HAX-1 is a multi-functional protein that is involved in the regulation of apoptosis, cell motility and calcium homeostasis. It is also reported to bind RNA: it associates with structural motifs present in the 3′ untranslated regions of at least two transcripts, but the functional significance of this binding remains unknown. Although HAX-1 has been detected in various cellular compartments, it is predominantly cytoplasmic. Our detailed localization studies of HAX-1 isoforms revealed partial nuclear localization, the extent of which depends on the protein isoform. Further studies demonstrated that HAX-1 is in fact a nucleocytoplasmic shuttling protein, dependent on the exportin 1 nuclear export receptor. Systematic mutagenesis allowed identification of the two nuclear export signals in the HAX-1 sequence. HAX-1 nuclear accumulation was observed after inhibition of nuclear export by leptomycin B, but also after specific cellular stress. The biological role of HAX-1 nuclear localization and shuttling remains to be established, but the HAX-1 transcript-binding properties suggest that it may be connected to mRNA processing and surveillance. In this study, HAX-1 status was shown to influence mRNA levels of DNA polymerase β, one of the HAX-1 mRNA targets, although this effect becomes pronounced only after specific stress is applied. Moreover, HAX-1 tethering to the reporter transcript caused a significant decrease in its expression. Additionally, the HAX-1 co-localization with P-body markers, reported here, implies a role in mRNA processing. These results suggest that HAX-1 may be involved in the regulation of expression of bound transcripts, possibly as part of the stress response.
cytoplasmic polyadenylation element-binding protein
chromosome region maintenance 1
mRNA-decapping enzyme 1A
human antigen R, ELAV-like protein 1
nuclear export signal
DNA polymerase β
protein PAT1 homolog 1
ATP dependent RNA helicase DDX6
HAX-1 was originally described as a factor interacting with hematopoietic cell-specific Src kinase substrate (HS-1). HS-1 is involved in B-cell receptor signaling, which includes its translocation to the nucleus upon phosphorylation . Subsequently, HAX-1 was shown to associate with a number of diverse proteins involved in several vital processes, such as apoptosis [2-5], cell migration [6, 7], regulation of calcium homeostasis [8, 9], interactions with viral proteins involved in cell survival [10-13] and ABC transporters . The number and scope of HAX-1 interactions suggests that its role in the cell must be significant, although probably complex . Its knockout in mice leads to postnatal lethality , while homozygous mutations in humans result in severe congenital neutropenia , a condition characterized by a paucity of neutrophils caused by maturation arrest in the bone marrow. Conversely, HAX-1 over-expression has been reported in several types of human cancer [7, 18, 19].
In addition to its other properties, which have hitherto attracted most attention, HAX-1 was also shown to interact with the 3′ UTRs of two transcripts: one encoding the cytoskeletal protein vimentin  and the second encoding an enzyme involved in base excision repair, DNA polymerase β (POLB) . In both cases, the interacting motif consisted of a stable hairpin, although there were no particular similarities in the structure. The transcript-binding properties of HAX-1 raise several questions regarding the role of these interactions in control of mRNA expression and the cellular compartment in which such binding takes place. HAX-1 has previously been described as localizing mainly in mitochondria [1, 4, 20], but also in the ER [9, 10], lamellipodia  and nucleus . One early report describes its localization in the nuclear membrane , but this observation has not been confirmed in later research. Data obtained by our group suggest that HAX-1 is instead associated with the nuclear matrix . Nuclear localization of HAX-1 was also demonstrated in systemic sclerosis fibroblasts  and breast cancer cells , which leads to the assumption that the observed accumulation may be disease-dependent.
The results presented here show that HAX-1 is in fact a nucleocytoplasmic shuttling protein. Nuclear localization was shown to vary depending on the isoform of the protein. The nuclear export of HAX-1 was shown to be dependent on exportin 1 (XPO1, a human homolog to yeast Crm1), and two nuclear export signals (NES) were identified in the HAX-1 sequence. HAX-1 nuclear accumulation was observed after specific cellular stresses, mainly after sodium arsenite treatment. Intriguingly, upon arsenite stress, HAX-1 status affected POLB mRNA levels, while in normal conditions this effect was not observed. These results and the results of the HAX-1 tethered assay suggest that the putative role of HAX-1 in post-transcriptional regulation of expression may consist of affecting mRNA levels of the bound transcripts, possibly as a response to specific cellular stress.
Different isoforms of human and rat HAX-1 protein display various degrees of nuclear localization
HAX1 is alternatively spliced, generating eight reported variants in humans [24, 25] and seven in rats . In the case of human transcripts, as variants I and V differ only by eight amino acids, and variants VI, VII and VIII contain a frameshift, our analysis was restricted to variants I–IV. Moreover, we demonstrated in a previous study  that only variants I and II display relatively high levels of expression, variants III and IV are expressed at low levels, and the expression of variant V is barely detectable, which further justifies the restriction of our analysis to variants I–IV. The human and rat splice variants used in this study are shown in Fig. S1.
To analyze the subcellular localization of protein isoforms corresponding to the various splice variants, HAX1 cDNA of human variants I–IV was fused to GFP and the localization of the isoforms was examined in the human HeLa cell line. Preliminary analysis revealed predominantly cytoplasmic localization for isoforms I, II and III, with sporadic localization to the nucleus, especially in the case of isoform III. In contrast, isoform IV (which lacks the C-terminal domain) was nucleocytoplasmic (Fig. 1A). The HAX-1 isoforms present in the cytoplasm were either diffuse or localized in unidentified granular structures.
