Induction of the Heat Shock Pathway during Hypoxia Requires Regulation of Heat Shock Factor by Hypoxia-inducible Factor-1*

Activation of heat shock proteins (Hsps) is critical to adaptation to low oxygen levels (hypoxia) and for enduring the oxidative stress of reoxygenation. Hsps are known to be regulated by heat shock factor (Hsf), but our results demonstrate an unexpected regulatory link between the oxygen-sensing and heat shock pathways. Hsf transcription is up-regulated during hypoxia due to direct binding by hypoxia-inducible factor-1 (HIF-1) to HIF-1 response elements in an Hsf intron. This increase in Hsf transcripts is necessary for full Hsp induction during hypoxia and reoxygenation. The HIF-1-dependent increase in Hsps has a functional impact, as reduced production of Hsps decreases viability of adult flies exposed to hypoxia and reoxygenation. Thus, HIF-1 control of Hsf transcriptional levels is a regulatory mechanism for sensitizing heat shock pathway activity in order to maximize production of protective Hsps. This cross-regulation represents a mechanism by which the low oxygen response pathway has assimilated complex new functions by regulating the key transcriptional activator of the heat shock pathway.

Activation of heat shock proteins (Hsps) is critical to adaptation to low oxygen levels (hypoxia) and for enduring the oxidative stress of reoxygenation. Hsps are known to be regulated by heat shock factor (Hsf), but our results demonstrate an unexpected regulatory link between the oxygen-sensing and heat shock pathways. Hsf transcription is up-regulated during hypoxia due to direct binding by hypoxia-inducible factor-1 (HIF-1) to HIF-1 response elements in an Hsf intron. This increase in Hsf transcripts is necessary for full Hsp induction during hypoxia and reoxygenation. The HIF-1-dependent increase in Hsps has a functional impact, as reduced production of Hsps decreases viability of adult flies exposed to hypoxia and reoxygenation. Thus, HIF-1 control of Hsf transcriptional levels is a regulatory mechanism for sensitizing heat shock pathway activity in order to maximize production of protective Hsps. This cross-regulation represents a mechanism by which the low oxygen response pathway has assimilated complex new functions by regulating the key transcriptional activator of the heat shock pathway.
In order to endure oxygen deprivation, most eukaryotes utilize a conserved set of cellular adaptations (1). Many of these changes are brought about by the activation of the transcription factor hypoxia-inducible factor-1 (HIF-1), 2 a heterodimeric complex composed of HIF-1␣ and HIF-1␤ subunits. When this complex is formed it binds to specific DNA enhancer sequences and regulates the activity of target genes. Both HIF-1␣ and HIF-1␤ are constitutively expressed in normal oxygen conditions (normoxia), but HIF-1␣ protein is quickly degraded before dimerization can occur with HIF-1␤ (2). Normoxic HIF-1␣ degradation is mediated by a series of hydroxylations and ubiquitinations that tag HIF-1␣ for disposal through the proteasome (3)(4)(5)(6).
The HIF-1 complex transcriptionally regulates a wide array of genes involved in anaerobic metabolism, growth, prolifera-tion, angiogenesis, and cell death (7,8). This multifaceted control of cellular and organismal physiological pathways is exploited by solid tumors through the natural hypoxic environment caused by rapid growth or genetic alterations that stabilize HIF-1␣ (9). Overexpression or activation of HIF-1␣ is often seen in a wide array of cancers and is correlated with patient survival (10), and studies have shown that targeting the HIF-1 pathway is a promising means of cancer therapy (11,12). Thus, HIF-1 is a central regulator of normal and pathological changes in response to low oxygen.
Although many genes that are up-regulated during hypoxia are known to be regulated by HIF-1, there are also diverse sets of genes up-regulated that have not been linked to the actions of HIF-1. Among these are the highly conserved heat shock proteins (Hsps) that are highly up-regulated during hypoxia but have not been linked to HIF-1 regulation (13). Hsps are known to act as cellular chaperones for proteins that are misfolded by cellular stresses (14). Heat shock factor (Hsf) was one of the first studied transcription factors, and its activation by stresses that promote the unfolding of proteins has been well characterized. When cells are unstressed Hsf is in a monomeric state, but cellular stress induces trimerization of the protein (15,16). The trimeric form of Hsf activates transcription of downstream genes such as Hsps (17,18). However, this study identifies a novel mode of regulation of heat shock pathway activity during hypoxia through a HIF-1-dependent increase in Hsf transcript levels. This up-regulation of Hsf is necessary for the full increase of Hsp transcripts normally observed during hypoxia and also during reoxygenation. These findings establish a novel regulatory link between two stress pathways previously thought to be independent in responding to hypoxia.

