Anti-tumor antibody produced by human tumor-infiltrating and peripheral blood B lymphocytes (original) (raw)

Anti-tumor antibody produced by human tumor-infiltrating and peripheral blood B lymphocytes

Cornelis J. A. Punt 1{ }^{1}, Jose A. M. Barbuto 1{ }^{1}, Hua Zhang 1{ }^{1}, William J. Grimes 2{ }^{2}, Kenneth D. Hatch 3{ }^{3}, Evan M. Hersh 1{ }^{1}
1{ }^{1} Section of Hematology and Oncology, Arizona Cancer Center, University of Arizona, Tucson, AZ 85724, USA
2{ }^{2} Department of Biochemistry, Arizona Cancer Center, University of Arizona, Tucson, AZ 85724, USA
3{ }^{3} Department of Obstetrics and Gynecology, University of Arizona, Tucson AZ 85724, USA

Received: 17 September 1993 / Accepted: 10 November 1993

Abstract

Cell suspensions from 69 human tumor biopsies and malignant effusions depleted of infiltrating T cells were incubated for 10-14 days with mitomycin-C-treated cells of the transformed T cell line MOT as feeder cells. B lymphocytes proliferated and differentiated as indicated by immunoglobulin (Ig) secretion in the culture supernatants ( B cell expansion). Ig was present in culture supernatants of tumor cell suspensions incubated without MOT feeder cells (non-expanded cells), but the addition of MOT feeder cells to these cultures invariably resulted in a significant increase in Ig concentration. While IgG, IgA, and IgM isotypes were all detected in supernatants of both expanded- and nonexpanded tumor cell suspensions, the increase in total Ig induced by MOT feeder cells was mainly due to an increase in IgG. Peripheral blood B lymphocytes (PBBL) from 15 cancer patients and 4 healthy individuals were also successfully expanded by the same method. In these it was shown that IgA was the predominant Ig isotype. Using a modified enzyme-linked immunosorbent assay, IgG of 25/25 / 36 expansions from tumor cell suspensions showed reactivity with autologous tumor targets, and that from 10/13 expansions reacted with allogeneic tumor targets of the same histological diagnosis. No reactivity was found against tumor targets of different histology. IgG of 4/104 / 10 expansions of PBBL from cancer patients showed reactivity with allogeneic tumor targets of the same histology, while no reactivity was demonstrated against tumor targets of different histology. IgG of expanded PBBL from healthy individuals showed no reactivity against tumor targets. This method allows detailed study of the specific humoral an-

[1]titumor immune response of intratumoral and peripheral blood B lymphocytes in cancer.

Key words: B lymphocytes - Tumor antibodies - Tumorinfiltrating lymphocytes - Immunoglobulin - Immunoglobulin isotypes

Introduction

Studies on tumor-infiltrating lymphocytes (TIL) have focused almost exclusively on T cells (reviewed in [1]). Tumor-infiltrating B lymphocytes (TIL-B) have been shown to account for 0−13%0-13 \% of the total lymphoid infiltrate [2-11]. Hybridomas producing tumor-specific antibodies have been produced from TIL [12, 13], which suggests a specific function of TIL-B. The main reason for the paucity of data on B lymphocytes from tumors is the limitation of assays that were available to study B cell function in vitro. These include the plaque forming cell assay, which does not allow expansion and further study of B lymphocytes [14], and Epstein Barr virus (EBV) transformation, which may be selective and has a low efficacy [15-17]. Recently Lipsky’s group [18, 19] demonstrated that peripheral blood B lymphocytes (PBBL) could be activated by culture in the presence of anti-CD3-stimulated human T cells, and that this response could be further enhanced by the addition of interleukin-2 (IL-2) or IL-6. This method induced Ig secretion of PBBL at a high frequency as well as production of multiple Ig-H chain isotypes by individual PBBL [20]. Subsequently we have demonstrated that the transformed T cell line MOT did not require any exogenous stimuli to induce maximal Ig secretion of B lymphocytes and that with MOT feeder cells even higher levels of B cell response could be achieved [21]. This facilitates the analysis of B cell function under pathological conditions such as cancer and AIDS, where an altered T cell population might negatively affect the stimulation of B cells [22-24]. Using this method of B cell expansion we demonstrate in the current

1. Work supported by grants from the Share and Concern Foundations and grant CA MOPP from the National Institutes of Health. C.J.A.P. is a visiting scientist from the University of Nijmegen, Department of Medical Oncology, Nijmegen, The Netherlands, and is supported by a Fulbright Senior Research Grant and grants from the Dutch Cancer Society and the Regional Cancer Center of the East Netherlands (IKO). J.A.M.B. is a visiting scientist from the University of Sao Paolo, Department of Immunology, Brazil, and is supported by grant 90/ 1844−41844-4 from the FAPESP
   Correspondence to: E.M. Hersh ↩︎

study that a humoral antitumor immune response may occur in cancer patients, since the Ig that was secreted among TIL-B and PBBL from cancer patients reacted to a considerable extent with autologous and allogeneic tumor targets of the same histological type.

