Albumin-mediated extracellular zinc speciation drives cellular zinc uptake

The role of the extracellular medium in influencing metal uptake into cells has not been described quantitatively. In a chemically-defined model system containing albumin, zinc influx into endothelial cells correlates with the extracellular free zinc concentration. Allosteric inhibition of zinc-binding to albumin by free fatty acids increased zinc flux.

. Isotopic purity of 68 Zn stock solutions (500 µM in 1× HEPES-buffered EBSS), determined by ICP-MS in He-gas mode. These isotopic abundances were considered in mathematical modelling of experimental data.

Analyte
Abundance (%) 64 Table S8. IC 50 concentrations for Zn 2+ (administered as ZnCl 2 ) determined in immortalised HUVEC cells in the presence of BSA (0-600 M) with defined supplementation with free fatty acids (FFAs): either none, C8:0 (as sodium octanoate), or C14:0 (as sodium myristate). Pvalues for the latter two conditions refer to the comparison to the fatty acid-free control at 60 M BSA and were calculated using a two-tailed t-test assuming unequal variances (Welch's t-test). Data are shown graphically in Figure S3. Tables S9-S16. Mathematical modelling results (data fits shown in Figure S3). Note that in cases where only the parameter is given, is equal to , as no change in total intracellular in out in Zn (Q) occurred.

Materials and Methods
Materials. Human umbilical vein endothelial cells (HUVECs) were immortalised using hTERT as described below and maintained in endothelial cell growth media (PromoCell GMBH, Germany

Immortalisation of primary human umbilical vein endothelial cells (HUVECs).
Early passage primary HUVECs were obtained from TCS Cellworks (Buckinghamshire, UK) and grown in endothelial cell growth media containing supplement mix (complete media; PromoCell, Germany). HUVECs were grown to 70% confluency and transduced with supernatant from ψ-CRIPpBABEpurohTERT cells (as described in reference 1 ) which were gifted by Prof. Andrew Riches (University of St Andrews). The ψ-CRIPpBABEpurohTERT cells were grown in DMEM supplemented with 2 mM glutamine and 10% foetal bovine serum (FBS) and reached confluency. The supernatant was collected, filtered (0.45 μm) and 10 μg mL -1 polybrene was added to enhance the binding of viral particles to the HUVEC membrane. Only a single transduction was performed. After 24 h, the supernatant was removed and HUVECs were given fresh complete medium. HUVECs were passaged and 0.2 μg mL -1 of puromycin was added to the medium for selection of cells.

Reverse transcription-polymerase chain reaction (RT-PCR).
To establish the expression of endothelial markers (genes encoding CD32, von Willebrand factor and VE-cadherin) in the immortalised HUVECs, RNA was extracted using TRIzol reagent (Invitrogen, UK) according to the manufacturer's instructions. RNA was precipitated using isopropanol and quantified using by NanoVue (GE Healthcare, UK). A 2 µg aliquot of total RNA was reverse transcribed into cDNA using Oligo(dT)-15 primer (Promega, UK). One µL of the resultant cDNA sample was subjected to PCR analysis using Bio-X-ACT short mix (Bioline, UK) and 250 nM of each primer. The cycling profile was: 95°C for 30 s, 58°C for 30 s and 68°C for 30 s for 35 cycles, preceded by a 2-minute enzyme activation at 95°C. Gene-specific primers for genes encoding CD31, vWF and VE-cadherin were designed and obtained from Eurofins Genomics (Ebersberg, Germany). PCR products were analysed on agarose gels composed of 1% (w/v) agarose, melted in Tris-acetate/EDTA-buffer (TAE-buffer). SyberSafe dye was added to the melted agarose to enable the detection of DNA in the gel. The gel was left to set and then was transferred into an electrophoresis tank containing TAE-buffer. Samples were prepared with a gel loading dye and Hyperladder I DNA-ladder was loaded (Bioline). Gels were run at 150 V for as long as required for the separation of bands. DNA was visualised in a Gel Doc XR+ Imager and analysed using Image Lab 2.0 software (Bio-Rad Laboratories, UK).  Table 1 and ESI Tables S2-S15).

