Spongospora subterranea root infection assessed in two potato cultivars differing in susceptibility to tuber powdery scab

Authors


E-mail: richard.falloon@plantandfood.co.nz

Abstract

Infection by Spongospora subterranea of roots of two potato (Solanum tuberosum) cultivars, either very resistant or very susceptible to powdery scab on their tubers, was studied in a glasshouse experiment. Plants grown in sand/nutrient solution culture were inoculated with S. subterranea sporosori 2 weeks after planting. Plant parameters, the intensity of zoosporangium infection in roots, numbers of Spongospora root galls and amounts of Spongospora DNA in roots, measured using quantitative PCR (qPCR), were assessed at sequential harvests. Inoculation with S. subterranea reduced water use (56 days after planting) by 26% in the tuber resistant cultivar compared with uninoculated plants, and by 60% in the susceptible cultivar. Inoculation did not affect growth of the resistant cultivar, nor shoot mass of the susceptible cultivar, but caused a 38% reduction in root mass of the susceptible cultivar. The intensities of zoosporangium development in both cultivars were similar. The susceptible cultivar had approximately four times more Spongospora root galls g−1 root mass than the resistant cultivar. Quantitative PCR detected S. subterranea DNA in roots 1 week after inoculation, and indicated a twofold greater amount of pathogen DNA in roots of the susceptible than the resistant cultivar. This study suggests that the S. subterranea zoosporangium stage in host roots is affected differently by host resistance factors than the sporosorus (root gall and tuber scab) stages. The study has also demonstrated the usefulness of qPCR for sensitive and consistent detection of S. subterranea across the duration of potato root infection.

Introduction

Powdery scab of potato (Solanum tuberosum) is caused by the plasmodiophorid (cercozoan) pathogen Spongospora subterranea f. sp. subterranea. This disease has previously been considered to be harmful mainly to potato tuber quality (Kole, 1954; Harrison et al., 1997), because surface lesions on tubers make them unacceptable for fresh supermarket sale, particularly as washed potatoes. In potatoes grown for processing, powdery scab-infected tubers require extra skin removal operations. Seed potato lines with powdery scab are downgraded or rejected by growers for establishment of new crops. Spongospora subterranea is also the vector of the Potato mop-top virus (Kirk, 2008), which causes internal damage in tubers (‘spraing’), making them unmarketable. Furthermore, the pathogen can also harm potato plant growth and productivity (Falloon et al., 1996, 2004; Lister et al., 2004; Houser & Davidson, 2010). Powdery scab and S. subterranea have increased in importance throughout the world where intensive production systems are used to grow potato crops (Merz & Falloon, 2009).

The S. subterranea life cycle on potato plants begins when zoospores infect root or stolon epidermal cells (Harrison et al., 1997; Merz, 1997, 2008). Zoosporangia of the pathogen then develop before the release of new zoospores which cause further root and stolon infections. These infections develop into creamy-white root galls, which later become brown and filled with many sporosori, each of which contains many resting spores of the pathogen. A similar process occurs on potato tubers (modified stolons), where sporosori fill powdery scab lesions on tubers. Sporosori and the resting spores they contain are the perennation stage of the pathogen life cycle, and these can survive in soil for many years. The life cycle completes when zoospores are released from resting spores to infect host root cells.

The root stage of infection by S. subterranea has been relatively little studied compared with tuber infection. Root infection has been associated with reduced water use and nutrient uptake, probably through zoosporangium development and zoospore release disrupting membrane selectivity in host epidermal cells (Falloon et al., 2004). The root infection stage is also very important for multiplication of the pathogen, both from heavy infestations of root cells by zoosporangia and through root gall development, which can produce large amounts of resting spore (sporosorus) inoculum to be released into soil. Thus, the root infection stages of the pathogen are important, both for effects on plant productivity and to increase soil inoculum for development of powdery scab (tuber disease) epidemics.

The use of potato cultivars with resistance to powdery scab is likely to be the most effective and sustainable method for managing the disease (Falloon et al., 2003; Falloon, 2008; Merz & Falloon, 2009). To date, no cultivar is known to be completely resistant to powdery scab on tubers. Furthermore, plant resistance is not similarly expressed in potato roots and tubers (Houser & Davidson, 2010; Merz et al., 2012), suggesting that resistance in these tissues involves different mechanisms under different genetic controls (Harrison et al., 1997; Falloon et al., 2003). The full spectrum of S. subterranea infection in resistant compared with susceptible potato cultivars is not well understood, with disease levels typically assessed based on tuber powdery scab or (less frequently) on intensity of root gall formation.

