Immunobiol., vol. 190, pp. 385-398 (1994) © 1994 by Gustav Fischer Verlag, Stuttgart

Department of Anatomy and Cell Biology, McGill University, Montreal, Canada

The Development of Natural Killer (NK) Cells from Thy-lIOLin-Sca-l + Stem Cells: Acquisition by NK Cells in vivo of the Homing Receptor MEL-14 and the Integrin Mac-l

SANDRA C. MILLER

October 27, 1993 . Accepted in revised form March 11, 1994

Abstract

The present study aimed to follow the development of natural killer (NK) cells in lethally irradiated BA mice reconstituted with 500 syngeneic Thy 1.11oLin-Sca-1+ stem cells. The proportions of NK 1.1 + lymphoid cells were assessed from smears of cell suspensions from the spleen and bone marrow by means of immunofluorescence microscopy, at 7, 9, 11,14,17,21,24 and 28 days after stem cell injection. At the same intervals, moreover, the proportions of NK 1.1 + lymphoid cells bearing either the homing receptor recog­nized by mAb MEL-14, or the integrin Mac-1 were recorded using double immuno­fluorescence microscopy, labelling variously with fluorescein isothiocynate and avidin­Texas Red. The results demonstrate that NK 1.1+ lymphoid cells re-appear by 11 (spleen) to 14 (bone marrow) days after injecting syngeneic Thy 1.11oLin-Sca-1+ stem cells. Moreover, in the absence of apparent stimulation, the newly developed NK 1.1+ lymphoid cells spontaneously express the homing receptor MEL-14 and the integrin Mac-1. The very similar patterns of acquisition of these latter 2 molecules on NK 1.1 + cells in the spleen during their recovery in the post-stem cell injection period suggests that MEL-14 and Mac-1 may co-express on the same NK 1.1 + cells. The absence, or low levels of both molecules on the newly developed NK 1.1 + cells while still in the bone marrow suggests that NK cells may progressively acquire these molecules outside that organ, en route to andlor within the vasculature of the spleen, their normal, primary destiny.

Abbreviations: BSA = bovine serum albumin; FA CS = fluorescence-activated cell sorter; FCS=fetal calf serum; HBSS = Hanks' balanced salt solution; PBS=phosphate-buf­fered saline; HEV = high endothelial venule; ASGM-J = ganglio-n-tetraosylceramide

386 . S. c. MILLER

Introduction

The origin of natural killer (NK) cells remains inconclusive. While some evidence suggests that NK cells belong to the T cell lineage (1-6), other, equally compelling evidence, indicates a monocyte-macrophage, or even a granulocyte origin (7-10). However, a vast literature has accumulated over several years which indicates that NK cells represent a separate lineage, co­evolving with, or preceding, the development of other immune and hemopoietic cell lineages. Another suggestion, nevertheless, has been pro­posed that natural killing is a phenomenon, or characteristic shared by several immune cell lineages (11, 12).

The currently accepted definition of NK cells holds that they are granule­bearing, small and/or large (depending on species) lymphoid cells, the mature forms of which always bear at least 2 known surface molecules, i.e., ASGM-1, and NK 1.1 (mice), or equivalent in other species. NK cells from mice also possess the homing receptor identified by the monoclonal anti­body, MEL-14 (13) and NK cells from various species have been shown to exhibit the integrins CDlla, CDllb (Mac-I, CR3), and CD11c (14-18) together with CD18.

It is not known whether the homing receptor and the integrins appear on NK cells only after appropriate stimuli, or whether NK cells possess these molecules constitutively. That the latter possibility may obtain, derives from previous observations that the glycosphingolipid, ASGM-1, is found on 100 % of newly-formed, unstimulated mature (lytic) and maturing NK cells (19-22). By contrast, this molecule can be found on activated, or stimulated (but not quiescent-unstimulated) T cells (22-24), where it is believed to be linked to functional activity. Since mature NK cells are by definition spontaneously functional, the rapid appearance on newly-formed NK cells of homing receptors and adhesion molecules may enable these cells, while still in the blood, to become spontaneously «homing ready» and destined for tissue entry under appropriate circumstances, without first engaging in the processes of producing and expressing several transendothe­lial migration molecules.

The aim of the present work was two-fold: (1) to establish whether NK cells can develop independently, as an autonomous cell lineage, by record­ing the earliest time of appearance of NK cells in the spleen and the bone marrow after Thy-110Lin-Sca-1 + stem cell injection in lethally irradiated host mice, and, (2) to establish whether certain molecules, specifically the MEL-14 homing receptor and the Mac-1 integrin, ever appear in vivo on NK cells in these stem-injected, unstimulated host mice.

The results point to an early appearance of NK cells, possibly arising directly from the Thy 1.11oLin-Sca-l + hemopoietic stem cell. Moreover, among the earliest molecules to appear on the developing NK cells in vivo are the MEL-14 homing receptor and the Mac-1 integrin. Both molecules are progressively acquired by spleen-localized, but not bone marrow­localized NK cells with time after stem cell injection.

