P. falciparum gametocyte density and infectivity in peripheral blood and skin tissue of naturally infected parasite carriers

Transmission of Plasmodium falciparum depends on the presence of mature gametocytes that can be ingested by mosquitoes taking a bloodmeal when feeding on human skin. It has long been hypothesised that skin sequestration contributes to efficient transmission. Although skin sequestration would have major implications for our understanding of transmission biology and the suitability of mosquito feeding methodologies to measure the human infectious reservoir, it has never been formally tested. In two populations of naturally infected gametocyte carriers from Burkina Faso, we assessed transmission potential to mosquitoes and directly quantified male and female gametocytes and asexual parasites in: i) finger prick blood, ii) venous blood, iii) skin biopsies, and in pools of mosquitoes that fed iv) on venous blood or, v) directly on the skin. Whilst more mosquitoes became infected when feeding directly on the skin compared to venous blood, concentrations of gametocytes in the subdermal skin vasculature were identical to that in other blood compartments. Asexual parasite densities, gametocyte densities and sex ratios were identical in the mosquito blood meals taken directly from the skin of parasite carriers and their venous blood. We also observed sparse gametocytes in skin biopsies from legs and arms of gametocyte carriers by microscopy. Taken together, we provide conclusive evidence for the absence of significant skin sequestration of P. falciparum gametocytes. Gametocyte densities in peripheral blood are thus informative for predicting onward transmission potential to mosquitoes. Quantifying this human malaria transmission potential is of pivotal importance for the deployment and monitoring of malaria elimination initiatives. IMPORTANCE Our observations settle a long-standing question in the malaria field and close a major knowledge gap in the parasite cycle. By deploying mosquito feeding experiments and stage-specific molecular and immunofluorescence parasite detection methodologies in two populations of naturally infected parasite carriers, we conclusively reject the hypothesis of gametocyte skin sequestration. Our findings provide novel insights in parasite stage composition in human blood compartments, mosquito bloodmeals and their implications for transmission potential. We demonstrate that gametocyte levels in venous or finger prick blood can be used to predict onward transmission potential to mosquitoes. Our findings thus pave the way for methodologies to quantify the human infectious reservoir based on conventional blood sampling approaches to support the deployment and monitoring of malaria elimination efforts for maximum public health impact.


72
Significant reductions in malaria burden in recent decades have stimulated malaria 73 elimination initiatives (1). It is widely accepted that malaria elimination with current tools is 74 unlikely for the majority of African settings (2). Therefore, novel interventions are needed and 75 approaches that specifically reduce malaria transmission may be of key importance (3). 76 Transmission of malaria depends on the presence of mature male and female gametocytes that 77 circulate in the bloodstream and may be ingested by mosquitoes from the subdermal 78 capillaries upon blood feeding. For P. falciparum, these circulating mature gametocytes are 79 the product of a prolonged developmental process that starts with commitment of asexual 80 parasites to the sexual pathway upon activation of AP2-G(4, 5). Developing gametocytes are 81 sequestered for 10-12 days, primarily to the bone marrow and spleen (6), until their release 82 into the blood circulation as mature gametocytes. Mosquitoes may become infected when 83 feeding and ingesting mature male and female gametocytes, even if their densities in the 84 peripheral blood are low (7). Interestingly, mosquito infections have been observed from 85 gametocyte donors whose low gametocyte density appears incompatible with transmission 86 (8). Mosquito infection rates are typically higher when mosquitoes feed directly on the skin of 87 gametocyte carriers, as compared to feeding on venous blood through an artificial membrane 88 (9, 10). In addition to a strategic adjustment of gametocyte sex-ratio to maximize transmission 89 success (7,11,12), gametocyte aggregation and sequestration may facilitate mosquito 90 infections from low gametocyte densities. Aggregation of gametocytes in blood meals has 91 been observed (13) and may increase the chance that both male and female gametocytes are in the skin is the best predictor of infectiousness (15)(16)(17)(18). 98 Indirect evidence for skin sequestration of mature gametocytes in the microvasculature 99 of the skin was first described following surveys conducted in the 1940s and 1950s in DR 100 Congo: gametocyte prevalence in a survey using skin scarification was 3-fold higher 101 compared to a survey 5 years earlier using finger prick blood (19). In a follow up study with 102 1243 paired samples, a more modest 13.4% increase in P. falciparum parasite prevalence and 103 15.6% increase in gametocyte prevalence was observed when blood and dermal fluids from 104 skin scarification were used for sample preparation instead of finger prick blood (20 (71.0%) membrane feeding experiments resulted in at least one infected mosquito (p=0.289).

