Dietary Microbes Modulate Transgenerational Cancer Risk

Inflammation-associated, metabolic, and neoplastic diseases have increased in frequency in the industrialized world due to numerous factors, including reduced activity levels, increased total energy intake, environmental carcinogens, and immune status (1, 2). Compelling hypotheses aim to explain this phenomenon and reshape preventive medicine. According to one microbe immune-centric hypothesis, inhabitants of developed countries have immune systems of reduced lifelong regulatory capacity due to societal practices in the form of refined diets, antibiotics, and Caesarian births with insufficient beneficial microbe exposure during perinatal life (1–3). Impaired regulatory capacity leads to uncontrolled immune inflammatory responses, obesity, and cancer later in life (2, 3). Our recent studies in mice have shown that Westernized diet (NWD)-associated obesity and cancer coincide with changes in gastrointestinal tract microbial communities and immunoregulatory capacity preventable by dietary enrichment with beneficial bacteria (4–7). We test here whether these postulated effects of gut microbes transcend generations. We discover that mother mice consuming NWD during pregnancy convey detrimental long-term consequences to subsequent generations, in particular high rates of lung and liver cancer with obesity and premature senescence. We show this transgenerational predisposition is mediated by intestinal bacteria, highlighting a key role for microbial flora in reducing the risk of cancer.

Outbred conventional or germ-free CD-1 Swiss stock mice (Charles River) were housed and handled in Association for Assessment and Accreditation of Laboratory Animal Care-accredited facilities with diets, experimental methods, and housing as approved by the Institutional Animal Care and Use Committee. A genetically outbred stock with robust breeding capacity and absent transgenic predilections to cancer was selected for these transgenerational studies. To test the transgenerational impact of dietary microbes on progeny, the experimental design was to expose F0 (grandmother) mice to diets starting at the age of 8 weeks. Special dietary treatment continued until the birth of their pups. Progeny were later examined to determine health risks to subsequent generations. Euthanasia was performed using carbon dioxide overdose at 1 year of age, unless otherwise specified. Because of early-life morbidity, F2 progeny (grandchildren) were euthanized at 6 months of age or younger according to institutional humane criteria and clinical disease. Tissues were collected upon necropsy and then examined histologically. Each experiment included 5 to 10 animals per sex per treatment group, performed in duplicate, as described in detail below.

Outbred Swiss CD-1 mice were placed on experimental diets starting at 8 weeks of age: control AIN-76A (Harlan-Teklad) and a NWD with high fat and low fiber with substandard levels of vitamins B and D (TD.96096; Harlan-Teklad; Supplementary Table S1) as previously described (7). Subsets of CD-1 mice who had been treated with NWD in utero subsequently received in their drinking water an anti-inflammatory strain of Lactobacillus reuteri ATCC-PTA-6475 cultivated as described elsewhere (8, 9), with live organisms supplied at a starting dosage of 3.5 × 10⁵ organisms/mouse/day in drinking water (4). Live bacterial counts in water bottles were calculated to be 1.4 × 10⁶ colony-forming units (CFU) per mouse on day 1, 4.1 × 10⁵ CFU on day 2, and 1.1 × 10⁵ CFU on day 3; and L. reuteri was detectable by PCR in feces and bowel of mice undergoing the dosing regimen, as described in detail in Lakritz and colleagues (4). Control mice received regular drinking water. Fresh drinking water for both groups of animals was replaced twice weekly throughout the experiments.

Six eight-week-old CD-1 female mice were fed ad libitum a NWD diet TD.96096 mimicking fast-food starting at 8 weeks of age (Fig. 1A). Females who ultimately served as mothers and grandmothers were immediately arranged into breeding pairs to induce pregnancy. Special diets were replaced with control AIN-76A chow upon birth of F1 progeny. To serve as controls, 6 separate 8-week-old CD-1 female mice were fed ad libitum a control chow AIN-76A and similarly arranged into breeding pairs to produce experimental control progeny animals. All animals had ad libitum access to diets with unrestricted exercise. Their progeny were also examined for obesity and cancer when they reached 1 year of age, including body weights along with tissue collections upon necropsy for histologic examination.

To test whether gut microbes were sufficient for transgenerational effects, 5 germ-free female Swiss CD-1 mice were fed by gastric gavage the feces from NWD-fed mother mice in Experiment 1 (Fig. 1B). Five germ-free mice received comparable stool collected from control diet-fed moms in Experiment 1 to serve as controls. To achieve this, the germ-free mice were dosed 3 times every other day by gastric gavage with 0.05 grams of fresh fecal slurry per mouse per dose. Subsequently, both groups of fecal transplant-recipient moms were co-housed with a male mouse in autoclaved caging and fed only control chow AIN-76A diet to produce progeny for future examination. Progeny were examined for cancer at 1 year of age when tissues were collected upon necropsy.