Natural Killer Cells from Pluripotent Stem Cells . 387

Materials and Methods

Animals

Male BA mice (C57Bl congenic), Thy 1.1, Ly 5.1, were used both as stem cell donors and recipients. Donors were aged 4 wk and recipients 8-12 wk. In control experiments, C57Bl/6 mice (Thy 1.2, Ly 5.2) served as recipients for either syngeneic stem cells or stem cells from BA donors. All mice were housed in micro-isolator units in the animal facility at Stanford University, Stanford, CA, USA, and provided with acidified water (pH 2.5) and food ad libidum.

Stem cell reconstitution of recipient mice

Mice were lethally irradiated (9 Gy) by a 250-kV x-ray machine at 100 rad/min in two doses (4.5 Gy each) with a 3 h interval. After irradiation, mice were maintained on antibiotic water containing 106 U/liter of polymyxin B sulfate and 1.1 g/liter of neomycin sulfate. The next day, 500 sorted bone marrow-derived stem cells (Thy 1.1!oLin-Sca-1 +) were intravenously injected in 200 f!l HBSS/mouse via the retro-orbital plexus under ether anesthesia. Reconstitution of mice injected with «100 Thy 1.1!oLin-Sca-1+ stem cells has been previously established (25, 26).

Stem cell preparation

Bone marrow cells were obtained by flushing femora and tibiae into HBSS + 2 % FCS and 10 mM Hepes buffer, as previously described (25, 27). Cells were then stained sequentially by incubating on ice for 20 min with lineage marker rat antibodies to B220 (RA3-6B2), CD4 (GK-1.5), CD8 (53.6.72), Gr-1 (RB6-8C5), Mac-1 (MI/70.15.11.5) and erythrocytes (antibody TER -119). The cells were then washed through a FCS cushion and exposed to phycoerythrin-conjugated goat anti-serum to rat immunoglobulin (Biomeda, Foster City, CA), washed as above and incubated on ice for 20 min in 20 % normal rat serum to block free binding sites. This was followed by the addition of biotinylated rat antibody to Sca-1 (antibody E13 161-7) and directly fluoresceinated mouse antibody to Thy-1.1 (antibody 19XE5). The cells were again washed and exposed to Texas Red-conjugated avidin (Cappel Laboratories, Malvern, PA, USA). After the final wash, cells were re-suspended in HBSS containing 1 f!g/ml propidium iodide. The labelled cells were analyzed and sorted with a modified (28) dual laser FACS (Becton Dickinson Immunocytometry Systems, Mountain View, CA, USA). Whole bone mar­row cells were separated on the basis of background levels of fluorescein (Thy-I. 1-) from all higher levels offluorescein (Thy-1.1+), background levels of Texas Red (Sca-1-) from high levels of Texas Red (Sca-1+), and background levels of phycoerythrin (Lin-) from high levels of phycoerythrin (Lin+). Dead cells were excluded from the analysis by propidium iodide staining. After sorting, the bone marrow fractions were subjected to a second analysis by FACS.

Preparation of spleen and bone marrow cells from stem cell recipients

Stem cell-injected BA recipients were killed at intervals of 7,9,11,14,17,21,24, and 28 days after stem cell injection. Spleens were removed and pressed through stainless steel screens into petri dishes containing ice-cold PBS + 1 % BSA + 0.1 % sodium azide. From the same mice, bone marrow was obtained from both femora by flushing through the bones repeatedly with ice-cold PBS (above). All spleen and bone marrow samples were suspended and layered onto approximately 1 ml FCS for 5 min to allow sedimentation of non-cellular clumps through the serum. The cell suspensions were removed and cen­trifuged for 7 min (100 g, 4°C) to yield a supernatant and a cell pellet which was then re-

388 . S. c. MILLER

suspended and centrifuged once more as above. After the final wash, 20--60 smears were made from the pooled spleens and pooled femoral bone marrow of 3-5 mice per sampling interval. Assessment of cells on smears was considered preferable to analyses on cytospots because the former provided a greater abundance of cells, collectively, and of NK cells, specifically, the numbers of the latter, accounting for only 1-5 % of all the hemopoietic cells found in the bone marrow and spleen, even in normal, unmanipulated mICe.

Differential analyses of hemopoietic cells

From the Giemsa-stained smears of the spleen and bone marrow, the nucleated hemopoietic cells were morphologically identified as fully described elsewhere (29, 30), and differential counts of 1000 and 2000 nucleated cells/organ/interval were performed. All non-lymphoid cells, including nucleated erythroid, monocyte-macrophage and granulocytic cells were collectively categorized. All lymphoid cells, morphologically defined previously (29-32), normally range in size from 6.5 flm-12.0 flm in nuclear diameter in smears and were separately categorized.

NK cell identification and enumeration

Immunofluorescence microscopy was the method selected to identify and enumerate NK cells for 2 reasons. Firstly, their potentially very low frequency, especially at the earliest post-stem cell injection intervals would have resulted in many or all NK cells being undetected in FACS profiles. The second advantage of immunofluorescence microscopy derives from the fact that the NK cell «specific» marker, NK 1.1, can be found, in certain abnormal situations, on cells other than classical (lymphoid) NK cells, i.e., on cells of the myeloid lineage during in vitro incubation (33), or, in vivo, on promonocytes in mice bearing the autoimmune (motheaten) defect (34). The possibility that such atypical NK 1.1 + cells might develop following lethal irradiation and stem cell reconstitution, therefore, could not be excluded. FACS analysis would not permit selective elimination of any potentially occurring NK 1.1+ non-NK cells.

