跳到正文
原文
Nature Biotechnology· Jefferson J. S. Santos·· 23 天前精选AI 评分67

H5 mRNA–LNP疫苗在泌乳奶牛H5N1攻毒实验中减轻乳腺炎并降低乳汁病毒载量

An mRNA–lipid nanoparticle vaccine against H5N1 influenza virus protects lactating dairy cows from infection

AI 导读

一种编码H5血凝素的mRNA–LNP疫苗在泌乳奶牛中诱导出血清和乳汁H5抗体,乳腺内攻毒后接种组乳汁H5N1病毒RNA峰值比未接种组低约1,000倍。研究在8头荷斯坦泌乳奶牛中按500 μg、间隔3周接种两剂,攻毒后接种组产奶量下降和临床乳腺炎更轻,接种乳区的乳腺纤维化评分也更低。

推荐理由

奶牛攻毒实验中接种组乳汁病毒RNA峰值约低千倍,为H5N1牛群防控提供疫苗数据。

正文

Main

Highly pathogenic avian influenza (HPAI) H5N1 clade 2.3.4.4b virus continues to cause widespread outbreaks among wild birds and mammals in North America, as well as on commercial poultry farms. In 2024, H5N1 clade 2.3.4.4b emerged in cattle in the USA and quickly spread to animals in 19 states1. The H5N1 clade 2.3.4.4b outbreak in cattle has already caused major economic losses in the dairy industry2 and represents a concern for human health because of the increased risk of mammalian adaptation and potential human-to-human transmission3.

H5N1 clade 2.3.4.4b virus causes limited respiratory signs in cattle. While the virus can be sporadically detected from oronasally inoculated calves under experimental conditions4,5,6, the respiratory tract is not likely the primary site of viral replication. Instead, the main site of H5N1 viral replication is the mammary tissue of infected lactating animals. Experimental intramammary inoculation with H5N1 clade 2.3.4.4b viruses leads to reduced feed intake and rumen motility, as well as drastic reductions in milk production because of severe mastitis4,5,7,8. H5N1 clade 2.3.4.4b viruses exhibit tropism for the mammary gland tissue of cows9,10, as confirmed by the presence of viral RNA and antigen in the alveolar milk-secreting epithelial cells and the interlobular space of infected animals8. Additionally, mastitis caused by the H5N1 clade 2.3.4.4b virus affects the color and consistency of milk and changes in the mammary tissue contribute to the dramatic reduction in milk production4.

In addition to enhanced biosecurity efforts, the use of HPAI clade 2.3.4.4b H5N1 vaccines in cattle has been considered11. Vaccination can potentially aid in controlling the current outbreak in lactating dairy cows, limiting the drop in milk production in exposed animals and reducing the risk of spillover events into humans. Some bovine vaccines elicit antibodies that are transferred into milk in lactating animals. For example, the Mycobacterium bovis-BoHV-1 combined vaccine effectively elicits antigen-specific maternal antibodies12 and an inactivated H5 avian influenza A virus vaccine elicits antibodies in both serum and milk13. While other vaccines against the HPAI clade 2.3.4.4b H5N1 virus have been tested in bovine animal models6,13,14,15,16, studies are needed to systematically evaluate the immunogenicity and protective efficacy of vaccines in lactating dairy cows against intramammary infection, as the mammary glands are the site for HPAI clade 2.3.4.4b H5N1 virus infection4,5.

We previously developed an mRNA–LNP vaccine encoding an HA from a clade 2.3.4.4b H5N1 virus and demonstrated that the vaccine elicited robust immune responses in mice and ferrets, preventing morbidity and mortality in ferrets following HPAI 2.3.4.4b H5N1 virus challenge17. We found that the same mRNA–LNP vaccine was immunogenic in calves, inducing a strong antibody and T cell-mediated immune response that reduces viral shedding from experimental HPAI 2.3.4.4b H5N1 virus infection6. In the present study, we evaluated the immunogenicity and efficacy of a clade 2.3.4.4b H5 mRNA–LNP vaccine administered to lactating dairy cows. We found that the H5 mRNA–LNP vaccine elicited strong antibody responses in both serum and milk. Additionally, the vaccine significantly reduced virus replication and disease, including clinical mastitis, associated with clade 2.3.4.4b H5N1 intramammary route of infection.

Results

H5 mRNA–LNP vaccines are well tolerated in lactating dairy cattle

We generated an mRNA vaccine that expresses the hemagglutinin (HA) of A/dairy cattle/Texas/24-008749-002/2024, which is representative of HPAI clade 2.3.4.4b H5N1 viruses currently circulating in dairy cattle in the USA4. We vaccinated four lactating dairy cows intramuscularly with 500 μg of H5 mRNA–LNP in saline and then boosted intramuscularly with the same vaccine dose 3 weeks later (Fig. 1a). As a control, we intramuscularly injected four additional lactating cows twice with saline (Fig. 1a). Animals in our study were obtained from an HPAI-monitored Holstein dairy herd in Pennsylvania and ranged from 2 to 8 years of age (Supplementary Table 1). We collected serum samples at 0, 7, 21, 28, 41 and 49 days after the initial vaccination and collected milk samples daily. Both unvaccinated and vaccinated animals were challenged with infectious clade 2.3.4.4b H5N1 virus through intramammary inoculation 49 days after the first vaccination.

Fig. 1: Vaccination study design, safety and mRNA–LNP vaccine immunogenicity in lactating dairy cows.

a, Overview of the vaccination and sampling scheme. Lactating dairy cows were vaccinated using a prime-boost regimen. Milk and blood samples were collected at different time points. At 49 days after vaccination and 28 days after boost, lactating dairy cows were challenged with HPAIV H5N1 and monitored daily for signs of disease and virus shedding. i.m.m., intramammary. Figure created BioRender; Lab, H. https://biorender.com/vb619x8 (2026). b,c, qPCR quantification of vaccine H5 mRNA from milk of vaccinated animals collected up to 8 days after prime (b) or boost (c) vaccination. Data shown are the mean of two replicates. d, H5-specific IgG antibody levels were measured by ELISA in sera from unvaccinated (open red triangles) and vaccinated (closed black circles) animals and endpoint titers are shown. e, Neutralizing antibodies were measured using a 90% FRNT; the reciprocal dilutions of serum required to inhibit 90% of virus infection are shown. f,g, HAI antibodies were measured by HAI assay in sera at 49 (f) and 63 (g) days after vaccination; reciprocal dilutions of serum required to prevent red blood cell agglutination are shown. h, H5-specific IgG antibody levels were measured by ELISA in milk and endpoint titers are shown. Data are the mean ± s.e.m. and individual data points (n = 4) are shown. Dashed lines indicate the lower limit of detection. Significant statistical differences (P < 0.05) between groups were determined using a log-normal two-tailed t-test with Welch correction and are indicated by an asterisk.

Source data

The vaccine was well tolerated, with no obvious adverse effects. Previous studies found trace amounts of vaccine-related mRNA in human milk in some individuals shortly after SARS-CoV-2 mRNA–LNP vaccination18,19,20. Therefore, we completed experiments to determine whether we could detect H5 mRNA–LNP in bovine milk samples obtained after vaccination. First, we established a sensitive quantitative PCR assay to measure H5 mRNA levels in milk. We spiked a milk sample from an unvaccinated cow with a known concentration of the H5 mRNA–LNP vaccine to assess the sensitivity of our PCR assay (Extended Data Fig. 1). We found that the detection limit of our assay was 65 mRNA copies per μl. We then used quantitative PCR (qPCR) to measure H5 mRNA levels in milk samples collected up to 8 days after prime or boost vaccination. None of the milk samples from vaccinated animals had detectable levels of H5 mRNA (Fig. 1b,c), demonstrating that vaccine-derived mRNA was not present at detectable levels in milk after intramuscular vaccination of lactating dairy cows.

