Inhibitors of HIF-1α and CXCR4 Mitigate the Development of Radiation Necrosis in Mouse Brain

Abstract
Purpose: There is mounting evidence that, in addition to angiogenesis, hypoxia-induced inflammation via the hypoxia inducible factor-1α (HIF-1α) / CXC chemokine receptor-4 (CXCR4) pathway may also contribute to the pathogenesis of late-onset, radiation-induced necrosis (RN). The present study investigates the mitigative efficacy of a HIF-1α inhibitor, topotecan, and a CXCR4 antagonist, AMD3100, on the development of RN in an intracranial mouse model. Methods and Materials: Mice received a single-fraction, 50-Gy dose of hemispheric radiation from the Leksell GammaKnife PerfexionTM and were then treated with either topotecan, a HIF- 1α inhibitor, from 1-12 weeks post-irradiation (PIR), or AMD3100, a CXCR4 antagonist, from 4- 12 weeks PIR. The onset and progression of RN were monitored longitudinally via noninvasive, in vivo MRI from 4-12 weeks PIR. Conventional hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC) staining were performed to evaluate the treatment response. Results: The progression of brain RN was significantly mitigated for mice treated with either topotecan or AMD3100, compared to control animals. MR-derived lesion volumes were significantly smaller for both of the treated groups, and histologic findings correlated well with the MRI data. By H&E staining, both treated groups demonstrated reduced radiation-induced tissue damage compared with controls. Further, IHC results revealed that expression levels of VEGF, CXCL12, CD-68, CD3 and TNF-α in the lesion area were significantly lower in treated (topotecan or AMD3100) brains vs. control brains, while Iba-1 and HIF-1α expression was similar, though somewhat reduced. CXCR4 expression was reduced only in topotecan-treated mice, while IL-6 expression was unaffected by either topotecan or AMD3100. Conclusions: By reducing inflammation, both topotecan and AMD3100 can, independently, mitigate the development of RN in mouse brain. When combined with first-line, anti–angiogenic treatment, anti-inflammation therapy may provide an adjuvant therapeutic strategy for clinical, post-radiation management of tumors, with additional benefits in the mitigation of RN development.

Introduction
Late-time-to-onset brain radiation necrosis (RN), a well-known, adverse effect following radiation therapy for central nervous system (CNS) tumors, is a serious clinical problem [1,2]. With combined chemo-radiation as the current standard of care for the majority of CNS tumors, the incidence of RN has substantially increased. [3] Treatment induced effects can appear from as few as three months to as many as ten or more years after radiotherapy [1]. Both medical and surgical management strategies have been used to treat RN. Current medical treatments, including corticosteroids [4], hyperbaric oxygen therapy [5], and anticoagulants or antiplatelet agents [6], are either associated with significant toxicity, or have limited efficacy. Surgical resection is typically reserved for symptomatic patients due to mass effect and is associated with potential morbidity and mortality. [7]To overcome this clinically challenging pathology, numerous studies have focused on unraveling the mechanisms underlying the development/progression of RN and discovering new molecular targets that might facilitate novel treatments [1,2,8-10].

In those studies, in addition to angiogenesis, inflammation via the hypoxia inducible factor-1α (HIF-1α) / CXC chemokine receptor-4 (CXCR4) pathway has been hypothesized as an important contributor to the pathogenesis of brain RN. Briefly, focal endothelial damage and associated local tissue hypoxia following irradiation induce hypoxia inducible factor-1α (HIF-1α) expression, which strongly mediates up-regulation of vascular endothelial growth factor (VEGF). VEGF plays an important role in both increased angiogenesis and vascular permeability. Meanwhile, radiation-therapy- associated increases in the cytokine signaling cascade lead to increased inflammation, with astrocytes that express CXC chemokine ligand 12 (CXCL12, also known as stromal derived factor-1, SDF-1) recruiting monocytes (i.e., microglia, macrophages) and lymphocytes expressing CXCR4 via CXCL12-CXCR4 chemotaxis. Furthermore, these monocytes and lymphocytes contribute to further inflammation by releasing inflammatory cytokines, including tumor necrosis factor (TNF)-α, interleukin (IL)-6, and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-kB). Increased angiogenesis and inflammation exacerbate tissue hypoxia and vasogenic edema, resulting in continued progression of CNS radiation necrosis.Bevacizumab (Avastin®, Genentech, San Francisco, CA), a monoclonal anti-VEGF antibody that prevents VEGF from reaching its endothelial targets [11,12], thereby reducing the associated treatment-related edema, has been recognized as an effective therapy for treating RN and reducing perilesional edema clinically [13,14].

