| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Experimental Therapeutics, Preclinical Pharmacology |
Center for Molecular Medicine and Immunology, Belleville, New Jersey 07109
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
4 x 106 in 48 h. This result raised a dilemma, in that the number of antigen sites per cell, on the cell surface, was much lower, by 10-fold. The purpose of this study was to explain the high level of accumulation. The initial possibilities considered were the following: (a Ab binding may induce a large up-regulation of antigen expression; or (b) there may be a large cytoplasmic pool of antigen, which recycles to the cell surface, as described for the TfR (6)
and other cell surface proteins. We show herein that anti-CD20 localizes to intracellular vesicles that are similar or identical to the site reached by Tf, and also that up-regulation does not occur. Therefore, the high level of Ab accumulation is probably caused by the presence of a large intracellular pool of antigen, although there is no evidence that the antigen is localized to this site in the absence of Ab binding. CD20 is a tetraspan membrane protein, with only a small loop present on the exterior of the cell (7) . All Abs recognize a single epitope, in competitive binding experiments, although different Abs differ in their ability to affect cell physiology (7) . The precise function of the molecule is not known, but it appears to play a role in lymphocyte activation, and evidence suggests that it may function as a Ca2+ channel (8) . Ab binding induces the rapid redistribution of nearly all of the CD20 to a low-density, detergent-insoluble plasma membrane compartment (9) . In previous studies of the processing of bound Ab using radiolabeled Ab, it was found that the Abs remained bound to the cell for days, with a low but detectable level of catabolism, perhaps dependent on the particular cell line used (10 , 11) .
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1 mCi/mg for DLT. Labeled preparations were analyzed by instant thin-layer chromatography or by gel filtration high-performance liquid chromatography, or both, by methods that have been described (15)
to determine the level of radioactivity not bound to the Ab, which was always <10% and usually <5%. Representative preparations of radiolabeled Abs, with each radiolabel, were tested for immunoreactivity (% bindable), by incubating with a large excess of cells. Control tubes had excess unlabeled Abs added, to block specific binding and to, therefore, indicate the level of nonspecific binding; specific binding was calculated by subtraction. The immunoreactivity was not significantly affected by the particular labeling method used. The range of specific binding was 41.8–50.6% for 1F5. 1F5 and a control nonreactive IgG2a Ab, TA99, were conjugated with fluorescein isothiocyanate (Sigma Chemicals) by standard procedures, with a conjugation ratio ranging from 8.9 to 12.5:1. Alexa Fluor 488 was conjugated to four Abs using methods provided by the supplier (Molecular Probes, Eugene, OR): 1F5, MX352a (a different nonreactive control IgG2a), MA103 (anti-CD147), and rituximab. Conjugation ratios ranged from 1.2:1 to 3.9:1. Alexa Fluor 488 was also conjugated to the Fab fragment of 1F5, with a conjugation ratio of 1.2:1. Fluorescein-conjugated B1 was purchased from Coulter Cytometry (Miami, FL), and fluorescein-conjugated L27 (Leu-16) from Becton Dickinson.
Saturation of Cell Surface Antigens, and Competitive Binding Experiments.
In duplicate, 106 cells in 50 µl of tissue culture medium were mixed with 50 µl either of medium containing excess unlabeled Ab or of control medium and were incubated for 15 min at room temperature. The excess unlabeled Ab was used to block specific sites and to, therefore, show the level of nonspecific binding. The radiolabeled Ab, in Dulbecco’s PBS (Life Technologies, Inc., Rockville, MD), 0.5% bovine albumin, and 10 mM NaN3, was then added, with 2.5 x 106 cpm per tube, with unlabeled Ab added to obtain the desired Ab protein concentration. The mixture was incubated for 1 h at the indicated temperature, with resuspension every 20 min, then was transferred onto a layer of 0.25 ml of phthalate oil mixture in a 400-µl polypropylene microcentrifuge tube (16)
. After spinning for 4 min at 10,000 rpm, at 4°C, the tube was cut and the pellets counted for radioactivity. Competitive binding experiments, to define CD20 epitopes, were performed similarly, except with RL lymphoma target cells. Blocking Abs were used at 50 µg/ml to inhibit the binding of 125I-labeled 1F5. The Abs that were tested, both of them anti-CD20, were L27 and B1-FITC.
Prolonged Ab Incubations.
Cells (5 x 106) and 7.5 x 107 cpm labeled Abs, with unlabeled Abs added to obtain the desired concentration, were mixed in 15 ml of tissue culture medium and plated in wells of 24-well plates, with 1.5 ml/well. At various times, in duplicate, cells were collected, pelleted, washed three times, transferred to a clean tube, and counted for radioactivity. Viable cell counts were obtained at each time point, using 30-µl samples diluted 2-fold into a trypan blue solution.
Antigen Modulation-Type Experiments.
