Quiescent and activated mouse granulocytes do not express granzyme A and B or perforin: similarities or differences with human polymorphonuclear leukocytes?

Perforin (perf) and granzymes (gzms) are major components of cytoplasmic granules from natural killer (NK) cells and cytotoxic T lymphocytes (CTLs) and are critical cytolytic effector molecules in NK/CTL-mediated apoptosis and are required for the control of intracellular pathogens and tumors.^1-10  In mice and humans, gzms, in particular gzmA and gzmB, and perf are produced mainly by the majority of NK cells and CTLs, a fraction of CD4⁺ T cells,^11-13  and to a lesser extent by related cell types, such as metrial gland cells,¹⁴  but not by polymorphonuclear leukocytes (granulocytes/neutrophils) and monocytes.^15-17 

Two recent studies provided evidence that gzms and perf are also produced by human polymorphonuclear neutrophils (PMNs).^18,19  These reports were intriguing in light of the known biologic effects of gzms and perf in NK/CTL-mediated immune responses^8,10,20,21  and the role of PMNs as a first-line defense against microbial pathogens^22,23  and tumors.^24,25  However, the assumption that PMNs use the same set of molecules as NK/CTL to combat pathogens and tumors was challenged by 2 subsequent studies.^26,27  To date, the discrepancy of the 4 studies, which used similar techniques for the enrichment of PMNs and to analyze the expression of GZM- and PRF1-specific transcripts and/or proteins, has not been resolved.

We have now analyzed in detail the expression of gzmA, gzmB, and perf in in vitro-generated and ex vivo-derived mouse granulocytes by combining cytology, correlative cell surface phenotypic analysis, intracellular flow cytometry, enzyme assays, immunoelectron microscopy (IEM), and reverse transcriptase-polymerase chain reaction (RT-PCR) analysis. In addition we have investigated the expression of gzmB in human PMNs from healthy individuals and a panel of patients with distinct diseases.

Inbred C57BL/6 (B6) and mouse strains deficient for gzmA (gzmA^-/-), gzmB (gzmB^-/-), and gzmA × B (gzmA × B^-/-) bred on the B6 background were maintained at the Max-Planck-Institut für Immunbiologie, Freiburg, Germany, and analyzed for their genotypes as described.^4,5  Male mice of 8 to 12 weeks of age were used in all experiments and were conducted in accordance with the ethical guidelines of the Federation of European Laboratory Animal Science Association.

The mouse cell lines 1.3E6SN (CTL; perf⁺, gzmA⁺, gzmB⁺²⁸) and EL4.F15 (thymoma; perf^-, gzmA^-, gzmB^-5) were used as control cells for the analysis of perf- and gzm-specific transcripts and proteins.

Cell lines were cultured in minimal essential medium (MEM) supplemented with 10% heat-inactivated fetal calf serum (FCS) and 2-mercapto ethanol (10^-5 M) at 7% CO₂, as described²⁹ ; for 1.3E6SN cells, medium was supplemented with 10% Concanavalin A supernatant (ConA SN). ConA SN was used as cytokine source for expansion of mouse T cells. To produce ConA SN the SN of rat spleen cells was previously stimulated with ConA (5 μg/mL, 2 days) and subsequently treated with 20 mg/mL α-methyl-D-mannopyranosid to block residual ConA activity.

In vitro-propagated alloreactive T-cell lines (H-2^b anti-H-2^d) were generated by weekly restimulation with irradiated spleen cells and ConA SN, as described.³⁰  For analysis of intracellular gzmA/gzmB expression, alloreactive T-cell lines (∼ 85%-90% CD8⁺) of B6, gzmA^-/-, and gzmB^-/- mice obtained after the third in vitro stimulation were used.

Blood was taken from 4 healthy donors (2 men, 2 women) and 4 patients. The patients were from the University Hospital Heidelberg, Germany. Patient 1 was a 28-year-old woman with Still syndrome, rheumatoid arthritis, acute inflammatory episode with fever, joint pains, and high C-reactive protein (CRP; 150 mg/L). Patient 2 was a 70-year-old woman with chronic lymphatic leukemia, fever, and tumor progression (CRP, 30 mg/L). Patient 3 was a 81-year-old man with B-non-Hodgkin-lymphoma with fever (no CRP increase, no infection). Patient 4 was a 56-year-old woman with acute progressive myeloid leukemia with fever, pneumonia, and sepsis (CRP, 86 mg/L). Cells from one healthy donor were used for cytospin analysis, cells of the other 3 donors were used for fluorescence-activated cell sorting (FACS) staining.