Six F1 female progeny with in utero exposure to NWD from Experiment 1 (above) were randomly arranged into breeding pairs at 8 weeks of age to produce offspring (grandchildren) for further observation (Fig. 1C). Six F1 progeny from control diet–fed moms in Experiment 1 were similarly arranged into breeding pairs to serve as controls. Both groups were fed control AIN-76A diet throughout Experiment 3. Unanticipated fertility problems arose in half (3 of 6) of the matings involving NWD-fed mother mice, whereas control mice had no such issues. Two litters with ancestral exposure to NWD also exhibited a scurfy-like thymic atrophy and failure-to-thrive in 9 of 20 (45%) infant mice and were euthanized at 8 weeks of age. Surviving F2 progeny (n = 10 males and n = 10 females) were examined for cancer starting at 6 months of age (earlier than their parent's cohort) due to unexpected morbidity requiring euthanasia. Tissues were collected upon necropsy and then examined histologically to confirm diagnoses.

To further test the microbe-centric hypothesis, 3-week-old progeny (F1) from NWD-fed CD-1 mother mice in Experiment 1 were randomly subdivided into groups of 20 mice (10 males and 10 females per treatment) and received in their drinking water L. reuteri ATCC-PTA-6475 as described elsewhere (8, 9) continuously until 1 year of age (Fig. 1D). Mice from Experiment 1 that received NWD while in utero that then got regular drinking water served as controls. Tissues were collected upon necropsy at 1 year of age.

Genomic DNA was extracted from stool samples using the Qiagen QIAamp DNA Stool Mini Kit. Samples for paired-end Illumina sequencing were constructed using a 2-step PCR amplicon approach targeting the V4 region of the 16S rRNA gene (U515F and E786R) and reads were quality filtered and clustered into operational taxonomic units (OTU) at 97% nucleotide identity as previously described (7).

For histologic evaluation, formalin-fixed tissues were embedded in paraffin, cut at 5 μm, and stained with hematoxylin or immunohistochemistry (IHC) as previously described (3, 10). Primary antibodies for IHC included rabbit antibodies against myeloperoxidase, p53 (ThermoFisher Scientific/Lab Vision), IL17 (Santa Cruz Biotechnology, Inc.), Thyroid transcription factor 1 and Ki-67 (TTF1, Abcam). Heat-induced antigen retrieval was performed with citrate buffer, pH 6, for myeloperoxidase, with EDTA buffer, pH 8, for TTF1 and p53 detection or with CC1 epitope retrieval solution (Ventana Medical Systems, Inc.) for Ki-67 and IL17. Primary antibody binding was detected with goat anti-rabbit polymer HRP (ZytoChem Plus). Color was developed with DAB substrate-chromogen system (ThermoFisher Scientific/Lab Vision) and tissues were counterstained with hematoxylin.

The Mann–Whitney U test was used for whole body and thymic weight analyses. Tumor incidence in experimental groups was compared with the Fisher exact test. All analyses were performed with the GraphPad Prism version 5.0 for windows, GraphPad software. P < 0.05 was statistically significant.

Lifestyle changes, including Westernized diet, appear to underlie many of the chronic inflammatory diseases, including obesity, diabetes, heart disease, autoimmune diseases, and cancer (11). To test whether NWD diet-induced obesity (7) and neoplasia (4) may be passed to subsequent generations, we fed NWD to mother mice (F0) during pregnancy and then examined health outcomes in their children reared eating a regular control chow diet (Fig. 1a). For these studies, we used outbred white Swiss CD-1 stock without additional exposure to carcinogens. Mother mice of generation F0 consuming NWD for several weeks during pregnancy were not themselves affected by cancer and obesity in this study. However, we found that F1 progeny of moms consuming NWD during pregnancy had more frequent cancers when compared with offspring of mother mice eating control diet during pregnancy (Table 1).

Specifically, the most profoundly increased cancer types arising in mice with in utero exposure to NWD included pulmonary adenoma (3 of 14) and adenocarcinoma (11 of 14; 14 of 20; P < 0.0001), liver hepatocellular carcinoma (12 of 20; P < 0.0001), and spleen or mesenteric lymph node lymphoma (11 of 20; P < 0.003; Table 1, Fig. 2A–C). Other types of neoplasia were increased, although to a lesser extent, including mammary carcinoma (Fig. 2D), reproductive organ malignancies such as ovarian theca and granulosa cell tumors and cystadenoma, and primary hepatic hemangiosarcoma, small intestine adenoma, and islet cell adenoma of pancreas (Supplementary Fig. S1). In addition, progeny of NWD-fed moms were more likely to suffer obesity than their counterparts whose mothers had eaten a control diet (Fig. 3A), earlier linked with high levels of cytokine IL17 (7). Using IL17A-specific IHC, we discovered extensive fat pyogranulomas displaying high levels of IL17 protein in situ in offspring when their mothers had eaten the fast-food-style NWD chow (Fig. 3A). We also found higher abundance of Firmicutes and lower abundance of Bacteroidetes in NWD-treated mice consistent with our previous findings and also those of other groups who have studied obesity and metabolic syndrome (Supplementary Fig. S2; refs. 7, 12). Knowing that gastrointestinal tract microbial communities change within days after eating NWD (7), we hypothesized that microbial dysbiosis in pregnant mother mice placed their infants at increased risk for cancer.

