Research ArticleCancer

Genetically modified lentiviruses that preserve microvascular function protect against late radiation damage in normal tissues

See allHide authors and affiliations

Science Translational Medicine  24 Jan 2018:
Vol. 10, Issue 425, eaar2041
DOI: 10.1126/scitranslmed.aar2041

Skin protection from radiation goes viral

With recent improvements in cancer therapy, an increasing number of people are living as cancer survivors, in many cases with long-term side effects caused by the cancer treatment. These effects include radiation-induced vascular dysfunction and fibrosis, which interfere with tissue reconstruction using skin flaps after mastectomy in breast cancer patients. Khan et al. developed a virus-based gene therapy approach to address this problem, up-regulating one gene to preserve skin flap volume and knocking down another to reduce radiation-induced skin contracture. The authors tested their approach in rat models of radiation therapy and skin flap reconstruction and also demonstrated that the gene therapy did not interfere with the anticancer effects of radiation.


Improvements in cancer survival mean that long-term toxicities, which contribute to the morbidity of cancer survivorship, are being increasingly recognized. Late adverse effects (LAEs) in normal tissues after radiotherapy (RT) are characterized by vascular dysfunction and fibrosis causing volume loss and tissue contracture, for example, in the free flaps used for immediate breast reconstruction after mastectomy. We evaluated the efficacy of lentivirally delivered superoxide dismutase 2 (SOD2) overexpression and connective tissue growth factor (CTGF) knockdown by short hairpin RNA in reducing the severity of LAEs in an animal model of free flap LAEs. Vectors were delivered by intra-arterial injection, ex vivo, to target the vascular compartment. LVSOD2 and LVshCTGF monotherapy before irradiation resulted in preservation of flap volume or reduction in skin contracture, respectively. Flaps transduced with combination therapy experienced improvements in both volume loss and skin contracture. Both therapies reduced the fibrotic burden after irradiation. LAEs were associated with impaired vascular perfusion, loss of endothelial permeability, and stromal hypoxia, which were all reversed in the treatment model. Using a tumor recurrence model, we showed that SOD2 overexpression in normal tissues did not compromise the efficacy of RT against tumor cells but appeared to enhance it. LVSOD2 and LVshCTGF combination therapy by targeted, intravascular delivery reduced LAE severities in normal tissues without compromising the efficacy of RT and warrants translational evaluation as a free flap–targeted gene therapy.


Ever improving cancer survival means that the long-term toxicities of cancer treatments have a major impact upon cancer survivorship. For example, for women diagnosed with small, mammography-detected breast cancers, it is the late complications of radiotherapy (RT), rather than the risk of locoregional recurrence (1, 2), that are the dominant hazard. Late adverse effects (LAEs) in the breast include breast shrinkage, hardening, discomfort, and skin changes. These changes (37) also occur in breasts that have been reconstructed with free flaps after mastectomy and augur a poorer long-term aesthetic outcome. Reoperation rates for flaps affected in this manner are high and, in some cases, salvage reconstruction with a second free flap remains the only option. Therefore, if adjuvant RT may be required, free flap reconstructions are delayed until 6 months after RT (delayed breast reconstruction after mastectomy) (8). Such delays incur more operations (9) and multiple hospital admissions (10) and result in aesthetically inferior reconstructive outcomes compared to immediate reconstructions (11, 12). Consequently, conferring radioprotection to a free flap offers the clinical benefit of allowing patients undergoing reconstructions after cancer excision to have an earlier reconstruction, which is more durable against the adverse effects of adjuvant therapies.

Irradiation of the vasculature causes endothelial dysfunction, sustained inflammation, pruning of the vascular architecture, and a prothrombotic milieu defined by overexpression of plasminogen activator inhibitor 1 (PAI-1) (1315). Fibrosis in LAEs may represent a compensatory response to tissue hypoxia caused by vascular damage (16). RT-induced reactive oxygen species (ROS) are implicated (17, 18) in the activation of profibrotic pathways [transforming growth factor–β1 (TGF-β1) and connective tissue growth factor (CTGF)] and appear to be central to the development of LAEs. Altered expression of cellular adhesion molecules [vascular cell adhesion molecule 1 (VCAM1)/intercellular adhesion molecule 1 (ICAM1)] (19), endothelial-to-mesenchymal transition (15, 2022), and overexpression of PAI-1 (1315) are thought to be mechanisms central to the development of LAEs. ROS are normally inactivated by the enzyme superoxide dismutase (SOD), which exists in cytoplasmic (SOD1), mitochondrial (SOD2), and extracellular (SOD3) forms. SOD2 gene therapy radioprotects normal cells in vitro and in vivo (19, 2330) and has been tested in a phase 1 clinical trial (31). SOD2 gene therapy represents an attractive therapeutic strategy because it targets an early step in the pathophysiological cascade from irradiation to LAEs. CTGF, a downstream effector of TGF-β1 signaling, is overexpressed in radiation-induced fibrosis (3234) and other pathological fibrotic states (3537). Therefore, we sought to develop a radioprotective therapy to target both oxidative stress (SOD2) and fibrotic (CTGF) pathways.

Radioprotective therapies should target specifically normal tissues and not tissues harboring microscopic residual disease. Clinically, such an opportunity arises when performing free tissue transfer [immediate breast reconstruction after mastectomy (3840)]. Free flaps are composite blocks of tissue harvested from distant donor sites with supplying blood vessels. Radioprotective gene therapy strategies using plasmid/liposomal and virally delivered SOD2 have been reported previously (24, 31, 4143), with vectors delivered either directly to target organs or intravenously. Given the central role of the vascular compartment in the development of LAEs, we sought to protect selectively the vasculature of normal tissues by delivering vectors intra-arterially into flaps during the ischemic interval. We hypothesized that overexpressing SOD2 and silencing CTGF would reduce the severity of LAEs in irradiated normal tissues without compromising the antitumor efficacy of RT. We investigated this using a free flap gene therapy (FFGT) model (40, 4448) because it permits exquisite anatomic control of viral delivery and offers potential for subsequent translation.

