We also examined the expression of FasL protein in thyroid follicular cells with the use of immunohistochemical staining (8) and Western blotting (9) techniques similar to those employed by Giordano et al. (1). Immunostaining of normal thyrocytes with a polyclonal, rabbit antibody to FasL did not detect FasL protein (10), but, consistent with the other studies, these cells demonstrated significant amounts of Fas antigen staining (2-4). Unexpectedly, a protein immunoblot of thyroid cell lysates performed with the mouse monoclonal antibody used by Giordano et al. (clone 33, Transduction Laboratories, Lexington, Kentucky) yielded an intense band at the appropriate size for FasL (Fig. 2). However, there was no difference in intensity of this band in lysates of PMA stimulated as opposed to untreated Jurkat cells, and this finding did not correlate with the changes in mRNA concentration observed in these cells (Fig. 1, A and B). This result brings into question the specificity of the clone 33 antibody for FasL.

Although there may be differences in the tissues examined by ourselves and Giordano et al., we have not observed the expression of mRNA for FasL in primary cultured thyrocytes from over 20 normal and thyroiditis tissue samples, unless there was also evidence of mRNA for rearranged immunoglobulin genes (10). The latter result suggests that, in those situations, the message for FasL came from lymphocyte contamination of the thyroid cells. More importantly, FasL-induced programmed cell death in thyroiditis is questioned by our recent finding that the Fas pathway in thyroid follicular cells is blocked by a labile protein inhibitor (5). It has also been found that the in vitro induction of the Fas pathway with soluble ligand or antibody may be less efficient than that achieved by cytotoxic T cells (11). Together, these considerations make it difficult to predict the relative importance of Fas-mediated apoptosis in thyroiditis. We hope that the findings presented in this comment will promote the research and discussion necessary to clarify the potential role of the Fas pathway in the pathogenesis of thyroiditis.

Theophil A. Stokes
Michal Rymaszewski
Patricia L. Arscott
Su He Wang
James D. Bretz
Jeffery Bartron
James R. Baker, Jr.
Division of Allergy, Department of Medicine,
University of Michigan Medical School,
Ann Arbor, MI 48109-0666, USA

