Dendritic Cells Route Human Immunodeficiency Virus to Lymph Nodes after Vaginal or Intravenous Administration to Mice

Abstract

We have developed a murine model to study the involvement of dendritic cells (DC) in human immunodeficiency virus (HIV) routing from an inoculation site to the lymph nodes (LN). Murine bone marrow-derived DC migrate to the draining LN within 24 h after subcutaneous injection. After incubation of these cells with heat-inactivated (Hi) HIV type 1 (HIV-1), HIV RNA sequences were detected in the draining LN only. Upon injection of DC pulsed with infectious HIV, the virus recovered in the draining LN was still able to productively infect human T cells. After a vaginal challenge with Hi HIV-1, the virus could be detected in the iliac and sacral draining LN at 24 h after injection. After an intravenous challenge, the virus could be detected in peripheral LN as soon as 30 min after injection. The specific depletion of a myeloid-related LN DC population, previously shown to take up blood macromolecules and to translocate them into the LN, prevented HIV transport to LN. Together, our data demonstrate the critical role of DC for HIV routing to LN after either a vaginal or an intravenous challenge, which does not require their infection. Therefore, despite the fact that the mouse is not infectable by HIV, this small animal model might be useful to test preventive strategies against HIV.

Understanding the early events that lead to establishment of a chronic human immunodeficiency virus (HIV) infection is essential for deciphering HIV pathophysiology and for designing therapeutic and vaccination strategies against this virus. HIV infection occurs through two main routes: sexual transmission, and direct intravenous inoculation. The establishment of a chronic infection requires that the contaminating virus productively infects one of its target cells, which are CD4-expressing T lymphocytes (22, 55), monocytes (13, 44), and dendritic cells (DC) (8, 23, 25, 53). Recently, a series of arguments has pointed to a scenario for the establishment of primary HIV infection through sexual transmission which involves CD4⁺ T lymphocytes and DC (23, 36, 43, 49, 50). It is assumed that DC at the inoculation site are first infected by HIV and then migrate to the draining lymph nodes (LN), where they transmit the virus to T cells in paracortical zones. Intense replication ensues, followed by dissemination of the virus to the whole lymphoid system. This scenario is inferred from (i) the fact that mucosae contain Langerhans cells (LC), immature DC known to migrate from their resident area to the draining LN upon exposure to antigens or to inflammatory signals (17–19, 24, 31, 32, 34, 46); (ii) observations of the nature and location of simian immunodeficiency virus (SIV)-infected cells at different time points after an experimental SIV vaginal inoculation in macaques (50); and (iii) in vitro observations in which conjugates of DC and T cells represent the optimal milieu for a productive HIV infection, because DC transmit HIV to T cells together with a vigorous activation signal that permits efficient infection and replication (9, 42). Although this scenario appears likely, the role of DC has not been definitively demonstrated. As pointed out previously (50), the first infected cells are seen in the lamina propria after an SIV vaginal inoculation, and it has not yet been explained how the virus crosses the multilayered epithelium that usually acts as a physical barrier to infectious agents. It is also not known how HIV reaches the LN in the case of an intravenous (i.v.) inoculation.

We aimed to analyze in more detail the role of DC in the early events of HIV infection. Because studies using primates and infectious virus are difficult to carry out, we investigated the possibility of using a mouse model. Indeed, although murine cells are not infectable by HIV, mice may serve as a good model to study HIV routing by DC. First, murine DC are well characterized and can be easily traced with specific antibodies (51). Second, after staining and reinjection, migration of these cells can be easily monitored (3, 16, 21, 26, 37), and there are solid experimental data on the circulation pathways of murine DC (2). Finally, it has previously been reported that murine DC, although not infectable by HIV, can transmit HIV to human T cells as efficiently as human DC (9). This indicates that DC can transmit to T lymphocytes HIV particles that they have not themselves produced but rather that they have adsorbed or internalized.

We show here that infectious HIV can indeed be transported by murine DC to the draining LN upon inoculation at the periphery and that DC are also essential for carrying HIV to LN upon i.v. administration. These results indicate that DC play a major role in the establishment of HIV infection which does not necessarily require their productive infection.

MATERIALS AND METHODS

Animals.

DBA/2 and (C57BL₆ × CBA/J)F₁ mice were purchased from IFFA-Credo (L’Arbresle, France). Human CD4 (hCD4)-transgenic mice derived from (C57BL₆ × CBA/J)F₁ mice were obtained from P. Lorès (29). Mice were 7 to 14 weeks of age at the time of the experiments. Thymidine kinase (TK)-chimeric mice were derived from LTR-TK-transgenic mice expressing the herpes simplex virus type 1 TK (HSV-1 TK) in DC as previously described (47, 48).