Materials and methods

Cell preparation. Histological verification of malignancy was obtained from all tumor samples. Fresh tumor specimens were minced with a scalpel and digested with 0.015%0.015 \% (w/v) DNAse (Sigma, St.Louis, Mo.) and 0.15%(w/v)0.15 \%(\mathrm{w} / \mathrm{v}) collagenase (Sigma, St.Louis, Mo.) for 1−2 h1-2 \mathrm{~h} at 37∘C37^{\circ} \mathrm{C}. Tumor cell suspensions from solid tissues and malignant effusions were cultured in tissue-culture flasks (Costar, Cambridge, Mass.) in RPMI1640 medium supplemented with 10%10 \% fetal calf serum (Tissue Culture Biologicals, Tulane, Calif.), 2 mM L-glutamine (Irvine Scientific, Santa Ana, Calif.), 100 U/ml penicillin and 100μ g/ml100 \mu \mathrm{~g} / \mathrm{ml} streptomycin (Irvine Scientific) (cRPMI) for 1 h overnight in a humidified atmosphere of 5%5 \% CO2\mathrm{CO}_{2} at 37∘C37^{\circ} \mathrm{C}. Mononuclear cells from the non-adherent fraction were isolated by centrifugation over Ficoll-Paque (Pharmacia, Piscataway, N.J.). T cells were removed from the mononuclear cell fraction by rosetting with sheep red blood cells (M.A. Bioproducts, Walkersville, Md.) treated with SS-(2-aminoethyl)isothiouronium bromide hydrobromide (AET) (Aldrich, Milwaukee, Wis.) [25], and the non-rosetting cells were isolated by centrifugation over Ficoll-Paque. The resultant BB-cell-enriched cell suspension from tumors will be further denoted as BEST.

Peripheral blood mononuclear cells were prepared from heparinized blood of cancer patients and healthy volunteers by centrifugation over Ficoll-Paque. Peripheral blood mononuclear cells were resuspended in serum-free RPMI-1640 medium and treated with L-leucine methyl ester (Sigma, St. Louis, Mo.) in order to deplete monocytes and natural killer (NK) cells [26]. B cells were then purified by rosetting with AET-treated sheep red blood cells followed by separation of nonrosetting cells by centrifugation over Ficoll-Paque [25].

MOT cells, a transformed T cell line derived from leukemic cells from a patient with hairy-cell leukemia, were a gift from Dr. Paul Fiorino (CDC, Atlanta, Ga.) and were used as feeder cells. MOT cells were resuspended in Hanks’ balanced salt solution (HBSS) at a concentration of 4×106−5×1064 \times 10^{6}-5 \times 10^{6} cells /ml/ \mathrm{ml}. Mitomycin C (Sigma, St. Louis, Mo.) was added to a final concentration of 40μ g/ml40 \mu \mathrm{~g} / \mathrm{ml} and cells were incubated in the dark under agitation for 30 min at 37∘C37^{\circ} \mathrm{C}, and then washed three times with cRPMI.

Melanoma cell lines used as tumor targets were DH903, a low (fewer than 15)-passage cell line established at the University of Arizona, and A375 M, a high(more than 30)-passage cell line. Ovarian carcinoma, non-small-cell lung cancer (NSCLC), colon carcinoma, and renal-cell carcinoma cell lines were low-passage cell lines and established from patients at the tissue-culture laboratory of the Arizona Cancer Center.

Preparation of plates with fixed cells. Fixing tumor cells to plates was performed after a modification of the method of Bishop and Hwang [27]. Flat-bottom 96 -well tissue-culture plates (Costar, Cambridge, Mass.) were incubated with poly-L-lysine hydrobromide, Mr>300000M_{\mathrm{r}}>300000 (Sigma, St. Louis, Mo.) diluted 5mg/ml5 \mathrm{mg} / \mathrm{ml} in phosphate-buffered saline (PBS), 100μl/100 \mu \mathrm{l} / well, for 30 min at room temperature. The fluid was removed and 150 000-200 000 tumor cells/well, suspended in HBSS, were plated. Plates were subjected to centrifugation at 100 g for 5 min , and the supernatant was carefully removed. Glutaraldehyde (Sigma, St. Louis, Mo.) 0.25%0.25 \%, freshly diluted in PBS ( 200μl/200 \mu \mathrm{l} / well), was added and incubated for 5 min at room temperature. After being washed gently once with 0.05%0.05 \% Tween 20 in TRIS-buffered saline (TBS-T) and once with TRIS-buffered saline (TBS) plates were stored in the dark at 4∘C4^{\circ} \mathrm{C} with 200μl/200 \mu \mathrm{l} / well PBS /0.1%/ 0.1 \% bovine serum albumin 0.02%0.02 \% sodium azide.

