Guinea pig (Cavia porcellus) blood in Alsever solution was obtained from Charles River Laboratories (Sulzfeld, Germany). Guinea pig hemoglobin was isolated and purified according to Paoli and Nagai (2004) based on the original protocol of Perutz [5], [28]. Briefly, erythrocytes were separated from plasma by centrifugation at 100 g for 15 min. Then erythrocytes were washed with isotonic saline (0.9% NaCl) and subsequently lysed by adding an equal amount of water, resulting in a release of hemoglobin from the cells. After addition of NaCl to a final concentration of 3%, cell debris was removed by centrifugation. As hemoglobin represents 98% of protein in the hemolysate, no further purification was necessary.

Crystallization of guinea pig hemoglobin was performed by hanging-drop vapor diffusion at 20°C. The drops contained 5 µl hemoglobin solution with a concentration of 10 mg/ml and were mixed with 5 µl reservoir solution. Then the drops were equilibrated against 1.0 ml reservoir solution (2.6 M (NH₄)₂SO₄, 100 mM sodium phosphate buffer, pH 6.5).

Prior to data collection the crystals were soaked in mother liquor containing 25% glycerol as cryoprotectant. Crystals were then flashed cooled in the gas stream of a cryostream system (Oxford Cryosystems, Oxford, United Kingdom), with a nitrogen gas temperature of 100 K. Data was collected using a Microstar rotating anode (Bruker AXS, Karlsruhe, Germany) and a “mar345” image plate detector (MARresearch, Norderstedt, Germany). Data was collected for 360 ° with an increment of 1.5 ° and a crystal to detector distance of 120 mm. Data was collected up to a resolution of 1.8 Å and processed with the XDS program package (Version: December 6th 2007) [29]. The space group was determined using the program POINTLESS from the CCP4 program suite [30].

Crystal parameters have been reported before [27]. The structure solution was obtained by molecular replacement with carbonmonoxy horse hemoglobin (PDB-code: 2D5X) as starting model, using the program PHASER implemented in the CCP4 suite [31]. The structure was refined using the program Coot/REFMAC [32]. Data collection and refinement statistics for the final model are presented in Table 1. The final model and structure factor were deposited in the Protein Data Bank with accession code 3HYU.

Molecular graphics were produced using PyMOL Molecular Graphics System (DeLano Scientific, USA).

The structure of guinea pig hemoglobin is a typical vertebrate hemoglobin tetramer made up from two α-subunits and two β-subunits of 141 and 146 amino acids, respectively (Fig. 1). The electron density clearly confirmed the sequence, which had previously been reported for the major hemoglobin component [26]. The α-chain of guinea pig hemoglobin shares 75% identical amino acids with human hemoglobin, while the β-chain is identical with its counterpart in human hemoglobin in 65% of the positions. Absorption spectroscopy of dissolved crystals indicated that guinea pig hemoglobin crystallized as met-hemoglobin [27]. The active site and its surrounding amino acids are typical in comparison with other hemoglobins in the met state.

Hemoglobins in the met-state usually crystallize in the R-state conformation and therefore a R-state conformation of the hemoglobin molecule was also expected for guinea pig met-hemoglobin. Unexpectedly the C-terminal four amino acids of the α-subunits had a very different conformation than normally found in the R-state. Superposition of guinea pig hemoglobin with human hemoglobin in T-, R- or R2-state revealed that guinea pig hemoglobin crystallized in the R2-state (Fig. 1). The R2-state of human hemoglobin features the same special C-terminal conformation of the α-subunit as observed in guinea pig hemoglobin [10]. Until now the R2-state has been only observed in hemoglobin crystals grown under low salt conditions. Hence it has been suggested that the R2-state may be the physiologically relevant liganded end state structure and that the R-state is an intermediate trapped between the R2 and T structures by the high-salt crystallization conditions [10], [11], [12]. Why does guinea pig hemoglobin crystallize in the R2-state under high-salt conditions?

