Turning up the heat: evaluating thermal treatments for phage management in dairy plants

Introduction

The dairy industry is a cornerstone of global food production, supplying a wide range of fermented and non-fermented products. The reliability of dairy fermentations, however, is continually threatened by microbial contaminants, among which bacteriophages (phages) represent the most persistent and economically damaging challenge (Garneau and Moineau, 2011; Mahony et al., 2012). In the dairy sector, phage infections are estimated to account for up to 10% of fermentation failures, primarily through the disruption of lactic acid bacteria (LAB) starter cultures (Doğan and Hinrichs, 2008; Moineau et al., 2002). These failures result in acidification delays, product downgrading or loss, and substantial financial penalties for producers (Fernández et al., 2017; Pujato, 2014; Szczepankowska et al., 2013).

Phages are ubiquitous in dairy production environments and can be detected in raw milk, whey, air, processing surfaces (including biofilms), and equipment. Their persistence is further exacerbated by modern manufacturing practices, particularly the widespread recycling of dairy by-products. In contemporary cheesemaking, whey components such as whey proteins and whey cream are frequently reincorporated into cheese milk to improve nutritional value and economic efficiency (Lawrence, 1993; Hinrichs, 2001). While advantageous from a processing standpoint, the reuse of native whey preparations substantially increases the risk of phage reintroduction into fermentation systems, with potentially severe consequences for subsequent production batches (Kleppen et al., 2011).

The industrial impact of phage contamination is further illustrated by representative case studies from large-scale dairy operations. Although detailed quantitative reports are rarely available in the published literature due to commercial sensitivity, such examples provide valuable insight into the scale and dynamics of phage-related events under real processing conditions.

In one example, a cheese plant processing approximately 2.31 million liters of milk per day across 60 fermentation vats experienced progressive phage accumulation during routine culture rotation. In the absence of a validated clean-in-place (CIP) strategy for phage control, low-level infections initiated early in the production day propagated during the first rotation cycle. This resulted in delayed acidification in 11 vats, affecting over 424,000 L of milk. The immediate financial loss exceeded €148,000 in downgraded milk alone, accompanied by quality defects in more than 42,000 kg of cheese. Reduced yields and price penalties generated an additional loss exceeding €17,000 for a single production day, excluding operating and remediation costs.

Another internal case study illustrating phage impact on fresh fermented technologies concerned phages found in a starch-based stabilizer (at minimal detectable levels) containing no milk components. These were produced on the same line as stabilizers containing milk proteins and were the cause of fermentation disturbances in the production of sour cream.

Such cases illustrate how even modest phage loads and seemingly innocent practices can rapidly escalate into large-scale economic events.

Thermal treatments have historically formed the backbone of microbial control strategies in dairy processing, with pasteurization representing the most established method. However, a growing body of evidence demonstrates that many LAB phages display exceptional thermal resilience. Phages isolated from whey powders and dried dairy ingredients are frequently able to survive standard pasteurization regimes and, in some cases, even more severe heat treatments (Guglielmotti et al., 2012; Wagner et al., 2017). The selective pressure exerted by repeated thermal exposure, combined with the extensive reuse of whey-derived ingredients, has likely contributed to the emergence and persistence of highly heat-resistant phage populations in modern dairy plants (Pujato, 2014).

An additional emerging factor under consideration concerns bacteriophages embedded within biofilms which can exhibit enhanced environmental persistence due to physical entrapment and protection by the extracellular polymeric substance (EPS) matrix.

Using fluorescence correlation spectroscopy, Briandet et al. (2008) demonstrated that biofilms can act as active phage reservoirs, in which viral particles penetrate the EPS but become partially immobilised through interactions with biomass and cell surfaces, thereby shielding phages from external stresses such as desiccation and sanitising treatments. Consistently, Agún et al. (2018) showed that bacteriophages targeting Staphylococcus aureus retained higher functional activity when associated with biofilms than in planktonic suspensions, with the biofilm architecture protecting phage particles from inactivation by chemical disinfectants. Together, these findings support the concept that biofilm entrapment enhances phage stability by buffering viral particles from harsh environmental conditions, a mechanism that likely contributes to the increased apparent thermostability of biofilm-associated phages relative to free virions.

To mitigate these risks, the dairy industry has adopted a multi-hurdle approach to phage control, combining thermal, chemical, physical, and biological interventions. Chemical disinfectants such as peracetic acid, sodium hypochlorite, and ethanol are widely implemented, although their efficacy remains highly variable and strongly dependent on phage type, organic load, and application conditions (Guglielmotti et al., 2012). Emerging technologies including photocatalysis (UV/TiO₂), high-pressure processing, and membrane filtration have also been investigated as complementary or alternative strategies for phage mitigation in air and liquid streams (Moroni, 2002; Bozoğlu, 2009; Michel et al., 2021). Despite these advances, no single disinfection method provides universal or fail-safe phage control, underscoring the need for integrated and context-specific management strategies. The use of thermal phage inactivation combined with very effective cleaning is still attractive to the dairy industry as the most natural method, safe for the employee and the consumer (minimal risk of residual allergens).

