Contribution of the Box 1 and Box 2 Motifs of Cytokine Receptors to Jak1 Association and Activation*

Kinases of the Jak family (Jak1/2/3 and Tyk2) interact with the membrane proximal domain of different cytokine receptors and play a critical role in the activation of cytokine and growth factor signaling pathways. In this report we demonstrate that both the Box 1 and Box 2 motif collaborate in the association and activation of Jak1 by type I interferons. Mutational analysis of the β chain of type I interferon receptor (IFNαRβL/IFNAR2) revealed that Box 1 plays a more significant role in activation than in the association with Jak1. On the contrary, the Box 2 motif contributes more to the association with Jak1 than to kinase activation. Additionally, the study of the Jak1 binding sites on the IL2 receptor β (IL2Rβ), IFNγRα/IFNGR1, and IL10Rα/IL10R1 chains suggests that cytokine receptors have two different kinds of interaction with Jak1. One form of interaction involves the Box 1 and the previously described Box 2 motif, which we now designate as Box 2A, characterized by the VEVI and LEVL sequences present in IFNαRβL/IFNAR2 and IL2Rβ subunits, respectively. The second form of interaction requires a motif termed Box 2B, which is present in the IFNγRα/IFNGR1 (SILLPKS) and IL10Rα/IL10R1 (SVLLFKK) chains. Interestingly, Box 2B localizes close to the membrane region (8–10 amino acids from the membrane) similar to Box 1, whereas Box 2A is more distal (38–58 amino acids from the membrane).

can associate with and be activated by different receptors. However, the absence of a distinct Jak cannot be compensated by another Jak as has been demonstrated by the cytokinespecific phenotypes observed in animals with targeted disruptions of these kinases (4 -8).
The interaction between Jak2 and single subunit cytokine receptors (i.e. erythropoietin receptor (EPOR), growth hormone receptor, and prolactin receptor (PRL-R)) as well as heterodimeric receptors (IFN␥R and granulocyte-macrophage colony stimulating factor receptor)/interleukin-3 and -5 receptors) has been extensively characterized (9,10). The membrane proximal region of cytokine receptors that interact with Jak2 contains a highly conserved motif termed Box 1 (see Fig. 1E), which is formed by a proline-rich sequence located 6 -10 amino acids after the transmembrane domain (11)(12)(13). The consensus sequence for the Box 1 motif present in single-subunit cytokine receptors can be described as follows: PX(I/V)PXP(E/K) ( ϭ hydrophobic residues). It is also important to point out that a region within the cytoplasmic domain C-terminal from Box 1 contains a second motif, Box 2, which is not so well defined. The amino acid sequence more frequently found in the Box 2 motif is (V/L)E(V/L)L represented by E (see also Fig. 1E). The Box 2 motif is also present in single-chain cytokine receptors and is required in some cases for full activation of Jak2 (14).
Interestingly, the binding sites for Jak1, Jak3, and Tyk2 are not so well defined. The Tyk2 binding site on the ␣ chain of IFN␣R (or IFNAR1) has only distant homology with other cytokine receptors (15,16). Jak1 is activated by IFNs (IFN␣/␤ and IFN␥) and several cytokines, including the IL6 (IL6, leukemia inhibitory factor, ciliary neurotrophic factor, oncostatin M) and IL2 families (IL2, IL4, IL7, IL9, and IL15) (17)(18)(19)(20)(21)(22)(23)(24)(25). The Jak1 binding site has been studied for a few cytokine receptors and appears to be rather heterogeneous. For example, the sequence 266 LPKS 269 of the ␣ chain of IFN␥R (or IFNGR1), which has at best distant similarity to Box 1, revealed that only Pro 267 was important for Jak1 binding (19). In the case of the IL2R␤ chain, a sequence with some similarity to Box 1 and Box 2 appear to be important for Jak1 binding (26). Mutations of the Box 2 motif in gp130 and IL2R␤ chains affect not only binding but also activity (26 -28). In comparison, initial studies of the Box 1 of the IFN␣R␤L (or IFNAR2) chain suggested that this motif by itself has a minimum contribution to Jak1 binding (21). Interestingly, the Box 1 and Box 2 motifs of IFN␣R␤L are separated by only 7 amino acids, whereas the distance between these two motifs in the IL2R␤ and gp130 is 38 and 31 residues, respectively. Although some of the cytokine receptor subunits that activate Jak1 have Box 1 and/or Box 2 motifs (i.e. gp130, IL2R␤, and IFN␣R␤L), others such as the IL10R␣, and possibly the IFN␥R␣, do not have easily definable Box 1 or Box 2 motifs. This variability in the Jak1 binding sites of cytokine receptors also appears to be accompanied by differences on the corresponding receptor binding surfaces of Jak1 (29). Although all cytokine receptors require the JH7 and part of the JH6 domains of Jak1 for activation, distinct cytokine receptors interact with a second region of Jak1 that is receptor-specific. For example, the second area of interaction for the IL10R␣ is JH3, whereas IL2R␤ and IL4R␣ interact with JH4 and JH5-6, respectively (29).