Assessment of S. subterranea root infection has been achieved by microscopic observation and scoring of zoosporangium infection in epidermal cells, and by visual assessment (scoring or enumeration) of root galls. An intensity scale for microscopy of zoosporangium infection was presented by Merz (1989), and a root gall severity scoring scale has also been outlined (http://www.spongospora.ethz.ch/LaFretaz/scoringtablegalls.htm). Although both methods effectively measure intensity of S. subterranea infection, these methods are likely to be subjective and dependent on observer capability. As in other plant pathology studies, more objective methods could be based on molecular technology. Qu et al. (2006) and Nakayama et al. (2007) used PCR techniques to quantify S. subterranea infection in roots, demonstrating advantages of increased sensitivity and rapidity for PCR relative to direct assessment methods.

The present paper reports an experiment where two potato cultivars with markedly different susceptibilities to powdery scab on tubers were grown under controlled glasshouse conditions, and S. subterranea infection of their roots was measured using direct observation and quantitative PCR (qPCR) technology. The experiment aimed to: (i) quantify the progression of S. subterranea throughout root infection, (ii) measure and quantify the relative susceptibility of the two cultivars to root infection by the pathogen, and (iii) assess the validity of using qPCR for detection and monitoring of the early stages of host infection by the pathogen.

Materials and methods

Plant growth

Two-week-old tissue cultured plantlets of the S. tuberosum cultivars Gladiator (very resistant to tuber powdery scab) and Iwa (very susceptible; Falloon et al., 2003) were planted into coarse river sand (1–2 mm grade) in 680 mL capacity plastic pots without drainage holes. The pots were placed in a greenhouse compartment where temperature was maintained at 18°C (± 2°C), with supplementary lighting (mercury lamps) to give daily cycles of 16 h light and 8 h dark. The sand in the pots was irrigated (by weight) to 90% of pot water holding capacity with nutrient solution (Falloon et al., 2003) and then irrigated to this amount between 08·00 and 09·00 h each Monday, Wednesday and Friday for the 56-day duration of the experiment. Pots were taken to a separate bench area in the greenhouse compartment for weighing and irrigation. Pots containing sand but without plants were included in the experimental design and also irrigated to 90% pot water holding capacity at each irrigation time. These were used for determination of amounts of nutrient solution used (‘water use’) by the plants.

Daily water use by plants

Water use for each plant was determined at each irrigation by subtraction of the weight of pots without plants from that for pots with plants. This was converted to ‘water use’ (g per day) using the numbers of days between the respective irrigations. Total pot weights were adjusted after each harvest (below) to allow for plant fresh weights.

Inoculation with Spongospora subterranea

Plants were grown for 2 weeks. Half of them were then inoculated with 10 mL of a standardized suspension of S. subterranea sporosori in nutrient solution. This applied 20 000 sporosori per plant, calculated as equivalent to 9·4 × 106 resting spores per plant (Falloon et al., 2011). The sporosori were obtained by scraping dry powdery scab lesions from field-grown tubers of cv. Agria potato and collecting sporosori after passage through a sieve (80 μm mesh).

Harvest of plants

Predesignated plants in the experiment were harvested on six occasions, at 0, 1, 2, 3, 4 and 6 weeks after inoculation. Roots and shoots of each plant were separated at the sand surface level, and fresh and dry weights were determined.

Experimental design

In total, 24 treatment combinations were applied in the experiment, comprising the two potato cultivars, two S. subterranea sporosorus inoculation treatments (uninoculated or inoculated), and the six harvests. Five replicates of these combinations were laid out in a 5 × 24 grid across two adjacent benches in the greenhouse compartment, using a randomized block design, where one row of 24 pots was a replicate. Two pots without plants were included within each replicate for water use determination (above).

