Role of Natural Killer Cells in Resistance against Friend Retrovirus-Induced Leukemia

Abstract

We have previously shown that immunization with a synthetic peptide that contains a single CD4⁺ T-cell epitope protects mice against immunosuppressive Friend retrovirus infection. Cells producing infectious Friend virus were rapidly eliminated from the spleens of mice that had been immunized with the single-epitope peptide. However, actual effector mechanisms induced through T-helper-cell responses after Friend virus inoculation were unknown. When cytotoxic effector cells detected in the early phase of Friend retrovirus infection were separated based on their expression of cell surface markers, those lacking CD4 and CD8 but expressing natural killer cell markers were found to constitute the majority of effector cells that lysed Friend virus-induced leukemia cells. Depletion of natural killer cells by injecting anti-asialo-ganglio-N-tetraosylceramide antibody did not affect the number of CD4⁺ or CD8⁺ T cells in the spleen, virus antigen-specific proliferative responses of CD4⁺ T cells, or cytotoxic activity against Friend virus-induced leukemia cells exerted by CD8⁺ effector cells. However, the same treatment markedly reduced the killing activity of CD4⁻ CD8⁻ effector cells and completely abolished the effect of peptide immunization. Although the above enhancement of natural killer cell activity in the early stage of Friend virus infection was also observed in mice given no peptide, these results have demonstrated the importance and requirement of natural killer cells in vaccine-induced resistance against the retroviral infection.

Understanding the types of immune responses associated with the control of viral infection is pivotal for the development of effective antiviral vaccine strategies. Several lines of evidence indicate that priming of virus-specific CD4⁺ T cells might result in protective immunity against immunosuppressive retrovirus infection in humans (6, 31, 35). However, the observed relationship between priming of CD4⁺ T cells and protection against retrovirus infections remains circumstantial. Using a mouse model of immunosuppressive Friend retrovirus infection, we previously showed that immunization with a synthetic peptide containing a single CD4⁺ T-cell epitope resulted in rapid elimination of virus-producing cells from the host and partial protection against the development of fatal leukemia (22). In these peptide-based experiments animals can be primed with a CD4⁺ T-cell epitope alone, without other components of the immune system being stimulated before virus inoculation, allowing us to critically analyze the role and mechanisms of CD4⁺ T-cell help in inducing protective immune effector functions in retrovirus infections.

Friend mouse retrovirus complex (FV) is composed of replication-competent Friend murine leukemia helper virus (F-MuLV) and defective spleen focus-forming virus (SFFV), the latter of which induces rapid growth and terminal differentiation of infected erythroid progenitor cells (2, 16). FV is known to induce fatal erythroleukemia associated with severe immunosuppression when injected into immunocompetent adult mice of susceptible strains (2, 27). Mice with a BALB/c background are especially susceptible to FV-induced disease because they lack immune and nonimmune mechanisms that render some other strains of mice resistant against FV replication and disease progression (5). Epitopes recognized by CD4⁺ helper T cells and CD8⁺ cytotoxic T lymphocytes (CTL) have been identified in the products of env and gag genes of F-MuLV (1, 15, 20, 32, 36), and a yet-unidentified immunoprotective CD4⁺ T-cell epitope has been mapped in the MA (p15) portion of the gag gene product (23). Among them, two env-derived CD4⁺ T-cell epitopes, the A^b-restricted N-terminal peptide DEPLTSLTPRCNTAWNRLKL (epitope fn) and the C-terminal peptide HPPSYVYSQFEKSYRHKR (epitope i) restricted by the hybrid E^b/d molecule, are effective in inducing protective immunity against FV challenge when injected into (B10.A × A.BY)F₁ mice (22). Production and class switching of virus-neutralizing antibodies after challenge inoculation with live pathogenic Friend retrovirus were accelerated in mice immunized with the synthetic peptides in comparison with unimmunized control mice, and the number of virus-producing cells in the spleen was drastically reduced between 7 and 11 days after the virus inoculation, along with the expansion of CD4⁺ T cells (22). To identify effector mechanisms activated after FV infection in mice immunized with the CD4⁺ T-cell vaccine, spleen cells were separated according to their expression of cell surface markers, and their cytotoxic activities against FV-induced leukemia cells were examined. Both CD8⁺ and CD4⁺ T cells constituted a part of effector cells that lysed FV-induced leukemia cells in vitro. However, to our surprise, cells expressing natural killer (NK) cell markers constituted the majority of cytotoxic effector cells in FV-infected mice, and they were required for vaccine-induced protection against FV infection.

MATERIALS AND METHODS

Mice and virus.

(BALB/c × C57BL/6)F₁ (CB6F₁) mice were purchased from Japan SLC, Inc., Hamamatsu, Japan. Male mice aged 8 to 11 weeks at the time of immunization were used throughout the present study. All the animal experiments were approved by and performed under the guidelines of Kinki University. A stock of B-tropic FV was originally given from Bruce Chesebro, Laboratory of Persistent Viral Diseases, National Institute of Allergy and Infectious Diseases, Hamilton, Mont. The stock used in the present study was prepared by inoculating nine female BALB/cAJcl mice purchased from Japan SLC, Inc., with 7,500 spleen focus-forming units (SFFU) of B-tropic FV. Nine days later the infected mice were killed, 20% spleen homogenate was prepared as described previously (22), and 1-ml aliquots were stored frozen at −80°C until use. SFFV and F-MuLV titers of the FV stock were determined as described previously (22, 29). The FV stock used in the present study had an SFFV titer of 9.2 × 10⁴ SFFU per ml and an F-MuLV titer of 4.3 × 10⁵ focus-forming units per ml. For inoculation into CB6F₁ mice, a dilution of the virus stock prepared with phosphate-buffed balanced salt solution (PBBS) containing 1% fetal bovine serum (FBS) was injected into the tail vein. Infected mice were observed at least twice a day, and the number of surviving mice was counted. Mice found dead were dissected, and their spleen weights were measured. Recombinant vaccinia viruses expressing either the F-MuLV env or gag gene have been described elsewhere (9, 24, 26).

Peptide synthesis and immunization.

