Thyroid Hormone Receptor-interacting Protein 1 Modulates Cytokine and Nuclear Hormone Signaling in Erythroid Cells*

Erythropoietin (Epo) and thyroid hormone (T3) are key molecules in the development of red blood cells. We have shown previously that the tyrosine kinase Lyn is involved in differentiation signals emanating from an activated erythropoietin receptor. Here we demonstrate that Lyn interacts with thyroid hormone receptor-interacting protein 1 (Trip-1), a transcriptional regulator associated with the T3 receptor, providing a link between the Epo and T3 signaling pathways. Trip-1 co-localized with Lyn and the T3 receptor α in the cytoplasm/plasma membrane of erythroid cells but translocated to discrete nuclear foci shortly after Epo-induced differentiation. Our data reveal that T3 stimulated the proliferation of immature erythroid cells, and inhibited maturation promoted by erythropoietin. Removal of T3 reduced cell division and enhanced terminal differentiation. This was accompanied by large increases in the cell cycle inhibitor p27Kip1 and by increasing expression of erythroid transcription factors GATA-1, EKLF, and NF-E2. Strikingly, a truncated Trip-1 inhibited both erythropoietin-induced maturation and T3-initiated cell division. This mutant Trip-1 acted in a dominant negative fashion by eliminating endogenous Lyn, elevating p27Kip1, and blocking T3 response elements. These data demonstrate that Trip-1 can simultaneously modulate responses involving both cytokine and nuclear receptors.

Erythropoiesis, the process of generating red blood cells, is controlled by hormones that bind to cytokine receptor and nuclear hormone receptor families (1,2). Two well characterized molecules that strongly influence erythropoiesis are erythropoietin (Epo) 1 and thyroid hormone (T 3 ). Epo binds to a cell surface receptor of the cytokine receptor family (3), initiating an intracellular signaling cascade that has been deciphered gradually over the past decade (reviewed in Refs. 4 -7). Numerous signaling proteins are activated by Epo, including JAK2, STAT5, Ras, phosphatidylinositol 3-kinase, phospholipase C␥, and MAP kinase (5,7); activation of negative regulators such as SHP1, SOCS1, and CIS also occurs after receptor engagement (8 -10).
T 3 binds to the intracellular thyroid hormone receptor (TR), a member of the nuclear hormone receptor family, which regulates gene expression (11). T 3 has a potent effect on erythropoiesis, especially in hypothyroid patients who are often anemic; however, erythroid hyperplasia can occur in individuals with hyperthyroidism (12,13). It is noteworthy that the TR␣ isoform is expressed preferentially in differentiating erythroid cells (14) and that the v-erbA oncogene involved in avian erythroleukemia represents a mutated form of TR␣ (15,16). Studies with whole animals have indicated that T 3 stimulates erythropoiesis (17), whereas in vitro assays have shown that T 3 inhibits colony formation by erythroid progenitors (18,19). The elegant studies of Beug and colleagues (19 -24) have demonstrated that the balance between proliferation and differentiation can be altered by the introduction of exogenous TR␣ (c-erbA) or v-erbA into immature avian red blood cells.
We have examined Epo-initiated signaling in J2E erythroid cells as they proliferate, remain viable, produce hemoglobin, and undergo morphological maturation in response to Epo (25)(26)(27). Following Epo stimulation of these cells, phosphorylation changes to the Epo receptor, JAK2, STAT5, Ras-GAP, phosphatidylinositol 3-kinase, phospholipase C␥, and MAP kinase are identical to the kinetics reported in other cell systems (28). The tyrosine kinase Lyn is crucial for Epo-induced differentiation of immature J2E and R11 cell lines (27). Lyn associates with the Epo receptor and can phosphorylate the receptor and STAT5 in vitro (28,29). As the most abundant Src family kinase in red blood cells (30), Lyn also phosphorylates key erythrocyte membrane proteins (31). Our recent data indicate JAK2 is the primary kinase that initiates Epo signaling and that Lyn acts as a secondary kinase to promote differentiation (32) In this study we attempted to identify downstream effectors of Lyn in erythroid cells using a yeast two-hybrid screen. HS1, a known target for Src family kinases including Lyn, was identified in this screen and its effects on erythroid differentiation demonstrated (33). Here, we report on the interaction between Lyn and thyroid hormone receptor-interacting protein 1 (Trip-1), a transcriptional regulator that associates with TR␣ (34 -36). The Lyn/Trip-1 association, therefore, provides a link between the Epo and T 3 signaling pathways.

