To summarize, we were able to demonstrate that exposure to HIV-1 directly decreased the transepithelial resistance across intestinal and genital epithelial monolayers. The reduction in TER correlated with significant decrease in tight junction protein expression and increased permeability, indicating functional impairment of the barrier. The effect was specific for HIV-1 and reached significant levels within 2–4 hours following HIV-1 exposure. Similar reduction in tight junction functioning was observed following treatment of ECs with HIV-1 envelope protein gp120 but not tat, a regulatory protein. Neutralization of gp120 and exposure to an Env− HIV significantly abrogated the impairment of epithelial barrier, indicating that the effect was mediated by HIV-1 envelope glycoprotein. We further determined that exposure of the epithelial monolayers to HIV-1 led to enhanced production of a number of inflammatory cytokines, including TNF-α, by both intestinal and genital epithelial cells. When epithelial cells were exposed to HIV-1 in presence of anti-TNF antibody, there was no significant decrease in TER, indicating that TNF played a major role in impairing the barrier functions. In experiments designed to determine whether the disruption of epithelial barrier function could be directly associated with microbial leakage across the mucosa, we found evidence for small but significant bacterial and viral translocation across epithelial monolayers following HIV-1 exposure.
To the best of our knowledge, this is the first study to demonstrate that HIV-1 can directly disrupt mucosal epithelial barrier functions that can lead to enhanced microbial translocation. Previous clinical studies have documented that in HIV-1 infected patients intestinal permeability is altered, characterized by diarrhea-induction 
. A recent study showed impairment of barrier function in intestinal biopsies of HAART-naïve patients compared to those on HAART treatment 
. Increased production of cytokines IL-2, IL-4, IL-5 and TNF-α was found in supernatants of cultured intestinal biopsies in this study. Their conclusion was that following infection, HIV replication in target cells leads to local increase of inflammatory cytokines in the intestinal mucosa, which induce barrier impairment. This supports previous studies where PBMCs co-cultured with HIV-infected macrophages resulted in increased production of a number of cytokines, including TNF-α, IL-1β, IFN-α and IFN-γ which were shown to compromise epithelial barrier function 
. The prevailing opinion from these studies is that the effect on epithelial barrier is likely mediated via immune cell activation due to viral replication 
. Of note are other studies that were unable to show that mononuclear cells isolated from colon of infected patients produce increased amount of cytokines 
. Thus far the cellular source of inflammatory cytokines that could lead to barrier disruption in HIV infected patients remains controversial 
. Based on our results, we would like to propose that the primary sources of the inflammatory cytokines that disrupt the mucosal barrier are the epithelial cells themselves. Our studies demonstrate that epithelial cells respond directly and rapidly to HIV envelope glycoprotein by production of increased levels of cytokines which lead to loss of barrier functions, rather than an indirect effect mediated by immune cells following HIV replication. This provides an alternate and more direct explanation as to why decrease in viral load following HAART treatment restores intestinal barrier functions 
. Our results that demonstrate that barrier dysfunction can allow bacterial translocation could also provide explanation for increased levels of immune activation during acute infection, an observation noted in a previous study which examined immune activation following HIV infection in North American cohorts 
. The mechanism demonstrated in the present study does not exclude the possibility that cytokines released from immune cells in the HIV-infected intestines could also contribute to further disruption of the barrier, more likely in the chronic phase of the infection.
That viral exposure could directly lead to compromised barrier function has been shown before 
. Many other viruses and even bacteria have been shown to directly compromise both epithelial and endothelial barrier integrity. Astrovirus, a single stranded RNA virus and a causative organism of common diarrhea was recently shown to increase epithelial barrier permeability in Caco-2 intestinal cells, modulated by its capsid protein, independent of viral replication 
. Coxsackievirus has also been shown to directly compromise endothelial tight junctions 
. Previous studies have shown that HIV-1 infection can compromise the blood-brain barrier thereby leading to progression of HIV-1 encephalitis (reviewed in 
). The functioning of the tight junctions between endothelial cells, that form the blood-brain barrier, is quite similar to those present between mucosal epithelial cells. However, the mechanism elucidated by these studies was not a direct effect of HIV, but facilitated by production of TNF-α during chronic infection that mediated opening of paracellular route in endothelial lining, for viral entry into the brain. Interestingly, a recent study elucidated that HIV-1 tat protein can directly compromise the retinal epithelial barrier function 
. Despite this evidence, no studies have so far examined the direct effect of HIV-1 exposure on mucosal epithelium. Our results show that increased permeability is mediated directly by HIV viral envelope glycoprotein. Further, given that significant disruption of tight junction proteins and decreased TERs occurred following treatment with UV inactivated virus, this phenomenon is independent of viral replication. Whether HIV-1 entry is required for the epithelial cell response is currently being examined.
