Implication of KRT16, FAM129A and HKDC1 genes as ATF4 regulated components of the integrated stress response

The ATF4 transcription factor is a key regulator of the adaptive integrated stress response (ISR) induced by various stresses and pathologies. Identification of novel transcription targets of ATF4 during ISR would contribute to the understanding of adaptive networks and help to identify novel therapeutic targets. We were previously searching for genes that display an inverse regulation mode by the transcription factors ATF4 and p53 in response to mitochondrial respiration chain complex III inhibition. Among the selected candidates the human genes for cytokeratine 16 (KRT16), anti-apoptotic protein Niban (FAM129A) and hexokinase HKDC1 have been found highly responsive to ATF4 overexpression. Here we explored potential roles of the induction of KRT16, FAM129A and HKDC1 genes in ISR. As verified by RT-qPCR, a dysfunction of mitochondrial respiration chain and ER stress resulted in a partially ATF4-dependent stimulation of KRT16, FAM129A and HKDC1 expression in the HCT116 colon carcinoma cell line. ISRIB, a specific inhibitor of ISR, was able to downregulate the ER stress-induced levels of KRT16, FAM129A and HKDC1 transcripts. An inhibition of ATF4 by RNAi attenuated the induction of KRT16, FAM129A and HKDC1 mRNAs in response to ER stress or to a dysfunctional mitochondrial respiration. The similar induction of the three genes was observed in another tumor-derived cervical carcinoma cell line HeLa. However, in HaCaT and HEK293T cells that display transformed phenotypes, but do not originate from patient-derived tumors, the ER stress-inducing treatments resulted in an upregulation of FAM129A and HKDC1, but not KRT16 transcripts, By a luciferase reporter approach we identified a highly active ATF4-responsive element within the upstream region of the KRT16 gene. The results suggest a conditional regulation of KRT16 gene by ATF4 that may be inhibited in normal cells, but engaged during cancer progression. Potential roles of KRT16, FAM129A and HKDC1 genes upregulation in adaptive stress responses and pathologies are discussed.


Introduction
One of the major adaptive signaling pathways of eukaryotic cells acts through a regulation of eukaryotic translation initiation factor 2 alpha (eIF2α). Various stressful signals converge to at least four protein kinases that phosphorylate eIF2 α leading to a global reduction in translation initiation (reviewed in [1]). Besides, the phosphorylation of eIF2 α is associated with switching to translation of selected transcripts that contain upstream open reading frames (uORF) [1,2]. One of such preferentially translated transcripts encodes Activating Transcription Factor 4 (ATF4) that participates in the reprogramming gene expression to endure stresses [3]. Because eIF2 α phosphorylation can direct translational control in response to diverse environmental stresses, the eIF2 α /ATF4 pathway is often referred to as the integrated stress response (ISR) [3].
ATF4 is basic leucine zipper transcription factor that binds to the C/EBP-ATF response elements (CARE) composed of a half-site for the C/EBP and a half-site for a member of the ATF family transcription factors [4]. Products of the ATF4-induced genes modulate a wide spectrum of cellular processes to endure diverse stresses. Activities of the ATF4 affect amino acid biosynthesis and transport [2,3], upregulate protein synthesis [5] and alleviate outcomes of oxidative stress [2]. The ATF4-mediated ISR draw particular attention for its role in many pathologies, including cardiac [6], neurodegenerative (Alzheimer [7], Parkinson [8]), pulmonary [9], renal [10], and several other. Various functions of ATF4 promote metabolic homeostasis and cancer cell survival in hypoxic-and nutrient-deprived regions, in particular by affecting amino acid uptake and biosynthesis, autophagy, redox balance and angiogenesis [11]. Because the ATF4 expression is increased in many tumors the ISR signaling pathway is being considered as an attractive target for anti-cancer therapy [11]. An upregulation of the ATF4 protein in response to different stress conditions usually occurs at both translation and transcription levels and is associated with the induction of ATF4 mRNA [12]. Further studies on identification of novel transcription targets of ATF4 could contribute to the development of new therapeutic strategies.