EXPERIMENTAL PROCEDURES
Cell Culture-Cells were grown in Dulbecco's modified Eagle's medium, 5% fetal calf serum, or serum depleted of T 3 (24). T 3 could not be detected in depleted sera by radioimmunoassay. The Epo-responsive J2E (25) and R11 (37) cell lines were derived from murine fetal liver cells transduced with retroviruses expressing v-Raf/v-Myc (J2E) or v-Raf (R11). Differentiation of J2E and R11 cell lines was initiated with Epo (5 units/ml). Nuclear hormones were used at a final concentration of 1 M. Viability was determined by eosin dye exclusion and hemoglobin synthesis by benzidine staining (28). Cell morphology was examined following cytocentrifugation onto glass slides and Wright-Giemsa staining (26). Proliferation was assayed by [ 3 H]thymidine incorporation (26). Fetal liver cells were plated in methylcellulose for CFU-E and BFU-E assays as described previously (38) before benzidine-positive colonies were enumerated. All graphical data are represented as the mean Ϯ S.D. (n Ն 3).
Retroviral Infection of Cells-Sense (Cs) and antisense (C␣) tTrip-1 cDNAs encoding amino acids 1-171 were generated by polymerase chain reaction, and the fragments were subcloned into the pMSCV2.2neo vector (42). Amphotropic and ecotropic retroviruses expressing the Cs and C␣ constructs were used to infect erythroid cells as described previously (33). The efficiency of fetal liver cell infection was at least 50% (43). Numerous independent clones were isolated, and representative clones are shown.
Transient Transfection Assays-Cells (10 7 ) were electroporated with 10 g of the TR reporter constructs pF2-Luc or pDR4-Luc (44) and 1 g of pRL-SV40 (Promega, Madison, WI) at 300 V/1000 microfarad using a Gene Pulser II (Bio-Rad). Transfected cells were harvested after 48 h of culture in the presence or absence of T 3 , and dual luciferase reporter assays were performed (Promega) on an Autolumat LB953 (EG&G Berthold, Oak Ridge, TN).

RESULTS
Trip-1 Associates with Lyn and TR␣-To identify specific binding partners of Lyn, a yeast two-hybrid screen was conducted using wild-type Lyn and the kinase-inactive Y397F mutant (33); Fig. 1A shows the association of Lyn with Trip-1. In addition to transcriptionally regulating TR␣ (34 -36), Trip-1 also possesses intrinsic helicase activity, and independently it has also been described as SUG1, a component of the 26-S proteasome (41,45).
In vitro studies with purified proteins revealed that Lyn and Trip-1 interacted directly (Fig. 1B) in a phosphotyrosine-independent manner (Fig. 1, A-C). Trip-1 also bound the kinaseinactive Y397F mutant of Lyn (Fig. 1, A and B), indicating that the enzymatic activity of Lyn was not required for this association. Deletion analyses showed that the SH2 and, to a lesser extent, the SH3 domains of Lyn were responsible for Trip-1 binding in vitro (Fig. 1, B and C); these observations were confirmed using lysates from erythroid cells (data not shown). The regions of Trip-1 that bound Lyn were then analyzed. Fig.  1D shows that the amino-terminal of Trip-1 but not the coiledcoil domain alone was required for Lyn binding. This region is distinct from the highly conserved ATPase/DNA helicase AAA (ATPase associate with a variety of cellular activities) domain of Trip-1 needed to bind the TR␣ (41). Co-immunoprecipitation experiments performed on lysates from erythroid cell lines show that Lyn does indeed associate with Trip-1 in vivo (Fig. 1E). Similarly, Trip-1 and TR␣ coimmunoprecipitated in these cells. These studies show that Trip-1 is able to bind both Lyn and TR␣ in erythroid cells, thereby connecting discrete signaling pathways involving Epo and T 3 .
