It is puzzling how autoreactive B cells that escape self-tolerance mechanisms manage to produce Abs that target vital cellular processes without succumbing themselves to the potentially deleterious effects of these proteins. We report that censorship indeed exists at this level: when the Ab synthesis in the cell is up-regulated in IL-6-enriched environments (e.g., adjuvant-primed mouse peritoneum), the cell dies of the increased intracellular binding between the Ab and the cellular autoantigen. In the case in which telomerase is the autoantigen, mouse hybridoma cells synthesizing such an autoantibody, which appeared to grow well in culture, could not grow in syngeneic BALB/c mice to form ascites, but grew nevertheless in athymic siblings. Culture experiments demonstrated that peritoneal cell-derived IL-6 (and accessory factors) affected the growth and functions of the hybridoma cells, including the induction of mitochondria-based apoptosis. Electron microscopy revealed an abundance of Abs in the nuclear chromatin of IL-6-stimulated cells, presumably piggy-backed there by telomerase from the cytosol. This nuclear presence was confirmed by light microscopy analysis of isolated nuclei. In two other cases, hybridoma cells synthesizing an autoantibody to GTP or osteopontin also showed similar growth inhibition in vivo. In all cases, Ab function was crucial to the demise of the cells. Thus, autoreactive cells, which synthesize autoantibodies to certain intracellular Ags, live delicately between life and death depending on the cytokine microenvironment. Paradoxically, IL-6, which is normally growth-potentiating for B cells, is proapoptotic for these cells. The findings reveal potential strategies and targets for immunotherapy.
When appropriately stimulated, B lymphocytes differentiate into plasma cells and produce Abs that arm the humoral immunity. Abs are normally produced against extrinsic agents such as viruses and allergens but sometimes, the responses are mounted against self-components. These autoreactive Abs (autoantibodies), which recognize a variety of cell surface and intracellular Ags, are found abundantly in the circulation of humans and animals inflicted with autoimmune disease (1). Whereas autoantibodies that bind to the readily accessible surface Ags of target cells can be patently pathogenic, much debate has been raised whether those directed to intracellular constituents are able to penetrate the living cell in the first place to cause harm. Proof of entry has been provided for isolated experimental instances (2, 3) but importantly, the intrinsic inhibitory potential of some of these Abs on cellular functions was demonstrated (4, 5).
How such potentially deleterious autoantibodies can be produced in the first place has not been questioned, but the mechanism must overcome two significant barriers. One is immunological self-tolerance, in which autoreactive T and B cells are normally eliminated by central and peripheral censoring mechanisms but which may be broken by newly generated T cell reactivation of normally quiescent autoreactive B cells (6, 7). Some autoreactive B cells are able to hide themselves by modulating their Ag-binding specificity through receptor editing (8). The second barrier against autoantibody production is at the molecular level: How do the cells, which produce autoantibodies that target vital physiological processes in the cell, such as those directed to GTP, which were found, for example, in high persistent amounts in a patient with systemic lupus erythematosus (9), manage to escape the potentially toxic effects of these Abs themselves? It can be argued that this action is possible because Igs are normally synthesized and transported in tight compartments separated from the rest of the cell. Thus, Ig H and L chains synthesized on separate ribosomes are released to the endoplasmic reticulum (ER)3 where they assemble into whole (H2L2) molecules with the help of ER chaperones and transported in COP II vesicles to the Golgi apparatus (reviewed in Ref. 10). In this organelle, they are processed for glycosylation and checked for correct folding and assembly. Mutant molecules are removed for degradation in the cytosol via the ER-Sec61 channel (11), whereas appropriately assembled proteins exit in vesicles to the cell membrane where they become embedded as membrane Ig or exteriorized as secreted Ig.
In this study, we reveal that the compartments for Ig synthesis and transport are not tight, and that cells that synthesize deleterious types of autoantibodies are actually deviant and are vulnerable to suicide depending on the amount of Abs produced. We base these observations on three examples, all are mouse hybridomas with an autospecificity for telomerase, GTP, or osteopontin (OPN). The suicide mechanism in the telomerase-specific hybridoma was elucidated. In this case, the target Ag, telomerase, is an important housekeeping enzyme without which, a cell eventually enters into senescence and dies (reviewed in Ref. 12). Telomerase replenishes the telomeric repeats that are lost at the ends of chromosomes after each cell division. The enzyme is composed of a catalytic protein subunit TERT (telomerase reverse transcriptase) and a telomerase RNA component TERC. TERT is synthesized on free ribosomes in the cytosol, and the nascent protein is bound by various chaperone proteins such as heat shock protein 90 and p23 (13), including 14-3-3 protein that directs TERT to the nucleus (14) where it assembles with TERC. We show that the Abs produced by this hybridoma can bind to TERT in the cytosol and thus transported to the nucleus. This activity is generally not lethal to the cell (the cell culture in fact appears normal), but when the Ab synthesis is up-regulated in environments enriched with IL-6, intrinsic apoptosis invariably ensues.
