High Prevalence of Rickettsia typhi and Bartonella Species in Rats and Fleas, Kisangani, Democratic Republic of the Congo

Abstract

The prevalence and identity of Rickettsia and Bartonella in urban rat and flea populations were evaluated in Kisangani, Democratic Republic of the Congo (DRC) by molecular tools. An overall prevalence of 17% Bartonella species and 13% Rickettsia typhi, the agent of murine typhus, was found in the cosmopolitan rat species, Rattus rattus and Rattus norvegicus that were infested by a majority of Xenopsylla cheopis fleas. Bartonella queenslandensis, Bartonella elizabethae, and three Bartonella genotypes were identified by sequencing in rat specimens, mostly in R. rattus. Rickettsia typhi was detected in 72% of X. cheopis pools, the main vector and reservoir of this zoonotic pathogen. Co-infections were observed in rodents, suggesting a common mammalian host shared by R. typhi and Bartonella spp. Thus, both infections are endemic in DRC and the medical staffs need to be aware knowing the high prevalence of impoverished populations or immunocompromised inhabitants in this area.

Introduction

The last two decades have been characterized by the emergence or re-emergence of various diseases on the world epidemiological scene, among which are a series of endemic zoonotic infections, most of them being rodent or vector-borne.¹ Endemic foci do usually refer to rural settings, but the focal distribution of pathogens may occur in urban areas offering a mosaic of habitats suitable to hosts,² vectors,³ pathogens survival,⁴ and exposed human population.⁵ At different spatial and temporal scales, the anthropogenic modifications directly and indirectly play a role in the host and vector dynamics and act on the probability for (re)emergence of zoonotic diseases. Factors leading to an increase in the incidence of illnesses caused by zoonotic bacteria in urban areas include societal changes ad intrinsic components of the natural history of these organisms that favor their survival in cities.⁶ It is thus crucial to understand interactions between land change, ecology, and dynamics of potential vectors, animal and human hosts constituting those so-called “pathogenic landscapes.”⁷ Kisangani is a major city that consists of a mosaic of landscapes or microhabitats that create an array of ecological niches, which in turn may increase the risk for specific pathogen-carrying hosts and vectors to establish. However, the problem of densely populated, major metropolis of sub-Saharan Africa has so far been overlooked.⁸

Typically, Rickettsia and Bartonella species are emerging zoonotic bacteria that occur in humans, rodents, and their ectoparasites^9–11; and they are increasingly reported worldwide. Rickettsioses are diseases caused by infection with obligate intracellular bacteria (genus Rickettsia) many of which are pathogenic for humans, and all are vector-borne (lice, ticks, fleas, mites). The Rickettsia species described fall into four groups of which the typhus group (TG), comprises Rickettsia typhi and Rickettsia prowazekii, the spotted fever group (SFG) comprises 24 rickettsial species including Rickettsia felis, Rickettsia canadensis group, and Rickettsia bellii group.¹²

