Viral Infection of the Lungs through the Eye^▿

Instillation of virus and inhibitors in the eye.

Female BALB/c mice, 6 to 8 weeks old, were purchased from Charles River Laboratories. RSV (Long strain, serotype A) was grown on HEp-2 cells and purified on sucrose layers to a concentration of 10¹¹ as described previously (6). Dilutions were done in phosphate-buffered saline (PBS) immediately before use as needed. A similarly diluted sucrose solution was used in sham-infected control mice. We followed the inoculation procedure optimized for other ocular pathogens such as herpes simplex virus (HSV) (42), which results in a consistent infection without causing “blepharitis” (eyelid inflammation). In brief, mice were anesthetized by intraperitoneal injection of pentobarbital (50 mg/kg), and the right eye was lightly scarified by two twists of a 2-mm corneal trephine. Various quantities of virus in 2 μl PBS was dropped onto the corneal surface and massaged in with closed eyelids. Control mice were inoculated with the same volume of PBS containing the same concentration of sucrose as the corresponding viral inoculum. The day of the inoculation was considered day 0.

In routine RNA interference (RNAi) experiments, 1 nmol of small interfering RNA (siRNA), complexed with TransIT TKO transfection reagent (Mirus), was applied in the eye at 30 min (unless otherwise stated) before instillation of RSV. The siRNA was based on the NA(N)₁₉NN design, corresponding to the sequence AAGCCCTATAACATCAAATTCAA of the P mRNA of RSV. Each strand of the siRNA was 21 nucleotides long and contained 3′-terminal dTdT extensions. The design followed the “asymmetry rule” (25, 37), with a ΔG difference of −6.7 kcal/mol between the two termini. Fluorescent siRNAs, labeled with Cy3 at the 5′ end of the sense strand, were custom designed and similarly applied into the eye. All siRNAs were commercially synthesized (Dharmacon). Negative control, scrambled siRNA of undisclosed sequence was purchased from Ambion.

When used, monoclonal rat antibody with neutralizing activity towards mouse IL-1α (R&D Systems) or TNF-α (BD Biosciences) was administered subconjunctivally at the same time as RSV. Just before use, the antibody was diluted in sterile PBS such that 25 μg antibody was present in 1 μl of the injected volume. The same amount of nonimmune mouse IgG was used as control.

Assays.

To detect Cy3-labeled siRNA (red), mice were sacrificed at 12 h, 24 h, 36 h, and 48 h after instillation of the liposomal siRNA, and the eyes were enucleated and prepared as 10-μm cryosections for fluorescence microscopy. Gross microscopic examinations were performed in a TE2000-E2 imaging station (Nikon) equipped with epifluorescence illumination and a Photometrics Coolsnap (Sony) image-capturing device.

When the eye or lung was to be used for reverse transcription-PCR immediately after organ collection, the tissue was mechanically homogenized in RNAwiz (Ambion) and RNA was purified according to the manufacturer's protocol. Quantitative real-time PCR was performed using standardized primers as described before (5, 43). In brief, first-strand cDNA was made using the GeneAmp RNA PCR Core kit (Perkin-Elmer-Applied Biosystems), and real-time PCR was performed on the iCycler iQ quantitative PCR system (Bio-Rad Laboratories) using the iQ SYBR green SuperMix. Gene expression measurements were calculated using the manufacturer's software; GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was used as an internal control. Infectious virus in the lung or eye was determined by serial dilution of the clarified tissue homogenate and plaque assay on HEp-2 monolayer as described previously (6). IL-1α and TNF-α in the homogenate were quantified by standard enzyme-linked immunosorbent assay (ELISA) (R&D Systems) (5, 6).

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting (Western blotting) using chemiluminescence-based detection of the horseradish peroxidase-conjugated secondary antibody were performed as described previously (43). Primary antibodies against RSV and mouse actin were from Chemicon International, and goat IgGs against mouse MIG, RANTES, and IP-10 were from R&D Systems.

Ocular and pulmonary pathology.

Ocular disease was evaluated with a slit lamp biomicroscope essentially as described previously (18). Pathology was scored on a scale of 0 to 4 as follows: 0, clear eye; 1, slight redness in the corners; 2, moderate redness and injection; 3, conjunctival and corneal injection with ciliary flush; 4, extensive injection, generally associated with some mucus. Eyes were examined in a coded fashion, with the reader unaware of the treatment given. Pulmonary pathology was enumerated as described before (6). In brief, lungs were perfused and fixed in 10% buffered formalin and embedded in paraffin. Multiple 4-μm-thick sections were stained with hematoxylin and eosin and examined by light microscopy. Inflammatory infiltrates were scored by enumerating the layers of inflammatory cells surrounding the vessels. Finding zero to three layers of inflammatory cells was considered normal. Finding more than three layers of inflammatory cells (surrounding 50% or more of the circumference of the vessel) was considered abnormal. The number of abnormal perivascular and peribronchial spaces divided by the total such spaces was the percentage reported as the pathology score.

