Exploring the Pathogen Profiles of Ancient Feces

Abstract

Analysis of ancient desiccated feces – termed paleofeces or coprolites – can unlock insights into the lives of ancient people. We collected desiccated feces from caves in the Rio Zape Valley in Mexico (725-920 CE). First, we extracted DNA with methods previously optimized for paleofeces. Then, we applied highly sensitive modern molecular tools (i.e., PCR pre-amplification followed by multi-parallel qPCR) to assess the presence of 30 enteric pathogens. We detected ≥1 pathogen associated gene in each of the ten samples and a mean of 3.9 pathogens per sample. The targets detected included Blastocystis spp. (n=7), atypical enteropathogenic E. coli (n=7), Enterobius vermicularis (n=6), Entamoeba spp. (n=5), enterotoxigenic E. coli (n=5), Shigella spp./enteroinvasive E. coli (n=3), Giardia spp. (n=2), and E. coli O157:H7 (n=1). The protozoan pathogens we detected (i.e., Giardia spp. and Entamoeba spp.) have been previously detected in paleofeces via enzyme-linked immunoassay (ELISA), but have not via PCR. This work represents the first detection of Blastocystis spp. atypical enteropathogenic E. coli, enterotoxigenic E. coli, Shigella spp./enteroinvasive E. coli, and E. coli O157:H7 in paleofeces. These results suggest that sensitive modern molecular tools, such as PCR, can be used to evaluate ancient materials for genes of interest.

Introduction

Ancient peoples left behind a wealth of evidence, but direct biological evidence is scarce and often highly decayed. Sensitive modern molecular methods offer the opportunity to advance our knowledge of these ancient civilizations. One piece of the puzzle is desiccated human fecal material, which has been recovered at cave sites across the world [1–5]. Much can be learned from these materials; ancient stool samples, termed paleofeces or coprolites, can offer insights into dietary practices, human migration, and pathogen exposures.

The ova of soil transmitted helminths (i.e., intestinal worms) have been a focus of ancient feces research for decades because they are highly persistent in the environment and are large enough to view via microscopy [6–10]. However, it is accepted in modern clinical practice that microscopy for helminth ova requires highly trained microscopists and there is potential for misclassification. The visual identification of ova in ancient feces – including from Ascaris and hookworm – has therefore generated much debate regarding the veracity of the claims [11–17].

Modern molecular methods – including metagenomics and polymerase chain reaction (PCR) – are highly sensitive and specific [18,19]. However, nucleic acid fragments, including microbial DNA in ancient feces are highly degraded and usually present at low concentrations [20]. Given the poor analytical sensitivity associated with metagenomic methods, analysis is limited to communities and generally not specific pathogens. PCR is highly sensitive and specific, but specialized methods are required to maximize recovery and prevent contamination in laboratory handling [20]. Refinement of existing methods, combined with state-of-the-science PCR platforms, offers the opportunity for sensitive detection of enteric pathogens in ancient feces. Such knowledge could expand our understanding of how ancient peoples lived. Our research objective was to: (1) apply modern stool based molecular diagnostics to ancient fecal samples; and (2) assess the enteric pathogen profile of these samples.

Methods

Study site and population

Previously collected ancient fecal material used in this study was from the Cave of the Dead Children (La Cueva de los Muertos Chiquitos) in Rio Zape Valley of Mexico (n=10). The cave system along the Rio Zape had been sealed until its excavation in the 1960s and exists in a low-humidity area, which slowed the degradation of the cave’s contents via desiccation. Analysis of the materials recovered suggest that the stool samples analyzed in this study are from 600-800 CE. While palaeofaeces are not subject to the Native American Graves Protection and Repatriation Act (NAGPRA) or other regulations, indigenous populations with strong cultural ties to the specimens were consulted as part of previous work with these samples [21].

