To investigate influenza infection in the mother-infant dyad, we intranasally inoculated 4-week-old infant ferrets with the 2009 H1N1 virus, A/California/07/2009 (Cal/07), at 10⁵ EID₅₀. Infants were housed with nursing-mothers before and after inoculation. Nasal washes (NW), temperature, weight, clinical assessment, and survival were collected daily as shown in the design schematic (Fig 1A). Necropsies were performed at Day 3, 4, 7, and 14 post-inoculation or if an animal was removed from the study (reached weight cut-off or had succumbed to illness). Inoculated infant ferrets developed illness leading to mortality starting with a high fever at 104% of baseline temperature (p<0.05) Day 2 post-infant-inoculation. Infants then developed hypothermia (Day 6) (Fig 1B). All infants reached weight cut-off (lost 20% of original weight; p<0.05) or died by Day 8 post-infant-inoculation (Fig 1C and 1D). Four days post-infant-inoculation, nursing-mother ferrets also developed influenza-like symptoms including increased body temperature (103% of baseline). Mother weights significantly decreased by 13% Day 4 post-infant-inoculation (Fig 1C). Weight did not recover by study end and one mother reached cut-off (Fig 1D). The weight and temperature of the mothers of mock inoculated control infants are presented in S1 Table which showed mothers did not develop symptoms post-mock-inoculation of infant ferrets.

As the mothers of inoculated infants began to display clinical signs of influenza including temperature increases, weight loss, and mortality, we went on to analyze the upper and lower respiratory tracts of the infants and their mothers to determine virus transmission and pathogenesis. Viral burden in infant NW was first detected Day 1 post-infant-inoculation (Fig 2A). The NW of nursing-mothers were positive for influenza virus Day 3 post-infant-inoculation. Viral titers remained high in mother NW Day 6–7 (Fig 2A) at more than 6 TCID₅₀/ml (Log₁₀). Residual virus persisted in the mothers’ NW by Day 10 (S1A Fig). Virus transmission between adult ferrets pair-housed (S1C Fig) were used as a comparison of transmission dynamics between infants and mother ferrets. For this control, one adult was intranasally inoculated under similar circumstances as the infant-mother inoculations and NW were collected for viral load assessment. Naïve cage mates of inoculated adult ferrets began shedding virus in NW Day 3 post-inoculation. These results showed that the influenza virus was able to be transmitted from direct intranasally inoculated infant ferrets to the upper respiratory tracts of their nursing-mothers on a similar time scale as that of adult ferrets.

We next determined if transmission of the virus from infants to mother led to lower respiratory tract infection. Examination of viral titers in the lower respiratory tract of the mothers found virus levels above 6 TCID₅₀/ml (Log₁₀) in both the trachea and lungs post-infant-inoculation in 2 out of 3 mothers investigated (Day 4 and 7 post-infant-inoculation) (Fig 2B). H&E (hematoxylin & eosin) staining of mother lungs revealed delayed, airway-localized inflammation (Fig 2C). Day 3/4 post-infant-inoculation, mother lungs showed areas of minimal infiltrating leukocytes (black arrows: diffuse areas; green arrows: dense infiltration) (Fig 2C). Increased leukocyte infiltration was observed Day 7 post-infant-inoculation where sites of inflammation were dense in mononuclear cells with lymphocyte-like morphology. Infected mother lungs were compared to the lungs of adult male ferrets directly inoculated with Cal/07 through the intranasal route (Fig 2C, right column). The lungs collected from mothers of mock inoculated infants were similar in architecture to that of mock inoculated adult male ferrets. The adult lungs collected Day 3/4 post-direct-inoculation had markedly more leukocyte infiltrates compared to the mother lungs at Day 3/4 post-infant-inoculation (Fig 2C). Taken together with the above data, these findings indicated that 2009 H1N1 transmission from infant to mother led to severe respiratory disease with upper and lower respiratory tract infection and accompanying lung pathology but on a delayed time course compared to direct adult infections.