EXPERIMENTAL PROCEDURES
Cell Culture and Hypoxia Treatments-Drosophila melanogaster Kc 167 tissue culture cells were obtained from the Drosophila Genomics Resource Center. Cells were maintained in Schneider's Drosophila medium (Invitrogen) supplemented with 5% heat-inactivated fetal bovine serum (Invitrogen). For hypoxia experiments, cells were incubated for 6 h in chambers flushed with 0.5% O 2 gas. The reoxygenation step consisted of a 15-min return to normal oxygen levels.
Hypoxia Reporter Construction-A small region of the Hsf intron containing the possible hypoxia regulatory elements was cloned into the Green H Pelican reporter vector. Kc 167 cells were transfected using the Effectene kit (Qiagen) with this reporter and put in normoxia, hypoxia, or hypoxia with HIF-1␣ RNAi. Images were taken using a Nikon Eclipse TE2000-U microscope and MetaVue image capture software.
Chromatin Immunoprecipitation-Kc 167 cells were transfected using the Effectene kit (Qiagen) with a pAc5.1/Sima plasmid that contains the full-length HIF-1␣ (sima) cDNA sequence with a c-terminal V5 epitope tag under the control of the Actin5c promoter. An equal quantity of mock transfected cells was used as a control, and all purification steps were carried out in parallel with the control and experimental cells. 24 h after transfection, control and experimental cells were incu-bated in hypoxia at room temperature for 24 h. DNA isolation and purification procedures followed standard V5 protocol.
Larval Survival-First instar larvae of wild-type or flies with only one copy of the Hsf gene were counted and placed 10/vial of food. These vials were split and placed into groups for normoxic or hypoxia and reoxygenation stresses. This experimental group was maintained at 0.5% oxygen for 23 h and then placed in ambient oxygen for 1 h before returning to hypoxia. This cycle was repeated daily, and after 2 weeks the vials were scored for survival.

RESULTS
Hsf Transcript Levels Increase during Hypoxia in a HIF-1␣dependent Manner-D. melanogaster Kc 167 cells were treated with GFP control double-stranded RNA (dsRNA) or dsRNA directed to eliminate transcripts of the Drosophila HIF-1␣ homologue, similar (20,21), through RNAi. After exposure of the treated cells to normoxic or hypoxic conditions, total RNA was isolated for semi-quantitative reverse transcription PCR in order to characterize the transcript levels of HIF-1␣, Hsf, and Actin5c as a control (Fig. 1A). Interestingly, we found that transcript levels of Hsf increased during hypoxia and that this upregulation was HIF-1␣ dependent. Cells lacking HIF-1␣ due to RNAi did not display a hypoxic increase in Hsf, instead maintaining Hsf levels more similar to control normoxic cells. Realtime PCR was then used to more accurately characterize these results (Fig. 1B) and further corroborated that Hsf transcripts increase under hypoxic conditions in a HIF-1␣-dependent manner. As an additional control we repeated the RNAi with an alternate dsRNA sequence targeting another area of HIF-1␣, which also showed a HIF-1␣-dependent hypoxic increase of Hsf (Fig. 1C). This control experiment confirms our results were not due to off-target effects of the original RNAi.
The Increase in Hsf Transcript Levels during Hypoxia Is Directly Regulated by HIF-1␣-The DNA recognition element to which HIF-1 binds during hypoxia contains a core 5Ј-RCGTG sequence (22). We had identified multiple instances of a related motif, 5Ј-TACGTGC, in the intron of the known HIF-1 target gene (23) HIF-1 prolyl hydroxylase and searched for this motif in the Hsf gene region. We identified two of these putative hypoxia response elements (HREs) in close proximity to one another in the second intron of Hsf. The two sites were 923 and 992 base pairs downstream from the transcriptional start site of Hsf, respectively. When this genomic region was aligned (24) with seven other Drosophila species these potential HREs were perfectly conserved ( Fig. 2A). The sequence conservation of the two HRE motifs strongly suggests that there is evolutionary pressure to maintain these specific sequences.
We next tested whether these conserved HRE motifs had a regulatory function during hypoxia. A portion of the second intron of Hsf containing the potential HREs was cloned upstream of a minimal promoter driving GFP in the Green H Pelican reporter vector (25). This reporter construct was then transfected into the Kc 167 cell line and put under normoxic or hypoxic conditions. The hypoxic cells showed a dramatic increase in GFP fluorescence compared with the normoxic cells (Fig. 2B). The hypoxic increase in GFP expression was eliminated by HIF-1␣ RNAi treatment. The original reporter vector lacking the Hsf intron showed no hypoxic activation of GFP (data not shown), confirming that it was the cloned intronic region of Hsf that was leading to the HIF-1-dependent induction of the reporter during hypoxia.
