Sonic Hedgehog (Shh)

First printed in R&D Systems' 2002 catalog.



Natural killer (NK) cells are an important element of the innate immune system as they are capable of killing tumor cells and virally-infected cells.1 NK cells express a large number of different cell surface receptors that deliver either activating or inhibitory signals.2-6 The relative balance of these signals regulates NK cell activity. Many of these receptors also occur on ab T cells. NKG2D is an activating receptor that has recently generated considerable interest. A number of NKG2D target ligands have been identified. The most intriguing of these are a pair of closely related proteins called MICA and MICB [major histocompatibility complex (MHC) class I chain-related].7,8 These are cell-surface molecules distantly related to MHC class I proteins, and the genes possess elements of heat shock promoters. MICA and MICB, therefore, are expressed during cell stress and are up-regulated in tumor cells and during viral infections. This receptor-ligand combination may play a critical role in the immune response to a variety of pathologies.

Human NKG2D was originally identified in 1991 as an orphan receptor that is expressed on NK cells and many T cells.9 The mouse,10,11 rat,12 and porcine13 homologues have also been identified. Interspecies amino acid (aa) sequence identities range from 52-78% for the entire protein (mouse and rat are the most closely related sequences) and from 72-90% within the lectin domain. Its function was first described in 1999 by two separate groups investigating MICA/MICB ligands14 or signal transduction through the DAP10 adapter protein.15 More recently, several additional ligands have also been reported.16-18

NKG2D is a type II transmembrane protein with an extracellular C-type (i.e. Ca2+-binding) lectin-like domain. Although the original prototype for this domain, mannose-binding protein, actually binds a carbohydrate ligand, many receptors having a C-type lectin-like domain, including NKG2D, lack the Ca2+ binding site and recognize protein ligands rather than carbohydrate.19

Fig. 1. NKG2D signaling requires association with the DAP10 adapter protein. Engagement of NKG2D on NK cells (e.g. via binding of the ligand MICA) can trigger cytolytic activity. It can also elicit cytokine production (e.g. MIP-1ß, TNF-a and IFN-?)

The NKG2D gene exists within the "NK complex" on human chromosome 12 and mouse chromosome 6.10,11 Genes within this complex encode structurally similar type II lectin-like receptors, and many of these genes are expressed primarily in NK cells. Other genes found in the human NK complex include CD69, CD94, NKR-P1A, and the NKG2 family. The mouse NK complex includes the NKG2 family, the Ly49 family, the NKR-P1 family, CD94, and CD69. Other members of the NKG2 family form heterodimeric receptors with CD9420 that recognize the non-classical MHC class I molecule, HLA-E.21-23 Although NKG2D was originally grouped with the NKG2 family, the NKG2D protein shares only 21% aa sequence identity with the other NKG2 family members, and it has a very different function.9


MICA/MICB: One of the first descriptions of NKG2D function was its role as a ligand for the cell surface proteins, MICA and MICB.14 These proteins are closely related and highly polymorphic.7,8,24 They are structurally similar to MHC class I proteins and possess a1, a2, and a3 domains, but they share only 27% aa identity with human MHC class I.18 Unlike MHC class I proteins, they do not associate with ß2-microglobulin or bind peptides.25 The genes are found within the MHC and occur in most mammalian species; however, they are absent from mice.26,27 There are 51 recognized human MICA alleles. The functional significance of this polymorphism remains unclear.8

Expression of the MIC proteins is confined to the gastrointestinal epithelium, endothelial cells, and fibroblasts.7,8,26 Little expression is observed in quiescent epithelial cells, but higher levels of expression occur in rapidly proliferating cells.28 Expression is also up-regulated in various transformed cells, particularly those of epithelial origin.8,28 The transcriptional control regions of the MICA and MICB genes contain sequences that are highly homologous to the heat shock elements of HSP70 genes.26 Expression of the MIC proteins can be up-regulated under conditions of “stress” such as heat shock and viral infection.29,30 Infection of fibroblasts with human cytomegalovirus (HCMV) resulted in a pronounced down-regulation of MHC class I with a simultaneous up-regulation of the MIC proteins. Expression was also observed in kidney allografts undergoing rejection episodes.31


ULBPs: Another group of ligands for human NKG2D was recently identified during a search for targets of an HCMV protein.18 HCMV encodes UL16, which is a predicted type I membrane glycoprotein that is not present in virions and is not essential for replication. Therefore, it was hypothesized that it might interact with a cellular protein. A UL16-Fc fusion protein bound to the human T cell line, HSB2, and to a Burkitt's lymphoma cell line. UL16-Fc was then used to identify cDNA clones that encode target ligands. The single cDNA clone identified from the HSB2 library encodes MICB. Probing of the Burkitt's lymphoma library yielded a previously unidentified protein that was designated ULBP1 (UL16 binding protein). Two similar sequences, designated ULBP2 and ULBP3, were identified from EST databases. The aa sequences of the three ULBPs are 55-60% identical to one another but share only 23-26% identity with the MIC proteins. Since MICB is a known ligand for NKG2D, each of the ULBPs was also tested and found to bind human NKG2D. ULBP1 and ULBP2 also bind to mouse NKG2D.32 UL16 bears no sequence or structural similarity to NKG2D.