Further, our preclinical studies havevalidated the mitigative effect of the analogous murine anti-VEGF antibody (B20-4.1.1) on radiation necrosis [15] in a murine model of late-onset RN [16]. However, many clinical studies have reported that the efficacy of bevacizumab in treating radiation necrosis is transitory, and RN may recur or progress after bevacizumab treatment is discontinued [17-19]. Treatment side effects, including deep-vein thrombosis and focal mineralization, have also been documented [20,21].It has been widely known that overexpression of HIF-1α correlates significantly with tumor invasion and metastasis [22-24]. Hypoxia-induced CXCR4 expression has also been implicated in many tumors [25,26]. Blocking the HIF-1α / CXCR4 signaling pathway, using a HIF-1 inhibitor or CXCR4 antagonist, can inhibit tumor growth and retard potential metastasis [24,27]. However, whether the HIF-1α / CXCR4 pathway plays a critical role in the progression of RN has not been elucidated conclusively. We have recently developed and characterized a mouse model of late-onset radiation necrosis that recapitulates all of the histological features of RN observed in patients [15]. In the present study, the mitigative efficacy of inhibiting the HIF-1α / CXCR4 signaling pathway of RN was evaluated by longitudinal, in vivo magnetic resonance imaging (MRI) and immunohistochemistry (IHC).All experimental procedures involving animals were approved by ×××× Institutional Animal Care and Use Committee and conformed to the NIH Policy on Responsibility for Care and Use of Animals. 7- to 8-week-old female BALB/c mice (Envigo, Indianapolis, IN) were used in the study.Mice were irradiated with a single fraction of 50 Gy (50% isodose) radiation, as previously described [15].

Unlike clinical radiotherapy, which seeks to avoid damage to normal brain tissue, this dose was chosen specifically to produce late time-to-onset radiation necrosis in mice in an experimentally tractable time frame (i.e., 4-5 weeks post-irradiation). Briefly, mice were anesthetized and restrained on a custom-made platform mounted to the stereotactic frame that attaches to the treatment couch of the Leksell Gamma Knife® PerfexionTM treatment unit (GK; Elekta, Stockholm, Sweden), a state-of-the-art device used for stereotactic irradiation of patients with malignant brain tumors. Mice were anesthetized with a mixture of ketamine (25 mg/kg),acepromazine (5 mg/kg), and xylazine (5 mg/kg), and injected intraperitoneally 5 min before the start of irradiation. A single radiation fraction of 50 Gy (50% isodose) was focused on the cortex of the left hemisphere, ~3 mm posterior to bregma. At this dose, the onset of RN typically occurs at approximately four weeks post irradiation (PIR) [28]. While comparisons are inexact, we note that laboratory mice have roughly half the radiation sensitivity (re LD50) of humans [29,30]. Thus, the 50-Gy mouse irradiation dose corresponds approximately to a 25-Gy human exposure.Topotecan (Cayman Chemicals, Ann Arbor, MI), a topoisomerase I inhibitor, non-selectively suppresses HIF-1α expression [31], and causes concomitant inhibition of HIF-1α target genes, angiogenesis, and tumor growth [32]. AMD3100 (Cayman Chemicals), a symmetrical bicyclam that is an antagonist of CXCR4, has been widely used to block the CXCL12-CXCR4 chemotaxis [33,34].

Both topotecan and AMD3100 have received FDA approval and have been used clinically in the treatment of brain tumors and other tumors. [32,35].Two treatment groups, designated A and B, were included in this study. Group A mice were used to investigate the efficacy of HIF-1α inhibition via topotecan as a treatment of RN. Following GK irradiation, mice were randomly divided into two cohorts. Mouse cohort A1 (n=8) received topotecan (10 mg/kg in DMSO), while mouse cohort A2 (n=7) received an equivalent volume of DMSO, intraperitoneally, twice weekly, from 1 to 12 weeks PIR.Group B mice were used to determine the efficacy of CXCR4 inhibition via AMD3100 on the treatment of RN. At week 4 PIR, mice were randomly divided into two cohorts. Mouse cohort B1 (n=7) received AMD3100 (5 mg/kg), while mouse cohort B2 (n=6) received the equivalent volume of phosphate-buffered saline (PBS, pH = 7.4), intraperitoneally, daily from 4 to 12 weeks PIR.Doses of topotecan and AMD3100 were chosen based on previously published studies in mouse brain tumors [36-38]. In experiment A, the anti-HIF-1 treatment started from week 1 PIR, as it has been shown that local hypoxia, which mediates the upregulation of HIF-1, was induced soon after irradiation[2]. In experiment B, the treatment with AMD3100 starts from week 4 PIR, which is the earliest time at which focal RN lesions can be observed on anatomical MR images [16].

