Avian Influenza Virus Inactivation in Beef Slurries in Response to Food-Relevant Conditions of pH and Temperature

Virus

Meat slurries were inoculated with a low pathogenic avian influenza virus (LPAIV) isolate designated A/rgGyrfalconHAxPR8/2014 H5N1 and referred to herein as strain rgGYR/14. Further details on the genetic makeup of this strain, as well as its propagation and maintenance, are described in prior studies (Spackman and Killian, 2020; Stephens and Spackman, 2017). A stock solution of this virus was prepared at the Charles C. Allen Biotechnology Laboratory (CABL), University of Delaware (Newark, DE, USA), via protocols approved by the University of Delaware Institutional Biosafety Committee. A high-volume, high-titer preparation of this LPAIV isolate was transported (under appropriate containment conditions and with the requisite permits) via a commercial courier to the United States Department of Agriculture (USDA) Agricultural Research Service (ARS) Eastern Regional Research Center (ERRC; Wyndmoor, PA, USA) and subsequently used to inoculate beef slurries (see below).

Preparation of beef slurry

In preliminary studies, slurries comprised of autoclaved tap water and various amounts (i.e., 5, 10, 20, and 30% meat) of lean raw ground beef [85% lean:15% fat (85:15)] were evaluated for viscosity, consistency, and ease-of-use (i.e., amenable to transfer via pipetting). In general, a 5% slurry did not provide sufficient beef substrate, while the 20 and 30% slurries were too viscous and presented difficulty when attempting to adjust pH, either of which scenario could potentially hinder accurate handling and possibly prevent consistent experimental outcomes or extrapolations. For these reasons, the experiments detailed herein were conducted using a 10% slurry prepared with one fat level of fresh ground beef (85:15) as purchased from a local supermarket. After storage at −20°C for 1–7 d, previously portioned aliquots (ca. 250 g each) of vacuum-packed ground beef were separately transferred to 4°C and thawed for ca. 12 h. A 10% (w/v) beef slurry was prepared by blending 200 g of ground beef with 1.8 L of autoclaved tap water for at least 5 min using a commercial blender (Waring CB15BU Food Blender Heavy-Duty Base; Stamford, CT, USA). The pH of this fresh ground beef slurry (ca. 5 mL per sample), as measured using a model 6,000 P pH/temperature electrode and a model 5500 pH meter (Daigger; Vernon Hills, IL, USA), was on average pH 5.8 ± 0.15 (N = 6, n = 1). Next, as detailed in Figure 1, a 500-mL volume of 10% slurry was transferred to each of 3 individual flasks to facilitate the adjustment of pH to ca. pH 4.4 or ca. pH 5.0 using 400 or 250 μL, respectively, of a reagent-grade lactic acid solution (≥85% in water; Sigma-Aldrich; St. Louis, MO, USA). The pH of the 10% slurry within the third flask was left unadjusted to serve as a positive control (ca. pH 5.8 ± 0.15; see above) because it falls within the pH range typical for raw ground beef (Haider et al., 2020). Note, the pH of slurries adjusted to pH 4.4 and pH 5.0 with lactic acid, as well as the pH of the control slurry (pH 5.8) that was not adjusted, remained relatively stable for up to 48 h at 4°C (Data not shown).

Figure 1.

Preparation, storage, and analysis of 10% meat slurries inoculated with avian influenza virus.

Inoculation and neutralization of beef slurry

The procedures herein for conducting research using beef slurry samples containing LPAIV were reviewed and approved by the USDA ARS Northeast Area (NEA) Institutional Biosafety Committee, and these experiments were conducted in a BSL-2 enhanced laboratory at ERRC. In brief, 500 mL of slurry were inoculated with 5 mL of LPAIV strain rgGYR/14 to achieve an average initial level of ca. 4.5 log₁₀ 50% embryo infectious doses (EID₅₀) per mL of slurry. The contents of the flasks containing the inoculated slurries were swirled manually for about 15 s to distribute the inoculum.

