A Critical Role for Perivascular Cells in Amplifying Viral Haemorrhage Induced by Dengue Virus Non-Structural Protein 1

Dengue is the most prevalent arthropod-borne viral disease affecting humans, with severe dengue typified by potentially fatal microvascular leakage and hypovolaemic shock. Blood vessels of the microvasculature are composed of a tubular structure of endothelial cells ensheathed by perivascular cells (pericytes). Pericytes support endothelial cell barrier formation and maintenance through paracrine and contact-mediated signalling, and are critical to microvascular integrity. Pericyte dysfunction has been linked to vascular leakage in noncommunicable pathologies such as diabetic retinopathy, but has never been linked to infection-related vascular leakage. Dengue vascular leakage has been shown to result in part from the direct action of the secreted dengue virus (DENV) non-structural protein NS1 on endothelial cells. Using primary human vascular cells, we show here that NS1 also causes pericyte dysfunction, and that NS1-induced endothelial hyperpermeability is more pronounced in the presence of pericytes. Notably, NS1 specifically disrupted the ability of pericytes to support endothelial cell function in a 3D microvascular assay, with no effect on pericyte viability or physiology. These effects are mediated at least in part through contact-independent paracrine signals involved in endothelial barrier maintenance by pericytes. We therefore identify a role for pericytes in amplifying NS1-induced microvascular hyperpermeability in severe dengue, and thus show that pericytes can play a critical role in the aetiology of an infectious vascular leakage syndrome. These findings open new avenues of research for the development of drugs and diagnostic assays for combating infection-induced vascular leakage, such as severe dengue. SIGNIFICANCE STATEMENT The World Health Organisation considers dengue one of the top ten global public health problems. There is no specific antiviral therapy to treat dengue virus and no way of predicting which patients will develop potentially fatal severe dengue, typified by vascular leakage and circulatory shock. We show here that perivascular cells (pericytes) amplify the vascular leakage-inducing effects of the dengue viral protein NS1 through contact-independent signalling to endothelial cells. While pericytes are known to contribute to noncommunicable vascular leakage, this is the first time these cells have been implicated in the vascular effects of an infectious disease. Our findings could pave the way for new therapies and diagnostics to combat dengue, and potentially other infectious vascular leakage syndromes.


23
Dengue is the most prevalent arthropod-borne viral disease affecting humans, with 24 severe dengue typified by potentially fatal microvascular leakage and hypovolaemic shock.

82
At a microvascular level, the precise mechanisms leading to vascular leakage in 83 severe dengue remain incompletely understood. A dysregulated cytokine response has been 84 proposed to be a major contributor to vascular hyperpermeability, especially in heterotypic 85 secondary infections with a different DENV serotype to the first infection (3, 5, 6). Antibodies 86 against the viral non-structural protein NS1 have also been shown to bind to endothelial cells 87 in secondary infections, causing their apoptosis (7-9). In addition, NS1 protein is abundantly 88 secreted into patient serum and has been shown to directly induce endothelial cell 89 hyperpermeability (10, 11). Serum NS1 concentrations in dengue patients vary widely.

90
Although concentrations as high as 15,000 ng/ml have been recorded in some patients, 91 concentrations of 10-1,000 ng/ml are more typical, with one study suggesting that levels above 92 600 ng/ml may be predictive of severe disease (12-14). NS1 enters endothelial cells through

113
Here, we use microvascular hyperpermeability induced by DENV NS1 as a model to 114 demonstrate a crucial role for pericytes in amplifying an infectious haemorrhagic syndrome.

115
We show that DENV-2 NS1 induces hyperpermeability in in vitro co-cultures of primary 116 pericytes and primary endothelial cells, and that the observed hyperpermeability is greater 117 than for endothelial cells cultured alone. NS1 does not broadly affect all pericyte functions, but 118 rather specifically reduces the capacity of pericytes to support endothelial cell function in 3D 119 microvascular models. Finally, we demonstrate that NS1-induced hyperpermeability is not  pericytes to support the barrier function of endothelial cells in this system. In contrast, 152 ovalbumin did not affect permeability of the co-culture (Fig 1D), confirming the specificity of 153 the observed pericyte dysfunction induced by DENV-2 NS1.