Rat variants I, II, V and VII partially correspond to human variants I, II, III and IV, respectively. Rat variants III, IV and VI have no human counterparts . All of the rat variants were fused with GFP and the cellular localizations of the corresponding isoforms were analyzed in rat pituitary cell line GH3. Cytoplasmic localizations were detected for isoforms I, II and III, while isoforms VI and VII display partial nuclear localization (Fig. S2). Localization of isoforms IV and V was difficult to determine due to low expression of the GFP-fused protein.
HAX-1 nuclear export depends on XPO1
The observation that HAX-1 is present in both the cytoplasm and the nucleus led to the question of whether it is actively shuttled between these two compartments. Although no obvious nuclear localization signal is present in HAX-1, a consensus NES [Φ-(X)2–3-Φ-(X)2–3-Φ-X-Φ] (F = L,I,V,F,M, and X = any amino acid)  was identified at the C-terminus, at position 260-270. The presence of an NES suggests that HAX-1 is exported via the XPO1-dependent pathway. To investigate this possibility, the effect of leptomycin B (LMB), a specific inhibitor of the XPO1 export pathway, was analyzed. LMB treatment (50 nm, 4 h) caused nuclear retention of all studied isoforms (GFP fusions), except for the already partially nuclear isoform IV (Fig. 1A). Nuclear retention after LMB treatment was observed and calculated for an HAX-1–GFP fusion (isoform I), endogenous HAX-1 and FLAG-tagged HAX-1 (isoform I). Quantification was performed using ImageJ software (for detailed description see: Experimental procedures), and nuclear/cytoplasmic fluorescence ratios were calculated for LMB-treated and untreated cells, showing statistically significant differences in all cases (Fig. 1B).
Precise quantitative analysis performed using a BD Pathway (BD Biosciences, San Jose, CA, USA) bioimaging system enabled assessment of the distributions of the various HAX-1 isoforms and the effect of LMB treatment. Changes in nuclear content were estimated using the RING segmentation method (bd attovision software, BD Biosciences), which automatically calculates the ratio of nuclear staining to staining of the arbitrarily assigned cytoplasmic ring in every analyzed cell. This ratio is relative and does not represent the real nuclear/cytoplasmic ratio, indicating only the extent of nuclear translocation. However, the results obtained by this method are very precise. Isoform I displayed the most pronounced increase in nuclear presence after LMB treatment, followed by a markedly smaller increase for isoform II and a minimal increase for isoform III. The observed range of responses to leptomycin reflects statistical differences in the distribution of each of the isoforms in untreated cells: isoform II is almost entirely cytoplasmic while isoform III is more nuclear. In contrast to the other isoforms, isoform IV is present in the nucleus both before and after LMB treatment. Thus, as expected, quantitative assessment of the effect of LMB demonstrated the most dramatic nuclear retention for variant I and no additional retention for variant IV (Fig. 1C).
To corroborate the microscopic data, LMB-treated and untreated cells were fractionated, and cytoplasmic and nuclear fractions were analyzed for FLAG- and GFP-tagged HAX-1. FLAG-tagged protein was analyzed in order to eliminate non-specific effects of the large GFP tag. Western blots of the fractions confirmed the increase in HAX-1 nuclear content after LMB treatment (Fig. 1D).
The association of HAX-1 and XPO1 was confirmed by co-immunoprecipitation. HeLa cells were transiently co-transfected with a FLAG-tagged HAX-1 expression plasmid and a hemagglutinin-tagged XPO1 expression plasmid. The protein extract was immunoprecipitated using antibody against FLAG-tag, western blotted, and probed using antibodies against both expression tags (Fig. 1E).
Identification of nucleocytoplasmic export signals in HAX-1
Several HAX-1 constructs with mutations/deletions were generated in fusion with GFP, and were analyzed together with four GFP fusions of the natural isoforms to test their effect on nuclear export (Fig. 2A). Mutagenic analysis of HAX-1 was aimed at removal of hydrophobic regions with putative NES. The whole sequence was scanned using the NetNES 1.1 NES prediction server , and only one strong consensus NES was identified (Fig. S3). The predicted NES and the neighbouring hydrophobic amino acids were subjected to alanine-scanning mutagenesis in order to inactivate the potential NES. Although this region after mutagenesis completely lacked hydrophobic residues typical of NES (construct 8), mutated HAX-1 was still exported to the cytoplasm (Fig. 2B). This result indicates that efficient nuclear export occurs without the predicted NES, hence there must be another XPO1-binding region. A large deletion of 71 amino acids (Δ81–152, construct 4) did not affect HAX-1 nuclear localization, but its combination with mutation of the consensus NES (construct 9) showed strong nuclear localization, indicating the presence of a non-consensus NES sequence in the deleted region (Fig. 2B). Only two clusters of hydrophobic residues are present in the deleted region, and both were subjected to mutagenesis (Fig. 2A, constructs 10 and 11). The first cluster of hydrophobic amino acids (amino acids 79-88) was identified as the relevant non-consensus NES sequence. The identified NES sequences are shown in Fig. 2C. None of the other constructs, alone or in combination with the consensus NES mutant, displayed nuclear retention of HAX-1 (Fig. S4).
Cellular localization of the protein encoded by construct 10 (F81A, L84A, F88A and NES consensus destroyed) was analyzed by western blot in cytoplasmic and nuclear fractions, showing reversal of normal cellular distribution (as shown for the untreated cells in Fig. 1D), with the mutant protein present predominantly in the nucleus (Fig. 2D).