Here we describe the identification of KRT16, FAM129A and HKDC1 genes as novel targets for the transcription regulation by ATF4. KRT16 gene encodes a cytokeratin protein associated with hyperproliferation of keratinocytes [13]. HKDC1 is a human hexokinase gene that is overexpressed in tumor tissues and is considered to be a promising target for a treatment of lung cancer [14]. FAM129A gene encodes a Niban protein that is overexpressed in many types of cancer and supposedly protects cells from apoptosis [15,16]. We show that the ATF4-mediated induction of KRT16, FAM129A and HKDC1 mRNAs occurs in response to endoplasmic reticulum (ER) stress and after an inhibition of mitochondrial respiration chain. We demonstrate some peculiarities in the responses that are potentially related to the process of selection of malignant cells and cancer progression.

Identification of new potential ATF4-regulated genes
Recently we have described an algorithm for searching novel ATF4-regulated genes induced in response to an inhibition of mitochondrial electron transfer chain (ETC) [17]. The algorithm is based on the experimental evidence that in the absence of the p53 response a dysfunction of mitochondrial respiration leads to increased expression of the ATF4 gene and of some known ATF4-regulated genes, while a simultaneous activation of p53 suppresses the ATF4 mediated transcription [18]. As the result, a transient inhibition of respiratory complex III is associated with an upregulation of ATF4 and ISR, though a more sustained dysfunction results in an activation of the p53 response due to the impaired pyrimidine biosynthesis by the ETCcoupled dihydroorotate dehydrogenase and prevents ATF4 upregulation [18][19][20]. The p53 response can be prevented by uridine supplementation leading to an uninterrupted activation of the ATF4-dependent ISR during the sustained inhibition of complex III [18]. A set of transcripts that are most heavily upregulated by the complex III inhibition in the absence of p53 activation were selected on the basis of transcriptome sequencing as products of putative ATF4 stimulated genes and the effect was verified by RT-qPCR in samples overexpressing ATF4 [17]. In the previous study we have revealed transcripts from ADM2, KRT16, FAM129A and HKDC1 genes as most responsive to the ATF4 overexpression [17]. In ATF4 overexpressing HeLa cells levels of ADM2, FAM129A, KRT16 and HKDC1 mRNAs were increased 11, 9.5, 5.2 and 3.4 fold, respectively. The role of ADM2 has been previously reported [17]. In the present study we address a significance of KRT16, FAM129A and HKDC1 genes in ISR and discuss their potential roles in pathologies.

Induction of mRNA from KRT16, FAM129A and HKDC1 genes in response to different stresses
The induction of KRT16, FAM129A and HKDC1 transcripts in HCT116 colon carcinoma cells was tested by RT-qPCR after the treatment with mitochondrial respiratory chain complex I inhibitor Piericidin A or complex III inhibitor Myxothiazol under the previously optimized conditions that result in the induction of ATF4 but do not activate p53 [18]. We observed a substantial upregulation of mRNAs from the genes (Fig 1A-1F) indicating that the inhibition of mitochondrial respiratory chain induces a coordinated stimulation of both ATF4 and transcripts from the KRT16, FAM129A and HKDC1 genes.
Accumulating evidences suggest that ATF4 may play a leading role in ER stress. The unfolded protein response (UPR), which is characteristic to ER stress, is induced by denatured proteins within the ER and initiates generalized signaling cascades [21,22]. The UPR fires signaling processes that include PERK/eIF2α/ATF4 pathway [21]. During the ER stress GRP78, the master regulator of UPR, binds to unfolded proteins and releases UPR mediator, the PKRlike endoplasmic reticulum kinase (PERK), which forms dimers and phosphorylates eIF2α. The phosphorylated eIF2α activates ATF4 to facilitate transcription of ATF4-responsive genes [22]. To find whether KRT16, FAM129A and HKDC1 genes can also respond to ER stress we tested levels of corresponding transcripts after inducing UPR by Tunicamycin and Brefeldin A, the treatments that were previously found optimal for the induction of ATF4 transcripts [17]. A substantial increase in KRT16, FAM129A and HKDC1 transcripts in HCT116 cells was observed after the treatment (Fig 2).
