The anemia-induced reversible switch from hemoglobin A to hemoglobin C in caprine ruminants: immunochemical evidence that both hemoglobins are found in the same cell

During anemic episodes, goats and certain sheep replace hemoglobin A (HbA = α₂β₂^A) with hemoglobin C (HbC = α₂β₂^C). Rabbit serum directed against either purified sheep HbA or purified sheep HbC was prepared. Both types were used to test whether the two hemoglobins are found in the same cell during switching by an indirect fluorescent antibody assay.Unabsorbed antisheep HbA cross-reacted extensively with goat HbA but to a lesser extent with goat or sheep HbC. Similarly, unabsorbed antisheep HbC reacted with these antigens in the order: Sheep HbC > goat HbC > sheep HbA > goat HbA. Cross-absorption resulted in sera specific either for sheep and goat HbA or for sheep and goat HbC. The specificities were confirmed by indirect fluorescent antibody staining of sheep and goat erythrocytes containing either at least 99% HbA or at least 99% HbC.Smears of erythrocytes from sheep and goats in the process of switching were reacted with one of the absorbed sera then with fluorescein conjugated antirabbit immunoglobulin G. The sum of the fractions stained both by anti-HbA and by anti-HbC exceeded 100% during the switch. Most strikingly when HbA was replacing HbC, nearly all cells stained for HbC while more than half stained for HbA. Thus, the two hemoglobins are found in the same cell during switching.     

Solution-active structural alterations in liganded hemoglobins C (beta6 Glu --> Lys) and S (beta6 Glu --> Val).

Based upon existing crystallographic evidence, HbS, HbC, and HbA have essentially the same molecular structure. However, important areas of the molecule are not well defined crystallographically (e.g. the N-terminal nonhelical portion of the alpha and beta chains), and conformational constraints differ in solution and in the crystalline state. Over the years, our laboratory and others have provided evidence of conformational changes in HbS and, more recently, in HbC. We now present data based upon allosteric perturbation monitored by front-face fluorescence, ultraviolet resonance Raman spectroscopy, circular dichroism, and oxygen equilibrium studies that confirm and significantly expand previous findings suggesting solution-active structural differences in liganded forms of HbS and HbC distal to the site of mutation and involving the 2,3-diphosphoglycerate binding pocket. The liganded forms of these hemoglobins are of significant interest because HbC crystallizes in the erythrocyte in the oxy form, and oxy HbS exhibits increased mechanical precipitability and a high propensity to oxidize. Specific findings are as follows: 1) differences in the intrinsic fluorescence indicate that the Trp microenvironments are more hydrophobic for HbS > HbC > HbA, 2) ultraviolet resonance Raman spectroscopy detects alterations in Tyr hydrogen bonding, in Trp hydrophobicity at the alpha1beta2 interface (beta37), and in the A-helix (alpha14/beta15) of both chains, 3) displacement by inositol hexaphosphate of the Hb-bound 8-hydroxy-1,3,6-pyrenetrisulfonate (the fluorescent 2,3-diphosphoglycerate analog) follows the order HbA > HbS > HbC, and 4) oxygen equilibria measurements indicate a differential allosteric effect by inositol hexaphosphate for HbC approximately HbS > HbA.     

Liquid-liquid phase separation in hemoglobins: distinct aggregation mechanisms of the beta6 mutants

Reversible liquid-liquid (L-L) phase separation in the form of high concentration hemoglobin (Hb) solution droplets is favored in an equilibrium with a low-concentration Hb solution when induced by inositol-hexaphosphate in the presence of polyethylene glycol 4000 at pH 6.35 HEPES (50 mM). The L-L phase separation of Hb serves as a model to elucidate intermolecular interactions that may give rise to accelerated nucleation kinetics of liganded HbC (beta6 Lys) compared to HbS (beta6 Val) and HbA (beta6 Glu). Under conditions of low pH (pH 6.35) in the presence of inositol-hexaphosphate, COHb assumes an altered R-state. The phase lines for the three Hb variants in concentration and temperature coordinates indicate that liganded HbC exhibits a stronger net intermolecular attraction with a longer range than liganded HbS and HbA. Over time, L-L phase separation gives rise to amorphous aggregation and subsequent formation of crystals of different kinetics and habits, unique to the individual Hb. The composite of R- and T-like solution aggregation behavior indicates that this is a conformationally driven event. These results indicate that specific contact sites, thermodynamics, and kinetics all play a role in L-L phase separation and differ for the beta6 mutant hemoglobins compared to HbA. In addition, the dense liquid droplet interface or aggregate interface noticeably participates in crystal nucleation.     

