Effects of sampling location and frequency on the detection of Mycoplasma gallisepticum populations in the respiratory tract of commercial layer pullets

Abstract

Mycoplasma gallisepticum (MG) is an economically important pathogen in layers, as it is responsible for chronic respiratory disease. Routine detection of MG is performed via swabbing of the trachea or choanal cleft. However, the possible impact of repeated sampling events on in vivo MG population dynamics within individual subjects has not been investigated. Therefore, eighty 11 wk-old Hy-Line W-36 layer pullets were infected with F-strain MG, and were individually caged in one of 4 identical rooms, with 20 birds per room. Two wk post-infection, 5 birds in each room were randomly assigned to either 2, 4, 8, or 16 d sampling interval schedules, and were maintained for an additional 32 d. At each sampling event, the birds were swabbed in the choanal cleft and trachea, and sample-associated MG populations were quantified using real-time PCR. A complete block experimental design was employed, with room as an experimental block containing multiple replicates of each sampling schedule. Data, within birds, was subjected to repeated measures analysis. Undetected bacterial populations were considered as a negative sample. There was no evidence of any difference in probability of detecting a positive sample or in trend over time of microbial abundance due to the different sampling schedules. However, there were ∼26 times as many genomic equivalents of MG in the choanal cleft than in the trachea across all the sampling intervals, and significantly greater probability of positivity in the choanal cleft at d 16 and d 32. It was concluded that the site of sampling influenced MG population with the sampling interval exerting no additional influence.

Introduction

Mycoplasma gallisepticum (MG) is an economically important pathogen of avian species (Ley, 2008), and is a major contributor to chronic respiratory disease in chickens and infectious sinusitis in turkeys (Edward and Kanarek, 1960). It colonizes the upper respiratory tract of infected hosts and is commonly associated with infraorbital sinuses, choanal cleft, trachea, air sacs, and nasal turbinates (Henderson and Jensen, 2006). Susceptibility and disease progression vary by host species and age, and once infected, birds are considered carriers of MG for life (Armour and Ferguson-Noel, 2015). Mycoplasma gallisepticum is highly transmissible and available means of its control are limited (Ishfaq et al., 2020). While antibiotics have been used to reduce the economic impact of MG, they are unable to clear MG from infected hosts, and biosecurity and bio-surveillance remain the primary barriers to pathogen propagation (Garmyn et al., 2019). Among commercial layers, live vaccines may be employed to reduce the economic impacts of MG, but no wholly effective means of control exist, and development of future means of control is limited by the availability of tools and a knowledge of MG and its associated pathobiology (Ishfaq et al., 2020).

Traditionally, swabs have served as sources for MG culture and labeled species-specific antibodies have been used for species identification (Limsatanun, 2022). With the advent of molecular technologies, real-time PCR is now commonly used to diagnose MG infections (Lysnyansky et al., 2005; Limsatanun, 2022). This technology, which relies upon the detection of MG DNA, is fast and sensitive and can be used to differentiate MG strains or isolates. In addition, real-time diagnostics may allow for the absolute quantification of MG by analyzing results against serially diluted standards (Mekkes and Feberwee, 2005).

In vivo tracheal and choanal cleft MG populations have been shown to vary widely among MG-infected poultry, but only limited information currently exists (Mekkes and Feberwee, 2005; Callison et al., 2006). The ability to accurately assess and quantify in vivo MG populations within a single host may afford researchers an important tool towards the development of further means of control and provide an additional metric for elucidating novel treatment effects. To this end, this research was undertaken to quantify the in vivo populations of MG associated with 2 commonly used anatomical sites (Branton et al., 1984). In addition, a variety of scheduled sampling interval treatments (TRT) were evaluated to determine the impact of repeated sampling on the in vivo MG populations of the choanal cleft and trachea within a single host.