This review critically examines the current state of knowledge on thermal disinfection strategies used in dairy plants for phage management. Particular emphasis is placed on the effectiveness and limitations of heat-based treatments against dairy-relevant bacteriophages, as well as documented case studies demonstrating phage thermal adaptation. We further sought to consolidate source data over decades worth of studies in order to provide additional insights to inform the rational design of robust, evidence-based thermal control strategies for modern dairy processing environments.

Thermal treatments in dairy processing and their role in phage control

Raw milk is the primary source of phages. Thermal treatments are fundamental to dairy processing, serving to ensure microbiological safety, extend shelf life, and stabilize product quality. These treatments span a wide range of time–temperature combinations, from mild processes intended to reduce microbial loads while preserving the functional and sensory properties of milk, to more severe treatments designed to achieve commercial sterility. Although such processes are primarily optimized for bacterial inactivation and enzyme control, their effectiveness against bacteriophages is considerably more variable and often insufficient.

Pasteurization, introduced in the early 20th century as a public health measure, remains the most widely applied thermal treatment in the dairy industry. Conventional pasteurization regimens include amongst others low-temperature long-time (LTLT; 63 °C for 30 min–mainly as a preliminary milk treatment before further processing steps), high-temperature short-time (HTST; 72 °C for 15–20s mainly for cheese milk), and higher-heat shorter-time (HHST; approximately 89 °C–100 °C for less than 1 s or alternative schemes such as 88 °C–98 °C from 2 to even 8 min for yoghurt milk) processes. LTLT and HTST treatments reliably eliminate pathogenic bacteria and inactivate alkaline phosphatase with minimal impact on milk quality (Pak, 2025). However, numerous studies have demonstrated that many lactic acid bacteria (LAB) phages survive standard pasteurization conditions, highlighting a fundamental mismatch between bacterial and phage thermal sensitivities (Guglielmotti et al., 2012; Lewis and Deeth, 2008; Madera et al., 2004).

More intensive heat treatments are applied to products requiring extended shelf life. Ultra-pasteurization typically involves temperatures of ∼138 °C for at least 2 s and is widely used for cream and similar products stored under refrigeration for up to 90 days. Ultra-high temperature (UHT) processing, performed at 135 °C–140 °C for 2–5 s followed by aseptic packaging, produces shelf-stable products with storage lives exceeding 6 months. While these treatments may induce slight changes in flavor and color, nutrient losses are generally minimal. Despite their severity, even UHT-adjacent treatments cannot be assumed to eradicate phages in all matrices, particularly when high organic loads or protective components are present (Burton, 2012; Mandal et al., 2019).

Additional thermal processes are also employed in dairy operations. In-container sterilization (approximately 115 °C–120 °C for 20–30 min) achieves true sterility but causes pronounced sensory degradation and has largely been replaced by UHT processing. Thermization, a mild pre-treatment of raw milk (63 °C–65 °C for ∼15 s), is used to reduce spoilage microorganisms prior to storage or further processing but provides limited phage reduction.

Bacteriophages have been known to challenge dairy fermentations since they were identified in the 1930s through their infection of LAB starter cultures, causing acidification delays and fermentation failures (Whitehead and Cox, 1936). Although pasteurization reduces phage loads in raw milk, thermal treatments designed for bacterial control are frequently inadequate for consistent phage inactivation. Consequently, mid-20th-century dairy plants adopted multi-layered control strategies combining heat-based sanitation (hot water, steam, and CIP systems) with operational measures such as culture rotation, phage-resistant starter strains, and controlled raw milk and whey handling. Whey, recognized as a major phage reservoir, was often heat-treated prior to reuse, yet the persistence of heat-resistant phage populations limited the effectiveness of these approaches.

By the 1980s and 1990s, the International Dairy Federation recommended a treatment of 90 °C for 15 min as a conservative benchmark for phage inactivation (Capra, 2013). This condition became widely established as a practical “phage kill step” in laboratories and production facilities for sanitizing media, whey derivatives, and fermentation systems, and for diagnostic testing. Over time, however, it became evident that such standardized treatments do not universally ensure phage control, particularly against thermoresistant strains.

The limitations of thermal phage control have been further amplified by modern dairy practices involving whey recycling and microparticulation. These processes combine heat treatment with high shear to convert whey protein concentrates into functional ingredients, such as protein-based fat replacers, or to reincorporate whey components into cheese milk for economic and nutritional optimization. Industrial microparticulation systems operate under a range of thermal conditions that may not guarantee phage inactivation if inadequately validated. Whey cream fractions with fat contents of 25%–30% are commonly reused for milk standardization, for which heat treatments of at least 93 °C for 30 min have been suggested (Marcó, 2009). In this context, thermal inactivation studies conducted specifically in whey matrices are essential, as inadequate heating of recycled whey streams can promote the accumulation and dissemination of highly concentrated, heat-resistant phage populations within dairy plants and finished products.

To minimize the risk of fermentation failure associated with whey recycling, it has been proposed that thermal treatments should be designed to achieve at least a 9-log reduction in phage titer, particularly in environments where thermoresistant phages are prevalent. Accordingly, thermal inactivation curves and corresponding 9-log reduction thresholds have been calculated and proposed to guide the design of robust heat treatment processes for whey and whey-derived ingredients (Atamer and Hinrichs, 2010). These findings underscore the necessity of tailoring thermal strategies specifically for phage control, rather than relying solely on bacteriological safety criteria.

