The skin is a significant but overlooked anatomical reservoir for vector-borne African trypanosomes

1Understanding the process of parasite transmission is essential for the design of rational control measures to break the disease cycle and requires the identification of all reservoirs of infection. In a number of vector-borne diseases, it is becoming evident that asymptomatic individuals, be they humans or animals, can represent a significant proportion of the infected population and therefore an important reservoir of disease that requires targeting by control measures. The recent identification in West Africa of asymptomatic individuals with human trypanosomiasis (long-term seropositives) but undetected parasitaemia, raises the question of what role these individuals play in disease transmission. Therapy is currently only directed towards microscopy-positive individuals and thus a proportion of the infected population remain untreated.

2There is convincing evidence that seropositive individuals with low or undetected parasitaemia contain transmissible trypanosomes. Xenodiagnosis experiments, in which tsetse flies are fed on microscopy-negative infected humans or, more recently, experimentally-infected pigs, have shown that these apparently aparasitaemic hosts contain the parasite since the tsetse flies became infected. It is uncertain where the trypanosomes reside in the host but, given the telmophagus (slash and suck) feeding habit of the tsetse fly, they could be skin-dwelling parasites ingested with the blood meal. Our findings suggest that parasites may be sufficiently abundant in the skin to allow transmission and therefore the skin may represent an anatomical reservoir of infection.

3Detection of trypanosomes in the skin is not well documented, although there are descriptions of cutaneous symptoms associated with African trypanosomiasis and distinct ‘trypanid’ skin lesions. Imaging data from mouse models of infection suggest that trypanosomes sequester to major organs such as the spleen, liver and brain and recent evidence has demonstrated trypanosomes in extravascular adipose tissue. These adipose-associated trypanosomes appear to be a new life-cycle stage with a distinct transcriptional profile and, while tsetse bite-site associated transmission has been suggested, and a historical study made a passing observation of localised deposition of trypanosomes in the skin matrix, the broader role of skin-dwelling trypanosomes in transmission remains unclear. In this paper we report the investigation of a possible anatomical reservoir in the skin of the mammalian host. We provide conclusive evidence of T.b. brucei, (a causative agent of animal trypanosomiasis) and the human-infective trypanosome, T.b. gambiense, invading the extravascular tissue of the skin (including but not restricted to the adipose tissue) and undergoing onward transmission despite undetected vascular parasitaemia. We also provide evidence of localisation of trypanosomes within the skin of undiagnosed humans. The presence of a significant transmissible population of T. brucei in this anatomical compartment is likely to impact future control and elimination strategies for both animal and human trypanosomiases.

4In order to investigate the potential for extravascular skin invasion by T. brucei, BALB/c mice were inoculated via IP injection with the low virulence STIB247 strain of T. brucei and skin sections were assessed over a 36-day time-course. The presence and relative quantities of extravascular parasites were evaluated by semi-quantitative scoring of the histological samples (Figure 1—source data 1) and compared to blood parasitaemia (Figure 1—source data 2). Extravascular parasites were first observed in the skin 12 days post-infection and remained throughout the experiment. Skin parasite numbers fluctuated to a lesser extent than blood parasitaemia and the apparent periodicity in the skin may be due to one particularly high data point on day 24 (Figure 1). Parasites were found in the dermis, subcutaneous adipose tissue (Figure 2) and in fascia beneath the panniculus carnosus muscle. We did not detect any particular clustering around dermal adipocytes. The presence of parasites in the skin was not associated with major inflammation (Supplementary file 1). To confirm that skin invasion by this parasite was not strain or sub-species specific, the more virulent TREU927 strain of T. brucei and the human-infective T.b. gambiense strain, PA, were used to infect mice. Extravascular skin invasion of the dermis, subcutaneous adipose tissue and fascial planes (Figure 2—figure supplement 1) was evident with associated mild to moderate inflammation (Supplementary file 2), in some instances the degree of skin invasion in the T.b. gambiense infected mice was far greater than T.b. brucei, perhaps suggesting a greater propensity for sequestration in this sub-species.

