Comparison of 2 sampling methods for molecular detection of bacteria or fungi from feline hair and scale specimens

Abstract

Skin diseases of cats are among the most frequent client motivations for a veterinary consultation. Both carpet and toothbrush sampling are commonly used to obtain hair and scale samples for microbiologic testing. Although molecular tests have become more accessible and more widely used by clinicians, the ideal collection method for clinical specimens is unclear. To assess their performance in retrieving microbial DNA from clinical samples, we compared the bacterial and fungal DNA load in hair and skin scale samples collected using carpet or toothbrush methods. We evaluated sample DNA yield using fluorometry, spectrophotometry, and quantitative PCR. Despite no measurable differences in sample weight, toothbrush samples yielded significantly higher bacterial (p = 0.028) and fungal (p = 0.005) DNA loads compared to carpet samples, regardless of disease status. The toothbrush method was more effective in harvesting microbial DNA from hair and skin scale samples.

Since ~2000, laboratory testing methods have shifted from culture-based methods to more sensitive molecular, culture-independent techniques.^3,18 Various skin-sampling tools have been used to collect material for laboratory analyses. Swabs are by far the most widely used sampling tool for molecular investigation of fungi or bacteria on skin.^13,15 However, swabs are not ideal for sampling species with dense fur because they do not also sample hair and skin scales. Thus, carpet and toothbrush hair-sampling techniques were developed and are now used widely in veterinary dermatology practice to collect samples for the detection of agents that infect hair and skin scales.^11,16 However, such techniques rely heavily on the performance of the sampling method. If the initial DNA amount is below a detection threshold for a given sample, inaccurate data may be generated, including false-negative results. ³ Detection is particularly challenging when probing hair and skin scale samples for infectious agents given their low microbial biomass. ¹

We compared the efficiency of the carpet and toothbrush techniques in collecting microbial DNA from hairs and skin scales as a pre-analytical step for culture-independent microbiologic testing. Additionally, we compared the bacterial and fungal DNA loads obtained by each technique between domestic cats diseased and subclinical for dermatophytosis (used here as a model disease given that the causative agent colonizes the hair primarily, not the skin). ¹¹ We hypothesized that there would be a difference in the amount of microbial DNA harvested between the toothbrush and carpet methods when sampling feline hair and skin scales.

Our study was approved by the Institutional Animal Care and Use Committee of our institution (protocol 8549140217) and the owners signed a written consent before enrollment in our study. We used a convenience sample of 45 cats enrolled in a feline dermatophytosis study for our prospective, randomized, and cross-sectional study (Suppl. Table 1). We collected samples from 1) 15 Persian cats with skin lesions characteristic of dermatophytosis (hair loss, erythema, scaling) and a positive Microsporum canis fungal culture (i.e., bright yellow colonies on Mycosel agar, BD Diagnostics; diseased group); 2) 15 Persian cats with no skin lesions and a positive fungal culture (subclinical group), and 3) 15 domestic short-haired cats with no history and no clinical signs of skin disease, and with a negative fungal culture (control group). Short-haired cats were recruited for the negative control group because Persian cats are seldom negative for dermatophytes and given the ease of recruiting enough animals.

A new pair of gloves was used to handle and sample each cat. All animals were sampled by the carpet and toothbrush methods, one method immediately after the other. The order of sampling was determined by a random choice generator (http://jklp.org/html/choose.html). Sampling began on the head followed by the neck, dorsum, trunk, ventrum, limbs, and tail. These sites were brushed once in the direction of, and, once against, hair growth. Samples were placed in individual plastic bags and stored at −80°C until processing. Toothbrush and carpet negative controls (n = 2) were processed similarly but without coming into contact with any animals. The toothbrush (2 × 2 × 1 cm; adult toothbrush, Condor) technique was performed according to a modified Mackenzie collection method. ¹¹ The carpet (5 × 5 × 0.3 cm; polyester carpet 0502220801, Tapetes) technique was performed as described previously. ¹⁶ Hair and scales obtained from the toothbrushes and carpets were, to the best of our ability, completely dislodged macroscopically and weighed (120 B2-Series; VWR, Avantor). Negative control samples were trimmed to fit the 1.5-mL tubes used for DNA extraction. Each sample was handled individually using a sterile pair of nitrile gloves in a biosafety cabinet.

