Dissemination of Cryptococcus neoformans via localised proliferation and blockage of blood vessels

Cryptococcus neoformans is an opportunistic fungal pathogen that can cause life-threatening cryptoccocal meningitis, predominantly within immunocompromised individuals. Cortical infarcts are observed in as many as 30% of cryptococcal meningitis cases, being particularly common in severe infection. Limited clinical case studies suggest infarcts are secondary to vasculitis and blood vessel damage caused by cryptococcal infection. However, the cause of infarcts in cryptococcal infection has not been determined. To examine potential causes of vascular damage and cryptococcal dissemination in cryptococcal infection, the zebrafish C. neoformans infection model was used. We demonstrate that spread of cryptococci from the vasculature occurs at sites where cryptococci grow within the blood vessels, originating from a single or small number of cryptococci. We find that cryptococcal cells become trapped within the vasculature and can proliferate there resulting in vasodilation. Localised cryptococcal growth in the vasculature is also associated with sites of dissemination – in some cases simultaneously with a loss of blood vessel integrity. Using a cell-cell junction protein reporter (VE-cadherin) we identified sites dissemination associated with both intact blood vessels and where vessel rupture occurred. Thus, we have identified a mechanism for blood vessel damage during cryptococcal infection that may represent a cause of the vascular damage and cortical infarction observed in cryptococcal meningitis. Author summary Human infection by the fungal pathogen, Cryptococcus neoformans, can lead to life-threatening cryptococcal meningitis. In severe cases of cryptococcal meningitis, a lack of blood supply can cause tissue death and a resulting area of dead tissue (infarct) in the brain. Although vasculature inflammation in known to occur in cryptococcal meningitis, the cause of infarcts in unknown. Using a zebrafish model of cryptococcal infection, the growth and dissemination of fungal cells was observed over time. We show that cryptococcal cells become trapped and proliferate in the vasculature, resulting in cryptococcoma that damage the blood vessels. We propose that vessel damage results from increased blood pressure caused by cryptococci blocking blood vessels suggesting that the vascular damage that ensues on cryptococcoma formation may in turn be a cause of infarct formation seen in cryptococcal meningitis.

On the other hand, host responses may also enable cryptococcal 79 dissemination into the brain. In vivo studies have suggested that cryptococcal cells use host 80 macrophages as Trojan horses to cross the BBB, (7). Whilst various factors may play a role 81 in C. neoformans ability to cross the blood brain barrier, each method requires the presence 82 of cryptococcal cells within brain blood vessels.

83
A small number of clinical studies have suggested that blood vessel damage and bursting 84 may also facilitate cryptococcal dissemination, however the mechanism of blood vessel 85 damage is not known. Case reports indicate that cortical infarcts are secondary to 86 cryptococcal meningitis, and suggest a mechanism whereby resulting inflammation may lead 87 to damage to blood vessels (8-10). This is supported by the observation that cerebral  105 including brain dissemination, recapitulated in this model (16,17). Furthermore, the 106 cryptococcal zebrafish model enables high quality imaging of host pathogen interactions, 107 including throughout the vasculature, that are of specific relevance for this study. Notably, a 108 high fungal burden within a particular zebrafish cranial blood vessel, is correlated with a 109 higher chance of tissue invasion into the brain (Tenor et al., 2015). Based on this finding 110 together with the observation that cryptococcal cells can get trapped in small blood vessels 111 in the brain, we decided to investigate the role of cryptococcal expansion within blood 112 vessels as a route of dissemination. We postulated that once trapped within a brain blood 113 vessel, C. neoformans continues to proliferate, leading to physical damage of the 114 vasculature and eventual dissemination and invasion of the surrounding tissue.

115
In this study we observed cryptococcal cells becoming trapped and then proliferating within 116 the vasculature in a manner similar to that seen in murine models.

145
GFP and mCherry-labelled cryptococci and found that single colour infections were rare.

146
Therefore, we decided to use a skewed ratio so that we could better quantify the likely hood 147 of a population "bottleneck" during the progression of cryptococcal infection. We injected a

186
Cryptococcal clonal expansion is more common in small blood vessels.

187
It has been demonstrated that cryptococcal cells can become mechanically trapped in small 188 blood vessels in the brain and subsequently disseminate (13). The mechanism by which 189 cryptococci disseminate in this case is unknown but has been suggested to be via 190 trancytosisis (13). We found that individual cryptococcal cells become trapped in the inter-191 segmental vessels (ISVs) (Fig. 4A). We quantified the distribution of cryptococcomas and

192
found that most (80.3%) were located in the smaller brain and trunk blood vessels (Fig. 4B).

193
As cryptococcoma formation at the start of infection is seen in the smaller blood vessels, we

212
(4.5um) did not lead to formation of large masses, although there was a small but significant 213 increase in vessel size at locations where beads did become trapped in the vasculature by 214 3dpi (Fig. 5E). Additionally, beads were observed stuck in the inter-segmental blood vessels 215 significantly less frequently than live cryptococcal cells, with 13.6% of blood vessels 216 containing beads compared to 89.0% containing cryptococcal cells. In addition, we imaged 217 the small vessels of the brain and found that infected blood vessels were larger relative to 218 blood vessels in the same location in control animals ( Fig. 5F; Fig. 6G). This indicated that 219 viable cryptococcal cells have an increased ability to become trapped and form masses 220 compared to inert but similarly sized beads, and further implicates cryptococcal proliferation 221 as a mechanism of inflicting vessel damage rather than a build-up caused by an initial 222 blockage. Our finding that vessel diameter was enlarged suggested that blood vessels were 223 vasodilating to reduce the total peripheral resistance to decrease blood pressure due to

259
Here we demonstrate a hitherto uncharacterised mechanism of dissemination from blood

11
Clinical studies and case reports indicate that vascular damage is caused by cryptococcal 264 infection. Vascular damage during cryptococcosis may result from inflammation in the small 265 blood vessels, seen predominantly in severe cases of cryptococcal meningitis (8-10,12).

266
Indeed, in one retrospective study, infarcts were observed in as many as 30% of

286
In addition, we demonstrate that the presence and subsequent growth of cryptococcomas in 287 the vasculature can lead to pathogen dissemination. This is consistent with post-mortem

288
reports showing cryptococcal cells invading the brain, located next to brain capillaries that Evidence that cryptococcal cells can become trapped in brain blood vessels was first 319 adduced using a murine model. Furthermore, trapped cryptococcal cells appeared to 320 transmigrate into the brain in real time (13). We observed cryptococcal trapping in small 321 inter-segmental blood vessels in the zebrafish via a conserved trapping mechanism in 322 similar sized blood vessels. C. neoformans trapping in small blood vessels may be urease 323 dependent, particularly in organs such as the brain, but the mechanism is unknown (28

352
Here, using a VE-cadherin reporter, we demonstrate in vivo that at sites of cryptococcomas

452
Image analysis

453
Image analysis performed to measure the size of cryptococcal masses, and blood vessel 454 width was completed using NIS elements. Fluorescence intensity of GFP and mCherry C.

455
neoformans for low infection analysis was calculated using ImageJ software.

456
Statistical analysis

457
Statistical analysis was performed as described in the results and figure legends. We 458 used Graph Pad Prism 6 (v7.02) for statistical tests and plots.

461
We thank Bateson Centre aquaria staff for their assistance with zebrafish husbandry. We 462 thank Timothy Chico (University of Sheffield, UK) and Robert Wilkinson (University of