Up-modulation of interferon-γ mediates the enhancement of spontanous cytotoxicity in prolactin-activated natural killer cells

Authors


Lina Matera, Department of Internal Medicine, University of Turin, Corso A.M. Dogliotti, 14. 10126 Turin, Italy.

Summary

Prolactin (PRL) has been shown to participate in lymphocyte activation. In particular, the constitutive natural killer (NK) and the lymphokine-activated killer (LAK) cytotoxicity of CD56+ CD16+ cells is increased by its physiological to supraphysiological concentrations. As PRL has been shown to up-regulate the production of interferon-γ (IFN-γ) by peripheral blood mononuclear cells, we studied its effect on IFN-γ production by NK cells as a possible mechanism of autocrine activation of cytotoxicity. Released and intracellular IFN-γ, as well as IFN-γ mRNA expression, were increased by pituitary and recombinant human PRL, which stimulated optimal NK and LAK cytotoxicity. Treatment with blocking anti-IFN-γ monoclonal antibody (mAb) selectively affected PRL-increased killing of K562 targets, demonstrating that PRL-mediated enhancement of spontaneous cytotoxicity depends, at least in part, on up-regulation of IFN-γ.

Abbreviations
LAK

lymphokine-activated killer

NK

natural killer

PRL

prolactin

Introduction

Prolactin (PRL), an immunomodulator hormone, acts as an autocrine factor during T-lymphocyte activation 1,2 (reviewed in 3–5). In addition, interleukin (IL)-2-driven lymphokine-activated killer (LAK) maturation of natural killer (NK) cells, present in peripheral blood mononuclear cells (PBMC), has been shown to depend on endogenously produced PRL. 6 When added to purified NK cells 7,8 or to PBMC, 9 PRL has been shown to mimic IL-2-induced cell growth, 8 NK enhancement and LAK development. 7–9 PRL and IL-2 receptors belong to the family of haemopoietin class I receptors 10,11 and share downstream intracellular signalling pathways 12,13 and, possibly, target genes. 14 Interferon-γ (IFN-γ) is the major factor involved in NK cytotoxic function and is induced by IL-2 in NK cells. 15,16 To determine whether PRL activates NK cells with the same mechanism, we studied its effects on IFN-γ production and the involvement of IFN-γ in the PRL-induced NK and LAK activation.

Materials and methods

Media

Minimal essential medium (MEM) containing 2 g/l hydrogen carbonate and 20 n m HEPES buffer (Sigma-Aldrich S.r.l., Milan, Italy) was used as bench medium throughout the lymphocyte separation procedures. RPMI-1640 containing 2 m m l-glutamine, 3 mg/ml sodium bicarbonate, 5 × 10–8 mβ-mercaptoethanol, penicillin (100 IU/ml)/streptomycin (100 UG/ml) and 10% heat-inactivated fetal calf serum (FCS), was used for cell line feeding. To avoid interference of serum growth factors, a FCS-free medium containing bovine insulin (sodium- and zinc-free; 5 mg/ml), human (holo) transferrin (5 mg/ml) and sodium selenite (5 ng/ml) (Redu-SerTM; UBI, Lake Placid, NY), referred to as STI-RPMI, was used for lymphocyte culture.

Growth factors and antibodies

Native pituitary human (nh) PRL was a gift from the National Hormone and Pituitary Program, National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), NIH (Baltimore, MD). Recombinant (r) hPRL was a gift from Genzyme Corporation (Framingham, MA). rIL-2 was kindly provided by Glaxo (IMB SA, Geneva, Switzerland). B73.1 monoclonal antibody (mAb), directed against the CD16 determinant on NK cells, was a gift from Dr G. Trinchieri (The Wistar Institute, Philadelphia, PA). Leu11b (anti-CD16) (immunoglobulin M, IgM) and Leu19 (anti-CD56) mAbs were purchased from Becton-Dickinson (Mountain View, CA). Anti-CD3 OKT3 mAb was purchased from Ortho Diagnostics (Raritan, NJ). The anti-IFN-γ mAb B133.3.1, used in immunocytochemistry experiments, was also a gift from Dr G. Trinchieri. The blocking mAbs anti-IFN-α (neutralizing activity 10 IU/μl) and anti-IFN-γ (neutralizing activity 1 IU/ng) were purchased from Boehringer Mannheim (Monza, Milan, Italy) and Genzyme.