Materials and Methods
Cells, cell culture, and peritoneal cIL preparation
All cells were maintained in RPMI 1640 plus 10% FCS (Invitrogen). G50 was made by fusing Sp2/0 cells with cells of BALB/c mice immunized with GTP human serum albumin, whereas 659 and 446 were generated using NSO myeloma cells and BALB/c mice immunized with OPN-GST (see below) (15). HL-60, Sp2/0, AB1-2, CRL 1640, and HB 8609 were obtained from American Type Culture Collection. Resident peritoneal exudate cells obtained from adjuvant-primed normal BALB/c mice by peritoneal lavage were cultured at 4 × 106 cells/ml in RPMI 1640 plus 20% FCS for 48 h. The supernatant obtained (cIL) was filtered (0.22 μm) and the cytokine content determined using Mouse Cytokine Ab Array III (RayBio). Chemiluminescence signals were quantified by densitometry and normalized against biotinylated IgG control.
mRNA analysis, plasmid constructions, and cell transfections
An IgVH 336-bp fragment was synthesized from 476 total RNA by RT-PCR (16). A TERT 462-bp fragment (Motifs T-2) was similarly synthesized using the primers 5-CCCGGCCTTGAGCAATG-3 and 5-AGCAGGTCGTCGCCCACT-3. Both fragments were ligated separately to pPCR-Script Amp SK(+) (Stratagene), sequenced, and subcloned in reversed orientation (antisense) to pcDNA3 (Invitrogen). Plasmid DNA (20–40 μg) of selected clones was introduced to 476 cells (1 × 107 per ml) using Gene Pulser (Bio-Rad). Geneticin-selected clones were examined for expression of antisense mRNA by RT-PCR or for Ab secretion. Small interfering RNA transfection of 476 cells (5 × 104 per ml) was performed with 80 pmol oligonucleotides (Invitrogen) specific for TERT (5-CUUAAAGAAGUUCAUCUCGTT-3) or TERC (5-AGGAAAGUCCAGACCUGCATT-3) using Lipofectamine 2000 (Invitrogen). 476-M was obtained by limiting-dilution culture of 476 cells recovered from the ascites of a nude BALB/c mouse.
Cell lysates, Western blot analysis, nuclei preparation, and staining
Whole cell and cytosolic extracts were prepared by incubating 107 cells/ml with ice-cold Nonidet P-40 lysis buffer for 30 min and digitonin lysis buffer for 30 s, respectively (17). Protein content in extract was determined using the Bicinchoninic Acid Protein Assay kit (Pierce). Western blot analysis was performed on extracts (10–100 μg/ml) separated on 8–15% SDS-PAGE gels transferred onto polyvinylidene difluoride membrane (15).
Intact nuclei were prepared as described (18). Briefly, 107 cells/ml were lysed (10 min, room temperature) in NE buffer (0.01 M HEPES (pH 7.9), 0.0015 M MgCl2, 0.01 M KCl), and the nuclei pelleted by centrifugation at 3300 × g, 1 min. The nuclei were incubated (6 min, room temperature) in NE buffer containing 0.06% Triton X-100 with gentle intermittent vortexing, washed (1500 × g, 4 min) in 0.9 M sucrose and 0.1% formaldehyde in NE buffer, and finally fixed (30 min, room temperature, vortex) in 2% formaldehyde in PBS. The nuclei were washed (3300 × g, 10 min) and incubated (overnight, room temperature) with 4% formaldehyde in PBS. For immunostaining, the nuclei preparation was spotted on microscope slides and allowed to dry, then washed with PBS and incubated (15 min, room temperature) with 1% BSA in PBS. The preparation was incubated (30 min, room temperature) with FITC-labeled goat anti-mouse IgG (1/20 dilution; BD Pharmingen) and propidium iodide-1% BSA (1 μg/ml; BD Pharmingen), washed three times in PBS and mounted with aqueous mountant.
Ags and Ab (Ig) detection
TERT-GST (15) and phosphorylcholine-human serum albumin (19) were prepared as described. OPN-GST was similarly produced from HEp2 OPN gene (aa 1–175) cloned into the pGEX-2T vector (Amersham Biosciences). GTP or GMP (Sigma-Aldrich) were coupled to human serum albumin, BSA (bovine albumin), or MSA (mouse albumin) by periodate oxidation. Ags (0.5–2 μg/ml) were coated on Immunon-2 plates (Dynex) and used in ELISA (16) to detect secreted Abs from culture supernatants or mouse body fluids. In the inhibition ELISA, human or mouse OPN (Sigma-Aldrich) was coincubated (overnight, 4°C) with the indicator 659 mAb before development (90 min, room temperature) with peroxidase-labeled goat anti-mouse IgG (Invitrogen). Ab quantitation was based on Ab standards purified on protein G-Sepharose (Sigma-Aldrich) and quantified using the bicinchoninic acid kit (Pierce). Intracellular Ig in whole cell extract was detected by Western blot analysis using peroxidase-labeled goat anti-mouse Ig (BD Pharmingen) or mouse anti-β-actin (Sigma-Aldrich). Intracellular Ig in 1.8% formaldehyde-fixed 0.5% saponin-permeabilized cells was detected by flow cytometry (FACSCalibur, CellQuest software; BD Biosciences) using FITC-labeled anti-mouse μ (BD Pharmingen). Cell numbers were based on trypan blue-excluded cells after 24 h culture.
Telomerase and immune complex detection
Telomerase enzymic activity in whole cell extracts was measured using the TRAP ELISA kit (Roche) (15). Telomerase expression in cells was detected by 1) in situ hybridization using a TERT RNA probe, developed with nitroblue tetrazolium or 2) immunohistology, using biotinylated mAb 476 (15). TERT protein in whole cell extract was detected by Western blot analysis using biotinylated mAb 476 (15). Ab-bound TERT in cells was detected as follows: 1 × 107 per ml cells were cultured with or without cIL for 24 h, then in methionine-free medium for 30 min, and later in dialyzed medium (BioSource International) containing FCS (Invitrogen) and [35S]methionine (100 μCi/ml; Amersham Biosciences) with or without cIL for 3 h. Whole cell extract (500 μg) was incubated with protein G-Sepharose (100 μl) at 4°C for 1 h. The Sepharose beads were collected by centrifugation, washed, resuspended in loading buffer (0.3% SDS, 2.5% 2-ME, 5% glycerol, 0.025% bromphenol blue, 75 mM Tris-Cl (pH 6.8)) and heated (100°C, 5 min), and electrophoresed on 8% SDS-PAGE gel. The gel was fixed, dried, and autoradiographed.