Cells in 24-well plates were incubated overnight with serial dilutions of 1F5, from 0.01 to 10.0 µg/ml, while control cells were incubated without Ab. They were collected and washed twice with medium. Half of the samples were treated with 100 µl of 1F5 at 20 µg/ml, for 45 min at 37°C and then were washed twice. All of the samples were then stained with 50 µl of fluorescein-conjugated goat antimouse IgG (Cappel-ICN, Costa Mesa, CA) at 0.25 mg/ml in tissue culture medium with 5% human serum and 10 mM NaN3, for 45 min at 0–4°C. After 1 wash with tissue culture medium, and 1 wash with Dulbecco’s PBS (Life Technologies, Inc.), 0.5% bovine albumin, 10 mM NaN3, the cells were suspended in the same medium containing 1 µg/ml propidium iodide (to stain dead cells red) and were examined for fluorescence by flow cytometry on a Becton Dickinson FACS II.
Immunofluorescence To Detect Intracellular Accumulation.
Abs, sterilized by filtration, were added to wells of 24-well plates containing 5 x 105 cells and 0.5 ml of culture medium. Intact Abs were used at a final concentration of 10 µg/ml, and the Fab fragment at a final concentration of 50 µg/ml. Rhodamine Red-X-human Tf was purchased from Jackson ImmunoResearch (West Grove, PA), and was used at a final concentration of 1:50. In preliminary experiments, serial dilutions of these reagents were tested for various time intervals to identify the optimal conditions. At various times, the cells were collected, washed twice with Phenol Red-free tissue culture medium, and examined. In some experiments using 1F5 only, cells were counter-stained with propidium iodide (1 µg/ml) to stain dead cells red, and viability was always >95%. Examination was on either of two microscopes. One was an Olympus BH2 fluorescent microscope with a 100-W mercury bulb, standard fluorescein filters, and excitation filter D545/30X (Chroma Technology, Brattleboro, VT) for rhodamine-Red-X. The second was an Olympus BX50W1 microscope equipped with a 100-W mercury bulb and standard filters for fluorescein and rhodamine. This microscope was also equipped as a Fluoview confocal microscope, with argon and krypton lasers. The emission filters used for confocal microscopy were BA510-540 for fluorescein/Alexa Fluor 488 and BA590 for rhodamine-Red-X. For the observation of rhodamine-Red-X only, light from the argon laser was blocked. All of the photographs used the x40 objective and were taken either with Kodak slide film or with a Sony 3CCD model DKC-5000 digital camera. Superimposition of photographs used Image-Pro software. Printing of digital photos was on a Tektronix Phaser 740 Plus printer.
The Golgi cisternae marker BODIPY-FL-C5-ceramide (Ref. 17 ; previously called C5-DMB-ceramide) was purchased from Molecular Probes and was used following instructions provided by the supplier. The bovine albumin used for preparing complexes was crystallized and fatty-acid free (Sigma Chemicals). Stained cells were examined both with the fluorescein filters, to detect total staining, and with the rhodamine filters, to detect only regions containing a high concentration of the label. In some experiments, cells were treated with BFA (Sigma Chemicals), at 5 µg/ml, which disperses the Golgi cisternae but not the TGN (18 , 19) .
Cell Extraction and SDS-PAGE.
For SDS-PAGE analysis of cell extracts, cells were extracted by the sample buffer, which contains 2.0% SDS, by heating at 100°C for 3 min. Insoluble material was removed by centrifugation; 104
cpm were run on 12% acrylamide gels, followed by autoradiography, by methods that have been described previously (20)
.
Other Methods.
Cells were spun onto slides using a cytocentrifuge, a Cytofuge 2 (Statspin, Norwood, MA), using a filter concentrator, spinning at 2200 rpm for 6 min. After air-drying, slides were fixed in methanol for 10 min, then stained with Wright’s stain by standard methods.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, maximum accumulation occurred at 24 h, with 2.4 x 106 Ab molecules bound per cell. The accumulation at 1 h was much less, by 3.7-fold, and accumulation was relatively slow, because values at 4 h were not much higher than those at 1 h. Accumulation with Abs at 10 µg/ml was only slightly greater than that at 5 µg/ml, which indicated that saturation was approached. Other experiments used Abs at 20 µg/ml, and results were essentially the same as those with 10 µg/ml (data not shown). Control experiments to demonstrate the specificity of accumulation used an Ab of the same subclass, labeled in the same way; nonspecific accumulation was very low, <0.05% of the total radioactivity at 24 h, and correcting for nonspecific accumulation would have only a small effect on the values shown, of 
4.3% of the Ab molecules bound. The apparent decrease in accumulation from 24 to 48 h, shown in Fig. 1
, is not understood and is shown primarily to demonstrate that accumulation does not increase after 24 h. This decrease is attributable to cell division because the absolute level of bound cpm increased somewhat but did not match the increase in the cell number, which approximately doubled in the 24-h period.
|
, because the catabolites rapidly leave the cell (21)
. But it was of interest to determine whether the use of a "residualizing" label would result in an even greater radiolabel accumulation. Therefore, identical experiments were performed with an 125I-DLT label, which cannot cross lysosomal membranes (21)
. As shown in Fig. 1B
, results with this label were virtually identical to results with the chloramine T label, which demonstrated that catabolism of bound 1F5 is not a significant factor in these experiments.
