Deregulated tumour growth and neovascularization result in an inadequate tumour blood supply, leading to areas of chronic hypoxia and necrosis. Irregular vascular structure and abnormal tumour physiology also cause erratic blood flow in tumour vessels. We reasoned that tumour stroma, including vascular endothelial cells, would consequently experience transient hypoxia that may allow transcriptional targeting as part of an antivascular gene therapy approach to cancer. To exploit hypoxia for transcriptional regulation, retroviral vectors were generated with modified LTRs: a 6-mer of hypoxia response elements in place of the viral enhancer produced near wild-type levels of expression in hypoxia but was functionally inert in normoxia. In a tumour xenograft model, expression was mainly around areas of necrosis, which were shown to be hypoxic; no expression was detected in tumour stroma. Time-course experiments in vitro demonstrated that expression was transient in response to a hypoxic episode, such that a reporter gene would be insensitive to acute hypoxia in vivo. In contrast, a significant therapeutic effect was seen upon ganciclovir administration with a vector expressing thymidine kinase (TK) in the tumour stroma. Expression of TK was more effective when targeted to acute hypoxia in the stroma compared to chronic hypoxia in the poorly vascularized regions of the tumour cell compartment. The data presented here are evidence that hypoxia in the stromal compartment does occur and that transient hypoxia constitutes a valid therapeutic target.
Hypoxia is a state of low oxygen tension and is a widespread phenomenon in many different tumours due to inadequacies in their vasculature.1 Studies using vascular casting techniques showed that tumour vessels are highly irregular, with arterial venous shunts, blind ends and twisting.2, 3 The blood flow within these vessels is also erratic, as dyes injected several minutes apart showed: tumour blood vessels could be labelled with either dye alone, implying that vessels were opening or closing during the interval between injections.4 Fluorescent labelling of red blood cells showed unstable flow in mammary adenocarcinomas and this caused changes in vascular oxygen tension.5 Helmlinger et al6 found large, well-perfused vessels with hypoxic pO2 values and that within a 1 h time period the oxygen tension (pO2) could fluctuate. Because of their disordered vasculature, growing tumours can quickly outstrip the available oxygen to cause areas of chronic hypoxia, increased acidity and nutrient depletion. Such cells remain viable and are resistant to radio- and chemotherapy;7, 8 although they are essentially nonproliferative, they constitute a pool of cells capable of repopulating the tumour. However, fluctuations in blood flow as described above additionally lead to cells experiencing transient hypoxia. It has recently been appreciated that acute hypoxia represents a more significant challenge in that cells are protected against standard therapies yet readily contribute to tumour progression or regrowth when the conditions change.8
The transcriptional response to hypoxia is mediated by the hypoxia-specific transcription factor, HIF-1 (hypoxia inducible factor-1). HIF-1 is composed of two subunits, HIF-1α and HIF-1β (also known as ARNT, the aryl hydrocarbon receptor nuclear translocator). HIF-1α is the main modulator of control for expression of HIF-1 responsive genes; three mechanisms of oxygen sensing and control of response have been elucidated. Firstly, in the presence of oxygen, HIF-1α becomes hydroxylated by the prolyl hydroxylases PHD1, 2 and 3, of which PHD2 was found to be the key oxygen sensor.9, 10, 11 Hydroxylation of prolines 564 and 402 within the oxygen-dependent degradation domain (ODD) allows association with pVHL (the von Hippel–Lindau tumour-suppressor protein). HIF-1α is then ubiquitinated by an E3 ubiquitin ligase complex containing pVHL, which targets it for destruction by the proteosome.12, 13, 14 Acetylation in normoxia of lysine 532 in the ODD of HIF-1α by ARD1 (ADP-ribosylation factor domain protein-1) increases its interaction with pVHL and consequent degradation.15 During hypoxia, hydroxylation does not occur and the protein is stabilized. Finally, hydroxylation during normoxia of asparagine 803 in the HIF-1α C-terminal activation domain by FIH-1 (factor inhibiting HIF-1) prevents interaction with p300/CBP coactivators and thus transcription of target genes.16 In hypoxia, active HIF-1α translocates to the nucleus and heterodimerizes with HIF-1β to activate various hypoxia-responsive genes via upstream or downstream enhancer elements. The hypoxia-responsive element (HRE) is the DNA recognition site for HIF-1 and consists of one or more copies of 5′-ACGTG-3′. HIF-1 binding to HREs stimulates transcription of many genes including erythropoietin (EPO), vascular endothelial growth factor (VEGF), glucose transporter-1 (GLUT-1) and phosphoglycerate kinase (PGK). The physiological effects of genes upregulated in this way include increased ventilation, increased cardiac output, induction of erythropoiesis, tissue neovascularization and the metabolic shift from oxidative phosphorylation to glycolysis.
Work in this laboratory is concentrating on the destruction of tumour vasculature as a strategy to eliminate solid tumours.17, 18 The dependence of tumours on a vascular network for growth and the greater accessibility of vessels compared to tumour cells make this an attractive target for gene therapy. To this end, we are developing vectors for systemic delivery to target tumour vasculature. As one means of imparting specificity, vectors for endothelial cell-specific expression have been generated.19, 20 In this study, we sought to evaluate whether transient episodes of hypoxia could be exploited for transcriptional targeting of vectors to the tumour blood vessels. As discussed above, disorganized vasculature and the physiological characteristics of tumours, such as high interstitial pressure,21 lead to erratic patterns of blood flow. We reasoned that periods of vessel collapse or stagnation of flow would result in transient exposure of endothelial and adjacent stromal cells to hypoxia that would be a tumour-specific phenomenon. Such periods of hypoxia, although short, may be of sufficient duration to induce hypoxia-dependent gene expression and so could be exploited for tumour specificity. In this strategy, the endothelial cells are accessible to vector during blood flow, gene expression is induced during subsequent periods of vascular shutdown and the therapeutic effect follows resumption of blood flow. We have generated retroviral vectors with modified LTRs that incorporate HREs from the PGK gene that have been shown to have good transcriptional activity in hypoxic cells.22 In vivo results suggest that transient hypoxia in the stromal compartment does occur and that it is a suitable target for a gene therapy strategy.
