Dynamics of shape recovery by stored red blood cells during washing at the single cell level

Abstract

BACKGROUND:

Hypothermic storage transforms red blood cells (RBC) from smooth biconcave discocytes into increasingly spherical spiculated echinocytes and, ultimately, fragile spherocytes. Individual cells undergo this transformation at different rates, producing a heterogeneous mixture of RBCs at all stages of echinocytosis in each unit of stored blood. Here we investigated how washing (known to positively affect RBC properties) changes morphology of individual RBCs at the single-cell level.

STUDY DESIGN AND METHODS:

We tracked the change in shape of individual RBCs (n = 2870; drawn from six 4-6-week old RBC units) that were confined in an array of microfluidic wells during washing in saline (n = 1095), 1% human serum albumin (1% HSA) solution (n = 999), and the autologous storage supernatant (control, n = 776).

RESULTS:

Shape recovery proceeded through the disappearance of spicules followed by the progressive smoothening of the RBC contour, with the majority of changes occurring within the initial 10 min of being exposed to the washing solution. About 57% of all echinocytes recovered by at least one morphological class when washed in 1% HSA (36% for normal saline), with 3% of cells in late-stage echinocytosis restoring their discoid shape completely. About 1/3 of all spherocytic cells lysed in either washing solution. Cells washed in their autologous storage supernatant continued to deteriorate during washing.

CONCLUSION:

Our findings suggest that the replacement of storage supernatant with a washing solution during washing induces actual shape recovery for RBCs in all stages of echinocytosis, except for spherocytes that undergo lysis instead.

Introduction

The functional properties of human red blood cells (RBCs) deteriorate progressively throughout the allowable duration of hypothermic storage ex vivo.^1,2 During storage, the RBC membrane develops protrusions that become increasingly spicular and ultimately leave the cell body as microvesicles.^3,4 This process results in a disproportionately greater loss of membrane surface area than cell volume, leading to a substantial increase in the sphericity of the cell. Affected RBCs transform from biconcave discocytes, through intermediate stages of echinocytosis, into spherocytes, and ultimately undergo lysis.^5–8 This storage-induced change in shape may significantly reduce the ability of stored RBCs to perfuse the microvasculature,^9–11 predispose the cells for removal from circulation soon after transfusion¹² and serve as a quantitative biomarker of the storage lesion overall.¹³

Previous studies have shown that washing either in a human serum albumin (HSA) solution,^8,15 or in normal saline^10,16–19 can significantly improve morphology of stored RBCs in vitro. The ability of stored RBCs to rapidly restore normal shape is known to correlate strongly with post-transfusion viability (24-hr recovery) in vivo.²⁰ Furthermore, animal studies have shown that washing of stored RBCs can significantly reduce plasma iron and severity of end organ damage in the guinea pig model,²¹ and significantly improve outcomes in the model of canine pneumonia.²² In all these studies RBC morphology was measured as a population statistic, by simply counting the fraction of cells of each morphological class, before and after the washing procedure. However, the extent to which individual RBCs undergo the echinocytic transformation varies greatly – after the first 2–3 weeks of storage, the population of stored RBCs in a single unit consists of a complex mixture of cells from every morphological class, even within the same unit.^6,14 It is therefore not clear whether washing improves RBC morphology by selectively destroying cells with specific morphologies (similarly to the selective lysis of spherocytes in hypotonic conditions¹⁷), or by actually restoring the shape of individual cells.

In this study, we investigated the dynamics of RBC shape recovery in response to washing with 1% HSA washing solution and in normal saline, by tracking the behavior of individual RBCs confined within an array of microfluidic wells. We found that a large fraction of RBCs in all stages of the storage-induced echinocytosis could recover their shape during washing, while spherocytic RBCs could not, undergoing selective lysis instead. Washing, therefore, appears to improve the overall morphology of the RBC population within the unit by both restoring the shape of relatively well-preserved RBCs, and destroying irreversibly-damaged RBCs.

