Hypoxia that develops in solid tumors stabilizes the hypoxia-inducible factor-1α (HIF-1α) subunit of the HIF-1 transcription factor, leading to up-regulation of dozens of hypoxia-regulated genes that increase glycolysis and oxygen delivery. HIF-1α and its downstream target gene CA9 have both been used as surrogate hypoxia markers, and, in general, high expression predicts for a poor response to treatment. Combinations of hypoxia markers offer the opportunity to measure changes in tumor oxygenation that may be relevant to tumor response to treatment. We compared the degree of colocalization of two endogenous markers for hypoxia, HIF-1α and carbonic anhydrase IX (CAIX), with a chemical marker for hypoxia, pimonidazole. Unexpectedly, expression of HIF-1α was reduced in the most hypoxic regions that border necrosis in xenograft tumors composed of SiHa cervical carcinoma, WiDr colon carcinoma, or M006 astrocytoma cells. Similar results were obtained for samples from three cervical cancer biopsies. However, CAIX was present in these perinecrotic cells that were also capable of metabolizing and binding a chemical marker for hypoxia, pimonidazole. In vitro experiments using tumor cells and tumor cubes incubated under anoxic conditions indicated that nutrient deprivation seems to be largely responsible for the lack of HIF-1α expression in perinecrotic regions. The half-life of CAIX was sufficiently long that, once formed, it remained for days in the absence of continued HIF-1α expression. These results have implications for the use of HIF-1α as an indicator of tumor hypoxia and aggressiveness as well as development of hypoxia-directed antitumor therapies based on the expression of HIF-1α.

Development of hypoxia in solid tumors is associated with tumor aggressiveness, metastasis, and recurrence following treatment (1). Recently, endogenous markers of hypoxia have been described that have the potential to help identify those tumors likely to respond poorly to treatment (2, 3). These markers offer an alternative for those tumors that are inaccessible to oxygen needle microelectrodes. In addition, endogenous markers do not necessitate an additional biopsy beyond that required for diagnosis and or treatment. However, questions remain concerning the hypoxia specificity of protein marker expression, whether one marker is likely to be superior to others and whether combinations of endogenous markers could be advantageous for detecting transient and chronic forms of tumor hypoxia (46).

The extent of colocalization between endogenous hypoxia markers and chemical hypoxia markers has been examined using human tumor biopsies (68). There is typically good colocalization, however, in most previous studies; sequential tumor sections were examined so that the extent of colocalization for individual tumor cells could not be determined. Analysis of colocalization of hypoxia markers using multiple fluorophor-conjugated antibodies can be used to address this question (810), although such studies are best done using fresh frozen sections.

The oxygen dependency of hypoxia-inducible factor-1 (HIF-1) transcription factor is controlled by the rapid stabilization of its HIF-1α subunit in response to decreases in cellular oxygen concentrations (11). Under aerobic conditions, HIF-1α is rapidly degraded (12). A downstream target gene of HIF-1, carbonic anhydrase 9, codes for a membrane-bound enzyme involved in acid/base regulation and ion balance. Carbonic anhydrase IX (CAIX) levels are increased under hypoxic conditions primarily through binding of HIF-1α to the hypoxia response element of the promoter and to the presence of the SP1/SP3 binding element in the promoter region (13). CAIX loss upon reoxygenation is slow with a reported half-life of 2 to 3 days (14, 15), so it is an indicator that hypoxia that has been present for a sufficiently long period to induce CAIX expression, but it may not still be present. In contrast, HIF-1α is an indicator of the current oxygen status because it is rapidly stabilized under hypoxia and rapidly degraded when oxygen is available (12). Chemical hypoxia markers, like pimonidazole, are metabolically reduced under hypoxia to produce adducts that can be detected using antibodies (16). They are reliable indicators of the tumor oxygenation in the recent past. Like CAIX, adducts of pimonidazole have been reported to be long-lived (17), but like HIF-1α, they can be detected within a few minutes of administration. Combining these markers in analysis of clinical biopsy material could allow confirmation of the presence and frequency of regions undergoing transient changes in perfusion and may be useful in examining tumor response to therapy. Toward this goal, patterns of hypoxia marker binding were examined in human tumor xenografts and in pretreatment cervical cancer biopsies. Quantitative image analysis methods were used to measure hypoxia marker colocalization.

Pimonidazole administration to patients. Three patients with invasive cervical cancer who were part of a clinical trial evaluating pretreatment tumor hypoxia were administered pimonidazole hydrochloride (Chemicon International, Inc., Temecula, CA) 18 to 20 hours before incisional tumor biopsy. This protocol and patient characteristics have been previously described (6). This protocol was approved by the local University of British Columbia ethics board. The biopsies were immediately (within 30 seconds) placed into ice-cold saline before freezing for sectioning or disaggregation and single-cell preparation for analysis of pimonidazole binding by flow cytometry.

Cell culture. SiHa cervical cancer cells and WiDr human colon carcinoma cells were obtained from American Type Culture Collection (Manassas, VA) and grown as monolayers with twice weekly subcultivation in MEM containing 10% fetal bovine serum (FBS). M006 human astrocytoma cells were obtained from Dr. Alan Franko (Cross Cancer Institute, Edmonton AB) and maintained similarly. Single-cell suspensions were incubated at 2 × 105 to 4 × 105 cells/mL in glass spinner culture flasks (Bellco, Vineland, NJ). For anoxic conditions, spinner flasks were gassed continuously with 95% nitrogen 5% CO2 for 1 hour before introducing cells and during incubation. For experiments using pimonidazole hydrochloride, single cells were incubated with 2 to 50 μg/mL pimonidazole in complete medium. Tumor cubes were prepared from SiHa xenografts 90 minutes after mice had received pimonidazole and Hoechst 33342. Cubes were prepared as previously described (18) and incubated under anoxic conditions in either serum- and glucose-free MEM or in complete medium containing 2% FBS.

