Dynamic phenotypic restructuring of the CD4 and CD8 T-cell subsets with age in healthy humans: a compartmental model analysis

Mechanisms of Ageing and Development 105 (1998) 241 – 264

Dynamic phenotypic restructuring of the CD4 and CD8 T-cell subsets with age in healthy humans: a compartmental model analysis Duaine R. Jackola a,*, Helen M. Hallgren b a

Department of Medicine, Section of Allergy, Uni6ersity of Minnesota Medical School, Box 434 UMHC, 420 Delaware St. S.E., Minneapolis, MN 55455, USA b Department of Laboratory Medicine and Pathology, Uni6ersity of Minnesota Medical School, Box 609 UMHC, 420 Delaware St. S.E., Minneapolis, MN 55455, USA Received 17 April 1998; received in revised form 20 July 1998; accepted 23 July 1998

Abstract In healthy humans, phenotypic restructuring occurs with age within the CD3 + Tlymphocyte complement. This is characterized by a non-linear decrease of the percentage of ‘naive’ (CD45RA + ) cells and a corresponding non-linear increase of the percentage of ‘memory’ (CD45R0 + ) cells among both the CD4 + and CD8 + T-cell subsets. We devised a simple compartmental model to study the age-dependent kinetics of phenotypic restructuring. We also derived differential equations whose parameters determined yearly gains minus losses of the percentage and absolute numbers of circulating naive cells, yearly gains minus losses of the percentage and absolute numbers of circulating memory cells, and the yearly rate of conversion of naive to memory cells. Solutions of these evaluative differential equations demonstrate the following: (1) the memory cell complement ‘resides’ within its compartment for a longer time than the naive cell complement within its compartment for both CD4 and CD8 cells; (2) the average, annual ‘turnover rate’ is the same for CD4 and CD8 naive cells. In contrast, the average, annual ‘turnover rate’ for memory CD8 cells is 1.5 times that of memory CD4 cells; (3) the average, annual conversion rate of CD4 naive cells to memory cells is twice that of the CD8 conversion rate; (4) a transition in dynamic restructuring occurs during the third decade of life that is due to these differences in turnover * Corresponding author. [email protected]

Tel.:

+1

612

6264624;

fax:

+1

612

6249188;

0047-6374/98/$ - see front matter © 1998 Elsevier Science Ireland Ltd. . All rights reserved. PII S0047-6374(98)00089-X

e-mail:

242 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

and conversion rates, between and from naive to memory cells. © 1998 Elsevier Science Ireland Ltd. . All rights reserved. Keywords: T Lymphocyte; CD4; CD8; CD45RA; CD45R0; Aging

1. Introduction Humans are continually exposed to antigenic challenges to which the immune system must adaptively respond. The dynamics of immune response change with age, especially among the elderly (\ 60 years of age). Age-related changes of immune response primarily center around T-lymphocyte function or functions that require mediation by T cells (reviews in Miller (1991) and Thoman and Weigle (1989)). However, there is no absolute correspondence between advanced age and immune-cell functional decline. Rather, there appear to be selective changes that occur within the T-cell complement. This led to the formulation of a ‘functional mosaic model’ (Thoman and Weigle, 1989; Miller, 1991). According to this model, some, but not all, T cells become functionally incompetent with age, while other cells remain as functionally robust in the elderly as comparable cells from the young. The functional mosaic hypothesis presupposes that lymphocytes are subject to cellular senescence. It also implies that lymphocytes are relatively long-lived, at least insofar as they can experience the effects of cellular aging. These proposals are difficult to investigate as there is no reliable marker for senescent immune cells and the lifespans of lymphocytes have proven difficult to define and measure (Freitas and Rocha, 1993). In addition, the functional mosaic model does not fully account for the fact that the T-cell complement is structured within a homeostatic framework. The proportions of thymus-derived lymphocytes (CD3 + , CD4 + and CD8 + ) do not change significantly with age among healthy humans (Reichert et al., 1991). There is a continual loss of cells, but there is also a continual renewal of functionally competent cells that continues even into advanced age (Steinmann, 1986). In order to account for altered immune function, an alternative to the functional mosaic model proposes a dynamic restructuring of the T-cell complement that occurs with age (Franceschi et al., 1993). The simplest demonstration of this restructuring involves differential expression of isoforms of the leukocyte common antigen, CD45R, on T-cell surfaces. In humans, T cells not thought to have been previously exposed to antigen (‘naive’ or ‘virgin’ cells), express a high molecular weight isoform of the leukocyte common antigen, CD45RA (Rogers et al., 1992). Upon exposure to antigen, T cells express a low molecular weight isoform, CD45R0, and will respond to recall antigens (‘memory’ function) (Akbar et al., 1991). The expressions of the CD45RA and CD45R0 isoforms are mutually exclusive on CD4 and CD8 T-cell subsets (Jackola et al., 1994).

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 243

Functionally, naive and memory cells are markedly different. They differ with regard to the stimuli to which they will respond (Akbar et al., 1991), they have different antigen-presenting cell and co-stimulation requirements (Inaba and Steinman, 1984; Geppert et al., 1990) and their responses differ, particularly with respect to their cytokine secretion patterns (Ernst et al., 1990). With regard to the dynamic restructuring of the T-cell complement, there is a decrease with age of naive (CD45RA + ) cells in the peripheral blood with a corresponding increase of memory (CD45R0 + ) cells among both the CD4 and CD8 T cell subsets (Cossarizza et al., 1992; Jackola et al., 1994). In order to provide a framework for dynamic phenotypic restructuring, we have devised a simple compartmental model to describe gains and losses in the age-dependent proportions and absolute numbers of circulating naive and memory T cells. To quantitate these changes, we derived a mathematical model and fit the resulting equations to previously published data (Cossarizza et al., 1992; Jackola et al., 1994). The solutions provide parametric estimates for the annual gains minus losses of naive and memory cells, as well as annual conversion rates of naive to memory cells. In addition, these analyses provide estimates for a mean compartmental residence time. This is an indication of the amount of time (in years) that a complement of cells with a particular phenotype may be present in the total T-cell complement. We use the results to demonstrate fundamental differences in dynamic restructuring between CD4 and CD8 T cells.

