PathogenMip Assay: A Multiplex Pathogen Detection Assay

The Molecular Inversion Probe (MIP) assay has been previously applied to a large-scale human SNP detection. Here we describe the PathogenMip Assay, a complete protocol for probe production and applied approaches to pathogen detection. We have demonstrated the utility of this assay with an initial set of 24 probes targeting the most clinically relevant HPV genotypes associated with cervical cancer progression. Probe construction was based on a novel, cost-effective, ligase-based protocol. The assay was validated by performing pyrosequencing and Microarray chip detection in parallel experiments. HPV plasmids were used to validate sensitivity and selectivity of the assay. In addition, 20 genomic DNA extracts from primary tumors were genotyped with the PathogenMip Assay results and were in 100% agreement with conventional sequencing using an L1-based HPV genotyping protocol. The PathogenMip Assay is a widely accessible protocol for producing and using highly discriminating probes, with experimentally validated results in pathogen genotyping, which could potentially be applied to the detection and characterization of any microbe.


INTRODUCTION
Multiplex assays to identify the presence of a wide variety of microbes within a specimen will have a profound impact on the efficient management of disease treatments and prevention, clinical follow-up studies, and the development of new therapies of prophylactic or therapeutic vaccines. Here we describe a complete protocol for a multiplex assay in single-tube reactions useful in pathogen diagnostics. We compare our results to those obtained from a conventional general-primer based amplification [1] approach when genotyping primary tumor genomic DNA extracts from cervical cancers.
Nucleic-acid based diagnostic techniques typically rely on unbiased PCR amplification [2] of the targets, which requires a set of general PCR amplification primers to hybridize upstream and downstream of the variable region, followed by either realtime PCR [3], hybridization assays [4] or template sequencing [5] for read-out interpretation. Multiplex PCR [6], has the potential to decrease cost, time and effort in pathogen diagnostics [7]. Engineering an efficient multiplex PCR requires laborious strategic planning in primer design, nucleotide concentrations, optimal salt and buffer conditions, and use of chemical adjuvants [8] and is rarely capable of achieving multiplex degrees greater than a 20-plex.
The field of molecular diagnostics is experiencing a revolution in regards to theranostics, the integration of diagnostic technologies with therapeutic applications [9]. As the number of biomarkers for microbial agents and disease markers [10] increases, simultaneous detection of multiple agents or risk factors implicated in particular clinical syndromes or diseases that share similar epidemiological features will become highly desirable. Groundbreaking research has been performed with the advent of a ligase-mediated gene detection technique [11] and the many technologies based on the work of Landegren et al [12]. Ligasebased technology has led to myriad promising techniques such as the padlock probe [13,14], and the proximity ligation assay [15], which have both been investigated as potential pathogen diagnostic methods [15,16]. Molecular Inversion Probe (MIP) technology [17], has contributed new features to padlock probes, resulting in an ultra high-throughput method for SNP detection [18] and further investigation as a potential quantitative method [19].
As a model assay for MIP pathogen diagnostics, we chose to target the Human Papillomavirus (HPV), which is well known for its cancer-associated cervical infections and for the existence of multiple genotypes [20]. Based on the oncogenic potential, the subtypes are classified as ''high-risk'' or ''low-risk'' HPVs. We developed a 24-plex PathogenMip assay to target HPV genotypes commonly associated with cervical cancer progressions [21]. The MIP assay was performed in four sequential enzymatic driven reactions ( Figure 1). Target recognition sites for the MIPs in this assay were designed using a previously described software tool, PathogenMIPer [22]. We also attempted to introduce a novel ligase-dependent probe production at equal or increased assay effectiveness, which would be less expensive than direct manufacture of full-length probes. Pyrosequencing [23] and hybridization [24] were applied to the HPV amplified L1-region to validate the results obtained from the PathogenMip Assay.

