Effect of Holder pasteurization and UV-C irradiation on bacteriophage titres in human milk

Abstract Human milk is the optimal nutrition source for infants and contains a complex mix of bioactive compounds and microorganisms. When unavailable, pasteurized donor milk may be provided, particularly to preterm infants. Holder pasteurization (HP) is typically implemented in human milk banks to prevent pathogen transmission. Given the impact of heat on milk bioactives, ultraviolet-C irradiation (UV-C) is an alternative being explored and has demonstrated effective bactericidal activity. In addition to bacteria, milk contains viruses, including primarily bacteriophages (phages) and which likely influence the developing bacterial microbiome of infants. However, the effect of pasteurization on human milk phages is unknown. This study assessed the effect of HP and UV-C on titres of exogenous bacteriophages inoculated into human milk. Ten donor human milk samples were tested in parallel with water controls. Milk samples or water controls were inoculated to a final concentration of 1 × 104 PFU/mL (±1 log) each of a thermotolerant Escherichia coli phage (T4) and a thermosensitive Staphylococcus aureus phage (BYJ20) and subjected to HP and UV-C treatments. UV-C inactivated both phages within milk and water controls, however, HP was ineffective against the thermotolerant T4 phages. Initial data suggest that UV-C treatment may eliminate phage with potential to affect preterm infant gut colonization. Further studies should extend this to other phages.


Introduction
Holder pasteurization (HP) is used by human milk banks globally to eradicate bacteria from donor human milk samples . T his technique , which in volves heating donor milk to 62.5 • C for 30 minutes, results in a 5-log 10 reduction of commensal and potentially pathogenic bacteria, including Esc heric hia coli , Staphylococcus epidermidis , Enterobacter cloacae , v egetativ e cells of Bacillus cereus , and Staphylococcus aureus (Czank et al. 2009 ). While the ability of HP to reduce bacterial numbers in human milk is well c har acterized, the effect of this treatment on non-bacterial microbes in milk remains understudied with r esearc h focused on human pathogens (Friis and Andersen 1982, Orloff et al. 1993, Walker et al. 2020, Pitino et al. 2021. The effect of HP on milk bacteriopha ges (pha ges) is curr entl y unknown.
Pha ges ar e bacterial viruses that ar e often highl y specific to their bacterial host. Both the human milk virome and the infant gut virome are dominated by phages (Lim et al. 2015, Pannaraj et al. 2018, with phage diversity being highest in early life (Lim et al. 2015 ). The infant gut phage-ome is initially dominated by pr opha ge, whic h is deriv ed fr om maternal sources (including milk) and induced within the infant gut (Duranti et al. 2017, Liang et al. 2020, Shamash and Maurice 2022. Phages modulate human health lar gel y via their effects on the bacterial microbiota (Weinbauer and Rassoulzadegan 2004 , Dahlman et al. 2021 ). Ho w e v er, despite the role of phages in human health, there is only limited data describing phage dynamics in early life . T his is a critical gap, given that early life is a k e y window for assembly and maturation of gut bacterial populations, with life-long consequences for host health (Stewart et al. 2018, Stiemsma and Michels 2018, Stinson 2020. The study of the infant gut phage-ome is limited by the large portion of uncharacterized viral diversity within infant gut samples, with 70% of infant viral taxa unable to be matched to gut viral databases (Shah et al. 2023 ). Given the predatory relationships between lytic phages and their bacterial hosts, phages may influence early life bacterial colonization patterns.
It is not known whether pasteurization of donor milk affects the viability of pha ges. Importantl y, giv en that pha ges can be thermotoler ant or thermosensitiv e (Jonczyk et al. 2011 ), HP ma y ha ve variable effects on different phage populations. Interestingly, data from the dairy industry suggest that milk itself may intrinsically provide thermal protection to phages. In one study of Leuconostoc phages, a 1-minute heat treatment at 70 • C on the test phage P808 resulted in a low phage reduction (1 log unit) when the phage was suspended in bovine milk, but a high phage reduction (4 log units) when the phage was suspended in water (Atamer et al. 2011 ). These data suggest that milk provides thermal protection to some pha ges, whic h may indicate that HP may be ineffective a gainst pha ge.
In addition to HP, ultraviolet C (UV-C) irradiation has been proposed as a novel form of treatment for donor human milk (Christen et al. 2013a ). UV-C treatment reduces milk bacterial numbers to a similar extent as HP, but has the additional benefit of pr eserving man y bioactiv e components of milk, whic h ar e destro y ed b y heating (Christen et al. 2013a , b ). While the impact of UV-C irradiation on phage titres in human milk is currently unknown, e vidence fr om the dairy industry suggests that UV-C light treatment may be more effective at destroying phages in bovine milk than heat pasteurization (Atamer et al. 2013 ).
Here, we aimed to characterize the effect of HP and UV-C irradiation on titres of thermosensitive and thermotolerant exogenous phages in human milk.