After inoculation, the ca. 500 mL of slurry from each pH treatment was divided equally between 2 flasks that were subsequently placed into circulating water baths (Model 2874; Thermo Fisher Scientific; Marietta, OH, USA) maintained at either 23°C or 37°C. At time 0, but before placing flasks into the circulating water baths, 2 ca. 15-mL portions of each treatment (pH 4.4, pH 5.0, and pH 5.8) were decanted separately into a 50-mL centrifuge tube. Likewise, at each sampling interval (3, 7, 9, and 15 h), flasks were removed from the water baths, and 2 ca. 15-mL portions of each treatment were also decanted separately into a 50-mL centrifuge tube. Next, the slurry samples were neutralized to ca. pH 7.0 using 4 mL of sterile Dey Engley (DE) neutralizing broth (Sigma-Aldrich) (ca. 19 mL total volume per sample). The DE broth also contained a penicillin-streptomycin solution [final concentrations in neutralized slurry samples were 10,000 IU per mL penicillin and 10,000 μg per mL streptomycin; Gibco, Thermo Fisher Scientific; Grand Island, NY, USA; Lombardi et al. (2008)] to lower the levels of the indigenous flora within the retail raw ground beef and to preclude their further amplification, since higher levels and certain types of this microbial community would subsequently cause false positives (e.g., embryo mortality attributed to bacteria rather than AIV) in embryonated chicken egg (ECE) assays. To neutralize each treatment to ca. pH 7.0, the concentration of the DE broth varied, with 4-fold (4X) DE broth required to neutralize slurries adjusted to pH 4.4, 2-fold (2X) DE broth for slurries adjusted to pH 5.0, and 1-fold (1X) DE broth for pH 5.8 slurries. In prefatory experiments, after adjustment to a target pH of ca. pH 7.0, the average pH of neutralized slurries was pH 6.93 ± 0.06 (range pH 6.82 to pH 7.02; 6 samples/pH level x 3 pH levels = 18 total pH readings: N = 1, n = 6 for each pH treatment).

After neutralization with DE broth, each slurry sample was thoroughly vortexed, and a ca. 4-mL portion of each sample to be assayed for infectious strain rgGYR/14 was aseptically transferred into cryovial tubes (SPL Lifesciences, Catalog No 43023, Gyeonggi-do, Korea) and appropriately sealed/prepared for subsequent transport to CABL. Samples were then held at ambient temperature for ca. 1 h to allow for the penicillin-streptomycin solution to cause untoward effects on the indigenous bacterial community. After overnight storage at 4°C, samples were placed on ice and driven by a certified ERRC team member to CABL for virological analyses following the requisite incident response protocols and attendant security plans approved by the biosafety office of the USDA ARS NEA for transporting LPAIV specimens.

The experimental matrix for slurries incubated at 23°C or 37°C consisted of 1 inoculation level × 1 fat level × 1 strain of avian influenza virus × 3 pH levels × 5 sampling points x 2 replicates per sampling point × 6 trials = 180 samples total per temperature. A noninoculated sample (15 mL) adjusted to pH 4.4 with lactic acid and neutralized with 4X DE (4 mL) at time 0 h served as a negative control.

Virus quantification

Infectious LPAIV or nucleic acid therefrom were recovered from beef slurries essentially as described (Alphin et al., 2009), with only minor modifications. In brief, a 4-mL portion of each slurry sample was transferred into microfuge tubes (Fisher Scientific; Pittsburgh, PA, USA) and clarified by centrifugation (1,200 × g for 5–10 min at 4°C; Allegra 6R, Beckman Coulter, Inc., Sharon Hill; PA, USA). A ca. 2-mL portion of the resulting supernatant was transferred into a second microfuge tube and subjected to another round of centrifugation (17,720 × g for ca. 10 min at 4°C). The clarified supernatant from the second centrifugation was transferred into a microfuge tube and held at ambient temperature for ca. 60 min to allow additional time for the aforementioned penicillin-streptomycin solution to be effective. Given that each EID₅₀ titration required a substantial number of embryonated eggs, titrations were performed only at the initial and final sampling points. Because it would have exceeded our allotted budget and capacity/throughput, aliquots of slurries from intermediate timepoints were not titered for infectious LPAIV (i.e., “not determined”). For titration of AIV levels, a 0.1-mL portion of clarified supernatant from each test sample was inoculated [with or without prior dilution in brain heart infusion (BHI; BD Difco; Franklin Lakes, NJ, USA) broth plus a penicillin-streptomycin-amphotericin B solution (concentrations of the antimicrobial stock solution were 5,000 IU per mL penicillin, 5,000 μg per mL streptomycin, and 25 mg per mL amphotericin B; Gibco) and holding for ca. 60 min at ambient temperature] into each of 3 or 4, 10-day-old specific pathogen-free (SPF) embryonated chicken eggs (ECE; AVSBio; Norwich, CT, USA). Inoculated ECE were incubated at 37°C and candled daily for up to 6 d to monitor embryo survival as detailed below. The limit of detection was ≤1.1 log₁₀ EID₅₀ per mL of slurry.