Hypoxia response in vitro of vectors with HRE-containing hybrid LTRs
In order to exploit hypoxia for transcriptional regulation, retroviral vectors were generated that contained HREs in place of the wild-type (control) enhancer within the LTR (Figure 1). Three or six tandem HREs were inserted, equally spaced at 24 bp intervals, to test the dependence on the number of HIF-1 binding sites. Equivalent vectors with enhancer deletion or containing mutated HREs were created as negative controls.
Viruses from bulk populations of amphotropic producer cells were titrated on TE671 cells and assayed histochemically for β-galactosidase expression under three different conditions. The cells were maintained in normoxia throughout or, in parallel, treated with 100 μM cobalt chloride (a hypoxia mimetic) or placed in 0.5% oxygen for 24 h prior to assay. All titrations were in triplicate for each condition and the whole experiment was repeated with three independent virus harvests (Figure 2a). Under normoxia, all vectors with deleted or modified enhancers showed a 50- to 100-fold decrease in histochemical titre relative to control. For the HRE-containing vectors, titre was fully restored when the infected cells were subjected to 0.5% oxygen. Treatment with the hypoxia mimetic restored titre for the vector containing six HREs, but the titre recovery was partial when only three sites were present. Vectors with mutated HREs did not respond to either treatment and were functionally equivalent to the enhancer-deleted vector.
The full recovery of titre upon induction by the two HRE-containing vectors is evidence that virus production and transduction efficiency were unchanged compared to the control. Moreover, Southern blot analysis showed that the copy numbers in the transduced target cells were comparable for all vectors (Figure 2b). This confirms that the differences in expression seen were due to transcriptional control exerted by the modified LTRs. In particular, transcription was dependent on both functional HREs and hypoxia, indicative of HIF-1 dependence. The stabilization and nuclear translocation of HIF-1α by treatment with 0.5% oxygen or 100 μM CoCl2 was demonstrated by cell fractionation and Western blotting (Figure 2c). HIF-1β also accumulated in the nucleus of treated cells. These data are consistent with the presence of functional HIF-1. The level of HIF-1 was greater under the more physiological conditions of true hypoxia than when using the mimetic, which binds directly to HIF-1α and prevents its association with pVHL.23 Hypoxia in all subsequent experiments was achieved using 0.5% oxygen.
Further investigation of the vectors was carried out by means of a quantitative, colorimetric assay for β-galactosidase expression (Figure 2d). The results mirror those of the titration assay (Figure 2a). However, the quantitative assay is sensitive to differences in activity above and below the threshold necessary for a histochemical read-out. Thus, it is evident that the vector containing six HREs had a higher level of expression in hypoxia than that containing only three sites. Both vectors in normoxia, and those with mutated HREs in either condition, had baseline expression comparable to the vector lacking an enhancer, showing that expression was dependent upon HIF-1. The degree of regulation exerted by hypoxia was 10-fold, with the induced level of expression reaching approximately 60% of that due to the control vector.
To determine how widespread was the ability of different cells to respond to hypoxia and to activate vector expression, five other human target cell lines were studied. These included those representative of fibroblasts and endothelial cells present in tumour stroma. Cells were transduced with the vector containing six HREs alongside controls with unmodified or enhancer-deleted LTRs. Similar to the results with TE671 cells, all showed a significant reduction in histochemical titre in normoxia and a 50- to 100-fold increase in hypoxia (Figure 3). Additionally, the whole panel of vectors was titrated on HMME7 and HMFD cells, conditionally immortalized from human breast microvascular endothelial cells and fibroblasts, respectively. The results were very similar to those shown in Figure 2a for TE671 cells (data not shown).
Assessment of in vivo expression of hypoxia-responsive vectors in a subcutaneous tumour model
Amphotropic producer clones for the vectors with three and six HREs were isolated and screened by histochemical staining and Southern blot analysis for proviral copy number. HRE3-47 and HRE6-38 were the highest-titre producer cell clones chosen for further study. In vivo activity was assessed in subcutaneous xenografts grown in nude mice, as described previously.19, 20 Mawi cells (human colorectal carcinoma) were chosen for coinjection with irradiated producer cells due to their ability to form good tumour structure in vivo while retaining a significant hypoxic fraction, as shown by staining for hypoxia-dependent pimonidazole adducts. Figure 4 shows typical sections of tumours stained with X-Gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside). Expression of the control vector occurred throughout the tumour and stroma whereas that with the deleted enhancer was limited and very faint (Figure 4a and b). The two HRE-containing vectors expressed mainly around areas of necrosis (Figure 4c and d). This colocalized with areas of chronic hypoxia identified on sections by staining for endogenous GLUT-1 expression (Figure 4e and f). No difference in the pattern of expression was observed between vectors containing three or six HREs.