Specific stress induces HAX-1 nuclear accumulation
HAX-1–GFP fusions sporadically localize to unidentified granular structures in the cytoplasm, resembling some type of RNA granule. To determine whether these structures were stress granules, stress-inducing treatment was applied. HeLa cells were treated with sodium arsenite (SA), an oxidative stress inducer. Additionally, thapsigargin (TG), an ER stress-inducer, was used, as it inhibits the sarcoplasmic/endoplasmic reticulum calcium ATPase 2 (SERCA2) calcium pump, a known HAX-1-interacting protein. The mouse NIH 3T3 cell line was used for the thapsigargin treatment, as TG is not effective in HeLa cells. To determine whether stress granules were formed, and to assess the potential localization of HAX-1 in these structures, cells were co-transfected with constructs encoding the stress granule fusion protein TIA-1–YFP and HAX-1–RM (HAX-1 tagged with DsRed-Monomer). Co-localization studies showed that stress granules were visibly distinct from HAX-1–RM localizations in both experiments (Fig. 3A).
Unexpectedly, it was observed that sodium arsenite treatment, initially performed to verify potential HAX-1 localization in stress granules, caused nuclear accumulation of the HAX-1–RM fusion protein (Fig. 3A). This result was confirmed for HAX-1–GFP (Fig. 3B). In contrast, TG treatment did not cause nuclear accumulation of HAX-1 fusion proteins (Fig. 3A,B). Subsequently, different stress inducers were used to determine whether HAX-1 nuclear accumulation is specific to stress type. HAX-1–GFP-transfected cells were subjected to heat shock, vinblastine (microtubule assembly inhibitor), H2O2 and UV treatment (Fig. 3B and Table 1). Of all the tested factors, only sodium arsenite and severe heat shock (45 °C) resulted in statistically significant nuclear accumulation of HAX-1 (Fig. 3C). Other stress inducers did not affect HAX-1 cellular localization. HAX-1 accumulation in the nucleus after stress is relatively fast and visible after 10 min of heat shock (Fig. 3B) or arsenite treatment, although arsenite-induced accumulation is more pronounced after 30 min (Fig. 3D and Movie S1).
|Stress inducer||Dose||Nuclear accumulation|
|Sodium arsenite||0.5 mm, 30 min||+|
|Heat shock||42 °C, 30 min||+/−|
|45 °C, 10 min||+|
|H2O2||125 μm, 30 min||−|
|200 μm, 30 min||−|
|500 μm, 30 min||−|
|2.5 mm, 30 min||−|
|5 mm, 30 min||−|
|Thapsigargin (TG)||1 μm, 60 min||−|
|1 μm, 24 h||−|
|Vinblastine||500 nm, 60 min||−|
HAX-1 status affects POLB mRNA levels after arsenite stress
HAX-1 nuclear accumulation, coupled with its ability to bind specific mRNAs, suggests that HAX-1 mRNA binding may occur in the nucleus and act as an element of an mRNA surveillance system. To determine whether this is the case, two stable cell lines were generated: a cell line in which HAX-1 has been silenced and a silencing control (Fig. 4A). POLB mRNA is one of the reported HAX-1 binding transcripts. POLB mRNA levels were measured in both studied cell lines by quantitative PCR to examine the influence of HAX-1 status on POLB expression, but the differences were not statistically significant. However, HAX-1 nuclear accumulation after arsenite treatment suggested that it may represent an element of the cellular stress response, possibly involved in the regulation of expression of specific mRNAs. To assess the possibility that HAX-1 regulates POLB expression in response to arsenite stress, POLB mRNA levels were quantified in the two studied cell lines after arsenite treatment, and a moderate, but statistically significant, increase was observed for the HAX-1-silenced cell line (Fig. 4B).
HAX-1 tethering decreases reporter expression at the mRNA and protein level
HAX-1 mRNA-binding properties [20, 21] and its nucleocytoplasmic shuttling and its effect on POLB mRNA levels (this study) raise questions concerning the role of HAX-1 in post-transcriptional regulation of expression. To test the ability of HAX-1 to influence reporter expression, the tethered function assay with reporter luciferase mRNA was used. HAX-1 was fused with MS2 coat protein and tethered to the 3′ UTR of a reporter RNA via MS2 coat protein binding sites (Fig. 5A). Two reporter constructs were used: with 4 and 12 MS2 binding sites. In both cases, tethered HAX-1 caused an approximately twofold decrease in luciferase expression compared to the MS2 control. A similar decrease was also observed at the mRNA level (Fig. 5B).
HAX-1 co-localizes with P-body markers
HAX-1–GFP fusions were observed to accumulate in cytoplasmic granular structures, resembling RNA granules, but, as shown above, distinct from stress granules. The data presented here indicate HAX-1 involvement in mRNA processing, suggesting that these structures may represent P-bodies, cytoplasmic foci that harbor factors associated with mRNA degradation and repression of translation. To identify these structures, co-transfection of HAX-1–GFP with a construct expressing an RFP fusion of P-body marker, mRNA-decapping enzyme 1A (Dcp1a), was performed. Significant co-localization was observed in co-transfected cells (Fig. 6A,C) and confirmed by immunostaining for endogenous HAX-1 (Fig. 6B). Silencing using HAX-1-targeted miRNA revealed that P-bodies were still present in the cells subjected to HAX-1 knockdown (Fig. 6C), indicating that HAX-1 is not required for P-body formation.
Although HAX-1 protein is known to be involved in processes as crucial as apoptosis, cell migration and the regulation of calcium homeostasis, its mechanisms of function are largely unknown. Its interaction with the 3′ UTR of some mRNAs indicates a role in post-transcriptional regulation, but this function has been rather poorly studied to date. Here we report HAX-1 nuclear localization, its nucleocytoplasmic shuttling, co-localization with P-bodies and its impact on bound mRNA levels, which may help to elucidate its role in post-transcriptional regulation of expression.