Embryo deaths within the first 24 h post-inoculation of SPF eggs were considered non-specific and, as such, these eggs/embryos were excluded from further analyses and appropriately discarded. Eggs displaying mortality as recorded between 2 and 6 d were held at 4°C for 24 h. The chorioallantoic fluid (CAF; 1–2 mL per egg) was harvested from each retained egg, aseptically transferred into separate microcentrifuge tubes, and stored at −80°C for ≤5 d, and thawed once at room temperature immediately before HA testing. CAF samples were subjected to only a single freeze-thaw cycle because repeated freeze-thaw cycles can appreciably reduce HA infectivity (WHO, 2011; WOAH, 2021).

The CAF was subsequently analyzed for hemagglutination (HA) activity following established protocols (Alphin et al., 2009; Swayne and Halvorson, 2008). Briefly, fresh chicken red blood cells (CRBC) were collected from birds housed on the University of Delaware campus farm and mixed with a 50% Alsever’s solution, centrifuged at 1,000 × g for 5 min at 4°C, and washed 3 times with phosphate-buffered saline (PBS; Gibco; Waltham, MA, USA). A 10% CRBC stock suspension was prepared in sterile PBS, and a final dilution was done to obtain a 1% CRBC suspension in sterile PBS. For the HA assay, 25 μL of CAF from each egg was combined with 25 μL of PBS, to which 25 μL of the 1% CRBC suspension was added to a sterile 96-well, U-bottom, microtiter plate (Fisher Scientific). The detection of HA activity in CAF (i.e., by the formation of a CRBC “shield”) from any one of the inoculated eggs was indicative of the presence of virus within a given sample. Note, since we did not confirm that HA activity was ascribable to AIV, other causes of hemagglutination, such as bacterial lectins (Neter et al., 1952) or chemical agglutinins present in the beef slurry supernatant (Luner et al., 1975), may also have been in play.

RNA extraction and real-time reverse transcriptase quantitative PCR

The RNA was extracted from 200-μL of the supernatant of twice washed/centrifuged slurry samples using the MagMAX CORE Nucleic Acid Purification Kit (Applied Biosystems; Waltham, MA, USA) per the manufacturer’s instructions. The RNA was eluted in 90-μL of nuclease-free water, with 8-μL used as the template for analyses by real-time reverse transcriptase quantitative PCR (RT-qPCR). A one-step RT-qPCR assay targeting the AIV matrix gene was performed using the USDA-licensed VetMAX^™-Gold SIV Detection Kit (Applied Biosystems) per the manufacturer’s instructions. Reactions were performed using an ABI 7500 Fast Real Time PCR System (Applied Biosystems), and results were expressed as cycle threshold (Ct) values. The RT–qPCR assay was performed as previously described (Spackman et al., 2002; Spackman et al., 2024b) using a standard curve (R² = 0.99) generated de novo to convert the Ct values into estimated titers in EID₅₀ per mL equivalents.

Chemical analyses

Proximate chemical analyses were conducted on a single ca. 225-g composite (representative) sample from each of 3 batches of retail raw ground beef and from the corresponding slurries (N = 3 batches of ground beef/slurries; n = 1 sample per batch). Chemical analyses of all ground beef and slurry samples were conducted by a commercial testing facility using methods approved by the AOAC (AOAC, 2023) as follows: fat (AOAC 960.39 for ground beef; AOAC 933.05 for slurries), ash (AOAC 920.153), moisture (AOAC 950.46A), protein (AOAC 992.15), carbohydrates (by difference), titratable acidity (AOAC 942.15), and water activity (AOAC 978.18).