Ocularly inoculated RSV infects the eye and the lung.

To first test whether RSV can infect the eye, we administered various amounts of purified RSV, ranging from 10⁴ to 10⁸ PFU, into the BALB/c mouse cornea as described in Materials and Methods. We then measured infectious progeny virus in the eye as well as in the lungs at different days afterward. While 10⁴ PFU produced no obvious infection, 10⁸ PFU resulted in severe ocular inflammation, mucus, conjunctivitis, and discomfort, suggestive of an excessive infection plane uncharacteristic of the human disease. With amounts between 10⁴ and 10⁸ PFU, a graded infection was seen. In this paper, we present results obtained using 2 × 10⁶ PFU RSV, which, as shown below, produced optimal lung infection, approximating that caused by nasally introduced RSV (6). Representative results show that infectious RSV titer peaked in the eye on day 2 and decreased subsequently (Fig. 1A). Infectious RSV was appreciable in the lung tissue as well. Pulmonary viral titer was measurable on day 3 after ocular inoculation, reached its peak on day 5, and then slowly subsided to the threshold of detection at around day 14 (Fig. 1C). In both tissues, the viral titer paralleled the viral protein level as measured by the amount of RSV N protein in immunoblots. Since robust protein expression by negative-strand RNA viruses requires de novo viral replication and since the input virus was undetectable on day 1, the results clearly indicate that virus replicated in both the eye and the lung tissues. We had previously shown that when introduced through the traditional nasal route, RSV attains the highest pulmonary growth on day 4 (6). Thus, it appears that the ocular route takes roughly an extra day (day 5) to establish maximal pulmonary growth. We speculate that this is due to two reasons: (i) the time taken for the virus to replicate in the eye, which amplifies the inoculum, and (ii) the travel time from eye to lung. As we show below, inhibition of viral replication in the eye by siRNA significantly reduces lung infection.

We then tested whether the ocular route produces ocular or pulmonary pathology. Examination of eye and lung tissue sections indeed revealed significant pathology in both as indicated by the pathology scores (Fig. 1A and C). In slit lamp examination, various degrees of teary, red eyes were observed, with minor ciliary flush and conjunctival injections. Some discharge was noticed in more severe cases at the peak of the inflammatory reaction. In representative examples, RSV proteins could also be detected in situ by indirect immunofluorescence of eye and lung sections (Fig. 1E). The chronology of the pathology scores generally paralleled the viral titer (Fig. 1A and C). Finally, the mice showed the typical outcomes of RSV disease (data not shown), such as rapid respiration, reduced activity, and weight loss (6). Clearly, once the virus reaches the lungs, its history and path become irrelevant; i.e., the nature and course of the pulmonary disease and subsequent clearance of the virus follow the same kinetics regardless of the initial tissue of administration. Together, these results establish that the eye can indeed act as a bona fide route of pulmonary infection in the mouse model.

Early intervention in ocular viral growth inhibits pulmonary infection.

In recent years RNAi has emerged as a key strategy to knock down gene expression (3, 19). We provided proof of concept that properly designed synthetic siRNA could act as a potent antiviral agent in cell culture as well as in the BALB/c mouse model (2, 4, 6). Once we found that pulmonary viral infection can occur through the eye, a logical question was whether this knowledge could be exploited for prevention by using RNAi in the eye. The optimal results, obtained with 1 nmol liposomal siRNA applied 30 min before RSV, revealed that the infectious viral titer was strongly inhibited (Fig. 1, compare panels A and B) and was below the threshold of detection of our assay, i.e., under 6 × 10² PFU per μg eye tissue. Small amounts of viral protein could be detected (Fig. 1B, day 1), some of which likely represent the input virus. Thus, the siRNA caused nearly complete inhibition of RSV replication in the eye.

Studies of the lungs in the same animals showed that ocularly applied siRNA also led to a significant reduction of pulmonary viral titer (Fig. 1, compare panels C and D). The animals also showed little discomfort or RSV disease as judged by the lack of the clinical correlates described above (data not shown). The small amount of viral growth in the lung (about 100-fold lower than in lungs not treated with siRNA) is explained by the travel of the ocular inoculum without amplification by replication in the eye. Evidently, this amount was too small to establish a significant viral load and pathology. In either case, we can conclude that ocular RNAi treatment can essentially prevent respiratory RSV infection through the eye.

Tissue distribution is a critical issue for a prospective drug because of its bearing on bioavailability and possible toxicity in various organs. Currently, two siRNA drugs are undergoing clinical trials for use in the eye to treat the wet form of age-related macular degeneration. Code-named sirna027 (Sirna Therapeutics) and cand5 (Acuity Pharmaceuticals), they target vascular epidermal growth factor and its receptor, respectively (3). However, no data are available regarding their transit outside the eye. To test the tissue distribution of ocularly applied siRNA, we used fluorescent Cy3-labeled siRNA and investigated its presence in four major organs (eye, lung, brain, and liver) at 12, 24, 36, and 48 h postinstillation. Large amounts of fluorescence were observed in the eye between 12 h and 36 h, which started to decline thereafter. In contrast, no fluorescence could be detected in all other organs tested at all times, and therefore, representative negative results are shown for lung only (Fig. 1F). These results suggest that ocularly applied siRNA does not leave the eye and thus should not cause systemic side effects. Importantly, its absence in the lung specifically suggests that the observed reduction in the lung viral titer was an indirect effect of inhibition of viral growth in the eye, further confirming the eye-to-lung travel.