In the present studies, NK 1.1 + lymphoid cells were identified on fixed smears of the spleen and bone marrow from experimental and control mice at each post-stem cell injection interval. Freshly prepared smears were thoroughly air-dried, pre-fixed in ice­cold methanol (100 %) for V2 h, placed in methanol/PBS (1: 1) for 5 min washed 3 times in ice-cold PBS + 1 % BSA + 0.1 % sodium azide for 5 min/wash. Slides were then placed horizontally, covered with a 1 :25 dilution of protein G purified, rat IgG2 monoclonal anti-mouse FcR antibody for V2 h at room temperature in 100 % humidity. This treatment blocked Fc receptors known to be present on many NK cells. Smears were subsequently washed 3 times in PBS/BSA (5 min/wash) in coplin jars, again placed horizontally in 100 % humidity at room temperature and covered for 2 h with a 1: 10 dilution of protein G purified, biotinylated mouse IgG2a monoclonal anti-mouse NK 1.1 antibody. After incubation, smears were washed 5 times for 5 min/wash in PBS/BSA, incubated for 8 min with a 1 :400 dilution of avidin Texas Red in PBS/BSA, and washed as above. Controls for specific labelling consisted of (1) incubating fixed spleen and bone marrow smears with avidin Texas Red only, the results of which confirmed the absence of biotin in hemopoietic cells (35), or, (2) staining fixed smears of a hemopoietic tissue devoid of NK cells, i.e., BA strain adult (3-4 mo) thymus from 5 individual mice, with biotinylated anti-NK 1.1 antibody and avidin Texas Red.

To assess the presence of Mac-Ion NK cells, fixed smears prepared from the cells of both the spleen and bone marrow, after staining with biotinylated anti-NK 1.1 antibody, . were exposed to a 1:5 dilution of protein G purified FITC labelled rat IgG2b monoclonal anti-mouse Mac-1 antibody, incubated for 2 h, washed 10 times for 5 min each in PBS/

Natural Killer Cells from Pluripotent Stem Cells . 389

BSA+0.1 % sodium azide, incubated with avidin Texas Red as above and again washed. Other smears were stained with a 1:5 dilution of FITC-labelled anti-NK 1.1 antibody for 2 h, washed as above, incubated for 2 h with a 1: 10 dilution of protein G purified, biotinylated rat IgG2a monoclonal anti-mouse lymphocyte homing receptor antibody, MEL-14, washed, incubated with avidin Texas Red and again washed.

All antibody-stained slides were mounted in a 1: 1 solution of glycerol in PBS, cover slipped and stored in the dark at 4°C for 1-2 days at which time they were analyzed by fluorescence microscopy (Nikon Microphot FXA). The proportions of NK 1.1+ cells of lymphoid morphology (confirmed by phase contrast microscopy), were obtained from differential counts of approximately 1000, or 2000 nucleated cells/interval!organ in the spleen and bone marrow, depending upon the interval sampled.

At all sampling intervals, by means of double immunofluorescence microscopy, the proportions of MEL-14 +, or Mac-1 + NK 1.1 + lymphoid cells were assessed from counts of 200-600 NK 1.1+ lymphoid cells from the bone marrow and spleen, respectively, from normal mice and from host mice sampled at the later post-stem cell-injection intervals (14-28 days). At the earlier intervals (9-11 days), 30-100 NK 1.1+ lymphoid cells, and 50-500 NK 1.1 + lymphoid cells were assessed in the bone marrow and spleen, respec­tively, for each marker.

Origin of NK 1.1 + lymphoid cells in irradiated, stem cell-injected mice

Control experiments were performed to ascertain the origin (host vs donor) of NK 1.1 + lymphoid cells found subsequently in the stem cell-injected recipient organs. Thus, each of 4 irradiated C57Bl!6 mice (Ly 5.2) received 500 BA (Ly 5.1) stem cells. At 40 days after stem cell injection, FACS analysis of the spleen cells of each stem cell-injected recipient was performed to assess the relative proportions of NK 1.1+ Ly 5.1+ cells (indicative of donor stem cell origin), and of NK 1.1+ Ly 5.2+ cells (indicative of host origin). Other control experiments consisted of injecting 500 C57Bl!6 (Ly 5.2) stem cells into syngeneic (Ly 5.2) irradiated hosts, followed 40 days later by FACS analysis of the frequency of NK 1.1+ Ly 5.1+ lymphoid cells in the hosts.

The spleen cells from all control mice were stained after red cell lysis, with biotinylated anti-Ly 5.1 (1:150) in first stage labelling, followed by FITC-labelled anti-NK 1.1 (1:100) and avidin Texas Red (1:400) prior to FACS analysis. The background levels established for each fluorochrome alone, read from the FACS profiles, averaged 0.05 ± 0.05 % for fluorescein and 0.07 ± 0.04 % for Texas Red.