135
To examine whether this higher infectivity in direct skin feeding assays was related to 136 higher ingested gametocyte densities, or to a higher gametocyte fraction in the blood meal, we 137 directly quantified gametocytes and asexual parasites in mosquito blood meals. The blood 138 content of individually fed mosquitoes was released into an RNA preservative 15 minutes 139 after starting the feeding; RNA was then extracted and quantified from pools of 4 mosquitoes. 140 We quantified asexual parasites by skeleton-binding protein 1 sbp1 qRT-PCR (31) and 141 gametocytes (Pfs25 and Pfmget qRT-PCR) in a median of 3 mosquito pools per participant, 142 each containing 4 individual mosquitoes, from skin-feeding (range=2-3) and 4 pools per 143 participant, each containing 4 individual mosquitoes, from membrane feeding (range=2-4). 144 We observed strong correlations between parasite quantities in pools of mosquitoes that fed 145 8 on skin or venous blood through artificial membranes for asexual ring-stage parasites 146 (r=0.921, p<0.0001), male (r=0.790, p<0.0001) and female gametocytes (r=0.655, p=0.0001) 147 ( Figure 1B). We also expressed gametocytes as a fraction of the total parasite biomass. This 148 fraction ranged from very low (<1% gametocytes in an individual with 21,086 ring-stage 149 asexual parasites/µL and 179 gametocytes/µL) to 100% in 3 individuals without asexual 150 parasites detected by qRT-PCR ( Figure 1C). We observed no tendency towards a higher 151 fraction of gametocytes in skin-fed mosquitoes or capillary blood compared to venous blood 152 ( Figure 1D).

153
In a complementary study, 9 adult gametocyte carriers participated in skin biopsy 154 sampling. After a screening visit, participants were seen on 2 occasions spaced 4 days apart. samples ( Figure 2B) and not significantly different between venous or finger prick blood 171 (p≥0.121) or between leg skin tissue or arm skin tissue (p≥=0.116). The same RNA aliquots 172 were also processed for analysis by Nanostring expression array, a highly sensitive probe-173 based expression platform that we have optimized for use in P. falciparum (32, 33). Using a 174 previously defined stage-specific marker set for asexual rings and mature gametocytes (33, 175 34), there was no evidence for higher gametocyte transcripts in skin samples compared to 176 blood samples ( Figure 2C). The two approaches to quantify gene expression also showed a 177 strong correlation for sbp1 and Pfs25 ( Figure 2D).

178
To directly detect gametocytes in subcutaneous tissue, skin biopsy samples that were (20). In the current study, we therefore not only collected venous blood and finger prick blood 219 but we also directly quantified parasite stage composition in skin tissue of naturally infected 220 11 donors and in blood meals of mosquitoes that naturally fed on the skin of the corresponding 221 donor. We used the absolute quantity of gametocytes and the fraction of the total parasite 222 biomass that is gametocyte as indicators of sequestration. In skin biopsy samples, we only 223 sporadically encountered gametocytes by histology. We chose a fluorescence imaging 224 protocol to image thick sections by confocal microscopy. This method allowed capturing of 225 entire parasites and three-dimensional reconstruction of parasite and surrounding tissues.

226
Using Pfs16 labelling we classified gametocytes by crescent shape, three-dimensional 227 structure (as opposed to non-specific speckles and autofluorescence, which is an inherent 228 issue of this approach), nuclear stain and presence of a surrounding red blood cell. The 229 frequency of immunofluorescence-detected gametocytes in our tissue samples was lower than 230 that by molecular methods in a tissue sample taken during the same visit. The quality of the 231 skin tissue, tested by analysing the tissue sections by haematoxylin and eosin staining, as well 232 as by labelling for endothelial cells, clearly indicates they were processed and preserved well.  While direct skin-feeding assays tend to result in higher infectivity compared that observed in 261 indirect feeding procedures using venous blood, our data demonstrate that any differences 262 observed are based on technical rather than biological differences in the feeding procedure.