Nucleic Acid Extraction

Several precautions were taken to prevent contamination of the samples. Extractions were performed in Class IIA biological safety cabinet that was disinfected with 10% bleach, 70% ethanol, and UV before and after use. All materials were either single use items that had been purchased sterile and only opened inside the biological safety cabinet, or had been autoclaved, washed, autoclaved again, and flame sterilized before use. In addition, we purchased new bottles of reagents for this work.

We adapted nucleic acid extraction “Method B” as described in Hagan et al. 2020 [20], which optimized extraction methods for the recovery of DNA from paleofeces. The samples from were high in fiber [22] which made them rigid and required an initial grinding step that was not previously described. After carefully breaking off small quantities of paleofeces inside a sterile whirl-pak bag (Nasco, Madison, Wisconsin), we transferred approximately 25-50 mg of paleofeces to a sterile 50mL tissue grinding tube (VWR, Radnor, Pennsylvania). The fecal material was ground into a powder by rotating the handle of the tissue grinding tube and then contents were carefully poured into a 2 mL PowerBead tube (Qiagen, Hilden, Germany) containing garnet beads. Additional material was ground and transferred until 200 mg of fecal material was achieved. We extracted 1-3 replicates of each sample depending on the quantity of fecal material available.

After loading bead tubes with paleofeces, we followed the methods described in Hagan et al. 2020 [20]. First, we added 400μL of 0.5 M EDTA (VWR, Radnor, Pennsylvania), 100μL of Proteinase K (ThermoFisher, Waltham, Massachusetts), and 750μL of PowerBead Solution (Qiagen, Hilden, Germany) to each sample in the PowerBead tube. Then, the bead beating tube was placed inside a 50mL tube and gently rotated on a tube roller four hours at room temperature. Next, we vortexed samples for 10 minutes using a Vortex-Genie mixer (Scientific Industries, Bohemia, New York) at maximum speed and then centrifuged samples at 11,000 x g for five minutes.

Next, we carefully but firmly pushed the bottom of a Zymo-Spin V reservoir (Zymo, Irvine, California) into a MinElute column (Qiagen, Hilden, Germany) and placed device into a sterile 50 mL centrifuge tube. We added 14 mL of Qiagen PB buffer and the entire supernatant from the bead beating tube into the chamber of the Zymo-Spin V reservoir. Then the tube was centrifuged for 4 minutes at 2000 x g, rotated 90°, and centrifuged at 2000 x g for an additional 2 minutes. Afterwards, we removed the MinElute column from the reservoir, inserted the column into a 2mL micro collection tube and centrifuged at 11,000 x g for two minutes to dry the column. Next, we pipetted 700 μL of Qiagen PE buffer (Qiagen, Hilden, Germany) into the MinElute column, centrifuged for 2 minutes at 11,000 x g, discarded the flow-through and then repeated this step with fresh PE buffer. Finally, we added 30 μL of Qiagen EB buffer (Qiagen, Hilden, Germany) to the MinElute column, incubated at room temperature for 5 minutes, centrifuged at 11,000 x g, retained the flow-through, and repeated this step. This process resulted in 60 μL of template, which was stored at 4°C for less than 24 hours and then stored at −80°C.

On each day of extractions, we included one negative extraction control to monitor for contamination. We did not spike in extraction control material to the samples to limit the potential for contamination.

Pre-amplification

We used the TaqMan PreAmp Master Mix Kit to pre-amplify our target sequences and lower our limit of detection (ThermoFisher, Waltham, Massachusetts). First, we pooled the 38 forward and 38 reverse primers (IDT, Coralville, Iowa) such that concentration of each primer was 0.18 μM, which equates to the final concentration required by the kit. Then we prepared 20 μL pre-amplification PCR reactions for each sample according to the manufacturer instructions, which included TaqMan PreAmp Mastermix (10μL), the diluted primer pool (5 μL), DNA template (5 μL). The resulting reaction was run on a Bio-Rad CFX 96 Touch thermocycler (Bio-Rad, Hercules, CA) with an initial activation at 95°C for 10 minutes, followed by 14 cycles of 95°C for 15 seconds followed by 60°C for 4 minutes.