The ULBPs are distantly related to MHC class I in their a1 and a2 domains, but they lack the a3 domain that is present in the extracellular portion of MHC class I molecules. The ULBPs do not associate with ß2-microglobulin or present peptide, and they are anchored to the membrane via a GPI (glycosylphosphatidylinositol)-linkage. The ULBP genes are not located in the MHC but map to human chromosome 6q25.

Expression of the ULBPs was examined with RT-PCR.18 The ligands were broadly expressed in a number of tissues and transformed cell lines, and high levels of expression were observed on normal human epithelial and endothelial cells of different origin.32 Furthermore, there was no general correlation between expression and tumor status.

Rae-1 and H-60: Ligands for mouse NKG2D were identified by their capacity to bind dimeric NKG2D-Fc fusion proteins16 or tetrameric NKG2D-streptavidin complexes.17 The NKG2D ligands identified by expression cloning were previously described cDNAs of unknown function. Two cDNAs that encode NKG2D ligands, Rae-1ß17 and Rae-1?16 , had been described previously as products of retinoic acid early inducible (Rae-1) transcripts.33,34 This family was known to contain at least three distinct loci encoding polypeptides that share 92-95% aa identity. Expression cloning also identified a fourth family member, Rae-1d16, which has 92% identity with Rae-1?. All Rae-1 family members were shown to bind NKG2D. Another related molecule identified by expression cloning as an NKG2D ligand is the minor histocompatibility antigen, H-60. Previous studies35 had reported that H-60 is expressed in BALB.B but not in B6 mice and that a peptide derived from H-60 was presented by the MHC class I molecule, H-2Kb, and served as an immunodominant minor histocompatibility antigen.

The Rae-1 proteins share only 25% aa identity with H-60, however, these protein have several similarities.17 The genes for Rae-1 and H-60 are not part of the MHC (found on mouse chromosome 17) but rather map to mouse chromosome 10.34,35 The syntenic human segments are located on human chromosome 6 in a region that also contains the ULBP genes.17 The Rae-1 proteins and H-60 are similar to the ULBPs in that they are distantly related to MHC class I and include only the a1 and a2 domains. The Rae-1 proteins are anchored to the membrane via a GPI-linkage, whereas, H-60 has a hydrophobic transmembrane segment.17 The Rae-1 proteins and H-60 do not have any substantial homology with MICA or MICB.

Experiments with soluble NKG2D proteins demonstrate that the mouse NKG2D ligands are expressed at appreciable levels on thymocytes but at low levels or not at all on most normal tissues.16,17 Expression is up-regulated, however, on many tumor cells. Rae-1 mRNA was detected throughout early embryos and in the brain/head region of day 10-14 embryos, but transcripts were not observed in day 18 embryos.16,34 Very low levels of Rae-1 transcripts were detected in adult spleen and liver, and transcripts were absent from adult brain and kidney.16 Rae-1 transcripts were present in several transformed cell lines.16 H-60 transcripts were also detected in a number of transformed cell lines although they were absent from any cell line derived from the B6 background. Low levels of H-60 transcripts were seen in adult spleen and embryonic tissue from BALB/c mice but were absent from adult brain, liver, and kidney.16


The expression of NKG2D has been assessed at the mRNA level,11,36 and at the protein level using antibodies against NKG2D14,17 and soluble versions of ULBP.32 NKG2D is expressed on most NK cells, aß T cells.14,17,36 It is generally absent from CD4+ αβ T cells. Expression of mouse NKG2D was also observed in lipopolysaccharide-activated peritoneal macrophages, but not in unstimulated macrophages.17 Increased expression is observed when NK cells are treated in vitro with IL-2 or IL-15.32 Other cytokines, including IL-12, IFN-a, IL-4, IL-6, IL-10, and IL-17 had no effect. In vivo treatment of SCID mice with IL-15 leads to up-regulation of NKG2D on splenocytes.32