Immunoblotting. After incubation to determine time for HIF-1α development under anoxia, cells were rapidly pipetted into buffer containing 100 μmol/L cobalt chloride to inhibit HIF-1α degradation. Cells were centrifuged and immediately lysed to limit exposure to cobalt chloride to <5 minutes. Cobalt chloride was not used in experiments that examined HIF-1α loss upon reoxygenation. Equal amounts of protein were loaded onto a 10% polyacrylamide precast gel (Bio-Rad, Mississauga, ON, Canada) followed by electrophoresis and transfer to a nitrocellulose membrane (Bio-Rad). The membrane was blocked using 5% nonfat dry milk in TBS-T [25 mmol/L Tris-HCl (pH 7.04), NaCl, KCl, 0.05% Tween 20] for 1 hour at room temperature with shaking. Membranes were next incubated in anti-mouse monoclonal HIF-1α antibody (1:1,000 dilution; BD Transduction Laboratories, Lexington, VA) overnight. After several washes in TBS-T, membranes were incubated in horseradish peroxidase anti-mouse antibody (1:5,000 dilution; Sigma, Oakville, ON, Canada) for 1 hour and developed using the ECL detection kit (Amersham, Oakville, ON, Canada). Membranes were exposed under the chemiluminescence imager (MultiImage Light Cabinet, Alpha Innotech Corp., CA) for 4 to 10 minutes. Images were analyzed for relative intensity using FluorChem software (v.3.04A, Alpha Innotech).

Analysis of pimonidazole and carbonic anhydrase IX binding by flow cytometry. SiHa cells in glass spinner culture flasks were incubated under anoxia for various times or returned to normoxic conditions after incubation under anoxia. Approximately 5 × 105 cells were fixed in 70% ethanol in PBS. Cells were rehydrated in PST (PBS with 4% FBS and 0.1% Triton X-100) for 15 minutes before incubation in antipimonidazole antibody (Chemicon, 1:200 dilution) for 1 hour. Alternatively, cells were incubated with anti-CAIX antibody (1:5,000 as previously described; ref. 6) for 60 minutes followed by incubation in Alexa-488 anti-mouse antibody (1:200; Molecular Probes, Eugene, OR) for 45 minutes with shaking. Cells were again centrifuged, washed twice in PBS, and incubated in 4′,6-diamidino-2-phenylindole dihydrochloride hydrate (1:50 dilution of 100 μg/mL stock in distilled water) for nuclear staining. Cells were then analyzed using a dual laser Epics Elite-ESP flow cytometer (Coulter Corp., Hialeah, FL). Gates were set to discriminate against debris on the basis of cell size and time of flight. List mode files were collected and post processed to determine parameters of interest using the WINLIST software package (Verity Software House, Inc., Topsham, ME).

Growth of xenograft tumors. Xenograft tumors were grown s.c. in the flanks of female nonobese diabetic severe combined immunodeficient mice that were bred and maintained in our facility in accordance with guidelines from the Canadian Council on Animal Care. The Animal Care Committee of the University of British Columbia approved protocols used in this study. Tumors were used when they reached a size a 0.4 to 0.6 g after ∼3 to 4 weeks.

Mice were injected i.p. with 100 mg/kg pimonidazole hydrochloride that was metabolized and bound to hypoxic cells within the tumor (19). At subsequent times after pimonidazole injection, mice were injected i.v. via the tail vein with 0.1 mL Hoechst 33342 (4 mg/mL stock; Sigma) to label nuclei of cells close to perfused tumor blood vessels. Ten minutes later, mice were sacrificed and tumors were removed within 30 seconds for embedding at −20°C in Tissue-TEK OCT embedding medium (Sakura Finetek, Torrance, CA).

Tumor sectioning and staining. Once frozen (∼20 minutes), 5 to 7 μm sections were prepared using a Cryostar HM560 cryostat (Microm International, GmbH, Walldorf, Germany). Frozen sections were air dried for 30 seconds and transferred to 2% paraformaldehyde for fixation for 15 minutes. A rapid staining procedure was used to preserve Hoechst 33342 localization within the same slide. Slides were dipped sequentially in PBS and methanol at −20°C with a final 5 minutes block of nonspecific binding in PTN (PBS containing 1% bovine serum albumin and 0.2% Tween 20). The area surrounding the tumor section was dried and 50 μL primary antibody (monoclonal anti–HIF-1α antibody from BD Transduction Laboratories, Oakville, ON, Canada) at a 1:100 dilution in PTN, or monoclonal anti-CAIX antibody at a 1:5,000 dilution (6), was placed over the section. The area was covered with a parafilm and the slide was placed in a humidified box for 30 minutes at ambient temperature. After rinsing, the slide was covered with 100 μL secondary fluorophor-conjugated antibody [1:100 dilution in PTN, Alexa 549–conjugated goat anti-mouse F(ab')2 fragment antibody from Molecular Probes] for 15 minutes at ambient temperature. After rinsing in PBS, the slides were incubated for 5 minutes in paraformaldehyde before proceeding with the next primary antibody. The second primary antibody typically detected pimonidazole adducts using FITC-conjugated antipimonidazole antibodies, 1:1,000 dilution, in PTN for 30 minutes. Alternatively, incubation with CAIX antibody for 30 minutes was followed by incubation for 15 minutes in a 1:100 dilution in PTN of Alexa 488–conjugated goat anti-mouse F(ab')2 fragment antibody from Molecular Probes. Finally, 10 μL Fluorogard mounting medium (Bio-Rad) was added and the section was covered with a cover slip. Slides were viewed immediately.

For patient samples, sequential sections were incubated with FITC-conjugated antipimonidazole antibodies and either mouse antihuman CD31 (1:50 dilution, DakoCytomation, Mississauga, ON, Canada) or HIF-1α antibodies. Resulting images were superimposed using pimonidazole binding patterns to define the correct orientation. Only the sections stained for both pimonidazole and HIF-1α were used for quantitative analysis.