Cells fixed in 2% paraformaldehyde 0.15% glutaraldehyde (15 min), embedded in 1% agar and impregnated with lowicryl HM20 resin, were processed as 90–120 nm sections. The mounted sections preblocked with 5% BSA Tris-borate buffer (30 min) were incubated with 10 nm colloidal gold-conjugated goat anti-mouse IgG (Dakopatts) at 4°C overnight. The sections were washed, fixed in 1% glutaraldehyde, counterstained, and viewed under a FEI/Philips Tecnai 12 BioTWIN electron microscope.
Studies on cell growth and death
Cells (2–5 × 104/ml) were incubated for 48–72 h with 2-6-bis[3-(N-piperidino)propionamido]anthracene-9,10-dione (PPA) (1.3 μM; Calbiochem). On occasions, the cells were 1) cocultured with anti-cytokine mAbs (10 μg/ml; R&D Systems) or 2) preincubated (1 h) with the following chemicals (Calbiochem): pan-caspases inhibitor I (Z-VAD-FMK, 100 μM), caspase-2 inhibitor (Z-VDVAD-FMK, 60 μM), caspase-8 inhibitor (Z-IETD-FMK, 100 μM), Bax-inhibiting peptide (H-VPMLK-OH, 100 μM), p53 inhibitor (pifithrin-α, 5 μM), or p38 MAPK inhibitor (SB203580, 10 μM).
Apoptosis was detected by 1) flow cytometry, using the Annexin V FITC Apoptosis Detection kit (BD Pharmingen); 2) Western blot analysis, using rabbit ab (Cell Signaling Technology) to various cleaved fragments, cytochrome c, poly(ADP-ribose) polymerase (PARP) and Bid, or mouse mAb to prohibitin (NeoMarkers) and actin (Sigma-Aldrich); or 3) DNA fragmentation assay (20). Expression of apoptosis-related genes was detected using the Mouse ApoptosisGene Array kit (SuperArray Bioscience). Signals obtained were normalized against those for β-actin.
Cell proliferative activity was measured using the MTT kit (Roche). Cell viability was estimated by counting trypan blue-treated cells in a hemocytometer. Cell-doubling time was estimated by counting the increase in viable cells after 48 h culture.
Normal and athymic nu/nu BALB/c mice 8–12 wk of age were i.p. injected with 0.5 ml of IFA (Sigma-Aldrich). After 1 wk, 5 × 106 hybridoma cells (0.5 ml) in PBS were injected and the animals examined regularly for ascites development by gross anatomy. Ascites formation was sometimes checked by examining blood samples for Ab activity. Results are expressed as the percentage of ascites formed based on the proportion of mice that produced ascites regardless of the amount of fluid produced. All procedures were approved by the Chinese University of Hong Kong Animal Experimentation Ethics Committee.
Results are presented as the mean ± SE of replicate experiments or of replicate determinations. Differences between groups were analyzed using the Wilcoxon rank test or ANOVA (PRISM, GraphPad software). Values for p < 0.05 were considered significant.
Hybridoma 476 produces an anti-telomerase autoantibody in culture but fails to grow as mouse ascites
Hybridoma 476 was derived from a normal BALB/c mouse hyperimmunized with a bacterially produced TERT Ag (15). The IgG1 Ab produced specifically reacted with the nuclear telomerase of human (HL-60) cells as well as murine hybridoma 476 and Mab2 cells in histological preparations (Fig. 1 A). Specificity of staining was demonstrated in various ways previously (15). The interspecies cross-reactivity between human and mouse is due to the high homology (69% identities, 82% positives) between these proteins in the selected region (www.ncbi.nlm.nih.gov).
Although hybridoma 476 grew as well in culture as other hybridomas, including Mab2 and the parental myeloma Sp2/0, the surprising finding was that it failed to grow repeatedly (8 attempts, 40 mice in total) in the adjuvant-primed peritoneum of normal syngeneic (BALB/c) mice. Thus, although all 40 mice inoculated with Mab2 cells formed ascites within 21 days, the 476-inoculated animals did not form ascites for as long as 4 mo and serum samples obtained from randomly selected animals showed absence of anti-TERT Ab activity. Hybridoma Mab2 is very similar to hybridoma 476 in having a BALB/c background and the same Sp2/0 fusion partner, including the fact that the Ab is an IgG1, but which is specific for phosphorylcholine (19). Coadministration of 476 cells or 476 Abs (150 μg/mouse) with Mab2 cells to adjuvant-primed mice did not affect growth of the Mab2 cells. Interestingly, however, when nude BALB/c mice were used, the 476 cells were able to grow and form ascites, though more slowly than Mab2 cells. Thus, 50% of the inoculated animals formed ascites after 9 days in the case of Mab2 (n = 20 mice), and 30 days for 476 (n = 21 mice).