Total DNA was extracted (PowerLyzer PowerSoil DNA isolation kit; Mo Bio) following the manufacturer’s recommendations. For quantification of DNA yield prior to quantitative PCR (qPCR) assessment, 2 independent methods were used. Fluorometry was conducted (PicoGreen dsDNA assay kit; Life Technologies) with a hand-held fluorometer (PicoFluor; Turner BioSystems). Spectrophotometry was conducted (ND-1000 spectrophotometer; NanoDrop Technology). All assays were performed according to the manufacturer’s instructions.

The qPCR analyses (LightCycler 1536 system; Roche) used 2 DNA-binding dye–based assays targeting the 16S rRNA and the fungal internal transcribed spacer 2 (ITS2) genes, which are universal targets for detecting bacteria and fungi. The 16S rRNA hypervariable region 4 (V4) was amplified using primers 515F (TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGTGCC) and 806R (GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGGACTACHVGGGTWTCTAAT). ⁵ ITS2 amplification was performed using primers 5.8SR’ (TCGTCGGCTCAGATGTGTATAAGAGACAGTCGATGAAGAACG) and ITS4_Nextera (GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGTCCTCCGCTTATTGATATGC). ⁴ Each 25-μL reaction contained 1× KAPA HiFi HotStart real-time kit (Roche), forward and reverse primers (0.8 μM, 1 μL of each), and 1 μL of DNA template. Amplification was performed for both genes using the following cycling parameters: 1 cycle of 95°C for 5 min, followed by 35 cycles of 98°C for 20 s, 55°C for 15 s, and 72°C for 1 min.

All standard curve qPCR reactions were run in triplicate, and all studied samples were run in duplicate. All samples were analyzed on a plate containing a no-template control and a standard curve composed of target-containing plasmids (ITS2 or 16S rRNA) at concentrations of 10⁰–10⁸ copies/reaction. All standard curve samples had a Ct of ≤ 30 (Suppl. Fig. 1). All reactions in which the duplicates differed by > 1 Ct were repeated. In addition, during sample handling in the biosafety cabinet, 2 tubes (sterile and DNA-free) were left open and then processed as above, serving as environmental controls for DNA extraction.

Data were analyzed using Prism v.8.3 (GraphPad). Data normality was assessed by the Shapiro–Wilk test. The nonparametric Wilcoxon–Mann–Whitney U-test was used to compare the sample-collection techniques (carpet vs. toothbrush). To evaluate the differences among clinical groups (diseased, subclinical, and negative control), the Kruskal–Wallis test with Dunn post-test was used; p < 0.05 was considered statistically significant.

We analyzed 92 samples. The x̄ weight of the 45 carpet-collected samples (16 ± 12 mg) and 45 toothbrush-collected samples (20 ± 11 mg) did not differ significantly (p = 0.221, Mann–Whitney U-test; Fig. 1A). In addition, no difference in sample weight was found between clinical groups (p = 0.811, Kruskal–Wallis test; Fig. 1B).

Figure 1.

Box plot of sample weights. A. Overall comparison between carpet and toothbrush methods, regardless of clinical group. B. Comparison of groups. Boxplots range from the lower quartile to the upper quartile; central line indicates the median; whiskers indicate minimum and maximum values (Kruskal–Wallis test).