B cell expansion assay. BEST or PBBL were resuspended in cRPMI and plated in sterile 96 -well U-bottom tissue-culture plates (Costar,

Cambridge, Mass.). Mitomycin-C-treated MOT cells were added as feeder cells at 50000 cells // well [21]. The final volume of each well was 200μl200 \mu \mathrm{l}, and 60 wells/plate were used. Twelve wells that were plated with MOT feeder cells served as controls for the production of Ig in the 48 wells that were plated with BEST or PBBL together with MOT feeder cells. In the majority of B cell expansions of tumors, 18 wells/plate were plated with BEST without MOT feeder cells in order to determine the presence and release of Ig from tumor cells, further denoted as endogenous Ig. Expanded BEST/PBBL will denote the MOT feeder cell-stimulated expansion of BEST/PBBL, and non-expanded BEST the culture of BEST without MOT feeder cells. After an incubation period of 10−1410-14 days at 37∘C37^{\circ} \mathrm{C} in a humidified atmosphere of 5%CO2,150μl5 \% \mathrm{CO}_{2}, 150 \mu \mathrm{l} cell-free culture supernatant was withdrawn from each well and analyzed for Ig.

Enzyme-linked immunosorbent assay (ELISA) for Ig detection. Falcon flexible 96 -well assay plates (Becton Dickinson, Lincoln Park, N.J.) were coated with 50μl/50 \mu \mathrm{l} / well goat polyclonal anti-(human Ig) (Tago, Burlingame, Calif.) diluted to 1μ g/ml1 \mu \mathrm{~g} / \mathrm{ml} in carbonate buffer, pH 9.6 , for a minimum of 4 h at room temperature. Wells were blocked overnight at 4∘C4^{\circ} \mathrm{C} with TBS 1%1 \% bovine serum albumin (BSA). Wells were washed once with TBS-T and once with TBS. Then 30μl/30 \mu \mathrm{l} / well culture supernatant from B cell expansions or human Ig standard was added, and plates were incubated for 1.5−2.5 h1.5-2.5 \mathrm{~h} at room temperature. After being washed three times with TBS-T and once with TBS, plates were incubated with a mixture of alkaline-phosphatase-conjugated goat polyclonal anti-(human κ\kappa and λ\lambda light chain) (Tago, Burlingame, Calif.) diluted 1:10 000 in TBS 1%1 \% BSA, 50μl/50 \mu \mathrm{l} / well, for 1.5−2.5 h1.5-2.5 \mathrm{~h} at room temperature. For detection of Ig isotypes (IgG, IgA, IgM) the appropriate alkaline-phosphatase-conjugated antibodies (Tago, Burlingame, Calif.) at 1:10 000 dilutions were used. After three washes with TBS-T and one with TBS, 50μl/50 \mu \mathrm{l} / well substrate pp-nitrophenyl phosphate (Gibco, Grand Island, N.Y.), diluted 1mg/ml1 \mathrm{mg} / \mathrm{ml} in carbonate buffer, pH 9.6 , was added. Plates were read after 1−2 h1-2 \mathrm{~h} in a microplate reader (Dynatech MR600) at 410 nm . Supernatants were considered positive for Ig by an absorbance that was more than 3 SD above background, established by determining the mean absorbance from the 12 wells/ plate containing MOT cells alone. In experiments in which endogenous Ig from tumors also served as a control, the Ig production by expanded BEST was considered positive when the absorbance was more than 2 SD above the mean absorbance of supernatants from non-expanded BEST.

Subtyping of infiltrating lymphocytes. The percentage of B lymphocytes in BEST was determined after immunohistochemical staining with anti-Leu12 (CD19) (Becton Dickinson, San Jose, Calif.). Field selection sought areas of highest CD19 expression evident by lowerpower scanning, and the number of positive cells among a total number of 500 cells were counted within high-power ( 40×40 \times objective) microscopic fields.

Calculation of responder cells. The frequency of responding cells (cells producing immunoglobulin) was calculated according to the Poisson distribution based on a linear regression curve constructed with the frequency of wells negative for Ig at each cell concentration [28]. The frequency of response [f(t)][f(t)] was calculated by plating at limiting dilutions of 78−5000078-50000 cells/well for the expansion of BEST, and 8−20008-2000 cells/well for the expansion of PBBL, 12-24 wells being plated with the same number of cells in each experiment.

Statistics. Experimental data were statistically evaluated using the paired and unpaired Wilcoxon test.