Failure of hybridoma 476 to grow is related to Ab specificity
We reasoned that the problem with hybridoma 476 was related to the specificity of the Ab and accordingly, we abrogated the Ab activity in these cells. Based on the VH sequence (Fig. 1,B), antisense stable transformants were constructed. Two clones, 476-S1 and 476-S2, were obtained that did not secrete any detectable Ab in culture due to the absence of the H chain (Fig. 1,C) and the corresponding VH-sense transcript (Fig. 1,D). In addition, a naturally mutated clone (476-M) derived from 476 cells recovered from the ascites of a nude mouse also did not produce any Ab in culture due to the absence of the L chain (Fig. 1 C) and the corresponding mRNA (data not shown), but the exact defect is not known.
All three mutant clones were able to grow in normal mice and produced ascites. Thus, 50% of the inoculated animals formed ascites after 34 (476-M, n = 11 mice), 42 (476-S2, n = 10 mice) or 52 (476-S1, n = 10 mice) days. Mice inoculated with 476 cells transfected with empty vector only (n = 5 animals) or with unmutated cells recovered from nude mouse ascites that still produced the original Ab (n = 5 animals), as expected, failed to grow. Whereas no Ab could be detected in the ascites of 476-S2 and 476-M, some activity was found with 476-S1. However, culture of the 476-S1 cells recovered from the ascites yielded no Ab activity, suggesting incomplete blockade of VH sense transcripts in the peritoneal environment. All three mutant clones also grew in nude mice (n = 4 animals each), at similar rates to the parental cells.
Peritoneal cells produce IL-6 that can replicate the adverse effects of the peritoneal environment
The failure of 476 parental cells to grow in the normal mouse peritoneum is probably due to apoptosis, as revealed by Annexin V FACS analysis. Thus, the number of apoptotic cells increased from 10.8 ± 2.5% at 24 h postinoculation to 16.2 ± 1.8% after 72 h (n = 3 mice). To elucidate the basis of the apoptosis, we performed in vitro studies using peritoneal exudate cells harvested from adjuvant-primed normal mice. The 48 h culture supernatant (cIL) of these cells was obtained and used. As shown (Fig. 2), cIL had a marked effect (p < 0.01) on the growth of 476 cells, but none whatsoever on Mab2 and Sp2/0. Thus, it induced apoptosis in 476 cells as early as 24 h following incubation, but more significantly after 48 h and much more after 72 h (Fig. 2,A). cIL also inhibited significantly (p < 0.01) the proliferative activity of these cells, and reduced their viability (Fig. 2 B).
Suspecting the active factor(s) in cIL might be some cytokines involved in B cell growth or Ab synthesis, we used Abs to IL-4, IL-5, IL-6, and IL-13 in neutralization studies with cIL. Anti-IL-6 Abs, but not the other Abs, were found to block significantly, albeit incompletely, the apoptotic (p < 0.01) and anti-proliferative activity (p < 0.05) of cIL on 476 cells (Fig. 2,C). Direct proof that IL-6 itself could induce apoptosis (p < 0.05, 24–28 h; p < 0.01, 72 h) and cell death, as well as inhibit the proliferative activity (p < 0.01, 72 h) in 476 cells, was obtained, although at the concentration used, the effect was not as pronounced as that mediated by cIL (Fig. 2, A and B). As expected, IL-6 was found abundantly in cIL, as well as cytokines or chemokines and adhesion molecules such as LIX, KC, MIP-1, CXCL16, PF-4, VCAM-1, and L-selectin (Fig. 2 D). Significantly, neither TNF-α nor Fas ligand, both extrinsic apoptosis-inducing agents, were detected in cIL (<10 pg/ml).
cIL or IL-6 affects important cell functions in 476 cells
To elucidate how cIL or IL-6 adversely affected the growth of 476 cells, we examined several cellular activities in the treated cells. First, both cIL and IL-6 up-regulated Ab synthesis 1.7- to 2.0-fold in terms of the intracellular Ig concentration, and the effect was seen as early as 12 h (Fig. 3,A). A similar effect (1.3- to 1.5-fold increase) was observed in Mab2 cells used as control. Second, measurement of the amount of Ab secreted into the culture medium showed, unexpectedly, that cIL decreased rather than increased the secretion in 476 cells, especially after 24 h (1.4-fold, p < 0.05) (Fig. 3,B). At the concentration used, IL-6 had a similar but less marked effect after 24 h (1.2-fold decrease, p > 0.05). In contrast, both cIL and IL-6 induced a marginal increase (1.1- to 1.2-fold) in the amount of Ab secreted by Mab2 cells (Fig. 3,B). Third, both cIL and IL-6 caused a significant (p < 0.05) decrease in the telomerase activity of 476 cells as early as 12 h posttreatment (Fig. 3,C). The cIL effect was significantly (p < 0.05) abrogated by anti-IL-6 Abs (Fig. 3,C), suggesting IL-6 as an important active factor in cIL. In contrast, neither cIL nor IL-6 had an effect on the telomerase activity of Mab2 (Fig. 3 C).
The effect of cIL and IL-6 on telomerase activity was further examined by gene expression analysis. TERT gene function was found to be unaffected in both the 476 and Mab2 cells (Fig. 3 D), suggesting that the effect was mediated at the protein level.
cIL induces apoptosis via the intrinsic pathway different from enzymic neutralization or induction by a chemical inhibitor
We investigated the mechanism of apoptosis in 476 cells. First, based on DNA fragmentation, apoptosis was apparent in the cIL-treated cells as early as 24 h following treatment (Fig. 4,A). Apoptosis-related genes up-regulated after 48 h included, most notably, Apaf-1, as well as Bnip3, DFF40/CAD, Mcl-1, TRAIL, and TRAF6 (Fig. 4,B). In contrast, in cells treated with a chemical inhibitor (PPA) of telomerase function, Apaf-1 was only modestly up-regulated, but caspase-11 was markedly so (Fig. 4 B).