Materials and methods

Experimental design

Eighty MG-free Hy-Line W-36 layer pullets were eye-drop vaccinated with Poulvac® Myco F vaccine (Zoetis, Parsippany-Troy hills, NJ) at 11 wk of age according to recommendations from the manufacturer. Two wk post-infection, 5 pullets were randomly assigned to one of 4 sampling interval TRT groups (5 birds/ TRT) in each of 4 separate rooms (20 total birds per room). The birds were individually caged and remained housed for 32 d. The TRT groups were 2-, 4-, 8-, or 16-d sampling intervals. Each cage provided > 1290.32 cm² floor space and each bird was separated by 3 empty cages horizontally and was vertically offset in order to reduce chances of horizontal MG transmission (Fig. 1). The birds were provided antibiotic-free feed and ad libitum water via a nipple drinker. The feed was formulated to meet or exceed NRC requirements (NRC, 1994). The temperature setpoint of each room followed that of the Hy-Line Management Guide (Hy-Line Red Book, Hy-Line International, 2020) and was set at 21°C. The bird handling and management practices were reviewed and approved by the USDA-ARS Animal Care and Use Committee (AUP 22-1).

Fig. 1.

A single room layout of the 4 sampling frequency treatments (TRT), with 5 birds belonging to each TRT in a room. The 5 birds belonging to each TRT were randomly arranged in the room, and at least one empty cage was used to separate the caged birds.

Sampling and PCR analysis

The trachea and choanal cleft of each bird were individually swabbed with sterile pre-wetted swabs (Puritan® sterile cotton tipped applicators, Puritan Medical Products Co LLC, Guilford, ME, REF 25-800 C 50) according to the sampling schedule and the swab-affixed material was transferred to sterile phosphate-buffered saline (100 µL). Each site was swabbed by the same individual for uniformity and birds were minimally handled to reduce stress. None of the birds showed any clinical signs of disease and no mortality was observed. Genomic DNA was extracted from a 100 µL volume of the above solutions using a BioSprint 96 One-For-All Kit (Qiagen, Valencia, CA, USA) following the manufacturer’s recommendations. All swab-derived DNA samples were run in duplicate using a QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). The strain F MG-specific primer/probe set (Table 1) and reaction conditions were described by Raviv et al. (2008).

Table 1.

Primers and probe used for DNA analysis.

Item		Base Sequence
Primer		5′-GTTCAAGAACCAACTCAACCA-3′
Primer		5′-GATTAAGACCGAATTGTGGATTG-3′
Probe		5′-FAM-CAACCAGGATTTAATCAACCTCA-BHQ1-3′

Reactions were performed using 25 µL solution volumes, with diluted primers and probe of 0.5 µM and 0.1 µM, respectively. Forward and reverse primers and the probe with volumes of 2.5 µL each, a 12.5 µL volume of Taqman Universal Master Mix (Applied Biosystems, Waltham, MA, USA), and 2.5 µL volume of sterile water (Raviv et al., 2008) were used in the reaction. This master mix volume of 22.5 µL was loaded into each well of a 96-well plate (MicroAmp® Optical 96-well reaction plate with barcode, applied biosystems by life technologies, Life Technologies Corporation, Waltham, MA, REF 4306737), followed by a 2.5 µL volume of sample DNA in each well. The thermocycler program was set at 95°C for 15 min with optics off, and 40 cycles of 94°C for 15 s, and 61°C for 1 min. Each reaction plate included non-template controls (NTCs) and a standard curve (10^-1, 10^-2, 10^-3, 10^-4, 10^-5 & 10^-6) derived from a unique stock solution of an F-strain of Mycoplasma gallisepticum (FMG) template.

Statistical analyses

The study was conducted using a complete block experimental design consisting of multiple replicates per block, with room as the blocking factor. All the sampling schedule TRT were assigned in the same pattern to each block. A Bayesian generalized linear mixed model was fit to determine whether the different sampling frequencies exhibited variation over time in MG population size. The calculation of genomic equivalents (GE) was described by Garmyn et al. (2017). The response variable in the model was MG population converted from C_t (critical threshold value) to units of GE, assuming a hurdle-lognormal distribution. The DNA concentration per mgc2 gene was determined to find the Genomic equivalence. This error distribution models the probability of observing a 0 (negative result for MG) as well as the variation in MG population size assuming a non-zero value is observed. The model included sampling frequency (categorical variable), sampling day (continuous variable), location (binary variable indicating choanal cleft or trachea), and their two-way interactions as fixed effects. A random intercept was fit to each room and to cage nested within room. The hurdle term was allowed to vary as a function of the fixed-effect main effect and interaction terms. Two models were fit, each with different repeated measures terms, a spline smoothing term and a factor smooth term varying by sampling schedule, to account for temporal autocorrelation in the response.