ELISA for detection of Ig reactivity against tumor cells. Plates with fixed tumor cells were washed once with TBS-T and once with TBS and incubated with supernatants from B cell expansions, 30−60μl30-60 \mu \mathrm{l} / well, for 3−4 h3-4 \mathrm{~h} at 37∘C37^{\circ} \mathrm{C}. After three washes with TBS-T and one with TBS, plates were incubated with goat anti-(human IgG), γ\gamma-chain-specific, urease-conjugated (Sigma, St. Louis, Mo.), diluted 1:1000 in TBS 1%1 \% BSA, 100μl/100 \mu \mathrm{l} / well, for 2−4 h2-4 \mathrm{~h} at 37∘C37^{\circ} \mathrm{C}. After plates had been washed twice with TBS-T and twice with distilled deionized water,

Fig. 1 A, B. Effect of expansion of B-cell-enriched suspension (BEST) derived from a melanoma (A) and an ovarian carcinoma (B) on Ig concentration in culture supernatants. For each cell number the mean Ig concentration in 24-48 supernatants of expanded, and in 8-16 supernatants of non-expanded cells is shown. The difference in Ig concentration between supernatants from non-expanded and expanded BEST was statistically significant ( P<0.02P<0.02 )
100μl/100 \mu \mathrm{l} / well urease substrate was added [29]. Plates were read after 1-6 h in a microplate reader (Dynatech MR600) at 570 nm . All assays were performed in duplicate. In each plate, 12 wells, incubated with supernatants from wells plated with MOT feeder cells only, served as controls for the 48 wells incubated with supernatants of Ig-positive wells from B cell expansions. Wells were considered positive when the absorbance was more than 2 SD above the absorbance of control wells in duplicate assays. Generally, from each B cell expansion a total of 24-96 Ig-positive supernatants were tested for reactivity.

Results

Expansion of BEST

A total of 69 human tumor specimens (solid tissues and malignant effusions) were processed. Ig was detected in the supernatants of B cell expansions from all 69 tumor specimens, and there was a good correlation between the Ig con-

Table 1. Histological diagnosis and frequency of response [f®] in expanded B-cell-enriched suspension (BEST) from 44 tumors

r2\mathrm{r}^{2} denotes correlation coefficient; NSCLC, non-small-cell lung cancer

Table 2. Frequency of response [f®] in CD19+ cells in expanded B-cell-enriched suspension (BEST) from 13 tumors

centrations and the number of plated cells, as shown in Fig. 1. In 41 tumor specimens the f(r)f(\mathrm{r}) of BEST was calculated (Table 1). The median f(r)f(\mathrm{r}) was 1/14561 / 1456 (range 1/147−1/180581 / 147-1 / 18058 ) with a median correlation coefficient (r2)\left(r^{2}\right) of 0.95 (range 0.72−0.990.72-0.99 ). The percentage of BB cells in the total number of cells from BEST was determined by immunohistochemical staining with anti-CD19 in 24 specimens: 11 ovarian carcinomas, 5 melanomas, 2 colorectal carcinomas, and 1 each NSCLC, neuroblastoma, soft-tissue sarcoma, and pancreas-, renal-cell-, and breast carcinoma. The median percentage of CD19+ cells in these specimens was 0.6%0.6 \% (range 0.2−14.60.2-14.6 ). In 13 of these specimens the f(r)f(\mathrm{r}) of CD19+ cells was calculated, and was 1/121 / 12 (range 1/1−1/911 / 1-1 / 91 ) (Table 2).

Fig. 2. Effect of expansion of peripheral blood B lymphocytes (PBBL) from cancer patients (A-C: pretreated-, D, E: not treated with chemotherapy) on Ig concentration in culture supernatants. For each cell number the mean Ig concentration in 12-24 supernatants is shown

Table 3. Frequency of response [f(r)][f(\mathrm{r})] in expanded peripheral blood B lymphocytes (PBBL) from 8 cancer patients

r2r^{2} denotes correlation coefficient

Culture supernatants of non-expanded BEST from 41 tumors were also examined in order to determine the presence of endogenous Ig released into culture supernatants. It was found in the supernatants of all 41 specimens (data not shown). Using this endogenous Ig as a control for B lymphocyte expansion instead of the supernatant of MOT feeder cells, it was still possible to identify multiple-Igpositive supernatants in all experiments. A typical example is shown in Fig. 1 where a significant difference (P<0.02)(P<0.02) is shown in Ig concentrations between non-expanded and expanded suspensions of BEST in two different tumors at different numbers of plated cells.

Expansion of PBBL from cancer patients and healthy individuals

PBBL were isolated from 19 individuals: 8 patients with melanoma ( 4 of whom showed no evidence of disease at the time of blood sampling), 6 patients with NSCLC, 1 patient with colorectal cancer (no evidence of disease), and 4 healthy individuals. After expansion of PBBL, Ig was detected by ELISA in the culture supernatants of all specimens, and again there was a good correlation between the Ig concentration in supernatants and the number of plated cells (Fig. 2). In the 8 PBBL specimens from cancer patients the f(r)f(\mathrm{r}) was calculated, and the median f(r)f(\mathrm{r}) was 1/671 / 67 (range 1/23−1/1981 / 23-1 / 198 ) with a

Fig. 3. Ig isotypes in supernatants from expanded BEST from an ovarian carcinoma (A) and two melanomas (B, C). Numbers of cells/ well are 10000(A,C)10000(\mathbf{A}, \mathbf{C}) and 20000(B)20000(\mathbf{B}). Results are expressed as the percentage (mean+SD) absorbance of the mean absorbance in control supernatants from wells plated with MOT feeder cells only ( 100%100 \% ). The figures below the bars represent the percentage of positive wells for that particular isotype

