ASB4 is expressed in the vascular lineage of differentiating ES cells.
In order to identify genes important during vascular development, we performed a set of experiments that compared the gene expression profiles of Flk1+
cells isolated during different stages of differentiating ES cells (described by Wang et al. [46
]). Out of 20,000 genes, ASB4 was the most highly differentially expressed gene in Flk1+
cells, compared to Flk1−
cells, at early time points of differentiation (Fig. ). This differential expression was confirmed by RT-PCR analysis showing that ASB4 mRNA is highly enriched in the Flk1+
population at 84 h, 95 h, and 192 h of differentiation but is undetectable at 72 h of differentiation (before Flk1 is expressed in this system) (Fig. ). Since Flk1 is expressed in early precursor cells of the endothelial, hematopoietic, vascular smooth muscle, and cardiomyocyte lineages (20
), we reasoned that ASB4 could be important during cardio- and/or hematovascular development. In support of this, supervised hierarchical clustering analysis with only genes demonstrating a ≥1.5-fold absolute mean fold difference in 84-h Flk1+
cells showed that ASB4 clusters closely with other genes known to be important in cardiovascular development, including Flk1, fibronectin, Gata2, and Gata4 (Fig. ).
FIG. 1. ASB4 is highly differentially expressed in the vascular lineage during ES cell differentiation (Diff). (A) Supervised hierarchical cluster of genes with a ≥1.5-fold absolute mean difference in 84-h Flk1+ cells from embryoid bodies. The (more ...) The SOCS box-containing protein ASB4 assembles with a ubiquitin ligase complex.
Since other SOCS box-containing proteins function as substrate adaptor proteins for elongin B/elongin C/cullin/Roc ubiquitin ligase complexes by binding the complex in a SOCS box-dependent manner, we investigated whether ASB4 exists in such complexes and can function as a ubiquitin ligase. In transient transfection assays with HEK-293T cells, Flag-ASB4, but not the EV or a mutant lacking the C-terminal SOCS box (ΔSOCS), coprecipitated endogenous elongin B, Cul5, Roc1, and to a lesser extent Cul2 (Fig. ). The signal for coprecipitation of Cul2 was greatly increased with lysates from COS7 cells infected with Flag-ASB4 adenovirus (Fig. ), indicating that ASB4 can use Cul2 or Cul5 complexes and that this preference is dependent upon the cellular context. Given that other ASBs associate with Cul5/Roc1 and Cul5/Roc2 complexes, further studies are needed to determine if the Cul2 interaction is unique to ASB4 or inherent to all ASBs in the appropriate context and if ASB4 function is affected by the cullin complex used. Notably, VHL associates with Cul2, suggesting that ASB4 may share molecular mechanisms with VHL (5
FIG. 2. ASB4 associates with a ubiquitin ligase complex. (A) HEK-293T cells were transfected with Flag-ASB4, Flag-ASB4ΔSOCS (flag-ΔSOCS), or the EV pCMV. Lysates were immunoprecipitated (IP) with anti-Flag-agarose beads and subjected to immunoblotting (more ...)
To evaluate whether ASB4 functions as part of a ubiquitin ligase complex, in vivo ubiquitination assays were performed. HEK-293T cells were transfected with Flag-ASB4 for 20 h, followed by treatment with the proteasome inhibitor MG-132 to allow polyubiquitin-tagged protein accumulation. Flag-tagged immunoprecipitates were then immunoblotted with anti-Flag and antiubiquitin antibodies. Strong ubiquitin immunoreactivity was detected in Flag-tagged immunoprecipitates of Flag-ASB4-transfected cells after MG-132 treatment (Fig. , part 1). We reasoned that this signal could represent ubiquitinated ASB4-associated proteins, ubiquitinated ASB4 itself, or both. A 7-kDa Flag-tagged immunoreactive laddering pattern was detected in Flag-tagged immunoprecipitates after treatment with MG-132, indicating that ASB4 itself is ubiquitinated (Fig. , part 2). Furthermore, Flag-ASB4 coprecipitates with myc-ASB4 in cotransfected HEK-293T cells, indicating that ASB4 complexes with itself (Fig. ). Since additional coimmunoprecipitation experiments with HEK-293T cells did not indicate that substrates are stably associated with ASB4 in this cell type (data not shown), these data indicate that ASB4 is most likely autoubiquitinated (which may represent a mechanism of self-regulation) and thus behaves like a ubiquitin ligase in this system.