Normal (0, 2) prior distributions were assigned to all fixed-effect coefficients on the response, and Normal (0, 1) prior distributions were assigned to all fixed-effect coefficients on the hurdle term. Posterior distributions were sampled using Hamiltonian Monte Carlo with 4 Markov chains, each running for 3000 warmup iterations and 1000 post-warmup sampling iterations. The R-hat (< 1.01) statistics for the 2 models indicated that the models converged (Vehtari et al., 2021). The 2 repeated measures terms were compared using leave-one-out cross-validation and it was found that the factor smooth was a better fit (difference in ELPD = 8.4). Therefore, results from the factor smooth model are presented. The posterior distribution of the expected value of MG abundance for all combinations of fixed effects and their 95% quantile credible intervals (QCI) were calculated. Statements of statistical evidence were based on the Bayesian probability of direction for the difference between posterior means, expressed as a two-sided p-value using the formula 2(1 - p_d).

The proportion of birds testing positive within each treatment on D 0, 16, and 32 were analyzed using generalized linear mixed model with a binomial distribution and logit link. Fixed effects were sampling schedule, site, day, and their interactions (all discrete). Random intercepts were fit to room (replicate) and room by sampling schedule interaction. Normal(0, 5) priors were placed on the fixed effects to mitigate complete separation. Joint Chi-squared tests were used to assess each model term’s significance within each of the three days. Marginal probabilities of positivity were estimated for each combination of site and treatment, and site and treatment overall, within each day, and separated using the Tukey HSD procedure, with differences considered significant at adjusted P < 0.05.

Analysis was done using R software version 4.2.2 (R Core Team, 2022) and Stan software version 2.30 (Stan Development Team, 2022), including the packages brms version 2.18.0, cmdstanr version 0.5.3, glmmTMB version 1.1.11, and emmeans version 1.8.2.

Results and discussion

Percentage positivity analysis

There was no significant main effect due to sampling interval TRT, and no significant sampling site × sampling interval TRT interaction on the probability of positivity for MG on D 0, 16, and 32 (Table 2). There was also no significant main effect due to sampling site TRT on the probability of positivity for MG on D 0 (Table 2). However, on D 16 and on D 32, significantly higher positivity rates for MG occurred in the choanal cleft than in the trachea (P = 0.0008 and P < 0.0001, respectively). In a previous report by Callison et al. (2006), the sensitivity of real-time PCR from MG-infected broilers compared to the conventional PCR assay was validated. However, in that study, the choanal cleft samples were taken for real-time PCR analysis, whereas tracheal samples were evaluated using conventional PCR. The average estimated choanal cleft template copies varied between 1.7 × 10² and 2.5 × 10⁸ from breeder flocks infected at one wk-of-age. In a study by Branton et al. (1984), it was substantiated that better isolation rates of FMG occurred in the choanal cleft than in the trachea. It could be speculated that the anatomical proximity of the nasal cavities (Nickel et al., 1977), along with reduced distress due to the greater accessibility of the choanal cleft (Branton et al., 1984), as well as less interference with PCR inhibitors, make the choanal cleft a more suitable site for MG isolation. In a study by Sprygin et al. (2010), it was mentioned that PCR inhibitors from samples were removed from the tracheal samples taken, however, nothing was mentioned about the intensity of PCR inhibitors present in the tissue sites they considered.

Table 2.

Main and interactive effect means (± SE) for swab sampling interval (2, 4, 8, and 16 d) and site of swabbing (trachea and choanal cleft) treatments for probability of positive Mycoplasma gallisepticum (MG) observations on D 0, 16, and 32¹.