H5 mRNA–LNP vaccine is immunogenic in lactating dairy cattle

We quantified the level of H5-reactive antibodies in the sera from vaccinated animals and performed functional assays to determine whether these antibodies could inhibit H5N1 virus infection. We detected low levels of H5-reactive IgG in sera as early as 21 days after vaccination that significantly increased after booster vaccination (Fig. 1d). Unexpectedly, we found that one cow from the unvaccinated group and one cow from the vaccinated group had low levels of H5-reactive serum IgG before the start of the experiment but were below the limit of detection in the H5 neutralizing assay (day 0; Fig. 1d,e). The source herd was under weekly HPAI surveillance before and during the study and remained HPAI negative, suggesting that the low levels of H5-reactive IgG in these animals was not because of H5N1 infection. A recent study reported that 2.6% of cattle in the USA have serological evidence of past exposures to human influenza A virus21; therefore, we hypothesized that the low levels of H5-reactive IgG in two of the animals in our study were because of cross-reactive antibodies elicited by human influenza virus infections. Consistent with this idea, the two animals that had low but detectable H5 IgG antibodies before the study also had detectable H1-reactive antibodies (Extended Data Fig. 2). All four vaccinated animals, including the one animal with low levels of preexisting antibodies, produced similar levels of H5 antibodies following vaccination (Fig. 1d,e).

We completed in vitro neutralization assays to determine whether serum antibodies could prevent infection of the A/dairy cattle/Texas/2024 H5N1 virus. Consistent with the IgG binding data, H5N1 neutralizing antibodies were detected as early as 21 days after vaccination and were significantly increased after booster vaccination (Fig. 1e). We also completed hemagglutination inhibition (HAI) assays using sera collected just before viral challenge to determine whether vaccine-elicited antibodies blocked viral attachment. Consistent with both the IgG binding and in vitro neutralization data, we found that antibodies in the sera from all four vaccinated animals blocked viral attachment to erythrocytes (Fig. 1f). Vaccine-elicited antibodies were boosted after H5N1 challenge, rising to levels observed in unvaccinated challenged animals (Fig. 1g).

We also quantified H5-reactive IgG levels in milk samples collected in our study. H5-reactive IgG was detected in only one vaccinated cow after the first vaccination (in one of the vaccinated animals without preexisting antibody) (Fig. 1h). We detected high levels of H5-reactive IgG in milk from all four vaccinated animals following the booster vaccination (Fig. 1h). Peak levels of H5-reactive IgG were higher in sera compared to milk and antibody levels declined more rapidly in milk compared to the sera (Extended Data Fig. 3). We were unable to perform in vitro neutralization with milk samples as nonspecific inhibitors in milk from unvaccinated animals inhibit virus in these assays.

Specificity of antibodies elicited by H5 mRNA–LNP vaccination

We performed pseudovirus deep mutational scanning experiments22 to identify HA epitopes targeted by neutralizing antibodies in sera from lactating dairy cows vaccinated with the H5 mRNA–LNP vaccine (Fig. 2). For these experiments, we incubated bovine sera with a pool of nonreplicative lentiviral particles that were pseudotyped with mutant HAs from the clade 2.3.4.4b A/American Wigeon/South Carolina/USDA-000345-001/2021 virus, as previously described23. This H5 deep mutational scanning library carries nearly all of the 10,773 possible amino acid substitutions in the HA23. We found that the neutralizing activity of the sera from all four vaccinated animals targeted the antigenic sites on the globular head of HA (Fig. 2a,b), including antigenic site A (residues 124–127 and 144–145, H3 numbering) and antigenic site B (residues 165, 167, 169 and 171, H3 numbering). There was a remarkable similarity in the amino acid sites targeted by the serum antibodies isolated from individual vaccinated cows. These data indicate that most neutralizing antibodies elicited by the H5 mRNA–LNP vaccine target neutralizing epitopes adjacent to the receptor-binding domain of HA. Importantly, the amino acid identities at the sites most strongly targeted by the vaccine-induced neutralizing antibodies are not highly variable in the HA of circulating clade 2.3.4.4b viruses (Fig. 2c).

Fig. 2: Effect of substitutions in HA on vaccinated cattle sera neutralization.

a, Neutralization decrease caused by substitutions at each position in HA. The top plot shows the mean total neutralization decrease for the four vaccinated cow sera and the plots below show the individual cow neutralization decrease. Sites causing the greatest neutralization decrease are labeled by amino acid position above the top plot. b, Mean effects of substitutions at each site of cattle serum neutralization overlaid on a surface representation of HA (Protein Data Bank (PDB) accession 4KWM). Sites are colored white if substitutions have no effect on neutralization and red if substitutions decrease neutralization. c, Amino acid conservation at HA sites where substitutions caused the greatest neutralization decrease by the vaccinated cow sera, calculated over all available H5 sequences, 2.3.4.4b clade sequences and B3.13 and D1.1 genotype sequences from dairy cows.

Source data

H5 mRNA–LNP vaccination protects lactating cows from H5N1 infection

To determine whether H5 mRNA–LNP vaccination can prevent virus shedding and reduce H5N1 disease, we experimentally challenged unvaccinated and vaccinated cows with clade 2.3.4.4b H5N1 (A/dairy cattle/Texas/24-008749-002/2024) through the intramammary inoculation route and monitored the cows for 14 days after inoculation. We then humanely euthanized the animals and performed macroscopic and microscopic lesion analyses.

We measured levels of H5N1 viral RNA in milk samples by quantitatice reverse transcription PCR (RT–qPCR) assays. In unvaccinated animals, H5N1 viral RNA in milk from inoculated mammary quarters was first detected at 1 day after inoculation and peaked at 2 days after inoculation; milk samples remained RT–qPCR-positive throughout the study period (Fig. 3a). In vaccinated animals, H5N1 viral RNA levels in milk from inoculated mammary quarters were significantly lower at 1–3, 5 and 8–13 days after inoculation, with ~1,000-fold lower peak titer compared to levels found in milk from unvaccinated animals (Fig. 3a). Most milk samples that were positive for H5N1 viral RNA by RT–qPCR with cycle threshold (Ct) < 35 contained infectious H5N1 virus (Supplementary Table 2). We detected either H5N1 viral RNA or infectious virus in milk from the uninoculated quarters of two unvaccinated animals (Extended Data Fig. 4 and Supplementary Table 2), suggesting that virus was able to migrate from an inoculated to uninoculated quarter in these animals. We detected viral antigen in mammary tissue from uninoculated quarters of both unvaccinated and vaccinated animals using immunohistochemistry (IHC) (Extended Data Figs. 4 and 5a,b); however, viral RNA in the milk from uninoculated mammary quarters in vaccinated animals was below the limit of detection, suggesting that the vaccine prevented further spread of the virus.

Fig. 3: Virus RNA detection in clinical samples from unvaccinated and vaccinated lactating cows after challenge.

a, The log10-transformed TCID50 per ml equivalent Ct detection of virus RNA in stripped milk from inoculated mammary quarters of unvaccinated (open red triangles) and vaccinated (closed black circles) lactating dairy cows. Data shown are the mean ± s.e.m. of n = 4. Significant statistical differences (P < 0.05) between groups were determined using a mixed-effects model (residual maximum likelihood model) followed by Tukey’s multiple-comparison test. Time points with a significant difference (P < 0.05) between unvaccinated and vaccinated are indicated by an asterisk. b, qPCR quantification of virus RNA from antemortem clinical samples of unvaccinated and vaccinated lactating dairy cows. A Ct below 35 was considered positive (red), a Ct between 35 and 38 was considered suspect (yellow) and a Ct above 38 was considered negative (white). LL, LeucoLOCK from whole blood; NA, not applicable; NS, nasal swab; OS, ocular swab.