REFERENCES AND NOTES

1. C. Giordano et al., Science 275, 960 (1997) .
2. L. J. Hammond et al., J. Pathol. 182, 138 (1997) [CrossRef] [Web of Science] [Medline] .
3. C. Tanimoto et al., Endocr. J. 42, 193 (1995) [Web of Science] [Medline] .
4. A. Kawakami, et al., Endocrinology 137, 3163 (1996) [Abstract] .
5. P. L. Arscott et al., ibid. 138, 5019 (1997) .
6. For RT-PCR, RNA was isolated with the use of Tri Reagent (Molecular Research Center, Cincinnati, OH). One microgram of total RNA was used in a first-strand cDNA synthesis with 100 ng of oligo(dT)18 and amplified by PCR. PCR was performed at 94°C for 2 min followed by 30 to 34 cycles of amplification. Each cycle consisted of 35 s of denaturation at 94°C, 35 s of annealing at 58°C, and 45 s for enzymatic primer extension at 72°C. After the final cycle, the temperature was held at 72°C for 10 min to allow re-annealing of the amplified products. PCR products were then size-fractionated through a 2% agarose gel and the bands visualized with the use of ethidium bromide. A pair of primers was used to amplify human -actin as a control. Primer sequences: FasL: forward primer: 5-ACAACCTGCCCCTGAGCC-3; reverse primer: 5-AGTCTTCCTTTTCCATCCC-3; -actin: forward primer: 5-CACGGCATTGTAACCAACTG-3; reverse primer: 5-TCTCAGCTGTGGTGGTGAAG-3.
7. For ribonuclease (RNase) protection, the RiboQuant MultiProbe RNase Protection Assay System (Pharmingen, San Diego, CA) was used for the detection and quantitation of specific mRNA species. ³²P-labeled antisense RNA probes were prepared with the use of the Human Apoptosis hAPO-3 Template Set (Pharmingen), which included Fas, FasL, and human glyceraldehyde 3-phosphate dehydrogenase (GAPDH). Probes were hybridized with 10 µg of RNA from treated thyrocytes, Jurkat cells, and peripheral blood lymphocytes. After hybridization, samples were subjected to RNase treatment followed by purification of RNase-protected probes. Protected probes were resolved on a 5% denaturing polyacrylamide gel.
8. Thyroid cells were grown on Falcon chamber slides (Becton Dickinson, Franklin Lanes, NJ) or on glass coverslips. For immunostaining, slides were washed twice with PBS, then fixed with methanol for 5 min at 4°C, briefly air-dried, placed into PBS and blocked with 5% normal goat serum. Slides were incubated with 1.0 µg/ml affinity purified rabbit antibody to Fas (C-20, Santa Cruz Biotechnology), antibody to FasL (Q-20, Santa Cruz Biotechnology), or rabbit antithyroglobulin (Dako Corp., Carpinteria, CA). After washing with PBS, slides were incubated with biotinylated Abs specific for rabbit IgG, followed by detection using an avidin-biotin complex detection kit with glucose oxidase substrate (Vectastain ABC-GO kit, Vector Labs., Burlingame, CA). Slides were briefly counterstained with eosin and mounted with permount (Fisher Scientific, Fair Lawn, NJ).
9. Thyrocytes were scraped from tissue culture dishes and collected by centrifugation at 200 g for 10 min. Cell pellets were lysed in cold lysis buffer (0.5% Triton X-100 in 50 mM tris, pH 7.6, 300 mM NaCl) containing protease inhibitors for 30 min on ice, then centrifuged at 15,000g for 20 min, and the Triton-soluble protein fraction was collected. Total protein concentrations were determined using BCA protein assay (Pierce Chemicals, Rockford, IL). As a positive control for expression of FasL, protein lysates were also prepared from Jurkat cells that were activated using PMA and ionomycin for 4 hours. Equal amounts of cell protein lysates were mixed with 2X sample buffer (4% SDS, 10% -mercaptoethanol, 20% glycerol in 0.125 M tris, pH 6.8), samples were boiled for 3 min, then separated by electrophoresis on a 12.5% SDS-polyacrylamide gel. Proteins were then electrophoretically transferred to nitrocellulose. Blots were first blocked in 5% milk in PBS with 0.02% sodium azide (PBS-A) for 2 hours at 25°C. Blots were incubated overnight at 4°C with mouse monoclonal anti-FasL (Clone 33, Transduction Laboratories, Lexington, KY). After washing with PBS with 0.02% sodium azide and 0.05% Tween-20 (PTA), blots were then incubated with anti-mouse IgG alkaline phosphatase conjugated antibody (both from Jackson ImmunoResearch Labs., West Grove, PA) diluted 1:2500 in 5% milk/PBS-A for 1 to 2 hours at 25°C. Blots were again washed and developed with a BCIP/NBT substrate.
10. T. A. Stokes et al., data not shown.
11. M. Tanaka et al., Nature Med. 4, 31 (1998) [CrossRef] [Web of Science] [Medline] .

Giordano et al. describe the constitutive expression of Fas ligand (CD95L) by thyrocytes that they isolated from the glands of patients with Hashimoto's thyroiditis or nontoxic goiter (1). On the basis of this unexpected result, Giordano et al. propose that the concommitant expression of Fas and its ligand induces programmed cell death of thyrocytes and that this might be a major pathological mechanism underlying many forms of hypothyroidism. To detect FasL expression, Giordano et al. used polymerase chain reaction (PCR), immunohistochemical stainings, and FACS analysis. For the latter experiments, they used two commercially available antibodies against FasL, C-20 and mAb33. C-20 (Santa Cruz Biotechnology, Santa Cruz, California) is a rabbit polyclonal IgG antibody against an extracellular FasL epitope corresponding to amino acid residues 260 to 279, and it is recommended for use in protein immunoblots and immunohistochemistry. Monoclonal antibody mAb33 (Transduction Laboratories, Lexington, Kentucky), is an IgG1 monoclonal antibody against the extracellular part (216-277) of human FasL. It is made for the study of human FasL by protein immunoblots, immunoprecipitation, and immunofluorescence.

With the use of monoclonal antibody 33 (mAb33), we analyzed FasL expression with protein immunoblots in a panel of human tumor cell lines covering different B cell, monocyte, and T cell lines. Unexpectedly, we found that all these cell lines express FasL, as shown by a single band on the blot at about 37 kD. To confirm these results, we tested these cell lysates with a FasL-specific rabbit polyclonal antiserum PE62 against an extracellular peptide of FasL (2). In contrast to the results obtained with mAb33, none of these cell lines now showed a FasL-specific signal (Fig. 1A).

We therefore examined further the FasL specificity of mAb33. We transiently transfected human 293T embryonic kidney carcinoma cells with a FasL expression vector that encodes human FasL that has an NH₂-terminal FLAG tag, and we performed a protein immunoblot, with the use of the FasL-specific antibodies mAb33, C-20, G247-4 (Pharmingen), and P62 (2), as well as M2 (Eastman Kodak, New Haven, Connecticut), which is specific for the NH₂-terminal FLAG tag (Fig. 1B).

