Ultraviolet Radiation Inhibits Mammary Carcinogenesis in an ER-Negative Murine Model by a Mechanism Independent of Vitamin D₃

Seven decades ago, Apperly postulated that the decreased incidence and decreased mortality of colorectal, breast, and several other cancers at lower latitudes are due to increased ambient sun light exposure (1), which the Garlands postulated acts by producing D₃ (2). This idea has contributed to the widespread use of oral D₃ supplementation in the hope of protecting against various maladies beyond abnormalities of bone mineralization (3). However, recent publications indicate increasing doubt about the range of action of D₃ (4, 5). Despite the reporting of numerous careful studies in this field, few have assessed ultraviolet radiation (UVR)'s effect on autochthonous murine carcinogenesis. We report here that, consistent with Apperly's hypothesis, UVR does inhibit murine mammary carcinogenesis in the well-established C3(1)/SV40 Tag model (6) but that, in this specific model the inhibitory effect is independent of D₃, thus suggesting the existence of a D₃-independent anticancer effect of UVR.

All mouse studies were done as per the protocol approved by the CHORI IACUC. FVB/NJ and NOD/SCID mice were from The Jackson Laboratory (JAX, Sacramento, stock nos. 001800 and 001303, respectively). C3(1)/SV40 Tag transgenic mice provided by J. Green (NCI NIH, Bethesda, MD; ref. 7) were maintained by breeding with FVB/NJ nontransgenic mice. Transgenic animals were identified by PCR analyses of genomic DNA extracted from ear punches using a mouse genotyping kit (KAPA Biosystems) and primer sequences and PCR conditions as previously described (7). Mice from each litter were assigned randomly to different treatment groups. We enrolled 6-week-old mice in all studies unless otherwise specified. All NOD/SCID mice and C3(1)/SV40 Tag not enrolled in dietary studies were maintained on a normal D₃/normal minerals diet—D₃ 1,500 IU/kg, Ca 1%, phosphate 0.7%; TD2018: Harlan. Mice were housed under standard conditions (fluorescent lighting 12 hours per day, room temperature 23°C–25°C, and relative humidity 45%–55%). Animal care and use were in compliance with protocols 197 and 201 approved by the Institutional Animal Care and Use Committee (IACUC) of Children's Hospital Oakland Research Institute (CHORI).

C3(1)/SV40 Tag transgenic mothers and pups were weaned onto and maintained on a D₃-depleted diet. At age 6 weeks, mice were enrolled randomly in one of four dietary groups: D₃-depleted, D₃ normal, D₃ rescue, or a high D₃ diet (Table 2).

We used a UVR irradiation unit (R2025-bx-0012; Daavlin Co.) with a bank of F40-T12 sunlamps arranged in parallel, emitting UVB (280–315 nm), and equipped with an electronic controller to regulate dosage. The dose was calibrated with an independent external UV light meter (925GOX96; Daavlin Co.). UVR treatments were delivered consistently as described (8). In brief, mice were shaved weekly and treated at the same time of the day in a cage without bedding placed below the same areas of the device with a distance between light sources and mouse skin of 20 cm.

Transgenic mice were treated 3×/week with 350 mJ/cm² UVR (i) starting at age 6 weeks until euthanasia or until the age when the first tumor became palpable or (ii) starting at the time of appearance of the first palpable tumor until euthanasia. NOD/SCID mice were treated 3×/week with 350 mJ/cm² UVR starting from the age when the longest tumor diameter reached 1 cm until euthanasia. To study UVR-induced cyclopyrimidine dimers, we exposed 4- to 5-month-old NOD/SCID mice to a single dose of UVR (350 mJ/cm²), euthanized mice 30 minutes later, and collected skin and allograft tumors for histologic examination.

Transgenic D₃-depleted 6-week-old mice were randomly assigned to groups treated with either topical D₃ or acetone vehicle control (n = 9 mice/group). We treated mice in the D₃ topical group by applying cholecalciferol (0.38 μg/day, Sigma-Aldrich) to a small shaved area of the dorsal neck, 5 days/week in 20 μL of acetone, age 6 to 12 weeks. Control mice received topical acetone 5 days/week for the same duration. We confirmed that mice do not lick the application site by monitoring the persistence of a water-soluble ink at this site and its disappearance at a lickable site. NOD/SCID mice received 3.8 μg/day of cholecalciferol starting from the age when the longest tumor diameter reached 1 cm until euthanasia. All enrolled mice were housed singly and were shaved dorsally weekly.