154
Notably, the absolute magnitude of the TEER decrease was significantly higher in the 155 co-cultures compared to endothelial cells cultured alone, indicating a stronger effect of NS1 156 on permeability in the presence of pericytes (Fig 1Ei). Even when we accounted for the higher

179
Treatment of 3D endothelial-pericyte co-cultures with 500 ng/ml DENV-2 NS1 caused 180 a reduction in branch width (Fig 2A), indicating a dysfunctional interaction between endothelial 181 cells and pericytes. Quantification of NS1's effect revealed a dose-dependent decrease in 182 branch width, with doses above 300 ng/ml completely abolishing the contribution of pericytes 183 to capillary-like structure formation (Fig 2Bi). In contrast, treatment with ovalbumin ("ova") did 184 not affect the endothelial-pericyte interaction in Matrigel (Fig 2Bii), demonstrating the 185 specificity of the NS1 effect. Furthermore, branch width was not impacted by NS1 treatment 186 when endothelial cells were cultured in the absence of pericytes, indicating that NS1 187 specifically interferes with the ability of pericytes to support the endothelium rather than the 188 intrinsic angiogenic capacity of endothelial cells (Fig 2Biii). Taken together, our data suggest 189 that the amplified effect NS1 has on endothelial permeability in the presence of pericytes is

199
In order to support the microvascular architecture, pericytes must be able to migrate 200 along capillaries. For this reason, we tested the intrinsic migratory capacity of pericytes upon 201 9 treatment with NS1. DENV-2 NS1 concentration as high as 1,000 ng/ml had no effect on 202 pericyte migration compared to the untreated control in a scratch assay measuring 'wound' 203 closure following manual disruption of the cell monolayer ( Fig 3B). These data further support 204 our finding that NS1 specifically affects the endothelial-pericyte interaction.

256
We propose a mechanism whereby, early during DENV infection, NS1 secreted into 257 patient serum induces well-described direct effects on endothelial cells causing local 258 permeability that allow NS1 to gain access to pericytes on the apical side of the endothelium 259 (Fig 4i-iii). Subsequent effects of NS1 on pericytes cause a dysregulation of paracrine 260 signalling between pericytes and endothelial cells that leads to pronounced and potentially 261 life-threatening hyperpermeability that manifests during defervescence (Fig 4iv). In patients 262 that recover, the integrity of the microvascular endothelium is restored as pericytes regain the 263 ability to support endothelial cell functions crucial to endothelial barrier formation and 264 maintenance (Fig 4v).

265
Concentrations as high as 15,000 ng/ml of NS1 have been observed in the serum of 266 dengue patients, however patients with severe dengue most commonly have serum NS1 267 levels ranging between 10-1,000 ng/ml (12-14). In our hands, NS1 affected the ability of 268 pericytes to support endothelial cell function in 3D microvascular co-cultures at concentrations 269 as low as 300 ng/ml (Fig 2B). To our knowledge, this is the lowest concentration at which 270 microvascular effects of NS1 have been demonstrated in vitro. Furthermore, in endothelial-271 pericyte co-cultures, the observed reduction in TEER upon treatment with 500 ng/ml of NS1 272 is dramatically larger compared to endothelial cells cultured alone (Fig 1E). We therefore

518
NS1 to access pericytes as NS1 levels peak. (iv) Pericyte function is modulated by NS1, 519 causing pericytes to secrete a paracrine factor ("factor X") that signals to endothelial cells to 520 further disrupt the endothelial barrier, resulting in enhanced vascular leakage after NS1 levels 521 peak. (v) Pericytes recover from transient NS1 effects and normal endothelial-pericyte 522 signalling is restored to allow cardiovascular recovery. All data N = 3; n = 4. Error bars 523 represent standard error of the mean. * P < 0.05; ** P < 0.01; *** P < 0.001. In Fig 4A, the 524 difference between "EC alone" and "untreated" (co-cultures) is significant at P < 0.05; 525 significant differences from "untreated" (co-cultures) are shown above the line for TNF-α 526 treatment and below the line for NS1 treatment.