Data driven analysis of phage thermostability study

The compiled phage thermostability dataset encompassed more than 1,000 individual heat-inactivation observations, representing unique combinations of phage, bacterial host, temperature, exposure time, and suspension medium. These observations were drawn from over 20 independent studies published between 1954 and 2022 that were found following extensive searches of PubMed, Google Scholar, Web of Science, and Semantic Scholar using keyword combinations such as “bacteriophage”, “thermal resistance”, “dairy phage” etc., and consulting the reference lists of pertinent articles. Data was manually extracted from the text, tables, or from figures if no other source was provided.

The resulting dataset comprises independent observations collected across diverse experimental systems and is intended to provide a representative, though not exhaustive, overview of phage thermal inactivation behavior. Efforts were made to minimize selection bias by including studies spanning multiple decades, phage types, and experimental conditions.

Seven bacterial host taxa were represented. Lactococcus lactis (including subspecies lactis, cremoris, and diacetylactis) accounted for the majority of records, followed by Streptococcus thermophilus, Leuconostoc spp (including L. pseudomesenteroides), and several Lactobacillus species (L. plantarum, L. casei, L. paracasei, L. delbrueckii subsp. Bulgaricus, and L. helveticus). Experiments were conducted in an out-of-host context in order to specifically quantify viral particle decay under conditions in the absence of the host strain.

The dataset included phages from several taxonomic groups but was mostly dominated by Skunavirus, Ceduovirus, P335-type phages, and moineauviruses. Many phages, particularly some infecting S. thermophilus, Lactobacillus, and Leuconostoc had no typing information that could be retrieved and were classified as not specified (NS). In several cases, data was only available in the form of graphical figures and best attempts were made to digitize this manually. The raw data collected for the study is provided in Supplementary Table S1 along with all of the source references also listed in this table (Atamer et al., 2011; Chen et al., 2018; Chopin, 1980; Ebrecht et al., 2010; García-Anaya et al., 2020b; Jończyk, 2011; Koka and Mikolajcik, 1970; Marvig et al., 2011; Mercanti et al., 2012; Müller-Merbach et al., 2005a; Müller-Merbach et al., 2005b; Murphy et al., 2014; Nichols and Wolf, 1945; Quiberoni et al., 1999; Şanlibaba et al., 2009; Suárez and Jorge, 2002). A high level summary of common thermal treatments and outcomes are provided in Table 1.

TABLE 1. Overview of common heat treatments and their efficiency in phage reduction.

Thermal inactivation patterns across temperature ranges

Across the entire dataset, thermal inactivation exhibited a sharply non-linear relationship with temperature (Figure 1; Supplementary Table S1). At lower temperatures (≤65 °C), even extended exposure times frequently resulted in negligible inactivation. At intermediate temperatures (70 °C–85 °C), outcomes were highly heterogeneous and strongly dependent on phage identity and medium composition. At high temperatures (≥90 °C), most phages were rapidly eliminated; however, a small subset of skunaviruses displayed extraordinary resistance, surviving prolonged exposure even near boiling temperatures. Globally, the data indicates that an extended treatment of 90 °C for at least 40 min would be required to adequately eliminate phages.

Early kinetic descriptions of phage heat inactivation in lactic streptococci date back to the mid-20th century (Wilkowske, 1954; Daoust, 1965; Zottola and Marth, 1966). Thermal survival curves of dairy bacteriophages consistently deviate from simple first-order (log-linear) kinetics, displaying instead a characteristic biphasic pattern: an initial rapid decline in infectivity followed by pronounced tailing of persistent survivors (Figure 1). This non-linearity has been documented across phages infecting Lactococcus lactis, Streptococcus thermophilus, and Lactobacillus spp., indicating it is a general phenomenon of dairy phage thermal inactivation rather than a species- or group-specific anomaly (Guglielmotti et al., 2012; Marcó, 2019). It is important to note however that dedicated kinetics studies remain limited. Mechanistically, two explanations have been advanced. First, intrinsic population heterogeneity: a phage suspension comprises subpopulations of differing inherent heat resistance, each following approximate first-order kinetics but with distinct rate constants (Hiatt, 1964). The majority of virions are heat-labile and destroyed during the steep initial phase; a minority of heat-tolerant particles persist, producing the slower terminal slope. Heat-resistant phages have been shown to occur naturally in all types of dairy environments as a low proportion of the total phage population (Atamer, 2009). Second, physical clumping of virions can shield interior particles from heat, yielding apparent tailing even in an otherwise homogeneous population. Both mechanisms predict survival curves that cannot be described by a single exponential, and they are not mutually exclusive (Guglielmotti et al., 2012; Marcó, 2019).

Concrete examples illustrate the practical magnitude of this effect. For L. lactis Ceduovirus P001, the T₉₉ at 69 °C in filtered whey is around 2.5 min, yet viable phage particles remained detectable after 30 min of continuous treatment (Figure 1). The Lb. delbrueckii phage Ib 3 presents another case: its T₉₉ at 72 °C is only 2.9 min, but complete inactivation required more than 45 min. For the highly thermoresistant skunavirus P680, heating at 90 °C for 15 min in whey achieved only a 6-log reduction from an initial titre of approximately 10⁹ PFU/mL, with infective particles still detectable in each medium tested (Supplementary Table S1). These observations confirm that the T₉₉ parameter captures only the behaviour of the most sensitive 99% of the population and does not reflect the time required for complete elimination; the minimum time for total inactivation may be substantially longer, and ideally both parameters should be determined for any phage of interest.