This variability in the Jak1 binding site among cytokine receptors suggests that Jak1 could interact with different motifs in distinct cytokine receptors. This type of model would be important to develop therapeutic agents that may block Jak1 binding and/or function within the context of some cytokine systems without disturbing Jak1 activation by others. In this report, we sought to address whether there are differences in Jak1 binding sites among cytokine receptors. We characterized the kinase binding domain for two different types of receptors that activate Jak1: (a) receptors such as the IFN␣R␤L and IL2R␤ that contain Box 1 and Box 2 motifs and (b) receptors that do not have either Box 1 or Box 2 motifs (i.e. IFN␥R␣ and IL10␣). Although in the case of the IFN␣R␤L, Box 2 appears to contribute to the majority of Jak1 binding, its mutation did not affect Jak1 activation by IFN␣. Surprisingly, some mutations of Box 1 such as Pro 289 and Pro 291 affected Jak1 binding and activation, whereas other alterations such as the addition of a Pro in position 292 (N292P) or the inversion of Box 1 of IFN␣R␤L did not affect Jak1 binding but impaired Jak1 activation. The Box 2 motif of the IL2R␤ was also critical for binding to Jak1, whereas Box 1 cannot interact with Jak1 in the absence of Box 2.
Interestingly, IL10R␣ and IFN␥R␣, which do not contain either Box 1 or Box 2 motifs, associate with Jak1 through a different motif that is localized closer to the transmembrane region. Thus, these data suggest that Jak1 interacts with cytokine receptors in two different manners. One form of interaction involves both Box 1 and Box 2A motifs (i.e. IL2R␤ and IFN␣R␤L chains). The Box 2A motif, previously described as Box 2, is represented by a sequence formed by hydrophobic, charged, hydrophobic, and hydrophobic amino acids (i.e. VEVI or LEVL). The second form of interaction, found in the IL10R␣ and IFN␥R␣/IFNGR1 chains, involves Box 2B, a motif that is different from Box1 and Box 2A.

MATERIALS AND METHODS
IFNs and Antibodies-Human recombinant IFN␣2 and IFN␥ were kindly provided by Drs. M. Brunda (Hoffman-La Roche) and Ronald Borden (Schering-Plough). The anti-phosphotyrosine antibody (4G10) and the anti-Jak2 serum were obtained from Upstate Biotechnology Inc. (Lake Placid, NY). The monoclonal antibodies against Jak1 and STAT1 were purchased from Transduction Laboratories (Lexington, KY).