The results were in agreement with the suggestion that the genes under the study may indeed be regulated by ATF4 and therefore belong to ISR. To test this possibility we treated the cells with ISRIB, a specific small molecule inhibitor of integrated stress response [23]. The ISRIB was previously shown to reverse the effects of eIF2α phosphorylation on translation; in particular, ISRIB recovers the total protein biosynthesis under stress conditions, while suppressing translation of some uORF-containing transcripts, including the ATF4 mRNA [23][24][25].
We found that in HCT116 cells ISRIB can substantially inhibit the induction of KRT16, FAM129A and HKDC1 mRNAs by the ER stress inducer Brefeldin A (Fig 3). The results were reproduced in HeLa cells (S1 Fig). The data is in favor of a major role of the integrated stress response in the induction of KRT16, FAM129A and HKDC1 genes during ER stress, presumably mediated by ATF4.

Suppression of ATF4 mRNA by RNAi affects the induction of KRT16, FAM129A and HKDC1 transcripts
We inhibited transcripts from the ATF4 gene by stable transduction of a lentiviral construct expressing specific shRNA into HCT116 cells [18]. Levels of mRNAs from ATF4, KRT16, FAM129A and HKDC1 genes were measured by real-time RT-PCR. There was a 2-fold suppression of the basal ATF4 mRNA level and a 3-fold suppression of the ATF4 mRNA induced by complex I inhibitor Piericidin A ( Fig 4A). The suppression of ATF4 did not affect basal levels, but partially suppressed the Piericidin A induced activation of FAM129A, KRT16 and HKDC1 genes' transcription ( Fig 4B-4D). This effect was more pronounced for FAM129A, but even though the reduction was more modest for KRT16 and HKDC1, it was still statistically significant. The results indicate that ATF4 is involved in the upregulation of FAM129A, KRT16 and HKDC1 genes during the inhibition of the mitochondrial respiration chain.
There was also observed a partial suppression of FAM129A, KRT16 and HKDC1 mRNA induction in response to ER stress inducer Brefeldin A ( Fig 5) further confirming the contribution of ATF4 in transcription regulation of the above genes during ER stress.

An ectopic expression of ATF4 and ER stress affect protein levels of FAM129A/Niban protein
We overexpressed ATF4 protein in HCT116 cells from the introduced plasmid construct and measured protein levels by Western analysis. In parallel, we treated HCT116 cells with Brifeldin A for 16 hours and analyzed the protein extracts (Fig 6). There was a strong increase in ATF4 protein in the treated samples, as compared with the control. Antibodies to FAM129A/ Niban protein also detected the increased expression of the protein in samples prepared from HCT116 cells overexpressing ATF4 or treated with Brifeldin A (Fig 6), as well as in HeLa cells treated with Brifeldin A and Tunicamycin (S2 Fig). The results indicate that the ATF4-medited upregulation of FAM129A/Niban also occurs at the protein level.

ER stress induces the ATF4-mediated transcriptional upregulation of FAM129A and HKDC1 genes but does not induce KRT16 gene in transformed cell lines that were not derived from tumors
The experiments described above were performed in HCT116 colorectal carcinoma cells. In the cervical carcinoma HeLa cells KRT16, FAM129A and HKDC1 transcripts were also induced in response to treatments associated with ATF4 upregulation (S3 Fig) and in response to ectopic overexpression of ATF4 [17]. To expand the observations we tested two additional cell lines. HaCaT cells, spontaneously transformed human skin keratinocyte cell line [26] was chosen as supposedly it represents a more suitable model for testing the regulation of cytokeratin- related KRT16 gene by ATF4. As expected, treatment of HaCaT cells with ER stress inducers Tunicamycin and Brefeldin A led to a statistically significant upregulation of ATF4 mRNA ( Fig 7A) and a substantial induction of transcripts from FAM129A and HKDC1 genes (Fig 7B  and 7C). Surprisingly, there was no induction of transcripts from the KRT16 gene, even a slight reduction in KRT16 mRNA level (Fig 7D). Similar results were obtained with the in vitro transformed human embryonic kidney cell line HEK293T (S4 Fig). The data obtained with these cell lines confirm the involvement of FAM129A and HKDC1 genes in ISR-related responses mediated by ATF4 but raises questions about the controversial mode of regulation of KRT16.