Real-Time PCR

We developed a custom TaqMan Array Card (TAC) according to Liu et al. 2013 [23] and 2016 [24] (Table S1, Table S2). The pathogens assessed included helminths (Ancylostoma duodenale, Ascaris lumbricoides, Enterobius vermicularis, Hymenolepis nana, Necator americanus, Strongyloides stercolaris, and Trichuris trichiura), protozoa (Acanthamoeba spp., Balantidium coli, Blastocystis spp., Cystoisospora belli, Cyclospora cayetanensi, Cryptosporidium spp., Enterocytozoon bieneusi, Encephalitozoon intestinalis, Entamoeba histolytica, Entamoeba spp., Giardia spp.), and bacteria (Campylobacter jejuni/coli, Clostridium difficile, E. coli O157:H7, enteroaggregative E. coli, enteropathogenic E. coli, enterotoxigenic E. coli, Helicobacter pylori, Shigella/enteroinvasive E. coli, Plesiomonas shigelloides, Salmonella spp., shiga-toxin producing E. coli, Yersinia enterocolitica). In addition, the card included assays for enteric 16S rRNA and phocine herpes virus (PhHPV) [24]. The enteric 16S rRNA assay – described in Rousselon et al. 2004 [25] – was designed to detect a cluster of phylotypes, called Fec1, corresponding to 5% of the human fecal microflora.

The TAC was prepared by combining 6.67 μL of pre-amplification product [26], with 31.3 μL of molecular grade water, 2 μL of a synthetic DNA sequence matching the PhHPV assay to monitor inhibition (10⁶ copies per μL), and 60 μL of AgPath-ID^™ One-Step RT-PCR Reagents (Applied Biosystems, Waltham, MA). We used the standard TAC cycling conditions with a 1 °C/s ramp rate between all steps: 45 °C for 20 minutes, 95 °C for 10 minutes, then 45 cycles of 95 °C for 15 seconds and 60 °C for 1 minute [23,24]. The TAC performance was evaluated using an 8-fold dilution series (10⁹-10² gene copies per reaction) of an engineered combined positive control that was developed using the methods from Kodani and Winchell 2012 [27]. The linearity and efficiency the targets were within normative standards (linearity: 0.97-1.0, efficiency: 87%-102%) (Table S2). Each day of TAC analysis, one PCR positive control and one negative extraction control was analyzed. Quantification cycle (Cq) values were determined by manual thresholding and included comparison of each assay’s fluorescent signal against the daily negative and positive controls (Figure S1). Any target that amplified past a Cq of 40 was categorized as negative to reduce the potential for false positives[24]. The theoretical limit of detection – a result of the dilutions used – was 60 gene copies per gram solids (Table S3).

Digital PCR

We performed digital PCR with a QIAcuity 4 instrument (Qiagen, Hilden, Germany) to assess the presence of human mitochondrial (mtDNA) DNA in the paleofeces samples to assess if fecal material was of human origin (Table S4) [28]. Human mtDNA is present in human feces because intestinal epithelial cells and leucocytes are shed from the intestinal lining into fecal matter. This assay has demonstrated high sensitivity (100%) and specificity (97%) to fresh human feces. We prepared reactions with QIAcuity Probe Mastermix, 200 nM forward and reverse primers, 800 nM probe, and 2 μL of raw template (i.e., no pre-amplification was performed). Thermocycling conditions were 95°C for two minutes, followed by 45 cycles of 95°C for 15 seconds and 55°C for 60 seconds. We included one positive and one negative control on each dPCR nanoplate. We differentiated positive and negative partitions by manual thresholding between the bands of the positive and negative controls. Samples with less than three positive partitions were classified as negative.

