Quantitation of DNA Sequences in Environmental PCR Products by a Multiplexed, Bead-Based Method

A first application of a multiplexed, bead-based method is describedfor determining the abundances of target sequences in an environmentalPCR product. Target sequences as little as 0.3% of the totalamount of DNA can be quantified. Tests were conducted on 16Sribosomal DNA sequences from microorganisms collected from contaminatedgroundwater.

INTRODUCTION

Top

Introduction

We are developing a multiplexed, bead-based method involvingpolystyrene microspheres, nucleic acid hybridizations, and flowcytometry instrumentation with an aim towards environmentalfield testing. High-throughput, quantitative methods which cansimultaneously measure the abundances of many microorganismswould help us gain a broader view of microbial activity involvedwith degrading, eliminating, or immobilizing contaminants inthe environment (4, 9). When PCR is used, a critical step isthe accurate measurement of the amount of individual sequencesin the PCR product derived from community DNA. In this paper,we developed a procedure for multiplexed quantitation of PCRproducts using the bead method in a manner which can lead tohigh-throughput testing. Measurements were made on 16S ribosomalDNA sequences obtained from groundwater contaminated with TCEand TKEBS at the Lawrence Livermore National Laboratory, Well-834-D3,Site 300.

Microorganisms were collected by passing groundwater throughfilters. A phenol chloroform-isoamyl alcohol method, adaptedfrom reference 1, was used to extract DNA. The 16S rRNA genewas amplified from community DNA or plasmid DNA obtained fromclones using bacterial primers 338Fbs2 (5"[Biotin]*T*C*C*T*ACGG GAG GCA GC) and 907R (5"CCG TCA ATT CMT TTR AGT TT; OligosEtc.) (3), where 338Fbs2 was synthesized with a 5" biotin modification,12-carbon linker, five phosphorothioate bonds (*), and reverse-phasehigh-performance liquid chromatography purification (Synthegen,Inc.). The terms "PCR mix," "environmental PCR product," and"pure PCR product" refer to PCR products prepared from no-templateDNA, community DNA, and plasmid DNA, respectively. Typical concentrationswere 50 to 60 fmol/µl for a pure PCR product and 12 or13 fmol/µl for the environmental PCR product.

Capture probes for four target sequences labeled d006, d011,d023, and Geothrix (Gx) and a universal probe 533FA (3) weredesigned, synthesized, and attached to beads, as described inreference 7. The bead probes were designated plain/d006, 154/d023,138/d011, 134/Gx, and 133/533FA, where 154/d023 means that thed023 capture probe was attached to bead type 154.

A "bead mix" and single-stranded amplicons were prepared accordingto reference 7. In each tube, 17 µl of single-strandedamplicons was combined with 34 µl of bead mix containingfive types of bead probes (10,000 beads of each type) and werehybridized at 46°C for 2 h. The beads were then washed in46°C 1x TMAC buffer (7). Twelve microliters of a 20-µg/mlstreptavidin-R-phycoerythrin (Molecular Probes) solution, preparedin 1x TMAC buffer at room temperature, was added to the hybridizationmix, vortexed, and incubated at 46°C for 10 min. The beadswere then washed two times in 46°C 1x TMAC buffer and resuspendedin 100 µl of 1x TMAC buffer at room temperature.

The beads were detected with a Luminex 100 flow cytometer. Theintensity values of the reporter signals were converted intounits known as molecules of equivalent soluble fluorochrome(MESF) using Quantum 27 (R-PE) Reference Standards (Bangs Laboratories,Inc.) according to standard procedures. Cytometry data wereanalyzed with FCS Express version 1.065 (De Novo Software).The mean intensity (I_s) of the reporter signal and intersamplestandard deviation (SD) were determined by running $<=$ 7 replicatetubes. A similar procedure was used for the background signal(I_b). The uncertainty in the fluorescence response F = I_s -I_b was calculated using the standard error SD in the differenceof means (5).

The quality of the bead probes was checked as follows. The backgroundsignal for each bead probe was determined using PCR mix as theanalyte. A typical value for plain/d006 was 203 ± 25MESF with higher values for the fluorescent beads. Sequencediscrimination was evaluated by exposing bead probes to noncomplementarypure or environmental PCR products diluted with PCR mix. Nocross-hybridization signal was found.

