i MOLECULAR MECHANISM OF HEMOLYTIC ANEMIA IN HOMOZYGOUS HEMOGLOBIN C DISEASE ELECTRON MICROSCOPIC STUDY BY THE FREEZE-ETcHING TECHNIQUE* YB LAWRENCE S. LESSIN,* M.D., WALLACE N. JENSEN, M.D., DNA ERIC PONDER, M.D. (From eht Institut ed eigolohtaP ,erialulleC France, Kremlin Bicetre, dna eht tnemtrapeD of Medicine, State Ohio University egelloC of Medicine, Columbus, )oihO D o (Received for publication 31 March )9691 wn lo a d e The substitution of lysine for glutamic acid in the sixth position of the hemo- d fro globin beta chain is the single molecular aberration in holnozygous hemoglobin m h C disease (2) and must account for the mild hemolytic anemia, splenomegaly, ttp target cells, microspherocytes, ,3( 4, 5) and the rare crystal-containing cells ://rup re seen in this disorder ,3( 6). Studies of the physical properties of erythrocytes ss.o from patients with homozygous hemoglobin C disease by Charache and co- rg /je workers have demonstrated decreased solubility of intracellular hemoglobin C m/a (relative to hemoglobin A) (7). They suggest that abnormally tight molecular rticle packing and altered molecular interaction may result from the substitution of -pd f/1 lysine for glutamic acidi n an exterior region of the beta chain, conferring a local 3 0 /3 difference in charge and, possibly, steric conformation. By increasing intra- /4 4 3 molecular attraction and fit, these alterations decrease solubility of hemoglobin /1 0 8 C, which in turn leads to increased viscosity and decreased deformability of 33 3 8 C-C erythrocytes as compared to normal cells. These factors produce accele- /4 4 3 rated erythrocyte "senescence" by reticuloendothelial pitting and fragmentation .p d with the production of microspherocytes and crystal cells which are sequestered f by g in the microcirculation and destroyed by reticuloendothelial elements. ue Electron microscopic freeze-fracture-replication techniques appear to be st on 1 capable of demonstrating hemoglobin molecules in situ in the erythrocyte 1 F e (8-12). Particles of6 5-70A have been identified in intraerythrocytic rat hemo- bru a globin crystals (11), in the cytoplasm of rat erythrocytes, and in helical poly- ry 2 mers of hemoglobin S in situ in sickled erythrocytes (12). The present study 02 3 attempts to gain morphological insight into abnormal interaction of hemoglobin C molecules in C-C erythrocytes by employing an in vitro system of gradual osmotic dehydration for production of crystal cells (7), and the freeze-etching technique for preparation for electron microscopy (13). Results suggest that hemoglobin C molecules exist intracellularly in a par- * A portion of this work has been published in abstract form (1). Recipient of National Heart Institute Special Fellowship 1-F3-HE 35, 777-01. Present Address: Duke University Medical Center, Durham, N. C. 344 444 MOLECULAR MECHANISM OF HEMOLYTIC ANEMIA tially aggregated state and that gradual water loss from C-C erythrocytes leads to a progressive decrease of intermolecular distance producing precrystalline molecular alignment and, finally, crystal formation. Materials and Methods Preparation of Crystal Cells.--Blood obtained by venopuncture from a 28 year old Negro male with homozygous hemoglobin C disease was drawn directly into heparinized blood bank tubing which was promptly heat-sealed and kept at low temperature until use. After 1-3 days storage, cells were washed twice in phosphate-buffered saline (NaC1, 8 g/llter; KC1, 0.3 g/liter; Na:HPO4, 0.184 g/liter; KH2PO4, 0.02 g/liter; glucose, 2.0 g/liter) buffered to pH 7.45. Samples taken after 4, 12, and 24 hr incubation were inspected with a Zeiss photomicroscope adapted for phase contrast, interference, Soret absorption, and polarization microscopy. D o "Unincubated" specimens were treated only by immersion in 25% glycerine in phosphate- wn buffered normal saline to prevent formation of ice-caused artifacts and disruption of cells loa d during freezing. Samples were centrifuged at 1200 rpm for 5 min and cells resuspended in 25% ed glycerine in phosphate-buffered saline for 20 rain. from Freeze-Etching.