Image analysis. Slides were viewed with 488 nm excitation, 594 nm excitation, and UV excitation using a 10× neofluor objective with a Zeiss Axioplan epifluorescence microscope. Three-color images were captured using an image-intensified CCD camera and an ITEX memory board (Imaging Technology, Inc., Bedford, MA). Thresholded images for HIF-1α, CAIX, and pimonidazole antibody-stained areas were prepared using Photoshop and analyzed using Scion imaging software. Software developed for Northern Eclipse 5.0 software (Empix, Toronto, ON, Canada) was also used to determine pixel colocalization. To correct for the nuclear staining of HIF-1α and membrane staining of CAIX, image resolution (typically 1 pixel/μm) was reduced 10-fold and contrast was enhanced before overlaying on the pimonidazole thresholded images. The fraction of total pixels that stained for pimonidazole only, HIF-1α only, pimonidazole and HIF-1α, or pimonidazole in the absence of HIF-1α were calculated. Similar comparisons were made between pimonidazole and CAIX and for CAIX and HIF-1α. The ratio of each region to the total cellular area of the image was then calculated. Three to five images were analyzed from three to six tumors (n = 15-30) and results were pooled to determine the mean and standard deviation. Significance was determined using ANOVA.

Results using three human cervical cancer biopsies provided the basis for subsequent experiments done with human tumor xenografts. Patient biopsies were collected in ice-cold saline, rapidly frozen and sectioned, then examined for pimonidazole binding in relation to HIF-1α expression (Fig. 1A). A separate sample from the same tumor was disaggregated and single cells were analyzed for pimonidazole binding by flow cytometry (Fig. 1B). As expected based on their differences in oxygen dependencies of binding or expression, antibodies against HIF-1α were identified closer to the CD31-labeled blood vessels than antibodies against pimonidazole adducts. On average, 33% of the pixels were positive for HIF-1α but only 18% exhibited pimonidazole. Results using flow cytometry of single cells from disaggregated tumors (Fig 1B) or analysis of results from frozen sections from the same tumors (Fig. 1C) indicated comparable fractions of pimonidazole-positive cells. Unexpectedly, HIF-1α expression was significantly reduced in cells that stained for pimonidazole but were adjacent to necrotic regions. For these three tumors, 24% to 38% of the pimonidazole-labeled pixels failed to express HIF-1α. This fraction was associated predominantly with perinecrotic pimonidazole binding. To be able to metabolize pimonidazole, these cells must be intact and metabolically active. Experiments were, therefore, done with single cells in vitro and with xenograft tumor in mice to confirm the lack of HIF-1α staining in perinecrotic regions and attempt to undercover a mechanism.

Figure 1.

Expression of HIF-1α and binding of pimonidazole in pretreatment human cervical cancer biopsies. Patients were administered 0.5 g/m2 pimonidazole hydrochloride as a 20-minute i.v. infusion 18 to 20 hours before tumor biopsy. Single-cell suspensions were prepared for analysis of pimonidazole binding by flow cytometry (B). A and C, frozen sections were also prepared and analyzed for the distributions of pimonidazole (green), CD31 (blue), and HIF-1α (red). A, “N” indicates regions of necrosis, distinct from perivascular unstained regions. C, analysis of pimonidazole and HIF-1α–positive pixels from each tumor; columns, mean; bars, SD (n = 4). The pimonidazole-positive fraction determined by flow cytometry in (B) is shown for comparison.

Figure 1.

Expression of HIF-1α and binding of pimonidazole in pretreatment human cervical cancer biopsies. Patients were administered 0.5 g/m2 pimonidazole hydrochloride as a 20-minute i.v. infusion 18 to 20 hours before tumor biopsy. Single-cell suspensions were prepared for analysis of pimonidazole binding by flow cytometry (B). A and C, frozen sections were also prepared and analyzed for the distributions of pimonidazole (green), CD31 (blue), and HIF-1α (red). A, “N” indicates regions of necrosis, distinct from perivascular unstained regions. C, analysis of pimonidazole and HIF-1α–positive pixels from each tumor; columns, mean; bars, SD (n = 4). The pimonidazole-positive fraction determined by flow cytometry in (B) is shown for comparison.

Close modal

Initial studies were done in vitro to examine the kinetics of development of marker binding or expression under anoxia and loss upon reoxygenation. SiHa cells incubated under anoxic conditions developed HIF-1α rapidly, reaching a half-maximal level within 2 hours. Loss half-time after return to normoxia was <2 minutes (Fig. 2A and B). CAIX developed much more slowly, and loss half-time after return to normoxia was ∼2 days, consistent with CAIX loss upon SiHa cell doubling. Detection of pimonidazole-positive cells was possible after only a few minutes of exposure of cells under anoxia; half-maximal expression was observed after an hour. Loss rate of pimonidazole adducts from SiHa cells was also slow. Table 1 lists the development and loss half-times for these three hypoxia markers in SiHa cells.

Figure 2.

Kinetics of development of expression or binding of hypoxia markers under anoxia and loss upon reoxygenation. A, SiHa cells were incubated in suspension culture under continuous gassing with 95% N2 and 5% CO2. Development of HIF-1α was determined by immunoblotting of protein lysates at various times after incubation. CAIX and pimonidazole expressions were measured by flow cytometry after antibody labeling. B, after 4-hour incubation under anoxia, single cells were returned to normoxic conditions. Samples were examined for pimonidazole (○) or CAIX (▵) binding by flow cytometry, or for HIF-1α (•) expression by immunoblotting. Results from two or more experiments were combined after normalization (minimum expression = 0, maximum expression = 1).

Figure 2.

Kinetics of development of expression or binding of hypoxia markers under anoxia and loss upon reoxygenation. A, SiHa cells were incubated in suspension culture under continuous gassing with 95% N2 and 5% CO2. Development of HIF-1α was determined by immunoblotting of protein lysates at various times after incubation. CAIX and pimonidazole expressions were measured by flow cytometry after antibody labeling. B, after 4-hour incubation under anoxia, single cells were returned to normoxic conditions. Samples were examined for pimonidazole (○) or CAIX (▵) binding by flow cytometry, or for HIF-1α (•) expression by immunoblotting. Results from two or more experiments were combined after normalization (minimum expression = 0, maximum expression = 1).

