Novel single-cell measurements suggest irreversibly sickled cells are neither dense nor dehydrated

Abstract

In sickle cell anemia, deoxygenation causes erythrocytes to distort, while reoxygenation permits them to recover a normal biconcave disk shape. Irreversibly sickled cells (ISCs) remain distorted when reoxygenated and have been thought to have among the highest intracellular hemoglobin concentrations of the sickle red cell population and therefore the greatest vulnerability to vasoocclusion. Using a new optical method, which we describe, we have made precise measurements of the intracellular hemoglobin concentration, and intracellular O₂ saturation, of ISCs, as well as oxygenated sickle cells with a normal biconcave disc shape, and cells with shapes distorted by the sickle fibers they contain. This method also provides good estimates of cell volumes, and hemoglobin per red cell. The concentration distribution of the ISCs is found to be similar to normal, discoid cells. Average ISC volumes exceed their discoid counterparts, with a much broader distribution, arguing against dehydration as their origin. The concentration distribution of the polymer-laden sickled cells is significantly higher in mean value, and their volume distributions indicate some dehydration. Previous assumptions about ISCs may have thus been colored by the presence of sickle cells that did contain polymer, and true ISCs may be much more benign than once thought, which underscores the importance of accurate measurement on individual cells. This method could be used to follow changes in individual cell properties under various specific perturbations, and where characterization by flow cytometry is infeasible.

Figure 1

Illustration of sample assembly (not to scale), with cell images obtained on it. (a) A top view shows the spread out drop of blood on the coverslip. Small, included regions remain uncovered by blood and are used for thickness measurements. The circular optical flats are clamped in a housing (dark gray) that is placed between microscope objectives. (b) A side view (top) illustrates the blood drop placed on the coverslip before the top flat is pressed down. A layer of immersion oil is between the coverslip and the bottom flat. The lower view illustrates the compressed final structure. Note that the coverslip allows the curvature of the drop’s edge to be less than it would be without the pedestal. (c) An illustration of a red cell compressed in the drop. The top drawing is the uncompressed cell. Note that when viewed this way the edge of the cell is approximately circular (hence the circle shown), and this is used in the calculation of cell volume. (d–g) Representative cells from this arrangement, imaged at 415 nm, viewed in intensity. For clarity, the range of gray dark pixels was made to cover the full span of grayscale, but linearity of response was not adjusted. This is the equivalent to plotting a graph so as to fill the available graph range. Not every cell in each image was subject to analysis. Note the variety of forms. Cells are characterized as discocytic when the ratio of their major axes is less than 2; otherwise cells are considered distorted. These cells were from the first three patients before an O₂ flushing protocol was adopted.

Figure 2

Typical reflectance spectrum and fit from Filmetrics thickness measurement. The spectrum is reflectance from 400 to 1000 nm, and is the rough blue line. This was fit from 500 to 800 nm, as shown by the smooth red curve, which yields the thickness. Here, the thickness is 1930 nm, illustrated in the upper right-hand corner.

Figure 3

Absorbance spectra measured on a 2.56 μm² area of each erythrocyte from three different cells illustrating different degrees of oxygenation, Y. (a) Y = 0.93. (b) Y = 0.53. (c) Y = 0.26. In all cases, the circles are the data points, and the solid curve is a fit using the sum of standard spectra for oxyHb, deoxyHb, and a scattering component that was featureless but had a λ⁻⁴ dependence.

Figure 4

Number of cells as a function of concentration for blood from the first three patients. (a) Cells showing no distortion (discocytes) (245 cells). (b) All distorted cells (305 cells). Representative cells are seen in Fig. 1. Discocytes are defined by having the ratio of their axes less than 2; otherwise cells are considered distorted (c). Cells from distribution shown in (b) for which O₂ saturation precludes polymer formation (36 cells). (d) Cells from distribution in (b) for which O₂ saturation and concentration rise above solubility and therefore should contain polymers (269 cells). While (b) shows a distribution that would be considered typical of ISCs, the dissection based on O₂ saturation argues that only the population of (c) should truly be regarded as ISCs. Note the strikingly shifted mean values of the distributions in (b) and (d) compared with (a) and (c).

Figure 5

Percent of cells at a given percent deoxygenation. (a–c) Are discocytes from patients 1, 2, and 3, respectively. (d–f) Are distorted cells from patients 1, 2, and 3. Percentages are calculated relative to a given category, e.g., percent of discocytes that have a given level of deoxygenation, rather than as a percentage of the whole number of cells. ISCs systematically show greater levels of deoxygenation, further confirming the presence of polymers in many cells thought to be ISCs from morphology alone.

Figure 6

Solubility as a function of solution fractional saturation. This red solid curve is based on an empirical formula expressing solubility c_s (g/cm³) = 0.183 + 0.0924 Y + 0.0980 Y² + 0.235 Y¹⁵ where Y is the fractional saturation. This accurately fit substantial data collected for both O₂ and CO and illustrates graphically that partially saturated solutions can have significant fractions of polymer. The dashed line is the mean concentration of the discocytes. Thus, for most sickle cells, polymers will persist until around 80% oxygenation and, for most ISCs shown in Fig. 1c, polymers will persist even beyond 90%, thanks to their higher concentration. This makes it very challenging to validate cells as truly polymer-free ISCs without direct oxygenation measurements of the cells themselves.

Figure 7

Number of cells at a given concentration for three classes of cells for all five blood samples: (a) discocytes and (b) ISCs all have oxygen saturations that are high enough to keep the Hb from polymerizing. Virtually all of these cells had been extensively flushed with pure O₂. (c) Polymerized cells that are distorted but have saturation low enough to allow polymerization. The discocytes of (a) have been fit by a Gaussian curve of mean 32.9 g/dL and the standard deviation is 4.9 g/dL. The ISCs of (b) are likewise fit by a Gaussian curve, with the fit from the discocytes superimposed as a dashed line. (The discocyte maximum has been adjusted for the difference in number of cells.) As is evident, the two curves almost overlap. In (c) the discocyte distribution is superimposed with a dashed line, and is in stark contrast to the high density of the sickled cells shown there.

Figure 8

Number of cells at a given volume: from top to bottom (a) discocytes, (b) ISCs, (c) polymerized cells. ISC volumes are notably broader than either discocytes or polymerized cells. Further information is provided in Fig. 3.

Figure 9

Number of cells at a given mass: (a) discocytes, (b) ISCs, (c) polymerized cells. Mass per cell was determined by combining the data of Fig. 3 (concentration) with that of Fig. 4 (volume), both of which were directly measured. The polymerized cells have dramatically greater mass per cell. ISCs have a very broad and flat distribution.