In addition to many other cellular localizations, HAX-1 has been detected in the nucleus [15, 21]. In this study, we have established that the nuclear presence of HAX-1 is much more common than previously described and varies between isoforms; most of the human and rat isoforms localize to the cytoplasm, but some (human isoform IV and its rat counterpart isoform VII as well as rat isoform VI) are present in the nucleus. The importance and role of the primarily nuclear isoforms may be modest, as expression of isoforms other than isoform I is low , but, even for the cytoplasmic isoforms, nuclear localization may be detected in a portion of the analyzed cells. Furthermore, nuclear retention of human isoforms I, II and III was observed after LMB treatment, suggesting that HAX-1 is a nucleocytoplasmic factor and its nuclear export is mediated by XPO1. This assumption was confirmed by HAX-1–XPO1 co-immunoprecipitation.
Intriguingly, some GFP-fused HAX-1 isoforms localized in discrete foci in the nucleus, especially after LMB treatment. This observation suggests that HAX-1 may be present in specific nuclear bodies. Identification of these structures may help to unravel the specific nuclear role of HAX-1.
The findings presented may shed new light on the known inhibitory effect that HAX-1 exerts on the Rev protein of human immunodeficiency virus (HIV) . Rev facilitates XPO1 (Crm1)-dependent nuclear export of viral unspliced mRNAs containing the Rev response element. HAX-1 interacts with Rev and inhibits nuclear export of Rev response element RNAs . In the same study, HAX-1 was found to alter the intracellular localization of Rev from nuclear to cytoplasmic. These results were interpreted to indicate Rev trapping in the cytoplasm mediated by direct interaction with HAX-1. HAX-1 nucleocytoplasmic shuttling and its interaction with XPO1 suggests that the inhibitory effect of HAX-1 on Rev may involve more than a simple sequestration in the cytoplasm, and may result from interference with formation of the Rev–XPO1 exporting complex.
Quantitative analysis of the nuclear presence of HAX-1-GFP fusions displayed statistically important differences between isoforms before and after LMB treatment. The localization of isoform I was most affected by LMB, while localization of isoform IV remained unchanged. More subtle differences were detected for localization of isoforms II and III. These results indicate that the pool of shuttling HAX-1 consists mostly of isoform I.
Interestingly, HAX-1 nuclear translocation is never complete, and even for the most nuclear isoform IV and/or localization of the other isoforms after LMB treatment, there is always a vast proportion of the protein that remains in the cytoplasm.
In this study, two NES in the HAX-1 sequence were identified: a consensus NES in the C-terminal part of the protein, and a second non-consensus NES, containing a cluster of hydrophobic amino acids at positions 79-88. The double mutant in which both signals are mutated is predominantly nuclear.
Additionally, the N-terminal part of HAX-1 appears to have some influence on nuclear/cytoplasmic shuttling. The N-terminus contains a stretch of hybrophobic amino acids (3-13 amino acids) that resembles an NES and is absent from the frequently nuclear human isoform III and nuclear rat isoform VI.
There is no apparent nuclear localization signal within the HAX-1 sequence, which is not that surprising as the majority of the protein localizes to the cytoplasm. Two of the known HAX-1 protein partners, HS-1  and prohibitin , are cytoplasmic proteins that translocate to the nucleus upon lymphocyte receptor activation or in the presence of estrogen, respectively. This suggests that HAX-1 may enter the nucleus in complex with other proteins and in response to external (or internal) stimuli.
Accordingly, HAX-1 nuclear accumulation was observed in response to specific stress, particularly arsenite treatment and, to a lesser extent, heat shock, while H2O2, vinblastine, UV light and thapsigargin (an ER stress inducer) did not affect HAX-1 nuclear accumulation. It may be hypothesized that arsenite-induced oxidative stress inhibits XPO1 function , resulting in HAX-1 accumulation. However, strong oxidative stress induced by H2O2 treatment had no effect, indicating that arsenite causes more specific alterations in the cellular metabolism, to which HAX-1 responds.
HAX-1 nuclear accumulation after specific stress suggested its possible involvement in post-transcriptional regulation as part of the stress response. The POLB transcript is a known HAX-1 mRNA target. POLB expression was shown to be affected by arsenite treatment , and POLB enzyme was reported to participate in DNA repair after arsenite-induced damage . In this study, HAX-1 status was shown not to affect POLB mRNA levels under normal conditions, but a statistically significant increase in POLB mRNA levels was detected in the HAX-1-silenced cell line after arsenite stress. This result indicates that HAX-1 binding to the specific mRNA causes a decrease in the amount of this mRNA. As reported previously, the binding of HAX-1 to the instability-conferring element in the POLB mRNA  further supports these deductions. As HAX-1 binds to the 3′ UTR of the transcripts and lacks a DNA-binding domain, its effect on mRNA is most likely post-transcriptional. The effect of HAX-1 on POLB mRNA appears to be modest, but one should bear in mind that POLB expression is tightly regulated and any alteration may be detrimental. As POLB over-expression leads to chromosomal instability and an increased mutagenic rate [33, 34], the existence of mechanisms to combat this effect seems probable. Accordingly, HAX-1-mediated down-regulation may be an element of post-transcriptional fine-tuning of the stress response.