ASB4 is expressed in the embryonic vasculature during a narrow time window.
To determine that ASB4 expression is high in anatomic locations known to harbor active vascular development and remodeling and to further define exactly which tissue(s) ASB4 expression is confined to, we analyzed its expression in embryonic and adult tissues. Global ASB4 mRNA expression is comparatively low in E7.5 embryos but quickly increases until E9.5 (Fig. ). Highly vasculogenic tissues such as the allantois, yolk sac, and placenta all express high levels of ASB4. However, while ASB4 expression in the adult is highest in the testis, ovary, and heart, it is undetectable in highly vascular organs such as the lung, kidney, and liver (Fig. ), suggesting that ASB4 function may be critical to proper vascular development but dispensable for the maintenance of adult vessels in some tissues. Furthermore, tissues containing high numbers of hematopoietic cells such as those of the spleen and bone marrow have undetectable levels of ASB4 mRNA expression.
FIG. 3. ASB4 mRNA expression in embryonic and adult tissues. (A) Northern blot analysis of ASB4 expression levels in whole embryos and dissected extraembryonic tissues. (B) Northern blot analysis of ASB4 expression levels in adult tissues. BM, bone marrow. Ethidium (more ...)
To further define the anatomic location of ASB4 expression during embryogenesis, we performed whole-mount in situ hybridization analysis with DIG-labeled ASB4 antisense riboprobes on mouse embryos at various gestational stages. In E9.5 embryos, ASB4 is expressed in the intersomitic vessels, dorsal aorta, forelimb buds, allantois/umbilical vessels, vitelline vessels, septum transversum, proepicardium, capillary plexi of the head and branchial arches, endocardium, and yolk sac vasculature (Fig. ). In E10.5 embryos, areas with high ASB4 expression levels include the forelimb and hind limb buds, intersomitic vessels, peripheral liver cells, and umbilical vessels (Fig. ). In E11.5 embryos, high levels of ASB4 expression are limited to the forelimbs and hind limbs and the most caudal (and most recently formed) intersomitic vessels (Fig. ). Notably, E7.5 embryos show no remarkable ASB4 staining (data not shown). Analysis with sense probes confirmed the specificity of the antisense signal (Fig. ).
FIG. 4. Localization of ASB4 mRNA during mouse embryogenesis. Whole-mount in situ hybridization with DIG-labeled RNA probes for ASB4 was performed on E9.5 to E11.5 mouse embryos. (A) E9.5 embryo. Arrow, allantois; arrowhead, forelimb; bar, 500 μm. (B) (more ...)
The most striking recurring pattern of ASB4 embryonic expression is its high levels in primitive capillary plexi, followed by downregulation as vessels mature. At E9.5, ASB4 is highly expressed in the capillary plexi of the head and branchial arches (Fig. ), but by E10.5 expression in these vascular beds is no longer detectable (Fig. ). Similarly, ASB4 expression is high in intersomitic vessels at E10.5 but by E11.5 is confined to only the most caudal (and thus recently formed) intersomitic vessels (Fig. ). This pattern is recapitulated in the placenta, with high ASB4 expression in the immature placenta at E9.5 compared to the mature placenta at E13.5 (Fig. ). Finally, ASB4 expression is undetectable in highly vascularized adult organs such as the kidney and lung (Fig. ). This dynamic temporal regulation of ASB4 expression suggests that its function is temporally limited. Notably, the endothelium is exposed to drastic increases in oxygen tension between E9.5 and E10.5 as the placenta develops and maternal-fetal blood gas exchange is initiated. Since another SOCS box-containing protein, VHL, is known to regulate the cellular response to changing oxygen concentrations and since ASB4 may share molecular mechanisms with VHL, we postulated that ASB4 may function during development in an oxygen-dependent manner to modulate an endothelium-specific oxygen response during a time at which oxygen tension drastically increases.
ASB4 binds to FIH by using a conserved motif.