Preparation of DC.

DC were obtained from bone marrow (BM) of DBA/2 or hCD4-transgenic mice as described previously (20), with slight modifications. Briefly, BM cells were collected from femurs and tibias, erythrocytes were lysed with ammonium chloride, and immunomagnetic depletion of lymphocytes and Ia-positive cells was performed with a mixture of hybridoma supernatants of anti-CD4 (GK-1.5; ATCC TIB207), anti-CD8 (H35 JP), anti-major histocompatibility complex (MHC) class II (M5/114; ATCC TIB120), and anti-B220 (RA3-6B2) monoclonal antibodies (MAb) and sheep anti-rat immunoglobulin G (IgG)-coated Dynabeads (M-450; Dynal, Oslo, Norway). After the depletion steps, cells were cultured at 7 × 10⁵ cells/ml in RPMI 1640 supplemented with 5% decomplemented fetal calf serum (FCS), 20 μg of gentamicin per ml, 50 μM 2-β-mercaptoethanol, and 2,000 U of recombinant granulocyte-macrophage colony-stimulating factor (GM-CSF) (Genzyme, Cambridge, Mass.) per ml in 24-well plates at 37°C. On days 2 and 4, 90% of the medium was replaced by GM-CSF-containing medium. On day 6, the medium was harvested, and loosely adherent aggregates of growing DC were dislodged by vigorous pipetting and then counted (47). DC were subcultured at 2 × 10⁵ to 3 × 10⁵ cells/ml in fresh GM-CSF-containing medium in six-well plates overnight and then harvested and counted.

LN were cut into small fragments and incubated in RPMI 1640 supplemented with 1.6 mg of collagenase (type IV; Sigma Chemical Co., Saint Quentin Fallavier, France) per ml and 200 μg of DNase I (Boehringer Mannheim, Mannheim, Germany) per ml at 37°C for 30 min. Cells were dissociated by repeated pipetting, reincubated at 37°C for 10 min, and washed. Cell suspensions were then incubated with 200 μg of DNase I per ml for 15 min at room temperature and resuspended in staining buffer (phosphate-buffered saline–3% FCS–0.02% azide) for flow cytometric analyses.

The BM-derived and LN DC populations were evaluated by flow cytometry analysis for the expression of CD11c and MHC class II molecules, classical markers of murine DC (35, 46, 54). For this purpose, 5 × 10⁵ BM-derived cells and 1 × 10⁶ LN cells were centrifuged at 350 × g for 5 min and resuspended at 10⁷ cells per ml in buffer, and single or double labeling with saturating concentrations of Ab was performed. CD11c expression was analyzed with unlabeled MAb N418 (hamster Ig; ATCC HB224) revealed by a phycoerythrin-conjugated F(ab′)₂ goat anti-hamster IgG (Caltag Laboratories, San Francisco, Calif.). MHC class II expression was analyzed with MAb M5/114 (rat Ig; ATCC TIB120) revealed by a fluorescein isothiocyanate (FITC)-conjugated F(ab′)₂ goat anti-rat IgG (Caltag Laboratories) or with MAb 14.4.4S (PharMingen, San Diego, Calif.), either FITC conjugated or biotin conjugated, revealed by streptavidin Tri-color (Caltag Laboratories). After the final wash, cells were fixed in 1% paraformaldehyde and analyzed on a FACScan flow cytometer (Becton Dickinson).

In some experiments, MHC class II and CD11c double-stained DC were sorted with a FACStar Plus (Becton Dickinson) and kept at 4°C throughout the procedure. Excitation was done at 488 nm with a 5-W argon laser (Coherent, Palo Alto, Calif.) operating at 150 mW. Cells were sorted at a rate of 3,000 events/s, with the abort rate being about 10% of the events.

Cell line.

The CD4⁺ human lymphoid cell line HUT-78 was purchased from the American Type Culture Collection (Rockville, Md.) and maintained in RPMI 1640 supplemented with 10% decomplemented FCS (Dutscher, Brumath, France).

HIV strains.

The lymphotropic HIV type 1 (HIV-1) strain HIV-1_LAI (52) and the macrophage-tropic HIV-1_Ba-L were obtained from Diagnostics Pasteur (Marne-la-Coquette, France) and from B. Asjö, respectively. The HIV-1_LAI titer, determined by infection of activated peripheral blood mononuclear cells (PBMC) with viral supernatant, was 10⁴ 50% tissue culture infective doses (TCID₅₀)/ml. The concentration of p24 antigen in the HIV-1_LAI viral supernatant was 5 μg/ml.