To determine if influenza infected nursing-mother ferrets could transmit the virus to their feeding-infants, we reversed the inoculation/infection design as shown in the schematic (Fig 3A). We intranasally inoculated nursing-mothers with Cal/07 at 10⁵ EID₅₀ and observed clinical symptoms as well as determined respiratory viral load and pathogenesis. Following inoculation, nursing-mother ferrets developed a biphasic fever which spiked at 104% (baseline temperature) (Fig 3B) Day 3 post-mother-inoculation. Weight loss was observed in the inoculated mothers which reached nadir on Day 7 post-mother-inoculation at 89% of original weight (p<0.05) (Fig 3C). Although mother mortality did not occur, the infants of inoculated mothers lost weight, had increased temperature, and all died or reached cut-off by Day 11 post-mother-inoculation (Fig 3D). The weight of infants of inoculated nursing-mothers began to decline on Day 3 (Fig 3C). Although infants did not have a significant increase in temperature they did become hypothermic as they reached mortality (Fig 3B). Virus was detected in mother NW Day 1 post-mother-inoculation (peak Day 2 at 6 TCID₅₀/ml (Log₁₀)) and infant viral shedding began Day 3 (Fig 3E). Live influenza virus was detected in the lungs of infants reaching weight cut-off or succumbing to illness between Day 5 and Day 8 post-mother-inoculation (3–8 TCID₅₀ (Log₁₀)) (Fig 3F). Significant pathology was also detected in lungs of infants feeding from 2009 H1N1 inoculated nursing-mothers Day 7 post-mother-inoculation. The lungs were characterized by areas of dense leukocyte infiltration and alveoli destruction (Fig 4). In summary, these data indicate that influenza transmission was also possible from mother to feeding-infants causing upper and lower respiratory tract infection and respiratory tract disease.

Mammary glands are a significant point of contact and possible pathogenic transmission within the mother-infant dyad [34]. To investigate the susceptibility of mammary glands to influenza infection and the mammary as a point of transmission, we determined virus presence within mammary tissue and virus shed in expressed milk as shown in the experimental design graphic (Fig 5A). All mammary glands were collected from mothers of directly influenza inoculated infants on Day 4 and Day 7 post-infant-inoculation (3 per time point). Live viral load was quantified from each mammary tissue. MDCK titration assay revealed live virus in 7 of the glands (Fig 5B) where some viral loads reached 6 TCID₅₀/g (Log₁₀) or higher. All mothers assessed had at least one gland positive for live influenza virus. Nipples of positive mammary glands also had significant amounts of live influenza virus on Day 4 post-infant-inoculation (3–7 TCID₅₀/g (Log₁₀)) (Fig 5C).

To determine if influenza virus positive mammary glands were able to shed virus in the milk expressed from the infected mammary gland, we measured both live virus and vRNA in expressed milk from mothers of infected infants. Live virus assessment in milk collected between Day 3 and Day 7 post-infant-inoculation revealed virus presence where some samples contained high virus levels between 6–7 TCID₅₀ (Log₁₀) (Fig 5D). vRNA analysis from milk collected on separate occasions showed vRNA was present in some milk samples as high as ~10,000 copies/5 ng totRNA. Most samples collected had between 10 and 1000 copies (Fig 5E). vRNA and live virus were not detected in milk collected at baseline (Fig 5D and 5E). High levels as well as moderate and low levels of vRNA were detected in infant ferret feces (Fig 5F). Negligible or no amounts of vRNA were found in the peripheral blood of intranasally inoculated infants or their mothers to account for viremia contributing to severe disease or virus shedding from milk (Fig 5G). Control directly inoculated adults were also negative for blood vRNA. This data suggested that mammary glands are able to be harbor live influenza virus and viral shedding can occur through milk expression.

We next visualized the virus in 2009 H1N1 influenza positive mammary glands (H1N1+ MG) to determine virus localization and gland pathology following infection. Histopathology and virus localization in H1N1+ MG revealed destruction of mammary architecture (Fig 6). The acini of control glands had either thick epithelial cell layers indicative of active milk production (green arrows) or acini filled with pink proteinaceous fluid (milk) (blue arrows) (Fig 6). Virus staining was not detected in Control mammaries (top right panel). At Day 4 post-infant-inoculation, the glands of nursing-mothers had retained tissue architecture although leukocyte infiltration (yellow arrows) was identified. Foci of virus staining (light brown in color) was also observed (Day 4 post-infant-inoculation). By Day 7, marked increases in leukocyte infiltration (yellow arrows) in and surrounding the lobules was noted in some mammary glands (Fig 6) where the lobules of these glands had lost typical mammary acini architecture (bottom left panel). Increased viral staining was seen by Day 7 post-infant-inoculation with loss of significant tissue structure (bottom right panel). Viral staining was more pronounced in epithelial cells (black arrows) (semi-quantitative analysis). These data provide evidence that the influenza virus is capable of infecting cells of the mammary gland.