Hemoglobin of the Antarctic fishes Trematomus bernacchii and Trematomus newnesi: structural basis for the increased stability of the liganded tetramer relative to human hemoglobin

Hemoglobins extracted from fishes that live in temperate waters show little or no dissociation even in the liganded form, unlike human hemoglobin (HbA). To establish whether cold adaptation influences the tendency to dissociate, the dimer-tetramer association constants (L(2,4)) of the carbonmonoxy derivatives of representative hemoglobins from two Antarctic fishes, Trematomus newnesi (Hb1Tn) and Trematomus bernacchii (Hb1Tb), were determined by analytical ultracentrifugation as a function of pH in the range 6.0-8.6 and compared to HbA. HbA is more dissociated than fish hemoglobins at all pH values and in particular at pH 6.0. In contrast, both fish hemoglobins are mostly tetrameric over the whole pH range studied. The extent of hydrophobic surface area buried at the alpha(1)beta(2) interface upon association of dimers into tetramers and the number of hydrogen bonds formed are currently thought to play a major role in the stabilization of the hemoglobin tetramer. These contributions were derived from the X-ray structures of the three hemoglobins under study and found to be in good agreement with the experimentally determined L(2,4) values. pH affects oxygen binding of T. bernacchii and T. newnesi hemoglobins in a different fashion. The lack of a pH effect on the dissociation of the liganded proteins supports the proposal that the structural basis of such effects resides in the T (unliganded) structure rather than in the R (liganded) one.     

Two-dimensional analysis of glycated hemoglobin heterogeneity in pediatric type 1 diabetes patients

Interindividual and ethnic variation in glycated hemoglobin levels, unrelated to blood glucose variation, complicates the clinical use of glycated hemoglobin assays for the diagnosis and management of diabetes. Assessing the types and amounts of glycated hemoglobins present in erythrocytes could provide insight into the mechanism. Blood samples and self-monitored mean blood glucose (MBG) levels were obtained from 85 pediatric type 1 diabetes patients. Glycated hemoglobin levels were measured using three primary assays (boronate-affinity chromatography, capillary isoelectric focusing (CIEF), and standardized DCA2000+ immunoassay) and a two-dimensional (2D) analytical system consisting of boronate-affinity chromatography followed by CIEF. The 2D system separated hemoglobin into five subfractions, four of which contained glycated hemoglobins. Glycated hemoglobin measurements were compared in patients with low, moderate, or high hemoglobin glycation index (HGI), a measure of glycated hemoglobin controlled for blood glucose variation. MBG was not significantly different between HGI groups. Glycated hemoglobin levels measured by all three primary assays and in all four glycated 2D subfractions were significantly different between HGI groups and highest in high HGI patients. These results show that interindividual variation in glycated hemoglobin levels was evident in diabetes patients with similar blood glucose levels regardless of which glycated hemoglobins were measured.     

Analysis of single erythrocytes by injection-based capillary isoelectric focusing with laser-induced native fluorescence detection

A modified version of capillary isoelectric focusing (cIEF) was developed to separate hemoglobin variants contained within single human erythrocytes. Laser-induced native fluorescence with 275 nm excitation was used to detect the separated hemoglobins. In this method, baseline fluctuations were minimized and detection sensitivity was improved by using dilute solutions of anolyte, catholyte, and carrier ampholytes (with methylcellulose). Since electrosmotic flow was used for mobilization of the focused bands, separation and detection were integrated into a single step. The capillary was first filled with only ampholyte solution, and the cell (or standard) was injected as in capillary zone electrophoresis. The ∼90 fl injection volume for individual cells is 7×10⁴ times lower than those previously reported. Adult (normal and elevated A₁), sickle (heterozygous), and fetal erythrocytes were analyzed, with the amounts of hemoglobins A₀, A_1c, S and F determined. The pH gradient for cIEF was linear (r² = 0.9984), which allowed tentative identification of Hb F_ac. Variants differing by as little as 0.025 pI units were resolved.     