Live virus supernatant was used either for subcutaneous (s.c.) injection (25 μl = 250 TCID₅₀) or for incubation experiments (100 to 300 μl [3 × 10⁶ to 10 × 10⁶ cells/ml] = 1,000 to 3,000 TCID₅₀).

Other experiments were performed after heat inactivation (Hi) of the virus at 56°C for 1 h. In this case, because the virus inoculum cannot be evaluated as TCID₅₀ per milliliter, viral amounts are given as the volume of either pure or diluted supernatant.

HIV-1_Ba-L, with a titer of 10⁵ TCID₅₀/ml determined by infection of activated PBMC with viral supernatant, was used after similar Hi.

In vivo tracing of DC. (i) s.c. injection of BM-derived DC.

The migration pattern of BM-derived DC was analyzed by injecting cells stained with PKH2 (Zynaxis, Malvern, Pa.), a nontoxic fluorochrome previously used in migration studies of murine antigen-presenting cells (37). BM-derived DC suspended at a final concentration of 5 × 10⁶ to 10 × 10⁶ cells/ml in a diluent (Zynaxis) were incubated for 5 min at room temperature with 2 μM (final concentration) PKH2. Two volumes of RPMI 1640 with 10% fetal calf serum was added to stop the reaction, and then the cells were washed three times.

PKH2-stained and control cells were injected s.c. at 5 × 10⁵ cells/25 μl into the footpads of anesthetized (2.5% tribromoethanol) DBA/2 mice. At 6 and 24 h after injection, ipsilateral and contralateral popliteal and inguinal LN were harvested, frozen in OCT medium (Labonord, Villeneuve d’Ascq, France), sectioned (5 μm) on a cryostat, and viewed under a fluorescence microscope.

(ii) FITC application on the vaginal mucosa.

The migration of resident LC present in the vaginal mucosa was investigated by using a previously published technique utilized for studying the migration of epidermal LC (31, 46). Twenty-five to 30 μl of 0.8% FITC (Isomer 1; Sigma Chemical Co.) dissolved in phosphate-buffered saline was applied atraumatically by inserting a 22-gauge 1 1/2 animal feeding cannula (Polylabo, Strasbourg, France) onto the vaginal vaults of anesthetized mice. The animals were maintained immobilized for 30 to 45 min. Vaginal mucosae and sacral, iliac, lumbar, and inguinal LN were harvested 12 and 24 h later for flow cytometry analysis.

Coculture experiments.

DC (3 × 10⁶ to 10 × 10⁶) were incubated with 100 μl of either infectious or Hi HIV-1_LAI supernatant in a final volume of 1 ml of complete medium for 2 h at 37°C and washed twice in RPMI, and then 5 × 10⁴ DC were cocultured with 5 × 10⁵ HUT-78 cells in RPMI 1640–10% FCS for 10 days. Controls were unpulsed DC.

Similarly, HIV-pulsed DC were cocultured with 5 × 10⁵ T cells isolated from PBMC by rosetting in the presence of 100 ng of staphylococcal enterotoxin A (SEA) per ml in RPMI 1640 supplemented with 10% AB+ normal human serum, glutamine (2 mM), and 1% antibiotics (GIBCO-BRL, Paisley, Scotland) for 10 days. Controls were pulsed T cells or unpulsed DC cocultured with unpulsed T cells.

Cells obtained after mechanical dilaceration of LN were cocultured with 5 × 10⁴ and 1 × 10⁵ HUT-78 cells for popliteal and inguinal nodes, respectively, in RPMI 1640–10% FCS with glutamine and antibiotics. Cells were expanded for 10 days and then maintained at 10⁵ cells by biweekly medium changes up to 80 days.

The kinetics of viral p24 production in the different culture supernatants was determined by enzyme-linked immunosorbent assay (ELISA) (Diagnostics Pasteur).

In vivo HIV-1 experiments. (i) s.c. injection.

In a first set of experiments, infectious or Hi HIV_LAI supernatant (25 μl/mouse) was injected s.c. into the footpads of anesthetized DBA/2 mice. At 24 h after injection, ipsilateral, contralateral popliteal, and inguinal LN were removed and frozen for RNA extraction.