Materials and Methods

Fabrication of microwell arrays

Microwell arrays were fabricated out of polydimethylsiloxane (PDMS) using soft lithography.¹⁷ Briefly, SU-8 photoresist (Microchem, Westborough, MA) was spin-coated onto a 4 inch silicon wafer (University Wafer, South Boston, MA) at a depth of approximately 50 μm. The wafer was then exposed to UV light through a photomask (Photo Sciences, Torrance, CA) containing a negative image of a 100 × 100 array of 50 μm × 50μm wells. Following development, the wafer was used as the master mold. To create a replica of the microwell array, PDMS (1:10 curing agent to base; Sylgard 184, Dow Corning, Midland, MI) was poured over the master wafer and allowed to cure at 65°C for 3 hours. Once cured, the array was carved out and plasma oxidized (100 seconds, PDC-3xG, Harrick Plasma, Ithaca, NY) to render the surface hydrophilic. The array was then placed onto a glass slide with a PDMS frame around the perimeter to keep the washing solution contained.

Samples of stored RBCs

RBC units (CPD>AS-1, leukoreduced, stored for 4-, 5- and 6-weeks, n = 2 for each storage duration) were purchased from the Gulf Coast Regional Blood Center (Houston, TX), and stored in a blood bank refrigerator (Helmer iB111, Helmer Scientific, Noblesville, IN) at 2-6°C until use. To perform an experiment, two samples from each RBC unit were obtained via the aseptic technique. One RBC sample was centrifuged at 225×g for 10 minutes (Microcentrifuge 22R, Beckman Coulter, Fullerton, CA) and the supernatant was removed. The second sample was then diluted to approximately 0.05% hematocrit (HCT) with the supernatant of the first sample and allowed to mix by gentle inversion for 5 minutes (Labquake, Barnstead Thermolyne, Dubuque, IA). Diluted samples were used immediately.

Washing of stored RBCs at single cell level

Human serum albumin (HSA; Sigma-Aldrich, St Louis, MO) was dissolved at 1% (w/v) in normal saline to create the 1% HSA washing solution (Thermo Fisher Scientific, Waltham, WA). A sample of diluted stored RBCs was deposited onto the microwell array, and RBCs were allowed to sediment to the bottom of the wells, loading the wells at a density of ~4 cells per well. Upon loading, images of the wells were acquired at 40× magnification on an inverted bright field microscope (IX79, Olympus American, Inc., Center Valley, PA) using a high-speed camera (MC1362, Mikrotron GmbH, Unterschleisheim, Germany). While continuously acquiring the images, about 1 mL of washing solution (either 1% HSA, normal saline, or autologous storage medium) was dispensed onto the microwell array, and the cells were allowed to remain in solution for a total of 20 minutes. Upon completion of the experiment, captured images were used to determine morphology of each individual RBC: flat biconcave cells were classified as discocytes (D), convex-concave shaped cells as stomatocytes (ST), flat cells with an irregular contour as echinocytes 1 (E1), flat cells with a few spicules on cell face as echinocytes 2 (E2), rotund cells with many spicules as echinocytes 3 (E3), spherical cells with short, thin spicules as sphero-echinocytes (SE), smooth spherical cells as spherocytes (S), and lysed cells as ghosts (G). Classification of images was performed by an expert observer who was not blinded to the identity of the samples. Morphological Index (MI) was calculated by assigning a numerical value for each morphology type (ST = −1, D = 0, E1 = 1, E2 = 2, E3 = 3, SE/S = 4, G = 5), adding these values together and dividing the sum by the total number of cells counted, as previously described.¹⁵

Statistical analysis

Statistical analysis was performed using MATLAB Statistical Toolbox (The MathWorks, Inc., Natick, MA). The one-tailed paired t-test was used to determine statistical significance of the differences between two different samples (p < 0.05 was considered significant). The results are presented as mean values ± standard deviation (SD).