Results

Following irradiation and stem cell injection, the relative proportions of lymphoid cells (most of which are mature small T and B lymphocytes), and of non-lymphoid cells in the spleen differed most widely from their respective control (day 0) levels at day 11 (Fig. 1a). The profound reduction in lymphoid cells reflects the turnover (loss) of small lymphocytes resulting from the radiation protocol. Between days 11-17, however, both major categories of cells in the spleen proceeded to normalize rapidly, followed by little relative change in the 2 groups between days 17 and 24 (Fig. la). Even by day 28 after stem cell injection, the relative proportions of lymphoid and non-lymphoid cells in the spleen had still not completely normalized (Fig. 1a), lymphoid cells having reached only 87 % of their normal levels in that organ by this time. This deficiency in the proportions of lymphoid cells

390 . s. c. MILLER

100

80

Q) 60 (J c: Q) "t:I ·u 40 ..:

20 ,/ ,

, ,

,

a)

oL-----------------------------__ ___ 100

b) I~ ....... ---- ....

, --_ 80 _____________ ,,' ...... ------.----.--____ •

~ Q) 60 (J c: Q)

~ 40

-= 20

o+-~--_r~~--~ __ --~--~~--_r--~ o 3 6 9 12 15 18 21 24 27 30

Days after stem cell injection

Figure 1. Relative proportions of cells in the spleen (a) and bone marrow (b) of irradiated host BA mice injected with 500 syngeneic stem cells/mouse. Nucleated cells were defined morphologically using light microscopy (XIOO) into 2 main categories. Data points represent the values obtained from the pooled organs of 3-5 mice/sampling interval. Lymphoid cells (----); Other (non-lymphoid) cells (----).

undoubtedly reflects an absolute deficiency in those cells due to the reduction in the full complement of antigen-experienced, recirculating T and B lymphocytes (radiation-killed) in this short period (28 days).

On the other hand, the relative fluctuations between the lymphoid and non-lymphoid cell populations in the bone marrow were much less pro­nounced throughout the 28 days following irradiation and stem cell recon­stitution (Fig. 1 b), indicating that the various hemopoietic precursor cells, of which most of the bone marrow is composed, were similarly affected by the radiation protocol. These relative changes were, again, very probably reflective of changes in absolute number of cells in the major categories of cells, at both the early (radiation-reduced) and later (stem cell reconstitut­ing) intervals during the 28 day assay period. In the bone marrow (Fig. 1 b), as in the spleen (Fig. 1a), the most vigorous attempts at normalization by both maj or categories of cells occurred between days 11-17 after irradiation and stem cell injection. Moreover, as in the spleen, the relative levels of lymphoid cells in the bone marrow still remained sub-normal (85 % of the day 0 value) by 28 days after stem cell injection (Fig. 1 b).

Natural Killer Cells from Pluripotent Stem Cells . 391

8

C 6

UJ

'ii u

+ 4 '"":

\0:: Z

2

0 0 3 6 9 12 15 18 21 24 27 30

Days after stem cell reconstitution

Figure 2. Proportions of NK 1.1+ cells of lymphoid morphology in the spleen (--) and bone marrow (----) of irradiated BA mice at various intervals after irradiation and reconstitution with 500 syngeneic stem cells/mouse. Data points represent the values obtained from the pooled organs of 3-5 mice/sampling interval.

Cells of lymphoid morphology, bearing the surface molecule, NK 1.1, i.e., classical NK cells, reached their nadir relative to other hemopoietic cells, in both the spleen and bone marrow by 7 days following lethal irradiation and stem cell injection (Fig. 2). NK 1.1 + cells were detectably present by days 9 and 11 in the spleen and bone marrow, respectively, and thereafter increased quickly (Fig. 2). In the normal spleen, NK 1.1 + cells comprised 6.5 % of all splenic lymphoid cells (Fig. la, 2). This proportion dropped sharply by day 7 after stem cell injection but NK 1.1 + lymphoid cells soon began to re-appear so that at the time when the total lymphoid population had reached its nadir relative to the other hemopoietic cells in the spleen (day 11), NK 1.1 + cells comprised 10.9 % of all the lymphoid cells in that organ. NK 1.1 + lymphoid cells continued to increase in relative frequency in the spleen so that by days 21-24, they comprised 13-14 % of the lymphoid population. By day 28 after stem cell injection, however, NK 1.1 + cells approached their normal relative levels within the splenic lym­phoid population, accounting for 6.4 % of all lymphoid cells observed in that organ at that time.

In the normal bone marrow, NK 1.1 + lymphoid cells constituted 6.6 % of all the lymphoid cells (Fig. 1b, 2), falling within the normal range of 1-3 % of all nucleated. cells in that organ. NK 1.1+ lymphoid cells in the bone marrow, like the spleen, became negligible by day 7 after stem cell injection, regenerated to reach their maximum proportions of 10.1-13.5 % of all lymphoid cells at days 21-24 and approximated their normal relative levels by 28 days.

Thus, in both the spleen and the bone marrow, the dynamics of recon­stitution of the NK cell lineage followed similar patterns. In both organs, moreover, the rate of reconstitution of the NK 1.1 + lymphoid cells appeared higher than that of the other lymphoid cells, accounting for the

ell o

392 . s. c. MILLER

higher relative proportions of NK 1.1+ vs other types of lymphoid cells with time after stem cell injection.

