CD66a Interactions Between Human Melanoma and NK Cells: A Novel Class I MHC-Independent Inhibitory Mechanism of Cytotoxicity

Cells and mAb

The cell lines used in this work are the class I MHC-negative human cell line 721.221 (23), the YTS NK tumor line, and various MHC class I-negative and -positive human melanoma cell lines. NK cells were isolated from PBL using the human NK cell isolation kit and the autoMACS instrument (Miltenyi Biotec, Auburn, CA). For the enrichment of CD66a-positive NK cells, isolated NK cells were further purified by depletion of CD16-positive NK cells using anti-CD16 mAb 3G8 and the autoMACS instrument. NK cells were grown in culture as previously described (24). The production and specificity of anti-NKp44 and NKp46 sera were previously described (16). The mAbs used in this work were mAb W632, directed against class I MHC molecules, the mAb anti-CD66 a,b,c,e Kat4c (purchased from DAKO, Carpenteria, CA), anti-CD66a mAb 5F4 (25), and the rabbit polyclonal anti-CD66a,c,e Abs (purchased from DAKO). The anti-CD99 mAb 12E7, used as a control, was a gift from A. Bernard (Hopital de L’Archet, Nice, France). The anti-CD56 mAb (BD PharMingen, San Diego, CA) was also used as control.

Cytotoxicity assay and Ig fusion proteins

The cytotoxic activity of YTS and NK cells against the various targets was assayed in 5-h ³⁵S release assays as described previously (24). In experiments in which mAb were included, the final mAb concentration was 10 μg/ml, or 40 μl/ml in cases where rabbit polyclonal Abs were used. Redirected lysis experiments were performed as previously described (11). The production of CD99-Ig, CD16-Ig, NKp30-Ig, NKp44-Ig, and NKp46-Ig fusion proteins by COS-7 cells and purification on a protein G column were previously described (15, 16, 26)

Quadruple staining

For quadruple staining, the following fluorochrome-conjugated mAbs were used: FITC-conjugated anti-CD66 Kat4c mAb (DAKO), PE-conjugated anti-CD56 mAb (BD PharMingen), and CyChrome-conjugated anti-CD3 (BD PharMingen). As the fourth color, biotinylated anti-CD16 mAb (Serotec, Oxford, U.K.) was used, followed by streptavidin-Cy5 (Jackson ImmunoResearch Laboratories, West Grove, PA) as a second reagent. To block nonspecific binding, cells were first incubated for 1 h on ice with 25% human serum and then incubated with the various Abs.

Generation of YTS and 721.221 cells expressing CD66a

The primers used for the amplification of CD66a cDNA needed for the transfection of 721.221 cells were as follows: 5′ primer, CCCAAGCTTGGGGCCGCCACCATGGGGCACCTCTCAGCC (including the HindIII restriction site), and 3′ primer, GGAATTCCTTACTGCTTTTTTACTTCTGAATA (including the EcoRI restriction site). cDNA was cloned into the pCDNA3 vector (Invitrogen, San Diego, CA) and transfected into 721.221 cells as previously described (24). For transfection into YTS cells, CD66a cDNA was amplified by PCR using the 5′ primer GGAATTCCGCCGCCACCATGGGGCACCTCTCAGCC (including the EcoRI restriction site) and the 3′ primer GCGTCGACTTACTGCTTTTTTACTTCTGAATA (including the SalI restriction site). For amplification of the CD66aTrunc cDNA, the same 5′ primer was used, with the 3′ primer GCGTCGACATCTTGTTAGGTGGGTCATT. Amplified fragments were cloned into the pBABE retroviral vector and transfected into YTS cells as previously described (26).