The results of HAX-1 tethering to the reporter mRNA corroborate the above conclusions. Again, any regulation observed in this reporter system is post-transcriptional. A decrease in the luciferase reporter was observed at the protein and mRNA levels, suggesting that HAX-1 affects mRNA stability. Intriguingly, HAX-1 was recently shown to interact with Pelota protein , a human homolog of yeast Dom34, which is involved in the ‘No-Go’ mRNA decay pathway. No-Go recognizes translation elongation stalling and targets the mRNA for endonucleolytic cleavage. A role of HAX-1 in this process remains to be established; it may be hypothesized that HAX-1 recruits Pelota to specific transcripts or mediates mRNA 5′ degradation after endonucleolytic cleavage, which may be linked to its co-localization with 5′-decay factors in P-bodies. However, at this point, no mechanism can be proposed to explain these findings, and further studies are required to clarify the role of HAX-1 in mRNA decay.
HAX-1 co-localization with the P-body marker further strengthens the possibility of its involvement in mRNA processing. Although their role remains controversial, P-bodies are thought to be involved in post-transcriptional activities such as decapping, degradation, miRNA silencing and storage of mRNA. Although no interaction with any P-body protein has been detected so far for HAX-1, its co-localization with these structures suggests involvement in mRNA processing, consistent with the data showing its influence on POLB and reporter mRNAs.
P-bodies are dynamically linked to and partially share protein content with stress granules , but, despite the co-localization with P-bodies, HAX-1 was found not to be present in stress granules. This result indicates that it has no specific role in storage and protection of translationally active mRNAs during stress.
The results presented above raise several questions concerning HAX-1 biology and its functions other than functions directly related to apoptosis, cell migration or calcium homeostasis. The functional significance of the nuclear presence of HAX-1 and its nucleocytoplasmic shuttling remains to be established. The HAX-1 transcript-binding properties suggest that its nuclear role may be associated with processing of specific mRNAs. The results obtained here suggest that this mRNA processing may consist of regulation of the stability of specific transcripts, possibly as an element of fine-tuning of the cellular stress response. This conclusion is supported by the presence of HAX-1 in P-bodies. There are several examples of nucleocytoplasmic shuttling proteins that are present in P-bodies and that play a role in post-transcriptional regulation: HuR , rck/p54 (DDX6) , DDX3 , the Rpb4p subunit of RNA polymerase II , CPEB  and Pat1b . HAX-1 may represent another example, and further studies are required to clarify its role in mRNA processing.
Plasmids and constructs
Constructs containing human HAX-1–GFP fusions were generated by cloning cDNA of the four HAX1 variants in-frame into the pEGFP–N1 vector (Clontech Laboratories, Palo Alto, CA, USA). cDNA was prepared from healthy human lung tissue or HeLa cells. Coding sequences of variants I–III were obtained using variant-specific primers: a forward primer containing an EcoRI restriction site and a reverse primer containing a BamHI restriction site (variants I and II: 5′-GGAATTCGGGAATGAGCCTCTTTG-3′ and 5′-CGGGATCCCGCCGGGACCGGAAC-3′; variant III: 5′GGAATTCAGGGATGACTCGAGAT-3′ and 5′-CGGGATCCCGCCGGGACCGGAAC-3′).PCR products were cloned directly into EcoRI/BamHI sites of pEGFP-N1. Variant IV was amplified using PCR with variant-specific primers 5′-CGGGAATGAGCCTCTTTGAT-3′ and 5′-CGGGATCCCGTTTCACCCACCAACCC-3′; the reverse primer contained a BamHI restriction site. The PCR product was cloned into the pGEM-T Easy vector (Promega, Fitchburg, WI, USA), and sub-cloned into the EcoRI site of the pEGFP-N1 vector. The resulting plasmid was digested with BamHI and re-ligated to fuse HAX-1 and GFP coding sequences in-frame.
A fusion of a red fluorochrome with human HAX1 (variant I) was obtained by cloning the HAX1 ORF obtained as above into pDsRed-Monomer-C1 (Clontech), using EcoRI and BamHI restriction sites.
Rat Hax-1–GFP fusions of variants I, II and III were generated in two steps: (a) sub-cloning the sequence encoding the C-terminal part of Hax-1 from pET201-Hax-1  into pUC19-Hax-1 var. I/II/III  using XbaI and HindIII restriction sites, generating Hax-1 coding sequences lacking a stop codon, and (b) sub-cloning the Hax1 var. I, II and III thus obtained into pEGFP-N1 (Promega), creating GFP fusions using BglII and HindIII restriction sites. Hax1 variants IV, V and VI were generated by PCR using rat liver cDNA, two forward primers (variants V and VI: 5′-CGTGGTTTATCTGCCTCCAT-3′; variant IV: 5′-TCTCCAGACTGGGGTGATGC-3′) and a reverse primer (5′-GGATCCTCGGGACCGAAACCAAC-3′). PCR products were cloned into pCR2.1-TOPO (Invitrogen, Life Technologies, Carlsbad, CA, USA), and then sub-cloned into pEGFP-N1 using the BamHI restriction site.
pCR3-FL2-HAX1, bearing FLAG-tagged HAX-1, with two consecutive copies of the FLAG epitope preceding the human HAX1 ORF (variant I), was constructed on the basis of the pCR3 plasmid (Invitrogen), which contains a synthetic linker bearing the appropriate restriction sites (a kind gift from Natarajan Mohen, University of Texas, TX, USA). HAX1 cDNA was obtained using the EcoRI site-containing forward primer as above and a reverse primer containing an EcoRV site (5′-CGGATATCAGGCTACCGGGACC-3′), cloned into the respective sites of pcDNA3.1(+) (Invitrogen), and sub-cloned into modified pCR3 using the BamHI and XbaI restriction sites.
