Use of Deep Amplicon Sequencing Based on the Cytochrome Oxidase I Gene to Quantify the Relative Percentages of Eimeria spp. Oocysts in Poultry Litter Open Access

Avian coccidiosis caused by Eimeria spp. and necrotic enteritis (NE) caused by Clostridium perfringens in association with Eimeria infection are regularly listed by poultry veterinarians as the top diseases affecting broiler performance. Accurate diagnosis of Eimeria infection is important not only to identify the cause of disease symptoms, but also to uncover possible Eimeria drug resistance or immunovariability. For instance, overt intestinal lesions or the presence of Eimeria oocysts in intestinal scapings (i.e., microlesion scoring) during drug treatment could indicate emerging drug resistance. Necropsy and examination of intestinal tissues for typical coccidian lesions and oocysts are helpful in initial evaluation of coccidiosis. However, lesion scoring requires technical expertise and cannot resolve mixed Eimeria infections, especially in regions of the gut in which more than one Eimeria species invade and replicate. As one way of overcoming the inherent ambiguity of this diagnostic method, researchers have developed molecular methods for identifying the Eimeria species in litter and intestinal tissue. These techniques typically use single or nested PCR to amplify the internal transcribed spacer 1 (ITS1) or internal transcribed spacer 2 (ITS2) ribosomal DNA followed by gel electrophoresis or capillary electrophoresis (1,2,3,4,5,6,7,8,9,10,11,12). Another popular technique utilizes sequence-characterized amplified region (SCAR) markers that can be used in a multiplex PCR followed by gel electrophoresis or in real-time PCR to simultaneously identify seven Eimeria species infecting chickens (13,14). PCR using ITS1, ITS2, or SCAR primers has been used by numerous groups to determine the Eimeria species composition in fecal droppings and litter (15,16,17,18,19,20,21,22,23,24,25,26,27,28). These methodologies are limiting because the copy number of ribosomal DNA and SCAR marker genes among and between different Eimeria spp. is thought to vary (29), so it is impossible to accurately quantify the relative amounts of each particular Eimeria sp. in a mixed sample. In this study, a deep amplicon sequencing approach was taken to overcome this problem by using generic primers targeting the single-copy cytochrome oxidase I (COI) gene on mitochondrial DNA. Our initial studies to test this approach on DNA extracted from a single species or a mixture of multiple Eimeria species oocysts validated the assay. This approach was applied to DNA extracted from a mixture of Eimeria oocysts present in litter from commercial broiler farms at 0, 2, and 4 wk after chick placement (growout). This technique avoids the labor-intensive ITS1-based PCR methodology and allows for estimation of relative Eimeria species abundances in complex poultry litter, which can help to uncover associations between disease and levels of Eimeria presence that cannot be distinguished from one another by microscopy.

Eimeria acervulina (strain APU1), Eimeria maxima (APU1), and Eimeria tenella (APU1) oocysts were isolated from fecal material at days 5–7 postinfection (depending on Eimeria sp.) using a standard propagation procedure (30). An aliquot of these oocysts was immediately stored at 4 C to prevent sporulation, whereas the remainder of oocysts were incubated in 2-L flasks with aeration at 29 C in a shaking water bath. Propagations and processing of each Eimeria sp. oocyst sample were conducted in different buildings and on different days to prevent cross-contamination. Sporulated oocysts of all three species were determined to have the same percent sporulation by McMaster chamber cell count before being pooled in equal oocyst numbers in the same tube for DNA extraction.

Unsporulated or sporulated E. acervulina, E. maxima, or E. tenella oocysts were pelleted by centrifugation at 1500 × g for 10 min, followed by sterilization with 6.5% sodium hypochlorite for 30 min at room temperature on a rocker to remove contaminating bacteria and organic material. After sterilization, the oocysts were subjected to four to five wash steps involving dilution in deionized H₂O and centrifugation at 1500 × g for 10 min. Eimeria acervulina, E. maxima, and E. tenella oocysts were extracted for DNA either singly or as an equal mixture of 10⁵ oocysts/tube. The mixing experiments with unsporulated or sporulated oocysts were done in triplicate, whereas analyses of individual Eimeria species were done singly. DNA extraction followed a standard protocol using the QIAmp Fast DNA Stool Mini Kit (Qiagen, Valencia, CA) with an additional bead-beating in 500 μl of InhibitEX buffer with approximately 200 mg of 0.5-mm glass beads (Bio-Spec Products, Bartlesville, OK) using a BioSpec Mini-Beadbeater for two cycles of 2 min to aid the degradation of the Eimeria oocyst and sporocyst walls. The Qiagen protocol was followed using 400 μl of bead-beating supernatant and dilution in AL buffer and EtOH. After column elution, a standard EtOH precipitation was performed, followed by 70% EtOH washing, air drying at room temperature, and resuspension in 10-mM Tris at pH 7.6 and 1-mM EDTA (TE) for quantification on a Qubit using the 1X dsDNA HS Assay kit (Invitrogen, Waltham, MA).