Blood was taken by venous puncture using EDTA (ethylenediaminetetraacetic acid)-coated tubes (Sarstedt, Nürnbrecht, Germany), and cells were separated on Ficoll Paque gradient (Amersham Pharmacia Biotech, Freiburg, Germany) according to established protocols. The interphase containing the peripheral blood mononuclear cells (PBMCs) was removed and stored on ice. The pellet containing the PMNs and the erythrocytes were mixed with phosphate-buffered saline (PBS) and 3% Dextran (Amersham Pharmacia Biotech; diluted in PBS) was added 1:1 and mixed. After sedimentation of erythrocytes at room temperature for 20 minutes the supernatant (SN) was removed, cells were resuspended in buffer containing 155 mM NH₄Cl/10 mM KH₂CO₃/0.1 mM EDTA in H₂O, and incubated for 15 minutes on ice to lyse resting erythrocytes. After centrifugation the SN was discarded, and the pellet containing neutrophils was resuspended in RPMI.

Approval was obtained from the Medizinische Klinki 5, University Heidelberg institutional review board for these studies. Informed consent was provided according to the Declaration of Helsinki.

Mouse granulocyte populations were generated from bone marrow (BM). BM cells were rinsed from femur and tibia (needle 23G; Terumo, Roma, Italy) with MEM/1 × nonessential amino acids/1 mM sodium pyruvate/1 mM l-glutamine/10^-5 M 2-mercapto ethanol (MEM-complete [MEM-comp]), adjusted to 1 × 10⁶ cells/mL, and cultivated in a 24-well plate (MEM-comp + 20% horse serum [HS]; Cell Concepts, Umkirch, Germany; 1 mL/well; 37°C, 7% CO₂) in the presence of 5 ng/mL granulocyte colony-stimulating factor (G-CSF; PeproTech, Rocky Hill, NJ) and 5 μM hydrocortisone (Sigma-Aldrich, Taufkirchen, Germany). After 5 days, 500 μL medium was replaced by 500 μL fresh MEM-comp + 20% HS for a further 4 days.

Mice were infected with 5 × 10³ colony-forming units (CFUs) Listeria monocytogenes (2.5 × 10⁴ CFU/mL; wild-type, strain EGD; PBS) or injected with 2 μg lipopolysaccharide (LPS; Salmonella abortus equi, smooth form; provided by C. Galanos, Max-Planck-Institut für Immunbiologie, Freiburg, Germany; 10 μg/mL in PBS) intraperitoneally. At indicated time points, the peritoneal cells were recovered by rinsing with PBS and stored on ice for further analysis.

Mice were infected with 1 × 10⁵ plaque-forming units lymphocytic choriomeningitis virus (LCMV)-WE intraperitoneally according to established protocols.^5,31  At day 8 after infection, spleens were removed for further analysis.

In vitro-enriched granulocytes (BM cells, day 9, G-CSF) were incubated in the presence of either LPS (Salmonella minnesota, rough form; provided by C. Galanos; 1 μg/mL) and recombinant lipoprotein outer surface protein A (Lip-OspA; 10 μg/mL; GlaxoSmithKline Beecham, Rixensart, Belgium) in MEM-comp + 20% HS for 8 hours or with 50 μg/L × 10⁵ cells zymosan A (Sigma-Aldrich) in MEM-comp + 20% HS for 90 minutes.

Mouse splenocytes were activated with 5 μg/mL ConA (Amersham Pharmacia Biotech) with or without 10% IL-2 (ConA SN) in MEM (10% FCS/10^-5 M 2-mercaptoethanol) for 2 to 3 days.

Cells were washed with PBS, and 1 × 10⁵ cells/150 μL PBS were centrifuged on microscope slides in a Cytospin 3 (Shandon Thermo Electron, Pittsburgh, PA). The cells were incubated with 150 μL May-Grünwald (Merck, Darmstadt, Germany) for 2 minutes, the same volume of H₂O was added for a further 3 minutes. After washing the slides twice with H₂O, cells were stained with Giemsa (diluted 1:20 in H₂O; Sigma-Aldrich) for 20 minutes, washed, and embedded with Entellan (Merck). The images were taken using a Zeiss Axioskop 10 microscope, a Zeiss Axiocam as analysis camera, and Zeiss Vision 3.0 as software (Carl Zeiss, Jena, Germany). The objective used was a Zeiss PlanNeofluor, with a magnification of 40 × and a numeric aperture of 0.75.