In another set of experiments, 3 × 10⁶ to 10 × 10⁶ BM-derived DC were first incubated with 300 μl of either infectious or Hi HIV_LAI or HIV_Ba-L supernatant in a final volume of 1 ml of complete medium for 2 h at 37°C. After two washes, 5 × 10⁵ cells/25 μl were injected s.c. into the footpads of anesthetized mice. At 24 h after injection, ipsilateral, contralateral popliteal, and inguinal LN were removed and frozen for RNA extraction. The amount of virus bound to BM-derived DC before their injection was determined by p24 ELISA.

(ii) Application on the vaginal mucosa.

Twenty-five microliters of Hi HIV-1_LAI supernatant was administered in the vaginal vaults of anesthetized mice as described above for FITC. Twenty-four hours later, vaginal mucosae and sacral, iliac, lumbar, and inguinal LN were harvested and frozen for RNA extraction.

(iii) i.v. administration.

One hundred fifty microliters of Hi HIV-1_LAI supernatant was administered i.v. into the retroorbital sinuses of anesthetized mice. Thirty minutes later, peripheral LN (axial, brachial, and inguinal) were harvested separately and then mixed into two pools corresponding to the left and right LN of mice. The kidney was also harvested as a control. All of the samples were frozen for RNA extraction.

RT-PCR.

Control and HIV-pulsed cells were lysed in RNA extraction solution (RNA-BTM; Bioprobe, Montreuil, France). Frozen LN and kidney were teased with a scalpel in the same solution.

For the reverse transcription-PCR (RT-PCR), 1 μg of total cellular RNA or half total cellular RNA of LN was reverse transcribed as follows. In a 20-μl reaction mixture, 1 mM each deoxynucleoside triphosphate (Pharmacia LKB Biotechnology), 0.04 U of random primer P(dN)6 (Pharmacia LKB Biotechnology), 40 U of RNase inhibitor (Pharmacia LKB Biotechnology), 10 mM dithiothreitol, and 200 U of Moloney murine leukemia virus reverse transcriptase (Gibco BRL, Gaithersburg, Md.) were used. RT reactions were performed at 42°C for 1 h. A 5-μl aliquot of the RT mixture was used directly for each amplification reaction. PCR was performed as follows. In a 50-μl reaction mixture, 0.5 μM primers SK38 and SK39 for HIV-1 gag sequences (Perkin-Elmer Cetus, Emeryville, Calif.) or 1 μM hCD4 primers (Eurogentec, Angers, France), 200 μM each deoxynucleoside triphosphate (Pharmacia LKB Biotechnology), 1.5 mM MgCl₂, and 1 U of Taq DNA polymerase (Goldstar DNA polymerase; Eurogentec, Seraing, Belgium) with 1× buffer were used. The hCD4 primers were as follows: sense primer, 5′-ATGAACCGGGGAGTCCCTTTT-3′; antisense primer, 5′-ACTATCCTGGAGCTCCAGCT-3′. Reaction mixtures were incubated in a Crocodile II DNA thermal cycler (Appligene, Illkirch, France). For the HIV-1 gag primers, after 5 min at 94°C, 35 cycles of denaturation for 60 s at 94°C, annealing for 60 s at 58°C, and extension for 60 s at 72°C were performed, followed by an extension of 10 min at 72°C at the completion of the run. For the hCD4 primers, after 5 min at 94°C, 35 cycles of denaturation for 30 s at 94°C, annealing for 30 s at 64°C, and extension for 60 s at 72°C were performed, followed by an extension of 10 min at 70°C. A second amplification was performed under the same conditions with 5 μl of 1:200 diluted PCR products as the template. The products of PCR were analyzed by Southern blot hybridization with the HIV-1 SK19 probe (Perkin-Elmer Cetus). For quantitative RT-PCR, the test Amplicor HIV-1 Monitor (Roche Diagnostic Systems, Neuilly-sur-Seine, France) was used with some modifications: this quantification was performed not with plasma but with 2 μg of RNA extracted from LN with RNA extraction solution (RNA-BTM; Bioprobe) and treated with RNase-free DNase (Boehringer Mannheim, Meylan, France).

RESULTS

Murine BM-derived DC efficiently transmit HIV-1 infection to human T cells.

Murine DC can routinely be generated from BM cells cultured in the presence of GM-CSF for 7 days (20). Under such culture conditions, 40 to 70% of the cells were DC as characterized by their morphology and phenotype, i.e., expression of the CD11c marker and of high levels of MHC class II molecules (20, 30, 47).

