Selective intracellular retention of virally induced NKG2D ligands by the human cytomegalovirus UL16 glycoprotein



Human cytomegalovirus (HCMV) has evolved a multitude of molecular mechanisms to evade the antiviral immune defense of the host. Recently, using soluble recombinant molecules, the HCMV UL16 glycoprotein was shown to interact with some ligands of the activating immunoreceptor NKG2D and, therefore, may also function as a viral immunomodulator. However, the role of UL16 during the course of HCMV infection remained unclear. Here, we demonstrate that HCMV infection of fibroblasts induces expression of all known NKG2D ligands (NKG2DL). However, solely MICA and ULBP3 reach the cellular surface to engage NKG2D, whereas MICB, ULBP1 and ULBP2 are selectively retained in the endoplasmic reticulum by UL16. UL16-mediated reduction of NKG2DL cell surface density diminished NK cytotoxicity. Thus, UL16 functions by capturing activating ligands for cytotoxic lymphocytes that are synthesized in response to HCMV infection.


Human cytomegalovirus


NKG2D ligand(s)


MHC class I-related chain A/B


UL16 binding protein




Human foreskin fibroblasts


Enhanced green fluorescent protein





1 Introduction

Human cytomegalovirus (HCMV) is a β herpesvirus that establishes lifelong infections. The infection is critically controlled by the cellular immune system and associated with severe morbidity in immunocompromised individuals 1. HCMV encodes a battery of immunomodulatory molecules that interfere with recognition of infected cells by NK and CD8 T cells. In a concerted action, the HCMV glycoproteins US2, US3, US6 and US11 down-regulate MHC class I surface expression to prevent recognition of HCMV-derived peptides by CD8 α β T cells (reviewed in 2, 3). As reduction of MHC class I molecules on the cell surface would render HCMV-infected cells susceptible to NK cell lysis, HCMV also expresses the MHC class I-like UL18 and an UL40 encoded peptide ligand for HLA-E. Both UL18 and HLA-E/UL40 suppress NK cell activity by engaging the inhibitory receptors LIR-1 and CD94/NKG2A, respectively 46.

Recently, evidence for an immunomodulatory function of the orphan HCMV UL16 glycoprotein was provided 7. UL16, a 50-kDa type I transmembrane glycoprotein, is dispensable forgrowth of HCMV in fibroblasts and lacks homology to other known proteins 8. In a search for cellular targets of UL16, Cosman et al. 7 identified the MHC class I-like molecules MICB, UL16 binding protein (ULBP) 1 and ULBP2, which function as ligands of the activating immunoreceptor NKG2D.

NKG2D is a member of the C-type lectin-like receptor family expressed by most human CD8 α β T cells, γ δ T cells and NK cells 9. NKG2D homodimers form an activating immunoreceptor complex with DAP10 adaptor molecules that transmit signals through recruitment of the p85 subunit of the phosphatidylinositol-3-kinase 10, 11]. Engagement of NKG2D leads to costimulation of CD8 α β T cells and triggers effector functions of NK cells [7, 9, 12, 13. NKG2D interacts with several MHC class I-like cell surface glycoproteins, hereafter collectively designated as NKG2D ligands (NKG2DL). In humans, there are two families of NKG2DL, the MIC and ULBP molecules. The MHC class I-related chains A/B (MICA and MICB) are highly polymorphic, MHC-encoded glycoproteins and their expression can be induced by cell stress 14, 15. In contrast to MHC class I molecules, MIC molecules do not associate with β2-microglobulin and antigenic peptides and their in vivo expression on healthy cells is highly restricted 14. However, MIC molecules were detected on many epithelial tumors and infected cells 12, 16, 17, which led to the hypothesis that they may act as danger signals 18, 19.

ULBP are atypical MHC class I-like molecules, since they lack an α3 domain and are attached to the cell surface via a glycosylphosphatidylinositol anchor. They are encoded by a multigene family, which apparently contains at least six functional proteins 20. However, NKG2D interaction has only been demonstrated for ULBP1, ULBP2 and ULBP3 21, 22. Like MIC molecules, they trigger effector functions of NK cells 7. Crystal structures of human NKG2D in complex with either MICA*;01 or ULBP3 show that NKG2DL interact in a similar fashion with NKG2D that is reminiscent of the T cell receptor binding to MHC molecules 23, 24. Thus far, ULBP protein expression has only been shown for cell lines and its regulation is largely unknown 7.

Soluble, recombinant UL16 binds ULBP1 and ULBP2, and also MICB, but not ULBP3 and MICA in vitro7. This was surprising since MICA and MICB are closely related by sequence, but only distantly related to ULBP. Binding of soluble UL16 to soluble ULBP1 impaired the ULBP1/ NKG2D interaction and thus, UL16 may block the NKG2D-mediated triggering of NK cell effector functions by MICB, ULBP1 and ULBP2 7.