Close modal
Table 1.

Development and loss of hypoxia markers

Hypoxia markerTime to half-maximum (mean ± SD)Loss half-time (mean ± SD)
Pimonidazole 1.2 ± 0.5 h 1.4 ± 0.2 d 
CAIX 16.8 ± 1.2 h 1.9 ± 0.8 d 
HIF-1α 2.2 ± 0.9 h 1.5 ± 0.6 min 
Hypoxia markerTime to half-maximum (mean ± SD)Loss half-time (mean ± SD)
Pimonidazole 1.2 ± 0.5 h 1.4 ± 0.2 d 
CAIX 16.8 ± 1.2 h 1.9 ± 0.8 d 
HIF-1α 2.2 ± 0.9 h 1.5 ± 0.6 min 

SiHa cervical carcinoma xenografts were compared for patterns of antibody labeling to CAIX, HIF-1α, and pimonidazole adducts (Fig. 3). As expected based on the reported oxygen dependencies of expression, the percentage of the tumor that stained for HIF-1α or CAIX was significantly larger than the percentage that was able to metabolize and bind pimonidazole. Again, HIF-1α was significantly reduced in perinecrotic regions. Yet, in spite of the lack of HIF-1α expression, CAIX, a target gene of HIF-1α, was present in these regions distant from blood vessels. Note that few cells in these xenografts were HIF-1α positive but CAIX negative, confirming that CAIX expression is associated with HIF-1α stabilization. The presence of CAIX in perinecrotic regions can be explained by the long CAIX lifetime (Fig. 2B). Although no new CAIX is produced in these regions in the absence of HIF-1α expression, residual CAIX is still present. Few cells expressed pimonidazole in the absence of CAIX. The greatest degree of mismatch in marker binding was observed between CAIX and HIF-1α.

Figure 3.

Comparison of patterns of binding of three hypoxia markers and a perfusion marker in SiHa xenograft tumors. Mice with SiHa xenografts were injected i.p. with pimonidazole 90 minutes before injection of Hoechst 33342 to label cells surrounding perfused blood vessels (blue-stained areas). Ten minutes later, tumors were excised. Frozen sections were prepared and analyzed for combinations of three markers. Representative images show the nuclear staining of Hoechst 33342 and HIF-1α, the membranous staining of CAIX, and the whole cell staining pattern of pimonidazole. N, regions of necrosis; bar, 250 μm. A, HIF-1α (red), pimonidazole (green). B, HIF-1α (green), CAIX (red). C, CAIX (red), pimonidazole (green). Columns, means for ≥15 tumor sections (×10 magnification) from three to five tumors; bars, SD.

Figure 3.

Comparison of patterns of binding of three hypoxia markers and a perfusion marker in SiHa xenograft tumors. Mice with SiHa xenografts were injected i.p. with pimonidazole 90 minutes before injection of Hoechst 33342 to label cells surrounding perfused blood vessels (blue-stained areas). Ten minutes later, tumors were excised. Frozen sections were prepared and analyzed for combinations of three markers. Representative images show the nuclear staining of Hoechst 33342 and HIF-1α, the membranous staining of CAIX, and the whole cell staining pattern of pimonidazole. N, regions of necrosis; bar, 250 μm. A, HIF-1α (red), pimonidazole (green). B, HIF-1α (green), CAIX (red). C, CAIX (red), pimonidazole (green). Columns, means for ≥15 tumor sections (×10 magnification) from three to five tumors; bars, SD.

Close modal

The presence of pimonidazole-positive, HIF-1α–negative cells in SiHa xenografts was confirmed in WiDr colon carcinoma xenografts and M006 astrocytoma xenografts that have higher and lower hypoxic fractions, respectively, compared with SiHa xenografts (Fig. 4). Although more of the cells of WiDr xenografts bound pimonidazole than SiHa or M006 xenografts, the fraction of the tumor cells that expressed HIF-1α was not significantly different for the three types of tumors. The average fraction of pimonidazole-positive cells that did not express HIF-1α was 0.25 for the M006, 0.20 for SiHa, and 0.25 for WiDr xenografts. Although most of the mismatch in staining could be attributed to the lack of HIF-1α expression in perinecrotic regions, rare areas (<5% of the images) contained regions positive for both pimonidazole and Hoechst 33342. These areas are likely to represent transient changes in perfusion that occurred between the time of pimonidazole administration and Hoechst 33342 administration 90 minutes later.

Figure 4.

Comparison between HIF-1α expression and pimonidazole binding for three xenograft tumors. A diagram illustrates the position of the various labeled populations relative to a functional Hoechst 33342–labeled blood vessel (BV). Columns, means for analysis of ≥30 sections from three to five tumors; bars, SD.

Figure 4.

Comparison between HIF-1α expression and pimonidazole binding for three xenograft tumors. A diagram illustrates the position of the various labeled populations relative to a functional Hoechst 33342–labeled blood vessel (BV). Columns, means for analysis of ≥30 sections from three to five tumors; bars, SD.

Close modal

To examine possible reasons for the lack of HIF-1α expression in perinecrotic regions, HIF-1α expression was first examined as a function of oxygen tension for SiHa cells incubated in vitro. SiHa cells showed increasing levels of HIF-1α for up to 12 hours, so anoxic conditions alone do not seem to be responsible (Fig. 2A). The oxygen dependency of HIF-1α expression for SiHa cells incubated for 4 hours under various oxygen gassing conditions confirms that maximum HIF-1α levels occur in the absence of oxygen, at least when nutrients are not limiting (Fig. 5A). Also, when mice were made to breathe 60% oxygen to increase the oxygen diffusion distance into the tumor without affecting the nutrient diffusion distance, expression of HIF-1α was still not observed in these perinecrotic regions (data not shown). Therefore, factors other than anoxia must be responsible for the lack of HIF-1α staining in these regions.