In vivo MRI experiments were performed on a 4.7-T small-animal MR scanner (Agilent/Varian, Santa Clara, CA) equipped with a DirectDriveTM console. MRI data were collected using an actively decoupled coil pair. Before being placed into the magnet, mice were injected intraperitoneally with 0.25 mL of MultiHance (gadobenate dimeglumine; Bracco Diagnostics Inc, Princeton, NJ) contrast agent, diluted 2:10 in sterile saline. (See supplementary material 1 for more detailed information).All mice in experiment A were imaged biweekly from 4 to 12 weeks PIR. In experiment B, mouse cohort B1 was imaged weekly from 4 to 12 weeks PIR, while cohort B2 mice were imaged biweekly from 4 to 12 weeks PIR. Note, contrast agent clears completely between imaging sessions. For both experiments, multislice, spin-echo, T2-weighted images were collected with the following parameters: time-to-repetition (TR) = 1200 ms, time-to-echo (TE) = 50 ms, number of transients (NT) = 4, field of view (FOV) = 15 x 15 mm2, matrix size = 128 x 128, slice thickness = 0.5 mm, and 21 slices to cover the whole brain. Multislice, spin-echo post- contrast T1-weighted images were acquired with the same FOV and slice coverage, with the following parameters: time-to-repetition (TR) = 650 ms, time-to-echo (TE) = 20 ms, number of transients (NT) = 4.All datasets were analyzed, as previously described [15,39], using custom-written Matlab software (The Mathworks, Natick MA).

Briefly, each mouse brain was divided along the midline into left (irradiated) and right (non-irradiated) hemispheres. The lesion was defined as the region of hyperintensity on post-contrast T1- or T2-weighted MR images. MR-derived lesion volumes were determined via a threshold segmentation algorithm, in which areas of the left hemisphere brighter than the 95th percentile of the right hemisphere were defined as lesions. (The lesion volume is then the sum of the lesion voxels multiplied by the voxel volume). Repeated measures two-way ANOVA with Bonferroni post-tests were used to compare the tumor volumes across groups at selected time points post-irradiation. Graphs and statistical analyses were performed in Prism (GraphPad Software, San Diego, CA) and Matlab.Histology and ImmunohistochemistryTissue sections were stained with hematoxylin and eosin (H&E), per standard protocols. In addition, IHC stainings for HIF-1α, VEGF, CXCR4, CXCL12, Iba-1, CD68, CD3, TNF-α and IL-6were performed following manufacturer’s procedures. Briefly, mice were sacrificed and intracardially perfused with 1% PBS, followed by 10% formalin, immediately after the final imaging session. Each mouse head was dissected and immersed in 10% formalin for 24 hours. Brains were extracted from the skulls and a 2-mm thick coronal block, centered at the irradiation site (~3 mm posterior to bregma), was obtained for each brain. The blocks were embedded in paraffin wax and sectioned from the center, at a thickness of 5 µm. All sections were deparaffinized and rehydrated; antigen retrieval was performed with citrate buffer (pH = 6.8) at 70°C overnight following one-hour, non-specific blo cking using Avidin-Biotin Complex (ABC) blocking kit (Life Technologies, NY, USA).All primary antibodies (see supplementary material 2 for more detailed information) were incubated with sections at 4°C overnight. All sections were processed with the Histostain Plus Broad System kit (Invitrogen Life Technology, Frederick, MD, USA) followed by a broad-spectrum secondary antibody for one hour. Three percent H2O2 was used to decrease the background. Histological sections were examined with the Hamamatsu NanoZoomer whole-slide imaging system (Hamamatsu, Hamamatsu City, Japan). For each animal, each IHC stain was graded on a 0 to 3 scale (e.g., 0 = No stain; 1 = Light stain; 2 = Moderate stain; 3 = Heavy stain) by an experienced histologist (L.Y.).

