Fms-like tyrosine kinase 3
modified vaccinia virus Ankara Bavarian Nordic
The cooperation between IFN-α/β and FL, the ligand of Fms-like tyrosine kinase 3 (Flt3), plays an important role in the defense against herpes simplex virus type 1 (HSV-1) in neonates. Treatment of neonatal mice with recombinant IFN-α has a short-term, FL-independent and a long-term, FL-dependent protective effect against HSV-1. In mice lacking FL, neonatal resistance against HSV-1 is very low and DC numbers in the spleen are reduced. The treatment of these mice with rIFN-α at day 6 resulted in an increased resistance against infection with HSV-1 at day 7. In C57BL/6 mice, treatment with rIFN-α at birth induced both FL and plasmacytoid DC (pDC), resulting in enhanced resistance against HSV-1 at day 7. In contrast, in mice lacking FL, IFN-α treatment at birth did not influence the splenic cell composition and had no effect on viral protection. The transfer of pDC to mice lacking FL enhanced viral resistance. Therefore, the induction and function of pDC, normally controlled by IFN-α/β and FL, are decisive for viral resistance in neonatal mice.
Neonates are highly susceptible to viral infections due to qualitative and quantitative differences of the immune system when compared with adults. This is due to low IFN-α/β production, low numbers of dendritic cells (DC) and T cells and immature lymph node structures 1–6.
An increase in resistance to viral infection can be achieved by treatment of neonatal mice with the hematopoietic growth factor FL, the ligand of Fms-like tyrosine kinase 3 (Flt3). FL treatment augments the number of conventional DC (cDC) 7 and plasmacytoid DC (pDC) in different organs 4. In FL-treated neonates, the cDC have an improved capacity of antigen presentation, and pDC, the ‘natural IFN-producing cells’, produced higher amounts of IFN-α than the respective cells from control animals 7.
An increase in neonatal viral resistance can also be achieved by virus-based vaccines to protect against homologous as well as heterologous virus 5, 8. To be effective, the vaccine virus needs to translocate its genetic information into the cell, followed by viral gene transcription and translation. However, replication of the vaccine virus is not necessary 5, 8. The modified vaccinia virus Ankara Bavarian Nordic (MVA-BN) is a vaccine virus able to infect cells but unable to replicate in mammalian cells. Treatment of newborn mice with MVA-BN increases the resistance against infection with the heterologous virus herpes simplex virus type 1 (HSV-1), 1 wk later. Protection was associated with increased levels of FL in serum and increased numbers of pDC in the spleen 5.
It is clear from studies using mice lacking type I IFN receptors that type I IFN plays a central role in the protective immune response 9, 10. IFN-α/β is immediately induced after viral attachment, recognition by Toll-like receptors, or after infection and replication in cells 11, 12. IFN-α/β induces at least a few hundred genes and orchestrates the defense 13–15via direct inhibition of viral replication 16, 17, the activation of NK cells 18, macrophages 19 and DC 20–23, and the enhancement of T cell stimulation 24, 25.
In this work, we have utilized mice deficient in FL (FL–/– mice) to delineate the role of IFN-α and FL in the defense against HSV-1 in neonatal mice. Our data show that there is an FL-independent, short-term and an FL-dependent, long-term protective effect of IFN-α. During the long-term effect, IFN-α induces FL, which is able to increase the number of pDC. Although protection against HSV-1 is not solely dependent on pDC, they play a critical role in the defense against HSV-1 in neonatal mice.
Neonatal mice lacking FL expression are highly susceptible to HSV-1 infection and are deficient in splenic DC populations
We have previously shown that treatment with FL or MVA-BN enhanced resistance to viral infections in neonatal C57BL/6 mice 4, 5. Likewise, treatment of C57BL/6 mice with recombinant (r)IFN-α at day 6 of age increased resistance against HSV-1 infection at day 7 4. The lethal dose (LD)50 of untreated, neonatal C57BL/6 mice is 103 PFU of HSV-1 4. This was increased tenfold to 104 PFU of HSV-1 after rIFN-α treatment (Fig. 1) and as previously shown 4. These data also indicated that IFN-α or MVA-BN treatment enhanced serum levels of FL. Thus, several lines of evidence suggested that FL and IFN-α/β played important roles in neonatal viral resistance, but it was not clear whether these factors acted synergistically or independently.