Figure 5.

Oxygen and glucose dependency of HIF-1α expression in SiHa cells. A, SiHa cells were incubated for 4 hours under various oxygen gassing conditions before analysis of HIF-1α expression by immunoblotting using actin as a loading control. B, SiHa cells were incubated under anoxic conditions for 4 hours in the presence or absence of glucose and serum before lysis and analysis for HIF-1α. Columns, means for three independent experiments; bars, SE.

Figure 5.

Oxygen and glucose dependency of HIF-1α expression in SiHa cells. A, SiHa cells were incubated for 4 hours under various oxygen gassing conditions before analysis of HIF-1α expression by immunoblotting using actin as a loading control. B, SiHa cells were incubated under anoxic conditions for 4 hours in the presence or absence of glucose and serum before lysis and analysis for HIF-1α. Columns, means for three independent experiments; bars, SE.

Close modal

When incubation of SiHa cells was conducted for 4 hours under anoxia in glucose- and serum-free medium, expression of HIF-1α was reduced ∼10-fold (Fig. 5B). This result suggests that nutrient deprivation prevents HIF-1α expression and could be responsible for the lack of HIF-1α expression in perinecrotic regions of solid tumors. To test this concept in vivo, 1- to 2-mm-sided tumor cubes were prepared from WiDr tumors and incubated for 2 hours in MEM containing 2% FBS equilibrated with 95% nitrogen and 5% CO2. Cubes were also incubated under anoxia in MEM lacking both glucose and serum. After incubation, cubes were rapidly frozen, sectioned, and analyzed for HIF-1α expression in relation to pimonidazole binding and Hoechst staining. In complete medium under anoxia, the edges of the tumor cubes showed higher expression of HIF-1α than the internal regions, consistent with greater nutrient availability at the surface of the cubes (Fig. 6A and C). However, HIF-1α levels were reduced or absent in pimonidazole-labeled regions at the surface of cubes incubated in medium lacking both glucose and serum (Fig. 6B and D). These results support the theory that inadequate delivery of nutrients leads to lack of HIF-1α expression in perinecrotic regions. However, when these cells are supplied with nutrients, they are capable of expressing HIF-1α.

Figure 6.

Expression of HIF-1α in tumor cubes prepared from SiHa xenografts. SiHa xenografts were excised 90 minutes after i.p. injection of mice with pimonidazole (green) and 10 minutes after i.v. injection of Hoechst 33342 (blue). Two-millimeter-sided cubes were prepared and transferred to pre-equilibrated spinner culture flasks for incubation in medium containing 2% serum or in medium lacking both serum and glucose. After 2-hour incubation under anoxic conditions, cubes were frozen, sectioned, and analyzed for the expression of HIF-1α (red) in pimonidazole-stained regions at the edges of the cubes. Analysis using NIH ImageJ software shows the intensity of pimonidazole (thin line) and HIF-1α (heavy line) through regions of the tumor cubes that exhibit pimonidazole labeling at the cube surface. A, cube after incubation under anoxia in complete medium, indicating high HIF-1α expression at the cube surface in the pimonidazole-stained area. B, cube after incubation under anoxia in serum- and glucose-free medium. Note the decrease in HIF-1α staining intensity at the cube edge. C, representative image of a cube incubated in complete medium under anoxia showing high levels of HIF-1α expression at the surface of the cube (red nuclear staining), independent of pimonidazole labeling (green). D, representative image of a cube incubated under anoxia in the absence of glucose and serum indicating lack of HIF-1α staining at the cube surface in regions stained for pimonidazole. Rectangle (C and D), regions analyzed in (A) and (B); arrows, edge of the cube; bar, 250 μm.

Figure 6.

Expression of HIF-1α in tumor cubes prepared from SiHa xenografts. SiHa xenografts were excised 90 minutes after i.p. injection of mice with pimonidazole (green) and 10 minutes after i.v. injection of Hoechst 33342 (blue). Two-millimeter-sided cubes were prepared and transferred to pre-equilibrated spinner culture flasks for incubation in medium containing 2% serum or in medium lacking both serum and glucose. After 2-hour incubation under anoxic conditions, cubes were frozen, sectioned, and analyzed for the expression of HIF-1α (red) in pimonidazole-stained regions at the edges of the cubes. Analysis using NIH ImageJ software shows the intensity of pimonidazole (thin line) and HIF-1α (heavy line) through regions of the tumor cubes that exhibit pimonidazole labeling at the cube surface. A, cube after incubation under anoxia in complete medium, indicating high HIF-1α expression at the cube surface in the pimonidazole-stained area. B, cube after incubation under anoxia in serum- and glucose-free medium. Note the decrease in HIF-1α staining intensity at the cube edge. C, representative image of a cube incubated in complete medium under anoxia showing high levels of HIF-1α expression at the surface of the cube (red nuclear staining), independent of pimonidazole labeling (green). D, representative image of a cube incubated under anoxia in the absence of glucose and serum indicating lack of HIF-1α staining at the cube surface in regions stained for pimonidazole. Rectangle (C and D), regions analyzed in (A) and (B); arrows, edge of the cube; bar, 250 μm.

Close modal

Another important factor influencing the degree of colocalization between HIF-1α and pimonidazole is the time between pimonidazole administration and tumor biopsy. In the previous experiments, mice were injected with pimonidazole 90 minutes before tumor biopsy. However, when the time between pimonidazole administration and biopsy was extended to 24 and 48 hours, the discrepancy between HIF-1α–stained regions and pimonidazole-stained regions increased dramatically (Fig. 7). Pimonidazole-labeled regions were clearly associated with necrotic regions at these later times and these areas exhibited a more honeycomb appearance indicating loss of nuclear staining. Hypoxic cells advanced into these regions probably as a result of tumor cord expansion of proliferating cells closer to blood vessels. Although 80% of the pimonidazole-positive regions also expressed HIF-1α when examined 90 minutes after pimonidazole injection, 48 hours later only 32% of the pimonidazole-labeled regions colocalized with HIF-1α (Fig. 7B). The percentage of regions that expressed both pimonidazole and HIF-1α was significantly reduced at 24 and 48 hours compared with 90 minutes. Therefore, HIF-1α expression in pimonidazole-stained regions decreased as time between administration of pimonidazole and tumor biopsy increase.