Fig. 4. Ig isotypes in supernatants from non-expanded and expanded BEST from a breast carcinoma. The difference in the level of IgG between non-expanded- and expanded BEST was statistically significant (P<0.05)(P<0.05). Differences in the levels of IgA and IgM were not statistically significant. For further details see legend of Fig. 3
median r2r^{2} of 0.96 (range 0.83−0.990.83-0.99; Table 3 ). The f(r)f(\mathrm{r}) was lower in PBBL from patients who previously had been treated with chemotherapy (Table 3). Four experiments in which PBBL were incubated without MOT feeder cells showed a complete absence of Ig in the culture supernatants. It is worth noting that the PBBL and BEST that we expanded were never obtained from the same patient.

Ig isotypes in B lymphocyte expansions

IgG,IgA\mathrm{IgG}, \mathrm{IgA}, and IgM isotypes were determined in the culture supernatants from expansions of 4 BEST ( 2 melanomas, 1 ovarian, and 1 breast carcinoma) and of 4 PBBL (2 NSCLC, 2 healthy individuals). In the culture supernatants from expansions of BEST IgG was present both at the highest level and at the highest frequency (Figs. 3, 4). When the effect of expansion of BEST on the Ig isotype

2000 cells/well

Fig. 5A,B. Ig isotypes in supernatants from expansions of PBBL from a healthy individual (A) and from a patient with non-small-cell lung cancer (B) at two cell concentrations. For further details see legend of Fig. 3
pattern was examined by comparing the culture supernatants of non-expanded BEST to those of expanded BEST, a significant increase ( P<0.05P<0.05 ) in the level of IgG was apparent after MOT feeder cell stimulation (Fig. 4). The Ig isotype pattern in the supernatant of expanded PBBL showed that IgA was secreted at the highest frequency, and often at the highest level as well (Fig. 5).

Reactivity of IgG from supernatants of expanded BEST with fixed tumor targets

Verification of the urease ELISA for detection of bound IgG to fixed cells was obtained as follows. Macrophages isolated from the peritoneal cavity of normal ICR mice, incubated for 90 min with 0,50,1000,50,100, and 200μ g/ml200 \mu \mathrm{~g} / \mathrm{ml} tetanus toxoid (Merieux Institute Inc., Miami, Fla.), were fixed to plates, and the plates were incubated with human serum (IgG concentration 12.5mg/ml12.5 \mathrm{mg} / \mathrm{ml} ) diluted 1:100 to 1:51 200, which was obtained from a healthy individual 11 days after a tetanus booster injection. The urease ELISA detected a

Table 4. Reactivity of IgG from supernatants of expanded BEST with autologous and allogeneic tumor targets

a Of the same histological diagnosis
b Of different histological diagnosis
significant difference ( P<0.003P<0.003 ) in bound IgG between tetanus-toxoid-sensitized macrophages and unsensitized macrophages at all concentrations of IgG and at all concentrations of tetanus toxoid (data not shown).

The reactivity of Ig-positive supernatants of non-expanded and expanded BEST was tested. The Ig-positive supernatants of non-expanded BEST from 27 different tumors ( 15 ovarian and 1 pancreatic carcinomas, 6 melanomas, 2 sarcomas, 1 Hodgkin’s lymphoma, 1 NSCLC, 1 neuroblastoma) were tested. This endogenous Ig reacted in 10(37%)10(37 \%) of these tumors ( 7 ovarian carcinomas, 2 melanomas, 1 Hodgkin’s lymphoma) with the autologous tumor. This reactivity was found only in supernatants obtained from wells incubated with high numbers of BEST (i.e. with the highest Ig concentrations), and the absorbance in the urease ELISA was less intense compared with the reactivity of supernatants from expanded BEST (data not shown). The Ig-positive supernatants of non-expanded BEST from 3 different tumors ( 2 colorectal and 1 ovarian carcinoma) were tested against allogeneic tumor targets, and none of these tested positive.

The Ig-positive supernatants from 43 expanded BEST were tested against autologous and allogeneic tumor targets (Table 4). IgG reactivity was detected in the supernatants of 25/36(69%)25 / 36(69 \%) expanded BEST to autologous tumor, and to allogeneic tumor of the same histological diagnosis in the supernatants of 10/13 (77%) expanded BEST. The supernatants of 6 expanded BEST ( 3 melanomas, 2 sarcomas, 1 ovarian carcinoma) were tested against both autologous and allogeneic tumor, and reactivity was found to both in all samples. The small volume (120μl)(120 \mu \mathrm{l}) of available culture supernatant from each well precluded more extensive testing of the same sample against multiple tumor targets. Ig-positive supernatants from expanded BEST of 9 different tumors were tested against allogeneic tumor of a different histological diagnosis, and no reactivity was found (Table 4). In the 29 expansions of BEST in which reactivity was found, reactive supernatants were observed at a median of 10%10 \% (range 2%−34%2 \%-34 \% ) of the total number of Ig-positive

Table 5.Reactivity of IgG from supernatants of expanded PBBL from cancer patients and healthy individuals against allogeneic tumor targets

a Derived from patients who were disease-free at the time of blood sampling
supernatants tested. As opposed to our results with Ig from supernatants of non-expanded BEST, we did not find a correlation between the Ig concentration in culture supernatants of expanded BEST and reactivity with tumor targets (data not shown).

