An amplification microarray combines asymmetric PCR amplification and microarray hybridization into a single chamber, which significantly streamlines microarray workflow for the end user. Simplifying microarray workflow is a necessary first step for creating microarray-based diagnostics that can be routinely used in lower-resource environments.
Simplifying microarray workflow is a necessary first step for creating MDR-TB microarray-based diagnostics that can be routinely used in lower-resource environments. An amplification microarray combines asymmetric PCR amplification, target size selection, target labeling, and microarray hybridization within a single solution and into a single microfluidic chamber. A batch processing method is demonstrated with a 9-plex asymmetric master mix and low-density gel element microarray for genotyping multi-drug resistant Mycobacterium tuberculosis (MDR-TB). The protocol described here can be completed in 6 hr and provide correct genotyping with at least 1,000 cell equivalents of genomic DNA. Incorporating on-chip wash steps is feasible, which will result in an entirely closed amplicon method and system. The extent of multiplexing with an amplification microarray is ultimately constrained by the number of primer pairs that can be combined into a single master mix and still achieve desired sensitivity and specificity performance metrics, rather than the number of probes that are immobilized on the array. Likewise, the total analysis time can be shortened or lengthened depending on the specific intended use, research question, and desired limits of detection. Nevertheless, the general approach significantly streamlines microarray workflow for the end user by reducing the number of manually intensive and time-consuming processing steps, and provides a simplified biochemical and microfluidic path for translating microarray-based diagnostics into routine clinical practice.
Early case detection and rapid treatment are considered the most effective control strategies to reduce Mycobacterium tuberculosis (MTB) transmission1, and there is now a broad consensus in the TB community that a point of care (POC) or near POC test to simultaneously diagnose TB and drug resistance (DR) is needed. Technologies such as Cepheid’s GeneXpert and other nucleic acid amplification tests reduce the time to diagnosis for many TB patients, and provide a rapid read-out indicating resistance to rifampin or selected mutations conferring resistance to other first or second line drugs2. Although real-time and isothermal nucleic acid amplification tests are designed to identify the drug resistance mutations that lead to MDR-TB, the spectrum of mutations they detect is often inadequate to design an individualized drug regimen corresponding to the drug resistance profile of the patient, and technical constraints related to optical cross-talk or the complexity of amplification and reporting chemistries3-7 may limit the number of loci or mutations that are detected. Thus, detection technologies with higher multiplexing capacity are required to address known gaps in MDR-TB POC diagnostics.
Microarrays and the WHO-endorsed Hain line probe assays can address the “multiple gene, multiple mutations” challenge of diagnosing MDR-TB8-29. Unfortunately, these hybridization-based, multiplexed detection platforms use multistep, complicated, and open-amplicon protocols that require significant training and proficiency in molecular techniques. The amplification microarray30 was designed to address some of these microarray work-flow and operational concerns. The simplifying fluidic principles are to amplify, hybridize, and detect nucleic acid targets within a single microfluidic chamber. The user introduces the nucleic acid and amplification master mix into a fluidic chamber with a pipette and starts the thermal cycling protocol. For the batch processing method shown here, microarrays are subsequently washed in bulk solution, dried, and imaged. This study demonstrates the functionality of an amplification microarray using an MDR-TB microarray test for rpoB (30 mutations), katG (2 mutations), inhA (4 mutations), rpsL (2 mutations), embB (1 mutation), IS1245, IS6110, and an internal amplification and hybridization control. At least one matched pair of microarray probes (wildtype (WT) and single-nucleotide mutant (MU)) is included for each mutation of interest. Purified nucleic acids from multi-drug resistant M. tuberculosis are from the TDR Tuberculosis Strain Bank31. Gel element microarrays are manufactured on glass substrates by copolymerization essentially as described elsewhere32, except that we use 4% monomer and 0.05 mM each probe in the polymerization mixture. Arrays are surrounded with a 50 ml gasket prior to use. After thermal cycling, hybridization, and wash steps, amplification microarrays are imaged on an Akonni portable analyzer. Background-corrected, integrated signal intensities are obtained from the raw .tif images using a fixed circle algorithm. Noise for each gel element is calculated as three times the standard deviation of the local spot backgrounds. Gene targets are typically considered detectable for signal to noise ratio (SNR) values ≥3. In order to determine the presence or absence of a specific mutation in each gene or codon, a discriminant ratio is calculated from the SNR values as (WT-MU)/(WT+MU). Discriminant ratios <0 are indicative of a drug-resistance mutation at the locus, whereas ratios >0 are indicative of the wild-type sequence.