Global observations of heat inactivation separated by host species and phage type reveal distinct patterns. For one, host thermophily is not indicative of thermal resistance in corresponding phages, with S. thermophilus phages showing non-remarkable heat resistance (Figure 2). Thermal inactivation also shows clear taxonomic patterns at the phage level with ceduoviruses occupying a relatively narrow window between 60 °C–80 °C and skunaviruses found across a large temperature window with many high temperature resistant representatives (Figure 3). Indeed, this group dominates the 90 °C temperature bracket along with some not specified representatives which could very likely belong to the skunavirus group. The following sections detail specific observations across different temperature brackets.

Low-temperature heat treatments (55 °C–65 °C)

Low-temperature, long-time (LTLT) treatments frequently failed to eliminate phages. For example, L. lactis Skunavirus bIL170 exhibited no detectable loss in infectivity after 60 °C for 30 min in whey-based medium (10⁶ PFU/mL no change), and Ceduovirus c2 showed identical resistance under the same conditions. In contrast, P335-type phages were markedly more sensitive: phage P335 was completely inactivated at 60 °C within 20 min (10⁶ to 0 PFU/mL), demonstrating some dependence of thermostability on phage taxonomy even at mild temperatures (Figure 3).

Limited inactivation was also observed for thermophilic and mesophilic lactobacilli. L. plantarum phages treated at 63 °C for 7.6–9.9 min in MRS broth, RSM, or TMG buffer showed only ∼2 log₁₀ reductions (10⁸ to 10⁶ PFU/mL). Similarly, S. thermophilus phages displayed limited sensitivity at this temperature: phage 021–4 required 5 min at 63 °C in ETS broth to achieve a 2 log₁₀ reduction (10⁶ to 10⁴ PFU/mL), while treatment durations up to 45 min yielded no further reduction in several isolates. These findings collectively demonstrate as has been previously reported on multiple occasions that temperatures ≤63 °C are insufficient for reliable control of dairy phages.

The potential temperature dependence observed across phage taxa raises an interesting question regarding phage ecology within dairy processing environments subjected to defined thermal treatments. Thermal processing conditions may contribute, among other factors, to the high prevalence of skunaviruses reported in some dairy facilities, particularly considering the relative sensitivity of other phage groups.

Moderate temperatures (70 °C–85 °C)

Treatments between 70 °C and 85 °C produced the greatest variability in outcomes. At 72 °C for 15 s (HTST-equivalent conditions), lactococcal phages exhibited heterogeneous responses (Supplementary Table S1; Figure 2). Some isolates were fully inactivated from initial titers of ∼10⁵–10⁶ PFU/mL, whereas others retained substantial viability. For example, phage 38 decreased from 6.8 × 10⁵ to 3.4 × 10³ PFU/mL (2.3 log₁₀ reduction), while phage 170 (L. lactis subsp. diacetylactis) showed almost no reduction (3.5 × 10⁵ to 3.0 × 10⁵ PFU/mL; 0.07 log₁₀). Comparable resistance patterns have been reported for Lactobacillus and Streptococcus phages (Binetti and Reinheimer, 2000; Quiberoni, 2003; Chen et al., 2017).

Small temperature increases within this range often caused dramatic shifts in outcome. For phage 170, increasing the temperature from 72 °C to 80 °C for 15 s reduced titers from 2.6 × 10⁵ to 15 PFU/mL (4.2 log₁₀ reduction), illustrating a steep thermal transition zone.

At 80 °C for 5 min, Atamer et al. (2013) reported outcomes ranging from complete inactivation to full survival among L. lactis phages. The most resistant isolate, Skunavirus P680, showed no detectable reduction (5 × 10⁷ PFU/mL no change), while other skunaviruses and multiple ceduoviruses were fully inactivated under identical conditions. This intra-group heterogeneity highlights inadequacy of species- or group-level assumptions.

Leuconostoc phages exhibited one of the most striking thermal discontinuities observed in the dataset. In milk at 80 °C for 1 min, multiple isolates (P758, P787, P820, P821) showed no detectable reduction (0 log₁₀). However, increasing the temperature by just 5 °C (85 °C for 1 min) resulted in catastrophic inactivation: P758 decreased from 5 × 10⁷ to 50 PFU/mL (≈5.7 log₁₀ reduction), P820 decreased from 10⁸ to 100 PFU/mL (6.0 log₁₀), and the majority of more than 60 isolates yielded no detectable survivors. This narrow thermal “cliff” highlights the sensitivity of Leuconostoc phages to small temperature increments.

High temperatures (≥90 °C)

At 90 °C, most phages across all host species were completely inactivated within 5 min. All S. thermophilus phages tested at 90 °C for 5 min yielded 0 PFU/mL regardless of medium (ETS broth, RSM, or TMG buffer). L. plantarum phages treated at 90 °C for 5 min showed either no detectable survivors or ≤10 PFU/mL from initial titers of 10⁷–10⁸ PFU/mL (≥6 log₁₀ reduction). Similar outcomes were observed for L. casei and L. paracasei phages. Notably phage lb3 infecting L. delbruekii required 45 min at 72 °C or 15 min at 90 °C for inactivation (Quiberoni et al., 2003).