Spectrophotometry results revealed that carpet-collected samples yielded an x̄ DNA concentration of 1.75 ng/μL (range: 0.5–32 ng/μL); the toothbrush-collected samples had an x̄ of 2.69 ng/μL (range: 0.4–118 ng/μL), and no statistically significant difference was observed (p = 0.0561, Mann–Whitney U-test). Similarly, there was no significant difference (p = 0.061, Mann–Whitney U-test) when comparing the x̄ DNA yield for the carpet (0.01 ng/μL, range: 0–23.6 ng/μL) and toothbrush (0.03 ng/μL, range: 0–70.8 ng/μL) sample methods when analyzed using the fluorometry test (Suppl. Table 2).

The 16S rRNA gene copies were significantly higher (p = 0.020, Wilcoxon–Mann–Whitney U-test; Fig. 2A) in samples collected using the toothbrush (x̄ of 1,916 gene copies/μL, range: 35–37,200 copies/μL), compared to carpet-collected samples (x̄ of 1,378 gene copies/μL, range: 13–31,900 copies/μL). The bacterial DNA load in samples collected using either method among the 3 groups revealed no significant difference (p = 0.090, Kruskal–Wallis test; Fig. 2B).

Figure 2.

16S rRNA gene qPCR assay results. A. Comparison of carpet and toothbrush methods, regardless of clinical group. B. Comparison of groups. Error bars illustrate 95% CIs (p < 0.05). The toothbrush method harvested significantly more 16S rRNA gene copies/µL than the carpet method (Wilcoxon–Mann–Whitney U-test).

The x̄ ITS2 gene copy number was significantly higher (p = 0.005, Wilcoxon–Mann–Whitney U-test; Fig. 3A) in samples collected using the toothbrush method (x̄ of 573 gene copies/μL, range: 11–17,000 copies/μL) compared to those collected using the carpet method (x̄ of 185 gene copies/μL, range: 7–19,500 copies/μL). No significant difference was found (p = 0.412, Kruskal–Wallis test; Fig. 3B) in collection methods among groups.

Figure 3.

Fungal DNA load based on qPCR targeting the ITS2 gene. A. Comparison of carpet and toothbrush methods, regardless of group. B. Comparison of groups. Error bars illustrate 95% CIs (p < 0.05). The toothbrush method collected significantly more ITS2 gene copies/µL than the carpet method (Wilcoxon–Mann–Whitney U-test).

Comparing microbial DNA loads between the methods, we found a 1.4 times higher bacterial load and a 3 times higher fungal load (p = 0.018, Kruskal–Wallis test) in toothbrush-collected samples. Considering both sampling methods together, the total bacterial DNA load was significantly higher (p = 0.017, Wilcoxon–Mann–Whitney U-test; Fig. 4A) than the fungal amount. We attributed this difference to a higher bacterial DNA load in the clinically diseased group (p = 0.040, Wilcoxon–Mann–Whitney U-test; Fig. 4B), compared to fungal DNA levels.

Figure 4.

Findings from qPCR assay targeting the 16S rRNA and ITS2 genes. A. Overall comparison of bacterial and fungal amounts, regardless of group. B. Comparison of groups regarding collection methods. Boxes depict interquartile range. Whiskers illustrate 95% CIs. The amount of bacterial DNA was significantly higher (Wilcoxon–Mann–Whitney U-test) than the amount of fungal DNA. The diseased group had more bacterial DNA than fungal DNA (Kruskal–Wallis ANOVA). No differences were found between the subclinical and negative control groups (Kruskal–Wallis test).

Negative control samples (n = 2 per sampling method) had no detectable amounts of DNA by spectrophotometry or fluorometry. The average 16S rRNA gene copies from negative control toothbrushes and carpets were 11/μL and 6/μL, respectively. The average ITS2 gene copies were 8/μL for toothbrushes and 12/μL for carpets. DNA extraction controls averaged 35 16S rRNA gene copies/μL and 0.5 ITS2 gene copies/μL.

We found that both techniques recovered enough bacterial and fungal DNA for detection using qPCR. However, the toothbrush technique yielded significantly higher levels of bacterial (1.4×) and fungal (3×) DNA compared to the carpet method. We suggest that the toothbrush method is more effective in collecting microbial DNA for molecular analysis.