The peptides used in the present study were synthesized and purified by Fmoc chemistry as described previously (15, 20, 36, 39, 40). Peptides representing the F-MuLV env-encoded T-helper cell epitopes, fn and i, have been described in detail (15, 36). Control peptides used include the following: fa₁₃RT (AAAAAARAATAAA), which contains the major histocompatibility complex (MHC) anchor residues of fn (36); i_e (HSPSYVYHQFERRAKYKR), which represents an endogenous retroviral env-derived sequence corresponding to F-MuLV peptide i (40); MHC class II A^b-binding pigeon cytochrome c-related peptides 50V (AEGFSYTVANKNKGIT) and 50A (AEGFSYTAANKNKGIT) (13, 14, 28); and H-2D^b-restricted influenza virus nucleoprotein peptide NP_366–374 (ASNENMETM) (38). The molecular weight of each peptide was confirmed by quadrupole mass spectrometry as described previously (15, 36, 39, 40). For immunization, peptide i was dissolved in PBBS and emulsified with an equal volume of complete Freund's adjuvant (CFA). Mice were injected intradermally with a total of 100 μl of the emulsion given as multiple split doses into the abdominal wall. Control mice were given an emulsion of PBBS and CFA that did not contain any peptide.

Cells and cytotoxicity assays.

CD4⁺ T-cell clones SB14-31 and F5-5 specific for the A^b-restricted N-terminal and E^b/d-restricted C-terminal epitopes of the F-MuLV env gene product, respectively, were maintained as described previously (15). Three other T-cell clones—FP3-10, FP8-7, and FP10-16—were established from CB6F₁ mice immunized with peptide i as described previously (40). Target cells used were as follows: an FV-induced leukemia cell line, FBL-3, established from a C56BL/6 mouse (H-2^b); another line of FV-induced leukemia cells, Y57-2C (7), established from a (C57BL/10 × A.BY)F₁ mouse (H-2^b); a chemically induced T-cell lymphoma line, EL-4, established from a C57BL/6 mouse; a Moloney murine leukemia virus (Mo-MuLV)-induced T-cell lymphoma line MBL-2 (H-2^b); an H-2^b/d hybridoma cell line, LB 27.4, exhibiting both class II A- and E-restricted antigen-presenting activities (17); a B-cell lymphoma line, A20, established from a BALB/c mouse (H-2^d) (18); and an A/Sn mouse-derived Mo-MuLV-induced lymphoma line, YAC-1, which is widely used as an NK target. Y57-2C cells were originally provided by Bruce Chesebro; FBL-3, MBL-2, EL-4, and YAC-1 cells were kindly provided by Kagemasa Kuribayashi, Mie University School of Medicine; and LB 27.4 and A20 cells were purchased from the American Type Culture Collection, Manassas, Va. EL-4 and MBL-2 cell lines have recently been shown to have a common origin (41).

Cytotoxicity assays were performed by using ⁵¹Cr-labeled target cells as described elsewhere (20, 30, 39). Target cells were incubated with 3.7 MBq of ⁵¹Cr/10⁶ cells for 1 h and washed three times. When necessary, cells were further incubated for 30 min with a synthetic peptide and washed. Labeled target cells (5 × 10³) were placed in each well of V-bottomed 96-well plates and mixed with effector cells at various effector-to-target (E:T) ratios. Four or twelve hours later, plates were centrifuged and supernatant was collected from each well for measurement of released radioactivity with a gamma counter. Killing activity was calculated by a standard equation: percent specific killing = [(⁵¹Cr release in a test well − spontaneous release)/(maximum release − spontaneous release)] × 100, where maximum release is the radioactivity obtained by adding 1% Triton X-100 into wells of labeled target cells, and spontaneous release is the radioactivity in the supernatant of target cells cultured without effector cells. Data are expressed as means ± standard errors of the means (SEM) of triplicate samples. Inhibition of cytotoxic activity by anti-CD4 or anti-CD8 monoclonal antibody (MAb) was performed by using culture supernatants of relevant hybridoma cells (8, 33).

Flow cytometry.

Flow cytometric analyses of cell surface markers were performed as described elsewhere (12, 19, 34, 40). Spleen tissue was dissociated in PBBS containing 2% FBS, and single-cell suspension was prepared by passing the dissociated tissue through a nylon mesh. Cells were incubated with a combination of the MAbs listed below, washed three times with PBBS containing 2% FBS and 0.05% NaN₃, and stained with 20 μg of 7-aminoactinomycin D/ml, which was used to exclude dead cells (34). Antibodies and their final concentrations used in the present study were as follows: fluorescein isothiocyanate (FITC)-conjugated anti-mouse CD4 (rat immunoglobulin G2b [IgG2b]; Seikagaku Corporation, Tokyo, Japan) at 0.5 μg/10⁶ cells, phycoerythrin (R-PE)-conjugated anti-mouse CD8 (rat IgG2a; Caltag Laboratories, Burlingame, Calif.) at 1 μg/10⁶ cells, FITC-conjugated anti-mouse CD69 (hamster IgG; PharMingen, San Diego, Calif.) at 1 μg/10⁶ cells, R-PE-conjugated anti-mouse B220 (rat IgG2a; Coulter Immunology, Hialeah, Fla.) at 0.5 μg/10⁶ cells, FITC-conjugated anti-NK1.1 (mouse IgG2a; PharMingen) at 2 μg/10⁶ cells, biotin-conjugated anti-mouse Pan-NK (DX5, rat IgM; PharMingen) at 1 μg/10⁶ cells, and allophycocyanin-conjugated anti-mouse TER-119 (PharMingen) at 0.2 μg/10⁶ cells. TER-119 reacts with a molecule associated with glycophorin A and marks the late erythroblasts and mature erythrocytes (19). To detect cells infected with F-MuLV, MAb 720 (29) reactive with F-MuLV gp70 but not with any other mouse retrovirus was purified and conjugated with biotin as described previously (12, 22, 29). R-PE-conjugated streptavidin (PharMingen) was used for staining with the biotin-conjugated antibodies. All staining reactions were performed in the presence of 0.25 μg of anti-mouse CD16/CD32 (PharMingen)/10⁶ cells as described previously (12), to prevent the binding of MAb to Fc receptor-expressing cells. Cells were also incubated with isotype-matched control antibodies purchased from the same suppliers, and staining patterns obtained with these negative control antibodies were used to draw demarcation lines that separate positively stained cells from those not stained. Multicolor flow cytometric analyses were performed with a Becton Dickinson FACScalibur and CellQuest software (Becton Dickinson Immunocytometry Systems, San Jose, Calif.). Mature erythrocytes and dead cells were excluded from the analyses by setting a polygonal gate in the dot plots showing intensities of forward scatter and the fluorescence for 7-aminoactinomycin D.

Purification of T-cell subsets and NK cells.