The effect of acetaldehyde concentrations on the relative rates of formation of acetaldehyde-modified hemoglobins

The effect of various concentrations of acetaldehyde (0, 0.05, 0.1, 0.25, 0.5, 1.0, and 5.0 mM) on the relative rates of formation of hemoglobin acetaldehyde adducts detected in fractions eluted from cation exchange high-pressure liquid chromatography (HPLC) was investigated. When the hemoglobin and acetaldehyde mixtures were incubated at 37 degrees C for various time intervals up to 24 hr, increased amounts of HbA1c could be observed after 2 hr incubation with 1 mM or greater concentrations of acetaldehyde, or after 4 hr incubation with at least 0.5 mM acetaldehyde. An increase in the HbA1a + b fraction was not observed with 4 hr incubation time until the acetaldehyde level reached 1 mM. The HPLC method detected no difference in minor hemoglobins from alcoholic and normal subjects. Incubation of red blood cells at 37 degrees C for 1 hr with six consecutive pulses of 0.05 mM [14C]acetaldehyde showed no differences in the amounts of minor hemoglobins determined chromatographically at various pulse intervals. However, the measure of the 14C-label incorporation into hemoglobin showed that adducts eluting in the HbA1a+b fraction were formed at a faster rate than those eluting in the HbA1c or HbA0 fraction, respectively. The specific activities of the HbA1a+b fractions at 2, 4, and 6 pulses were 34, 128, and 949 cpm/mg hemoglobin; those of the HbA1c fraction were 15, 58, and 174 cpm/mg hemoglobin. This evidence of modification of hemoglobin by physiological levels of acetaldehyde from 14C-label incorporation suggests that an assay more sensitive than chromatographic separation of adducts might be clinically useful in detecting alcoholism or monitoring alcohol detoxification programs.     

Intermolecular interactions,nucleation, and thermodynamics of crystallization of hemoglobin C

The mutated hemoglobin HbC (beta 6 Glu-->Lys), in the oxygenated (R) liganded state, forms crystals inside red blood cells of patients with CC and SC diseases. Static and dynamic light scattering characterization of the interactions between the R-state (CO) HbC, HbA, and HbS molecules in low-ionic-strength solutions showed that electrostatics is unimportant and that the interactions are dominated by the specific binding of solutions' ions to the proteins. Microscopic observations and determinations of the nucleation statistics showed that the crystals of HbC nucleate and grow by the attachment of native molecules from the solution and that concurrent amorphous phases, spherulites, and microfibers are not building blocks for the crystal. Using a novel miniaturized light-scintillation technique, we quantified a strong retrograde solubility dependence on temperature. Thermodynamic analyses of HbC crystallization yielded a high positive enthalpy of 155 kJ mol(-1), i.e., the specific interactions favor HbC molecules in the solute state. Then, HbC crystallization is only possible because of the huge entropy gain of 610 J mol(-1) K(-1), likely stemming from the release of up to 10 water molecules per protein intermolecular contact-hydrophobic interaction. Thus, the higher crystallization propensity of R-state HbC is attributable to increased hydrophobicity resulting from the conformational changes that accompany the HbC beta 6 mutation.     

The acetylation state of human fetal hemoglobin modulates the strength of its subunit interactions: long-range effects and implications for histone interactions in the nucleosome

The source of the 70-fold increased tetramer strength of liganded fetal hemoglobin relative to that of adult hemoglobin between pH 6.0 and 7.5 reported earlier [Dumoulin et al. (1997) J. Biol. Chem. 272, 31326] has been identified as the N-terminal Gly residue of the gamma-chain, which is replaced by Val in adult hemoglobin. This was revealed by extending the study of the pH dependence of the tetramer-dimer equilibrium of these hemoglobins into the alkaline range as far as pH 9. From pH 7.5 to 9.0, the 70-fold difference in the association equilibrium constant between hemoglobins F and A lessened progressively. This behavior was attributed to the difference in the pK(a) 8.1 of Gly-1(gamma) compared to the pK(a) 7.1 value of Val-1(beta) of hemoglobins F and A, respectively. Evidence for this conclusion was obtained by demonstrating that natural hemoglobin F(1), which is specifically acetylated at Gly-1(gamma) and hence unable to be protonated, behaves like HbA and not HbF in its tetramer-dimer association properties over the pH range studied. An increased degree of protonation of the gamma-chain N-terminus of hemoglobin F from pH 9.0 to 8.0 is therefore suggested as responsible for its increased tetramer strength representing an example of transmission of a signal from its positively charged N-terminal tail to the distant subunit allosteric interface where the equilibrium constant is measured. An analogy is made between the effects of acetylation of the fetal hemoglobin tetramer on the strength of its subunit interactions and acetylation of some internal Lys residues within the N-terminal segments of the histone octamer around which DNA is wrapped in the nucleosome.     