5To confirm that the extravascular distribution of parasites was not an artefact of the route of inoculation, infections by natural vector transmission were carried out using a bioluminescent T.b. brucei strain, AnTat1.1E AMLuc/tdTomato. Mice were infected by a single infective bite of an individual G.m. morsitans. After 4 to 11 days and up to the end of the experiment, parasites were observed in the skin with a dynamic distribution (Figure 3A) and a variable density (Figure 3B). Parasites were first detected in the blood between 5 and 19 days after natural transmission and parasitaemia remained lower than 10 parasites/ml. Observed bioluminescence directly reflects the total number of living parasites in the entire organism, including blood and viscera, but the intensity of the signal decreases with tissue depth. Therefore, at the end of each experiment, mice were sacrificed and their organs were checked for bioluminescence. The presence of extravascular parasites in cutaneous and subcutaneous tissues was first demonstrated by bioluminescence imaging in entire dissected skins (Figure 3C) and was not necessarily localised to the bite site. In addition to the skin, only the spleen, some lymph nodes and adipose tissue were observed to be positive for bioluminescence in several individuals (Figure 3—figure supplement 1). This suggests that the observed bioluminescence is likely to originate predominantly, but not solely, from parasites in the skin. In addition, only mild inflammation was observed after 29 days (Figure 3—figure supplement 2).

6To confirm that trypanosomes in the skin are a viable population, fluorescent parasites were monitored by two intravital imaging methods following IP injection and natural transmission (Figure 4). First, IP-injected fluorescent T. brucei STIB247 were imaged in vivo using 2-photon microscopy (Figure 4A). Extravascular trypanosomes observed in the dermal layer of dorsal skin were highly motile, consistent with viability (Video 1). Second, naturally-transmitted fluorescent T. brucei AnTat1.1E AMLuc/tdTomato were imaged in vivo using spinning-disk confocal microscopy in the C57BL/6J-Flk1-EGFP mouse line that has green fluorescent endothelial cells in the lymphatic and blood vessels (Figure 4B). Extravascular trypanosomes were observed in the dermal layer of the ear and were highly motile (Videos 2, 3 and 4).

7Differentiation of dividing trypanosomes to non-dividing stumpy forms is essential for transmission to the tsetse. Using the relative transcript abundance of the stumpy marker Protein Associated with Differentiation 1 (PAD1) to an endogenous control, Zinc Finger Protein 3 (ZFP3), we estimated that approximately 20% of skin-dwelling parasites were stumpy (Supplementary file 3). To directly determine the proportion of stumpy forms in skin sections, histological staining for PAD1 was also performed (Table 1—source data 4). PAD1-positive cells were observed in variable proportions (from 8 to 80%) in all bioluminescent skin samples examined after IP injection (Supplementary file 4). Following natural transmission, up to 38% of parasites detected using VSG surface markers (Figure 3—figure supplement 3A–B) also expressed PAD1 (Figure 3—figure supplement 3C–D) (n = 441 cells from 8 skin sections). In all skin sections, stumpy parasites were homogenously distributed in the dermis and subcutaneous adipose tissues.

8We next assessed the ability of skin-dwelling parasites to infect tsetse flies. Teneral flies (immature flies that have not yet taken a blood meal) were fed on different regions of skin from mice infected with AnTat1.1E AMLuc/TY1/tdTomato with differing levels of bioluminescence across the skin (Table 1—source data 1 and Table 1—source data 2). This was repeated in 20 mice with differing levels of parasitaemia. Flies were dissected and checked for the presence of fluorescent trypanosomes after two days (Table 1). In mice with undetectable or low parasitaemia (<5 × 104 parasites/ml), no parasites were found in flies fed on non-bioluminescent regions. Conversely, a median of 36% (± 23%, n = 70) of flies that fed on bioluminescent regions of low parasitaemic mice became infected (Table 1 and Table 1—source data 3). This demonstrates that skin parasites, from mice without visible parasitaemia, can contribute to tsetse infection, possibly reflecting human asymptomatic infections. When parasites were detected in the blood and skin, tsetse infection rates increased up to 100% (median of 61% ± 22%, n = 120) (Table 1 and Table 1—source data 3). Parasites taken-up from the skin were able to further develop through the life-cycle to early procyclic forms in the fly as evidenced by GPEET-procyclin marker on the parasites surface (n = 721 cells from 16 flies) (Table 1—source data 4 and Supplementary file 4).