Ligation-based probe construction
For cost-efficient production of MIPs, two shorter DNA constructs (denoted A and B) were produced in-house and ligated through a ligation-based production assay ( Figure 2). Several different assay protocols were examined, in which: i) dCTP, dTTP and a polymerase (Stoffel fragment) were added to the ligation reaction, ii) bridge concentrations were varied from ratios of 2:1 down to 1:4 with respect to the concentrations of the A and B constructs, and iii) Ampligase was replaced with a T4 DNA ligase based kit (Fast-Link DNAligation kit, Epicenter Biotechnologies; Madison, WI). Gel-electrophoresis was used as read-out for all optimizations (data not shown).
Thus, best results with regards to yield and purity were achieved by inclusion of Stoffel fragment, dCTP and dTTP, and use of equal concentrations of bridge, A, and B constructs. No noticeable differences were seen with the use of different ligase-enzymes, and Ampligase was used in subsequent tests. Using this protocol, the ligation step produced clear DNA gel-bands at a 100 base pairs, which were isolated and purified. All 25 probes (including the reference probe, rMIP, which targets the b-globin region in human DNA) were pooled together into a mixture of roughly equal probe concentrations with a total concentration of 5 fmol/probe.

PathogenMip Assay validation
Our primary goal was to specifically genotype each target while using a complete 25-plex MIP mixture (24 HPV probes plus the control rMIP), without generation of false-positives or negatives. To enhance sensitivity and selectivity, we explored a variety of initial probe concentrations and thermoprofiles for the initial ligation and gap-filling step ( Figure 1B). All optimizations of the PathogenMip Assay were based on UV-visualization of DNA band intensities after size separation by gel electrophoresis (data not shown). The initial probe concentrations yielded an appropriate signal when using 5 fmol of each contributing probe. Cycle times Figure 1. Schematic overviews of molecular inversion probe technology. A) Synthetic oligonucleotide containing following four regions; i) H1 and H2: homology regions comprised of unique continuous 40-50 base pair fragments for target recognition ii) BARCODE: molecular barcode comprised of a 20 base pair DNA tag for target identification iii) U1 and U2: universal primer regions for inverted probe amplification, and iv) R: restriction site for probe linearization. B) Upon target recognition, a DNA polymerase fills the missing gap in between the juxtaposition of the probes' flanking ends, and through the activity of a DNA ligase the probe is circularized. In all cases the missing nucleotide is a ''G''. C) Circular DNA enrichment through DNA degradation by enzymes Exonuclease I and III. D) Probe linearization restriction site cleavage. E) All reacted and inverted probes are amplified with universal primers, of which one is biotinylated for subsequent amplicon validation. doi:10.1371/journal.pone.0000223.g001 Figure 2. Ligase-based probe construction for the PathogenMip Assay. Seen in the figure is the sequence for probe-16 prior to inversion. Two shorter fragments were synthesized and hybridized to a bridge complementary to the universal primer regions, common for all probes used in the assay. The coupling of the shorter fragments was mediated by polymerase and ligase activity to achieve maximum yield. doi:10.1371/journal.pone.0000223.g002 and annealing temperatures in the initial reaction step yielded an efficient protocol, comprising 5 cycles of denaturing and two annealing times at 50uC and 37uC.
Using the optimized conditions, all 21 HPV plasmids and 3 oligonucleotide targets (see Materials and Methods, DNA materials) were detected with the PathogenMip Assay, and sequencing individually validated each result. The positive control, rMIP, was successfully validated through a reaction with a commercially available human genomic DNA, without positive matches to any of the HPV genotypes.

PathogenMip Assay performance
To further investigate assay sensitivity, we used a fixed concentration of the MIP mixture at 5 fmol/probe in the ligation step, while varying the amount of target DNA. This experiment was performed in two versions. The first set of reactions contained pure plasmids. The second set of reactions contained, in addition to the target plasmids, a constant amount (200 ng) of non-reactive, commercially available human genomic DNA. By spiking with human genomic DNA we hoped to approximate more realistic sample environments for which the assay was constructed. So that the experiments would be of significant value, the probe mixture did not include the rMIP, since rMIP would react with the human genomic DNA and give false quantification data. The plasmid concentrations ranged from 100 ng to 10 fg through a 10-fold dilution series. The original detection limit goal for the assays was 1 pg in the pure plasmid environment and 1 ng for plasmid in the presence of 200 ng human genomic DNA (data not shown), when cycling the PCR reaction at 35 cycles. By using an alternative PCR protocol, working at a higher annealing temperature (58uC) for 60 cycles, we were able to improve the sensitivity of the assay. No background was observed with either of the PCR conditions, but a 10-fold higher sensitivity was obtained with the more extensive 60 cycle-protocol. The new detection limit values were 100 fg for the pure plasmid and 100 pg for plasmid in the presence of 200 ng human genomic DNA (data not shown). Targets in all reactions were correctly typed as judged by L1-pyrosequencing validation.
In multiple HPV-infections, the viral load of each contributing genotype can differ by orders of magnitude [5]. To obtain a deeper understanding of the assay's capabilities when faced with such challenges, we set up a series of reactions to mimic this possibility. Here we chose a two-plasmid detection model in which plasmid HPV-45 was present at a fixed amount of 100 ng, while the second plasmid HPV-59 was present in a wide concentration range. In this scenario, we were able to obtain interpretable data for both genotypes for up to 2-logs difference in target DNA concentration, using the PathogenMip Assay on the different in parallel with pyrosequencing to confirmresults ( Figure 3).