263
Our findings also indicate that gametocyte levels in venous or finger prick blood can be used 264 to predict onward transmission potential to mosquitoes. Our findings thus pave the way for 265 methodologies to quantify the human infectious reservoir based on conventional blood 266 sampling approaches to support the deployment and monitoring of malaria elimination efforts 267 for maximum public health impact.   le Paludisme (CNRFP) in Ouagadougou for membrane feeding and skin feeding. Immediately 298 after venipuncture in lithium heparin and EDTA tubes (BD Vacutainer™), 400-500µL of 299 heparinized blood in duplicate (for infectivity) and 400-500µL EDTA blood (for gametocyte 300 quantification in blood meals) was offered to 60 starved 4-5-day-old female An. coluzzii 301 mosquitoes via an artificial membrane attached to a water-jacketed glass feeder maintained at 302 37°C (28). After exactly 15 minutes of feeding in the dark, fully fed mosquitoes from heparin 303 blood were transferred to storage cups by aspiration and maintained with glucose solution at 304 27-29°C for 6-8 days before dissection with 1% mercurochrome staining and examination for 305 oocysts by two independent microscopists. From mosquitoes that fed on EDTA blood, 16 306 fully fed mosquitoes were sacrificed after feeding for exactly 15 minutes by sharp needle 307 puncture of their midguts to release the blood contents into 50µl of RNAprotect cell reagent; 308 blood meal material was stored for individual mosquitoes at -80°C. Immediately following 309 membrane feeding, direct skin feeding took place. The participant's calves were exposed to 310 60 mosquitoes distributed over 2 paper cups that were allowed to feed for exactly 15 minutes.

311
From this group, 12 fully fed mosquitoes were immediately sacrificed and their midguts 312 punctured as described above. Remaining mosquitoes were maintained on glucose solution 313 before dissections for oocyst presence, as above. In addition to the membrane and direct skin 314 feeding assays, K2EDTA blood was collected by venipuncture (BD Vacutainer™) and finger 315 prick (BD Microtainer®). stages. For NanoString analysis, 5 l of purified total RNA was used for initial hybridization 362 reaction. RNA from each sample was allowed to hybridize with reporter and capture probes at 363 65°C for 20 hours according to the nCounter gene expression assay protocol (NanoString 364 Technologies). RNA-probe complexes were immobilized to nCounter cartridge followed by 365 scanning in the nCounter Digital Analyzer. Data was first normalized by applying background 366 subtraction and then normalized to expression of housekeeping genes using the R package 367 "NanoStringNorm". The dataset was then quantile normalized using the R package 368 "aroma.light" and rank scaled. Mature gametocyte and asexual marker genes, as defined in 33 , 369 were then averaged per patient, per tissue and per visit. were generated to act as positive and negative controls. Asexual and mixed asexual-immature 409 gametocyte clots and mature gametocyte clots were generated as described previously (6). For the paired skin feeding-membrane feeding study, we assumed an average of 15% infected 419 mosquitoes in patent gametocyte carriers with a standard deviation of 20% and a within 420 19 subject correlation of the outcome of 0.5 (9, 46, 47). If we then expected two-fold higher 421 mosquito infection rates in direct skin feeding, 17 paired membrane feeding and skin-feeding 422 experiments on patent gametocyte carriers would give 80% power to detect this difference at 423 an alpha of 0.05. Sample size justification for skin-biopsy sampling was based on a paired 424 comparison of the proportion of the total parasite population that is mature gametocyte. We 425 expected that 73% of the skin snip biopsy samples had higher gametocyte concentrations, 426 based on a meta-analysis that demonstrated enhanced infectivity following skin feeding 427 compared to venous blood membrane-feeding (9). When assuming that 70% of infected adults