Letters to the Editor 356 CD56dimCD16 þ NK cells downregulate CD16 following target cell induced activation of matrix metalloproteinases

Leukemia (2007) 21, 356–359. doi:10.1038/sj.leu.2404499 nature of the CD56dimCD16À NK cell subpopulation reported previously. To investigate this we have employed various methods of NK Natural killer (NK) cells are immune effector cells that recognize þ both virally infected and malignant target cells. Unlike T and B cells purification: positive magnetic selection of CD56 cells followed by exclusion of CD3 þ cells (by means of electronic cells, NK cells, do not belong to the adaptive , þ and their activation and cytotoxicity is regulated via a series of gating), positive selection of CD56 cells from a CD3 depleted non-rearranged cell surface receptors. These receptors transmit peripheral blood fraction (MACS beads, Miltenyi Biotech, either activating or inhibitory signals, the summation of which is Auburn, CA, USA), as well as isolation of NK cells by depletion thought to dictate whether NK cell activation occurs or not. of other cell subsets (negative isolation – Rosette sep, Stem Cell Functionally NK cells are able to kill target cells without prior Technologies, Vancouver, BC, Canada). The results of these three approaches consistently showed that in positively selected sensitization and release potent pro-inflammatory cytokines þ dim À such as interferon (IFN) and tumor necrosis factor (TNF). CD56 cells (Figure 1a), vast majority of the CD56 CD16 þ 7 Discrete NK cell subsets can be identified in peripheral blood fraction were CD3 T cells (93.6 4.2%, n ¼ 3) (Figure 1b). In freshly isolated peripheral blood NK cells (by definition CD3À), on the basis of CD56 and CD16 staining (reviewed by Cooper dim À et al.1). CD56brightCD16À/ þ cells are more potent producers of the contribution of CD56 CD16 cells is marginal B 7 IFN-g in response to monokine stimulation and reciprocally ( 0.74 0.27%, n ¼ 5) (Figure 1c). interact with dendritic cells, leading to the activation of both cell NK cells were cultured either with or without K562 cells in the types, whereas CD56dimCD16 þ NK cells show higher cytotoxi- presence of fluorescein isothiocyanate -conjugated CD107a city against malignant targets. By definition, NK cells lack the specific mAb. After 4 h, cells were stained with CD3-PE, CD16- þ þ PerCpCy5.5 and C56-APC and analyzed by fluorescein-activated expression of CD3 and the T-cell . CD3 CD56 cells þ are present in the peripheral blood (PB) and as these cells belong cell sorter. Upon co-incubation with K562 targets a CD107a subset of NK cells was seen, indicative of cytotoxic to the T-cell lineage, they share phenotypic and functional dim properties with NK cells.2 When studying NK cells the (Figure 2b, middle). As well a fraction of CD56 NK cells with interference of these NK-like T cells should be excluded. reduced or even negative CD16 staining emerged following One of the main mechanisms of target cell killing is through incubation with K562 (Figure 2b, left). Interestingly, the pre- the granule release pathway. Following target cell conjugation ponderance of cells expressing CD107a showed reduction of and NK activation, cytotoxic granules move toward the cell CD16 staining intensity and the reduction was not complete in the majority of degranulating (CD107a þ ) cells (Figure 2b, right). This surface, fuse with the cell membrane and the granule contents dim þ (including perforin and granzymes) are expelled into the suggested that CD56 CD16 cells downregulate CD16 upon interaction with target cells (Figure 2a vs b, left). synaptic junction between the effector and target cells. Until dim þ recently, it has not been possible to identify individual effector In order to confirm that the CD56 CD16 population is the source of cells with reduced CD16 staining after target cells that have extruded their granule contents. However, recent þ þ studies show that specific membrane , such as CD107a encounter, purified CD56 CD16 cells were cultured either (lysosome-associated membrane 1, LAMP-1) are located with or without K562 cells in the presence of a CD107a-specific within the cytotoxic granule membrane and are transiently mAb as described above. In the absence of targets, these cells expressed on the cell surface following degranulation. Mono- did not express CD107a (Figure 3a, left) and retained their clonal antibody (mAb) against CD107a identifies cells that have phenotype (Figure 3a, middle). In contrast, after 4 h of co-culture recently released their granule contents.3 with K562, a proportion of cells showed CD107a staining Penack et al.4 have used CD107a to investigate NK cell (Figure 3b, left) and there was a continuous reduction in CD16 staining intensity, giving rise to NK cells with dim CD16 staining subsets responsible for target cell cytolysis. They report that þ À dim À and an emergence of CD56 CD16 cells (Figure 3b, middle). following co-culture with tumor targets, a CD56 CD16 NK dim þ cell subset underwent de-granulation as these cells expressed Thus, upon co-culture with K562 cells, some CD56 CD16 CD107a. They concluded that this NK cell subpopulation cells gained CD107a on the cell surface and downregulated CD16 expression (i.e., became CD56 þ CD16dim or even was exclusively responsible for target cell cytotoxicity given þ À that other NK cell subsets (CD56brightCD16À/ þ and CD56 CD16 ). Both events (granule exocytosis and loss of CD56dimCD16 þ ) did not express CD107a. Following this, CD16) are dependent upon the presence of target cells. Zimmer et al.5 commented that the method of NK cell isolation However the two phenotypic changes – gain of CD107a and 4 loss of CD16 – do not necessarily coexist at a single-cell level used by Penack et al. (CD56 positive selection) results in þ contamination by CD56 expressing T cells (CD3 þ CD56 þ ) and (Figure 3b, right) as not all cells loosing CD16 were CD107a that the CD56dimCD16À cell fraction was almost completely and conversely CD107a could be expressed by cells that did not absent after gating on the CD3À fraction. Stemming from the downregulate CD16, suggesting that these two events are not same argument (that the majority CD56dimCD16À cells in completely linked. peripheral blood are CD3 þ T cells), we have queried the true Previous studies demonstrated that following NK cell stimula- tion, CD16 shedding occurs through the activation of matrix matalloproteases (MMPs).6 As it was possible that target cell engagement leads to a similar sequence of events, we Correspondence: Dr MR Verneris, Department of Pediatrics, Uni- investigated the role of MMPs in target-cell induced CD16 versity of Minnesota, CCRB 660, 425 East River Rd, Minneapolis, MN þ þ 55455, USA. downregulation. CD56 CD16 NK cells were incubated with E-mail: [email protected] K562 targets in the presence of increasing concentrations of a Letters to the Editor 357 abcCD56+ positively CD56dimCD16- CD56+CD3- isolated fraction gated population fraction CD56 CD56 CD56 CD56 APC CD56 APC CD56 APC