Microbiological analyses

Total aerobic plate counts (TPC), as well as levels of total lactic acid bacteria (LAB), Enterobacteriaceae, and pyschrotrophic bacteria, were determined by separately analyzing a single 25-g portion of ground beef from 4 of the 6 batches of non-inoculated ground beef (N = 4, n = 1) essentially as described (Jung et al., 2018; Porto-Fett et al., 2008). Samples were spread plated onto BHI agar (BD Difco) for TPC, De Man Rogosa Sharpe (MRS; BD Difco) agar for LAB, MacConkey (MAC; BD Difco) agar for Enterobacteriaceae, and Tryptic Soy (TSA; BD Difco) agar for psychrotrophic bacteria. The BHI and MAC agar plates were incubated for 24 h at 37°C, and the TSA agar plates were incubated for 7 d at 7°C, whereas MRS agar plates were incubated anaerobically (10% carbon dioxide, 5% hydrogen, and balance nitrogen; Model DG250; Don Whitley Scientific, West Yorkshire, United Kingdom) for 48 h at 37°C. Results were expressed as log₁₀ CFU per g. Samples (25 g; N = 2, n = 1) were also tested for the presence/absence of Salmonella spp. and the so-called “Big Six Non-O157” regulated serotypes of Shiga toxin-producing Escherichia coli (STEC), namely serotypes O26, O103, O45, O111, O121, and O145, by a commercial testing facility following methods AOAC 2004.03 and AOAC RI 091301, respectively, as detailed elsewhere (AOAC, 2023). The limit of detection for pathogen recovery was 0.04 log₁₀ CFU per g by enrichment.

Statistical analyses

Statistical analyses were performed using GraphPad Prism software version 8.4.3 (La Jolla, CA, USA). One-way analysis of variance followed by Tukey’s multiple comparison test (Keselman and Rogan, 1977) was used to analyze viral titration data to assess the effects of a single pH level across multiple time points at 2 temperatures (23°C and 37°C), as well as to assess the effects of a single temperature across 3 pH levels (pH 4.4, pH 5.0, and pH 5.8) at multiple time points. Adjusted P values from Tukey’s multiple comparison test were considered statistically significant at ≤0.05. Data are expressed as means ± standard error of the mean.

Microbial and proximate composition analyses

The microbial profile of the 4 batches of retail ground beef tested was as follows: TPC of 5.32 ± 1.6 log₁₀ CFU per g, LAB levels of 5.16 ± 1.7 log₁₀ CFU per g, Enterobacteriaceae levels of 4.18 ± 1.6 log₁₀ CFU per g, and psychrotrophic bacteria counts of 5.66 ± 1.7 log₁₀ CFU per g. These data agree with studies by other investigators who reported TPC ranging from 4.6-8.5 log₁₀ CFU per g (Djordjević et al., 2019; Ercolini et al., 2009; Jay et al., 2003; Pao and Ettinger, 2009), levels of LAB ranging from 3.2-5.1 log₁₀ CFU per g (Djordjević et al., 2019; Jay et al., 2003), levels of Enterobacteriaceae ranging from 3.2-3.9 log₁₀ CFU per g (Djordjević et al., 2019; Jay et al., 2003), and levels of psychrotrophic bacteria ranging from 6.0-7.4 log₁₀ CFU per g (Ercolini et al., 2009; Pao and Ettinger, 2009). Note, cells of Salmonella spp. or “Big Six Non-O157” STEC were not detected (≤0.04 log₁₀ CFU per g) within either of the 2 batches of commercial ground beef tested that were utilized in this study to prepare slurries.