As a first step in the mechanistic characterization of ASB4, we attempted to identify ASB4 binding partners with the yeast two-hybrid system. We used an ASB4 mutant lacking the C-terminal SOCS box (ASB4ΔSOCS) as bait to avoid reconfirmation of interactions with known SOCS box binding partners elongin B and elongin C. With a human heart oligo(dT)-primed pretransformed yeast library, we screened more than 1.5 × 107 independent clones and identified FIH as an interacting protein under stringent conditions (positive clones grew on plates lacking tryptophan, leucine, histidine, and adenine). The FIH prey clone encoded all but the five most N-terminal amino acids of the human FIH protein. This interaction was confirmed in a wheel assay in which yeast cells were transformed with various combinations of ASB4ΔSOCS bait, FIH prey, EVs (V1, V2), or mismatched nonspecific controls (Fig. ). p53- and T-antigen (T)-containing plasmids were used as positive controls for an interaction in this system. Flag-tagged immunoprecipitates of HEK-293T cells overexpressing Flag-ASB4 but not Flag-ASB1 coimmunoprecipitated endogenous FIH, confirming the specificity of this interaction in mammalian cells (Fig. ).
FIG. 5. ASB4 binds to FIH through a conserved motif in AR6. (A) Yeast wheel assay. Yeast cells transformed with different combinations of bait (ASB4ΔSOCS) and prey (FIH) constructs were spread onto a high-stringency selective plate lacking tryptophan, (more ...)
FIH is an asparagine hydroxylase that is known to hydroxylate at least four proteins, HIF1α, HIF2α (27
) IκBα, and the NF-κB precursor protein p105 (3
). Since the hydroxyl group is derived directly from atmospheric dioxygen, this reaction is oxygen dependent and thus FIH is often referred to as a cellular “oxygen sensor.” This raised the intriguing possibility that ASB4 is a hydroxylation substrate of FIH and may therefore be regulated by atmospheric oxygen tension. FIH has been shown to bind to and hydroxylate its substrates on the β carbon of a leucine-flanked asparagine residue within ARs (3
). The flanking leucines bind first to position the asparagine for optimal binding in the FIH catalytic cleft via a process called “induced fit” (7
). Sequence alignment of ASB4 with four other known FIH substrates revealed a conserved leucine-flanked asparagine in AR6 (Fig. ). This alignment also indicates highly conserved residues surrounding the asparagine, including perfect conservation of Val 245. Notably, all of these residues (leucines, valine, and asparagine) have previously been shown to be involved in FIH-substrate interactions (29
). To test whether AR6 is necessary for interaction with FIH, a variety of N-terminally Flag-tagged mutants were generated that sequentially lack different domains of the ASB4 protein (Fig. ). Since the hydrophobic interactions between helices of adjacent ARs are crucial for proper folding, we carefully positioned the borders of the deletions in order to best conserve the modular structure of the mutants. These mutants were transiently transfected into HEK-293T cells, Flag immunoprecipitated, and immunoblotted with anti-FIH antibody to detect coprecipitating endogenous FIH protein (Fig. ). ASB4 AR6 and -7 are necessary for interaction with FIH since mutants lacking either AR6 or -7 were unable to coprecipitate FIH. Mutants lacking other ARs (AR1, -4, -5, or -8, -9) or the SOCS box were still able to coprecipitate FIH, indicating that these domains are dispensable for FIH binding and that deletion of single ARs does not disrupt the tertiary structure of the FIH-interacting motif.
To test the involvement of the leucine-flanked asparagine and surrounding residues in FIH binding, we generated a number of point mutants in and around this motif to test their involvement in the FIH interaction. Each Flag-tagged point mutant was transiently transfected with N-terminal myc-FIH, immunoprecipitated with anti-Flag-conjugated agarose beads, and immunoblotted with anti-myc antibody to detect coprecipitating myc-FIH. Consistent with the “induced fit” model, point mutation of any of the flanking leucines to aspartate residues (L238D, L239D, and L256D) reduced coprecipitating myc-FIH to undetectable levels (Fig. ). Single-point mutation of the asparagine residue (N246A) reduced binding to undetectable levels, while mutation of the adjacent glutamate (E244A) or valine (V245T) residue reduced but did not abolish binding. Mutation of the EVN residues to ATA abolished binding to undetectable levels. Taken together, these data indicate that ASB4 binds to FIH through a conserved hydroxylation motif in AR6 and suggest that ASB4 may be a novel FIH hydroxylation substrate.