To test our microbe-centric hypothesis, we used fresh fecal slurry containing gut bacteria from NWD-fed mother mice delivered by gastric gavage to recipient 8-week-old germ-free Swiss female mice subsequently arranged in matings (Fig. 1B). Control germ-free mice received comparable stool collected from control diet–fed moms. Subsequently both groups of stool-recipient moms ate only the standard control chow diet. When progeny were examined 1 year later, we found that mice receiving NWD-donor microbiota had higher rates of lung (P < 0.0001) and liver (P < 0.0001) cancers, at levels similar to those in offspring from NWD-consuming moms (Table 1). Conversely, offspring of germ-free mothers receiving control stool gavage rarely exhibited neoplasms in the target organs. It remains to be determined whether in utero activities of specific microorganisms result entirely from increased inflammatory tone or other mechanisms (13). Nonetheless, these findings demonstrated that transplantable fecal microbes were sufficient for increased risk of cancer in progeny mice.

Recognizing extensive pathology existing in F1 mice, we next tested whether subsequent generations are also at increased risk for cancer arising from ancestral in utero dietary indiscretions. Thus, we performed mating experiments among the F1 children of mother mice who had consumed NWD or control chow during pregnancy (Fig. 1C). Initially, infertile mating of Swiss F1 progeny [3 of 6 (50%) vs. 0 of 6; not significant] made it difficult to produce the necessary F2 grandchildren from NWD-fed grandmother mice. Upon further examination of this generation, F1 female nulliparous littermate progeny exhibited not only obesity but also polycystic ovaries (P < 0.001) and endometrial disease (Fig. 3B). In these same female mice, elongated anogenital distances (data not shown) and urogenital atypia suggestive of masculinization were not extensively characterized. Another feature in daughters of NWD-fed mothers, hepatocellular carcinoma, is more typical of male animals with high levels of IL6 (14). Specific causes of impaired fertility in these mice remain to be investigated.

Grandchildren (F2 progeny) emerging from fertile F1 mating ultimately exhibited lung (P < 0.0001) and liver (P < 0.0001) cancers with neoplasms arising at younger ages than those seen in their parents (Table 1). Similar early-onset transgenerational phenomena have been described in F2 progeny of rats undergoing experimental stressors while in utero (15), raising the possibility that in utero microbial “stress” is linked with underlying epigenetic mechanisms. Ancestral exposures to NWD were also manifested as failure-to-thrive (9 of 20 vs. 0 of 20; P < 0.05) in 8-week-old F2 progeny (Fig. 3C), affecting both male [50% (5 of 10)] and female [40% (4 of 10)] animals. The syndrome included thymic involution and inflammatory infiltrates throughout vital organs resembling “scurfy” mice with immunoregulatory deficits (16).

Recognizing that pathogenic gut bacteria trigger accumulations of neutrophils (17), we examined non-neoplastic tissues from descendants with gut dysbiosis and found increased numbers of intravascular and extravasated neutrophils, an observation supported by myeloperoxidase (MPO)-specific IHC in liver, adipose tissue pyogranulomas, and mesenteric arteries (Fig. 3C). In contrast, the presence of neutrophils in control diet- and LR-treated mouse tissues was minimal. It was previously shown that reciprocal relationships exist between neutrophils and host regulatory capacity (10, 18). Additional studies are needed to ascertain whether perinatal changes in microbe communities serve to modulate immunoregulatory capacity later in life. Taken together, these findings led us to conclude that microbial dysbiosis leads to transgenerational effects, including cancer, obesity, and accelerated aging. Knowing NWD-associated obesity and cancer are preventable by dietary enrichment with beneficial bacteria (4, 7), we hypothesized that oral microbe therapy may similarly reduce transgenerational cancer burdens.