Isolation of NK cells

Purified NK cells were isolated from peripheral blood of normal donors, as previously described. 6 PBMC were obtained after separation of heparinized blood of normal donors on Ficoll–Hypaque. NK-enriched cell populations consisting of > 70% CD16+ CD56+ , < 2% CD3+ cells were obtained from PBMC after depletion of monocytes by adherence on plastic flasks, depletion of B cells by passage on a nylon wool column and depletion of high-affinity E-rosette-forming cells (EHa), as previously described. 6 In some experiments, a passage on immunocomplex monolayers was included to obtain highly purified (> 95% CDI6+ CD56+) NK cells. 6 Purifed (> 95% CD3+) T cells were obtained from the EHa after osmotic lysis of sheep red blood cells.

Biological assay for IFN-γ

Purified NK and T cells were cultured with the indicated concentrations (see the Results) of nhPRL or rhPRL. At the time-points specified (see the Results), the supernatant was collected, filtered and stored. Antiviral activity was determined in culture supernatants by a cytopathogenic effect (CPE) inhibition assay employing 3-(4,5 dimethylthiazol-2-yl)-2,5 diphenyltetrazolium bromide (MTT) metabolic reduction as an indicator of cell viability. IFN activity was determined by preincubation of each sample for 2 hr at 37° with anti-IFN-γ mAbs or with medium alone. Briefly, test supernatants, as well as the hIFN-γ reference (NIBS, Hertfordshire, UK; 250 IU/ml), were diluted in serial twofold dilutions in 96-well microtitre plates containing a confluent monolayer of WISH cells (a human amniotic cell line). After incubation for 4 hr at 37° in an atmosphere of 5% CO2, plates were washed three times with serum-free Eagle’s minimal essential medium (EMEM) and each well received vesicular somatitis virus (VSV) at a multiplicity of 1 in serum-free EMEM containing 0·25% bovine serum albumin (BSA). After overnight incubation, when the microscopically assessed CPE in control wells was > 90%, 25 μl of a 5 mg/ml MTT solution was added to each well, and incubation was carried out for an additional 2·5 hr. The quantity of formazan was measured photometrically at 570 nm after overnight solubilization with 10% sodium dodecyl sulphate (SDS). Non-linear regression curves were generated for test samples and for standard IFN-γ preparations by plotting each dilution against its optical readout. The reciprocal of the dilution that reduced the viral CPE by 50% was considered as the IFN titre. Each test sample titre was adjusted by comparing its regression-generated 50% end-point value with that of the IFN standard.

Immunocytochemistry

At specified time-points during culture with hPRL (see the Results), enriched NK cells were removed, spun onto glass slides using a cytocentrifuge (Heraeus Sepatech GmbH, Labofuge A, Am Kalkberg, Germany) and air-dried. Following fixation with methanol/acetone (1 : 1) for 3 min, slides were blocked with 1% BSA in phosphate-buffered saline (PBS) for 30 min and incubated for 2 hr with the anti-IFN-γ mAbs described above or with normal mouse serum. Immunocytochemical staining was performed using an alkaline phosphatase (AP) streptavidin–biotin system (Multilink; Biogenex Laboratories, San Ramon, CA).