Purification of T-cell subsets and NK cells from the spleens of FV-infected mice was performed by using Ab-conjugated magnetic microbeads and a magnetic cell sorter I (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). Spleen cells were first treated with Tris-buffered ammonium chloride solution to lyse erythrocytes and incubated with anti-B220 MAb-conjugated magnetic beads to remove B cells by passing them through a negatively selecting CS column. To purify CD4⁺ T cells, B220⁻ cells were incubated with anti-CD8 MAb-conjugated magnetic beads, passed through a CS column to remove CD8⁺ cells, and then incubated with anti-CD4 MAb-conjugated microbeads to positively select CD4⁺ cells by passing them through a VS column. Cells not retained in the column were collected as a CD4⁻ CD8⁻ (double-negative) population. Multicolor flow-cytometric analyses revealed that the resultant CD4⁺ T-cell preparation was 99.2% positive for CD4. CD8⁺ cells were similarly purified from B220⁻ cells by positively selecting CD8⁺ cells. This preparation was 97 to 98% CD8⁺ in repeated experiments. To purify cells expressing the NK marker DX5, the B220⁻ population of spleen cells was incubated with anti-mouse Pan-NK (DX5) MAb-conjugated microbeads, and cells expressing this marker were positively selected as a DX5⁺ population.

Depletion of NK cells in vivo.

Anti-asialo-ganglio-N-tetraosylceramide (anti-asialo-GM₁) rabbit Ab (11) and control normal rabbit serum were purchased from Wako Pure Chemicals (Osaka, Japan). CB6F₁ mice immunized once with peptide i 25 days before FV inoculation were injected intravenously with 400, 160, or 60 μg of anti-asialo-GM₁ Ab/dose at 1 day prior to FV inoculation and 2, 5, 8, and 11 days after the virus infection. NK cell activity was tested on days 7, 9, and 11 postinoculation. Since administration of any of the above three amounts of anti-asialo-GM₁ Ab resulted in undetectable NK cell activity, 60 μg of anti-asialo-GM₁ Ab/dose was adopted for analyzing the effect of NK cell depletion on protective immunity induced with the peptide vaccine.

Assays for proliferative responses of T cells.

CD4⁺ T cells were purified from FV-infected CB6F₁ mice as described above. CD4⁺ T cells (2 × 10⁵) were incubated with 5 × 10⁵ syngeneic spleen cells that had been irradiated with 45 Gy of X-irradiation and various amounts of a peptide. After 3 days of culture in RPMI 1640 medium supplemented with 10% FBS, cells in each well of 96-well tissue culture plates were pulsed with 18.5 kBq of [³H]thymidine (DuPont NEN, Boston, Mass.) for the last 8 h and harvested onto a glass fiber filter as described previously (15, 40). Incorporated radioactivity was measured with a multiplate scintillation counter (TopCount, Packard Instruments Co., Meriden, Conn.). Antigen-specific proliferation was expressed as the change in counts per minute, calculated by subtracting the average incorporation of [³H]thymidine in wells containing CD4⁺ T cells and irradiated spleen cells but no peptide from the average incorporation of [³H]thymidine in wells containing responder CD4⁺ T cells, irradiated spleen cells as antigen-presenting cells, and a peptide. Data are expressed as means ± SEM of triplicate samples.

RESULTS

Protection of highly susceptible CB6F₁ mice against Friend retrovirus infection with a single-epitope CD4⁺ T-cell vaccine.

CB6F₁ mice are highly susceptible to FV-induced leukemia, and most mice of this strain died by 60 days after inoculation of 5, 15, or 50 SFFU of FV (A. Niwa, N. Iwanami, H. Uenishi, N. Tabata, H. Yamagishi, and M. Miyazawa, submitted for publication). When control CB6F₁ mice that had been given CFA emulsion without a peptide were infected with 150 SFFU of FV, >95% of them died by postinoculation day (PID) 60 (Fig. 1a). On the other hand, >90% of CB6F₁ mice immunized only once with 10 μg (5 nmol) of peptide i survived longer than 60 days after infection with the same dose of FV. When infected mice were killed at PID 44 in a separate experiment, none of the mice immunized with peptide i showed splenomegaly, while the control mice that had died by that time or were killed at PID 44 had an enlarged spleen weighing >2.5 g (Fig. 1b). Thus, it was clear that immunization with the single-epitope peptide i prevented the development of FV-induced leukemia and protected highly susceptible CB6F₁ mice from leukemic death.

FIG. 1.

Development of fatal leukemia in CB6F₁ mice after inoculation of FV and its prevention by immunization with the single-epitope peptide i. (a) CB6F₁ mice (12 per group) were either immunized with 10 μg of peptide i each (○) or given a CFA emulsion without a peptide (●). Four weeks later they were inoculated intravenously with 150 SFFU of FV and monitored for the development of leukemia. (b) Comparison of spleen weights of CB6F₁ mice at 44 days after FV inoculation. Mice were either immunized only once with 10 μg of peptide i/mouse (○) or given CFA alone (●) and were inoculated with FV 4 weeks later. Control mice found dead before PID 44 were dissected within 10 h of their death.

Kinetics of the activation of immune mechanisms in FV-infected mice.

CB6F₁ mice were either immunized with 10 μg of peptide i/mouse or given CFA emulsion without a peptide and infected with FV 4 weeks later. Groups of mice were killed at regular intervals, and the numbers of cells expressing the late erythroid cell marker TER-119, F-MuLV gp70, CD4, CD8, and the early activation marker CD69 were measured by multicolor flow cytometry. TER-119 marks late erythroblasts and mature red blood cells but not erythroid burst-forming units (BFU-E) or erythroid CFU (CFU-E) (19). Erythroblasts detectable with TER-119 increased slowly between 3 and 7 days after FV inoculation, and these cells showed an abrupt increase in number at PID 8 in the control mice (Fig. 2a). A similar pattern of increase in the number of F-MuLV-infected cells was also observed in the control mice (Fig. 2b). Dual-color analyses revealed that >80% of TER-119⁺ cells were F-MuLV gp70⁺ after PID 5 in the control mice given CFA without a peptide. Since the SFFV component of FV induces proliferation and terminal differentiation of late BFU-E and CFU-E by stimulating them through erythropoietin receptor, and since mouse bone marrow BFU-E stimulated with erythropoietin usually produce erythroid cell bursts within 5 to 8 days (16), the observed rapid expansion of TER-119⁺ cells in the FV-infected control mice quite possibly reflects the burst formation from initially infected BFU-E. In contrast to the rapid expansion of FV-infected erythroid cells in the control mice, TER-119⁺ cells showed only a transient increase peaking at PID 5 in mice immunized with peptide i (Fig. 2a). The number of F-MuLV-infected cells also stayed at a low level in the immunized mice. These results, along with the previous demonstration that numbers of F-MuLV-producing cells detected as infectious centers decreased between 7 and 11 days after FV infection in (B10.A × A.BY)F₁ mice immunized with peptide i (22), suggest that some effector mechanisms were activated in the immunized mice at around 5 to 7 days after FV infection to reduce the number of FV-infected erythroid cells.