KRT16 promoter is strongly induced by transcription factor ATF4
To clarify a role of ATF4 in the regulation of KRT16 promoter, we have constructed a reporter plasmid by placing firefly luciferase gene under control of the KRT16 gene promoter. There are two potential ATF4 binding sites within the upstream region of KRT16 gene, at position 171-163 and 482-474 upstream of the translation start site (Fig 8A), one (proximal) exactly matching the consensus sequence for CARE (C/EBP-ATF responsive element) and the other (distal) being different in only one nucleotide [4]. The 508 bp upstream region spanning from the translation start and composed of the promoter region comprising two putative ATF4-binding sites and the 5'-UTR, was amplified by PCR from total human DNA and inserted into pGL3-Basic vector upstream of firefly luciferase reporter gene. The obtained plasmid pGL3-KRT16(508) was transfected into HeLa cells. A reporter plasmid expressing beta-galactosidase under control of the constitutive CMV promoter was used as normalization control. In response to overexpression of ATF4 we observed almost a 10-fold increase in activity of the reporter gene luc in the transfected cells (Fig 8B), indicating that ATF4 can activate KRT16 promoter. We also made a construct pGL3-KRT16(471) containing only the proximal putative ATF4-responsive element by insertion of genomic fragment spanning from the upstream nucleotide 471 to translation start site (Fig 8A). These two constructs demonstrated similar induction of luciferase activity in response to the ectopic ATF4 expression (Fig 8B) suggesting that the distal site may not be involved into the ATF4-mediated regulation of KRT16 promoter. We then constructed and tested reporter plasmid pGL3-pKRT16 (471)-mut in which the proximal ATF4 binding site TGATGAAAT was mutated to inactive TTATTATAT sequence (Fig 8A). The mutations not only prevented the induction of a reporter gene in response to ectopic expression of ATF4, but also even led to a reliable 2-fold decrease in reporter activity below the basal levels ( Fig 8B). The similar and even more pronounced effects were observed in transfected HCT116 cells (Fig 8C). We conclude that the promoter region of the KRT16 gene does indeed contain the ATF4 responsive element that apparently is responsible for the upregulation of transcription as part of the ISR.

Discussion
The results obtained in this study indicate that a dysfunction of mitochondrial respiration, as well as ER stress result in a partially ATF4-dependent stimulation of KRT16, FAM129A and HKDC1 expression. The stimulation of the genes can be part of ISR, which is supported by the fact that a treatment of HCT116 cells with the specific inhibitor of the integrated stress response ISRIB resulted in a decrease of ER stress-induced levels of KRT16, FAM129A and HKDC1 transcripts. Similarly, the downregulation of ATF4 by RNAi in HCT116 cells has attenuated the induction of KRT16, FAM129A and HKDC1 mRNAs in response to ER stress or dysfunctional mitochondrial respiration. The same three genes demonstrate a transcriptional activation in response to ectopic expression of ATF4, and for FAM129A/Niban the upregulation was also observed at the protein level.
The similar mode of regulation of KRT16, FAM129A and HKDC1 genes by ER stress and dysfunctional mitochondrial respiration was observed in cervical carcinoma HeLa cells. However, the regulation was different in two human transformed epithelial cell lines HaCaT and HEK293T that do not originate from patient-derived tumors. The HaCaT cells represent spontaneously transformed human keratinocytes [26], and HEK293T are human embryonic kidney cells transformed in vitro by oncogenes of adenovirus type 5 and SV40 [27]. In response to Tunicamycin and Brefeldin A in these cell lines we also observed the substantial upregulation of FAM129A and HKDC1 transcripts, but there was no effect or even a downregulation of KRT16 mRNA.