M28 and M6 mouse mammary carcinoma cell lines derived from C3 (1)/SV40 Tag mice (9) were cultured in Dulbecco's modified Eagle Medium supplemented with 10% (v/v) FBS (UCSF Cell Culture Facility) and penicillin/streptomycin. At 70% confluence, 1 × 10⁶ cells or 2 × 10⁶ cells (100 μL; as indicated on the figures) were resuspended in DMEM/Matrigel (1/1, v/v) and injected (100 μL/per injection) subcutaneously into each flank of 6-week-old shaved NOD/SCID mice fed a D₃ normal diet (TD2018, Harlan Teklad). Twenty-six mice received M28 cells, and 6 received M6 cells. When a single tumor had reached 1 cm in longest diameter (after approximately 4 weeks, i.e., at ∼10 weeks of age), NOD/SCID mice were assigned randomly to one of the four treatment groups: UVR or untreated control and topical D₃ or topical acetone. Individual body weights and tumor sizes were measured weekly as described below.

C3(1)/SV40 Tag mice develop DCIS as early as age 9 weeks and palpable mammary tumors at approximately age 16 weeks. To analyze the effect of UVR on progression during the early stage of carcinogenesis, we euthanized C3(1)/SV40 Tag mice at ages 9, 11, 13, 14, and 15 weeks and collected all 10 mammary glands from each mouse (n = 3–5 mice/group). Histologic analysis of H&E-stained sections was performed blindly by the breast cancer pathologist (P. Du), and the percentages of mammary glands with atypical hyperplasia, DCIS, and invasive carcinoma were determined.

In studies of the autochthonous model, we recorded body weights weekly as well as time to first palpable tumor (subsequently confirmed histologically as a mammary tumor), numbers of palpable tumors, and sizes of tumors starting at age 10 weeks. In studies of mice bearing allograft tumors, we recorded mouse body weights and tumor sizes weekly starting at age 6 weeks (i.e., at the time of allograft placement). We used an electronic caliper (10) to measure the length, width, and depth consistently, and we calculated tumor volume using the following formula: length × width × depth × 0.52 (11). Mice were euthanized when a single tumor reached a longest diameter of 2 cm or the sum of multiple tumors approached 9% of body weight, as per our IACUC-approved protocol. At euthanasia, blood was taken by cardiac puncture for D₃ analysis; tumors, mammary glands, and skin samples were collected for histology, flow cytometry, and cell sorting.

Collected tissues were fixed in paraformaldehyde 4% overnight at 4°C. Prior to commercial automated paraffin processing (Redwood Dermatopathology Laboratory), tissues were washed twice with phosphate-buffered saline and transferred to isopropanol 70%. Four-micrometer sections were subjected to standard immunostaining procedures as described previously (12). The following primary antibodies (Ab) were used: anti-SV40 TAG mouse monoclonal Ab (1:100, clone PAb 101, #554149; BD Pharmingen), anti–Ki-67 rabbit monoclonal Ab (RM-9106-R7; Thermo Fisher Scientific), and anti-CPD Ab (MC-062; Kamiya Biomedical Company).

D₃ and 25(OH)D₃ levels were measured commercially (Heartland Assays LLC & Metabolic Technologies, Inc.). Radioimmunoassay (RIA) was used to quantify 25(OH)D₃, while D₃ was measured by LC/MS/MS using an Agilent 1290/6460 Series Triple Quadrupole System and a deuterated internal standard.

We detected apoptosis and necrosis in M28 allografts using Alexa Fluor 488 labeled Annexin V and propidium iodide (PI) staining by flow-cytometric analysis, following the manufacturer's protocols (Invitrogen). Briefly, collected allograft tumors were digested in 0.1% collagenase IV (C5138, Sigma-Aldrich) to obtain single-cell suspensions. Cells were washed in PBS and resuspended in buffer (50 mmol/L HEPES, 700 mmol/L NaCl, 12.5 mmol/L CaCl₂, pH 7.4). After addition of 5 μL Annexin V and 1 μL PI, cells were incubated at 37°C in light-protected vials for 20 minutes. Samples were diluted with binding buffer and were analyzed using the BD LSR-Fortessa (BD Bioscience). Results were analyzed using FlowJo single-cell analysis software 8.5.2 (TreeStar).

Kaplan–Meier survival analyses, determination of significance of difference between groups using log-rank test, Gehan–Wilcoxon test and Student t tests were carried out using GraphPad Prism 6 software. Values are presented as mean ± SD or mean ± SEM; those with P ≤ 0.05 were considered to be statistically significant. All studies were designed to yield at least n = 30 for each group of transgenic mice and n ≥ 3 for each group of NOD/SCID. Data were collected from multiple litters; mice from each litter were randomized into different treatment groups. Each data point represents a biological replicate. The investigators were not blinded to allocation during experiments and outcome assessment. For all experiments presented in this study, the sample size was large enough to measure the effect size. All statistics were done in consultation with the CHORI statistician G. Gildengorin.