We next investigated the effect of decreasing the telomerase level in 476 cells on the induction of apoptosis (Fig. 4,C). As previously noted, apoptosis was induced in cells treated with cIL or IL-6 in which the telomerase level was lowered to 59.6–70.2% of control. In cells treated with PPA, the telomerase level dropped even more drastically to 28.3%, and expectedly, these cells readily became apoptotic. In accordance at the other extreme, small interfering RNA transformants derived from TERT or TERC inactivation in which the telomerase activities were only minimally affected (86.5–75.7% of control), apoptosis was unaffected. However, an antisense transformant (476-T) in which TERT sense transcripts were targeted (Fig. 4 D), showed a marked depression of telomerase activity (43.3% of control) that was lower than that of cIL treatment, but surprisingly, the apoptosis of these cells was not affected. This argues against the possibility that compromising the telomerase function of the 476 cells led to the demise of these cells.
We further compared the apoptotic pathways induced in 476 cells by cIL and PPA. In cIL-treated cells, immunoblotting of the cell lysates with appropriate Abs revealed substrate cleavage by caspase-3 and caspase-9, as well as translocation of cytochrome c to the cytosol, all as early as 24 h posttreatment (Fig. 4,E). PARP was also activated early but more pronouncedly at 48 h (Fig. 4,E). Bid was activated late (48 h). Specific enzyme inhibitors were also used to delineate the apoptotic pathway at 48 h by Annexin V FACS analysis. Pan-caspases inhibitor I significantly (p < 0.01) blocked cIL-induced apoptosis (Fig. 4,F). This cocktail of inhibitors act against caspase-1, caspase-3, caspase-4, and caspase-7, but the likely target is caspase-3 because caspase-1 and caspase-4 are related to Fas-mediated apoptosis, and caspase-7 is normally granzyme-mediated. Inhibitors targeting caspase-2, caspase-8, p53, p38, and bax had no effect on the apoptosis (Fig. 4 F).
Hybridoma 476 is deviant even in normal culture but mildly so
We examined 476 cells grown under normal growth conditions to see whether these were normal and found indeed that these cells had a significantly (p < 0.01) longer population doubling time (17.4 ± 0.9 h) than Mab2 (12.4 ± 0.3), Sp2/0 (12.6 ± 0.3) or AB1–2 (11.2 ± 1.1) (Fig. 5). AB1-2 is an IgA anti-idiotypic hybridoma derived from NS-1 myeloma cells and A/J mouse spleen cells. In addition, the telomerase activity of 476 cells (48.2 ± 7.9 relative units) was significantly (p < 0.01) lower than those of Mab2 (98.9 ± 13.6), Sp2/0 (99.1 ± 1.9) and AB1-2 (95.3 ± 10.3). Moreover, significantly (p < 0.01) more apoptotic cells were found in 476 cell culture (15.9 ± 0.5%) than in mAb (10.9 ± 1.5) or Sp2/0 (10.5 ± 1.6).
Ab-bound TERT can be found in 476 cells especially when cells are stimulated
Because, taken together, the results suggested that the intracellular binding between the Abs and the TERT protein was responsible for the demise of 476 cells, we sought to find evidence of preformed immune complexes in these cells. Thus, both 476 and Mab2 cells were biosynthetically labeled with [35S]methionine, the cell lysate incubated with protein G-Sepharose to trap the IgG Abs present, and the bound protein examined by gel analysis to determine whether telomerase was copurified. Presence of the TERT protein (108 kDa) was indeed found in the (unstimulated) 476 cells, together with the Ig γs and γm chains as expected, but not from Mab2 cells (Fig. 5,B). Another protein (82 kDa) was also present, presumably an accessory protein normally associated with telomerase. Significantly, both this protein and TERT were increased by 1.4-fold in the cIL-treated 476 cells (Fig. 5 B).
Immunoelectron microscopy reveals presence of Abs in the nucleus and mitochondria of stimulated 476 cells, confirmed by light microscopy analysis of isolated nuclei
To locate the immune complexes in 476 cells, we used electron microscopy to detect IgG as an indirect indicator. Morphologically, naive 476 and Mab2 cells were similar to each other under electron microscopy and, following stimulation with cIL for 24 h, both cells developed vacuoles but were otherwise intact, including the nuclear membrane. Vacuolation was more extensive in 476 cells and, unlike Mab2 cells, these also showed chromatin condensation, an early sign of apoptosis (Fig. 6). As expected, in both cells either cIL-treated or unstimulated, Ig deposits were found abundantly in the cytoplasm in the ER and Golgi apparatus. Significantly, cIL-treated 476 cells but not Mab2 cells (cIL-treated or naive) also revealed presence of Ig in the nucleus in significant amounts, and to a lesser extent, in the mitochondria. In this organelle, the Ig deposits were found largely in the condensed chromatin. Ig was also found in the nucleus of unstimulated 476 cells in scanty amounts, but not in the mitochondria.