The HRE-containing region was also tested by chromatin immunoprecipitation for direct binding by HIF-1. We transfected Kc 167 cells with an epitope-tagged HIF-1␣ expression vector (26). DNA bound to HIF-1 protein during hypoxia was immunoprecipitated, and the Hsf intron genomic region as well as control genomic regions were PCR amplified to test for enrichment compared with DNA immunoprecipitated from a mock transfection. We found distinct enrichment of a 260-bp fragment encompassing the two HREs of the Hsf intron (Fig.  2C). As a positive control we showed an enrichment of a genomic region containing an HRE within the intron of the known HIF-1 target Hph. A negative control fragment located in an Actin5c intron showed no enrichment between the HIF-1␣ pulldown and the untransfected pulldown. These data indicate that Hsf is a direct target of HIF-1␣ through the binding of an intronic region containing two HREs that act as an enhancer of transcription during hypoxia.
Full Induction of Hsps during Hypoxia Is Dependent on HIF-1␣ Regulation of Hsf-The functional impact of the upregulation of Hsf by HIF-1␣ on Hsp induction during hypoxia was then assayed. Kc 167 cells were exposed to normoxia and hypoxia after treatment with control and HIF-1␣ RNAi, and reverse transcription PCR assayed transcript levels of various Hsps. All Hsps examined were dramatically up-regulated under hypoxia, and this increase was partly HIF-1␣ dependent (Fig.  3A). Hsp transcripts were not completely eliminated in hypoxic cells treated with HIF-1␣ dsRNA, presumably because the hypoxic stress activated the basal (normoxic) levels of Hsf protein already present in the cells. No HREs were found near any of the Hsp genes; therefore it is unlikely that HIF-1 was directly up-regulating these genes during hypoxia.
We tested whether the up-regulation of Hsps during hypoxia was dependent on Hsf. Cells were treated with control or Hsf RNAi and placed in normoxic and hypoxic conditions. When Hsf was removed through RNAi, Hsp transcripts were eliminated completely compared with the strong induction seen in cells treated with control dsRNA (Fig. 3B). Real-time PCR was used to more accurately quantify the results from both of the RNAi experiments. HIF-1␣ RNAi reduced the up-regulation of Hsps during hypoxia, yet Hsf RNAi completely removed Hsp transcripts (Fig. 3C). From these results, we can discern that Hsf regulates Hsps, while HIF-1 regulates Hsf.
The lack of strong Hsp up-regulation in hypoxic HIF-1 knockdown cells suggests that the HIF-1-mediated increase in Hsf transcript levels is an important step in regulating the sensitivity and activity of the heat shock response pathway. The functional impact of an increase in Hsf transcript levels in hypoxia was tested by assaying the response to hypoxia of a fly heterozygous for the null Hsf 1 mutation (27) and therefore containing only a single wild-type copy of Hsf. After exposure to hypoxia, these flies had reduced levels of Hsf transcripts compared with wild-type Oregon R flies as measured by real-time PCR (Fig. 4). The heterozygous flies with a reduction in Hsf transcripts also showed a strong reduction in Hsp26, Hsp27, and Hsp68 transcript levels compared with the control flies, although two Hsp70 genes had normal levels of induction. These findings suggest that Hsf abundance impacts the up-regulation of some Hsps in a dose-dependent manner during hypoxia. Lower Hsf transcript abundance than the levels normally achieved during hypoxia are insufficient for the full upregulation of Hsps.
Full Induction of Hsps and Viability during Reoxygenation Is Dependent on Increased Hsf Levels-During the return to normal oxygen conditions, Hsp levels remain high and are critical to tissue survival during this reoxygenation (28,29). The effect of the HIF-1-dependent increase in Hsf level on Hsp expression persists during reoxygenation. Kc 167 tissue culture cells with HIF-1␣ knocked down by RNAi had little increase in Hsp expression after hypoxia treatment and a reoxygenation period (Fig. 5A). Thus, the up-regulation of Hsf during hypoxia is critical to the high levels of Hsp transcripts during reoxygenation, as well as hypoxia.
Furthermore, we examined the functional importance in vivo of increased Hsf transcript abundance by assaying larval survival under hypoxia and reoxygenation stress. First instar larvae were reared in a regimen of alternating hypoxia and reoxygenation. The Hsf 1 heterozygotes had greatly reduced survival compared with larvae reared in normoxia (Fig. 5B). Control wild-type larvae showed no significant difference in survival between normoxia and the hypoxia and reoxygenation environments. These findings demonstrate the dosage importance of Hsf transcript levels for coping with hypoxia and reoxygenation at the organismal level.