Further resolution of adult chick hemoglobins by isoelectric focusing in polyacrylamide gel.

Evidence is presented that adult chick hemoglobins exist in four types separable by isoelectric focusing on polyacrylamide gels instead of the two hemoglobin types previously resolved by other methods. These are hemoglobin A1 (HbA1), hemoglobin A2 (HbA2), hemoglobin D1 (HbD1), and hemoglobin D2(HbD2). Their pI values are 7.53 +/- 0.02, 7.37 +/- 0.02, 6.92 +/- 0.04 and 6.72 +/- 0.05, respectively, constituting about 63, 14, 18 and 5% of the total hemoglobin from adult chick erythrocytes, respectively. HbA1 and HbA2 ar identical in size, as determined on sodium dodecyl sulfate gels and similar in their amino acid composition and tryptic peptides. The molecular weight and amino acid composition of HbD1 and HbD2 are also identical although there are differences in their tryptic peptides. Experiments were done to show that the existence of four hemoglobin types is not due to genetic heterogeneity of the experimental animal, nor to artifacts of oxidation of carboxyhemoglobin to methemoglobin tetramers. Care was exercised to eliminate deamination and modification of side chain amino groups by using freshly prepared hemolysates and to minimize the "plateau phenomenon" peculiar to isoelectric focusing by controlling the duration of electrophoresis. The use of cyanmet form of (thus liganded) hemoglobin in this study reduced the chance of heterotetramer formation. Furthermore, consideration was given to possible anomalies caused by ampholytes. In the face of negative evidence for artifacts, it is concluded that adult chicken has more than the two hemoglobin types previously reported.     

The interaction of inositol pentaphosphate with the hemoglobins of highland and lowland geese.

We have studied the binding of inositol pentaphosphate (IPP) to the hemoglobins from two species of goose living at low and high altitudes, using the proton absorption method. Measurements were done at 25 and 37 degrees C in a pH range between 6.0 and 8.8. The bird hemoglobins show a high affinity and a binding stoichiometry of 1 IPP molecule/hemoglobin tetramer both in the ligated and unligated state, indicating the same binding site for IPP in oxy- and deoxyhemoglobin. The results indicate that the interaction of IPP with both geese hemoglobins is very similar. For the deoxyhemoglobins of both species the IPP-binding constant shows a strong pH dependence extending over a wide pH range (i.e. +/- 2 x 10(6) M at pH 8.8 and +/- 6 x 10(10) M at pH 6.0). The binding constant of IPP for the oxyhemoglobins shows a much weaker pH dependence (i.e. +/- 4 x 10(4) M at pH 8.8 and +/- 3 x 10(6) M at pH 6.0), indicating that the interaction of IPP with the goose hemoglobin is strongly dependent on the state of ligation of the protein. The IPP binding constants for the oxy- and deoxyhemoglobins are found to be in good agreement with the IPP-induced change in oxygen affinity of both hemoglobins as estimated from oxygen binding curves.     

A Koelliker hemoglobin in chick erythrocytes

1. Adult chicken hemoglobins were analysed by ion exchange chromatography and isoelectric focusing and a minor hemoglobin fraction (HbK) was isolated. 2. Analysis of the constituent chains shows that HbK differs from the two major hemoglobins HbA and HbD in the alpha globin. 3. The amino acid composition, the tryptic peptide maps, the results of carboxypeptidase digestion and the functional properties show that the HbK alpha globin is quite similar to that of HbA except that the C-terminal amino acid Arg 141 is lacking. 4. HbK must then be considered a Koelliker-type hemoglobin.     

Gold nanoparticles-coated magnetic microspheres as affinity matrix for detection of hemoglobin A1c in blood by microfluidic immunoassay