To uncover the molecular mechanisms of influenza infection within the mammary glands we performed gene expression analysis on RNA isolated from H1N1+ MGs. Global gene expression analysis by hierarchical clustering using Pearson correlation revealed increased expression of innate immune and cell remodeling genes, as well as downregulation of metabolism and lymphocyte activation genes (Fig 7A).

Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway analysis showed pronounced alterations in signaling gene expression in H1N1+ MGs (Fig 7B; Ingenuity Pathway Analysis S2 Fig, Gene Enrichment Scores S2 Table). Jak-STAT (Janus Kinase-Signal Transducers and Activators of Transcription) signaling was prominently affected, highlighted by increased STAT3 and involution signaling (STAT3, PIAS4, and SOCS5). Corresponding decreases were seen in STAT5A/B milk production and STAT associated gene networks PRL (Prolactin), STAT5A/B, SOCS1 (Suppressors of Cytokine Signaling 1), JAK3, EPO (Erythropoietin), and EPOR (Erythropoietin Receptor)) [35–38]. As well as PRL, other genes involved in milk production [39], LPO (Lactoperoxidase), LPL (Lipoprotein lipase), ATP2B2 (ATPase for Ca2+ transport during milk production), SLC34A2 (protein component of milk), LALBA (alpha-lactalbumin), and CSN2 (β-casein), were also significantly downregulated (Fig 7B).

Cancer Related and Cell Cycle genes including Wnt, p53, and TGFβ signaling pathways, as well as cell attachment (Focal Adhesions and Adherens Junctions) gene networks, were regulated in H1N1+ glands. Significant expression of M2 macrophage genes were found including CD163, CLEC7A (C-type lectin domain family 7, member A), MSR1, CD209, LGMN (Legumain), and MRC1 (Mannose receptor C type 1) transcripts (Fig 7B). MMP1, MMP2, MMP8, MMP11, and MMP14 matrix metalloproteases were also increased along with collagen genes COL5A2, COL3A1, COL1A1, COl1A2, COL4A1, COL6A3, and COL4A6 (Fig 7B and S2 Fig). Upregulation of Integrins (ITGA4, ITGB5, ITGB1, ITGB6) and other genes involved in Focal Adhesions and Adherens Junctions (CAV1 (Calveolin 1), RAP1A, PAK1 (p21 Activated Kinase), ROCK1 (Rho Kinase), VAV1 (guanine nucleotide exchange factor), WASL (Wiskott-Aldrich Syndrome-Like), SPPL1) was also noted (Fig 7B and S2 Fig). Gene networks involved in Cell Cycle and Cell Proliferation/Stress were also prominent. These included MAP Kinase associated protein genes (MAP2K3, MAP2K6, MAPK8, MAPK13, MAPK1, MAP2K1, MAP2K4, JUN, and FOS) and Cyclin regulation genes (CDC16, CD4, CDKN2D, CDKN2D, CDKN1A, CDC26, CDK7, CDC45) (S2 Fig). Many of these genes have known roles in cancer gene networks such as the proto-oncogenes JUN and FOS, RAP1A (RAS-related protein 1A), ITGB5, as well as BRCA2 (Breast cancer 2, early onset) and PCNA (Proliferating cell nuclear antigen) (Fig 7B and S2 Fig). The P53 gene network included both pro- and anti-apoptotic gene upregulation (Bax, CASP9 (Caspase 9), CASP3 (Caspase 3), PTEN (Phosphatase and Tensin homolog), CDK4, and MDM4 (Mouse Double Minute 4)).

Functional annotation analysis by KEGG classification of significantly upregulated or downregulated genes similarly suggested pronounced transcription-level changes in cellular proliferation, remodeling, and metabolic pathways in H1N1+ MG. Genes associated with cell growth, morphology, and catabolism were significantly enriched among upregulated genes while genes implicated in lipid and protein metabolism were dominant among downregulated gene subsets (S2 Table). Seven signaling cascade gene classifications exhibited statistically significant enrichment among upregulated or downregulated genes for at least one time-point (Fig 7D). Among upregulated genes, cell growth (Wnt, TGF-β) and stress response (p53) signaling pathway genes were significantly enriched by Days 6/7 with a trend of increasing enrichment over time. A similar profile was observed for TLR signaling-associated genes. Conversely, genes implicated in Jak-STAT and PPAR signaling were significantly enriched among downregulated genes, with potential implications for milk production (Fig 7D). Microarray gene expression was validated by qRT-PCR for 9 ferret specific primer sets including STAT3, STAT5, FOS, CSN2, and CXCL10 (S3 Fig, Ferret Primer Sequences S3 Table).