We also tested the alternative hypothesis that ASB4 inhibits FIH activity either by ubiquitin-mediated proteasomal degradation or by enzymatic blockade via a competitive binding mechanism. We found that ASB4 has no effect on FIH ubiquitination and that overexpression of ASB4 does not result in the activation of HIF-responsive luciferase reporter genes or in the upregulation of multiple HIF target genes in endothelial cell lines (data not shown).
ASB4 is hydroxylated on asparagine 246 by an oxygen-dependent mechanism.
To determine if the leucine-flanked asparagine is positioned in an appropriate conformation to allow binding to and hydroxylation by FIH, we generated a structural model of ASB4 AR6 and -7 (Fig. ). This model shows that the asparagine and flanking leucine residues of ASB4 are optimally positioned to allow an induced fit of the asparagine into the catalytic cleft of FIH and prompted further studies to examine the possibility of FIH-mediated ASB4 hydroxylation.
FIG. 6. ASB4 is hydroxylated on asparagine 246 by an oxygen-dependent mechanism. (A) Structural model of ASB4 AR6 and -7 (yellow ribbon backbone) positioned above the FIH (gray stick backbone)/HIF peptide (purple ribbon backbone) crystal structure (PDB accession (more ...)
To test the hypothesis that ASB4 is an FIH hydroxylation substrate, adenovirus-overexpressed Flag-ASB4 was Flag immunoprecipitated from COS7 cells grown under normoxic culture conditions, in-gel trypsinized, and analyzed by MALDI-TOF MS. The predicted mass/charge ratio (m/z) of the Asn 246-containing trypsin-digested peptide TLLDNNAEVNAR is 1,329, and hydroxylation of this peptide will result in a 16-Da shift to 1,345. By MALDI-TOF analysis, we detected the 1,329-Da peptide and a 1,345-Da peptide that did not match the predicted m/z value for any of the peptides from trypsin-digested Flag-ASB4, suggesting that the m/z 1,345 peak represents the hydroxylated form of the 1,329-Da peptide (Fig. ). This was confirmed through MALDI-TOF-TOF MS/MS sequencing analysis of the 1,329- and 1,345-Da peptides, separately. Alignment of the MS/MS spectra of the 1,329- and 1,345-Da peptides demonstrates that the 16-Da shift due to hydroxylation is present only in peptide ions containing Asn 246, indicative of FIH-mediated asparagine hydroxylation of ASB4 (Fig. ).
Under normoxic conditions (21% oxygen), the ratio of unhydroxylated to hydroxylated ASB4 peptide is 3.1:1, indicating that under normoxic conditions, only ~25% of ASB4 is hydroxylated (Fig. ). To test if this was due to saturation of endogenous FIH enzymatic activity by supraphysiologic concentrations of transfected ASB4, we coexpressed ASB4 and FIH in HEK-293T cells, which resulted in the hydroxylation of nearly all (89%) of the ASB4. These results emphasize the importance of the stoichiometric ratio of FIH and its substrates for hydroxylation activity, and further studies that evaluate what percentage of endogenous ASB4 is hydroxylated are needed. To confirm that ASB4 hydroxylation was indeed dependent on endogenous FIH, we cotransfected cells with siRNA duplexes against either FIH or GAPDH as a negative control. Immunoblot analysis confirmed the successful knockdown of FIH, and the percentage of hydroxylated ASB4 in cells lacking FIH (7%) was threefold less than that in the negative controls (21%). Since FIH-mediated hydroxylation of known substrates is O2 dependent (and thus inhibited by hypoxia), we tested whether hypoxic conditions would decrease ASB4 hydroxylation. Importantly, when HEK-293T cells transfected with Flag-ASB4 were cultured under hypoxic conditions (1% oxygen), the percentage of hydroxylated ASB4 peptide dropped to 7%. Similar results were obtained when these cells were treated with chemical hypoxia mimetics (CoCl2 and dipyridyl) or hydroxylase inhibitors (dimethyloxalylglycine) (data not shown). Together, these data demonstrate that FIH hydroxylates ASB4 at Asn 246 via an oxygen-dependent mechanism and suggest that ASB4 function may be regulated by oxygen concentrations.