Results

Controls

Assays for the four PCR positive controls exhibited positive amplification as expected (Cq ~ 18). There was no positive amplification for any target in the four negative extraction controls, except for the 16S assay. Microbial DNA contamination of Taq polymerase in PCR mastermix has been documented [29,30] including for the 16S rRNA gene [31,32]. According to the quality control of the manufacturer, the 20 μL pre-amplification reactions contained ≤2 copies of the 16S rRNA gene[32]. The pooled primers in the reaction included primers for the 16S assay, which if present, may have amplified this contamination. The spiked inhibition control – which was a synthetic DNA sequence – amplified consistently (Cq ~20) for all samples.

Molecular Results

We detected ≥1 enteric pathogen in each of the ten paleofeces from Mexico and a mean of 3.9 pathogens per sample out of the 30 pathogens assessed (Table 1). The targets detected in the ten samples included Blastocystis spp. (n=7), atypical enteropathogenic E. coli (n=7), Enterobius vermicularis (n=6), Entamoeba spp. (n=5), enterotoxigenic E. coli (n=5), Shigella spp./enteroinvasive E. coli (n=3), Giardia spp. (n=2), and E. coli O157:H7 (n=1). One sample was positive for human mtDNA via dPCR at a high concentration (approximately 10⁵ gene copies/gram paleofeces) and was only positive for Blastocystis spp. While, human mtDNA was not detected in the remaining nine samples, all 10 samples were positive for the enteric 16S rRNA target.

Table 1.

Prevalence of molecular targets

Target	Prevalence	N (out of 10)
enteric 16S rRNA†	100%	10
Blastocystis spp	70%	7
Enteropathogenic E. coli (atypical)	70%	7
Enterobius vermicularis (pinworm)	60%	6
Entamoeba spp.	50%	5
Enterotoxigenic E. coli	50%	5
Enteropathogenic E. coli (typical)	30%	3
Shigella/EIEC	30%	3
Giardia spp.	20%	2
E. coli O157:H7	10%	1
human mtDNA	10%	1*

Note: The following pathogens were not detected: Acanthamoeba spp., Ancylostoma duodenale Ascaris lumbricoides, Blantidium coli, Cystoisospora belli, Cyclospora cayetanensi, Campylobacter jejuni & coli, Clostridioides difficile B, Cryptosporidium spp., Enterocytozoon bieneusi, Encephalitozoon intestinalis, Entamoeba histolytica, Hymenolepis nana, Helicobacter pylori, Necator americanus, Plesiomonas shigelloides, Salmonella spp., Strongyloides stercolaris, Trichuris trichiura, Yersinia enterocolitica, Enteroaggregative E. coli, Shiga-toxin producing E. coli. EIEC = Enteroinvasive E. coli

^*The positive sample exhibited a strong positive signal

^†This assay was designed for human feces, but may cross react with some animal feces[25]

We analyzed replicates from nine of the ten samples. This included six samples in duplicate and three samples in triplicate, which was limited by the mass of material available. Among these nine samples, there were 40 instances where a pathogen associated gene was detected in a sample and could be compared against detection in the other replicate samples. We observed perfect concordance (i.e., both duplicates or all three triplicates positive) among 60% (n=24/40) of replicates, moderate concordance (two out of three triplicates positive) among 5.0% (n=2/40), and poor concordance (one detection out of two or three replicates) among 35% (n=14/40).

Discussion

We detected diverse enteric pathogens in 1,100-1,300 year-old paleofeces from Mexico. The 60% prevalence of pinworm (i.e., Enterobius vermicularis) we observed is greater than the 34% (n=34/100) prevalence determined via microscopy on paleofeces from the same cave in Mexico. PCR has greater sensitivity than microscopy, but our small sample size suggests cautious interpretation. Protozoan pathogens we detected, including Giardia spp. [33] and Entamoeba spp. [34] have been previously detected in paleofeces via enzyme-linked immunoassay (ELISA), but have not via PCR. In addition, this work represents the first detection of Blastocystis spp., atypical enteropathogenic E. coli, enterotoxigenic E. coli, Shigella spp./enteroinvasive E. coli, and E. coli O157:H7 in paleofeces.