Participants and sample collection
Mothers of infants aged 0-12 months were invited to donate milk samples ( n = 11; 200-800 mL) for this study. Participants expressed and stored the samples in their home freezers prior to donation ( −20 • C, maximum 9 months storage). This study was approved by the University of Western Australia's Human Research Ethics Committee (RA/4/1/2369) and all participants provided written informed consent.

Milk characteristics
Total fat, total protein, and total solids wer e measur ed in each human milk sample prior to inoculation and pasteurization. Total fat was measured using the validated creamatocrit method (Lucas et al. 1978, Du et al. 2017. Total pr otein was measur ed in skim milk in duplicate using the Bradford method (Bradford 1976 ). Total solids content was measured by pre-drying samples on a boiling water bath, follo w ed b y e v a por ation and in a drying ov en at a temper ature of 102 • C (Standardization, 2018 ).

Bacteriophage prepar a tion
Tw o phages w ere selected based on their thermostability characteristics and included thermo-resistant dsDNA phage Escherichia coli phage T4 (select as a reference phage) and thermo-sensitive dsDN A Staph ylococcus aureus phage BYJ20 (laboratory phage isolated fr om waste water, selected due to the high abundance and pr e v alence of Staphylococcus in human milk). The pr opa gating hosts included Esc heric hia coli B (r efer ence str ain) and Staphylococcus aureus SSCC 61935 (clinical isolate) and were cultured overnight in Tryptic Soy Broth (TSB; Becton Dickinson, USA) at 37 • C, 250 r pm. High titr e stoc ks of 10 8 PFU/mL bacteriopha ges wer e pr opa gated as per standard ov erlay methods described previousl y (Furfar o et al. 2020 ) and purified using centrifugal filtr ation (Amicon 100 KDa, Merck, Germany) in Sodium Magnesium (SM) buffer (pH 7.5) (Bonilla et al. 2016 ).

Inoculation of human milk samples and water controls
For each experiment ( n = 10), 800 mL of human milk was used. Given that this volume was not always obtainable from a single donor, milk samples were pooled from two donors for one experiment. These samples were combined in a sterile 1-L Schott bottle and gently mixed by in verting. T he remaining nine experiments consisted of milk from single donors. Samples were inoculated with Esc heric hia coli pha ge T4 and Staphylococcus aureus phage BYJ20 to a final concentration of 1 × 10 4 PFU/mL ( ±1 log) each and again gently mixed (Fig. 1 a).
To assess any potential thermal/UV-C protection that milk may provide for phages, a water control was included alongside each experimental batch. A volume of 800 mL of MilliQ w ater w as inoc-ulated with phages as above and processed in an identical manner to the milk samples.

HP and UV-C irr adia tion
Each milk sample ( n = 10) and water control ( n = 10) was divided into two 400 mL aliquots. One aliquot was Holder pasteurized by heating to 62.5 • C in a water bath for 30 minutes. Temper atur e was gauged using a thermal probe (Thermocouple Thermometer, Delta Ohm, Ital y). UV-C irr adiation was performed as pr e viousl y described (Christen et al. 2013a ). Briefly, samples were placed into a sterile 500-mL beaker in a sterile laminar flow hood. A germicidal UV-C lamp (95% of UV-C output at 253.7 nm; Infralight Pty Ltd, Helensburgh, NSW, Australia) was placed diagonally into the beaker, so that the uncov er ed portion of the lamp was submerged in the sample (Fig. 1 b). Samples were stirred with a magnetic stir bar (500 rpm) to create a low velocity laminar flo w v ortex of milk around the lamp, ensuring all of the sample was exposed to the light throughout the duration of the treatment (18 minutes). UV-C r adiance was measur ed for eac h experiment using a UV-C light meter (X9 11 μ UV -C Meter , Gigahertz-Optik, Türkenfeld, German y). UV-C dosa ge (fluence) was calculated as radiance (W/m 2 ) × exposure time (seconds). Following HP or UV-C irradiation, samples wer e immediatel y aliquoted and pha ge titr e determined via plaque assa ys .
Given that human milk has antimicrobial properties (Lonnerdal 2003 ), there was the possibility that phage titres may be affected by the milk per se , regardless of pasteurization method. To account for this, phage inoculated untreated aliquots of milk and w ater w er e stor ed at r oom temper atur e for the duration of each experiment and then titred to assess survi ving n umber of pha ges (non-tr eatment contr ols).