Source data

We measured H5N1 RNA levels in other clinical samples, including nasal swabs, ocular swabs, urine, saliva, serum and whole blood. We detected H5 RNA in multiple sample types from all four unvaccinated animals (Fig. 3b and Extended Data Fig. 4). By contrast, few samples obtained from vaccinated animals possessed detectable levels of H5 RNA (Fig. 3b and Extended Data Fig. 4). We also collected fresh mammary tissue, supramammary lymph nodes and a section of diaphragm at the necropsy 14 days after inoculation. H5 RNA was detected in mammary tissue from inoculated quarters in all unvaccinated cows, in an uninoculated quarter of one unvaccinated cow and only in inoculated quarters of two vaccinated cows, as well as in an uninoculated quarter of one vaccinated animal using IHC (Extended Data Fig. 4). Supramammary lymph node tissue was positive for H5 RNA in three unvaccinated cows and in two vaccinated cows and diaphragm liquid was suspect in one unvaccinated cow (Extended Data Fig. 4).

Following H5N1 virus infection, unvaccinated cows displayed decreased rumen motility 1–7 days after inoculation, with a significant decrease at 3 and 4 days after inoculation (Fig. 4a). Milk production also significantly decreased in unvaccinated animals following H5N1 infection, beginning at 2 days after inoculation, with a dramatic drop at 3 days after inoculation that remained significantly reduced throughout the study at 4–8 and 10–12 days after inoculation (Fig. 4b). Rumen motility and milk production were less impacted in vaccinated cows following H5N1 infection (Fig. 4a,b). We observed severe changes in color (Fig. 4c) and consistency (Fig. 4d) indicative of mastitis in milk from the inoculated mammary quarters of unvaccinated cows following H5N1 infection. By contrast, milk color and consistency from inoculated mammary quarters of vaccinated cows were less affected by H5N1 infection (Fig. 4c,d). Uninoculated mammary quarters of unvaccinated cows also displayed changes in milk color and consistency, albeit to a lesser extent compared to inoculated mammary quarters (Fig. 4c,d). Furthermore, unvaccinated cows showed signs of mastitis in inoculated quarters and, to a lesser extent, in uninoculated quarters, as determined by California mastitis test (CMT) (Fig. 4e). Inoculated quarters from unvaccinated animals consistently scored above 2 (score range: 0–3) in the CMT, which indicates moderate to severe inflammation and leukocyte infiltration. Conversely, inoculated and uninoculated mammary quarters of vaccinated cows frequently scored a CMT below 1 following H5N1 virus infection (Fig. 4e), indicating little to no signs of mastitis. The clinical mastitis scores in uninoculated quarters were primarily contributed by two unvaccinated cows (Fig. 4c–e). We found that unvaccinated cows had a significant association between loss in milk production, clinical signs, and viral RNA detections in milk of inoculated quarters. By contrast, vaccinated cows did not present a significant association among loss in milk production, clinical signs and viral RNA in milk, except for rumination time (Extended Data Fig. 6a,b). No significant changes in rectal temperatures between unvaccinated and vaccinated cows were observed throughout the study.

Fig. 4: Clinical signs in unvaccinated and vaccinated lactating dairy cows after challenge.

a,b, Unvaccinated (open red triangles) and vaccinated (closed black circles) lactating dairy cows were challenged with HPAI clade 2.3.4.4b H5N1 and monitored daily for rumination (a) and milk production (b). i.m., intramuscular. Data are the mean ± s.e.m. and individual data points (n = 4) are shown. Significant statistical differences (P < 0.05) between groups were determined using a two-way repeated-measures ANOVA followed by Tukey’s multiple-comparison test. Time points with a significant difference (P < 0.05) between unvaccinated and vaccinated are indicated by an asterisk. c–e, Collected milk samples were evaluated by quarter for mastitis by milk color (c), milk consistency (d) and CMT (e). Darker red shades represent higher scores; lighter to white shades represent lower or zero scores. Data shown are the mean of n = 4 per treatment group.

Source data

H5 mRNA–LNP vaccination of lactating cows significantly decreased mammary fibrosis

We completed macroscopic evaluation of the mammary glands from inoculated and uninoculated quarters for all lactating dairy cows at necropsy. Consistent with the clinical observations, inoculated mammary quarters from unvaccinated cows exhibited significantly higher macroscopic lesion scores compared to vaccinated cows (P < 0.05; 95.4 and 8.1; Fig. 5a). By contrast, macroscopic lesion scores in uninoculated mammary quarters were low and did not differ significantly between unvaccinated and vaccinated cows (P > 0.05; 12.1 to 10.0; Supplementary Table 3). A representative image of normal mammary tissue in an uninoculated quarter is shown in Fig. 5b and an image of fibrotic mammary tissue in an inoculated quarter of an unvaccinated cow is shown in Fig. 5c. Consistent with the macroscopic scores, microscopic scores of mammary gland fibrosis using Picrosirius red staining demonstrated less fibrosis in inoculated glands from vaccinated animals compared to unvaccinated animals (Fig. 5d–h and Supplementary Tables 4 and 7). A histologically unremarkable photomicrograph of mammary tissue is shown in Fig. 5e for comparison and representative photomicrographs are shown with tissue from a vaccinated uninoculated cow (Fig. 5f), a vaccinated inoculated cow (Fig. 5g) and an unvaccinated inoculated cow (Fig. 5h).

Fig. 5: Macroscopic and microscopic lesions in mammary tissue after challenge.

a, The percentages of macroscopic lesions in the mammary tissue of unvaccinated and vaccinated lactating dairy cows were measured in inoculated quarters. Data shown are the calculated average percentage (0–100) of the rostral, mid and caudal regions for inoculated quarters (mean ± s.e.m. of n = 4). b, Macroscopic photo representative of a normal mammary tissue producing milk (cow 2, unvaccinated rear right uninoculated quarter). c, Representative photo of mastitis-affected mammary tissue with fibrosis (cow 2, rear left inoculated quarter). d, The percentage of microscopic lesions in the mammary tissue of unvaccinated and vaccinated lactating dairy cows. Data shown are the calculated median score of the rostral, mid and caudal regions for inoculated quarters (scale 0–144; n = 4). e, Picrosirius red stain of mammary gland highlighting connective tissue and fibrosis (red). A histologically unremarkable photomicrograph was included for comparative purposes. Thin bands of supporting connective tissue (arrowhead) separate lobules and no to minimal intralobular collagen (cow 6, front left quarter, vaccinated, uninoculated, fibrosis score = 0). Representative photomicrographs were selected on the basis of their alignment with the mean fibrosis score for the respective group. f, Extensive interlobular fibrosis with lobular separation (arrowhead) and dense intralobular collagen (arrow) with partial lobular disruption (cow 5, front left quarter, vaccinated uninoculated, fibrosis score = 9). g, Thick interlobular collagen bundles with compression of adjacent lobules (arrowhead) and increased intralobular collagen (arrow) with mild distortion of lobular architecture (cow 7, front right quarter, vaccinated inoculated, fibrosis score = 9). h, Extensive interlobular fibrosis with lobular separation (arrowhead) and dense intralobular collagen (arrow) with near complete lobular disruption (cow 4, rear left quarter, unvaccinated inoculated, fibrosis score = 14).

Discussion

The recent widespread outbreaks of HPAI clade 2.3.4.4b virus in dairy cattle, along with its association with milk loss and severe mastitis, require effective countermeasures for the dairy industry. In one study, decreased milk production, mortality and early herd removal of lactating cows clinically affected by HPAI clade 2.3.4.4b virus resulted in economic losses of 950 USD per clinically affected cow, totaling approximately 737,500 USD for the infected herd that was studied2. Vaccines that reduce clade 2.3.4.4b H5N1 virus in milk might be useful in preventing the spread of clade 2.3.4.4b H5N1 viruses in lactating cattle. Combined with robust viral testing programs, a 2.3.4.4b H5N1 vaccine has the potential to reduce virus transmission, clinical signs and mastitis in H5N1-infected lactating dairy cattle, while limiting potential viral spillovers into humans and other susceptible wild and domestic animal populations.