In our previous, similar experiments using an 111In label, the accumulation of Ab molecules per cell was even higher, by 2-fold. The peak accumulation, which was at 48 h, was 5.0 x 106 Ab molecules per cell (3)
. This difference can be attributed primarily to the increase in the size of the cells in the 111In experiments, which resulted from the toxic radiation dose delivered.
1F5 Binding Sites per Cell.
The preceding experiments show two notable results. First, the accumulation increased markedly from 1 to 24 h. Second, the cumulative binding appeared to be much higher than the published values for CD20 binding sites per cell for Raji cells (22)
or for other B-lymphoma cell lines (10)
. It was, therefore, important to redetermine the number of sites per cell. Results of antigen saturation experiments are shown in Fig. 2
. High concentrations of 1F5, up to 33 µg/ml, were required to obtain saturation, as indicated by a plateau in the level of binding. Moreover, although initial experiments were done with a 4°C incubation, which is generally considered optimal in studies of this type, a binding plateau was never reached nor clearly approached at that temperature, even at these high Ab concentrations (Fig. 2A)
. We, therefore, performed experiments at 37°C, which produced a clear plateau (Fig. 2B)
at 2.4 x 105 Ab molecules specifically bound per cell. The maximum specific binding at 37°C was 3.4-fold higher than the highest binding obtained at 4°C. The nonspecific binding was almost the same at both temperatures; therefore, the ratio of specific:nonspecific binding was much higher at 37°C. We cannot explain the difficulty in obtaining saturation at 4°C (this is discussed below), but the results obtained at 37°C are adequate for our present purpose, for two reasons. First, the reason for generally preferring a 4°C incubation is that internalization or membrane recycling of the antigen is prevented. That is, binding at 37°C is considered to possibly overestimate the number of sites on the cell surface. Therefore, our conclusion, that the total accumulation per cell over 24 h is much greater than the number of sites per cell, may be understated by performing saturation experiments at 37°C. Second, we note that the previous determinations of CD20 sites per cell (10)
, which we use for comparison, were determined at room temperature for reasons that were not given. By comparison with Fig. 1
, we conclude that the Ab accumulation in 24 h is 10-fold higher than the number of antigen sites on the cell surface.
|
Confirmation of the Specificity of 1F5.
The high level of accumulation of 1F5, not previously reported for anti-CD20 Abs, induced us to check the specificity of the Ab. This was done by competitive blocking with other CD20 Abs, namely B1 and L27. There is only one epitope of CD20 (7)
. These two Abs blocked the binding of 125I-labeled 1F5, whereas unrelated Abs had no inhibitory effect (data not shown).
Is CD20 Up-Regulated by the Ab?
One possibility was that the level of CD20 antigen expression on the cell surface was up-regulated as a result of Ab binding. This question was investigated by classical antigen modulation experiments, in which cells were preincubated with a saturating concentration of the Ab for 24 h, then tested by immunofluorescence for antigen expression on the cell surface. This experiment is usually performed to demonstrate down-regulation (modulation), but it is equally useful to detect up-regulation. Immunofluorescence was determined with a fluorescein-conjugated second Ab, goat antimouse IgG, to detect mouse Ab bound to the cell surface. The results were that CD20 antigen expression was unchanged (data not shown). We conclude that the high accumulation of 1F5 is not caused by up-regulation.
Form of the Retained CD20 Ab.
To investigate the form of the retained Ab, extracts from solubilized cells were analyzed by SDS-PAGE. Material extracted after 24 h or 48 h of Ab incubation looked essentially the same as the control Ab, with most of the label being on the heavy chain and only very weak labeling of the light chain (data not shown).
1F5 Accumulation Intracellularly, as Detected by Immunofluorescence.
Another possibility was that 1F5 accumulated intracellularly. If true, the accumulation could not be in lysosomes, because the Ab would be rapidly catabolized in lysosomes, with subsequent rapid release of the radioiodinated catabolites (21)
. To detect intracellular accumulation, Raji cells were incubated overnight with 1F5 conjugated to a fluorescent dye, then examined on a fluorescent microscope. This incubation resulted in a bright spot of intracellular fluorescence in the JN region. Similar experiments were performed with both fluorescein and with the new fluorophore, Alexa Fluor 488, conjugated to 1F5. Essentially the same results were obtained with both fluorescent markers, but the Alexa Fluor 488 displayed much less fading, as reported by the supplier (Molecular Probes), and was, therefore, used in most of the subsequent experiments. To demonstrate that formation of a JN spot is unique to 1F5, a control Ab, MA103 (anti-CD147), was similarly conjugated to Alexa Fluor 488 and tested. This Ab, which reacts with a typical, slowly internalizing cell-surface antigen (11)
, displayed only ringed staining after an overnight incubation, with no indication of a JN spot. Nonreactive control Abs of the same subclass, labeled and used in the same way, produced no significant staining. To investigate whether bivalent Ab binding was required to induce Ab delivery to the JN spot, similar experiments were performed with an Alexa Fluor 488 conjugate of the Fab fragment of 1F5. After overnight incubation, a JN spot was seen in most Raji cells, but it was very faint and less bright than the cell-surface staining of the same cells (data not shown). Considering that the avidity of a Fab fragment will be much lower than that of an intact Ab, it is not clear how to interpret these results. We can conclude that the efficient delivery of the Ab to the JN spot depends on cross-linking, but a small amount of monovalently bound Ab may also reach this site.