Preparation of isotope-enriched
Zinc uptake experiments. Briefly, 5 × 10 6 HUVECs were seeded in P145 dishes using Endothelial cell growth medium (PromoCell, Germany) and incubated until achieving >90% confluence. (310 K, 5% CO 2 humidified atmosphere). The supernatant medium was then removed by aspiration, and cells were treated with physiological concentrations of BSA (600 µM) and natural abundance Zn 2+ (20 µM) prepared in HEPES-buffered EBSS for 24 hours (NB the growth of human umbilical vein endothelial cells is contact-inhibited). Cells were then washed thoroughly (3×) with PBS, and then treated with test solutions: isotope-enriched 68 Zn 2+ (20 µM) with defined bovine serum albumin supplementation (0-600 µM). Cell pellets were collected in a time-dependent manner using 1 mL Trypsin/EDTA (0.25%) and were resuspended in PBS to obtain a single cell solution, from which a cell count was obtained using a hemocytometer. Cell pellets were obtained by centrifugation and repeated washing with PBS (3 × 1 mL). Cells were then re-suspended in 200 μL ultra-pure 72% nitric acid and digested at 351 K. After 24 h, solutions were diluted with milli-Q water to achieve a final working nitric acid concentration of 3.6% v/v (4 mL). Zinc content ( 64 Zn, 66 Zn, 67 Zn, 68 Zn, 70 Zn) was determined using an Agilent 7900 series ICP-MS in He-gas mode. The instrument was calibrated using freshly prepared Zn standard solutions (0.1-1000 ppb) in 3.6% v/v nitric acid with an internal standard of 166 Er (50 ppb). Data were processed using MassHunter (Agilent Technologies, Inc., UK). Each timepoint is the result of three biological repeats. For the experiments involving FFAs, isotope-enriched 68 Zn 2+ (20 µM) media were prepared in the presence of 60 µM BSA with 5 mol. equiv. (300 µM) of FFAs, octanoate (C8:0), myristate (C14:0) or palmitate (C16:0), which were pre-incubated at 310 K for 24 h. Fatty acids were not co-administered in the pre-conditioning step of any experiment.
Determination of Zn 2+ toxicity. Briefly, 5  10 4 HUVECs were seeded per well in a 96-well plate using 0.15 mL of endothelial cell growth medium per well and incubated for 48 h (310 K, 5% CO 2 atmosphere). After this time, the supernatant was removed and cells were treated with defined concentrations of Zn 2+ (as ZnCl 2 , typically 0.001-10 mM) in HEPES-buffered EBSS, in the absence of FFAs and in the presence of either 0, 6, 30, 60, or 600 M BSA for 24 h. After this time, the supernatant was removed, and cells were washed with PBS (2  200 L). Cells were then fixed by addition of 50% trichloroacetic acid (50 L per well) and incubated at 277 K for 1 h. The supernatant was removed by sequential washing with water, and cells were stained using sulforhodamine B (0.4% dye content in 1% acetic acid solution) for 30 min. After this time, excess dye was removed by sequential washing with 1% acetic acid solution, and then dye was liberated by addition of Tris base (pH 10.5, 10 mM, 200 L per well) for 1 h. Absorbance was measured using a Thermo Scientific SkanIt microplate reader fitted with a 492 nm filter. Absorbance measurements were normalized relative to untreated negative control wells to determine cell survival. IC 50 value determinations were carried out as duplicate of triplicate experiments, and the IC 50 values were calculated using Origin for Windows to fit sigmoidal curves to experimental data. Data are reported as the average of each duplicate experiment with associated standard deviation. The experiment was repeated with the following modification: Zn 2+ toxicities were determined as described above using a fixed concentration of 60 M bovine serum albumin (equivalent to 10% fetal calf serum routinely used in cell culture experiments) in the presence of FFA (0.3 mM, 5 mol. equiv., C8:0 octanoate or C14:0 myristate) for 24 h.

Calculation of free zinc concentrations. Free [Zn 2+
] was estimated based on affinity data for the primary binding site, namely stoichiometric stability constants of log K = 7.0 and = 7.6. 2, 3 Using a published pK a = 8.2, 2 these were converted to conditional stability constants valid at pH 7.4, giving log K' = 6.13 and 6.73. These were then used, together with total BSA and Zn 2+ concentrations, to calculate free [Zn 2+ ], for both constants separately, to give an upper and lower limit. The data plotted in Figure 3(b) are the average of these two values. It may be noted that the resulting concentrations are likely an over-estimate, as they do not take binding to site B into account. We have refrained from attempting to include site B, as no stoichiometric or suitable conditional stability constant is available for this site.
The pro rata assumption yields the kinetics for each isotope: Applying the quotient rule and substituting the above expressions for the rates of change of and , we find: It is useful to switch to the auxiliary variable significant amount of zinc during the experiment. If the null hypothesis of a slope equal to zero could not be rejected, the cells did not change their total content, meaning that fluxbalance was given, and the equations valid for were used. The influx was ϕ in = ϕ out ≡ ϕ estimated by non-linear least-squares fitting of the data to the model given above; for the flux-balanced case, Q 0 was set equal to 52 fgcell -1 . If the null hypothesis was rejected, the parameters and were estimated by non-linear least-squares fitting of the 66 Zn/ 68 Zn ϕ in ϕ out data to the general model given above. . To obtain a parameter estimate in this case, the quantity (  was set equal to the ϕ out ϕ in ϕ out ) estimated slope of the linear regression to the accumulation data, and only was ϕ in subsequently estimated via the normal procedure.

Justification for assuming constant fluxes.
To validate the assumption that fluxes were constant over time, an empirical reconstruction approach was used. Briefly, we have the following explicit expressions for the fluxes as a function of time: and ϕ out (t) = -Q(t) G ' (t) with G(t) = r 66,exr 68,ex R(t) r 66,ex -66,ex -(r 68,ex -̃r 68,ex )R(t) where 66 Zn/ 68 Zn. This means that the fluxes can in fact be reconstructed as a function of ≡ time if R and Q have been sampled sufficiently frequently in time, with sufficient accuracy, to warrant the fitting of a suitable smooth curve. A drawback is that the expressions involve differentiation, which tends to exacerbate the effects of any inaccuracies. However, this approach was merely taken to test whether the assumption of time-constant influx and efflux over the course of the experiment was warranted. By way of representative example with relatively high influx and non-constant Q(t), let us consider the accumulation and isotope ratio data for the [BSA] = 40 M experiment. The figure below shows smooth curve fits to the data, giving R(t) and Q(t) empirically, as well as the resulting reconstructed fluxes. The latter are broadly consistent with the assumption of time-constant fluxes, although there is evidence that the efflux experiences a transient at the beginning of the experiment.
The assumption of constant fluxes is an idealisation that results in excellent goodness of fit, whereas the empirical-reconstructive approach suggests that the cells go through an initial period of acclimatisation. Inasmuch as the data do not clearly favour one or the other hypothesis, we have preferred to adhere to the simplicity of the mechanistic analysis presented in the main text.