To confirm that KRT16 gene is directly responding to ATF4 we tested luciferase reporter constructs containing upstream regions of the KRT16 gene and identified a single active ATF4-specific DNA element TGATGAAAT located at a distance of 171 bp from the translation start. Identical ATF4 binding sites were previously identified in the gene for 4E-BP1 protein, a suppressor of the eukaryotic translation initiation factor eIF4E [4,28] and the ATF4 mediated induction of 4E-BP1 was shown to protect insulin-producing pancreatic beta cells against ER stress [28].
Our results indicate that although KRT16 gene contains a functional ATF4-responsive DNA element, its regulation by ATF4 is conditional, and in particular, the gene responds to ATF4 in the two patient-derived cancer cell lines, but not in the cell lines transformed in vitro. It is possible that the ATF4 regulation of KRT16 is inhibited in normal cells, but is somehow activated during cancer progression processes. KRT16 is not expressed in healthy epidermis, but the expression is associated with activation and hyperproliferation of keratinocytes [13]. The Keratin 16 protein participates in innate barrier function of the skin [29], and hereditary mutations in KRT16 gene are associated with pachyonychia congenita, hyperkeratotic lesions of the skin, nails and oral epithelium [30]. KRT16 expression is upregulated in certain types of cancer [31][32][33][34]. Keratins form filamentous structures by a formation of heteropolymeric pair between type I (acidic) and type II (basic) molecules [35]. In stratified epithelia type-I KRT16 is co-expressed with the type-II KRT6 as the keratin pair, and both genes are coordinately induced by stresses, injuries and in pathologies [35]. However, we found that the upregulation of KRT16 mRNA by inducers of ER stress was not accompanied by the upregulation of KRT6A transcripts (S5 Fig). This result points to the potential non-canonical functions of KRT16 during stress responses.
Previous studies has identified a participation of several transcription factors that control KRT16 in response to different signals, such as stimulation with growth factors (EGF and TGFalpha) [36,37], or oxidative stresses [38]. Here we show that KRT16 gene is also linked to ISR through the action of ATF4. However, the participation of KRT16 in ISR seems to have a rather complex regulation. The failure to induce KRT16 mRNA in HaCaT and HEK293T cells, as opposed to the robust induction in HCT116 and HeLa cells may be explained by certain steps during cancer progression and selection of cells that are more competitive in tumors. The process may eliminate certain existing limitations that hinder the induction of KRT16 transcription in response to ATF4. This hypothesis is substantiated by the reported upregulation of KRT16 in tumors of the salivary glands [31], laryngeal [33], squamous skin [32] and breast cancer [34].
The two other genes identified here as ATF4-responsive members of ISR, FAM129A and HKDC1, were also implicated in adaptive processes and pathologies. FAM129A has previously been found to be inducible by ER stress [39], and here we show that levels of FAM129A transcripts and of Niban protein depend on ATF4. FAM129A encodes Niban protein that is overexpressed in many types of cancer. Niban can protect cells from apoptosis induced by genotoxic stress [15,16,40]. It positively regulates overall translation by affecting the phosphorylation of eIF2 alpha, S6 kinase and 4E-BP1 [39]. UV irradiation induces the phosphorylation of Niban by Akt protein kinase leading to an association of Niban with nucleophosmin NPM / B23 and dissociation of nucleophosmin from the complex with Mdm2 [41]. The released Mdm2 may stimulate the proteasomal degradation of p53 leading to an attenuated apoptotic response to stresses. In opposite, a suppression of Niban expression promotes the stabilization of p53 and increases apoptosis [41]. The two transcription factors ATF4 and p53 appear to act in an antagonistic manner. We found that p53 can suppress the induction of ATF4-dependent transcription during a sustained inhibition of mitochondrial Complex III [18]. In contrast, ATF4 can negatively regulate p53 by suppressing p14ARF, the inhibitor of the p53-Mdm2 interaction [42]. As mentioned above, Niban can also promote the Mdm2-assisted proteasome degradation of p53 [41]. Therefore, the ATF4 dependent upregulation of FAM129A can confer a pro-survival function during ISR, through the attenuation of p53-dependent apoptotic responses.