GST Fusion Proteins-The following GST fusion proteins encoding full-length or truncated forms of the cytoplasmic domains of cytokine receptor subunits were used for this study (see Fig. 1). 1) GST-IFN␣R␤L265-515, GST-IFN␣R␤L265-375, and GST-IFN␣R␤L300 -375, encoding full-length and the previously described Jak1 binding site with and without Box 1 of IFN␣R␤L (Fig. 1A), respectively (21). 2) GST-IL2R␤ encodes the full-length cytoplasmic domain of the IL2R␤ chain, whereas GST-IL2R␤322 and GST-IL2R␤265 encode truncations at amino acids 322 and 265 that containing Box 1 and Box 2, and Box 1 alone, respectively (Fig. 1B). 3) GST-IFN␥R␣ and GST-IFN␥R␣s (IFN␥R␣, short) encode the entire cytoplasmic domain of the IFN␥R␣/ IFNGR1 chain and an N-terminal truncation that starts at amino acid 270 after the LPKS sequence important for Jak1 binding (Fig. 1C). 4) GST-IL10R␣ and GST-IL10R␣299 correspond to the entire cytoplasmic domain of the IL10R␣ chain and a form truncated at amino acid 299, respectively (Fig. 1D). Mutations introduced into the Box 1 motif of IFN␣R␤L chain (Fig. 1A) include Pro to Ala in positions 289, 291, and 294 ( 287 AWPFPNLPP 295 , mutated residues are underlined). We also produced mutations that make Box 1 of IFN␣R␤L ( 287 AWPFPNLPP 295 ) closer to that observed in receptors that bind Jak2 such as the introduction of a Pro at position 292 (N292P, ( 287 AWPFPPLPP 295 ), inversion of Box 1 motif from PXPXXP to PXXPXP (RV mutation, from 287 AW-PFPNLPP 295 to 287 AWPLNPFPP 295 ) and partial substitution of Box 1 of IFN␣R␤L for the EPOR ( 287 AWPFPNLPP 295 to 287 AWPGIPSPP 295 ). Point mutations were generated by PCR using the overlap extension method or the QuikChange kit (Stratagene) and were confirmed by sequencing. All GST fusion proteins were produced as described previously (21), and the amount used in each experiment was determined by Coomassie Blue staining and/or immunoblotting with an anti-GST monoclonal antibody (Transduction Laboratories).
Expression of Mutant Forms of the ␤L Subunit of the Type I IFN-R in Mouse L-929 Cells-A mutation of the Box 1 and Box 2 motifs were generated as described above and subcloned into the retroviral vector pLXSN. LpR␣ cells corresponding to mouse L-929 cells, which had been stably transfected with the ␣ subunit of the IFN␣R (21), were infected with the pLXSN␤L viruses packaged in PA317 or BOSC23 cells. Clones were selected and grown in medium containing G-418 (500 g/ml) and hygromycin B (500 g/ml). Receptor expression was assessed by fluorescence-activated cell sorting analysis after staining with the specific anti-␤L antibody IFNaR␤1 (30).

Identification of a Jak1 Binding Site on the IFN␣R␤L
Chain-We previously determined that the region delineated by amino acids 300 -346 contains the Jak1 binding site (21). In addition, the membrane proximal region of the IFN␣R␤L intracellular domain truncated after Asp 315 maintained functional interaction with Jak1 (31). To identify critical residues for the formation of the Jak1 binding site, we performed a modified alanine scanning by introducing two or three mutations into the GST␤L300 -375 wild type construct. It is worth mentioning that this construct lacks the Box 1 motif of IFN␣R␤L (Fig. 1A, residues 289 -295), which was previously found to have a small contribution to the interaction with Jak1 (21). GST␤L300 -375 constructs carrying these mutations were used to pull-down Jak1 from U-266 cell lysates. Fig. 2A (upper panel) shows that only a mutation involving residues 303-305 (lane 2, EVI) has a significant effect on the ability of these fusion proteins to interact with Jak1. The lower panel (Coomassie) shows that equal amounts of the different GST fusion proteins were used. It is important to point out that the VEVI sequence resembles the Box 2 motif that regulates Jak1 activity of other cytokine receptors (see below) (26 -28).
To define precisely the individual amino acids involved in the interaction with Jak1, we made GST␤L300 -375 constructs carrying individual mutations to Ala of residues 303-305, as well as the flanking residue 302. Fig. 2B shows that mutation of Val 302 (lanes 4 and 5) almost eliminated Jak1 binding, whereas Ala mutations of Val 304 (data not shown) and Ile 305 (lanes 3) had a less significant effect on the interaction between IFN␣R␤L and this kinase. Mutation of Glu 303 (lanes 6 and 7) had only a modest effect on the interaction with Jak1. The lower panel (Coomassie) shows that equal amounts of the different GST fusion proteins were used. Thus, the VEVI motif appears to be critical for the interaction between Jak1 and IFN␣R␤L. This finding was unexpected, because this motif partially overlaps with the RACK1 binding site (residues 302-304 and 314 -327) and expression of full-length IFN␣R␤L containing mutations of Box 2 affects activation of STAT1 and STAT2 but not Jak 1 (32). Thus, the finding that Jak1 binding is affected by mutation of Box 2 in the context of residues 300 -375, but not in the context of the full-length receptor, suggests that one or more regions within IFN␣R␤L other than amino acids 300 -375 should also contribute to Jak1 binding and activation.