All monoclonal antibodies (mAbs) for surface staining, except anti-(α) CD4 mAb, αCD8 mAb (both from Coulter, Marseille, France; used for the healthy individuals 2 and 3), and αCD66b mAb (Coulter, Marseille, France) were obtained from BD Pharmingen, Heidelberg, Germany. Splenocytes, mouse granulocytes, and human PBMCs and PMNs were washed with FACS-washing buffer (PBS, 5% FCS, 0.1% NaN₃) as previously described³¹  and incubated with 10 μL mAbs for 20 minutes. αMouse (m) mAb used were fluorescein isothiocyanate (FITC)- and phycoerythrin (PE)-labeled αB220 (clone RA3-6B2), PE-labeled αCD8 (clone 53-6.7), FITC- and PE-labeled αGr-1 (clone RB6-8C5), FITC-labeled αMac-1 (clone M1/70), FITC- and PE-labeled αNK1.1 (clone PK136), and FITC- and PE-labeled αThy1.2 (clone 53-2.1). All fluorescence-conjugated αm mAbs were diluted in an αFc receptor antibody (clone 2.4G2), 1:50 for FITC-labeled mAb and 1:100 for PE-labeled mAb. αHuman (hu) mAbs used were FITC-labeled αCD4 (clone RPA-T4), PE-labeled αCD8 (clone RPA-T8), and FITC-labeled αCD66b (clone 80H3; Coulter) for all patients and the healthy individual 1; FITC-labeled αCD4 (Coulter), PE-labeled αCD8 (Coulter), and FITC-labeled αCD66b (clone G10F5) for the healthy individuals 2 and 3. The αhu mAbs were diluted in washing buffer. Stained cells were washed twice and fixed in 100 μL PBS containing 1% paraformaldehyde (PFA), examined in a FACSCalibur (Becton Dickinson), and analyzed with CellQuest software (Becton Dickinson, Franklin Lakes, NJ).

For analysis of intracellular gzmA and gzmB, 1.3E6SN, EL4.F15, splenocytes from LCMV-immune mice (day 8 after infection), mouse granulocytes, and human PBMCs or PMNs were stained with surface markers as described in “Flow cytometry” and then fixed with 100 μL PBS containing 2.5% PFA for 15 minutes at 4°C. Subsequently, cells were incubated in 100 μL permeabilizing buffer (PBS, 5% FCS, 0.1% NaN₃, 0,1% Saponin; Roth, Karlsruhe, Germany) for 10 minutes at 4°C, then stained for 45 minutes at 4°C with 40 μL rabbit αgzmA immune serum (IS) (diluted in permeabilizing buffer) or with allophycocyanin (APC)-labeled mouse αgzmB mAb (clone GB12; Caltag, Burlingame, CA; diluted in permeabilizing buffer), and subsequently washed with permeabilizing buffer; in the case of gzmA, cells were stained with 40 μL FITC-labeled IS goat αrabbit immunoglobulin G (IgG) as secondary Ab (Jackson Dianova, West Grove, PA; diluted in permeabilizing buffer; 1:500). As isotype controls, rabbit IgG (Jackson Dianova) for gzmA and APC-labeled mouse IgG (BD Pharmingen) for gzmB were used. Stained cells were washed twice in permeabilizing buffer, fixed in 100 μL PBS containing 1% PFA, examined in a FACSCalibur, and analyzed with CellQuest software.

For cell sorting, the enriched granulocytes (BM, day 9, G-CSF) were washed with FACS-washing buffer and incubated with 50 to 100 μL FITC/PE-labeled αGr-1 mAb (diluted 1:50/1:100 in αFc receptor SN [clone 2.4G2]) 20 minutes at 4°C. Cells were washed in PBS, homogenized with a separation filter (Miltenyi Biotec, Bergisch Gladbach, Germany), and positively sorted with the cell sorter MoFlo (Cytomation, Freiburg, Germany).