The donor stem cell origin of NK 1.1 + cells re-appearing in the irradi­ated' stem cell reconstituted hosts was established by F ACS analysis. When irradiated host C57Bl/6 mice bearing the Ly 5.2 phenotype were given donor BA mouse stem cells bearing the Ly 5.1 phenotype, the resulting NK 1.1+ spleen cells which bore the Ly 5.1 phenotype at 40 days after stem cell injection averaged 7.73 ± 0.51 % of all spleen cells. By contrast, NK 1.1+ cells not bearing the donor (Ly 5.1) phenotype, indicating background, or host origin, accounted for 0.61 ± 0.28 % of the total spleen cells at 40 days. In other control experiments, where C57Bl/6 (Ly 5.2) stem cells were given to syngeneic (Ly 5.2) irradiated hosts, 99.51 % of NK 1.1+ cells were negative for the Ly 5.1 phenotype, as determined by FACS analysis.

As NK 1.1 + lymphoid cells progressively increased in relative frequency in the spleen following stem cell injection, they also progressively acquired both the homing receptor recognized by mAb MEL-14 (Fig. 3a) and the integrin, Mac-1 (Fig. 3b). In the spleen, NK 1.1 + lymphoid cells, as detected in double immunofluorescence microscopy, acquired detectable MEL-14 homing receptor and Mac-1 very early in the lineage reconstitution process (Fig. 3a, b). In the spleen, progressive acquisition of the homing receptor and the integrin followed similar patterns with time after stem cell injection (Fig. 3a, b).

On the other hand, in the bone marrow, the increase in NK 1.1 + lymphoid cell frequency after stem cell injection was not mimicked by a progressive increase in the proportions of NK 1.1 + lymphoid cells bearing either the homing receptor or the integrin molecules (Fig. 3a, b). U ndoub-

40 a) b)

0 ap,"" 0 apl"" • bone marrow • bon.mlrrow r-30

r-

~ 20 ~ ·u ~

10

o I'l:l ~. ~- ~. ~. ~ .. ~. n I • .. • I ~ 7 9 11 14 17 21 24 28 N 7 9 11 14 17 21 24 28 N

Days after stem cell injection Days after stem cell Injection

Figure 3. Proportions of NK 1.1 + cells of lymphoid morphology bearing the homing receptor MEL-14 (a) or the integrin, Mac-l (b), at various intervals after irradiation and reconstitution of BA mice with 500 syngeneic stem cells/mouse. Data points represent the values obtained from the pooled organs of 3-5 mice/sampling interval.

Natural Killer Cells from Pluripotent Stem Cells . 393

tedly, the increased proportions of NK cells reflect absolute increases in NK 1.1 + lymphoid cell numbers in both organs in the non-steady state, post-stem cell reconstitution period of exponential growth.

Discussion

The cell characterized by Thy-11oLin-Sca-1 + is now well established as pluripotent, capable of reconstituting the hemopoietic cell lineages of the bone marrow and spleen of lethally irradiated animals (25-27). The cell identified by these surface molecules can, moreover, restore both the Band T cell lineages, with their respective functions, leading to full repopulation of the secondary lymphatic organs. To date, however, no attempt has been made to assess the ability of the Thy-11oLin-Sca-1 + stem cell to reconstitute the lineage responsible for spontaneous cell-mediated immune responses, i.e., that of NK cells.

The present analysis of NK cell population dynamics during stem cell reconstitution has revealed the presence of NK 1.1 + lymphoid cells at intervals earlier than would be expected had such cells needed first to be thymus-processed, or were T cell lineage-derived. Nevertheless, the possi­bility of a common NK/T progenitor (restricted stem cell) cannot be excluded.

Increases with time in the relative frequency of NK 1.1 + lymphoid cells, as with other cell types, undoubtedly reflected even greater increases in absolute numbers, since the total cellularity of the spleen and bone marrow was increasing during the post-irradiation, stem cell reconstitution period of intense hemopoietic cell regeneration.

That the NK 1.1 + lymphoid cells detected after stem cell injection were simply not residual, radiation-resistant, host-derived mature NK cells is supported by our previous observations (36) that functional NK cells are continuously renewed from the spleen and the bone marrow exponentially with half renewal times (T 112) of 24 hand 9 h, respectively. Rapidity of turnover, short life span and an absence of recirculation capacity or memory are demonstrated hallmarks of cells of the NK lineage (36-40). Nevertheless, the possibility that host-derived, radioresistant, late pre-NK cells bearing the NK 1.1 surface marker may have been recorded among the NK 1.1 + lymphoid cells cannot be excluded. Recent evidence provided by F ACS analysis indicates that at the earliest post-stem cell-injection inter­vals, host-derived cells bearing NK 1.1 can be found included among the regenerating, donor-derived NK 1.1 + cells (Dr. LEO AGUILA, Stanford University, unpublished observations). These disappear as the donor stem cells reconstitute at subsequent intervals.

Most NK cells produced in the bone marrow appear to leave that organ prior to acquiring the NK 1.1 marker, the latter developing en route to and/ or within the spleen, the normal destiny of newly-formed, bone marrow­derived NK cells. Moreover, we have recently shown that another marker

394 . S. c. MILLER

of NK cell maturity, ASGM-1, also develops on NK lineage cells predo­minantly after they leave their bone marrow birth site (41, 42). NK cell generation may become an autonomous function of the spleen, in conjunc­tion with the bone marrow, under the present conditions which stimulate total hemopoietic cell population expansion. The capacity of the spleen for autonomous NK cell production has already been indicated (43).