Likewise, bartonelloses are diseases caused by infection with the facultative intracellular bacteria. Species of the genus Bartonella are associated with an increasing array of human diseases, among which bacillary angiomatosis and peliosis hepatitis in immunocompromised patients.¹³ Mammalian reservoirs of Bartonella include small carnivores and rodents,¹⁴ their transmission rate depends on the Bartonella strain, host, and specific vector. Historical data on rickettsiosis and bartonellosis in the Democratic Republic of the Congo (DRC) (former Zaire) are fragmentary, but serological tests have reported Rickettsia conorii, Rickettsia prowazekii, and R. typhi as early as the 1950s.^15,16 These data are currently obsolete, and there is a real urgency to estimate the identity and prevalence of Rickettsia and Bartonella species in the DRC. Moreover, recent surveys in DRC revealed that 1) in the Ituri district (north east of the Orientale province), Congolese patients (4.5%) where seropositive for Bartonella henselae, Bartonella quintana, or Bartonella clarridgeiae,¹⁷ but not for Rickettsia¹⁸; that 2) domestic arthropod harbor Rickettsia spp. and Bartonella spp.¹⁹; that 3) rodents harbor Bartonella spp. closely related to Bartonella elizabethae, or Bartonella tribocorum²⁰; and 4) in Kinshasa, that Rickettsia africae and R. felis circulate in arthropod vectors.²¹ A preliminary study was carried out in Kisangani, where the relatively recent stability has changed the profile of the city into a fast-growing economical center in eastern DRC with 1,310,587 inhabitants.²² It seems likely that the increasing numbers of refugees or migrants have augmented the pressure on the ecosystems surrounding Kisangani. Because of this intensifying human encroachment into natural ecosystems, the risk of importing exogenous zoonotic pathogens could raise.^23,24 Moreover, traders and consumers of the central market claimed to regularly consume rodents captured in their own homesteads (Falay A, unpublished data). Finally, because of its fluvial connection with Kinshasa, the potential influx of rats infested with Xenopsylla cheopis fleas, a known vector of murine typhus (R. typhi) is a true concern.²⁵ Our survey pursued three major objectives: 1) to estimate the prevalence and identity of Bartonella spp. and Rickettsia spp. in Rattus spp. populations of Kisangani town; 2) to identify the potential vectors and/or carriers of those zoonotic agents; and 3) to elaborate realistic recommendations to improve the monitoring and control of urban rat populations and their ectoparasites in Kisangani, which could eventually be implemented throughout the DRC and elsewhere.

Material and methods

Sites and habitats.

In collaboration with a team of zoologists from the Faculty of Sciences (Kisangani University, DRC) a rodent and flea survey was conducted in several urban and peri-urban areas of Kisangani town in April and June 2011. Kisangani (0°30′N; 25°11′E) is the principal city of the Orientale province, mainly situated between the Congo River right bank and the Tshopo River. It covers 1910 km² and consists of six administrative districts (Makiso, Tshopo, Kisangani, Mangobo, Kabondo, and Lubunga) and an estimated 80% of the population lives below the poverty line. The economy of the city is primarily based on artisanal mining, “small trade businesses” and agriculture. The sampling sites were selected in relation to the occurrence of rats observed or reported by people and to their potential threat to public health; namely, a slaughterhouse (SH), the central market (CM); the International Aeronautic Transit (IAT) market (IM), and the Tshopo market (TM). The slaughter house (0°32′N, 25°10′E) is located near the left bank of the Tshopo river in Mangobo district, a suburban-rural neighborhood where the main income is based on agriculture. This site is situated in a savannah habitat, scattered with corn or cassava fields. The IAT market (0°30′N; 25°10′E) is located in the Makiso district, at the city's secondary port where merchants from the surrounding villages trade and reside temporarily. The CM (0°30′N; 25°11′E), located in Makiso district, is surrounded by shops, warehouses (rice, wheat flour, cassava or corn flour, fry fish, smoked bushmeat, etc.) and homes. Because of its strategic position, this market provides the majority of the sylvatic products (including bushmeat) sold in town.²⁴ The TM (0°31′N, 25°11′E), consists of open air stalls and warehouses and is situated in the popular Tshopo district (100,000 inhabitants). Hygienic conditions are on the whole poor (lack of proper drainage of sewage, garbage disposal, or septic tanks) for cohabiting families living on and around the market places.

Rodent trapping.

Rodents were trapped with Tomahawk collapsible live traps (dimensions 490/178/173 mm, N = 10, 6 nights/site, 60 trap nights/site) baited with salted and smoked fish. Traps were set at sunset and checked the next morning. The location of each positive trap was recorded using a global positioning system (GPS) (Garmin 60Cx, Southampton, Hampshire, UK). Rodents were taken to the Faculty of Sciences (University of Kisangani), where they were anaesthetized with isofluorane and combed above a white tray to collect ectoparasites. Every rat was identified using morphological characteristics, and were given a unique label and stored in 70% ethanol. For each rat, a series of external measurements was recorded and biopsies were stored in 96% ethanol. Ectoparasite species were identified at the University of Antwerp, pooled per host, flea species, day and location of trapping, and crushed in dry ice before DNA extraction.