In situ hybridization

For in situ hybridization, a polymerase chain reaction (PCR) product of human IFN-γ, spanning nucleotides 475–1069, was prepared and cloned into the pGEMT vector (Promega, Madison, WI). The orientation of the insert was determined by partial sequencing. Digoxigenin-labelled RNA was prepared using the Dig-RNA labelling kit (Boehringer Mannheim, Mannheim, Germany). One microgram of restriction enzyme-digested plasmid was used to prepare sense and antisense digoxigenin-labelled RNA by reverse transcription with either T7 or Sp6 RNA polymerase, according to the manufacturer’s protocol. Cytospin cell preparations were fixed by a 15-min incubation in PBS containing 4% paraformaldehyde, and then washed for 30 min in PBS. The digoxigenin-labelled RNA probe (100 μl), diluted to 100 ng/ml in hybridization solution, was applied to the slides and incubated at 42° overnight. Stringent washes were conducted at room temperature for 30 min in 2× sodium chloride/sodium citrate buffer (SSC), then three times for 15 min at 45° in 2× SSC/50% formamide and twice for 15 min at room temperature in 1× SSC. For detection of bound digoxigenin-labelled probes, the Dig Nucleic Acid Detection Kit (Boehringer Mannheim) was used, according to the manufacturer’s protocol.

Treatment of PRL-stimulated NK cell cultures with a blocking anti-IFN-γ mAb

Enriched NK cells were cultured in STI-RPMI with different concentrations of PRL, discussed above, in the presence or absence of the anti-IFN-γ mAb or normal mouse serum at a final concentration of 50 ng/ml (chosen on the basis of the results obtained in pilot experiments). After 5 days of culture, cells were tested for cytotoxicity against the K562 and HL60 cell lines in a 4-hr 51Cr-release test, as previously reported. 6 NK activity was expressed as lytic units (LU)/108 effectors, which reduce the 51Cr-release response to linearity. 17 A LU is the number of effector cells required to produce 30% specific cytotoxicity of 4 × 103 target cells.

Statistical analysis

Significance of the difference between untreated and treated cells was assessed using the test of Wilcoxon.

Results

Release of IFN-γ by NK cells

Purified NK and T cells were cultured in serum-free medium, with or without increasing concentrations (12, 25 and 50 ng/ml) of nhPRL or rhPRL. These concentrations were chosen on the basis of our previous studies. 6 The IFN activity of the supernatants was studied after 6, 48 and 72 hr of culture. The effect of PRL on NK cells was maximally evident at the earliest sampling time (control versus 12 ng/ml PRL, P = 0·02 at 6 hr) ( Fig. 1). Both nhPRL and rhPRL modulated IFN-γ release by NK cells ( Fig. 2), according to a bell-shaped pattern, with a significant increase at 12 ng/ml (P = 0·04 for nhPRL and P = 0·01 for rhPRL) ( Fig. 2). A marked, although not significant, increase of IFN-γ activity was detected in the supernatant of T cells after 6 hr of culture with 50 ng PRL ( Fig. 1), which might be attributed to residual NK cells. However, this response followed a different pattern from that observed with purified NK cells, where an increase of IFN-γ activity was not observed after 12 hr of culture (data not shown) or at later time-points ( Fig. 1). The data presented in Fig. 1 represent the mean ± SD of four experiments with different donors. In three of these experiments, the specificity of the IFN activity was assessed by preincubating the test supernatants with blocking anti-IFN-α/γ Abs ( Fig. 3).

Figure 1.

Early and late modulation of interferon (IFN) activity in the supernatant of natural killer (NK) and T cells stimulated with prolactin (PRL). Supernatant from purified NK (▴) and T (▪) cells cultured for 6, 48 and 72 hr with the indicated concentrations of natural pituitary hormone (nh)PRL or recombinant (rh)PRL was tested for antiviral activity using a cytopathogenetic effect (CPE) inhibition assay (see the Materials and methods). Results represent the mean ± SD of four donor blood samples treated with nhPRL. Each determination was repeated four times. (*P = 0·02).

Figure 2.

