| Background and PurposeHemoglobin (Hb), the oxygen-carrying moiety of erythrocytes, is α polypeptide tetramer of two α-like chains (ζ or α) and two β-like chains (ε, Gγ, Aγ,δ or (3) surrounding a heme molecule. During embryonic development, the ζ, α, ε or γ chains contribute to embryonic Hbs such as Hb Gower1(ζ2ε2), Hb Gower2(α2ε2) and Hb Portland (ζ2γ2).The α and γ chains contribute to Hb F (α2γ2) during fetal development. In normal adults, the major Hb type is Hb A, which consists of two a and two (3chains, while the minor Hb fractions are Hb A2(α2δ2) and Hb F. In normal newborns, Hb F fraction is approximately70%hile that of Hb A is approximately30%.Thalassemias are inherited disorders of hemoglobin synthesis that are found at high frequencies in countries, which are historically afflicted with endemic malaria, including Southern China. The geographical correlation of the disease distribution with the historical endemicity of malaria suggests that thalassemias have risen in frequency through natural selection by malaria because of genes conferring genetic resistance to malaria in humans. The two main types of thalassemia are α and β. Point mutations or deletions cause low or no expression of the globin gene leading to imbalance between the α-like and β-like chains, which is the etiology of thalassemias. The ultimate method for the diagnosis of different types of thalassemias is DNA analysis based on polymerase chain reaction (PCR). Thalassemia caused by deletion mutations is usually detected by gap-PCR while reverse dot blot (RDB) is useful for point mutations. Many other strategies such as denaturing HPLC, multiplex ligation-dependent probe amplification, and high-resolution melting have been applied. However, a direct DNA approach without a precise biochemical hematological indication can be time-consuming, expensive and subject to false negative results or misinterpretations. The complex relationship between genotype and phenotype make the diagnosis difficult. Thus, a combination of different tests is required for accurate diagnosis.In fact, phenotype studies still occupy a key position in the diagnosis of hemoglobin disorders despite the development of a number of molecular biology techniques. Several markers such as mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), and Hb A2are recommended as complementary tests. Globin chain analysis is always helpful for detecting thalassemias and abnormal Hbs. Proper identification of an unknown Hb variant is important to determine if it is causing a clinical abnormality or is simply a polymorphism. Reversed-phase high-performance liquid chromatography (RP-HPLC) could be useful for this approach. Globin chain analysis is also useful for monitoring gene therapy and other related hematologic studies.To date, techniques such as cellulose acetate electrophoresis, sodium dodecyl sulfate polyacrylamide gel electrophoresis and capillary zone electrophoresis have been used for globin chain detection. However, these methods have long analysis times or use extracted denatured globin chains. RP-HPLC, a generally applied tool for the separation, purification and quantification of proteins from biological materials, has greatly influenced and stimulated research in the analysis of human globin chains. This methodology has a number of advantages:it is fast and accurate, uses minute amounts of material, has good resolution and reproducibility, and most importantly, can be completely automated. Based on differences in hydrophobicity, a variety of RP-HPLC gradient programs with long analytical times have been used to analyze human globin chains. This method reveals differences in hydrophobicity, improving the separation of Hb, which is difficult to achieve with electrophoretic systems (e.g., urea-triton electrophoresis and isoelectrofocusing). The separation is usually based on the application of an increasingly hydrophobic environment to the chromarographic column, which is filled with an interacting lipophilic stationary phase. Practically, the condition of the dissociated heme group could be well created at a pH below3using a constant concentration of trifluoroacetic acid and the order of elution, therefore, primarily depends on the hydrophobicity of the individual chains.Rapid separation of human globin chains by RP-HPLC has been reported, but the technique still required time-consuming globin precipitate extraction and the resolution was limited. The analytical time and sample pre-treatment of this method were the two main problems for its clinical application. Therefore, a simple, fast and high-resolution method is required. Furthermore, a challenge to analysis of human globin chains is the selection of initial separation