Ecotropic producer clones for the vectors with six intact or mutated HREs were also generated to examine the response to hypoxia in the stromal compartment (which includes the vasculature). The in vivo activity was assessed in subcutaneous producer cell coinjection xenograft models with Mawi cells or SW620 cells (human colorectal carcinomas). Expression within the stroma was minimal and the HRE-containing vectors did not produce higher levels of expression than the baseline seen with their mutated controls (Figure 4g and h).
Kinetic studies of β-galactosidase expression in response to hypoxia
The activity of the reporter gene was studied at various times following induction to determine the kinetics of the response to hypoxia. Initially, the time course of HIF-1α stabilization in various cell lines was determined (Figure 5a). In TE671 cells, HIF-1α was stabilized within half an hour of hypoxia, with maximal stabilization achieved by 1–2 h. Other cell lines achieved maximal stabilization by 2–4 h. Upon returning to normoxia, HIF-1α protein was degraded within half an hour in all the cell lines. Figure 5b shows the β-galactosidase activity of TE671 cells infected with hypoxia-responsive and control vectors that were subjected to various periods of hypoxia or normoxia. Despite HIF-1α being rapidly stabilized in these cells upon exposure to 0.5% oxygen, hypoxia-responsive vector expression required 8–16 h of exposure. The high level of expression continued for at least 4 h of normoxia after the cells had experienced 24 h of hypoxia, whereas HIF-1α was degraded within 0.5 h. The β-galactosidase activity was half-maximal 24 h after returning to normoxia: this slow decay reflects the protein's intrinsic half-life. The expression kinetics for vectors with three or six HREs were the same. The wild-type vector was generally unaffected by hypoxia treatment although the amount of β-galactosidase/μg protein was slightly less in hypoxia than in normoxia.
The time lag of around 4–8 h from HIF-1α degradation in normoxia to a drop in β-galactosidase activity suggests that protein synthesis (ie completion of transcription, translation, folding and tetramerization) continues in normoxia following the initial induction of transcription in hypoxia. To test this, the hypoxia time-course experiment was repeated but included a normoxic ‘chase’ period before lysis of the cells to determine β-galactosidase activity (Figure 5c). Induction of expression was seen within 4–8 h of hypoxia exposure followed by 4 h of normoxia, indicating that β-galactosidase production involves a time lag of 4–8 h for completion of ongoing synthesis initiated by HIF-1-activated transcription during hypoxia.
Therapeutic targeting of hypoxic regions in vivo
The time-course experiments suggested that the β-galactosidase-expressing vectors would not efficiently reflect transient hypoxia in the in vivo tumour model. Moreover, they would at best provide only a ‘snapshot’ of reporter gene expression 2 weeks after tumour induction, with no indication of activity at earlier stages of tumour growth. This may explain the lack of expression seen in xenografts with the ecotropically targeted HRE-containing vector (Figure 4g). A therapeutic gene would give a better indication of a dynamic response, since tumour size over time would show the cumulative therapeutic effect of numerous transient episodes. Therefore, vectors were constructed for expression of the enhanced HSV-TK (TK: thymidine kinase) mutant SR39.24 High-titre amphotropic and ecotropic producer clones were generated and used in SW620 xenograft coinjection tumour experiments, as above. We have previously used this model for assessing the effect of transcriptional control of HSV-TK to sensitize tumour endothelial cells to ganciclovir (GCV), in the course of which we established that viral producer cells were cleared by the time of prodrug administration and that treatment of control tumours without TK expression was ineffective.18 Hypoxia within SW620 tumours was detectable both by pimonidazole adduct formation and by staining for endogenous GLUT-1 upregulation (Figure 6a).
3T3 or sEND1 cells were infected with virus from the selected ecotropic hypoxia-regulated vector producer clone (EH6SR48), or an equivalent wild-type TK vector,17 followed by in vitro treatment with various doses of GCV for 5 days. Assessment of viability by MTT assay showed the SR39 mutant to require approximately 10-fold lower prodrug concentrations to kill the cells, consistent with the enhanced performance reported by Black et al24 (data not shown).
Two groups of mice were coinjected with SW620 tumour cells and irradiated EH6SR48 producer cells. Following tumour establishment, one group received GCV. There was a significant difference in tumour growth rate between these groups (Figure 6b). A smaller effect was observed using an amphotropic producer clone (AH6SR14). Since transduction using amphotropic vectors occurs mainly in the tumour cell compartment,17 AH6SR14 shows the effect of targeting hypoxia mainly in the tumour cells. In contrast, with clones of control vectors with unmodified LTRs, the amphotropic control was more efficient than the ecotropic control (clones ASR29 and ESR9; Figure 6c).
Tumours in the amphotropic control group (ASR29) did not progress for the duration of GCV treatment nor for the next 10 days but growth subsequently resumed, probably due to all the virally infected cells having being cleared leaving a small number of viable tumour cells. The amphotropic HRE-regulated vector group (AH6SR14) showed a statistically significant but much less marked effect than the positive control. The small effect contrasts with the extensive staining shown with the equivalent lacZ vector (Figure 4). However, the likely explanation is that the perinecrotic expressing cells are quiescent in addition to being hypoxic, which would prevent metabolism and the S-phase toxicity of GCV.
The EH6SR48 group had markedly delayed tumour growth compared to the untreated group, similar to that for its positive control, ESR9 (Figure 6b and c). Thus, the HRE-controlled expression of TK(SR39) from HRE6 is of comparable efficacy to that of the wild-type LTR. The result indicates that the stroma has experienced hypoxia sufficient to activate gene expression and that continuous drug exposure is a more sensitive assay to detect this rather than the single snapshot that the β-galactosidase vectors provided. Moreover, this hypoxia must be intermittent to allow the cells to continue cycling and so be sensitive to GCV. Comparison with AH6SR14 and the controls, as discussed above, illustrates that chronic versus acute hypoxia is relatively ineffective for therapeutic targeting. The lack of detectable β-galactosidase expression in Figure 4g is consistent with the conclusion that hypoxia in the stroma is transient.