Reactivity of IgG from supernatants of expanded PBBL with allogeneic tumor targets

The Ig-positive supernatants from expanded PBBL derived from 10 cancer patients and 3 healthy individuals were tested against allogeneic tumor targets (Table 5). IgG reactivity was demonstrated in the supernatants of 4/104 / 10 ( 40%40 \% ) expanded PBBL when supernatants were tested against tumor targets with the same histological diagnosis, and in the supernatants of 0/40 / 4 expanded PBBL when tested against tumor targets of different histology. In the 4 expansions of PBBL from cancer patients in which reactivity was found, reactive supernatants were observed at 2%−3%2 \%-3 \% of the total number of Ig-positive supernatants tested. The supernatants from expanded PBBL from 3 healthy individuals all tested negative against 4 different tumor targets.

Discussion

We have successfully expanded the B lymphocytes derived from 69 human tumor specimens and from 19 peripheral blood samples from cancer patients and normal subjects as indicated by Ig secretion in culture supernatants. The frequency of Ig-producing cells varied greatly. We found Ig in the supernatants of non-expanded BEST as well. When this endogenous Ig was used as a control for Ig secretion by expanded cells, we still could demonstrate a significant increase in the Ig concentration in each tumor specimen, indicating a true expansion of tumor-derived B lymphocytes.

Determination of Ig isotypes in 4 expansions of BEST showed that all three main isotypes, IgG⁡,IgA⁡\operatorname{IgG}, \operatorname{IgA}, and IgM⁡\operatorname{IgM}, were present in the supernatants of both expanded and nonexpanded cells, and that the increase in total Ig induced by expansion appeared to be mainly attributable to an increase in IgG. By contrast, IgA was the predominant isotype in B
cell expansions of PBBL, indicating that the B lymphocytes that were expanded from peripheral blood and tumor sites are different populations. However these observations will have to be confirmed since only a small number of samples were tested, and BEST and PBBL were not derived from the same donors.

To determine whether IgG secreted by expanded TIL-B and PBBL showed reactivity to tumor cells, autologous and allogeneic tumor targets were incubated with Ig-positive supernatants from BEST and PBBL. Since mammalian tissues may contain considerable endogenous alkaline phosphatase and/or horseradish-peroxidase-like activity [30, 31], we chose the bacterial enzyme urease as a conjugate to the second antibody for detection of tumor-bound IgG reactivity [29]. Reactivity of isotypes other than IgG was not tested since urease-conjugated antibodies for these isotypes were not available. The urease ELISA has been proven useful for the detection and relative quantification of cell-surface antigens [27], and we have confirmed this observation by showing that this method detected reactivity of highly diluted human serum from a tetanus-vaccinated individual with fixed tetanus-toxoid-sensitized macrophages. Using this assay, IgG from the supernatant of expanded BEST was shown to react in 69%69 \% with autologous tumor targets and in 77%77 \% with allogeneic tumor targets of the same histology. No reactivity was observed with allogeneic tumor targets of different histology. IgG secreted by expanded PBBL from cancer patients reacted in 40%40 \% of cases with allogeneic tumor targets of the same histology but not of different histology. No reactivity was found of IgG derived from healthy individuals with tumor targets. Of the 4 patients whose PBBL-secreted IgG showed reactivity with tumor, 2 had had their primary tumor removed 3-6 months earlier and were disease-free at the time of blood sampling. One of these patients was undergoing adjuvant chemotherapy at the time of blood sampling. These observations indicate that a humoral antitumor response may be recalled after the tumor has been removed, and that chemotherapy may not affect the reactivity of these antibodies despite our finding that it does decrease the frequency of the B cell response. Although it could be argued that these results may be explained in part by anti-HLA reactivity induced by previous blood transfusions or pregnancies, this is not likely since 7 of 14 patients, whose IgG from expanded BEST or PBBL reacted with allogeneic tumor targets, were men who had never been transfused. The reactivity of IgG with both autologous and allogeneic tumor targets is additional evidence for the existence of shared antigens on tumor cells from different origin, as has been reported by others [32, 33]. Furthermore these results show that a significant proportion of cancer patients can mount a humoral antitumor response. Antitumor reactivity in sera from a subset of cancer patients against autologous and allogeneic tumors has been reported by several authors (reviewed in [32]), and our findings are in accordance with this. Our findings are unique in that we have demonstrated that both peripheral and tumor-derived B lymphocytes may be the source of this antitumor reactivity. The tumor-derived B cells were expanded in the presence of autologous tumor cells, but the PBBL were not. We presume that the PBBL that produced specific antibodies have been in