Trip-1 Co-localizes with Lyn and TR␣ in the Cytoplasm of Erythroid Cells-Having established a biochemical interaction between Lyn and Trip-1, the subcellular localization of these proteins was ascertained in uninduced erythroid cells. Lyn was found primarily in the cytoplasm of erythroid cells, with significant concentration at the plasma membrane ( Fig. 2A). As Trip-1 was also distributed in the cytosol and plasma membrane, appreciable co-localization between Lyn and Trip-1 was observed ( Fig. 2A). As described by Zhu et al. (46), TR␣ was detected in both the cytoplasm and nucleus. Consequently, Trip-1 co-localized with TR␣ in the cytoplasm, but not the nucleus, of these unstimulated erythroid cells (Fig. 2B). Comparable results were obtained with other erythroid cells (data not shown). These experiments demonstrate that the association of Trip-1 with Lyn and TR␣ occurs in the cytoplasm/ plasma membrane of erythroid cells.
T 3 Inhibits Epo-induced Differentiation-As both Epo and T 3 affect erythropoiesis (1, 2), and Trip-1 was shown to associate with Lyn and TR␣ (Figs. 1 and 2), biological evidence for interplay between these pathways was sought. To this end, T 3 and Epo concentrations were manipulated and cellular responses monitored. Fig. 3A shows that the removal of T 3 enhanced Epo-induced hemoglobin production, whereas the addition of T 3 severely impeded synthesis of the oxygen carrier. These effects were T 3 -specific, as reverse T 3 (Fig. 3A) and a variety of nuclear hormones (data not shown) had no effect on differentiation. The inhibitory effects of T 3 on hemoglobin syn-thesis were concentration-dependent with an IC 50 of 100 pM. Furthermore, morphological maturation was severely retarded in the presence of T 3 ; nuclear condensation, cytoplasmic acidophilia, reduced cell size, and enucleation were all restricted by T 3 (Fig. 3B). Conversely, removal of T 3 accelerated the appearance of erythroid cells with a more mature phenotype, in particular orthochromatic erythroblasts and reticulocytes. Identical results were obtained with other Epo-responsive cell lines (data not shown).
In marked contrast with the inhibition of Epo-initiated differentiation, T 3 promoted [ 3 H]thymidine uptake (Fig. 3C). When T 3 was removed from cultures, DNA synthesis almost ceased; however, replication resumed upon re-introduction of T 3 . This observation was supported by monitoring cell numbers in these cultures (data not shown).
To extend the analysis of erythroid cells beyond cell lines, the Epo/T 3 axis was examined in red cell progenitors from murine fetal livers. Fig. 4A shows that the inhibitory effects of T 3 on differentiation were not restricted to immortalized cells, as increasing the T 3 concentration reduced Epo-induced colonies from normal progenitors. However, the effect was more pronounced on erythroid colony-forming units (CFU-E) than erythroid burst-forming units (BFU-E). These data indicate that T 3 also had an inhibitory effect on Epo-induced differentiation of normal erythroid cells.
To determine whether T 3 affected fetal liver erythroid progenitors before exposure to Epo, cells were pre-incubated with T 3 and then treated with Epo. Intriguingly, when T 3 was added prior to Epo, both the BFU-E and CFU-E numbers rose (Fig.  4B). This expansion of the erythroid progenitor compartment by pre-incubation with T 3 indicates that the effects of T 3 are stage-specific. Taken together these results show that T 3 promotes proliferation and the expansion of immature erythroid cells at the expense of maturation, whereas Epo favors terminal differentiation toward a nonreplicating state.
T 3 Affects Erythroid Transcription Factors and p27 Kip1 -To identify the biochemical mechanism for the effects of T 3 on erythroid proliferation and differentiation, an immunoblot analysis was performed on key transcription factors and cell cycle regulators. The effects of T 3 were quite striking as the levels of erythroid-restricted transcription factors EKLF, NF-E2, and GATA-1 rose 3-10-fold when T 3 concentrations were reduced (Fig. 5, left). In addition, withdrawal of T 3 resulted in a 20-fold increase in p27 Kip1 , a cell cycle inhibitor important for the maturation of erythroid cells (47,48). In contrast, no change was observed in p57 Kip2 (Fig. 5, right). Interestingly, the Lyn and TR␣ content doubled when cells were cultured in T 3 -depleted media. These observations provide biochemical explanations for enhanced differentiation and restricted replication after T 3 withdrawal i.e. raised GATA-1, EKLF, and NF-E2 facilitate red cell maturation, whereas elevated p27 Kip1 enables cells to exit the cell cycle and enter the terminally differentiated state. At this stage it is unclear whether the addition of T 3 increases the frequency of immature cells containing less EKLF, NF-E2, and GATA-1 or directly reduces the expression of these factors in erythroid cells.
