Hls5 regulated erythroid differentiation by modulating GATA-1 activity

The erythroblastoid J2E cell line responds to erythropoietin (Epo) by morphologic maturation and hemoglobin synthesis.¹  However, on rare occasions, these cells have undergone a spontaneous lineage switch, and display features of monoblastoid cells, which do not respond to Epo.²  Thus, regulated expression of structural/functional genes involved in commitment to the erythroid lineage can be breached, as previously demonstrated with B cell to macrophage lineage switching.³ 

Hls5 and Hls7 were isolated as genes markedly up-regulated in the J2E monoblastoid variants.^2,4  Hls7 is the murine orthologue of Myeloid Leukemia Factor 1 (Mlf1), a gene involved in the t(3;5), associated with acute myeloid leukemia.⁵  Importantly, ectopic expression of Hls7/Mlf1 in J2E cells imposes a dramatic phenotypic change on the cells, rendering them monoblastoid.²  Hls5 is a recently identified member of the RING finger, B box, coiled coil (RBCC)⁴  or Tripartite motif (TRIM) family,⁶  which includes PML, a gene involved in acute promyelocytic leukemia. Hls5 is expressed in a wide variety of hemopoietic cell types, including fetal liver progenitors.⁴  The role of Hls5 in erythroid maturation was investigated in this study.

Cell lines were cultured with or without Epo, as described previously.^1,2  Assays for differentiation, cell-cycle progression, clonogenicity, confocal microscopy, cytocentrifugation, and staining have been detailed elsewhere,^1,2,4  as have methylcellulose colony assays.^1,2,7  Immunoblotting, immunoprecipitation, and in vitro pulldown assays were used as reported previously.^1,2  Yeast 2 hybrid analyses were described by Ingley et al,⁷  and electrophoretic mobility shift assays by Spadaccini et al.⁸  The M1α construct was used to determine GATA-1 activity,⁹  while chromatin immunoprecipitation experiments used primers for the β^maj globin promoter and HS2 regions of the β-globin locus.¹⁰ 

To determine the effects of elevated Hls5 on normal erythroid progenitors, fetal liver cells were infected with retroviruses expressing wild-type Hls5, myc-tagged Hls5 or empty vector control. Methylcellulose assays revealed that both burst forming units-erythroid (BFU-E) and colony forming units-erythroid (CFU-E) were suppressed between 40% and 70% by Hls5; in contrast, Hls5 increased myeloid colony formation (Figure 1A). Thus, elevated Hls5 levels inhibit the development and maturation of erythroid progenitors, which could be due to cell death, as over-expression of Hls5 can induce apoptosis.⁴ 

To examine the effects of Hls5 on immature erythroid cells in greater detail, J2E cells¹  were infected with vector alone, or Hls5-expressing retroviruses and termed J-Hls5 or J-Hls5-myc (Figure 1B). Confocal microscopy showed that Hls5 was present primarily in granular structures throughout the cytoplasm, a feature characteristic of RBCC/TRIM family members⁶ ; however, the protein was also detected in nuclear structures (Figure 1C).

Enforced expression of Hls5 did not alter the erythroblastoid morphology phenotype of the cells (Figure 1D). This observation contrasts markedly with J2E cells expressing Hls7/Mlf1, which appeared much larger and monoblastoid.²  It was concluded that, unlike Hls7/Mlf1, Hls5 does not primarily affect structural genes that control cellular morphology. It is noteworthy that Hls5 and Hls7/Mlf1 do not induce expression of each other (data not shown). Increasing Hls5 levels in J2E cells induced a slight reduction in clonogenicity (Figure 1E) and cell growth (Figure 1F), which could be attributed to reduced transit through G₁/S of the cell cycle (Figure 1G). This is consistent with our previous observations that elevated Hls5 perturbs cell-cycle progression in transiently transfected COS and HeLa cells.⁴ 

Strikingly, Epo-induced hemoglobin synthesis was almost completely abolished in Hls5-expressing cells (Figure 1H). Globin and GATA-1 transcripts were reduced in these cells (Figure 1I), as were protein levels (Figure 1J). Hls5, therefore, had a significant effect on globin synthesis and hemoglobin production, which could be caused, in part, by reduced GATA-1 levels.¹¹  However, the elimination of globin proteins indicates that post-transcriptional factors also influence globin content in J-Hls5 cells (Figure 1J).

Although the Hls5-expressing cells were unable to manufacture hemoglobin, they were still able to mature morphologically, when induced to differentiate⁷  (Figure 1K). The cells progressed beyond the proerythroblast/basophilic erythroblast boundary displaying decreased cell volume, nuclear condensation, and even nuclear extrusion. Elevated Hls5, therefore, inhibited hemoglobin synthesis, but not morphologic maturation. Similarly, suppression of the transcription factor EKLF has been shown to reduce hemoglobin production, but not interfere with morphologic differentiation.⁸  From these experiments it was concluded that the effects of Hls5 and Hls7/Mlf1 on erythroid differentiation are quite distinct: while Hls7/Mlf1 mainly affects cellular morphology, Hls5 influences biochemical aspects of erythroid differentiation such as globin production.