Crystal structures have been determined for mouse NKG2D37 and human NKG2D in complex with MICA.19 The basic folding pattern is similar to other C-type lectin domains despite relatively low sequence homology. The complex with MICA reveals that an NKG2D dimer binds a MICA monomer. Similar surfaces on each NKG2D monomer interact with distinct surfaces on either the a1 or a2 domains of MICA. The surfaces on MICA that contact NKG2D are analogous to those on MHC class I that bind to the aß-T cell receptor. Although there is little obvious sequence similarity between the human and mouse NKG2D ligands, there appears to be considerable structural similarity as evidenced by the ability of mouse NKG2D to react with the human ligands, MICB, ULBP1, and ULBP2.32 Structural conservation is also indicated by the fact that, although the MICA and MICB proteins of various primate species are highly diverse, they are readily recognized by receptors on human intestinal epithelial ?d T cells.29

Although NKG2D lacks signaling elements in its cytoplasmic domain, it delivers an activating signal that requires association with the DAP10 adapter protein.15 DAP10 exists as a disulfide-bonded homodimer that associates with NKG2D via charged residues in the transmembrane domains. Cell surface expression of NKG2D requires co-expression of DAP10.15,18 The short cytoplasmic region of DAP10 includes a YXXM motif that serves as an SH2 domain-binding site for the p85 subunit of PI 3-kinase.15


Engagement of NKG2D on NK cells and γδ T cells can trigger cytolytic activity. This was demonstrated in transfectants that express the MIC proteins,14,18 the ULBPs,18 the Rae-1 proteins,16,17 and H-60.16,17 The parent cell lines were resistant to NK lysis while the transfectants were readily susceptible to NK lytic activity. The results also indicate that the NKG2D activating signal can overcome the inhibitory signals resulting from MHC class I recognition.14,18 Populations of ?d T cells that occur in tumors and in the intestinal epithelium also kill target cells that express the MIC proteins.28,29 This lytic activity is strongly inhibited by antibodies against NKG2D.14

The activating signal of NKG2D can also elicit cytokine production. NK cell production of MIP-1ß, TNF-a and IFN-? was enhanced upon treatment with soluble ULBP.32 A much more striking effect of ULBP was observed when IL-12 was included in the culture. Similarly, HCMV-specific γδ T cells produced IFN-?, TNF-a, IL-2, and IL-4 when exposed to target antigen presented by C1R cells. Cytokine secretion was strongly enhanced if the C1R cells were transfected with MICA.30

NKG2D appears to function in some cells as a co-stimulatory molecule. T cell co-stimulation refers to activation of signaling pathways that are complementary to those activated through the antigen-specific T cell receptor. The potent T cell co-stimulatory molecule, CD28, has a cytoplasmic YXXM motif that activates PI 3-kinase via the DAP10 adapter protein15 suggesting a similar co-stimulatory capacity. CD8+CD28- aß-T cell clones specific for HCMV antigens were tested for their capacity to kill HCMV-infected fibroblasts.30 Experiments with blocking antibodies reveal that NKG2D engagement contributed to cytolytic activity only during later stages of the infection when MHC class I expression was down-regulated and the MIC proteins were up-regulated. Lytic activity absolutely required the T cell receptor, and NKG2D provided a co-stimulatory signal. Similarly, these same T cell clones secreted increased levels of cytokines in response to their target antigens when simultaneous NKG2D engagement occurred. Co-stimulation was also observed in mouse macrophages.17 Transfectants expressing H-60 or Rae-1ß were used to stimulate activated macrophages. TNF-a transcription and nitric oxide production were enhanced in response to the signal, but only in the presence of lipopolysaccharide which also provides an activating signal to macrophages.

NKG2D-mediated tumor rejection was demonstrated in a murine system using the tumor-forming RMA cell line.38 RMA transfectants expressing Rae-1 proteins were rejected. In vivo depletion experiments demonstrated that tumor rejection was mediated by NK cells, and studies in CD1 deficient mice demonstrated that CD1-restricted NKT cells were not involved in the rejection. No immunity was generated against the non-transfected parent cell line in these experiments. A second report using similar methods confirmed that ectopic expression of the NKG2D ligands, Rae-1ß or H60, in several tumor cell lines resulted in potent rejection of the tumor cells by syngeneic mice.39 Cell depletion experiments demonstrate that both NK cells and CD8+ T cells are involved in this rejection. These authors also report a potent immune memory in contrast to the report in reference 38. The parent tumor cell lines that do not express NKG2D ligands were rejected in a secondary challenge, and this immunity failed to develop in Rag 1-/- mice or in mice that had been depleted for CD8+ T cells. These results suggest that NKG2D ligands might be useful in the design of tumor vaccines.