The average values and ranges for select intrinsic analytes of retail ground beef and slurries as determined by a commercial testing facility were as follows: fat (18.3%; range from ca. 15.2–20.8%), protein (17.4%; range from ca. 16.4–18.3%), moisture (62.8%; range from ca. 61.1–65.0%), ash (0.79%; range from ca. 0.73–0.85%), carbohydrate (1.3%; range from ca. <0.1–1.9%), and titratable acidity (0.91%; range from 0.88-0.94%). The average fat content (18.3%) measured herein is typical for 85:15 ground beef available at retail, which reportedly ranges between 14-18% (Cook et al., 2013; Leheska et al., 2008; Purohit et al., 2022). The observed values for protein, ash, and moisture herein are also within the expected range for 85:15 retail ground beef as published elsewhere (range = 11.9–19.2% protein, 0.4–1.3% ash, and 64.7–70.4% moisture) (Leheska et al., 2008; Purohit et al., 2022). As expected, the fat content and protein levels were significantly lower for beef slurries compared to retail ground beef. The fat content of 10% slurries for all pH treatments was, on average, ca. 1.3%, with a range from ca. 0.5–2.6%. For each treatment, the fat content of the pH-adjusted slurries was, on average, as follows: pH 4.4 (1.5%; range from ca. 1.2–1.6%); pH 5.0 (1.4%; range from ca. 0.5–2.6%); and pH 5.8 (0.9%; range from ca. 0.5–1.4%). Likewise, protein levels of 10% slurries for all pH treatments were, on average, ca. 2.0%, with a range from ca. 0.7–3.3%. For each treatment, the protein levels of the slurries were, on average, as follows: pH 4.4 (2.5%; range from ca. 2.2–2.8%), pH 5.0 (1.7%; range from ca. 0.7–2.5%), and pH 5.8 (1.8%; range from ca. 0.7–3.3%). In general, however, the levels and types of the microbial flora and the chemical composition of the ground beef were within acceptable ranges and did not vary appreciably among batches. Thus, the microbial profile and intrinsic parameters of the meat used herein likely had little to no direct effect among trials on the viability of LPAIV under the food-relevant conditions that were evaluated for AIV-inoculated beef slurries.

Infectivity of LPAIV in beef slurries under food-relevant conditions

Significant reductions of strain rgGYR/14 were observed in slurries as assessed via embryo mortality in ECE and HA assays, especially for lower pH levels, longer incubation times, and the higher incubation temperature (Tables 1 and 2). Data on mortality of LPAIV via ECE assays attributed to challenge by pH and temperature in beef slurries mirrored results recorded for HA assays. Both embryo mortality and HA activity decreased over time, which may indicate a reduction in infectious/intact viral particles and/or functional hemagglutinin, respectively, that remained after exposure of strain rgGYR/14 to food-relevant conditions of pH and temperature in beef slurries. However, because embryo death can also occur for reasons unrelated to influenza virus, these outcomes are interpreted as supportive observations rather than as direct measures of infectivity. Since HA activity was not confirmed to be influenza-specific, results from HA assays are also presented only as descriptive reductions in RBC-agglutinating activity and not as evidence of virus presence or viability. In this study, the infectivity of LPAIV is defined solely by EID₅₀ titers.

For egg mortality results specifically, the number of embryos adversely affected (i.e., % mortality) decreased from time 0 through to 15 h for all 3 pH levels and 2 temperatures tested, presumably due to lower numbers of infectious/intact rgGYR/14 particles being present in slurries. However, these results should be interpreted with some consideration given the potential for non-viral causes of embryo death. For slurries incubated at 23°C, greater (P < 0.05) mortality (i.e., fewer embryos surviving) was observed for pH 4.4 and pH 5.0 slurries compared to pH 5.8 slurries (Table 1). In contrast, no differences (P > 0.05) in mortality were observed among any pH levels for slurries incubated at 37°C (Table 2). Overall and as expected, at time 0, there were no significant (P > 0.05) differences in egg mortality among the 3 pH treatments for a given temperature or between the 2 storage temperatures. Similarly, no differences (P > 0.05) in egg mortality were observed between slurries adjusted to pH 4.4 or pH 5.0 following storage at 23° or 37°C for 15 h. In contrast to the pH 5.8 slurries, differences were observed (P = 0.0275) after incubation for 15 h at 23°C compared to 37°C. Lastly, it should be stated that the pH of beef slurries adjusted to pH 4.4, 5.0, and 5.8 and held at 23° and 37°C remained relatively stable (≤0.5 pH unit decrease) over the 15–h incubation period (Data not shown).