Cell lysates from 1.3E6SN, EL4.F15, positively sorted (αCD8 Microbeads; Milteny, Biotec; according to manufacturers instruction) CD8⁺ LCMV-immune splenocytes (day 8 after infection) and from in vitro-enriched granulocytes (BM, day 9, G-CSF) were prepared in 0.1% Triton X-100/10 mM Tris [tris(hydroxymethyl)aminomethane]-HCl in H₂O, pH 7.5 (1 hour on ice) and tested as described³⁰  on the following substrates: Ac-Ile-Glu-Pro-Asp-pNA (gzmB; Bachem, Weil am Rhein, Germany), Suc-Phe-Pro-Phe-pNA (CathepsinG; Bachem). The absorbance was examined in a spectrophotometer SpectraMax 190 (Molecular Devices, Ismaning/Munich, Germany; wavelength 405 nm/690 nm) and analyzed with SOFTMax Pro software (Molecular Devices).

In vitro-enriched granulocytes and 1.3E6SN cells were fixed in 4% PFA in 0.2 M PHEM buffer (240 mM PIPES [Piperazine-1,4-bis(2-ethanesulfonic acid)]/100 mM HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid]/40 mM EGTA [ethyleneglycoltetraacetic acid]/8 mM MgCl₂, pH 6.), stored in 0.5% PFA in 0.1 M PHEM buffer, and then processed for ultrathin cryosectioning as previously described.³²  Cryosections (45 nm) cut at 125°C using diamond knives (Drukker, Cuijk, The Netherlands) in an ultracryomicrotome (Leica Aktiengesellschaft, Vienna, Austria) were transferred with a mixture of sucrose and methylcellulose onto formvar-coated copper grids.³³  The grids were placed on 35-mm Petri dishes containing 2% gelatin. For double immunolabeling, a published procedure³⁴  was followed with 10- and 15-nm protein-A-conjugated colloidal gold probes, using αmgzmA mAb.³⁰  After immunolabeling, the cryosections were embedded in a mixture of methylcellulose and uranyl acetate and examined with a Philips CM 10 electron microscope (Philips, Eindhoven, The Netherlands). Images were captured on negatives, and prints were scanned and processed with Adobe Photoshop (Adobe, San Jose, CA). Original magnification, 12 500 ×.

Intracellular gzmA was determined in cell lysates from magnetic-activated cell sorting (MACS)-sorted LCMV-immune CD8⁺ cells of B6, gzmA^-/-, and gzmB^-/- mice by Western blotting under nonreducing conditions. The αmgzmA mAb (clone 7.1) was generated as decribed,³⁰  the αmβ-actin mAb (clone C-11) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Then the blot was stained with peroxidase-conjugated secondary antibodies, goat αrat or rabbit αgoat, respectively, from Jackson Dianova. Subsequently, a peroxidase substrate for enhanced chemiluminescence (ECL) from Pierce (Rockford, IL) was used for detection. Purified gzmA protein (50 μg/mL)³⁵  was used as positive control.