--Droplets of the glycerinated, centrifuged "crystal cell" preparations http (from 0, 4, and 12 hr samples) were placed on centrally scratched 3 mm copper discs and im- ://ru mersed successively into liquid dichloromonofluorethane (--150°C) and liquid nitrogen pre (--196°C). The discs bearing the frozen droplets were then placed on the liquid nitrogen- ss.o cooled stage of a Balzers 500M high vacuum plant with freeze-etching device and freeze rg fractured with platinum-carbon replication carried out by the method of Moor and Muhle- /jem tshivaelly er flo(13). ating Replicas on commercial thus obtained bleach were and cleaned7 0% sulfuric of adherent acid, and cells they and were fragments triple-washed by succes- in /article -p distilled water. They were then placed on neoprenized 200-mesh copper grids and viewed with df/1 Philips EM-200 and Associated Electronic Industries EM-6B electron microscopes. 30 Reflection Demitometry of Electron Micrographs.--Reflection densitometry of electron /3/4 4 micrographs was carried out with a Photovolt Reflection Density Unit (Model 53) and photom- 3/1 0 eter (501A). Strips, 3 cm wide, of electronmicrographs of magnifications from 179,000 to 8 3 3 217,000 were scanned by the 0.40 cm spot at 0.25 cm intervals, with light incident at 45 .° 3 8 Readings were given in units of "reflection density", defined as logl~(Ro/Rt) where oR is the /44 3 light reflected by a white surface and R, that reflected by the test surface. Tracings, except .p d in one test case, were made parallel to the shadowing direction of the replica. Mean reflection f b densities and their standard deviations were used as a measure of uniformity of a given electron y gu e micrograph; more specifically, here, these data are used to compare variations of homogeneity st o in portions of two adjacent cells in the same micrograph. Because the size of the scanning spot n 1 1 exceeded that of the interval scanned, the curves do not reflect minute variations in density, F e but represent an integration of density maxima and minima, relating respectively to interag- bru a gregate noitcerroC spaces of and Dimensional particle aggregates. Distortion Produced by eht Replication Process.--The correction ry 20 2 system used is adapted from that of Misra and Das Gupta, (14) calculated for heavy metal 3 shadowing of specimens of known size, including latex and colloidal gold particles. The system permits correction of distortions due to shadowing angle and variations in heavy metal depo- sition if the shadowing angle is known. Schema I indicates the basic trigonometric method used, where 0 is the shadowing angle (in our case, always 45 °) AB, the true height diameter of the particle in a plane perpendicular to the horizontal (the fracture plane) and AC = EF, the measured "height" of the structure. Hence, AB = AC cos 0. Schema II shows the techniques for correction of width diameter, d, of a spherical particle L. S. LESSIN, W. N. JENSEN~ AND E. PONDER 445 where measured width, d, (perpendicular to the shadowing direction, arrow) and shadow length, l, are determined from the electron micrograph, and shadowing angle 0 is known (45°). It has been shown that the relationships of shadowing angle, 0, particle diameter, d, and shadow length, l, are given by the formula: d(l - d/2) Tan 0 = - - - l(l -- d) From this, one can determine d as given by: d' - (2(/)(tan 0)) Ad = d' -- d = (l + cos 0) D Since minor deviations of the fracture plane from the true horizontal and variation in heavy ow n metal deposition determining the size of metal caps (shaded areas) could not be well controlled, lo a the precision of these corrections is variable. de d fro m h ttp ://ru p re ss.o rg /je m /a rticle -p d f/1 3 0 /3 /4 4 3 /1 0 8 3 3 3 8 /4 4 3 .p d f b y g u e st o n 1 1 F ATAMEHCS I DNA II. eb ru in laStcehreamla projectionI , shows the where trigonometric d' is the obcsoerrrveecdt ion height of dimensional of a shadowed distortion sphere, due and to shadowd ing, the true ary 20 2 height. Schema II shows the method used for correction of shadowing error in width, based 3 on an empirical formula (see text). Adapted from Misra and Das Gupta 04). RESULTS Optical Microscopy.