Although we did not find statistical differences in sample weight collected by either method, the toothbrush method harvested numerically higher amounts of hair and scale target DNA loads than the carpet method. Although the surface area of a toothbrush is smaller than the carpet, the toothbrushes had a longer bristle length. This may help explain the differences in microbial DNA material collected. We hypothesized that toothbrushes, with their longer bristles compared to carpets, were able to sample closer to the skin where conditions may be more favorable for microbial growth and the concentration of sloughed-off epithelial cells is higher. ⁶

The concentration of DNA detected by spectrometry and fluorometry was low in both sampling techniques, as expected for skin samples compared to mucosal sites. ⁸ Spectrophotometry is a rapid, inexpensive method to evaluate DNA load and purity within samples for analyses. ¹⁴ Compared to fluorometry, spectrometry lacks specificity, and results are generally overestimated because of contaminants, such as proteins; DNA extraction residues (ethanol) or other large molecules are also detected, including those of host origin. ¹⁹ This lack of specificity is likely the cause of the difference in values obtained between spectrophotometry and fluorometry. Spectrophotometry results were an overestimate, presumably as a result of the detection of molecules other than DNA in the extracted samples as evidenced by the 260/280 nm and 260/230 nm absorbance values reported. These values have a wide range, suggesting the low purity of the DNA extracted. Similar findings have been reported when comparing the performance of fluorometry to spectrophotometry and qPCR using different initial samples. ¹² Fluorometry is used widely to quickly assess double-stranded or single-stranded DNA concentration. Various studies have indicated that, compared to spectrophotometry, DNA fragmentation led to an underestimation of the nucleic acid content by fluorometry and PCR.^7,17 Although we did not assess DNA fragmentation, it may have had an impact on the DNA load detected.

The successful qPCR performed using samples in our study suggests that the microbial DNA load obtained by either sampling method was sufficient for amplification at an acceptable threshold (Ct < 30), even though the amounts detected by spectrophotometry and fluorometry were numerically low. The qPCR assays used targeted only microbial DNA (bacterial or fungal), whereas the non-PCR techniques used are nonspecific and detected DNA from any source (including contaminants). Hence, both techniques that we used are suitable for detection. It is accepted that real-time PCR assays have a dynamic detection range and a very high analytical sensitivity (< 5–10 target DNA copies). ¹⁰ Therefore, it is not surprising that levels of fungal and bacterial DNA were detectable by PCR.

Interestingly, our findings suggested that cats with dermatophytosis (diseased group) harbored 3 times more bacteria than fungi in their hair and scales. This was different from the 2 clinically normal groups. Although similar data have not been reported for domestic cats, studies of human normal skin found a similar pattern throughout different body sites. ² It is known that each bacterial genome harbors on average four 16S rRNA gene copies, and individual fungal genomes have an average of 113 ITS copies.^9,20 Thus, there is inherent overestimation of the bacterial and fungal load in each sample when using these genes for qPCR analysis. Even though the fungal loads that we found were largely overestimated compared to the bacterial load, we still detected more bacterial than fungal DNA in the diseased group samples, compared to the controls. Although the absolute numbers may not be precise, the fungi:bacteria ratio that we found is still remarkable. We postulate that cats with clinical disease did not have an increased fungal load of M. canis outcompeting and limiting the growth of other fungi. This remains to be confirmed by future research.

Although diseased and subclinical animals in our study were all long-haired, the negative control group consisted of short-haired animals. This could bias the results, but hair length did not seem to significantly affect the amount of sample or microbial DNA recovered given that we did not observe differences among the 3 groups. Although the swab is the most widely used sampling tool for molecular investigation of fungi or bacteria communities in skin and it collects primarily scales,^13,15 we suggest that the toothbrush is also a suitable tool for hair and skin scale collection, providing a higher load of genetic material than the carpet.