FIG. 2.

Changes in the numbers of TER-119⁺ erythroid cells (a), F-MuLV gp70-expressing cells detected with MAb 720 (b), and CD4⁺ (c) and CD8⁺ (d) cells in the spleens of mice inoculated with FV. CB6F₁ mice were either immunized with peptide i (○) or given CFA without a peptide (●). Four weeks later, they were inoculated with 150 SFFU of FV. A group of three or four animals were killed at each indicated point, and their spleen cells were subjected to flow-cytometric analyses. Data presented here are means ± SEM. The dashed line in panel b indicates the limit of detection by the flow-cytometric analysis.

In fact, numbers of both CD4⁺ and CD8⁺ T cells increased between PID 5 and 9. As shown in Fig. 2c, a significant increase in the absolute number of CD4⁺ T cells was observed at PID 7 in both immunized and control CB6F₁ mice, and the number of CD4⁺ T cells continued to increase until PID 9 in immunized mice. A similar increase in the number of CD8⁺ T cells that peaked at PID 7 was observed in the control mice given CFA alone (Fig. 2d). In the immunized mice the peak of CD8⁺ T-cell number was lower, but this population of T cells also continued to increase until PID 9. It is interesting that only in immunized mice was another phase of increase in the number of CD4⁺ and CD8⁺ T cells observed at PID 13 (Fig. 2c and d). In uninfected CB6F₁ mice only 3% of CD8⁺ cells expressed the early activation marker CD69. The CD69⁺ population among CD8⁺ T cells increased transiently to 7% at PID 8 in the control mice given CFA alone but rapidly decreased to 2% at PID 11. In contrast, 13% of CD8⁺ cells were CD69⁺ at PID 8 in immunized mice, and the percentage of CD69⁺ activated cells among CD8⁺ T cells still increased to 14% at PID 11 in mice immunized with peptide i (data not shown). Thus, proliferation and activation of T cells were observed in both immunized and unimmunized control mice after FV infection, but T-cell proliferation and activation of CD8⁺ T cells was apparently prolonged in mice immunized with peptide i.

Detection and identification of cytotoxic effector cells in FV-infected mice.

Cytotoxicity assays were performed by using spleen cells from FV-infected mice as effector cells without in vitro restimulation and FV-induced leukemia cells as target cells. In repeated preliminary experiments, a peak in activities of cytotoxic effector cells capable of lysing FV-induced leukemia cells were observed at around PID 7 and 9 in mice immunized with peptide i (data not shown). This was in accordance with the reduction in the number of TER-119⁺ erythroid cells in immunized mice observed between 5 and 7 days after FV inoculation and coincidental proliferation of CD4⁺ and CD8⁺ T cells in vivo (Fig. 2). Thus, in the following experiments effector cells were prepared at PID 7 and 9 and separated into CD8⁺, CD4⁺, and double-negative populations using a magnetic cell sorter. In six repeated experiments performed at PID 7 and 9, CD8⁺ T cells isolated from mice immunized with peptide i always showed low but consistent levels of cytotoxicity against FV-induced leukemia cell lines FBL-3 and Y57-2C (Fig. 3). However, chemically induced T-lymphoma cells of the same H-2 genotype, EL-4, were not lysed by these effector cells. Interestingly, similar levels of cytotoxic activity of CD8⁺ T cells were also detected after FV inoculation in the control mice given CFA alone. The same two lines of FV-induced leukemia cells were also lysed in a dose-dependent manner by CD4⁺ effector cells isolated from peptide-immunized CB6F₁ mice in four of the six repeated experiments. Also, similar killing activities of CD4⁺ effector cells isolated from peptide-immunized, FV-infected CB6F₁ mice were detected in additional experiments (see Fig. 7). CD4⁺ T cells isolated from the control mice showed no or only marginal killing activities in repeated experiments (Fig. 3).

FIG. 3.

Detection of cytotoxic effector cells in FV-infected CB6F₁ mice. Mice were either immunized with 10 μg of peptide i/mouse or given CFA emulsion without a peptide. B220⁻ spleen cells were separated into CD8⁺, CD4⁺, and CD4⁻ CD8⁻ populations, and their cytotoxic activities against FBL-3 (○), Y57-2C (□), and EL-4 (●) cells were tested by incubating the effector and labeled target cells for 12 h. Representative data obtained from a set of experiments performed at PID 9 are shown here, and the results obtained from the six repeated experiments were consistent with these charts.

FIG. 7.

In vivo depletion of NK cell activity by injection of anti-asialo-GM₁ Ab. (a and b) CB6F₁ mice immunized with peptide i were injected either with 60 μg of anti-asialo-GM₁ Ab each (b) or with normal rabbit serum (a) and were infected with FV. Spleen cells were obtained at PID 9, and the NK cell activity of the B220⁻ population was tested by using YAC-1 (▵) and EL-4 (●) target cells. Data from two separate experiments are shown together here. Injection of higher doses of anti-asialo-GM₁ Ab gave the same results when B200⁻ cells were similarly tested for their YAC-1-killing activities. (c and d) Flow cytometric analyses for the expression of the NK cell markers on spleen cells obtained from mice injected with normal rabbit serum (c) or anti-asialo-GM₁ Ab (d). Experiments were performed twice and gave essentially the same results as those shown here. (e through j) Cytotoxicity assays using different cell populations isolated from spleen B220⁻ cells of peptide-immunized, FV-infected mice. CD8⁺, CD4⁺, and CD4⁻ CD8⁻ populations were purified as described for the experiments shown in Fig. 3 from CB6F₁ mice injected with anti-asialo-GM₁ Ab (f, h, and j) or from those injected with control rabbit serum (e, g, and i). The experiments were performed twice at PID 7 and 9, and the results from the repeated experiments were consistent with the representative data shown here. Target cells used were YAC-1 (Δ), FBL-3 (○), and EL-4 (●).