We show that free flap irradiation causes clinically relevant contracture and volume loss due to irreversible damage in the vascular compartment resulting in tissue hypoxia and fibrosis. We generate therapeutic lentiviral vectors that mediate durable transgene expression in vivo and show that SOD2 overexpression and CTGF silencing achieve differential effects in ameliorating LAEs after irradiation (after RT). Combination therapy with both vectors protects irradiated flaps against both volume loss and flap contracture. Finally, we show that SOD2 overexpression in normal tissues does not compromise the efficacy of RT and may enhance it.


Irradiation of superficial inferior epigastric artery flaps with 50 Gy produces a clinically relevant LAE phenotype

We report a model of free flap irradiation using the superficial inferior epigastric artery (SIEA) flap in rodents (fig. S1A) that we developed to test our hypotheses. Briefly, we trialed three biologically equivalent fractionation schedules and characterized the LAEs that developed in vivo after irradiation (fig. S1B). SIEA flaps irradiated with 50 Gray (Gy)/3 fractions developed a clinically relevant LAE phenotype (fig. S1, C to O) characterized by skin paddle contracture (Fig. 1A), volume loss, and pathognomonic skin changes (altered pigmentation and telangiectasia). Skin paddle surface areas were reduced in irradiated flaps compared to unirradiated controls (Fig. 1, A and B). Irradiated flaps exhibited acute RT-induced toxicities (Fig. 1C) ranging from dry desquamation to confluent, moist desquamation (fig. S1C). Acute toxicities were evident at day 7, peaked at day 14 after RT, and resolved by 35 days (Fig. 1C). Irradiated flaps attained progressively higher scores on the Radiation Therapy Oncology Group (RTOG) LAE criteria (Fig. 1C), which stabilized by 140 days after RT [mean score of 5.85 (SEM, 0.10)] at 180 days. LAEs were first evident in the skin followed by the subcutaneous tissues and, ultimately, the knee joint (Fig. 1C). Mean passive range of movement at irradiated knee joints was reduced compared to nonirradiated controls (Fig. 1D). Magnetic resonance imaging (MRI) of irradiated and matched unirradiated controls in vivo revealed reductions in flap volume at 6 months (Fig. 1E).

Fig. 1 Irradiation with 50 Gy/3 fractions generates LAE characterized by SOD2 depletion and CTGF overexpression.

(A) Representative photographs of bilateral superficial inferior epigastric artery (SIEA) flaps (edges tattooed in Indian ink) in Fischer (F344) male rats showing characteristic late adverse effect (LAE) features observed in irradiated flaps including contracture, induration of the skin, telangiectasia, and hair loss (n = 6 per group). (B) Changes in skin paddle surface area of irradiated and control SIEA flaps [±95% confidence interval (CI)]. (C) Acute and late RTOG scores including component subscores by tissue type: skin, subcutaneous (s/c), and joint. ROTG, Radiation Therapy Oncology Group. (D) Passive range-of-movement (ROM) measurement at the knee joint in irradiated and nonirradiated hindlimbs (n = 6 per group). (E) T2-weighted magnetic resonance imaging (MRI) of bilateral SIEA flaps showing an irradiated flap (L) and unirradiated control (R) and quantification of volumetric changes (mean ± 95% CI, n = 6 per group). fx, fractions. (F) Connective tissue growth factor (CTGF) enzyme-linked immunosorbent assay (ELISA) and superoxide dismutase 2 (SOD2) biochemical activity assay performed on flap tissues taken from irradiated and control flaps (n = 6 per group). (G) Western blotting of control and irradiated SIEA flaps for SOD2, phospho–glycogen synthase kinase 3β (GSK3β) (Ser9), β-catenin, and LEF1, with β-actin used for loading control. (H to J) Immunohistochemical analysis and quantification of collagen deposition [(H) Masson’s trichrome (collagen is green)], CTGF expression (I), and hematoxylin and eosin (H&E) staining for the assessment of fat necrosis [(J), red circle)]. Graphs represent means ± SEM of counts of collagen fibrils (H) or cells with membranous CTGF expression (I) per high-powered field (hpf) or % of section exhibiting fat necrosis (J) (n = 6 per group; scale bar, 50 μm). (K) Real-time quantitative polymerase chain reaction (RT-qPCR) for Ctgf, Col1a2, and Col3a1 gene expression (mean fold increase in gene expression ± SEM) in irradiated flap tissues (n = 6 per group). All tissue-based analyses were performed on flaps harvested at 180 days after RT. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We observed CTGF overexpression (Fig. 1F) and functional SOD2 depletion (Fig. 1F) in irradiated flaps. Western blot analyses demonstrated a reduction in SOD2 protein and activation of WNT signaling in irradiated flaps (Fig. 1G). Histological analysis (Masson’s trichrome staining) revealed a dense fibrotic reaction, increased CTGF expression, and greater fat necrosis in irradiated flaps (Fig. 1, H to J). Real-time quantitative polymerase chain reaction (RT-qPCR) analyses demonstrated significant reductions in Ctgf and Col3a1 (P < 0.05) expression and increased Col1a2 gene expression (P < 0.001) (Fig. 1K). Together, these data suggest that the LAE phenotype in our model exhibits clinical, anatomical, volumetric, and biological changes similar to those seen in a clinical setting (18, 49).