Figure 7.

Movement of pimonidazole-labeled cells into necrotic regions. SiHa tumor sections were stained and analyzed for the degree of colocalization between HIF-1α and pimonidazole as a function of time after pimonidazole injection. A, representative SiHa xenograft tumor examined for HIF-1α (red), pimonidazole (green), and Hoechst 33342 (blue) 48 hours after pimonidazole injection. Note that the necrotic regions (N) show extensive pimonidazole labeling. B, fractions of total pixels for various combinations of markers are shown as a function of time after pimonidazole injection. Note that the decrease in HIF-1α–stained regions colocalized with pimonidazole is not a result of a change in either HIF-1α expression or pimonidazole binding over this period.

Figure 7.

Movement of pimonidazole-labeled cells into necrotic regions. SiHa tumor sections were stained and analyzed for the degree of colocalization between HIF-1α and pimonidazole as a function of time after pimonidazole injection. A, representative SiHa xenograft tumor examined for HIF-1α (red), pimonidazole (green), and Hoechst 33342 (blue) 48 hours after pimonidazole injection. Note that the necrotic regions (N) show extensive pimonidazole labeling. B, fractions of total pixels for various combinations of markers are shown as a function of time after pimonidazole injection. Note that the decrease in HIF-1α–stained regions colocalized with pimonidazole is not a result of a change in either HIF-1α expression or pimonidazole binding over this period.

Close modal

Hypoxia stabilizes HIF-1α, leading to up-regulation of >70 genes that aid in adaptation to hypoxia (20). Both HIF-1α and its downstream targets are indicators of the presence of hypoxia within solid tumors leading to the proposal that HIF-1α would be a good therapeutic target (21). In sections from SiHa, WiDr, and M006 xenografts, 75% to 80% of pixels positive for pimonidazole were also positive for HIF-1α. However, only 40% to 60% of the HIF-1α–labeled regions contained cells with adducts to pimonidazole. Lack of complete colocalization was anticipated for several reasons. First, pimonidazole adducts are formed only below 1% O2 with half-maximal binding around 0.1% O2 (22), whereas CAIX and HIF-1α are expressed below ∼2% O2 (12). Results in Fig. 5 indicate that 2% oxygen gassing may produce as much as 30% of the maximum signal in SiHa cells, and previous results have shown that the oxygenation of cells bordering blood vessels in SiHa xenografts is close to 2% (23). HIF-1α and CAIX might be viewed as more sensitive indicators of hypoxic microregions within tumors because they are expressed at higher levels of oxygen than those that allow pimonidazole metabolism and binding. However, the ability to distinguish differences in hypoxic fractions for the three xenografts was better for pimonidazole staining than HIF-1α expression (Fig. 4). In the few clinical studies where colocalization between two hypoxia markers has been compared, the degree of colocalization was generally good (2, 6, 7, 24), although one marker may predict outcome following therapy better than another (8).

Lack of complete marker colocalization can also occur because of differences in kinetics of marker binding or expression when cells become hypoxic or undergo reoxygenation. Although the HIF-1 transcription factor responds rapidly to changes in oxygenation, CAIX responds much more slowly. These differences can contribute to the lack of complete colocalization seen between HIF-1α and CAIX (25). They also suggest that patterns of hypoxia marker binding could be used to identify regions of tumors undergoing changes in perfusion. A motivation for this current study was to determine whether combinations of hypoxia markers could be used to identify regions undergoing transient changes in perfusion. However, in the SiHa xenograft tumor, regions that were labeled for both pimonidazole and Hoechst 33342 were rarely seen when only 90 minutes separated the injections of pimonidazole and Hoechst. Instead, there was a large variation from one tumor cord to another in the distance between Hoechst-stained perivascular regions and regions that expressed the hypoxia markers, a more subtle indicator of possible transient changes in perfusion in this tumor. These results are consistent with the finding that SiHa xenografts may undergo changes in perfusion but these typically occur over several hours (26). To eliminate transient changes in perfusion as a confounding factor in understanding the reason for lack of HIF-1α in perinecrotic regions, subsequent experiments were restricted to the SiHa tumor.

A third reason for lack of colocalization in the progression of pimonidazole-labeled or CAIX-expressing cells into necrotic regions as a result of continuing tumor growth. The time between administration of pimonidazole and analysis of hypoxic fraction was chosen based on plasma half-life of pimonidazole. Analysis 90 minutes after injection of pimonidazole is appropriate for a drug with a plasma half-life in rodents of 30 minutes (27). The 6-hour half-life in humans (28) was responsible for the choice of the 18- to 24-hour biopsy time post–pimonidazole administration. Reduced colocalization seen in the patient biopsies (25-38% of the pimonidazole labeled cells did not express HIF-1α) may be attributed in part to the long time between pimonidazole delivery and biopsy. Earlier biopsy times, such as that used by Nordsmark et al. (29), might be preferable to avoid loss of pimonidazole-labeled cells that are displaced into necrotic areas by proliferating tumor cells. The suggested use of an p.o. pimonidazole formulation (30) could also reduce this problem.

The decrease in the fraction of tumor cells expressing both HIF-1α and pimonidazole with time after pimonidazole administration seems to provide an indication of hypoxic cell lifetime. Rate of loss of pimonidazole was not useful in the absence of a definitive method of separating viable from necrotic cells. The rate of decrease in HIF-1α and pimonidazole colocalization with time after pimonidazole administration may also be a useful indicator of potential tumor doubling time as well as response to treatment. This possibility is currently being examined by comparing the change in HIF-1α colocalization with pimonidazole after cell cycle perturbations by anticancer treatments. The indication that HIF-1α expression is increased following exposure of tumors to ionizing radiation (31) may be related in part to posttreatment increases in nutrition of surviving perinecrotic cells.