Mutations of the Box 1 Motif of IFN␣R␤L Affect Jak1 Activation-The logical candidate to contribute or collaborate with Box 2 in Jak1 binding and activation is the Box 1 motif. The minimal Jak1 binding previously detected with the Box 1 motif (21), in the absence of Box 2, could be sufficient to support Jak1 binding and activation in the context of a full-length receptor containing Box 2 mutations. Therefore, we next assessed the contribution of the Box 1 motif to Jak1 binding by introducing mutations within a GST␤L265-375 that also contains the Box 2 motif. Fig. 3 shows that mutation of Pro 289 (lane 3) and Pro 291 (lane 4) to Ala in Box 1 eliminated or decreased the association of IFN␣R␤L with Jak1, respectively (compare lane 2 with lanes 3 and 4), whereas the mutation Pro 294 to Ala had no effect (compare lanes 2 and 6). Mutations Asn 292 to Pro, EPOR, or RV, which we used in an attempt to make Box 1 similar to the one that binds Jak2 (PXXPXP), did not have an effect on Jak1 binding (lanes 5, 7, and 8, respectively). The effect on Jak1 binding observed by introducing mutations within Box 1 confirm that this motif plays a role in the interaction with Jak1 (21).
To determine the role of Box 1 on Jak1 activation, mutations of Box 1 were introduced into full-length IFN␣R␤L and ex- The conserved residues are in boldface. The conserved prolines (Pro) within Box 1 are numbered 1-3, and the kinase associated with the receptor is indicated on the right column. Notice that Pro #1 or #2 are always present in cytokine receptors that associate with Jak2, whereas either one is absent in cytokine receptors that bind only Jak1 such as IFN␣R␤L, IL4R␣, and IL2R␤. pressed using a retroviral system. Clones were selected using G-418, and expression of the different constructs was confirmed by fluorescence-activated cell sorting analysis using a specific anti-IFN␣R␤L monoclonal antibody. Fig. 4 shows representative examples of the clones obtained. We next examined the activation of Jak1 in response to human IFN␣ (huIFN␣). Fig. 5A shows that mutation of Pro 289 or Pro 291 to Ala produced a significant decrease in activation of Jak1 by huIFN␣ when compared with control muIFN␣4 that activates the endogenous mouse receptor (compare lanes 2 with 3, and 5 with 6). The decrease in Jak 1 activation correlates with the decrease in binding observed in GST pull-down experiments (see Fig. 3). As expected, mutation of Pro 294 did not affect Jak1 activation by huIFN␣2 (compare lanes 8 and 9). Surprisingly, there was no correlation between Jak1 binding and activity for mutation Asn 292 to Pro and RV, because they did not affect binding but clearly affected kinase activity. The activation of STAT1 and STAT2 (Fig. 5A, bottom panels) correlated with the activation of Jak1 in the different mutants. The finding, that mutations of Box 1 such as N292P do not affect Jak1 binding in pull-down experiments but have an impact in Jak1 activation, prompted us to determine whether this mutation impairs the association with the kinase in vivo. We studied the interaction between IFN␣R␤L and Jak1 in cells stably expressing mutation N292P by performing immunoprecipitation with anti-␤L antibodies followed by immunoblotting with an anti-Jak1 monoclonal antibody. Fig. 5B demonstrates that there are no significant dif-ferences in the amount of Jak1 associated with IFN␣R␤L mutated in Asn 292 or a mutation of the receptor that does not affect Jak1 activation (Fig. 5B, 292.9 and EPOR.4, respectively). Moreover, mutation Asn 292 to Pro did not affect binding of STAT1 to IFN␣R␤L (data not shown) ruling out that this mutation affects the interaction with other signaling components. The finding that mutations that do not have a major impact in Jak1 association produced a decrease in kinase activation suggest that, although the Box 1 motif participates in Jak1 binding, it plays a more significant role in kinase activation (see "Discussion").