CD16 PerCpCy5.5 CD3 PE CD16 PerCpCy5.5 CD16CD3 CD16

Figure 1 CD56dimCD16À cell population in PB is mainly made up of CD3 þ CD56 þ T cells. (a) CD56 þ selected cells (MACS) from PB, stained with CD56 and CD16. Gate designates the CD56dimCD16À population. (b) CD3 and CD56 staining of the gated population in panel (a). (c) CD56 positively selected population from a CD3À depleted fraction of PB, showing the paucity of CD56dimCD16À NK cells.

a 2.6% 0.6% 0.3%

media CD56 CD56 CD16 CD56 APC CD56 APC CD16 PerCpCy5.5

CD16 PerCpCy5.5 CD107a FITC CD107a FITC CD16 CD107a CD107a b 6.5% 20.2% 16%

+ K562 CD56 CD56 CD16 CD56 APC CD56 APC CD16 PerCpCy5.5

CD16 PerCpCy5.5 CD107a FITC CD107a FITC CD16 CD107a CD107a

Figure 2 K562 target cells lead to CD107a expression and downregulation of CD16 in purified NK cells. (a) NK cells incubated without targets (control). (b) NK cells incubated with K562 cells at 1:1 ratio for 4 h. During incubation CD107a mAb was present. After incubation cells were stained with the stated mAbs. Results are representative of three donors.