Genetic analysis.

Total DNA was extracted from samples (Rattus spp. organs: spleen, heart, liver and left kidney, and flea pools) using the commercial kit QIamp DNA mini kit instructions (Hilden, Germany) according to the manufacturer's recommendations. The DNA was eluted in 100 μL of buffer AE.

Detection of Bartonella in host samples.

Regular polymerase chain reaction (PCR) was performed using the primers 1400F and 2300R²⁶ targeting 852 base pairs (bp) of the Bartonella DNA polymerase beta subunit (rpo B).²⁶ A sample was considered as positive when a ~900 bp fragment was visible. Positive controls consisted of rat DNA found positive for Bartonella sp. and B. elizabethae in previous studies.^20,27

Detection of Rickettsia in host samples and flea pools.

The PCR assay was performed using the primers 120-M59′F and 120-807′R targeting 833 bp of the ompB Rickettsia protein-encoding gene.²⁸ Amplifications were performed in 20 μL reaction volume containing 10 μL of PCR mix (Qiagen multiplex PCR kit no. 206143), 0.5 μL of each primer (10 μM), and 2 μL of total DNA. A sample was considered as positive when a ~850 bp PCR product was visible. Positive controls consisted of flea DNA from domestic fleas' specimen extracted in toto, and collected in locations (households), which had been previously found positive for Rickettsia sp. in a former study.¹⁹

Genetic identification of rodent, Rickettsia, and Bartonella species/strains.

For each rat, DNA was extracted from a mixture of the four organ biopsies to 1) confirm the host species identity using mitochondrial DNA cytochrome b sequencing using L7 and H6 primers²⁹; obtained sequences were compared with known sequences of the Africanrodentia database (http://projects.biodiversity.be/africanrodentia/), 2) detect the presence and estimate the prevalence and strain identity of Bartonella spp. and Rickettsia spp. To identify the Bartonella spp. and Rickettsia spp. strains, positive PCR products were purified and sequenced by VIB Genetic Service Facility (University of Antwerp, Belgium) using the same primers as used for the PCR. The Bartonella and Rickettsia-positive DNA or PCR products from rodents and fleas were sent to URMITE, Marseille, France. All samples were screened for all spotted fever group rickettsiae (SFGR) using real-time qPCR targeting a fragment of gltA gene (RKND03 system),³⁰ and by R. typhi-specific qPCR targeting a fragment of RTB9991CWPP_01310 gene coding for a hypothetical protein.³¹ The DNA R. montanensis and DNA R. typhi were used as positive controls. The concentration of rickettsial DNA is estimated through the number of cycles that is required before the fluorescence signal intensity exceeds the detection threshold (Ct). The Bartonella-positive DNA samples were tested by qPCR targeting an internally transcribed spacer (ITS).³² Sequences were checked and aligned using MEGA 5.05. Sequences were compared with other Bartonella spp. and other Rickettsia spp. using the basic local alignment search tool (BLAST, NCBI) algorithm. During the checking process, clear double peaks (distinct from the baseline noise) were visible on the sequencing chromatograms of four of the flea samples. These may be the result of mixed infections or—caused by the pooling of several flea specimens—to the mixing of several specimens each being infected with a different strain of Rickettsia.

Statistical analysis and models.