Most L. lactis skunaviruses were likewise fully inactivated at 90 °C for 5 min (e.g., CB13, CB14, CB19, CB20, p2, SL4: 10⁹ to 0 PFU/mL). However, two skunaviruses, P680 and P1532, exhibited extraordinary thermal resistance and represented the most resistant phages in the dataset.

P680 retained 5 × 10⁶ PFU/mL after 90 °C for 5 min in SM (10⁷ to 5 × 10⁶; 0.3 log₁₀ reduction) and still showed 10⁴ PFU/mL after 15 min at the same temperature. Complete inactivation required 40 min at 90 °C. In filtered whey at 85 °C, P680 survived for up to 360 min, with titers declining gradually from 7.3 × 10⁸ to 463 PFU/mL (∼6.2 log₁₀ reduction over 6 h). It is well acknowledged that this phage exists as a benchmark for thermal treatment efficacy (Atamer, 2022).

P1532 proved even more resistant. At 90 °C for 5 min in LM17, titers decreased only from 10⁹ to 8 × 10⁷ PFU/mL (1.1 log₁₀ reduction). After 15 min at 90 °C in SM, P1532 retained 10⁵ PFU/mL, and complete elimination similarly required 40 min. Remarkably, P1532 retained measurable infectivity after treatment at 95 °C–97.5 °C in GM17, lactoferrin, and β-lactoglobulin, with survivors ranging from 10⁴ to 10⁶ PFU/mL from initial titers of 10⁸ PFU/mL. No other phage in the dataset exhibited comparable survival at these temperatures, identifying P1532 as the most thermally resistant isolate.

Impact of the food matrix on thermal inactivation of dairy phages

The efficacy of thermal treatments against bacteriophages is strongly governed by the composition and physicochemical properties of the surrounding food matrix (Figure 3). For a given phage, thermal tolerance can vary by several orders of magnitude depending on whether inactivation occurs in water, buffer, milk, whey, or high–total solids dairy products. Compared with aqueous systems or simple salt solutions, milk and other protein- and fat-rich dairy matrices consistently confer pronounced protection to phages during heat treatment. Identical time–temperature conditions that result in substantial lethality in water often yield minimal reductions in milk. For example, a Leuconostoc phage exhibited an approximately 4-log₁₀ reduction at 70 °C for 1 min in water, but only ∼1-log₁₀ under the same conditions in milk, corresponding to a matrix-derived protection factor of roughly three orders of magnitude. Numerous studies confirm that pasteurization-level treatments effective in buffers systematically underperform in milk matrices (Atamer et al., 2010; Wagner, 2017).

This pronounced matrix effect is reflected in highly variable thermal inactivation kinetics among dairy phages. Reported times to achieve a 99% reduction (t₉₉) at 72 °C span several orders of magnitude depending on both phage identity and matrix, ranging from minutes to several hours. Under milk-based conditions, t₉₉ values approaching 300 min have been reported for particularly thermotolerant phages, highlighting that a single temperature setpoint cannot be assumed to deliver consistent lethality across matrices. These findings highlight the danger of extrapolating thermal resistance data obtained in simplified systems to real dairy applications.

Matrix-associated protection extends beyond milk to a broad range of protein-rich and high-total-solids dairy systems. Inactivation kinetics measured in whey, whey protein concentrates (WPC), and whey cream frequently resemble those observed in milk, implicating proteinaceous components and total solids as dominant stabilizing factors. Quantitatively, protein- and fat-rich dairy matrices often confer 2–6 log₁₀ higher phage survival compared with simple solute or salt solutions under identical thermal challenges, with the magnitude of protection increasing at higher temperatures (Figure 3). At 70 °C for 10 min, phage P008 exhibited a ∼7 log₁₀ reduction in CaCl₂, concentrated NaCl, or dilute whey protein solutions, but only a ∼3 log₁₀ reduction in 2.8% casein. At 85 °C for 10 min, the protective hierarchy became even more pronounced: all dairy matrices tested (skim milk, cream, casein, whey protein, permeates, and retentates) limited inactivation of phage P680 to ∼1 log₁₀, whereas simple solutes yielded complete (>7 log₁₀) inactivation. Even after extended heating at 85 °C for 30 min, survivors persisted at 10⁵–10⁶ PFU/mL in dairy matrices while remaining at detection limits in non-protective media (Supplementary Table S1).

Heat-resistant lactococcal phages, among the most problematic in industrial fermentations, remain particularly difficult to inactivate across these matrices. For example, thermal D-values of 76.8 min and 277 min have been reported for phage P680 in milk and filtered whey respectively at 75 °C (Michel et al., 2021; Wagner, 2017).

The challenge is further exacerbated in dried systems: phages isolated from whey powders are often among the most thermotolerant reported (Capra, 2013; Wagner, 2018; Sadiq, 2019). In some cases, no measurable reduction in phage titer has been observed after approximately 9 months of storage in milk powder, demonstrating remarkable stability in low-moisture environments. These “powder-selected” phage populations can persist through reconstitution and subsequent heating steps, representing worst-case scenarios for downstream thermal control.