A final reason for lack of complete colocalization is the reduction of HIF-1α expression in perinecrotic regions. This seems to be a result of insufficient nutrients. Discrepancies between HIF-1α and CAIX staining have been previously reported, with HIF-1α expression seen less frequently than CAIX (32). Another transcription factor identified as hypoxia-inducible, DEC1, was also found to be absent in regions of breast cancers that were positive for CAIX (33). However, DEC1 was also expressed by normal cells, including endothelial cells, macrophages, and pneumocytes. For HIF-1α, he differences observed in perinecrotic staining are fairly subtle and could go unnoticed when sequential sections are stained and compared. An internal ribosomal entry site in the 5′ untranslated region was shown to spare HIF-1α from the overall reduction in translation normally observed under hypoxia and it was shown to provide some protection against serum starvation (34). However, the duration of this protection seemed to be limited. Perinecrotic regions maintain the ability to metabolize and bind pimonidazole, consistent with a report indicating that low glucose levels have little effect on the binding of a closely related nitroimidazole, misonidazole (35). An important question is whether these pimonidazole and CAIX positive but HIF-1α negative tumor cells are clonogenic. Cell sorting experiments using the SiHa tumor model previously showed that CAIX-positive cells most distant from blood vessels remain clonogenic following tumor disaggregation (6), although loss of perinecrotic cells during the disaggregation procedure could not be ruled out. When mixtures of HIF-1α+/+ and HIF-1α−/− embryonic stem cells were mixed and injected into mice, resulting tumors showed a preponderance of HIF-1α−/− cells in regions distant from the blood supply (36). Although contrary to current thinking, loss of HIF-1 transcriptional/translational activity may be advantageous to survival in hypoxic regions that are severely depleted of nutrients.

The extent of nutrient depletion in cryostat sections from tumor xenografts and clinical biopsies has been examined using an elegant quantitative bioluminescence method that can measure microregional variations in tissue ATP, lactate, and glucose levels (37). Glucose levels below the detection limit of 0.5 μmol/g were commonly observed in regions of tumors where adjacent regions contained up to 3.5 μmol/g glucose. Although perinecrotic regions may not be completely lacking glucose, a significant reduction in glucose and ATP through the tumor cord can be expected. This metabolic mapping method is currently being combined with pimonidazole and HIF-1α staining (38) and this approach should be able to measure the level of glucose in perinecrotic regions.

The lack of HIF-1α expression in regions distant from the vasculature is relevant to the application of HIF-1α targeting strategies in cancer treatment. Development of peptides that hinder interaction between HIF-1 and the coactivator p300/CREB, antisense technology leading to down-regulation of HIF-1α, and high-throughput screening of small-molecule inhibitors of the HIF-1 transcriptional pathway have been proposed (21). However, these approaches might be limited if hypoxic cells most distant from blood vessels fail to express this target protein. Conversely, HIF-1 targets with slow turnover rates continue to be expressed by perinecrotic hypoxic cells so that strategies using downstream targets such as CAIX (39) may be more successful.

Grant support: Canadian Institutes of Health Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Susan MacPhail, Laura Sinnott, and Denise McDougal for expert technical assistance.