Deletions and mutagenesis
Deletions in the HAX1 sequence were generated by PCR amplification of the whole pEGFP-N1-HAX-1 (variant I) plasmid sequence except for the parts encoding chosen regions to be deleted, using primers 5′-GAGAGTGATGCAAGAAGTGAATC-3′ and 5′-GCCGAAGTTATCGTGGAAACG-3′, and 5′-AGCAGTCCTAGGGGTGATCC-3′ and 5′-TTTGGGCTGGGGCTGTAGAAC-3′ for Δ81–152 and Δ206–246, respectively, and subsequent blunt-end ligation of the product. PfuPlus! high-fidelity polymerase (EURx, Gdańsk, Polska) was used, and PCR conditions were as follows: 94 °C for 5 min, followed by 30 cycles of 94 °C for 30 s, 60 °C for 30 s and 72 °C for 6 min, then 72 °C for 7 min.
Mutagenesis of the hydrophobic amino acids of the putative nuclear export sequence (L261, F265, L268, L270, L272 and F276) was performed using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA), according to the manufacturer's instructions. Mutagenesis of the consensus NES was performed in three rounds using primers (lower-case letters indicate mutation): 5′-TCCAGCCgcGGATGATGCCgcTTCCAT-3′ and 5′-ATGGAAgcGGCATCATCCgcGGCTGGA-3′ (round 1), 5′-TGCCGCTTCCATCgcGGACgcATTCCT-3′ and 5′-AGGAATgcGTCCgcGATGGAAGCGGCA-3′ (round 2), and 5′-CGCATTCgcGGGACGTTGGgcCCGGTC-3′ and 5′-GACCGGgcCCAACGTCCCgcGAATGCG-3′ (round 3). Mutagenesis of the two leucines (L196 and L200) was performed using primers 5′-GGAGGGTgcTGGCCCGGTTgcACAGCC-3′ and 5′-GGCTGTgcAACCGGGCCAgcACCCTCC-3′. Mutagenesis of F141, V144 and L145 was performed using primers 5′-CAGGATCgcTGGGGGGGcCgcGGAGAG-3′ and 5′-CTCTCCgcGgCCCCCCCAgcGATCCTG-3′. Mutagenesis of F81, L84 and F88 was performed using primers 5′-CTTCGGCgcTGATGACgcAGTACGAGATgcCAATAG-3′ and 5′-CTATTGgcATCTCGTACTgcGTCATCAgcGCCGAAG-3′.
Tethered assay constructs
pSL-MS2-12X and pMS2-GFP MS2 plasmids were obtained from Robert Singer (Albert Einstein College of Medicine, New York, NY, USA). MS2 hairpin elements were cloned from plasmid pSL-MS2-12X into the pCMLuc vector containing firefly luciferase cDNA, obtained from Dominique Weil (André Lwoff Institute, Paris, France) into a BamHI restriction site. pCMLuc-MS2-4X was obtained by incomplete KpnI digestion of pCMLuc-MS2-12X and self-ligation. MS2 coat protein was cloned in-frame with HAX-1 protein in pcDNA3.1(+). MS2 coat sequence was amplified by PCR from pMS2-GFP using primers 5′-GGATCCGCTAGCCGTTAAAATGGCTTC-3′ and 5′-GAATTCTCGCGTAGATGCCGGAGTTTG-3′, and cloned in-frame with HAX-1 into the pcDNA3.1-HAX-1 construct, using BamHI/SpeI/Klenow fragment. The same PCR product was used to generate pcDNA3.1-MS2 control plasmid.
Plasmids expressing markers of stress granules (TIA-1) and P-bodies (Dcp1a) were obtained from Paul Anderson (Harvard Medical School, Boston, MA, USA) (pEYFP-TIA-1, mRFP-Dcp1a). The hemagglutinin-tagged XPO1-encoding pXPO1-HA vector was obtained from Hisatoshi Shida (Institute for Genetic Medicine, Hokkaido University, Japan).
Cell culture and transfections
The rat pituitary clonal cell line GH3, the mouse embryonic fibroblast cell line NIH 3T3 and the HeLa human cervical carcinoma cell line were used. Cells were grown on Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen). Transfections were performed using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions.
Fluorescence microscopy, live-cell imaging, high-content imaging and data analysis
HeLa cells were grown on Lab-Tek standard cover slips (Nunc, Roskilde, Denmark) and transfected with plasmids expressing HAX-1–GFP fusions as indicated. Twenty-four hours after transfection, cells were fixed in 4% formaldehyde, washed with PEM buffer (80 mm PIPES, 5 mm EGTA, 2 mm MgCl2), quenched with 0.1 m ammonium chloride and permeabilized using 0.5% Triton X-100 in PEM buffer. Immunostaining was performed using antibody against HAX-1 (1 : 100; BD Biosciences). Samples were blocked using 5% low-fat milk solution (Tris-buffered saline containing 5% low-fat milk and 0.1% Tween-20) for 30 min, incubated with primary antibody at 4 °C overnight, washed (5 min each wash) five times with blocking buffer, incubated for 30 min with secondary antibody (goat anti-mouse IgG(whole molecule)-FITC (Fluorescein Isothiocyanate), Sigma-Aldrich, St. Louis, MO, USA), washed (5 min) five times with Tris-buffer saline with 0.1% Tween and once with PEM buffer. For the specificity control, only secondary antibody was used. Cells were stained with DAPI (4′,6-Diamidino-2-Phenylindole, Sigma-Aldrich), mounted and observed using standardized settings on a Nikon Eclipse E-800 microscope with a 100 × oil immersion objective (Plan Fluor; Nikon, Tokyo, Japan). Deconvolution of images was performed using ImageJ image-processing software , which was also used for quantitative analysis of LMB-induced retention and stress-related accumulation. Images were split into RGB channels, auto-threshold was applied for the blue channel representing DAPI staining, and the appropriate mask was created and used to isolate nuclear staining in the green channel. Cytoplasmic and nuclear staining intensities were measured for every image, and the nuclear/cytoplasmic ratio was calculated.