Expression of various lysis ligands, CD66a, and class I MHC proteins on human melanoma cells

The roles of NKp30, NKp44, NKp46, and CD16 receptors in NK recognition of various melanoma cells deficient in class I MHC expression (except from LB33melA1, used as a control) were studied by production of fusion proteins in which the extracellular domains of NKp30 NKp44, NKp46, and CD16 were fused to the Fc portion of Ig. cDNA encoding the extracellular domains of CD99 fused to the human IgG1 DNA was used as a control. The Ig fusion proteins were incubated with the various melanoma cells and analyzed for binding by indirect immunostaining as previously described (15). In general, the highest staining of the melanoma cells was observed with the NKp30-Ig and NKp44-Ig fusion proteins (Table I⇓). Little staining of all Ig fusion proteins was observed with LB33melA1 cells, a cell line that is hardly killed by NK cells (data not shown). All other cell lines that can be killed by NK cells were stained to various degrees with the Ig fusion proteins (Table I⇓).

We have previously reported that IL-2-activated CD16-negative NK cells inefficiently kill 1106mel cells (15). Surprisingly, 1106mel cells probably express the ligands for all the NK lysis receptors tested, including CD16, NKp30, NKp44, and NKp46 (Table I⇑). Thus, killing of 1106mel cells was expected to occur even when CD16-negative NK cells, which express the NKp44 and NKp46 receptors (data not shown), were used. We therefore hypothesized the existence of a class I-independent mechanism of inhibition of NK cell cytotoxicity that controls the lysis of 1106mel cells. This inhibitory mechanism should include a protein that is expressed mainly on the surface of IL-2-activated CD16-negative NK cell and is expected to deliver an inhibitory signal via the ITIM. The CD66a (carcinoembryonic Ag CAM1) protein was reported to follow the above criteria. It is expressed primarily on CD16-negative NK cells (22), it contains ITIM sequences, and it can bind in a homotypic/heterotypic manner to various CD66 proteins (19, 20). The inhibitory effect of the CD66a protein on human NK cell cytotoxicity was never investigated.

We therefore tested whether the expression of CD66a molecule can be detected on the surface of various melanoma cell lines and NK cells. Remarkably, all seven class I MHC-negative melanoma cell lines tested expressed the CD66a protein at moderate or high levels (Table II⇓). No expression of MHC class I protein (detected with the W632 mAb) was observed among these cell lines, except from the LB33melB1 cell line that expresses the HLA-A24 protein only (27). In contrast, 40% of the class I MHC-positive melanoma lines tested showed little or no staining for the CD66a protein (Table II⇓).

Recognition of CD66a expressed on 1106mel by CD66a on CD16-negative NK cells leads to inhibition of lysis