We verified that these BM-derived DC could efficiently transmit productive HIV infection to human T cells as previously reported for murine splenic DC (9). After incubation with either Hi or infectious HIV-1_LAI, DC were cocultured with HIV-susceptible human HUT-78 cells. The cells became productively infected (Fig. 1A), demonstrating that the virus bound to murine BM-derived DC remained infectious. Similarly, DC could transmit an HIV infection to human resting T lymphocytes when these cells were cocultured in the presence of a superantigen (Fig. 1B). In contrast, resting T lymphocytes incubated with HIV-1 and a superantigen in the absence of DC did not replicate HIV-1.

FIG. 1.

Murine BM-derived DC transmit a productive HIV infection to human T-lymphoblastoid cells and to human T lymphocytes. The kinetics of HIV replication, measured by ELISA for p24 production, in cocultures of BM-derived DC with either HUT-78 cells (A) or human T lymphocytes (B) was determined. (A) HUT-78 cells were cocultured with Hi or infectious HIV-1_LAI-pulsed DC or with unpulsed DC as a control. (B) T lymphocytes were cocultured with HIV-1_LAI-pulsed DC or T lymphocytes in the presence of SEA. The control was unpulsed DC with T lymphocytes in the presence of SEA.

Murine BM-derived DC migrate to the draining LN after s.c. injection.

Peripheral DC are known to migrate to the draining LN upon antigen stimulation. BM-derived DC can similarly migrate to the draining LN after s.c. injection. Fluorescent BM-derived DC labeled with the PKH2 tracer were already found in the draining ipsilateral popliteal LN 6 h after injection, and their number increased at 24 h (Fig. 2A). No fluorescent cells could be detected in the ipsilateral inguinal or the contralateral popliteal and inguinal LN. Thus, BM-derived DC generated in vitro retain their ability to migrate from an s.c. injection site to the draining LN.

FIG. 2.

Murine BM-derived DC route HIV to the draining LN after s.c. injection. (A) Section of a draining popliteal LN observed under a fluorescence microscope 24 h after injection of murine BM-derived, PKH2-stained DC into the footpads of DBA/2 mice. Note the presence of numerous fluorescent cells. Magnification, ×200. (B) Detection of HIV (left) and hCD4 (right) RNAs by RT-PCR in LN of nontransgenic mice, 24 h after injection of 25 μl containing 5 × 10⁵ Hi HIV-1_LAI-pulsed DC derived from hCD4-transgenic mice into the footpad. Pop, popliteal LN; Ing, inguinal LN.

Murine BM-derived DC transport HIV to draining LN after s.c. injection.

We then repeated these experiments with DC incubated with the lymphotropic Hi HIV-1_LAI. At 24 h after s.c. injection of Hi HIV-1_LAI-pulsed DC into the footpad, HIV gag RNA sequence was detected in the ipsilateral popliteal nodes of three of five mice, but it was not found in the ipsilateral inguinal or in the contralateral popliteal and inguinal LN (Table 1). In experiments performed with the macrophage-tropic Hi HIV-1_Ba-L, HIV gag RNA sequence was also found only in the ipsilateral popliteal LN (data not shown).

TABLE 1.

Detection of HIV gag RNA in draining LN 24 h after s.c. injection of Hi HIV_LAI-pulsed DC or Hi HIV_LAI supernatant

Treatment	No. of positive micea/no. of tested mice
	gag				hCD4
	Homolateral		Contralateral		Homolateral		Contralateral
	Popb	Ingb	Pop	Ing	Pop	Ing	Pop	Ing
Unsorted HIV-pulsed DC from normal mice	3/5	0/5	0/5	0/5	NAc	NA	NA	NA
HIV-pulsed DC from hCD4-transgenic mice
Unsorted	3/7	0/7	0/7	0/7	6/7	0/7	0/7	0/7
Sorted	2/3	0/3	0/3	0/3	1/3	1/3	0/3	0/3
Total	8/15	0/15	0/15	0/15	7/10	1/10	0/10	0/10
HIV-1 supernatant
Undiluted	6/6	NDd	ND	ND	NA	NA	NA	NA
Diluted 1:300	0/7	ND	ND	ND	NA	NA	NA	NA

^aNumber positive for gag or hCD4 RNA by RT-PCR. 

^bPop, popliteal LN; Ing, inguinal LN. 

^cNA, not applicable. 

^dND, not done. 