An example of the flow cytometer output is shown in Fig. 1,demonstrating the fluorescence response of the bead probes afterhybridization to 115 fmol of environmental PCR product. A visualcomparison between Fig. 1a and b shows that the reporter signalsfor 138/d011, 154/d023, and 133/533FA in the environmental sampleare higher than the background, indicating the presence of d011and d023. For a given bead population, the standard error ofthe mean intensity was $<=$ 3% of the mean intensity; however, intersamplevariability was larger. For a measurement of abundance, thefluorescence response must be related to the actual amount ofa sequence in the PCR product.

View larger version (19K):
[in this window]
[in a new window]
FIG. 1. Fluorescence detection of the bead-probes with the Luminex 100 flow cytometer. The dot plots show the orange reporter signal (abscissa) and the red bead classification color (ordinate). (a) Analyte containing 16S ribosomal DNA from the contaminated well. (b) Negative control (PCR mix).

Since the hybridization efficiency between probes and targetscan vary considerably, standard concentration curves were developedfor each probe. We have found that the concentration responsein the environmental PCR product is different from that in thepure PCR product. Therefore, quantitation was accomplished byadding known amounts (C_i) of a target (measured in, for example,fmoles) to a fixed amount of environmental PCR product, containingan unknown amount (C_e) of that target. The total amount of targetin the analyte was therefore C = C_e + C_i.

Model.

Top

Model.

To determine C_e, a model equation was developed. Estimates inreference 7 showed that not more than a few tenths of a percentof the total amount of target DNA in the analyte actually hybridizeto capture probes on the bead surface. In this case, the fluorescenceresponse of a specific target sequence may be described by aScatchard-type equation for a unireactant system (probe-target)(6): F_i = [k_aI_max(C_e + C_i)]/[1 + k_a(C_e + C_i)]) (equation 1),where F_i is the fluorescence response of a particular targetin the sample (with background subtracted); k_a = K_a/V; K_a isthe association constant; V is the reaction volume, and I_maxis the maximum response corresponding to complete saturationoccurring at k_a(C_e + C_i) >> 1. In practice, k_a, I_max, and C_emust be determined experimentally. Equation 1 represents a systemof equations where i = 0, 1, 2, 3... This model assumes (i)that there is no cross-hybridization, consistent with our observations,and (ii) that parameters k_a and I_max are constant for all C_i.

At minimum, three spiking values are required to determine C_e,k_a, and I_max. When C_o = 0 (no spike), equation 1 yields a simple,exact solution: C_e = [ ${alpha}$ C₁C₂]/[C₂(1 - ${alpha}$ ) - C₁] (equation 2), where ${alpha}$ = F₀(F₂ - F₁)/[F₁(F₂ - F₀ )]. Parameters I_max, k_a, and C_e canalso be determined by directly fitting equation 1 to the experimentaldata {C_i, F_i}.

Verification of model.

Top

Verification of model.

The model was tested by spiking equimolar mixtures of pure PCRproducts for d006, d011, and d023 and Gx into 106 fmol of environmentalPCR product, using PCR mix as diluent. The C_e value for eachsequence was calculated from equation 2 using C_i = 0, 8, and16 fmol. The uncertainties in C_e were determined from S_D andconventional error propagation formulas (2). C_e was found tobe 10.0 ± 4.7 fmol (9.4%) for d023, 4.3 ± 1.2fmol (4.1%) for d011, and 0.31 ± 0.17 fmol (0.3%) ford006, where the percentage refers to the relative abundanceout of 106 fmol of environmental PCR product. In the case of138/Gx beads, F₀ was -16 ± 24 MESF, indicating that thissequence was not detected (0%) in the environmental PCR productwithin the error. Except for d023, the relative abundances ofd011, d006, and Gx were in agreement with cloning and sequencingresults based on 63 clones. The cloning results were 28.6% (17.9to 41.4) for d023; 4.8% (0.99 to 13.3) for d011; 1.6% (0.04to 8.53) for d006; and 0% (0.00 to 4.64), where the bracketednumbers represent the 95% confidence interval determined byusing Minitab software (J. H. Klotz, L. M. Leemis, and K. S.Trivedi, Letter, Am. Statistician 50:388-389, 1996).