Severity of root infection by Spongospora subterranea

Intensity of zoosporangium infection

A small sample (c. 0·1 g) of fresh roots was stained for 2 min in 0·1% trypan blue in lactophenol (Waller et al., 2002) and then stored in sterile reverse osmosis (RO) water. Assessment of S. subterranea zoosporangium infection of root epidermal (including root hair) cells was carried out for samples of 10 root segments (each c. 10 mm long) from each plant, using a compound microscope (× 200 magnification, and confirmation of infection at × 400 or × 1000 magnification). Intensity of zoosporangium infection in each root segment was assessed using the following six point score: 0 = no zoosporangia; 1 < 10% root hair cells containing zoosporangia; 2 = 10–20% of root hairs infected; 3 = 21–40% of root hairs infected; 4 = 41–60% of root hairs infected; and 5 ≥ 60% of cells infected. Scores were applied by comparison with micrographs (× 200 magnification) of 0·9 mm lengths of roots representing each of the six scores.

Severity of root galling

The total numbers of S. subterranea root galls on each plant were counted at each harvest. Numbers of root galls g−1 root dry weight were also determined, to account for differences in root mass between the two cultivars.

Spongospora subterranea DNA quantification in roots

DNA extraction

The root system of each plant was freeze-dried, then cut into small pieces and mixed to obtain a homogenized sample. Approximately 10 mg of each sample was disrupted with steel beads and DNA was extracted using a CTAB method (Russell & Bulman, 2005). DNA samples were resuspended with 10 mm Tris-HCl elution buffer (pH 8·5). The quantity and quality of DNA was checked by gel electrophoresis.

Quantitative PCR

A ribosomal internal transcribed spacer (ITS) probe/primer set, primers SPO10 (5′- GGTCGGTCCATGGCTTGA-3′) and SPO11 (5′- GGCACGCCAATGGTTAGAGA-3′), TaqMan probe SPOPRO1 (5′ FAM-CCGGTGCGCGTCTCTGGCTT-BHQ 3′), was used to detect S. subterranea. The primer/probe set was designed using primer express v. 3.0 (Applied Biosystems). qPCR reactions (1 μL of DNA in 20 μL of reaction) were completed in an ABI StepOnePlus machine. Yields of DNA appeared consistent when measured by gel electrophoresis (not shown). Triplicate reactions were used for all samples and standards. A non-template control (NTC) was added to each plate. qPCR was carried out in a 40 cycle run (95°C for 15 s, 60°C for 20 s and 72°C for 20 s). To further test the accuracy of the measurements, four complete sets of qPCR repeats were carried out for each sample (i.e. a total of 12 replicates per sample).

Amounts of Spongospora DNA in samples were calculated by an absolute quantification technique, using a standard curve constructed from plasmid DNA containing the S. subterranea ITS gene. To obtain the plasmid, a copy of the S. subterranea ITS gene was amplified by PCR, using the primers ITS5 and ITS26 (Bulman & Marshall, 1998), then cloned using the TOPO TA Cloning Kit (Invitrogen). The correct insert was identified by PCR and DNA sequencing, then the plasmid was purified using the Plasmid Mini Kit (QIAGEN).

A standard curve was constructed from tenfold dilutions of the plasmid, ranging from a concentration of 4·7 × 10−5 to 4·7 × 10−1 ng μL−1. From these regressions, the efficiency factor (Eff) was calculated as Eff = 10−1/slope. It was found to range from 1·83 to 2·04.

Statistical analysis

Water use over time was first explored graphically, and formal analysis of water use was carried out for the data at 56 days. A mean zoosporangium score was calculated for each plant. When root galls were not observed at individual harvests, these data were excluded from the statistical analysis to avoid underestimating variation. To stabilize the variances, numbers of root galls and numbers of root galls g−1 root dry weight were square root transformed, and estimated DNA quantities were log transformed before analysis. Data were analysed with analysis of variance, including contrasts to compare between cultivars, inoculation treatments, harvests (except for water use) and interactions between these factors. All statistical analyses were carried out with genstat (GenStat Committee, 2010).

Results

Water use

At 56 days after planting, cv. Gladiator plants were using approximately four times the amount of water (mean = 8·23 g per day) of cv. Iwa plants (2·52 g per day; = 0·001). Water use by plants inoculated with S. subterranea was less than for uninoculated plants (= 0·088). Mean water use by uninoculated Gladiator plants at the end of the experiment was 9·47 g per day, while that for the inoculated plants (42 days after inoculation) was 6·98 g per day (26% reduction due to inoculation). Mean water use by uninoculated Iwa plants at the same time was 3·59 g per day, and for inoculated plants was 1·45 g per day (60% reduction due to inoculation).