With the use of mAbs C247-4 and M2, the FasL-transfected 293T cells showed FasL-specific signals that were absent in the untransfected control cells. Although the polyclonal rabbit antibodies C-20 and PE62 showed some background staining in untransfected 293T cells, they also revealed CD95L-specific signals in the lysates of the transfectants, as would be expected from earlier reports (2-4). In contrast, mAb33 detected a 37-kD signal (similar to the band in Fig. 1A) in both transfected and untransfected cells. To further characterize the specificity of mAb33, we immunoprecipitated FLAG-tagged FasL from lysates of transfected 293T cells with the use of the FLAG-specific antibody M2. The immunoprecipitates were tested by protein immunoblotting with G247-4 and mAb33. FasL expression was detected with G247-4, but not with mAb33. However, mA33 (but not G247-4) produced a strong 37-kD signal in the supernatant of the immunoprecipitate (Fig. 1C).

Thus, C-20, PE62, and G247-4--but not mAb33--seem to be suited for the analysis of CD95L expression by protein immunoblotting. However, mAb33 does not stain human FasL, but a different protein expressed in many cell types. It is therefore questionable whether the signals shown in figure 4 of the report by Giordano et al. correspond to FasL expressed by thyroid cells.

With the use of 293T cells transfected with FLAG-tagged FasL, we then analysed several antibodies (NOK-1, NOK-2 (5), and G247-4 (PharmIngen, San Diego, California), mAb33, C-20, and MIKE 2 (Alexis, San Diego, California) for FasL specific staining with the use of flow cytometry and immunoflorescence. Untransfected cells served as specificity controls. With the use of a fluorescent activated cell sorter (FACS), only NOK-1, NOK-2, and MIKE-2 detected FasL expressed on the 293T cell transfectants, whereas C-20, mAb33, and G247-4 did not show any specific signals (6). G247-4 is known to work only in Western blots, but C-20 was also reported by Giordano et al. to stain FasL-expressing thyrocytes with the use of flow cytometrical and immunohistochemical analyses. In the immunoflurescence studies, FasL-specific signals were easily detected with NOK-1, G247-4, and with the FLAG-specific antibody M2. The peptide-specific rabbit polyclonal antibody C-20, however, gave a high background staining already with untransfected cells. FasL transfectants showed stronger signals, but also a similar background staining similar to the controls. Both stainings were not seen when the blocking peptide was added (6).

On the basis of these results, we conclude that C-20 is suited for analyzing CD95L expression by protein immunoblotting, but not by flow cytrometry, immunofluorescence, or immunohistochemistry. Because C-20 detects in such blots a signal of about 65 kD, which does not correspond to FasL (Fig. 1B), it may also stain by immunofluorescence antigens unrelated to FasL that may bear epitopes similar to the peptide used to generate the antibody. Flow cytometrical experiments using C-20 may also reveal cells expressing such epitopes that may be absent in 293 T cells but not in thyroid cells. We therefore cannot recommend the application of C-20 in flow cytrometrical and immunofluorescence studies designed to show FasL expression.

In comparison to the other FasL-specific antibodies, mAb33 detects a different intracellular protein that seems to be expressed ubiquitously. Previous studies that used this antibody should be interpreted with caution. In conclusion, our studies describe a panel of antibodies that are specific for FasL and that may be used either in flow cytrometrical, immunofluorescence, or Western blot analysis. They also show that the two antibodies C-20 and mAb33 may bear additional specificities or might not be specific for FasL. Because both antibodies are commercially available, they seem to be used frequently by many investigators. We suggest that results from such studies (1, 7) should be reinterpreted keeping in mind the specificity of these antibodies, or repeated with the use of reagents that are known to be specific for FasL.

Petra Fiedler
Christian E. Schaetzlein
Hermann Eibel
Clinical Research Unit for Rheumatology,
University Hospital Freiburg,
D-79106 Freiburg, Germany
E-mail: eibel{at}nz11.ukl.uni-freiburg.de

REFERENCES

1. C. Giordano et al., Science 275, 960 (1997) .
2. M. Hahne et al., Intern. Immunol. 7, 1381 (1995).
3. M. Hahne et al., Eur. J. Immunol. 26, 721 (1996) [Web of Science] [Medline] .
4. M. Hahne et al., Science 274, 1363 (1996) .
5. N. Kayagaki et al., J. Exp. Med. 182, 1777 (1995) [Abstract/Free Full Text] .
6. P. Fiedler and H. Eibel, data not shown.
7. G. A. Niehans et al., Cancer Res. 57, 1007 (1997) [Abstract/Free Full Text] .
8. We thank J. Tschopp and M. Hahne for providing the rabbit polyclonal antibody PE62 and for helpful discussions. Part of the work was supported by grant Nr. Ei235/4-1 to H.E.