9To demonstrate that skin invasion by trypanosomes can occur in humans as well as the murine model, slides of historical skin biopsies, taken as part of a diagnostic screening programme for Onchocerca microfilaria in the trypanosomiasis-endemic Democratic Republic of the Congo, were examined for trypanosomes. At the time of sampling, the incidence of trypanosomiasis was 1.5–2% and we hypothesised that some individuals may have harboured undiagnosed infections. Re-examination of 1121 skin biopsies by microscopy revealed six individuals with trypanosomes in their skin (Figure 5). These individuals had not previously been diagnosed with human trypanosomiasis by clinical signs or the presence of blood parasites.

10We have shown that there exists a significant yet overlooked population of live, motile, extravascular T. brucei in the dermis and subcutis of animal models infected by artificial routes or by vector transmission. It is likely that once injected the parasites disseminate via the lymph and blood to the skin where they are ingested during tsetse fly pool-feeding and readily initiate cyclical development in the vector. Given the relative volume of the skin organ compared to the vasculature, it is possible, depending on the density of skin invasion, that more parasites exist in the skin than in the blood. The skin, therefore, represents an unappreciated reservoir of infection. The extravasation of trypanosomes was described previously in major organs such as liver, spleen and visceral adipose tissue, but the importance of these parasites in transmission was not investigated. Here we show that these skin-dwelling trypanosomes contribute to transmission and could explain the maintenance of disease foci, despite active screening and treatment. Skin invasion for enhanced transmission is likely a powerful evolutionary force driving extravasation, suggesting that the generalised tissue penetration underlying pathogenesis (i.e. splenomegaly, hepatomegaly, CNS invasion) is a secondary epiphenomenon. A skin reservoir also presents a novel target for diagnostics (e.g. skin biopsies), allowing the prevalence of infection to be accurately determined and the identification of any previously undetected animal reservoirs of human disease.

11The skin as an anatomical reservoir of parasites is a recurring theme in arthropod-borne human diseases such as Leishmania and Onchocerca. Here we present evidence of trypanosomes in the skin of hitherto undiagnosed individuals. This anatomical reservoir may serve to explain how HAT foci re-emerge and persist despite low numbers of reported cases even in the absence of an animal reservoir.

12HAT was once widespread across much of sub-Saharan Africa but concerted control efforts brought it close to elimination during the 1960s. However, disease foci persisted, with a resurgence in the number of reported cases to over 300,000 in the 1990s. Currently, HAT is approaching elimination in many areas. Understanding how HAT evaded elimination in the latter half of the 20th century and how it continues to persist is vital to efforts to eliminate the disease. For example, our results indicate that it may be necessary to develop novel therapeutics capable of accessing extravascular compartments. The current policy in most endemic countries is not to treat serologically positive individuals unless they demonstrate active infection, due to the long duration and high toxicity of treatment and the low predictive value of the serological tests. We suggest that this policy should be reconsidered in light of our compelling evidence that they represent a carrier population which may maintain HAT foci and explain previously thwarted efforts to eliminate this major pathogen.

13All strains used in this study are pleomorphic. STIB247 is a low virulence T.b. brucei strain that induces a chronic infection and was isolated from a hartebeest in the Serengeti in 1971. TREU927 is the genome reference strain for T. brucei and is more virulent than STIB247. This strain was isolated from a tsetse fly in Kenya in 1969/1970. PA is a human-infective group 1 T.b. gambiense strain isolated from a patient in the Democratic Republic of the Congo in 1975. mCherry STIB247 was created by transfection of STIB247 with pHD1034-mCherry. This strain constitutively expresses fluorescent mCherry from the ribosomal RNA promoter and its expression is stable over repeated passage.