Results

Tracking changes in RBC shape at the single cell level

Figure 1 illustrates the experimental setup we used to track the change in shape of individual RBCs. A 4μL sample of stored RBCs diluted down to ~0.05% HCT with autologous storage medium (obtained from the same unit) was deposited onto the array of microfluidic wells, and the RBCs were allowed to sediment to the bottom of the wells (Fig. 1a). After the cells settled completely (typically, within 5 min), 1 mL of a washing solution (1% HSA, saline, or autologous storage medium as control) was carefully deposited on top of the microwell array to initiate washing. The microfluidic wells were sufficiently deep (~50μm) for the cells to remain within their wells even as washing solution was flushed over the array. Image acquisition started after the cells were loaded in the wells, and continued at a rate of one frame per second for another 20 min. Because cells were confined to their specific wells throughout the experiment, and their position within the wells did not change significantly between the consecutive frames, we were able to track the morphology change while preserving the identity of each individual RBC (Fig. 1b).

Figure 1:

Experimental setup used for tracking RBC shape change at single cell level. (a) A dilute suspension of stored RBCs was placed onto the microwell array, and cells were allowed to sediment to the bottom of the wells. The microwell array was then covered with an excess of washing solution, and images of RBCs in the wells were acquired every second for 20 minutes. (b) A typical sequence of images showing the change in shape of two different RBC over time (dashed arrows). Scale bar is 20 μm.

Morphology of stored RBCs prior to washing

A total of 2870 cells from all units were tracked in this study (n = 987 from 4-week old units, n = 921 from 5-week old units, and n = 962 from 6-week old units). As expected, initially each unit contained a highly heterogeneous mixture of RBCs in all stages of morphological deterioration. Prior to washing, <1% of all cells in each unit were stomatocytes (ST), and on average 3% were either sphero-echinocytes (SE) or spherocytes (S), regardless of storage duration. The distribution of cells between discocytes (D) and echinocytes (E), however, depended on the duration of storage. The fraction of D cells decreased by about 10% with each additional week of storage, from 34% in 4-week old units, to 25% in 5-week and 15% in 6-week old units. The fraction of E cells increased by about 10%, from 60% in 4-week old units, to 71% in 5-week and 81% in 6-week old units. Overall, about 25% of all cells tracked in this study were D, and 71% were cells in various stages of E, prior to washing.

Recovery of RBC shape during to washing

Our direct visual observations showed that recovery of echinocytic RBCs occurred through first the disappearance of spicules, and then the progressive smoothening of the RBC contour when cells were introduced to a washing solution (please see Supplementary Video). Figure 2 illustrates the change in shape of individual RBCs after a 10-min incubation with a washing solution. For 1% HSA, more than half of all E cells recovered shape by at least one morphological class (Table 1A), with nearly a quarter of E2 and even some E3 cells undergoing a complete restoration all the way to D (Fig. 2a). E2 appeared to be the category most susceptible to morphological recovery, with only a small fraction retaining their initial shape after washing in 1% HSA (Table 1A). Morphological recovery was less pronounced when normal saline was used as the washing medium, with about one third of all E cells recovering by at least one morphological class (Table 1B), and only some of E2 (and none of E3) cells restoring their shape to D (Fig. 2b). Once again, E2 was the category most susceptible to morphological recovery, with only about a third of all E2 cells retaining their initial shape after washing in normal saline (Table 1B). None of the E cells washed in the autologous storage medium recovered shape in this study (Table 1C).

Figure 2:

Shape recovery map for washing stored RBCs with (a) 1% HSA washing solution (n = 999), and (b) normal saline (n = 1095). Arrows indicate the direction of change, numbers indicate % of cells with a specific initial morphology that changed to other morphological classes (e.g., 63% of E2 cells recovered to E1, and 24% recovered to D, when washed in 1% HSA).

Table 1:

Change in shape by individual RBCs after washing for 10 min with (A) 1% HSA (n = 999), (B) normal saline (n = 1095), and (C) the autologous storage medium (n = 776).

