where MHb = molecular weight of hemoglobin (64.5 kDa), AFry = Fry-corrected absorbance (absorbance/cm) at 410 nm, εHb = hemoglobin extinction coefficient, and N = absolute number of cells. Pyridine hemochromogen assay
Samples of mouse blood and cell culture resuspensions were kept on ice and disrupted by sonication. An equal volume of 50% (v/v) pyridine: 0.2 N NaOH solution was added to the hemoglobin standards or homogenized cell samples to prepare the pyridine hemochrome derivative. Heme content was measured from reduced minus oxidized difference spectra as described by Berry and Trumpower (23) and based on the original work of Paul et al. (7). Scans were carried out on a Cary 1G UV-visible spectrophotometer, whose settings were controlled using Cary WinUV software (Agilent, Santa Clara, CA). The absorbance difference between the reduced peak at 556 nm and oxidized trough at 540 nm (Supplementary Figure S1A) was converted to heme concentrations using the extinction coefficient 23.98 µM-1 cm-1 (23). Fluorescence heme assay
Fluorescence assays were carried out as previously published (8,13), with minor modifications. Briefly, aliquots of hemoglobin standards or sonicated cells were diluted 40-fold in concentrated oxalic acid and then split. One sample was boiled at 99°C for 30 min in a Robocycler PCR machine (Stratagene, La Jolla, CA). The other solution was treated in the same way except that it was maintained at room temperature throughout to control for the presence of endogenous porphyrins. All samples were then analyzed on a CLARIOstar microplate reader (BMG LABTECH, Offenburg, Germany) at excitation and emission wavelengths of 400 nm and 662 nm, respectively. Sensitivity and limits of detection and quantitation
Calibration data for hemoglobin standards were pooled to derive a limit of detection (LOD) for the CLARiTY, pyridine hemochromogen assay, and fluorescence heme assay using the limit-of-blanks approach summarized elsewhere (24). Limits of quantitation (LOQ) and upper limits of linearity (LOL) were also determined. The LOQ was defined as the lowest concentration at which the calibration curve is linear and the coefficient of variance (CV) of replicate samples is <20%. The LOL was identified as the highest concentration of standard that remained in the linear range of the curve. Limits were established for raw sample concentrations (e.g., LOQR) and normalized to original, pre-derivatized concentrations (e.g., LOQN). Finally, analytical sensitivity was defined as the slope of the linear region of the curve divided by the standard deviation of the response variable. Statistics
Extinction coefficients and red blood cell indices for mouse blood and cell culture samples were determined as mean ± SD. Statistical significance between slope values was evaluated with GraphPad Prism 6 (GraphPad Software, LaJolla, CA). Significance between two sets of cell culture conditions was established with the two-tailed Student's t-test. Multiple treatments were analyzed with one-way ANOVA followed by Tukey's post-hoc analyses using Statistica freeware (StatSoft, Tulsa, OK). Results and discussion Detection parameters in pyridine hemochromogen, CLARiTY, and fluorescence assays of hemoglobin standards
Hemoglobin standards were used to derive calibration curves and to compare the detection parameters for the pyridine hemochromogen, CLARiTY, and fluorescence assays, including LOD, LOQ, LOL, and analytical sensitivity. The pyridine hemochromogen assay served a dual purpose in that the results were also used to check the heme content of the commercially available hemoglobin, which was not expected to contain four bound heme molecules per hemoglobin tetramer due to impurities, the unknown impact of oxidation (i.e., conversion to methemoglobin), and molecular structure deformities inherent to the lyophilization process. Based on a comparison of slope values from pooled plots of reduced minus oxidized absorbances against both hemoglobin and heme concentrations (Figure 2A), it was found that the lyophilized hemoglobin contained 2.13 heme groups per hemoglobin tetramer. The detection parameters listed in Table 1 were defined in terms of heme concentrations using this value.
Corrected CLARiTY absorbances for the hemoglobin standards at 400 nm (Figure 2B, Supplementary Figure S1B) were pooled and yielded a heme extinction coefficient (slope) of 0.266 ± 0.004 µM-1 cm-1. The CLARiTY calibration curve was sigmoidal in shape, and the linear range of response was determined by maximizing R2 values between visually identified bend points in the curve (Table 1). Although the LODR for the CLARiTY was approximately one-third the LODR for the pyridine hemochromogen assay, the LOQR for the CLARiTY was more than twice the LOQR for pyridine hemochromogen. Because the CV dropped below 20% at 0.11 µM heme, we defined LOQR for the CLARiTY system by the lowest heme concentration in the linear region of the curve. This 2-fold difference in LOQR was mirrored closely by the analytical sensitivities of the two methods.
Fluorescence heme assays also generated nonlinear calibration data across the range of standard concentrations (Figure 2C). However, unlike the CLARiTY, two of the three series maintained linearity through the origin. The slope of the line (505 ± 8 µM-1) is referred to as Kheme, a proportionality constant in the expression,