2. Methods

2.1. Donors and cell surface marker analyses We used the results of previously published work that employed identical methods for donor selection and cell surface marker analyses (Cossarizza et al., 1992; Jackola et al., 1994). Donors ranged in age from newborns to centenarians. These individuals were in excellent health with no histories of acute illnesses or chronic conditions known to influence immune-cell function according to the SENIEUR protocol (Ligthart et al., 1984). This report and others discuss at length the importance of choosing healthy individuals for studying age-related changes of immune component functions. Heparinized whole blood was obtained by sterile venipuncture and PBMCs were isolated using Ficoll – Hypaque as previously described (Jackola et al., 1994). The cells were stained with flourescently tagged monoclonal antibodies (mAbs) for two-color analysis by flow cytometry (Jackola et al., 1994). Cells were initially mixed with a mAb either for CD4 or CD8, followed by mixing with a mAb for CD45RA (Leu-18) or CD45R0 (UCHL1). Using a flow cytometer, electronic gates were set for forward and side scatter to include small, CD3 + lymphocytes. Only the cells within this region were analyzed. This criterion included the CD4 and CD8 T-cell subsets, and excluded other cells

244 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

that might express isoforms of the leukocyte common antigen, CD45R. For each donor, determinations were made for the percentages of CD4 + /CD45RA + , CD4 + /CD45R0 + , CD8 + /CD45RA + and CD8 + /CD45R0 + T cells. Although the relative proportions of CD3, CD4 and CD8 T cells in the peripheral circulation do not change with age (Reichert et al., 1991), there is an apparent linear decline in the absolute numbers (c cells/ml whole blood) of these subsets with age (Sansoni et al., 1993). This observation has been confirmed by subsequent studies (Hulstaert et al., 1994; Stulnig et al., 1995). For our analyses, we used the results reported by Sansoni et al. (1993) for the age-dependent, linear decline of absolute numbers of CD4 and CD8 T cells in the peripheral blood circulation.

2.2. Compartmental model Compartmental modeling describes dynamic processes involving the continuous exchange of materials (metabolites, cells, etc.) between separate, but physiologically interconnected, compartments (Jacquez, 1985). Our intent is to describe the age-dependent gains and losses of the absolute numbers (c cells/ml blood) of circulating naive and memory CD4 and CD8 T lymphocytes, as well as the rates of conversion of naive to memory cells. Fig. 1 shows a schematic diagram of our model. The model comprises two independent compartments: one compartment for naive cells and one compartment for memory cells. In Fig. 1, arrows represent flows

Fig. 1. Schematic diagram of the compartmental model. Definitions and explanations are given in Section 2.2.

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 245

into or out of compartments. The letters next to the arrows describe compartmental inputs and rates of flow between compartments. For modeling purposes, we assume that CD4 and CD8 pathways are independent of one another. We also assume that naive or memory cells are homogeneously distributed without regard to anatomical structures (lymph nodes, spleen, etc.). That is, there is no sequestration of significant numbers of naive or memory cells in preferred anatomical locations, and samples from the peripheral blood are representative. This assumption is predicated upon experimental observations for recirculation kinetics of lymphocytes in mammals (Westermann et al., 1988) and theoretical analyses for lymphocyte redistribution (Farooqi and Mohler, 1989). These and other studies demonstrate the rapid recirculation of labeled lymphocytes (: 24–48 h) in mammals. In conjunction with choosing donors who met the criteria of the SENIEUR protocol (Ligthart et al., 1984), it is reasonable to assume that in the absence of chronic inflammatory processes, which might induce site-specific lymphocyte sequestration, the peripheral blood is a representative sample. Naive cell numbers may increase or decrease by several mechanisms. First, external to the naive compartment, functionally mature T cells will enter the naive compartment by export from the thymus at an amount that decreases with age (see mathematical description later). Second, within the naive cell compartment, we define three separate mechanisms reflective of intrinsic cellular properties. Naive cell numbers may grow at a constant rate, gN, or the cells may die at a constant rate, d. Also, naive cell numbers may decrease due to conversion to memory cells at a constant rate, k. These mechanisms are independent of one another and, by assumption, are also age independent for healthy humans. We describe the naive to memory cell conversion as unidirectional. Other authors have suggested that memory cells may reconvert to naive cells in vivo (Michie et al., 1992). However, this study did not take into account the possibility of input from the thymus. Because cells from the thymus can be continually exported throughout the lifespan, albeit at greatly diminished amounts with advanced age (Steinmann, 1986), we do not consider memory to naive cell reconversion to appreciably contribute to naive cell input. Memory cell numbers may also increase or decrease by several independent mechanisms. First, external to the memory cell compartment, memory cells may increase due to naive cell conversion at a constant rate, k. Healthy humans will be continually exposed to antigenic challenges from the environment. However, save for periodic occurrences of increased antigen load, such as annual influenza outbreaks, the average yearly challenge should be relatively constant and, thus, naive to memory conversion should be constant. Second, two mechanisms reflective of intrinsic cellular properties are defined. Memory cells may grow at a constant rate, gM, or decrease at a constant rate, f. As with naive cells, these mechanisms are assumed to be independent of one another and age independent. The proportions of the major T-lymphocyte subsets (CD3, CD4, CD8) found in the peripheral blood do not appreciably change with age (Reichert et al., 1991). Because naive (CD45RA + ) and memory (CD45R0 + ) T cells are reciprocal subsets

246 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

(Jackola et al., 1994), the percentage of naive (CD4 or CD8) T cells plus the percentage of memory (CD4 or CD8) T cells is always 100% (Cossarizza et al., 1992; Jackola et al., 1994). There is, however, an apparent linear decline in the absolute numbers of T cells found in the peripheral circulation that occurs with age (Sansoni et al., 1993; Hulstaert et al., 1994; Stulnig et al., 1995). It has been proposed that this decline in cell numbers may be due to hematopoietic dysregulation among the elderly (Rothstein, 1993). Unfortunately, there is currently no compelling experimental evidence to support or reject this hypothesis, although this does emphasize the complex nature of the immune system remodeling that occurs with age (Franceschi et al., 1993). Thus, among healthy humans, the parameters defined in our model are adequate to describe the age-dependent changes that occur in the CD4 and CD8 subsets.

2.3. Asymptotic limits to changes in nai6e and memory cell percentages There are dramatic changes in the percentages of naive and memory cells during the first two to three decades of the lifespan, followed by a leveling-off phenomenon (Cossarizza et al., 1992) (see, also, Fig. 2). These results suggest asymptotic limits to the overall percentage changes of naive and memory cells in humans. To quantitate these changes, we assume some continuous function of age, X(A). X(A) is the amount (percentage) of something within a defined compartment. The amount X(A) may increase due to input at a constant amount, a, irrespective of the amount of X(A) currently present. Also, X(A) may decrease at a constant rate, b, but the total amount of decrease is proportional to the amount currently present. This is mathematically described by: dX/dA = a− bX. A general solution is Ce − bA +(a/b), where C is an arbitrary constant of integration. To find particular solutions, we assume certain initial values of X(A) when age A= 0. First, when A = 0, X(0) = 100%. Solving for the integration constant gives C=100 − (a/b). X(A) = (a/b) +[100 −(a/b)] e − bA

(1)

This is a generic equation to describe the asymptotic limit for naive cell percentage decrease. The asymptotic limit is the ratio (a/b). Second, we assume that when A= 0, X(0)=0%. Now, solving for C: C= − (a/ b). X(A) = (a/b)(1 − e − bA)

(2)

This is a generic equation describing memory cell percentage increase, with an asymptotic limit of (a/b). Note: the ratios (a/b) for naive and memory cell asymptotes will not have the same values when fit to the data, as they were found using different initial values when resolving the general solution equation.

