HIV-1 can be detected in body fluids such as breast milk or seminal plasma despite control of the viral load in the blood with effective ART (Neveu et al, 2011
; Zhang et al, 1998
) and provides a continuous source of virus particles to easily spread the infection. DCs or Langerhans cells of the intestinal and vaginal mucosa were repeatedly described to be the first target for HIV-1 and the vehicle for the virus to reach replication competent cells (Hladik et al, 2007
; Hu et al, 2000
; Shen et al, 2010
; Spira et al, 1996
). Whether DCs/Langerhans cells have an active role in capturing the virus through the epithelial cell layer or are infected following viral release into the lamina propria, can have important implications in prevention strategies. Here, we finally demonstrate that DCs rapidly migrate through a monostratified intestinal epithelium only in response to CCR5-using HIV-1, independently of genetic subtype, in an ex vivo
organ culture and an in vitro
Caco-2/DCs polarized culture. Whether such mechanism of DCs migration occurs also in pluristratified epithelia will require further analysis. We provide evidence that DCs migration across a tight intestinal epithelium is triggered by the R5 viral env itself engaging DCs via the CCR5 molecule. Accordingly, DCs migration is abrogated by treatment of the cells with CCR5-binding ligands. These results further support the use of CCR5 inhibitors for prevention, as already suggested by the Rhesus Macaques vaginal infection model, where vaginal application of CCR5 inhibitors prevented transmission of chimeric Simian-HIV (Lederman et al, 2004
; Veazey et al, 2005
). It is conceivable, that analogous interventions can be implemented also for prevention of infection of rectal tissue, where drugs were shown to be more efficient than in vaginal tissues (Herrera et al, 2011
The migration process of DCs is R5 viral env mediated and is further restricted to the V3 loop, suggesting that an env-CCR5 interaction is necessary to initiate DCs migration across the epithelium. Interestingly only one (SF162) out of 6 R5 viruses was not able to trigger migration but was transcytosed through epithelial cells. If this is an exception to the rule or if other viruses may have its same restrictions will need further testing and possibly help to identify specific determinants. In addition, viral envelopes with the intrinsic characteristic to induce migration of and be captured by DCs may be selected and exploited in mucosal vaccination strategies to favour antigen presentation.
The thorough tracing of the virus travel through the epithelial monolayer revealed that the release of virus through a transcellular pathway was not sufficient to engage DCs and, that the virus can penetrate in between adjacent epithelial cells and localize in intrajunctional spaces close to the adherent junctions. We propose that HIV-1, similarly to other viruses (Nava et al, 2004
; Wang et al, 2011
), induces a transient opening of the TJs to gain access through the epithelium. It may be plausible that intraepithelial virions can be also released laterally into the interepithelial spaces and thus, create a paracellular viral gradient attracting DCs. The role of the virus itself is also supported by the absence of detectable levels of CCR5-binding chemokines in the culture systems together with the previously published observation that HIV-1 R5 env is chemotactic for DCs (Lin et al, 2000
). Analogously it was described that HIV-1 gp120 binding to CCR5 activates a signaling cascade leading to phosphorylation of the nonreceptor tyrosine kinase Pyk2, and downstream activation of p38MAPK, which ultimately leads to DCs migration (Anand et al, 2009
; Harman et al, 2006
; Wilflingseder et al, 2004
). Thus, it is conceivable that also in our model signaling through CCR5 may be required to induce DCs migration across the epithelium.
We observed the formation of junction like structures between DCs and epithelial cells, which may initiate the haptotaxis of DCs, namely their directional motility towards a gradient of cellular adhesion sites. We showed previously that blocking the TJs proteins on DCs inhibits the opening of adjacent epithelial cell junctions (Rescigno et al, 2001
; Rimoldi et al, 2004
). We may envisage that the close contact between DCs and epithelial cells may also favour cell-to-cell viral spreading of intraepithelial virions. It is tempting to speculate that during transmission of mixed R5 and X4 viruses these latter ones, internalized in the epithelial cells and unable to attract DCs themselves, may hijack migrated DCs for their own purpose.
Some studies showed a decreased barrier function and enhanced permeability following HIV exposure using in vitro
cultures of intestinal epithelial cells (Nazli et al, 2010
; Schmitz et al, 2002
). We provide several evidences that in our experimental setting the epithelial barrier is instead completely preserved. This discrepancy could be ascribed to different factors: we used lower virus input and shorter time of virus exposure compared to the other studies, which makes our experimental setting closer to physiological conditions. However, also in our hands the morphology of the epithelial monolayer is altered with long-term exposure, which underlines the necessity of severely controlled settings in the mucosal culture systems. In our ex vivo
organ culture, we confirmed that the virus is heavily trapped by the mucus present on the epithelial cells, as previously shown (Maher et al, 2005
). It would be relevant to understand how infected body fluids, such as milk and semen, may impact on epithelial integrity or cellular cross-talk. Indeed, some studies indicated that seminal plasma can lead to inflammatory response and facilitate HIV-1 transmission (Planchon et al, 1999
), other studies showed instead an enhanced epithelial barrier function mediated by seminal plasma components (Gorodeski & Goldfarb, 1998
). Unfortunately, the frailty of these organ culture models does restrict very much its use.