Alternatively, cells were observed using a Leica (Leica Microsystems GmbH, Wetzlar, Germany) TCS SP2 confocal laser scanning inverted microscope with a Plan Apo 63 × 1.4 NA oil-immersion objective. An argon laser (488 nm) was used for GFP and FITC excitation, and fluorescence was detected within the range 500–535 nm. Images were acquired at 0.5 μm Z-steps, and the maximum projection of the images is shown.
Rat GH3 cells were grown and fixed as above and imaged using Deltavision high-resolution deconvolution (image restoration) microscopy system (Applied Precision, Issaquah, WA, USA) with a Zeiss 68 Axiovert S100 microscope (Carl Zeiss, Thornwood, NY, USA). Images were taken at 0.15 μm Z-steps using a Plan Apo 63 × 1.4 NA oil-immersion objective.
Live cell imaging
HeLa cells were grown on Lab-Tek chambered coverglass dishes (Nunc) and transfected as indicated. Cells were observed on a Leica TCS SP5 confocal laser scanning inverted microscope within the environmental chamber, allowing maintenance of optimal temperature (37 °C) and CO2 (5%) conditions. A Plan Apo 63 × 1.4 NA oil-immersion objective was used. GFP fluorescence was monitored using 488 nm excitation with an argon laser, and 500–550 nm emission.
Quantitative analysis was performed using a BD Pathway 855 bioimager (BD Biosciences) equipped with UApo20×/340 (Universal Apochromat, numerical aperture 0.75) and UApo40×3/340 (Universal Apochromat, numerical aperture 0.90) objectives (Olympus, Tokyo, Japan). Cells were grown on Lab-Tek standard cover slips, transfected and fixed as described above. DAPI was excited at 380/10 nm, and its emission fluorescence was directed through a 400 nm dichroic long-pass and a 435 nm long-pass filters. GFP was excited at 488/10 nm, and the emission beam was passed through Fura/FITC and 515 nm long-pass filters. The images were recorded using an Orca digital camera (Hamamatsu Herrsching, Germany). All imaging data were collected and analyzed using the bd attovision 1.6 software package (BD Biosciences). Each sample was visualized at least 10-12 times. Nuclear regions were identified by DAPI staining, and the cytoplasm was defined as a narrow ring (6 μm) around the previously marked nuclei. Accumulation of GFP-tagged protein was monitored separately in the nuclear and cytoplasmic areas, and presented as the nuclear/cytoplasmic ratio of GFP signal intensities (using the Ring [2 Outputs]Band segmentation method, attovision). The standard error of the mean (SEM) was calculated, and a t test was performed to assess the statistical significance of the results.
Cytoplasmic and nuclear fractions were isolated from 106 transfected HeLa cells using the PARIS (protein and RNA isolation system) kit (Ambion-Life Technologies, Carlsbad, CA, USA), according to the manufacturer's instructions. Briefly, cells were washed with NaCl/Pi and resuspended in cold Cell Fractionation Buffer (PARIS Kit). Then cells were incubated on ice (10 min) and centrifuged (5 min, 500 g, 4 °C), and the cytoplasmic fraction was collected. The nuclear fraction was lysed using Cell Disruption buffer (PARIS Kit) and sonicated. Protein concentrations were determined by the Bradford assay (Bio-Rad, Hercules, CA, USA). Fractions were analyzed by western blot using horseradish peroxidase-conjugated antibodies against FLAG-tag (Sigma) and GFP (Abcam). Fractionation was confirmed using antibodies against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Sigma) and histone H3 (Abcam, Cambridge, UK).
HeLa cells were transiently co-transfected with pCR3-FL2-HAX1 and pXPO1-HA plasmids. The pcDNA3.1-HAX1 and pXPO1-HA pair of constructs was used as a control. Cells were harvested for immunoprecipitation 24 h following transfection, lysed in RIPA buffer (50 mm Tris/HCl pH 8.0, 150 mm NaCl, 1 mm EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mm phenylmethanesulfonyl fluoride) supplemented with Complete protease inhibitor cocktail (Roche, Basel, Switzerland). After pre-incubation with Protein A+G-coated agarose beads (Dynabeads; Invitrogen), equal amounts of protein (200 μg) from cell lysates were immunoprecipitated using 40 μl EZview Red ANTI-FLAG M2 Affinity Gel (Sigma). The immune complexes were heat-denatured in loading buffer (50 mm Tris-HCl, 0.01% bromophenol blue, 1.75% 2-mercaptoethanol, 11% glycerol, 2% SDS), and analyzed for the presence of tagged XPO-1 by western blot. After separation by 8% SDS/PAGE, the complexes were transferred onto Hybond-C Extra membrane (Amersham, General Electric Company, Fairfield, CT, USA). Blots were blocked for 45 min using 5% low-fat milk solution, incubated overnight at 4 °C in blocking solution containing mouse monoclonal primary antibody against hemagglutinin (Abcam), and detected using a horseradish peroxidase-conjugated goat anti-mouse secondary antibody (Pierce-Thermo Scientific, Waltham, MA, USA).