The appearance of the homing receptor recognized by mAb MEL-14, and the integrin, Mac-Ion NK 1.1 + lymphoid cells in unprimed, unstimu­lated, stem cell-injected mice suggests that newly-produced NK 1.1+ lymphoid cells may be constitutively transendothelial cell migration-ready, in keeping with their state of functional readiness. That both the surface molecule, gp90MEL-1\ and Mac-l are virtually co-incident on the regenerat­ing NK 1.1 + lymphoid cells in the spleen suggests that the same cells may bear both molecules simultaneously, although it is, nevertheless, mathemat­ically possible for different NK 1.1 + lymphoid cells to bear either one but not both molecules. The presence or absence of gp90MEL-14 on NK cells in the spleen is, however, inconsequential to the functional activity of these cells (13). The observation that this molecule remains on NK cells during target cell lysis (13), suggests that it, and possibly the integrins, may never be down-regulated during the short, functional life span of these cells (36).

In the bone marrow, the fact that the regenerating NK 1.1+ lymphoid cells do not progressively develop gp90MEL-14 or Mac-l suggests that these molecules remain either unexpressed, or masked, at least until the cells exit the bone marrow. The progressive acquisition of these molecules may take place en route to and/or within the splenic vasculature. In contrast to the present observations, derived by means of immunofluorescence micros­copy, FACS analysis has recently shown, at selected post-stem cell-injec­tion intervals, little or no difference in the proportions of NK 1.1 + cells which bear Mac-l between the spleen and bone marrow (Dr. L. Aguila, unpublished observations). However, this discrepancy may be explained by the presence in the bone marrow of abundant granuloid and monocytic cells, many of which bear Mac-I, and, as considerable evidence now indicates, can display the NK 1.1 surface marker as well (34, 44, 45). These lineages cannot be morphologically distinguished by FACS from marker­bearing lymphoid (true NK) cells, suggesting the possibility of contamina­tion of the latter by granuloid/monocytic cells.

Unlike T and B lymphoid cells, it is not known if, or where, NK 1.1+ lymphoid cells make use of their homing receptor, gp90MEL-14

• The lymph nodes (whose high endothelial venules are well established sites of the MEL-14 ligand), are organs wherein NK cells normally exist only in very low numbers. NK cell entry into the spleen, which is the organ of their greatest concentration under normal conditions, mayor may not be medi­ated by the MEL-14 receptor-ligand interaction. The specialized high endothelial venules (HEV), by which gp90MEL-14+, antigen-experienced T and B lymphoid cells enter the lymph node parenchyma, have not been observed in the spleen. In the latter organ, NK 1.1 +, gp90MEL-14+ lymphoid

Natural Killer Cells from Pluripotent Stem Cells . 395

cells may bind to morphologically distinct regions of cuboidal endothelium which may bear a complementary ligand. Alternatively, the appropriate ligand may be present only on certain scattered, yet unremarkable, endothelial cells. Given the paucity of NK cells, even in the spleens of normal, young animals, the latter mechanism exclusively, may readily accommodate intra-parenchymal entry of all migrating, blood-borne NK cells. Flat-walled venules functioning like HEV, which can mediate lym­phocyte migration in the lamina propria have already been described (46). Thus, morphology may not be directly linked to function.

The homing receptor identified by mAb MEL-14 does not, however, provide strong cell binding and subsequent transendothelial migration of immune cells. The latter events are mediated by the integrins, and in accordance with the present study, several (CDlla-c/CDI8) have already been identified on NK cells of both mice and man (15-18, 47). When and where NK cells employ these and possibly other molecules which may direct their trafficking and intra-organ migration are currently under inves­tigation. Whether or not the intensity of expression of homing receptors and integrins is modified in vivo on NK cells in the presence of carcinogens, tumors, or other stimuli which may alter their normal trafficking and migration patterns, remain topics of future study.

Acknowledgments

The author is grateful to Dr. N. UCHIDA for carrying out the stem cell separation procedures, to Mrs. LIBUSE JERABEK for excellent technical assistance, and to Dr. L. AGUILA for helpful discussions. This work was supported by grants from the Howard Hughes Medical Institute and the National Institutes of Health to Dr. 1. L. WEISSMAN, Stanford University, in whose laboratory this study was completed.

References

1. KAPLAN, J. and D. M. CALLEWAERT. 1978. Expression of human T lymphocyte antigen by natural killer cells. J. Nat!' Cancer Inst. 60: 961.

2. BROOKS, C. G. 1983. Reversible induction of natural killer cell activity in cloned murine cytotoxic T lymphocytes. Nature 305: 155.

3. BORN, W., A. BEN-NUN, U. BAMBERGER, M. NAKAYAMA, V. SPETH, D. SUN, and M. THORNTON. 1983. Killer-cell lines derived from mouse thymus, resembling large granular lymphocytes and expressing natural killer-like cytotoxicity. Immunobio!. 165: 63.

4. TANG,]., D. c. DELONG, L. D. BUTLER, P. MARDER, and E. W. ADES. 1986. Murine thymocytes mediate a natural killer-like activity against Herpes virus-infected target cells but not YAC-1 target cells. Scand. J. Immun. 24: 115.