contact with the tumor cells, either in the tumor itself or through circulating tumor cells. This methodology will enable further characterization of both the antitumor antibodies and the B cells that produce these antibodies. Another observation worth discussing is the presence of Ig in the supernatants of non-expanded tumor cells. Two explanations may be given for the presence of this endogenous Ig. First, plasma cells may have invaded the tumor and actively secreted Ig into the culture supernatant. Second, tumor cells may have bound Ig, which is subsequently released into the culture supernatant. Support for the latter comes from our own observations that Ig can be eluted from membrane preparations of tumors by lowering the pH and subsequently passing the preparation through a protein G column (unpublished results). Similar observations have been made by others [34]. Manson recently reviewed the data on local humoral immune response in tumors [35]. In a murine tumor model he demonstrated that IgM accumulation on tumor cells concurred with a diminished ability of tumor cells to bind anti-(MHC class I) antibodies and a resistance of tumor cells to cytotoxic effector cells. This coating by IgM protected tumor cells from cytotoxic effector cells without inducing lysis by complement, antibody-directed cellular cytotoxicity, or macrophages since the growth of these murine tumor cells was not impaired. He hypothesized that tumor-specific antigens on the surface of tumor cells stimulate both a cell-mediated and a humoral response, the latter producing IgM that coats the tumor cells. This antibody-mediated epitope masking might explain why a tumor can grow progressively despite the apparent immune response that it triggers. Our results differ from these observations in that IgG and not IgM was the predominant isotype of endogenous Ig. The less intense reactivity with tumor targets that we found for endogenous Ig compared with Ig from expanded tumor-derived B cells might be explained by a difference in affinity for tumor epitopes. Occupation of IgG binding sites on fixed tumor targets by endogenous IgG is unlikely since background activity was low in the urease ELISA, and this would also preclude reactivity of IgG from expanded B cells. Further studies are necessary to explain the biological significance of this endogenous Ig.

In summary we have demonstrated that B cells from cell suspensions of tumors and peripheral blood of cancer patients can be expanded to produce Ig. The IgG subfraction showed considerable binding to autologous and allogeneic tumor targets of the same histology. Isotype patterns in Ig expanded from BEST differed from those in Ig expanded from PBBL. Using this methodology it will be possible to define further the role of this local and peripheral humoral antitumor immune response in the interplay of all the immune responses in the tumor-bearing host.

Acknowledgements. The authors which to thank Dr. J. Donald Capra, University of Texas, Southwestern Medical School, for initially alerting us to this methodology, and Dr. Thomas M. Grogan, Department of Pathology, University of Arizona, for expert assistance in reviewing the anti-CD19-stained slides.

References

1. Topalian SL, Rosenberg SA (1991) Adoptive cellular therapy: basic principles. In: DeVita VT Jr, Hellman S, Rosenberg SA (eds) Biologic therapy of cancer. Lippincott, Philadelphia, p 187

2. Hurlimann J, Saraga P (1985) Mononuclear cells infiltrating human mammary carcinomas: immunohistochemical analysis with monoclonal antibodies. Int J Cancer 35: 753

3. Whiteside TL, Miescher S, Hurlimann J, Moretta L, Von Fliedner V (1986) Clonal analysis and in situ characterization of lymphocytes infiltrating human breast carcinomas. Cancer Immunol Immunother 23: 169

4. Ferguson A, Moore M, Fox H (1985) Expression of MHC products and leucocyte differentiation antigens in gynaecological neoplasms: an immunohistological analysis of the tumour cells and infiltrating leucocytes. Br J Cancer 52: 551

5. Wolf GT, Hudson JL, Peterson KA, Miller HL, McClatchey KD (1986) Lymphocyte subpopulations infiltrating squamous carcinomas of the head and neck: correlations with extent of tumor and prognosis. Otolaryngol Head Neck Surg 95: 142

6. Schoorl R, Brutel de la Rivière A, Borne AEGKr von dem, Felt-kamp-Vroom TM (1976) Identification of T and B lymphocytes in human breast cancer with immunohistochemical techniques. Am J Pathol 84: 529

7. Husby G, Hoagland PM, Strickland RG, Williams RC (1976) Tissue T and B cell infiltration of primary and metastatic cancer. J Clin Invest 57: 1471

8. Stevens A, Klitter I, Roggendorf W (1988) Inflammatory infiltrates and natural killer cell presence in human brain tumors. Cancer 61: 738

9. Vose BM, Moore M (1979) Suppressor cell activity of lymphocytes infiltrating human lung and breast tumours. Int J Cancer 24: 579

10. Bröcker EB, Zwadlo G, Holzmann B, Macher E, Sorg C (1988) Inflammatory cell infiltrates in human melanoma at different stages of tumor progression. Int J Cancer 41: 562