Irradiated flaps exhibit abnormal vascular function characterized by loss of endothelial function and fibrosis

Quantitative functional imaging of irradiated and unirradiated flaps demonstrated changes in the MRI transverse relaxation rate R2* of flap tissues (Fig. 2, A to C). Briefly, the presence of paramagnetic deoxyhemoglobin in erythrocytes creates magnetic susceptibility perturbations around blood vessels, increasing R2* of the surrounding tissue in proportion to the tissue deoxyhemoglobin concentration. This noninvasive approach offers a sensitive method of monitoring the dynamics of vascular modeling, function, and regression in vivo (5052). Before irradiation, flaps situated on the left hindlimb (that were due to be irradiated) had a faster R2* than flaps on the contralateral, control hindlimb (P < 0.05) (Fig. 2B). In irradiated flaps, the mean absolute R2* decreased significantly after RT [two-way analysis of variance (ANOVA), P = 0.0004; Fig. 2B], and post hoc analyses revealed slower R2* values in irradiated flaps at all time points (P < 0.001) (Fig. 2B).

Fig. 2 LAEs are characterized by vascular dysfunction, loss of endothelial perfusion and permeability, and perivascular hypoxia.

(A) Parametric R2* maps of control and irradiated flaps overlaid on T2-weighted images acquired 6 months after RT (image representative of n = 6 animals per group). (B) Absolute transverse relaxation rates (R2*) in irradiated versus control flaps at 6 months after RT and relative changes R2* (ΔR2*) after RT (n = 6 animals per group) (mean ± SEM). ns, not significant. (C) Vascular staining for perfused endothelium [Hoechst 33342 (H33342)], vascular permeability [Evans blue (EB)], and immunofluorescent (IF) staining for stromal hypoxia (pimonidazole adduct formation) in irradiated and control flaps at 180 days after RT (scale bars, 100 μm; images representative of n = 3 animals per group). Images depict whole-section H&E (inset left) and whole-section tile scan (inset right), and main image represents higher-magnification view of area depicted by red box. (D) Quantification of H33342 uptake, EB leakage, and pimonidazole adduct formation (mean ± SEM). (E) Fluorescence microscopy for H33342 staining in irradiated and control flaps at day 180 after RT. (F) Fluorescence microscopy for EB staining. (G) Immunofluorescent staining for pimonidazole adduct formation in irradiated and control flaps at 180 days after RT, with H&E staining of images depicting perivascular fibrosis (black arrow; all images representative of n = 3 animals per group). (E to G) Scale bars, 20 μm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To correlate vascular function and structure, we performed immunofluorescence analyses using markers for perfused (functional) blood vessels [Hoechst 33342 (H33342)], vascular permeability [Evans blue (EB)], and hypoxia [pimonidazole adduct formation (P)]. Irradiated flaps demonstrated significant reductions in functional vasculature (control 34.9% versus irradiated 16.5%; P < 0.001) and vascular permeability (control 19.4% versus irradiated 3.8%; P < 0.0001) (Fig. 2, C to G), and greater hypoxia within the stromal compartment (control 5.2% versus irradiated 16.7%; P < 0.001). Areas of hypoxia were observed surrounding vessels exhibiting a dense fibrotic reaction (Fig. 2G). In summary, LAEs are associated with vascular dysfunction characterized by a reduction in R2*, indicating impaired hemodynamics, associated with loss of patent vasculature, perivascular fibrosis, and hypoxia.

LVSOD2 and LVshCTGF transgene expression provides durable radioprotective efficacy

To investigate the time course of SOD2 depletion after RT, we irradiated fibroblasts and found dose-dependent reductions in SOD2 activity (Fig. 3A). We generated stable SOD2-overexpressing fibroblast cell lines and showed that these cells retained greater SOD2 activity after RT (Fig. 3B and fig. S2A). To investigate whether this translated into a survival benefit, we performed 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays at 120 hours in endothelial cells (YPEN1) after RT with three biologically equivalent fractionation doses (5 Gy/single fraction, 6.4 Gy/2 fractions, and 7.2 Gy/3 fractions). No survival benefit was observed in fibroblasts (fig. S2B). SOD2-overexpressing cells demonstrated greater survival in all fractionation schedules trialed (Fig. 3C). This was validated using three-dimensional spheroid assays (Fig. 3D), where greater preservation of spheroid volume after RT was observed in cells overexpressing SOD2. To investigate whether this was attributable to SOD2 overexpression, we performed transient SOD2 knockdown in YPEN1 cells using small interfering RNA 48 hours before RT (Fig. 3E) and demonstrated a reversal of the survival benefit observed previously. PCR demonstrated reductions in SOD2 gene expression using this approach (Fig. 3F). Clonogenic assays in tumor cell lines (HeLa and FaDu) overexpressing SOD2 demonstrated increased survival at some, but not all, radiation doses (Fig. 3, G and H). To investigate whether radioprotection was mediated through mitochondrial or cytosolic localization of SOD2, we performed immunofluorescence for SOD2 and mitochondrial proteins (cytochrome C oxidase, MTCO1) (Fig. 3I) and observed colocalization (Fig. 3I). SOD2 function assays on mitochondrial and cytosolic fractions validated increased SOD2 activity in the mitochondrial compartment only (Fig. 3J).

Fig. 3 SOD2 overexpression preserves ROS scavenging capacity after RT.