Fat content and microstructure introduce additional layers of protection for microbial inactivation (Verheyen et al., 2020; Liu et al., 2025). Although detailed mechanistic data remain limited, emulsified fat droplets and milk fat globule membrane interfaces are regarded as potentially providing physical shielding to phage virions. Fat globules may also promote virion aggregation, resulting in increased apparent D-values and non-linear inactivation kinetics characterized by shoulders and persistent tails. Consistent with this interpretation, homogenization which reduces fat globule size and disrupts interfacial structures has been shown to diminish fat-associated protection, indicating a direct relationship between fat microstructure and phage heat resistance. That being said, studies show that proteins seem to provide the dominant protective effect against thermal treatment for LAB phages (Marcó, 2019). Interestingly, milk components in particular fat globules, can negatively impact phage stability in some cases, providing a non-significant barrier in the application of phages to target spoilage organisms (Garcia-Anaya et al., 2020a).

Mineral composition and pH further modulate thermal stability and must be carefully controlled when interpreting inactivation data. Calcium ions (Ca²⁺), commonly added to media to enhance plaque formation in phage assays, are well known to stabilize many tailed phages and can inadvertently increase apparent thermal tolerance. Similarly, the near-neutral pH of milk provides conditions favorable to phage stability, whereas acidification toward pH 4–5 imposes an additional stress that may enhance thermal inactivation. This distinction is particularly relevant when comparing sweet whey (typically pH > 6.3) from ripened cheese manufacture with sour whey (pH < 4.6) drained from fresh cheese processes, where differences in acidity and protein composition can yield markedly different thermal outcomes. Interestingly, lactoferrin in whey has been demonstrated to have a pH dependent protective effect on phage P1532 (Geagea and Ahmed, 2017).

Beyond matrix composition, effective thermal exposure is also influenced by the thermal properties of processing equipment. Many studies report lethality based on nominal hold temperatures without accounting for temperature ramp-up dynamics. However, vessel material can significantly affect heating rates: stainless steel containers have been shown to reach target temperatures at least twice as fast as glass vessels at 90 °C, resulting in greater overall lethality despite identical setpoints. Neglecting such factors may lead to systematic underestimation of process variability and control risk (Capra, 2004). This factor also renders comparative analyses of thermostability studies challenging.

A final perspective that is important to consider is that of the impact of any given treatment on the starter culture. Inadequate thermal processing of whey may leave viable host cells that serve as incubators for a domino effect on the phage population despite what may have been a seemingly effective treatment.

Collectively, these matrix-related effects frequently result in deviations from classical log-linear inactivation kinetics. Thermal treatments in complex dairy systems often display an initial shoulder followed by a pronounced tail of persistent survivors. In such cases, Weibull-type models generally provide a superior empirical description compared with first-order kinetics by more accurately capturing non-linear behavior. Taken together, matrix composition, fat structure, mineral content, pH, total solids, and thermal delivery all play decisive roles in determining phage survival during heat treatment. Failure to account for these factors risks creating sub-lethally treated niches that facilitate phage persistence and adaptation, underscoring the necessity of matrix-aware strategies for robust, industrially relevant thermal phage control.

Mechanisms of viral adaptation to thermal stress

Bacteriophage thermotolerance in dairy and industrial settings arises from a convergent set of structural, genomic, and biochemical adaptations, shaped by both natural and process-driven selection pressures. Recent investigations have identified specific genetic determinants underlying these adaptations, providing a mechanistic foundation for predicting and managing heat resistance in industrially relevant phages (Figure 4).

Structural protein mutations constitute the best-characterised determinants of phage heat resistance. Among Lactococcus lactis phages of the skunavirus group the tape measure protein (TMP) has been directly implicated in thermal stability (Geagea and Labrie, 2018). Comparative genomic analysis revealed that the highly thermoresistant phages P1532 and P680, which remain infectious after exposure to 97 °C for 5 min and 95 °C for 5 min respectively, carry truncated TMP genes encoding a protein of 916 amino acids, compared with 999 amino acids in heat-sensitive phages p2 and CB14. Laboratory selection of phage CB14 under sequential heat challenges independently produced the same 120-bp in-frame deletion in the tmp gene, removing 40 amino acids from the protein (Geagea and Labrie, 2018). The resulting mutants (CB14a and CB14b) exhibited markedly improved heat survival at 70 °C for 5 min: wild-type CB14 suffered a 1.6 log₁₀ titer reduction, whereas CB14a lost only 0.7 log₁₀ and CB14b only 0.4 log₁₀, with no detectable cost to infectivity (titers ≈10¹⁰ PFU ·ml^-1) or plaque size (≈2.5 mm). Critically, all three phages were completely inactivated at 80 °C for 5 min (<10 PFU·ml^-1), and TMP truncation alone did not confer the extreme thermotolerance of P1532 or P680, indicating that additional structural proteins contribute to the full phenotype. Additional 8 sequence diversity was observed in TMP residues 370–520 and in a 3-amino-acid deletion at position 521, both of which correlate with heat resistance across the skunavirus group (Geagea and Labrie, 2018).