Regarding agglutination assay results, HA activity was not observed in the negative control (i.e., noninoculated slurry adjusted to pH 4.4, neutralized to ca. pH 7.0, and sampled at 0 h) across 4 trials (0 of 16 embryo deaths; N = 4, n = 1 sample per each of 4 trials using a set of 4 SPF eggs per each sample). For samples incubated at 23°C, HA activity decreased from 81.8% (27 of 33 embryo deaths) at 0 h to 70% (14/20) after 15 h in pH 5.8 slurries. For slurries adjusted to pH 5.0 and pH 4.4 that were incubated at 23°C, HA activity decreased from 93.8% (30/32) to 40% (8/20) and from 100% (28/28) to 0% (0/11) from 0 to 15 h, respectively (Table 1). At 37°C, HA activity was reduced from 100% (28/28) to 0% (0/11) within 15 h for slurries adjusted to pH 4.4 and from 93.8% (30/32) to 4.8% (1/21) for slurries adjusted to pH 5.0, whereas for pH 5.8 slurries, HA activity decreased from 81.8% (27/33) to 11.8% (2/17) within 15 h. Although inferred, because HA activity was not confirmed as influenza-specific, these declines are interpreted as reductions in RBC-agglutinating activity and not as direct evidence of a decline in virus presence or infectivity.

As expected, RT-qPCR confirmed the presence of AIV RNA in all inoculated samples based on Ct values, albeit at lower levels after 15 h compared to 0 h. Baker et al. (2025) reported that elevated Ct values are indicative of lower RNA quantities; however, lower levels of AIV nucleic acid are not necessarily indicative of the infectivity of the virus per se, presumably due to some amount of viral RNA degradation rendering it innocuous, yet present/detectable by RT-qPCR. Herein, we also calculated the RT-qPCR-based quantity estimates for viral RNA from Ct values. For pH 5.8 and pH 5.0 slurries, there was a 1.0- to 1.3-log₁₀ EID₅₀ per mL reduction in RT-qPCR-based quantity estimates of viral RNA after 15 h at 37°C, compared to 23°C, which showed a <1-log₁₀ EID₅₀ per mL reduction after 15 h (Data not shown). Based on Ct values generated via RT-qPCR, for slurries adjusted to pH 4.4 there was a ca. 2.0-log₁₀ EID₅₀ per mL reduction at 37°C and a ca. 1.3-log₁₀ EID₅₀ per mL reduction at 23°C after 15 h (Data not shown). Regardless of the pH or temperature, the continued detection of viral RNA by RT-qPCR from 0 through 15 h of incubation despite a complete lack of strain rgGYR/14 infectivity as inferred from ECE and HA assays illustrates the limitations of PCR-based detection methods for assessing public health risks attributed to the presence of virus particles. Across numerous platforms, from infectious disease surveillance to foodborne pathogen risk assessment, nucleic acid-based detection methods often show little to no correlation with levels or presence of infectious AIV particles. Note, our results further highlight the critical need for viability-based assays to assess actual risk to the population from viral pathogens of concern, particularly following virus challenge in foods, or in response to food-relevant conditions, or because of food processing.

We also obtained viral titers of strain rgGYR/14 present in samples at the beginning and at the end of the 15-h incubation period. The initial titer of strain rgGYR/14 in 10% beef slurries, as determined by ECE mortality data using the EID₅₀ assay, was 4.47 ± 0.08 log₁₀ EID₅₀ per mL on average for all 3 pH treatments. After 15 h at pH 4.4, and regardless of the temperature, there was substantial inactivation of LPAIV (reductions of ca. ≥3.4 log₁₀ EID₅₀ per mL), with viral titers decreasing to below the detection limit (1.1 log₁₀ EID₅₀ per mL). For pH 5.0 and pH 5.8 slurries incubated at 23°C, reductions of 3.0 and 2.1 log₁₀ EID₅₀ per mL, respectively, were observed after 15 h. For pH 5.0 and pH 5.8 slurries incubated for 15 h at 37°C, however, reductions of 3.2 and ≥3.4 log₁₀ EID₅₀ per mL in levels of rgGYR/14 were observed, respectively. The greater negative impact on LPAIV observed at lower pH levels, as measured by EID₅₀ values, agrees, in general, with results published by Beato et al. (2012) wherein incubation at pH 5.0 reduced infectivity of both low pathogenic and highly pathogenic strains of AIV in spiked poultry meat between 1–2 d at 4°C and 20°C, respectively, compared to 2–5 d at pH 7.0. As one explanation, low pH likely induces conformational changes in viral hemagglutinin protein(s) that subsequently disrupt membrane integrity, thus exposing AIV RNA to environmental degradation, and ultimately resulting in loss of infectivity (Hamilton et al., 2012; Junankar et al., 1986; Lenz-Ajuh et al., 2025; Remeta et al., 2002; Sato et al., 1983). The consistency between our findings in beef slurries and those of Beato et al. (2012) for poultry meat underscores the efficacy of acidic conditions, such as those achieved by fermentation, to mitigate AIV risks in meat products. From a food safety perspective, the conditions herein elaborated to simulate key intrinsic parameters of fermentations delivered appreciable reductions in AIV infectivity; however, because the slurry model does not incorporate all variables of commercial fermented or dried meat processes (e.g., salt, curing agents, starter cultures, drying, or water activity changes), these findings should be interpreted only as preliminary insight into how AIV responds to acidic, fermentation-like conditions rather than as evidence of safety for finished products such as salami or pepperoni. Accordingly, any potential mitigation of LPAIV risk in actual fermented meats remains to be confirmed in product-based AIV challenge studies.