Total RNA was extracted from up to 5 × 10⁶ cells, using the QIAshredder spin columns, the RNeasy Mini Kit, and the RNase-free DNase Kit (all from Qiagen, Hilden, Germany) according to the manufacturer's instructions. mRNA was transcribed by incubating total RNA with random hexamer primer (660 ng/μL; Pharmacia, Freiburg, Germany), RNasin inhibitor (20 U; Promega, Madison, WI), deoxyribonucleoside triphosphate (0.5 mM; Qbiogene, Heidelberg, Germany), and Omniscript reverse transcriptase (4 U; Qiagen) as advised by the manufacturer. Reverse transcription was performed in a Thermocycler PTC-200 DNA-Engine (MJ Research, Waltham, MA). The transcription reaction profile was as follows: 37°C for 60 minutes and 70°C for 10 minutes. The resulting cDNAwas used as a template for RT-PCR hypoxanthin:guanine phosphoribosyl-transferase (Hprt1), Prf1, Gzma, and Gzmb amplification. Polymerase chain reaction (PCR) for Hprt1, Prf1, Gzma, and Gzmb was carried out in a Thermocycler PTC-200 as described in “Probing for mRNA transcription.” The PCR reaction profile for Hprt1, Prf1, Gzma, and Gzmb was as follows: 1 cycle at 94°C for 2.5 minutes as an initial denaturation step; then denaturation at 94°C for 20 seconds; annealing at 56°C for 20 seconds for Hprt1 or 55°C for 20 seconds for Prf1, Gzma, and Gzmb; extension at 72°C for 20 seconds for Hprt1 or 72°C for 30 seconds for Prf1, Gzma, and Gzmb (35 cycles); followed by further incubation for 3 minutes at 72°C (1 cycle). The PCR reaction profile for the cathepsin G gene (Ctsg) was as follows: 1 cycle at 95°C for 1 minute as an initial denaturation step, then denaturation at 94°C for 30 seconds, annealing at 48°C for 30 seconds, extension at 70°C for 1 minute (35 cycles), followed by further incubation for 8 minutes at 70°C. The following primers were used for amplification: Hprt1-specific primers, sense (5′-GCT GGT GAAAAG GAC CTC T-3′) and antisense (5′-CAC AGG ACT AGA ACA CCT GC-3′ amplifying a 249-base pair (bp) segment; Prf1-specific primers, sense (5′-GAG CCC CTG CACACATTACTG GAA-3′) and antisense (5′-ACA TTC TCA AAG TCC ATC T-3′) amplifying a 380-bp segment; Gzma-specific primers, sense (5′-GGG GAT CTA CAA CTT GTA CGG-3′) and antisense (5′-ATT GCA GGA GTC CTT TCC ACC AC-3′) amplifying a 291-bp segment; Gzmb-specific primers sense (5′-TCA GGC TGC TGA TCC TTG ATC G-3′) and antisense (5′-ATG AAG ATC CTC CTG CTA CTG C-3′) amplifying a 135-bp segment; Ctsg-specific primers, sense (5′-CAT CCA AAT GCG AGA AAG GAC-3′) and antisense (5′-CAT CTG CAC TCT TAG CTG CAC-3′) amplifying a 272-bp segment. The primers for Hprt1, Gzma, Gzmb, and Prf1 were synthesized by Hermann (Freiburg, Germany), for Ctsg by Thermo Electron (Ulm, Germany). The PCR products were analyzed by gel electrophoresis (2% agarose), stained with ethidium bromide, and analyzed using an UV-transluminator (Vilber Lourmat, New Ark, NJ). As size markers, 1-kilobase (kb) DNA marker was used (BioLabs, Frankfurt/Main, Germany).

Granulocytes were generated from mouse BM by in vitro cultivation with G-CSF (day 9) and further enriched by FACS sorting with mAb to Gr-1. This protocol leads to a highly purified granulocyte population (Figure 1A): freshly isolated BM cells consisted of approximately 50% Gr-1⁺, 3% NK1.1⁺, 44% B220⁺, 6% Thy1.2⁺, and 48% Mac-1⁺ cells (Figure 1Ai), with approximately 94% of Gr-1⁺ cells staining in addition for Mac-1 and approximately 2% to 3% for NK1.1, B220, or Thy1.2. The cell population propagated in G-CSF consisted of greater than 93% Gr-1⁺Mac-1⁺ (Figure 1Aii) and on further enrichment by FACS of greater than 99% Gr-1⁺Mac-1⁺ with only a minority (∼ 1%) also staining for NK1.1, B220, or Thy1.2 (Figure 1Aiii). Cytospin analysis revealed that the majority of Gr-1-sorted cells are granulocytes (Figure 1B, BM-derived versus in vitro-propagated and FACS-sorted cells). The majority of granulocytes generated under these conditions are Gr-1⁺Mac-1⁺, with only few, if any, expressing markers for NK, B, and T cells. However, the concern that the Gr-1 marker is not restricted to mouse granulocytes^36,37  (and information provided by manufacturer) is supported by phenotypic data obtained with ex vivo-derived or in vitro mitogen-stimulated spleen cells. From the approximately 7% of Gr-1⁺ cells detected in ex vivo-derived spleen cells (day 8 after infection), approximately 16% to 20% also stained for NK1.1 or B220 and greater than 75% for Thy1.2 (Figure 1C). Furthermore, spleen cell populations sensitized with ConA in vitro to enrich for activated T cells contained approximately 15% Gr-1⁺ cells, of which approximately 2%, approximately 35%, and approximately 60% also stained for NK1.1, B220, and Thy1.2, respectively. When the same cell population was cultured in the presence of ConA plus ConA SN, supporting polyclonal T-cell expansion, from the approximately 32% Gr-1⁺ spleen cells recovered, approximately 17% and approximately 90% also stained for B220 and Thy1.2, respectively. These observations are critical for a meaningful evaluation of data regarding expression of gzms and perf in mouse granulocytes and underscore the necessity for a correlative phenotypic analysis of the Gr-1⁺ cells, in particular for NK and T-cell-associated cell surface markers, so as to exclude gzm/perf-expressing contaminating cells.