--C-C erythrocytes, after suspension for 20 rain in the glycerine-normal saline solution at 37°C, showed minor morphologic altera- tions including spicule formation, and partial loss of target cell morphology (Fig. ; 3) microspherocytes showendo observable change. After 4 hr of dehydra- 446 MOLECULAR MECHANISM OI ~ REMOLYTIC ANEMIA tion by incubation in % 3 NaC1 solution, the C-C erythrocytes demonstrated shrinkage of cell contents into central masses surrounded by halos of membrane. Some central masses assumed polygonal form and showed we~k birefringence. After 21 hr of incubation, 25 to 50% of the cells contained crystalloid inclusions, occasionally several in a single cell, in the form of hexagonal and tetrag~nal plates and prisms. Elongate crystals often produced marked distortion of the cell membrane by protraction about the extremities of the crystal. Crystalloid D o w n lo a d e d fro m h ttp ://ru p re ss.o rg /je m /a rticle -p d f/1 3 0 /3 /4 4 3 /1 Fie. .1 densitometric Reflection curve of an electron hpargorcim strip fo reticulocyte dna 0 8 3 adjacent erythrocyte replica from Fig. .5 peak Curves indicate increased frequency ni the 3 3 enarbmematxuj snoiger fo (RBC) erythrocyte the dna the of irregularity density the etagergga 8/4 4 ni krowhsem the reticulocyte with greater mumlnim-mumixam density Reflection distances. 3.p d (RD) with standard ,snoitaived P-value dna mean ,mumixam-mumixam muminim-mumixam f b secnatsid given are ni I. Table y g u e st o inclusions uniformly showed negative birefringence (Fig. 4) and an angle of n 1 1 extinction (determined by rotating stage) of about 45 .° Birefringence indicates F e b noncubic molecular alignment within the crystals (15). Sorer absorption (ob- ru a tained by monochromating the cadmium arc light source with 415 m~ exclusion ry 2 0 2 filter) by the crystalloid inclusions was strong, confirming a high concentration 3 of berne-containing protein in the crystals. nortcelE Microscopy.--Freeze-etched replicas of unincubated cells made after glycerinization in a medium, normotonic with respect to NaC1, showed generally rounded cell contours, although a few cells developed spicules (Fig. 3, 0 hr N 7000). Fracture planes passing through the cell interiors revealed relatively uniform coarse granular patterns. At high magnifications, this granular ap- pearance was seen to be composed of aggregates of 2-10 closely applied 70 A 447 L. S. LESSIN~ W. N. JENSEN, AND E. PONDER particles (Figs. 5 and 6). These aggregates were separated by spaces measuring up to 250 A. Average interaggregate spacing was greatest at the center of the cell interior and least adjacent to the cell membrane, where particles appeared more tightly packed, and in some juxtamembrane regions showed linear align- ment. Although the majority of unincubated C-C cells showed relatively wide interaggregate spacing, rare cells showed packing of particles both in cell center and periphery. Reficulocytes encountered in these preparations (Fig. 6), identi- fiable by the presence of simple and autophagocytic vacuoles, showed greater interaggregate spacing and lesser aggregation of particles. In these cells, small aggregates 3 to 5 particles across, were distributed more or less uniformly throughout the cell interior, showing very little packing in the juxtamembrane D o regions. wn lo Replicas of cells freeze-etched after 4 hr of hypertonic incubation showed ad e d greater variation in the granular patterns of their cytoplasm (Figs. 8 and 9). fro m Adjacent cells, varying in degree of aggregation of cytoplasmic particles, were h ttp frequently encountered (Fig. 8). Microphotoreflectometry (Fig. )1 was used to ://ru quantitate size and statistical variation of this spacing within a given cell and pre in adjacent cells (see below). ss.o rg Replicas of cells freeze-etched after 21 hr of incubation in hypertonic medium /je m revealed the presence of polygonal inclusions in about 50-75 % of erythrocytes /a included in the fracture plane. The remaining cells showed tighter packing of rticle -p cytoplasmic particles with coalescence of particle aggregates and minimal df/1 3 interaggregate spacing. Again, packing of particles was most marked in the 0/3 juxtamembrane regions where areas of linear alignment were found in the cell /44 3 /1 interior, well removed from visible membrane structures. Cells manifested a 0 8 3 spectrum of variation in particle aggregation and spacing, with juxtamembrane 33 8 regions consistently showing the tightest packing of particles. /44 3 Crystalloid