Mitogen-activated protein kinase kinase kinase kinase 4 as a putative effector of Rap2 to activate the c-Jun N-terminal kinase.

Little is known about the specific signaling roles of Rap2, a Ras family small GTP-binding protein. In a search for novel Rap2-interacting proteins by the yeast two-hybrid system, we isolated isoform 3 of the human mitogen-activated protein kinase kinase kinase kinase 4 (MAP4K4), a previously described but uncharacterized isoform. Other isoforms of MAP4K4 in humans and mice are known as hematopoietic progenitor kinase (HPK)/germinal center kinase (GCK)-like kinase and Nck-interacting kinase, respectively. MAP4K4 belongs to the STE20 group of protein kinases and regulates c-Jun N-terminal kinase (JNK). MAP4K4 interacted with Rap2 through its C-terminal citron homology domain but did not interact with Rap1 or Ras. Interaction with Rap2 required the intact effector region of Rap2. MAP4K4 interacted preferentially with GTP-bound Rap2 over GDP-bound Rap2 in vitro. In cultured cells, MAP4K4 colocalized with Rap2, while a mutant MAP4K4 lacking the citron homology domain failed to do so. Furthermore, Rap2 enhanced MAP4K4-induced activation of JNK. These results suggest that MAP4K4 is a putative effector of Rap2 mediating the activation of JNK by Rap2.

its specific effector, like the other Ras family members sharing the identical effector region.
The effector region of Rap2 differs from those of Ras and Rap1 by a single amino acid: amino acid 39 in Rap2 is Phe, while in Ras and Rap1 it is Ser. This difference could confer on Rap2 an ability to interact with its specific effectors and play signaling roles distinct from those of Rap1. Here, we report the identification of a novel Rap2-interacting protein, human MAP kinase kinase kinase kinase (MAP4K) 4 isoform 3, a previously described but uncharacterized isoform of MAP4K4 (2)(3)(4).
MAP4K4 exhibits structural homology to hematopoietic progenitor kinase (HPK) and germinal center kinase (GCK) (for a review, see Ref. 5). Thus, human MAP4K4 was designated HPK/GCK-like kinase (HGK) (2). Mouse MAP4K4 was discovered as a kinase that interacts with Nck, an adaptor protein composed of one Src homology (SH) 2 and three SH3 domains, and hence designated Nck-interacting kinase (NIK) (3) (Fig.  1A). MAP4K4 belongs to a large group of kinases related to Saccharomyces cerevisiae MAP4K, STE20. The STE20 group is divided into two families (for a review, see Ref. 6). One is the p21-activated kinase family, whose members are activated by the Rho family small GTP-binding proteins Cdc42 and Rac. The other is the GCK family, which consists of eight subfamilies. GCK and HPK, also known as MAP4K2 and MAP4K1, respectively, belong to the GCK-I subfamily; MAP4K4 belongs to the GCK-IV subfamily. Members of both GCK-I and -IV subfamilies are upstream of c-Jun N-terminal kinase (JNK), a "stressactivated" MAPK. In the present study, we show that Rap2 enhances activation of JNK by MAP4K4. We postulate that Rap2 plays signaling roles distinct from those of Rap1 by regulating JNK. * This work was supported by Grant-in-aid for Scientific Research on Priority Areas (13216082) from the Ministry of Education, Culture, Sports, Science and Technology of Japan and by Grant-in-aid for Scientific Research (C) (14570125) from Japan Society for the Promotion of Science. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) NM_145687 (MAP4K4 isoform 3).

EXPERIMENTAL PROCEDURES
Two-hybrid Screening-A cDNA fragment encoding amino acids 1-168 of human Rap2A was subcloned into plasmid pBTM116 to yield pBTM116-Rap2A. pBTM116-Rap2A and the human fetal brain cDNA library cloned in pACT2 (Clontech, Palo Alto, CA) were co-transformed into the S. cerevisiae L40 reporter strain. Approximately 4.7 ϫ 10 6 co-transformants were examined for HIS3 and ␤-galactosidase expression as described previously (7).

