CELLULAR IMMUNOLOGY 117,4 14-424 ( 1988)

CD4+ CD8+ Cells Are Rare among in Vitro Activated Mouse or Human T Lymphocytes

KATHERINEKELLY,*LINDAPILARSKI,*,-~KENSHORTMAN,* AND ROLAND~COLLAY*

“The Walter and Eliza Hal/ Institute of Medical Research, Melbourne, Victoria 3050, Australia, and tDepartment ofImmunology, University ofAlbetia, Edmonton, Canada

Received April 4, 1988; accepted August 15, I988

The predominant cell type in the thymus expresses both of the function-associated T cell surface markers, CD4 and CD8, but CD4+ CD8+ cells are rare or absent outside the thymus. Double expression has therefore been assumed to be an indication of immaturity. However, recent reports have suggested that CD4+ CD8+ cells can appear in cultures of activated mature cells. We have therefore activated human peripheral blood lymphocytes and mouse spleen cells, lymph node cells, and cortisone-resistant thymocytes using a number of different stimulation regimes, and we have analyzed them at various times for CD4 and CD8 expression. In all cases, upon analysis of cultured cells by flow cytometry, CD4+ CD8+ cells were rare. A combination of microscopic analysis, cell sorting followed by microscopic analysis, and careful staining controls demonstrated that even when flow cytometry showed some CD4+ CD8+ cells, most of these were artifacts in the form of doublets or clumps of single positive cells or dead cells. Taking this into account, CD4+ CD8+ cells made up less than 1% in the mouse and less than 3% in the human T cell cultures at any time periods. We therefore found no evidence for the generation of large numbers of CD4+ CD8+ cells in cultures of mouse or human T cells. o 1988 Academic

Ress, Inc.

INTRODUCTION

Cell surface phenotype can reflect both the degree of maturity and the activation state of a cell. Some markers tend to be more useful in determining activation (e.g., IL-2 receptor, transferrin receptor), while others seem to be related primarily to lin- eage or maturity. The molecules CD4 and CD8 fall into the latter category, and it has generally been believed that they are not changed by activation. Thus, early thymo- cytes are CD4- CD8- (double negative), cortical thymocytes are CD4+ CD8+ (double positive), and mature T cells are CD4+ CD8- or CD4- CD8+ (single positive). This applies to both dividing and nondividing cells within these categories ( 1,2). However, several recent publications have suggested that mature single positive T cells from human (3) and rat (4) can, upon activation in vitro, take on a double positive pheno- type. Since this raises questions about the source of double positive cells in the thymus and since we had rarely detected such cells despite extensive experience culturing mouse lymphocytes, we have investigated the question in detail in the mouse, in order to see which cell types might change upon stimulation, and what stimuli were required. We also analyzed the phenomenon using human peripheral blood lympho-

414

0008-8749/88 $3.00 Cornnight 8 1988 by Academic Press, Inc. All rights of repmducticm in any form reserved.

CD4 AND CD8 ON ACTIVATED T LYMPHOCYTES 415

cytes in an attempt to reproduce the published findings (3). In no case, in either mouse or human, were large numbers of double positives generated, and most of those which appeared to be double positive when analyzed by flow cytometry proved upon further investigation to be clumps or dead cells.

MATERIALS AND METHODS

Mice. All mice were bred at The Walter and Eliza hall Institute under specific pathogen-free conditions; 6- to 7-week-old, male CBA/&H mice were used throughout.

Cortisone-resistant thymocytes (CRT).’ These were obtained 48 hr after intraperi- toneal injection of 2.5 mg of hydrocortisone acetate (Merck, Sharp, and Dohme, Pty. Ltd., Australia).

Mouse lymphoid cell suspensions. Mice were killed with COP and organs were re- moved immediately to cold balanced salt solution containing 10% fetal calf serum (BSS-FCS). Cell suspensions were made by teasing the tissue through a wire mesh into BSS-FCS and washing through FCS. Where there were many red blood cells present, cells were teased into 0.168 M NH&l (5), underlayed with FCS, and incu- bated for 5 min on ice, before centrifugation. Cells were resuspended in dead cell removal buffer (5) and filtered through a cotton-wool column into 1 ml BSS-FCS, spun, and resuspended in BSS-FCS or medium. Viable cell numbers were determined by eosin exclusion. Cell suspensions of cultured cells for flow cytometric analysis were depleted of dead cells by a density cut using metrizamide of density 1.085 g/ml (6).