Results
Figures 1A and 2A show representative post-contrast, T1- and T2-weighted MR images at weeks 4, 8 and 12 PIR for each mouse cohort. Heterogeneous, hyperintense areas in these images correspond to RN lesions in the brain. The DMSO-treated and PBS-treated brains show larger lesions at week 8 PIR, and the lesions progressed significantly by week 12 PIR. By contrast, topotecan-treated and AMD3100-treated brains show minimal hyperintense areas until week 8 PIR. Figures 1B and 2B are plots of RN lesion volume vs. weeks PIR derived from T1- and T2-weighted MR images. For the mice treated with topotecan, these plots show significantly smaller RN lesion volumes at weeks 8, 10, 12 PIR, compared to mice treated with carrier only. In a similar manner, AMD3100 treatment significantly decreased RN volume, beginning six weeks PIR, compared to the PBS-treated mice. H&E staining proves that both topotecan and AMD3100 mitigate RN in irradiated brain tissue Figures 3A and 3B show representative H&E histologic images for treated and control mice from both experiments A and B at twelve weeks PIR. The irradiated hemispheres of the DMSO- treated (Fig. 3A, left column) and PBS-treated (Fig. 3B, left column) mice demonstrate histologic hallmarks of RN, including fibrinoid vascular necrosis (yellow arrow), vascular telangiectasia (black arrow), hemorrhage (blue arrows), loss of neurons, and edema (green arrows). By contrast, the irradiated hemisphere of the topotecan-treated (Fig. 3A, middle and right column) and AMD3100-treated (Fig. 3B, middle and right column) mice showed only mild- to-modest tissue damage.

These histologic findings support the MR data shown in Fig. 1A and Fig. 2A, demonstrating the mitigative effect of HIF-1α and CXCR4 inhibition on RN development and progression. The mixed pathological features seen in the H&E-stained sections also explain the heterogeneous signals observed on post-contrast T1W and T2W images. IHC staining demonstrates that both topotecan and AMD3100 reduce inflammation in irradiated mice Figure 4 shows representative IHC staining for molecular markers of microglia and immune cells (macrophage and lymphocyte) in mouse brains from experiments A (Fig. 4A) and B (Fig. 4B). The number of Iba1-labeled microglia was similar, although somewhat lower, in the brains of topotecan- or AMD3100-treated mice, comparing to those of the carrier-treated animals. Additionally, there were no clusters of positively Iba1-stained cells in either topotecan- or AMD3100-treated RN brain slices, while such clusters could be seen clearly in the brains of control mice (data not shown). CD68-labeled macrophages and CD3-labeled lymphocytes were observed, predominantly on the edges of RN lesions and surrounding the damaged vascular vessels.

Both topotecan and AMD3100 dramatically reduced the numbers of CD68- and CD3- positive cells in treated mice. Figure 5 shows representative immunohistochemistry staining for the expression of HIF-1α, VEGF, CXCR4, CXCL12, TNF-α, and IL-6 in mouse brains from both experiments A (Fig. 5A) and B (Fig. 5B). The results of scoring the IHC-stained slides, as described in Methods and Materials, are reported in Table 1. Values in this table represent the average score across all animals within each treatment or control group. Compared with the control group, treatment with either topotecan or AMD3100 resulted in reduced expression of VEGF in radiated mouse brain, while HIF-1α expression was largely unchanged. Expression of CXCL12 was clearly reduced in the brains of either topotecan- or AMD3100-treated animals relative to their control counterparts, while CXCR4 expression was reduced only in topotecan-treated mice. Treatment with either topotecan or AMD3100 reduced the expression of the pro-inflammatory cytokine, TNF-α, but had no measurable effect on IL-6 expression.