Trip-1 Affects Responsiveness to Epo and T 3 -To determine whether Trip-1 could simultaneously regulate differentiation signaling by Epo and proliferation promoted by T 3 , a truncated Trip-1 (tTrip-1) encompassing the Lyn-binding domain (Fig.  1D) was introduced into J2E cells. Numerous independently isolated transfectants were termed JCs cells, whereas the antisense controls were labeled JC␣. Significantly, tTrip-1 had a marked inhibitory effect on Epo-induced hemoglobin production and morphological maturation; very few hemoglobin-synthesizing cells were detected in JCs cultures, and the cells were incapable of proceeding beyond the proerythroblast/basophilic erythroblast boundary of maturation (Fig. 6, A and B). Moreover, T 3 -induced proliferation was severely impeded (Fig. 6C).
In contrast, the antisense controls behaved like control cells or those infected with the retroviral vector alone (data not shown). These data demonstrate that the truncated Trip-1 acted in a dominant negative manner and had a profound impact on both Epo-induced differentiation and T 3 -promoted cell division.
The effect of tTrip-1 on normal erythroid progenitors was investigated by infecting fetal liver cells and observing colony formation. Consistent with the effects of tTrip-1 on cell lines, the truncated Trip-1 decreased the number of Epo-induced BFU-E and CFU-E (Fig. 7A). Furthermore, tTrip-1 greatly diminished the ability of T 3 pre-incubation to expand the progenitor compartment (Fig. 7B). It was concluded from these experiments that Trip-1 plays an important role in regulating the responses of normal as well as immortalized erythroid cells to Epo and T 3 .
tTrip-1 Affects Lyn and p27 Kip1 Levels-The biochemical mechanisms by which tTrip-1 inhibited Epo and T 3 action were then investigated. The inhibition of Epo-induced differentiation by tTrip-1 coincided with the elimination of Lyn but not other tyrosine kinases such as Fyn, Src, Syk, Lck, and Hck ( Fig.  8A and data not shown). Conversely, the elevated levels of p27 Kip1 in JCs cells correlated with reduced proliferation (Fig.  8A). However, the inability to differentiate was not caused by a reduction in GATA-1, EKLF, NF-E2, globins, or endogenous Trip-1 (Fig. 8A), nor was it due to restricted phosphorylation of the Epo receptor or STAT5 (data not shown). It is noteworthy that the antisense control did not affect Trip-1 protein levels, validating its use as an additional control (Figs. 6 and 7).
To determine whether tTrip-1 also interfered with T 3 -induced transcription, activation of T 3 response elements was examined in JCs cells. The cells were transfected with either direct repeats or inverted palindromes of the T 3 response elements and were then exposed to T 3 . Significantly tTrip-1 negated the T 3 responsiveness of both elements (Fig. 8B and data  not shown). It was concluded from this series of experiments that tTrip-1 blocked erythroid differentiation and proliferation by suppressing endogenous Lyn, increasing p27 Kip1 and interfering with the transcriptional activity of T 3 response elements.
Trip-1 Translocates to the Nucleus during Erythroid Differentiation-Trip-1 localized primarily in the cytoplasm of uninduced erythroid cells (Fig. 2). The subcellular localization of this protein was then examined in cells stimulated with Epo or when T 3 was withdrawn from culture. Strikingly, cytoplasmic Trip-1 translocated to discrete nuclear regions after 30 min of exposure to Epo or removal of T 3 (Fig. 9, A and B). Altering the compartmental balance of Trip-1, therefore, coincided with enhanced differentiation and reduced replication. Trip-1 also relocated to the nucleus in JCs cells (data not shown), indicating that the inhibitory effects of tTrip-1 were not due to impaired nuclear translocation. DISCUSSION In this article we have demonstrated that Trip-1, a transcriptional regulator of TR␣ (34,36), associates with the tyrosine kinase Lyn in erythroid cells. The interaction between Lyn and Trip-1 was initially identified by a yeast two-hybrid screen and was confirmed by in vitro binding, co-immunoprecipitations, and intracellular co-localization ( Figs. 1 and 2). Because Lyn is involved in the Epo signaling cascade (27,29,32), Trip-1 provides a link between pathways mediated by cytokine receptors and nuclear hormone receptors.