To confirm the observed cytotoxic effector function exerted by CD4⁺ T cells, CD4⁺ T-cell clones specific for F-MuLV-encoded antigens were tested for their killing activities. SB14-31 cells that recognize the N-terminal epitope represented by peptide fn induced significant lysis of FBL-3 leukemia cells in vitro (Fig. 4a). Syngeneic H-2^b/d hybridoma cells (LB 27.4) possessing MHC class II-restricted antigen-presenting ability were killed by this CD4⁺ T-cell clone only when they were incubated with the antigenic peptide, fn. On the other hand, cells of the H-2^d lymphoma line A20 that lack the restricting MHC class II molecule, A^b, were not lysed even when they were incubated with peptide fn. Interestingly, MBL-2 cells that share a homozygous H-2^b haplotype with FBL-3 were not lysed by the CD4⁺ T cells even when they were incubated with peptide fn. MBL-2 cells are now believed to have a common origin with EL-4 cells, which were not killed by bulk CD4⁺ T cells in the experiments whose results are shown in Fig. 3. Antigenic specificity of the CD4⁺ cytotoxic cells was further confirmed by incubating H-2^b/d LB 27.4 cells with several different peptides. Among four different peptides that were known to bind to the MHC class II A^b molecule, fn alone induced killing of target cells by the CD4⁺ T-cell clone. LB 27.4 target cells were also lysed when they were infected with a recombinant vaccinia virus that expressed F-MuLV env gene but not when they were infected with a vaccinia virus recombinant expressing the gag gene. Thus, FV-induced FBL-3 leukemia cells, but not MBL-2 lymphoma cells, are killed by antigen-specific CD4⁺ T cells.

FIG. 4.

Cytotoxic activity of a CD4⁺ T-cell clone, SB14-31, specific for an F-MuLV env-encoded epitope. (a) SB14-31 cells were incubated with various target cells with or without preincubation with peptide fn. ⁵¹Cr release during 4 h of incubation at an E:T ratio of 20 was measured. (b) LB 27.4 target cells were either incubated with the indicated A^b-binding peptides after ⁵¹Cr labeling or infected with the indicated recombinant vaccinia virus for 16 h at a multiplicity of infection of 10 and then labeled. Pretreated LB 27.4 cells were then incubated with SB14-31 cells for 4 h at an E:T ratio of 20:1. Experiments were performed at least twice at various E:T ratios, and the results were consistent with the representative data shown here.

Similar killing activity was also demonstrated with four independent CD4⁺ T-cell clones specific for the E^b/d-restricted C-terminal epitope represented by peptide i. All four clones tested lysed the H-2^b/d target cells when they were incubated with peptide i, and the killing activity was almost completely blocked by the addition of an anti-CD4 MAb but not an anti-CD8 MAb (Fig. 5). It should be emphasized that three of the four CD4⁺ T-cell clones specific for peptide i were established from CB6F₁ mice immunized with this particular peptide. Thus, although the killing activity of bulk CD4⁺ T cells detected from the immunized CB6F₁ mice after FV inoculation was low, some clones of CD4⁺ T cells recognizing the C-terminal epitope do exhibit cytotoxic activities against the target cells that present the F-MuLV env-derived antigenic peptide.

FIG. 5.

Cytotoxic activities of four different CD4⁺ T-cell clones specific for peptide i. T-cell clones F5-5 (a), FP3-10 (b), FP8-7 (c), and FP10-16 (d) were tested for their ability to lyse LB 27.4 target cells by incubation at the indicated E:T ratios for 3 h. LB 27.4 cells were incubated either with peptide i (□, ○, ▵) or with the control peptide of the same length, i_e (●). Killing assays were performed in the absence (○, ●) or presence of anti-CD4 (□) or anti-CD8 (▵) MAb. Assays were performed at least twice, and the results were consistent with the representative data shown here.

Unexpected dominance of cells expressing NK markers among cytotoxic effector cells in FV-infected mice.

When CD4⁻ CD8⁻ cells from FV-infected CB6F₁ mice were tested as effectors, they showed unexpectedly high cytotoxicities against FV-induced leukemia cells at lower E:T ratios than CD8⁺ and CD4⁺ T cells in all six repeated experiments performed at PID 7 and 9 (Fig. 3). However, EL-4 lymphoma cells that shared a homozygous H-2^b haplotype with the FV-induced leukemia cells were not lysed effectively by this population of spleen cells prepared from FV-infected mice. The double-negative populations of spleen cells from both immunized and control mice showed similar levels of cytotoxicity against FV-induced leukemia cells. Fluorescence-activated cell sorting analyses of the double-negative population revealed that in both immunized and unimmunized mice, 10 to 30% of these cells were positive for the NK cell marker NK-1.1. Thus, to further identify the characteristics of effector cells in the double-negative population, NK cells were purified from B220⁻ cells based on their expression of another surface marker defined by MAb DX5 (Pan-NK). Positive selection of DX5-expressing cells was performed by using B220⁻ cells as the starting material, instead of CD4⁻ CD8⁻ cells, because the yield of B220⁻ CD4⁻ CD8⁻ cells was usually around 2% of the total spleen cells, and it was impractical to obtain a large enough number of DX5⁺ effector cells from such a small starting cell number of the double-negative population. A large proportion of the cells selected for the expression of the Pan-NK marker were positive for both NK cell markers DX5 and NK-1.1 (Fig. 6b). These cells, obtained from immunized CB6F₁ mice at both 7 (Fig. 6c) and 9 (Fig. 6d) days after FV inoculation, showed strong killing activity against standard NK target YAC-1 cells and two lines of FV-induced leukemia cells at very low E:T ratios. However, NK-resistant EL-4 cells were not efficiently lysed by the same effector cells. A similar level of killing of YAC-1 cells and FV-induced leukemia cells was observed when the DX5⁺ NK cell population was obtained from the control mice given CFA alone and tested for cytotoxicity at PID 7 and 9 (data not shown). On the other hand, B220⁻ cells depleted of the DX5⁺ population lost most of the killing activity against YAC-1 cells and showed a low level of cytotoxicity against the FV-induced leukemia cells at higher E:T ratios, confirming that DX5⁺ cells constitute the bulk of cytotoxic effector cells in the spleens of FV-infected mice at this early stage.

FIG. 6.

Cytotoxic activity of DX5⁺ cell population isolated from the spleens of FV-infected mice. (a and b) Flow-cytometric analyses of B220⁻ (a) and B220⁻ DX5⁺ (b) populations. Cells selected with the anti-Pan-NK MAb were highly enriched for the expression of both NK cell markers, NK1.1 and DX5. (c and d) Cytotoxic activity of DX5⁺ cells separated from mice immunized with peptide i at PID 7 (c) or 9 (d). Target cells used were YAC-1 (▵), FBL-3 (○), Y57-2C (□), and EL-4 (●). Experiments were repeated by using two groups of animals for each PID, and results obtained from the repeated experiments were consistent with the representative data shown here.

Role of cells expressing NK markers in vaccine-induced resistance against FV infection.