A further point with the EH6SR48 group is that half the mice showed little or no increase in tumour growth over the course of the experiment, whereas the other half grew tumours at a rate up to that of the untreated control (Figure 6d). Such a heterogeneity of response was not seen for any other groups and, in particular, not with the untreated group consisting of tumour cells coinjected with the same producer cells. For the tumours that did respond, the growth delay was more significant than that obtained for the control vector ESR9. Response correlated with starting tumour size, whereby the smaller tumours (around 50 mm3 or less) did not grow once treatment began, whereas tumours larger than this did not respond to GCV. This is not merely a case of having to reach a certain size for the tumours to become viable, as the AH6SR14 and untreated groups contained tumours for which treatment began at around the 50 mm3 volume and which grew substantially during the course of the experiment.
Eventual therapeutic application for targeting tumour vasculature necessitates the development of efficient vectors for gene delivery by systemic administration to the patient, and the incorporation of appropriate measures to achieve tumour specificity. The aim of this study was to evaluate the performance of a retroviral vector incorporating tandem HREs within the LTR, in place of the viral enhancer. We reasoned that transient episodes of hypoxia would be experienced by tumour vascular endothelial cells as a consequence of the disorganized architecture and erratic blood flow, and that this would be a tumour-specific phenomenon. Transcriptional control is only one part of our overall strategy to target tumour vasculature, and other levels of restriction, such as targeted delivery, will also be necessary. It is important to distinguish systemic accessibility of the vector to the vascular endothelial cells during periods of normal blood flow from subsequent periods of vascular shutdown and consequent hypoxia that we hope to exploit for control of gene expression.
The study builds on earlier work from this laboratory regarding transcriptional control using retroviral vectors18, 19, 20 and from elsewhere regarding the use of HREs for hypoxia regulation.22, 25 Transcriptional regulation by hypoxia has been well studied using plasmids and transient transfection systems,22 including the combination with tissue-specific control elements to target ischaemic muscle.25 Moreover, the latter indicated the potential for brief periods of hypoxia in vivo to enhance reporter gene expression. We have evaluated the use of retroviral vectors both in vitro and using a tumour xenograft model in vivo. The latter experimental system does not represent a delivery model but provides a setting in which to address issues pertinent to vectors following systemic delivery. Thus, issues of specificity and therapeutic efficacy can be addressed independently of the critical but as yet unsolved issues of delivery. The appropriate choice of ecotropism and the TK/GCV therapeutic read-out underpin our intention to target acute hypoxia within the tumour stroma, as elaborated below.
Incorporation of tandem HREs into the LTR resulted in a 50- to 100-fold difference of titre in normoxia versus hypoxia determined by histochemical staining, comparable to that reported previously.22 Using a quantitative assay, there was a 20-fold reduction in gene expression in normoxia by LTR modification. Hypoxia induction was dependent upon intact HREs and corresponded with nuclear accumulation of HIF-1α. The efficiency of retroviral transduction was unaffected by the LTR modification. Similar data were obtained for six different human cell lines. Moreover, a preparation of primary human endothelial cells, transduced with virus from the producer cell clones owing to their relatively poor infectivity, also showed a similar response (data not shown).
HIF-1 induction with CoCl2 was significantly lower than with 0.5% oxygen (Figure 2c), consistent with the relative degree of induction of expression for the vector containing three HREs (Figure 2a). The ability of the vector with six HREs to respond maximally under both conditions indicates that the greater number of HIF-1 binding sites allows more efficient activation of transcription when the available protein is limiting. Similarly, the greater response to hypoxia in the quantitative expression assay with six HREs (Figure 2d) indicates that more (occupied) HIF-1 binding sites enable more transcription. The greater sensitivity to HIF-1 and greater maximal expression shown with more HREs could be important features for targeting transient or low-level hypoxia, and for achieving greater therapeutic gene expression in vivo.
Within tumours, HRE-containing vectors expressed the reporter gene in regions of chronic hypoxia adjacent to necrotic areas, co-localizing with areas marked using pimonidazole or expressing GLUT-1. Pimonidazole can also identify areas of acute hypoxia at the time of administration and we have observed occasional staining within the tumour stroma, in the absence of GLUT-1, that is suggestive of the transient periods that we wish to target (data not shown). Vascular endothelial cells that are hypoxic during pimonidazole administration due to vessel shutdown are, however, unlikely to be exposed to the reagent and so cannot be identified in this way. We were also unable to identify hypoxic stromal cells when tumours were established using ecotropic virus producer cells to restrict transduction to the stroma. On the basis of time-course experiments for β-galactosidase expression in response to hypoxia, we reasoned that a reporter gene assay was unlikely to be of value in assessing transient hypoxia because it represented a single ‘snapshot’ of activity at a late time point. In contrast, periodic expression of a therapeutic gene would be ‘registered’ and so provide a more robust indication of the occurrence and potential utility of acute hypoxia: in the case of TK/GCV, the continuous presence of prodrug would enable successive periods of TK expression to exert a cumulative effect on tumour growth rate. This hypothesis was validated, as discussed further below. It is important to note that it is the cumulative nature of the assay rather than the kinetics of TK induction by individual episodes of hypoxia (which are expected to be similar to, but may differ in detail from, β-galactosidase) that distinguishes the therapeutic approach; an equivalent reporter gene approach would require a mechanism, such as genetic recombination (eg Cre-mediated removal of a floxed ‘stuffer’ fragment), to maintain expression following induction.