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 247

Fig. 2. Raw data and asymptotic limits of phenotypic restructuring. For CD4 T cells (upper panel) and CD8 T cells (lower panel), naive cell (CD45RA + ) percentages are given as filled circles (“) and memory cell (CD45R0 + ) percentages are given as open circles (). Solid curves are best fit approximations to asymptotic equations derived in Section 2.3. Data from Jackola et al. (1994) and Cossarizza et al. (1992).

2.4. Mathematical model We derive mathematical descriptions for our compartmental model in order to quantitate age-dependent changes in the numbers of naive and memory cells, and to estimate parameters with which to compare the dynamics of restructuring among CD4 and CD8 naive and memory cells. The mathematical strategies used here can be found in any standard text on differential equations.

248 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

2.4.1. State 6ariables Exchange of materials between compartments occurs by mass transfer. In order to quantitatively analyze these exchanges, the variables of interest must be expressed as concentrations or total numbers, and are thus variables of state (Jacquez, 1985). The appropriate state variables here are the absolute numbers of naive or memory cells per unit body mass. Of course, this cannot be estimated with any reasonable accuracy. The only accessible means to quantitate lymphocyte numbers derives from measures in the peripheral blood. For our purposes, we assume that measures taken using peripheral blood samples, with cell concentrations expressed as c cells/ml whole blood, are representative of the state variables of interest. A complication arises, however, since the concentrations of peripheral blood lymphocytes do not remain constant with age (Sansoni et al., 1993; Hulstaert et al., 1994; Stulnig et al., 1995). Two different experimental strategies have been employed to assess the age-dependent changes of T-cell phenotypes and concentrations. One strategy assessed the proportions of naive and memory CD4 and CD8 T cells (Cossarizza et al., 1992; Jackola et al., 1994). These experiments used lymphocytes isolated from peripheral blood samples. The second strategy measured the concentrations (c cells/ml) of CD3, CD4 and CD8 cells using anti-coagulated whole blood (Sansoni et al., 1993). These results show a linear, age-dependent decline in cell concentrations of CD3, CD4 and CD8 T cells. (At any age, the concentrations of CD4+ CD8= CD3.) Using the results of Sansoni et al. (1993), best estimates for the age-dependent decreases of CD4 and CD8 concentrations are: c CD4/ml =1263.1 − 6.14 × age

(3)

c CD8/ml =827.6 − 4.02 × age

(4)

To estimate the concentrations of naive or memory cells at any age, the results of Eq. (3) or Eq. (4) are multiplied by the estimated changes in proportions of naive or memory cells (Eq. (1) or Eq. (2)).

2.4.2. Thymic input The total thymic mass in humans is greatest at puberty and declines thereafter. However, the mass of the thymic cortex, in which T-cell maturation occurs, is greatest at birth and declines exponentially thereafter (Steinmann, 1986). We assume that the number of mature T cells exported from the thymus at any age is proportional to the mass of the thymic cortex. Cells exported from the thymus are rapidly distributed throughout the body, including the peripheral blood circulation (Westermann et al., 1988; Farooqi and Mohler, 1989). We assume that the number of cells exported from the thymus and recirculated in the peripheral blood declines at a constant rate, i. The number of cells due to thymic input at any age is: I(A) (c cells/ml). The rate of change of I(A) is proportional to the amount currently present: dI/dA = −iI. The solution of this equation is:

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 249

I(A) = I(0) e − iA

(5)

where I(0) is the c cells/ml exported at age 0. Eq. (5) describes the input into the naive compartment from the thymus in Fig. 1.

2.4.3. Nai6e cells As described in our model, the number (concentration) of naive cells in the peripheral blood at any age, N(A), can increase at a constant rate, gN, due to naive cell growth; or, these numbers can decrease due either to cell death (rate d) or conversion to memory cells (rate k). The age-dependent change of N(A) will be proportional to the amount currently present: dN/dA = (gN − d−k)N. In addition, the concentration of naive cells can increase due to import from the thymus (Eq. (5)). Thus, an equation describing the overall age-dependent change of naive cells is: dN/dA = (gN −d − k)N + I(A). One objective of our analysis is to determine dynamic changes in the T-cell complement due to total differences in gains minus losses of cells. We simplify the equation for naive cells by defining a naive cell parameter: l=gN − (d+ k). That is, the parameter l describes the annual overall rate of change of naive cells due to rate gains minus rate losses. dN = lN+ I(0) e − iA dA

(6)

A general solution of this equation is N(A) = C1 elA −G e − iA

(7)

where C1 is an arbitrary constant of integration and G is a constant consisting of a collection of parameters: G = − I(0)/(l + i ).

2.4.4. Memory cells Memory cell concentrations at any age, M(A), can increase due to cell growth at a constant rate, gM, or decrease due to cell death at a constant rate, f. We define a memory cell parameter: m =gM − f. Also, memory cells can increase due to conversion from naive cells at a constant rate, k, and this increase is proportional to the current concentration of naive cells, N(A). dM = mM+ kN dA

(8)

A general solution is C2 emA +C3 elA +C4 e − iA

(9)

C2 is an arbitrary constant of integration, and C3 and C4 are collections of parameters (see later).

250 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

2.4.5. Initial 6alues The dynamic behavior of our compartmental model is described by Eqs. (7) and (9) for naive and memory cells respectively. The system behavior will be determined by three parameters: l, m and k. Because we assume a unidirectional flow between naive and memory compartments, the rate k must be positive (see Fig. 1). However, the parameters (rates) l and m, reflecting annual gains minus losses, may be less than, equal to or greater than zero. In order to estimate these parameters, we find particular solutions to Eqs. (7) and (9) by eliminating the two constants of integration: C1 and C2. We do this by introducing the following initial values. At age zero, the number of naive cells in the peripheral blood, N(0), will be given by the intercepts of the equations previously given for the age-dependent changes in absolute numbers of cells in the peripheral blood (Eq. (3) for CD4 cells or Eq. (4) for CD8 cells, where we are assuming that all CD3 T cells are naive at age zero). From Eq. (7), N(0)= C1 − G and rearranging gives C1 = N(0)+ G. Incorporating into Eq. (7), N(A) = (N(0) +G) elA −G e − iA

(10)

We also assume that there are no memory cells found in the peripheral blood at age zero: M(0) = 0 = C2 +C3 +C4. Solving for C2: C2 = − (C3 + C4). Incorporating into Eq. (9): M(A) = C3(elA −emA) +C4(e − iA − emA)

(11)

After some straightforward algebra, it is easily shown that the two remaining constants are collections of parameters: C3 =k/(l − m)(N(0) + G)

C4 = kG/(i +m)

where, from above, G = − I(0)/(l + i ).