Human lymphoid cell populations. Human peripheral blood mononuclear cells (PBL) were isolated from healthy normal adult donors by Ficoll-Paque (Pharmacia, Dorval, Quebec) density centrifugation, followed by two washes in RPM1 1640 me- dium containing 10% fetal calf serum. PBL from six different donors, provided by the Melbourne Red Cross Blood Transfusion Service, were studied.

Culture conditions for murine T cell activation. Cultures were set up in Nunc or Costar 24-well culture dishes in a volume of 1.5 ml per well, with at least 10 replicate wells for each treatment. Replicates were pooled at the time of assay. Cells were grown in modified RPM1 1640 medium containing 10% FCS and were cultured at an initial concentration of 3 X 1 06/ml in a humidified, 10% COz-in-air incubator at 37°C for l- 5 days. Stimulants and growth factors were included in the medium. Two stimulation regimes were used: (1) 1 rig/ml phorbol my&ate acetate (PMA) (Sigma, St. Louis, MO) +0.25 pLg/ml ionomycin (Calbiochem, La Jolla, CA) + 100 U/ml human recom- binant IL-2 (Cetus Corp., Emeryville, CA) +supernatant from concanavalin A (Con A)-activated spleen cells (CAS) - 10%; (2) 2.5 &ml Con A + 10% CAS.

Culture conditions for human T cell activation. PBL were cultured at a density of IO6 cells/ml in 30 ml of RPM1 1640 medium modified to include 25 mM Hepes buffer, 0.3 1% sodium bicarbonate, and 10% fetal calf serum in Corning 75-cm2 tissue culture flasks. All cultures were maintained at 37°C in a 10% CO*-in-air mixture for 3-6 days. Since we wished to reproduce the culture conditions used by Blue et al. (3),

’ Abbreviations used: Av, avidin; BSS, balanced salt solution; Con A, concanavalin A; CAS, Con A- activated spleen cell supematant; CRT, cortisone-resistant thymocytes; FCS, fetal calf serum; FITC, fluo- resczin isothiocyanate; PBL, peripheral blood lymphocytes; PE, phycoerythrin; PHA, phytohemagluti- nin; PMA, phorbol my&ate acetate; TCGF, T cell growth factor.

416 KELLY ET AL.

initially cultures were stimulated using 60 pg/ml of Con A. However, in our hands, this concentration was toxic. Therefore PBL were stimulated with either 5-25 &ml of Con A or 1% PHA as indicated, together with T cell growth factor (TCGF). TCGF was the supernatant of PBL which had received 1500 rad of gamma irradiation to curtail their proliferation, and which were stimulated with 10 pg/ml of Con A for 3 days. This preparation had sufficient lymphokines to support the growth of PHA- stimulated human thymocytes at concentrations of l- 10%. To mimic the conditions of Blue et al. (3), TCGF where added was used at a final concentration of 10%.

Monoclonal antibodies used. All anti-mouse antibodies used for fluorescent stain- ing were prepared from supernatants of hybridomas grown in this laboratory. They were GK1.5 (anti-CD4) used as a biotin conjugate (7) or phycoerythrin conjugate (purchased from Becton Dickinson, Mountain View, CA) and 53.6.7 (anti-CD8) used as a fluorescein conjugate (8). Phycoerythrin avidin (PE-Av) (Serotec Laboratories, Bicester, England) was used as a second stage for the biotinylated antibody. For the staining of human cells, FITC-conjugated Leu 2 (CD8) and Leu 3 (CD4) and isotope control antibodies (IgG1-PE and IgGi-FITC) were purchased from Becton Dickinson. T8-RDl (PE-conjugated antiCD8), and TCRDl (PE-conjugated antiCD4) conju- gates were purchased from Coulter Immunology (Hileah, FL).