To further validate the gene expression results, we analyzed the protein expression of prominent regulators of gene networks discovered in our data analysis. IHC analysis of STAT3 expression in control and H1N1+ MGs (Fig 8, left panels) showed increased STAT3 protein expression in virus positive glands with specific increases within the nucleus and diminished cytoplasmic staining of the mammary gland cells (shown in the inset magnified picture). STAT5 protein was prominent in the control tissue but decreased in the H1N1+ samples (Fig 8, right panels). Together, our gene expression profiling and validation suggests influenza virus infection may regulate milk cessation, breast involution, and oncogenic microenvironment related gene networks.

Since H1N1+ MG had significant differentially regulated gene pathways, we also investigated the expression profiles of Bystander mammary glands to determine if these glands also had evidence of inflammation. We defined Bystander mammary glands to be mammary glands that were negative for virus by qRT-PCR but present on a mother who had H1N1+ MGs. It is not known if Bystander glands were previously virus infected and were able to clear the virus prior to collection and gene profiling. Parallel analyses of Bystander mammary glands (H1N1- at time of collection) partially recapitulated responses detected in H1N1+ MG (S4 Fig; S4 Table). Both H1N1+ and Bystander gland response signatures were characterized by upregulation of proliferation, remodeling, and immune response associated genes (S5 Table). Downregulation of sugar/lipid metabolism genes and STAT5A/B was also a shared feature of both H1N1+ and Bystander glands (S4 Fig; S5 Table). However, Bystander gland responses differed from H1N1+ MG as the PPAR/Hedgehog/Adipocytokine pathway was downregulated and the immune responses were less robust. Catabolism, apoptosis, and TGF-β/p53-associated gene responses characteristic of H1N1+ gland responses were also less pronounced in Bystander glands (S4 Fig).

We next sought to define the immune response regulation within the H1N1+ MGs compared to the canonical gene pathways that have been determined in influenza infected lungs. Clustal analysis of H1N1+ MGs alongside infected lungs revealed the common upregulation of antiviral response pathways but also tissue specific regulation of other immune genes (Fig 7C). Induction of genes typical of the influenza antiviral response [40] such as CXCL10 and CCL2 were seen Day 6/7 post-infant-inoculation in H1N1+ MG (Fig 7C). The immune response also differed in the mammary gland compared to the lung. The response in mammary tissue was characterized by decreased expression of lymphocyte-associated genes such as CD3e, IL2Ra, CD4, AICDA (Activation Induced Cytidine Deaminase) [41]; IL4 and IL1β upregulation; and unchanged IL6 expression (Fig 7C). These results suggest that influenza infected mammary glands are capable of producing an antiviral and immune response that is unique compared to the immune response mounted during infection in the respiratory tract.

Above we determined that lactating mammary glands are able to host the influenza virus but it was not discerned if the mammary glands were directly susceptible to influenza infection. To determine if mammary glands are able to be infected with the influenza virus infection and act as a vessel of transmission, we inoculated active glands of lactating mother ferrets with Cal/07 (10⁵ EID₅₀) or PBS through the lactiferous ducts (mammary gland inoculation experimental design, Fig 9A). This also allowed us to investigate the direct responses of influenza infection in the mammary gland as well as the role of the mammary gland in virus transmission.

Following direct inoculation with Cal/07, mothers and their infants developed significant disease characterized by temperature increases, weight loss, and infant mortality which was not observed in the vehicle inoculation groups (Fig 9). H1N1 mammary inoculated mothers developed a significant fever at 104% of original temperature Day 2 post-mammary inoculation not seen in the mock inoculation control mothers (Fig 9B, left panel). Virus inoculated mothers also lost between 7–10% of their original weight and did not recover by the end of study (Fig 9B, right panel). Infants feeding on virus inoculated mammary glands also had significant weight loss with a 30% survival rate by Day 7 post-mammary-inoculation (Fig 9C, right panel and D). Live virus, average 6 TCID₅₀/ml (Log₁₀), was detected in the NW of infants feeding from virus inoculated glands Day 4 post-mammary-inoculation (Fig 9E). Virus was detected in mother NW after detection in the infants (Day 7 pi). Live virus was present in expressed milk, Day 2 and Day 4 post-mammary-inoculation, between 3 and 13 TCID₅₀/ml (Log₁₀) there by confirming successful inoculation and infection (Fig 9F). As virus was detected in mother NW after in infant NW, these results suggested that infants developed respiratory infection directly from virus shed from the inoculated mammary gland.