Taken together, these experiments show the sequential order and importance of the hypoxia response. During hypoxia, HIF-1 directly up-regulates Hsf, which in turn upregulates the whole family of Hsps. Without the HIF-1-reg-ulated increase in Hsf, Hsps transcript levels never reach full induction during hypoxia or reoxygenation and organismal viability is reduced.

DISCUSSION
Up-regulation of Hsps during hypoxia is part of the canonical low oxygen stress response seen in Drosophila (30), Caenorhabditis elegans (13), and mammalian tissues (31). This study provides evidence that the up-regulation of Hsf during hypoxia surprisingly requires the activity of HIF-1, the effector of the low oxygen response. The transcriptional control of Hsf by HIF-1 has a functional impact on the activity of the heat shock response during hypoxia and the return to normal oxygen levels. Cells lacking HIF-1 or with reduced dosage of Hsf only increase Hsp transcript production slightly during low oxygen exposure and reoxygenation. The decreased production of Hsps reduces viability in flies experiencing hypoxia and reoxygenation, demonstrating that the full induction of the heat shock response is essential to counter the diverse physiological stresses associated with low oxygen.
Thus, we propose a model where HIF-1 directly up-regulates Hsf during hypoxia and the increased Hsf abundance in turn allows Hsf to further up-regulate Hsps during low oxygen exposure and also after the return to normal oxygen levels. The regulation of Hsf by HIF-1 provides a clear example of how cross-regulation between physiological stress response pathways can allow one pathway to sensitize the second and elicit a response under conditions where normally it would not be activated.

Complex Regulation of Physiological Response Pathways-
Cross-regulation between physiological pathways appears to be a feature of the low oxygen response. It has been shown that the insulin pathway can dramatically affect the HIF-1 pathway (32). Through the actions of the phosphatidylinositol 3-kinase/Akt pathway, HIF-1␣ translation is increased in a manner that outpaces the naturally normoxic degradation of HIF-1␣ (33). This leads to HIF-1 activation even when oxygen is present and up-regulating its downstream targets. Recently it has been shown that transforming growth factor-␤1 activates the HIF-1 pathway by reducing the levels of prolyl hydroxylases that tag HIF-1␣ for degradation. Interestingly, it is also known that Hsp90 plays a role in stabilizing HIF-1␣ (34,35). This mechanism is independent of the canonical oxygendependent regulation of HIF-1␣ and was the first evidence of any link between the heat shock and hypoxia stress pathways.
The cross-regulation between HIF-1 and Hsf found here is a new type of control, where the transcriptional effector of the low oxygen response directly regulates the transcript level of the effector of the heat shock response in order to sensitize the pathway. Interestingly, it has been already shown that HIF-1 and Hsf pathways have regulatory interactions, but in response to heat. Studies using C. elegans and rats showed that HIF-1 activity was essential for heat acclimation (36,37). Our findings may explain the mechanism behind this phenomenon in that the increase in metabolic activity during high temperature may cause oxygen scarcity, thus stabilizing HIF-1 and increasing Hsf transcript levels.
Transcriptional Control of the Heat Shock Response-The activity of the heat shock pathway has been shown to be controlled by the trimerization and post-translational modification of Hsf protein subunits (18). Our results indicate that transcriptional control of Hsf is a means of further regulation of heat shock pathway activity. This transcriptional regulatory step is controlled by HIF-1, supporting a model in which the HIF-1 pathway causes increased Hsf transcription during hypoxia as a means to increase the cellular abundance of Hsf and increase the sensitivity of the heat shock pathway. In addition, the control of heat shock response sensitivity by HIF-1, the regulator of the low oxygen response, suggests that stress response pathways can assimilate complex new functions by regulating the transcriptional activators of other stress pathways.
Disease Implications-It has been shown that the increases in Hsp levels are critical for cell survival during hypoxia and the subsequent reoxygenation (29,38). Our results indicate that it is through the HIF-1 pathway that the cell achieves this Hsp increase and is a means to protect against the stress of hypoxia. HIF-1 accumulation and activity have been linked to tumor  Real-time PCR of flies heterozygous for a null Hsf mutation show a significant reduction in Hsf, Hsp26, Hsp27, and Hsp68 transcript abundance compared with wildtype flies after hypoxia (p Ͻ0.05). This demonstrates that Hsf transcript abundance is critical to the magnitude of Hsp production. S.E. of the mean is shown.
progression, and various Hsps have also been shown to be crucial to cancer survival (39); thus, the hypoxic and heat shock response pathways play important roles in the pathophysiology of cancer. Our finding that the activity of HIF-1 controls the output of the heat shock pathway offers possible therapeutic approaches for mitigating hypoxic tissue damage and tumor growth by targeting this novel regulatory link.