Since these findings suggested a suppression of NKG2D-mediated immunosurveillance of HCMV-infected cells by UL16, it was unexpected when another group reported that HCMV infection caused induction of MIC expression on fibroblasts allowing recognition by HCMV-specific CD8 T cells in spite of reduced MHC class I levels 12. Accordingly, MIC molecules appear to function as potent enhancers of TCR-mediated recognition of HCMV-infected cells, thereby compensating for HCMV-mediated down-regulation of MHC class I. At this state, it was of interest to reconcile these contrasting findings on the role of NKG2DL for immunosurveillance of HCMV. Therefore, we scrutinized the induction of individual NKG2DL upon HCMV infection, the interference of UL16 with virally induced NKG2DL and its consequences for NKG2D-mediated recognition.

2 Results

2.1 Induction of NKG2DL expression upon HCMV infection

HCMV infection has been reported to induce MIC cell surface expression on fibroblasts 12. However, it remained unclear whether HCMV infection induces surface expression of both human MIC species, MICA and MICB, and how it affects ULBP expression. To investigate HCMV-induced expression of all known NKG2DL, we established NKG2DL-specific real-time PCR using sets of oligonucleotides specific for MICA, MICB, ULBP1, ULBP2, and ULBP3, respectively. Human MRC-5 fibroblasts were infected with HCMV strain AD169 and harvested at 3, 12, 24, 48 and 72 h post-infection (p.i.). Relative NKG2DL mRNA levels of infected fibroblasts were quantified by real-time PCR. Expression of all NKG2DL was strongly induced after HCMV infection, and increased mRNA levels were detected already 12 h p.i. (Fig. 1) However, kinetics and induction rates substantially varied between the individual ligands. Both, MICA and MICB were induced about 10-fold, but whereas MICA levels gradually reached a maximum at 48 h and slightly dropped at 72 h, kinetics of MICB were faster with a maximum at 24 h followed by a sharp decline. ULBP3 levels rose to about 6-fold at 12 h p.i. and remained stable thereafter. In contrast, ULBP1 mRNA gradually accumulated over 72 h to about 30-fold above background. ULBP2 mRNA levels were most dramatically induced to about 50-fold at 24 h p.i. and remained high until 72 h p.i. Altogether, all NKG2DL were clearly induced already 12 h p.i., but their maxima varied between 12 and 72 h p.i. To corroborate these results, we also analyzed mRNA levels of HCMV-infected primary human foreskin fibroblasts (HFF). Similarly to MRC-5, expression of all NKG2DL was strongly induced by HCMV infection and kinetics characteristically varied between the individual NKG2DL, as described above (data not shown).

Thus, HCMV infection concomitantly induces mRNA expression of all NKG2DL resulting in drastically increased NKG2DL mRNA levels throughout early and late stages of HCMV infection.

Figure 1.

Induction of NKG2DL mRNA expression upon HCMV infection. MRC-5 monolayers were infected with HCMV AD169 3, 12, 24, 48 and 72 h, respectively, before RNA extraction. Relative copy numbers were determined by real-time PCR. Increase in relative copy numbers was calculated by defining relative copy numbers of mock-infected cells as 1, except for UL16 where relative copy numbers at 3 h p.i. were defined as 1.

2.2 Restricted NKG2DL surface expression on HCMV-infected fibroblasts

Having shown that HCMV infection is paralleled by a pronounced increase in NKG2DL mRNA levels, we next investigated NKG2DL cell surface expression by HCMV-infected fibroblasts. In order to survey expression of NKG2DL, we generated a panel of mAb specific for MICA, MICB, ULBP1, ULBP2, and ULBP3, respectively, by immunizing BALB/c mice with P815 cells transfected with the respective NKG2DL cDNA. Specificity of the antibodies was demonstrated on P815, COS, and C1R cells transfected with the respective antigens. For MICA- and MICB-specific mAb, direct and mutually exclusive binding to recombinant MICA and MICB, respectively, was demonstrated by ELISA (data not shown).

HFF and MRC-5 cells were infected with HCMV strain AD169 and surface NKG2DL expression analyzed at 24, 48, and 72 h p.i. by flow cytometry. For both, HFF and MRC-5, low constitutive expression of MICA and ULBP2 was detected, but not for MICB, ULBP1 and ULBP3 (Fig. 2A and data not shown). No significant changes were observed 24 and 48 h p.i., respectively (data not shown), but 72 h p.i. MICA surface expression was clearly increased, whereas ULBP2 was no more detectable. No MICB and ULBP1 molecules were detected at any time p.i. on the cell surface, whereas a pronounced expression of ULBP3 was detected 72 h p.i.