Mammary glands were stained with H&E to determine tissue architecture from direct inoculation of 2009 H1N1 and subsequent virus replication. Histopathology showed destruction of acini glandular architecture and absence of milk staining (Fig 10A). Gene expression profiling by qRT-PCR revealed Cal/07 directly inoculated mammary tissues had significant decreases in the milk production genes STAT5A, STAT5B, LPL, and CSN2 (Ferret Primer Sequences S3 Table; Fig 10B). The β-casein protein was decreased in milk samples collected from inoculated tissue when it was able to be collected (S5 Fig). Taken together, direct mammary gland inoculation suggests influenza transmission is possible through breastfeeding and influenza virus infection in mammary tissue leads to pathogenesis and milk cessation.

We next sought to determine if human breast cells were also susceptible and permissible to influenza virus infection. To this end we inoculated three cell lines of cultured human epithelial breast cells with Cal/07 strain (2009 H1N1) (in the absence of exogenous proteases) to visualize the virus life-cycle in inoculated cells, assess the viral kinetics, and determine cell viability post-inoculation (Fig 11). “Normal” non-tumorigenic (MCF-10A) and adenocarcinoma (MCF-7 and MCDA-MB-231) human epithelial breast cell lines were used to eliminate single cell type biases. To visualize the virus within the cell and determine virus subcellular localization, inoculated cells were stained for Influenza A Virus (IAV) NP (blue), nucleus (green), and actin filaments (red) (Fig 11A). Positive NP stain was used to determine if viral replication may be occurring within the nucleus. At 24 h post-inoculation influenza NP was detected in the nucleus (white arrows) of all three cell types, MCF-7, MDA-MB-231, and MCF-10A, indicated by co-localization of the nuclear stain (green) with NP staining (blue) (Fig 11A). Furthermore, marked punctate NP accumulation at the plasma membrane was observed in MCF-10A cells (yellow arrows) and to a smaller degree in the other cell types. vRNA was significantly increased in all cells types between 3 and 24 h post-inoculation by ~10 fold (Fig 11B). Cell viability analysis showed significant drops in cell viability of MCF-10A and MDA-MB-231 post-inoculation reaching ~35% viability at the 72 h time point compared to uninoculated cells at the same incubation (Fig 11C). To assess productive infection, live virus was quantified from collected supernatant at each time point. Live viral titers ranged between 3 and 4 TCID₅₀/ml (Log₁₀) for all cell types (Fig 11D). Baseline control (BC) wells had minimal or no detectable live virus. Together, these results suggest that the 2009 H1N1 virus was able to enter “normal” human breast cells leading to virus replication and productive infection.

All animal work was conducted in strict accordance with the Canadian Council of Animal Care (CCAC) guidelines. The protocol license number AUP 1031 was assigned by the Animal Care Committee of the University Health Network (UHN). UHN has certification with the Animals for Research Act including for the Ontario Ministry of Agriculture, Food and Rural Affairs, Permit Number: #0132–01 and #0132–05, and follows NIH guidelines (OLAW #A5408-01). The animal use protocol was approved by the UHN Animal Care Committee (ACC). All efforts were made to minimize animal suffering and infections and sample collections were performed under 5% isoflurane anesthesia.

The 2009 H1N1 virus strain, A/California/07/2009 (Cal/07), in chicken egg allantoic fluid was provided by the Influenza Reagent Resource, Influenza Division, WHO Collaborating Center for Surveillance, Epidemiology and Control of Influenza, Centers for Disease Control and Prevention, Atlanta, GA, USA. TCID₅₀ and EID₅₀ determinations were done as previously described [59]. All virus work was performed in a BSL-2+ facility as previously described [29]. Four-week-old-male and female ferrets, female Jill ferrets (aged 5 months to 1 year) and male ferrets 8 months old (adult) were bred in an on-site SPF ferret colony (University Health Network, Toronto, ON, Canada). Ferrets were shown to be seronegative by haemagglutination inhibition (HI) assay against currently circulating influenza A and B strains before infection. Infant ferrets were housed in litters with nursing-mother ferrets. Litters were housed according to UHN ferret colony specifications.