A yeast 2-hybrid screen was conducted to identify partner proteins that may elucidate the function of Hls5 in erythroid cells. FOG-1 was isolated in this screen as an Hls5-interacting protein. The FOG-1 clones that associated with Hls5 comprised amino acids 817 to 980, a region spanning FOG-1's seventh and eighth zinc fingers (Figure 2A). Figure 2B demonstrates that FOG-1 bound to full-length Hls5 but not to truncation mutants lacking the B Box; thus, the B Box of Hls5 is crucial for the interaction with FOG-1. Hls5 also associated with FOG-1 in mammalian cells, as demonstrated by coimmunoprecipitation (Figure 2C). The specific interaction of Hls5 with zinc fingers 7 and 8 of FOG-1 is interesting, as other fingers (1, 5, 6 and 9), but not 7 and 8, associate with GATA factors.⁹ 

It has recently been shown that zinc finger 3 of FOG-1 binds TACC3, a protein that also contains a Coiled Coil domain.¹²  As FOG-1 is a transcriptional cofactor for GATA-1,¹³  and TACC3 inhibits the function of FOG-1,¹²  the impact of Hls5 on GATA-1 activity was investigated. Data presented in Figure 2D show that Hls5 modestly enhanced the FOG-1 repression of the GATA-1 activity. Together, these data indicate that Hls5 associates with FOG-1 and affects GATA-1 transactivation.

A possible interplay between Hls5 and GATA-1, in the absence of FOG-1, was then explored. Interestingly, GATA-1 imposed a marked relocation of Hls5 from cytoplasmic granules⁴  into the nucleus, where it colocalized with GATA-1 (Figure 2E); furthermore, GATA-1 coimmunoprecipitated with Hls5 (Figure 2F). Yeast 2-hybrid experiments showed that this interaction occurred in the absence of FOG-1, and deletion mutant analysis revealed that the B Box, Coiled Coil domain of Hls5 associated primarily with the N-terminal finger of GATA-1 (Figure 2G,H). Significantly, Hls5 was able to repress GATA-1 transcriptional activity (Figure 2I), and interfere with GATA-1 binding to DNA (Figure 2J), in a dose-dependent manner. Deletion of the B Box, Coiled Coil domain abrogated the Hls5 inhibition of GATA-1 transactivation (data not shown). While these data indicate Hls5 can modulate GATA-1 transactivation independently, it is likely that Hls5 forms a complex with FOG-1 and GATA-1. The B Box, Coiled Coil domain of Hls5 appears crucial for these interactions that influence GATA-1 binding to DNA. As numerous GATA-1 partners have been identified,¹⁴  it is possible that different pools of GATA-1 exist that interact with specific proteins to influence expression of discrete subsets of genes.

Chromatin immunoprecipitation assays revealed a significant reduction in GATA-1 binding to the β globin locus of Hls5-expressing cells (Figure 2K); consequently, J-Hls5 cells had decreased transcripts for GATA-1 target genes, α and β globin (Figure 1I). Thus, Hls5 modulates GATA-1 by reducing mRNA and protein levels (Figure 1I, 1J) and by influencing DNA binding/transactivation (Figure 2I,J,K). However, some GATA-1 responsive genes (FOG-1 and Epo-receptor) were not affected by the increased Hls5 levels. These data suggest either concentration-dependent effects of GATA-1 on transcription of different genes, or site-specific effects of Hls5 on distinct GATA-1 elements.

The biologic consequences of elevated Hls5 in erythroid cells, viz severely impaired globin synthesis and restricted cell growth (Figure 1F-I), are consistent with Hls5 inhibiting the activity of GATA-1, because GATA-1 regulates globin gene transcription and cell cycle progression.^15,16  Hls5 might, therefore, play an important role in the maturation of red blood cells by modulating GATA-1 activity and expression of lineage-restricted functional genes.^{9,12,–14,17}  The presence of numerous GATA binding sites in the Hls5 promoter,⁴  adds another potential layer of cross-regulation between these molecules. Interestingly, other RBCC/TRIM family members (Herf1 and TIF1γ) have been implicated in transcriptional regulation in erythroid cells.^18,19  Moreover, another family member, PML, has been shown to associate with, and influence the activity of, GATA-2.²⁰  These observations suggest that RBCC/TRIM proteins represent a class of molecules that can modulate the activity of GATA factors.

Hls5 was originally isolated as a gene up-regulated during an erythroid-to-myeloid lineage switch,⁴  and independently as a proapoptotic gene activated during macrophage maturation.²¹  The suppression of GATA-1 activity by Hls5 could potentially promote maturation along the myeloid pathway, as inhibition of GATA-1 favors myelopoiesis over erythropoiesis.²²  Collectively, these data suggest that during the J2E erythroid/myeloid lineage switch, spontaneous activation of Hls5 suppresses globin synthesis, while Hls7/Mlf1 contributes to the altered morphology of the cells. We postulate that Hls5 and Hls7/Mlf1 act cooperatively to promote this lineage transition. These data add to the body of knowledge building on the genes involved in lineage switching.^2,3,23,–25 

This work was supported by the National Health and Medical Research Council (No. 458693), the Medical Research Foundation of Royal Perth Hospital, and the Cancer Council of Western Australia.

Contribution: R.E., I.J.M., and J.-P.L. designed and performed experiments, analyzed data, and contributed to writing the manuscript. L.W., J.G.B., A.S., R.S., and E.L. designed and undertook experiments, while M.C. and S.P.K. supported the research, designed experiments, analyzed data, and contributed to writing the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: S. Peter Klinken, Western Australian Institute for Medical Research, Level 6, Medical Research Foundation Building, Rear, 50 Murray Street, Perth WA 6000, Australia; e-mail: pklinken@waimr.uwa.edu.au.