Recent studies have evaluated the immunogenicity of HPAI clade 2.3.4.4b H5N1 inactivated13,16, DNA-based16 and mRNA–LNP15 vaccines in lactating dairy cows. These studies have demonstrated that different vaccine platforms are capable of eliciting H5-reactive antibodies in cows but more in-depth comparative challenge studies are needed to evaluate how different vaccine platforms affect milk production, rumination and virus-induced mastitis. Our study provides a comprehensive and quantitative assessment of an H5 mRNA–LNP vaccine against intramammary HPAI clade 2.3.4.4b H5N1 virus infection in lactating dairy cows. We found that our H5 mRNA–LNP vaccine induced neutralizing antibodies in lactating dairy cows and protected against intramammary HPAI clade 2.3.4.4b H5N1 virus infection. We found that rumination and milk production dropped more in unvaccinated cows relative to vaccinated cows following H5N1 infection. The vaccine was well tolerated and animals showed no apparent adverse events following vaccination. We did not detect H5 mRNA in milk collected after intramuscular vaccine administration, supporting its safety for cows and their milk. Furthermore, the H5 mRNA–LNP reduced peak H5N1 viral titers in milk ~1,000-fold. While it is reasonable to suspect that this large reduction in viral titers will limit animal-to-animal H5N1 transmission, additional large-scale field experiments will be required to test this possibility. Previous studies have demonstrated that intramuscular immunization of HA mRNA–LNP or conventional influenza vaccines can reduce influenza virus transmission in ferrets, pigs and guinea pigs24,25,26,27 and future experiments should evaluate whether our H5 mRNA–LNP vaccine reduces H5N1 transmission in lactating dairy cows.

The H5 mRNA–LNP vaccine elicited antibody responses in both serum and milk. Antibody levels in milk were consistently lower compared to those in serum before challenge. This was expected, considering the varying concentrations of IgG present in milk after the periparturient and colostrogenesis periods. Future studies are needed to investigate the impact of different routes of immunization, vaccination schedules and vaccine doses on the magnitude and longevity of antibody responses elicited by mRNA–LNP vaccines in cattle. It is critical to continue optimizing vaccine regimens that elicit sustained protective levels of antibodies in the milk of lactating cattle, with the goal of optimizing a vaccine that elicits protective antibody titers after a single immunization. Additionally, it will be important to consider a differentiating infected from vaccinated animals (DIVA) strategy for diagnostics and HPAI outbreak investigation in vaccinated livestock. As our H5 mRNA–LNP vaccine elicited only antibody responses to the HA protein, the detection of NP antibodies would indicate a natural infection and could be used as a DIVA strategy.

HPAI H5N1 2.3.4.4b has a high tropism for the mammary gland and infects milk-secreting mammary epithelial cells lining the alveoli, resulting in a viral-induced mastitis8. Experimental inoculation of lactating cows with HPAI H5 2.3.4.4b through the intramammary route recapitulates field observations, leading to severe acute mammary gland infections with necrotizing mastitis5. In affected mammary tissue, 15–50% of the secretory alveoli and ducts, both intralobular and interlobular, are replaced by fibrous connective tissue4. In our studies, we used a special collagen stain to directly compare fibrosis in mammary tissue caused by H5N1 infection. The vaccinated cows developed significantly less mammary fibrosis in inoculated quarters compared to unvaccinated cows. Although we were unable to confirm a prior history of mastitis for the cows used in this study, mild fibrotic changes to the mammary gland of dairy cattle are not unexpected nor do they severely affect milk production. Vaccinated cows showed mild mammary fibrosis that could be due to a previous mastitis episode or our H5N1 2.3.4.4b virus inoculation but this did not affect milk production as compared to the unvaccinated cows. Our work highlights that the H5 mRNA–LNP 2.3.4.4b vaccine provided protection against severe fibrosing mastitis induced by H5N1 infection that will likely protect milk yields across subsequent lactations.

Previous experimental challenge studies reported that the 2.3.4.4b H5N1 genotype B3.13 virus did not migrate to neighboring uninoculated mammary glands4,5,16. However, in our study, we found evidence that virus migrated from inoculated to uninoculated mammary quarters of unvaccinated cows following 2.3.4.4b H5N1 infection. Viral RNA was detected in milk samples and mammary gland tissue, virus was isolated from milk samples and viral antigen was detected by IHC in mammary gland tissues of uninoculated quarters. Additionally, we observed mild clinical signs of mastitis in corresponding uninoculated quarters of unvaccinated cows. Although we also detected viral antigen by IHC in an uninoculated quarter of one vaccinated cow, we did not detect evidence of virus or viral RNA in milk or in any other sample type from this animal. These data suggest that vaccination prevented further spread of virus within the uninoculated quarter and, furthermore, prevented spread of virus outside of the mammary gland and draining lymph nodes. Our study reports 2.3.4.4b H5N1 genotype B3.13 virus migration between mammary quarters and we found that vaccination reduced viral migration in our studies. Additional studies should evaluate whether other H5N1 vaccine candidates prevent the migration of 2.3.4.4b H5N1 to neighboring uninoculated mammary glands.

It will be important to monitor future antigenic changes in clade 2.3.4.4b viruses in cattle. Most of the serum neutralizing antibodies elicited by H5 mRNA–LNP vaccination in our study targeted antigenic sites A and B in the globular head of HA near the receptor-binding site. These sites have remained antigenically stable within clade 2.3.4.4b viruses and continued surveillance studies will be useful to identify any future changes in these sites. While we did not assess T cell-mediated immunity in the current study, we previously found that mRNA–LNP vaccination induces a potent CD8+ T cell response in calves6. Thus, our H5 mRNA–LNP vaccine is well matched to currently circulating clade 2.3.4.4.b viruses and also elicited T cell responses that may be protective if these viruses evolve to escape antibody responses in the future.

In conclusion, our study demonstrates that the H5 mRNA–LNP 2.3.4.4b vaccine is safe and immunogenic in lactating dairy cows, eliciting robust antibody responses in sera and milk. Moreover, the H5 mRNA–LNP vaccine reduced clinical disease, particularly milk loss, and significantly reduced virus shedding in milk and mammary gland fibrosis in vaccinated lactating cows. The H5 mRNA–LNP 2.3.4.4b vaccine in dairy cows provides an additional tool to use in combination with current control measures to limit the outbreak on farms, reduce spread between farms and alleviate the economic losses to the dairy industry.

Methods

Ethical compliance

The Institutional Animal Care and Use Committees of the University of Pennsylvania Office of Animal Welfare and the US Department of Agriculture (USDA) Agricultural Research Service (ARS) National Animal Disease Center (NADC) reviewed and approved all studies.

Study design

Eight Holstein lactating cows at 2–8 years of age (Supplementary Table 1) were split into two groups at the New Bolton Center Marshak dairy herd at the University of Pennsylvania and received either saline (unvaccinated) or 500 μg of A/dairy cattle/Texas/24-008749-002/2024 (H5N1) HA mRNA–LNP in saline vaccine (vaccinated) intramuscularly, directly into the muscle of the neck in front of the shoulder and below the nuchal ligament. The vaccine regimen consisted of two doses of the mRNA–LNP vaccine administered with a 3-week interval between doses. Dose and vaccine regimen were selected on the basis of our previous studies in calves6. The vaccine was created using standard techniques as previously described17. For unvaccinated cows, saline was used instead of an irrelevant or empty mRNA–LNP as our previous murine studies demonstrated no differences between saline and irrelevant mRNA–LNP controls28,29. Blood samples were collected at 0, 7, 21, 28, 41 and 49 days after vaccination, while milk samples were collected daily. Vaccinated cows were milked segregated away from the herd and their milk did not enter the human food supply. At 42 days after vaccination, cows were transported from the New Bolton Center at the University of Pennsylvania to the USDA ARS NADC in Ames. At 49 days after vaccination, cows were inoculated with 1 ml of 1 × 105 tissue culture infectious dose 50 (TCID50) per ml of A/dairy cattle/Texas/24-008749-002/2024 (H5N1) through the intramammary route in each of two quarters, using a teat canula4. The inoculation occurred after milking and standard teat disinfection and the teats were additionally disinfected with an alcohol wipe immediately before the procedure. After infusion, the inoculum was manually moved upward into the teat sinus with gentle massage.