inclusions showed characteristic polygonal contours. The presence .pd f b and dimension of definable periodicity was dependent upon the direction of the y g u fracture plane and the degree of etching (Figs. 01 and 11). Approximately one- est o third of these inclusions showed a simple linear and laminar periodicity of n 1 1 particles. The period was variable, measuring from 70-100 A (corrected). A F e b few crystals permitted more detailed examination of the crystal structure and ru a measurement of at least two dimensions of the molecular subunits (Fig. )11 ry 2 0 2 and their mode of arrangement in the crystal lattice (see paragraph below on 3 crystal structure). noitagerggA of cimsalpotyC Partides.--In none of the preparations were the 70 A cytoplasmic particles found completely dissociated from adjacent particles. Reticulocytes presented the least degree of particle aggregation and the greatest spacing. Except in the juxtamembrane areas, where some degree of packing was nearly always seen, the cytoplasmic particles were disposed in chains 1 to 5 particles thick, branching to adjacent chains to form a meshwork of chain-like 448 MOLECULAR MECHANISM OF HEMOLYTIC ANEMIA aggregates of particles interspersed with open spaces. These open spaces, regions relatively devoid of platinum-carbon deposition in the replica, are interpreted to represent, primarily, regions of depression between chains of particles from which water molecules were sublimated during the etching process, (16) and, secondarily, shadows of the particles and aggregates themselves. The chains, measuring from 70-350 A (corrected) are separated by open spaces up to 500 A (corrected) in width. The other extreme of particle aggregation and packing was seen in 4 and 21 hr incubated specimens (rarely seen in unincubated ones) in which the chain- space meshwork coalesced into a more tightly packed, homogenous pattern. Juxtamembrane regions of tightly packed particles manifested varying degrees D o w of organization and alignment (Fig. 8). n lo a Intermediate stages showed coalescence of single and complex chains into d e d larger aggregates with greater aggregation and tighter packing as dehydration fro m progressed (Fig. 9). h Crystal Structure.--Crystalloid inclusions observed in C-C cells incubated in ttp://ru hypertonic medium for 21 hr and longer were negatively birefringent. Prepara- pre tions further dehydrated between slide and cover slip overnight at 37°C de- ss.o rg veloped many large extracellular crystals (measuring up to 001 ~) manifesting /je m the same shape and birefringence as the smaller intracellular crystalloid inclu- /a sions (Fig. 4). Replicas of freeze-etched intracellular crystals revealed molecular rticle -p d subunit structure and manifested two patterns of periodicity. The pattern and f/1 3 period seen was dependent upon the direction of the fracture plane relative to 0/3 /4 the long axis of the crystal, and corresponded to the three-dimensional aspects 43 /1 of the crystal. Two distinct patterns were seen. The first was a linear-laminar 08 3 period of 60-100 A (P1) in which a single direction of alignment could be dis- 33 8 /4 cerned. The second (P2, Fig. 10), seen within the same crystals, was a 300-350 A 4 3 .p period resembling sheets of juxtaposed layers composed of 07 A particles and d f b appearing to represent fractured ends of laminae seen in pattern 1P (Fig. 11). y g u At higher resolution subunit particles measuring about 56 A by 001 A could est o easily be identified. These particles were arrayed in a noncubic monoclinic n 1 1 pattern resembling either a tetragonal or hexagonal arrangement, depending F e b on the region of the crystal viewed (Fig. ,11 t and h). ru a Comparison ot Normal Erythrocytes.