Radiolabeled Diatom Cultures
Natural seawater with a salinity of 35, collected from the Gulf of Mexico, was sequentially filtered through a
0.2 μm polycarbonate cartridge and ultrafiltered with a 1000 amu cutoff membrane to remove particulate
and colloidal organic matter [Guo et al., 1995; Roberts et al., 2009]. The <1000 amu ultrafiltrate fraction was
then used for later experiments. The 234Th tracer was milked and purified from a 238U solution [Alvarado
Quiroz et al., 2006; Quigley et al., 2002]; 233Pa, in equilibrium with 237Np, was obtained from Pacific Northwest
National Laboratory; 210Pb, in 1 mol L-1 nitric acid (HNO3), was purchased from Eckert & Ziegler Isotope
Products, and the 7Be tracer solution (in 0.1 mol L-1 hydrochloric acid, HCl) was manufactured at the Paul
Scherrer Institute, Switzerland [Schumann et al., 2013].
Autoclaved f/2 media (50 ml) were added to preconditioned clear polyethylene containers, and then ~10 to
15 Bq of each radionuclide tracer (234Th, 233Pa, 210Pb, and 7Be) was added. In each radiolabeled medium, 1ml
of laboratory axenic culture, Phaeodactylum tricornutum (UTEX 646), was then added and incubated at a temperature of 19 ± 1°C with a light:dark cycle of 14 h:10 h under an irradiance of 100 μmol quanta m-2 s-1.
Incubation experiments were carried out in duplicate. The growth status of P. tricornutum was monitored
by changes in optical density at 750nm (OD750) in a parallel nonlabeled culture. When P. tricornutum reached the stationary phase observed by its OD750, cells were harvested for further extraction and analyses. The total incubation
time was 35 days.
Exopolymeric Substance (AEPS and NAEPS) Extraction
AEPS and NAEPS extractions were performed following the procedures described in Chuang et al. [2014, 2015]. Briefly, for NAEPS, laboratory cultures were centrifuged (2694 g, 30 min) and filtered (0.2 μm). The filtrate was desalted via diafiltration with a 1000 amu cutoff cross-flow ultrafiltration membrane, followed by freeze drying
for later use. For the AEPS extraction, diatom cells were collected after centrifugation from the previous step. Then,
the pellet was soaked with 0.5mol L -1  sodium chloride (NaCl) solution for 10 min, followed by centrifugation at
2000 g for 15 min to remove the medium and weakly bound organic material on the cells. The pellet was
then resuspended in a fresh 100 ml, 0.5mol L-1 NaCl solution and stirred gently overnight at 4°C. The resuspended
particle solution was ultracentrifuged at 12,000 g (30 min, 4°C), and the supernatant was then filtered through a 0.2 μm polycarbonate membrane. The filtrate was further desalted via diafiltration with a 1000 amu cutoff ultrafiltration membrane and subsequently freeze dried for later use. 

Intracellular and Frustule BF Extraction
Procedures for frustule biopolymers extraction adapted from Scheffel et al. [2011]. Briefly, the
clean diatom cells from the previous AEPS extraction step were resuspended in 10 ml, 100mmol L-1 EDTA
(pH 8.0) at 4°C overnight. EDTA solution was used to extract the intracellular material after cell lysis. The supernatant
was collected after centrifuging at 3000 g for 10 min, defined as EDTA extractable BF. Subsequently,
the pellet was placed in 10 ml, 1% SDS in 10mmol L-1 Tris (pH 6.8) solution and heated at 95°C for 1 h.
The resulting frustules were collected by centrifugation (2500 g, 10 min), washed with 10 ml milli-Q water 3
times, and then were freeze dried for later use. The supernatant from this step was collected and defined
as SDS extractable BF, mostly composed of soluble cell-membrane-associated materials. These two fractions
represent intracellular biopolymers lysed after cell breakage.

HF digestion was applied to help extract the diatom frustule biopolymers. HF is a nonoxidizing acid commonly
used to convert SiO2 to volatile SiF4 during wet digestion [Scheffel et al., 2011; Šulcek and Povondra,
1989]. Hence, frustule biopolymers could be separated from the digested solution by a 3 kDa cutoff membrane. However, high-concentration HF would also liberate A type metal radionuclides (Th, Pa, and Be in
this study) from any complex by frustule biopolymers [e.g., Burnett et al., 1997]. Furthermore, deglycosylation
might also have occurred during a HF digestion [Mort and Lamport, 1977]. Therefore, the <3000 amu fraction
represents the sum of silica frustules and broken down frustule biopolymer residues.

Subsequently, 5 ml, 52% HF was added to the frustules in a 15ml plastic centrifugation tube, and the mixture
solution was incubated on ice for 1 h. Hydrogen fluoride was then evaporated under an N2 stream to reduce
the volume to dryness. The remaining material was neutralized with 3ml Tris–HCl (250mmol L-1, pH 8.0) and
followed by centrifugation at 11,000 g for 15 min with 3000 amu Microsep centrifugal filter tubes (Milipore).
The filtrate was collected and defined as the fraction of digested silica with <3000 amu frustule BF residues.
The supernatant (defined as>3000amu HF soluble BF, e.g., silaffin) was concentrated to 250 μL and rinsed with
milli-Q water. The pellet from this step was then washed by a 3 ml, 200mmol L-1 ammonium acetate solution
twice with centrifugation at 3000 g for 20 min. The pellet was then resuspended in a 2 ml, 100mmol L-1
ammonium acetate solution and was sonicated for 20 s until the mixture solution appeared homogenized.
After ultracentrifuging the mixture solution at 12,000 g for 5 min, the pellet (>3000 amu HF insoluble BF,
e.g., cingulin) was collected and freeze dried for later use. Combined BF from all three HF fractions represented
frustule-embedded biopolymers.

Activity concentrations of 234Th, 233Pa, 210Pb, and 7Be were measured by counting the gamma decay energies at 63.5 keV, 312 keV, 46.5 keV, and 477.6 keV, respectively, on a Canberra ultrahigh purity germanium well detector. The 210Po activity was analyzed by liquid scintillation counting (Beckman Model 8100 Liquid Scintillation Counter).
Concentrations of total carbohydrate (TCHO) were determined by the TPTZ (2, 4, 6-tripyridyl-s-triazine) method using glucose as the standard and [Hung and Santschi, 2001]. Protein content was determined using a modified Lowry protein assay, using bovine serum albumin as the standard (Pierce, Thermo Scientific). uronic acids (URA) were measured by the metahydroxyphenyl method using glucuronic acid as the standard [Hung and Santschi, 2001].

Elemental contents of carbon (C) and nitrogen (N), were determined by a Perkin Elmer CHN 2400 analyzer, using cysteine (29.99% C, 11.67% N) as a standard.