5. LAUZON, R., K. A. SIMINOVITCH, and]. C. RODER. 1986. T cell receptor gene rearrangements in cells against natural killer activity in the mouse. Immuno!. Res. 5: 191.

6. CARE, A., P. G. PELICCI, E. MECCIA, M. FAGIOLI, U. TESTA, E. CICCONE, L. MORETTA, and C. PESCHLE. 1990. Natural killer cells carry the germline configuration

396 . s. C. MILLER

of the T cell receptor delta chain gene and heterogeneously express six distinct delta transcripts. Eur. J. Immunol. 20: 939.

7. BREARD, J. and S. F. SCHLOSSMAN. 1982. Presence of a myeloid differentiation antigen on human peripheral blood NK cells. In Natural Killer Cells, vol. 4. B. SERROU, C. ROSENFELD, and R. B. HERBERMAN, eds. Elsevier Biomedical Press, Amsterdam, p. 69.

8. GALLI, S. J., A. M. DVORAK, T. ISHIZAKA, H. D. NABEL, H. SIMONIAN, H. CANTOR, and H. DVORAK. 1982. A cloned cell with NK function resembles basophils by ultrastructure and expresses IgE receptors. Nature (London) 298: 288.

9. CLAESSON, M. H., L. OLSSON, L. MARTINSEN and P. BRIX-POULSEN. 1982. Bone marrow derived diffuse colonies: Their cytotoxic potential, morphology and anti­genic phenotype. Exp. Hematol. 10: 708.

10. BACCARINI, M., L. HAO, T. DECKER, and M. L. LOHMANN-MATTHES. 1988. Mac­rophage precursors as natural killer cells against tumor cells and microorganisms. Nat. Immun. Cell Growth Regul. 7: 316.

11. BYKOWSKY, M. J. and o. STUTMAN. 1986. The cells responsible for murine natural cytotoxic (NC) activity: A multi-lineage system. J. Immunol. 137: 1120.

12. REYNOLDS, C. W. andJ. R. ORTALDO. 1987. Natural killer activity: The definition of a function rather than a cell type. Immunology Today 8: 172.

13. BALLAS, Z. K. and W. RASMUSSEN. 1990. NK 1.1+ thymocytes. Adult murine CD4-, CD8- thymocytes contain an NK 1.1+, CD3+, CDShi, CD44hi TCR-VB8+ subset. J. Immunol. 145: 1039.

14. NISHIMURA, T. and T. ITOH. 1988. Higher level expression of lymphocyte function­associated antigen-1 (LFA-1) on in vivo natural killer cells. Eur. J. Immunol. 18: 2077.

15. TIMONEN, T., M. PATARROYO, and C. G. GAHMBERG. 1988. CD11a--c/CD18 and GP84(LB-2) adhesion molecules on human large granular lymphocytes and their participation in natural killing. J. Immunol. 141: 1041.

16. RAMOS, O. F., M. PATARROYO, E. YEFENOF, and E. KLEIN. 1989. Requirement of leukocytic cell adhesion molecules (CD11a-ciCD18) in the enhanced NK lysis of iC3b-opsonized targets. J. Immunol. 142: 4100.

17. WERFEL, T., W. WITTER, and o. GOTZE. 1991. CD11b and CD11c antigens are rapidly increased on human natural killer cells upon activation. J. Immunol. 147: 2423.

18. ALLAVENA, P., C. PAGANIN, 1. MARTIN-PADURA, G. PERI, M. GABOLl, E. DE]ANA, P. C. MARCHISIO, and A. MANTOVANi. 1991. Molecules and structures involved in the adhesion of natural killer cells to vascular endothelium. J. Exp. Med. 173: 439.

19. KASAl, M., M. IWAMORI, Y. NAGAI, K. OKUMURA, and T. TADA. 1980. A glycolipid on the surface of mouse natural killer cells. Eur. J. Immunol. 10: 175.

20. YOUNG, W. W., S. HAKOMORI, J. M. DURDIK, and C. S. HENNEY. 1980. Identifica­tion of ganglio-N-tetraosylceramide as a new surface marker for murine natural killer (NK) cells. J. Immunol. 124: 199.

21. BECK, B. M., S. GILLIS, and C. S. HENNEY. 1982. Display of the neutral glycolipid ganglio-N-tetraosylceramide (Asialo GM1) on cells of the natural killer and T lineages. Transplantation 33: 118.

22. STOUT, R. D., G. A. SCHWARTING, andJ. SUTTLES. 1987. Evidence that expression of Asialo-Gm1 may be associated with cell activation. J. Immunol. 139: 2123.

23. HARGROVE, M. F. and C. TING. 1988. Asialo GM1 as an accessory molecule activating the function and reactivity of cytotoxic T lymphocytes. Cell. Immunol. 112: 123.

24. YUASA, H., D. A. SCHEINBERG, and A. N. HOUGHTON. 1990. Gangliosides of T lymphocytes: Evidence for a role in T-cell activation. Tissue Antigens 36: 47.

Natural Killer Cells from Pluripotent Stem Cells . 397

25. SPANG RUDE, G. J., S. HEIMFELD, and 1. L. WEISSMAN. 1988. Purification and characterization of mouse hematopoietic stem cells. Science 241: 58.