Antemortem samples such as nasal swab, ocular swab, urine, saliva, blood serum (BD vacutainer SST) and blood in EDTA (BD EDTA K2) were collected from 0 to 7 days after inoculation. Milk samples from each quarter were collected daily up to 14 days after inoculation. After milking, each cow’s bucket of milk was weighed daily from −6 to 14 days after inoculation to evaluate milk production. Lactating cows were monitored continuously for rumination using an ear-tag accelerometer sensor from −6 to 14 days after inoculation (Cow Manager SensOor, Agis Automatisering). Milk production and rumination were expressed as the percentage change relative to the pre-inoculation average. Cows were recorded daily for rectal temperature and visually monitored for clinical signs, including cough, nasal and ocular discharge and fecal consistency. Milk samples were collected by quarter and evaluated for mastitis by the CMT (ImmuCell), for consistency and for color using a visual scorecard on a scale of 0–12 (ref. 4). Samples were also collected from the bucket following milking. At 14 days after inoculation, cows were sedated using an α2-adrenergic receptor antagonist, xylazine (Rompun, Dechra), at 0.1 mg kg−1, administered i.m. Animals were then anesthetized using ketamine (Zetamine, VetOne), at 2 mg kg−1 administered intravenously (i.v.), and humanely killed by i.v. pentobarbital (Fatal-Plus, Vortech) administration per manufacturer’s dosage recommendation. At necropsy, mammary tissue from each quarter, supramammary lymph nodes and the diaphragm were collected for pathologic evaluation and virus detection by PCR as described below. The study was carried out in a working dairy farm facility at the University of Pennsylvania during the vaccination stage and in an Agricultural Biosafety level 3 (BSL3-Ag) facility at the USDA during the challenge stage. The Marshak dairy at the University of Pennsylvania was enrolled in the Pennsylvania Lactating Dairy Cattle Health Monitoring Program, which is the state’s voluntary PCR-based surveillance program that tests for H5N1 infections in dairy cattle. Weekly bulk milk tank testing was implemented before and during the study period and all samples were negative.

Vaccine mRNA undetected in milk

After prime and boost immunizations, milk samples were collected and tested for the presence of residual H5 mRNA from the vaccine. RNA was isolated from 50 μl of milk using the MagMAX-96 viral RNA isolation kit (Thermo Fisher Scientific). The RT–qPCR assay was performed in a QuantStudio 3 real-time PCR instrument (Thermo Fisher Scientific) using the TaqMan fast virus one-step master mix for qPCR (Thermo Fisher Scientific) following the manufacturer’s instructions and the primers/probe designed with the PrimerQuest Tool (Integrated DNA Technologies). The position and sequence of primers and probe used for the assay are reported in Supplementary Table 5. A standard curve was generated by spiking a milk sample with a known concentration of the H5 mRNA–LNP vaccine. A tenfold serial dilution curve, ranging from 8 × 10−0.1 to 1010.9 mRNA copies per μl of template, was used to correlate Ct values with the amount of mRNA in milk. The limit of detection of the assay was assessed as 8 × 100.9 mRNA copies per μl of template (65 mRNA copies per μl). All samples and standards were tested in duplicate. mRNA copy numbers per μl were calculated as follows:

$${\mathrm{Copies}}\,{\mathrm{per}}\,{{\mu}l}=\frac{{\mathrm{mRNA}}\,{\mathrm{amount}}\,\left({\mathrm{ng}}\,{{\mu}l}^{-1}\right)\times 6.022\times {10}^{23}}{{\mathrm{Length}}\,\left({\mathrm{bp}}\right)\times 1\times {{10}^{9}}\times 340}$$

Recombinant HA (rHA) protein expression

The extracellular domain of A/dairy cattle/Texas/24-008749-002-v/2024(H5N1) HA was synthetically fused to the trimeric FoldOn of T4 fibritin, the Avitag sequence and a hexahistidine affinity tag30 and cloned into the pCMV Sport6 expression vector. HA expression plasmids were cotransfected with a plasmid encoding cell surface expressed A/Puerto Rico/8/1934 neuraminidase (NA) into Expi293F cells (Thermo Fisher) with PEI Max reagent (Polysciences), following the manufacturer’s instructions. NA cleaves sialic acid and facilitates the release of the rHA into the supernatant, increasing protein yields. After 4 days, the supernatant was clarified by centrifugation. HA proteins were purified using Ni-NTA agarose resin (Qiagen) and gravity flow chromatography columns (Bio-Rad). Purified HA was buffer-exchanged and concentrated using centrifugal filter units with a 30-kDa molecular weight cutoff (Millipore). Purified HA proteins were stored at −80 °C.

ELISA

ELISA was performed on 96-well Immulon 4HBX extra-high-binding flat-bottom plates (Thermo Fisher). Plates were coated with PBS or recombinant HA protein from A/dairy cattle/Texas/24-008749-002-v/2024(H5N1) at 2 μg ml−1 in DPBS at 4 °C overnight. Plates were washed three times with 1× PBS + 0.1% Tween-20 and then blocked for 1 h at room temperature using blocking buffer (1× TBS, 0.05% Tween-20 and 1% BSA). Bovine serum samples were heat-inactivated at 56 °C for 1 h and then serially diluted fourfold in dilution buffer (1× TBS, 0.05% Tween-20 and 0.1% BSA). Bovine milk samples were serially diluted twofold in dilution buffer (1× TBS, 0.05% Tween-20, and 0.1% BSA). After blocking, plates were washed three times with 1× PBS + 0.1% Tween-20. Diluted milk samples were then transferred to ELISA plates and incubated for 2 h at room temperature. After washing three times with 1× PBS + 0.1% Tween-20, peroxidase-conjugated goat anti-bovine IgG polyclonal antibody (Jackson ImmunoResearch Laboratories) was added to the corresponding wells at a concentration of 1:5,000 in dilution buffer for 1 h at room temperature. After washing three times with 1× PBS + 0.1% Tween-20, all plates were developed by adding SureBlue TMB peroxidase substrate (SeraCare) for 5 min at room temperature, followed by stopping the reaction with 250 mM HCl solution. The absorbance was measured at 450 nm using a SpectraMax ABS Plus plate reader (Molecular Devices). Background optical density (OD) values from the plates coated with PBS were subtracted from the OD values from plates coated with recombinant HA protein. Endpoint titers were calculated as the reciprocal serum dilution that generated an equivalent OD of 0.5.

Focus reduction neutralization assay

Focus reduction neutralization (FRNT) assays were carried out using conditionally replicative influenza viruses that encode GFP instead of PB1, as previously described31. Conditionally replicative viruses were generated encoding the HA and NA from A/dairy cattle/Texas/24-008749-002-v/2024(H5N1). Bovine serum samples were pretreated with receptor-destroying enzyme (RDE) (Denka Seiken) for 2 h at 37 °C, followed by heat inactivation at 56 °C for 30 min. RDE-treated sera were serially diluted in neutralization assay medium (NAM; medium 199 supplemented with 0.01% heat-inactivated FBS, 0.3% BSA, 100 U per ml penicillin, 100 µg ml−1 streptomycin, 100 µg ml−1 calcium chloride and 25 mM HEPES), mixed with virus and incubated at 37 °C for 1 h. After incubation, 2.5 × 104 MDCK-SIAT1-CMV-PB1-TMPRSS2 cells were added to each well. Plates were incubated for 40 h at 37 °C followed by measurement of GFP fluorescence intensity on an EnVision multimode microplate reader (PerkinElmer) using an excitation wavelength of 485 nm and an emission wavelength of 515 nm. Neutralization titers were expressed as the reciprocal of the highest dilution of sera that inhibited 90% of GFP expression relative to control wells with no sera.