DCs fail to migrate in response to X4 viruses both in the in vitro
and ex vivo
culture models used in our study. In the in vitro
system, it may be ascribed to low levels of CXCR4-expression on DCs used, however, we show that DCs expressing a comparable level of the two co-receptors remain unresponsive to X4 viruses. In human colon, DCs express constitutively CXCR4 and CCR5 (68.6
16.7 and 36.7
19.35%, respectively). It can be envisaged, that the stromal derived factor-1, which is expressed by intestinal epithelial cells (Agace et al, 2000
), may interphere with the binding of the virus to CXCR4, thereby preventing the migration of DCs in tissues.
We show with both models that DCs migrate and capture R5 virus as fast as within 30
min, which indicates that they are the first cells that capture HIV-1 in the intestine. However, the extent to which DCs in the intestine are infected is still unknown and would need a thorough analysis. Interestingly, the migration process is specific for lamina propria resident DCs and not Mϕ, which instead remained confined within the subepithelial layer. An involvement of Mϕ in the subsequent passage of HIV-1 within the intestinal mucosa may not be excluded, however, human intestinal Mϕ were shown to be poorly infected (Smith et al, 1994
), and did not support efficient viral replication in vitro
(Li et al, 1999
; Meng et al, 2000
; Shen et al, 2011
). Furthermore, in contrast to Mϕ, DCs can migrate into draining lymph nodes and thus, may play an active role in viral spreading from the very first moments after infection.
Interestingly, we showed that DCs migration is reversible, thus DCs sensing a chemokine gradient may migrate back into the lamina propria to further spread infection. This observation is in line with a recent study showing that in a penile foreskin culture LCs move from the epidermis to the dermis after 4
h from exposure to HIV-1 infected PBMC and form conjugates with T cells (Zhou et al, 2011
). The relevance of DCs in disseminating further the virus, as early as 24
h after infection, to genital lymph nodes was demonstrated in intra-vaginally inoculated animals (Hu et al, 2000
; Masurier et al, 1998
; Miller et al, 1994
). The question if virus loaded DCs of the intestine will also migrate to the lymphatic system or just to the lamina propria cannot be concluded in vitro
but will need in vivo
studies. In the mouse intestine, it was described that CD103+
DCs migrate to lymph nodes whereas the CX3CR1+
ones are deputed to extend dendrites to sample bacterial antigens in response to fractalkine released by the epithelium (Niess et al, 2005
; Schulz et al, 2009
). The DCs used in our in vitro
model express CX3CR1, but their migration was not abrogated by pre-treatment with its ligand fractalkine (data not shown). Thus, it will be interesting to understand whether these two functionally distinct sub-sets of antigen presenting cells have also in human similar functions.
There is plenty of evidences that to spread the infection the DCs transfer virus to CD4+
T cells, which sustain actively viral replication, via both vesicular uptake (called trans
infection) and de novo
production (called cis
infection) (Garcia et al, 2008
; Geijtenbeek et al, 2000
; Hladik et al, 2007
; Hu et al, 2004
; Pope et al, 1994
; Turville et al, 2004
). Here, we show that also those DCs, which sampled virions through an epithelial barrier, maintain the ability to spread the infection to CD4+
T cells for at least 4 days from infection. These same DCs, when cultured alone, were showing low levels of virus release, which may either represent active viral replication or release of membrane-attached virions. R5 HIV-1 was rapidly internalized by DCs and redirected into an intracellular compartment, which neither co-localized with ovalbumin-positive retention vesicles nor with acidic lysosomes. However, the GFP+
virus was lost within 24
h suggesting that it did not remain intact inside the cells. Our observation that transfer to CD4+
T cells could occur after several days of culture, indicates that the pathway we observed lead, at least in part to virus integration. The low expression of CXCR4 on DC would instead hamper their direct infection and favour the trans
pathway. These results are in line with previous reports suggesting a bi-model transmission of virus (Cavrois et al, 2007
; Turville et al, 2004
in the short period virus was transferred in trans
whereas in the long period through de novo
production. In conclusion, our data indicate that DCs can act as reservoir and become a source of virus detrimental for the surrounding cells and affecting innate and adaptive immune responses.
In , we present a schematic model of the different pathways used by R5 and X4 viruses to access the intestinal mucosa and disseminate within the host. Despite most of the mechanisms used are similar for the two viral phenotypes, R5 viruses are the only capable of selectively recruit DCs to migrate through the epithelium to capture luminal virions. This pathway, which may possibly give to R5 viruses an advantage in migrating to lymph nodes to further spread the infection, could be targeted to design new prevention strategies.
Schematized proposed mechanism of HIV-1 penetration across the intestinal epithelium and DCs recruitment.