To further dissect the interdependence of IFN and FL, we analyzed the role of IFN-α in the protection of FL–/– mice from HSV-1 infection.
First, it was necessary to determine the susceptibility of FL–/– mice to HSV-1 infection. As FL is present in C57BL/6 neonates and as exogenous FL increases protection against viral infections 4, we hypothesized that mice lacking FL would be more susceptible to viral infection. Indeed, 100% of 7-day-old neonatal mice lacking FL expression were killed by as few as 50 PFU of HSV-1.
To define which cell populations were responsible for the high susceptibility, we performed phenotyping of spleen cell populations of FL–/– neonates at day 7. Similar to adult FL–/– mice 26, DC were highly reduced in cell numbers (Fig. 2, 3). In comparison to C57BL/6 mice, many of the cDC present had an immature appearance with lower expression of CD11c and MHC class II. In contrast to C57BL/6 neonatal spleen cDC that were mainly CD8+, most of the cDC of FL–/– mice were CD4–CD8–. We could not detect any difference in NK and B cell numbers (Fig. 3).
Treatment with human FL (huFL) or murine FL (muFL) reconstituted DC to similar numbers as seen in untreated C57BL/6 mice. After FL treatment, the FL–/– cDC remained CD4low/CD8low. CD4 but not CD8α expression was induced on pDC (Fig. 2). The relative number of NK cells (Fig. 3) was increased.
We next tested if treatment with huFL or muFL not only increased DC numbers in FL–/– neonates (Fig. 2, 3) but also restored viral resistance to C57BL/6 levels. After treatment with huFL for 1 wk, FL–/– neonates had an increased LD50 of 104 PFU of HSV-1, while treatment with muFL resulted in an LD50 of 103 PFU of HSV-1, which equals the resistance of untreated C57BL/6 neonatal mice (Fig. 4A). By contrast, all of the control mice lacking FL expression were killed by as few as 50 PFU of HSV-1 (Fig. 4A). Thus, cells such as DC derived from FL-dependent progenitors play an important role in the defense against HSV-1 in neonatal mice.
rIFN-α induces FL-independent short-term protection of neonates against HSV-1
We next tested whether treatment with IFN-α had any effect on the susceptibility of FL–/– mice to infection.
Treatment with rIFN-α at day 6 of age increased resistance, so that 80% of the FL–/– mice survived an infection with 5 × 103 PFU of HSV-1 at day 7 (Fig. 4B), while all untreated FL–/– neonates died. Thus, rIFN-α conferred protection during an infection with HSV-1 even in the absence of FL and in an environment with greatly reduced DC numbers.
We have previously shown that MVA-BN, an efficient inducer of IFN-α/β 27, 28, is able to increase resistance to neonatal infection with HSV-1 in an IFN-dependent manner 5. Therefore, we expected that treatment with MVA-BN at day 5 of age also induced protection against HSV-1 in FL–/– neonates. Indeed, in a dose-dependent manner MVA-BN was able to protect against HSV-1. A dose of 2.5 × 106 MVA-BN protected 80% of the neonatal mice while a dose of 2.5 × 104 MVA-BN did not have any protective effect (Fig. 4B and data not shown). Thus, we conclude that rIFN-α or MVA-BN both have a short-term, FL-independent effect on the resistance against HSV-1 in neonatal mice.
The long-term protection of rIFN-αis FL dependent
We have shown that MVA-BN treatment within 24 h of birth was able to protect against infection with HSV-1 at day 7 when the IFN-α/β system was intact. Furthermore, the treatment was also able to elevate FL concentrations in serum and pDC numbers in the spleen in neonatal mice 5. Since in this case induction of IFN-α was very early and viral challenge 1 wk later, we speculated that MVA-BN induced a long-term protection against viral infections that was dependent on MVA-BN-induced IFN-α and FL. To dissect the relative contribution of IFN and FL to this long-term effect, we examined if treatment with rIFN-α or MVA-BN at birth could protect FL–/– mice against HSV-1 infection 1 wk later.