Since NKG2DL mRNA induction was observed for all NKG2DL, we wondered whether MICB, ULBP1 and ULBP2 molecules were selectively retained in HCMV-infected cells. To test interference of HCMV with NKG2DL expression, 293T cells which constitutively express all five NKG2DL on the cell surface were infected with HCMV strain AD169. In contrast to HFF and MRC-5 fibroblasts, infection of 293T cells was only partial as judged by detection of viral antigens and cytopathic effects. This is also reflected by partial down-regulation of MHC class I molecules (Fig. 2B). Analysis of NKG2DL cell surface expression 3 and 5 days p.i. revealed that MICA and ULBP3 expression remained unaffected by the HCMV infection. However, in parallel to MHC class I down-regulation, MICB and ULBP2 vanished on a subpopulation of 293T cells. HCMV-associated down-regulation was most impressive for ULBP2 where two distinct populations became apparent 5 days p.i.

These results demonstrate that HCMV infection is associated with increased cell surface expression of MICA and ULBP3, and a loss of cell surface MICB and ULBP2.

Figure 2.

Selective NKG2DL surface expression upon HCMV infection. The human primary MRC-5 fibroblasts (A) and the 293T kidney fibroblast cell line (B) were infected with HCMV AD169 or mock-infected, and NKG2DL surface expression monitored with NKG2DL mAb in flow cytometry. The following mAb were used: AMO-1 (MICA), BMO-1 (MICB), AUMO-1 (ULBP1), BUMO-1 (ULBP2), CUMO-1 (ULBP3) and W6/32 (MHC I). The FACS histograms are shown in black, respectively, and were overlaid with isotype control stainings (in gray).

2.3 UL16 inhibits cell surface expression of MICB, ULBP1 and ULBP2

Selective down-regulation of MICB and ULBP2, but not MICA and ULBP3, perfectly matches the binding preferences described for a recombinant soluble variant of the HCMV glycoprotein UL16 7. The UL16 ORF encodes a heavily glycosylated 230-amino acid (aa) protein of 50 kDa 8. It is predicted as a type I transmembrane protein with an N-terminal endoplasmic reticulum (ER) signal sequence, a large luminal domain (163 aa) containing eight potential N-linked glycosylation sites followed by a short transmembrane domain (17 aa) and positively charged cytoplasmic domain (25 aa).

To investigate whether down-regulation of MICB and ULBP2 surface expression can be attributed to UL16, we generated an UL16 expression construct where the putative full length UL16 ORF was fused at the C terminus to enhanced green fluorescent protein (EGFP). 293T cells were stably transfected with plasmids encoding UL16-EGFP or EGFP, respectively.

For controlled assessment of changes in NKG2DL expression, we used neomycin-resistant 293T cells transfected with the UL16-EGFP construct only partially expressing UL16-EGFP. Cell surface expression of MICB, ULBP1 and ULBP2 was strongly reduced on UL16-EGFP-expressing transfectants versus UL16-EGFP-negative transfectants (Fig. 3A). In contrast, expression of MICA, ULBP3 and MHC class I molecules did not differ between UL16-EGFP-expressing and non-expressing cells. Surface expression of NKG2DL was not altered on 293T-EGFP transfectants and comparable to untransfected 293T cells (data not shown). Failure to detect MICB, ULBP1 and ULBP2 on the cell surface of 293T-UL16-EGFP transfectants may be either due to their intracellular retention and/or degradation, or to masking of the respective mAb epitopes by UL16. To address this issue, we analyzed intracellular NKG2DL expression of permeabilized 293T-UL16-EGFP cells. Staining intensities for MICA and MHC class I did not significantly differ between intact and permeabilized UL16-EGFP-positive and -negative cells (Fig. 3B and data not shown). In contrast, intracellular stainings of UL16-EGFP-positive cells detected significantly higher MICB and ULBP2 levels as compared to cell surface analysis and were almost (MICB) or fully equivalent (ULBP2) to levels of UL16-EGFP-negative cells. These results strongly suggested that MICB and ULBP2 are intracellularly retained by UL16 and do not reach the cell surface.