Identification of Jak1 Binding Sites on the IL2R␤-Zhu et al. (26) previously reported that some residues within Box 1 and the first Leu within Box 2 affected Jak1 binding. We next determined whether Box 1 and Box 2 of the IL2R␤ play the same role in the association with Jak1. Because Box 1 of IFN␣R␤L alone is unable to bind substantial amounts of Jak1, we truncated a GSTIL2R␤ after the Box 1 motif to determine if Box 1 of IL2R␤ alone was able to associate with Jak1. Fig. 6 shows that Jak1 interacts with the full-length cytoplasmic domain of IL2R␤ (lane 2) and with a fusion protein truncated at residue 322 (lanes 3 and 4), which contains the Box 1 and Box 2 motifs. However, Jak1 does not interact with a more proximal truncation (amino acid 265) that only includes the Box 1 motif (lane 5). This result indicates that Box 1 of the IL2R␤, like Box 1 of IFN␣R␤L, is not sufficient for Jak1 binding.
Because the Box 2 motifs of IL2R␤ and IFN␣R␤L are very similar (LEVL and VEVI, respectively), we also wanted to assess the contribution of the different amino acids of Box 2 of IL2R␤ to Jak1 binding. Fig. 6B (lane 2) shows that as in the case of IFN␣R␤L, mutation of the first hydrophobic residue of this motif (Leu 299 ) results in a marked decrease in Jak1 binding as previously reported (26), whereas mutations of the two other hydrophobic amino acids ( 301 VL 302 ) or Glu 300 have a less significant effect (lanes 3-5) on the interaction with the kinase. The lower panel (Fig. 6B, WB: GST) shows that similar amounts of GST fusion proteins were used. These data suggest that the function of the Box 1 and Box 2 motifs of IL2R␤ and IFN␣R␤L is very similar. Box 1 by itself does not sustain efficient Jak1 binding, and equivalent amino acids within Box 2 are important for the association with Jak1. However, they differ in the impact that mutations of Box 2 have on Jak1 activation, because substitution of Leu 299 of IL2R␤ completely abrogated Jak1 activation (26,28), whereas the same mutation of IFN␣R␤L did not have any effect (see "Discussion").
The IFN␥R␣ and IL10R␣ Chains Use a Different Jak1 Binding Motif-We finally sought to address which regions are important for Jak1 binding in receptors that do not contain Box 1 or Box 2 motifs. The previously reported Jak1 binding site of the IFN␥R␣ is localized very close to the transmembrane region and does not resemble the Box 1 motif of receptors that activate Jak2 or even the less conserved Box 1 of IFN␣R␤L or IL2R␤ (19). We produced GST fusion proteins encoding the entire cytoplasmic domain and a deletion of the cytoplasmic region containing the previously described Jak1 binding site (19) of the IFN␥R␣ to confirm that the Jak1 binding site was indeed unique. Deletion of the region containing the LPKS sequence (IFN␥R␣s) completely abrogated Jak1 binding (Fig.  7A, compare lanes 1 and 2), confirming that the binding site for this kinase resides within the first 15 amino acids of the cytoplasmic domain of IFN␥R␣. This result suggests that Jak1 does not interact in the same manner with all cytokine receptors. For instance, IFN␥R␣ has a Jak1 binding site that is very close to the transmembrane region (ϳ10 amino acids) and is different from the typical Box 1 responsible for Jak2 binding, the less conserved Box 1 observed in IL2R␤ and IFN␣R␤L (Fig. 1E), and the Box 2 motif.
We finally determined the Jak1 binding sequence used by the IL10R␣ chain, which, like the IFN␥R␣, contains neither a Box 1 nor Box 2 motif. Fig. 7B shows that GST fusion proteins, encoding the full-length IL10R␣, associate with Jak1 (Fig. 7, A  and B, lanes 3 and 2, respectively). A fusion protein containing only the first 39 amino acids of the cytoplasmic domain of IL10R␣ bound Jak1 as efficiently as the full-length receptor (Fig. 7B, lanes 3 and 4). These data suggest that, like in the case of IFN␥R␣, the Jak1 binding site in the IL10R␣ may be close to the transmembrane region.