Genotyping of primary tumor genomic DNA extracts
The ultimate objective was to screen samples with unknown viral load. A total of 20 commercially available primary tumor genomic DNA extracts were collected from patients with cervical cancers. The DNA samples were screened through the HPV PathogenMip Assay and detected using both methods (pyrosequencing and the in-house barcode-chip assay). The results were validated by genotyping the samples in parallel using the conventional general primer based nested PGMY09/11 and GP5+/6+ amplification protocol. For all tumor-drived samples, the agreement among the methods (PathogenMip Assay and conventional genotyping) was 100%, with 10 samples positive for HPV-16, 5 for HPV-18, two for HPV-45, two for HPV-59 and one sample returning as HPVnegative. When using 200 ng of genomic DNA extract for the PathogenMip assay, equal levels of detection efficiency were obtained for both methods (Figure 4).

Multiplexing capabilities
We also performed the PathogenMip assay screening a mixture of multiple HPV plasmids set at equal concentrations of a 100 ng. Many patients have multiple HPV infections and, though rare, such patients are known to carry as many as five subtypes [5], so we set as our goal to prove that it is feasible to detect four simultaneous infections at equal viral loads using the PathogenMip Assay. HPV plasmids of four high-risk HPV genotypes (-16, -18, -34, and -59) were pooled in equal amounts at a 100 ng/plasmid, and screened simultaneously in triplicate using the complete 25plex-probe mixture. The reactions were then validated by both pyrosequencing and hybridizations methods. All four HPV subtypes were detected at similar detection intensities differing by 620% from the median value, when hybridized through the barcode to an in-house barcode-chip ( Figure 5).