To corroborate these results, we analyzed soluble MICA (sMICA) and MICB (sMICB) in the supernatants of 293T cells. We observed recently, using a highly sensitive ELISA for sMICA, that MICA molecules are shed from the surface of tumor cells by metalloproteinases 25. To extend these studies, we also established a sandwich-ELISA for MICB with a comparable range of sensitivity. Here, we employed these ELISA to analyze interference of UL16 with the release of MIC molecules. In full agreement with our other observations, we found no differences for release of sMICA between 293T, 293T-EGFP and 293T-UL16-EGFP cells (Fig. 4). However, supernatants of 293T-UL16-EGFP cells contained 10–20-fold less MICB in comparison to 293T and 293T-EGFP cells, again suggesting that cell surface transport of MICB molecules is selectively inhibited by UL16.

Figure 3.

UL16 impairs surface expression of MICB, ULBP1 and ULBP2. Cell surface (A) and total NKG2DL expression (B) of intact and permeabilized UL16-EGFP-transfected 293T cells, respectively, were detected with NKG2DL-specific mAb. Following mAb were used: AMO-1 (MICA), BMO-1 (MICB), AUMO-1 (ULBP1), BUMO-1 (ULBP2), CUMO-2 (ULBP3) and W6/32 (MHC class I). Strong staining with antibodies specific for the ER-resident PDI demonstrated cell permeability, whereas lack of staining indicated cell integrity (B). ULBP1 staining was omitted due to the low staining intensity using AUMO-1 for 293T cells.

Figure 4.

UL16 selectively blocks release of sMICB. Release of sMICA and sMICB in culture supernatants by 293T, 293T-EGFP and 293T-UL16-EGFP was determined by an sMICA and an sMICB-specific sandwich-ELISA, respectively.

2.4 Subcellular localization of UL16

We used confocal microscopy to visualize the different fate of MICA and MICB in UL16-expressing cells and to localize UL16 intracellularly. Confirming flow cytometric results, MICA surface and intracellular staining was observed for both UL16-EGFP-positive and -negative 293T cells (Fig. 5A). In contrast, surface and bright intracellular MICB staining was detected only for UL16-EGFP-negative 293T cells, whereas only residual intracellular MICB was detected in UL16-EGFP-positive cells (Fig. 5B).

These data further substantiated selective intracellular retention of MICB, but not MICA by UL16. To address cellular localization of UL16, we analyzed co-localization of UL16-EGFP with the ER-resident protein-disulfide-isomerase (PDI), the lysosomal lamp-1 protein and surface MHC class I, respectively, using confocal microscopy. There was a prominent co-localization of UL16-EGFP with PDI expression (Fig. 5D), a partial co-localization with lamp-1 (data not shown), but none with cell surface MHC class I molecules (Fig. 5C, E and F). These results suggest that UL16 is mainly resident in the ER, which is in concordance with predictions based on the UL16 primary sequence and the intracellular retention of MICB and ULBP2. They also indicate that UL16 does not reach the cell surface to block NKG2DL interaction with NKG2D in a direct fashion as proposed elsewhere 26. At present, we do not know whether lysosomal localization is a feature of UL16 or simply an experimental artifact due to overexpression of the UL16-EGFP fusion protein. Unambiguous organelle-assignment of UL16 has to be addressed in future studies when UL16-specific reagents become available.

Figure 5.

Subcellular localization of UL16, MICA and MICB in 293T-UL16-EGFP cells. Confocal microscopy was implemented for co-localization studies of MICA (A), MICB (B), surface MHC class I (C), PDI (D) (all in red) with UL16-EGFP (green). Co-localization appears in yellow. For demonstration of mutually exclusive surface MHC class I and UL16-EGFP localization, single color photographs showing only surface MHC I (E) or UL16-EGFP stainings (F) were added. In (A), (B) and (D) approximately 50% of the cells were UL16-EGFP-negative 293T transfectants.

2.5 UL16-mediated NKG2DL down-regulation reduces NK cell reactivity

The sole function known for MICB, ULBP1 and ULBP2 consists in their capacity to ligate NKG2D and, thereby, to trigger NK cells and costimulate CD8 T cells 7, 9, 12, 13. To address the impact of UL16 on NKG2D-mediated immune responses, we analyzed binding of soluble human NKG2D tetramers to 293T-UL16-EGFP transfectants. As expected, binding levels of sNKG2D were significantly reduced on UL16-EGFP-expressing as compared to UL16-EGFP-negative transfectants (Fig. 6A). Low staining intensity of 293T-UL16-EGFP cells by sNKG2D, in spite of their considerable surface expression of MICA and ULBP3, is likely due to the lower avidity of sNKG2D as compared to mAb, and to the fact that ULBP3 and some allelic MICA variants exhibit lower affinity to NKG2D as compared to MICB, ULBP1 and ULBP2 21, 22. This may also explain that there was no significant staining of HCMV-infected fibroblasts with sNKG2D (data not shown).