(A) SOD2 activity in rat fibroblasts (RF) at 2, 6, and 24 hours after irradiation with 0, 8, or 16 Gray (Gy) of radiation. (B) After irradiation changes in SOD2 activity in RF cells overexpressing SOD2 (RF-LVSOD2) compared to vector (RF-LVGFP) and naïve (RF) controls. (C) 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays in YPEN1 and YPEN1_SOD2 cells at 120 hours after RT across a variety of biologically equivalent fractionation schedules. (D) Three-dimensional spheroid assays using YPEN1 and YPEN1_SOD2 after RT across a range of biologically equivalent fractionation schedules. Graph shows mean relative spheroid area (± SEM) and representative spheroid images (n = 3 repeats). (E) MTT assay (120 hours) investigating the effect of transiently silencing SOD2 expression with small interfering RNA (siRNA) in cells overexpressing SOD2. Control (YPEN1) endothelial cells (ECs) and ECs overexpressing SOD2 were irradiated and evaluated for survival. (F) Confirmation of SOD2 knockdown using siRNA by RT-qPCR. (G) Clonogenic assays (bottom) and quantification (top, shown as mean surviving fraction ± SEM) for HeLa and HeLa-LVSOD2. (H) Clonogenic assays (bottom) and quantification (top, shown as mean surviving fraction ± SEM) for FaDu and FaDu-LVSOD2 at different radiation doses. Images are representative of n = 3 repeats. (I) Confocal IF microscopy of RF and RFSOD2 (MTCO1, anti-cytochrome C oxidase antibody; SOD2, anti-SOD2 antibody; scale bars, 10 μm; images representative of n = 3 repeats). (J) Biochemical SOD2 activity (mean ± SEM) in mitochondrial and cytosolic compartments of YPEN1 and YPEN1_SOD2 cells. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We cloned several plasmids encoding short hairpin RNA (shRNA) against rat CTGF and generated lentiviral particles for each to create stable cell lines (rat fibroblasts) to assess the efficacy of each hairpin. Using enzyme-linked immunosorbent assay, we identified a lead candidate shRNA that knocked down both basal and stimulated CTGF to about 50% of that seen in naïve control cells (fig. S2, C and D).

To investigate whether SOD2 overexpression and CTGF knockdown could be achieved in vivo, we infected SIEA flaps with LVSOD2 [108 transducing units (TUs)] and LVshCTGF (108 TUs) and harvested flap tissues 30 days postoperatively—the time point at which flaps would have been irradiated. We found increased SOD2 protein expression in the SIEA pedicle of LVSOD2-infected flaps compared to sham-infected flaps (Fig. 4A). We confirmed significantly greater SOD2 activity in LVSOD2-infected flaps compared to controls (P < 0.001) (Fig. 4B) and reductions in CTGF expression in flaps infected with LVshCTGF (Fig. 4C). Because our therapeutic strategy targets the vascular system, we sought to demonstrate increased SOD2 activity within this compartment. We dissected out SIEA pedicles from flaps infected with LVSOD2 and demonstrated greater Sod2 gene expression and SOD2 function compared to sham flaps (Fig. 4D).

Fig. 4 Lentiviral transgene expression penetrates tissue and provides durable effects in vivo.

(A) Immunohistochemical staining for SOD2 protein expression in SIEA of flaps infected with LVSOD2 [108 transducing units (TUs)] compared with sham [phosphate-buffered saline (PBS)] controls at 180 days after RT. Representative images of n = 3 animals per group are shown. (B) Biochemical assessment of SOD2 activity in flaps infected with LVSOD2 and controls (mean ± SEM) (n = 3 per group). (C) ELISA for CTGF expression in flaps infected with LVSOD2 or LVSOD2 + LVshCTGF (n = 3 per group) compared to controls (mean ± SEM). (D) RT-qPCR for Sod2 gene expression and biochemical SOD2 activity (mean ± SEM) in flap pedicles (artery and vein only) infected with LVSOD2 and controls (n = 3 per group). (E) Immunofluorescent (IF) staining for green fluorescent protein (GFP) expression at 6 months in flaps infected with LVeGFP (108 TUs) and PBS-sham controls (PBS), depicting macrovascular (SIEA and superficial inferior epigastric artery expression; L, lumen; scale bars, 50 μm) and microvascular expression (CD31 colocalization with GFP; scale bars, 25 μm; images representative of three animals per group). Nuclei counterstained with 4′,6-diamidino-2-phenylindole (DAPI). (F) IF staining for extravascular GFP expression (FABP4 colocalization with GFP; scale bar, 50 μm) in flaps infected with LVeGFP (108 TUs; images representative of three animals per group). Nuclei counterstained with DAPI. (G) qPCR for viral gag gene expression in flap tissues infected with LVeGFP at 6 months after infection (n = 3 animals per group). *P < 0.05, **P < 0.01.

Given the duration over which LAEs manifest in our model, we investigated the durability of transgene expression in vivo. SIEA flaps were infected with a green fluorescent protein (GFP)–expressing lentiviral vector (LVeGFP), and flap tissues were harvested at 180 days after infection. We observed sustained GFP expression in large (SIEA and superficial inferior epigastric vein) and small vessels (Fig. 4E). In addition, we observed GFP expression in extravascular tissues, notably adipose tissues (Fig. 4F). qPCR performed on flap tissues 180 days after infection found 104 copies of the lentiviral gag gene per 100 ng of DNA (Fig. 4G).