Bacteriophage quantification
Bacteriophage activity was quantified using the spot test method as described pr e viousl y with modifications (Furfaro et al. 2020 ). Briefly, double agar o verla ys were prepared using Tryptic soy agar (TSA; Becton Dickinson, USA), whereby 4 mL of molten TSA (0.5%) and 250 μL of pr opa gating host bacteria ( Esc heric hia coli B or Staphylococcus aureus clinical isolate SSCC 61935) were layer ed ov er a solid TSA plate. Neat and dilutions (in SM buffer) of the experimental milk and water samples were spotted onto the o verla ys in triplicate 10 μL spots. Plates were incubated at 37 • C under atmospheric conditions overnight and individual plaques quantified after 24 h. All r esults wer e standardized based on the quantification of bacteriophages resulting from immediate inoculation prior to treatment for each sample (original titre postinoculation).

Sta tistical anal ysis
Statistical analyses were performed using IBM SPSS Statistics for Windows, version 28.0.1.0 (IBM Corporation, USA). Descriptive statistics r e v ealed non-normal distribution resulting in use of the non-parametric independent samples Kruskal-Wallis test. Bonferr oni corr ection was used to account for multiple tests and significance was set at an alpha value of 0.05.

Milk characteristics
Total fat, total protein, and total solids for the tested samples were within expected ranges for mature human milk (Table 1 ;

Holder pasteurization
HP had little impact on the thermo-resistant phage T4 with titres within 1 log of the inoculated untreated milk ( P = 1.000; Fig. 2 ) and minimal difference in T4 titre was observed between the non-tr eatment contr ol and HP-tr eated water samples ( P = 0.274). Additionall y, plaques wer e mor e defined and lar ger for T4 after HP compared to pre-treatment. In contrast, Staphylococcal phage BYJ20 (thermo-sensitive) was consistently inactivated by HP methods ( P < 0.001).

UV-C irr adia tion
An av er a ge UV-C dose of 1879.2 J/m 2 was ac hie v ed during the 18minute treatment time (SD ± 1058 J/m 2 ). UV-C irradiation displayed complete consistent inactivation across almost all treatments of both T4 and BYJ20 phages ( P < 0.001; Fig. 2 ), noting T4 pha ge withstood UV-C inactiv ation in two human milk samples ( ∼50% reduction in titre) and BYJ20 phage in one water sample (30% reduction in titre). There was no observed major difference between pha ge titr es corr esponding to suspension in water or human milk.

Non-treatment controls
To test the impact of the milk itself, non-tr eatment contr ols wer e included in the study. There was no significant difference between the controls and the inoculated pre-treatment groups ( P > 0.05). Similarly, non-inoculated milk was tested for endogenous phage activity against the propagating hosts used in this study and only one of the 10 milk samples had observ ed pha ge activity a gainst E. coli B host. This was purified and plaque production confirmed, ho w e v er, the low titr e pr esent in the raw milk (single plaque on neat) did not impact the results obtained.