14The AnTat 1.1E clone of T.b. brucei was derived from a strain originally isolated from a bushbuck in Uganda in 1966. Bloodstream forms were cultivated in HMI11 medium supplemented with 10% foetal calf serum at 37°C in 5% CO2. This strain was genetically engineered to produce two strains continuously expressing the red-shifted luciferase (PpyRE9H) and the tdTomato red fluorescent protein either individually or in combination.

15For the first AnTat 1.1E strain, the pTbAMLuc plasmid (M. Taylor, London School of Hygiene and Tropical Medicine, UK) was used for continuous cytosolic expression of the red-shifted luciferase PpyRE9H. The tdTomato coding sequence was cloned with HindIII and BamHI into the pTSARib vector, generating the final pTSARib-tdTomato construct. The two plasmids were linearised with KpnI and SphI restriction enzymes, respectively, and used to transform procyclic parasites with an Amaxa Nucleofector (Lonza). After 24 hr, transfected cells were selected by the addition of blasticidin or puromycin (10 μg/ml). After one week, the population was examined (i) for red fluorescence by fluorescence microscopy and (ii) for both red fluorescence and bioluminescence by using a fluorimeter Infinite 200 (Tecan, Switzerland). Cells were sub-cloned by limiting dilution, and clone selection was performed after 15 days by measuring both bioluminescence in a microplate reader Infinite 200 (Tecan) and fluorescence with a Muse cell Analyzer (Merck-Millipore). This strain, named AnTat1.1E AMLuc/tdTomato, was used for natural transmission experiments.

16The second AnTat 1.1E strain expressing the 3.1 Kb chimeric multiplex reporter protein PpyRE9H::TY1::tdTomato was named AnTat 1.1E AMLuc/TY1/tdTomato. This cytoplasmic reporter is composed of three distinct markers: the red-shifted luciferase (PpyRE9H) is fused to the tdTomato red fluorescent protein by a TY1 tag. Briefly, the 1.68 Kb optimised version of the North American firefly Photinus pyralis luciferase was fused with a 10-bp sequence known as TY1-tag and cloned into the pTSARib plasmid by using XhoI and HindIII restriction enzymes to obtain the pTSARib-PpyRE9H-TY1 plasmid. Finally, the 1.4 Kb coding region of the tdTomato fluorescent protein was inserted downstream using HindIII and BamHI. The resulting 8.9 Kb vector, containing a blasticidin S resistance cassette, was linearised with SphI to integrate the rDNA promoter locus. Bloodstream parasites were transformed with an Amaxa Nucleofector (Lonza), sub-cloned by limiting dilution, and clone selection was performed by measuring both bioluminescence in a microplate reader Infinite 200 (Tecan) and fluorescence with a Muse cell Analyzer (Merck-Millipore). The selected AnTat 1.1E AMLuc/TY1/tdTomato sub-clone was comparable to the parental wild-type strain in terms of growth rate (Table 1—source data 1A), pleomorphism (Table 1—source data 4C), tsetse infectivity and virulence in mice (Table 1—source data 2). In order to verify the reliability of the bioluminescent marker as well as to define the bioluminescence detection threshold of the AnTat 1.1E AMLuc/TY1/tdTomato selected sub-clone, a parasite density / bioluminescence intensity analysis was performed in 96-micro-well plates with an IVIS Spectrum imager (Perkin Elmer). Parasite density and bioluminescence intensity were correlated when bioluminescence levels were higher than 104 p/s/cm²/sr, corresponding to about 103 parasites, allowing estimation of the parasite density from in vivo imaging over this threshold (Table 1—source data 1B). This correlation was verified by quantification in a microplate reader Infinite 200 (Tecan) at the very beginning of the first in vivo experiment as well as the end of the last one, demonstrating the stability of the triple reporter expression in the AnTat 1.1E AMLuc/TY1/tdTomato selected sub-clone over time, especially after at least one full in vivo parasite cycle in the tsetse fly and the mammalian host (Table 1—source data 1C). This strain was used for xenodiagnosis experiments and quantification of parasite densities.