We used another technique to ascertain that the electron microscopy signal found in the 476 nuclei was not a technical artifact due to contamination by cytoplasmic Abs. Thus, we first isolated the nuclei from the cells and then stained these for the presence of mouse IgG. The fluorescence light microscopy results displayed in Fig. 6 B show a total absence of IgG in the nuclei of Mab2 cells both treated with cIL or untreated. In untreated 476 cells, 5% of the cells (801 cells counted) were stained positive (green or yellow), some more strongly stained than others. Remarkably, many more of these positive cells (39%, of 731 cells) were found in the cIL-treated 476 cells.
The G50 hybridoma that produces an IgM anti-GTP Ab also has difficulty growing in mice
To verify that the abnormality of 476 was a general biological phenomenon, we constructed other hybridomas and found a similar clone, G50, which secretes an IgM anti-GTP Ab in culture. This Ab is specific for guanosine irrespective of the number of phosphate groups attached (Fig. 7,A). The G50 cells grew normally in culture but when introduced to the adjuvant-primed peritoneum of normal mice, growth inhibition was observed. Thus, in a representative of three experiments, all except 1 of 24 mice inoculated failed to produce ascites over a 48-day period (Fig. 7,B). In the ascites-producing animal (Fig. 7 B, mouse no. 5), ascites fluid drawn on day 14 had high anti-GTP activity, whereas the day-14 serum of this animal also contained high Ab levels. Surprisingly, in seven other animals that had no ascites formation, significant levels of serum Abs to GTP were also found on day 14. This suggested that the G50 cells were not totally eliminated from these animals, only their growth was suppressed.
In contrast, all five nude mice injected with G50 cells formed ascites by day 14 that contained high levels of anti-GTP Abs, and the circulation of these animals also had high Ab levels (Fig. 7 B).
In the normal mouse (no. 5) that formed ascites, the ascites formation was unusual for two reasons. First, the formation was small, containing <3 ml of fluid, compared with >8 ml of fluid found in nude mice. This suggested growth suppression, albeit incomplete. Second, the ascites in the animal spontaneously disappeared completely within the following 2 wk and remained so for months. The serum also had no Ab activity. This regression was not aberrant because it was also found in two other animals injected with G50 cells in a separate experiment. In these cases, the ascites, which were also very small, were both found on day 16 and both disappeared completely within 11 days. Thus, it appeared that in a very small percentage of animals, the inoculated G50 cells managed to escape immediate growth suppression to form ascites initially but then succumbed to suppression subsequently.
One normal mouse (no. 6) with no obvious ascites formation had moderate levels of anti-GTP Abs in the circulation on day 14 that persisted over the 70-day period. This suggested incomplete but continual suppression of cell growth in this animal. In two other mice (nos. 2 and 9) that had no ascites or serum Abs all along, ascites appeared unexpectedly on day 70 (Fig. 7,B). Again, both ascites were small and contained only moderate levels of anti-GTP activity, but no serum Abs were found from either animal (Fig. 7,B). Ascites cells recovered from both mice were cultured and examined for the presence of Ig μ chain. The FACS results (Fig. 7 C) revealed that the recovered cells of the two animals after culturing for 1 day had only 21.9–48.8% of the μ+ cell activity of the parental G50 cells, but these activities increased on further culture to 38.9–58.5% and 61.9–82.3% after a total of 5 and 20 days, respectively. This suggested acute but incomplete suppression of growth of μ+ cells in the peritoneum, which allowed μ− Ab-deficient mutants that spontaneously arose to populate the peritoneum and form ascites. In culture, successive dilution of the putative suppressive factors from the peritoneum presumably allowed the μ+ cells to grow uninhibitedly and dominate over the μ− cells.
Other hybridomas screened include clone 659, which produces an IgG Ab to OPN, also fails to form ascites
We screened other mouse hybridomas made in our laboratory for their ability to form ascites. Two clones, 659 and 446, both reactive with human OPN, are particularly interesting (Fig. 8,A). Although the 446 IgM mAb does not bind to mouse OPN, the 659 IgG mAb binds, albeit much less than to human OPN (Fig. 8,B). The cross-reactive epitope recognized by mAb 659 is located at the thrombin-sensitive site of the molecule (data not shown). Interestingly, whereas the 446 cells readily formed large ascites in normal mice (n = 5 animals), the 659 cells failed to grow in the peritoneum or produce serum Abs for more than 50 days (n = 5 mice) or in a repeat experiment, 25 days (n = 11 mice). In accordance, when treated with cIL, only the 659 cells were significantly affected in their proliferative activity (p < 0.01) (Fig. 8 C).
We also obtained two autoantibody-producing hybridomas from American Type Culture Collection for similar screening: HB 8609, which produces an IgG Ab to bovine La/SS-B that reportedly can cross-react with the human Ag, and CRL 1640, which produces an IgG to human DNA polymerase α. Both clones were able to grow normally as ascites, and their proliferation was not affected by cIL treatment (Fig. 8 C).
Hybridoma 476 was produced in mice against human TERT and not murine TERT thus overcoming tolerance restrictions in the mouse, but the Ab cross-reacts with the murine protein and is thus a true autoreactive Ab in the mouse. The 476 cells were actually deviant but exhibited normal growth and function in culture. The defect became apparent only when attempts were made to grow the 476 cells in the peritoneum of immunologically normal mice. These cells are representative of normal cells, however, and not just idiosyncratic for the following reasons. First, another hybridoma with anti-telomerase specificity was reported also to be unable to produce ascites (21). Second, we searched for other hybridomas with other autospecificities for a similar growth abnormality and found two such clones, G50 and 659, which produced autoantibodies to GTP and OPN, respectively. In contrast, all other hybridomas in our hands (with 9 different non-self specificities) were able to produce ascites. Third, Ab specificity is crucial to the problem because, as observed of the 476 and G50 cells, loss of or decrease in Ab function resulted in a reversal of the impairment. In addition, no known extrinsic proapoptotic factors such as TNF-α and Fas ligand, were found in the peritoneal environment that could account for the problem.