(A)		Initial shape
		ST	D	E1	E2	E3	SE/S
Shape after washing	ST	2	2	0	0	0	0
	D	0	257	161	29	9	0
	E1	0	0	120	76	95	0
	E2	0	0	0	15	29	0
	E3	0	0	0	0	159	0
	SE/S	0	0	0	0	0	28
	G	0	0	1	0	0	16
(B)		Initial shape
		ST	D	E1	E2	E3	SE/S
Shape after washing	ST	11	2	0	0	0	0
	D	0	283	108	3	0	0
	E1	0	0	189	100	23	0
	E2	0	0	0	62	36	0
	E3	0	0	0	0	228	0
	SE/S	0	0	0	0	0	32
	G	0	0	0	0	1	17
(C)		Initial shape
		ST	D	E1	E2	E3	SE/S
Shape after washing	ST	5	0	0	0	0	0
	D	0	142	0	0	0	0
	E1	0	28	280	0	0	0
	E2	0	0	13	62	0	0
	E3	0	0	0	16	212	0
	SE/S	0	0	0	0	4	11
	G	0	0	0	0	0	3

Deterioration of RBC shape during washing

No shape recovery was observed for SE/S and ST cells in either washing solution (Fig. 2). A substantial fraction of SE/S cells – more than a third for both 1% HSA and normal saline – underwent lysis during the initial 10 min of washing (Fig. 2, Table 1). Lysis of cells with morphology other than SE or S was a very rare occurrence – only one E1 cell lysed during washing in 1% HSA (Table 1A), and one E3 cell lysed during washing in normal saline (Table 1B). Additionally, we observed only two D cells transforming into ST after 10 min of washing in either 1% HSA solution or normal saline, accounting for less than 1% of the D population in both cases (Fig. 2, Tables 1A and B). In stark contrast, the morphology of RBCs incubated in their autologous storage solution continued to deteriorate further over the same time period (Table 1C). We observed about 16% of D cells deteriorate to E1, 4% of E1 – to E2, 21% of E2 – to E3, 2% of E3 – to SE/S, and 21% of SE/S undergoing lysis after 10 min of incubation in autologous storage solution (Table 1C).

The effect of prolonged incubation on RBC shape recovery

Figure 3 shows the change in morphological index (MI) calculated for RBCs prior to washing and after incubation in 1% HSA, normal saline, and autologous storage medium for 10 and 20 min. Initially, the MI was 1.57 ± 0.31 for RBC samples washed with 1% HSA (n = 6), 1.54 ± 0.21 for samples washed with normal saline (n = 6), and 1.48 ± 0.21 for those washed in their autologous storage medium (n = 6), and were not significantly different from each other. Washing in 1% HSA decreased the MI from its initial value to 1.03 ± 0.18 (p = 0.00651) after 10 min, and further down to 0.99 ± 0.20 by the end of the full 20-min incubation (all differences were statistically significant). Washing in normal saline followed a similar trend, with the MI decreasing to 1.28 ± 0.18 (p = 0.00014) during the initial 10 min, and further down to 1.21 ± 0.19 after 20 min of incubation (all differences were statistically significant). In stark contrast, MI of samples washed with their autologous storage media increased to 1.57 ± 0.19 during the initial 10 min, and further to 1.58 ± 0.18 after full 20 min of incubation in autologous storage solution (all differences were statistically significant). Although there was no difference between RBC samples prior to washing, the difference between MI of RBC samples washed with 1% HSA and normal saline were statistically significant, and both were significantly different from samples washed in autologous storage media after 10 min of washing. After 20 min of incubation, the MI of RBC samples washed in 1% HSA was no longer significantly different from that of samples washed in normal saline (p = 0.063).

Figure 3:

The change in morphological index over time for RBCs washed in (Δ) autologous storage medium, (□) normal saline, and (○) 1% HSA. Statistically significant differences are denoted with * (p-value < 0.05), and ** (p-value <0.01).

Discussion

To the best of our knowledge, this study is the first to document washing-induced recovery of RBC shape at the level of individual cells. The progressive change of RBC shape from healthy, flexible discocytes to spiculated echinocytes and ultimately to non-deformable, fragile spherocytes is a hallmark of the storage-induced deterioration of RBCs quality.^4,7,13,23 By the end of the allowable storage duration, the population of RBCs in a typical unit contains a highly heterogeneous mixture of cells at all stages of echinocytosis.^6,14,15 Previous studies of the effect of washing on properties of stored RBCs have shown improvement of overall RBC morphology, as indicated by an aggregate morphology index, or a change in the percent-fraction of cells belonging to different morphological classes.^8,15 These studies, however, evaluated morphology by classifying images of a few hundred cells withdrawn from a RBC sample before washing, and of a few hundred different cells withdrawn from the RBC sample after washing. Because of the morphological heterogeneity and the loss of identity of cells classified to score morphology it was impossible to tell whether the observed improvement of the overall RBC morphology due to washing was caused mostly by preferential lysis of cells in the latest stages of echinocytosis, or the shape of individual cells actually recovered. And if true shape recovery did occur, which stages of echinocytosis and to what extent participated in this process. Our single-cell washing system was able to answer these fundamental questions by tracking the evolution of shape change for individual RBCs via time-lapse microscopy throughout the entire washing process.