The amphotropic TK(SR39) vector with a wild-type LTR had a marked effect on tumour progression when GCV was administered, while the equivalent ecotropic vector caused a more moderate growth delay. Targeting hypoxia using an amphotropic vector caused a growth delay over that of the untreated control group; however, the effect was small in comparison to the presence of significant regions of hypoxia. This likely reflects the quiescence of cells in chronic hypoxia limiting the efficacy of GCV metabolism. A more significant response was seen when targeting hypoxia with an ecotropic vector, in keeping with the cells experiencing hypoxia for induction and normoxia for therapy. The sensitivity in the therapeutic system yet lack of expression with a reporter gene using ecotropic hypoxia-responsive vectors is evidence that transient hypoxia does occur in the stroma (as argued above) and that it occurs frequently enough to produce a delay in tumour growth. Although these events are not sufficient to sustain expression, the presence of the prodrug exploits it when it does occur and the tumour growth rate reflects the cumulative effect of many such occurrences. While the amphotropic, hypoxia targeted vector would similarly respond to acute hypoxia, the greater efficacy with the ecotropic vector reflects the therapeutic advantage of targeting the stromal compartment.17, 18
A different therapeutic approach for which toxicity is independent of cell cycle (eg nitroreductase/CB1954) would be expected to show efficacy towards cells in chronic hypoxia, in contrast to the S-phase-dependent activity of TK/GCV; however, cell cycle dependence does not limit the exploitation of acute hypoxia, the subject of our study. Moreover, regions of chronic hypoxia are likely to be inaccessible for the clinically relevant route of systemic vector delivery, so use of TK/GCV to avoid their contribution in the present experimental model system is a more appropriate reflection of eventual application.
Close analysis revealed that the EH6SR48 group was split equally between responders and nonresponders. This result is comparable to a study in which tumours established from stably transfected HT1080 cells containing HRE-driven P450 reductase were targeted with the prodrug RSU1069 and radiotherapy. In this case, five out of 10 tumours were totally eradicated by the treatment, whereas the other five grew as fast as the control tumours. The authors speculated that one factor was the number of cells within the tumours at oxygen tensions low enough to inhibit killing by radiation but insufficiently low to activate P450 reductase.26 From our data, the heterogeneous response in the group with HRE-driven stromal TK(SR39) correlated with starting tumour volume. Nonresponders grew similarly to the group that did not receive GCV and started at a mean volume of approximately 100 mm3. The responders started at approximately 40 mm3 and demonstrated virtually no growth for the duration of the experiment. Thus, a greater response correlates with a smaller starting volume. It is possible that if the tumours accumulate additional stroma in the later stages of tumour formation, after the early viral infection period, then the modified cells become ‘diluted’. In this case, there would be a larger proportion of uninfected cells than if the majority of the stroma derived from that recruited early in tumour development, and strategies targeting the stroma would be less effective.
The consequences of hypoxia in tumours include limiting the use of chemotherapeutic drugs that operate via free radicals and by the lack of ‘fixation’ of free radical damage from ionizing radiation.27, 28 Additionally, HIF-1 expression has been associated with increased aggressiveness and poor prognosis, and has possible links with increased metastases.27, 29, 30, 31 For these reasons, the targeted destruction of hypoxic cells is an important strategy in combating tumours and tumour progression, as one component of a combined therapeutic approach.26 Transcriptional targeting of hypoxia is an attractive way to restrict expression mainly to tumour cells. The data presented here show large areas of hypoxia-dependent expression within the tumour: these are the areas that are difficult to kill by radiotherapy or chemotherapy. However, in the light of studies demonstrating that transient hypoxia due to variable blood flow is more relevant as a means of tumour resistance to therapy,8 targeting acute hypoxia is a more important concept. There remains, however, the problem of vector delivery to such target regions. The therapeutic results presented here suggest that targeting transient hypoxia within the vasculature and stroma is a valid approach for tumour destruction, with the additional advantage that these areas are much more accessible to vector and prodrug delivery. In terms of efficacy, acute hypoxia is a comparable alternative to the use of an endothelial cell-specific promoter for transcriptional targeting.18
Materials and methods
TE671 (human rhabdomyosarcoma), EAhy926 (human endothelial cell/lung carcinoma fusion), Mawi and SW620 (human colorectal carcinoma), and TE-FLY-A8 (amphotropic) and MO (ecotropic) retroviral packaging cells,32 3T3 (murine fibroblast) and sEND1 (murine skin-derived endothelial cell line) were maintained in DMEM supplemented with 10% FCS. HUVEC-C (human umbilical vein-derived endothelial cell line) and HMEC-1 (human dermal microvascular endothelial cell line) were maintained in human microvascular endothelial growth medium (Clonetics Corp., obtained from TCS Biologicals). HMFD and HMME7 (conditionally immortalized human fibroblast and microvascular endothelial cells, respectively)33 were a kind gift from C Clarke and M O'Hare and were maintained in DMEM/10% FCS and endothelial growth medium, respectively. For hypoxic induction, a gas mixture containing 0.5% O2, 5% CO2 and 94.5% N2 was flushed through a humidified chamber containing the plates of cells for 2 min. This was then sealed and placed at 37°C for the specified time period (typically 24 h).