Discussion
Numer ous studies hav e highlighted the abundance of bacteriopha ges pr esent in human milk (P annar aj et al. 2018 , Liang et al. 2020 , Mohandas and P annar aj 2020 , Dinleyici et al. 2021 ), demonstrating the need for assessment of the impact that pasteurization tec hniques hav e on these types of viruses. Her e, we demonstr ate that pha ge ar e differ entiall y impacted by UV-C irr adiation and HP of human milk. HP was able to er adicate thermosensitiv e S. aureus phage (BYJ20), ho w ever, thermotolerant E. coli phage (T4) remained viable with minimal reduction in titre. In contrast, UV-C tr eatment er adicated both T4 and BYJ20 pha ges in 8/10 samples, with no observable activity following treatment.
Removal of viral agents in human milk may be beneficial when considering pathogenic viruses such as Zika virus, cytomegalo virus , and human immunodeficiency virus (Black 1996 , Michie andGilmour 2001 , Blohm et al. 2018 ); ho w ever, the importance of bacteriophages in the early life microbiome is not well understood. A r ecent anal ysis of the human milk virome found that 92% of all human milk samples tested had viruses detected and further observed differences in bacteriophages predominance with respect to lactation period, preterm birth, mode of delivery, and infant birth weight (Dinleyici et al. 2021 ). For the safety of preparing donor milk, UV-C treatment is a promising method for inactivation of viruses such as bacteriophages, howe v er, giv en their abundance in healthy human milk (P annar aj et al. 2018, Liang et al. 2020, Mohandas and P annar aj 2020, Dinleyici et al. 2021, one might speculate of their role in the early stages of microbiome development. Indeed, vertical transmission of Bifidobacteria phages has been demonstrated from mother to infant via human milk (Duranti et al. 2017 ). If bacteriophages play a role in shaping the early life bacterial microbiome, their inactivation by commonly used milk bank pasteurization methods may Figur e 2. T he effect of HP and UV-C treatment (UV-C) compared to non-treatment controls (Control) on: (A) E. coli phage T4 and (B) S. aureus phage BYJ20 in both water (Water) and human milk (Milk). A total of 10 samples were tested and are represented as individual columns for each variable. Significant differences between HP and UV-C treatment of phage T4 ( * P < 0.003). BYJ20 phage sho w ed significant difference between Control and both tr eatment gr oups (UV-C and HP) ( * * P < 0.01). Ov er all, no differ ence was observ ed betw een w ater and human milk samples.
impact early microbiome establishment in donor milk fed infants. In particular, donor milk is fr equentl y fed to preterm infants, whose gut microbiomes are known to vary from those of full term infants (Aguilar-Lopez et al. 2021 ). These infants are particularl y vulner able to bacterial infections, suc h as E. coli , leading to necr otizing enter ocolitis. Faecal filtr ate tr ansplants suggest that bacteriophages may play a role in protection from gut bacterial infections (Ott et al. 2017, Brunse et al. 2021. Ho w e v er , the curr ent e vidence on bacteriopha ge populations in human milk and the early life gut is sparse. More work is needed to understand phage-bacteria dynamics in infants. Given that infant gut phage profiles are determined by breastfeeding (Liang et al. 2020 ), and that gut phages modulate the bacterial microbiota (Liang et al. 2020 ) and exert direct host effects (Gorski et al. 2018, Fluckiger et al. 2020, we suggest that destruction of milk phages by pasteurization may impact infant health. Indeed, eradication of thermo-sensitive phages in Holder pasteurized donor milk may contribute to the differences seen in the gut microbiome and health outcomes of donor milk fed and mother's own milk fed infants (P arr a- Llorca et al. 2018, Pineiro-Ramos et al. 2021 ). The present study highlights that different milk processing tec hniques can hav e effects on pha ge populations. Understand-ing these effects is an important first step to w ar ds examining endogenous milk phages in future studies to further contextualize our findings.
While milk phages may act to maintain a balanced infant gut micr obiome, pha ges may also pose a risk to vulnerable preterm infants. In fact, transfer of antimicrobial genes via human milk has been acknowledged (Das et al. 2019 ), potentially implicating milk phages as mobile genetic elements and a potential source of gene transfer among milk taxa such as Staphylococcus aureus (temper ate pha ges in particular). Ther efor e, ther e may be incidences wher e er adication is warr anted. Further, pasteurization may r esult in induction of temperate phages. Bacterial stress response activation can lead to the induction of pr opha ges, with UV a wellknown inducing agent (Klaenhammer and McKa y 1976 ). T herefore, the effects of UV treatment on endogenous bacteria and subsequent pr opha ge induction in human milk is an important topic for future study. Our study provides vital information to enable the field to assess the various parameters that impact the milk microbiome and the potential effects of processing techniques on bacteriophage activity.
