All materials were cleaned by soaking overnight with heating (1.5% Citrad®, by Decon Labs, Inc) in deionized water, rinsed in RO water, and soaked in 10% TMG HCl (Fisher) in ultrapure water for 30 days, then rinsed with ultrapure water, let dry in an AirClean 400 work station overnight, and double-bagged in polyethylene bags (Mellett et al. 2018). M9 minimal media for bacterial cultures was made using ultrapure water (18.2 MΩ cm), containing final concentrations of 33.7 mM Na2HPO4·2H2O (Sigma-Aldrich ≥99.0% Titration), 22 mM KH2PO4 (ACS Reagent ≥99% purity), 8.56 mM NaCl (Certified ACS ≥99.0% purity), 18.7 mM NH4Cl (Fisher Scientific ≥99.0% purity FCC), 0.1 M MgSO4 (Sigma-Aldrich ≥99.99% trace metal purity), 0.1 M CaCl2 (Alfa Aesar from Fisher Scientific 99.99% metals basis), 1 µg/ml Thiamine HCl (Fisher Scientific 99% purity), and 0.5% Glucose (Fisher Scientific 99% purity) in 1 L of Milli-Q (Kutter and Sulakvelidze 2004, Table 1; see Supplemental Documents below).

M9 minimal media was spiked with 57FeSO4 (final concentration: 10 µM), then filtered through a 0.02 µm Whatman Anotop syringe filter that had been rinsed with Milli-Q; the first few drops of media were discarded (Sample 2). Some un-spiked media was reserved for bacterial pellet rinses (Sample 1). A volume of 20 mL media was inoculated with a frozen culture of the E. coli strain ZK126 (Betty Kutter, Evergreen State College) which was grown over eight generations exclusively on M9 minimal media with 57FeSO4. The culture was placed into a polyethylene bag and vented to avoid contamination, then incubated while shaking overnight at 37 °C to reach late-logarithmic growth. The following morning, three 45 mL aliquots of 10 µM 57FeSO4 spiked and 0.02 µm filtered media samples (one 45 mL blank sample remained un-inoculated with E. coli) were weighed into 250 mL acid-cleaned polycarbonate flasks. Each of the three aliquots were inoculated with 10% of the overnight bacterial culture. The three bacterial cultures as well as the blank samples were placed in a vented polyethylene bag and incubated while shaking at 37 °C. Once the culture reached mid-logarithmic growth, as indicated by an absorbance (OD600)  measured on a spectrophotometer between 0.200-0.500 (Figure 1), the cultures were divided into 20 mL aliquots (Figure 2, see Supplemental Documents, below).

To rinse excess 57Fe label from the surface of the bacterial cells, the aliquoted cultures were transferred to a 50 mL falcon tube and centrifuged at a speed of 6500 x g. The bacterial cells pelleted to the base of the tube, and the supernatant was discarded. A volume of 10 mL of fresh Fe-less (not spiked with 57FeSO4) 0.02 µm filtered media was added and vortexed for 1 minute. The rinsed bacterial cells were again centrifuged 6500 x g for 5 minutes to pellet, and the cell rinsing was repeated three times. The final bacterial pellet was re-suspended in 20 mL of Fe-less media. Samples designated A, B, and C were T4 phage lysates, samples D, E, and F remained uninfected by phage and served as bacterial control cells that were burst open by treatment with chloroform, and samples G, H, and I were blanks of Fe-less media. For all rounds (Round 1, 2, 3, 4, and 5 samples A, B, and C (Figure 3) were infected with 5 µl of T4 bacteriophage (Betty Kutter, Evergreen State College) at a titer of 1.3 x 1011 phage/ml. For Rounds 2, 3, 4, and 5, samples M, N, and O were infected with 5 µl of T5 bacteriophage (ATCC® 11303-B5™) at a titer of 8.2 x 1011 phage/ml. Samples D, E, and F (bacterial controls) as well as G, H, and I (blanks) remained uninfected and instead were stored at 4 °C overnight to be used as phage-free lysis controls. Phage-infected cultures A, B, and C were left shaking at 37 °C overnight.

On the next day of the experiment, the uninfected bacterial controls D, E, and F were treated with 20% chloroform (Fisher Scientific, Mobile phase for HPLC applications ≥99.8% purity) and vortexed for one minute to burst open the bacterial cells without viral lysis. The necessity for this viral lysis-free control is to account for colloids from within the bacterial cells that contain 57Fe and would purify with T4 phage. For Round 5 all phages (A, B, C, M, N, O) and blanks (G, H, I) were also treated with 20% chloroform and vortexed for one minute. All the samples were then centrifuged in 50 mL Falcon tubes at a speed of 9500 x g for 5 minutes to pellet the remaining bacterial debris. The supernatant, which contained the T4 phage progeny for samples A, B, and C and the T5 phage progeny for samples M, N, and O, was then filtered using a 0.22 µm Sterivex PVDF syringe filter (EMD Millipore), with a Milli-Q pre-rinse and the first few drops of sample discarded. The filtrate was the fraction containing phage and any soluble or colloidal 57Fe within the dissolved size fraction (<0.22 µm). The subsequent filtration of a subsample of the dissolved fraction through a 0.02 µm Whatman Anotop syringe filter (with a Milli-Q pre-rinse and the first few drops of sample discarded) was collected for the soluble fraction (<0.02 µm). The difference between the dissolved and the soluble fractions are used to calculate iron within the colloidal fraction (0.2 µm-0.02 µm).

The phage within the dissolved fraction were further purified using a sucrose cushion, which is a density-dependent technique used to concentrate and purify phage by precipitating viral particles below a dense layer of sucrose (Hurwitz et al. 2013). To do so, a 2.5 ml layer of 38% sucrose (Fisher Scientific) in SM buffer (100 mM NaCl, 8 mM MgSO4, 50 mM Tris-HCl in Milli-Q, pH 7.5 and 0.02 µm filtered) was added to the bottom of the ultracentrifuge tube (Beckman Coulter), followed by 1 ml of sample and 10.5 ml of SM Buffer by carefully tilting the tube so as not to disturb the dense sucrose layer. The samples were spun in a Beckman Coulter SW40Ti swinging bucket ultracentrifuge, for 3 hours and 15 minutes at 175,000 x g (37,200 rpm). The sucrose and SM buffer layers were then discarded, and the tubes were dried in a laminar flow clean hood (Air Clean) for 20 minutes. The pelleted phages from samples A-C and M-O, including any potential bacterial colloids of the same density as the phage as accounted for in samples D-F, were then resuspended in 1 mL of SM Buffer.

All the samples were dialyzed using Float-A-Lyzer 100 kDa dialysis devices (Fisher Scientific) in 1 L of dialysis buffer (10 mM NaCl, 50 mM Tris-Cl pH 8.0, 10 mM MgCl2) for a total of 6 buffer changes over 4 days. Bacterial and viral counts were performed throughout for samples using SYBR nucleic acid stain under epifluorescence microscopy (Noble & Furman 1998).

Metal concentrations were quantified using Element XR ICP-MS (Thermo) after 50-fold dilution with 5% nitric (Fisher Scientific, Optima) containing 10 ppb rhodium as internal standard, and using external standard calibration curves. Blank values after rhodium correction are listed in Table 2 (see Supplemental Documents below).