conditions and subsequent optimization of the appropriate experimental parameters for a column without a ready gradient program.Our main aim was to established a simple and rapid chromatographic procedure for human globin chain separation and to discuss the value of the established RP-HPLC method in diognosis of thalassemia and hemoglobin variant. In addition, Our experience in developing this RP-HPLC method for the rapid separation of human globin chains could be of use for similar work.Materials and MethodsStudy subjectsA total of301samples with normal (n=106), a thalassemia silent/trait (n=60, including2ones with Hb WS), Hb H disease (n=59), β thalassemia carriers (n=46, including2ones with Hb E and3co-inheritence of a thalassemia ones with Hb WS) and P thalassemia intermediate/major (n=30, including11ones with Hb E and2Hb WS) were collected using EDTA as anticoagulant in this study. Full blood counts and a hemoglobin test of direct DNA analysis was used as described previously. The nondeletional forms of a thlasassemia and P thalassemia mutations were identified by RDB assay while the deletional forms of a thalassemia were determined by gap-PCR.Instrumentation and chromatographyAlthough any conventional HPLC machine could be used, we used a Shimadzu LC-20AT chromatographic system (Shimadzu, Kyoto, Japan) with a CBM-20A system controller, a LC-20AT binary pump, a CTO-20A column oven, a SPD-20A UV-Vis detector, a SIL-20A auto-injector and an LC-SOLUTION work station. Chromatographic separation was with a Jupiter C18HPLC column (4.6mm×250mm,5μm,300A, Phenomenex, Torrance, CA, USA) and a SecurityGuard C18column (4.0mm×3.0mm,5μm,300A, Phenomenex, Torrance, CA, USA), that is shielded ahead of Jupiter C18HPLC column for filtering some small particles. Elution was obtained at40℃, a temperature chosen for lower pressure, better resolution and higher reproducibility than could be obtained at uncontrolled room temperature. Solvent A was a mixture of acetonitrile-methanol, at90:10(v/v). Solvent B was0.5%TFA in deionized water at pH2.6adjusted by a few drops of ION NaOH. A flow rate of2.0ml/min was applied during a linear gradient of acetonitrile-methanol. The UV detection wavelength was280nm as previously reported, and the injection volume was10μl. By summing the change in the law of hydrophobicity (increase between25%and60%acetonitrile) according to previous similar works, the gradient was designed with a high percentage of solvent B at the beginning and ended with a lower percentage during several minutes within the determined scope of hydrophobicity. Our goal was obtaining a procedure for satisfactory separation of human globin chains. Relative quantification was carried out by measuring the percentage of the peak area of the heme and globin chains. The gradient program could be modified according to the machine, the geometry of the column, and the separation to be achieved. Resolution was adjusted by changing the gradient slope and run time. Simple pre-treatment of blood samplesStatistical analysisBivariate correlation analysis conducted by using the spearman correlation coefficient and linear regression analysis, with adjustment for assuming the actual a peak, were used to evaluate the independent correlation between Hb F levels and the ratio of Gγ+Aγ/α. Using an analysis of variance (ANOVA) test, we compared normal groups with patient groups for area ratio of α/pre-β+β. The ROC curve was plotted using SPSS17.0(IBM, Armonk, New York, USA) to select the best indicator and to determine suitable cut-off values for the prediction of thalassemia patients. For all tests, statistical significance was defined as P<0.05.Results and discussionMethod development and optimization For the separation of y subunits, a Hb solution from a normal newborn (high Hb F) was used for method optimization. A gradient starting with51%solvent B and ending with40%solvent B in12min was found to be the best separation procedure. The prepared Hb solution was injected, and the gradient started and developed. Heme groups and four globin chains (β, α, Gγ, and Aγ) were separated clearly. The elution pattern showed the appearance of the most hydrophilic heme groups at5.0min, followed in7.8min by the β peak (6.9min for pre-(3),11.4min by the a chain,10.7min by the Gy chain and12.0min by the Ay chain. The most suitable gradient started with51%solvent B, continued for1min, and then changed from51%to50%in1min, then from50%to40%solvent B (with a corresponding increase in solvent A from49%to60%). Re-equilibration (51%solvent B) was required for the next run. As the total y chains contributed to Hb F,30samples with different Hb F levels from β thalassemia intermediate/major group were analyzed to verify the