To analyze the correlates of host and environmental factors with the probability that an individual host was infected with either Bartonella spp. or Rickettsia spp., we used a generalized linear model with binomial family and a logit link to construct a multiple logistic model in R. The generalized linear model was fit to relate the logit of either Bartonella spp. (2 levels, infected or healthy) or Rickettsia spp. (infected or healthy) occurrence to host species (2 levels: Rattus rattus or Rattus norvegicus), month (2 levels: April or June), sampling sites (4 levels: SH, IM, CM, TM), X. cheopis infection (in a first step, flea number, continuous, and in a second step, 2 levels: infected versus not infected), mite infection (2 levels: infected versus not infected), and weight (continuous). Model selection was based on the akaike information criterion and likelihood-ratio tests and we used a top-down protocol.³² Statistical significance was declared at P values of lower than 0.05. In summary, the generalized linear starting model included the following sets of variables: Bartonella infection or Rickettsia infection, host species, month, locality, X. cheopis infestation, mite infestation, and weight. The final model over Bartonella was finally run on a reduced data set containing only two sites (CM and IM), and the factor species (χ² = 141.5, degrees of freedom [df] = 6, P < 0.0001), whereas the Rickettsia model included the factors species and weight (χ² = 7.8, df = 1, P = 0.0052).

Results

Rat and flea distribution.

A total of 240 traps (60 per site) were placed during the study period (from 12011/04/13 to 2011/12/06) with a daily variable trapping success, ranging from 100% (IM) to 0% in TM and SH site. In total, 126 sexually mature Rattus spp. were trapped; the sample consisted of 106 R. norvegicus and 20 R. rattus. The oriental rat flea X. cheopis (N = 188) was collected off 56 R. norvegicus (N = 164) and 7 R. rattus (N = 24), whereas Xenopsylla brasiliensis (N = 5), the African rat fleas was collected off only two R. norvegicus. Table 1summarizes the number of rodents trapped per site and the number of fleas collected off them.

Table 1.

Site surveyed, trap success, number of Rattus sp. caught, number of X. cheopis fleas, and X. cheopis flea index on Rattus norvegicus and Rattus rattus^*

Site	Trap nights	N Rattus sp.	N XcRn	FI Rn	N XcRr	FI Rr
Slaughterhouse	60	4.8%	2	0.0	4	0.8
Central market	60	37.3%	45	2.3	2	1.5
IAT market	60	50.0%	56	1.0	7	0.7
Tshopo market	60	7.9%	3	1.3	7	1.9
Total	240	126	106	1.5	20	1.2

Trap success: number of positive traps; Number of Rattus sp. caught: N Rattus sp.; Number of X. cheopis fleas: N _XcRn.

^*Flea index: number of fleas per host.

Flea index on Rattus norvegicus: FI _Rn.

Number of X. cheopis fleas: N _XcRr.

Flea index on Rattus rattus: FI _Rr.

IAT = market: the International Aeronautic Transit.

In summary, the predominant host species and flea species were R. norvegicus (84%) and X. cheopis (97.4%). For a similar trapping effort, significantly more Rattus spp. were caught in the IAT market (50%) and CM (37.3%) than in the SH (4.8%) and TM (7.9%) area (χ² = 38.7, df = 3, P < 0.0001). In all four sites, R. norvegicus was significantly more often caught than R. rattus (χ² = 39.3, df = 3, P < 0.0001).

Prevalence of Bartonella and Rickettsia species in Kisangani Rattus species.

In the sampled rats, the overall Bartonella spp. and Rickettsia spp. prevalences were 16.7% and 12.7%, respectively. We trapped fewer R. rattus than R. norvegicus, and the proportion of R. rattus infected with Rickettsia (20%) or Bartonella (25%) was similar to the prevalence of Rickettsia (11.3%) or Bartonella (15%) in R. norvegicus. No Bartonella-positive animals were found in SM and TM sites, whereas Rickettsia-positive rodents and fleas were detected in all four sites. However, the model including overall infection with Bartonella spp. and host species identity—removing the site because of low capture rate- revealed that the probability of being infected with Bartonella spp. was significantly higher for R. rattus (P = 0.5) than R. norvegicus (P = 0.16) (χ² = −1.56, df = 1, P < 0.001). Four rodents (2 R. norvegicus and 2 R. rattus) were coinfected with Bartonella spp. and R. typhi and all came from the IAT market near market along the Congo River. In the Rickettsia model, the only relevant factor, namely weight, was significantly and negatively correlated across species, to the probability of having Rickettsia spp. (P = 0.98 + 0.99 × weight; df = 1, P = 0.02).