C57BL/6 mice were treated with rIFN-α at day 0/1 and challenged with HSV-1 at day 7. rIFN-α-treated mice showed an increased resistance with an LD50 of 105 PFU of HSV-1 (Fig. 1). To determine the dependence on FL, we treated FL–/– neonates with rIFN-α at day 0/1 and then infected them with 5 × 103 PFU of HSV-1 at day 7. rIFN-α treatment did not show any effect and all treated mice died within the same time interval as the untreated controls (Fig. 4C). Furthermore, treatment of FL–/– mice with different doses of MVA-BN within 24 h of birth also did not show any protective effect (Fig. 4C and data not shown). Similar results were obtained with 500 PFU of HSV-1 as a challenge (data not shown).
Thus, we conclude that rIFN-α has a long-term, FL-dependent effect on the resistance against HSV-1 in neonatal mice.
The absence of FL impairs IFN-α/β productionin vitro
The high susceptibility of FL–/– neonates to HSV-1 could have been caused by an FL-dependent paucity of pDC and other cells able to produce IFN-α/β after direct stimulation with virus 12. Therefore, we investigated the capacity of bone marrow, spleen and liver cells from 7-day-old mice to produce IFN-α. After cultivation with CpG, HSV-1 or MVA-BN, cells of FL–/– neonates produced 10–80 times less IFN-α than cells from C57BL/6 neonates (Fig. 5). This drastic reduction was evident in cell suspensions from the whole organ just depleted of red blood cells as well as in dendritic cell-enriched cell suspensions. As CpG oligonucleotides induce IFN-α from pDC exclusively 12, the drastically reduced production of this cytokine correlates with the low number of pDC in the spleen (Fig. 2, 3). The data likely reflect similar low levels of pDC in liver and bone marrow of FL–/– mice and may explain the low resistance of FL–/– mice against viral infections.
rIFN-α treatment of C57BL/6 mice increasesthe number of pDC
Treatment with rIFN-α or MVA-BN increased protection against viral infection. This could be a direct effect of IFN-α/β or an indirect effect by induction of FL 4, 5. Therefore, we treated C57BL/6 and FL–/– mice with rIFN-α at birth and investigated the relative and absolute cell numbers of DC subpopulations in the spleen at day 7.
Treatment of C57BL/6 neonates at day 0/1 with rIFN-α increased the proportion of pDC to cDC to 1.2 : 1. In absolute cell numbers, pDC numbers increased twofold and cDC numbers decreased twofold in the spleens of these animals (Fig. 6). By contrast, treatment of FL–/– mice with rIFN-α at day 0/1 had no appreciable effect on the splenic cell composition (data not shown). Expression of CD8α, CD4 and MHC class II was not changed after any rIFN-α treatment (data not shown).
Thus, rIFN-α treatment leads to an increase of pDC numbers, which is FL dependent.
rIFN-α induces increased FL production
Since treatment with rIFN-α increased the numbers of pDC in C57BL/6 neonates with an intact FL system, we determined if IFN-α itself could induce FL production in these mice. Untreated 1–4-day-old mice had moderate levels of FL in the serum (450 pg/mL). rIFN-α treatment, at day 1, increased these levels. As early as 6 h after treatment, the serum levels were 600 pg/mL, which were increased after 12 h to 1000 pg/mL and peaked at 24 h with 1400 pg/mL. The FL serum levels started to decrease at 48 h to 1300 pg/mL, but were still well above control levels after 96 h (700 pg/mL) (Fig. 7).
Thus, rIFN-α is able to induce a higher production of FL in the serum of neonatal C57BL/6 mice.
Induction of pDC by FL is IFN-α/β independent
Since rIFN-α was able to increase the number of pDC, we investigated if development of pDC was IFN-α/β dependent. For this, we used IFN-α/β receptor gene-deleted (A129) neonatal mice. Flow cytometry showed that these mice had similar numbers of splenic pDC as 129Sv/Ev neonates. Treatment with huFL or muFL increased the number of pDC and cDC two- to threefold, a similar increase as observed in C57BL/6 neonates 4.
Thus, although rIFN-α induces increased numbers of pDC, the development of pDC is not IFN-α/β dependent and can be augmented by exogenous FL treatment.
pDC play the decisive role in the defense against HSV-1 in neonates
IFN-α and pDC seem to play an important role in the defense against HSV-1 in neonatal mice. To test the effect of pDC directly, we transferred pDC from FL-treated adult C57BL/6 mice into 6-day-old FL–/– mice and infected them at day 7 5. Transfer of 5 × 106 pDC into FL–/– mice increased the rate of survival, so that 7 out of 13 mice survived infection for over 3 wk (Fig. 8). Control mice usually died between days 5 and 9 (Fig. 4). By contrast, when 5 × 106 cDC (Fig. 8), CD11c+/CD45RA–, or CD11c–/CD45RA– cells, the NK cell-enriched fractions from adult C57BL/6, or pDC from adult A129 animals (data not shown) were transferred, all mice died within a time frame of 8 days.