It was previously reported that the CD66a isoform is expressed on CD16-negative NK cells (22). We next tested whether 1106mel cells would be protected from lysis by CD16-negative NK cells expressing the CD66a protein. For the generation of CD16-negative NK cells expressing CD66a, NK cells were first isolated from PBL of various healthy donors and then depleted from CD16-positive NK cells by using the anti-CD16 mAb 3G8 (as described in Materials and Methods). Of 63 CD16-negative clones tested, 28 (45%) expressed the CD66a protein (data not shown). In rare cases (2%) CD66a expression could be detected on the surface of CD16-positive NK clones (see Fig. 5⇓). The percentage of CD16-negative, CD66a-positive NK cells can vary among different donors after activation. Various NK clones were then tested in killing assays against 1106mel cells. Efficient killing of 1106mel cells was observed with CD16⁺CD66a⁻ NK clones (Fig. 1⇓, A and B). The addition of anti-CD66 polyclonal Abs or the control 12E7 mAb (incubated with either effector or target cells) had little or noeffect (Fig. 1⇓, A and B). Similar results were obtained when CD16⁻CD66a⁻ NK clones were used (see Fig. 5⇓). In agreement with our previous observations (15), little killing of 1106mel cells was observed when CD16⁻CD66a⁺ NK cells (for example, clone 163) were used (Fig. 1⇓, C and D), whereas effective killing (32.4%; data not shown) of the CD66-deficient NK-sensitive 721.221 cell line was observed. The low rate of 1106mel killing observed was the result of the inhibition mediated by the CD66a homotypic interactions, as lysis of 1106mel cells was restored when either the effector or the target cells were incubated with anti-CD66a polyclonal Abs (Fig. 1⇓, C and D). The anti-CD66 polyclonal Abs specifically stained all cells that were positive for CD66a expression (NK clones, melanomas, and various transfectants) and did not stain cells that were negative for CD66a expression (for example, CD66a-negative NK clones; data not shown). The controls, 12E7 mAb or polyclonal Abs from rabbit immunized with purified ubiquitin, had little or no effect (Fig. 1⇓ and data not shown). Similar results were obtained when CD16⁺CD66a⁺ NK clones were used (see Fig. 5⇓). Reversal of the CD66a-mediated inhibition was also observed even when the LB33melB1 cell line was used as a target. This cell line expressed the lowest levels of CD66a proteins among all seven class I-negative melanoma cells tested (Table II⇑ and data not shown). No inhibition of lysis by CD66a-positive NK cells was observed when these clones were incubated with the 1259mel melanoma line, a cell line that is efficiently killed by CD16-negative NK cells (Fig. 1⇓E). The reason why 1259mel cells are so sensitive to NK killing is not yet understood. It is especially surprising because the expression levels of the lysis ligands for CD16, NKp30, NKp44, and NKp46, detected by the Ig fusion proteins were similar to those of 1106mel cells. One possible explanation is that other lysis ligands for other lysis receptors (yet to be identified) exist on the surface of 1259mel cells, and the combined effect of all lysis receptors overcomes the CD66a-mediated inhibition. In conclusion, the results presented above demonstrate a novel class I-independent inhibitory mechanism of human NK cell cytotoxicity.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Killing of melanoma lines by NK clones. Lysis of 1106mel cells (A–D) and 1259mel cells (E) by CD16⁺CD66a⁻ NK clone (A and B) or CD16⁻CD66a⁺ NK clone (C–E) was performed as described in Materials and Methods. The anti-CD99 mAb (12E7) and anti-CD66a polyclonal Abs were incubated with the target cells (A and C) or with the effector cells (B, D, and E). The E:T cell ratio was 3:1.

721.221 cells expressing the CD66a protein are protected from lysis by CD66a-positive NK and YTS cells

To directly test the role of the CD66a protein in inhibition of NK cell cytotoxicity, 721.221 target cells and YTS effector cells (both deficient for CD66a expression; Fig. 2⇓) were transfected with the CD66a cDNA. Several clones of 721.221 cells expressing various levels of CD66a protein (.221/CD66a) were obtained. Two representative clones, expressing either low (.221/CD66a^low) or high (.221/CD66a^high) levels, are shown in Fig. 2⇓. YTS cells expressing either CD66a (YTS/CD66a) or CD66a in which the cytoplasmic tail of the molecule was truncated not to include the ITIMs (YTS/CD66aTrunc) were also generated (Fig. 2⇓). Importantly, the expression level of the CD66a protein on YTS cells was similar to the physiological level of expression on an average primary NK clone (median fluorescence intensity (MFI) was >2-fold over the background; a representative clone is shown in Fig. 2⇓). YTS transfectants expressing higher levels of CD66a protein could not be obtained. Transfectants were next tested in cytotoxicity assays. Moderate, but significant, inhibition of YTS/CD66a killing was observed when cells were incubated with .221/CD66a^high in all E:T cell ratios tested (Fig. 3⇓). The percentages of killing of other targets, including the parental 721.221 or the .221/CD66a^low were similar. Similar results were obtained with other YTS and 721.221 cells expressing similar levels of CD66a (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Expression of CD66a on various cell types. Transfectants were generated as described in Materials and Methods. Shown is CD66a staining of transfected .221 and YTS cells with the anti-CD66a mAb Kat4c (dark line) overlaid on the staining of the parental cells (.221 and YTS) with the same mAb (light line). Staining of a representative NK clone by Kat4c (dark line) overlaid on the staining of the same NK clones with the control FITC-conjugated goat anti-mouse Abs (light line) is also shown. The figure shows one representative experiment of three performed.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Killing of various 721.221 transfectants by various YTS transfectants. Killing assays were performed as described above. The various YTS transfectants are indicated in each histogram. The figure shows one representative experiment of six performed.

