asp 73 Ј in the E-helix, whereas Glu 90 and Arg 94 Ј interact more weakly. The main chain amide and carbonyl oxygen atoms of Val 15 form hydrogen bonds with the side chains of Glu 82 Ј and Arg 89 Ј, respectively. A number of additional side chains are located within the interface and make contact with residues in the other subunit, including Val 73 with Val 73 Ј and Ala 86 with Val 15 Ј and Leu 69 Ј.

DISCUSSION
Recently, the crystal structure of Ascaris hemoglobin domain one was reported (6). However, because D1 alone does not form multimers in solution (9), this structure has not defined regions of the full-length molecule involved in octamer formation. The two tandem globin folds are highly homologous with the exception of a highly charged, 23 residue COOH-terminal extension on D2, which contains 22 ionizable residues, including four direct HKEE repeats (4, 5). Perutz and colleagues (5) have suggested that this tail may be capable of forming an eightstranded ␤-barrel. Thus, we targeted the Ascaris hemoglobin tail as a potentially important structure required for octamer formation.
We first tested the ability of the tail to directly promote multimerization of a structurally related but monomeric protein. Myoglobin served as an appropriate candidate, due to its homologous compact, highly ␣-helical structure (6,12). The engineered myoglobin, Mgbϩt, was incapable of forming multimers. This raised the possibility that the myoglobin either contained regions that interfered with multimerization or lacked certain motifs necessary for multimer formation. In order to address this, we constructed a mutant of a completely unrelated molecule, glutathione S-transferase, which upon addition of the charged tail was also incapable of forming multimers (data not shown). Taken together, these data suggest that interaction of other regions in the Ascaris hemoglobin molecule are essential for multimerization. Indeed, construction and analysis of D1ϩt demonstrated that the tail alone is unable to promote multimer formation of the monomeric, Ascaris hemoglobin domain. In contrast, full-length hemoglobin without a tail, as well as D2 without a tail, were still capable of forming the full complement of multimers, albeit in a poorly controlled fashion. The most abundantly observed multimer in these tail-less hemoglobins was a dimer, suggesting that the regions responsible for dimer formation lie within the globin fold of D2. Based on these data, we propose that the ability to multimerize is intrinsic to the globin region of D2 and that the COOH-terminal tail serves to stabilize the Ascaris hemoglobin octamer.
Previously, Darawshe and Daniel (3) proposed that Ascaris hemoglobin adopts a two-layered arrangement with four monomers in each layer. Using cross-linking analysis, the distribution of crosslinked species corresponded to theoretical values predicted from a two-layered arrangement with one type of interaction within a layer and another type between layers. Based on these results, the investigators suggested that a disruptive agent could possibly discriminate between these interactions, yielding tetramer and dimer intermediates. Additionally, the polar zipper model of Perutz and co-workers (5, 10) aligns four monomers on either side of the ␤-barrel, consistent with the arrangement proposed by Darawshe and Daniel (3). A previous investigation using pH extremes showed dissociation of Ascaris hemoglobin directly into dimers and monomers (3). In our studies, a chaotropic salt, NaSCN, served as a more selective reagent for disruption of the octamer. If only one type of interaction was responsible for octamer formation, then the multimer should dissociate directly into monomers. However, gel filtration analysis yielded elution times corresponding to dimer as well as tetramer, indicating that several types of interactions exist within the octamer. In our NaSCN experiments, it is possible that some subunit reassociation occurred during gel filtration analysis. Regardless, the detection of intermediate species indicates the existence of multiple types of interactions between subunits. These data, combined with our analysis of the tail-less mutants, suggest that interactions between the globin folds, in addition to those between the charged tails, are important for octamer assembly.
Although D2 is certainly capable of multimerization, its ability to form an octamer is reduced in comparison to the fulllength molecule, demonstrated by the presence of detectable intermediates. Furthermore, upon dilution, D2 readily dissociated, whereas the full-length molecule remained octameric (data not shown). These results suggest that the first domain in native Ascaris hemoglobin may contribute additional interactions required for the formation of a stable octamer, despite the inability of separated D1 with the tail to multimerize. The fact that D1 forms a weak dimer in the crystal provides further support for this hypothesis. Distinct from vertebrate hemoglobin tetramer interactions (12), the D1 interface is formed by the E-and F-helices along with the AB loop. Examination of surface residues at the D1 dimer interface revealed numerous favorable hydrophobic and electrostatic interactions. This dimer may represent actual molecular interactions that occur between D1 subunits within native Ascaris hemoglobin in solution. Furthermore, an analogous interface may serve as a major site of D2 subunit interactions. Interestingly, Val 73 and Ala 86 in D1 are replaced with leucine and methionine, respectively, in D2. Both of these substitutions result in side chains with greater hydrophobic surface areas than D1, which could serve as a strong driving force promoting multimer formation.
In this study, mutagenesis experiments demonstrated that the COOH-terminal tail is required for stable octamer formation, but is not sufficient to promote multimerization alone. Dissociation experiments and analysis of the D1 crystalline dimer provided further insight into subunit interactions. Unquestionably, solution of the crystal structure of full-length Ascaris hemoglobin or D2 will provide the most direct approach for elucidating the structural basis of octamer formation. This will allow direct visualization of quaternary structure, including relevant contacts between globin domains, as well as interactions between the charged COOH-terminal tails.