In any therapy, knowledge of the “window of opportunity” is important. To this end, we wanted to know whether and to what extent the ocular siRNA will inhibit pulmonary infection if administered after RSV. In these studies, the anti-RSV siRNA formulation was added to the eye at different times before or after RSV inoculation. The results (Fig. 2) reveal that siRNA, added a day before or at about the same time as the virus, prevents virus migration to the lungs, confirming the results in Fig. 1. In contrast, siRNA added progressively later offered little or no protection. These results additionally confirm eye-to-lung travel of the virus and the preventive role of siRNA in the eye, but only if used promptly.

Activation of proinflammatory cytokines and chemokines in the RSV-infected eye.

Having shown that the eye can serve as a gateway to lung infection, the next query was whether the eye itself is affected by the virus. Pathogens that are known agents of ocular infection, such as HSV, often cause severely impaired vision and blindness (8). In contrast, there is no report of long-term vision loss or blindness in patients with RSV infection. Thus, we reasoned that the ocular effect, if any, would be short term and perhaps would be manifest as a cytokine-mediated inflammatory response. We carried out a rapid real-time PCR screening of the common cytokines, chemokines, and selected receptors in RSV-infected eye tissue over the full course of infection. These studies led to a number of interesting observations. First, genes activated by at least fourfold (Fig. 3A) were mainly those for CC and CXC chemokines and their receptors. The most strongly activated genes were those for MIG/CXCL9 (monokine induced by gamma interferon), RANTES/CCL5 (regulated upon activation, normal T-cell expressed and secreted), IP-10/CXCL10 (gamma interferon-inducible protein of 10 kDa), MCP-3/CCL7 and -5/CCL12 (monocyte chemotactic protein), and MIP-1α/CCL3 (macrophage inflammatory protein). We confirmed the activation of MIG, RANTES, IP-10, MCP-5, and MIP-1α by ELISA (data not shown); these results were essentially identical to the transcriptional data. Second, although no two genes had exactly identical kinetics of activation, their expression generally peaked between 4 and 8 days after instillation of the virus in the eye and then gradually subsided to basal or near-basal levels by day 14 (the longest time point at which samples were taken). The kinetics of activation (Fig. 3) trailed that of the virus load (Fig. 1) by 1 to 2 days, suggesting that the chemokine profile was a response to viral replication. Expression of a number of chemokine receptors was also up-regulated, suggesting further amplification of the signaling process. The protein levels of the three most highly activated chemokines, namely, MIG, RANTES, and IP-10, were found to be induced; the kinetics of induction were indeed very similar and were comparable to their transcriptional activation (Fig. 3B).

As mentioned earlier, both IL-1 and TNF play important roles in inflammatory responses of various tissues, including the eye. IL-1 is an early-warning cytokine that, together with TNF, activates corneal keratocyte growth factor and promotes neovascularization of the inflamed eye and apoptosis of the corneal stroma at low concentrations (21, 24, 45). Both are activated in essentially all ocular injuries and inflammatory conditions such as conjunctivalized cornea, uveitis, and recurrent herpes stromal keratitis (9, 21, 24, 33, 45). Another commonality is that IL-1β and TNF are activated and/or released by proteolytic processing of their precursors (26); IL-1α, in contrast, is activated transcriptionally. As transcriptional activation did not reveal any significant activation of these genes, we measured the proteins by ELISA and found that both IL-1α and TNF were significantly activated (Fig. 4A) but that IL-1β was not (data not shown).

Anticytokine therapy reduces pathology of the RSV-infected eye.

As revealed in Fig. 4A and B, the kinetics of the ocular chemokine response paralleled the ocular pathology, with maximal pathology discernible at around 3 days postinfection. To test the hypothesis that specific cytokine antagonists may provide relief of ocular inflammation in RSV infection, we treated the eye with neutralizing anti-TNF and anti-IL-1α antibodies along with RSV. Both antibodies were highly effective in reducing the ocular pathology (Fig. 4B) to about 50% or less of that in the untreated infected eye.

However, as the inflammatory reaction could be a manifestation of an antiviral response, we tested the effect of anti-inflammatory treatment on viral growth by using IL-1α as a representative target. When applied to the eye, anti-IL-1α did not inhibit viral replication. In fact, the vial titer was slightly higher not only in the eye (Fig. 4C) but in the lung as well (Fig. 4D). Thus, ocular anti-inflammatory treatment relieved inflammation but provided no inhibition of the actual process of infection.