Figure 2 illustrates the experimental data {C_i, F_i} for d023and a weighted, nonlinear fit of equation 1 using the calculatedC_e to obtain k_a and I_max (Table 1). For this and other sequences,the fits using the model equation are excellent.

View larger version (19K):
[in this window]
[in a new window]
FIG. 2. Application of model to determine the amount (C_e) of d023 in the environmental PCR product (106 fmol). F_i, indicated by ${circ}$ , is shown for C_i at 0, 2, 8, and 16 fmol. C_e was determined from equation 2 by using errors given by S_D. The solid line is a weighted, nonlinear fit to equation 1. The "+" symbols and the dotted line are the F_i and fit for pure d023 PCR product diluted in PCR mix.

View this table:
[in this window]
[in a new window]
TABLE 1. Fitting parameters associated with the standard curve for each target sequence in an analyte consisting of either pure or environmental PCR product^a

The values of k_a and I_max are different in the pure and environmentalPCR products, indicating different hybridization conditions(see dotted and solid curves in Fig. 2). Concentration curvesfor pure PCR products were obtained with equimolar mixturesof d023, d011, and d006 and Gx diluted in PCR mix. Parametersk_a and I_max were calculated by a weighted, nonlinear fit ofequation 1 with C_e = 0 (Table 1). In general I_max is greaterin the pure PCR product than in the environmental PCR product.As shown in Fig. 3 for d006, the excellent fit to the experimentalconcentration curve spanning 3 orders of magnitude in the amountof target DNA in the tube again supports the model equationfor this system.

View larger version (16K):
[in this window]
[in a new window]
FIG. 3. Concentration curve of pure d011 PCR product diluted in PCR mix. The reporter signal I_s is displayed on the ordinate axis indicating that I_b = 348 ± 51 MESF for 138/d011 beads. The solid line is a weighted, nonlinear fit to the model. The error bars represent intersample SD.

We sought further validation of the model by comparing our valuesfor the dissociation constant K_d (= 1/K_a) to those in the literature.In our experiments the K_d values for a 570-nucleotide targetDNA hybridized to a 20-mer capture probe ranged from 2.5 x 10^-9M for d006 to 8.7 x 10^-9 M for Gx. Detailed studies (8) of hybridizationson microparticle surfaces between a short target (22-mer) anda particle-bound 22-mer capture probe obtained K_d values of $~$ 10^-9 M at 50°C. Our values are similar but somewhat higher;we speculate that this may be due to a decrease in hybridizationefficiency for long-target DNA. We note, however, that the designof the capture probe can affect the hybridization response morethan the target sequence length.

As expected, there is substantial variation of the coefficientsk_a and I_max for different target sequences, most likely dueto secondary structure. In practice, for assays targeting abroad array of microorganisms, it will be impossible to designprobes with the desired specificity and identical coefficients.Good quantitation will therefore require separate determinationof k_a and I_max for each control strain. Correspondingly, finequantitation with universal probes is probably not achievable,as demonstrated by the data for 133/533FA beads in which thetotal amount of DNA was found to be 17.6 ± 10.5 fmol,compared to 106 fmol determined by gel-based quantitation.

The coefficients k_a and I_max are different between an environmentalsample and a pure PCR product. Cross-hybridization is unlikelyto be the cause since none was observed in our experiments.(For example, the Gx capture probe subjected to noncomplementarypure or environmental PCR products produced no signal.) At thistime, we suggest two possible causes: (i) chemicals in the groundwateror extraction process which reduce the efficiency of hybridizationor labeling and (ii) many similar sequences in the PCR productwhich interfere with each other. Tests on more types of environmentalsamples will be required to elucidate this point.

In summary, we have developed and verified a quantitative methodfor multiplexed determination of abundances in environmentalPCR products.

ACKNOWLEDGMENTS

We acknowledge B. Methe and D. Lovley for providing Geobacterand Geothrix clones. This work was supported by DOE NABIR grantDE-FG02-99ER62868.

FOOTNOTES

* Corresponding author. Mailing address: Physics Department, Loyola College in Maryland, 4501 N. Charles St., Baltimore, MD 21210. Phone: (410) 617-2709. Fax: (410) 617-2646. E-mail: mlowe{at}loyola.edu.

REFERENCES

Top