--A-A erythrocytes collected by veno- ry 2 0 2 puncture in heparin anticoagulant and subjected to osmotic dehydration, 3 failed to show intracellular crystalloid inclusions during up to 27 hr of incuba- tion at 37°C in a medium identical to that used for C-C cells. Moreover, the majority of the normal cells became crenated and hemolyzed, showing signifi- cantly less resistance to the hyperosmotic medium than the C-C cells. It was possible, however, to produce crystals in normal cells by simply allowing a drop of blood to air dry between slide and cover slip on the lab bench overnight; the degree of crystallization could be augmented by adding a drop of 1% sodium L. S. LESSIN, W. N. JENSEN, AND E. PONDER 449 bisulfite, thus increasing the proportion of methemoglobin; see Bessis, et al (17). Replicas of normal erythrocytes, freeze-etched after washing and immersion in a mixture of 25 % glycerine in phosphate-buffered (pH 7.4) normal saline show a number of significant differences from C-C cells. The aggregation-spac- ing phenomenon seen in replicas of the majority of freeze-etched C-C cells is much less apparent in the normal cells. The latter present, to a greater extent, the pattern of random dispersion found in the 70 A cytoplasmic particles. Although the particles appear less closely packed, regions of open-space measur- ing up to 150 A can be defined, distributed randomly throughout the cell in- TABLE I D o ~R noiae~ ¢irtemotisneD sesylanA wn lo a d Mag Mean corrected Mean corrected ed Figure Subject nific~ tion Mean density reflection 4- sD ecnacifingiS maximum maximum -- minimumm aximum -- from × lC distance distance h ttp RBC (perpendicu- 712 64.)~ 4- 0.080* P<0.1 562 -q 58 *A 671 4- 06 *A ://rup Same xal to RBC )gniwodahs (par- 712 52.)E 4- 0.075 452 4- 06 A 701 -4- 34 A ress.org allel to shadow- /je m )gni /a rticle Reticulocyte 971 12.P 4- 0.095 P < 520.0 372 4- 601 A 871 -4- 68 A -pd RBC (adjacent 971 0.16.4- 0.078 392 4- 36 A 741 4- 48 A f/13 0 to retieulocyte) /3 /4 4 3 RBC )a( 181 0.32 4- 0.145 P < 500.0 443 4- 78 A 442 -4- 591 A /10 8 3 aRC )b( 181 0.27 4- 0.056 563 4- 502 A 502 + 85 A 3 3 8 /4 4 * Standard deviation. 3.p d f b terior, but excluding the juxtamembrane region. In this region, tight packing y gu e and a degree of alignment of particles similar to that seen in the C-C cells is st o n commonly found. A few normal cells did present a picture of tight packing, both 11 F in the juxtamembrane region and throughout the cell interior, resembling the eb ru oCf- C normal cell preparation cells did afmtaenri fest 4 hr of aggregation hypertonic of dehydration. cytoplasmic This particles, small proportion but to a ary 20 2 3 lesser degree than their C-C counterparts. At no time, did replicas of freeze- etched normal cells show particulate alignment of a paracrystalline or crystal- line type. Reflection Densitometric Studies of Electron Micrograpks.--Reflection densi- tometry was used in this study to assess density variation patterns per unit distance of a given electron micrograph. This distance measurement initially made in millimeters on the micrograph is easily converted to Angstroms of the replica. Fig. 1 is a representative study and points up several quantitative 450 MOLECULAR MECHANISM OF HEMOLYTIC ANEMIA aspects of the analysis which corroborates the qualitative impression of the micrograph strip (Table I). The reflection density curve for the reticulocyte (same as that shown in Fig. 6) shows a more marked variation of maxima and minima, with wider spacing of peaks in comparison to the older RBC on the right, where density variation and mean density are less. The curves, integrals by nature of the scanning technique, reflect the closer packing of particles and reduced interaggregate spacing in the juxtamembrane area. These integrated differences in density distribution of replicas of adjacent cells in the same elec- tron micrograph confirm in quantitative terms visual impressions of interag- gregate spacing, aggregate size, and particle packing gained from the micro- graphs. Table I presents data from such studies. The two scans of the same cell D o (in Fig. 8) in directions perpendicular and parallel to the shadowing direction wn lo are included to show the effect of shadowing direction of mean density and ad e d minimum-maximum spacing.The first measurements refer to the reticulocyte fro m and RBC shown in Figs. 6 and 7; the third to adjacent erythrocytes in Fig. 9. h ttp ://ru DISCUSSION p re cimsalpotyC Particles dna nibolgomeH Molecules.