Plant parameters

Mean shoot dry weights increased through the sequential harvests (< 0·001), as expected. Mean shoot dry weight of each cultivar was similar for uninoculated and inoculated plants (> 0·10). Shoot dry weight was greater for Gladiator than for Iwa (= 0·021). Mean root dry weights also increased through the sequential harvests (< 0·001), and were similar for both cultivars and for the two S. subterranea inoculation treatments at all harvests (> 0·1). However, at the final harvest an effect of inoculation on Iwa plants was indicated. The inoculated plants had a mean root dry weight of 0·56 g, compared with that of 0·91 g for uninoculated plants.

Root infection by Spongospora subterranea

Intensity of zoosporangium infection

Low incidences of zoosporangium infections were observed in a few uninoculated plants at 4 and 6 weeks for Gladiator, and 3, 4, and 6 weeks for Iwa (Fig. 1).

Figure 1.

Mean zoosporangium intensity scores for roots of the potato cultivars Gladiator and Iwa, which were either uninoculated or inoculated with sporosori of Spongospora subterranea and harvested at 1 week intervals up to 6 weeks after inoculation. Bars indicate LSDs (= 0·05, d.f. = 54); upper bar to compare non-zero means, lower bar to compare zero with non-zero means.

For plants inoculated with S. subterranea, zoosporangia were first observed in plants 1 week after inoculation, but infection was rare at this stage. Mean zoosporangium scores were very similar in both cultivars, reaching a maximum at 3 weeks after inoculation (Fig. 1). Zoosporangium numbers decreased in both cultivars at later harvests, 4 and 6 weeks after inoculation for Gladiator and 6 weeks for Iwa.

Zoosporangia were obvious in root epidermal cells as multicelled structures (Fig. 2a,b). Under high magnification, tetrads of heavily stained subunits were obvious within individual units in zoosporangia (Fig. 2b). These tetrads have been shown to be four zoospores (Clay & Walsh, 1990). At the later harvests, many root epidermal cells in these plants contained zoosporangia (with zoospores; Fig. 2b,c), but some zoosporangia were free of cell contents (Fig. 2d,e), indicating that their zoospores had been released. Occurrence of empty zoosporangia was accompanied by reduction in mean severity score (Fig. 1) at the later harvests, 4 and 6 weeks after inoculation.

Figure 2.

Micrographs of zoosporangia in stained roots of potato plants (cv. Iwa), grown in sand/nutrient solution and inoculated with Spongospora subterranea sporosori. (a) Root with heavily infected root hairs (severity score 4; see text), 4 weeks after inoculation; (b) root surface cells with zoosporangia, 4 weeks after inoculation; (c) root hair with zoosporangium, 4 weeks after inoculation; (d) and (e) root epidermal cells with zoosporangia that have released zoospores, 6 weeks after inoculation.

Numbers of root galls

No root galls were observed on any of the uninoculated plants at any harvest (Fig. 3). For Iwa, galls were observed on roots of the S. subterranea inoculated plants 3 weeks post-inoculation, and for Gladiator, 4 weeks post-inoculation. At the later harvests (3 to 6 weeks post inoculation), Iwa had approximately four times more root galls g−1 root dry weight than Gladiator (< 0·001, Fig. 3b).

Figure 3.

(a) Mean numbers of root galls per plant, and (b) mean numbers of root galls g−1 root dry weight, for potato plants of cvs Gladiator and Iwa, either uninoculated or inoculated with Spongospora subterranea and harvested at 1 week intervals up to 6 weeks after inoculation. Bars indicate LSDs (= 0·05, d.f. = 16), lower bar uninoculated, upper bar inoculated.

Quantification of S. subterranea DNA using qPCR

Quantitative PCR detected small amounts of S. subterranea DNA in roots at 1 week after inoculation (Fig. 4). Amounts of DNA increased with harvest time for all plants (< 0·001 for the overall effects of harvests), but the amount of DNA in roots of the uninoculated plants was substantially less (< 0·001) than in the plants inoculated with S. subterranea sporosori throughout the experiment. Amounts of pathogen DNA reached maxima in both cultivars at 4 weeks and then slightly decreased at 6 weeks after inoculation. Iwa had approximately twice the amount of S. subterranea DNA in roots compared with Gladiator at 3, 4 and 6 weeks after inoculation (Fig. 4b). However, differences between cultivars were not statistically significant (= 0·858).