Cell fractionation and Western blotting
Nuclear extracts were created using a Nuclear Extraction kit (Active Motif, Rixensart, Belgium) following the protocol for a 100 mm plate. Briefly, cells were washed in PBS/phosphatase inhibitor buffer, scraped off the dish in the same buffer and pelleted. The pellet was resuspended in hypotonic buffer and shaken at 150 r.p.m. on ice for 30 min. The samples were pelleted and the supernatant removed (this was the cytoplasmic fraction). The pellet was lysed in complete lysis buffer, vortexed and centrifuged. The supernatant contained the nuclear fraction and was frozen in aliquots (typically 15–20 μl) at −80°C. Whole-cell extracts were also created using the kit.
To quantify the amount of protein in each extract, a standard curve of BSA was set up and the Bio-Rad (Bradford) Protein Assay Reagent was added. Absorbance was read at 595 nm after 5 min incubation at room temperature. Protein concentration in the extract was then calculated from the standard curve.
For detection of HIF-1α and HIF-1β protein, 25 μg of nuclear extract was electrophoresed through a 6% resolving gel. After electrophoretic transfer to a Hybond™ ECL nitrocellulose membrane (Amersham Biosciences UK Ltd, Little Chalfont, UK), the primary antibody was diluted in TBS/0.1% Tween-20/5% nonfat dried milk and allowed to hybridize to the membrane. A mouse monoclonal to HIF-1α (H1sup67, Abcam Ltd, Cambridge, UK) was used at 1:1000 and a rabbit polyclonal to HIF-1β (Abcam Ltd, Cambridge, UK) was used at 1:2000. Proteins were visualized by chemiluminescence after exposure to peroxidase-conjugated anti-mouse/rabbit secondary antibodies.
Construction of retroviral vectors
The plasmid pNeoMFGnlslacZ and derivative lacking the NheI/XbaI region of the 3′ LTR for generating the control unmodified and enhancer-deleted vectors, respectively, and the methodology for 3′ LTR modification have previously been described.19, 20 Briefly, modifications were transferred as NheI/NotI fragments after digestion of pNeoMFGnlslacZ with NheI (partial) and NotI. To generate such HRE-containing LTR fragments, trimers of the PGK HRE were synthesized as oligonucleotides with NheI and XbaI cohesive ends (see Figure 1), annealed, phosphorylated and inserted between these sites of the LTR in pΔBN. Oligonucleotides with mutated HREs were also synthesized as a control. Insertion of a second trimer generated an LTR with six tandem wild-type or mutated HREs. All LTR modifications were confirmed by sequencing. TK(SR39) vectors, in other respects equivalent to the lacZ vectors, were generated by amplifying the coding sequence24 for insertion into pMFG as an NcoI/BamHI fragment, prior to transfer of the NheI fragment into pNeoMFGnlslacZ and replacement of the 3′ LTR as detailed above.
The packaging cell line TE-FLY-A8 was used to generate amphotropic retrovirus following CaPO4-mediated transfection and selection in 1 mg/ml G418 for 2 weeks. The producer cell transfection efficiency was equivalent for all vectors (data not shown). Bulk (ie polyclonal) lacZ virus was harvested from confluent monolayers overnight in fresh medium and filtered through a 0.45 μm filter to remove cellular debris and titrated as described below. Titres were confirmed by analysis of DNA from TE671 cells infected with concentrated virus from the bulk producer cells. For concentration, 15 ml of virus was filtered, pelleted by low-speed centrifugation (2500 g) overnight and resuspended in 500 μl medium. NheI-digested DNA (20 μg) was analysed by Southern blotting using a [α-32P]dCTP-labelled lacZ probe. High-titre virus producer cell clones were generated by picking colonies of stable transfectants and screened by testing the viral harvests on cells subsequently treated with hypoxia. Ecotropic producer clones were similarly generated using TE-FLY-MO packaging cells and screening on 3T3 cells. All selected high-titre clones were analysed by Southern blotting on TE671 or 3T3 cells.
High-titre amphotropic and ecotropic producer clones of TK(SR39)-expressing vectors were similarly generated. The initial screening was performed by slot blot analysis of producer cell supernatant: viral RNA was co-precipitated with yeast tRNA, transferred to a nitrocellulose membrane and probed using a [α-32P]dCTP-labelled TK(SR39) fragment.34 Clones with significant levels of vector production were subsequently screened by Southern blot analysis, as above.
In vitro β-galactosidase analysis
Cells were plated at 2.5 × 104 cells/well in a 24-well plate the day before infection with 50 μl of virus in the presence of 8 μg/ml polybrene. At 24 h after infection, the cells were placed in the hypoxic chamber. In some experiments, hypoxia was mimicked by addition of 100 μM CoCl2 to the medium. After a 24 h period of hypoxia, the cells were fixed in 0.5% gluteraldehyde and stained with 1 mg/ml X-Gal in buffer (2 mM MgCl2, 0.01% sodium deoxycholate, 0.02% NP-40, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide). Histochemical titre was determined by colony counting. For the β-galactosidase enzyme assay, cells were infected with retrovirus and subjected to hypoxia as described above. They were then lysed in 350 μl of 0.1% Triton X-100 and 250 mM Tris pH 8.0 and the lysate was frozen at −80°C. A 50 μl portion of this lysate was combined with 50 μl of PBS containing 0.5% BSA. To this, 150 μl of 1 mg/ml chlorophenol red galactopyranoside (CPRG, Roche) in buffer (60 mM Na2HPO4 pH 8.0, 1 mM MgSO4, 10 mM KCl, 50 mM β-mercaptoethanol) was added and the lysates were placed at 37°C. The resulting colorimetric changes were read at 578 nm and compared with a standard curve of β-galactosidase activity. The results were then standardized for the amount of protein in each lysate using the Bradford protein assay reagent as described for nuclear extract preparation above.