In vitro investigations of the 476 cells implicate IL-6, a proinflammatory cytokine found abundantly in the adjuvant-primed mouse peritoneum, with the possible help of other peritoneal factors, to be indirectly responsible for the demise of the cells in vivo. IL-6 does not kill the 476 cells directly. Rather, by up-regulating the Ab synthesis in these cells, it increases the intracellular formation of immune complexes between the Abs and the enzyme that are toxic to the cell. Similar binding occurs too in the unstimulated cell but this level of activity is appears to be innocuous. The Abs appear to escape from the ER-Golgi complex (or transport vesicles) to the cytosol where they bind to the naked, nascent TERT protein (15). Evidence of binding is shown by the presence of preformed immune complexes in the cell and, indirectly, by the lower-than-expected secretion of Abs to the culture medium following IL-6 stimulation. This interaction reduces the telomerase activity of the cell but not the telomerase gene function. Neutralization of enzymic activity could itself trigger the cell to apoptose but two observations argue against this. First, the onset of apoptosis was rapid (24 h). Second, although the treated cells became more apoptotic, they still retained 60% of the telomerase activity of untreated cells, which is higher than telomerase activity (43% control) of a TERT-antisense transfectant (476-T) that exhibited a normal rate of apoptosis. Moreover, it is known that cells can survive for generations with low or negligible levels of telomerase (22, 23).
Instead, the immune complexes formed inside the 476 cell appear to be the toxic factor. These complexes apparently entered the nucleus and, to a lesser extent, the mitochondria, as inferred from the presence of Ig in these organelles. TERT is normally chaperoned to the nucleus by a 14-3-3 protein (14). Abs bound to TERT, presumably involving just a single molecule of each due to the monovalency or bivalency of these proteins, can thus be piggy-backed to this organelle, and like “Trojan horses,” be similarly transported across the nuclear membrane. Although such transportation is unprecedented, it is known that macromolecules as large as IgM-nucleoplasmin (35–40 nm) (24) and gold-protein complexes (39 nm) (25) can enter the nucleus through specific pores in the organelle that are expandable in size (26, 27). Entry is not passive but mediated by the glycine- and phenylalanine-rich pore proteins (nucleoporins) that interact with the cargo-carrier complex in a process orchestrated by Ran GTPase. Nonspecific entry in the 476 cells is also unlikely because the nuclear membrane of the stimulated or unstimulated cell was intact, while the possibility of a technical artifact is ruled unlikely because no nuclear presence of Ig was found in Mab2 cells. Indeed, the Abs found in the 476 nucleus appeared TERT-bound because they colocalized with the chromatin. A mitochondria-targeting sequence located at the N-terminal leader in mouse TERT (MTRAPRCPAVRSLLRSRYRE) (www.cbs.dtu.dk), similar to that found in human TERT (28), could similarly direct the migration of Ab-bound TERT to the mitochondria with the help of another 14-3-3 carrier protein (29) and mitochondrial translocases (30).
The abnormal presence of Abs or Ab-bound TERT in the nucleus or the mitochondria probably stresses the organelle (31) and triggers a cascade of events leading to intrinsic apoptosis. Thus, DNA damage initiated by nuclear stress can activate PARP-1 and lead to the translocation of Apoptosis-inducing factor from the mitochondria to the nucleus, which then induces chromatin condensation and DNA fragmentation (32). Mitochondrial stress arising from the nuclear cross-talk or directly from the imported Abs can result in the cytosolic translocation of cytochrome c and the activation of caspase-9 and caspase-3, but not caspase-8 (31). Caspase-8 normally responds to external apoptotic stimuli and such activation was seen only when the 476 cells were treated with PPA, a chemical that intercalates the telomeric DNA to form a binary complex with the G-quadruplex structure at the ends of telomeres, thus inhibiting telomerase binding (33). PPA also differs from IL-6 in activating other genes in 476, which again supports the contention that the proapoptotic effect of IL-6 is not due to the sheer abrogation of telomerase activity.
The mechanism of death has not been elucidated in the G50 cells but from preliminary observations, a similar increase in Ab binding to sensitive cytosolic targets in the peritoneal environment seems likely, the increase again mediated by IL-6 because suppression was absent in nude mice. However, in this study, disruption of vital cell processes rather than the formation of toxic immune complexes may be responsible as many of the GTP autoantigens are cellular structures involved in signal transduction. These Ags include GTP and GDP (di-phosphate) molecules associated with various cell structures such as tubulin and the various GTPases (34), as well as GMP (mono-phosphate) epitopes that are present, for example, in ribosomal RNA and tRNAs. It is intriguing whether Rab1 GTPase which is located in the Golgi apparatus, including the family of golgin proteins (35) similarly located, are directly accessible to the anti-GTP and anti-golgin Abs synthesized in this organelle that are found in some autoimmune disorders (9, 36).