Combinatorial gene therapy with LVSOD2 and LVshCTGF in free flaps mitigates the LAEs of RT

Flaps transduced with LVSOD2 and LVshCTGF exhibited less pigmentation change and contracture (Fig. 5A), even in the presence of typical LAE changes such as telangiectasia arising in adjacent skin, outside the zone of viral vector delivery (Fig. 5A). There was no mitigation of RT-induced effects with the scrambled vector control (LVsh-scram) (Fig. 5A). Skin paddle contracture improved significantly in flaps transduced with LVSOD2 (relative surface area, 53.8%; P < 0.05) or LVshCTGF monotherapy (relative surface area, 70.8%; P < 0.0001). The greatest improvement was observed with combination therapy (relative surface area, 86.5%; P < 0.0001) (Fig. 5, B and C).

Fig. 5 LVSOD2 and LVshCTGF therapy reduces volume loss and skin contracture after RT.

(A) Phenotypic appearance of irradiated (50 Gy/3 fractions) SIEA flaps (dashed white line, skin paddle outline) at 180 days after RT (scale bars, 10 mm). Note the appearance of LAEs such as telangiectasia in adjacent tissues into which neither LVSOD2 nor LVshCTGF were delivered (red arrows; images representative of n = 6 animals per group). (B) Quantification of the relative skin paddle surface area changes (mean ± 95% CI). (C) Comparison of relative skin paddle surface area means at day 180 after RT (mean ± SEM). (D) RTOG scores for acute toxicities after RT in flaps infected with LVSOD2 and/or LVshCTGF compared to PBS sham and LVsh-scram controls (mean ± SEM). (E) RTOG severity scoring for LAEs in irradiated flaps (left) and RTOG score component breakdown (right) for flaps infected with LVSOD2 and/or LVshCTGF compared to PBS sham and LVsh-scram controls (mean ± SEM). *P < 0.05, **P < 0.01, ***P < 0.001.

Flaps transduced with LVSOD2, alone or in combination, exhibited significant reductions (P < 0.01) in the duration, but not maximum severity, of acute toxicities after RT (Fig. 5D) and significant reductions in late RTOG severity scores (P < 0.05) (Fig. 5E). Flaps infected with LV-shCTGF alone showed transient improvements in late RTOG severity scores (Fig. 5E). Analysis of the RTOG component scores for skin, subcutaneous tissues, and joints showed that LVSOD2-infected flaps exhibited preservation of subcutaneous volume (Fig. 5E) and a more modest improvement in skin severity scores. Conversely, LVshCTGF-infected flaps had greater improvements in skin component scores but scored poorly with regard to volume loss (Fig. 5E). Combination therapy resulted in improvements in both skin paddle contracture and volume loss (Fig. 5E). All cohorts scored maximum severity points for joint-related changes, as would be expected given that joint tissues were extrinsic to the zone of vector delivery. These data demonstrate differential improvements in skin paddle contracture and flap volume loss attributable to LVshCTGF and LVSOD2, respectively, suggesting different mechanisms for RT-induced skin paddle contracture and volume loss.

In vivo MRI demonstrated significant improvements in flap volume only in flaps transduced with LVSOD2, either alone or in combination (P < 0.0001) (Fig. 6, A and B) compared to controls. Analysis of functional MRI data also demonstrated that flaps transduced with LVSOD2, alone or in combination, demonstrated normalization of R2* toward that seen in unirradiated flaps (Fig. 6C). These data suggest that LVSOD2 therapy is responsible for the preservation of flap volume after RT and that this is attributable to vascular preservation. Western blot analyses demonstrated sustained overexpression of SOD2 in irradiated flaps compared to those infected with LVshCTGF or controls (Fig. 6D).

Fig. 6 Combination therapy with LVSOD2 and LVshCTGF preserves flap volume and reduces fibrosis after RT.

(A) T2-weighted in vivo MRI of irradiated flaps (red arrows) at 180 days after RT (images representative of n = 3 animals per group). (B) MRI-derived volumetric analysis (mean ± SEM) comparing volume changes between groups (n = 3 animals per group). (C) Relative transverse relaxation rate (R2*) differences between groups at 180 days after RT (mean ± SEM) (n = 3 animals per group). (D) Western blot for SOD2 in irradiated and control flap tissues taken from different animals at 180 days after RT with 50 Gy/3 fractions (β-actin loading control). (E) Masson’s trichrome staining for collagen deposition in flap tissues taken at 180 days after RT from each therapeutic group. Images are representative of the larger cohort (n = 6), and whole sections are presented inset (top left). Scale bars, 100 μm. (F) Quantification of Masson’s trichrome staining shown in (E) expressed as total percentage of section exhibiting fibrotic change (mean ± SEM) and correlative RT-qPCR for Col1a2 gene expression in flap tissues 180 days after RT. (G) Multiplexed IF staining for GFP and red fluorescent protein (RFP) expression 180 days after RT in flaps infected with LVSOD2-RFP and LVshCTGF-GFP (both at 108 TUs) or negative controls (PBS) (scale bars, 100 μm). GFP and RFP colocalization is observed within vessel walls (L, lumen). Nuclei are counterstained with DAPI. Cells demonstrating either GFP expression alone (white arrows) or RFP expression alone (red arrows) are also highlighted (images representative of n = 6 animals per group). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Masson’s trichrome staining of irradiated flaps demonstrated greater fibrosis in control flaps compared to those infected with LVSOD2 and LVshCTGF, either alone or in combination (P < 0.0001) (Fig. 6, E and F). Post hoc analysis revealed that flaps infected with both LVSOD2 and LVshCTGF combination therapy exhibited significantly less fibrosis (% fibrosis, 1.7%) than those receiving either agent as a monotherapy (P < 0.05) and were not different significantly from unirradiated flaps (Fig. 6F). RT-qPCR revealed significant reductions in Col1a2 gene expression in flaps infected with LVSOD2, either alone or in combination (P < 0.01), but not in flaps infected with LVshCTGF compared to controls (Fig. 6F). To investigate the extent to which both viral vectors colocalized within the vascular compartment of the flap, we performed multiplexed immunofluorescent staining for GFP (LVshCTGF) and red fluorescent protein (LVSOD2) (Fig. 6G). Dual infection of cells was observed most commonly, with single-vector infection being observed less frequently (Fig. 6F, white and red arrows).