	Modeled probability of positive sample (%)
Treatment	D 0	D 16	D 32
Sampling interval (d)
2	99.5 ± 0.7	93.8 ± 5.7	86.8 ± 7.9
4	99.8 ± 0.5	85.8 ± 8.1	74.1 ± 10.1
8	96.3 ± 3	87.7 ± 6.6	91.8 ± 9.1
16	98.8 ± 1.7	80.3 ± 8.5	80.9 ± 10.2
Sampling site
Trachea	98.2 ± 1.5	71.3a ± 8.7	51.7a ± 10.2
Choanal cleft	99.6 ± 0.6	95.4b ± 3	96.6b ± 2.6
Sampling interval within site
Trachea
2	98.7 ± 1.6	69.1 ± 12.9	61.2 ± 13.9
4	99.5 ± 1	61.1 ± 13.9	55.6 ± 14.3
8	95.8 ± 4.1	81.8 ± 10.1	45.1 ± 14.5
16	96.6 ± 3.6	70.8 ± 12.6	44.5 ± 14.3
Choanal cleft
2	99.8 ± 0.4	99 ± 1.7	96.5 ± 3.8
4	99.9 ± 0.4	95.9 ± 4.3	86.8 ± 8.4
8	96.8 ± 3.6	91.9 ± 6.4	99.4 ± 1.5
16	99.6 ± 1	87.2 ± 8.2	95.7 ± 4.4
P-value2
Interval	0.51	0.59	0.56
Site	0.32	0.0008	< 0.0001
Interval × Site	0.84	0.31	0.37

^a,bDifferent superscript letters within the Site category denotes a significant (P < 0.05) difference.

¹In each sampling interval and site, modeled probability of positive MG observations (N = 4) based on 5 birds in each of 4 replicate rooms (blocks).

²P-values based on joint Chi-squared tests, df = 3 for interval, df = 1 for site, df = 3 for interval × site.

Genomic equivalents analysis

The quantitative analysis in the current study provides strong evidence that the choanal cleft (mean GE 1.6 × 10⁵, 95% QCI [4.9 × 10³, 5.6 × 10⁵]) has a greater MG abundance than the trachea (mean GE 6.5 × 10³, 95% QCI [1.8 × 10², 1.2 × 10⁴]), with a 25.9-fold difference in GE and a 95% QCI ranging between 17.6 and 37.7 (P < 0.001; Table 3). Mekkes and Feberwee (2005) demonstrated from research concerning tracheal swabs taken from 66-wk-of-age layers that MG populations ranged between 6.31 × 10¹ and 5.01 × 10⁵ CFU. However, their experiments were conducted only on 3, 5 or 10 birds on various d between D 0 and 35. There was no evidence for a trend over time in the in vivo MG populations of the choanal cleft and trachea between D 0 and 32 in the current study (Fig. 2), suggesting that the frequencies of swabbing across the 2 tissue sites employed in this study led to no appreciable differences in in vivo MG populations. It is speculated that the 4, 8, and 16 d interval swabbing events may not have impacted the in vivo MG populations. The only exception occurred in the 2-d schedule, in which the average microbial abundance was approximately threefold lower than that in the 8-d and 16-d schedules (Table 3). However, the evidence for this difference was relatively weak, with only a 90% credible interval or less excluding the zero difference (P = 0.10; Table 3). The majority (17 of 20) of the observed GE values of the choanal cleft samples on D 6 were zero (Fig. 3).

Table 3.

Modeled ratios of abundance means averaged across time, between pairs of sampling frequency treatments within each sampling location and averaged over location, and overall between the two sampling locations, with 95% quantile credible intervals and Bayesian p-values.