To characterize the region stained with 1F5, costaining was performed with rhodamine-conjugated Tf, which labels endosomes and the ERC. The Tf was added for the last 45 min of the overnight incubation with Alexa Fluor 488–1F5 because it was determined in preliminary experiments that staining with Tf reached a steady state after 45 min. The Tf staining was very similar to that of 1F5, as shown in Fig. 3
. Close observation indicated that there were slight differences in the appearance of the JN spot seen with the two markers in some cases, but in many cases the distributions were entirely aligned. The JN spot was not a circle but generally had some irregularity in shape, and the irregularities were usually, but not always, the same with both markers. The most prominent difference between the two markers was in the brightness of membrane staining. Ab 1F5 stained the membrane fairly brightly (although less brightly than the JN spot), whereas Tf produced faint, barely detectable staining of the cell surface, as shown (Fig. 3)
. In addition, small vesicles in the cytoplasm outside of the bright JN spot were more evident with Tf than with 1F5. In some experiments, cells were also examined by confocal fluorescent microscopy. Observations on the confocal microscope supported the conclusions that have been stated, and, in particular, demonstrated that the JN spot was inside the cell. We have chosen to present photos taken by standard microscopy because this shows cell membrane staining more clearly.
|
shows cells spun onto a slide with a cytocentrifuge and stained with Wright’s stain. The vesicles appear to be the same as those seen by electron microscopy in Raji cells, and characterized as lipid droplets (23)
. In addition to the vesicles, a pale area is seen between the nucleus and the vesicles, which is characteristic of the Golgi area (24)
. The nuclei frequently, but not always, have a cleft, and this is where the Golgi region is located. Fig. 5
shows a similar cell preparation in which the cells were stained with Alexa Fluor 488–1F5. The cells were illuminated by very faint visible light as well as by UV light to allow both the fluorescent spot and the vesicles to be seen simultaneously. As shown, the fluorescent spot was generally just inside the cluster of vesicles. Thus, the area stained by 1F5 is very close to the pale JN Golgi area (although it is probably the TGN and/or the ERC rather than the Golgi cisternae, as discussed below).
|
|
There was some capping in the experiments described above, but it was at a relatively low level, with no definite relationship to Ab internalization. Almost all of the caps were large caps, covering 20–50% of the cell surface. Most JN spots occurred in cells that were not capped. We note that very prominent, long microvilli, stained with Alexa Fluor 488–1F5, were evident in some of these experiments (data not shown). Such microvilli have been described before on normal B-lymphocytes, as detected by immunofluorescence (25 , 26) . They were particularly prominent in our experiments because of the brightness of Alexa Fluor 488. The microvilli developed most extensively in the presence of 10 mM NaN3. This was demonstrated in controlled experiments with or without azide, with all of the other parameters constant, and is consistent with previous results (25 , 26) . In the absence of azide, under normal tissue culture conditions, microvilli were still detected, but were much less prominent.
To compare the JN spot stained by 1F5 and Tf with the Golgi region, Raji cells were stained with BODIPY-FL-C5-ceramide, which stains the Golgi cisternae, more specifically the trans-Golgi cisternae, when examined for red fluorescence (17
, 27)
. Green fluorescence, which is much brighter, shows staining of other membranes as well as those in the Golgi region (Fig. 6A)
. Rhodamine filters showed a JN spot that appeared similar to that seen with the two previous markers (Fig. 6B)
. This Golgi spot, like the JN spot stained by 1F5, was just beneath the cluster of vesicles noted above. However, a distinction between the markers was demonstrated by treating cells with BFA which is known to disperse the Golgi cisternae but not the TGN (18
, 19)
. BFA in fact caused dispersal of BODIPY-FL, as expected, as shown in Fig. 6C
but did not significantly affect staining by Alexa Fluor 488–1F5 (data not shown). Although BFA often induces markers of the TGN to form transient tubular structures before condensation at a JN site (18
, 19)
, this was not observed with the JN spot stained with 1F5.