Similar experiments were also conducted with DC generated from the BM of transgenic mice expressing hCD4. These mice were used with the idea that hCD4 expression on the cell surface might favor the binding of HIV, although it cannot render these cells HIV infectable (29). In addition, the transgene may serve as a congenic marker to trace the cells after injection into nontransgenic mice. After injection of hCD4 DC previously incubated with Hi HIV-1_LAI, both HIV gag and hCD4 RNAs were detected in the ipsilateral popliteal LN of three of the seven injected mice (Table 1; Fig. 2B). Under similar experimental conditions, no viral RNA could be detected after injection of purified murine LN T cells incubated with Hi HIV_LAI (data not shown).

These results were also confirmed by using purified DC obtained by flow cytometry cell sorting. Under these conditions, two of three mice were positive for HIV gag RNA in the ipsilateral popliteal LN, and one of these mice was also positive for hCD4 RNA (Table 1). Together, these findings indicate that murine BM-derived DC incubated in vitro with HIV-1 can route the virus to draining LN.

To further verify the role of DC in HIV-1 transport to LN, similar experiments were performed by injecting Hi HIV-1_LAI supernatant alone at different concentrations into the footpads of mice and analyzing ipsilateral and contralateral popliteal and inguinal LN 24 h later for the presence of HIV RNA. Table 1 shows that HIV gag RNA signal was detected in the popliteal draining LN of all six mice injected with a highly concentrated Hi HIV-1_LAI supernatant. By contrast, no signal was detected in the draining LN of all seven mice injected with a 1:300 dilution of this Hi HIV-1_LAI supernatant. This dilution was assayed because it corresponds to the amount of virus bound to DC after incubation with Hi HIV-1_LAI supernatant and two washes, as determined by p24 ELISA. This demonstrates the essential role of DC, which appear to be capable of routing very small amounts of virus to LN.

HIV is still infectious after transport to LN.

Similar experiments with infectious virus allowed us to evaluate whether it remained infectious upon arrival in the LN. After injection of infectious HIV-1_LAI, draining popliteal LN cells were harvested and cocultured with HIV-susceptible human HUT-78 cells. A productive HIV infection of these cells was detected in one of four mice injected with HIV-1_LAI-pulsed DC and in one of four mice injected with 25 μl of highly concentrated HIV-1_LAI supernatant (Fig. 3). Thus, HIV can remain infectious 24 h after transport from its injection site to the draining LN in mice.

FIG. 3.

HIV remains infectious after its routing to the draining LN by DC. Twenty-five microliters of HIV-1_LAI supernatant (squares) or 25 μl containing 5 × 10⁵ HIV-1_LAI-pulsed DC (triangles) was injected into the footpads of normal mice. Twenty-four hours later, LN were harvested and the cells were cocultured with HUT-78 cells. The kinetics of HIV replication was measured by ELISA for p24 production.

HIV is transported to draining LN after vaginal application.

Previous reports showed that LC of the vaginal mucosa can pick up antigens and migrate to the draining LN (39, 41). In agreement with these studies, we observed FITC-positive cells only in draining sacral and/or iliac LN 24 h after FITC was applied on the vaginal mucosa. No positive cells were observed in inguinal LN (Fig. 4A). The stained cells were detected within an LN DC subset (l-DC) (Fig. 5) previously shown to derive from LC (46).

FIG. 4.

HIV is routed to draining LN after vaginal application. (A) Fluorescence-activated cell sorter analysis of LN cells 24 h after FITC application on the mouse vaginal mucosa. FITC-fluorescent cells could be detected almost exclusively within a subpopulation of LN DC (l-DC) characterized by expression of CD11c and MHC class II molecules, as previously described (46) (also see Fig. 5). Data are presented as FITC fluorescence histograms for l-DC from an iliac LN of a noninjected mouse (left panel) and from iliac and inguinal LN of an FITC-inoculated mouse (middle and right panels, respectively). Results are from the analysis of 2 × 10⁵ to 3 × 10⁵ LN cells. FITC staining is detected in about 30% of the l-DC-draining iliac LN. Results from one representative experiment of four independent experiments with two or three mice per group are shown. (B) Detection of HIV RNA by RT-PCR in the vaginal mucosae and inguinal, iliac, and sacral LN of mice, 24 h after inoculation of 25 μl of Hi HIV-1_LAI supernatant into the vaginal vault.

FIG. 5.

Specific abrogation of s-DC in transgenic mice. Fluorescence-activated cell sorter analysis of inguinal LN of control and DC-depleted mice is shown. s-DC depletion was obtained by a 7-day GCV treatment of transgenic mice expressing the suicide gene for HSV-1 TK preferentially in DC (47, 48). Analysis of s-DC and l-DC subpopulations based on expression of the CD11c marker and MHC class II molecules was performed as previously described (46).