HAX-1 was silenced in HeLa cells using the BLOCK-iT PolII miR RNAi expression vector system (Invitrogen). The miRNA sequence (5′-CCAAATCCTATTTCAAGAGCA-3′) was chosen to target all four HAX1 variants. Double-stranded oligonucleotide encoding pre-miRNA was prepared and cloned into the pcDNA6.2-GW/EmGFP-miR vector (Invitrogen) according to the manufacturer's instructions. HeLa cells were transfected with the silencing plasmid using Lipofectamine 2000 (Invitrogen). The pcDNA6.2-GW/EmGFP-miR-neg plasmid (Invitrogen) was used as a negative control. To confirm silencing, cells were harvested 24 and 48 h after transfection, lysed in RIPA buffer, and 65 μg of total protein extracts were used for the western blot. Protein extracts were separated by 12% SDS/PAGE and transferred onto Hybond-C Extra membrane (Amersham). Membranes were blocked using 5% low-fat milk solution, and detected using a mouse monoclonal primary antibody against HAX-1 (BD Biosciences) and a horseradish peroxidase-conjugated goat anti-mouse secondary antibody (Pierce). To confirm equal loading of proteins, blots were re-probed using a rabbit polyclonal primary antibody against GAPDH (Sigma) and a horseradish peroxidase –conjugated goat polyclonal secondary anti-rabbit antibody (Abcam).
Generation of stable cell lines
HeLa cells were transfected as above. Twenty-four hours after transfection, cells were detached and placed in 24-well plates for selection of clones growing in the presence of 7.5 μg·mL−1 blasticidin. After ∼ 2 weeks, cells derived from single colonies were tested by western blot of HAX-1 expression. Two stable cell lines were established, siHAX-1, with the highest degree of HAX-1 silencing, and a control line, with no change in HAX-1 level. To confirm silencing, cells from both lines were harvested, lysed in RIPA buffer and used for western blotting, performed as described above.
Quantification of POLB mRNA
Stable cell lines with silenced HAX-1 and the silencing control were plated in four repeats in six-well plates and left untreated or subjected to sodium arsenite treatment (0.5 mm, 3 h). Cells were harvested and used for total RNA preparation (PureLink RNA mini kit; Invitrogen), treated with recombinant DNase I (Roche), and 2 μg of the obtained RNA was used for cDNA synthesis using Superscript III (Invitrogen). cDNA was quantified by quantitative PCR on an ABI Prism 7500 real-time PCR system using Power SYBR Green PCR Master Mix (Applied Biosystems, Life Technologies, Carlsbad, CA, USA) and primers amplifying 158 bp of the POLB transcript (forward 5′-AGATT CGGCAGGATGATACGAG-3′; reverse 5′-CCCAA TTCGCTGATGATGGTTC-3′) and, as a reference, 226 bp of the GAPDH transcript (forward 5′-GAAGGTGAAGGTCGGAGTC-3′; reverse 5′-GAAGATG GTGATGGGATTTC-3′). PCR was performed using the following conditions: pre-cycling hold at 95 °C for 10 min, followed by up to 40 cycles of 95 °C for 30 s and 60 °C for 60 s. The ΔΔCT method was used for calculating mRNA expression levels.
The MS2–HAX-1 fusion (pcDNA3.1-MS2-HAX-1) was tethered to MS2 hairpins cloned downstream of the firefly luciferase reporter gene (pcMLucMS2 4X and 12X). An MS2 coat construct was used as a control (pcDNA3.1-MS2). Unmodified Renilla luciferase was used as an internal control (phRL-CMV; Promega). HeLa cells were plated in six-well plates (three repeats), and co-transfected with: (1) reporter plasmid containing MS2 hairpins, (2) MS2-HAX-1 fusion or MS2 control plasmid and (3) Renilla-encoding plasmid. Cells were harvested, and one-twentieth of the suspension was used for the microplate luciferase assay (four repeats) (Dual-Glo; Promega). The rest of the material was used for total RNA preparation (PureLink RNA mini kit; Invitrogen), treated with recombinant DNase I (Roche), and used for cDNA synthesis with Superscript III (Invitrogen). cDNA was quantified by quantitative PCR (ABI Prism 7500) using Power SYBR Green PCR Master Mix (Applied Biosystems) and primers amplifying 209 bp of the firefly luciferase transcript (forward 5′-TCGTTGACCGCCTGAAGTCT-3′; reverse 5′-GGCGACGTAATCCACGATCT-3′), and, as a reference, 232 bp of the Renilla luciferase transcript (forward 5′-TGGAGCCATTCAAGGAGAAG-3′; reverse 5′-TTCACGAACTCGGTGTTAGG-3′). PCR was performed using the following conditions: a pre-cycling hold at 95 °C for 10 min, followed by up to 40 cycles at 95 °C for 30 s, 56 °C at 30 s and 72 °C at 30 s. The ΔΔCT method was used for quantity calculations.
We thank Antek Laczkowski for assistance with statistical analysis, Paul Anderson (Harvard Medical School, Boston, MA, USA), Dominique Weil (Andre Lwoff Institute, Paris, France), Robert Singer (Albert Einstein College of Medicine, New York, NY, USA) and Hisatoshi Shida (Hokkaido University, Japan) for the constructs, Anna Balcerak, Lukasz Szafron and Michael Mancini (Baylor College of Medicine, Houston, TX, USA) for technical support, and Ania Wilczynska (University of Leicester, UK) for helpful comments. This work was supported by the Polish National Science Centre (grant numbers N301 317439 and 2011/01/B/NZ1/03674).