Flaps infected with LVSOD2, either alone or in combination, demonstrated significantly greater H33342 uptake (P < 0.0001) and EB leakage (P < 0.0001) and significantly less P adduct formation (P < 0.0001) (Fig. 7, A to D). Flaps infected with LVshCTGF alone demonstrated H33342 uptake and P staining similar to sham controls but greater EB leakage (Fig. 7, B to D); however, this was still less than observed with flaps treated with LVSOD2 (P < 0.001). Together, these data suggest that LVSOD2 preserves vascular function as demonstrated by improvements in perfused vasculature and vascular permeability after RT and the associated reduction in hypoxia. The contribution of LVshCTGF is to preserve vascular permeability, albeit to a lesser degree than LVSOD2, but this did not translate to a reduction in tissue hypoxia after RT.

Fig. 7 LVSOD2 acts through the preservation of microvascular function.

(A) Immunofluorescent imaging of perfused vasculature (H33342 and EB) and hypoxia [pimonidazole (P)] in flap tissues on day 180 after RT. Images are presented as merged composites (whole section in the upper panel and hpf of area depicted by white box shown in the middle panel) and split channels (lower panel) (scale bars, 100 μm; images representative of n = 6 animals per group). (B to D) Thresholded imaging analysis of IF staining for H33342 (B), EB (C), and P (D) showing percentage of section stained (mean ± SEM) (n = 6 animals per group). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

LVSOD2 therapy in normal tissues does not compromise the efficacy of RT

Because of concerns that SOD2 overexpression in stably transduced flap tissues might increase the survival of microscopic residual disease through release of SOD2, we performed conditioned medium experiments taking supernatant from HeLa-LVSOD2 “producer” cells and using it to inoculate naïve HeLa “target” cells. We found no evidence that SOD2 was transmitted between cell types (Fig. 8A).

Fig. 8 SOD2 overexpression in normal tissues does not compromise the cytotoxic efficacy of RT.

(A) Conditioned medium experiment to assess transmissibility of SOD2 overexpression after exposure of target cells (HeLa) to supernatant (HeLa-SOD2 supernatant) taken from producer cells (HeLa_SOD2) (blot representative of n = 3 biological repeats). (B) Schematic of the in vivo tumor recurrence model showing tumor (red arrow) growing in an SIEA flap (paddle outline dashed white). Animals with tumors greater than 2 cm in diameter were deemed to have exceeded the experimental severity limit and euthanized. (C) Tumor volume growth (mean ± SEM) for Mat B III tumors after RT in flaps infected with LVSOD2, LVeGP, or sham (PBS) (n = 5 animals per group). (D) Individual growth curves for tumors growing in sham (PBS)–infected and unirradiated, sham (PBS)–infected and irradiated (20 Gy/5 fractions), LVSOD-infected and irradiated (20 Gy/5 fractions), or LVeGFP-infected and irradiated (20 Gy/5 fractions) flaps (n = 5 animals per group). (E) Kaplan-Meier plot of survival to severity limit (2- cm tumor diameter) for animals with Mat B III tumors grown in sham (PBS)–, LVSOD2-, or LVeGFP-infected flaps. Median survival is shown at the bottom right. n/a, not applicable.

To elucidate the effects of tumor irradiation through a flap expressing radioprotective genes, we developed an in vivo model of tumor recurrence (Fig. 8B). Untreated tumors growing in sham-transduced flaps grew rapidly to the experimental severity limit, and RT induced a growth delay (Fig. 8, C to E). Tumors irradiated in LVSOD2-infected flaps exhibited reductions in tumor growth (Fig. 8, C to E), with four of five tumors achieving remission to 40 days. We repeated the experiment using a lentiviral vector (LVeGFP) to control for viral infection and found that tumor growth mirrored that seen in the sham group (Fig. 8, C to E). Median survival was not attained by 40 days for tumors that grew in LVSOD2-infected flaps (Fig. 8E).

In summary, these data support our hypothesis that SOD2-mediated radioprotection of normal tissues does not compromise the antitumor efficacy of RT. Normal tissue radioprotection may improve the RT response through maintenance of the normal tissue microenvironment.


The development of LAEs within normal tissues poses a challenging clinical problem; therefore, radioprotecting free flaps represents an attractive therapeutic strategy because it may offer patients an earlier reconstruction that is more durable. In conjunction with the clinical imperative, the technical procedure of performing a free flap offers a distinct therapeutic window for delivering viral gene therapies under exquisite anatomical control. This concept of FFGT was first described using a liposomal vector carrying the VEGF gene (53) but later characterized extensively by Gurtner and colleagues (45, 46). Unlike systemically delivered viral gene therapies, FFGT has fewer potential barriers to success because vectors are less susceptible to immune neutralization and can be delivered in higher titers to target tissues (54). Here, therefore, we used a combinatorial radioprotective strategy in an FFGT model.

