Inhibition of on-chip PCR using PDMS–glass hybrid microfluidic chips

Abstract

In the course of developing a microfluidic analytical platform incorporating the polymerase chain reaction (PCR) and subsequent capillary electrophoresis (CE) analysis for a variety of bio-assays, we examined PCR inhibition through surface interactions with the chip materials. Our devices perform PCR in a three-layer chip, a glass–poly(dimethylsiloxane)–glass sandwich in which the poly(dimethylsiloxane) (PDMS, a silicone rubber) layer is used for pneumatic membrane pumping and valving of the PCR reagents. Initial on-chip PCR–CE tests of BK virus replicated in multiple uncoated chips showed variable results, usually yielding no detectable product at the target sample concentrations used. Subsequent “chip-flush” experiments, where water or reagents were flushed through a chip and subsequently incorporated in off-chip PCR, highlighted bovine serum albumin (BSA) amongst other pre-treatments, chip materials and PCR recipes as being effective in mitigating inhibition. When the BSA channel pre-coating was applied to on-chip PCR–CE experiments, a substantial improvement (10× to 40×) in signal-to-noise (S/N) of the CE product peak was conferred, and was shown with high confidence despite high S/N variability. This is the first study to quantitatively examine BSA’s ability to reduce inhibition of PCR performed on PDMS chips, and one of very few microfluidic PCR inhibition studies of any kind to use a large number of microfluidic chips (~400). The simplicity and effectiveness of our BSA coating suggest that passivating materials applied to microfluidic device channel networks may provide a viable pathway for development of bio-compatible devices with reduced complexity and cost.

Appendices

Appendix 1

Product peak quantitation, described briefly previously, is described next in more detail and illustrated in Fig. 2. Plot A of Fig. 2 shows a typical electropherogram for the PCR output, with a peak for each of the primer and product peaks; plot B shows the area of interest for S/N calculations at and before the product peak for an on-scale peak; plot C shows the same as plot B but for an off-scale product peak; and plot D shows a magnification of the baseline segment used to determine both peak height and magnitude of noise. After manually bracketing the time window for each product peak, all data processing was automated using Microsoft Excel 2003 or 2010 spreadsheet functions to remove human bias. Electropherogram data were loaded in Excel directly from the μTK exported text files without smoothing or alteration. For each electropherogram, the product peak maximum was located, and 2 s of baseline data, from 7 to 5 s before the peak maximum, was used to determine the baseline DC and noise magnitude (standard deviation or σ) values. Because the baseline was sloping down from the primer peak tail, the 2 s of baseline data was regressed linearly and the slope subtracted to produce ‘corrected baseline’ data (Fig. 2d). The DC value of this zero-slope corrected baseline was set to match the value at 5 s before the peak maximum, i.e. the data at 7 s before the peak were corrected (lowered) the most, while that at 5 s before was not corrected at all. The signal was calculated as the height difference between the peak maximum and corrected baseline, while the noise was calculated as the standard deviation of the corrected baseline.

In many instances, strong signal produced off-scale electrophoretic peaks. For these peaks, dual calculations were performed to generate both minimum and estimated S/N values based on minimum and estimated signal values (noise values were as described above). The minimum signal value was as described above, where the clipped peak maximum value (5 V) was used to calculate the peak’s understated signal value. The estimated signal value was determined as the difference between the corrected baseline and the interpolated peak maximum. The interpolated peak maximum was calculated as 84.1 % of the height from the corrected baseline to the intersection of linear extrapolations of the clipped peak’s front and tail. The peak’s front and tail were linearly regressed from 65 to 95 % of on-scale peak height to reduce the impact of product peak tailing on the signal estimate. Additionally, a reducing factor (0.841) was required to account for the fact that peaks are nominally Gaussian in nature, not triangular, and thus the extrapolation intersection just described is an over estimate of the peak maximum. The 84.1 % fraction is the height of an ideal Gaussian peak relative to the height determined by the method above: the intersection of linear extrapolations (from 40 to 60 % of peak height) of the peak’s front and tail (data not shown). Comparison of this graphically interpolated estimate of peak height to actual peak heights for on-scale peaks showed the interpolated values to be conservative in all cases.

Appendix 2

The statistical approach (Snedecor and Cochrane 1980) used to evaluate the distinctness of averages in a comparison of Table 3 data groups, some of small sample size and most highly variable, is described next. We wish to determine whether two average S/N values, A and B, with corresponding standard deviations s _A and s _B and sample sizes N _A and N _B, are statistically different from each other at a chosen confidence level (CL), assuming a normal (Gaussian) distribution of errors. If they are distinct from each other at that CL, then the following inequality applies:

$$ \left| {A - B} \right| > t_{\text{CL,DF}} \times \sqrt {\frac{{s_{\text{A}}^{2} }}{{N_{\text{A}} }} + \frac{{s_{\text{B}}^{2} }}{{N_{\text{B}} }}}. $$

(1)

where t _CL,DF is the Student’s t value for the chosen two-sided CL and number of degrees of freedom, DF, for the difference between the two averages. The latter is evaluated as:

$$ {\text{DF}} = \frac{{\left( {\frac{{s_{\text{A}}^{2} }}{{N_{\text{A}} }} + \frac{{s_{\text{B}}^{2} }}{{N_{\text{B}} }}} \right)^{2} }}{{\frac{{\left( {\frac{{s_{\text{A}}^{2} }}{{N_{\text{A}} }}} \right)^{2} }}{{\left( {N_{\text{A}} - 1} \right)}} + \frac{{\left( {\frac{{s_{\text{B}}^{2} }}{{N_{\text{B}} }}} \right)^{2} }}{{\left( {N_{\text{B}} - 1} \right)}}}} $$

(2)

where DF is rounded down to the nearest integer value.

In practice, we evaluated all the data group comparisons shown in Table 4 at each confidence level (50, 60, 70, 80, 90, 95, 98, 99, 99.5, 99.8 and 99.9 %), and reported the highest CL for which A and B could be considered distinct per Eq. (1) as a measure of the strength of the distinction.