a 0.4% 0.2% 0.4%

media CD56 CD56 CD16 CD56 APC CD56 APC CD16PerCpCy5.5 0.2% <0.1%

CD107a FITC CD16PerCpCy5.5 CD107a FITC CD107a CD16 b 15.2% 3.2% 13.2%

+ K562 CD56 CD56 CD56 APC CD16 CD56 APC CD16PerCpCy5.5 1% 2%

CD107a FITC CD16PerCpCy5.5 CD107a FITC CD107a CD16 CD107a

Figure 3 CD56 þ CD16 þ NK cells acquire CD107a and downregulate CD16 upon co-culture with target cells. CD16 þ cells were isolated from purified PB NK cells and were cultured either without (a) or with K562 cells at a E:T ratio of 1:1 (b). Under both conditions mAbs against CD107a was present during the 4 h incubation, following this the cells were stained with mAbs against CD3 (not shown), CD56 and CD16. Results are representative of three donors. Letters to the Editor 358 a DMSO GM6001 0.3 uM 3uM 30uM

media CD56

+K562

CD16 b

media CD16

+K562

CD107a

Figure 4 MMP activation leads to target cell induced CD16 shedding in NK cells. Purified CD56 þ CD16 þ NK cells were co-cultured with increasing amounts of GM6001 or vehicle (dimethyl sulfoxide (DMSO)) in the absence (top panels a and b) or the presence of K562 cells (bottom panels a and b) for 4 h. CD107a was present from the beginning of the assay (b) and cells were stained with CD3 (not shown), CD16 (a and b) and CD56 (b) at the end of the incubation period. Results are representative of two donors.

broad-spectrum MMP inhibitor, GM6001 (Gallardin, Calbio- CD56dim population (not shown); thus, the property of target chem, San Diego, USA). GM6001 inhibited CD16 shedding cell-induced granule exocytosis is not unique to the CD56dim induced by target cells in a dose-dependent manner (Figure 4a, subset. Collectively, these results are in line with Nagler et al.8 bottom panels). The MMP inhibitor did not affect the phenotype and support the assertion that freshly isolated CD56dimCD16 þ of NK cells incubated in the absence of targets (Figure 4a upper NK cells are the main cytotoxic NK cell population in the PB. panels). This result shows that MMP activation, resulting from The finding reported by Penack et al., can be explained on the co-incubation with target cells, is responsible for the observed grounds of presented data in the following way: in the cells that CD16 downregulation. Inhibition of MMPs prevented CD16 were CD56 þ positively selected from PB, the CD56dimCD16À loss, however, it did not reduce degranulation (as measured by fraction contains mainly CD56 þ T cells. Cells with CD107a staining, Figure 4b), allowing us to dissociate the two CD56dimCD16À phenotype found to be uniquely CD107a þ phenotypic changes induced by targets. after co-incubation with tumor targets were likely PMA induced shedding of CD16 by NK cells is not dependent CD56dimCD16 þ NK cells that lost CD16 expression owing to on tyrosine protein kinases nor protein kinase C,7 which are its shedding induced by MMPs. Although the exact teleological major downstream pathways of NK cell activation. Co-incuba- explanation for the loss of CD16 from NK cells (following target tion of NK cells with K562 targets induces release of cytotoxic cell encounter) requires further study, one role of MMPs is to granule contents (as evidenced by CD107a staining, Figures 2– facilitate tissue invasion. Activation of this mechanism following 4) and granule release likely involves protein kinase activation. malignant target cell contact as reported herein may be of Our data shows that target-cell induced CD16 shedding occurs significance in the biology of NK cells and their applicability in via MMP activation yet, this MMP activation is neither unique to adaptive immunotherapy. de-granulating NK cells (Figure 4b, bottom left) nor is it necessary for degranulation (Figure 4b, bottom right). dim þ B Grzywacz, N Kataria and MR Verneris We also find that in addition to CD56 CD16 cells, Division of Blood and Marrow Transplantation, Department of CD56brightCD16À NK cells show CD107a labeling after co- Pediatrics, University of Minnesota, Minneapolis, MN, USA incubation with K562, albeit at lower percentage than the E-mail: [email protected] Letters to the Editor 359 References 5 Zimmer J, Andres E, Hentges F. NK cell subsets and CD107a mobilization assay. Leukemia 2005; 19: 1849–1851; author reply 1857. 1 Cooper MA, Fehniger TA, Caligiuri MA. The biology of human 6 Harrison D, Phillips JH, Lanier LL. Involvement of a metalloprotease natural killer-cell subsets. Trends Immunol 2001; 22: 633–640. in spontaneous and phorbol ester-induced release of natural killer 2 Lu PH, Negrin RS. A novel population of expanded human cell-associated Fc gamma RIII (CD16-II). J Immunol 1991; 147: CD3+CD56+ cells derived from T cells with potent in vivo 3459–3465. antitumor activity in mice with severe combined immunodefi- 7 Borrego F, Lopez-Beltran A, Pena J, Solana R. Downregulation ciency. J Immunol 1994; 153: 1687–1696. of Fc gamma receptor IIIA alpha (CD16-II) on natural killer 3 Rubio V, Stuge TB, Singh N, Betts MR, Weber JS, Roederer M et al. cells induced by anti-CD16 mAb is independent of protein Ex vivo identification, isolation and analysis of tumor-cytolytic T tyrosine kinases and protein kinase C. Cell Immunol 1994; 158: cells. Nat Med 2003; 9: 1377–1382. 208–217. 4 Penack O, Gentilini C, Fischer L, Asemissen AM, Scheibenbogen C, 8 Nagler A, Lanier LL, Cwirla S, Phillips JH. Comparative studies of Thiel E et al. CD56dimCD16neg cells are responsible for natural human FcRIII-positive and negative natural killer cells. J Immunol cytotoxicity against tumor targets. Leukemia 2005; 19: 835–840. 1989; 143: 3183–3191.

