Study sites
We collected fleas in five districts of Uganda, namely Jinja (Eastern Uganda), Kampala (Capital of Uganda), Kasese (Western Uganda), Gulu (Northern Uganda), and Luwero (Central Uganda) (Figure 1) between April 2017 and September 2018. The selected districts are considered major economic hubs in their respective regions and are both geographically and culturally diverse, with high levels of economic heterogeneity. Fleas were collected from livestock (cattle, goats, sheep, pigs, rabbits), companion animals (dogs and cats), rodents, and directly from the environment. In each of the study districts, 18 villages were selected, and from each village four animal owning households were selected for sample collection. In total, 360 households were visited over 18 months. Informed consent was obtained from the head of household for inclusion in the study.
Flea collection and Pooling
Most mammalian fleas do not attach to hosts during feeding and can easily escape when disturbed, so attempts to restrain and immobilize the fleas were made [15]. Pets with phlegmatic temperament and small ruminants were hand restrained. Aggressive pets were sedated with Xylazine 1mg/kg (Xyla, Interchemie, Holland) and Ketamine 6mg/kg (Rotex Medica, Tritau, Germany). Animals were placed on white plastic sheets, and the surface of their fur was sprayed with 70% ethanol to immobilize fleas. After one to four minutes, fleas under the fur emerged to the surface and were brushed onto the white sheet, picked with forceps, and transferred into 70% ethanol for preservation. Rodents were trapped using Sherman traps in houses occupied or frequently visited by people in the selected homesteads. Traps were baited and set in the evening with a mixture of ground peanuts and smoked fish, with captured rodents recovered and processed in the morning as previously reported [18]. Traps were transported in white aerated gunny bags to prevent fleas from escaping. The rodent species were identified by morphological characteristics including the length of their body, tail, ear, hind foot, weight [21]. Fleas were collected by combing the fur with a brush on to white gunny bags and placed in 70% ethanol for preservation. The flea index (# of fleas/host examined) for the animals was determined using established methods [22]. Fleas were also collected from earthen floor houses inhabited or frequented by people in the homesteads. We used trays containing water with grease or Vaseline smeared on the sides to prevent fleas from crawling as previously described [8]. Fleas were identified to the species level using published morphologic taxonomic keys under a stereomicroscope [16, 23-24]. A total of 14,641 individual fleas were collected and then pooled into 714 sample pools according to flea species, host, collection area and date of collection. Pool sizes ranged from 1-281 fleas.
Homogenization of the flea pools
Flea pools were placed in Eppendorf tubes containing RNA later (Sigma Life Science, Darmstadt, Germany) and disrupted using sterile disposable pestles attached to a motorized grinder (HLD-12, Ryobi, China). The fleas were then homogenized by passing them through 20-gauge needles. The homogenates were stored at -80°C until DNA extraction. A subset of 172 pools (2,457 fleas) consisting of 68 rodent flea pools (three pools of E. gallinacea, 65 X. cheopis) and 104 pools (pool size 1-63 fleas) from pets, livestock, and the environment were tested for Y. pestis. The flea pools selected for Y. pestis testing comprised the five flea species collected in the five districts and two seasons (dry and rainy seasons). Of the 172 pools, a subset of 62 pools (353 individual fleas) with pool sizes of 1-28 fleas selected based on seasons (dry and wet) and subsequently tested for Rickettsia spp.
DNA Extraction and PCR
Total DNA was extracted from all the flea homogenates using the Qiagen DNeasy Blood and Tissue kit (Qiagen, Hilden, Germany), according to the manufacturer’s protocol. Duplicate positive and negative controls were included during every batch of DNA extraction. The 62 flea pool DNA samples were screened for flea-borne Rickettsia spp. with primers amplifying the 74-bp citrate synthase (gltA) gene as previously described [25-26]. The primers were CS-F (5-TCGCAAATGTTCACGGTACTTT-3) and CS-R (5-TCGTGCATTTCTTTCCATTGTG-3). A second Rickettsia genus-specific qPCR amplified a 115-bp segment of the 17kDa and ompA genes to confirm the initial PCR results as described previously [27]. R. felis DNA (provided by Walter Reed Army Institute of Research, Silver Spring, MD) was used as a positive control and ultrapure water as a negative control. To detect Y. pestis from flea pools, qPCR targeting the plasminogen activator gene (Pla) was performed using primers Yper_PLA_F (59- ATG-GAG-CTT-ATA-CCG-GAA-AC-39) and Yper_PLA_R (59-GCG-ATA-CTG-GCC-TGC-AAG-39) and probe Yper_PLA _P (6- FAM-TCC-CGA–AAG-GAG-TGC-GGG-TAA-TAGG- TAMRA) [28]. Y. pestis CSUR P100 strain DNA was used as positive control.
Sequencing and Phylogenetic Analysis
A 539 base pair amplicon for 17kDa and a 650 base pair amplicon for ompA gene was amplified as previously described [29] using Platinum Taq (Thermo Fisher Scientific). PCR products were purified using the QIAquick PCR Purification Kit (Qiagen) according to the manufacturer’s instructions. Agarose gel purified amplicons of the 17kDa and ompA genes were sequenced on the SeqStudio (Thermo Fisher Scientific) using the BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific) according to the manufacturer’s recommendations. Forward and reverse reads were aligned using CLC Genomics Workbench (Qiagen), and a consensus sequence for each gene was generated for BLAST analysis. Sequences of 17kDa and ompA genes and references from GenBank were imported and aligned in Geneious Prime 2022.11.0.14.1. The sequences were MAFFT aligned and exported to MEGA 10.2.6 [30]. Maximum likelihood trees were created using the Tamura-Nei model with bootstrap iterations set at 1,000.
Mapping
Descriptive maps with the collection sites, tick species, and pathogen location were created in QGIS 3.28 [31]. The Uganda district shapefile was accessed at https://data.unhcr.org/en/documents/details/83043.
Statistical analysis
The pooled prevalence rates, maximum likelihood estimation (MLE), and minimum infection rates (MIR) were calculated by collection district and flea species to assess the probability of Rickettsia spp. detection from the flea pools as described before [32]. MIR was calculated by dividing the number of infected pools by the total number of fleas in all of the pools and expressed as the number of infected pools per 100 fleas tested assuming that each positive pool only had one positive flea. The maximum-likelihood estimate (MLE) is the infection rate most likely observed given the test results and an assumed probabilistic model; which is a binomial distribution of infected individuals in a positive pool. The CDC’s Mosquito Surveillance Software (https://www.cdc.gov/westnile/resourcepages/mosqSurvSoft.html) was used in Excel to obtain MLE and MIR estimates with their corresponding 95% confidence intervals accounting for individual pool sample size. A Pearson chi-squared test was used to detect any differences between the distributions of outcomes in different groups, with a p-value of <0.05 considered significant. Data were analyzed using STATA software, version 16.1 (StataCorp, College Station, TX).