To the Editor:

Decades ago it was observed that young cohorts of radiolabeled human red blood cells (RBC) became progressively enriched in the more dense cell fractions as the cohorts aged.1 It was also shown that the most dense cells were short-lived in circulation after reinfusion into the donor.2 Consequently, density gradient centrifugation has served as the cornerstone methodology for the isolation and study of senescent human RBC. However, such studies have not quantitatively assessed the degree of enrichment for old cells in dense fractions, and the observed dispersion of RBC cohorts of any particular age throughout the entire density gradient has been bothersome to many.3

Quantitative studies of this issue in animals have generally intensified the controversy. Experiments on hypertransfused rats have shown that a progression of cells to higher densities (measured by mean corpuscular hemoglobin concentration [MCHC]) occurs early in the RBC life span, suggesting that density changes in rats are associated primarily with maturation rather than senescence.4 Similarly, hypertransfused mice have been shown to display minimal age-associated changes across RBC density fractions.5,6 Analysis of biotinylated rabbit RBC (B-RBC) cohorts during their last 10 days of survival has shown an enrichment of only twofold to threefold over unmarked (younger) cells in the most dense fractions,7 and these senescent B-RBC comprise only $13$ of the RBC in the dense fraction. Such results have tended to raise questions regarding the validity of density fractionation as a tool for isolating senescent RBC.

We have been evaluating RBC senescence in the dog based on the conviction that the dog is a more appropriate model of human RBC senescence than are small laboratory mammals. This assumption is justified by strong similarities between human and dog RBC in important features such as their comparatively long life spans (∼110 to 120 days) and their low amounts of random RBC loss. We have validated the biotinylation system for use in the dog and have shown that cells greater than 104 days exhibit many characteristics of senescent human cells (manuscript submitted), including increased autologous IgG binding.8 We have now used this unequivocal method of retrieving aged dog RBC to address two important questions: (1) at what cell age do RBC become a significant component of the most dense fraction of cells? and (2) what is the approximate age distribution of the 1% most dense cells in circulation? We predicted that cells in all fractions would be ∼100% biotinylated at day 1 and that, with time, the percentage of nonbiotinylated cells would increase, beginning in the lightest fraction and progressing to the most dense fraction.

As shown in Fig 1, isolation of dog RBC on arabinogalactan density gradients resulted in a crude cell age-based separation, with younger unlabeled cells usually but not invariably entering fractions of increasing density consecutively. Further, unlabeled RBC entering circulation after biotinylation were generally excluded from the most dense fraction for much of the RBC life span (see hatched squares in Fig 1). Thus, dog 331 showed ≤6% nonbiotinylated cells in this fraction until after day 72, but contained ∼25% of the younger cells at day 86 (Fig 1A, Table 1). Dog 794 showed ≤5% nonbiotinylated cells until day 86 and wasn't significantly contaminated with younger cells until between days 93 and 100 (Fig 1B, Table 1).

Fig. 1.

Two mature male beagles were infused intravenously with N hydroxysuccinimido (NHS)-biotin (35 mg/kg body weight in dimethyl sulfoxide) to biotinylate all circulating RBC (99% each), as described.9 One dog (794) was recovering from an idiopathic anemia and had a low normal hematocrit of 37% at labeling that increased to 52% by day 48, resulting in a partial cohort effect. The other dog (331) was assessed to be completely normal. Blood was taken for density gradient centrifugation at roughly 2-week intervals initially and then weekly from day 86 to 105 to determine the proportions of B-RBC in each density fraction (by flow cytometry) with increasing mean cell age. Fraction 4 contains the densest 1% of RBC.

Fig. 1.

Two mature male beagles were infused intravenously with N hydroxysuccinimido (NHS)-biotin (35 mg/kg body weight in dimethyl sulfoxide) to biotinylate all circulating RBC (99% each), as described.9 One dog (794) was recovering from an idiopathic anemia and had a low normal hematocrit of 37% at labeling that increased to 52% by day 48, resulting in a partial cohort effect. The other dog (331) was assessed to be completely normal. Blood was taken for density gradient centrifugation at roughly 2-week intervals initially and then weekly from day 86 to 105 to determine the proportions of B-RBC in each density fraction (by flow cytometry) with increasing mean cell age. Fraction 4 contains the densest 1% of RBC.

Table 1.