HAI assay

For the HAI assay, 50 µl of bovine serum was treated with 150 µl of RDE (Denka Seiken) with an overnight incubation at 37 °C in a water bath, followed by 300 µl of saline solution (0.85%) and heat inactivation at 56 °C for 30 min. RDE-treated serum samples were adsorbed with 100% rooster red blood cells for 1 h at 4 °C to remove nonspecific agglutinins. RDE-treated serum samples were run in HAI assays using 0.5% rooster red blood32.

Mapping of antibody specificity by deep mutational scanning

Production of H5 HA deep mutational scanning libraries has been described previously23. The libraries are based on the HA of a 2.3.4.4b clade A/American Wigeon/South Carolina/USDA-000345-001/2021 strain, with mutants of this HA expressed on the surface of pseudotyped lentiviral particles (pseudoviruses). Importantly, these pseudoviruses can only undergo a single round of cell entry and are, therefore, not fully infectious pathogens capable of causing disease, thereby providing a safe way to study the impacts of substitutions in HA. To identify substitutions that affect neutralization by the cattle sera, deep mutational scanning libraries were incubated with serum concentrations that were sufficient to neutralize between 50% and 90% of the pseudovirus libraries, as estimated by neutralization standard used in deep mutational scanning experiments. After 45 min of incubation at 37 °C, the serum–virus mix was used to infect 293T cells. At 15 h after infection, nonintegrated viral genomes were recovered and prepared for Illumina sequencing as described previously23. A nonneutralizable standard that was added to the libraries was used to determine effects of mutations on neutralization23. Neutralization escape for individual libraries was calculated using a biophysical model implemented in polyclonal package33. For each serum experiments were performed with two biological library replicates. Before use, sera were treated with RDE for 2 h at 37 °C and heat-inactivated for 30 min at 56 °C.

The serum neutralization data and data analysis pipeline for deep mutational scanning are available on GitHub (https://github.com/dms-vep/Flu_H5N1_American-Wigeon_2021_HA_DMS_cattle_Penn_sera). An interactive map of the serum neutralization data is available on the Deep Mutational Scanning of Viral Entry Proteins repository (https://dms-vep.org/Flu_H5N1_American-Wigeon_2021_HA_DMS_cattle_Penn_sera/htmls/cattle_sera_escape_faceted.html). H3 numbering of the mature HA peptide is used in this analysis.

To assess the conservation at key amino acid sites targeted by the neutralizing activity of the sera, we downloaded all H5 HA proteins in GISAID as of October 27, 2025 (acknowledgments in GISAID EPI_SET_251028bx, EPI_SET_251028no and EPI_SET_251028mu). We then calculated the frequency of each amino acid at each site among all H5 HAs, clade 2.3.4.4b HAs and dairy cow HAs.

Viral RNA detection in clinical samples

Viral RNA was extracted from nasal swabs, ocular swabs, urine, saliva, sera and liquid collected from thawed diaphragm tissue using the MagMAX CORE nucleic acid purification kit (Thermo Fisher Scientific) with an extra wash using 80% ethanol. A piece of approximately 10 mm of mammary gland tissue from each quarter and supramammary lymph nodes were processed separately with the clarifying solution of the MagMAX CORE mechanical lysis module for tissue homogenization in a GentleMACS Octo dissociator (Miltenyi Biotec). Following tissue dissociation, viral RNA was extracted with the same protocol described above. Viral RNA was extracted from milk samples using MagMAX CORE mastitis and pan-bacteria module (Thermo Fisher Scientific). Whole-blood samples were processed for RNA extraction using LeukoLOCK total RNA isolation system (Thermo Fisher Scientific).

All RNA extracted were assessed by PCR for viral RNA detection using a commercial RT–qPCR VetMAX-Gold SIV detection kit (Thermo Fisher Scientific). Milk samples that were PCR positive with Ct ≤ 35 were submitted to the USDA Animal and Plant Health Inspection Service (APHIS) National Veterinary Servies Laboratories (NVSL) lab to perform virus isolation in 10-day-old embryonated chicken eggs to determine the presence of viable virus. Ct values from RT–qPCR of tenfold diluted stock virus of known TCID50 titer were extrapolated into a virus titration curve using GraphPad Prism (version 10.3.0) and reported as log10 TCID50 per ml equivalent Ct.

Pathological assessment of mammary gland tissues

At necropsy, mammary gland quarters were identified (front left, front right, rear left and rear right) by color markings and the entire gland was dissected from the abdominal wall. For macroscopic evaluation, each quarter was dissected into cranial, middle and caudal sections that were assigned a percentage of fibrosis score (percent of section that was fibrotic), which was then averaged by quarter. From each cranial, middle and caudal section, three representative histologic sections were taken for a total of nine histologic sections per quarter and 36 sections per cow. Formalin-fixed tissues were transferred to 70% ethanol 48 h after collection and processed routinely for microscopic evaluation. Tissues were stained by hematoxylin and eosin stain and Picrosirius red stain per manufacturer’s instructions (AbCam) to highlight fibrosis34. Each section (36 per cow) was scored using a dual-component (intralobular and interlobular fibrosis) histologic scoring rubric that was developed to assess mammary fibrosis. Each component was independently scored on the basis of two criteria: histologic features (scale 0–4) and the percentage of tissue affected (scale 0–4). These scores were summed to produce a component-specific score ranging from 0 to 8. The overall fibrosis score was the combined total of both components (intralobular and interlobular fibrosis), yielding a maximum possible score of 16. The results by quarter were summed for a possible total of 144 and then the median by status (uninoculated or inoculated) of gland within a cow (Supplementary Tables 6 and 7). IHC staining targeting the NP antigen (GeneTex) of the influenza A virus was performed on all sections that originated from an uninoculated quarter that had viral RNA detected at any time point, as previously described35.

Statistical analysis

Antibody data analyses were performed using GraphPad Prism Software version 10 (GraphPad Software) using a log-normal two-tailed t-test with Welch correction. The percent changes in rumination, milk production and milk PCR in the uninoculated quarters were analyzed with a two-way repeated-measures ANOVA followed by a Tukey’s multiple-comparison test to compare vaccinated versus unvaccinated groups across days after inoculation. Viral RNA detection by RT–qPCR in the inoculated quarters were analyzed by a mixed-effects model (residual maximum likelihood model) followed by Tukey’s multiple comparisons test. Differences in mammary microscopic lesions (fibrosis) and mammary macroscopic lesions between vaccinated and unvaccinated cows were evaluated using the Mann–Whitney U-test. Statistical significance was evaluated using an α threshold of 0.05. P values greater than 0.05 were considered not significant. Pearson’s correlation coefficients in unvaccinated and vaccinated lactating cows were calculated between Ct values (milk bucket and milk from inoculated and uninoculated quarters) and parameters: milk production (n = 56), rumination time (n = 58), CMT scores from inoculated quarters (n = 60) and uninoculated quarters (n = 60), milk color from inoculated quarters (n = 60) and uninoculated quarters (n = 60) and milk consistency scores from inoculated quarters (n = 60) and uninoculated quarters (n = 60), across all time points, using GraphPad Prism Software version 10 (GraphPad Software). Correlations were considered significant at P < 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Source data are provided with this paper.

References

  1. Nguyen, T. Q. et al. Emergence and interstate spread of highly pathogenic avian influenza A (H5N1) in dairy cattle in the United States. Science 388, eadq0900 (2025).

    Article  CAS  PubMed  Google Scholar 

  2. Pena-Mosca, F. et al. The impact of highly pathogenic avian influenza H5N1 virus infection on dairy cows. Nat. Commun. 16, 6520 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. Mostafa, A. et al. Avian influenza A (H5N1) virus in dairy cattle: origin, evolution, and cross-species transmission. mBio 15, e0254224 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  4. Baker, A. L. et al. Dairy cows inoculated with highly pathogenic avian influenza virus H5N1. Nature 637, 913–920 (2025).

    Article  CAS  PubMed  Google Scholar 

  5. Halwe, N. J. et al. H5N1 clade 2.3.4.4b dynamics in experimentally infected calves and cows. Nature 637, 903–912 (2025).