We have tested the hypotheses of Apperly (1) and of the Garlands (2) using C3(1)/SV40 Tag mice (6), in which a promoter active in the mammary gland drives expression of the SV40 T antigen, which inactivates both p53 and RB proteins. Mammary carcinogenesis in these mice appears most closely to resemble the triple negative, basal form of human breast cancer (13–16). We exposed the shaved dorsal skin of these mice to UVR (350 mJ/cm²) three times per week starting from age 6 weeks. We found in multiple independent experiments that UVR treatment significantly delayed the appearance of the first palpable mammary tumor (age 18.9 vs. 16.8 weeks, combined P = 0.00074; Fig. 1A and C), increased overall survival (age 24.8 vs. 22.8 weeks, combined P = 0.0037; Fig. 1B and D), and reduced the number of palpable mammary tumors per mouse at age 25 weeks (4.3 ± 2.1 vs. 6.7 ± 1.0 tumors, P = 0.048; Fig. 2B, red dashed vs. black solid lines). UVR's inhibitory effect occurred in mice fed either a D₃-depleted/high minerals diet or a diet with normal amounts of D₃ and minerals (Tables 1 and 2). In mice fed a diet with normal D₃/high minerals (D₃ rescue diet), UVR delayed the development of palpable tumors (Fig. 1C) but, inexplicably, had no significant effect on either overall survival (Fig. 1D) or tumor multiplicity. UVR did not change expression of the oncogenic T antigen (Fig. 1E). As expected (8), this UVR dose produced skin tumors of the squamous cell carcinoma (SCC) lineage in these mice starting at age 25 weeks.

The antitumor effect of UVR occurred early during carcinogenesis: (i) UVR started at age 6 weeks prolonged overall survival and reduced the number of palpable mammary tumors equally well if continued until death or if stopped at the time when the first tumor became palpable (Fig. 2A and B in red; Table 1); (ii) UVR treatment started after the first tumor became palpable had little to no effect on overall survival (P = 0.5) or on the number of tumors (Fig. 2A and B in black); and (iii) UVR started at age 6 weeks reduced the percentage of mammary glands with DCIS at ages 9 to 13 weeks but not at ages 14 weeks and later (Fig. 2C). UVR significantly decreased proliferation in histological samples with DCIS at ages 10 to 15 weeks as measured by Ki67 protein expression.

We next compared the effects of UVR with those of orally ingested D₃. We weaned female pups from mothers fed D₃-depleted chow onto diets containing D₃ in amounts varying from none to a level 20-fold higher than normal (Tables 1 and 2). In mice not exposed to UVR, these dietary changes failed to affect significantly the number of palpable mammary tumors that formed (Fig. 3A), the age at first palpable tumor (Fig. 3B), or the age at which IACUC guidelines mandated euthanasia. This lack of phenotypic effect occurred despite elevation of circulating 25(OH)D₃ caused by oral D₃ ingestion to levels similar to those induced by UVR treatment (D₃ normal diet; 31 ± 8 vs. 40 ± 9 ng/mL, respectively; P = 0.33) or to levels 4-fold higher (high D₃ diet; 31 ± 8 vs. 112 ± 31, P ≤ 0.01; Fig. 3C).

To more closely mimic the effects of D₃ produced in the skin, we applied 15.2 IU (0.38 μg)/day D₃ (cholecalciferol) to the shaved dorsal skin of the neck (a site that mice cannot self-lick) of singly housed mice that were fed a D₃-depleted diet. This normalized circulating 25(OH)D₃ and increased circulating D₃ to levels similar to those achieved by UVR treatment (Fig. 3D). Nonetheless, topical D₃ failed to affect early-stage mammary carcinogenesis as assessed by histologic analysis of mammary glands from 12-week-old C3(1)/SV40 Tag mice (Fig. 3E). Of note, the same dose of topical D₃ applied to the back of the neck robustly delayed the development of visible BCCs in Ptch1^+/− mice (17), thus indicating its successful absorption and distribution to other tissues.

Finally, we produced mammary tumor allografts on NOD/SCID mice using cell lines derived from a histologically normal mammary gland from an 8-week-old FVB/NJ-Tag transgenic mouse (M28) and from an invasive mammary carcinoma from an older mouse (M6; ref. 9). NOD/SCID mice themselves and their mothers were fed diets containing normal amounts of D₃ (Table 1). The M28 allograft tumors from the early-stage cell line were sensitive to inhibition of growth by UVR (Fig. 4A). By contrast, M6 allograft tumors from the invasive carcinoma cell line were resistant to UVR (Fig. 4B), mirroring the stage-specific effects on autochthonous tumors in UVR-treated mice outlined above. UVR inhibition of the growth of the M28 allograft tumors occurred despite UVR's failure to increase serum D₃ in host NOD/SCID female mice (Fig. 4C), confirming that UVR has a D₃-independent effect on tumorigenesis. The tumor inhibitory effect of UVR was not due to direct irradiation of the allografts because (i) UVR produced abundant cyclobutane pyrimidine dimers (CPD), classic indicators of UV-induced DNA damage, in the overlying epidermis but none in the tumor allografts or in the dermis (Fig. 4D) and (ii) UVR inhibited the growth of the allograft tumors underlying the ventral (unexposed; Fig. 4E) as well as the dorsal (exposed; Fig. 4A) skin.