2.5. Parameter estimation Eqs. (1) and (2), describing asymptotic limits to percentage changes, and Eqs. (10) and (11), describing age-dependent changes in cell concentrations, are all non-linear. In order to fit these equations to the data, we used a non-linear Levenberg – Marquardt fitting algorithm provided by the software program(s) in MATHEMATICA (Wolfram Research, Champaign, IL). Starting from user-provided initial value estimates for the parameters of interest, an iterative search was done until a minimum for the sum of squares for error was found. The fitting program also provided the parameters’ coefficients of errors whose values were least when the best fit was found. For each data set to be tested, we employed a range of initial parameter estimate values until these conditions were met.

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 251

3. Results

3.1. Dynamic restructuring is different for CD4 and CD8 T cells The raw data for the age-dependent changes of the phenotypic percentages are given in Fig. 2 for CD4 T cells (upper panel) and CD8 T cells (lower panel). These data were collated from two separate studies that used identical methods for donor selection and cell surface marker analyses (Cossarizza et al., 1992; Jackola et al., 1994). The plotted points are from 110 individuals ranging in age from newborns to 105 years. As previously discussed (Jackola et al., 1994), the expressions of CD45RA and CD45R0 were mutually exclusive on CD4 and CD8 cells. For each individual, percentage naive cells plus percentage memory cells was approximately 100%. For both CD4 and CD8 cells, there is a rapid decline of percentage naive cells starting at birth and continuing to about age 30 years, followed by a ‘leveling off’. Conversely, for memory cells there is a rapid increase beginning at birth followed by a ‘leveling off’ after age 30. In order to depict these qualitative differences, we fit the data using the equations describing asymptotic (limiting) behavior (Eqs. (1) and (2)). These results are also plotted as solid lines in Fig. 2. Table 1 gives the best fit estimates for the parameters, the parameters’ standard errors and the lower and upper bounds of the 95% confidence intervals for the parameters. These were found using the non-linear fitting algorithm. In addition, to assess the overall quality of the fits, we calculated simple Chi-square statistics Table 1 Parameters characterizing asymptotic limits to change Cells CD4+ CD45RA+

CD45R0+

CD8+ CD45RA+

CD45R0+

a

a (per year)a

b (per year)

(a/b)

1.342% (9 0.2139) [1.176, 1.381] 4.361% (9 0.3670) [4.300, 4.422]

0.0579 (9 0.0056) [0.0570, 0.0588] 0.0568 (9 0.0056) [0.0559, 0.0577]

23.3%

2.360% (9 0.5115) [2.545, 2.715] 2.472% (9 0.3226) [2.419, 2.526]

0.0508 (9 0.0082) [0.0949, 0.0522] 0.0513 (9 0.0082) [0.0499, 0.0527]

51.8%

76.7%

48.2%

a, yearly percentage increase in cells; b, yearly rate of decrease of cells; (a/b), asymptotic limiting percentage. Values in parentheses are one S.E. for the parameters. Values in square brackets are lower and upper bounds for the 95% confidence intervals for the parameters.

252 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

(Bevington, 1969). These calculated values were compared to tabulated values that give the probability that a set of random data points would yield a Chi-square result equal to or greater than that provided by the experimental data. These results were: naive CD4, P B0.001; memory CD4, PB0.025; naive CD8, P B 0.01; memory CD8, P B 0.001. There are clear differences in the limiting values between CD4 and CD8 cells. For CD4 cells, the asymptotic value for naive cells is 23.3% and the value for memory cells is 76.7%. In contrast, the limiting value for naive CD8 cells is 51.8% and the value for CD8 memory cells is 48.2%. Interestingly, even in advanced age, significant proportions of CD4 or CD8 T cells maintain a ‘naive’ phenotype. However, there are obvious differences in the dynamic restructuring of the CD4 and CD8 T-cell subsets that occur with age.

3.2. Compartmentation describes dynamic restructuring The data in Fig. 2, expressed as percentages, were converted to cell concentrations (c cells/ml) using the results of Sansoni et al. (1993) (see Section 2.4.1). These results are shown in Fig. 3. The converted data were fit to our quantitative model (Eq. (10) for naive cells and Eq. (11) for memory cells; see Section 2.4.5). Different initial values of the parameters to be estimated were used and, proceeding iteratively, best fit searches were done using the non-linear optimization algorithm until the following conditions were met simultaneously: a minimum value for the total sum of squares for error and minimum values for the parameters’ coefficients of error. The results of these searches are plotted as solid lines in Fig. 3. There are five parameters whose estimates were simultaneously determined. The values of these parameters and comparisons between CD4 and CD8 cells are given in Table 2. The parameters estimated were: (a) I(0) (c cells/ml), input to the naive cell compartment from the thymus at age zero; (b) i (per year), the yearly rate of decline of cells imported from the thymus; (c) l (per year), naive cell parameter for yearly gains minus losses; (d) m (per year), memory cell parameter for yearly gains minus losses; (e) k (per year), yearly rate of naive to memory cell conversion. As a procedural note, we are aware of the difficulties that can arise when attempting to fit data to an equation with two (or more) exponential terms (Jacquez, 1985). For example, it is possible that the values of the exponential coefficients for CD4 naive cells (parameters l and i ) could be ‘reversed’ by an order of magnitude (e.g. l : −0.01355 and i :0.0798). We took this contingency into account when entering initial parameter values into the fitting algorithm. The results invariably showed that the parameteric estimates given in Table 2 were the best statistical estimates. There are minor differences between CD4 and CD8 cells with respect to the parameters for import from the thymus (I(0) and i; see Eq. (5) and values in Table 2). These results lend support for our model assumption of independence between CD4 and CD8 pathways. Our results for the yearly decline in cells imported from the thymus (parameter i ) are also in accordance with estimates made for the percentage yearly decline of thymic cortical mass made by Steinmann (1986).

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 253

Fig. 3. Corrected raw data and dynamics of phenotypic restructuring. For CD4 T cells (upper panel) and CD8 T cells (lower panel), naive cell (CD45RA + ) numbers/ml are given as filled circles (“) and memory cell (CD45R0 + ) numbers/ml are given as open circles (). Data from Fig. 2 (percentages) were converted to c cells/ml using Eqs. (3) and (4) and results reported by Sansoni et al. (1993). Solid curves are best fit approximations to Eqs. (10) and (11).