Since cells lacking CD4 and CD8 and expressing DX5 constituted the majority of cytotoxic effector cells at the point when numbers of FV-infected erythroid cells were being reduced in immunized mice, and since FV-induced leukemia cells were shown to be susceptible to killing by DX5⁺ cells, the possible role of NK cells in vaccine-induced resistance against FV infection was tested by depleting NK cells from immunized mice. CB6F₁ mice were first immunized with peptide i and were injected with anti-asialo-GM₁ Ab before and after FV inoculation to deplete cells expressing this NK cell marker in vivo. Effects of the injection of various amounts of anti-asialo-GM₁ Ab were first assessed by testing YAC-1-killing activity of the spleen B220⁻ cells. After four repeated injections of 60 μg of anti-asialo-GM₁ Ab each, NK cell activity of spleen B220⁻ cells at PID 9 became undetectable (Fig. 7b), while mice injected with control rabbit serum retained NK cell activity (Fig. 7a). Flow cytometry analyses demonstrated that the DX5⁺ NK-1.1⁺ population of cells in the spleen of the mice injected with anti-asialo-GM₁ Ab was markedly reduced (<0.4% [Fig. 7d]) in comparison with the readily discernible cluster of DX5⁺ NK-1.1⁺ cells among the spleen cells of the control mice (Fig. 7c). Cytotoxic activities of different cell populations were assayed by separating effector cells from the peptide-immunized and anti-asialo-GM₁ Ab-injected mice with the magnetic cell sorting system. In two repeated experiments performed at PID 7 and 9, both CD8⁺ and CD4⁺ cell populations showed low cytotoxic activity against FV-induced leukemia cell line FBL-3, regardless of whether the mice were injected with anti-asialo-GM₁ Ab or normal rabbit serum (Fig. 7e to h). These results were consistent with those shown in Fig. 3, confirming the low but reproducible level of cytotoxic activity of CD8⁺ and CD4⁺ effector cells induced after FV infection in mice immunized with peptide i. In addition, the CD4⁻ CD8⁻ population of the spleen cells from the control mice injected with normal rabbit serum showed strong cytotoxic activities against both YAC-1 NK target cells and FLB-3 FV-induced leukemia cells (Fig. 7i). EL-4 cells were not lysed by any of the above effector cells. However, the killing activities of the CD4⁻ CD8⁻ population were markedly reduced in mice injected with the anti-asialo-GM₁ Ab (Fig. 7j), confirming that the majority of effector cells detectable as the double-negative cells were cells expressing the NK markers. Importantly, however, the injection of anti-asialo-GM₁ Ab did not affect the number or function of CD4⁺ and CD8⁺ populations of T cells in peptide-immunized mice. Flow-cytometric analyses revealed no significant difference in percentages of CD4⁺ and CD8⁺ cells in the spleen and the total nucleated-cell number between the anti-asialo-GM₁-injected and normal rabbit serum-injected groups of peptide-immunized CB6F₁ mice at PID 7 and 9 (Fig. 8a and b). Furthermore, CD4⁺ T cells purified from the anti-asialo-GM₁-injected mice showed antigen-specific proliferative responses upon stimulation with various concentrations of peptide i that were not significantly lower than the responses exerted by the same T-cell subpopulation isolated from the control mice (Fig. 8c and d). Nevertheless, when the mice depleted of NK cell activity were monitored for the development of FV-induced leukemia, >95% of the immunized mice injected with the anti-asialo-GM₁ Ab died within 60 days after FV inoculation, showing a survival curve similar to that of unimmunized control mice (Fig. 8e). On the other hand, >75% of the immunized control mice given normal rabbit serum survived past PID 60. These results clearly show that NK cells are required for vaccine-induced resistance against FV infection.

FIG. 8.

In vivo depletion of asialo-GM₁⁺ cells and its effect on T cells and protective immunity against FV infection induced by peptide immunization. (a through d) Mice used for the experiments whose results are shown in Fig. 7 were also analyzed for the presence of CD4⁺ and CD8⁺ T cells in the spleen and their ability to mount viral-antigen-specific CD4⁺ T-cell responses. Flow-cytometric analyses for the expression of CD4 and CD8 were performed by using pooled whole spleen cells obtained from the mice injected with anti-asialo-GM₁ Ab (b) or normal rabbit serum (a). Experiments were performed twice at PID 7 and 9, and results obtained from the repeated experiments were consistent with the representative data shown here. Numbers indicate percentages of CD4⁺ and CD8⁺ cells among live nucleated spleen cells. B220⁻ CD8⁻ CD4⁺ T cells purified for the experiments whose results are shown in Fig. 7g and h were also tested for their proliferative activities in response to stimulation with peptide i. CD4⁺ T cells purified from the mice injected with anti-asialo-GM₁ Ab (d) and those purified from control mice given normal rabbit serum (c) were incubated with X-irradiated syngeneic spleen cells and the indicated amount of peptide i (○). As controls, the CD4⁺ T cells purified from the anti-asialo-GM₁ Ab-injected mice were also stimulated with an endogenous retroviral env-derived peptide i_e (●) and the influenza virus nucleoprotein-derived peptide NP_366–374 (▵). Experiments were performed twice, and results obtained from the repeated experiments were consistent with the representative data shown here. (e) Development of FV-induced leukemia in CB6F₁ mice immunized with peptide i. Mice were either immunized with 10 μg of peptide i each (○, ▵, □) or given CFA alone (●). Two groups of the immunized mice were then injected with anti-asialo-GM₁ Ab (▵) or control rabbit serum (□), while the remaining group (○) was not injected with any Ab. All mice were inoculated with 150 SFFU of FV.

DISCUSSION

FV causes fatal erythroleukemia in immunocompetent adult mice through a multistep process. In the first step, SFFV gp55 expressed on cell surfaces interacts with the erythropoietin receptor and induces mitogenic activation and differentiation of infected erythroid progenitor cells in a polyclonal fashion (16, 21). This is followed by repeated integration of proviral DNA into the chromosomes of expanded erythroid cells and the consequent emergence of an immortalized leukemia cell clone (16). The primary targets for SFFV-induced polyclonal proliferation are late BFU-E and CFU-E, which are responsive to erythropoietin.

Several host genes affect the development and progression of FV-induced disease. These include the genes affecting the entry and replication of F-MuLV in target cells, those regulating and interfering with erythropoietin receptor-induced growth potentiation of erythroid progenitor cells, and those regulating immune responses against the viral antigens (2, 16). Immune mechanisms affecting FV infection and their genetic regulation have been studied mainly in mice with a (C57BL × A)F₁ background, because these mice show spontaneous resistance against FV infection depending on their composition of alleles at MHC loci (2, 24, 25). On the other hand, mice possessing a BALB/c background are extremely susceptible to FV-induced disease. In fact, when CB6F₁ mice used in the present study are inoculated with a low dose of FV, 95 to 100% die within 60 days postinoculation. This is striking because (B10.A × A/WySn)F₁ mice, which have typically been used as a strain susceptible to FV, show mortality rates of 70 to 80% at 90 to 100 days after FV inoculation (24). It should be emphasized, therefore, that a single immunization with the CD4⁺ T-cell epitope vaccine, peptide i, induced almost complete protection against FV infection even in this highly susceptible strain of mice.