11. Withford P, Mallon EA, George WD, Campbell AM (1990) Flow cytometric analysis of tumor infiltrating lymphocytes in breast cancer. Br J Cancer 62: 971

12. Sikora K, Alderson T, Philips J, Watson JV (1982) Human hybridomas from malignant gliomas. Lancet I: 11

13. Sikora K, Alderson T, Ellis J, Philips J, Watson J (1983) Human hybridomas from patients with malignant diseases. Br J Cancer 47: 135

14. Jerne NK, Henry C, Nordin AA, et al (1974) Plaque forming cells: methodology and theory. Transplant Rev 18: 130

15. Henderson E, Miller G, Robinson J, Heston L (1977) Efficiency of transformation of lymphocytes by Epstein-Barr virus. Virology 76: 152

16. Aman P, Ehlin-Henriksson B, Klein G (1984) Epstein-Barr virus susceptibility of normal human B lymphocyte populations. J Exp Med 159: 208

17. Kozbor D, Croce CH (1985) Fusion partners for production of human monoclonal antibodies. In: Engleman EG, Foung SKH, Larrick J, Raubitschek A (eds) Human hybridomas and monoclonal antibodies, Plenum, New York, p 21

18. Hirohata S, Jelinek DF, Lipsky PE (1988) T cell-dependent activation and differentiation of B cell proliferation by immobilized monoclonal antibodies to CD3. J Immunol 140: 3736

19. Amoroso K, Lipsky PE (1990) Frequency of human B cells that differentiate in response to anti-CD3-activated T cells. J Immunol 145: 3155

20. Kelly PJ, Pascual V, Capra JD, Lipsky PE (1992) Anti-CD3-stimulated T cells induce the production of multiple Ig⁡H\operatorname{Ig} \mathrm{H} chain isotypes by individual human peripheral B lymphocytes. J Immunol 148: 1294

21. Barbuto JAM, Verastegui EL, Hersh EM (1993) The use of transformed T cell lines for clonal expansion of human B cells from peripheral blood, spleen, and tumor-infiltrating lymphocytes. Hybridoma 12: 115

22. Hersh EM, Mavligit GM, Gutterman JU (1976) Immunodeficiency in cancer and the importance of immune evaluation of the cancer patient. Med Clin N Am 60: 623

23. Miescher S, Whiteside TL, Carrel S, Von Fliedner V (1986) Functional properties of tumor-infiltrating and blood lymphocytes in patients with solid tumors: effects of tumor cells and their supernatants on proliferative responses of lymphocytes. J Immunol 136: 1899

24. Lane HC, Fauci AS (1985) Immunologic abnormalities in the acquired immunodeficiency syndrome. Annu Rev Immunol 3: 477

25. Saxon A, Feldhaus J, Robins RA (1976) Single step separation of human T and B cells using AET treated sheep red cells. J Immunol Methods 90: 221

26. Thiele DL, Kurosaka M, Lipsky PE (1983) Phenotype of the accessory cell necessary for mitogen-stimulated T and B cell responses in human peripheral blood: delineation by its sensitivity to the lysosomotropic agent, L-leucine methyl ester. J Immunol 131: 2282

27. Bishop GA, Hwang J (1992) Use of a cellular ELISA for the detection of cell surface antigens. Biotechniques 12: 326

28. Henry C, Marbrook J, Vann DC, Kondlin D, Wofsy C (1980) Limiting dilution analysis. In: Mishell BB, Shiigi SM (eds) Selected methods in cellular immunology. Freeman, San Francisco, p 138

29. Chandler HM, Cox JC, Healy K, MacGregor A, Premier RR, Hurrell JGR (1982) An investigation of the use of urease-antibody conjugates in enzyme immunoassays. J Immunol Methods 53: 187

30. Douillard JY, Hoffman T, Herberman RB (1980) Enzyme-linked immunosorbent assay for screening monoclonal antibody production: use of intact cells as antigen. J Immunol Methods 39: 309

31. Suter L, Bruggen J, Sorg C (1980) Use of an enzyme-linked immunosorbent assay (ELISA) for screening of hybridoma antibodies against cell surface antigens. J Immunol Methods 39: 407

32. Hellström KE, Hellström I (1991) Principles of tumor immunity: tumor antigens. In: DeVita VT Jr, Hellman S, Rosenberg SA (eds) Biologic therapy of cancer. Lippincott, Philadelphia, p 42

33. Kawakami Y, Zakut R, Topalian SL, Stötter H, Rosenberg SA (1992) Shared human melanoma antigens. Recognition by tumorinfiltrating lymphocytes in HLA-A2.1-transfected melanomas. J Immunol 148: 638

34. Sjögren HD, Hellström I, Bansal SC, Warner GA, Hellström KE (1972) Elution of “blocking factors” from human tumors, capable of abrogating tumor-cell destruction by specific immune lymphocytes. Int J Cancer 9: 274

35. Manson LA (1991) Does antibody-dependent epitope masking permit progressive tumour growth in the face of cell-mediated cytotoxicity? Immunol Today 12: 352