Litter samples were collected from a total of 20 individual broiler houses encompassing six different commercial poultry farms on the Maryland Eastern Shore during an anticoccidial drug program (starter: 0.0125% nicarbazin; grower: 0.0125% zoalene). These farms were chosen based on incidence of NE (Farms NE-1, NE-2, NE-3 = NE negative; Farms NE+1, NE+2, NE+3 = NE positive) as defined by chick mortality. Aside from one farm (Farm NE+2, with 0 and 2 wk placement only), litter was collected at 0, 2, and 4 wk after chick placement (growout). Approximately 50-g samples of litter were collected into 250-ml polypropylene bottles from 15 different locations in the brooder area of each house. The litter samples were weighed in order to calculate Eimeria oocysts/g, saturated with tap water to 250 ml, and incubated overnight at 4 C. The Eimeria oocysts were enumerated using standard procedures (31). An aliquot of the litter slurry was processed for isolating Eimeria oocysts by sucrose flotation followed by sterilization and DNA extraction and quantification as above.

The Eimeria species composition of oocysts recovered from broiler house litter was determined using standard ITS1 PCR (4). The reactions were electrophoresed on polyacrylamide gels and visualized by EtBr staining; image capture was done using a GelLogic 200 Imaging System (Kodak, Rochester, NY). The density of the target band was estimated visually and scored on a scale from 0 to 3.

The metagenomic protocol employed was a modified version of the 16S Metagenomic Sequencing Library protocol by Illumina (no. 15044223 Rev. B). Amplicon locus-specific primers with additional overhang adapter sequences (mEimeria COI-F1 and mEimeria COI-R) were designed by aligning 32 Eimeria COI sequences collected from complete mitochondrial sequences in GenBank and searching for conserved sequences among all the Eimeria spp. Each individual sample was subjected to 35 cycles of primary amplification according to the protocol, resulting in an ∼260-bp product. The mEimeria COI-F1 primer sequence TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGGTTAAGTACTTATATGGGWAATCC and mEimeria COI-R sequence GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGATCCAATAACCGCACCAAGAG contained Illumina tags (underlined). Illumina unique dual (UD) indexes were added for sample identification, bead purified, quantified again by Qubit with a 1X dsDNA HS Assay kit (Invitrogen), normalized, and diluted to a 4-nM final concentration. A pooled, denatured, and diluted 6pM library was run with 5% PhiX control on an Illumina MiSeq V3 cartridge with 300-bp paired-end reads from all of the individual libraries of pooled samples. The read length allowed complete overlap between forward and reverse reads.

For each sample, raw forward and reverse reads in fastq format were quality checked using FastQC version 0.12.1. Adapters were trimmed from reads using the BBDuk script and adapters reference file from BBTools ver. 39.01. Next, the primers GGTTAAGTACTTATATGGGWAATCC and ATCCAATAACCGCACCAAGAG were removed from reads using BBDuk. Cleaned reads were imported to QIIME 2 Amplicon Distribution ver. 2024.2 (32) and denoised using the DADA2 plugin, which performs quality filtering, read merging, and chimera filtering and is able to correct Illumina-sequenced amplicon errors (33). DADA2 truncation settings were set to 176 for forward reads and 180 for reverse reads to exclude positions where median quality scores dropped below Q20. DADA2 produced 71 amplicon sequence variants (ASVs), and the ASVs were input to blastn in BLAST+ ver. 2.15.0 as queries against a custom database of COI sequences from seven Eimeria species known to infect chickens, including E. acervulina, E. maxima, E. tenella, Eimeria brunetti, Eimeria mitis (and those labeled as Eimeria mivati), Eimeria necatrix, and Eimeria praecox, and three cryptic Eimeria operational taxonomic units (OTUs) that infect chickens currently referred to as Eimeria lata (OTU-X), Eimeria nagambie (OTU-Y), and Eimeria zaria (OTU-Z) (34,35). Sequences were obtained by searching the National Center for Biotechnology Information nucleotide database, and the majority of sequences were previously reported in phylogenetic analysis (36). The custom database of Eimeria COI sequences included 10 sequences from E. acervulina, one from E. brunetti, 54 from E. maxima, three from E. mitis, four from E. mivati, two from E. necatrix, two from E. lata (OTU-X), two from E. nagambie (OTU-Y), three from E. zaria (OTU-Z), two from E. praecox, and 40 from E. tenella. The best target sequence result in BLAST was used to assign an Eimeria species to the query ASV. ASVs with the same species assignment were summed to calculate the total read count and relative abundance of each Eimeria species in a sample. Despite the possibility of cross-mapping, denoising sequences to ASVs with the DADA2 bioinformatics package in our study resulted in no BLAST hits to E. necatrix. We found utilization of DADA2 to be most effective in avoiding cross-mapping during bioinformatics analysis, likely because the package specializes in correcting amplicon sequence errors from Illumina sequencing (33).