Discussion
In a recent retrospective review of IHC analyses of surgical human RN specimens designed to elucidate the molecular mechanisms underlying brain RN, Yoritsune, et al. [2], found that both angiogenesis and inflammation may be caused by the upregulation of HIF-1α post-radiotherapy. HIF-1α not only contains a transactivation domain for VEGF, but is also an important regulator of the CXCL12-CXCR4 chemokine axis [2,40,41], whose activity is thought to be significantly upregulated under both hypoxia and inflammation [42,43]. Accumulated pro-inflammatory cytokines in the peri-necrotic area would, in turn, aggravate hypoxia, and, consequently, further upregulate HIF-1α (i.e., positive feedback). CXCR4 activation by CXCL12 plays a key role within hypoxic areas of tumors by enabling increased cell growth, invasiveness, and recruitment of endothelial cell progenitors, leading to tumor angiogenesis. A CXCR4 antagonist demonstrated inhibition of intracranial glioblastoma xenograft tumor-cell growth by increased apoptosis through acting on the CXCL12-CXCR4 axis [35,44]. VEGF and CXCR4 antagonists are potential therapeutic agents that may be used to both mitigate RN and inhibit glioma tumor-cell growth. Thus, in the present study, we sought to investigate the role of hypoxia and inflammation on the progression/development of Gamma Knife induced late-time-to-onset RN via the use of a HIF-1α inhibitor, topotecan, and a CXCR4 antagonist, AMD3100.

The onset and progression of RN in mouse brain were characterized by heterogeneous, hyperintense regions detected on either post-contrast T1W or T2W images (Figs, 1A and 2A). Differences between T1WI- and T2WI-defined lesion boundaries, thus volumes, reflect the different contrast mechanisms of these MRI protocols, BBB breakdown and edema, respectively. Compared with controls, irradiated brains of mice treated with either topotecan or AMD3100 demonstrated significantly smaller lesion volumes, as measured longitudinally by in vivo MRI, reduced swelling, and decreased RN-related tissue damage, assessed in H&E-stained tissue sections. These results demonstrate clearly the efficacy of both topotecan and AMD3100 in mitigating the progression of RN in mice.To validate that treatments with topotecan or AMD3100 reduce inflammation via the hypoxia- HIF-1α-CXCR4 pathway, IHC staining of brain slices for HIF-1α, VEGF, CXCR4, CXCL12, Iba-1, CD68 and CD3 was performed. Initially, we focused on detecting changes in microglia, macrophages and lymphocytes, cells that are widely recognized to be involved in theprogression of inflammation in the central nervous system. The decreased expression of CD68, and CD3, and somewhat lower expression of Iba-1, measured via IHC staining at twelve weeks PIR, suggests that both topotecan and AMD3100 treatments attenuate microglia activation, and macrophage and lymphocyte infiltration.

Topotecan is a negative regulator of the CXCL12-CXCR4 chemokine axis, while AMD3100 is an antagonist of CXCR4. Treatment with either topotecan or AMD3100 in our study markedly impaired CXCL12 expression, while topotecan reduced expression of CXCR4, thereby providing strong evidence that both topotecan and AMD3100 can deactivate the CXCL12-CXCR4 signaling pathway. (As an antagonist of CXCR4, AMD3100 blocks the action of CXCR4, without reducing its expression.) However, HIF-1α expression was largely unaffected by either treatment, suggesting that both topotecan and AMD3100 act predominantly on downstream expression of VEGF and CXCL12, an effect that might be due to the reduced positive feedback effect of pro-inflammatory cytokines. To test this notion, the expressions of pro-inflammatory cytokines, including TNF-α and IL-6, were evaluated. Consistent with the staining results for immune cells (i.e., macrophages and lymphocytes), which are able to produce these cytokines, both topotecan and AMD3100 treatment resulted in reduced TNF-α expression. However, there was no obvious decrease of IL-6 expression after treatment. This suggests that signaling pathways such as NF-kB-associated increases in IL-6 activation with associated Janus kinase (JAKs), and signal transducer and activator of transcription (STAT)-3 transcription factor, may play a role in the progression of RN. However, further study is needed to investigate and validate these observations.

Conclusion
Radiation is a powerful and ubiquitous treatment for brain tumors. However, its outcome may be complicated by the appearance of delayed RN. Blocking the HIF-1α/CXCR4 pathway axis with topotecan or AMD3100 results in significant anti-tumor activity against many types of cancer in vitro and in vivo and, importantly, inhibits the development of metastases, cancer tumor-cell growth and invasion [32,45-52]. The results of the present study demonstrate that treatment with topotecan or AMD3100 can also significantly inhibit the HIF-1α/CXCR4 axis, HIF inhibitor thereby reducing the progression of RN. Targeting the HIF-1α/CXCR4 pathway may be a promising therapy for treating recurrent tumor post-RT, with the additional benefit of mitigating the progression of RN.