Next, we investigated whether UL16-mediated reduction of surface NKG2DL levels modulates NKG2D-mediated recognition by NK cells. In first experiments, we analyzed recognition of 293T-EGFP and 293T-UL16-EGFP by polyclonal NK cells from healthy donors. However, both 293T transfectants were highly susceptible to lysis and addition of anti-NKG2D mAb did not greatly reduce lysis, suggesting that 293T cells express several other activating ligands besides NKG2DL recognized by the various activating receptors on polyclonal NK cells (data not shown). Therefore, we used the NK cell line NKL which has been well characterized for its NKG2D expression and NKG2D-mediated effector functions 9. In cytotoxicity assays, both 293T and 293T-EGFP cells were lysed by NKL whereas lysis of 293T-UL16 cells was strongly reduced (Fig. 6B and data not shown). As MICA and ULBP3 remain the only known NKG2DL being surface-expressed on 293T-UL16-EGFP cells, we addressed NK cytotoxicity in the presence of anti-MICA and anti-ULBP3 antibodies. Blocking interaction of NKG2D with MICA and ULBP3 completely abrogated lysis of 293T-UL16 cells and reduced lysis of 293T and 293T-EGFP cells by NKL (Fig. 6C and data not shown).

Figure 6.

Reduced NKG2DL expression impairs NK cytotoxicity. (A) Binding of soluble NKG2D tetramers is reduced on UL16-EGFP-expressing as compared to UL16-EGFP non-expressing 293T cells (right panel). No staining was observed with streptavidin-PE (SA-PE) only (left panel). Lysis of 293T-EGFP and 293T-UL16-EGFP pretreated with AMO-1 and CUMO-1 (C) or irrelevant isotype-matched Ab (B) by NKL cells. Susceptibility of untransfected 293T cells to lysis by NKL was similar to 293T-EGFP (data not shown).

3 Discussion

HCMV implements diverse molecular strategies to escape anti-viral NK and T cell responses. The concerted action of US2, US3/US11 and US6 proteins, for example, results in down-regulation of MHC class I molecules on HCMV-infected cells minimizing CD8 T cell reactivity 2, 3. In addition, products of the UL18 and UL40 ORF are thought to provide protection against NK cell killing by constituting ligands engaged by inhibitory receptors on NK cells 46. The HCMV UL16 glycoprotein now emerges as antagonist of another immune surveillance principle: the surface expression of cell-stress-induced ligands of activating NK cell receptors. Our data demonstrate that UL16 intracellularly retains ligands of the activating receptor NKG2D which are induced in the course of HCMV infection. Sutherland et al. 26 proposed two models of action for UL16: intracellular retention of NKG2DL or masking NKG2D binding sites of NKG2DL on the cell surface. Although our data clearly support the first model, we cannot rule out that some UL16 molecules reach the cell surface.

We also demonstrate that expression of all five known NKG2DL is strongly induced upon HCMV infection. Only three of them (MICB, ULBP1, ULBP2) are retained by UL16, whereas MICA and ULBP3 reach the cell surface (Fig. 7). These data reconcile the description of UL16 as an HCMV-encoded NKG2D antagonist with findings that CD8 T cell recognition of HCMV-infected fibroblasts is augmented by virally induced MIC expression compensating for MHC class I down-regulation 7, 12. The selective binding of UL16 to NKG2DL is intriguing from a structural point of view, since MICA and MICB are closely related by sequence (∼85% amino acid identity), but quite dissimilar from ULBP (∼25% identity), which are also more divergent amongst each other (∼55% identity). One might speculate that UL16, as does NKG2D, rather recognizes a common structural element of NKG2DL, which is altered in MICA and ULBP3, than an array of particular amino acid side chains. MICA and MICB appear to be products of a recent gene duplication 27 and may have consequently evolved differently under selective pressure, one of which may be UL16 binding. Interestingly, MICA and ULBP3, but not MICB, ULBP1 and ULBP2, contain an N-linked glycosylation site at position 8 that is located in the center of the β-pleated sheet platform 28 and may be involved in precluding UL16 binding.

Figure 7.

Model of UL16 function. Upon HCMV infection, expression of all five NKG2DL is induced, but only MICA and ULBP3 reach the cell surface. HCMV-encoded UL16 selectively binds to MICB, ULBP1, and ULBP2, retaining them in the ER.

