This study was approved by the institutional ethics committee and followed current protocols for research involving humans and the handling of biological samples (Declaration of Helsinki). Non-leukoreduced whole blood units were obtained from six healthy volunteer blood donors using triple CPD-Adsol® bags (Optipac, Fenwall, Baxter S.A., Lake Zurich, IL, USA). Whole blood units were spun at high speed (4000RPM, equivalent to 4100×g for 10min at 22°C) in a Beckman J6B centrifuge with rotor JS-5.2 (Beckman Coulter Inc., Palo Alto, CA, USA). Top and bottom technology was used for automatic fractionation (Optipress II, Fenwall, Baxter S.A., Deerfield, IL, USA). The resulting buffy coat-depleted RBC units in AS-1 were stored in standard blood bank conditions for six weeks. RBC units entered the study 72–96h (0.5 weeks) after collection due to the implementation of routine screening for infectious diseases, according to local regulations for blood processing. For the experiments, 10mL aliquots of each unit were obtained after 0.5, 2, 4 and 6 weeks of storage using an aseptic technique.

Considering that the reticulocyte count in blood bank RBC units is lower than in peripheral blood and the number decreases further during refrigerated storage as we have previously reported,9 a reticulocyte-enrichment procedure was designed to obtain culture inoculums with a reticulocyte count not below that usually expected in peripheral venous blood. Thus, before inoculation in culture bottles, reticulocyte-enriched fractions were obtained from RBC unit aliquots using Percoll density gradient separation (Sigma–Aldrich, St. Louis, MO, USA). Samples obtained from the RBC units were diluted 1:1 in a buffered solution containing 9mM Na2HPO4, 1.3mM NaH2PO4, 140mM NaCl, 5.5mM glucose and 0.8g/L bovine albumin at pH 7.40. Diluted samples were incubated at room temperature (25±2°C) for 30min. Continuous density gradients were prepared using 36.4% and 62.53% (v/v) Percoll solutions in 154mM NaCl. Subsequently, 6mL of each Percoll solution were dispensed in a gradient former (Bio-Rad Gradient Former Model 385, Catalog # 165-2000, Bio-Rad Laboratories, Hercules, CA, USA). Gradients were prepared in 15mL polypropylene centrifuge tubes (Falcon, Becton Dickinson Company, San Jose, CA, USA) and were stabilized by centrifuging at 900×g (2500RPM) for 5min in a 45° fixed-angled rotor centrifuge (Sorvall Biofugue Prime R, with a Highconic® rotor #75007588, Sorvall, Newtown, CT, USA). Six hundred microliters of diluted sample was added to each gradient and spun at 900×g (2500RPM) for 20min at room temperature. Percoll density marker beads (Sigma–Aldrich, St. Louis, MO, USA) were used to determine the density of continuous gradients. Flow cytometry with thiazole orange (Retic-Count, Becton Dickinson Company, San Jose, CA, USA) was used to measure the total reticulocyte count and reticulocyte subtype concentrations before and after separation as described below. The yield of the reticulocyte-enrichment procedure is presented in the Results Section.

Culture bottles were inoculated with a reticulocyte-enriched concentrated aliquot from each RBC unit obtained at the different sampling times. The enriched culture medium was based on RPMI-1640 with the addition of 23mM HEPES, 9mM inosine 5-monophosphate, 5mM adenosine, 4mM glutamine, 22mM NaHCO3, 100IU/mL penicillin, 100μg/mL streptomycin and 0.25μg/mL amphotericin B (all chemicals were obtained from Sigma–Aldrich, St. Louis, MO, USA). Osmolarity was adjusted to 295mOsm/kg and pH to 7.3 before sterile filtration. NaHCO3 was added as an extra buffer to keep pH stable at 5% pCO2 in the incubator. The pH was 7.45 after pCO2 equilibrium. To prevent cell adhesion, culture bottles were pre-incubated at 37°C for 30min with autologous plasma to achieve 20% (v/v) final concentration. Each concentrated aliquot was used for inoculation of a set of culture bottles that were incubated at 37°C in 5% CO2 atmosphere. Total RBC concentration in the bottles was 3.0±0.2×109cells/L. After incubation at 1- to 4-h intervals, one bottle was taken from each set for reticulocyte analysis.