Fluorescence staining. When staining for analysis, 0.5-3.0 X lo6 cells were sus- pended in 30 ~1 of the appropriate reagent and incubated for 30 min on ice. When staining for cell sorting, lo7 cells in 100 ~1 were stained per tube. Cells were washed after each stage by diluting in BSS-FCS and centrifuging through FCS. After the last wash murine cells to be analyzed were resuspended in BSS-FCS containing 5 pg/ml propidium iodide (Calbiochem, La Jolla, CA) to allow electronic exclusion of dead cells. Cells to be sorted were suspended in BSS-FCS without propidium iodide. Since our flow cytometry facility requires fixation of all human material, human cells were fixed in 1% formaldehyde in phosphate-buffered saline with 2% glucose and 5 mM sodium azide. For this reason propidium iodide could not be used for human cell analysis.

Flow cytometric analysis and cell sorting. A modified FACS II (Becton Dickinson FACS Systems) was used for sorting and analysis. Forward light scatter, red fluores- cence (from PE conjugates and propidium iodide), and green fluorescence (from flu- orescein conjugates) were measured using a single laser, exciting at 488 nm. Red blood cells were excluded electronically on the basis of forward light scatter. Dead cells were excluded on this basis and, in the case of murine cells, by propidium iodide fluorescence (which registers off scale in the red for dead cells); lo-25,000 cells were analyzed per file and contours plotted at three levels representing threefold differences in cell numbers, e.g., for a 25,000 cell file 3, 10, and 30 cells per square on a grid of 64 X 64 squares.

RESULTS

Generation ofApparent CD4+ CD8+ Ceils in Cultures of Murine Lymphocytes

In order to look for the generation of CD4+ CD8+ cells in cultures of mouse lym- phocytes, we cultured cells from spleen and lymph nodes and from cortisone-treated thymus. The population of CRT contains a sample of the mature single positive cells from the thymus and was included as a possible source of intrathymic generation of double positive cells. All these populations were stimulated by the mitogen Con A,

CD4 AND CD8 ON ACTIVATED T LYMPHOCYTES 417

b 3 I 3 Oays

IL culture aD 100 2 $.i:l :;j _i::,

<:.:. .‘. j:::. :,

,o

UL

&_ :..(-yilj :_:: .r:

,:-: ‘~~:::: :‘:‘, : ; ,:_:,::i... :jj,:_

._::j..::::::!:..i :.. ii, 2;$zjj? $..

, iij:i_.Y. :::

5 oiys culture

-1000

-500 s

E

5 -0 L

iti ul

-low 3

-500

-0

- 1000

-500

2oo"

CD4 Fluorescence Forward Light Scatter

FIG. 1. Analysis of CD4 and CD8 expression and forward light scatter of lymph node cells cultured in Con A and CAS for l-5 days. The left hand boxes show two-color immunofluorescence contour plots of CD4 and CD8 expression. The right hand boxes show the low-angle (fotward) light scatter distribution of single-parameter histograms of the same cells.

believed to involve the T cell receptor, or bypassing their receptor by using PMA and ionomycin. Cultures were followed for 5 days and at various times assessed for viable and nonviable cell numbers, cell size (determined by low-angle light scatter) as an indication of activation, and expression of CD4 and CD8 antigens. The results were similar in all cases. Cultured lymph node cells are shown in Figs. 1 and 2 as examples, and the data from all populations are summarized in Table 1.

The viable cell recovery after 1, 3, and 5 days of culture was around 50, 55, and 40% of the input cells, respectively. The original population contained approximately 80% T cells; this proportion increased during culture to >90% by Day 5. Cell size, as determined by forward light scatter analysis (Figs. I and 2) increased markedly during culture, indicating that the cells had been activated. At all time points the predomi- nant cell type was either CD4+ CD8- or CD4- CD8+ (see Figs. 1 and 2). Some drop in the levels of expression of CD4 and CD8 in the PMA- and ionomycin-stimulated

418 KELLY ET AL.

CD4 Fluorescence Forward Light Scatter

FIG. 2. Analysis of CD4 and CD8 expression and forward light scatter of lymph node cells cultured in PMA, ionomycin, and recombinant IL-2 for 1-S days. Details as for Fig. 1.

cultures was apparent and will be discussed in detail elsewhere. Small numbers of apparently CD4+ CD8+ cells could be seen in many cultures, at all time points. The levels of such apparent double positives are summarized in Table 1, and in no case did they exceed 5.6% of viable cells. A series of experiments in which allogeneic cells were the stimulus (mixed lymphocyte cultures) gave similar low values of apparent double positives (data not shown).