Selective binding of UL16 to NKG2DL may also have been evolved as a consequence of different functions of the respective NKG2DL. For example, MIC and ULBP molecules may also serve as ligands for receptors other than NKG2D. In fact, a recent report provides direct evidence for binding of MICA tetramers to γ δ T cell receptors of some Vδ1 γ δ T cells 29. Of note, some allelic MICA variants and ULBP3 apparently have a lower affinity for NKG2D as compared to MICB, ULBP1 and ULBP2 21, 22, which may also explain poor NK cell recognition of 293T-UL16-EGFP cells as compared to 293T-EGFP cells (Fig. 6). Thus, UL16 is selective for higher affinity NKG2DL, which may be sufficient to dampen a NKG2D-mediated anti-viral response. In this context, it will be of interest to compare anti-HCMV immune responses in individuals homozygous for low versus high affinity MICA allelic variants.

An urgent question concerns the redundancy of NKG2DL. Besides differences in affinity for NKG2D and the interaction of MIC molecules with some γ δ T cell receptors, there are no features known distinguishing NKG2DL with respect to their function. In particular, little is known about differences in regulation of gene expression. We show that expression of all NKG2DL is induced upon HCMV infection, but kinetics and induction rates varied considerably between the individual ligands. For example, induction of MICB was faster and shorter-lived as compared to MICA. This is in parallel with previous findings on induction of MICA and MICB by heat-shock 14. Future studies have to establish whether NKG2DL are distinct in their expression kinetics. It also remains to be clarified what molecular events drive induction of NKG2DL expression in response to HCMV infection. It is known that infection of adenoviruses and herpesviruses is accompanied by a cell-stress response, and since promoters of MIC genes contain a heat-shock response element, expression of MIC genes during HCMV infection has been suggested to be regulated via heat-shock factors 12, 30.

Taken together, our data provide another example for the delicate balance between host immune defense mechanisms and viral evasion strategies. They also strengthen the notion of the NKG2D/NKG2DL system as a surveillance system for dysfunctional cells.

4 Materials and methods

4.1 Cell lines, transfectants

The human embryonic kidney-derived cell line 293T and the mouse mastocytoma cell line P815 were cultured in IMDM supplemented with 10% FCS, transfected 293T cells in 10% FCS-IMDM with 2.0 mg G418/ml. HFF and MRC-5 fibroblasts were cultured in alpha-MEM supplemented with 10% FCS. The NK cell line NKL, kindly provided by M. J. Robertson (Indiana University School of Medicine, Indianapolis,IN), was grown in RPMI 1640 with 15% FCS and 200 U/ml IL-2 (Proleukin, Chiron, Ratingen, Germany). About 14 days prior to cellular cytotoxicity assays, NKL cells were cultured without IL-2. All media were supplemented with penicillin (100 IU/ml)/streptomycin (100 μg/ml) (Life Technologies, Karlsruhe, Germany), 2 mM L-glutamine and 1 mM sodium pyruvate. Transfections were performed with the FuGENE 6 reagent according to manufacturer's instructions (Roche, Mannheim, Germany).

4.2 HCMV infections

HFF, MRC-5, or 293T were grown to confluence and infected with HCMV AD169 at an multiplicity of infection of 5 to achieve an infection efficiency of ∼100%. Infections were performed with cell-free supernatant from productively infected HFF cultures with 100% late-stage cytopathic effects. Mock infections were done by using the same supernatants after removal of infectious virus by ultracentrifugation at 80,000×g for 70 min. Briefly, media were replaced by infectious or non-infectious supernatant, incubated for 2 h at 37°C, and finally replaced with fresh media. Cells werethen incubated at 37°C in 5% CO2 for time intervals (p.i.) as indicated. The exact efficiency of infection in the respective experiment was determined at 24 h p.i. by immunodetection of viral immediate-early antigen using monoclonal antibody E13 (Biosoft, Paris, France) or at 72 h p.i. by the appearance of characteristic nuclear inclusions.

4.3 UL16-EGFP expression construct

Viral DNA was prepared from HCMV AD169-infected HFF monolayers 4 h p.i. by phenol/chloroform extraction. The complete HCMV-UL16 ORF excluding the stop codon was amplified using the oligonucleotides 5′-ATGGAGCGTCGCCGAGGTACGGTAC-3′ (sense) and 5′-GTCCTCGGTGCGTAACCGCTGGTAT-3′ (antisense) by PCR. PCR products were cloned into the EcoRI site of the EGFP-N3 vector (Clontech, Palo Alto, CA) in front of the EGFP ORF and verified by sequencing. The resulting UL16-EGFP ORF contains 17 additional codons between the last UL16 codon (GAC) and the first EGFP codon (ATG).