In vitro-enriched BM-derived granulocytes were analyzed by FACS for intracellular expression of gzmA and gzmB by using αmgzmA IS or αhugzmB mAb, the latter is known to crossreact with mouse gzmB.³⁸  The specificity of both antibodies for the respective gzm was verified by using the gzmA/gzmB-positive CTL line, 1.3E6SN, and the gzmA/gzmB-negative thymoma line, EL4.F15, as well as ex vivo-derived LCMV-immune spleen cells (day 8 after infection) from B6, gzmA^-/-, and gzmB^-/- mice. As expected, only 1.3E6SN, but not EL4.F15, cells stained with αmgzmA IS and αhugzmB mAb (Figure 2A). In addition, αmgzmA IS stained a fraction of CD8⁺ T cells from LCMV-immune B6 and gzmB^-/- but not gzmA^-/- mice and, conversely, αhugzmB mAb stained a fraction of CD8⁺ T cells from LCMV-immune B6 and gzmA^-/- but not gzmB^-/- mice (Figure 2Bi). The lower intensities seen with T cells stained for gzmB compared with gzmA may be due to the fact that a monospecific mAb rather than a polyspecific IS was used in the former case. However, using in vitro-propagated alloreactive T-cell lines, known to express much higher levels of both gzms, it is clearly shown that both, αmgzmA IS or αhugzmB mAb, are specific and differentially stain the respective cell populations from B6, gzmA^-/-, and gzmB^-/- mice (Figure 2Bii). Most importantly, none of the in vitro-enriched granulocyte populations from B6, gzmA^-/-, and gzmB^-/- mice stained with αmgzmA IS and/or αhugzmB mAb (Figure 2Ci), even after earlier activation with either LPS or the lipoprotein Lip-OspA (Figure 2Cii; only shown for B6 granulocytes). Similar results were obtained with in vitro-enriched B6 granulocytes previously stimulated (90 minutes) with zymosan (Figure 3). The combined data indicate that in vitro-generated mouse granulocytes do not express gzmA and gzmB, even after activation and expansion.

The data obtained by FACS analysis were verified by IEM (Figure 4, only shown for gzmA; αmgzmA mAb; 7.1). Only specimens of 1.3E6SN but not EL4.F15 (data not shown) were heavily stained with αmgzmA mAb (7.1) in particular in cytoplasmic granules. In contrast, no or only few grains were seen in sections of in vitro-enriched mouse neutrophils or eosinophils. The specificity of the αmgzmA mAb (7.1) has been described before³⁰  and was verified again by WB analysis, using cell lysates from MACS-sorted CD8⁺ LCMV-immune cells from B6, gzmA^-/-, and gzmB^-/- mice as well as purified gzmA (Figure 4B).

Cell lysates from in vitro-enriched mouse granulocytes as well as from 1.3E6SN and EL4.F15 cell lines were tested for proteolytic activities of gzmB and cathepsin G, a granulocyte-specific protease,³⁹  by using appropriate chromogenic substrates. 1.3E6SN lysates contained gzmB- (19 U/10⁶ cells) but not cathepsin G-related activity (0 U/10⁶ cells), whereas EL4.F15 lysates did not harbor either of the 2 enzymatic activities (0 U/10⁶ cells). In addition, lysates from ex vivo-derived and enriched LCMV-immune CD8⁺ T cells from B6 (3 U/10⁶ cells) but not gzmAxB^-/- mice (0 U/10⁶ cells) expressed gzmB activity, although at much lower levels (3 U versus 19 U/10⁶ cells, as expected from previous studies; M.M.S. unpublished data, February 2004) when compared with the long-term cultured T-cell line, 1.3E6SN. Both ex vivo-derived T-cell populations expressed low but significant levels of cathepsin G or cathepsin G-like activity (1.3 U/10⁶ cells). In contrast, lysates from in vitro-enriched granulocytes, unstimulated or LPS-stimulated, expressed only cathepsin G-related (17 U/10⁶ cells), but not gzmB-related (0 U/10⁶ cells), enzymatic activities.