The G50 cells appeared to be less severely inhibited than the 476 cells. Thus, in ∼8% of animals, the inoculated G50 cells managed to escape initial suppression and formed ascites, whereas in the other animals in which suppression appeared complete from the beginning, small numbers of G50 cells managed to persist undetectably in the animal for months. These cells presumably survived in a quiescent state or replicated continually but had their numbers kept limited by host suppression, and were discovered only because Ab-deficient mutants arose from them which then grew and form ascites months later. The weaker growth suppression of G50 cells compared with 476 cells may reflect a weaker Ab or an inhibition of cellular function that is incomplete or reversible, which regardless could be less damaging than the induction of organelle stress. This, in fact, is consistent with the high prevalence of anti-guanosine Abs found in autoimmune patients (37) compared with anti-telomerase Abs (our unpublished observations).
The findings from the two OPN-specific hybridomas are instructive. They underscore the importance of distinguishing mouse cells that produce Abs to mouse autoantigenic epitopes, including those that are part of a human protein, from those that produce Abs to human-only epitopes. This reason may be why HB 8609 and CRL 1640 were able to grow in vivo because of their species specificity. We have not checked this possibility, however, but there are other factors governing cell vulnerability to suicide even if the Ab is appropriate, including the affinity or fine specificity of the Ab and the accessibility of the target Ag.
The vulnerability of the 659 cells was unexpected. OPN is a pleiotropic cytokine produced by many types of cells, which is implicated in diverse physiological processes such as bone remodeling, cell migration, cancer metastasis and inflammation. Its role in immune functions has been increasingly recognized, including the polyclonal activation of B cells (38) and the promotion of survival of activated T cells (39). OPN normally functions as a secreted glycoprotein but the intracellular protein has been reported to exact important roles in cell migration (40) and, in plasmacytoid dendritic cells, IFN-α production (41). Its intracellular function in B cells is not known, but we suspect from the present findings that this is either very important to the growth or survival of the cell or, as in the case of the 476 cells, Ab complexes formed by it inside the cell are extremely toxic. We know OPN is present in B cells from the cytoplasmic staining of NSO cells by mAb 659 (data not shown), but this raises many interesting questions about its presence and function here, including where it trafficks inside the cell.
Our postulation that IL-6 is proapoptotic may seem paradoxical in view of previous findings demonstrating that this cytokine clearly promotes growth and survival of B cells and plasmacytomas (42, 43, 44). Indeed, adjuvant-preparation of the peritoneum is long known to be a prerequisite for ascites formation (45) due essentially to the growth-promoting activities of IL-6 (46, 47), consistent with the fact that plasmacytomas do not grow in IL-6-deficient mice (48). IL-6 presumably acts through the STAT3 signaling pathway in these cells, a mechanism recently found for the survival of B cells in the peritoneum, particularly the B-1 subset that normally resides here and in which, interestingly, STAT3 is inherently activated (49). For this reason, we do not expect the 476 cells or the G50 and 659 cells to be able to grow in IL-6-deficient mice. Paradoxically, however, these same cells would also succumb to IL-6 if too much of it is present. (This property presumably does not apply to B-1 cells.) This effect is due to another function of IL-6, the augmentation of Ig synthesis (50). Increased Ab production leads to more toxic immune complexes being formed, which overrides any beneficial effect IL-6 has on the growth or survival of the 476 cells. For other cells without such an autospecificity (i.e., intracellular binding), by contrast, the antiapoptotic effect of IL-6 would not be negated. The 476 cells were able to grow in nude mice because smaller amounts of IL-6 are produced in these animals that lack T cells, which are the main producer, with macrophages and endothelial cells being the secondary producers. Based on the in vitro demonstration that purified IL-6 appeared less potent than the peritoneal mixture, other factors may be present in the inflamed peritoneum that enhance the proapoptotic effect of IL-6. However, none of the chemokines and adhesion molecules found has a known effect on Ig synthesis, and proteins such as OPN were not screened.
Thus, a cell that produces a potentially deleterious Ab can exist in nature indefinitely so long as the Ab synthesized is below a critical concentration, for example, as resting B cells. However, when these cells become activated by IL-6 in the splenic or peritoneal environment or differentiate to become production factories, the ensuing suicide could be mistaken for natural death as in the case of short-lived plasma cells, presumably such cells do not become long-lived plasma cells (51). effect could thus represent a previously undiscovered mechanism to prevent excessive production of self-inflicting Abs in the body. This mechanism is different from other control previously described that silence autoreactive B cells in the periphery through extracellular, rather than intracellular, autoantigen interacting with the BCR (52, 53). A fine balance thus exists between life and death in the autoreactive cells, and the cell population conceivably waxes and wanes with time according to local changes in cytokine concentration. Hybridomas 476, G50 and 659 are presumably snapshots of a spectrum of cells with varying propensities for suicide, which depend on factors such as Ab affinity or fine specificity and the Ig biosynthetic rate. Indeed, if not for their immortalization as hybridoma cells, these cells would have escaped notice as deviant cells.
The findings described in this study may have therapeutic implications. Both telomerase and OPN appear to be ideal targets for Ab-mediated therapy, but the Abs will have to be specially delivered, for example, in the form of intrabodies (54). This system can complement the arsenal available to inhibit telomerase function in cancer cells (55). An intriguing question for autoimmune disease is: Can IL-6 be judiciously used to kill off undesirable autoreactive B cells?
We thank Michael Jiang (Department of Biology and Chemistry, City University of Hong Kong) and Fernand Lai and Janet Cheng (Academy of Pathology, The Chinese University of Hong Kong) and for help in electron microscopy, and Peggy Fung for secretarial assistance.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Abbreviations used in this paper: ER, endoplasmic reticulum; OPN, osteopontin; PARP, poly(ADP-ribose) polymerase; PPA, 2-6-bis[3-(N-piperidino)propionamido]anthracene-9,10-dione.