--The identification of the ss.o rg cytoplasmic particles characteristic of freeze-etched erythrocytes with hemo- /je m globin molecules is based upon the presence of particles of identical size, 65-70 A /a (corrected), in intraerythrocytic crystals showing absorption in the Soret 412 rticle -p m~ light band, specific for heine. Although particles of this type have not been df/1 3 visualized in red cells studied by standard electron microscopic techniques, due 0/3 probably to the denaturation and polymerization of hemoglobin by aldehyde /44 3 /1 and osmic acid fixation,(18, )91 they are consistently present in freeze-fracture 0 8 3 replicas of erythrocytes (8-12). In techniques in which a minimum of "etching" 33 8 (removal of intermolecular and, possibly, intramolecular water from the frac- /44 3 ture surface by sublimation in a high vacuum at low temperature) is employed, .pd f b cytoplasmic particles are larger, 100-150 A (uncorrected), and more variable in y g u size (8-10). The Moor-Muhlethaler technique (13) used in this study employs est o a 1 min sublimation at --100°C and 01 -~ torr. The actual measurement of n 1 1 replica cytoplasmic particles is 90-100 A. Correction of this value by the method F e b outlined above yields a value of 63-70 A, comparable to the X-ray diffraction ru a measurements of Muirhead, Perutz, et al. for human oxyhemoglobin, 63.4 A X ry 2 0 83.6 A X 53.9 A (per unit cell of two molecules (20). Hemoglobin molecules 23 identified in freeze-etched crystals of rat hemoglobin show similar dimensions, and 3 or 4 subunits of 20-35 A may be discerned in these molecules (11). Similar 65-70 A particles have been shown to organize into helical polymers in sickle cells freeze-etched after sickling with sodium bisulfite (12). Membrane as- sociated particles (Figs. 3 and 5) vary in size from 56 A-150 A and from spheri- cal to oval in shape. On internal erythrocyte membrane surfaces, such particles may be seen in continuity with those of the cytoplasm, and they are more uni- L. S. LESSIN, W. N. JENSEN, AND E. PONDER 154 form in size, measuring about 70 A. This relationship suggests that some internal surface particles may in fact be hemoglobin which remains adherent to the internal membrane surface when the remainder of the cytoplasm is fractured away. The external surface is more complex, manifesting a greater number and variation of particles which could represent structural proteins and antigenic glycoproteins (21). Correlation of Morphologic and Biophysical Abnormalities in C-C Cells.-- Peripheral blood films of individuals homozygous for hemoglobin C are com- posed predominantly of two cell types, the target cell and the microspherocyte. Reticulocytes, biconcave discs, and a variety of poikilocytes are present in relatively small numbers. Careful inspection of such films yields 2-5 per 1000 D o ceils containing crystalloid inclusions (3, 6). In splenectomized C-C patients, wn lo the number of "crystal cells" may increase to 3 % in dried films (6). After ad e d ultracentrifugal separation, Charache et al. found that the mean corpuscular fro m hemoglobin concentration and viscosity were greater in the "older", more dense h ttp microspherocytes, which crystallized with less osmotic dehydration, than in the ://ru lighter, less viscous, "young" target cells (7). In addition, these authors found pre that the mean corpuscular hemoglobin concentration was greater in C-C target ss.o rg cells than in normal cells of comparable age. They found that whole blood vis- /je m cosity was higher, and that C-C cells passed much less readily through 3 tt /a millipore filters. The latter finding suggests decreased deformability and in- rticle -p creased rigidity of erythrocytes containing C-C hemoglobin as compared to df/1 3 those with A-A hemoglobin. Morphologically, the present study presents 0/3 /4 evidence for pathologic aggregation of hemoglobin C molecules within C-C ceils, 4 3 /1 even at the reticulocyte stage. Aggregates form a meshwork, interspersed with 08 3 spaces containing primarily water and electrolytes. This aggregation leads to a 33 8 decrease in freedom of molecular movement, and thus an internal rigidity of /44 3 .p C-C cells. This rigidity accounts for the increased mechanical fragility j of d f b these ceils (decreased ability to deform when compressed by glass beads) and y g u decreased filtrability of such cells through miUipore filters (7). Such rigid cells est o have been found to lose membrane by fragmentation in passage through the n 1 1 microcirculation. Simultaneously, the