The percentages of reads mapping to E. acervulina, E. maxima, or E. tenella in the oocyst mixing studies were compared for statistical differences using ANOVA GraphPad InStat Software (GraphPad Software, San Diego, CA) with significance at P = 0.05. No published sequences differed by more than 0.5% within any species, while no two species differed by less than 2.1% (E. necatrix and E. tenella), supporting high specificity for this assay. For NE-negative farms (NE-1, NE-2, NE-3), the average total number of reads was 81,280 (NE-1: 80,709 reads; NE-2: 79,438 reads; NE-3: 81,280 reads). For NE-positive farms (NE+1, NE+2, NE+3), the average total number of reads was 80,784 (NE+: 76,361 reads; NE+2: 70,480 reads; NE+3: 95,510 reads). For the litter sample Eimeria oocysts, a Pearson’s correlation coefficient (R) was calculated using InStat by comparing the percentage reads mapping to each Eimeria species to NE status. Samples from NE-negative farms were assigned a value = 0, while samples from NE-positive farms were assigned a value = 1.

Metagenomic analysis of samples containing a single Eimeria sp. of unsporulated or sporulated oocysts showed nearly 100% mapping of sequences to the appropriate Eimeria sp. COI (Table 1). Mapping of reads to the respective COI sequence of DNA from sporulated and unsporulated E. acervulina oocysts were similar (99.7% and 99.3%), as well as DNA from sporulated and unsporulated E. tenella oocysts (100.0% and 99.9%). It was interesting that for sporulated E. maxima oocysts, virtually all reads (99.7%) mapped to E. maxima, but unsporulated oocysts mapped 96.9% to E. maxima and 3.0% to E. acervulina (Table 1). Examination of the E. maxima suspension by microscopy did not reveal any contamination with small Eimeria oocysts, so it remains unclear why there was a difference in the read mapping between unsporulated and sporulated E. maxima oocysts.

Deep amplicon sequencing analysis of mixtures of equal numbers of E. acervulina, E. maxima, and E. tenella oocysts, whether sporulated or unsporulated, produced reads mapping in nearly the expected percentages (Table 1). For mixtures of unsporulated Eimeria spp. oocysts, a slightly higher percentage (35.5%) of reads mapped to E. tenella and slightly lower percentage (31.1%) mapped to E. maxima. A different pattern was observed with sporulated Eimeria spp. oocysts, with the highest percentage of reads mapping to E. maxima (36.9%) and lowest percentage mapping to E. tenella (29.5%) (Table 1). Pooling the read mapping of unsporulated and sporulated Eimeria spp. oocysts produced species distributions in about equal percentages (E. acervulina 33.5%, E. maxima 34.0%, E. tenella 32.5%) and not significantly different from each other (P > 0.05). One possible explanation is that there exists partially sporulated oocysts in the unsporulated oocyst sample and incompletely sporulated oocysts in the sporulated oocyst sample. This would lead to be a mixture of ploidy number among E. acervulina, E. maxima, and E. tenella oocysts. Although the majority of Eimeria oocysts recovered from litter are unsporulated, it is inevitable that oocysts are present in various stages of sporulation, and thus the numbers of mitochondria/oocysts may slightly vary.

This finding indicates that amplification of the COI sequence using the CoX-F1/R primers is equally efficient and thus useful for quantitative analysis of Eimeria oocysts in complex mixtures. The only caveat is that because the other Eimeria spp. (E. brunetti, E. mitis, E. necatrix, E. praecox) are not available in pure form, it is not possible to determine if PCR using the Cox-F1/R primer pair is similarly efficient with Eimeria species other than E. acervulina, E. maxima, and E. tenella. However, the design of both the Cox-F1 and Cox R primers was done by first aligning the COI sequences from the seven named Eimeria species and searching for conserved regions across all species, leading to amplification of an ∼600-bp product. Moreover, E. acervulina, E. maxima, and E. tenella are more phylogenetically distant from each other, and the other four species are interspersed between them on a phylogenetic tree (37,38,39). Thus, there is no reason to believe that the COI sequences of E. brunetti, E. mitis, E. necatrix, and E. praecox would not amplify as efficiently as with E. acervulina, E. maxima, and E. tenella. It should be noted that the most likely cross-mapping would be expected between E. tenella and E. necatrix, which only differ by three single nucleotide polymorphisms (SNPs) across the amplified segment of COI. All other species pairs differ by at least 14 SNPs, limiting the likelihood of mismapping. However, the dada2 algorithm can robustly distinguish amplicons that differ by as little as one nucleotide (32), and testing with a clustering method such as VSEARCH gave results identical to denoising, with 0% mapping to E. necatrix. Nevertheless, caution is warranted when E. tenella and E. necatrix are both observed in a sample. In the present study, ITS1 PCR, which can differentiate E. tenella from E. necatrix, did not reveal the latter in any litter sample. This suggests that the percentage of E. tenella as estimated by COI deep amplicon sequencing is accurate. Nevertheless, more exhaustive analysis of cross-mapping is recommended in this situation to ensure that proportions of E. tenella and E. necatrix are accurate.