A novel microfluidic immunoassay system for specific detection of hemoglobin A1c (HbA1c) was developed based on a three-component shell/shell/core structured magnetic nanocomposite Au/chitosan/Fe(3)O(4), which was synthesized with easy handling feature of Fe(3)O(4) by magnet, high affinity for gold nanoparticles of chitosan and good immobilization ability for anti-human hemoglobin-A1c antibody (HbA1c mAb) of assembled colloidal gold nanoparticles. The resulting HbA1c mAb/Au/chitosan/Fe(3)O(4) magnetic nanoparticles were then introduced into microfluidic devices coupled with a gold nanoband microelectrode as electrochemical detector. After that, three-step rapid immunoreactions were carried out in the sequence of HbA1c, anti-human hemoglobin antibodies (Hb mAb) and the secondary alkaline phosphatase (AP)-conjugated antibody within 20 min. The current response of 1-naphtol obtained from the reaction between the secondary AP-conjugated antibody and 1-naphthyl phosphate (1-NP) increased proportionally to the HbA1c concentration. Under optimized electrophoresis and detection conditions, HbA1c responded linearly in the concentration of 0.05-1.5 μg mL(-1), with the detection limit of 0.025 μg mL(-1). This system was successfully employed for detection of HbA1c in blood with good accuracy and renewable ability. The proposed method proved its potential use in clinical immunoassay of HbA1c.     

Hemoglobin of the lung fish Clarias lazera: isolation and oxygen equilibrium studies

The hemoglobin of the lung fish Clarias lazera has a single component on starch gel electrophoresis. The hemoglobin has a molar mass of c. 68,000 similar to HbA on column chromatography. Clarias hemoglobin has a high oxygen affinity with a low Bohr effect. There is a haem-haem interaction, n, which is pH dependent. The R-state is more stable than the T-state, unlike in most fish hemoglobins.     

Studies on subfractions of hemoglobin A1b and hemoglobin A1c in diabetic subjects

Hemoglobin (Hb) obtained from the hemolysate of normal subjects and diabetic patients was separated into HbA1a1, HbA1a2, HbA1b, HbA1c and HbA0 (major Hb) by Bio-Rex 70 cation exchange column chromatography. The glycosylated Hbs were further separated reproductively by cation exchange high performance liquid chromatography (HPLC), using 50 mM sodium phosphate buffer pH 5.80 with 0-0.2 M NaCl linear gradient system. HbA1b and HbA1c were separated into two subfractions (HbA1b1 and HbA1b2) and three subfractions (HbA1c1, HbA1c2, HbA1c3), respectively. The percentages of each subfraction except HbA1c1 in diabetic patients were significantly higher than those in normal subjects. Furthermore, HbA1c1, HbA1c2 and HbA1c3 correlated well with fasting blood glucose levels in the prior 5 month period, while subfractions in HbA1b revealed no significant correlation with blood glucose levels. The percentages of each subfraction of HbA1c in patients either with diabetic cataracts or with diabetic neuropathy were almost the same as those in the patients without complications. However, the percentages of each of the three groups were markedly higher than those of the normal subjects. These results suggest that glycosylation of hemoglobin in diabetic patients may be increased in various sites of the molecule in parallel with the blood glucose levels during the preceding 4-5 months.     

Transition of hemoglobin between two tertiary conformations: The transition constant differs significantly for the major and minor hemoglobins of the Japanese quail (Cortunix cortunix japonica)

We demonstrate that 5,5'-dithiobis(2-nitrobenzoate) - DTNB - reacts with only CysF9[93]beta and CysB5[23]beta among the multiple sulfhydryl groups of the major and minor hemoglobins of the Japanese quail (Cortunix cortunix japonica). K(equ), the equilibrium constant for the reaction, does not differ very significantly between the two hemoglobins. It decreases 430-fold between pH approximately 5.6 and pH approximately 9: from a mean of 7+/-1 to a mean of 0.016+/-0.003. Quantitative analyses of the K(equ) data based on published X-ray and temperature-jump evidence for a tertiary structure transition in liganded hemoglobin enable the calculation of K(rt), the equilibrium constant for the r<---->t tertiary structure transition. K(rt) differs significantly between the two hemoglobins: 0.744+/-0.04 for the major, 0.401+/-0.01 for the minor hemoglobin. The mean pK(a)s of the two groups whose ionizations are coupled to the DTNB reaction are about the same as previously reported for mammalian hemoglobins.     

The stability of the heme-globin linkage in some normal, mutant, and chemically modified hemoglobins.