Reply to Dr Grzywacz

Leukemia (2007) 21, 359. doi:10.1038/sj.leu.2404498 performed the CD107a assay with peripheral blood NK cells, RAJI cells (burkitt cell line) and 10mg/ml of the anti- 4 We would like to thank Dr Grzywacz and co-workers for their CD20 mAb rituximab, as described previously. As a result of well taken remarks1 concerning our article in Leukemia.2 We these experiments we could confirm Dr Grzywacz’s findings. have presented a novel flow cytometric assay that detects Figure 1 visualizes the CD16 loss of NK cells upon degranula- natural killer (NK) cell degranulation as a reliable indicator of tion with target cells. Taken together, we are happy that our publication in NK cell activation. Using this method, we found that NK cells, 2 after degranulation, are CD56dim/CD16neg upon interaction with Leukemia led to scientific discussions and consequently to Dr tumor targets. In a previous work, Nagler et al.3 showed that Grzywacz’s interesting findings. The CD107a assay is a highly CD56dim/CD16pos cells are potent mediators of natural cyto- sensitive method to assess NK cell degranulation, and its results toxicity in the peripheral blood. Dr Grzywacz is now able to are strictly correlated with NK cell cytotoxicity. Its simple and provide an explanation for both phenomena observed. In a rapid performance makes it a suitable method to assess NK cell series of experiments he shows that CD16pos NK cells lose CD16 activity in immunocompromised patients. Currently, we are expression upon interaction with NK sensitive target cells. using the assay successfully to detect NK cell degranulation as Inspired by Dr Grzywacz’s findings, we performed experi- marker of NK cell activity in patients undergoing hemopoietic ments in our lab in order to visualize the CD16 loss of NK cells stem cell transplantation. upon degranulation. To obtain optimum NK cell activation we O Penack and L Uharek Department of Hematology, Oncology, and Transfusion Medicine, Charite´-Campus Benjamin Franklin, Berlin, - Rituximab + Rituximab Germany E-mail: [email protected]

R3 R3 References

5%R2 R4 9.5% R2 R4 1 Grzywacz B, Kataria N, Verneris M. CD56dimCD16+ NK cells downregulate CD16 following taret cell induced activation of

CD 56 PE CD 56 PE matrix metalloproteinases. Leukemia 2006 (in press). 2 Penack O, Gentilini C, Fischer L, Asemissen AM, Scheibenbogen C, 15% 10.5% Thiel E et al. CD56dimCD16neg cells are responsible for natural cytotoxicity against tumor targets. Leukemia 2005; 19: 835–840. CD 16 FITC CD 16 FITC 3 Nagler A, Lanier LL, Cwirla S, Phillips JH. Comparative studies of human FcRIII-positive and negative natural killer cells. J Immunol Figure 1 NK cells lose CD16 expression upon degranulation. 1989; 143: 3183–3191. Normal donor’s peripheral blood NK cells were coincubated with NK 4 Fischer L, Penack O, Gentilini C, Nogai A, Muessig A, Thiel E et al. resistant RAJI lymphoma cells (left). Owing to application of 10 mg/ml The anti-lymphoma effect of antibody-mediated immunotherapy is Rituximab, the RAJI cells become NK sensitive. After degranulation, a based on an increased degranulation of peripheral blood natural considerable amount of NK cells have lost CD16 expression (right). killer (NK) cells. Exp Hematol 2006; 34: 753–759.