Genetic diversity of Bartonella DNA sequences.

All 17 Bartonella-positive rodent samples by standard PCR were confirmed positive by qPCR targeting an internally transcribed spacer with the mean Ct value with SD 30 ± 2.1 (URMITE). Out of 11 readable Bartonella sequences (±800–852 bp) retained (six sequences discarded because of poor quality), four of them were similar (96–100% homology) to the reference strain Bartonella sp. 1-1C (GenBank EU551156.1), a Bartonella rochalimae-like strain described from R. norvegicus in Taiwan. Interestingly, whereas Bartonella sp. was more prevalent in R. rattus, the four B. rochalimae-like sequences from this study were only found in R. norvegicus adult males. Four sequences (±825 bp) were similar with 97–100% homology to an uncultured Bartonella sp. isolate RRB047N (GenBank GU143495) and one sequence showed 100% homology with the reference strain RRB170N (GenBank GU143502), both detected in R. rattus brunneusculus from Nepal. One sequence shared 100% homology with the Bartonella queenslandensis strain (GenBank EU111790) identified in Australian rats belonging to the genera Melomys and Rattus. Bartonella elizabethae was identified in a R. norvegicus (CM), showing 100% homology with the reference strain (GenBank AF165992).³¹

Identification of Rickettsia.

Out of 18 flea pools, 13 (72.2%) were positive for the targeted ompB fragment of Rickettsia spp. Xenopsylla cheopis was the only flea species infected with Rickettsia spp. The positive controls consisted of one Pulex irritans and one Ctenocephalides felis strongylus from Ituri district,¹⁹ and showed 99.8% (787 of 788) homology with Rickettsia sp. R14 (GenBank HM370113) and Rickettsia sp. cf9 (GenBank DQ379483) detected in fleas from India and the United States, respectively, and only 94% (727 of 777) similarity with R. felis (GenBank GQ329879) suggesting that this is a Rickettsia felis-like bacterium. In the study of Sackal and others,¹⁹ it was identified as Rickettsia felis by real-time multiplex PCR assay suggesting that the molecular system used was not specific. The positive samples in this study were identified as R. typhi showing 100% homology with the reference strain R. typhi str. B9991CWPP (GenBank CP003398). Similarly, in the readable sequences obtained from R. norvegicus and R. rattus organs, all Rickettsia identified were showing between 98% and 100% homology with the R. typhi str. B9991CWPP (GenBank CP003398). All Rickettsia-positive samples were confirmed positive by R. typhi-specific qPCR.³¹ In the R. typhi-specific qPCR, the mean Ct value in rodent and flea DNA specimens with standard deviation (± SD) was 31.53 ± 3.60 and 26.03 ± 7.94, respectively. All these DNA samples tested for SFGR-specific qPCR were negative.

Discussion

In this study, we report a high prevalence of Bartonella spp. (17%) with at least five genotypes, and R. typhi (13%) with a single strain, in a small urban rat population in the city of Kisangani. This study also draws attention to the relatively high prevalence of R. typhi DNA in R. rattus (20%), R. norvegicus (11%), and in fleas (72%, 13 of 18 pools) in the country. The prevalence of R. typhi in fleas and rodents DRC is one of the highest from the world. In the last years, R. typhi was identified in 10.8% of X. cheopis pools (4 of 16) in Indonesia,³³ in tissues from three rats and in 10% of fleas from each animal in California,³⁴ in 4% X. cheopis and 6.6% in Leptopsylla segnis in Cyprus,³⁵ and in 3.2% flea pools in Korea.³⁶ The flea remains infected with R. typhi for life with good fitness and the horizontal and vertical transmission increase the infected populations in this endemic area.³⁷ Moreover, the rapid spread of flea-borne pathogens to human populations is caused by the frequent feeding behavior and extraordinary mobility of fleas, and to the abundance of cosmopolitan rat species, that once were infected by R. typhi, remain infective for life, whereas their lifespan and reproductive capacity are unaffected by the infection.³⁷ This high prevalence of R. typhi may have serious public health consequences. Usually, most patients presented mild illness with high fever, headache, chills, and nausea.^31,39 The rash often appears a week after the onset of fever and is discrete; however, it is not always found. In endemic areas, murine typhus was identified in 10% of patients with undifferentiated febrile illnesses⁴⁰; although, sometimes severe complications can occur (neuropsychiatric disorders, renal, hepatic, pulmonary and cardiac dysfunction) with < 5% mortality rate in large series.³⁸