Radiation-induced ROS mediate their effects over a hyperacute time frame, and SOD2 overexpression represents a very proximal point for therapeutic blockade of LAEs (19, 2326, 30, 41). Previous work has demonstrated the efficacy of a SOD2 gene therapy strategy delivered topically as a plasmid/liposome (23, 25, 26, 31, 55), packaged in an adenoviral vector (30) or delivered systemically (43, 56). In contrast, CTGF exerts its effects over a longer time course, as seen by its involvement in fibrotic disease states (57, 58). CTGF overexpression has been demonstrated in tissues exhibiting LAEs, but this has been mediated by Rho/Rho-associated protein kinase (ROCK) (rather than TGF-β1–SMAD) signaling (18, 34), and pharmacological intervention (such as statins as Rho/ROCK inhibitors) has reversed the established LAE phenotype in animal models (59, 60). CTGF-targeted therapies aim to reduce the drive to fibrosis and represent a distal therapeutic blockade of LAEs (37). Having observed reductions in SOD2 expression/activity and CTGF overexpression in our rodent model, we designed a dual-targeted preventative approach to augment the oxidative stress response and block fibrosis.

We showed that irradiating flaps with 50 Gy/3 fractions recapitulated the clinical development of LAEs and was associated with impaired vascular function and tissue hypoxia. Hence, this model represents an excellent platform in which to test preventative and therapeutic strategies. We confirmed that mitochondrial SOD2 overexpression is associated with increased endothelial cell survival after RT. In validating this strategy in vivo, we demonstrated that SOD2 is overexpressed and CTGF concentrations are reduced after lentiviral infection. SOD2 overexpression particularly was evident within the vascular compartment of flaps, and dual infection resulted in colocalization of transgenes.

Using conditioned-medium experiments, we confirmed that SOD2 overexpression was not transmissible between cells. To test the efficacy of RT against tumors growing in a radioprotected environment, we developed a tumor recurrence model and showed that LVSOD2 infection resulted in a more favorable response to RT. This was specific to SOD2 overexpression rather than viral infection. These data confirm that radioprotecting normal tissues should not compromise the efficacy of RT. The mechanisms underlying this observation are at present unclear and will require further evaluation. Previous studies have reported direct antitumor effects of SOD2 overexpression that are thought to be mediated by metabolic intermediaries (such as superoxide or hydrogen peroxide), inhibition of vascular endothelial growth factor, and epigenetic modification (6165). However, we favor an immunological hypothesis. That is, in the presence of normal tissue radioprotection, fewer distracting immunological signals may be generated after RT, leading the immune response to target the tumor cells, which are not radioprotected.

Although both LVSOD2 and LVshCTGF individually improved the LAE phenotype, the greatest gains were achieved by combining the two agents. The two vectors appeared to exert differential improvements on the LAE phenotype: LVSOD2 was more efficacious in preserving flap volume and vascular function, whereas LVshCTGF was more effective in preventing skin/flap contracture. These improvements were associated with reductions in fibrosis when delivered as monotherapies, and they were additive when using the combination. Finally, we confirmed that improvements in flap volume with LVSOD2 therapy (either alone or in combination) were due to preservation of vascular function and the subsequent reduction in stromal hypoxia. We showed that LVshCTGF monotherapy improved vascular permeability but did not reduce hypoxia after RT. These data suggest that CTGF blockade reduces stromal and perivascular fibrosis but this, on its own, is insufficient to prevent the development of hypoxia because endothelial dysfunction persists.

At the microvascular level, previous studies have shown sustained, acute inflammation of vasculature after irradiation associated with nuclear factor κB activation and overexpression of PAI-1 (13, 15), as well as differential regulation of SOD2 expression in irradiated arteries (66). Furthermore, a study using recombinant mice with homozygous deletions negative for endothelial adhesion molecules (ICAM1, VCAM1, E-selectin, and L-selectin) has shown that L-selectin−/− mice had increased survival and reduced pulmonary fibrosis after thoracic irradiation (67). This work demonstrates a central role for the endothelial compartment in mediating cellular responses to RT, and future studies might seek to investigate how SOD2 overexpression modulates adhesion molecule expression to differentially regulate immune and hematopoietic cell trafficking after RT. Therefore, we postulate that the intra-arterial delivery of a viral vector is the most effective way of achieving vascular radioprotection (48) because it achieves the greatest multiplicity of infection for the cells of the vascular tree, which are the cells of first contact for the viral particles.

In conclusion, these data provide preclinical evidence to support a combination strategy to augment proximal oxidative stress responses within the vascular compartment and mitigate distal end-stage fibrosis reactions in post-radiation LAEs in irradiated free flaps. These therapeutic goals can be combined additively and can be achieved without compromising the antitumor efficacy of RT. Translation of this concept should be feasible within the constraints of the technical considerations of free flap surgery and may allow for patients to be offered earlier reconstructions that are more durable against the side effects of adjuvant cancer therapies.


Study design

We identified a radiation fractionation regimen that generated a clinically relevant phenotype of LAEs and characterized this phenotype using clinical, imaging, immunohistochemical, and molecular end points. We validated the LVSOD2 and LVshCTGF vectors in vitro and evaluated their efficacy in radioprotecting endothelial cells. We assessed the therapeutic efficacy of the candidate vectors by infecting rodent free flaps before irradiation and followed the temporal changes in these flaps using the end points described above. To elucidate the mechanism behind the therapeutic effects, we performed functional vascular analyses using vascular stains for endothelial cells, vascular permeability, and hypoxia. We assessed the durability of transgene expression, tissue penetration, and colocalization of vectors in our in vivo model. To evaluate whether normal tissue radioprotection might compromise the efficacy of RT, we engrafted tumor cells into flaps infected with the LVSOD2 vector and performed tumor volume growth measurement after tumor irradiation. A single investigator performed surgical procedures and flap irradiation, and after RT measurements were performed in a blinded manner. Sample sizes were calculated on the basis of the primary outcomes (skin paddle contracture and flap volume) for the study and were derived from preliminary data generated by the fractionation pilot study.