Although there are differences in the inputs to the compartmental model from the thymus for CD4 and CD8 cells, these inputs are ‘external’ to the compartments. These factors do not contribute to (or explain) the dynamics of restructuring. Restructuring occurs within the compartments due to intrinsic cellular properties (parameters l, m, and k in the model). Before discussing these differences, it is instructive to consider first the components that contribute to the dynamic changes of naive and memory cells.

254 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

3.3. A transition age for dynamic restructuring According to our quantitative model (Eqs. (10) and (11)), dynamic restructuring is due to several components. Mathematically, for naive cells there are two exponential terms, while for memory cells there are three exponential terms. The additive results of these components provide the overall dynamic restructuring changes shown as solid lines in Fig. 3. To demonstrate how these components contribute to dynamic restructuring, we have plotted the results in Figs. 4 and 5. For naive cells, the results for CD4 and CD8 cells are given in Fig. 4. In this figure, solid lines are the total concentrations of naive cells at each age, as in Fig. 3. From Eq. (10), the dashed lines are the components due to input from the thymus [I(0)/(l + i )e − iA]. The broken lines are the components due to naive cell gains minus losses (N(0) + G)elA; G= − I(0)/(l + i ) and l= gN − (d+ k). Even though the input from the thymus declines with age, the net contribution by this component does not appreciably change with age. The calculated values for the parameter(s) i (see Table 2) are relatively small and, therefore, the net contribution by the exponential term only changes negligibly with advancing age. In contrast, the component owing to naive cell gains minus losses (parameter l in Table 2) shows a sharp decline from birth to about age 30, and then ‘levels off’. Thus, the major contributing factor resulting in naive cell decline with age is the net gains minus losses for these cells in both the CD4 and CD8 subsets. The results for memory CD4 and CD8 cells are given in Fig. 5. According to Eq. (11), there are three exponential terms contributing to these results. Instead of Table 2 Best fit parameter estimates Parametera

CD4

CD8

CD4/CD8

I(0)

33.95 (9 2.365) [30.41, 36.48] 0.0136 (9 0.0036) [0.0065, 0.0206] −0.07982 (9 0.0181) [−0.1156, −0.0440] −0.0199 (9 0.0011) [−0.0221, −0.0177] 0.0440 (9 0.0016) [0.0409, 0.0471]

33.78 ( 93.101) [30.02, 35.69] 0.0093 (9 0.0024) [0.0045, 0.0141] −0.0683 ( 90.0357) [−0.1392, −0.0335] −0.0299 (9 0.0228) [−0.0369, −0.0228] 0.0230 ( 90.0020) [0.0191, 0.0269]

1.01

i

l

m

k

a

1.45

1.17

0.67

1.91

Parameters are defined in Sections 2.2 and 3. Values are best fit estimates. Values in parentheses are one S.E. for the parameters. Values in brackets are lower and upper bounds on the 95% confidence intervals for the parameters.

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 255

Fig. 4. Components of naive cell restructuring. The components contributing to the age-dependent restructuring of CD4 T cells (upper panel) and CD8 T cells (lower panel). Solid lines are results for total naive cell restructuring defined by Eq. (10) described in Section 3.3. Dashed lines are contributions due to cell input from the thymus (see Section 3.3). Broken lines are contributions due to naive cell turnover (see Section 3.3).

plotting these terms separately, it is more informative to group terms as done in Eq. (11). In Fig. 5, the dashed lines are the contributions resulting from the collection of terms C4(e − iA −emA) (C4 is a constant/collection of parameters; see Section 2.4.5). Cells are first imported from the thymus into the naive compartment. Some of these cells are then converted to memory cells. The overall gain of memory cells due to this process is shown by the dashed lines. For both CD4 and CD8 memory cells, there is a gradual increase with age that contributes to dynamic restructuring. In contrast, there is a dramatic change in dynamic restructuring due to the collection of terms from Eq. (11) given by C3(elA − emA) (C3 is a constant/collection

256 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

of parameters; see Section 2.4.5). These results are plotted as broken lines in Fig. 5. Again, the parameter l is reflective of net gains minus losses for naive cells, while the parameter m is reflective of net gains minus losses for memory cells (see values in Table 2). These curves show the net contribution due to differences between naive cell gains and losses and memory cell gains and losses. For both CD4 and CD8 cells, there is an age where an apparent transition occurs. Beginning with a rapid increase at age 0 to about age 30, the curve suddenly changes and is followed by a gradual decline. Using the properties of smooth curves

Fig. 5. Components of memory cell restructuring. The components contributing to the age-dependent restructuring of CD4 T cells (upper panel) and CD8 T cells (lower panel). Solid lines are results for total memory cell restructuring defined by Eq. (11) described in Section 3.3. Dashed lines are contributions due to cell input from the thymus followed by conversion to naive cells (see Section 3.3). Broken lines are contributions due to differences between naive cell turnover and memory cell turnover (see Section 3.3). Transition age is the age at which phenotypic restructuring changes due to differences in naive and memory cell turnover rates (see Section 3.3).

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 257 Table 3 Derived parameters Parameter

CD4

CD8

CD4/CD8

gN−d (gN−d)/(gM−f ) l/m MCRT −1/l (years) −1/m (years) Transition age (years)

−0.0358 1.80 4.01

−0.0453 1.52 2.28

0.79 1.18 1.76

12.5 50.3 23.2

14.6 33.4 21.5

0.85 1.51 1.08

Parameters gN−d, (gN−d)/(gM−f ) and l/m described in Section 3.4. MCRT, mean compartmental residence time, defined in Appendix A. Transition age described in Section 3.3.

and the differential calculus, we can estimate the age at which this transition occurs for CD4 and CD8 cells. The broken lines in Fig. 5 show a local maximum value. From the differential calculus, for some continuous function of age, X(A), the derivative of the function with respect to age is zero at a local maximum: dX/dA = 0. For the collection of terms represented by the broken lines in Fig. 5, this derivative is: lelA − memA = 0. (Because the constant C3 is a common factor it does not contribute to finding the local maximum.) Rearranging this equation: (l/m)=(emA/elA). Taking the natural logarithm of both sides: ln(l/m) = mA−lA. Then, solving for A, A= [ln(l/m)/(m − l)]. Using the estimated parametric values in Table 2, the ‘transition age’ for CD4 cells is about 23.2 years, while the ‘transition age’ for CD8 cells is about 21.5 years. These values are marked by arrows in Fig. 5 and labeled ‘transition age’ (see also Table 3). The dynamic restructuring of the CD4 and CD8 T-cell subsets changes dramatically during the third decade of life. However, even though there are qualitative similarities in dynamic restructuring for CD4 and CD8 cells, there are fundamental quantitative differences between these subsets (compare Figs. 2 – 5). We turn our attention now to describing these differences.