Lysis of all target cells tested against YTS cells transfected with the pBABE vector alone (YTS/MOCK) was similar. The CD66a inhibitory signal is probably transduced via the ITIM sequences, as no inhibition of lysis by YTS/CD66aTrunc was observed, even when these cells were incubated with 721.221/CD66a^high cells (Fig. 3⇑). The low level of CD66a expression on target cells (.221/CD66a^low) did not confer protection (Fig. 3⇑). The inhibition of lysis of .221/CD66a^high cells by YTS/CD66a was the result of the CD66a interactions, as lysis was restored when anti-CD66 Abs were included in the assays (incubated either with the effector cells (Fig. 4⇓A) or with the target cells (data not shown)). The controls, 12E7 mAb or rabbit polyclonal Abs directed against ubiquitin, had no effect (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Killing of .221/CD66a^high cells by NK clones. Killing of target cells by YTS/CD66a (A), CD66-positive NK clone (B), and CD66-negative NK clone (C) incubated with or without anti-CD66a polyclonal Abs. The figure shows one representative experiment of five performed.

Lysis experiments were also performed with NK clones positive or negative for the expression of CD66a. NK clones were prepared as described above and tested against .221/CD66a^high cells. CD66a-dependent inhibition of lysis of .221/CD66a^high cells was observed when CD66a-positive NK cells were used (a representative clone is shown in Fig. 4⇑B). In contrast, no inhibition of lysis of .221/CD66a^high cells was observed when CD66a-negative NK clones were used (representative clone is shown in Fig. 4⇑C).

Levels of CD66a expression on both effector and target cells are important for effective inhibition

One likely explanation for the moderate inhibition observed when .221/CD66a^high cells were incubated either with YTS/CD66a or with CD66a⁺ NK cells is the level of CD66a expression on both target and effector cells. Indeed, no inhibition of lysis was observed when .221/CD66a^low cells were used (Fig. 3⇑), moderate inhibition was observed when .221/CD66a^high cells were used (Fig. 3⇑), and strong inhibition of lysis was observed when 1106mel cells were used (Fig. 1⇑, C and D). The 1106mel cell line expresses the CD66a protein at a level 10-fold higher than that of the .221/CD66a^high transfectants (Table II⇑ and Fig. 2⇑). 721.221 cells expressing the CD66a protein at a higher level than the transfectant presented in Fig. 2⇑ could not be obtained (data not shown). Thus, the level of CD66a expression on target cells is important for effective inhibition of both YTS and NK cells.

To correlate the level of expression of CD66a on NK cells and the strength of inhibition, various NK clones (positive or negative for CD16) expressing different levels of CD66a were used. The redirected lysis of P815 cells was induced with either anti-CD16 mAb or anti-NKp44 and -NKp46 sera depending whether the NK clone tested expressed the CD16 protein. A direct correlation was observed between the level of CD66 expression on the surface of the NK clones and the percentage of inhibition of redirected lysis (Fig. 5⇓). The level of CD66a expression had to be at least 2-fold above the background staining for efficient inhibition to occur (Fig. 5⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