Ferrets were maintained and monitored in litters or as previously described (adults) [29,32]. Briefly, prior to inoculation, 4-week-old nursing ferret litters were randomly selected and housed in appropriate litter boxes contained in bioclean portable laminar-flow clean-room enclosures (Lab Products, Seaford, DE, USA) in the BSL-2+ animal holding area. Baseline body temperature and weight were measured on Day -1 and 0 for each animal. Temperatures were measured by using a subcutaneous implantable temperature transponder (BioMedic Data Systems, Inc., Seaford, DE, USA) for mothers and adults. Infant temperature was measured rectally by DT-610B Single Input K-Type Thermometer (ATP Instrumentation, Leicestershire, United Kingdom).

Nasal washes from the inoculated ferrets were collected on specified days post-inoculation in nasal wash buffer (1% BSA and 100 U/ml penicillin, 100 μg/ml streptomycin in PBS) as previously described [29]. One ml (adults) or 500 μl (infants) of nasal wash buffer was used for each collection. Penicillin and streptomycin were obtained from Invitrogen Canada (Burlington, ON, Canada) and BSA was from Wisent Inc. (Saint-Bruno, QC, Canada). For milk collection, mammary glands were cleaned and a warm press was applied to increase the easy of milking the gland. Milk was collected and pooled from all mammary glands of one mother. Milk for vRNA and live viral load assessments were collected at different milking sessions. Milk was stored in RLT Buffer or directly frozen at -80^°C. Tracheas, lungs, mammary glands, and nipples were collected on specified days post-inoculation from the euthanized ferrets. Prior to mammary gland collection, nipples were externally cleaned with ethanol and then the nipples were removed. Internal mammary gland tissue was harvested subsequent to nipple removal. Mother ferrets can have a variable number of active mammary glands per pregnancy/postpartum (~5–8). Therefore, a variable number of mammary glands was assessed for viral loads per time point. As many mammary glands were run as possible to acquire the maximum amount of data. Although the number of glands was variable, the number of mother ferrets was always three ferrets per time point. Feces from infant ferrets were collected at the time of anal temperature collection. Each feces was placed directly in RLT buffer or placed directly at -80^°C from the animals’ anus to avoid surface contamination. Samples were stored in -80^°C before processing for viral loads or RNA. Samples for RNA were either placed in RNA later (milk and feces) or homogenized in TRIzol (Tissues). Sample aliquots for histopathology were stored in formalin until processing.

Viral replication in the upper, middle, and lower respiratory tract and mammary tissues were assessed by endpoint titration of nasal washes or homogenized tissue samples from the inoculated ferrets in MDCK cells (TCID₅₀) using haemagglutination as the readout for positive wells as previously described[29]. Briefly, nasal washes were diluted 1:10 in vDMEM (Dulbecco's modified Eagle's medium containing 1% BSA, 25mM glucose, 1mM sodium pyruvate, 4mM glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 μg/ml gentamycin, and 1 μg/ml TPCK-Trypsin) followed by the half log serial dilution in quadruplicate with vDMEM on MDCK cells in 96-well flat-bottom plates (SARSTEDT, Inc., Saint-Leonard, QC, Canada). Tissues were processed similarly and homogenized 1:10 (w/v) in sDMEM (vDMEM excluding 1% BSA) on ice and centrifuged before titration of the homogenates on MDCK cells. Before inoculation, MDCK cells were maintained in the log-phase at low-passage numbers and grown in cDMEM (DMEM containing 10% fetal bovine serum, 25 mM glucose, 1 mM sodium pyruvate, 6 mM glutamine, 1 mM non-essential amino acids, 100 U/ml penicillin, and 100 μg/ml streptomycin). The day before viral sample titration, 2x10⁴ MDCK cells were seeded into each well to reach 95% confluence on the next day. After a two-hour incubation of titrated viral samples with MDCK cells at 37^°C, 5% CO₂, samples were aspirated and replaced with fresh vDMEM followed by a six-day incubation at 37^°C, 5% CO₂. On Day 6 post-incubation, supernatants were examined for presence of virus by haemagglutination of 0.5% turkey erythrocytes (Lampire Biological Laboratories, Pipersville, PA, USA). The viral titers were determined as the reciprocal of the dilution resulting in 50% HA positivity in the unit of TCID₅₀/ml represented in log₁₀ base. Viral titers in cell culture supernatants were also determined to investigate capacity of mammary epithelial cells to produce infectious virus in vitro. Cell culture supernatants were loaded directly and serially diluted as described above. Assay upper and lower limits of detection depended on sample availability and/or expected viral titer and varied between experiment types. Lower limits of detection are indicated in all viral load figures by a dashed line. Upper limits of detection were 10^6.75 TCID₅₀/ml for mammary epithelial cell supernatant assessments, 10^7.75 and 10^13.75 for milk assessments in Fig 5D and Fig 9F, respectively, and 10^7.25 for all nasal wash and tissue assessments except for S1C Fig (10^7.75). Cell culture reagents were obtained from Invitrogen Canada or Wisent except for TPCK-Trypsin (Sigma-Aldrich Canada Ltd., Oakville, ON, Canada).