DISCUSSION
A complete pathogen detection assay based on MIP technology is under construction. Here, we describe a successful model assay targeting 24 HPV genotypes most commonly associated with the progression to cervical cancer. Our pilot subset can now be extended to include all known sequenced HPV genotypes, as multiplexing up to 20,000 using the MIP method for human SNP genotyping, has been demonstrate to be feasible (www.affymetrix. com/technology/mip_technology.affx), and this would more than cover the existing variability in the HPV genome.
In the case of a double HPV infection, the fact that the PathogenMip Assay detected and distinguished between two contributing viral loads that differed by 2-logs at an input level of 200ng genomic DNA is quite promising. While DNA yields from tumors vary widely, this level is in the mid-range of recovery in our experience. Exploration of a more full matrix of DNA doubleloads and inputs is under way.
Primary tumor genomic DNA extracts were efficiently genotyped for a wide dynamic range of viral loads ( Figure 4). The reference probe (rMIP) provided a positive reference control in each performed reaction ( Figure 4) and, if used to normalize viral loads against total human genomic DNA content, may allow the researcher a tool to compare viral loads from genomic DNA extracts. The results of the PathogenMip Assay genotyping were in 100% agreement with the conventional genotyping methods.
Our goal was to provide a complete, experimentally validated assay in a ready-to-use format, which could be applied generically to any pathogen based on genome target recognition. Pathogen-Miper [22] serves as an in silico multiplex test system, whereby one can check the feasibility of designing a MIP assay for the specific targets one might want to include in a multiplex detection assay.
Microbial genomes commonly incur variations from insertions, deletions, or mutations that can generate false-negative results if the variation resides within a probes' targeted region. A more complete diagnostic approach would be to include multiple probes per genotype; this would also provide deeper epidemiological insights in functional genome patterns and variations, i.e. mutations coding for drug resistance [25]. Conventional PCRbased genotyping methods are restricted to reside within conserved genomic regions, which limits the choice of genotyping targets [1,26]. These methods are not robust to genomic changes, a problem that can be at least partially addressed by the multiplexing possible in the PathogenMip Assay.
In the current incarnation of the assay, we have incorporated a ligation step to facilitate probe generation. An oligonucleotide complementary to the universal primer region, which is present in each probe, can be used as a bridge so that one longer probe can be ligated together from two smaller constructs, and length requirements for probe synthesis are therefore cut in half. Longer oligonucleotides result in higher initial costs, and have been known to compromise sequence quality [27]. With our new ligation-based probe production, we introduce cost-efficient probe production with possibilities for innovation and potential increase in efficiency. Probe production using the ligation method can easily be performed in any academic or other research institution with budget constraints, using only standard labware and avoiding core facility based research. We are currently investigating parameters for increased ligation efficiency and probe production. The goal of such efforts would be a more homogeneous distribution of probe concentrations, thus enabling one to perform high-throughput production via one-tube reactions and use of fully automated workstations, e.g. Biomek FX (Beckman Coulter, Palo Alto, CA). Parameters that could be varied to this end include ligation at higher temperatures (preferred 65uC) to allow improved incubation profiles, and higher cycle numbers [28]. Use of modified nucleotides, e.g. locked nucleic acids (LNA) [29] in the bridge would allow for higher melting temperatures (T m ) for the universal primer regions. Also by replacing the conventional gel-purification protocol with one based on single-strand DNA separation, higher purity of ligated versus non-ligated DNA can be achieved, resulting in a more sensitive assay (data not shown). The introduction of a second universal primer region would require a ligation-based production assay based on three contributing constructs and two bridges complementary to the two universal primer regions. By designing the new probes so that the two primer regions flank the barcode region, a three-way ligation currently under investigation would result in new design features regarding assay possibilities and efficiency.
Although our preferred readout for the PathogenMip Assay uses sequencing or hybridization of the identifying barcode, we also used sequencing of the target-homologous regions of the PathogenMip inverted probes to double check target identity. In addition, we performed conventional HPV identification through the sequencing of the L1 regions. All three assays (PathogenMip/ barcode, PathogenMip/target region, and HPV L1 sequencing) produced the same final result with respect to target identification, regardless of whether the readout was performed by pyrosequencing or hybridization methods. Pyrosequencing of the ID-tag was more time and cost-efficient than either hybridization to the inhouse barcode-chip or the L1-hybridization-assay. Multipleprimer pyrosequencing, however, is limited to reactions of 20plex or lower. A higher degree of multiplexing in a sequencingbased approach might be achieved through use of a highthroughput configuration of the pyrosequencing method; for example, the one offered by 454 Life Sciences (http://www.454. com/applications/index.asp). However, when high capacity is required, hybridization-based methods show more potential than sequencing methods, and here the PathogenMip Assay has the advantage of providing more discrimination through its targeted probes than would a standard hybridization to highly conserved regions. The use of ID-tag markers incorporated in the probes simplified sequence read-out in the current PathogenMip Assay, but a more forward looking approach involves refinement of the in-house barcode-chip hybridization piloted here. We anticipate that a high-throughput configuration of the current PathogenMip assay will be achievable through use of the commercially available Affymetrix TAG4-array [30], which is based on detection of barcodes through hybridization similar to the process described here. The TAG4-array covers a total of 80K features, which includes a total of 16,000 unique barcodes in five replicates.
Sensitivity levels obtained with the current assay were sufficient for a functional assay and were improved with the use of a more extensive amplification protocol. Previously described ligationbased genotyping methods [16,31] have shown sensitivity ranging down to nanograms of genomic DNA for accurate genotyping. These methods have placed careful emphasis on the amplification step by either rolling circle amplification method [32] or stringent reaction cleanup with reversible magnetic biotin-streptavidin separation [33], followed by design of a PCR run at high annealing temperatures and high cycle numbers, which also could be performed with the use of LNA universal primers [34]. Further consideration of these parameters should allow improvement in sensitivity of a next-generation PathogenMip Assay.
The potential uses for an assay that simultaneously detects multiple organisms based on differences in DNA sequences are almost unlimited. MIP-based diagnostics have already progressed in the field of human SNP-detection and cancer diagnostics [10] with ongoing clinical trials. Microarray techniques have gained ground in the pathogen diagnostic arena, with various genotyping chips such as the HPV-assay PapilloCheck (www.greinerbioone. com) upcoming. By combining the benefits of a high-throughput multiplex MIP-assay prior to PCR amplification and of a post-PCR hybridization based chip read-out, researchers can expand the scope of their surveillance to reach new insights into epidemiology and clinical management. We therefore encourage researchers with expertise in various pathogenic microbes to adapt the PathogenMip Assay to accomplish similar detection screenings as performed in this study. By combining the different models, we can reach a common goal toward a standardized virus-, bacteria-, or fungi-chip for routine diagnostics and epidemiological studies.