Effects of native pituitary hormone prolactin (nhPRL) and recombinant (rh)PRL on release of interferon (IFN) activity. Natural killer (NK) (▴) and T (▪) cells were cultured with either nhPRL or rhPRL at the indicated concentrations, and the IFN activity was assessed in the supernatant after 48 hr of culture. (*P < 0·05, see text).

Figure 3.

Specificity of the interferon (IFN) activity. The IFN activity of the natural killer (NK) supernatant was assessed with anti-IFN-α (◆) or anti-IFN-γ (●) blocking monoclonal antibodies (mAbs), or with no mAb (▴).

Intracellular IFN-γ

Enriched NK cells were incubated with or without rhPRL (12, 25 and 50 ng/ml) for 6 and 48 hr. Immunocytochemistry with two anti-IFN-γ mAbs (see the Materials and methods) produced a comparable pattern of reactivity in three experiments. The number of cells that reacted with the two mAbs increased markedly (as early as 6 hr) during culture with 12 (from 18 ± 3 to 43 ± 5) and 25 (from 18 ± 3 to 38 ± 5) ng/ml PRL. As shown in Fig. 4, the intensity of the staining (using mAb B133.3.1) was also increased.

Figure 4.

Modulation of intracellular interferon-γ (IFN-γ) by prolactin (PRL). Enriched natural killer (NK) cells cultured in the absence (a) and (c) or presence (b) of PRL (12 ng/ml) were cytocentrifuged and incubated with the anti-IFN-γ B133.3.1 mAb (a) and (b) or with a control mouse serum (c). Staining was performed using alkaline phosphatase (AP) streptavidin–biotin. Similar results were obtained with another mAb (see the Materials and Methods) (data not shown).

IFN-γ mRNA expression was also visualized at a single cell level by in situ hybridization ( Fig. 5). Constitutive, weak expression of IFN-γ mRNA was observed with unstimulated NK cells, whereas 8·5% and 21% of these cells were heavily stained following stimulation with 12 and 25 ng/ml PRL, respectively. The positive control, rIL-2, gave 65% positivity.

Figure 5.

Interferon-γ (IFN-γ) mRNA expression on unstimulated and prolactin (PRL)-stimulated natural killer (NK) cells. In situ hybridization with an IFN-γ RNA probe showed a weak background staining on NK cells cultured with medium (a) and the appearance of a heavy staining on 21% of PRL (25 ng/ml)-stimulated cells (b). In the negative control (c), the probe was omitted, whereas the positive control (d) was represented by NK cells treated with interleukin-2 (IL-2) as a standard IFN-γ inducer (65% positivity).

Neutralization of PRL-induced NK/LAK activities by a blocking anti-hIFN-γ mAb

As shown in Fig. 6, the K562 cytotoxicity of enriched NK cells cultured with 12 ng/ml nhPRL was inhibited by the anti-IFN-γ mAb (P = 0·002). By contrast, newly generated cytotoxicity against the NK-resistant HL60 cell line was unaffected, thus showing that the inhibitory effect observed with NK activity was not the result of non-specific mAb toxicity. Furthermore, no inhibition was observed in cultures treated with normal mouse serum (data not shown).

Figure 6.

Residual prolactin (PRL)-induced natural killer (NK) and lymphocyte-activated killer (LAK) activities after neutralization of interferon-γ (IFN-γ). Enriched NK cells were cultured with the indicated concentrations of PRL in the presence or absence of the blocking anti-IFN-γ mAb B133.3.1, and cytotoxicity against NK (K562) or LAK (HL60) target cell lines was tested on days 5–6 in a 4-hr 51Cr-release assay. Normal mouse serum had no effect (data not shown). Results represent the pattern seen in three experiments. (*P = 0·002). LU, lytic units.