Actual effector mechanisms activated after FV inoculation in mice immunized with the single-epitope peptide i seem complex. A transient increase in the number of CD4⁺ and CD8⁺ T cells in the spleen and the expression of early activation marker CD69 on CD8⁺ T cells coincided with the suppression of the growth of TER-119⁺ erythroid cells and F-MuLV-infected cells in immunized mice (Fig. 2). However, this transient increase in the number of CD4⁺ and CD8⁺ T cells was also observed in unimmunized control mice after FV infection, although the increase in number and activation of T cells in vivo was prolonged in mice immunized with peptide i. Despite the apparent proliferation and the expression of the early activation marker, cytotoxic activity of CD8⁺ T cells at the time of the reduction of erythroid cell number in immunized mice was not high (Fig. 3). Moreover, similar levels of cytotoxic activities were observed when CD8⁺ T cells were separated after FV infection from the control mice given CFA alone. The observation that CD8⁺ CTL were activated in the control mice after FV infection is not surprising, because FV-specific CD8⁺ CTL have been detected in unimmunized (B10 × A.BY)F₁ mice during their spontaneous recovery from FV-induced splenomegaly (3, 30). Furthermore, CTL activities were also detected in H-2^a/b (B10.A × A.BY)F₁ mice after inoculation with a low dose of FV (2). Thus, induction of CD8⁺ CTL is commonly observed in FV-infected mice irrespective of whether they have been immunized with an anti-FV vaccine prior to the virus inoculation. For this reason, it is unlikely that the effect of peptide immunization, especially the suppression of the growth of FV-infected erythroid cells observed as early as 5 to 7 days after virus inoculation, can be explained by the activation of CD8⁺ CTL alone. In this regard, a peak in CTL activities was observed at around 2 weeks after FV inoculation both in mice showing spontaneous recovery from FV infection and in those immunized with the recombinant vaccinia virus expressing the F-MuLV env gene (3, 9, 30). This is 1 week later than the point at which the absolute number of CD4⁺ and CD8⁺ T cells increased and numbers of FV-infected erythroid cells were reduced in mice immunized with peptide i (Fig. 2) (22). Thus, previously documented CTL activities peaking at PID 14 or later may not be the cause of the control of FV infection but might reflect the cytokine-induced expansion of CD8⁺ effector cells resulting from earlier immune responses that are actually related to the containment of virus infection. Given this, CD8⁺ CTL may also be involved in the control of leukemia development at a later stage, perhaps by suppressing the emergence or growth of monoclonal erythroleukemia cells.

In addition to CD8⁺ T cells, CD4⁺ T cells were also shown to exert killing activities against FV-induced leukemia cells (Fig. 3 to 5 and 7). This killing by CD4⁺ T cells was blocked by the addition of anti-CD4 MAb and was dependent on both the presence of a viral antigenic peptide and the presence in target cells of MHC haplotypes associated with the presentation of the antigenic epitope. This evidence indicates that CD4⁺ effector cells specifically recognize a viral antigenic peptide presented by MHC class II molecules on the surfaces of target cells. Some FV-induced leukemia cells, including FBL-3, are known to express detectable amounts of MHC class II molecules (4, 43), while MBL-2 cells that were resistant to killing by cloned CD4⁺ T cells are known to express an extremely low level of MHC class II molecules (42). Thus, in mice immunized with peptide i, some clones of CD4⁺ T cells specific for this antigenic epitope may lyse FV-infected target cells that express MHC class II molecules. The actual protective role of CD4⁺ cytotoxic effector cells in mice immunized with peptide i is difficult to assess, because both depletion and transfer of CD4⁺ T cells may affect not only the presumable cytotoxic effector function but also helper functions of CD4⁺ T cells. However, results from our preliminary experiments suggest a role of CD4⁺ effector cells in vivo, because mice deficient in β₂ microglobulin and thus lacking CD8⁺ T cells were nevertheless protected against a low dose of FV when immunized with peptide i (A. Niwa and M. Miyazawa, unpublished observation). More detailed analyses of the frequency and kinetics of the induction of CD4⁺ and CD8⁺ cytotoxic effector cells may correlate these effector mechanisms with the vaccine-induced elimination of FV-infected erythroid cells in the future.

To our surprise, cells expressing NK cell markers constituted the majority of cytotoxic effector cells detected in FV-infected mice when the number of FV-infected erythroid cells was reduced. In fact, when cytotoxic effector cells capable of killing FV-induced leukemia cells were separated from the B220⁻ population of spleen cells, CD4⁻ CD8⁻ cells showed higher cytotoxic activities at E:T ratios lower than those required for the killing of the same target cells by CD4⁺ or CD8⁺ effector cells (Fig. 3). The double-negative population contained 10 to 30% DX5⁺ cells. DX5⁺ cells separately isolated from the B220⁻ spleen cell population of the peptide-immunized and FV-infected mice showed effective killing of the FV-infected leukemia cells at E:T ratios three- to sixfold lower than those required for the double-negative population (Fig. 6). These results suggest that the majority of the effector cells contained in the double-negative population are probably DX5⁺ cells. Furthermore, these DX5⁺ cells also killed the standard NK target YAC-1 cells very efficiently. Expression of both NK-1.1 and DX5 and efficient killing of YAC-1 target cells are phenotypic and functional markers of mouse NK cells. The NK cell nature of the CD4⁻ CD8⁻ effector cells was further confirmed by depleting asialo-GM₁⁺ cells by injecting the relevant Ab into peptide-immunized and FV-infected mice. When anti-asialo-GM₁ Ab was repeatedly injected into CB6F₁ mice that had been immunized with peptide i and infected with FV, the CD4⁻ CD8⁻ population showed markedly reduced killing activity for both YAC-1 and FBL-3 target cells (Fig. 7). Thus, it is reasonable to conclude that the majority of effector cells contained in the B220⁻ CD4⁻ CD8⁻ population of the spleen cells were positive for multiple NK cell markers, and these NK cells effectively kill both YAC-1 target cells and two independent lines of FV-induced leukemia cells, FBL-3 and Y57-2C.