RESULTS
To identify a specific effector of Rap2, we carried out a yeast two-hybrid screening for Rap2-interacting proteins using a human fetal brain cDNA library. From this library, three independent positive clones, pACT2-H-1, -2, and -3, were isolated which encoded the C-terminal portions of MAP4K4 (Fig. 1A). MAP4K4 contains the N-terminal kinase domain bearing homology to STE20, the intervening region that includes a coiledcoil domain, and the C-terminal regulatory domain termed the CNH domain. The CNH domain was encoded by all the three clones, suggesting that it interacts with Rap2. Several Rasinteracting proteins contain regions of about 100 amino acids that mediate interactions with Ras; the Ras-binding domain and the Ras-associating domain (1). However, these domains were not found within the CNH domain of MAP4K4. When the CNH domain was divided into N-and C-terminal portions, each portion did not interact with Rap2 (Fig. 1A). This suggests that the entire CNH domain is required for interaction with Rap2. Alternatively, a region mediating the interaction may overlap the boundary between the two portions.
Next, we examined the interaction of MAP4K4 with Rap2 mutants and other small GTP-binding proteins by the twohybrid assay (Table I). MAP4K4 strongly interacted with Rap2 and its activated mutant, Rap2(G12V), while it interacted weakly with the dominant negative mutant, Rap2(S17N). MAP4K4 barely interacted with the effector region mutant, Rap2(F39S). MAP4K4 failed to interact with Rap1, its activated mutant, Rap1(G12V), or Ras. In comparison, Raf-1 failed to interact with only the dominant negative mutant, Rap2(S17N). Both MAP4K4 and Raf-1 failed to in- MAP4K4 as an Effector of Rap2 15712 teract with RhoA (data not shown). These results indicate that, unlike Raf-1, MAP4K4 interacts with only Rap2. They also indicate that the interaction requires the GTP-bound form of Rap2 and the intact effector region of Rap2.
We further examined this interaction in vitro (Fig. 1, B and C). For this purpose, HA-tagged full-length MAP4K4 (HA-MAP4K4) or its C-terminal deletion mutant (HA-MAP4K4⌬ CNH, amino acids 1-909) expressed in 293T cells was examined for interaction with immobilized GST fusion protein of Rap2 or Ras bound with GTP␥S (Fig. 1B) To assess whether Rap2 interacts with MAP4K4 in cells, we examined their subcellular localizations by indirect immunofluorescence microscopy (Fig. 2). When expressed alone, Myctagged Rap2 (Myc-Rap2A) showed a vesicular staining pattern that was particularly intense in the perinuclear region of the cell (panel a). Unlike Myc-Rap2A, HA-MAP4K4 showed a diffuse staining pattern in the cytoplasm when expressed alone (panel b). Strikingly, when co-expressed with Myc-Rap2A (panels c-e), HA-MAP4K4 co-localized almost completely with Myc-Rap2A. In contrast to full-length HA-MAP4K4, HA-MAP4K4⌬CNH failed to co-localize with Myc-Rap2A (panels f-h); Myc-Rap2A again showed a vesicular perinuclear staining pattern (panel f ), while HA-MAP4K4⌬CNH showed a diffuse cytoplasmic staining pattern (panel g). The staining pattern of HA-MAP4K4⌬CNH in cells co-expressing Myc-Rap2A was the same as that in cells expressing HA-MAP4K4⌬CNH alone (data not shown). The absence of co-localization was also observed when HA-MAP4K4 was co-expressed with Myc-Rap2A(S17N) (panels i-k) or Myc-Ha-Ras (panels l-n). HA-MAP4K4 showed a diffuse cytoplasmic staining pattern (panels j and m), while Myc-Rap2A(S17N) showed a fine vesicular cytoplasmic staining pattern (panel i), and Myc-Ha-Ras showed a perinuclear and plasma membrane-associated staining pattern (panel l). The staining pattern of Myc-Rap2A(S17N) or Myc-Ha-Ras in cells co-expressing HA-MAP4K4 was the same as that in cells expressing Myc-Rap2A(S17N) or Myc-Ha-Ras alone (data not shown).
MAP4K4 activates JNK (2, 3). We therefore tested whether Rap2 affects this process. FLAG-tagged JNK (FLAG-JNK2) was expressed alone, with HA-MAP4K4, with Myc-Rap2A, or with both of them in 293T cells (Fig. 3). Activation of JNK was examined by assessing the extent of phosphorylation of JNK within cell homogenates by Western immunoblotting with anti-  c and d). Co-localization of these proteins resulted in a yellow signal in the merged image (panel e). B, absence of co-localization. NIH3T3 cells were co-transfected with pCIneo-Myc-Rap2A and pCIneo-HA-MAP4K4⌬CNH, with pCIneo-Myc-Rap2A(S17N) and pCIneo-HA-MAP4K4, or with pCIneo-Myc-Ha-Ras and pCIneo-HA-MAP4K4 and examined as described in the legend to A. Bars, 10 m.

DISCUSSION
The functional difference between Rap2 and its close relative Rap1, sharing 60% amino acid identity, has been unclear. In other words, it has not been clear why Rap2 in addition to Rap1 exists. One possible answer is that Rap1 and Rap2 interact with different effectors and regulate different signaling pathways. Consistent with this, we have identified in the present study MAP4K4 as a putative effector of Rap2 that does not interact with Rap1. The effector region of Rap2 is different from that of Rap1 by a single amino acid, Phe-39. The importance of this amino acid in specific interaction with MAP4K4 was demonstrated by the substitution of Phe with Ser, present in the corresponding position in Rap1, which resulted in severe attenuation of interaction. The interaction may be conserved among multicellular organisms, as the C. elegans ortholog of MAP4K4 (MIG-15) interacts with the C. elegans ortholog of Rap2 (C25D7.7 protein) but not with orthologs of Rap1 (C27B7.8 protein) or Ras (LET-60) 2 .
Rap2 interacts with the CNH domain of MAP4K4. The CNH domain was originally found in citron that interacts with Rho and Rac (11). Subsequently, it was found in Rom2p, a GDP/ GTP exchange factor for S. cerevisiae Rho protein (Rho1p), and the myotonic dystrophy kinase-related Cdc42-binding kinase (MRCK␣) (12,13). Although all three proteins interact with Rho family small GTP-binding proteins, their CNH domains do not mediate the interactions directly. The CNH domain is also present in the C-terminal portions of GCK-I and -IV subfamilies of the STE20 group kinases (5,6). GCK/MAP4K2, which belongs to the GCK-I subfamily, interacts with Rab8 small GTP-binding protein through its C-terminal portion containing the CNH domain (14). GCK/MAP4K2 interacts with the GTPbound form of Rab8 preferentially over the GDP-bound form, and the interaction requires the intact effector region of Rab8. This suggests that GCK/MAP4K2 might act as an effector of Rab8 (14). Since our present study indicated that the CNH domain in MAP4K4 interacts with Rap2, it is interesting to hypothesize that some of the CNH domains in the STE20 group kinases may share a region that mediates interactions with small GTP-binding proteins.