To further define the residues involved in Jak1 binding, we performed mutagenesis using constructs in which 3 residues were simultaneously mutated to Ala. Two of these GST fusion proteins carrying mutation of residues 269 -271 and 272-274 showed a significant decrease in Jak1 binding (Fig. 7C, lanes 3  and 4). A decrease in Jak1 binding observed with a construct containing mutations of amino acids 266 -268 (lane 2) is due to lower amounts of this GST used in this experiment (Fig. 4C,  lower panel). The sequence SVLLFKK in the 269 -274 region of the IL10R␣ is similar to SIILPKS (amino acids 263-269) found in the IFN␥R␣. These motifs are 8 and 10 amino acids from the membrane, respectively. In the previous characterization of the Jak1 binding site of IFN␥R␣ (19), the first two hydrophobic residues (Ile 281 and Ile 282 ) were not included in the mutational analysis. We performed single point mutations of the first two hydrophobic amino acids present in this motif ( 269 VL 270 ) and Phe 272 in the IL10R␣. Phe 272 is in the same position as Pro 267 in the IFN␥R␣, which is critical for Jak1 binding. Fig. 7D shows that mutations of Leu 270 and Phe 272 (lanes 3-5), but not substitution of Val 269 (lane 2), in IL10R␣ significantly diminished  3 and 4), were used to pull down Jak1 from U-266 lysates as in previous experiments. C and D, characterization of the amino acids important for Jak1 binding in the IL10R␣ using constructs containing mutations of three (C) or one (D) amino acid to alanine. The indicated mutations were introduced in a GST fusion protein encoding the full-length cytoplasmic region of the IL10R␣ chain. Lower panels (A-D), filters were stripped and reprobed with an anti-GST monoclonal antibody. The arrows in B show degradation products of the full-length GSTIL10R␣. Jak1 binding. These data suggest that IL10R␣ and IFN␥R␣ interact with Jak1 in a similar manner, which differs from the interaction of Jak1 with IFN␣R␤L or IL2R␤. DISCUSSION Ligand-induced dimerization of receptor subunits brings Jak kinases into proximity allowing resting or low activity kinases to become fully activated. This probably occurs through transphosphorylation of tyrosines in the activation loop. Therefore, the position of the kinases on cytokine receptor chains may be critical to allow the proper conformation after receptor dimerization. The Box 1 motif is a proline-rich motif (33) in a hydrophobic context present in most cytokine receptor chains that interact with Jak2. This motif is invariably positioned 6 -12 residues from the transmembrane domain (Fig. 1E). The Box 1 motif on single-subunit cytokine receptors (i.e. GH-R, EPOR, and PRL-R) (Fig. 1) consists of two hydrophobic amino acids, proline #1, any amino acid, a valine or isoleucine, proline #2, any amino acid, proline #3, and a charged residue. In the case of multichain cytokine receptors that interact with Jak2 such as IL3R␣, IL5R␣, GMCSFR, and IFN␥R␤/IFNGR2, proline #3 may not be present (see Fig. 1, granulocyte-macrophace colony stimulating factor receptor, IL3R␣, and IFN␥R␤).
The interaction between Jak1 and cytokine receptor subunits does not appear to be so well defined. In most cases, the Box 1 motif is not as conserved (i.e. IL2R␤ and IL4R␣; see Fig.  1E, Box 1) or not present (i.e. IL10R␣). In the case of the IFN␥R␣ chain, analysis of a region proximal to the membrane (ϳ15 residues) revealed that only mutation of Pro 267 , within a motif with little homology to Box 1, affects Jak1 binding and activation. In contrast, the initial mapping of the Jak1 binding site on the IFN␣R␤L chain revealed that a region more distal from the membrane was more important for Jak1 binding than the Box 1 motif (21). These differences led us to question whether the interaction between cytokine receptors and Jak1 follows different rules than those for the association between single subunit receptors and Jak2.
Interaction between Jak1 and IFN␣R␤L or IL2R␤-The Box 1 and Box2 motifs of the IFN␣R␤L chain, unlike in the IL2R␤, are very close to each other and probably contribute to the formation of the binding site. It is likely that the Box 2 motif has the largest area of contact with Jak1, because GST fusion proteins that do not contain the Box 1 motif bind Jak1 almost as efficiently as wild type (21). However, the contribution of Box 1 to the formation of the Jak1 binding site is demonstrated by the finding that mutations of Pro 289 or Pro 291 decreases the interaction with this kinase even in the presence of Box 2. However, even though both motifs collaborate in Jak1 binding, Box 1 appears to be more important for the regulation of kinase activity. This is supported by the finding that kinase activation was affected even by mutations of Box 1 that did not affect Jak1 binding.