Subcutaneous tumour xenograft model
High-titre virus producer cell clones were X-irradiated (20 Gy) and mixed at a 5:1 (lacZ studies) or 10:1 ratio with 1.5 × 106 tumour cells. These were injected subcutaneously into the right flanks of 5- to 7-week-old male MF1 nude mice (Harlan UK Ltd, Bicester, UK) in a volume of 150 μl. For the lacZ studies, the tumours were allowed to grow for 2–3 weeks until they reached a volume of approximately 450 mm3, at which point the mice were killed and the tumours removed for immunohistochemical analysis. For the therapeutic studies, administration of GCV (Cymevene, Roche, UK) began 10–14 days after injection: 50 mg/kg was delivered intraperitoneally, twice-daily for 28 days. Tumours were measured with callipers every 2 days, from which volumes were calculated as (length × width2)/2. All animal experimentation was performed to UK Home Office Regulations. A Mann–Whitney U-test was used to show statistical significance between tumour growth in different groups.
Tumours were cut into two or four pieces and each piece was either fixed in 4% paraformaldehyde for staining with X-Gal or snap-frozen. For staining with X-Gal, tumours were washed in PBS-A and incubated twice for 20 min with pre-stain (2 mM MgCl2, 0.01% sodium deoxycholate, 0.02% NP-40). They were then incubated in X-Gal containing stain (2 mM MgCl2, 0.01% sodium deoxycholate, 0.02% NP-40, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 1 mg/ml X-Gal) at 37°C overnight. After this, they were washed in PBS-A twice before being placed in 70% ethanol at 4°C. Processing, embedding and sectioning were carried out and the sections were counterstained with haematoxylin or eosin.
For the GLUT-1 immunohistochemistry, sections stained with X-Gal were deparaffinized and rehydrated. Microwave retrieval in 0.01 M citrate buffer (pH 6.0) was performed and endogenous peroxidase was blocked. The sections were then incubated at room temperature for 1 h with 1:200 rabbit anti-human GLUT-1 antibody (DAKO Corporation, California, USA), washed and incubated with HRP-conjugated swine anti-rabbit Ig secondary antibody (DAKO) at 1:100 for 30 min at room temperature. The signal was amplified with rabbit PAP (DAKO) at 1:100 for 30 min and developed with DAB and the sections were eosin-counterstained. Representative images of the sections were captured using Image Pro Express software on an Olympus microscope.
For detection of hypoxia by staining for pimonidazole adducts, mice were injected intraperitoneally with 25 mg/kg pimonidazole (Hypoxyprobe™, Chemicon International Inc., Temecula, CA, USA) 90 min before killing. Tumours were formalin-fixed and paraffin-embedded. Sections (3 μm) were processed for antigen retrieval and endogenous peroxidase was blocked, as above. Avidin/biotin blocking and staining were carried out using a mouse-on-mouse immunodetection kit (Vector® MOM™ peroxidase, Vector Laboratories Inc., Burlingame, CA, USA). Murine IgG was blocked and sections were incubated for 1 h at room temperature with 1:100 anti-pimonidazole antibody (Hypoxyprobe™, Chemicon International Inc.). Sections were subsequently washed and incubated with MOM™ biotinylated anti-mouse IgG for 10 min, followed by amplification of the signal with Vectastain ABC for 5 min and development with DAB.
Vaupel P, Thews O, Hoeckel M . Treatment resistance of solid tumors: role of hypoxia and anemia. Med Oncol 2001; 18: 243–259.
Skinner SA, Frydman GM, O'Brien PE . Microvascular structure of benign and malignant tumors of the colon in humans. Dig Dis Sci 1995; 40: 373–384.
Grunt TW, Lametschwandtner A, Karrer K, Staindl O . The angioarchitecture of the Lewis lung carcinoma in laboratory mice (a light microscopic and scanning electron microscopic study). Scan Electron Microsc 1986 (pt 2): 557–573.
Trotter MJ, Chaplin DJ, Durand RE, Olive PL . The use of fluorescent probes to identify regions of transient perfusion in murine tumors. Int J Radiat Oncol Biol Phys 1989; 16: 931–934.
Kimura H et al. Fluctuations in red cell flux in tumor microvessels can lead to transient hypoxia and reoxygenation in tumor parenchyma. Cancer Res 1996; 56: 5522–5528.
Helmlinger G, Yuan F, Dellian M, Jain RK . Interstitial pH and pO2 gradients in solid tumors in vivo: high-resolution measurements reveal a lack of correlation. Nat Med 1997; 3: 177–182.
Gray LH et al. The concentration of oxygen dissolved in tissues at the time of irradiation as a factor in radiotherapy. Br J Radiol 1953; 26: 638–648.
Durand RE, Aquino-Parsons C . Non-constant tumour blood flow – implications for therapy. Acta Oncol 2001; 40: 862–869.
Epstein AC et al. C. elegans EGL-9 and mammalian homologs define a family of dioxygenases that regulate HIF by prolyl hydroxylation. Cell 2001; 107: 43–54.
Bruick RK, McKnight SL . A conserved family of prolyl-4-hydroxylases that modify HIF. Science 2001; 294: 1337–1340.