actual two y-globin peaks and α-globin peak. An obvious positive correlation was obtained between the ratio of Gγ+Aγ/α (α as a reference) and Hb F levels (rs=0.887, P<0.01). In our assay, this correlation could not be observed assuming the actual a chain was eluted before Gγ chain like in other previous similar works (rs=-0.787, P<0.01) or at12.0min (rs=0.267, P>0.05), and therefore the peak at a11.4min retention time was the actual α chain. The similar results of the correlation were obtained by linear regression analysis according to our data. It was also identified by comparing the chromatograms of a normal newborn, a normal adult, a mixture of them with equal volumn and a sample of Bart’s. Our results of the elution order of Gy and Ay peaks is similar to the other previous works, Gy had a faster retention time than Ay in the present method, thus indicating the corresponding to the separation mechanism of due to less hydrophobic for Glycine than Alanine under an increasingly hydrophobic condition. When the Hb solution of a normal adult was injected, the δ chain was obtained between the a and β chains and the pre-(3chain was observed between the heme groups and the β peak. We injected the same sample five times, obtaining a reproducibility of retention time from0.11%to1.29%. The peak area CV was between0.32%and4.86%. The CVs were all below5%, thus indicating the good reproducibility of our establised method. The a and β chains were separated without the heme peak after heme-depleted globin powder was injected, but the peak shape was poor. Degradation of the globin chains caused by acidified acetone might have been a possible reason. The plasma sample eluted before3min without an observable peak at the retention time of the globin chains or heme groups. The elution pattern of the sample with simple pre-treatment was similar to the prepared Hb solution. The plasma protein fractions had no influence on the separation of globin chains.The high flow rate reduced the retention time, but caused higher pressure and decreased the resolution of detection. To maintain the resolution, save solvent and protect the column, a2.0ml/min flow rate was used. A lower pH mainly enabled a faster retention time for heme, while a shorter gradient time achieved a faster elution time for globins. Wavelength was important factor for the absorption area. The area ratio of α/β between210and220nm was about one, which was specific for the peptide bond, and gave markedly better detection limits compared to other wavelengths. However, the slope of the baseline increased. This problem was not seen at280nm, which corresponds to the aromatic amino acids tryptophan and tyrosine and gave a higher absorption for the β chain than for the a chain. Therefore, the wavelength was set at280nm for quantitative analysis of human globin chains.Application studySignificant differences (P<0.001) among groups (normal, Hb H and β thalassemia) were found in the area ratio of α/pre-β+β applying the rapid elution procedure, while P≥0.05was obtained between the normal and a thalassemia silent/trait group.Based on the results of multiple comparisons, data sets from106normal persons and46β thalassemia carriers applying the established procedure were used to construct ROC curves. The areas under the ROC curve (AUC) were1.000for δ/β,1.000for δ/pre-β+β,0.999for8,0.980for α/β,0.879for α/pre-β+β and0.605for α/pre-β+β+β. The δ/β ratio was selected as the best indicator for the prediction of β thalassemia carriers. Based on the ROC curve analysis, the optimal cut-off value for indicating β thalassemia carriers using the ratio of δ/β was0.026with a sensitivity of100.0%and a specificity of100.0%. The same analysis was employed in plotting a curve for the ratio of α/β and α/pre-β+β for the prediction of Hb H based on ANOVA results. The ROC curve area was0.981for α/β and0.958for α/pre-β+β. The ratio of α/pre-β+β was a better marker for Hb H disease, and the proposed cut-off value was0.626with a sensitivity of96.6%and specificity of89.6%.Among the301subjects, each of7Hb WS and of13Hb E were well identified by the respective specific peaks using this RP-HPLC method. The abnormal αWS chains were eluted in13.6min while the βE chains eluted in7.3min.DiscussionHere we present a rapid method for the separation of human globin chains with simple sample preparation. Hemoglobin was released from red blood cells because of osmotic pressure. The presence of0.5%TFA in solvent B created a pH below3.0, and was required as the ion-pairing agent. Under these conditions, heme groups were removed from tetramers, and the constituent globins dissociated. Pre-β,β,δ, α, Gγ, and Aγ were clearly separated based on different hydrophobicities, but unlike the