3.4. Turno6er rates and con6ersion rates differ for CD4 and CD8 T cells In our model, dynamic restructuring is due to three mechanisms: net gains minus losses for naive cells, net gains minus losses for memory cells, and conversion of naive to memory cells. The quantitative aspects are determined by three parameters reflective of intrinsic cellular properties: l, m, and k (see values in Table 2). With regard to ‘turnover rates’ for naive and memory cells, we consider the following. The naive cell parameter in our model is defined by l= gN − d− k (growth rate− death rate− conversion rate). Using the fitting algorithm, the parameters l and k can be estimated (Table 2). Then, l+ k=gN − d. (Using our procedures with

258 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

the available data, it is not possible to find independent estimates for gN, gM, d and f.) The difference, gN −d, is a ‘turnover rate’ for naive cells due to cellular properties within the compartment (growth−death). For memory cells, the parameter reflecting gains minus losses is m=gM − f. We compare the turnover rates for naive and memory cells by calculating the ratio: (gN − d)/(gM − f ). As shown in Table 3 for derived parameter estimates, the average yearly turnover rate of naive cells is about 1.5 to 2 times that of the yearly turnover rate for memory cells, for both CD4 and CD8 T cells. In contrast, there are distinct differences when comparing the turnover rates between CD4 and CD8 T cells. The turnover rate for naive cells (gN − d) is due only to growth and death processes by these cells. However, the overall turnover rate for naive cells is defined by parameter l. Upon comparing the ratio of naive cell turnover rate to memory cell turnover rate (l/m), there are considerable differences between CD4 and CD8 cells (Table 3). The l/m ratio for CD4 cells is nearly twice that for CD8 cells. As shown in Table 2, there is essentially no difference when comparing the naive cell parameter (l) for CD4 and CD8 cells. However, the memory cell parameter (m) for CD4 cells is about two-thirds the value of the memory cell parameter for CD8 cells. Thus, naive CD4 and CD8 cells tend to ‘reside’ within their respective compartments for about the same amount of time. In contrast, the ‘residence times’ for memory CD4 and CD8 cells are different. The amount of time that a complement of cells ‘resides’ within its compartment is a function of its net ‘turnover rate’. The more quickly the cell complement passes through its compartment, the less time it resides within it. An estimate of the mean compartmental residence time (MCRT in Table 3) can be found. As shown in Appendix A, the MCRT for naive cells is − i/l and the MCRT for memory cells is − i/m (the negative inverses of the resepective turnover rates). The MCRTs for naive CD4 and CD8 cells are approximately equal. However, the MCRT for CD4 memory cells is 1.5 times the MCRT for CD8 memory cells. It is important to note that the mean compartmental residence time used here is not equivalent to lymphocyte lifespan. We are making estimates for the total complements of CD4 and CD8 T cells. The annual change in these complements is relatively small (parameters l and m are only fractional changes per year). Thus, the overall complement structure does not change very rapidly. However, factors that influence the aging of individual cells, or cell cohorts, may still be operative within the limits of the model we have described. Finally, the naive to memory cell conversion rate, k, is also fundamentally different when comparing CD4 and CD8 cells. As shown in Table 2, the estimated value of k for CD4 cells is about twice that of the estimated value of k for CD8 cells. Thus, although naive CD4 and CD8 cells tend to reside within their respective compartments for about the same amount of time, CD4 naive cells convert to memory cells at twice the rate of CD8 naive to memory conversion. In addition, CD4 memory cells tend to reside within their compartment about 1.5 times longer than CD8 memory cells within their compartment.

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 259

4. Discussion Dynamic phenotypic restructuring occurs with age in healthy humans within the T-lymphocyte complement. This restructuring is manifest by changes in the relative proportions of previously unchallenged ‘naive’ (CD45RA + ) cells and cells that can respond to recall antigens (‘memory’, CD45R0 + ). For both CD4 and CD8 T cells, restructuring is characterized by a decrease in the percentage of naive cells with a corresponding increase in the percentage of memory cells. Due to fundamental differences of activation requirements and physiological responses (Inaba and Steinman, 1984; Ernst et al., 1990; Geppert et al., 1990; Akbar et al., 1991), phenotypic restructuring could infer corresponding functional restructuring of the immune system with age (Franceschi et al., 1993). This might contribute to some of the changes in immune responses (responsiveness) that occurs in elderly humans. Qualitatively, the restructuring that occurs among the CD4 and CD8 subsets is similar (Fig. 2). Beginning at birth to about age 30, there is a rapid decrease of naive cells with a subsequent leveling off and a corresponding rapid increase with a subsequent leveling off among memory cells. However, there are substantial quantitative differences when comparing CD4 and CD8 cells (Figs. 2–5). In order to characterize these differences, we derived a simple compartmental model and an evaluative set of differential equations to describe the changes of the components within the compartments as a function of age. These quantitative differences contribute to phenotypic/functional restructuring. The principle elements involved with this restructuring are as follows. (1) The memory cell complement tends to ‘reside’ within its compartment for a longer time than the naive cell complement within its compartment. The net negative yearly gains minus losses for naive cells is greater than that of memory cells, either for CD4 or CD8 T cells (l/m ratios in Table 3). Irrespective of the input into the compartment, the gains and losses defined in our model are due to intrinsic cellular properties. This does not imply that individual memory cells are longer-lived than individual naive cells. Our results are yearly averages, and a component of ‘longer-lived’, senescing cells may be present. Thus, our results are not at odds with the ‘functional mosaic model’ (Thoman and Weigle, 1989; Miller, 1991) which infers an age-dependent increase in the proportions of functionally incompetent cells due to cellular aging. Indeed, our model is complementary to the functional mosaic model. From our results for the mean compartmental residence time (MCRT in Table 3), it is clear that significant proportions of the memory cell complements can ‘reside’ within their compartments for several decades. As defined here, the MCRT reflects the amount of time that a proportion of cells with a particular phenotype may be present. This is not a parameter reflective of individual cells’ length of residence (lifespan). In the context of homeostatic mechanisms that maintain relatively constant proportions of CD3, CD4 and CD8 T lymphocytes throughout the lifespan (Reichert et al., 1991), it is apparent that maintenance of memory cell complements overrides maintenance of naive cell components. On a per cell basis,