Berra E et al. HIF prolyl-hydroxylase 2 is the key oxygen sensor setting low steady-state levels of HIF-1alpha in normoxia. EMBO J 2003; 22: 4082–4090.
Ivan M et al. HIFalpha targeted for VHL-mediated destruction by proline hydroxylation: implications for O2 sensing. Science 2001; 292: 464–468.
Jaakkola P et al. Targeting of HIF-alpha to the von Hippel–Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science 2001; 292: 468–472.
Masson N et al. Independent function of two destruction domains in hypoxia-inducible factor-alpha chains activated by prolyl hydroxylation. EMBO J 2001; 20: 5197–5206.
Jeong JW et al. Regulation and destabilization of HIF-1alpha by ARD1-mediated acetylation. Cell 2002; 111: 709–720.
Lando D et al. Asparagine hydroxylation of the HIF transactivation domain a hypoxic switch. Science 2002; 295: 858–861.
Mavria G, Porter CD . Reduced growth in response to ganciclovir treatment of subcutaneous xenografts expressing HSV-tk in the vascular compartment. Gene Therapy 2001; 8: 913–920.
Mavria G, Harrington KJ, Marshall CJ, Porter CD . In vivo efficacy of HSV-TK transcriptionally targeted to the tumour vasculature is augmented by combination with cytotoxic chemotherapy. J Gene Med 2004; published online 15 Nov 2004, doi:10.1002/jgm.662.
Richardson TB, Kaspers J, Porter CD . Retroviral hybrid LTR vector strategy: functional analysis of LTR elements and generation of endothelial cell specificity. Gene Therapy 2004; 11: 775–783.
Mavria G, Jager U, Porter CD . Generation of a high titre retroviral vector for endothelial cell-specific gene expression in vivo. Gene Therapy 2000; 7: 368–376.
Jain RK . The Eugene M. Landis Award Lecture 1996. Delivery of molecular and cellular medicine to solid tumors. Microcirculation 1997; 4: 1–23.
Boast K et al. Characterization of physiologically regulated vectors for the treatment of ischemic disease. Hum Gene Ther 1999; 10: 2197–2208.
Yuan Y, Hilliard G, Ferguson T, Millhorn DE . Cobalt inhibits the interaction between hypoxia-inducible factor-alpha and von Hippel–Lindau protein by direct binding to hypoxia-inducible factor-alpha. J Biol Chem 2003; 278: 15911–15916.
Black ME, Kokoris MS, Sabo P . Herpes simplex virus-1 thymidine kinase mutants created by semi-random sequence mutagenesis improve prodrug-mediated tumor cell killing. Cancer Res 2001; 61: 3022–3026.
Prentice H et al. Regulated expression of a foreign gene targeted to the ischaemic myocardium. Cardiovasc Res 1997; 35: 567–574.
Patterson AV et al. Oxygen-sensitive enzyme-prodrug gene therapy for the eradication of radiation-resistant solid tumours. Gene Therapy 2002; 9: 946–954.
Brizel DM et al. Tumor hypoxia adversely affects the prognosis of carcinoma of the head and neck. Int J Radiat Oncol Biol Phys 1997; 38: 285–289.
Fyles AW et al. Oxygenation predicts radiation response and survival in patients with cervix cancer. Radiother Oncol 1998; 48: 149–156.
Zagzag D et al. Expression of hypoxia-inducible factor 1alpha in brain tumors: association with angiogenesis, invasion, and progression. Cancer 2000; 88: 2606–2618.
Graeber TG et al. Hypoxia-mediated selection of cells with diminished apoptotic potential in solid tumours. Nature 1996; 379: 88–91.
Hockel M et al. Association between tumor hypoxia and malignant progression in advanced cancer of the uterine cervix. Cancer Res 1996; 56: 4509–4515.
Cosset FL et al. High-titer packaging cells producing recombinant retroviruses resistant to human serum. J Virol 1995; 69: 7430–7436.
O'Hare MJ et al. Conditional immortalization of freshly isolated human mammary fibroblasts and endothelial cells. Proc Natl Acad Sci USA 2001; 98: 646–651.
Porter C . Retroviral vectors for suicide gene therapy. Methods Mol Med 2004; 90: 91–106.
We thank Demelza Bird for her assistance with the in vivo work and the Breast Cancer Pathology Core Unit for the processing, embedding and sectioning of the tumours. The TK SR39 cDNA is proprietary to Celltech R&D Inc.; we are grateful to Margaret Black and Celltech for the provision of this material. We are grateful to Mike O'Hare and the Ludwig Institute for Cancer Research for providing the immortalized fibroblast and endothelial cell cultures, HMFD and HMME7, and to the Centers for Disease Control and Prevention, Atlanta for the HMEC-1 cell line. This work was funded by the Institute of Cancer Research.
About this article
Cite this article
Ingram, N., Porter, C. Transcriptional targeting of acute hypoxia in the tumour stroma is a novel and viable strategy for cancer gene therapy. Gene Ther 12, 1058–1069 (2005). https://doi.org/10.1038/sj.gt.3302504
- tumour stroma
- cancer gene therapy
- transcriptional targeting
Adenovirus-mediated hypoxia-targeted gene therapy using HSV thymidine kinase and bacterial nitroreductase prodrug-activating genes in vitro and in vivo
Cancer Gene Therapy (2011)
Role of cell surface molecules and autologous ascitic fluid in determining efficiency of adenoviral transduction of ovarian cancer cells
Cancer Gene Therapy (2010)
Complement insufficiency limits efficacy in a xenograft model of hyperacute rejection for cancer therapy
Cancer Immunology, Immunotherapy (2007)