Beyond the tail, experimental evolution of Escherichia coli phage T7 under escalating heat stress demonstrated that point mutations in the tail connector protein (gene 8) and DNA injection apparatus genes (genes 14–16) were the primary targets of thermal adaptation, collectively raising virion survival by more than 10 °C across six independent populations (Cole et al., 2021). The repeated, parallel fixation of specific gene 8 substitutions across independent lineages implicates tail-core stability as a critical bottleneck for phage viability at elevated temperatures.

Genome-level features further modulate thermal resilience. Elevated G + C content strengthens base-pairing and raises the DNA melting temperature. Thermus thermophilus phage P23-77, with a genome G + C content exceeding 66%, possesses three major coat proteins (VP11, VP16, VP17) that remain correctly folded above 80 °C, an unusually thick capsid wall (∼6 nm), and a host with an optimal growth temperature of 70 °C–75 °C (Kheirvari and Tumban, 2025). An extreme DNA-protection strategy is illustrated by the archaeal virus SIRV2 (Sulfolobus islandicus rod-shaped virus 2), which encapsidates its dsDNA genome entirely in the A-form helix, a compact conformation more resistant to thermal denaturation than canonical B-form DNA, enabling the virion to remain infectious at 70 °C–80 °C and pH 3 (DiMaio et al., 2015). Although SIRV2 is technically an archaeal virus, this mechanism exemplifies a general principle relevant to food microbiology: tight genome packaging and intimate DNA–protein association can protect nucleic acid integrity under heat stress.

During capsid assembly and maturation, certain phages introduce covalent modifications that irreversibly harden the particle. The most thoroughly characterised example is coliphage HK97, whose 385-residue capsid subunits (gp5*) undergo autocatalytic isopeptide bond formation between Lys169 and Asn356, creating an interlocking “chainmail” topology (Ross et al., 2005). Differential scanning calorimetry demonstrated that crosslinking raises the capsid denaturation temperature from 86.3 °C (Prohead II) to 96.2 °C (Head II)—a gain of approximately 11 °C—and nearly doubles the calorimetric enthalpy from 165 to 278 kcal·mol^-1 (Ross et al., 2005). A K169Y mutant incapable of crosslinking showed no thermal stabilisation upon expansion (Tp = 84.5 °C), confirming that crosslinking, not expansion alone, is the essential stabilisation step and is required for HK97 viability (Ross et al., 2005). By contrast, phage T4 relies primarily on capsid expansion for stabilisation, reinforced by the dispensable accessory protein Soc, whereas other phages (T7, P22) lack accessory proteins entirely (Ross et al., 2005). These contrasting maturation strategies illustrate that no single stabilisation mechanism is universal: the relative contribution of covalent crosslinks, expansion, and accessory proteins varies across phage families. Importantly, genetically encoded thermostability must be distinguished from reversible, matrix-dependent heat protection: phage survival is consistently higher when heated in milk or whey than in buffer, owing to the physical shielding conferred by dairy proteins and fats. Such media-dependent protection does not reflect underlying genetic adaptation and can mislead assessments of intrinsic thermostability (Atamer, 2010).

The dairy processing environment functions as a potent selective crucible for thermotolerant phages. Pasteurisation regimes combined with the practice of recycling heat-treated whey and whey protein concentrates (WPC) into fresh cheese milk, subject surviving phages to repeated thermal challenges (Geagea and Labrie, 2018). Over successive cycles, this enriches for genotypes carrying any thermal advantage, a process that has driven the emergence of phages like P680 and P1532 in commercial cheese operations (Geagea and Labrie, 2018). Indeed, as we previously mentioned, the relative abundance of specific phage groups with intrinsic thermal resistance could be attributable in part to the nature of applied thermal treatments.

Controlled evolutionary experiments corroborate the field observations. Serial heat selection of T7 yielded populations with heritable heat tolerance increases exceeding 10 °C (Cole et al., 2021), while laboratory selection of CB14 produced TMP deletions identical to those found in wild dairy isolates, establishing a direct genotype–phenotype link validated both experimentally and observationally (Geagea and Labrie, 2018). Screening for tmp deletions in newly isolated skunavirus phages has been proposed as a practical early indicator of emerging thermoresistance in cheese factories (Geagea and Labrie, 2018). Together, these findings underscore the multifactorial nature of phage thermotolerance spanning structural gene mutations, genome composition, maturation chemistry, and environmentally driven selection and highlight the value of integrating structural virology with applied dairy phage management to anticipate and counter emerging heat-resistant phage threats.

Perspectives

The compiled dataset confirms that thermal inactivation of dairy bacteriophages is governed by a combination of intrinsic viral properties and extrinsic process variables, leading to strongly non-linear and often discontinuous survival behaviour. Across phage groups and matrices, thermal survival frequently exhibited sharp “cliff” effects, whereby modest increases in temperature resulted in catastrophic loss of infectivity. These observations are consistent with the metastable nature of tailed phage particles, which depend on a finely balanced network of capsid–tail interactions and the maintenance of high internal DNA pressure. Experimental work on bacteriophage λ has demonstrated that heat-induced genome ejection can occur prior to capsid denaturation, providing a mechanistic explanation for abrupt inactivation thresholds that precede overt structural collapse (Qui, 2012). DNA ejection was also observed for phage P008 under heat treatment with the heat resistance in P680 partly attributed to increased stability of DNA packaging (Atamer, 2010).