intracellular water content decreases as F e b does the surface:volume ratio of the cell (22). In the present study, freeze- ru a etched replicas of C-C cells in which water loss has been produced by gradual ry 2 0 2 osmotic dehydration show a progressive increase in molecular packing, with 3 concomitant decrease in intermolecular spacing, to the point where the inter- molecular distance can no longer be resolved by this technique. Studies of low angle X-ray scattering by human red ceils after incubation in normotonic, hypotonic, and hypertonic solutions suggest a mean (center to center) distance of 56 A for normal ceils, 58 A in osmotically swollen cells, and 51 A in osmoti- cally shrunken cells; i.e., a progressive decrease in intermolecular distance, and 1 Lessin, L. S. Unpublished data. 452 MOLECULAR MECHANISM OF I~MOLYTIC ANEMIA increase in mean corpuscular hemoglobin concentration as intracellnlar water decreases (23). In osmotically dehydrated C-C cells, in the present study, it appears that not only do hemoglobin molecules approximate into a tightly packed system, but when a critical level of packing is achieved, alignment of molecules begins, initially in the juxtamembrane region, and subsequently in nidi within the cell interior. This may represent a precrystaUine alignment and give rise to very small, rigid paracrystals, which by virtue of their surface loca- tion, can be "pitted" from these cells within the splenic microcirculation by a mechanism analogous to that described for Heinz bodies (24). With further loss of intracdlular water, and consequent tighter packing of molecular aggregates, a second critical decrease in intermolecular distance occurs, that which cor- D o w responds to solubility saturation point of hemoglobin C, and crystallization n lo a takes place. These crystal cells appear to be extremely short-lived in the patient, d e d probably destroyed in a single passage through the microcirculation, since very fro m few circulation "crystal ceils" can be found in dried peripheral blood films of h ttp nonsplenectomized C-C patients and, after splenectomy, they may increase to ://ru as high as 3 % (6). In a single passage through a 3/~ millipore filter in normotonic pre buffer, 83 % of C-C cells become entrapped in the filter mesh, whereas only 47 % ss.o rg normal cells fail to cross the filter. This difference is accentuated when cells are /je m rendered A raluceloM more rigid Model by for osmotic eht lacigolohtaP dehydration roivakeB (7). of nibolgomeH C-C -orhtyrE /article -p cytes.--Aggregation, precrystalline alignment, and intracellular crystallization df/1 3 of hemoglobin C molecules appear to be the characteristic fine structural ab- 0/3 /4 normalities of hemoglobin C molecules observed in freeze-etched preparations 43 /1 of serially dehydrated C-C erythrocytes. Moreover, this pathological behavior 08 3 must derive directly from the anomaly of primary structure of hemoglobin C, 33 8 /4 namely, the substitution of lysine, with its two amino groups and net positive 4 3 .p charge, for glutamic acid, a dicarboxyl amino acid with net negative charge (2). d f b The external position of the positively charged polar group in the A-helix of the y g u beta chain, distant from areas of intramolecular alpha-beta interaction, makes est o this site readily available for formation of a polar bond with a negative external n 1 1 site for an alpha chain of a neighboring hemoglobin C molecule (of. Perutz and F e b Lehmann (25), once a critical intermolecular distance is overcome. Thus, patho- ru a logical polymerization accounts for the increase in red cell rigidity encountered ry 2 0 2 in C-C erythrocytes. This progresses from an aggregation meshwork in young 3 cells to crystallization in the aging cell which, by symmetrical loss of membrane lipid (26) decreases its surface to volume ratio, loses intracellular water (22), and reduces intermolecular distance. A comparable mechanism of hemolysis has been described for the rat paracrystalline erythrocyte (27), in which hemo- globin exists in a "metastable state of incipient crystallization" (28). In sickle cells hemoglobin S molecules polymerize into a six-stranded helix (29, )21 rather than a crystal, similarly a function of alteration of primary structure of