Figure 4.

Mean amounts (ng μL−1) of Spongospora subterranea DNA detected using qPCR in roots of cvs Gladiator and Iwa, which were either uninoculated or inoculated with sporosori of Spongospora subterranea and harvested at 1 week intervals up to 6 weeks after inoculation. (a) Log10 transformed means; bar indicates LSD (= 0·05, d.f. = 92); (b) back-transformed means.

Discussion

This study has used conventional and molecular methods for determining intensity of S. subterranea infection of two potato cultivars. The cultivars were chosen to represent opposite ends of the spectrum of susceptibility to powdery scab on tubers. Inoculation with S. subterranea caused reduced water uptake in both cultivars, but this effect was greater for cv. Iwa (very susceptible to tuber disease) than for cv. Gladiator (very resistant). This suggests that intensity of root infection (which causes reduced root function) may not be directly related to tuber infection. Shah et al. (2012) recently demonstrated that reduced water use by Iwa was related to the amount of Spongospora inoculum in soil. Other plant characteristics may affect root function during S. subterranea infection, including root form, mass or absorptive capacity.

Quantitative PCR has been used quite widely for the quantification of S. subterranea in soil (van de Graaf et al., 2003; Brierley et al., 2009), but only rarely to track S. subterranea infection in plants (van de Graaf et al., 2007). In the present study, qPCR detected DNA of the pathogen in roots 1 week after inoculation, and small numbers of zoosporangia were subsequently directly observed in samples from the same time points. Merz (1997) reported that tomato (Solanum lycopersicum) roots exposed to S. subterranea sporosorus inoculum in a laboratory bioassay became infected by zoospores within 5 h of exposure, demonstrating that the first stages of infection can occur very quickly in suitable conditions. In a similar bioassay, Qu et al. (2006) detected S. subterranea by PCR in tomato and tobacco (Nicotiana debneyi) roots at 2 days after inoculation. The present study further emphasizes the ability of the pathogen to develop rapidly and multiply in host root epidermal cells at the zoosporangium stage.

The two cultivars in this experiment had similar amounts of S. subterranea zoosporangium development in root epidermal cells, as indicated by zoosporangium severity scores. Intensity of zoosporangium infection increased to a maximum at 3 to 4 weeks after inoculation in both cultivars, indicating that the pathogen was probably completing secondary cycles of infection, from zoospores released from zoosporangia. Zoosporangium infection in the two cultivars contrasts to their relative susceptibilities to tuber powdery scab. Furthermore, Falloon et al. (2003) detected a much greater difference in numbers of zoosporangia in roots of these two cultivars than was detected in the present study. This indicates that zoosporangium counts alone may not be reliable for determining cultivar or variety differences in susceptibility to the pathogen.

In contrast, amounts of S. subterranea DNA measured with qPCR and counts of Spongospora root galls on root systems indicated a divergence between the cultivars with increasing time after inoculation. As with the zoosporangium assessments, gall counts and qPCR also showed maximum infection at 3–4 weeks post-inoculation. However, unlike zoosporangium severity assessments, gall counts and qPCR both indicated greater infection in Iwa than in Gladiator. The difference was especially pronounced for numbers of root galls, with gall formation occurring later, and reaching only about 20% of the numbers in Gladiator compared with Iwa. Quantitative PCR demonstrated that there was approximately twice the amount of Spongospora DNA present in roots of Iwa than in Gladiator. Thus, the measured amounts of DNA appeared to increase in concert, both with increasing density of zoosporangium infection of root epidermal cells 2–3 weeks after inoculation, and with later formation of root galls 3–4 weeks after inoculation. Amounts of DNA increased most sharply at the time of gall formation, when the total numbers of Spongospora cells would be expected to be greatest. Overall, integration of qPCR into this glasshouse experiment allowed improved temporal tracking of total pathogen in roots, compared with the more traditional root infection assessments using zoosporangium and gall counts.