We then investigated whether resident LC of the vaginal mucosa may route HIV-1 to draining LN after vaginal challenge. HIV-1_LAI was detected 24 h after vaginal challenge in the draining LN of two of four mice but was not detected in any of the inguinal and lumbar nodes (Fig. 4B). Together, these results indicate that LC from the vaginal mucosa can route HIV to their draining LN.

DC participate in the transport of HIV to LN after i.v. inoculation.

We recently showed that a myeloid-related DC subset of peripheral LN (s-DC) (Fig. 5) has the ability to take up blood macromolecules and transport them to LN (46). We thus aimed to analyze if they could also route viral particles from the blood compartment to LN.

We injected Hi HIV-1_LAI supernatant i.v. into normal mice, and 30 min later, we studied HIV gag RNA in pools of peripheral LN, as well as in the kidney, by RT-PCR. Because in this case the viral inoculum would be immediately diluted within the blood compartment, we injected a larger volume of viral supernatant (150 μl) than in the experiments described above. The HIV gag RNA signal was strong in six of eight and weak in two of eight pools of peripheral LN (Table 2). In all cases, the kidney, which was chosen as a positive control tissue because of its high blood content, was also found to be positive for HIV gag RNA. Although this is unlikely, the positive LN signals could have been due to blood contamination. To rule out this hypothesis, we repeated these experiments using mice specifically depleted of s-DC. We used our model of transgenic mice preferentially expressing an HSV-1 TK gene in DC (1, 27, 28, 46–48). Ganciclovir (GCV), which is specifically metabolized into a toxic analog in HSV-1 TK-expressing cells, kills dividing HSV-1 TK-expressing DC. As shown in Fig. 5, a 7-day GCV treatment dramatically affected s-DC, which was reduced approximately 10-fold (from 0.2 to 0.4% of total LN cells in control mice to 0.03 to 0.05% in DC-depleted mice) (46). By contrast, this treatment did not affect l-DC, in accordance with their turnover, and the overall LN architecture was not affected (27a). To ascertain the role of s-DC in Hi HIV_LAI routing to LN, we performed i.v. injections of Hi HIV_LAI in s-DC-depleted mice. As shown in Table 2, no HIV gag RNA was detected in five of eight LN cell pools, while a weak signal was detected in two of eight pools and a strong signal was detected in one of eight pools. These results were confirmed by using a quantitative RT-PCR kit used for clinical HIV-1 detection, with RT-PCR performed on the total peripheral LN cells from control and DC-depleted mice (Table 2). Similar results were obtained in a second, independent experiment.

TABLE 2.

Detection of HIV gag RNA in peripheral LN of control and DC-depleted mice 30 min after i.v. injection of 150 μl of Hi HIV_LAI supernatant

Mice	gag RT-PCRa		Quantitative RT-PCR (HIV RNA copies/10 μg of total RNA)b
	Pool L	Pool R
Control	+	+	461
	+	+/−	1,523
	+	+	762
	+/−	+	299
DC depleted	+	−	Undetectable
	+/−	−	Undetectable
	−	+/−	Undetectable
	−	−	Undetectable

^aLeft and right axial, brachial, and inguinal LN were harvested independently and mixed into two pools (L and R) before RT-PCR analysis. +, strong signal; +/−, weak signal; −, no signal. 

^bQuantitative RT-PCR was performed on total LN (pool L plus pool R) of each mouse. 

Together, these results indicate that DC can take up HIV-1 in the blood and transport it to LN.

DISCUSSION

Intravaginal inoculation of SIV in macaques (50) revealed a role for DC in the initiation of a primary infection. These experiments clearly showed that after an initial viral challenge, the virus is rapidly (in less than 2 days) detectable in the draining LN. On day 2, the virus is already disseminated, as illustrated by its detection in the spleen and nondraining LN (50). Thus, the kinetics and the localization of these events correlate well with the known DC migration pattern and physiological interaction with T cells. By inference, it was therefore assumed that DC at the inoculation site became HIV infected, migrated to the LN, and transmitted HIV infection to local T cells that thereafter became the seeding point for viral dissemination. However, these experiments also raised questions regarding the precise role of DC. First, productively SIV-infected cells were detected only in the lamina propria of the cervicovaginal mucosa and not in the epithelium. Thus, how did the virus penetrate into the lamina propria, and are other cells involved in this initial step of SIV infection? Second, these infected cells have not be formally identified as DC. Third, SIV could be detected in the vagina-draining LN in only one of the four animals studied. Therefore, it remains to be formally demonstrated if and which DC are the initially infected cells after sexual exposure to HIV and are responsible for its systemic dissemination. However, these experiments are difficult to carry out and expensive with large animals and infectious virus. This led us to investigate the possibility of using mice to analyze some of these events in more detail.