Detailed Analysis of in Vitro Generated Apparent CD4+ CD8+ Cells

The apparent levels of CD4+ CD8+ cells generated by stimulation in culture were clearly low, as shown in Table 1. It was important to determine whether these low

CD4 AND CD8 ON ACTIVATED T LYMPHOCYTES 419

TABLE 1

Apparent Percentage of CD4+ CD8+ Cells in Cultures of Murine Lymphocytes

Percentage apparent CD4+ CDS+ cells

Stimulus and Source of growth factor T cells

Uncultured Days of culture control (*SD) 1 2 3 5

Con A + CAS Lymph node 1.3kO.5 5.4 4.5 1.7 4.5 Spleen 1.6kO.8 5.3 -2.0 ND 4.0 CRT 1.2 f 0.3 4.2 ND 5.5 0.7

PMA/ionomycin Lymph node 1.5kO.9 1.1 1.8 4.0 3.1 + IL-2 Spleen 1.3 +0.8 4.7 1.9 3.8 3.9

CRT 1.3 kO.2 0.4 ND 2.4 1.3

Note. The level of CD4+ CD8+ cells was assessed by two-color immunofluorescence and flow cytometry after various periods in culture. Background values have been subtracted, based on two sets of single-color staining and second stage alone controls. Each time point represents a single determination. Uncultured control values are means of two to four determinations, since each set of culture times was composed of several experiments.

values did indeed reflect a low level of real double positive cells. Since it is impossible to do the perfect staining controls in two-color experiments [e.g., the measured back- ground in green only controls may or may not overlap the background in red only controls (see Fig. 3)], the only way to unequivocally establish this was to separate these cells out on the cell sorter and reanalyze them, either by examining the “puri-

Anti-CD4

J

Anti-CD8

/

CD4 Fluorescence

FIG. 3. ExampIes of two-color fluorescence contour profiles of cultured lymph node cells stained with both anti-CD4 and anti-CD8, with each antibody alone, or no antibodies (but second stage added). Cultures were stimulated for 3 days with PMA, ionomycin, and recombinant IL-2.

420 KELLY ET AL.

CD4 Fluorescence

FIG. 4. Sorting of CD4+ CD8+ cells from 5&y cultures of lymph node cells stimulated with Con A and CAS. The windows defined by the broken line were used for sorting the double positives. The proportions in each category are as follows: left box, CD& CD8-, 1.5%; CD4- CDS+, 28%; CD4+ CD8-, 67%; CD4+ CD8+ (boxed), 3.1%; right box, CD4- CD8-, 2.6%; CD4- CD8+, 37%; CD4+ CD8-, 42%; CD4+ CD8+, 18%.

fied” double positives on a fluorescence microscope, or by rerunning the sorted cells on the flow cytometer. When this was done with thymocytes, which contain about 75-80% real double positive cells, our cell sorter yielded a population which was, upon reanalysis, greater than 96% double positive (data not shown). When we took a population of lymph node cells, activated for 5 days with Con A and CAS, which contained 3.1% apparent double positive cells, the result was very different. Most of the sorted cells were, upon reanalysis, single positive with double positives being extremely rare (Fig. 4). This strongly suggests that the original apparent double posi- tives were mainly clumps or doublets of single positive cells, a fact which would ac- count for the larger size (forward light scatter) of the apparent double positive cells (data not shown). This would also explain why the reanalyzed cells contained roughly equal numbers of single positives (37% CD8+, 42% CD4+), while the original popula- tion did not (28 and 67%). However, it could also be that the cell sorter was very inefficient at sorting double positive cells when they are a small minority, and the reanalysis reflected this. We therefore sorted and then reanalyzed a mixture of 96% uncultured lymph node cells and 4% thymocytes, so that there would be at least a few percent of “real” double positives present. The result is shown in Fig. 5. The sorted population (4.7% double positives in the original sort) upon reanalysis con- tained a mixture of double positive cells (42%) and single positive cells, exactly as would be expected from a mixture of real double positives (from the added thymus) and doublets or clumps from the lymph node cells. This strongly supports the validity of the data shown in Fig. 4.