The parallels between the Trip-1 association with Lyn and TR␣ and the interaction of p62 ORCA with Lck and COUP-TFII are striking. Like Trip-1 and Lyn, p62 ORCA associates with Lck (the closest Src kinase family member to Lyn) through phos- photyrosine-independent binding of p62 ORCA to the SH2 domain of Lck (49). Furthermore, the Trip-1/TR␣ association (34,36) is similar to p62 ORCA interacting with the nuclear hormone receptor COUP-TFII (50), which has been implicated in the regulation of globin genes (51). Marcus et al. (50) proposed "a role for ORCA and related factors in mediating cross-talk among distinct signal transduction pathways important for cellular growth and differentiation." Here we have provided the biological and biochemical evidence to support this proposition and have shown that Trip-1 can modulate pathways activated by Epo and T 3 .
The dominant negative effects of the truncated Trip-1 were quite profound, as it simultaneously inhibited Epo-induced differentiation in immortalized and normal erythroid cells and suppressed T 3 -mediated proliferation. The introduction of tTrip-1 resulted in a complete loss of endogenous Lyn, increased p27 Kip1 , and an inability to activate T 3 response elements. As Trip-1 is also a component of the 26-S proteasome (45), perhaps the truncated Trip-1 fosters proteasomal degradation of Lyn. The importance of Trip-1 to the T 3 pathway was illustrated by the suppression of T 3 response elements with the truncated Trip-1.
It is noteworthy that Trip-1 localized with Lyn and TR␣ in the cytoplasm/plasma membrane of uninduced erythroid cells and then translocated to discrete nuclear foci within 30 min of Epo stimulation or after withdrawal of T 3 . To our knowledge this is the first description of Trip-1 relocation between subcellular compartments correlating with enhanced differentiation. Studies are currently under way to determine the nature of these nuclear structures and their function during erythroid maturation.
Epo and T 3 had major effects on erythroid maturation in both cell lines and normal progenitors. Whereas Epo promoted differentiation (manifest by hemoglobin synthesis and morphological maturation), T 3 stimulated proliferation at the expense of terminal differentiation. Similarly, it has been shown that T 3 prevents hemoglobin production by NFS-60 cells and increases the erythrocyte yield from erythroblasts (18,20). Our data with normal erythroid progenitors confirm that T 3 inhibits colony formation (18). However, we also demonstrated that T 3 is able to expand the erythroid progenitor compartment prior to Epo stimulation, which supports the notion that the effects of T 3 may be stage-specific (18,19,24).
Withdrawal of T 3 from erythroid cells produced large increases in erythroid-restricted transcription factors GATA-1, EKLF, and NF-E2, which have been strongly implicated in the control of red cell maturation, especially in hemoglobin production (52)(53)(54)(55)(56). Increasing the concentration of these transcription factors, therefore, promotes differentiation. Cross-regulation of these nuclear proteins may also be important because TR␣ and COUP-TFII together suppress GATA-1 transcription (57), whereas TR␣ is able to associate directly with NF-E2 (58). Altering the concentration, activity, or combinations of these DNA proteins can have a major impact upon expression of genes required for red cell maturation. Removal of T 3 also caused a marked elevation in p27 Kip1 , which is significant because entry of erythroid cells into the noncycling, terminally differentiated state involves increasing p27 Kip1 levels (48). Thus, terminal differentiation was enhanced by the combination of cell cycle exit and elevated transcription factor levels.
It is conceivable that the Epo/T 3 axis provides a complementary mechanism for expanding erythroid progenitors and increasing cell numbers before terminal differentiation to generate the correct number of functionally mature red blood cells. As TR␣ has been proposed to act as a switch between proliferation and differentiation (24), Trip-1 may be a vehicle for coordinating the biological responses initiated by T 3 and Epo.