Although murine cells cannot be infected by HIV, the mouse model seemed appropriate to address some of these questions, in part because the migration patterns of murine BM-derived DC and chimpanzee DC are similar (5). Indeed, our results indicate that even in the absence of a productive infection, murine DC are able to transport HIV, regardless of the site of administration, from the periphery to the draining LN.

Several points should be emphasized. First, the role of DC in HIV routing after s.c. injection was demonstrated with DC purified by cell sorting. Second, the fact that the HIV-1 RNA was not detected in 100% of injected mice is most probably due to the sensitivity of the detection technique, which might be improved in the future. Furthermore, it should be pointed out that we never detected any viral signal in a single nondraining LN throughout the 10 independent experiments performed. This is also true for the experiments carried out with infectious HIV-1_LAI, which could be found in the draining LN of two of eight animals tested. The detection, even on a single occasion, of live infectious HIV-1 in the LN demonstrates that it can be transported from the periphery to the LN while retaining its infectivity. Third, HIV transport by DC was shown for both lymphotropic and macrophage-tropic strains; the latter are presumed to be the predominant strains mediating primary infection in humans (4, 14). Fourth, our data indicate that DC could play a major role in HIV-1 dissemination without being productively infected. This is in agreement with previous in vitro observations that also suggested a role for DC in HIV transmission to T cells in the absence of their infection (9, 11). Finally, our results provide a possible explanation for the absence of infected DC in the superficial vaginal mucosae of the SIV-challenged macaques. We suggest that in primates susceptible to HIV or SIV infection, these viruses can be taken up by resident mucosal LC that would carry them to LN without being infected.

It is noteworthy that murine DC can bind and transport HIV-1 without abrogating virus infectivity. Interestingly, in vitro experiments showed that HIV incubated with DC could escape trypsin treatment and remain infectious for T cells (6, 10). It is thus possible that DC, either human or murine, bind and retain the virus on the cell membrane, where it could be sequestered within compartments formed by cell surface dendritic processes (6). Alternatively, the virus might be internalized into cytoplasmic vesicles.

It is noteworthy that the binding of HIV on murine DC did not require the hCD4 receptor. In addition, its presence did not significantly influence HIV transport by DC, as shown by results obtained with transgenic hCD4⁺ murine DC. Therefore, the HIV binding on DC must involve other mechanisms, possibly related to the glycosylation of the HIV envelope proteins and/or the existence of mannose receptors on DC (45). It has recently been shown that HIV-1 is able to bind to LC by a CD4-independent mechanism (6).

While HIV-1 transport from the mucosae to the LN involves the LC or DC, we show for the first time that HIV-1 transport from blood to LN appears to involve another recently characterized DC subset. These cells have an immature phenotype and appear to be myeloid related (46). Their unique capacity to take up blood molecules and to carry them to LN suggests that they play a major role in the immune response to blood antigens and microorganisms. In this case, it appears that this property is subverted by HIV to facilitate virus dissemination and the establishment of a chronic infection.

Together, our data strongly suggest that noninfected DC mediate the initial steps of HIV-1 infection after exposure through sexual transmission or direct i.v. inoculation. Furthermore, depending on the site of inoculation, different DC subsets are involved in HIV-1 uptake: local LC-derived DC (l-DC) upon sexual transmission or blood DC (s-DC) after i.v. injection. In both situations, the virus would be carried to LN, where extensive viral replication would occur upon its transmission to CD4⁺ T cells, in accordance with several reports (12, 15, 38). The resulting viral production would then lead to HIV dissemination throughout the lymphatic system.

We have developed a murine model to study the involvement of DC in HIV-1 routing. This model uses a small, widely available, and inexpensive animal. Moreover, the analysis of HIV-1 transport to LN does not require infectious HIV-1, thus dramatically simplifying the experimental conditions. In addition, it should now be possible to analyze HIV transmission from DC to LN T cells in a murine model. Indeed, T lymphocytes from double-transgenic mice expressing the hCD4 receptor and CCR5 coreceptor are susceptible to HIV infection (7). Therefore, murine models might be useful to test different factors influencing HIV-1 transmission (such as hormonal impregnation) (33, 40), to screen molecules that upon mucosal application could prevent HIV-1 transmission, and to test vaccines for their ability to trigger a mucosal immunity to attack HIV-1 at this early step of infection.