For laboratories that follow universal PCR precautions, it is operationally more efficient to include several amplification microarrays and gaskets per substrate and wash all amplification microarrays simultaneously in a bulk container, as described here. Consumable formats are available for performing post-amplification microarray washing steps in an entirely sealed, closed amplicon test, as reported elsewhere30,33.
1. Setup
Reagent | Per- Sample Volume (µl) | Final [ ] |
Mulitplex PCR Buffer with HotStar Taq Plus | 25 | 1x |
Bovine serum albumin (BSA) | 0.55 | 0.6 mg/ml |
Formamide | 3.8 | 7.6% |
Additional Taq polymerase | 0.8 | units/µl (4 units total) |
MDR-TB primer mix | 15.75 | – |
RNase-free H2O | 2.1 | – |
Amplification/Inhibition Control | 1 | 5.0 fg/µl |
Total | 49 |
Table 1. MDR-TB amplification microarray master mix composition.
2. Load Amplification Microarrays
3. Thermal Cycling
Thermal Cycling Steps | ||
1 | 88 °C | 5 min |
2 | 88 °C | 30 sec |
3 | 55 °C | 1 min |
4 | 65 °C | 30 sec |
5 | Repeat steps (2-4) for 50 cycles | |
6 | 65 °C | 3 min |
7 | 55 °C | 3 hr |
4. Wash and Dry
5. Imaging
The asymmetric MDR-TB primer mix generates Cy3-labeled amplicons. Gel element microarrays can be imaged with any standard microarray imager capable of imaging Cy3. The following imaging procedure is specific for the MDR-TB master mix, Dx2100 field portable imager, and automated analysis software provided by Akonni.
Qualitative image analysis can provide insight into sources of experimental noise or variability that are challenging to identify in data tables generated by automated image analysis software. Thus, it can be useful to visually ascertain that 1) all gel elements are intact and undamaged, 2) the global background is free from fluorescent artifacts that might affect individual signal to noise ratio (SNR) values, 3) there is no evidence for bubble formation or nonuniform amplification/hybridization across the array, and 4) that the software accurately identified all spots on the microarray image. Example MDR-TB amplification microarray images are shown in Figure 1, illustrating a high-quality array and artifacts that may arise due to physical damage or bubbles during thermal cycling. Standard microarray image analysis software is usually able to place a grid and extract integral intensity and background data even from poor quality images such as shown in Figure 1B, so it is incumbent upon the researcher to establish quality control criteria and metrics for accepting or rejecting microarray images from further analysis. Probe redundancy may help ameliorate these issues. However, a fully automated, diagnostic MDR-TB amplification microarray reporting algorithm would otherwise declare the test from Figure 1B as “invalid”.
Amplification microarray SNR values for a dilution series of wildtype H37Ra genomic DNA are shown in Table 2. At 6.25 pg input genomic DNA per reaction or approximately 1.25 x 103 cell equivalents, all wildtype probes are readily detected. SNR values for each of the internal amplification and inhibition controls ranged from 98.48 (rpoB) to 967.24 (Akonni-supplied internal positive control), indicating that all gene targets in the asymmetric multiplex amplification reaction are amplified and detectable. No template controls were negative (SNR <3) for all probes in the amplification microarray. Genotyping ratios at 6.25 pg input DNA ranged from 0.18-1.00 for all WT/MU probe pairs, which demonstrates correct behavior of WT and MU probes and appropriate genotyping.
Representative genotyping data for a panel of World Health Organization reference isolates of known genotype are shown in Table 3. If a single nucleotide polymorphism (SNP) is represented on the gel element microarray, then the amplification microarray test accurately detects the corresponding mutation in the MDR-TB genome. Some of the amplification microarray probes are also sensitive to near-neighbor mutations that are not explicitly represented on the array as a unique probe. For example, isolate TDR-0129 contains an S531E mutation in the rpoB gene, but there is not a specific probe on the amplification microarray for S531E. Nevertheless, five other SNP probes targeting codon 531 indicate that a mutation is present at codon 531 ― the amplification microarray therefore correctly indicates that this isolate is rifampin resistant. Similar sensitivity to the rpoB S513W mutation is evident for isolate TDR-0148. On the other hand, some SNP probes are not sensitive to near neighbor mutations, as illustrated by isolate TDR-0148 and the rpoB S512G mutation, and isolate TDR-0011 and the katG S315R mutation. We suspect that this type of probe behavior is a consequence of secondary and tertiary structure in the single-stranded amplicon and/or probe34, and is not something that can be predicted a priori. Nevertheless, probe sensitivity to near neighbor mutations is analogous to the use of sloppy molecular beacons to detect multiple drug resistance mutations with a minimal set of real-time PCR probes35,36, and can be diagnostically advantageous provided the test does not generate false positives relative to the phenotypic drug susceptibility.