We next tested whether ASB4 hydroxylation affects ASB4 stability or autoubiquitination by repeating the in vivo ubiquitination experiments described in Fig. with unhydroxylatable mutants (ΔAR6 and EVN→ATA) or coexpression of FIH. Steady-state levels of ASB4 and levels of autoubiquitination were not affected by these parameters (data not shown).
ASB4 functions to promote ES cell differentiation into the vascular lineage in an oxygen-dependent manner.
Since we were unable to identify any effects of ASB4 over- or underexpression in multiple cell lines (data not shown) and since other SOCS family proteins are known to regulate cell differentiation, we investigated the effects of ASB4 overexpression on ES cell differentiation into the vascular lineage and whether these effects can be regulated by oxygen concentration.
First, to determine the effects of ASB4 overexpression on this system, mouse ES cells were electroporated with linearized 3×Flag-tagged ASB4, the unhydroxylatable ASB4 ΔAR6 and EVN→ATA mutants, or the p3Xflag-CMV EV and placed in G418 to select for stable transfected clones. (We also attempted RNAi-mediated knockdown of ASB4 with six different short hairpin RNA sequences from integrated lentivirus constructs but were unable to attain greater than 40% silencing [data not shown], indicating that the ASB4 transcript may be inherently resistant to RNAi-mediated downregulation.) After 14 days, surviving colonies were picked and expanded and expression was confirmed by immunoblotting. Multiple clones were chosen for initial experiments to control for nonspecific effects due to random integration. Upon differentiation, ASB4-expressing clones, but not unhydroxylatable mutants, demonstrated a significant increase in Flk1+ cells at 96 h of differentiation compared to EV clones (Fig. ), indicating that ASB4 expression leads to an expansion of the hematovascular lineage and that this effect is dependent on the hydroxylated Asn 246 in AR6. To examine the causes of this Flk1+ cell increase and to investigate the ramifications of this increase for downstream lineage commitment, real-time RT-PCR analysis was performed with stage- and lineage-restricted genes. Since Flk1+ cells arise from mesoderm cells during differentiation, we investigated the overall levels of brachyury (T) expression as a marker for mesoderm. At day 4 of differentiation, brachyury levels were slightly, but significantly, increased in ASB4-expressing clones, suggesting increased mesodermal commitment (Fig. ). Since Flk1+ cells are progenitors for both hematopoietic and vascular cells in this system, we investigated the global expression levels of a variety of hematopoietic (Gata1, Scl/Tal) and vascular (Tie2, VE-cadherin) markers. All of these markers were tested at time points representing the peak of their expression in this system (day 6 for hematopoietic markers, day 8 for vascular makers). Interestingly, compared to EV clones, ASB4-expressing clones exhibited increased expression of vascular markers and decreased expression of hematopoietic markers, suggesting that enforced expression of ASB4 causes preferential commitment of stem cells to the vascular lineage (Fig. ). Together, these data suggest that ASB4 induces the formation of vascular precursors from mesoderm and promotes commitment to the vascular lineage.
FIG. 7. ASB4 overexpression in ES cells increases the percentage of Flk1+ cells and vascular commitment via an oxygen-dependent mechanism. ES cells were electroporated with constructs encoding cytomegalovirus promoter-driven 3×Flag-ASB4, unhydroxylatable (more ...)
On the basis of our hypothesis that ASB4 is likely to be regulated by oxygen concentration, we predicted that its effects on vascular lineage commitment are oxygen dependent. To test this hypothesis, we differentiated the stably transfected ES cells described above under hypoxic conditions (1% atmospheric oxygen). The kinetics of differentiation are similar to normoxic differentiation (data not shown), so the same time points were used for analysis of lineage-restricted markers. Interestingly, the differences in the commitment to the Flk1+ cell population, as well as the expression of hematopoietic and vascular lineage-specific markers observed with ASB4 overexpression under normoxic conditions, were completely abrogated by hypoxic treatment (Fig. ), indicating that the ability of ASB4 to promote the differentiation or maturation of the endothelial lineage is oxygen dependent and is inhibited by hypoxia.