Several of the non-detect results contrast with previous findings. However, a non-detect does not indicate that the target was not initially present in the sample. It does indicate that we were unable to detect it with the methods used. For example, 61% (55/90) of paleofeces from the same cave in Mexico exhibited a strong positive signal for Cryptosporidium parvum via ELISA [35]. There are several possible explanations for this discrepancy. It is possible that the Cryptosporidium antigens measured had greater environmental persistence than the nucleic acids inside the oocysts. Second, our analysis was limited to ten samples. Analysis of other paleofeces samples from this cave may have detected Cryptosporidium. Likewise, there is debate whether helminths other than pinworm were circulating among people in the Americas before the Columbian exchange began in 1492 [11]. The molecular detection of pinworm supports the argument that finding of hookworm and Ascaris ova in paleofeces from the Americas pre-1492—which may have been degraded after hundreds of years—were simply misidentified pinworm ova. However, definitive conclusions should not be drawn from a small sample size of paleofeces from a single cave.

Cautious interpretation is also warranted for the infrequent detection of human mtDNA. These paleofeces were previously identified as human based on morphology and size [1,35], but such analysis is subjective. The detection of human mtDNA at a high concentration from one sample – and the detection of the enteric 16S rRNA assay in all samples – provides confidence that at least some of these paleofeces are from humans. Human mtDNA is often found at lower concentrations in human feces than genes from other microbial organisms [28,36] and its persistence in the environment, relative to the pathogenic genes we detected, is unclear. If some or all these feces are from humans, then this study suggests poor sanitation among the Loma San Gabriel culture from 600-800 CE resulted in exposures to fecal wastes in the environment. Human and animal feces may contain enteric pathogens, which are transmitted via drinking water, soils, food, flies, and fomites [37,38]. Most of the pathogens we detected are zoonotic, meaning that they can be shed by animals as well as humans. However, Shigella spp. is considered specific to humans[39].

We used PCR to detect the presence or absence of genes that are found in specific microorganisms [19]. PCR requires prior knowledge of the target sequence, which limited our analysis to specific pathogen associated genes. Our pre-amplification and PCR methods were also limited to the specific target of interest and were unable to provide information about other DNA sequences that may have been present. Metagenomics is an alternative method that involves sequencing all the genetic material in sample [18]. It provides the genetic code of the entire microbial community, including reads for known and unknown organisms. Metagenomic methods have been used to characterize microbial communities and reconstruct ancient microbial genomes from paleofeces[21,40,41]. Sequencing methods, however, have limitations in identifying microbial DNA at the genus or species level as we have done here. Metagenomics has a higher limit of detection than PCR, and sequencing may miss low abundance genomes. If genomes are similar, then bioinformatics pipelines may be unable to resolve species or genus-level details. Accurate identification also relies on the quality of reference databases. If these databases include misidentified genomes, or if specific genomes are not well represented in the database, then it may be challenging to accurately identify the aligned reads.

There are several limitations associated with this work. First, we analyzed a small number of samples from a single cave system. Analysis of additional samples may have detected other pathogen-associated genes. Second, the persistence of the gene targets in paleofeces is not well characterized. Evidently some pathogen associated genes persisted in the samples for 1,100-1,300 years, but it is unclear if other genes may have decayed beyond our ability to detect them. Finally, PCR inhibitors may have unique impacts on different PCR assays, and assay specific inhibition may have occurred given the large targets assays we used[42]. While our extraction methods had been optimized for paleofeces, we prioritized recovery of small DNA fragments by, in part, omitting inhibitor removal steps that would typically be used in fresh stool extractions [20].

We detected genes associated with nine enteric pathogens, many of which have never been detected before in paleofeces. These results indicate that modern molecular techniques are an effective tool to screen paleofeces – and potentially other ancient samples – for multiple gene-based targets of interest. The application of these methods to other ancient samples offers the potential to expand our understanding how ancient peoples lived and the pathogens that may have impacted their health.