260 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

however, factors that control for various outcomes among these cells are poorly understood (reviewed in Miller, 1996). These possible outcomes include continual proliferation, differentiation to terminal effector cells, survival as memory cells or apoptotic cell death. When these factors and the mechanisms controlling them are realized, accounting will also have to be made for phenotypic and functional restructuring among the major T-cell subsets. (2) A fundamental difference is present upon comparing ‘intrinsic cellular parameters’ between CD4 and CD8 cells. The ratio of gains minus losses for naive CD4 cells to naive CD8 cells is about the same (parameter l in Table 2) as is the mean compartmental residence time (MCRT in Table 3). In contrast, the amount of time that the memory CD4 complement resides within its compartment is about 1.5 times that of the CD8 complement (parameter m in Table 2 and MCRT in Table 3). Given that the inputs from the thymus for CD4 and CD8 are roughly the same (parameters I(0) and i in Table 2), the net effect ‘in the periphery’ is to have a longer-lasting complement of memory CD4 cells than memory CD8 cells. (3) The ‘conversion rates’ of naive to memory cells are fundamentally different upon comparing CD4 and CD8 cells. Naive CD4 cells convert to memory CD4 cells at an annual, average rate about twice that of the rate of naive CD8 to memory CD8 conversion (parameter k in Table 2). There is a higher annual turnover rate for CD8 cells (gains minus losses in the compartments) and a lower naive to memory conversion rate (flow between compartments). (4) The dominant feature of phenotypic restructuring is its non-linear character. As shown in Fig. 5, an apparent transition in dynamic restructuring occurs during the third decade of life for both CD4 and CD8 T cells. This transition occurs because of differences in ‘turnover rates’ between naive and memory cells. Up to about 30 years (‘rapid change phase’), restructuring is dominated by intrinsic naive cell parameters. After 30 years of age (‘leveling off phase’), restructuring is dominated by memory cell parameters. This transition occurs in the decade following puberty. The total thymic mass is greatest at puberty in humans (Steinmann, 1986), and it is possible that total cell export from the thymus is also greatest during this period, as has been demonstrated in mice (Scollay et al., 1980). For quantitative purposes, a ‘more realistic accounting’ for the input of cells into the naive compartment from the thymus would have to take this into account. We used different model terms for input from the thymus based upon the possible age-dependent alterations of variable thymic mass. Qualitatively, the results were not too different from those already presented here. This resulted as, in our model, thymic cell input is ‘extra-compartmental’, while the parameters of interest used here to describe dynamic phenotypic restructuring are ‘intra-compartmental’. That is, the restructuring parameters (l, m, and k) do not include contributions due to thymic input. Furthermore, thymic cell input does not appreciably contribute to the dynamic changes (parameters I(0) and i in Table 2). The ‘transition age’ is a fundamental characteristic difference between naive cell gains and losses and memory cell gains and losses (broken lines in Fig. 5 whose behaviors are dominated by the difference in the terms: elA − emA).

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 261

Also, the limited, diminished input from the thymus does not account for the apparent linear decline with age of the numbers of T lymphocytes in the peripheral blood (Sansoni et al., 1993). While this may be due to a hematopoietic dysregulation among the elderly (Rothstein, 1993), it seems more likely that other factors are involved. Even though the donors used in these studies were in excellent health, it is possible that conditions exist whereby large numbers of T lymphocytes become sequestered at anatomical sites due, possibly, to low-grade chronic conditions. This would diminish the absolute number of cells recirculating through the peripheral blood. An alternative to this might be that cell adhesion molecule function is altered with age so that passage of cells to the peripheral circulation is hindered. In any case, we cannot account for this phenomenon in our model. Finally, in the paper from which we have utilized data (Cossarizza et al., 1992), two important questions are asked for which we can provide partial answers. First, why are the sharp differences observed in the expressions of the CD45R isoforms mainly observed during the first two to three decades of life? (As a corollary, why do a consistent number of T cells show a virgin phenotype in advanced age?) Our answer is that there is a persistent, dynamic renewal of the T-cell complement even with advanced age. As a percentage of the total cell complement, the thymus provides a continuous, albeit small, source of T cells. The dynamics of restructuring change, however, and are characterized by a ‘transition age’. Prior to this age, naive cell gains minus losses dominate, while after this age memory cell gains minus losses dominate. Second, why are cells with a virgin phenotype more represented among CD8 than among CD4 T lymphocytes? Based upon our analyses, there is a combination of contributing factors. The memory cell complement tends to reside within its compartment for a longer time than the naive cell complement within its compartment, for both CD4 and CD8 cells. However, the average annual turnover rate for CD8 cells is shorter than for CD4 cells. And, the naive to memory cell conversion rate is higher among CD4 cells than CD8 cells. CD8 cells are ‘processed through’ the peripheral compartments more quickly than CD4 cells. Our model provides a framework to describe how dynamic phenotypic restructuring of the T-cell complement occurs with age in healthy humans. Unfortunately, it does not explain why these kinds of changes occur, particularly with regard to differences in restructuring between the CD4 and CD8 T-cell subsets. The most likely explanation is that these differences arise due to the fundamental differences in the processing and presentation of antigens to CD4 and CD8 cells. CD4 cells respond to exogenously processed antigens via interactions/binding of the T-cell receptor (TCR) with MHC-II molecules on APCs, while CD8 cells respond to endogenously processed antigens and subsequent interactions/binding of the TCR with MHC-I molecules (Abbas et al., 1994). The affinities between TCR – MHC-II complexes on CD4 cells and TCR–MHC-I complexes on CD8 cells are undoubtedly different. It is not known if, or how, the affinities of these binding reactions might change with age. Differences in these reactivities could significantly contribute to the dynamics of restructuring presented here.

262 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264

Because of fundamental differences when reacting to antigens and conditions optimal for responses between naive and memory cells (Inaba and Steinman, 1984; Ernst et al., 1990; Geppert et al., 1990; Akbar et al., 1991), it would seem that dynamic restructuring could play some role in altered immune responses (responsiveness) in elderly humans. In our simple model, we have proposed three parameters that control phenotypic restructuring: l, gains minus losses of naive cells; m, gains minus losses of memory cells; k, conversion rate of naive to memory cells. These are separate, and separable, variables that can be studied independently. These parameters, especially the naive to memory conversion rate, k, could well vary with age. The framework provided here can be used to compare and contrast evolving knowledge of immune functional changes with age in healthy humans. Appendix A. Mean compartmental residence time In the analyses of material exchanges between compartments, it is evident that some materials (cells) will reside within their compartments for a period of time before moving to another compartment (naive to memory conversion) or leaving the system (cell death). An estimate of this ‘residence time’ is made by determining the mean compartmental residence time (MCRT), defined as the ratio of the amount that may be present over all time to the amount present due to an impulsive input at some arbitrary time zero (Jacquez, 1985, pp. 125–133). In order to distinguish between the average time spent within a defined compartment and the independent variable of chronological age used in our analyses, we assume a continuous function of time, y(t). y(t) is the amount of something inside a defined compartment, where t can range from 0 to . In this context, the independent variable, t, is the amount of time in years that a complement of cells may be present and is not the same as the independent variable of chronological age, A, of the study population. The MCRT is defined by: MCRT =

&



y(t) dt/y(0)

0

Considering the naive cell compartment singly, and without regard for extracompartmental inputs, the amount present at any time is N(t)= N(0)elt, where N(0) is the amount present when t = 0 and l is a parameter defined in the text. Then, because lB 0, it is straightforward to show: MCRT = − 1/l

for naive cells

(A1)

Similarly, considering the memory cell compartment singly, and without regard for extra-compartmental inputs: M(t)= M(0)emt. Because mB0, it follows that: MCRT = − 1/m

for memory cells

(A2)

Here, N(0) and M(0) are the initial amounts present at some arbitrary time 0, and are not the same as the initial value conditions used in the developing model (Eqs. (10) and (11)) when age A =0.