The immune system relies on many signaling molecules to detect danger, and NKG2D appears to be an important contributor to this recognition. Up-regulation of the MIC proteins in transformed cells and upon virus infection will elicit a response via NKG2D. It is clear that NK cells play a major role in the immune response to some transformed cells and some viral infections.1 The role of NK cells in the control of HCMV infection has been particularly well documented.1 The HCMV-encoded protein, UL-16, probably serves to down-modulate the innate immune response by interfering with the recognition of NKG2D target ligands. The extensive allelic polymorphism of the MIC proteins may be driven by interaction with a rapidly evolving microbial protein such as UL-16.18 Different alleles of MICA show large differences in binding affinity for NKG2D, and it is therefore also possible that the polymorphism of NKG2D ligands might define genetic determinants for HCMV susceptibility.1,40

The cytolytic and immunomodulatory activities of NK cells are regulated by inhibitory signals that detect MHC class I on "normal" cells and by a variety of activating signals. It was previously speculated that the activating signals might be turned on constitutively and that NK cell function was actually controlled by dominant inhibitory signals. The results with NKG2D demonstrate that NK cells also respond directly to signals of stress and that the distribution and expression characteristics of the numerous NKG2D ligands play an important role in the innate immune response.


  1. Cerwenka, A. & L. Lanier (2001) Nature Immunol. in press.
  2. Lanier, L. (2001) Nature Immunol. 2:23.
  3. Moretta, A. et al. (2001) Annu. Rev. Immunol. 19:197.
  4. Ravetch, J. & L. Lanier (2000) Science 290:84.
  5. Lopez-Botet, M. et al. (2000) Human Immunol. 61:7.
  6. Lanier, L. (2001) Curr. Opin. Immunol. 13:326.
  7. Bahram, S. et al. (1994) Proc. Natl. Acad. Sci. USA 91:6259.
  8. Stephens, H. (2001) Trends Immunol. 22:378.
  9. Houchins, J. et al. (1991) J. Exp. Med. 173:1017.
  10. Vance, R. et al. (1997) Eur. J. Immunol. 27:3236.
  11. Ho, E. et al. (1998) Proc. Natl. Acad. Sci. USA 95:6320.
  12. Berg, S. et al. (1998) Int. Immunol. 10:379.
  13. Yim, D. et al. (2001) Immunogenetics 53:243.
  14. Bauer, S. et al. (1999) Science 285:727.
  15. Wu, J. et al. (1999) Science 285:730.
  16. Cerwenka, A. et al. (2000) Immunity 12:721.
  17. Diefenbach, A. et al. (2000) Nature Immunol. 1:119.
  18. Cosman, D. et al. (2001) Immunity 14:123.
  19. Li, P. et al. (2001) Nature Immunol. 2:443.
  20. Lazetic, S. et al. (1996) J. Immunol. 157:4741.
  21. Braud, V. et al. (1998) Nature 391:795.
  22. Lee, N. et al. (1998) Proc. Natl. Acad. Sci. USA 95:5199.
  23. Borrego, F. et al. (1998) J. Exp. Med. 187:813.
  24. Petersdorf, E. et al. (1999) Immunogenetics 49:605.
  25. Pingwei, L. et al. (1999) Immunity 10:577.
  26. Groh, V. et al. (1996) Proc. Natl. Acad. Sci. USA 93:12445.
  27. Steinle, A. et al. (1998) Proc. Natl. Acad. Sci. USA 95:12510.
  28. Groh, V. et al. (1999) Proc. Natl. Acad. Sci. USA 96:6879.
  29. Groh, V. et al. (1998) Science 279:1737.
  30. Groh, V. et al. (2001) Nature Immunol. 2:255.
  31. Hankey, K. et al. (2000) Transplantation 69:S143.
  32. Kubin, M. et al. (2001) Eur. J. Immunol. 31:1428.
  33. Zou, Z. et al. (1996) J. Biochem. (Tokyo) 119:319.
  34. Nomura, M. et al. (1996) J. Biochem. (Tokyo) 120:987.
  35. Malarkannan, S. et al. (1998) J. Immunol. 161:3501.
  36. Yabe, T. et al. (1993) Immunogenetics 37:455.
  37. Wolan, D. et al. (2001) Nature Immunol. 2:248.
  38. Cerwenka, A. et al. (2001) Proc. Natl. Acad. Sci. USA 98:11521.
  39. Diefenbach, A. et al. (2001) Nature 413:165.
  40. Steinle, A. et al. (2001) Immunogenetics 53:279.