Response: We appreciate the comments by Stokes et al. and Fiedler et al. about our earlier report (1). With regard to detecting FasL in primary thyrocyte cultures with RNA protection assay and PCR, such detection would be difficult if the thyrocytes were not freshly excised and immediately analyzed, because FasL expression is labile ex vivo (mRNA or protein).

Concerning Fas expression, Stokes et al. used thyrocytes from controlateral lobes of thyroid cancers as their "normal" controls (2). We also observed variable Fas expression in uninvolved thyrocytes from thyroids with cancer, but we did not use such samples for normal controls. We obtained more normal samples from thyroid sections from patients undergoing laringectomy for laringeal cancer. These thyrocytes consistently expressed very low amounts of Fas, similar to thyrocytes from patients with nontoxic goiters (NTG), which we used in our report (1).

Prompted by concerns raised about the specificity of the polyclonal antibody to FasL C20 (Santa Cruz, Biotechnology, Santa Cruz, California) and mAb 33 (Transduction Laboratories, Lexington, Kentucky) (see below), we repeated an immunohistochemical study with the NOK-2 antibody (PharMingen, San Diego, California) on thyroid sections from laringectomy patients (above). Most thyrocytes from normal thyroids showed detectable FasL expression (3) (Fig. 1). In situ hybridization would further address this issue. Moreover, FACS analysis of ex vivo thyrocytes from NTG patients confirmed the constitutive expression of FasL on most thyrocytes, with the use of both the NOK-2 antibody and the H11 (Alexis, San Diego, California) antibody (Fig. 2A).

Finally, (i) thyroid tissue from patients with a laringectomy or NTG and (ii) hematopoietic cell-depleted thyrocytes from patients with NTG expressed FasL, as detected by Western blot analysis with 33 mAb (Transduction Laboratories) or with G247-4 mAb (PharMingen) (Fig. 2B). We detect a single band in our blots, similar to some investigators (4), while others detect multiple bands, which might represent glycosylated and unglycosylated forms (5).

Although we agree that C20 antibody may give a relatively high background signal as compared with other commercially available reagents, these new data support our earlier conclusion that normal thyrocytes express substantial amounts of FasL in vivo.

Fiedler et al. conclude that mAb 33 recognizes a protein different from FasL. We performed similar experiments with the use of three different cellular systems: (i) 293T cells transiently transfected with human FasL, (ii) COS-7 cells transiently transfected with human FasL (6), and (iii) NIH 3T3 stably transfected with murine FasL (7). To detect the expression of FasL by FACS analysis, we used NOK-1 mAb (PharMingen), C20, G247-4 mAb, and 33 mAb. Species and isotype-matched antibodies were used as control primary reagents. All of these antibodies gave a specific staining only in transfected cells (Fig. 3; only data on NIH 3T3 are shown), although with different distributions.

It is not clear why Fiedler et al. did not detect FasL expression by FACS analysis (with the use of the 33, C20, and G247-4 antibodies), or why the FLAG-FasL immunoprecipitate reacts with G247-4, but not with the mAb 33. We used a human FasL construct; they used a human FasL-NH₂-terminal FLAG construct. Perhaps the FLAG is not unharmful to some FasL epitopes.

Fielder et al. used untransfected cells as specificity controls for their FACS analysis. This procedure may lead to an underestimation of specific signals, as our data show that all the antibodies tested are crossreactive between human and murine FasL. Because many cell lines express low levels of FasL constitutively, or under certain culture conditions, specificity controls should be performed with isotype-matched antibodies and mock-transfected cells, as opposed to untransfected cells.

We also analyzed protein immunoblots of 293T cells transiently transfected with human FasL, COS-7 cells transiently transfected with human FasL, and NIH 3T3 stably transfected with murine FasL. To reveal FasL, we used mAb 33 and mAb G247-4. Both mAbs detected an ~37 kD band only in transfected cells (Fig. 4) and in PMA-treated, but not in PMA-untreated, Jurkat cells (8).

Fielder et al. find a uniform strong band reactive with the mAb 33 by protein immunoblot analysis in untransfected murine 293T cells, as well as in other human cell lines. As we detect mAb 33-reactive signals in transfected cells or in Jurkat only after PMA exposure, their finding is intriguing, and we have no explanation for it.