While pr e vious work in bovine milk suggested that milk provides thermal protection to certain phage (Atamer et al. 2011 ), this did not appear to be the case here, potentially due to differences in the physiochemical composition of human and bovine milk. Ov er all, human milk did not appear to influence the stability of the pha ges a gainst eac h tr eatment as the r esults fr om the water contr ols wer e not statisticall y differ ent to those of milk. Milk deri vati ves such as skim milk have been assessed as microbial cryopr eserv ation a gents pr e viousl y, howe v er, with v ariable outcomes (Cody et al. 2008 ). It has been suggested that skim milk may affect the fatty acid content of the bacterial cell membranes, which may, in turn, change the viscosity of membranes and help to stabilize cell enzymes (Guo et al. 2020 ). Ho w e v er, in the pr esent study, where whole milk was used, this did not appear to be the case in stress conditions such as heat (HP) and UV-C irradiation. Similarl y, this r esult is promising for UV-C methods, highlighting penetration was not an issue in this study, with the continuous movement and direct UV-C exposure sufficient to eradicate both phages within human milk, which is a high opacity solution.
UV-C irradiation of human milk has previously been shown to inactivate the eukaryotic virus cytomegalo virus , whilst retaining the bioactivity of the milk itself (Lloyd et al. 2016 ). Our study further supports this result, with UV-C treatment observed to inactiv ate both pha ges used, while HP was onl y able to inactiv ate the thermosensitiv e pha ge (BYJ20). Whilst milk bioactivity was not assessed in this study, pr e vious studies (Christen et al. 2013b, Lloyd et al. 2016, Martysiak-Ż urowska et al. 2017, Almutawif et al. 2019 ) hav e demonstr ated minimal impact of UV-C on milk pr oteins and bioactive components, including secretory IgA, lactoferrin, and lysozyme. As a result, UV-C treated milk has been shown to induce better weight gain and intestinal health in preterm piglets (Li et al. 2017 ). This makes UV-C treatment a promising novel pasteurization technique for donor human milk. Ho w e v er, we note that data from studies of human (Martysiak-Ż urowska et al. 2017 ) and animal (Urgu-Ozturk 2022 ) milk suggest that UV-C irradiation may result in lipid oxidation, which may produce undesirable organoleptic effects. Further optimization of UV-C radiance and treatment time may identify a UV-C tr eatment pr otocol that eliminates micr oor ganisms while minimizing oxidation and other unwanted effects.
The impact of UV-C on bacteriophages has been well studied in gener al, with earl y r e ports of effecti ve use of UV-C to r emov e Streptococcus lactis bacteriophages from commercial dairy plants (Greene and Babel 1948 ). Using a lamp source above rather than submerged within the filtrate, they found that the wattage of the lamp, distance from the sample, and concentration of bacteriophages influenced the time required to inactivate the bacteriophages. In comparison to our methods, the recommended time to destroy all bacteriophages at a titre of 10 3 PFU/mL of pha ges, 3 inc hes fr om the lamp source (279 μW) was 21 minutes (Greene and Babel 1948 ). Our study has the benefit of the UV-C lamp being submerged and the steady flow of the solution allowing the entire sample to be dir ectl y exposed to the UV-C irradiation, whic h has r esulted in complete inactiv ation in 18 minutes (av er a ge UV-C dosa ge 1879.2 J/m 2 ). The titr e of pha ges used in the current study was moderate (10 4 PFU/mL); ho w ever, despite the high likelihood of other endogenous pha ges pr esent natur all y within the milk (targeting different hosts), we still observed inactivation of the phages present. To assess the scalability of this a ppr oac h, futur e studies may r equir e assessment of v arious titres.
Here, we describe the impact of UV-C compared to HP on phages in human milk. We show that while UV-C irradiation efficientl y destr oys both thermotoler ant and thermosensitiv e pha ge, HP is onl y effectiv e a gainst thermosensitiv e pha ge. Our r esults hav e br oad implications giv en the potentiall y beneficial r ole phages may play in the infant gut microbiome. While exogenous pha ge wer e tested in this proof-of-concept study, the effect of UV-C and HP treatment on the endogenous human milk 'phageome' should be assessed to better understand the impact of donor milk treatment on the developing infant gut microbiome. In particular, if human milk phage play a role in protecting preterm infants fr om neonatal necr otizing enter ocolitis and other bacterial infections, the impact of donor milk pasteurization techniques on milk pha ge m ust be consider ed.

Ac kno wledgements
The authors would like to acknowledge Dr Jeremy Barr for providing T4 phage and thank all of the mothers that participated in this study by donating breastmilk. We acknowledge Matthew Payne for the use of lab space for this study and ar e gr ateful to Shar on Perr ella, Zoya Gridne v a, and Ashleigh Wardern for their assistance in recruiting participants for this study. Graphical abstr act cr eated with BioRender.com. Medela AG (Switzerland), administered by the University of Western Australia had no role in the design of the study, collection/anal ysis/inter pr etation of data, writing of the manuscript, or in the decision to publish the results