17BALB/c and C57BL/6J mice were used as models for chronic disease. In addition, to allow for further 3D intravital imaging of the lymphatic and blood systems, C57BL/6J-Flk1-EGFP mice expressing a GFP tagged Kdr (Flk1) gene encoding the vascular endothelial growth factor receptor 2 (VEGFR-2) were used.

18This study was conducted under Home Office and SAPO regulations in the UK and in strict accordance with the recommendations from the Guide for the Care and Use of Laboratory Animals of the European Union (European Directive 2010/63/UE) and the French Government. The protocol was approved by the 'Comité d’éthique en expérimentation animale de l’Institut Pasteur' CETEA 89 (Permit number: 2012–0043 and 2016–0017) and undertaken in compliance with Institut Pasteur Biosafety Committee (protocol CHSCT 12.131).

19A total of 36 eight-week old BALB/c mice (Harlan, UK) were inoculated by intra-peritoneal (IP) injection with 104 parasites of strain STIB 247. Parasitaemia was assayed on each subsequent day using phase microscopy. Twenty-four uninfected animals served as controls. Every three days for 36 days, three infected animals and two uninfected animals were culled and 2 cm2 skin samples removed from the dorsum. Skin samples were fixed in 10% neutral buffered formalin prior to histological analysis.

20Tsetse flies (Glossina morsitans morsitans) were maintained, infected and dissected at the Institut Pasteur as described previously. Flies were infected with AnTat1.1E AMLuc/tdTomato parasites. Positive flies were selected first by screening the abdominal fluorescence (midgut infection) 15 days after the infective meal, and then by a salivation test (mature salivary gland infection) after one month. Single flies with salivary gland infections were used to infect the abdomen of mice anaesthetised by IP injection of ketamine (Imalgene 1000 at 125 mg/kg) and xylazine (Rompun 2% at 12.5 mg/kg) and feeding was confirmed by visual observation of the fly abdomen full of blood. Control mice were either not bitten or bitten by uninfected flies. The presence and density of parasites in the blood was determined daily by automated fluorescent cell counting with a Muse cytometer (Merck-Milllipore, detection limit 5.102 parasites/ml) or by direct examination under a phase microscope with standardised one-use haemocytometers (Hycor Kova, detection limit 104 parasites/ml), according to the manufacturer’s recommendations.

21A total of 20 seven-week-old male C57BL/6J Rj mice (Janvier, France) were IP injected with 10 AnTat 1.1E AMLuc/TY1/tdTomato bloodstream forms. Parasitaemia was assayed daily by automated fluorescent cell counting with a Muse cytometer (Merck-Millipore, detection limit 5.102 parasites/ml) according to the manufacturer’s recommendations.

22Three BALB/c were infected with T.b. brucei strain TREU927 and culled at day 11. The mice were perfused and a 2 cm2 region of skin removed from the flank. Skin sections were lysed using a Qiagen Tissuelyzer LT and RNA extracted using a Qiagen RNAeasy kit following the manufacturer’s instructions. 100 ng of RNA from each sample was reverse-transcribed using an Invitrogen Superscript III RT kit. qPCR was performed on each sample using 5 µl of cDNA using a protocol and primers validated previously on an Agilent Technologies Stratagene Mx3005P qPCR machine. The ratio of PAD1 to ZFP3, and hence the proportion of cells transcribing the PAD1 gene, was estimated using the Agilent Technologies MXPro software.