    Article  CAS  PubMed  Google Scholar 

  6. Souza, C. K. et al. H5 influenza virus mRNA–lipid nanoparticle (LNP) vaccination elicits adaptive immune responses in Holstein calves. Preprint at bioRxiv https://doi.org/10.1101/2025.05.01.651548 (2025).

  7. Burrough, E. R. et al. Highly pathogenic avian influenza A (H5N1) clade 2.3.4.4b virus infection in domestic dairy cattle and cats, United States, 2024. Emerg. Infect. Dis. 30, 1335–1343 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  8. Caserta, L. C. et al. Spillover of highly pathogenic avian influenza H5N1 virus to dairy cattle. Nature 634, 669–676 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. Song, H. et al. Receptor binding, structure, and tissue tropism of cattle-infecting H5N1 avian influenza virus hemagglutinin. Cell 188, 919–929 (2025).

    Article  CAS  PubMed  Google Scholar 

  10. Nelli, R. K. et al. Exploring influenza A virus receptor distribution in the lactating mammary gland of domesticated livestock and in human breast tissue. J. Dairy Sci. 109, 1891–1906 (2026).

    Article  CAS  PubMed  Google Scholar 

  11. Center for Veterinary Biologics, USDA. Notification of U.S. Department of Agriculture’s (USDA) Request for Information for Highly Pathogenic Avian Influenza (HPAI) Vaccines for Use in Cattle. Notice No. 24-09. (2024).

  12. Zhang, S., Liu, G., Guo, A. & Chen, Y. Maternal antibody transfer efficiency: the impact of M. bovis-BoHV-1 combined vaccine. Virology 611, 110656 (2025).

    Article  CAS  PubMed  Google Scholar 

  13. Abousenna, M. S., Shafik, N. G. & Abotaleb, M. M. Evaluation of humoral immune response and milk antibody transfer in calves and lactating cows vaccinated with inactivated H5 avian influenza vaccine. Sci. Rep. 15, 4637 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Wiggins, J., Madapong, A. & Weaver, E. A. Dual-route H5N1 vaccination induces systemic and mucosal immunity in murine and bovine models. NPJ Vaccines 11, 123 (2026).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Kong, H. et al. Protective efficacy of a hemagglutinin-based mRNA vaccine against H5N1 influenza virus challenge in lactating dairy cows. Research 9, 1104 (2026).

    Article  PubMed  PubMed Central  Google Scholar 

  16. Shi, J. et al. H5N1 virus invades the mammary glands of dairy cattle through ‘mouth-to-teat’ transmission. Natl Sci. Rev. 12, nwaf262 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Furey, C. et al. Development of a nucleoside-modified mRNA vaccine against clade 2.3.4.4b H5 highly pathogenic avian influenza virus. Nat. Commun. 15, 4350 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Low, J. M. et al. Codominant IgG and IgA expression with minimal vaccine mRNA in milk of BNT162b2 vaccinees. NPJ Vaccines 6, 105 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Hanna, N. et al. Detection of messenger RNA COVID-19 Vaccines in human breast Milk. JAMA Pediatr. 176, 1268–1270 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

  20. Hanna, N. et al. Biodistribution of mRNA COVID-19 vaccines in human breast milk. EBioMedicine 96, 104800 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Lang, Y. et al. Detection of antibodies against influenza A viruses in cattle. J. Virol. 99, e0213824 (2025).

    Article  PubMed  PubMed Central  Google Scholar 

  22. Dadonaite, B. et al. A pseudovirus system enables deep mutational scanning of the full SARS-CoV-2 spike. Cell 186, 1263–1278 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Dadonaite, B. et al. Deep mutational scanning of H5 hemagglutinin to inform influenza virus surveillance. PLoS Biol. 22, e3002916 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Hatta, M. et al. Influenza mRNA vaccine reduces pathogenicity and transmission of A (H5N1) virus in a ferret model. NPJ Vaccines 10, 263 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  25. Romagosa, A. et al. Vaccination of influenza a virus decreases transmission rates in pigs. Vet. Res. 42, 120 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Lowen, A. C. et al. Blocking interhost transmission of influenza virus by vaccination in the guinea pig model. J. Virol. 83, 2803–2818 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. van der Goot, J. A., Koch, G., de Jong, M. C. & van Boven, M. Quantification of the effect of vaccination on transmission of avian influenza (H7N7) in chickens. Proc. Natl Acad. Sci. USA 102, 18141–18146 (2005).

    Article  PubMed  PubMed Central  Google Scholar 

  28. Pardi, N. et al. Nucleoside-modified mRNA vaccines induce potent T follicular helper and germinal center B cell responses. J. Exp. Med. 215, 1571–1588 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Pardi, N. et al. Nucleoside-modified mRNA immunization elicits influenza virus hemagglutinin stalk-specific antibodies. Nat. Commun. 9, 3361 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

  30. Whittle, J. R. et al. Flow cytometry reveals that H5N1 vaccination elicits cross-reactive stem-directed antibodies from multiple Ig heavy-chain lineages. J. Virol. 88, 4047–4057 (2014).

    Article  PubMed  PubMed Central  Google Scholar 

  31. Doud, M. B., Hensley, S. E. & Bloom, J. D. Complete mapping of viral escape from neutralizing antibodies. PLoS Pathog. 13, e1006271 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  32. Kitikoon, P., Gauger, P. C. & Vincent, A. L. Hemagglutinin inhibition assay with swine sera. Methods Mol. Biol. 1161, 295–301 (2014).

    Article  PubMed  Google Scholar 

  33. Yu, T. C. et al. A biophysical model of viral escape from polyclonal antibodies. Virus Evol. 8, veac110 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

  34. Farris, A. B. & Alpers, C. E. What is the best way to measure renal fibrosis?: A pathologist’s perspective. Kidney Int. Suppl. (2011) 4, 9–15 (2014).

    Article  PubMed  PubMed Central  Google Scholar 

  35. Arruda, B. et al. Divergent pathogenesis and transmission of highly pathogenic avian influenza A (H5N1) in swine. Emerg. Infect. Dis. 30, 738–751 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We thank the staff at the New Bolton Center at the University of Pennsylvania. In particular, we acknowledge the technical support provided by K. Basiilio, M. Roney and P. Vermillion at the New Bolton Center Marshak Dairy Herd and New Bolton Center laboratory support from E. Yvonne Miller-Lux and B. Osborne. We thank the USDA NADC and NVSL leadership and animal resource and facilities and engineering units. In particular, we thank K. Young, E. Love and S. Anderson for laboratory technical assistance and T. McNunn for compliance, D. Chapman, M. Crocheck, J. Rasmussen, T. Williams, J. Gardner, J. Peterson, E. Hay, E. Pratt and multiple personnel from NADC for their expertise, support and assistance with animal studies in the high containment BSL-3Ag facility. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the US Government. USDA is an equal-opportunity provider and employer.

Funding

This work was supported in part by the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services (contract numbers 75N93021C00015), the USDA ARS (project number 5030-32000-231-000-D), the USDA APHIS (contract number 3200-231-112-I) and the Commonwealth of Pennsylvania (C.A.H., L.A.M.). D.B.G., G.C.Z. and C.A.S. were supported by an appointment to the USDA ARS Research Participation Program administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the US Department of Energy and the USDA under contract no. DE-AC05-06OR23100.