Contrast	Median estimated ratio	95% QCI	P-value
Choanal cleft
2 / 4	1.78	[0.61, 5.36]	0.30
2 / 8	0.64	[0.22, 1.94]	0.42
2 / 16	0.61	[0.2, 1.98]	0.40
4 / 8	0.36	[0.11, 1.25]	0.10
4 / 16	0.34	[0.1, 1.26]	0.10
8 / 16	0.93	[0.27, 3.37]	0.92
Trachea
2 / 4	1.56	[0.52, 4.76]	0.44
2 / 8	0.61	[0.19, 1.93]	0.40
2 / 16	0.89	[0.28, 3.25]	0.87
4 / 8	0.39	[0.11, 1.38]	0.14
4 / 16	0.57	[0.15, 2.4]	0.43
8 / 16	1.49	[0.38, 6]	0.58
Overall
2 / 4	1.77	[0.6, 5.28]	0.30
2 / 8	0.64	[0.22, 1.92]	0.42
2 / 16	0.61	[0.2, 1.99]	0.41
4 / 8	0.36	[0.11, 1.24]	0.10
4 / 16	0.35	[0.1, 1.26]	0.11
8 / 16	0.94	[0.28, 3.37]	0.94
Choanal cleft / Trachea	25.91	[17.63, 38]	< 0.001

Fig. 2.

Modeled predicted trends of Mycoplasma gallisepticum bacterial abundance in the trachea (left panel) and choanal cleft (right panel) over time period (Day 0 to 32). Solid lines represent the posterior median trend and dashed lines represent the upper and lower bounds of the 95% credible interval for each sampling frequency within each sampling location. The figure illustrates the lack of statistical evidence for abundance differences among sampling frequencies within a location.

Fig. 3.

Observed Mycoplasma gallisepticum microbial abundance in genomic equivalents (GE) over time for each sampling frequency in the trachea (left panel) and choanal cleft (right panel).

Raw data points are plotted as hollow circles, jittered in the x direction so that overlapping data points are visible. Medians for each sampling frequency at each time point are plotted as solid circles connected with lines. The y-axis is transformed with a log(x + 1) transformation to allow zero values to be plotted.

Error bars represent the central 50% quantile range around each median.

The impact of repeated different sampling events on the in vivo MG populations at 2 tissue sites (choanal cleft and trachea) were compared. In conclusion, the 2-d interval could be used as an ideal swabbing interval since it statistically validates the results with adequate data. Compared to other studies, in which different birds were used to swab the choanal cleft and trachea (Shingade et al., 2020), the same birds were used in the present study. This enabled MG colonization between both sites to be distinguished and compared. In a study by Callison et al. (2006), it was detected that 94% of samples on average possessed 1.9 × 10⁶ template copies/ reaction from palatine cleft swabs of 171 MG positive samples.

As expected, birds in the current study did not show any clinical signs of infection. Nevertheless, a large difference in overall MG populations was observed between the choanal cleft and the trachea. At this time, no swabbing methods have been developed to reduce the potential for tissue damage and distress to the birds (Abundo et al., 2020). Wilson et al. (2021) confirmed that vaccines elicited an increased tracheal inflammatory response when compared to control groups. These present results will provide future information toward avoiding possible confounding factors which might interfere with other controlled vaccine challenge studies. However, future studies comparing the positivity rates in both sites using the same bird, and by using different diagnostic tests, would enable a better understanding of the variation in positivity rates among different assays. Moreover, a study design that includes PCR inhibitor removal might also need to be employed to further remove this as a confounding factor. Antibody tests such as the Serum Plate Agglutination and the Enzyme-Linked Immunosorbent assays could also be performed to confirm the presence of antibodies against MG and the severity of infection (Ewing et al., 1996). In future studies, novel intervention strategies, including choanal cleft swabbing in 2 d intervals, could be used to lead to a better understanding of the mechanisms behind MG infections.

CRediT authorship contribution statement

S.K. Sasidharan: Writing – original draft, Visualization, Supervision, Methodology, Investigation, Data curation, Conceptualization. K.E.C. Elliott: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Investigation, Data curation, Conceptualization. J.D. Evans: Writing – review & editing, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Q.D. Read: Writing – review & editing, Formal analysis, Data curation. E.D. Peebles: Writing – review & editing, Supervision, Project administration, Methodology, Investigation, Data curation.

Disclosures

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.