Cell survival was monitored by measuring free hemoglobin released to the culture medium by spectrophotometry (HemoCue low Hb, HemoCue, Ängelholm, Sweden). RBC counts were determined by impedance cytometry and total hemoglobin by the sodium lauryl sulphate method in an automated hematology analyzer (Sysmex-R480, Sysmex America Inc., Mundelein, IL, USA). Percentage hemolysis was calculated as free hemoglobin percentage regarding total hemoglobin.

Reticulocyte percentage was determined by flow cytometry using thiazole orange staining. Two microliters of samples were added to 1mL of thiazole orange solution (Retic-COUNT, Becton Dickinson Company, San Jose, CA, USA). The mixture was stirred and incubated at 37°C in the dark for 30min. An unstained control was simultaneously prepared for each sample and analyzed by adding 2μL of the sample to 1mL of phosphate buffer solution and 0.1% sodium azide (w/v). Fluorescence intensity was measured at 530nm in a flow cytometer (FACSCalibur, Becton Dickinson Company, San Jose, CA, USA). Acquisition parameters were set using reticulocyte controls in whole blood (BD ReticCount Control, Becton Dickinson Company, San Jose, CA, USA); the results agreed with those expected by the manufacturer. Precision profile changes according to reticulocyte concentration had 28.2%, 1.8% and 13.3% variation coefficients (VC%) for 0.2%, 1.2% and 2.5% reticulocyte percentages, respectively. Determination coefficient (R2) test–retest was 0.99. The performance of the technique was thus close to that of fully-automated, clinical hematology analyzers (7.6–54.6% VC).11 One hundred thousand events were acquired for each sample at 1000events/s. The gate for RBCs was set from light scatter cytograms and fluorescence distribution scatterplots were analyzed using CellQuest Pro software (Becton Dickinson Company, San Jose, CA, USA) excluding auto-fluorescent cells. Reticulocytes were classified according to the intensity of their fluorescence as high (HFR), medium (MFR) and low fluorescence (LFR) subtypes (Figure 1).12 Light scatter and fluorescence intensity were plotted to analyze reticulocyte subtype size and fluorescence changes during culturing.13 For estimates of kinetics, reticulocyte counts expressed as percentage of total red cells were used, as were the absolute reticulocyte counts calculated from percentages and RBC counts.

Figure 1.

Reticulocyte analysis by flow cytometry using thiazole orange of a red blood cell unit stored for 0.5 weeks in AS-1. Fluorescence scatterplot from the red blood cell gate after excluding autofluorescent cells. Total reticulocytes were classified into subtypes having low (LFR), medium (MFR) and high fluorescence (HFR).12 FL1: fluorescence intensity; FSC: forward light scattering.

(0,08MB).

The results are expressed as means±1 standard deviation (SD). The calculated sample size was six RBC units for observing a difference in 0.1±0.05% in reticulocyte percentage or 10.0±4.6×109/L in reticulocyte counts, considering a power of 0.8, a 0.05 α level and comparing four time points during storage (0.5, 2, 4 and 6 weeks). The Friedman test was used to determine significance; the Statistical Package for the Social Sciences software (IBM SPSS) v23.0 was used for statistical analysis.

The Percoll densities were 1.070–1.088g/mL and densities of reticulocyte-enriched fractions were 1.0826–1.0844g/mL when fresh blood samples were used. With the enrichment procedure, fractions of from 3.7% to 5.3% reticulocytes were obtained for culture inoculum (Table 1) depending on the reticulocyte content in the RBC units at different blood bank storage periods.

The reticulocyte count became significantly and progressively lower during culturing while percentage hemolysis was low (0.2±0.2%); in all assays the percentage of hemolysis was less than the drop in the reticulocyte percentage. Figure 2 shows the patterns of total reticulocyte counts and reticulocyte subtypes during culturing for samples obtained from RBC units with different storage times in AS-1 after subtracting hemolysis. These graphical representations of data showed that the disappearance of reticulocytes cultured in vitro was a negative exponential function. The corresponding equation is:

Figure 2.

Reticulocytes enriched from red blood cell units stored in AS-1 had a reducing pattern in culture. HFR: high fluorescence reticulocytes; MFR: medium fluorescence reticulocytes; LFR: low fluorescence reticulocytes. The results are presented as the mean±1 standard deviation of six independent assays.

(0,34MB).