In conclusion, we find evidence that mAb 33 does recognize FasL in three different cell types, transiently or stably transfected with human or murine FasL, both by FACS and protein immunoblot analysis. Nevertheless, because of possible differences in specificity among the various available antibodies, we recommend the simultaneous use of several anti-FasL reagents. Constitutive FasL expression on in vivo and ex vivo normal thyrocytes has now been described by five different antibodies against FasL.

Giuliana Papoff
Department of Immunobiology,
Institute of Cell Biology,
National Research Council, (CNR),
00016 Rome, Italy
Giorgio Stassi
Laboratory of Immunology,
University of Palermo,
90100 Palermo, Italy
Ruggero De Maria
Department of Experimental Medicine and
Biochemical Sciences,
University of Rome "Tor Vergata,"
00133 Rome, Italy
Carla Giordano
Aldo Galluzzo
Laboratory of Immunology,
Institute of Clinical Medicine,
University of Palermo
Marcello Bagnasco
Department of Allergology
and Clinical Immunology,
University of Genova,
16100 Genova, Italy
Giovina Ruberti
Department of Immunobiology,
Institute of Cell Biology,
CNR, Rome
Roberto Testi
Department of Experimental
Medicine and Biomedical Sciences,
University of Rome "Tor Vergata"
E-mail: tesrob{at}flashnet.it

REFERENCES AND NOTES

1. C. Giordano, et al. Science 275, 960 (1997) .
2. P. L. Arscott et al., Endocrinology 138, 5019 (1997) [Abstract/Free Full Text] .
3. Thyroid fragments (0.5 cm) were snap-frozen in isopentane, chilled at 150°C, and kept at 80°C until used. Serial cryostat thyroid sections (4 µm) were allowed to equilibrate to room temperature and exposed to acetone for 10 min before starting the peroxidase anti-peroxidase staining. Bound mAbs anti-FasL (NOK-2, PharMingen) and isotype matched control IgG, were detected by the labeled streptavidin-biotin staining technique (Vecstain quick universal Kit, Vector, Labs, Burlingame, CA), with the use of a biotinylated antibody to mouse immunoglobulins prediluted in TBS at room temperature for 20 min, followed by incubation for 10 min at room temperature with an peroxidase-conjugated steptavidin. Primary antibodies were added, without washing, to the tissue preparations after incubation for 20 min with antibody to human serum. Binding was revealed by 3,3-diaminobenzidine tetrahydrochloride (DAB) colorimetric substrate for 2 min. Haematoxylin aqueous formula was used as a counterstain.
4. M. Hahne et al., Science 274, 1363 (1996) ; P. A. Kiener et al., J. Immunol. 159, 1594 (1997) [Abstract] .
5. M. Tanaka, T. Suda, T. Takahashi, EMBO J. 14, 1129 (1995) [Web of Science] [Medline] .
6. COS-7 and 293T cells were transiently transfected respectively by the DEAE-dextran and the Calcium-phospate methods with a pcDNA3FasL expression vector coding for a full-length human FasL or with a pcDNA3 empty vector. Transfection cells were stained 48 hours later with anti-FasL antibodies NOK-1 (PharMingen, Lot. M021271) (5 µg/ml), C20 (Santa Cruz Biotechnology, Lot. H145) (1 µg/ml), G247-4 (PharMingen, Lot M021163) (5 µg/ml), clone 33 (Transduction Laboratories, Lot 2) (5 µg/ml), or an isotype-matched control antibody, followed by a FITC-coupled sheep antibody to mouse or a FITC-coupled donkey antibody to rabbit Ig-conjugated antibody (Amersham). Cells were analyzed with a FACScan (Becton Dickinson).
7. D. P. M. Hughes and L. N. Crispe, J. Exp. Med. 182, 1395 (1995) [Abstract/Free Full Text] .
8. We also used aliquots of the transfected cell lines in protein immunoblot experiments. Briefly, cells were lysed (lysis buffer: 50 mM tris, pH 7.6, 150 mM NaCl, 0.5% NP-40, 2 µg/ml aprotinin, 2 µg/ml leupeptin, 1 mM PMSF) and an equal amount of proteins (50 to 100 µg) for each sample was loaded on a 12% SDS-PAGE, blotted against nitrocellulose filter, and analyzed with the CD95L antibodies clone 33 (0.3 µg/ml) or clone G247-4 (2 µg/ml) with the use of HRPO-coupled sheep antibody to mouse Ig and enhanced chemoluminescence (Amersham).