B6 mice were injected intraperitoneally with LPS or infected intraperitoneally with L monocytogenes. Twenty hours and 48 hours after infection, respectively, cells were recovered from the peritoneal cavity and analyzed by FACS for surface phenotype and the expression of intracellular gzmB. Approximately 45% of cells from LPS-challenged mice stained Gr-1⁺, although with differential intensity. Only approximately 3% to 4% of cells expressing Gr-1 were also positive for NK1.1 and Thy1.2, and mainly in the Gr-1^low cell population. When tested for the expression of gzmB, only few Gr-1^hi but significantly more Gr-1^low cells stained for gzmB. In contrast to Gr-1⁺ cells, a much higher proportion from Thy1.2⁺ cells (∼ 13%, total) and NK1.1⁺ cells (∼ 3%, total), expressed gzmB (∼ 10% and 20%, respectively) (Figure 5Ai). As expected, none of the ex vivo-derived LPS-sensitized Gr-1⁺ cells from gzmB^-/- mice expressed gzmB (Figure 5Aii).

The peritoneal cell population from L monocytogenes-infected mice contained approximately 34% Gr-1⁺ cells, with approximately 74% and approximately 6% being also positive for NK1.1 and Thy1.2, respectively (Figure 5B). However, none of the Gr-1⁺ cells expressed gzmB, whereas approximately 4% of the NK1.1⁺ cells (∼ 43%, total) and approximately 9% of the Thy1.2⁺ cells (∼ 22%, total) were positive for gzmB. Similar results were obtained for gzmA (data not shown). Together, these data indicate that ex vivo-derived Gr-1⁺ mouse granulocytes do not express gzmA and gzmB, even during infection or on challenge with pathogen-derived agents (LPS).

In vitro-enriched and FACS-sorted Gr-1⁺ BM-derived granulocyte populations from B6 and gzmB^-/- mice were assayed by RT-PCR for the expression of Gzma, Gzmb, and Prf1-specific transcripts. None of the 2 cell populations expressed Gzma, Gzmb, or Prf1-specific mRNA, irrespective of whether they were stimulated with LPS or left untreated (Figure 6A). As expected, mRNA from 1.3E6SN, but not EL4.F15, cells expressed transcripts for Gzma, Gzmb, and Prf1. In contrast, when in vitro-enriched but unsorted granulocyte populations were used, Gzmb-specific transcripts were detectable, the level of which was enhanced on pretreatment with LPS (Figure 6B; lanes 2 and 3). No Gzm- or Prf1-specific transcripts were seen in in vitro-generated and FACS-enriched Gr-1⁺ cells (Figure 6B; lane 1). In contrast, specific bands for the granulocyte-specific Ctsg were only detected in samples of granulocytes, but not in 1.3E6SN and EL4.F15 (Figure 6B). Similar results were obtained when mRNA preparations were subjected to quantitative RT-PCR using the light cycler system (data not shown).

In summary, these results clearly demonstrate that mouse granulocytes, even after in vitro activation or on encounter with pathogens in vivo, do not express gzmA, gzmB, or perf. The present data obtained by cytologic, intracellular flow cytometric, enzymatic, IEM, and semiquantitative RT-PCR analyses, support earlier studies, indicating that in mice the 3 cytolytic effector molecules, gzmA, gzmB, and perf, are produced mainly by NK and CTL,^12,15,28  and to some extend also by metrial gland cells,¹⁴  but not by other cells of the lymphoid or myeloid lineages.