One important factor that may contribute to crystallization in the R2-state is that the R2-state in guinea pig hemoglobin seems to be better stabilized than the R2-state in human hemoglobin. A comparison with R2-states of other species is not possible, since the R2-state has only been reported in human hemoglobin until now. Analysis of the salt bridges present in the tetramer reveals that guinea pig hemoglobin is stabilized by a total of 44 salt bridges in the R2-state, while human hemoglobin in comparison is only stabilized by 41 salt bridges in the R2-state. Out of these salt bridges both hemoglobins have 32 in common, while the remainder is unique to either one of them. The majority of salt bridges is formed between residues of the same subunit (intra-subunit salt bridges) and serves to stabilize a certain subunit. Only a few salt bridges stabilize interactions between adjacent subunits (intersubunit salt bridges) and thus are important for stabilization of the quaternary structure. Both, guinea pig and human hemoglobin, share one conserved salt bridge at the α1/β1-, α2/β2-, α1/β2- and α2/β1-interface in the R2-state. In contrast the β1/β2-interface of human hemoglobin in the R2-state is devoid of intersubunit salt bridges, while in guinea pig hemoglobin two intersubunit salt bridges are present (Fig. 2). They connect the N-terminus of the β1-subunit with the C-terminus of the β2-subunit and vice versa. Specifically they are formed between the N-terminal amino group (Val1) of one β-subunit and the C-terminal carboxyl group (His146) of the adjacent β-subunit. The distance between both charged groups is 2.6 Å. It was not possible to pinpoint a specific amino acid or set of amino acids responsible for the slightly different conformations of the C- and N-termini of the β-subunits in human and guinea pig hemoglobin. Nevertheless, we propose that the two salt bridges at the β1/β2-interface are an important factor, which increases the stability of the R2-state of guinea pig hemoglobin in comparison to human hemoglobin.

Besides stabilization of the R2-state an additional factor that might contribute to the oxygen binding behavior of guinea pig hemoglobin might be a destabilization of the T-state. Baldwin and Chothia described in their analysis of the hemoglobin tetramer flexible joint and switch regions [9]. The switch region involves residues 38–44 of the αl-subunit (C helix and CD corner) together with residues 97–102 of the β2-subunit (FG corner and G helix). In the course of the conformational change His97 of the β2-subunit slides in its position along helix C of the α1-subunit. Specifically, in the T-state His97 of the β2-subunit is positioned between Thr4l and Pro44 of the αl-subunit. Upon transition to the R-state, His97 of the β2-subunit moves one turn along the C helix of the α1-subunit to a position between residues Thr38 and Thr41 [9]. In the R2-state His97 of the β2-subunit slides even further in the same direction and is positioned opposite to Thr38 of the α1-subunit [10]. In human hemoglobin the switch region of the α1-subunit is stabilized by a salt bridge between Glu30 and His50, which connects helix C and helix E.

In the switch region of guinea pig hemoglobin two important differences exist in comparison to human hemoglobin (Fig. 3). Firstly, the salt bridge stabilizing the switch region in human hemoglobin is not present due to two amino acid exchanges (Glu30Thr30 and His50Pro50) in the α1-subunit of guinea pig hemoglobin (Fig. 3). The absence of this salt bridge most probably renders the switch region more flexible in guinea pig hemoglobin. Secondly, in the T-state His97 of the β2-subunit in human hemoglobin is positioned opposite to Pro44 of the α1-subunit. In guinea pig hemoglobin a bulky histidine instead of proline is found at position 44 of the α1-subunit, which inevitably will result in a steric hindrance between His44 and His97 of the β2-subunit (Fig. 3). This steric hindrance renders the T-state of guinea pig hemoglobin less stable and therefore will result in a higher oxygen affinity since the equilibrium between T-, R- and R2-state will be shifted towards the relaxed states (either R- or R2-state), which have a higher oxygen affinity. Human hemoglobin mutants confirm this hypothesis, since the mutants Milledgeville (Pro44Leu) and Kawachi (Pro44Arg) have a strongly increased oxygen affinity [33], [34]. This may explain why Pro44 of the α-subunit is highly conserved in vertebrate hemoglobin. Exceptions are hemoglobins of rat (Rattus norvegicus) and several fish species which have exchanged Pro44 of the α-subunit to a serine. However, no increased oxygen affinity has been reported for the exchange of Pro44Ser most probably due to the fact that serine is much smaller than leucine, arginine or histidine and therefore no steric hinderance exists with His97 of the β2-subunit.

In conclusion we propose that the increased oxygen affinity of guinea pig hemoglobin can be explained by two factors, namely a decreased stability of the T-state and an increased stability of the R2-state. Both factors together serve to shift the equilibrium between the conformational states towards the high affinity relaxed states resulting in an increased oxygen affinity of guinea pig hemoglobin. It is remarkable that guinea pig hemoglobin crystallizes in the R2-state under oxy high salt conditions, which in all other vertebrate hemoglobins provokes formation of crystals in the R-state. This may indicate that the R2-state in guinea pigs has an increased stability and thus is the physiological relaxed state, while in human hemoglobin the R-state or a mixture of R- and R2-state seems to be present as the relaxed state [13]. Surely the increased oxygen affinity of hemoglobin is not the only adaptation to high altitude in guinea pigs, but increased blood oxygen affinity has been proven to be advantageous for animals living at high altitudes, because it increases the oxygen saturation of the arterial blood [1], [2], [3], [19], [35].