All the above conclusions were confirmed by fluorescence microscopy on the CD4- and CD8-labeled cultured cells. Real double positives were less than 1% in the stained, unsorted cultures, and in some cases were difficult to find. Side by side analy- sis of thymocytes gave the expected values of 75-80% double positive. Fluorescence microscopy also confirmed our conclusions on the nature of the cells sorted as appar- ent double positives. In a series of experiments where 3-4% of cultured cells sorted into the apparent double positive category, 70-80% of these were found to be single

CD4 AND CD8 ON ACTIVATED T LYMPHOCYTES 421

CO4 Fluorescence

FIG. 5. Sorting of CD4+ CD8+ cells from a mixture of normal lymph node cells (containing l-2% “dou- ble positives”) and 4% thymocytes (containing 80% double positives). The cells sorted are indicated by the dotted box. The proportions in each category are as follows: left box, CD4- CD8-, 16%; CD4- CD8+, 2 1%; CD4+ CD8-, 58%; CD4+ CD8+ (boxed), 4.7%; right box, 15,20,23,42%, respectively.

positives (in a 1: 1 ratio), lo-20% were dead cells (often autofluorescent), l-2% were persistent doublets, and only l-5% were true double positives.

We therefore concluded that even the low values of generated CD4+ CD8+ cells shown in Table 1 were a considerable overestimate and that real double positives (and there were indeed some in the microscopic analysis) made up less than 1% of the cultured cells.

Generation ofApparent CD4+ CD8+ Cells in Cultures of Human Lymphocytes

In order to repeat the published experiments with human cells, we used initially the conditions as described, including the 60 pg/ml Con A (3). However, in our hands, at most 1% of the starting cells were viable at Day 3 when these concentrations of Con A were used. We therefore reduced the Con A concentration. In order to remain as close as possible to the original method, we used 25 pg/ml which gave a viable cell recovery at Day 3 of 3-30%. However, we also used Con A at 5 and 10 pg/ml and PHA at I%, which gave recoveries of 30-80% at 3-6 days of culture. Since most of the cells at these times were large, we assumed adequate activation had been achieved under these conditions. The cells were double stained with antibodies to CD4 (green) and CD8 (red) and the number of double positive cells was determined by flow cytometry. In this case the control was the same population of cells stained with antibodies to CD4 alone (green) or CD8 alone (red), the two being mixed together after staining, i.e., a mixture of two single-stained populations (which should not include any real double positives). The results are shown in Fig. 6 and Table 2, the values shown having had the background ( 1- 1 l%, determined from the mixture of single-stained cells) subtracted. The same staining protocol gave 52 + 15% (n = 11) double positives in fresh, uncultured thymus cell preparations. The proportion of double positives in the cultured PBL ranged from 1 to 9%, and this level did not change with different stimuli or different culture times. In several samples these dou- ble positives were sorted. The sorted population was found by fluorescence micros- copy to include O-25% real double positives, but also many single positives and con- siderable debris. Fluorescence microscopic analysis of cells from an unsorted culture,

422 KELLY ET AL.

CD4 Fluorescence Control Fluorescence

FIG. 6. Two-color immunofluorescence contour plots of human PBL cultured for 6 days in 5 &ml Con A + TCGF. Box (a) shows the two-color staining, box (b) shows the mixture of cells stained with each antibody alone, and box (c) shows the staining when both primary antibodies were replaced with isotype- matched control antibodies. The proportion of double positives in each case (window marked by broken lines in boxes (a) and (b)) are 4.8, 1.1, and 0.6%, respectively.

which gave an apparent value of 8% double positives by flow cytometry, revealed only 2.7% real double positives, while a second culture which gave an apparent value of 9%, revealed 0.8% real double positives by microscopy. Thus, as in the mouse, the values determined by flow cytometry for apparent double positives in cultured cells was a considerable overestimate. However, there do seem to be some real double positives (perhaps O-2%) present in some cultures. Cultures of PBL enriched by nega- tive selection for CD4+ CD8- or CD4- CD8+ cells, when analyzed at Day 5 by flow cytometry and fluorescence microscopy, also did not contain real CD4+ CD8+ cells at levels greater than O-2% (data not shown).