The Box 2 motifs in IL2R␤ and IFN␣R␤L are very similar, and Ala mutations of either motif appear to have equivalent effects on Jak1 binding. Moreover, the use of GST fusion proteins that contain only Box 1 revealed that neither Box 1 of IL2R␤ nor IFN␣R␤L could bind substantial amounts of Jak1 by itself. However, unlike IL2R␤ (26,28), mutations of Box 2 of IFN␣R␤L did not affect Jak1 activation. It is possible that the distance separating the Box 1 and Box 2 motifs in these receptors (7 and 38 residues for IFN␣R␤L and IL2R␤, respectively) results in a different type of interaction with Jak1. This type of model is supported by the finding that slightly different regions of Jak1 are required to associate with the IFN␣R␤L and IL2R␤ chains (29).
IFN␥R␣ and IL10R␣ Use a Different Jak1 Binding Motif-Unlike IFN␣R␤L and IL2R␤, IFN␥R␣ and IL10R␣ do not have an identifiable Box 1 or Box 2 motif. This prompted us to further characterize the interaction between Jak1 and these receptors. Our results indicate that the IL10R␣ has a Jak1 binding site close to the transmembrane region that has some degree of similarity with IFN␥R␣. Moreover, mutation of Phe 272 of IL10R␣, which is the equivalent to Pro 267 in IFN␣R␥, significantly decreased the interaction with Jak1. Altogether, these data suggest that there are at least two different types of interactions between Jak1 and cytokine receptors. First, cytokine receptors such as IL2R␤ and IFN␣R␤L interact with Jak1 through the Box 1 and Box 2 motifs that we will call Box 2A characterized by a hydrophobic-charged-hydrophobic-hydrophobic sequence. The Box 1 in these receptors is slightly different from Box 1 present in single subunit receptors that activate Jak2 (see Fig. 1E). Second, other cytokine receptors such as IL10R␣ and IFN␥R␣ use a different motif that is close to the membrane that we will call Box 2B. The Box 2B motif is characterized by a core of 4 hydrophobic residues flanked by a Ser and charged residues. However, mapping of the regions of Jak1 involved in the interaction with cytokine receptors strongly suggest that a region different from the Box 2B motif of the IL10R␣ also interacts with Jak1 (29). In the case of IFN␥R␣, a GST fusion protein, encoding the entire cytoplasmic domain except Box 2B, failed to interact with Jak1. These data suggest that Box 2B, at least in IFN␥R␣, should be sufficient to sustain Jak1 binding and probably activation.
Contrary to what was observed with the interaction between Jak2 and Box 1, the position of the Jak1 binding domain within the cytoplasmic region of cytokine receptors was more variable. This raises the hypothesis that the positioning of the kinase within the cytoplasmic domain may be important for correct kinase alignment after receptor dimerization, which would allow transphosphorylation of the tyrosines in the activation loop. In single subunit receptors, ligand binding results in homodimerization and, therefore, Jak2 is always at the same position in both members of the dimer. However, in heterodimeric receptors, positioning may be more critical. In this scenario, the Jak1 binding site in IFN␥R␣ is closer to the membrane to produce a correct alignment with Jak2, which also interacts with a binding site on the IFN␥R␤ chain that is close to the membrane. This suggests that the proximal Jak1 binding site on IFN␥R␣ evolved to pair the Jak2 site present in IFN␥R␤. The Tyk2 (34) and Jak1 binding sites on the ␣ and ␤L chains of IFN␣R are also at similar distances from the membrane (32 and 38 amino acids, respectively). This model would predict that the Tyk2 binding site on the IL10R␤ (CRFB4) (35,36) should be proximal to the membrane to match the membrane proximal Jak1 site present on the IL10R␣ chain. It should be considered that what we refer as "distance from the membrane" is only a relative measurement, because we do not know the secondary and tertiary structure of the cytoplasmic domain of these receptors. The finding that dimerization of the EPOR per se is not enough for activation may support the role of kinase orientation or stereo conformation in receptor activation (37).