Animals were euthanized on appropriate days and the lung or mammary gland tissues were harvested, perfused and fixed in formalin followed by paraffin embedding and sectioning as described [29]. Tissue slides were then stained with hematoxylin and eosin for histopathology assessment. Nursing-mothers of mock-inoculated 4-week-old infants were used for health and age control. Immunostaining was performed using a polyclonal goat anti-influenza antibody (1:2000) (Abcam (ab20841), Cambridge, United Kingdom). STAT3 protein was detected with a rabbit polyclonal anti-STAT3 (LSBio, Cat # LS-B4693) using CIT6 Antigen Retrieval at 1:300 primary antibody dilution for 1 hour incubation. STAT5 protein was detected with a rabbit polyclonal anti-STAT5 (Abcam, Cat # ab68465) using TE9 Antigen Retrieval at a concentration of 1:100 primary antibody dilution and a 1 hour incubation. High resolution scans were performed using an Aperio ScanScope XT, Leica Biosystems, Nußloch, Germany.

Milk was expressed from the mammary glands of lactating mother ferrets. Ten μg of milk was load per lane and proteins were separated by SDS-PAGE and transferred to PVDF membrane. To visualize beta-casein protein, the membrane was blotted with 1:500 rabbit polyclonal anti-bovine beta-casein (Cat# 251309, Abbiotec) and followed with 1:2000 anti-rabbit IgG-HRP (Santa Cruz sc-2030). To ensure equal loading of each sample, a protein estimation was conducted using a Pierce BCA Protein Assay Kit (Life Technologies). Proteins were detected with a chemiluminescence protocol and exposed. Images were exposed to Thermo Scientific CL-XPosure Film (Waltham, MA, USA) and developed using a Konica Minolta Model SRX-101A (Tokyo, Japan) machine.

Collected lung and mammary tissue was placed directly in TRIzol reagent (Life Technologies, Burlington, Ontario) for immediate homogenization and subsequently processed according to the manufacturer’s instructions as previously done [32]. Milk RNA was extracted using an RNEasy Kit (Qiagen). Fecal RNA was extract using QIAshredder (Qiagen) before subsequent processing by RNEasy Kit (Qiagen, Hilden, Germany). Total RNA was converted into cDNA using ImProm-II reverse transcription system (Promega Inc., Madison, WI, USA) followed by qRT-PCR. qRT-PCR was performed on the ABI Prism 7900HT system (Applied Biosystems, Foster City, CA, USA). Blood was collected in PAXgene Blood RNA tubes followed by extraction using a PAXgene Blood RNA Kit (PreAnalytiX, Hombrechtikon, Switzerland), per the manufacturer’s instructions. Quadruplicate PCR reactions were run for specified target genes, each containing 4 μl 1.25 ng/μl cDNA, 0.5 μl 4μM forward and reverse primers, and 5 μl SYBR green PCR master mix (Applied Biosystems Inc., Carlsbad, CA, USA). PCR reactions were run at an annealing temperature of 60^°C for 35 cycles for blood, tissue, and cell lysates. PCR reactions for milk and fecal RNA were done at 60^°C for 40 cycles due to the difficulty of extracting RNA from milk and feces. Applied Biosystems Sequence Detection Systems Version 2.4 software was used for raw data collection (Applied Biosystems, Foster City, CA, USA). Target gene expression levels were normalized to house-keeping gene β-actin, and quantified by the relative standard curve method.