Bartonella elizabethae, which causes endocarditis was previously identified in rat samples in DRC,²⁰ but B. queenslandensis, Bartonella sp. 1-1C and 2 uncultured Bartonella spp. with unknown pathogenicity were identified first in rat samples. No Bartonella spp. were detected in fleas from Kisangani, however B. clarridgeiae, B. vinsonii, B. rochalimae-like, and a new Bartonella genotype were previously¹⁹ reported in fleas collected from rural areas of DRC. The circulation of several Bartonella species may have a high risk in impoverished populations or immunocompromised inhabitants of the neighborhoods near our sample sites. The prevalence of human immunodeficiency virus (HIV) patients in Kisangani in 1999 was 6%, and among sex workers—for the country, between 1985 and 1997—was 25.4–38%.⁴¹ To date, no human case of bartonellosis has been reported for the DRC, or Kisangani in particular, but evidence of the recent exposure to Bartonella was established in patients from the Ituri district in the North of the same province.¹⁷ We found four co-infections with Bartonella spp. and Rickettsia spp. in rat specimens and recently, the < 1% co-infection with Bartonella spp. and R. felis was described in fleas,¹⁹ suggesting a common vector and mammalian host shared by both pathogens.

The incidence and prevalence of human rickettsiosis and bartonellosis are probably underestimated in DRC. There are three major reasons for this situation. First is the lack of awareness for those emerging and neglected vector-borne and rodent-borne infectious diseases. Second, the difficulty to diagnose them because of the high diversity of the clinical manifestations and symptoms, and finally the lack of modern diagnostic tools, often combined with an unreliable electricity supply and the collapse of the local health infrastructure. In the University Clinics of Kisangani for example, the burden of emerging or neglected diseases is unknown and their epidemiological records revealed that 17% of the admitted patients are never given a proper diagnostic (University Clinics of Kisangani; Falay, unpublished). As a consequence, patients do not receive adequate treatment—if any treatment at all—or are treated blindly with a combination of large spectrum antibiotics, which are known to be usually inefficient against Rickettsia or Bartonella infections.

Our results suggest that the urban Rattus populations living in Kisangani only carry cosmopolitan fleas (X. cheopis mainly), and that the flea burden did not affect the probability of infection. A similar situation—a nearly monospecific flea infestation—has been observed in Kinshasa,³⁰ and along the Congo river (Laudisoit A, unpublished data). In addition, we found that rodent weight was negatively correlated with the probability of being infected with R. typhi. The fact that heavier individuals—and probably older individuals—being less likely infected may reflect acquired immunity against Rickettsia as a result of their lifelong and constant exposure to the agent. The rpoB gene used in this study is one of the most potent genes for Bartonella identification.⁴² However, culture techniques and the analysis of other specific genes or genomes should be done to better characterize the Bartonella species identified in DRC.

In conclusion, this study has revealed the urgency 1) to address the issue of vector-borne diseases in major cities in the DRC and cities elsewhere in sub-Saharan Africa; 2) to develop local surveillance and control teams consisting of medical personnel and academicians, and 3) to perform interviews and sociological surveys to identify areas, socio-economic conditions, and times of the year where the transmission risk is the highest to develop adequate rodent and flea control policies.

Disclaimer: The authors declare they have no conflict of interest.