23Mice were first anaesthetised by IP injection of ketamine (Imalgene 1000 at 125 mg/kg) and xylazine (Rompun 2% at 12.5 mg/kg). Batches of 10 teneral male tsetse flies (from 8 to 24 hr post-eclosion) were then placed in 50 ml Falcon tubes closed with a piece of net through which they were allowed to feed directly on mouse skin regions of interest for 10 min. The selection of the skin regions for fly feeding was based on mice bioluminescence profiles and parasitaemia. Unfed flies were discarded and fed flies were maintained as previously described. Anaesthetised mice were finally sacrificed by cervical dislocation and their skin was dissected for controlling bioluminescence with an IVIS Spectrum imager (Perkin Elmer). All the flies were dissected and checked for the presence of trypanosomes either 2 or 14 days after their meal on mouse skin by two entomologists blinded to group assignment and experimental procedures. Dissections were performed as previously described, entire midguts were scrutinised by fluorescence microscopy to detect and count living red fluorescent parasites, and positive midguts were further treated for IFA. A total of 420 flies were used in 3 independent xenodiagnosis experiments.

24To perform the parasite density / bioluminescence intensity assay with AnTat 1.1E AMLuc/TY1/tdTomato bloodstream forms, parasites were counted, centrifuged and resuspended in fresh HMI11 medium at 10.106 cells/ml. Then, 100 μl (or 106 parasites) of this suspension were transferred into black clear-bottom 96-well plates and serial 2-fold dilutions were performed in triplicate adjusting the final volume to 200 μl of HMI11 with 300 μg/ml of beetle luciferin (Promega, France). Luciferase activity was quantified after 10 min of incubation with a microplate reader Infinite 200 (Tecan), following the instructions of the Promega Luciferase Assay System. After background removal, results were analysed as mean ± SD of three independent experiments.

25Infection with bioluminescent parasites was monitored daily by detecting the bioluminescence in whole animals with the IVIS Spectrum imager (Perkin Elmer). The equipment consists of a cooled charge-coupled camera mounted on a light-tight chamber with a nose cone delivery device to keep the mice anaesthetised during image acquisition with 1.5% isofluorane. D-luciferin potassium salt (Promega) stock solution was prepared in phosphate buffered saline (PBS) at 33.33 mg/ml, filter-sterilised and stored in a −20°C freezer. To produce bioluminescence, mice were inoculated IP with 150 μl of D-luciferin stock solution (250 mg/kg). After 10 min of incubation to allow substrate dissemination, all mice were anaesthetised in an oxygen-rich induction chamber with 2% isofluorane, and images were acquired by using automatic exposure (30 s to 5 min) depending on signal intensity. Images were analysed with Living Image software version 4.3.1 (Perkin Elmer). Data were expressed in average radiance (p/s/cm²/sr) corresponding to the total flux of bioluminescent signal according to the selected area (total body of the mouse here). The background noise was removed by subtracting the bioluminescent signal of the control mouse from the infected ones for each acquisition.

26Intravital multi-photon microscopy studies were carried out using a Zeiss LSM7 MP system equipped with a tuneable titanium:sapphire solid-state two-photon excitation source (4W, Chameleon Ultra II, Coherent Laser Group) coupled to an Optical Parametric Oscillator (Chameleon Compact OPO; Coherent). Movies were acquired for 10 to 15 min with an X:Y pixel resolution of 512 × 512 in 2 μm Z increments producing up to 40 μm stacks. 3D tracking was performed using Volocity 6.1.1 (Perkin Elmer, Cambridge, UK). Values representing the mean velocity, displacement and meandering index were calculated for each object. Mice were anaesthetised IP using medetomidine (Domitor 0.5 mg/kg) and ketamine (50 mg/kg) and placed on a heated stage. Following removal of hair with a depilatory cream, dorsal skin was imaged. An intravenous injection of non-targeted quantum dots (Qdot705) (Life Technologies, UK) prior to imaging allowed visualisation of blood vessels.