|
3–10) smaller cytoplasmic spots, rather than one large JN spot (Fig. 7)
. This difference is not fundamental, because, even with Raji, it was clearly seen that the JN spot, in fact, was composed of many smaller spots that were clustered closely together. As with Raji, red and green vesicles matched almost perfectly. There was, however, a significant difference between Ramos and Raji in the rate of uptake of Tf. With Raji cells, as noted above, 45 min was sufficient to reach a steady-state distribution. With Ramos cells, in contrast, 4 h was required. After a 45–60-min incubation of Ramos cells with Tf, the cytoplasm was filled with many very small red vesicles, which gradually condensed into fewer, medium-sized vesicles over the next 3 h. Daudi cells were generally similar to Raji cells in their staining pattern, but Alexa Fluor 488–1F5 staining was not as bright and homogeneous (Fig. 8)
. Staining of a JN spot with 1F5 was clear in only approximately one-half of the cells, and there were many borderline cells in which a hint of staining was observed. In contrast, Tf staining of Daudi was very similar to that of Raji, with a large, distinct JN spot in essentially all cells. All of the spots stained with 1F5 were also stained with Tf, whereas the inverse was not true. RL cells also were unique. These displayed very little intracellular staining with 1F5, although there was some weak staining in a minority of the cells, although cell surface staining was fairly bright. In contrast, Tf stained a bright JN spot in RL cells, very similar to the spot stained in Daudi or Raji cells. The other lymphoma cell lines were also stained with BODIPY-FL-C5-ceramide, the Golgi marker; Ramos and Daudi showed a single bright JN spot, similar to the spot described above for Raji, whereas RL had a heterogeneous pattern of cytoplasmic staining. By phase-contrast microscopy, Daudi had secretory vesicles similar to those in Raji, Ramos had a smaller number of the vesicles, and RL had few, scattered vesicles. These results suggest that RL has basic differences from the other cell lines in the cytoplasmic organization of membrane synthesis and transport.
|
|
With Ramos cells, capping was much more prominent than with Raji or the other cell lines tested. After a 45-min Ab incubation, 30–60% of the cells were capped, and after overnight Ab incubation, 20–30% of the cells were capped. In contrast to the results with Raji, the intracellular fluorescent spots present after overnight incubation were invariably beneath the cap, as exemplified in Fig. 7
. Because capping preceded internalization of the fluorescence into spots beneath the cap, it appears likely that in many cells, the internalized Ab was derived from the cap. However, it should be emphasized that, for both Ramos and Raji, many cells with bright intracellular fluorescence had no indication of capping; thus, capping was not required for internalization. This capping on Ramos cells was caused by bivalent Ab binding, because it did not occur with Alexa Fluor 488 Fab of 1F5, and was also inhibited by 10 mM NaN3 and by temperatures of 0°C to 4°C (data not shown). As with Raji cells, the caps remained quite large.
Cytotoxicity with 125I-labeled 1F5 on Four B-Lymphoma Cell Lines.
It seemed possible that cytotoxicity with radiolabeled 1F5 might be correlated with the accumulation of the Abs at an intracellular site. To investigate this possibility, the three other cell lines were tested for toxicity medicated by 125I-labeled 1F5 or a nonreactive control Ab. [Raji was tested previously (3)
]. The control Ab used, MN-14, was an IgG1, a different subclass from the IgG2a Ab 1F5; however, Ab accumulation experiments were previously done with nonreactive Abs of both subclasses, and there was no significant difference in accumulation, which was very low (data not shown), indicating that nonspecific toxicity is attributable primarily to radioactivity in the medium, rather than to cell-bound radioactivity. Therefore, the Ab subclass should not be a significant factor for nonspecific toxicity in vitro. Fig. 9
shows the cytotoxicity results with the four cell lines. Raji, in fact, was relatively resistant to specific cytotoxicity, with Daudi being substantially more sensitive. There were also, on the part of these cell lines, major differences in susceptibility to the nonspecific toxicity of the nonreactive radiolabeled Abs. Raji and RL, of the cell lines tested, were the least sensitive to nonspecific toxicity, whereas Daudi was the most sensitive, and Ramos was intermediate. The specificity of killing can be estimated by a Specificity Index, determined at the level of 3 logs of cell kill (chosen arbitrarily), and defined as the initial concentration of nonreactive Ab required for this level of killing, divided by the initial concentration of 1F5 required. The Specificity Index was only 1.9 for Ramos (meaning that the nonreactive Ab was almost as potent as 1F5) but was 16.8 for Raji, 17.9 for Daudi and 55.9 for RL. The X-axis scale of Fig. 9
, selected for optimal clarity, does not extend far enough to show 3 logs of nonspecific kill of Raji and RL, but higher Ab concentrations were tested so that this level of killing was obtained. For Raji, this required 698 µCi/ml starting concentration, and for RL, 777 µCi/ml. Unlabeled 1F5 at the same or higher protein concentrations was tested on all of these cell lines; the test did not result in detectable cytotoxicity. By comparing these results with the results of immunofluorescence, it appears that there is no correlation between internalization of the Ab and susceptibility to toxicity with 125I-labeled 1F5. That is, the cells with relatively little internalization, as detected by immunofluorescence, were killed more effectively and more specifically than the cells that had the most prominent internalization. This does not prove that internalization has no effect on toxicity, but rather that other factors are generally more important.
|
, but the staining was not as bright or homogeneous as with 1F5, and there were more cells with very little or no staining of the JN spot. In addition, there was very bright staining of cellular extensions [possibly uropods (28)
] on 5–10% of the cells, and bright staining of intercellular junctions on the rare cells that were in clusters (not shown). Such staining patterns were also observed occasionally with 1F5 but appeared to be more common with rituximab. By FACS analysis, there was no significant increase in fluorescence as the Ab incubation, at a saturating concentration, increased from 1 to 24 h, which suggests that the level of intracellular uptake was much less than with 1F5. Cytotoxicity was also tested with 125I-labeled rituximab on Raji, to determine whether other anti-CD20 Abs would also be potent toxic agents; killing was very similar to that obtained with 125I-labeled 1F5; and, again, the unlabeled Abs, tested at an equal or greater concentration, had no toxic effect (data not shown).