In light of the published conflicting results on the expression of gzms and perf in human neutrophils,^18,19,26,27  we tested a panel of PBMC and PMN populations from 4 healthy individuals (cells of 1 individual were used for cytospin, cells of the other 3 individuals were stained intracellularly) and 4 patients with either an autoimmune disorder or B-cell chronic lymphatic leukemia, B-cell non-Hodgkin lymphoma, or acute myelogenous leukemia. PBMCs and PMNs were separated by standard methods and subsequently applied to cytospin and FACS analysis. Both cell populations were stained for surface markers CD4, CD8, and CD66b as well as intracellularly for gzmB using αhugzmB mAb (GB12). PBMCs from the healthy individual consisted of approximately 44% CD4⁺ and approximately 20% CD8⁺ T cells, whereas PBMCs from the 4 patients had CD4⁺/CD8⁺ T-cell ratios, ranging from approximately 74%:6% (patient 4) to approximately 30%:30% (patient 3), respectively (Figure 7A). PBNs from all individuals consisted of greater than 98% CD66b⁺ cells with no contaminants of CD4⁺ or CD8⁺ T cells. The polymorphonuclear nature of CD66b-enriched cells of a healthy individual was also revealed by cytospin analysis in an independent experiment (Figure 7B). When tested for the intracellular expression of gzmB, a significant number of CD8⁺ PBMCs (ranging from 16% to 71%) but also a fraction of CD4⁺ PBMCs (ranging from 0.5% to 17%) stained positive for the enzyme. In contrast, none of the PBN populations from the healthy individuals and the 4 patients expressed intracellular gzmB.

Our findings that only human PBMCs, in particular fractions of CD4⁺ and CD8⁺ T cells, but none of the enriched PMN populations from the healthy individuals and 4 patients with various disease symptoms expressed gzmB are in line with 2 recent studies,^26,27  but challenge 2 others.^18,19  Final proof whether human PMNs are able to produce gzms and/or perf under certain conditions is still pending. A cautionary note regarding proof of principle needs to be added to avoid erroneous conclusions. This pertains, in particular, to enrichment of granulocytes, via mAb: We found that the majority of cytologically defined mouse granulocytes are Gr-1⁺, but that they may differentially express NK- and T-cell-specific markers, depending on their activation state. However, fractions of ex vivo-derived spleen cells or spleen cells, previously activated in vitro with ConA or ConA plus ConA SN were double positive for either NK1.1 and Gr-1 or Thy1.2 and Gr-1 (Figure 1C).

Phenotype analysis of human PBMCs and PMNs led to similar observations. Thus, highly enriched PBMC populations contained small fractions of CD56⁺ or CD3⁺ cells, which also expressed CD66b and, vice versa, highly enriched PMNs (> 95% CD66b) contained a small fraction of cells, which expressed either CD56 or CD3 in addition to CD66b. This also holds true when another marker, CD15, mainly associated with human PMNs,^18,26  was applied (data not shown). Thus, granulocyte populations/PMNs selected solely by Gr-1 (mouse) or CD66b/CD15 (humans) may either contain granulocyte/neutrophil subpopulations expressing NK/T-cell markers or may be contaminated with cells from other lineages, in particular NK, NKT (natural killer T), and T cells, which are the main producers of gzms and perf.

The data presented here provide strong supportive evidence that mouse granulocytes do not express gzmA, gzmB, and perf under physiologic or pathophysiologic conditions. This is supported by the lack of Prf1- and Gzma-/Gzmb-specific mRNA and the respective proteins in in vitro-generated or ex vivo-derived granulocyte populations, as revealed by cytologic, intracellular flow cytometry, enzymatic, IEM, and RT-PCR analyses. Gzms and perf could not be detected in granulocytes previously activated either in vitro and/or in vivo, by L monocytogenes or by pathogen-derived material, such as LPS, Lip-OspA, or the yeast product, zymosan. This indicates that mouse granulocytes do not use the known NK/CTL-associated cytotoxic effector molecules, gzmA, gzmB, and perf, as defense against pathogens and tumors. Our finding that fractions of CD4⁺ and CD8⁺ T cells, but not highly purified PMNs from 4 patients and 3 healthy individuals, expressed gzmB is in support of recent work from 2 independent groups,^26,27  but contrasts with other related studies.^18,19  Thus, it remains to be verified whether the discrepant results obtained using human PMNs are merely due to technical differences and whether mouse and/or human PMNs or fractions thereof are able to produce gzms and/or perf, but only under stringent conditions unlike the expression in NK/CTLs.

We thank Jero Calafat from The Netherlands Cancer Institute, Division of Cell Biology, Amsterdam, for performing the IEM experiments and for her critical comments, and Verena Weber, Claudia Luckner, Christiane Brenner, and Aynur Ekiciler for expert technical assistance.