Age Distribution and Purity of In Vivo Aged, Biotinylated RBC in the Most Dense Fraction of Arabinogalactan Fractionated Dog Cells

Most Dense Biotinylated RBC Population Post-Biotinylation Day
72  86  93  100 105
Age range (d):  72-115  86-115  93-115  100-115 105-115
Estimated mean cell age (d)* 94  101  104 108  110
Purity (% biotinylated cells)
Dog 331 (random age)  94  76  56  42  33
Dog 794 (partial cohort)  98  95  84  60  45
Most Dense Biotinylated RBC Population Post-Biotinylation Day
72  86  93  100 105
Age range (d):  72-115  86-115  93-115  100-115 105-115
Estimated mean cell age (d)* 94  101  104 108  110
Purity (% biotinylated cells)
Dog 331 (random age)  94  76  56  42  33
Dog 794 (partial cohort)  98  95  84  60  45

*The value for the partial cohort would be shifted slightly to a younger age, although the exact value is unknown.

The separations between days 72 and 105 allowed us to characterize the approximate age distribution of the 1% most dense RBC (Table 1). For instance, analysis of dog 331 cells at postbiotinylation day 86 showed that the most dense fraction contained 76% biotinylated RBC, indicating that only 24% of the RBC were younger than 86 days old. A sharper age dependency was obtained when a younger cohort of cells was biotinylated, as with dog 794, where the most dense fraction was 95% B-RBC at the same time point. Even at day 93, the densest fraction of dog 794 RBC was still 84% B-RBC, demonstrating a remarkable enrichment of old RBC in this highly dense fraction.

Our results clearly provide greater justification for use of density gradient centrifugation for isolation of senescent RBC than previous studies using rabbit biotinylated cells have indicated.7 At least part of the difference between the two studies may be due to the different methods used for identifying B-RBC. The purity of the rabbit's most dense fractions was likely underestimated, because the bead-binding assay used in that study is much less sensitive than flow cytometry to the presence of the biotin label.9 Also, the high percentage of random RBC removal in the rabbit and most rodents10,11 raises questions whether the senescent cell recognition mechanism is similar to humans and dogs in these smaller species. Finally, differences in the centrifugation media (Percoll-Hypaque v arabinogalactan) could have significantly affected the results, because in our hands arabinogalactan gradients yield a higher percentage of senescent cells in the dense fraction than do Percoll gradients.

In summary, these results confirm that the ≤1% most dense RBC from dogs represent a predominantly aged population of cells. Labeling a cohort of young cells further enhances the enrichment for old RBC in this fraction. To the extent that canine RBC serve as an accurate model for human RBC, it can be concluded that the changes that occur late in the human RBC life span and trigger RBC removal can be studied by fractionating the cells on arabinogalactan density gradients.

## ACKNOWLEDGMENT

This work was funded in part by National Institutes of Health Grant No. GM24417.

## REFERENCES

REFERENCES
1
Borun
ER
Figueroa
WG
Perry
SM
The distribution of Fe59 tagged human erythrocytes in centrifuged specimens as a function of cell age.
J Clin Invest
36
1957
676
2
TenBrinke
M
Regt
JD
51Cr-half life of heavy and light human erythrocytes.
Scand J Haematol
7
1970
336
3
Clark
MR
Senescence of red blood cells: Progress and problems.
Physiol Rev
68
1988
503
4
Ganzoni AM, Oakes R, Hillman RS: Red cell aging in vivo. J Clin Invest 50:1373, 1971
5
Morrison M, Jackson CW, Mueller TJ, Huang T, Dockter ME, Walker WS, Singer JA, Edwards HH: Does cell density correlate with red cell age? Biomed Biochim Acta 42:s107, 1983 (suppl)
6
Mueller
TJ
Jackson
CW
Dockter
ME
Morrison
M
Membrane skeletal alterations during in vivo mouse red cell aging, increase in the 4.1a:4.1b ratio.
J Clin Invest
79
1987
492
7
Dale
GL
Norenberg
SL
Density fractionation of erythrocytes by Percoll/hypaque results in only a slight enrichment for aged cells.
Biochim Biophys Acta
1036
1990
183
8
Christian
JA
Rebar
AH
Boon
GD
Low
PS
Senescence of canine biotinylated erythrocytes: Increased autologous immunoglobulin binding occurs on erythrocytes aged in vivo for 104 to 110 days.
Blood
82
1993
3469
9
Christian
JA
Rebar
AH
Boon
GD
Low
PS
Methodological considerations for the use of canine in vivo aged biotinylated erythrocytes to study RBC senescence.
Exp Hematol
24
1996
82
10
Brown IW Jr
GS
An analytical study of in vivo survival of limited populations of animal red blood cells tagged with radioiron.
J Gen Physiol
36
1953
327
11
Burwell
EL
Brickley
BA
Finch
CA
Erythrocyte life span in small animals: Comparison of two methods employing radioiron.
Am J Physiol
172
1953
718