RNA extracted from mammary glands of nursing-mothers (2009 H1N1 and mock inoculated infants) and influenza infected adult lungs were subjected to gene expression analysis by microarray. Mammary glands were determined to be virus infected glands if qRT-PCR found vRNA >10 copies per 5 ng of RNA and histological evidence of virus infection. Mammary glands were determined to be Bystander glands if vRNA was undetectable by qRT-PCR. Total cellular RNA from tissues was collected and purified using TRIzol reagent and methods, as described above. Complementary RNA (cRNA) was generated using 3’ IVT Express Kit (Affymetrix, Santa Clara, CA, USA) and microarray analysis was performed using Affymetrix GeneChip Canine Genome 2.0 Array (Affymetrix, Santa Clara, CA, USA), as previously described [40]. We have previously used and validated canine arrays for the investigation of ferret global gene expression profiling [60–64]. We demonstrated high levels of homology (average of 89% identity) between 30 canine and ferret nucleotide sequences [60]. Data sets for H1N1+ mammary glands, bystander mammary glands, and lungs were pre-processed independently with quantile normalization, variance stabilization, and log₂ transformation. For genes represented by multiple probes on the array, the probe with highest total signal in H1N1+ mammary gland and lung datasets were selected for analysis. Two parallel analyses were applied A H1N1+ and bystander mammary gland datasets. Global clustering analyses were performed by one-way hierarchical clustering (Pearson’s correlation) of all significantly differentially regulated genes (p-value<0.05 Student’s t-test unpaired, equal variances, fold change ≥ |1.5| fold) and functional annotation of generated clusters. Global pathway gene enrichment analyses were performed by functional annotation of genes exhibiting significant changes in expression at either Days 3/4 or 6/7 in either dataset, using the KEGG Pathway database. DAVID Bioinformatics Resource v6.7, described previously [40] was used for functional annotation and enrichment score profiling. Clustergrams in Fig 7B were manually selected as representative genes for signaling pathways and gene networks prominently represented in global analyses. Genes included in clustergrams were identified as significantly differentially regulated members of related KEGG-defined signaling pathways with the exception of Macrophage and Milk Production clustergrams which were generated by manual selection of representative genes. Full clustergrams of all significantly differentially regulated genes for each KEGG-signaling pathway identified in Fig 7 are included in S2 Fig

Accompanying data can be found in the Supplementary Materials. Raw (.CEL files) and normalized microarray data were compiled in accordance with Minimum Information About a Microarray Experiment (MIAME) guidelines and uploaded to the public data repository Gene Expression Omnibus (GEO) for public dissemination (www.ncbi.nlm.nih.gov/geo/). These datasets can be found under the accession numbers GSE63082 (mammary gland), GSE63083 (lung) and GSE63084 (both) at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE63084.

In vitro studies were performed similarly as reported previously [65]. MCF-10A, MFC-7, and MDA-MB-231 cells were purchased from ATCC. The cells were seeded at 50,000 cells/well in 6 well plates for RNA and viral load analysis. Cells were seeded at 75,000 cells per well in Nunc Lab-Tek II Chamber Slide System, 2 well glass slides (Thermo Scientific) for confocal microscopy. Cells were inoculated with A/Cal (H1N1) at a dose of 10⁴ (ELISA assay calculation for MO1 = 1) TCID₅₀ (determined by MDCK titration) per 10⁴ cells (10^6.5 TCID₅₀ HA assay calculation). For confocal, cells were fixed in 4% PFA in PBS at 24 hours post-inoculation and stained for filamentous actin using Phalloidin (Molecular Probes), DNA using SYTOX green (Molecular Probes), and influenza A virus NP protein using a mouse monoclonal [AA5H] to Influenza A Virus Nucleoprotein (Abcam; ab20343) at 1:100 for 60 min. This was followed by staining with Alexa Fluor 647 Donkey Anti-Mouse secondary (Jackson ImmunoResearch). Prior to staining cells were permeablized using Triton X-100 in PBS. Slides were imaged by confocal microscopy using a Zeiss LSM 710 NLO, using objective C-Apo 63x/1.4 NA (oil). For RNA and viral load, cell lysates and supernatants were collected at 3, 24, 48, and 72 h post-inoculation for RNA extraction and live viral load quantification, respectively. The baseline control was determined by incubating the inoculum in assay wells without cells for 72 h. to assess virus decay and viral attachment to the wells. qRT-PCR and viral load were performed as described in above methods section.

Unpaired, unequal variance, two-tail Student’s t-test or one-way ANOVAs were conducted. The Log-rank (Mantel-Cox) Test was used to determine significant differences in animal survival. A p value of ≤ 0.05 was considered statistically significant.