Experimental design
Our goal was to design an assay to simultaneously distinguishing among over 100 HPV genotypes that differ only slightly in their genomic sequences (up to 90% homology). The critical regions of the probes that recognize and distinguish among the HPV subtypes were generated using PathogenMIPer, a previously described software tool [22]. Each target sequence was designed through multiple, successive steps of evaluation of candidate sequences, based on user-defined criteria, followed by a BLAST search against non-target genomes potentially present in the sample that could otherwise cause background noise or interfere with signal. For the study, we included 24 HPV genotypes (Table 1 and Figure 6A) most commonly associated with cervical cancer progression [20]. Our design strategy allowed the probes to target any region of the approximately 8000 bp viral genome. Probes that targeted the E6/E7 genes were preferred, since these genes are most commonly associated with cervical cancer progression [35]. The HPV-PathogenMip Assay also included a reference probe (rMIP) targeting the b-globin region in human DNA (Table 1). Validation was performed by either ID-tag based multiple-primer DNA sequencing [5] or barcode hybridization, using our in-house microarray chip setup, the in-house barcodechip. Each ID-tag was carefully designed so that it was easily distinguishable when up to four probes were sequenced in parallel during the validation reaction, without primer cross hybridization (Table 2 and Figure 6B). A conventional genotyping assay was run in parallel, based on nested PCR amplification ( Figure 6C). The PCR amplicons were followed by sequencing or hybridization interpretation ( Figure 6D-E).

DNA materials
All oligonucleotides used in the assay (Table 3)  . The HPV plasmids were normalized at 100-ng/ml using a ND-100 Spectrophotometer (NanoDrop, Wilmington, DE). Because we lacked plasmids containing HPV subtypes 58, 68, and 69, we manufactured synthetic oligonucleotide targets, denoted as TEMP-58, -68 and -69 to complement the homologous regions of our probes for these genotypes (Table 3B). Twenty commercially available primary tumor genomic DNA extracts were obtained from Oncomatrix, Inc. (www.oncomatrix.com). Commercially available human genomic DNA G3041 (Promega, Madison, WI) was also used for assay validation.

Ligation-based probe construction
The full-length probes were designed to be of ,110 base pairs. Each probe was synthetically manufactured in two shorter fragments, A and B, at about 50 base pair lengths, and with 59phosphorylated modifications (Table 3A). HPV-complementary portions of the probes were designed using PathogenMIPer, a previously described software tool for designing unique and specific probes, targeting any type of genome based on its sequence data in a FASTA file format [22]. The universal primer region was chosen for the point of the A and B fragment separation, 10 base pairs upstream (59-39 directional viewpoint) of the first dUTP residue. A complementary bridge oligonucleotide was designed (Table 3B) (Fermentas, Hanover, MD). A GeneAmp PCR system 9700 (Applied Biosystems, Foster City, CA) thermocycler was used for initial heating at 95uC for 10 min, followed by 5 cycles of denaturation and hybridization/ligation at 95uC for 2 min and 50uC for 2 min. The ligation products were then run out on a Sub-Cell GT Agarose Electrophoresis System (Bio-Rad, Hercules, CA) using a 2.5% Ultra pure L.M.P Agarose gel (Invitrogen, Carlsbad, CA) with 5 mg/ml Ethidium Bromide (Sigma-Aldrich, St. Louis MO) staining. Gel bands of fragments at about a 100 bases were carefully cut out and dissolved in 0.4 M NaCl at 70uC. DNA extraction of the gel bands was based on a phenol and chloroform (Sigma-Aldrich, St. Louis, MO) organic extraction protocol. The aqueous phase was collected after organic phase separation with equal volumes of: i) phenol, ii) 1:1 phenol/chloroform, and iii) chloroform. This procedure was followed by ethanol precipitation at 220uC over night. Glycogen (Sigma-Aldrich, St. Louis, MO) was added to facilitate DNA precipitation. The purified probes were normalized for probe concentration using a ND-100 Spectrophotometer (NanoDrop, Wilmington, DE).