Multiple factors are known to contribute to the echinocytic transformation of RBCs during storage, including changes in the actin-spectrin network,^24,25 band 3 clustering,^26–28 loss of phospholipid asymmetry,²⁹ and oxidative damage.^7,30–32 As cells age in storage, they lose ATP which is necessary for maintaining proper membrane configuration.^33,34 Oxidation damages lipids and proteins of the cells, triggering protein aggregation and blebbing of the membrane in the form of characteristic spicules that ultimately leave the cell body as microvesicles.^35–38 The progressive loss of membrane surface area through microvesiculation is an irreversible process which ultimately transforms the RBC into a virtually spherical cell (SE/S). Our study shows that microvesiculation does not preclude shape recovery even for cells at advanced stages of echinocytosis, with a significant fraction of E3 cells being able to recover shape all the way to E1 and even D morphologies (particularly in the presence of HSA). None of SE and S cells were able to recover shape when incubated with either washing solution, confirming that the loss of excess membrane surface area by these cells is too great to allow shape recovery. Interestingly, the shape of cells ‘washed’ in their autologous storage medium continued to deteriorate throughout the experiment, while cells from the same RBC units washed in either normal saline or 1% HSA solution recovered shape. All washing experiments were carried out at room temperature (not at 4°C), regardless of the washing solution. This observation suggests that composition of the washing solution rather than temperature was likely the cause of the continued deterioration of RBCs washed in autologous storage medium. Further research is needed to discern whether the removal of detrimental factors in the autologous storage medium, or the addition of protective factors with a fresh washing solutions, or both were the reason for our observation. Our findings further confirm that although RBC shape recovery was certainly enhanced by HSA, its presence was not required, which agrees well with our previous studies.^8,10,15–18

A majority of shape changes occurred within the initial 10 min of RBCs being exposed to the washing solution, which is on par with our previous studies^15–18 and agrees well with the time it typically takes to wash a unit of RBCs using a conventional centrifugation-based cell processor.³⁹ Outside of hypothermic storage, normal RBCs can be transformed into echinocytes by incubation with sodium salicylate, a well-known echinocytogenic agent, and such a transformation is fully reversible upon exchange of the suspending medium.^5,9 Unlike exposure to sodium salicylate, the effect of hypothermic storage on shape of individual RBCs in a population is highly heterogeneous.^9,14 Nevertheless, the ability of a sizable fraction of stored RBCs in E3 morphology to fully recover their shape in a manner that appears remarkably similar to the reversal of the shape changes induced by sodium salicylate is intriguing. More research is needed to elucidate the mechanisms of RBC shape recovery under these rather different circumstances.

It is noteworthy that the mere act of replacing the autologous storage medium with an isotonic washing solution was sufficient to cause lysis of about 1/3 of all spherical (SE/S) cells in the first 10 min, without any need for excessive mechanical forces (e.g. due to centrifugation¹⁹), or osmotic stress (e.g. during washing in mildly hypotonic buffer¹⁷). Lysis affected predominantly SE/S cells and occurred significantly more in fresh washing solutions (at about the same rate for both normal saline and 1% HSA) than for autologous supernatant control. These observations suggest that our single-cells washing method was sufficiently gentle to avoid mechanically damaging fragile RBCs, and that the difference in the rate of lysis was due entirely to the differences in composition of the washing solutions and the autologous storage supernatant. There is of course a rather substantial difference between our single-cell washing method and how RBC units would be washed in practice – in the current research and clinical settings washing of stored RBCs is typically performed using a high-volume centrifuge or a centrifugation-based automated cell processor.³⁹ Centrifugation subjects stored RBCs to substantial mechanical forces, damaging the cells and causing excessive hemolysis. Although this mechanical washing is known to improve the overall morphology of RBC units, damaged cells continue to release free Hb and potassium into the suspending medium, reaching the pre-wash levels within 24 hours after the procedure.^19,22,40,41 Our current findings and previous research highlight the need for further development of novel centrifugation-free washing systems^16,18 capable of replacing the storage medium with washing solution (and thus conferring the beneficial effects of washing), while minimizing damage to washed RBCs.^16,18,19