4.4 Production of soluble MICA*;04, MICB*;02 and NKG2D tetramers

Soluble MICA was produced and purified as previously described 21. In brief, the cDNA encoding the MICA*;04 ectodomain (Glu 1 through Lys 276) in pET20b (Novagen, Madison, WI) was introduced in E. coli, and MICA*;04 production induced by addition of IPTG. Inclusion bodies were solubilized and successively dialyzed against decreasing concentrations of urea. Refolded MICA*;04 was purified by gel filtration, dialyzed against PNEA (50 mM PIPES, pH 7.0; 0.15 M NaCl; 1 mM EDTA; 0.02% NaN3) and eventually examined by SDS-PAGE and immunoblotting. Accordingly, the MICB*;02 ectodomain was amplified with oligonucleotides 5′-ACATGCATATGGAGCCCCACAGTCTTCG-3′ (sense) and 5′-CTGACTCGAGCTTCCCAGAGGGCACAGGGTG-3′ (anti-sense), cloned in pET21a (Novagen) via NdeI and XhoI, and soluble MICB*;02 produced as described above. The cDNA fragment encoding the extracellular portion of human NKG2D (Asn 80 through Thr 147) was amplified with oligonucleotides 5′-ATCATATGGAAAGTTACTGTGGCCCATGTC-3′(sense) and 5′-ATAAGCTTACTCGTGCCACTCGATCTTTTGAGCCTCGAAGATGTCGTTCAGAGTCCTTTGCATGCAGATGTATG-3′ (anti-sense) adding a BirA recognition sequence at the C terminus and ligated into pET20b. Soluble NKG2D was produced in E. coli as described above. Biotinylation of NKG2D was done overnight at 27°C with the BirA enzyme (Avidity, Denver, CO) according to manufacturer's instruction. Excess biotin was removed by FPLC with a Superdex S75 HL 26/60 column and biotinylated NKG2D tetramerized by gradual addition of phycoerythrin-conjugated streptavidin (Molecular Probes, Leiden, Netherlands) over 4 h.

4.5 Generation of mAb

The murine P815 mastocytoma cell line was transfected using FuGENE 6 with full-length cDNA of MICA*;01, MICA*;04, MICB*;02, ULBP1, ULBP2 and ULBP3 in RSV.5 neo, respectively 21. Transfectants were selected with 1 mg/ml G418/ml, and expression of the respective NKG2DL mRNA was verified by RT-PCR. BALB/c mice were immunized either with a mixture of MICA*;01-, MICA*;04-, and MICB*;02- or ULBP1-, ULBP2-, and ULBP3-expressing P815 cells. Splenocytes of immunized mice were fused with P3×63Ag8.653 myeloma cells. Hybridoma supernatants were tested by indirect immunofluorescence using a FACSCalibur (Becton Dickinson, Heidelberg, Germany) for selective binding to P815-NKG2DL transfectants and to COS cells transiently transfected with the various NKG2DL cDNA. Hybridomaproducing NKG2DL-specific mAb were subcloned twice and immunoglobulins isotyped using an isotyping kit (Roche). BAMO-1 (IgG1) and BAMO-3 (IgG2a) recognize MICA and MICB, AMO-1 (IgG1) is MICA-specific, BMO-1 (IgG1) and BMO-2 (IgG2a) are MICB-specific, AUMO-1 (IgG1) is ULBP1-specific, BUMO-1 (IgG1) is ULBP2-specific, and CUMO-1 and CUMO-2 (IgM) are ULBP3-specific.

4.6 Real-time RT-PCR

Total fibroblast RNA was prepared using TRIzol (Life Technologies) followed by DNase I treatment and reverse transcription using SuperScript RTII (Invitrogen, Karlsruhe, Germany) according to the manufacturer's protocol. The resulting cDNA was amplified with NKG2DL, UL16 and 18S rRNA-specific primer pairs in duplicates (40 cycles, 95°C for 15 s, 60°C for 1 min) using SYBRGreen chemistry on the ABI PRISM 7000 Sequence Detection System (Applied Biosystems, Weiterstadt, Germany). Samples were independently analyzed two to three times. Primers were selected to flank an intron, where possible, and specificity was validated using cloned NKG2DL. Data analysis was done by using the ΔCT method for relative quantification. Similar amplification efficiencies for NKG2DLand 18S were demonstrated by analyzing serial cDNA dilutions with values of the slope of log cDNA amount vs. ΔCT<0.1. Oligonucleotide sequences (forward; reverse) were for 18S rRNA: 5′-CGGCTACCACATCCAAGGAA-3′; 5′-GCTGGAATTACCGCGGCT-3′; MICA: 5′-CCTTGGCCATGAACGTCAGG-3′; 5′-CCTCTGAGGCCTCRCTGCG-3′; MICB: 5′-ACCTTGGCTATGAACGTCACA-3′; 5′-CCCTCTGAGACCTCGCTGCA-3′; ULBP1: 5′-GTACTGGGAACAAATGCTGGAT-3′; 5′-AACTCTCCTCATCTGCCAGCT-3′; ULBP2: 5′-TTACTTCTCAATGGGAGACTGT-3′; 5′-TGTGCCTGAGGACATGGCGA-3′; ULBP3: 5′-CCTGATGCACAGGAAGAAGAG-3′; 5′-TATGGCTTTGGGTTGAGCTAAG-3′; UL16: 5′-TAATCGAGCGCCTCTACGTCC-3′; 5′-AAGGTCGCGGACAGTTCCTCG-3′. PCR products were analyzed on 3% agarose gels for purity.