Discussion

PRL has been shown to activate NK cells. Both increase of spontaneous cytotoxicity against NK-sensitive targets and generation of LAK activity against NK-resistant targets stemmed from PRL-stimulated PBMC 9 or purified CD56+ CD16+ cells. 7 In agreement with previous reports 18,19 and in keeping with the dimerization model proposed for the growth hormone (GH) family receptors, 20 these effects were bell shaped with a peak at physiological concentrations (12–25 ng/ml) of PRL, close to the Kd of the PRL receptor on T, B 21 and NK 22 cells. Here we show that the same PRL concentrations up-modulated the synthesis and release of IFN-γ by CD56+ CD16+ cells. In our experiments, 12 ng/ml PRL induced optimal IFN-γ release, whereas maximal IFN-γ mRNA expression was observed with 25 ng/ml PRL.

Induction of IFN-γ by PRL has been reported in total PBMC. 23 Here we show that an increased production of IFN-γ mediates, in an autocrine manner, the enhancement of CD56+ CD16+ spontaneous cytotoxicity induced by PRL. Treatment with an α-IFN-γ mAb, in fact, almost completely abrogated the increased anti-K562 killing. Inhibition was specific for NK activity because LAK cytotoxicity, generated in vitro against the NK-resistant target HL60, was not affected. Conceivably, other cytokines, also induced by PRL, may up-regulate the expression of surface molecules involved in the recognition of LAK target-cell structures and triggering or inhibition of the cytotoxic machinery. 24,25 Our preliminary observations (L. Matera, A. Galetto, M. Geuna et al., manuscript in preparation) point to T-cell derived IL-2 as a candidate for this effect. The fact that IFN-γ is constitutively expressed by NK cells ( Figs 2 and 3) is in line with its prominent importance in the enhanced anti-K562 cytotoxicity, as opposed to the in vitro-generated anti-HL60 cytotoxicity, and is indicative of its role in the physiological function of NK cells as effectors of innate, not adaptive, immunity in vivo.

Production of IFN-γ by NK cells is stimulated by several cytokines, namely IL-2, IL-12, IL-15 and IL-18, 16,26,27 whose receptors share a common subunit (β-chain) and form part of the family of haemopoietin receptors to which PRL and GH also belong. 10,11 In particular, IL-2 and PRL share many features. They stimulate the growth and cytotoxic function of NK cells and phosphorylate the transcription factors, Stat-5b and IRF-1, which are both involved in the cytotoxicity of NK cells. 12–14,28 Thus, it is not surprising to find that IFN-γ is a target gene for both PRL and IL-2 receptors.

Evidence has accumulated for a role of PRL in the development or maintenance of autoimmune diseases. Abnormalities in the levels of serum, as well as lymphocyte-derived PRL, have been reported in systemic lupus erythematosus (SLE), rheumatoid arthritis and experimental autoimmune encephalomyelitis, and administration of dopaminergic compounds improves their clinical course. 29–34 As autoimmune diseases are characterized by a shift towards the T helper 1 (Th1) cytokine profile, 35 it is tempting to speculate that PRL acts as a Th1-differentiating cytokine. This possibility is presently being evaluated in our laboratory and is supported by the following observations: glucocorticoids, which inhibit PRL, 36 induce T helper 2 (Th2) maturation; 37 and GH prevents full activation of the Th2-cytokine phenotype in mice following infection with bacterial enterotoxin. 38

Whatever the position and the role of PRL in the Th1/Th2 cytokine network, its ability to promote both the survival 39 and the spontaneous cytotoxicity of NK cells 7 under serum-free conditions suggest that it acts independently of other cytokines in maintaining constitutive NK cell antiviral and antitumor functions in vivo. Up-regulation of the IFN-γ gene, as described above, may be one of the means by which this goal is accomplished.f

Acknowledgments

We wish to acknowledge the generous gifts of recombinant hPRL and standard hPRL by Genzyme (Cambridge, MA) and the National Hormone and Pituitary Program of the NIDDK, respectively. This work was supported by the Italian Ministry of University and Technology Research (MURST).

Ancillary