Finally, to further test our microbe-centric hypothesis involving detrimental NWD-induced changes in gastrointestinal tract microbe communities, we applied a dietary enrichment strategy using beneficial bacteria originally isolated from human breast milk. Experiment 4 used F1 progeny of NWD-fed mothers (Fig. 1D), such that upon weaning at 3 weeks of age, experimental mice consumed a model microbial organism L. reuteri continuously in their drinking water. Control mice consumed regular water throughout the study. Upon later examination at 1 year of age, F1 children consuming L. reuteri had significantly lower risk for obesity (Fig. 3A) and cancer than did their untreated control counterparts (Table 1; Supplementary Fig. S3). It is promising but unknown to what extent other microbe cocktails or life stage interventions may be effective, although oral administration of purified L. rhamnosus was previously shown to suppress obesity in human adults (19). Mouse models and diets do not precisely mimic human conditions; however, these findings further supported a gut microbe immune-centric hypothesis and the possibility that diet-associated bacteria may be engineered to mitigate emerging inflammatory disease epidemic in Westernized societies.

In summary, we show that in utero effects of gut microbes may transcend generations. We discover that mother mice consuming a Westernized diet during pregnancy convey to their children and grandchildren detrimental long-term consequences, including high risk of obesity and cancer. In humans, there is precedent for maternal dietary folate (vitamin B9) deficiency leading to leukemia in offspring (20); although, the cancer outcome in the present study was reproducible using microbes alone, and mothers of the most severely affected F2 progeny were actually eating a nutritionally balanced control AIN76A diet during pregnancy. Finding polycystic ovaries, endometrial disease, and infertility in F1 female mice implicates sex hormones together with immune dysregulation in transgenerational effects; however, this remains to be explicitly tested. Importantly, fecal microbiome transplant was sufficient for the transgenerational cancer effect. Grandchildren were most severely affected, exhibiting a premature aging scurfy-like syndrome and high rates of cancer at a young age, when their grandmother had consumed Western diet during her pregnancy. The scurfy-like syndrome suggests immune T regulatory (Treg) cell insufficiency (16) previously implicated in obesity and cancer (4–7). Taken together, our data suggest that dietary stress leads to gut microbe dysbiosis that elevates the risk of heritable cancers. Intestinal bacteria were sufficient to remedy the transgenerational effect, pointing to novel microbe-based personal or public health preventative medicine strategies. Intestinal bacteria were sufficient to remedy the transgenerational effect observed in mice, and further studies of novel microbe-based mechanisms in the role of human cancer and obesity prevention are warranted.

In conclusion, it is now clear that profound changes in our environment have contributed to a societal health crisis. Our previous data have built upon the perinatal microbe-driven concept to include neoplastic diseases (3, 5). Our current data expand this microbe-centric view further to include mammalian progeny and suggest that the microbial flora of a mother animal affects the predisposition of her progeny to cancer. We have shown here that a key mediator of diet-induced transgenerational disease predisposition is the gut microbial flora. Consequently, the effect was reversible by enriching the progeny gut flora with beneficial bacteria. Another well-substantiated transgenerational hypothesis suggests that westernized dietary habits during pregnancy shape disease susceptibility profiles of her descendants via epigenomics (21–23). It remains unproven that maternal microbes contribute to epigenetic changes upon their unborn progeny, and how results from an animal model in a highly controlled environment translate into free-living human populations. Our data also suggest that the dietary indiscretions of earlier generations may be reversible by targeting more healthful gut microbes early in life. This suggests new possibilities for decreasing the risk of cancer, obesity, and other pathologies that appear more prevalent in modern societies, using targeted gut flora enrichment strategies. This suggests new possibilities for decreasing the risk of cancer, obesity, and other pathologies, using targeted gut flora enrichment strategies that remain to be validated in humans.

No potential conflicts of interest were disclosed.

Conception and design: T. Poutahidis, B.J. Varian, S. Mirabal, S.M. Kearney, E.J. Alm, S.E. Erdman

Development of methodology: B.J. Varian, S. Mirabal, S.E. Erdman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Poutahidis, T. Levkovich, S. Mirabal, Y.M. Ibrahim, S.E. Erdman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Poutahidis, C. Kwok, S.E. Erdman

Writing, review, and/or revision of the manuscript: T. Poutahidis, B.J. Varian, S.M. Kearney, S.E. Erdman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Poutahidis, T. Levkovich, J.R. Lakritz, S.M. Kearney, A. Chatzigiagkos, S.E. Erdman

Study supervision: S.E. Erdman

Other (performed the experiments): C. Kwok

The authors thank James Versalovic for the gift of ATCC-PTA-6475 L. reuteri and special thanks to James G. Fox for encouragement and support.

This work was supported by NIH grants P30-ES002109 (pilot project award to S.E. Erdman and E.J. Alm), RO1CA108854 (S.E. Erdman), and U01 CA164337 (S.E. Erdman and E.J. Alm).