DISCUSSION

The purpose of this study was to analyze the cells and signals involved in the gener- ation of CD4+ CD8+ (double positive) cells from mature CD4- CD8+ or CD4+ CD8- (single positive) cells in vitro, since reports on human (3) and rat (4) T cells suggested

TABLE 2

Apparent Percentage of CD4+ CD8+ Cells in Cultures of Human PBL

Percentage apparent CD4+ CD8+ cells

Stimulus Uncultured

control

Days of culture

4 5 6

PHA(l%) 1.5 6.9 ND ND Con A (25 &ml) 3.1 3.6 ND 8.0 Con A (25 &ml) + TCGF 0.7, 1.5 4.0,7.2 6.2 f 2.9 ND Con A ( 10 &ml) + TCGF ND ND ND 9.0 Con A (5 &ml) + TCGF ND ND ND 3.7

Note. The level of CD4+ CD8+ cells was asses& by two-color immunofluorescence and flow cytometry after various periods in culture. Background values have been subtracted (see text). Single numbers repre- sent single determinations, two values represent determinations on two separate cultures, and values with standard deviations are the means of four determinations.

CD4 AND CD8 ON ACTIVATED T LYMPHOCYTES 423

that this could occur. However, we have been unable, either in mouse or human T cell cultures, to generate large numbers of double positive cells, despite the use of a range of cell sources, stimuli, and culture times. Our data show that flow cytometric analysis can lead to considerable overestimation of the number of double positive cells, especially in “difficult” preparations, such as cultures containing dead cells and/ or agglutinating stimulants, where clumps, doublets, or dead cells will often fall into this category. These problems are particularly acute when double (or multiple) posi- tive populations are analyzed; less so for double negatives or single positives. When all these possible artifacts were taken into account and the appropriate controls per- formed, we found the proportion of “real” double positives to range from 0 or unde- tectable to about 2% in cultures of human PBL, and to be well under 1% in cultures of murine cells. Thus we could find no evidence for the generation of large numbers of double positive cells in cultures of mature single positive cells. This conclusion is supported by the rarity of double positive cells among cloned (i.e., activated) human PBL T cells, although CD4+ CD8+ clones have occasionally reported (9, 10).

The reasons for our inability to contirrn the previously published studies are not clear. Species differences do not appear to provide an explanation. Our data using human cells closely followed the published experiments (in which only flow cytomet- ric analysis was used), except the Con A concentration was reduced in order to re- cover enough viable cells for analysis. We can only postulate that some of the artifacts mentioned above played some part or that there is some difference in the cells or culture conditions that are not readily apparent from the published report. The sug- gestion in the previous study (3) that both CD4+ CD8 and CD4- CD8 single posi- tives need to be present for the generation of double positive cells, and that the pres- ence of lectin is also necessary, would support the possibility of clumps or doublets of cells giving rise to the apparent high levels of double expression. These might also appear as large cells on scatter analysis.

The experiments using rat T cells were done with a completely different approach, using two-label immunohistochemistry on cytospin preparations of mitogen-stimu- lated cells (4). We can find no obvious explanation for the difference between these results and our own.

The expression of CD4+ CD8+ cells among freshly isolated lymphoid cells is also of interest. In the experiments on mouse T cells reported here, and in many others conducted in our laboratory, double positive cells are rare in spleen or lymph nodes (or in periphery), with values ranging from 0 to 2% in flow cytometric analysis of normal cells. We have not always subjected these analyses to the rigorous criteria used in this paper, so the values can only be considered an upper limit. Double posi- tive cells are also rare in mouse blood ( 11). Blue et al. (3) reported levels of CD4+ CD8+ cells in fresh human blood of 0.5-8% (flow cytometric analysis), rather similar to our own in this report (0.7-3.1% by flow cytometry). However, in a large number of fresh blood samples analyzed by fluorescence microscopy, double positive cells were always less than 1% of T cells (L. Pilarski, unpublished data). Double positive cells also appear to be rare in normal rat tissues (12). Thus it would seem that the widely held view that double positives are rare outside the thymus in normal rats, mice, and humans remains correct. There is one report of large numbers of double positive cells in the blood of pigs ( 13).