The evidence here points to similar amounts of primary infection (zoosporangium development) in the two cultivars, but a divergence between the cultivars at later stages, possibly during secondary infection cycles and definitely at root gall formation. Gladiator was earlier classified as ‘very resistant’ to tuber disease and among the most resistant to powdery scab of a large number of cultivars assessed, while Iwa was classified as ‘very susceptible’ and among the more susceptible of the cultivars (Falloon et al., 2003; Genet et al., 2011). Root galling is a precursor to sporosorus development in S. subterranea, and is probably similar to the development of powdery scab lesions on tubers. Galls are often observed in the early stages of pathogen development on tubers (sometimes described as the ‘cauliflower’ symptom). It therefore follows that root galling and tuber powdery scab may be closely related stages of development of the pathogen, but occurring on different plant organs. Similar zoosporangium infection in the two cultivars but a different transition to gall formation is consistent with prior plasmodiophorid research showing apparent onset of resistance at this point in the pathogen cycle. For example, several plasmodiophorid species can perform primary infection of plants other than their usual hosts, but these infections do not proceed to full-scale secondary infection (Ludwig-Muller et al., 1999; Desoignies et al., 2010). In the case of S. subterranea, some atypical hosts such as tomato rarely or never develop the sporosorus (resting spore) stage of the pathogen, but can become heavily infected with zoosporangia (Merz, 1989; Andersen et al., 2002; Qu & Christ, 2006). It is possible that infecting plasmodiophorids become more affected by host defence systems in the transition from epidermal to cortical infection. Alternatively, plasmodiophorids may only be capable of successfully subverting the growth regulatory systems of host cells in certain plant–plasmodiophorid combinations. These continue to be matters of speculation while studies of the resistance mechanisms of potato cultivars to Spongospora remain at a very early stage (Baldwin et al., 2008).

Having made the case that root galling and tuber powdery scab formation are similar in Iwa and Gladiator, in some potato cultivars this relationship is poor (Falloon et al., 2003; Merz et al., 2012). The genetics of host resistance to S. subterranea is therefore likely to be complex, and may be under different genetic control for the different phases of the infection and pathogen development (zoospore penetration of root cells, zoosporangium development, root gall formation or tuber powdery scab). The cellular and biochemical mechanisms of sporosorus formation by plasmodiophorids remain to be elucidated.

In the present study, qPCR provided earlier detection of S. subterranea infection than microscopic examination of zoosporangia, and also detected low amounts of infection in uninoculated plants, which was only later confirmed by detailed microscopic observations of roots. The uninoculated plants had very low mean zoosporangium severity scores, small amounts of S. subterranea DNA and no root galls, indicating that cross-contamination was slight. Anecdotal reports indicate that such contamination of uninoculated plants is a frequent and ongoing problem in Spongospora experiments. This contamination occurred despite the general protocols for carrying out the glasshouse experiment being designed to prevent cross-contamination between plants. If the contamination derived from aerosols within the glasshouse, then elimination of this problem may be difficult if uninoculated pots continue to be placed in close proximity to other treatments because of the requirement to use statistically valid experimental designs.

Traditional methods for assessing Spongospora infection, such as microscopic enumeration of zoosporangia and counts of root galls, are labour intensive, require significant experience, and are likely to be subject to variation between users (van de Graaf et al., 2003). Quantitative PCR requires no specific experience with Spongospora and results/protocols are comparatively easy to benchmark with those of other laboratories. The present study has demonstrated several advantages of qPCR measurements, including an ability to use a single technique across the duration of the infection period, and detection of low amounts of infection (and contamination) that might otherwise have not been detected in uninoculated samples. To the authors' knowledge, this is the first use of qPCR to monitor the Spongospora infection process in potato cultivars of differing resistance characteristics. New processes are presented for sample preparation and DNA extraction that produce consistent measurements of Spongospora in potato roots. With further optimization and streamlining, these techniques may be integrated into wider screenings of potato breeding material. Nevertheless, the two-fold difference in amounts of S. subterranea DNA in roots was not statistically significant. It is noteworthy that it is often difficult to demonstrate a two-fold difference in DNA using qPCR, and many studies rely on detection of larger fold changes. More rigorous experimental technique will be required to estimate infection of different potato cultivars, when differences in DNA amounts are small.

Acknowledgements

Discussion with Dr Alison Lees helped to initiate this study. The research was carried out with financial support from the Ministry of Science and Innovation (Programme LINX0804), and Horticulture New Zealand. Dr Farhat Shah provided assistance in this study. The authors declare no conflict of interest.

Ancillary