Mutations that subtly alter DNA packaging pressure or tail-core stability, analogous to those described in thermophilic or laboratory-selected phages, are therefore likely contributors to the extreme thermotolerance observed in certain dairy isolates.

Phage taxonomy emerged as the strongest predictor of thermal resistance at the population level, with lactococcal skunaviruses consistently ranking above ceduoviruses and P335-type phages. However, variability within the skunavirus group spanned several orders of magnitude, clearly demonstrating that taxonomic assignment alone is insufficient to infer process safety. This highlights the need for integrated studies combining large-scale thermal phenotyping with genome sequencing to resolve within-group diversity and identify genetic determinants of thermal resistance. However, genomic data are currently unavailable for a substantial proportion of the phages included in this study, precluding phylogenetically informed analyses of this relationship.

Notably, host physiology was not predictive of phage behavior: despite the thermophilic nature of Streptococcus thermophilus, its phages were uniformly eliminated by treatments that failed to fully inactivate several Lactococcus lactis phages. These findings reinforce the necessity of isolate-specific assessment, particularly in high-risk matrices such as whey, whey protein concentrates, and fat-rich streams.

Taken together, the data indicate that thermal phage control in dairy systems cannot be reduced to fixed “kill steps” defined by nominal time–temperature combinations. Instead, effective control must be conceptualised as a process risk envelope defined by the interaction between temperature, residence time, matrix composition, heating dynamics, and phage population structure. Reliance on single-point validation (e.g., 90 °C for 15 min) risks creating sub-lethal niches that permit the survival and progressive enrichment of thermotolerant variants. In this context, dairy plants function not only as sites of phage exposure but as long-term evolutionary environments in which repeated thermal challenges, combined with whey recycling practices, exert strong directional selection on phage populations.

The heterogeneity observed across the compiled studies further highlights the urgent need for harmonised, phage-aware validation and reporting standards. Differences in initial titre, detection limits, heating geometry, vessel material, and temperature ramp rates markedly influence apparent inactivation kinetics and complicate cross-study comparison. While starting titre did not significantly affect log reduction across the global dataset (Pearson r = −0.04, p = 0.25, R ² = 0.002), lower titres were associated with a higher probability of complete inactivation (Welch t-test, p < 0.01), underscoring the importance of conservative experimental design and transparent reporting. Future work should prioritise time-to-total-inactivation metrics alongside conventional T₉₉ values and explicitly account for non-log-linear behaviour through appropriate kinetic models. An additional point here can also be made regarding the out-of-host nature of such experiments. In reality, thermal treatments are applied in a context where host cells may be present. This could change the dynamic somewhat, for example, one could speculate that adsorption to a host cell could provide a shielding effect and result in productive infection that would inflate final titres compared to ideal experimental conditions. This would be even more so the case regarding the presence of host cells within biofilms. One should always therefore be prudent in translating experimental outcomes to the real world.

From a practical standpoint, the growing understanding of genetic and structural determinants of phage thermotolerance opens new avenues for proactive control strategies. Genomic features such as truncations in the tape-measure protein and mutations affecting tail-core or capsid stability represent potential molecular markers for early detection of emerging thermoresistant phage populations. Integrating such surveillance approaches with thermal validation and upstream barriers such as membrane filtration, pH management, or selective whey treatment, offers a more robust alternative to further intensification of heat alone. Thermal control strategies will continue to have their place in many dairy facilities but are insufficient as a standalone safeguard in milk and whey matrices, where sublethal niches can promote long-term phage persistence and adaptive evolution (Polyanskaya and Prosekov, 2025). A further point that merits study is that of the potential translational nature of thermoresistant adaptations. It is unclear to what extent some routes to thermotolerance also provide heightened resistance to chemical treatments and vice versa, necessitating further care in avoiding sub-optimal application. It has been observed that some LAB phages that survive pasturisation are also less sensitive to disinfectants but this is not conclusive of an interdependence (Marcó, 2019). Interestingly, it is known that certain capsid mutations in coxsackieviruses render them thermotolerant but with no impact with respect to chlorine-based disinfectants (Torii et al., 2024).

Conclusion

Thermal phage control in dairy processing is inherently multifactorial and dynamic. While heat treatments remain a cornerstone of microbial control, the evidence compiled here demonstrates that they are not sufficient as standalone safeguards against bacteriophage persistence, particularly in complex dairy matrices where non-linear kinetics, matrix protection, and population heterogeneity enable survival and adaptation. Effective phage management therefore requires moving beyond static thermal “kill steps” toward integrated strategies that combine thermal treatments with complementary physical and chemical interventions. Within such multi-hurdle frameworks, heat should be considered as one component of a broader, structured control system that incorporates matrix-aware validation, process engineering, and upstream preventive measures. By adopting this more holistic approach, the dairy industry can better mitigate fermentation failure risks while limiting the selection and propagation of increasingly thermoresistant phage populations.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frfst.2026.1894844/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Kuntal Ghosh, Midnapore City College, India

Reviewed by: Xiao-Lin Chu, Beijing Normal University, China

Fatmawati Lambuk, University Sains Malaysia, Malaysia