In our study, both the intestinal cell line and primary genital epithelial cells showed similar response to different strains of HIV-1: disruption of tight junctions, and increased permeability. However, we found the profile of cytokines produced constitutively by intestinal and genital epithelial cells was quite distinct. While the intestinal cell line T84 did not constitutively produce TNF-α, IL-6, IL-8 and MCP-1, there was significant induction of these cytokines following HIV-1 exposure. Primary genital epithelial cultures, on the other hand, constitutively produced TNF-α, IL-6, IL-8 and MCP-1 and production of TNF-α and IL-6 was significantly upregulated following HIV-1 exposure. Both types of ECs secreted minimal levels of IL-10 and IL-1β which was upregulated following HIV-1 exposure only in primary genital epithelial cells. The differences in the constitutive cytokine profile between genital and intestinal epithelial cells could be due to distinct characteristics of primary cells compared to cell lines. Alternatively, intestinal epithelial cells are likely to be more quiescent in terms of baseline cytokine production given their microenvironment where a variety of commensal organisms are always present in the lumen 
. In comparison, upper genital tract epithelium exist in a relatively sterile environment and are known to actively secrete an array of cytokines 
. Nevertheless, following exposure to HIV-1 both types of ECs responded with enhanced induction of inflammatory cytokines that mediated disruption of tight junctions. This indicates that as long as the viral load and exposure times are sufficient, HIV can likely disrupt any mucosal barrier in the body, independent of infection and replication.
Among the cytokines that were upregulated, the direct effect of TNF-α on disruption of intestinal epithelial tight junction and increased permeability has been extensively characterized 
. TNF-induced increase in permeability of Caco-2 cells is known to be mediated by NF-kB activation that downregulates ZO-1 protein expression 
. ZO-1 proteins are integral part of the tight junction assembly and function as a scaffolding protein critical in maintaining the integrity of the tight junctions. The results from ZO-1 quantification () indicate that the disruption of tight junctions following HIV-1 exposure likely happens in two stages: initially there may be a displacement of ZO-1 that leads to disruption of tight junction integrity followed by marked reduction in the amount of ZO-1 and other tight junction proteins due to decreased transcription. Thus, TNF-α produced by the ECs in response to HIV-1 envelope glycoprotein could induce NF-kB activation and subsequent downregulation of tight junction proteins, including ZO-1. Our ongoing studies show that NF-kB translocation occurs within 1 hour of HIV-1 exposure (Nazli and Kaushic, unpublished). Whether there are distinct steps in this process that have discrete mechanisms is currently under investigation. Regardless of the detailed mechanism, the outcome of tight junction disruption is decrease in TER and leakage across the epithelial barrier.
The finding that disruption of barrier function can result in small but significant amount of both viral and bacterial translocation across ECs following exposure to HIV-1 has profound implications. Although previous studies have demonstrated presence of LPS in serum of HIV infected patients and correlated it with immune activation in North American cohorts, the inference that microbial flora in the intestines was the source of LPS was indirect 
. Our studies provide direct evidence that both bacteria and virus present on the apical side of mucosal epithelial cells during HIV-1 exposure could leak through because of the impairment of epithelial tight junctions and increased permeability. This could allow HIV-1 access to target cells located in the lamina propria of the mucosa as well allow bacterial translocation that could cause local immune activation.
While the viral-epithelial interactions described here are novel, further investigation is needed to determine what role increased barrier permeability plays in initiating HIV-1 infection. HIV-1 transmission across intestinal and genital mucosa occurs predominantly via infected semen; currently the role of seminal plasma in HIV-1 transmission is far from clear. Recent studies indicate that seminal plasma can lead to inflammatory responses and facilitate HIV-1 transmission 
. However, seminal plasma components such as TGF-β and HGF also enhance epithelial barrier functions 
. Further, given the low efficiency of viral translocation seen in the ex-vivo model described here, the ability of HIV-1 to cross over in significant numbers, in vivo, would be depend presence of high viral load in the semen, most likely in acute phase of infection. While plasma and semen loads show overall correlation, compartmentalization between genital and blood viral loads is well recognized and more recent studies show that seminal plasma viral load can persist following treatment with HAART 
. While the results from the present study elucidate a new mechanism that could lead to viral translocation across the epithelial barrier, more information is needed to understand how other factors like seminal plasma, stage of the infection and viral load may influence any viral leakage across the mucosal barrier. If under physiological conditions, viral leakage does occur because of barrier disruption, it could play a critical role in initiation of infection, especially in the presence of existing inflammation from other viral or bacterial co-infections 
In conclusion, the current study provides evidence for the first time that HIV-1 exposure at the mucosal surface leads to direct response by the mucosal epithelium, seen by production of inflammatory cytokines. This response is rapid, independent of viral infection and likely plays a key role in initiation of mucosal damage. This information will be critical for strategies to target control of mucosal damage.