The rates and equilibria of heme exchange between methemoglobin and serum albumin were measured using a simple new spectrophotometric method. It is based on the large difference between the spectrum of methemoglobin and that of methemealbumin at pH 8-9. The rate of heme exchange was found to be independent of the albumin concentration and inversely proportional to the hemoglobin (Hb) concentration. Taken together with the finding that the rate was 10 times greater for Hb Rothschild, which is completely dissociated into alpha beta dimers and 10 times smaller for two cross-linked hemoglobins, the subunits of which cannot dissociate, this showed that the rate of dissociation of heme from alpha beta dimers is very much greater than from tetramers. Conditions were found for the attainment of an equilibrium distribution of hemes between beta globin and albumin. The equilibrium distribution ratio, R = methemealbumin/albumin/methemoglobin/apohemoglobin, for hemoglobin A was 3.4 with human and 0.005 with bovine serum albumin. Both the rates of exchange and the R values of HbS and HbF were the same as that for HbA. The equilibrium distribution ratio for Hb Rothschild was 7 times greater than that for HbA whereas that of one but not the other of the cross-linked hemoglobins was 10 times smaller. The structural bases for these differences are analyzed.     

Bioelectrocatalytic detection of glycated hemoglobin (HbA1c) based on the competitive binding of target and signaling glycoproteins to a boronate-modified surface

We developed an electrochemical glycated hemoglobin (HbA(1c)) biosensor for diagnosing diabetes in whole human blood based on the competitive binding reaction of glycated proteins. Until now, no studies have reported a simple and accurate electrochemical biosensor for the quantification of HbA(1c) in whole blood. This is because it is very difficult to correctly distinguish HbA(1c) from large amounts of hemoglobin and other components in whole blood. To detect glycated hemoglobin, we used electrodes modified with boronic acid, which forms a covalent bond between its diol group and the cis-diol group of the carbohydrate moiety of glycated proteins. For accurate HbA(1c) biosensing, we first removed blood components (except for hemoglobin) such as glycated proteins and blood glucose as they interfere with the boronate-based HbA(1c) competition analysis by reacting with the boronate-modified surface via a cis-diol interaction. After hemoglobin separation, target HbA(1c) and GOx at a predetermined concentration were reacted through a competition onto the boronate-modified electrode, allowing HbA(1c) to be detected linearly within a range of 4.5-15% of the separated hemoglobin sample (HbA(1c)/total hemoglobin). This range covers the required clinical reference range of diabetes mellitus. Hence, the proposed method can be used for measuring %HbA(1c) in whole human blood, and can also be applied to measuring the concentration of various glycated proteins that contain peripheral sugar groups.     

Differences in cattle globins

1. Comparisons have been made between electrophoretic mobilities of the cattle haemoglobins, HbA, HbB, HbC, HbD and HbF, at pH8.9. The fastest was HbB, then came in decreasing order HbF, HbC, HbA and HbD. 2. Globins were prepared from the main fractions of the five haemoglobins by CM-cellulose chromatography and investigated by starch-gel electrophoresis in four different buffer systems. In three of these (pH2.0 and 11.8) the globins appeared as two bands on stained starch gels. The slowest bands, the alpha-chains, showed the same rate of migration in all five globins. The faster bands, the non-alpha-chains, differed, that of HbF being the fastest and that from HbC the slowest. The other three were intermediate with, however, very small difference between the non-alpha-chains from HbA and HbD. 3. At pH1.8 in an acetate-phosphate-hydrochloric acid-urea buffer three bands appeared in all five globins of which the two slowest were indistinguishable in rates of migration, whereas the rates of migration of the third and fastest bands differed. Explanations for the occurrence of three bands are discussed.     

Phase separation and crystallization of hemoglobin C in transgenic mouse and human erythrocytes

Individuals expressing hemoglobin C (β6 Glu→Lys) present red blood cells (RBC) with intraerythrocytic crystals that form when hemoglobin (Hb) is oxygenated. Our earlier in vitro liquid-liquid (L-L) phase separation studies demonstrated that liganded HbC exhibits a stronger net intermolecular attraction with a longer range than liganded HbS or HbA, and that L-L phase separation preceded and enhanced crystallization. We now present evidence for the role of phase separation in HbC crystallization in the RBC, and the role of the RBC membrane as a nucleation center. RBC obtained from both human homozygous HbC patients and transgenic mice expressing only human HbC were studied by bright-field and differential interference contrast video-enhanced microscopy. RBC were exposed to hypertonic NaCl solution (1.5-3%) to induce crystallization within an appropriate experimental time frame. L-L phase separation occurred inside the RBC, which in turn enhanced the formation of intraerythrocytic crystals. RBC L-L phase separation and crystallization comply with the thermodynamic and kinetics laws established through in vitro studies of phase transformations. This is the first report, to the best of our knowledge, to capture a temporal view of intraerythrocytic HbC phase separation, crystal formation, and dissolution.