This equation expresses the number of reticulocytes surviving as a function of time with RET0 being the initial number of reticulocytes, e the basis for the Naperian logarithms and RET the number of reticulocytes present at any time t during the culture period; k is the reticulocyte maturation constant for each aliquot. Expressed in logarithmic form, Eq. (1) is as follows (Figure 3A):

Figure 3.

(A) Fit of Naperian logarithm of reticulocyte concentration measured as cells×106/μL (Ln#RET) vs. time. The figure gives an example of a reticulocyte sample obtained from a red blood cell unit after 0.5 weeks of blood bank storage. The linear regression model of the Naperian logarithm of reticulocyte count (Ln#RET) compared to time fitted the data for all the samples analyzed (R2=0.77–0.96: p-value=0.0001). (B) The time for disappearance in culture conditions also may be calculated by analyzing size and shape changes. Forward (FSC) and side (SSC) light scattering signals of reticulocytes (RET) decrease progressively until they become equivalent to those of mature red blood cells (RBC) with the time for disappearance corresponding to the moment at which the SSC RET/RBC and FSC RET/RBC ratios are equivalent to 1.0. Results obtained by this method correlated with those calculated by the method based on fluorescence (reticulocyte count) (R2=1.0; p-value <0.001).

(0,12MB).

Disappearance time as calculated from Eq. (2) is:

As the reticulocyte disappearance in culture was a mono-exponential function as we described above, some kinetic parameters were calculated by mathematical analysis. It was observed that the reticulocyte disappearance rate became faster as the blood bank storage time increased. The reticulocyte disappearance time was 144.0±46.3h for RBC units stored for 0.5 weeks and declined significantly for units stored for 2, 4 and 6 weeks (73.3±59.1; 27.1±29.0; and 15.1±14.3h; respectively; p-value <0.001 – Table 2).

In addition to the changes in fluorescence observed during cultivation, reticulocytes also showed changes in shape, manifested as a progressive decrease in side (SSC) and forward light scatter (FSC). The time for reticulocyte disappearance, defined by the moment at which the ratio of SSC of reticulocytes to mature RBCs (SSCRET/RBC) and the ratio of FSC of reticulocytes to mature RBCs (FSCRET/RBC) equaled 1.0, was calculated. Results obtained using this method were strongly correlated with the time for disappearance calculated by the fluorescence-based kinetic analysis (R2=1.0; p-value <0.001) (Figure 3B).

Moreover, reticulocyte subtypes showed distinctive disappearance kinetics; the time was faster for HFR and slower for MFR and LFR. Indeed, HFR completely disappeared after 42.5±8.5h, MFR at 73.4±20.8h and LFR at 269.9±98.8h for RBC units stored for 0.5 weeks. These times became significantly reduced for each reticulocyte subtype in RBC units having longer blood bank storage times. The kinetics of disappearance for HFR in units stored for six weeks could not be determined because they were undetectable.

Furthermore, the half-life and constant disappearance rate could be calculated from Eq. (2):

These additional kinetic parameters also showed differences related to blood bank storage time. The half-life of total reticulocytes obtained from RBC units stored for 0.5 weeks was 27.9±13.9h; this decreased significantly by the end of storage to 2.5±2.4h (p<0.001). HFR from RBC units after storage for 0.5 weeks had a half-life of 12.9±7.4h in culture, while this was 15.6±4.5h for MFR and 43.1±42.3h for LFR. These values decreased significantly as the RBC unit storage time increased (Table 2). The constant reticulocyte disappearance rate of RBC units at the beginning of storage −0.03±0.02h−1 increased to −0.74±0.66h−1 by the sixth week (p-value=0.001). HFR had a disappearance constant of −0.07±0.03h−1 at Week 0.5 which increased to −1.22±2.17h−1 by the 4th week of storage (p-value=0.01). For MFR, it was −0.05±0.02h−1 at Week 0.5 and −0.95±0.79h−1 by the end of storage (p-value=0.004). In the case of LFR, it was −0.03±0.02h−1 at Week 0.5 and −0.71±0.63h−1 by Week 6 (p-value <0.001 – Table 2).

To summarize, an exponential decay for reticulocytes in culture was observed in all six RBC units studied at different blood bank storage times in AS-1. However, the kinetic parameters revealed significant differences according to the length of storage: reticulocyte disappearance time in culture became shorter with increasing blood bank storage time for all aliquots analyzed. Bivariate regression was used to analyze the relationship between reticulocyte disappearance time during culturing and previous blood bank storage duration giving the following model: