Riems influenza a typing array (RITA): An RT-qPCR-based low density array for subtyping avian and mammalian influenza a viruses

Rapid and sensitive diagnostic approaches are of the utmost importance for the detection of humans and animals infected by specific influenza virus subtype(s). Cascade-like diagnostics starting with the use of pan-influenza assays and subsequent subtyping devices are normally used. Here, we demonstrated a novel low density array combining 32 TaqMan® real-time RT-PCR systems in parallel for the specific detection of the haemagglutinin (HA) and neuraminidase (NA) subtypes of avian and porcine hosts. The sensitivity of the newly developed system was compared with that of the pan-influenza assay, and the specificity of all RT-qPCRs was examined using a broad panel of 404 different influenza A virus isolates representing 45 different subtypes. Furthermore, we analysed the performance of the RT-qPCR assays with diagnostic samples obtained from wild birds and swine. Due to the open format of the array, adaptations to detect newly emerging influenza A virus strains can easily be integrated. The RITA array represents a competitive, fast and sensitive subtyping tool that requires neither new machinery nor additional training of staff in a lab where RT-qPCR is already established.

Scientific RepoRts | 6:27211 | DOI: 10.1038/srep27211 system based on reverse transcription quantitative PCR (RT-qPCR) in a multiwell layout that is easily prepared and runs on standard lab equipment. By using a hydrolysis probe technique (TaqMan ® ), one of the most powerful and widespread methodologies in diagnostic microbiology 12 was applied. The format is referred to as the Riems Influenza A Typing Array, abbreviated RITA. The RITA assay is the first system that combines simple duplex TaqMan ® reactions into one diagnostic tool, without the need for further equipment for detection or identification of 14 HA and 9 NA subtypes of influenza A viruses. More than 400 IAV isolates and clinical material from 63 swab samples were successfully subtyped to qualify this protocol as eminently suitable for routine procedures.
A typical lay-out of the 32 wells is depicted in supplemental Fig. S1, available online. During the setup of a plate, RNase-free water (0.5 μ l), the IAV specific primer-probe-mix (1 μ l) and the primer-probe-mix for the internal control (1 μ l) were added to each individual well. At this stage of preparation, storage of the plate at − 20 °C is possible, and bulk production of plates is advisable. After thawing the required amount of plates, the master mixes for each sample were prepared, which were composed of the RT-PCR buffer, the enzyme mix, the internal control RNA, residual water and the extracted viral RNA. Ten microliters of this mixture was added to each of the 32 wells of the test using a multi-channel pipette. Thus, a total of 80 μ l of extracted RNA is required per sample to feed the 32 wells for a complete characterization. For PCR amplification, the following temperature profile was applied: 10 min at 45 °C (reverse transcription) and 10 min at 95 °C (inactivation of the reverse transcriptase/activation Taq polymerase), followed by 45 cycles of 15 sec at 95 °C (denaturation), 20 sec at 56 °C (annealing) and 30 sec at 72 °C (elongation). Fluorescence values (FAM, HEX) were collected during the annealing step. In order to confirm the integrity of all of the target-and internal control-specific primer-probe-mixtures a positive control containing a mixture of RNA of all H and N subtypes targeted (panIAV-PC) is recommended. The panIAV-PC is a mixture of IAV-RNA from all IAV subtypes analysed in the array. However, based on the internal control system and the inclusion of the pan-influenza A IAV-M1.2 assay 15 , the application of RITA without the co-analysis of panIAV-PC is also valid. The analysis of RNase free water as a no template control (NTC) may identify cross-contamination of the master mix. Analysis of the NTC and the panIAV-PC was required after a new batch of arrays was produced, or the arrays were stored in the freezer for a prolonged period of time (more than 4 weeks) before use.

Results
RT-qPCRs and construction of a 32-well PCR array. By in silico analysis of published sequence data, primer and probes for the generic detection of HA subtypes H1 to H13 and H16 and all 9 NA subtypes were selected with the aim of detecting the broadest possible spectrum within a given subtype. To this end, various RT-qPCRs were newly developed, and their performance, analysed. In addition, published RT-qPCR subtyping protocols were evaluated, as well. Finally, the best performing assays were combined into the low density array referred to as "Riems Influenza A Typing Array" (RITA). The final makeup of RITA consisted of single assay detection of subtypes H4, H6, H8, H9, H11, H12, H13 and H16; two assays for H1, H2, H3, H5 and H10 detection; and three assays for H7 detection. NA subtyping was performed by single assays, except for the N3 subtype, which required two assays. This process resulted in the ultimate 32 well format RITA, which was evaluated using a selected set of 404 IAV isolates (detailed in supplemental Table S1). All samples were also run in the pan-influenza IAV-M1.2-assay 15 included in the 32 well RITA to verify the presence of IAV RNA in general and to obtain an estimation of the viral genome load of each individual sample by generic amplification of a highly conserved region of genome segment 7 of all IAV isolates 15 .
Analytical sensitivity of RT-qPCRs. The analytical sensitivity of each single assay of RITA was evaluated using a dilution series of matching viral RNAs and compared with the Cq-values of the respective RNA obtained by the generic-influenza IAV-M1.2 test. The analyses showed that all subtype specific assays were similar sensitive as the generic-influenza assay. At the maximum 10-fold less viral RNA was detected, using the serotype specific assays compared to the generic pan-influenza test. In some cases, a better analytical sensitivity for the serotype specific assay was measured (data available upon request).
Analytical specificity of RITA. Analytical specificity was assessed by the determination of inclusivity and exclusivity on a validation panel of 404 different IAV isolates. The results are summarized in Table 4. As an example, "H1-subtype mix 3" detected 51 out of 54 H1 virus isolates correctly, whereas "H1 mix 27" detected 53 out of 54 H1 RNAs. Isolates that were not detected differed between the two assays: "H1 mix 3" did not detect  1951 (H1N1). Therefore, with a combination of both H1 mixes, all H1 isolates represented in the validation panel were correctly identified as representatives of the H1-subtype. While "H1 mix 3" proved to be exclusively specific for H1, "H1-mix 27" showed a weak cross-reactivity with 37 H6 isolates and three isolates of the H8 subtype and therefore had a specificity of only 89.9%. However, it should be noted that there was a difference in the Cq-values (Δ C q ) between the true specific and the false positives of more than 10 units when comparing the pan IAV-M1.2 assay to the false positive signals, and 9 units when comparing the specific assay to the false positive reacting assay results. Both H2-detecting assays were specific for subtype H2 and identified 18 out of 18 H2 isolates correctly, and an unspecific cross-reactivity was recorded for only one of the H5 isolates for the "IAV-H2-Mix 10" (Δ C q > 10 generic assay, Δ C q > 8 specific assays; Table 4). Furthermore, from a total of 40 H3 isolates of the validation panel, 35 were detected by "IAV-H3-Mix 3", and 40 of 40 scored positive with the "IAV-H3-Mix 14" (Table 4). Additionally, all 17 H4 isolates were correctly identified by "IAV-H4-Mix 15" (Table 4). Specific detection of all H5 sequences was achieved by "IAV-H5a-Mix 1" (88/88) and also approximated by the "IAV-H5-Mix 1" assay (87/88). Subtype H6 sequences (in a total of 61 isolates) were all correctly evaluated by the "IAV-H6-Mix 8" assay (Table 4). Interestingly, for the complete identification of the H7 subtype sequences, three assays had to be implemented, providing the following coverage: "IAV-H7-CODA" detected 33 out of 35 correctly, with cross-reactivity with an H5 isolate, an H10 isolate and an H15 isolate; IAV-H7-Mix 2 detected 35 out of 35 correctly, with cross-reactions with six of the H10 isolates; "IAV-H7-Mix 2.2" detected all 35 isolates correctly. The nonspecific reactions again showed, in all cases, a Δ C q > 10 compared to the specific reactions. Finally, the assays designed to detect the H8, H9, H11, H12, H13 and H16 subtypes reacted with the respective isolates and showed no cross-reactivity with any other subtypes. Nevertheless, both H10 assays exhibited nonspecific reactions in addition to the exact identification of all H10 subtypes: "IAV-H10-Mix 5" scored positive with 30 out of 40 H3 isolates with Cq-values that were at least three units higher than the values of the "IAV-M1.2" assay and the specific assay; "IAV-H10-Mix 9" scored positive with 30 out of 35 H7 isolates with a Δ Cq value of 5 compared to the IAV-M1.2 assay and Δ Cq value of > 4 compared to the specific assay. However, with the redundancy of the two tests, all H10 isolates could be correctly identified.
The reactivity of the NA-subtype specific assays showed a 100% match to the applied isolates, with no cross-reactivity (Table 5).
Diagnostic performance of RITA. The evaluation of diagnostic sensitivity was conducted using RNA from swab samples of wild birds or pigs. In general, subtyping was successful for those samples that exhibited Cq-values < 34 with the pan IAV M1.2 assay. Out of 45 swab samples originating from wild aquatic birds, 45 samples exhibited (at least one) HA result and 44 samples (at least one) NA results by using RITA (Table 6). In 16 samples, additional reactivity with further subtypes was evident. Eleven of these additional reactivities apparently paralleled those already known from the validation panel; thus, no mixed IV infections were assumed for these samples. In contrast, mixed influenza infection was suspected in four samples because similar Cq-values for two or more HA and NA subtypes were obtained from multiple subtype assays. Verification of subtypes identified by RITA was achieved for the majority of bird samples by Sanger sequencing technology. However, samples showing Cq-values > 30 in the generic IAV M1.2 assay did not regularly yield specific amplicons, especially for the HA, that were suitable for Sanger sequencing. The individual reaction patterns of all diagnostic samples are summarized in Table S2. Nasal swabs from swine were also used for subtype detection. For 15 out of 17 IAV positive samples, a subtype was unambiguously assignable by RITA (Table 6). One sample showed additional (cross-) reactivity, as previously seen from the validation panel. Specific subtypes identified by RITA were confirmed by subsequent Sanger sequencing (except for one sample for which no NA amplicon was obtained; Table S2). For one nasal swab sample, a mixed infection with more than one subtype was suspected (Table 6).

Discussion
The objective of this study was to develop and validate an RT-qPCR based subtyping tool for rapid and reliable direct subtyping of influenza A viruses from original sample material of avian and mammalian hosts. The finally validated 32-well RT-qPCR array, named RITA, allows the differentiation of 14 HA and 9 NA subtypes, as well as generic IAV RNA detection and an internal control system. The rare subtypes H14 and H15 were not implemented in this RITA version due to their minor relevance and to maintain an easy to manage format (e.g. using multiples of eight).
RT-qPCR-based nucleic acid detection techniques are routinely implemented throughout diagnostic laboratories, and the RITA protocol is easily established in such laboratories. The single tube duplex assay structure of RITA provides very high sensitivity for the individual assay, as well as a very good handling versatility,  because all 32 wells can be pipetted using a multiwell pipette and a single master mix per sample. This format also offers complete flexibility to modify existing or implement new assays targeting specific clades, HA cleavage sites (pathotyping for H5 and H7 subtypes;) 17 , new geographical variants or assays for the identification of relevant pathogens of differential diagnostic importance. In the light of this flexibility, RITA can easily be adapted to the effects of evolutionary drift that may lead to IAV variants not only escaping vaccine-induced immunity but also routine diagnostics 18 . The current plate design resembling a low density PCR array format allows analysis of three samples in parallel per 96 well plate. The plates can be prepared in advance and stored at − 20 °C. A storage time of nine months did not lead to a loss of sensitivity. Using plates from the freezer, the overall setup and run time of RITA was less than three hours. Recently, Elizalde and colleagues described a molecular subtyping approach for HA subtyping of IAV 19 , and the use of locked nucleic acids (LNA) within the probe sequence in this method rendered the RT-qPCR assays highly subtype-specific 19 . LNA probes are specific to such an extent that single nucleotide polymorphism (SNP) genotyping is possible 20 . However, with this implement, mismatches within the LNA probe binding region are less well tolerated, and the likelihood of false negative reactions increases. In addition, LNA-containing probes are more expensive than Taqman ® or minor groove binding (MGB ® ) probes. Using SYBR green as a non-specific RT-qPCR  H1  H2  H3  H4  H5  H6  H7  H8  H9  H10 H11 H12 H13 H14 H15 H16   IAV-H1-Mix 3  51  ---------------IAV-H1-Mix 27  53  ----37*  -3*  --------IAV-H2-Mix 4  -18  --------------IAV-H2-Mix 10  -18  --1*  -----------IAV-H3-Mix 3  --35  -------------IAV-H3-Mix 14  --40  -------------IAV-H4-Mix 15  ---17 -  Table 4. Reactivity of HA subtype-specific RT-qPCR assays used in RITA with the validation panel of 404 viral isolates. * Δ C q between pan assay and false positive reactivity ≥ 10 and Δ C q between specific assay and false positive reactivity ≥ 8. # Δ C q between pan assay and false positive reactivity ≥ 3 and Δ C q between specific assay and false positive reactivity ≥ 3. † Δ C q between pan assay and false positive reactivity ≥ 5; and Δ C q between specific assay and false positive reactivity ≥ 4.  IAVs. Yet, the flexible basis of the array system allows the adaptation/inclusion/substitution of single RT-qPCRs to create RITA formats that serve specific needs e.g. according to geography (Eurasian versus American genotypes) or epidemiology (outbreak-adapted assays) or differential diagnostics. RITA was validated using 404 influenza A virus isolates from both avian and mammalian hosts, and unequivocal subtype identification was completely successful on virus isolates. The need to design primers and probes that guaranteed the broadest within-subtype reactivity and the inherent prerequisite of RITA to apply the same cycling conditions to all 32 assays implemented produced few trade-off effects with respect to specificity for some of the assays. Slight tendencies for the co-amplification of non-targeted HA subtypes were observed for some of the selected RT-qPCRs. However, these false amplicons were usually detected with a much lower sensitivity and thus could easily be identified as false-positives when comparing the true positive subtype-specific reactions to the Cq-values obtained in the pan IAV M1.2 assay. Nevertheless, this propensity of the RITA assays reduces only the reliability to detect minor virus species (particularly H1, H10) in mixed IAV infections. The necessity of discerning co-amplification from mixed infection became apparent when examining clinical material from swine and avian species. The mixed infection with subtypes H1 and N1 and N2 detected in one porcine nasal swab demonstrated the extraordinary capacity of RITA to identify mixed infections in one sample. Sample materials originating from aquatic wild birds also gave evidence of mixed infections by diverse influenza A virus subtypes when examined by RITA (Table 6). Mixed infections are a precondition of genome reassortment 28 , but discerning mixed infections by standard Sanger sequencing technologies is seriously hampered 29 . The current RITA protocol in principal allows the identification of mixed IAV infections, although the interpretation of data may require certain expertise and experience to distinguish true co-infections with different IAV subtypes from unintended co-amplification or sample contamination. For instance, samples that exhibit similar Cq values for two or more HA and NA subtypes likely represent true co-infections. In contrast, if a single major subtype is identified and one or more very minor subtypes are present in the sample, no clear decision is possible unless other subtyping methods (virus isolation, next-generation sequencing) produce corroborating results. Therefore, and to keep in touch with the evolution and geographic dissemination of IAVs, the single RITA assays will have to be reassessed and improved on a regular basis.

RT-qPCR
Rapid and unequivocal IAV subtype identification, ideally even in clinical samples, is the prerequisite for monitoring the evolutionary dynamics and diversity of IAVs, with special emphasis on the One Health concept. With the use of RITA, the standard diagnostic cascade recommended for the detection and characterization of IAV in avian samples is reduced to a pretesting using a pan-influenza IAV-assay to verify presence of IAV RNA in the sample and a single RITA run. Future versions of RITA are intended to hold an "avian RITA", which includes cleavage site-specific assays for H5/H7 pathotyping and the detection of avian pathogens of differential diagnostic relevance. An envisaged "mammalian Rita" can include assays for the differential diagnosis of influenza-like illnesses, while certain "avian" subtypes can be excluded from the array.