27AnTat1.1E AMLuc/tdTomato parasites were monitored in the ear of Kdr (Flk1) C57BL/6J Rj mice by spinning-disk confocal microscopy as described previously. Briefly, mice were first anaesthetised by IP injection of ketamine (Imalgene 1000 at 125 mg/kg) and xylazine (Rompun 2% at 12.5 mg/kg). Mice were wrapped in a heating blanket and placed on an aluminium platform with a central round opening of 21 mm in diameter. A coverslip was taped on the central hole and the mouse was positioned so that the ear was lying on this oiled coverslip. Imaging was performed using an UltraView ERS spinning-disk confocal system (Perkin Elmer) with a x40 oil objective (1.3 numerical aperture). Movies were acquired by an EM-CCD camera (Hamamatsu) controlled by the Volocity software (Perkin Elmer) with an exposure time of 500 ms for a total of 30 to 120 s. Images were analysed using ImageJ 1.48v and its plugin Bio-formats importer (NIH).

28Paraformaldehyde-fixed skin samples were trimmed and processed into paraffin blocks. Sections were stained with Haematoxylin and Eosin (HE). Additional serial sections were processed for immunohistochemical staining using a polyclonal rabbit antibody raised against the invariant surface glycoprotein 65 (IGS65) (M. Carrington, Cambridge, UK) using a Dako Autostainer Link 48 (Dako, Denmark) and were subsequently counterstained with Gill’s Haematoxylin.

29The extent of cutaneous inflammatory cell infiltration was assessed in haematoxylin and eosin stained sections with a semi-quantitative scoring system applied by two pathologists blinded to group assignment and experimental procedures. The extent of mixed inflammatory cell infiltration in the dermis and/or subcutis was assessed on a 0 to 3 grading scale (0 = no inflammation or only few scattered leukocytes; 1 = low numbers of inflammatory cells; 2 = moderate numbers of inflammatory cells; 3 = large numbers of inflammatory cells). Ten high-power fields were scored for each skin sample. An inflammation score calculated as the average of the scores in the 10 high-power fields was determined for each sample.

30Parasite burden was assessed in skin sections stained with anti-IGS65 antibody by two pathologists blinded to group assignment and experimental procedures. Presence of parasites defined as intravascular (parasites within the lumen of dermal or subcutaneous small to medium-sized vessels) and extravascular (parasites located outside blood vessels, scattered in the connective tissue of the dermis or in the subcutis) was evaluated in 5 randomly selected high-power fields at x40 magnification with a 0 to 3 semi-quantitative grading scale (0 = no parasites detectable; 1 = low numbers of parasites (<20); 2 = moderate numbers of parasites (20 < 50); 3 = large numbers of parasites (>50). An average parasite burden score was calculated for each sample.

31Cells were treated for immunofluorescence after paraformaldehyde or methanol fixation as described previously. Parasites were stained with one or two of the following antibodies: (i) the anti-CRD polyclonal rabbit antibody (1:300) to label the cross-reactive determinant of the glycosylphosphatidylinositol anchors of proteins, predominantly the variant surface glycoproteins, (ii) the anti-PAD1 polyclonal rabbit antibody (1:100) targeting the carboxylate-transporter Proteins Associated with Differentiation 1 (PAD1) (Keith Matthews, Edinburgh, UK), (iii) the anti-GPEET mouse IgG3 monoclonal antibody (1:500) targeting the T. brucei GPEET-rich procyclin (Acris Antibodies GmbH, San Diego, USA), (iv) the L8C4 mouse IgG1 monoclonal antibody labelling an epitope of the PFR2 protein. Specific antibodies with minimal cross-reactions with mice and coupled to AlexaFluor 488, Cy3 or Cy5 (Jackson ImmunoResearch, USA) were used as secondary antibodies. DNA was stained with 4,6-diamidino-2-phenylindole (DAPI). IFA image acquisition was carried out on a Leica 4000B microscope with a x100 objective lens using a Hamamatsu ORCA-03G camera controlled by Micro-manager and images were normalised and analysed with ImageJ 1.49v (NIH).

32Historical human skin samples were collected from 1991 to 1995 as part of The National Onchocerciasis Task Force (NOTF). Of this collection, 1121 paraffin embedded skin samples were cut into 2.5 micron sections and stained with Giemsa (Sigma-Aldrich). Slides were screened for the presence of parasites by two pathologists independently and representative images taken at x100 magnification.