26. UCHIDA, N. and 1. L. WEISSMAN. 1992. Searching for hematopoietic stem cells: Evidence that Thy-1.11°Lin-Sca-1+ cells are the only stem cells in C57BlIKa-Thy-1.1 bone marrow.]. Exp. Med. 175: 175.

27. IKUTA, K., T. KINA, 1. MACNEIL, N. UCHIDA, B. PEAULT, Y. H. CHIEN, and 1. WEISSMAN. 1990. A development switch in thymic lymphocyte maturation potential occurs at the level of hematopoietic stem cells. Cell 62: 863.

28. PARK, D. R. and L. A. HERZENBERG. 1984. Fluorescence-activated cell sorting: theory, experimental optimization and applications in lymphoid cell biology. Methods Enzymol. 108: 197.

29. MILLER, S. C. and D. G. OSMOND. 1974. Quantitative changes with age in bone marrow cell populations of C3H mice. Exp. Hemat. 2: 227.

30. MILLER, S. C. and D. G. OSMOND. 1976. Quantitative studies of lymphocytes and other cell populations in the bone marrow of neonatally thymectomized C3H mice. Anat. Rec. 184: 325.

31. MILLER, S. C. and D. G. OSMOND. 1975. Lymphocyte populations in mouse bone marrow: Quantitative, kinetic studies in young, pubertal and adult C3H mice. Cell Tissue Kinet. 8: 97.

32. MILLER, S. c., M. KAISERMAN, and D. G. OSMOND. 1978. Small lymphocyte production and lymphoid cell proliferation in mouse bone marrow. Experientia 34: 129.

33. Koo, G. c., J. R. PEPPARD, and W. H. MARK. 1984. Natural killer cells generated from bone marrow cultures. J. Immunol. 132: 2300.

34. Koo, G. c., C. L. MANYAK, J. DASCH,]. ELLINGSWORTH, and L. D. SHULTZ. 1991. Suppressive effects of monocytic cells and transforming growth factor-B on natural killer cell differentiation in autoimmune viable moth eaten mutant. J. Immunol. 147: 1194.

35. WOOD, G. S. and R. WARNKE. 1981. Suppression of endogenous avidin-binding activity in tissues and its relevance to biotin-avidin detection systems. J. Histochem. Cytochem. 29: 1196.

36. MILLER, S. C. 1982. Production and renewal of murine natural killer cells in the spleen and bone marrow. J. Immunol. 129: 2282.

37. GREENBERG, A. H. and M. GREEN. 1976. Non-adaptive rejection of small tumour inocula as a model of immune surveillance. Nature 264: 356.

38. SEAMAN, W. E., T. D. GINDHARDT,]. S. GREENSPAN, and M. A. BLACKMAN. 1977. Natural killer cells, bone and bone marrow: studies in estrogen-treated mice and in congenitally osteopetrotic (mi/mi) mice. J. Immunol. 122: 2541.

39. STUTMAN, o. and E. C. LATTIME. 1981. Natural cell mediated cytotoxicity against tumors in mice: a heterogenous system. Trans. Proc. 13: 752.

40. FRESA, K. L., R. KORNGOLD, and D. M. MURASKO. 1985. Induction of natural killer cell activity of thoracic duct lymphocytes by polyinosinic-polycytidylic acid (Poly I:C) or interferon. Immunology 91: 336.

41. CHRISTOPHER, F. L., 1. DUSSAULT, and S. C. MILLER. 1991. Population dynamics of natural killer cells in the spleen and bone marrow of normal and leukemic mice during in vivo exposure to Interleukin-2. Immunobiol. 184: 37.

42. MILLER, S. c., F. L. CHRISTOPHER, and 1. DUSSAULT. 1992. Population dynamics of natural killer cells and T lymphocytes in murine spleen and bone marrow during the development of erythroleukemia: The effect of indomethacin. Nat. Immun. 11: 78.

43. MILLER, S. C. and A. C. SHATZ. 1991. Relationship between large and small tumor­binding cells in the spleen and bone marrow. Nat. Immun. Cell Growth Regul. 10: 320.

398 . s. c. MILLER

44. POLLACK, S. B. and c. ROSSE. 1987. The primary role of murine bone marrow in the production of natural killer cells: A cytokinetic study. Jour. Immuno!. 139: 2149.

45. LI, H., R. SCHWINZER, M. BACCARINI, and M. L. LOHMANN-MATTHES. 1989. Cooperative effects of colony-stimulating factor 1 and recombinant interleukin 2 on proliferation and induction of cytotoxicity of macrophage precursors generated from mouse bone marrow cell cultures. J. Exp. Med. 169: 973.

46. JEURISSEN, S. H., A. M. DUI]VESTI]N, Y. SONTAG, and G. KRAAL. 1987. Lymphocyte migration into lamina propria of the gut is mediated by specialized HEV-like blood vessels. Immunology 62: 273.

47. MIGLIORATE, G., R. MORACA, 1. NICOLETTI, and C. RICCARDI. 1992. IL-2-depend­ent generation of natural killer cells from bone marrow: Role of MAC-1-, NK 1.1-precursors. Cell Immuno!. 141: 323.

Dr. SANDRA C. MILLER, Department of Anatomy and Cell Biology, Room 2128, McGill University, 3640 University Street, Montreal, Quebec, Canada H3A 2B2