PathogenMip Assay
Prior to the assay, genomic DNA was fragmented by digestion with EcoRI and Xba I according to manufacturer's instructions  For each probe, information is provided regarding which HPV genotype (or human gene) it targets, NCBI accession numbers, and the genes where the recognition sites lie.  (Table 3B). A 10-min incubation step at 95uC was followed by 35 cycles of amplification with a thermocycler GeneAmp PCR system 9700 (Applied Biosystems, Foster City, CA). Each cycle included a denaturation step at 95uC for 45 sec, and an annealing step at 52uC for 30 sec. A final extension was done at 72uC for 5 min. For three randomly chosen reactions, an alternative PCR protocol was run in parallel, which was comprised of 60 cycles with an increased annealing temperature at 58uC. The latter protocol was used to test whether increased sensitivity of the assay could be achieved with a more extensive PCR amplification.

Multiple-primer DNA pyrosequencing of barcodes
The B-F-CHPV-U1 primer used for universal probe amplification was biotinylated, and the leading strand could therefore be used as a sequencing template. Single strand template preparation was performed as previously described [23], and oligonucleotides complementary to barcodes included in the probes were used as sequencing primers (Table 3B). Each of the 24 HPV targeting probes contained one out of four ID-tags, which lay the basis for a rapid multiple-primer based screening by sequencing [5]. The primers were pooled into six mixes of four primers with different ID-tags (Table 2). Pyrosequencing was performed with a cyclic de novo sequencing dispensation (ACGT) using a PSQ TM HS96A DNA sequencing system.

Microarray barcode-chip template preparation
For more efficient microarray hybridizations, single stranded DNA was generated from the first PCR amplification. Amplification was carried out in a total reaction volume of 50 ml, containing 1 ml from the first PCR reaction, 0.1256 BD Titanium Taq DNA polymerase (Clontech, Mountain View, CA), 1X BD Titanium Taq PCR buffer (Clontech, Mountain View, CA), 0.2 mM dNTPs (Fermentas, Hanover, MD) and 0.2 mM of each universal primer B-F-CHPV-U1. A 10 min incubation step at 95uC was followed by 35 cycles of amplification with a thermocycler GeneAmp PCR system 9700 (Applied Biosystems, Foster City, CA). Each cycle included a denaturation step at 95uC for 45 sec, and annealing at 54uC for 30 sec. A final extension was performed at 68uC for 5 min.

Barcode microarray preparation
Oligonucleotides were synthesized at the Stanford Genome Technology Center. Oligonucleotides used as probes on the array (Table 3B) consisted of (from 59 to 39): a 59-amino group (for attachment to the array), a 5 bp poly-T sequence, Linker A, and a 20 bp complementary to barcode sequence. Probes were attached to the microarray essentially as described previously [24]. Each probe was printed in quadruplicate or quintuple, and two complete arrays were present on each chip. The post-printing processing of the microarrays was performed as recommended by the slide manufacturer (Amersham Biosciences, Piscataway, NJ). Control oligonucleotides used to verify array quality included a poly-T (20 bp) with 59-amino and internal-biotin modification as a labeling control (amino-B), a oligonucleotide complementary to Linker A with internal-Cy3 (amino-P) as an internal control for each spot's quality, and a 59-amino modified poly-T (20 bp) as a DNA spacer.

Array hybridizations
In the hybridization step, the biotinylated single-stranded target and amino-P were applied to the microarray. For the initial target hybridization step, we used 50 ml of target DNA in 16 hybridization buffer (100 mM MES, 1 M [Na+], 20 mM EDTA, 0.01 Tween20), 1.256 Denhardt's solution, and 5 nM amino-P. The hybridization was performed at 42uC for 12-16 hours. After hybridization, the microarray was washed 3 times in wash buffer and then labeled for 10 min at 50uC with a solution containing streptavidin-allophycocyanin (1mg/ml final concentration), 66 SSPE, 16 Denhardt's solution, and 0.01% Tween-20. The microarray was scanned for fluorescence intensity at 535 nm and 635 nm using a GenePix 4000 fluorescent scanner (Axon Instrument, Foster City, CA) set to scan at 450 PMT. GenePix Pro software was used to determine the total fluorescence signal from each spot on the array.

Validation and genotyping of DNA samples
For validation of the PathogenMip Assay, the 20 genomic DNA extracts were characterized for HPV presence and subtypes with nested PCR amplification using general primers PGMY09/11 and GP5+/6+ [1] according to a previously described protocol [26] with an initial amount of 40 ng/ml genomic DNA, followed by sequencing as described previously [5].