Figure 1. (A) Example amplification microarray image for 100 pg wildtype M. tuberculosis H37Ra genomic DNA and a 100 msec exposure time. Cy3 beacons (for automated gridding and segmentation software) are on the periphery of the image. (B) Image of an amplification microarray showing a fluorescent halo artifact resulting from bubble formation during thermal cycling, and damaged gel elements/debris. Automated image analysis software is still able to place a grid and extract data from this image. Please click here to view a larger version of this figure.
Company | Catalog Number |
Akonni Biosystems | Inquire |
Qiagen | #206143 |
Qiagen | #201207 |
Qiagen | #206143 |
Thermo Fisher Scientific, Inc. | #BP227-500 |
Sigma-Aldrich | #3B6917 |
Akonni Biosystems | Inquire |
Akonni Biosystems | Inquire |
Thermo Fisher Scientific, Inc. | #BP1328-4 |
Thermo Fisher Scientific, Inc. | #BP151-500 |
Current Technologies, Inc. | #BRSPRAY128 |
Decon Labs, Inc. | #8416 |
Company | Catalog Number |
Akonni Biosystems | 100-20011 |
Akonni Biosystems | 100-10021 |
ArrayIt | HTW |
Thermo Fisher Scientific, Inc. | #10-300 |
VWR | #3365040 |
VWR | #93000-196 |
Central Pneumatic | #95630 |
Table 2. Required Materials and Equipment.
Codon | Probe ID |
WT Probe SNR Values | ||||||
500 pg | 100 pg | 50 pg | 25 pg | 12.5 pg | 6.25 pg | |||
rpoB | 507 | 1 | 900.22 | 644.46 | 523.17 | 431.08 | 364.49 | 287.47 |
510 | 3 | 1016.37 | 699.38 | 600.16 | 525.07 | 446.51 | 361.64 | |
511 | 5 | 867.67 | 611.97 | 529.90 | 451.73 | 403.76 | 307.68 | |
512 | 8 | 810.69 | 544.04 | 488.85 | 436.89 | 391.62 | 257.46 | |
513 | 11 | 824.36 | 547.04 | 496.00 | 472.01 | 408.96 | 242.25 | |
515 | 15 | 715.48 | 536.59 | 416.96 | 335.81 | 267.48 | 189.10 | |
516 | 17 | 723.83 | 496.44 | 402.95 | 329.09 | 270.10 | 201.98 | |
522 | 27 | 674.37 | 413.33 | 360.40 | 316.75 | 236.19 | 153.40 | |
524 | 29 | 269.46 | 153.69 | 117.71 | 118.02 | 110.13 | 51.22 | |
526 | 31 | 136.37 | 109.90 | 82.63 | 76.76 | 53.61 | 37.76 | |
531 | 44 | 219.92 | 136.31 | 109.16 | 96.18 | 80.58 | 26.44 | |
533 | 51 | 130.54 | 83.60 | 76.03 | 63.47 | 61.35 | 41.54 | |
rpsL | 43 | 54 | 10.41 | 12.75 | 53.30 | 767.18 | 5.39 | 362.42 |
88 | 56 | 655.54 | 542.79 | 527.43 | 690.34 | 403.86 | 456.05 | |
katG | 315 | 58 | 1037.03 | 873.46 | 816.64 | 974.83 | 811.79 | 876.47 |
embB | 306 | 66 | 940.81 | 781.13 | 788.75 | 837.65 | 787.05 | 696.69 |
inhA | 8 | 82 | 1069.43 | 934.38 | 862.58 | 936.88 | 809.85 | 682.56 |
15 | 85 | 1111.66 | 931.88 | 918.22 | 957.87 | 795.72 | 723.16 | |
17 | 87 | 1114.49 | 926.03 | 920.60 | 970.70 | 854.15 | 744.41 |
Table 3. Signal to noise ratios for wildtype probes and a dilution series of M. tuberculosis H37Ra genomic DNA. SNR values > 3 are considered detectable.
Table 4. Representative genotyping data from MDR-TB isolates of known genotype at 25 pg per amplification microarray reaction. Asterisks identify mutations that are not represented by a unique probe on the gel element microarray. WT = wildtype nucleic acid sequence. Negative values (bold and shaded) are indicative of a mutation, hence phenotypic drug-resistance. Please click here to view a larger version of this figure.