Deep amplicon sequencing analysis provides a rapid single-tube way to determine the composition and relative abundance of Eimeria spp. in a sample. Indeed, this assay allowed for estimation of the Eimeria species composition in litter at 0, 2, and 4 wk growout (Fig. 1A–C). However, it cannot give an estimate of the absolute number of a particular Eimeria species. This is only possible by microscopy, which is limited to counting of E. maxima and non–E. maxima oocysts. Used together though, the level of a particular Eimeria species can be estimated by multiplying the relative number (%) of that species by the total oocyst concentration. Preliminary studies using the COI primers revealed a sensitivity of about 100 oocysts, suggesting that Eimeria existing as 1% of the population would be detected in this assay (unpubl. observations by Peter Thompson). Typical litter from commercial broiler houses contains on average 10⁴ oocysts/g. In our experience, Eimeria oocyst levels in litter above 10⁵ oocyst/g reflect either drug resistance or incomplete immunity. The deep amplicon sequencing technology described in this paper extends our understanding of increased oocyst levels by pinpointing a particular Eimeria species that may be expanding at a greater rate.

Another application of deep amplicon sequencing using Eimeria COI is that it provides a means of correlating the relative numbers of all Eimeria species and NE incidence, and thereby it provides insight on predisposing factors (e.g., drug resistance or immunovariability in E. maxima) contributing to NE. In the present study, three farms were documented as having chronic NE problems (increasing mortality [>2/1000] at about 2–3 wk growout), and another three showed no evidence of NE. Our application of COI sequencing did not reveal a correlation between NE status and the relative levels of E. maxima (Table 2), although, similar to past work (28), a correlation between both E. maxima counts (R = 0.46) and E. maxima ITS1 intensity (R = 0.46) and NE was found. This is not unexpected because several authors have found E. maxima infection to be predisposing to Clostridium perfringens–associated NE (40,41). The reason for a lack of association between E. maxima COI amplicon sequencing/mapping and NE is unknown. A factor of significance, though, is the strong correlation between NE and higher relative numbers of E. mitis at all three collection times (0, 2, and 4 wk) as indicated by ITS1 PCR and especially COI deep amplicon sequencing (ITS1, 0.34–0.58; COI, 0.52–0.64). Poultry diagnosticians point out that the excess mucous formation in the lower intestine associated with an E. mitis infection creates an ideal environment for C. perfringens growth and toxin release (D. Bautista, C. Hofacre personal comm.). A characteristic effect of E. mitis is the absence of gross lesions, but sloughing of the intestinal epithelium, leading to increased fluid and mucous accumulation in the lower small intestine, may promote NE in chickens (40,41,42,43,44,45). Without being able to quantify the relative numbers of E. mitis oocysts in a sample, this relationship between E. mitis abundance and NE would have remained anecdotal. Although this finding provides a compelling case for the role of this eimerian in NE, more research is necessary before it can be conclusively stated that E. mitis is a major predisposing factor for NE development.

ITS1 PCR, though showing some agreement with COI deep amplicon sequencing, is much more labor-intensive, and, although highly sensitive, it is difficult to obtain with any certainty a measure of the relative abundance of each Eimeria sp. in a sample.

In conclusion, deep sequencing analysis using COI-based PCR DNA from Eimeria spp. oocysts in litter from commercial broiler houses is an efficient way to determine the relative levels of each Eimeria species present, which may have been missed with oocyst counting or ITS1 PCR. Used together with Eimeria oocyst counts, this sequencing method provides a means of identifying trends in the Eimeria population dynamics during growout and may be helpful in diagnosing reasons for poor performance and increased chick mortality due to coccidiosis. This deep amplicon sequencing approach provided support for anecdotal evidence showing that E. mitis is highly predisposing to NE.

Supplemental data associated with this article can be found at https://doi.org/10.1637/aviandiseases-D-25-00006.s1.

ASV =

amplicon sequence variants;

COI =

cytochrome oxidase I;

ITS1 =

internal transcribed spacer 1;

ITS2 =

internal transcribed spacer 2;

NE =

necrotic enteritis;

OTU =

operational taxonomic units;

SCAR =

sequence-characterized amplified region