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Another pattern is more diffuse, with the Golgi cisternae, e.g., being distributed surrounding the nucleus but still relatively close to the nucleus. Anti-CD20 has been previously considered to be noninternalizing (10) or very slowly internalizing (11) . The discordant conclusions can probably be attributed to differences in the cell lines used, and/or to differences in the methods. (This earlier work was done with the same Ab used herein, 1F5). The Daudi cell line, which was primarily used by Press et al. (10) , in fact, displays much less intracellular uptake than does Raji or Ramos, as detected by immunofluorescence, although a low level of uptake was observed in our experiments. In their electron microscopy studies in Daudi cells, Press et al. used the Fab fragment, which we have found to reach the JN spot, much more faintly than an intact Ab. Thus, it is not surprising that this method did not detect Ab internalization. The other assay used by Press et al., low pH elution, has a basic problem that was not previously recognized, the fact that it lyses most of the cells, under the conditions that are usually used (30) . Therefore, it can potentially extract material from inside the cell as well as on the cell surface. In fact, we confirmed the previous results that virtually all bound 125I-labeled 1F5 is extracted from Raji cells by low pH, under conditions in which a large fraction of the bound Ab is known to be at the JN spot.4 Milder elution conditions (pH 3.0) did not efficiently remove the Ab from the cell surface. From our previous analysis of the low pH elution method, it was concluded that the method is successful in many cases, despite cell lysis, probably because the low pH buffer acts as a fixative and fixes intracellular protein ligands faster than they are eluted (30) . However, it is clear that extraction by low pH cannot be considered to be proof that an Ab is on the cell surface. For example, we demonstrated previously that low-molecular-weight catabolic products trapped within lysosomes were efficiently released from cells by low pH treatment (30) . This example does not explain the results with 1F5, which remains intact inside the cell, but it does show that intracellular material can be extracted by low pH treatment. We cannot currently explain why 1F5 behaves differently from many other internalized Abs in its extraction by low pH, but the major possibilities appear to be (a) 1F5 is transported to an unusual intracellular site (which is, in fact, the case); or (b) 1F5 may be more susceptible to low pH elution than most other Abs (thus, it is eluted from intracellular sites faster than it is fixed). In our previous study, Ab internalization was monitored by lysosomal catabolism of the bound Ab (11) . Although this is a reliable assay of internalization for most Abs, which are delivered efficiently to lysosomes after internalization, it clearly is not useful for Abs that are transported to a noncatabolic compartment, like 1F5.
The intracellular spots illuminated by Alexa Fluor 488–1F5 were also illuminated by rhodamine red-X-Tf, which suggests that the Ab is following the endocytic recycling pathway. However, significant differences between the uptake of 1F5 and Tf were also noted. First, the rate of uptake was markedly different. 1F5 uptake was slow, with a 16–18-h incubation required for maximum uptake into the intracellular spots. At 4 h, there was only partial and relatively faint uptake. In contrast, uptake of Tf into the JN spot reached a peak in 45 min with Raji cells, and in 4 h with Ramos cells. A second difference is that a higher fraction of the CD20 antigen is on the cell surface, because ringed staining was still prominent after an overnight incubation. In contrast, Tf showed much weaker cell surface staining, in comparison with the brightness of the intracellular spots, after incubations of 30 min or longer. The gradual and continuous accumulation of fluorescent 1F5 over a 24-h period, as assayed by immunofluorescence and by FACS analysis, is similar to the accumulation of radiolabeled Abs. This correlation strongly suggests that the high level of accumulation of radiolabeled Abs by Raji cells can be attributed to transport to the JN spot.
A basic question is whether the CD 20 antigen is normally present in the intracellular spots observed, or whether transport to the site is dependent on bivalent Ab binding. There are several possible approaches to address this question, but it has not yet been answered. One approach is to test a Fab Ab fragment, which will indicate whether bivalent binding is required. Our experiments with the Fab fragment of 1F5 showed much less delivery to the intracellular spots than with intact Abs. This can be considered to suggest a need for bivalent binding but might also be attributable to lower avidity. Another approach is to stain intracellular CD20 antigen in fixed and permeabilized cells. We have attempted this, but staining has not yet been achieved, which might be attributable to technical problems with the choice of fixative and permeabilizing agent. In the case of TfR, it is known that the molecule remains inside the cell for a relatively long period after the release of iron into the cytoplasm, presumably because the sorting of molecules from the ERC back to the cell surface appears to be the rate-limiting step in the transport pathway of this molecule (31) .