The extent of multiplexing with an amplification microarray is ultimately dictated by the efficiency of multiplex asymmetric PCR, not the microarray. In our experience, 10-12 unique primer pairs can be readily multiplexed in an amplification microarray format. Conventional primer and probe design criteria therefore apply to new assays, except that one also needs to consider potential interactions between solution-phase nucleic acids and immobilized microarray probes, the thermal efficiency of the thermal cycler, and probe hybridization behavior in a PCR buffer that also contains PCR primers and low molecular weight amplification artifacts. Highly multiplexed amplification approaches such as whole genome amplification are not conducive to a single-chamber amplification microarray protocol for a number of technical reasons, including interference between random hexamers and immobilized probes, and inefficient hybridization of long, double-stranded amplicons. There is also an inter-relationship between ease-of-use, total analysis time, and limits of detection that individual users will need to consider when designing custom assays or using the MDR-TB test described here. For example, reducing the number of amplification cycles and even eliminating the post-amplification hybridization step will significantly reduce total analysis time, but at the expense of test sensitivity. On the other hand, achieving a reproducible 10 gene copy detection limit may require more amplification cycles or hybridization time, thus extending the total analysis time beyond a single work shift.
The genes, mutations, imager, analysis software, and reporting algorithm described here are illustrative of the amplification microarray method and system, rather than a report on their analytical performance and clinical utility. For example, the starting sample can be sputum, decontaminated sediment, solid- or liquid cultures ― the requirement for the amplification microarray is simply that the extracted nucleic acids are amplifiable by asymmetric, multiplexed PCR. Some researchers or clinicians may find little to no clinical value in detecting rpsL or embB mutations that confer resistance to streptomycin and ethambutol, respectively, only resistance to rifampin and isoniazid define MDR-TB. Thus, these PCR primers and microarray probes can be replaced with primers and probes that target genes and mutations conferring resistance to other first- or second-line antibiotics. It is possible to write a reporting algorithm to provide the user with detailed information on the specific mutations that are detected in a given specimen, rather than a simple “resistance detected” or “resistance not detected” result. For those researchers interested in using this specific assay to analyze M. tuberculosis isolates, it is also possible to provide the genotyping data even if the IS6110 element is not detected in the specimen.
Regardless of the possible assay and reporting permutations, the amplification microarray and protocol described here is designed to significantly simplify microarray workflow by combining multiple molecular biology steps into a single biochemical reaction and microfluidic chamber. The representative data demonstrate the efficacy of approach by amplifying nine MDR-TB genomic regions and detecting those asymmetric amplicons with a low-density gel element microarray. The protocol described here uses a bulk washing procedure that requires the user to physically remove the gasket and cover slip from the amplification microarray before washing, and is therefore more appropriate for research-use-only applications rather than clinical diagnostics. As described elsewhere, however, it is certainly possible to incorporate a washing step on-chip using lateral-flow fluidics33 and therefore retain an entirely closed amplicon consumable, including one that can be readily incorporated into a sample-in answer-out diagnostic system that is appropriate for point-of-care applications.
The authors have nothing to disclose.
This work was supported by the National Institutes of Health (NIH) under grant RC3 AI089106.
MDR-TB nucleic acids were provided by the United Nations Children's Fund/United Nations Development Programme/World Bank/World Health Organization Special Programme for Research and Training in Tropical Diseases (TDR), Geneva, Switzerland.
We thank Dr. Tom Shinnick of the U.S. Centers for Disease Control and Prevention for guidance on the specific genes and mutations to include in the prototype assay.
MDR-TB amplification microarrays, with applied gasket and pre-cut cover slips | Akonni Biosystems | Inquire | |
Multiplex PCR kit, containing 2X PCR buffer with HotStar Taq plus | Qiagen | #206143 | |
Taq polymerase | Qiagen | #201207 | |
RNAse-free water | Qiagen | #206143 | |
Formamide | Thermo Fisher Scientific, Inc. | #BP227-500 | |
20 mg mL-1 non-acetlyated bovine serum albumin (BSA) | Sigma-Aldrich | #3B6917 | |
5X concentrated MDR-TB primer mix | Akonni Biosystems | Inquire | |
500 fg uL-1 amplification and inhibition control | Akonni Biosystems | Inquire | |
20X SSPE | Thermo Fisher Scientific, Inc. | #BP1328-4 | |
Triton X-100 | Thermo Fisher Scientific, Inc. | #BP151-500 | |
Disinfecting Spray | Current Technologies, Inc. | #BRSPRAY128 | |
70% Isopropyl Alcohol | Decon Labs, Inc. | #8416 | |
Equipment | Company | Catalog Number | |
Microarray imager, with automated image and data analysis software | Akonni Biosystems | 100-20011 | |
Thermal cycler with flat block insert | Akonni Biosystems | 100-10021 | |
High-throughput wash station and slide holder | ArrayIt | HTW | |
Dissecting forceps | Thermo Fisher Scientific, Inc. | #10-300 | |
Mini Vortexer | VWR | #3365040 | |
Mini-centrifuge | VWR | #93000-196 | |
Airbrush Compressor Kit | Central Pneumatic | #95630 |