previous report, in which a chain was eluted before Gγ, in our procedure, a chain was recorded between Gγ and Aγ, and the actual a chain was verified by bivariate correlations regression analysis between Hb F levels and the ratio of Gγ+Aγ/α. Small peaks observed before2min with the prepared Hb solution (Figure1b) may be carbonic anhydrase and super oxide dismutase, that should exist in red blood cells. Our RP-HPLC profiles were similar with results observed from other similar reports and we may need to do further work on identifying the actual peaks of them in chromatogram if necessary. In fact, they have little influence on the analysis of results because of the significantly lower absorption compared with Hb components. Of note, small quantities of globins would be easily denatured under low pH conditions (using solvent B as a buffer with0.5%TFA), resulting in a poor chromatogram map and a lower detection sensitivity. However, this was not a problem when deionized water was used as a buffer. The pre-β peak observed before the normal (3-peak in this procedure was a posttranslationally modified protein, presumably containing a P-globin glutathione adduct as its major component of β chains.Because of the down-regulation of P chain or relative elevation of5chain, the area ratio of δ/β was selected as the best indicator for β thalassemia carriers. It also had a high sensitivity and specificity in the procedure presented here for β thalassemia intermediate/major in our study. To evaluate this marker, a larger number of samples are required for a validation study. β thalassemia heterozygote with normal Hb A2may be detected by this marker, thus a comparison between this marker and Hb A2for β thalassemia would be interesting. ANOVA tests for the area ratio of α/pre-β+β showed no significant difference between the normal and a thalassemia silent/trait, possibly because of the compensatory over-expression of the a-globin gene or the degradation of excess globin chains. As a result, the sensitivity of this direct method was lower than globin chain synthesis as described previously. Therefore, this method cannot be used as the sole diagnostic tool for α thalassemia silent/trait because of the considerable overlap of α/pre-β+β ratio ranges.Despite these limitations, the ratio of α/pre-β+β had a good sensitivity and specificity for Hb H disease prediction in this simple and rapid method. In fact, for monitoring gene therapy for β thalassemia major or Hb H disease, the sensitivity of this established procedure would be sufficient.For abnormal hemoglobin, Hb WS is difficult to detect using the Bio-Rad Variant Ⅱ Beta-Thal short program (Bio-Rad, Hercules, CA, USA), but was easy in the method described here. Furthermore, the retention time on the Bio-Rad VARIANT Ⅱ makes discriminating between Hb E and HB A2difficult, but the βE and δ globin chains, representing Hb E and HB A2, were clearly separated in the RP-HPLC method, demonstrating its high resolution.Hb WS and Hb E occur frequently in the Chinese population. Because of the degradation of abnormal chains or the limited resolution of this rapid chromatographic program, Hb CS and Hb QS, also common in China, were undetected. If globins from adult lysates without detectable amounts of Hb F were studied, a gradient that saves time (less than8min for the separation of heme, pre-β, β,δ, and α in this study) and solvents could be designed by changing the gradient program slope for a shorter time or by increasing the flow rate. However, the presence of a variant chain may escape observation in these conditions. To obtain additional variants or more hydrophobic chains (i.e. ζ) based on the gradient described here, the run time and hydrophobicity might need to be extended. This method might also be useful for the separation of globin chains in non-human blood, which is important for studying gene therapy.The design of the gradient program might depend on the protein column. Using a new RP column without a gradient program for globin chain separation and establishing new conditions is time consuming. In summary, in separating human globin chains by RP-HPLC, we found that hydrophobicity between25%and60%acetonitrile is effective for various protein columns. The scope of hydrophobicity described here has reference value for similar work.Discrimination between the two types of γ chains in this procedure could be of interest for diagnosis of the hereditary persistence of fetal hemoglobin or δβ thalassemia, or for suggesting a haplotype in cases of sickle cell anemia. This method might also be useful for the separation of globin chains in non-human blood, which is important for studying gene therapy in animal experiment. |
PDF Full Text Request