Our findings are consistent with previous AIV inactivation studies conducted in poultry, beef, dairy, and milk products, which generated foundational data and relevant comparisons for understanding viral infectivity and persistence in a variety of food matrices. However, the present work is a simplified, model system rather than a full replication of commercial fermentation processes. Regarding the former, Shahid et al. (2009) observed that a H5N1 isolate within amnio-allantoic fluid (AAF) harvested from inoculated ECE lost infectivity after 30 min at 56°C or within 24 h at 28°C, but remained viable for over 100 d at 4°C. These same authors also noted that exposure of the infected AAF to extreme pH levels (i.e., pH 1, pH 3, pH 11, or pH 13) was virucidal within 6 h, whereas infectivity persisted for over 24 h at pH 7.0 and at pH 9.0. Yamamoto et al. (2010) reported that levels of H5N1 within feathers derived from the body of infected ducks decreased by ca. 4.0 log₁₀ EID₅₀ per mL after 15 d at 20°C compared to a decrease of ca. 3.0 log₁₀ EID₅₀ per mL after 160 d at 4°C. As additional examples, Thomas et al. (2008) reported a 6.0-log₁₀ EID₅₀ per g reduction of H5N1 in poultry breast meat (harvested from infected birds) when heated to 70°C within 1 s in a thermocycler block, while Luchansky et al. (2024) reported a ≥5.6-log₁₀ EID₅₀ per 300 g reduction of LPAIV H5N1 spiked into ground beef patties (ca. 300 g each) when cooked on a gas grill to 71.1°C (come-up time = 24 min). Crossley et al. (2025) observed that acidification of whole raw milk with citric acid to pH 4.1 reduced levels of H6N2 LPAIV and H5N1 HPAIV by ca. 4.0 log₁₀ viral RNA copies per μL as measured by RT-qPCR, whereas complete loss of infectivity was achieved within 6 h during storage at ambient temperature (i.e., 20°C). Collectively, these data align with our results for 10% beef slurries adjusted to pH 4.4 and, in general, confirm the efficacy of acidification of protein-containing matrices to lower the risk of transmission of bird flu to humans. However, because our slurry model did not address all process parameters inherent to commercial processing of fermented meats, extrapolation of our findings to finished products should be made prudently. Although the incubation temperatures and target endpoint pH values evaluated herein were selected to approximate those used during fermentation of salami (23°C and pH 5.0–5.3) and pepperoni (37°C and pH 4.4–4.6), the slurry model does not capture the full complexity of these processes. Importantly, although several intrinsic and extrinsic parameters can influence AIV stability in meat systems, the primary determinants for fermented products are pH, temperature, and dwell time: such conditions were the focus of the present study. The data collected in our model system should be interpreted as mechanistic rather than as a validation of the fate of AIV in commercial fermented or dried meat products.