Our previous work has shown that advanced echinocytosis significantly diminishes the ability of stored RBCs to perfuse microvascular networks in vitro,⁹ which can be partially restored by washing stored RBCs in either isotonic¹⁰ or hypotonic¹⁷ saline, or in 1% HSA solution.¹⁵ The results of the current study suggests that this rheological improvement could, at least in part, be due to the rapid restoration of morphology of individual RBCs when storage supernatant is replaced with a fresh suspending medium. If a similar level of shape recovery also occurs once stored RBCs are transfused into the body of the recipient, the resulting effect on blood rheology could potentially mitigate the negative effects of impaired RBC deformability typically associated with storage lesion. Irrespective of RBC shape recovery, however, patients may still benefit from the removal of harmful mediators contained in the RBC storage supernatant by washing.^42–45 Our study further showed that SE/S cells were the only morphological class of RBCs that were irreversibly damaged by storage. Our previous work and current findings suggest that SE/S cells undergo lysis readily, even upon relatively minor perturbations such as medium exchange with isotonic or slightly hypotonic saline,¹⁷ and conventional centrifugation.¹⁹ It is likely therefore that SE/S cells will undergo hemolysis soon after transfusion into the recipient, potentially contributing to the adverse outcomes that have been associated with transfusion of stored blood.^46–49 (And those SE/S cells that don’t, will be rapidly removed from the circulation by the spleen.¹²) Washing stored RBCs in 1% HSA solution or even in normal saline could substantially reduce the fraction of irreversibly damaged SE/S cells transfused, and thus potentially improve transfusion outcomes.

There are several factors that may limit interpretation of results obtained in this study. First of all, our single-cell washing method is obviously different from what would be done by a blood bank to wash stored RBCs. The conventional washing procedure involves adding washing solution to the RBC unit which is then centrifuged to replace the supernatant with a fresh solution, and is typically performed using a high-volume centrifuge or an automated cell processor.³⁹ The differences in HCT of the suspension, the mechanical forces applied, the washing solution used, as well as the temperature at which washing is performed all could have an impact on how the shape of individual RBCs evolves throughout this process. Thus, care should be taken when extending interpretation of our findings in the context of shape recovery for RBCs preserved with alternative storage systems, washed in different solutions and/or under different conditions.

Additionally, the morphological classification was performed by an un-blinded expert, and although all of the classified images were preserved and all classification decisions were reviewed for accuracy by a second expert, some observer-related bias in classifying shape of individual RBCs is still possible. Automated morphology classification approaches that are free from the observer-related bias^14,50 could provide a more accurate quantitative assessment of RBC shape recovery, which would be particularly useful for screening different washing solutions for how efficiently they may restore RBC properties.

Finally, multiple recent clinical trials have shown that transfusion of RBC units stored longer (up to 35 days) does not in general worsen clinical outcomes.^51–56 However, questions about safety of the oldest RBC units (stored for 35-42 days), particularly in cases of massive / frequent transfusions to certain susceptible patients, remain unanswered.^57,58 Patients with sickle cell disease receiving the exchange or simple (top-up) RBC transfusion therapy are one such group for whom safety and efficacy of stored RBCs requires further improvement.⁵⁹ Further improvement of stored RBCs as a therapeutic product may come from developing low-cost technologies for identifying and removing RBCs that had been damaged irreparably by storage, as well as from continuing to develop new methods for preventing damage during storage. The results of this study have the potential to contribute to both directions.