On the other hand there are reports that in antigen-stimulated rats, double positive cells appear in rather large numbers in the spleen (12) and in blood (14). In one of

424 KELLY ET AL.

these reports ( 14), the double expressers in the blood carried low levels of the antigen, ruling out the possibility of clumping artifacts. The low levels could possibly be ac- counted for by absorption of soluble CD8 (15) to initially CD4+ CD8- cells. However, this is an unlikely explanation for our own small proportion of CD4+ CD8+ cells, as the CD8 levels on these were high.

In conclusion, we can find no evidence for the existence of large numbers of CD4+ CD8+ cells outside the thymus of mice or humans, either in normal populations or in cells activated in vitro. We therefore feel it is unlikely that activated mature cells contribute in a significant proportion to the CD4+ CD8+ cells found in large numbers in the thymus. It remains possible, however, that the rare CD4+ CD8+ cells isolated from the peripheral lymphoid organs have a unique function or properties, and that they may sometimes be found within the thymus.

ACKNOWLEDGMENTS

We thank Anne Wilson and Angela D’Amico for help with some of the experiments, and Frank Battye, Justin Beall, and Mark Cozens for assistance with cytometric analysis. The Melbourne Red Cross Blood Transfusion Centre generously provided samples of blood from randomly chosen normal individuals.

This work was supported in part by the National Health and Medical Research Council of Australia, by the Medical Research Council of Canada, by Grant AI 173 10 from the National Institutes of Health, USA, and by the C. H. Warman Research Fund, Australia. L. Pilarski is the recipient of an Eleanor Roosevelt International Cancer Fellowship from the International Union Against Cancer. This work was performed while L. Pilarski was on Study Leave from the University of Alberta at The Walter and Eliza Hall Institute of Medical Research.

REFERENCES

1. Scollay, R., Bartlett, P., and Shortman, K., Immunol. Rev. 82,79, 1984. 2. Reinherz, E. L., and Schlossman, S. F., Cell 19,82 1, 1980. 3. Blue, M.-L., Daley, J. F., Levine, H., and Schlossman, S. F., J. Zmmunol. 134,228 1, 1985. 4. Bevan, D. J., and Chisholm, P. M., Immunology 59,62 1, 1986. 5. Shortman, K., Williams, N., and Adams, P., J. Zmmunol. Methods 1,273, 1972. 6. Von Boehmer, H., and Shortman, K., J. Zmmunol. Methods 2,293, 1973. 7. Dialynas, D. P., Wilde, D. B., Marrack, P., Pierres, A., Wall, K. A., Havran, W. G., Otten, G., Loken,

M. R., Pierres, M., Kappler, J., and Fitch, F. W., Zmmunol. Rev. 74,29, 1983. 8. Ledbetter, J. A., and Herzenberg, L. A., Zmmunol. Rev. 47,63, 1979. 9. Pilarski, L. M., Mant, M. J., Reuther, B. A., Carayanniotis, G., Otto, D., and Krowka, J. F., Blood66,

1266, 1985. 10. Fazekas de St. Groth, B., Gallagher, P. F., and Miller, J. F. A. P., Proc. Natl. Acad. Sci. USA 83,2594,

1986. 11. Scott, B. M., Hogarth, P. M., Scollay, R., and McKenzie, I. F. C., J. Zmmunogenet., in press. 12. Spickett, G. P., and Mason, D. W., Eur. J. Zmmunol. 13,785, 1983. 13. Saalmiiller, A., Reddehase, M. J., Biihring, H.-J., Jonjic, S., and Koszinowski, U. H., Eur. J. Zmmunol.

17,1297,1987. 14. Godden, U., Herbert, J., Stewart, R. D., and Roser, B., Transplantation 39,624, 1985. 15. Fujimoto, J., Levy, S., and Levy, R., J. Exp. Med. 159,752, 1983.