The extensive and detailed overlap of internalized 1F5 and Tf strongly suggests that they localize to the same compartment, although we have not shown directly that they are in the same vesicles. Whereas the Golgi cisternae have essentially the same JN appearance in Raji cells, as demonstrated by staining with BODIPY-FL-C5-ceramide, the results with BFA strongly suggest that 1F5 does not localize to the Golgi cisternae, because the 1F5 JN spot was not dispersed. Moreover, in Ramos cells, the Golgi region stained with BODIPY-FL-C5-ceramide was distinct from the pattern stained with either 1F5 or Tf. Tf, by definition, labels the ERC (32) , but the relationship of this compartment to the TGN should be discussed. We have not tested markers for the TGN because the more basic question is whether these two compartments are distinct. Slight differences between localization of two markers does not necessarily allow the conclusion that they reflect two distinct compartments. As an example, we note that the most prominent difference between the localization of 1F5 and Tf, in Raji and Ramos cells, was the much brighter cell-surface staining with 1F5. Cell-surface staining with Tf was barely detectable. Yet it would be clearly erroneous to conclude that Tf does not go to the cell surface, because the nature of the TfR is well established (6) , and the results can be attributed to two factors: (a) Tf uptake is much faster, so that in the time required to wash away unbound ligand and prepare the slide for observation, most of the Tf originally on the cell surface is already internalized; and (b) probably a smaller fraction of the total TfR is present on the cell surface, relative to CD20. Although some experiments have been interpreted to suggest that the TGN and ERC are distinct (32, 33, 34) , the apparent differences can potentially be explained by differences in the rates of uptake or in the steady-state distribution of the two markers. Moreover, Johnson et al. (33) used C6-NDB-ceramide as a marker for the TGN, but this, in fact, is primarily a marker for the trans-Golgi cisternae (27 , 35) .
The most common destination for Abs binding to the cell surface is lysosomes, leading to catabolism (36) . For many or all ligands binding to the cell surface, there appears to be a minor fraction that escapes catabolism and instead is routed to the TGN or the ERC (37) . However, a small group of Abs accumulate predominantly in the TGN after binding to the cell surface. There are no reports, to our knowledge, of Abs accumulating in the ERC. It is interesting to note that Abs to TfR do not follow the same pathway as Tf, but rather are mostly delivered to lysosomes, probably as a consequence of bivalent binding (6) . Examples of Abs that go to the TGN are Abs to TGN38, metallocarboxypeptidase D, furin, and the MAL protein (19 , 32 , 34 , 38 , 39) . Some of these are enzymes known to be active in the TGN (34 , 38) . The copper transporter MNK also cycles between the cell surface and the TGN but has, to date, only been detected on the cytoplasmic side of the plasma membrane (40) . This molecule has considerable similarity to CD20, being a transport protein with multiple transmembrane domains. The distribution of MNK between the TGN and the cell surface is regulated by ligand binding (40) , which, therefore, provides a precedent to suggest that CD20 may transit the same pathway in different cell lines, yet have a different steady-state distribution within that pathway. The cytoplasmic domain of CD20 has a large cluster of acidic residues similar to that on MNK and furin, which was identified as a signal for internalization of furin from the cell surface to the TGN (41) . CD20 also has the di-leucine sequence that was identified as the signal for the transport from cell surface to TGN of the MNK protein (42) . Some of these proteins, like CD20, are present selectively in glycolipid-enriched membrane microdomains (39) . The full significance of the cellular distribution of CD20 can only be appreciated when the function of the molecule is known.
The difficulty in saturating CD20 sites on the cell surface at 4°C should be noted. This may simply reflect a slow binding reaction. The CD20 antigen has only a small loop exposed on the exterior of the cell and, thus, may be somewhat inaccessible to Ab. However, we have now documented problems with 4°C-Ab-binding with Abs to three different antigens (13 , 43) . The effect was most dramatic with the anti-CD22 Ab LL2, which bound well at 37°C but was essentially unreactive at low concentrations at 4°C (43) . Thus, the widespread assumption that 4°C simply "freezes" the cell in time may not be justified. We note that previous investigations of CD20 sites per cell also did not use 4°C incubations, although the reason for this was not given. Press et al. (10) used room temperature, and the temperature used by Vervordeldonk et al. (22) was not given. In any case, the binding that occurs under physiological conditions, i.e., 37°C, is most important.