Author information

Author notes

  1. These authors contributed equally: Jefferson J. S. Santos, Carine K. Souza.

Authors and Affiliations

  1. Department of Microbiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

    Jefferson J. S. Santos, Reilly K. Atkinson & Scott E. Hensley

  2. National Animal Disease Center – United States Department of Agriculture, Agricultural Research Service, Ames, IA, USA

    Carine K. Souza, Giovana C. Zanella, Débora B. Goulart, Bailey Arruda, Paola Boggiatto, Mitchell V. Palmer, Celeste A. Snyder & Amy L. Baker

  3. Department of Clinical Studies – New Bolton Center, School of Veterinary Medicine, University of Pennsylvania, Kennett Square, PA, USA

    Michaela A. Kristula, Charlene Dickens, Terry L. Webb & Gary C. Althouse

  4. Basic Sciences Division and Computational Biology Program, Fred Hutchinson Cancer Center, Seattle, WA, USA

    Bernadeta Dadonaite & Jesse D. Bloom

  5. Division of Infectious Diseases, Philadelphia, USA

    Garima Dwivedi & Drew Weissman

  6. Penn Institute of RNA Innovation, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

    Garima Dwivedi, Mohamad-Gabriel Alameh & Drew Weissman

  7. Department of Pathology and Laboratory Medicine, Children’s Hospital of Philadelphia, Philadelphia, PA, USA

    Mohamad-Gabriel Alameh

  8. Department of Pathobiology, School of Veterinary Medicine, University of Pennsylvania, Philadelphia and Kennett Square, Philadelphia, PA, USA

    Christopher A. Hunter & Lisa A. Murphy

  9. Howard Hughes Medical Institute, Seattle, WA, USA

    Jesse D. Bloom

Authors

  1. Jefferson J. S. Santos
  2. Carine K. Souza
  3. Giovana C. Zanella
  4. Débora B. Goulart
  5. Bailey Arruda
  6. Paola Boggiatto
  7. Mitchell V. Palmer
  8. Celeste A. Snyder
  9. Michaela A. Kristula
  10. Charlene Dickens
  11. Terry L. Webb
  12. Reilly K. Atkinson
  13. Bernadeta Dadonaite
  14. Garima Dwivedi
  15. Mohamad-Gabriel Alameh
  16. Christopher A. Hunter
  17. Lisa A. Murphy
  18. Jesse D. Bloom
  19. Drew Weissman
  20. Gary C. Althouse
  21. Amy L. Baker
  22. Scott E. Hensley

Contributions

Conceptualization, S.E.H., A.L.B., G.C.A., J.J.S.S. and C.K.S. Sample collection, J.J.S.S., C.K.S., G.C.Z., D.B.G., B.A., P.B., M.V.P., C.A.S., M.A.K., C.D., T.L.W., R.K.A. and B.D. Investigation, J.J.S.S., C.K.S., G.C.Z., D.B.G., B.A., P.B., M.V.P., C.A.S. and B.D. Formal analysis, J.J.S.S., C.K.S., D.B.G., B.A., B.D. and J.D.B. Visualization, J.J.S.S., C.K.S., B.A., B.D. and J.D.B. Writing—original draft, J.J.S.S., C.K.S., G.C.A., A.L.B. and S.E.H. Writing—review and editing, J.J.S.S., C.K.S., G.C.Z., D.B.G., B.A., P.B., M.V.P., C.A.S., M.A.K., C.D., T.L.W., R.K.A., B.D., G.D., M.G.A., C.A.H., L.A.M., J.D.B., D.W., G.C.A, A.L.B. and S.E.H. Supervision, S.E.H., A.L.B., G.C.A. and D.W. Funding acquisition, S.E.H., A.L.B., C.A.H. and L.A.M.

Corresponding authors

Correspondence to Gary C. Althouse, Amy L. Baker or Scott E. Hensley.

Ethics declarations

Competing interests

S.E.H. and D.W. are coinventors on patents that describe the use of nucleoside-modified mRNA as a platform to deliver therapeutic proteins and as a vaccine platform. D.W. is also named on patents describing the use of lipid nanoparticles and lipid compositions for nucleic acid delivery. S.E.H. reports receiving consulting fees from Sanofi, Pfizer, Lumen, Novavax and Merck. JDB consults for Apriori Bio, Invivyd, GSK, Pfizer and the Vaccine Company. J.D.B. and B.D. are inventors on Fred Hutch licensed patents related to viral deep mutational scanning. C.A.H. has equity in Coherus Oncology and CytomX Therapeutics and has received financial compensation as a member of scientific advisory boards for Coherus Oncology, Synthekine, CytomX Therapeutics, Flagship Medicine and AnaptysBio, as well as research support from Incyte and Coherus Oncology. C.A.H also has patent disclosures that relate to the development of mRNA vaccines. The remaining authors declare no competing interests.

Peer review

Peer review information

Nature Biotechnology thanks Shuo Su and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Absence of vaccine mRNA detection in milk after vaccination.

Standard curve of known concentration of mRNA spiked in milk and threshold cycle (Ct) measured using a real-time qPCR for H5 mRNA vaccine. Data shown are mean of two replicates.

Source data

Extended Data Fig. 2 Evidence of influenza virus pre-exposure in sera from lactating dairy cows in the study.

Sera collected at day 0 from animals in the study were quantified by ELISA against influenza H1, H3, and H5 rHAs, as well as SARS-CoV-2 RBD for each animal.

Extended Data Fig. 3 Comparison of binding antibodies in sera and milk of vaccinated animals.

Sera (closed black circles) or milk (closed gray circles) collected at days 21, 28, 41, and 49 from vaccinated cows were quantified by ELISA. Data are mean ± SEM, and individual data points (n = 4) are shown. Dashed lines indicate the lower limit of detection. Significant statistical differences (P < 0.05) between groups were determined using a lognormal two tailed t-test with Welch correction and are indicated by an asterisk.

Source data

Extended Data Fig. 4 Evidence of virus migration from an inoculated to uninoculated mammary quarters.

Red represents positive, yellow suspect and white negative for each test, including qPCR on milk, virus isolation on milk; VI, mammary tissues, other clinical samples and IHC. SMLN, supramammary lymph node; DL, Diaphragm liquid; NS, nasal swab; OS, ocular swab; LL, LeucoLOCK from whole blood.

Extended Data Fig. 5 Detection of highly pathogenic avian influenza virus by immunohistochemistry in uninoculated mammary quarters.

Immunolabeling of numerous secretory epithelial cells (arrowhead) and cellular debris in the lumens of secretory alveoli (arrow) (a; Cow 7, vaccinated, rear right quarter, uninoculated). Immunolabeling of a secretory epithelial cell (arrowhead) and cellular debris in the lumen of a secretory alveoli (arrow) (b; Cow 4, front left quarter, unvaccinated uninoculated).

Extended Data Fig. 6 Correlation between virus RNA detection and clinical signs in unvaccinated and vaccinated lactating cows.

Black open boxes highlighted the significant (P < 0.05, two-sided) Pearson’s correlation coefficients in unvaccinated (a) and vaccinated (b) lactating cows are shown between Ct values (milk bucket, milk from inoculated and uninoculated quarters) and milk production, rumination time, CMT scores from inoculated and uninoculated quarters, milk color from inoculated and uninoculated quarters, and milk consistency scores from inoculated and uninoculated quarters, across all time points.

Supplementary information

Reporting Summary (download PDF )

Supplementary Table 1-7 (download XLSX )

Supplementary Table 1. Date of birth and age of lactating cows in the study. Supplementary Table 2. Virus isolation from milk samples with qPCR Ct values < 35.5. Supplementary Table 3. Macroscopic lesion score in uninoculated mammary quarters. Supplementary Table 4. Microscopic lesion score of uninoculated mammary quarters of vaccinated and unvaccinated lactating cows. Supplementary Table 5. Primers and probe sequences for A/dairy cattle/Texas/24-008749-002/2024(H5N1) HA (co-Tx/24 H5 HA) mRNA. Supplementary Table 6. Microscopic lesion scoring system for fibrosis. Supplementary Table 7. Microscopic lesions scores for fibrosis (0-144).

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Santos, J.J.S., Souza, C.K., Zanella, G.C. et al. An mRNA–lipid nanoparticle vaccine against H5N1 influenza virus protects lactating dairy cows from infection. Nat Biotechnol (2026). https://doi.org/10.1038/s41587-026-03343-6

Download citation

  • Received: 27 February 2026

  • Accepted: 14 September 2026

  • Published: 28 October 2026

  • Version of record: 28 October 2026

  • DOI: https://doi.org/10.1038/s41587-026-03343-6

来源:Nature Biotechnology · nature.com