4.7 Flow cytometry

Cells were incubated with NKG2DL-specific mAb or appropriate mouse immunoglobulin isotypes and then, after washing, with goat anti-mouse-PE conjugate (1:100) (Jackson ImmunoResearch Laboratories, West Grove, PA) as secondary reagent. PE-labeled NKG2D tetramers were used at 5 μg/ml. Intracellular staining was done using Cytofix/Cytoperm® following the manufacturer's protocol (BD PharMingen, Heidelberg). Samples were analyzed on a FACSCalibur.

4.8 Confocal microscopy

293T-UL16-EGFP cells were seeded on round coverslips in 12-well culture plates and grown to ∼50% confluency. Coverslips were washed in PBS and cells fixed with methanol/acetone (1:1, –20°C) for 20 min. After a washing step, cells were incubated for 20 min with the respective primary mAb. In case of Fig. 5C, E and F, cells were first stained with W6/32 (anti-MHCclass I), then washed and fixed. As organelle-specific markers for ER and lysosomes, an anti-PDI mAb (Stressgen Biotechnologies, Victoria, Canada) and anti-lamp-1 (kindly provided by M. Fukuda, La Jolla Research Center, La Jolla, CA), respectively, were used. For detection, fixed cells were incubated with AlexaTM 546-conjugated anti-mouse IgG (Molecular Probes). Microscopy was done with a ZEISS LSM 510 laser scanning microscope (Carl Zeiss, Göttingen, Germany). "Bleeding" of emission into other detection channels was excluded using the multitracking modus of the LSM 510. Thickness of the optical plane was adjusted by the pinhole to be <0.7 μm.

4.9 Cellular cytotoxicity assays

Cytotoxicity was analyzed in a 4-h 51Cr-assay. Target cells were labeled with 50 μCi of 51Cr (Amersham, Freiburg, Germany) for 2 h at 37°C and washed three times. In blocking experiments, mAb were added at 10 μg/ml during labeling procedure. Effector cells were titrated on target cells and incubated for 4 h at 37°C. Spontaneous release of target cells alone was less than 15% of the maximum release taken from target cells lysed in 1% Triton X-100. Percentage of lysis was calculated as follows: 100×[(experimental release – spontaneous release)/(maximum release – spontaneous release)]. Experiments were performed in duplicates.

4.10 ELISA

For the detection of sMICA, plates were coated with the capture anti-MICA mAb AMO-1 at 2 μg/ml in PBS, then blocked by addition of 15% BSA-PBS for 2 h at 37°C and washed. Next, the standard (recombinant MICA*;04 in 7.5% BSA-PBS) and samples were incubated for 2 h at 37°C, plates were washed. The detection mAb BAMO-3 was added at 5 μg/ml in 7.5% BSA-PBS and incubated for 2 h at 37°C. Plates were then washed and incubated with anti-mouse IgG2a-HRP (1:8,000 in 7.5% BSA-PBS) for 1 h at 37°C. Finally, plates were washed, developed using the TMB Peroxidase Substrate System (KPL, Gaithersburg, MD), and the absorbance measured at 450 nm. Results are shown as means of duplicates using recombinant sMICA*;04 as a standard. A similar ELISA was established to detect sMICB by using the anti-MICA/B mAb BAMO-1 as capture antibody and the MICB-specific mAb BMO-2 as detection antibody. For standardization, recombinant sMICB*;02 was used.

For analysis of release of sMICA and MICB, confluent 293T, 293T-UL16 and 293T-EGFP cells were cultured for 5 days, supernatants were removed, then they were tenfold concentrated using Centriprep (Amicon, Bedford, MA) and subjected to ELISA.


We would like to thank Oliver Schoor and Toni Weinschenk for eminent support with the real-time PCR. We also thank Iris Kehrer for excellent technical assistance. This work was supported in part by the fortüne programme of the Faculty of Medicine of the Eberhard-Karls-Universität Tübingen to A.S. (1010-0-0).


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5


  6. 6


  7. 7