Similarly high NK cell activities were detected in macaques at 1 to 2 weeks after infection with a pathogenic strain of simian immunodeficiency virus (10). Interestingly, the transient increase in NK cell activity was found to be inversely correlated with plasma antigenemia levels when cytotoxic activity against NK target K562 cells and levels of viral p27 in plasma in individual macaques were compared (10). That report suggested that NK cell killing of virus-infected cells might be involved in the containment of retroviral infection in its earlier stage. This hypothesis has been directly examined in the present study by depleting NK cells in FV-infected mice. In fact, vaccine-induced resistance against FV infection was totally abrogated when YAC-1-killing NK cell activity was eliminated by treating immunized mice with anti-asialo-GM₁ Ab. Slifka et al. (37) recently showed that in a model of lymphocytic choriomeningitis virus infection in C57BL/6 mice, ≥90% of virus antigen-specific CD8⁺ T cells and nearly 90% of CD4⁺ T cells responding to the viral antigen express asialo-GM₁, and 30 to 40% of virus-specific CD8⁺ cells express DX5 at 8 days after lymphocytic choriomeningitis virus infection. If this observation can be generalized to any virus infection, one can argue that in our experiments whose results are shown in Fig. 7 and 8, we might have depleted asialo-GM₁-expressing CD8⁺ and/or CD4⁺ T cells by injecting anti-asialo-GM₁ Ab into peptide-immunized mice. In addition, the DX5⁺ population of effector cells which we used in the experiment whose results are shown in Fig. 6 might have included DX5⁺, CD8⁺, and/or DX5⁺ CD4⁺ effector cells, and thus, we might have detected killing activities of such virus-specific effector T cells in addition to NK cell activities. However, these possibilities are unlikely for several reasons. First, if viral-antigen-specific T cells expressed NK cell markers in FV-infected mice, they must have been coseparated into the CD8⁺ and CD4⁺ populations of the effector cells in the experiments whose results are shown in Fig. 3 and 7. Thus, the double-negative population, which was actually ≥92% negative for CD4 and CD8 expression as confirmed by flow-cytometric analyses, contained very few, if any, T cells expressing the NK cell markers. Nevertheless, the majority of the cytotoxic activity was detected in the double-negative population, indicating that in the early stage of FV infection non-T NK cells, rather than CD8⁺ or CD4⁺ effector T cells expressing NK cell markers, exert the most efficient killing activities against the virus-infected cells. Second, in a few preliminary experiments (results not shown), we purified DX5⁺ cells from the B220⁻ CD4⁻ CD8⁻ population of spleen cells. Because of the very small number of the effector cells finally obtained, cytotoxicity assays were performed with limited E:T ratios. However, these B220⁻ CD4⁻ CD8⁻ DX5⁺ cells showed a high efficiency of killing of both YAC-1 and FBL-3 target cells. Third, when the minimal required amount of anti-asialo-GM₁ Ab was used for the depletion of NK cell activity, the total nucleated-cell number and percentages of both CD4⁺ and CD8⁺ T cells in the spleen did not change significantly in comparison with those in the control mice given normal rabbit serum (Fig. 8). Furthermore, the viral epitope-specific proliferative responses of CD4⁺ T cells were not significantly affected in the mice depleted of asialo-GM₁-expressing cells. Thus, at least the number and the above viral antigen-specific function of T cells were not affected by injection of anti-asialo-GM₁ Ab. In addition, the low but reproducible killing activity of CD8⁺ T cells against FV-induced leukemia cells was still detectable in the mice depleted of asialo-GM₁⁺ NK cells. Thus, it is appropriate to conclude that the complete abrogation of the peptide-induced protection against FV infection observed in mice injected with anti-asialo-GM₁ Ab (Fig. 8) is mainly attributable to the elimination of NK cell functions exerted by the CD4⁻ CD8⁻ cells.

Since NK cell activity against FV-infected leukemia cells was detectable in both immunized and control mice after FV infection, detected NK cell function alone cannot explain the effectiveness of peptide immunization. Rather, it is possible that some other effector mechanisms activated only in immunized mice cooperate with NK cells to effectively eliminate FV-infected cells. These may include accelerated production and class switching of virus-neutralizing antibodies detectable at as early as 7 days after FV inoculation in vaccinated mice (22), and CD4⁺ cytotoxic cells detected in the present study (Fig. 3 to 5 and 7). NK cells might also be required as a source of cytokines necessary for the activation of other effector mechanisms. Enhanced NK cell activity in the early stage of FV infection detected in the present study may be due to the administration of CFA before FV inoculation, since the control mice were given CFA emulsion without a peptide. However, in our preliminary experiments, B220⁻ CD4⁻ CD8⁻ cells purified from unmanipulated CB6F₁ mice at 7 and 9 days after FV infection showed very efficient killing of YAC-1 and FV-induced leukemia cells at low E:T ratios. In fact, the double-negative cells prepared at PID 9 from the mice that had received no adjuvant showed 50 and 35% killing of YAC-1 cells at E:T ratios of 150:1 and 75:1, respectively, and 32 and 28% killing of FBL-3 cells at the same respective E:T ratios. These killing activities were comparable to those shown in Fig. 3 and 7. Thus, it is possible that infection with FV itself, without stimulation of the immune system with CFA, induces enhanced NK cell activity.

Detection of cytotoxic effector cells from FV-infected mice has been performed without in vitro restimulation of the effector cell population (2, 3, 7, 30). Because of this, an incubation of effector cells with labeled target cells for a longer period than in other systems was required to detect significant killing activities. Although it has not been formally documented, many researchers, including us, have attempted to restimulate spleen cells from FV-infected mice with irradiated FV-induced leukemia cells in vitro, without getting selective augmentation of antigen-specific killing activity. In our experience, restimulation of spleen cells from FV-infected mice always results in a nonspecific killing activity against uninfected target cells like P815 mastocytoma cells, causing difficulties in detecting MHC-restricted, antigen-specific cytotoxicity. It is clear now that FV-induced leukemia cells are recognized effectively by NK cells, and NK cells are activated in FV-infected mice. It is, therefore, possible that in attempts to restimulate cytotoxic effector cells in vitro by coculturing spleen cells with FV-induced leukemia cells, NK cells are expanded and stimulated to exert a high killing activity.

In conclusion, FV-induced leukemia cells are recognized and killed by NK cells, and NK cells are activated in the early stage of FV infection. NK cells are necessary for vaccine-induced resistance against FV infection, but NK activity alone cannot explain the rapid elimination of FV-infected erythroid cells that takes place in mice immunized with the CD4⁺ T-cell vaccine. Thus, antiretroviral vaccine strategies may require not only direct priming of CD8⁺ effector cells and induction of virus-neutralizing antibodies but also proper stimulation of NK cell activities, which may also lead to the induction of T-helper type 1 responses advantageous for the control of retroviral infections.