SIEA flap irradiation

Flap irradiation was performed 30 days postoperatively under anesthesia [fentanyl (10 mg/ml) and midazolam (5 mg/ml), intraperitoneally]. Rats were placed in a radioprotective lead shield (6 mm) containing an aperture under which the flap could be positioned. Irradiation was performed using an orthovoltage (250 kV, 11 mA) x-ray machine (AGO) at a dose rate of 4.5 Gy/min. Animals were maintained for serial clinical measurements for skin paddle contracture and RTOG severity scores for 180 days after RT. At 180 days, flap tissues and contralateral control tissues were harvested for analyses under terminal anesthesia by performing intracardiac injection of 0.3-ml pentobarbital sodium with phenytoin sodium.

Tumor recurrence model

Animals underwent SIEA flap procedures as described above. Rat Mat B III breast adenocarcinoma cells (American Type Culture Collection) were trypsinized, pelleted, and counted. Cells (1 × 107) were prepared for injection into each animal. Cells were washed and centrifuged three times in phosphate-buffered saline and resuspended in a volume of 250 μl for injection. On the 26th postoperative day, cells were injected directly into the SIEA flap and monitored daily for tumor growth. Tumors were irradiated with 20 Gy in five consecutive daily fractions when they reached a diameter of 1 cm. Maximum tumor dimensions were measured, and volumes were calculated (width × length × depth × 0.5).

In vivo MRI

For MRI, rats were anesthetized with isoflurane and positioned within a birdcage 1H coil (inner diameter, 64 mm) in a 7-T horizontal bore microimaging system (Bruker BioSpin). Morphological multislice rapid acquisition with relaxation enhancement (RARE) fat-suppressed T2-weighted axial images were first acquired for both localization and subsequent determination of the flap volume. Contiguous 1-mm-thick images were acquired over a 6 × 6 cm field of view, using a 128 × 128 matrix, a repetition time (TR) of 4500 ms, an echo time (TE) of 33 ms, and 4 averages, giving an overall acquisition time of 4 min and 48 s. To quantify R2*, multiple gradient echo (MGE) images were then acquired from three 1.56-mm-thick transverse slices through the pedicle, using eight averages of 128 phase encoding steps over a 6 × 6 cm field of view, giving a temporal resolution of 3.5 min. MGE images were acquired using a train of eight echoes spaced 3.14 ms apart, initial TE = 6.21 ms, flip angle α = 45°, and TR = 200 ms. Tumor R2* maps were calculated from the MGE images by fitting a single exponential to the signal intensity TE curve on a voxel-by-voxel basis using a Bayesian maximum a posteriori approach (68). All post hoc MRI processing and image reconstructions were performed using either OsiriX DICOM viewer (OsiriX Foundation) or in-house software (ImageView, developed in IDL; ITT Visual Information Systems).

Statistical analysis

All data were presented as means ± SEM, with the exception of skin paddle surface area, which was reported as mean ± 95% confidence intervals. Differences among multiple group means were analyzed using one- or two-way ANOVA with post hoc analysis using Bonferroni’s correction. Differences between two group means were tested using a two-tailed t test. All statistical analyses were performed using GraphPad Prism 7.0a (GraphPad).


Materials and Methods

Fig. S1. Development and validation of a model of free flap LAEs.

Fig. S2. Validation of SOD2 overexpression using LVSOD2 and CTGF knockdown using LVshCTGF.

Table S1. Antibodies used, dilutions, and suppliers.

References (6971)


Acknowledgments: We would like to thank the following people for assistance in carrying out this work: F. Daley, A. McCarthy, D. Roberts, C. Gregory, A. Fletcher, L. Baker, and H. Barker. The HIVSiren plasmid was donated by G. Towers, (University College London) and is derived from a parent plasmid, CSGW (A. Thrasher, University College London). Rat fibroblast cells were donated by S. Irshad (King’s College, London). Funding: A.A.K. and J.T.P. were funded by the Wellcome Trust (WT098937MF and 200175/Z/15); The Royal College of Surgeons of England; British Association of Plastic, Reconstructive and Aesthetic Surgeons; and the Masons Medical Research Foundation. V.R., J.N.K., and K.J.H. were funded by the Institute of Cancer Research (ICR)/Royal Marsden Hospital National Institute for Health Research Biomedical Research Centre and the Rosetrees Trust. K.J.H. also received funding from the Oracle Cancer Trust and the Anthony Long Trust. S.P.R. acknowledges the support received for The Institute of Cancer Research’s Cancer Research UK and Engineering and Physical Sciences Research Council Cancer Imaging Centre (grant no. C1060/A10334) and to the Cancer Research Cancer Imaging Centre (grant no. C1090/A16464) in association with the Medical Research Council and Department of Health (England). M.H. was funded by the Cancer Research Funds of Radiumhemmet, the Stockholm County Council, and the Swedish Society of Medicine. Author contributions: A.A.K., K.J.H., and P.A.H. contributed to the conceptualization and study design. A.A.K., M.M., J.T.P., J.N.K., M.J.W., T.P., D.M., and V.R. designed and executed in vitro studies. A.A.K., R.S., J.T.P., and T.P. executed in vivo surgical studies. J.K.R.B. and S.P.R. performed imaging analyses. A.A.K. and M.H. performed data analyses. A.A.K. and K.J.H. wrote the manuscript. N.S., H.S.P., R.G.V., A.A.M., and P.A.H. performed critical review of the data and oversight. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Materials can be obtained from the Institute of Cancer Research (ICR) via a material transfer agreement.

Stay Connected to Science Translational Medicine

Navigate This Article