Although our findings provide new information on the fate of AIV in a model beef system, the present study has some minor shortcomings that may limit its broader applicability to related substrates and/or its extrapolation to other food-relevant conditions. More specifically, only one AIV strain (i.e., rgGYR/14) was tested, thus preventing insights into (the potential for) strain-to-strain variation. As addressed in a companion study on inactivation of strain rgGYR/14 in yogurt made from raw milk (Harrell et al., 2026a), however, given the absence of published data on differences related to thermal or acid stability among type A influenza virus, strain rgGYR/14 should be representative of such virus, and results obtained with this strain would be especially suitable and widely generalizable. Also, the present study evaluated a limited number and range of pH (i.e., pH 4.4, pH 5.0, and pH 5.8) and temperature (i.e., 23°C and 37°C) conditions. Additionally, the experimental duration of 15 h aligns with the lower end of fermentation times (e.g., 6–48 h) typical for some pepperoni and salami products, whereas complete processing of fermented meats, which may also include a drying step, often extends over several weeks (Feiner, 2016). Another potential limitation relates to the neutralization and sample handling steps used before virological analyses. Because slurry samples at different starting pH values required a single volume (4 mL) but varying concentrations of DE broth (4X, 2X, or 1X) to achieve a final ofpH 7.0 to obviate any dilution effects on LPAIV recovery, the resulting post-neutralization effects, if any, may have differed to a varying extent across treatments. These differences, along with the standardized holding period for ca. 1 h at room temperature and overnight storage at 4°C at ERRC before refrigerated transport of samples for analyses to CABL, could have influenced virus recovery in a treatment–dependent manner. Although DE broth has been widely used as a neutralizing medium in other influenza studies (Alphin et al., 2009; Lombardi et al., 2008) and was used to effectively neutralize acidic matrices in our own research on fate of AIV in dairy products made from raw milk (Harrell et al., 2026a), the specific effects of DE concentration and post–neutralization handling on AIV infectivity have not been fully investigated. Thus, future work should be conducted to evaluate slurry samples after neutralization to pH 7.0 to determine how DE broth strength/volume and subsequent holding conditions may affect virus recovery independent of the initial pH and temperature treatment.

Since beef slurries were spiked with AIV, this model system may not fully replicate the behavior of the virus present in naturally infected tissues, wherein factors such as tissue localization and immune components can influence viral stability (Guan et al., 2025). Such host-associated factors may impact viral load and distribution within the live animal and alter how the virus responds once this tissue becomes a protein source/substrate for either consumption or experimentation. A virus embedded within a specific tissue in naturally infected meat could exhibit different stability, protection, or inactivation attributes during subsequent acidification, fermentation, or other food processing conditions compared with virus added exogenously to homogenized slurries. To this end, recent work using milk spiked with H5N1 compared to milk collected from dairy cattle naturally infected with HPAIV demonstrated that treatment outcomes can differ between inoculated and naturally contaminated matrices (Guan et al., 2025), highlighting the need for astute analyses of all available data when extrapolating results from spiked systems. In the Guan et al. (2025) study, HPAIV (H5N1) present in raw milk from infected cows remained infectious for at least 5 weeks at 4°C and, in some cases, for up to 22 weeks, whereas the same, but laboratory passaged virus, when spiked into raw milk from healthy cows, became undetectable within 2-3 weeks. Similarly, during heat treatment, the virus in naturally infected milk showed greater thermal stability (half-life of 6 s for naturally infected milk vs 1 s for spiked milk), particularly at 63°C, than the same virus strain spiked into milk, despite equivalent starting titers. As reported (Guan et al., 2025), these differences were statistically significant and indicate that the virus produced within the mammary gland is more stable in milk than the same virus introduced exogenously into milk. With respect to the present study, in addition to potential differences in viability in response to pH and temperature in beef slurries, these differences may also influence LPAIV infectivity during overnight refrigerated storage at ERRC and subsequent refrigerated transport to CABL: virus associated with naturally infected tissues may or may not degrade or persist differently than the same virus spiked into slurries. Although the slurry model provides preliminary, controlled insight into AIV behavior under acidic conditions at food-relevant fermentation temperatures, it does not substitute for studies conducted in actual fermented or dried meat systems using AIV-infected meat/tissue harvested from contaminated animals. Therefore, further research should include experiments using naturally infected tissues or products when feasible to determine whether the virus present within intact tissues and host-derived components behaves differently from the laboratory-passaged virus subsequently spiked into slurries as detailed herein. Such studies are needed to assess whether the inactivation patterns observed in spiked or model systems accurately mirror those in naturally contaminated matrices. Further research is also needed to ground-truth findings generated in slurries with data collected in our ongoing studies involving fermentation and drying of raw beef that may be contaminated (or inoculated) with diverse AIV strains, including studies to quantify on the inactivation of highly pathogenic AIV strains directly in foods.