Considering that radiolabeled anti-CD20 has been extensively used for therapy of B-cell lymphoma in patients (2
, 44)
, there may be important clinical implications of these results. Whereas high-energy ß-particle emitters have been generally selected for treatment of macroscopic disease, we have argued previously that Auger electron emitters may have distinct advantages for treatment of micrometastases. The localization of 1F5 at a site close to the nucleus will enhance the delivery of radiation from low-energy electrons to the nucleus (which is the important target). Although the advantage of cytoplasmic versus cell-surface localization of a radionuclide is 2-fold for the isotopes that we have used (3)
, this advantage would be expected to be somewhat greater for JN localization. The magnitude of this effect was analyzed for a generally similar case of Ab localization to "macropinosomes," and was estimated to be 3-fold (45)
. Despite these speculations, it should be emphasized that, in our experiments, there was no correlation between 1F5 internalization and susceptibility to killing with 125I-labeled 1F5 for the four cell lines tested. The cell line that had weak specific killing, Ramos, was one of the two cell lines that showed brightly stained intracellular vesicles; and the cell line with very little intracellular uptake, RL, was very sensitive to killing and had the highest specificity index. Therefore, intracellular uptake of the Ab does not appear to be required for high levels of cytotoxicity. However, it is clear that many diverse factors affect the sensitivity of a cell line to killing by 125I-labeled 1F5, and it remains likely that a high level of intracellular uptake will enhance toxicity. Our results also suggest that the potential use of drug or toxin conjugates of anti-CD20 Abs should be reevaluated. Because these Abs have been regarded as noninternalizing, such conjugates were not considered to be promising. Some negative results with toxin conjugates of anti-CD20 were reported (46)
. However, it seems worthwhile to attempt to use other conjugates that may be effective when delivered to a noncatabolic intracellular site. Finally, the differences between anti-CD20 Abs should be noted. Whereas rituximab showed some internalization, the strongest internalization was with 1F5, of the four Abs tested. (However, only 1F5 and rituximab were tested as Alexa Fluor 488 conjugates, which provided the most sensitive assay.) Differences between 1F5 and other CD20 Abs were reported previously (7)
. Radiolabeled Ab 1F5 was used in some of the early clinical trials with anti-CD20 (47)
and appeared to be comparable in efficacy to B1.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported in part by NIH Grant CA87059 and Department of Energy Grant FG02-01ER63191 (to M. J. M.). 
2 To whom requests for reprints should be addressed, at Center for Molecular Medicine and Immunology, 520 Belleville Avenue, Belleville, NJ 07109. Phone: (973) 844-7013; Fax: (973) 844-7020; E-mail: mjmattes{at}gscaheer.org
3 The abbreviations used are: Ab, antibody; TGN, trans-Golgi network; ERC, endocytic recycling compartment; JN, juxtanuclear; DLT, dilactitol-tyramine; Tf, transferrin; TfR, Tf receptor; FACS, fluorescence-activated cell sorting; BFA, brefeldin A.
Received 3/ 4/02; revised 5/25/02; accepted 5/30/02.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. Perosa, E. Favoino, C. Vicenti, A. Guarnera, V. Racanelli, V. De Pinto, and F. Dammacco Two Structurally Different Rituximab-Specific CD20 Mimotope Peptides Reveal That Rituximab Recognizes Two Different CD20-Associated Epitopes J. Immunol., January 1, 2009; 182(1): 416 - 423. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Mattes, R. M. Sharkey, H. Karacay, M. S. Czuczman, and D. M. Goldenberg Therapy of Advanced B-Lymphoma Xenografts with a Combination of 90Y-anti-CD22 IgG (Epratuzumab) and Unlabeled Anti-CD20 IgG (Veltuzumab) Clin. Cancer Res., October 1, 2008; 14(19): 6154 - 6160. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. Qu, D. M. Goldenberg, T. M. Cardillo, V. Shi, H. J. Hansen, and C.-H. Chang Bispecific anti-CD20/22 antibodies inhibit B-cell lymphoma proliferation by a unique mechanism of action Blood, February 15, 2008; 111(4): 2211 - 2219. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. B. Michel, A. V. Rosario, P. M. Andrews, D. M. Goldenberg, and M. J. Mattes Therapy of Small Subcutaneous B-Lymphoma Xenografts with Antibodies Conjugated to Radionuclides Emitting Low-Energy Electrons Clin. Cancer Res., January 15, 2005; 11(2): 777 - 786. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C.-L. Law, C. G. Cerveny, K. A. Gordon, K. Klussman, B. J. Mixan, D. F. Chace, D. L. Meyer, S. O. Doronina, C. B. Siegall, J. A. Francisco, et al. Efficient Elimination of B-Lineage Lymphomas by Anti-CD20-Auristatin Conjugates Clin. Cancer Res., December 1, 2004; 10(23): 7842 - 7851. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Jilani, S. O'Brien, T. Manshuri, D. A. Thomas, V. A. Thomazy, M. Imam, S. Naeem, S. Verstovsek, H. Kantarjian, F. Giles, et al. Transient down-modulation of CD20 by rituximab in patients with chronic lymphocytic leukemia Blood, November 15, 2003; 102(10): 3514 - 3520. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. B. Michel, M. W. Brechbiel, and M. J. Mattes A Comparison of 4 Radionuclides Conjugated to Antibodies for Single-Cell Kill J. Nucl. Med., April 1, 2003; 44(4): 632 - 640. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Pagel, N. Hedin, K. Subbiah, D. Meyer, R. Mallet, D. Axworthy, L. J. Theodore, D. S. Wilbur, D. C. Matthews, and O. W. Press Comparison of anti-CD20 and anti-CD45 antibodies for conventional and pretargeted radioimmunotherapy of B-cell lymphomas Blood, March 15, 2003; 101(6): 2340 - 2348. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
). Results shown are means ± SDs of duplicates, and are representative of two to three experiments with each radiolabel.
, specific binding, which was calculated by subtraction: total - nonspecific. A, results at 4°C; B, results of an identical experiment except at 37°C. Values shown are means ± SDs of duplicates and are representative of two to three experiments, using slightly different Ab concentrations.