D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 263

References Abbas, A.K., Lichtman, A.H., Pober, J.S., 1994. Antigen processing and presentation to T lymphocytes. In: Cellular and Molecular Immunology. W.B. Saunders, Philadelphia, PA. pp, 115 – 135. Akbar, A.N., Salmon, M., Janossy, G., 1991. The synergy between naive and memory T cells during activation. Immunol. Today 16, 12–16. Bevington, P.R., 1969. Data Reduction and Error Analysis for the Physical Sciences. McGraw-Hill, New York. Cossarizza, A., Ortolani, C., Paganelli, R., Monti, D., Barbieri, D., Sansoni, P., Fagiolo, U., Forti, E., Londei, M., Franceschi, C., 1992. Age-related imbalance of virgin (CD45RA + ) and memory (CD45R0 + ) cells between CD4 + and CD8 + T lymphocytes in humans: a study from newborns to centenarians. J. Immunol. Res. 4, 118 – 126. Ernst, D.N., Hobbs, M.V., Torbett, B.E., Glasebrook, A.L., Rehse, M.A., Bottomly, K., Hayakawa, K., Hardy, R.R., Weigle, W.O., 1990. Differences in the expression profiles of CD45RB, Pgp-1 and 3G11 membrane antigens and in the patterns of lymphokine secretion by splenic CD4 + T cells from young and aged mice. J. Immunol. 145, 1295 – 1302. Farooqi, Z.H., Mohler, R.R., 1989. Distribution models of recirculating lymphocytes. IEEE Trans. Biomed. Eng. 36, 355–362. Franceschi, C., Monti, D., Sansoni, P., Cossarizza, A., 1993. The immunology of exceptional individuals: the lesson of centenarians. Immunol. Today 16, 12 – 16. Freitas, A.A., Rocha, B.B., 1993. Lymphocyte lifespans: homeostasis, selection and competition. Immunol. Today 14, 25–29. Geppert, T.D., Davis, L.S., Gur, H., Wacholtz, M.C., Lipsky, P.E., 1990. Accessory cell signals involved in T-cell activation. Immunol. Rev. 117, 5 – 66. Hulstaert, F., Hannet, I., Deneys, V., Munhyeshull, V., Reichert, T., DeBruye`re, M., Strauss, K., 1994. Age-related changes in human blood lymphocyte subpopulations. II. Varying kinetics of percentage and absolute count measurements. Clin. Immunol. Immunopath. 70, 152 – 158. Inaba, K., Steinman, R.M., 1984. Resting and sensitized T lymphocytes exhibit distinct stimulatory (antigen-presenting cell) requirements for growth and lymphokine release. J. Exp. Med. 160, 1717–1735. Jackola, D.R., Ruger, J.K., Miller, R.A., 1994. Age-associated changes in human T cell phenotype and function. Aging Clin. Exp. Res. 6, 25 – 34. Jacquez, J. 1985. Compartmental Analysis in Biology and Medicine. University of Michigan Press, Ann Arbor, MI. Ligthart, G.J., Corberand, J.X., Fournier, C., Galanaud, P., Hijmans, W., Kennes, B., Mu¨ller-Hermelink, H.K., Steinmann, G.C., 1984. Admission criteria for immunogerontological studies in man: the SENIEUR protocol. Mech. Ageing Dev. 28, 47 – 55. Michie, C.A., Mclean, A., Alcock, C., Beverley, P.C.L., 1992. Lifespan of human lymphocyte subsets defined by CD45 isoforms. Nature 360, 264 – 265. Miller, R.A., 1991. Aging and immune function. In: Jeon, K.W., Friedlander, M. (Eds.), International Review of Cytology, Vol. 124. Academic Press, San Diego, CA, pp. 187 – 215. Miller, R.A., 1996. The aging immune system: primer and prospectus. Science 273, 70 – 74. Reichert, T., DeBruye`re, M., Deneys, V., To¨tterman, T., Lydyard, P., Yuksel, F., Chapel, H., Jewell, D., Van Hove, L., Linden, J., Buchner, L., 1991. Lymphocyte subset reference ranges in adult caucasians. Clin. Immunol. Immunopath. 60, 190 – 208. Rogers, P.R., Pilapil, S., Hayakawa, K., Romain, P.L., Parker, D.C., 1992. CD45 exon expression in murine and human CD4 + T cell subsets. J. Immunol. 148, 4054 – 4065. Rothstein, G., 1993. Hematopoiesis in the aged: a model of hematopoietic dysregulation? Blood 82, 2601–2604. Sansoni, P., Cossarizza, A., Brianti, V., Fagnoni, F., Snelli, G., Monti, D., Marcato, A., Passeri, G., Orotlani, C., Forti, E., Fagiolo, U., Passeri, M., Franceschi, C., 1993. Lymphocyte subsets and natural killer cell activity in healthy old people. Blood 82, 2767 – 2773.

264 D.R. Jackola, H.M. Hallgren / Mechanisms of Ageing and De6elopment 105 (1998) 241–264 Scollay, R.G., Butcher, E.C., Weissman, I.L., 1980. Thymus cell migration. Quantitative aspects of cellular traffic from the thymus to the periphery in mice. Eur. J. Immunol. 10, 210 – 218. Steinmann, G.C., 1986. The human thymus: histophysiology and pathology. In: Mu¨ller-Hermelink, H.K. (Ed.), Current Topics in Pathology, Vol. 75. Springer-Verlag, Berlin, Germany, pp. 43 – 87. Stulnig, T., Maczek, C., Bo¨ck, G., Majdic, O., Wick, G., 1995. Reference intervals for human peripheral blood lymphocyte subpopulations from ‘healthy’ young and aged subjects. Int. Arch. Allergy Immunol. 108, 205–210. Thoman, W.L., Weigle, W.O., 1989. The cellular and subcellular bases of immunosenescence. In: Dixon, F.J. (Ed.), Advances in Immunology, Vol. 46. Academic Press, San Diego, CA, pp. 221 – 261. Westermann, J., Puskas, Z., Pabst, R., 1988. Blood transit and recirculation kinetics of lymphocyte subsets in normal rats. Scand. J. Immunol. 28, 203 – 210.

.

.