We identified HKDC1 gene as a novel candidate component of ISR that responds to ER stress and dysfunctional respiration chain. The response only partially depends on ATF4 (Figs 3-5). Of note, in all cell lines studied the magnitude of HKDC1 mRNA upregulation by the inducers loosely correlated with the value of induction for transcripts from the ATF4 gene. So, unlike the ATF4 transcripts, the induction of HKDC1 mRNA by Tunicamycin was more prominent than by Brefeldin A in HaCaT (Fig 7A and 7C) and HeLa (S3 Fig) cells, suggesting that an additional signaling pathway may also be involved in regulation of HKDC1 transcription in response to ER stress. The possibility looks plausible as at least three known signaling pathways (PERK/eIF2α/ATF4, IRE1/XBP1 and ATF6) participate in the UPR/ER stress response [21,22]. Details and specific role of IRE1/XBP1 and ATF6 pathways in the regulation needs to be further explored.
The human HKDC1 gene encodes one of the five known hexokinases that phosphorylate hexose sugars [14,43,44]. Since phosphorylation is the first step of glucose catabolism, hexokinases may represent the critical step in regulation of energy metabolism that particularly limits the rate of cell growth and proliferation [45]. However, among the hexokinases only HKDC1 was shown to be dramatically overexpressed in the tumor tissues [14], which suggested a pivotal role of HKDC1 in cancer and particularly, a potentially promising therapeutic target for lung cancer [14]. Possibly, the upregulation of HKDC1 during ISR helps to overcome the deficiency of energy during the stress by stimulating the intracellular retention of glucose and driving it to glycolysis.
In conclusion, we obtained evidences for three novel components of ISR mediated by the ATF4 transcription factor and suggested their potential pro-survival roles during ER stress and dysfunctional mitochondrial respiration. However more functional data for each of the genes is required for the holistic understanding of their biological roles in homeostasis and response to stresses in normal and cancerous cells.
The mitochondrial electron transfer chain inhibitors Myxothiazol or Piericidin A (Sigma-Aldrich, MO) were used at final concentrations of 1 and 2 μM, respectively; for induction of ER stress the cells were treated with 1 μg/ml Tunicamycin or 0.5 μg/ml Brefeldin A (Sigma-Aldrich, MO). ISRIB (Sigma-Aldrich, MO) was added to the final concentration 200 nM when indicated. The treatment time for each experiment is indicated in the legends to figures.
The lentiviral vectors expressing ATF4 shRNA or the control, scrambled shRNA, were prepared and introduced into HCT116 cell line as described [18].
The ectopic expression of ATF4 from transfected plasmid pHM-ATF4 was described in our recent report [17].

Plasmid construction
For construction of the reporter plasmids containing two or one putative ATF4 binding sites, the DNA fragment (556 bp) comprising KRT16 promoter and 5'UTR was amplified by PCR on a total DNA from HeLa cell line with the primers KRT16_forward (5'-GTGGGACAT GTGGG ATCCCCA-3') and KRT16_reverse (5'-CATGACCACCTGCAGCCGCCAA GCTTAAT-3'). The PCR product was inserted into the plasmid pUC19 between BamHI and HindIII sites for sequencing, then recloned as BamHI-HindIII or SacI-HindIII fragments into the pGL3-Basic vector (Promega,WI) resulting in formation of pGL3-KRT16(508) and pGL3-KRT16(471) constructs, encoding 508 or 471 nucleotides upstream of KRT16 translation start codon, respectively.
The potential ATF4-binding site was inactivated by introducing three mutations by megaprimer PCR. The first PCR was performed with primers KRT16_forward and KRT16_mut_rev (5'-ACCGGGAGTTATTATATCCAGAGGGGAAC -3'), the plasmid pUC19 encoding KRT16 gene promoter and 5'-UTR was used as a template. The PCR product was utilized as a forward primer in the second PCR with the other primer KRT16_reverse. The PCR product was sequenced and recloned in pGL3-Basic vector as SacI-HindIII fragment forming the plasmid pGL3-KRT16(471)mut.
The target ATF4, KRT16, FAM129A and HKDC1 transcripts were quantified using the reference 18 S rRNA for normalization.