1
Vaupel P, Kelleher DK, Hockel M. Oxygen status of malignant tumors: Pathogenesis of hypoxia and significance for tumor therapy.
Semin Oncol
2001
;
28
:
29
–35.
2
Airley RE, Loncaster J, Raleigh JA, et al. GLUT-1 and CAIX as intrinsic markers of hypoxia in carcinoma of the cervix: relationship to pimonidazole binding.
Int J Cancer
2003
;
104
:
85
–91.
3
Beasley NJ, Wykoff CC, Watson PH, et al. Carbonic Anhydrase IX, an endogenous hypoxia marker, expression in head and neck squamous cell carcinoma and its relationship to hypoxia, necrosis, and microvessel density.
Cancer Res
2001
;
61
:
5262
–7.
4
Bussink J, Kaanders JH, van der Kogel AJ. Tumor hypoxia at the micro-regional level: clinical relevance and predictive value of exogenous and endogenous hypoxic cell markers.
Radiother Oncol
2003
;
67
:
3
–15.
5
Begg AC. Is HIF-1α a good marker for tumor hypoxia?
Int J Radiat Oncol Biol Phys
2003
;
56
:
917
–9.
6
Olive PL, Aquino-Parsons C, MacPhail SH, et al. Carbonic anhydrase 9 as an endogenous marker for hypoxic cells in cervical cancer.
Cancer Res
2001
;
61
:
8924
–9.
7
Vukovic V, Haugland HK, Nicklee T, et al. Hypoxia-inducible factor-1α is an intrinsic marker for hypoxia in cervical cancer xenografts.
Cancer Res
2001
;
61
:
7394
–8.
8
Kaanders JH, Wijffels KI, Marres HA, et al. Pimonidazole binding and tumor vascularity predict for treatment outcome in head and neck cancer.
Cancer Res
2002
;
62
:
7066
–74.
9
Bussink J, Kaanders JH, Rijken PF, et al. Multiparameter analysis of vasculature, perfusion and proliferation in human tumour xenografts.
Br J Cancer
1998
;
77
:
57
–64.
10
Ljungkvist AS, Bussink J, Rijken PF, et al. Vascular architecture, hypoxia, and proliferation in first-generation xenografts of human head-and-neck squamous cell carcinomas.
Int J Radiat Oncol Biol Phys
2002
;
54
:
215
–28.
11
Semenza GL. HIF-1 and mechanisms of hypoxia sensing.
Curr Opin Cell Biol
2001
;
13
:
167
–71.
12
Jiang BH, Semenza GL, Bauer C, et al. Hypoxia-inducible factor 1 levels vary exponentially over a physiologically relevant range of O2 tension.
Am J Physiol
1996
;
271
:
C1172
–80.
13
Kaluz S, Kaluzova M, Stanbridge EJ. Expression of the hypoxia marker carbonic anhydrase IX is critically dependent on SP1 activity. Identification of a novel type of hypoxia-responsive enhancer.
Cancer Res
2003
;
63
:
917
–22.
14
Turner KJ, Crew JP, Wykoff CC, et al. The hypoxia-inducible genes VEGF and CA9 are differentially regulated in superficial vs invasive bladder cancer.
Br J Cancer
2002
;
86
:
1276
–82.
15
Rafajova M, Zatovicova M, Kettmann R, et al. Induction by hypoxia combined with low glucose or low bicarbonate and high posttranslational stability upon reoxygenation contribute to carbonic anhydrase IX expression in cancer cells.
Int J Oncol
2004
;
24
:
995
–1004.
16
Raleigh JA, Chou SC, Arteel GE, et al. Comparisons among pimonidazole binding, oxygen electrode measurements, and radiation response in C3H mouse tumors.
Radiat Res
1999
;
151
:
580
–9.
17
Azuma C, Raleigh JA, Thrall DE. Longevity of pimonidazole adducts in spontaneous canine tumors as an estimate of hypoxic cell lifetime.
Radiat Res
1997
;
148
:
35
–42.
18
Olive PL, Vikse C, Trotter MJ. Measurement of oxygen diffusion distance in tumor cubes using a fluorescent hypoxia probe.
Int J Radiat Oncol Biol Phys
1992
;
22
:
397
–402.
19
Olive PL, Durand RE, Raleigh JA, et al. Comparison between the comet assay and pimonidazole binding for measuring tumour hypoxia.
Br J Cancer
2000
;
83
:
1525
–31.
20
Semenza GL. Hydroxylation of HIF-1: oxygen sensing at the molecular level.
Physiology (Bethesda)
2004
;
19
:
176
–82.
21
Semenza GL. Targeting HIF-1 for cancer therapy.
Nat Rev Cancer
2003
;
3
:
721
–32.
22
Rasey JS, Hofstrand PD, Chin LK, et al. Characterization of [18F]fluoroetanidazole, a new radiopharmaceutical for detecting tumor hypoxia.
J Nucl Med
1999
;
40
:
1072
–9.
23
Olive PL, Banath JP, Durand RE. The range of oxygenation in SiHa tumor xenografts.
Radiat Res
2002
;
158
:
159
–66.
24
Wykoff CC, Beasley NJ, Watson PH, et al. Hypoxia-inducible expression of tumor-associated carbonic anhydrases.
Cancer Res
2000
;
60
:
7075
–83.
25
Olive PL, Aquino-Parsons C. Measurement of tumor hypoxia using single-cell methods.
Semin Radiat Oncol
2004
;
14
:
241
–8.
26
Bennewith KL, Durand RE. Quantifying transient hypoxia in human tumor xenografts by flow cytometry.
Cancer Res
2004
;
64
:
6183
–9.
27
Walton MI, Bleehen NM, Workman P. Effects of localised tumour hyperthermia on pimonidazole (Ro 03-8799) pharmacokinetics in mice.
Br J Cancer
1989
;
59
:
667
–73.
28
Allen JG, Dische S, Lenox-Smith I, et al. The pharmacokinetics of a new radiosensitiser, Ro 03-8799 in humans.
Eur J Clin Pharmacol
1984
;
27
:
483
–9.
29
Nordsmark M, Loncaster J, Aquino-Parsons C, et al. Measurements of hypoxia using pimonidazole and polarographic oxygen-sensitive electrodes in human cervix carcinomas.
Radiother Oncol
2003
;
67
:
35
–44.
30
Bennewith KL, Raleigh JA, Durand RE. Orally administered pimonidazole to label hypoxic tumor cells.
Cancer Res
2002
;
62
:
6827
–30.
31
Moeller BJ, Cao Y, Li CY, et al. Radiation activates HIF-1 to regulate vascular radiosensitivity in tumors: role of reoxygenation, free radicals, and stress granules.
Cancer Cell
2004
;
5
:
429
–41.
32
Tomes L, Emberley E, Niu Y, et al. Necrosis and hypoxia in invasive breast carcinoma.
Breast Cancer Res Treat
2003
;
81
:
61
–9.
33
Turley H, Wykoff CC, Troup S, et al. The hypoxia-regulated transcription factor DEC1 (Stra13, SHARP-2) and its expression in human tissues and tumours.
J Pathol
2004
;
203
:
808
–13.
34
Lang KJ, Kappel A, Goodall GJ. Hypoxia-inducible factor-1α mRNA contains an internal ribosome entry site that allows efficient translation during normoxia and hypoxia.
Mol Biol Cell
2002
;
13
:
1792
–801.
35
Koch CJ, Giandomenico AR, Iyengar CW. Bioreductive metabolism of AF-2[2-(2-furyl)-3-(5-nitro-2-furyl)acrylamide] combined with 2-nitroimidazoles. Implications for use as hypoxic cell markers.
Biochem Pharmacol
1993
;
46
:
1029
–36.
36
Yu JL, Rak JW, Carmeliet P, et al. Heterogeneous vascular dependence of tumor cell populations.
Am J Pathol
2001
;
158
:
1325
–34.
37
Walenta S, Schroeder T, Mueller-Klieser W. Metabolic mapping with bioluminescence: basic and clinical relevance.
Biomol Eng
2002
;
18
:
249
–62.
38
Zips D, Adam M, Flentje M, et al. Impact of hypoxia and the metabolic microenvironment on radiotherapy of solid tumors. Introduction of a multiinstitutional research project.
Strahlenther Onkol
2004
;
180
:
609
–15.
39
Lim HY, Ahn M, Chung HC, et al. Tumor-specific gene therapy for uterine cervical cancer using MN/CA9-directed replication-competent adenovirus.
Cancer Gene Ther
2004
;
11
:
532
–8.