In marked contrast to UVR-induced inhibition, topical application of D₃ (cholecalciferol) to the back of the neck of NOD/SCID host mice at a dose of 152 IU/day (3.8 μg/day), a 10-fold higher dose than used in our autochthonous tumor studies, failed to blunt the rate of growth of the tumor allografts (Fig. 4F), despite raising circulating D₃ levels (Fig. 4C). Correspondingly, UVR significantly increased apoptosis in allograft tumors (P = 0.0001), whereas topical D₃ had no effect on apoptosis (Fig. 4G).

In consonance with our findings, a very rarely cited paper from Apperly (18) reported in 1945 that UVR, but not orally administered D₃, reduced mammary cancer in a spontaneously cancer-prone strain of A/J mice. Thus, the anti-mammary cancer activity of UVR in these two mouse models is consistent despite the passage of seven decades and the use of different sources of radiation and different strains of mice. These data, moreover, are consistent with the beneficial effects of UVR on experimental murine autoimmune encephalomyelitis, a model of human multiple sclerosis in which the effect is also unrelated to vitamin D₃ levels (19). UVR's inhibitory effect on murine mammary (extracutaneous) cancer differs markedly from its stimulation of cutaneous SCC carcinogenesis, an effect that is due, at least in part, to immune inhibition (20). The maintenance of efficacy of UVR versus the Tag mammary carcinoma allografts in host NOD/SCID mice significantly depleted in T and B cells, suggests that UVR's anti-carcinogenesis carcinogenesis mechanism is independent of the adaptive immune system. These data identify the FVB/NJ C3(1)/SV40 Tag mouse as a tractable model for the elucidation of a D₃-independent mechanism of UVR's anticancer effect but do not in any way conflict with the sizable previous body of data indicating that, in other preclinical cancer models, vitamin D₃ has significant anticancer effects.

To the degree that these murine findings reflect the human condition, they may be highly significant for public health—they support the idea that sun exposure may reduce the incidence of cancers of the breast and perhaps of the colon and other organs (21–23). Indeed, some data suggest that more sun exposure is associated with reduced overall mortality (24, 25). However, unlike the majority of past studies, one recent epidemiologic study failed to find any effect of exposure to sunlight on human breast cancer development (26). Our findings indicate that although sun protection to prevent skin cancers might increase extracutaneous cancer risk, neither oral nor topical D₃ replacement will reverse whatever portion of the increased risk is mediated by non-D₃ mechanisms. However, more happily, our findings suggest that elucidation of the full mechanism of UVR's anticancer effect could uncover new preventive approaches to reduce the incidence of breast cancer in women living at higher latitudes to that of women living closer to the equator and perhaps even further in women everywhere (27).

No potential conflicts of interest were disclosed.

Conception and design: F. Frascari, E.H. Epstein Jr

Development of methodology: A.M. Makarova, P. Davari, F. Gorouhi, G. Wang, E.H. Epstein Jr

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.M. Makarova, F. Frascari, P. Davari, F. Gorouhi, L. Wang, G. Wang, J.E. Green, E.H. Epstein Jr

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.M. Makarova, F. Frascari, F. Gorouhi, P. Dutt, A. Dhawan, J.E. Green, E.H. Epstein Jr

Writing, review, and/or revision of the manuscript: A.M. Makarova, F. Frascari, P. Davari, F. Gorouhi, E.H. Epstein Jr

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Wang, A. Dhawan, G. Wang, E.H. Epstein Jr

Study supervision: A.M. Makarova, E.H. Epstein Jr

We thank G. Gildengorin for discussions on statistical analysis, Y. Khaimsky for mouse colony management, J. Dolorito and E. Libove for mouse work, and for help with animal monitoring, etc. UC Berkeley undergraduate and post-graduate student assistants: Ting Deng, Micah Fry, Ankur Gupta, Subheksha KC, Amanda Weston, and Carla Nicole Wood.

This work was supported by NIH R01CA142879 (E.H. Epstein Jr), The American Institute for Cancer Research (AICR) 10A103, and the joint UCB-CHORI T32 training grant (A.M. Makarova).

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.