The high level of CD66a expression on NK clones correlates with efficient inhibition of redirected lysis of P815 cells. The CD66a expression on NK clones was monitored by FACS. To correctly compare the level of CD66a expression among different NK clones, and because the background staining F(ab)′₂ of FITC-conjugated goat anti-mouse IgG Abs of each NK clone might be different, the level of CD66a expression in each clone was determined by dividing the MFI of the CD66a staining on a given clone with the MFI of the background staining of the same clone. The fold increase in CD66a staining above the background of each clone is indicated in brackets. The percent inhibition of each clone was calculated by dividing the percentage of specific lysis of the NK clone incubated with anti-CD66 mAb by that of the clone incubated with no mAb. Similar results were obtained when the specific lysis of each NK clone incubated with anti-CD66 mAb was divided by the percent specific lysis of the same NK clone incubated with control mAb. The NK clones are presented in the figure in the order of the fold increase in CD66a above background. The figure shows CD16⁻CD66⁻ clones (24, 89, and 98), CD16⁻CD66⁺ clones (21, 79, 84, and 100), CD16⁺CD66⁻ clones (25, 47, 48, 63, and 64), and CD16⁺CD66⁺ clones (1, 2, 3, 9, 10, 13, 17, 30, 32, 34, 43, 44, 49, 58, 61, 65, 69, 70, 71, 73, 75, and 96). When CD16⁻ NK clones were used, anti-NKp44 and NKp46 sera were included to stimulate the redirected lysis experiments. When CD16⁺ NK clones were used, anti-CD16 mAb was included in the redirected lysis experiments. The figure shows NK clones generated from one healthy donor YF that contains an unusually high number of CD16⁺CD66⁺ NK clones.

Elevation of CD66a expression on NK cells derived from melanoma patients

The in vivo significance of the CD66a interactions was studied by the staining of NK cells derived from either metastasized lymph nodes or peripheral blood. The lymph node of patient M-169 was infiltrated with melanoma cells, highly positive for the CD66a expression (MFI, 247; data not shown). The lymph node was surgically removed, and lymphocytes in direct contact with the tumor cells were obtained after digestion and density gradient separation. Quadruple staining of the lymphocytes was performed for the expression of the CD16, CD3, CD56, and CD66 receptors. Remarkably, 12.85% of the NK cells (CD56⁺CD3⁻) obtained from M-169 lymph node expressed the CD66a protein (Fig. 6⇓A). A total of 10.5% of the NK cells obtained were CD16⁻CD66⁺, and 2.35% were CD16⁺CD66⁺ (data not shown). Importantly, the MFI of the CD66a-positive NK population was 8-fold above background, which is sufficient for effective inhibition (an MFI >2-fold above background is needed; see Fig. 5⇑). Similar results were obtained when peripheral blood NK cells derived from patient 3 were analyzed with the same quadruple staining; 14.8% of the NK cells were CD66a positive, and the MFI of this NK population was 7.5-fold above background (Fig. 6⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

CD66a expression on NK cells derived from healthy donors and melanoma patients. Lymphocytes were obtained from surgically removed lymph nodes derived from two different melanoma patients, infiltrated with melanoma metastases positive (A) or negative (C) for CD66a expression. Lymphocytes were also obtained from peripheral blood of another melanoma patient (B) or from peripheral blood of representative healthy donor (D). Lymphocytes were stained for expression of CD3, CD16, CD56, and CD66 as described in Materials and Methods. The figure shows CD66a expression on NK cells.

In contrast, little or no CD66a expression was observed among NK cells derived from the metastasized lymph node of patient M-172 (Fig. 6⇑C). Strikingly, the infiltrating M-172 melanoma cells did not express the CD66a protein (data not shown). Furthermore, no CD66a expression was observed among NK cells derived from the peripheral blood of 10 other melanoma patients with no clinical evidence of active disease (data not shown).

PBL from eight healthy donors were also obtained, and the expression of CD66a on NK cells was analyzed using the same quadruple staining as that described above. Very little or no CD66a staining was observed among all NK cells tested (a representative healthy donor OM is shown in Fig. 6⇑D). This is in agreement with a previous observation (22) demonstrating the expression of the CD66a molecule on activated NK clones only.

In conclusion, the results presented in this study suggest that the CD66a interactions are used by some melanoma cells as a mechanism of defense to avoid attack by CD66a-positive NK cells.