Therefore, transferred adult pDC, but not cDC or other cells, provide enhanced resistance against viral infection in FL–/– neonatal mice.
In this report, we define that IFN-α administration to neonates provides anti-HSV protection in two distinct phases. The first, ‘early’ phase we define as protection elicited by IFN-α administration 24–36 h before viral infection. The second, ‘late’ phase we define as protection elicited by IFN-α administration 6–7 days before viral infection.
The ‘early’ phase is clearly independent of the DC poietin FL. Treatment of FL–/– mice with rIFN-α was protective when administered 24 h before challenge with HSV-1.
IFN-α induces a few hundred genes that are effective in the control of viral replication, such as GTPases, in virtually all cells of the body 29–31. In addition, rIFN-α boosts innate and specific immune responses 32. It is reported that IFN-α activates DC in vivo and enhances the DC induction of CTL 33, 34. The FL–/– mice were clearly deficient in both cDC and pDC, relative to C57BL/6 neonates. Moreover, spleen, liver or bone marrow preparations from FL–/– mice that were enriched for DC exhibited a severe decrease in IFN-α production (Fig. 5). Thus, administration of IFN-α shortly before viral challenge is possibly able to counteract the lack of IFN-producing cells in these mice.
Treatment with MVA-BN 36 h before heterologous viral challenge was also protective in FL–/– mice. Interestingly, co-administration of MVA-BN with the challenge virus had no protective effect (data not shown), indicating that the protective effect required some time for induction. Even though we cannot exclude the possibility of NK (or DC) activation 32 for increased survival 35 after MVA-BN stimulation, the data can best be explained by the need of IFN type I production and release by pDC. The low number of pDC in FL–/– mice correlates with low protection against viral infection. Increased numbers of these cells induced either by FL treatment as shown here or previously 4 or directly by transfer of pDC into naive mice was associated with protection. Transfer of cDC, NK cells or pDC from animals without functional IFN receptors had no protective effect (Fig. 8, data not shown and 5).
In contrast to the short-term antiviral effect of IFN-α, the long-term effect was strictly dependent on the presence of FL. There was no difference in antiviral protection between FL–/– mice that were left untreated, treated with IFN-α or MVA-BN (Fig. 4). Moreover, neither treatment of FL–/– mice had any effect on the splenic DC composition analyzed at 7 days of age.
However, treatment of C57BL/6 mice at birth with IFN-α increased serum levels of FL over a 4-day period and increased pDC twofold and actually decreased cDC numbers (Fig. 6), and concomitantly correlated with viral protection. This pDC bias in DC increase differs from the FL-induced total DC increase seen in C57BL/6 mice 4. Although the single IFN-α treatment induced FL serum increases over 4 days, this is a much shorter period than our normal FL treatment. We hypothesize that although both FL and IFN-α treatment of C57BL/6 neonatal mice can increase pDC numbers, short FL bursts, as obtained as a consequence of IFN-α treatment, preferentially lead to an increase in pDC.
By contrast, treatment of FL–/– animals with huFL or muFL from birth for 7 days increased the number of DC to levels similar to those of untreated C57BL/6 mice. This was associated with protection and thus highlights the importance of FL for increased viral protection in neonates.
The numbers of DC appeared to be associated with increased protection against viral infection 4, 5. It was therefore important to determine which DC population conferred resistance against viral infection. FL–/– mice at 7 days of age have limited numbers of cDC and pDC and proved ideal for transfer experiments with purified cell populations followed by viral challenge. Among the cell populations tested, only adult pDC induced protection, whereas all the other cells had no effect. pDC appear crucial to control viral replication before T cells, and perhaps other specific cells, can provide long-term protection.
As shown in Fig. 2, the DC subsets of neonatal FL–/– mice, even after FL treatment, differed in phenotype from DC of C57BL/6 mice. In particular, there was an absence of CD8+ pDC and ‘normal’ CD8+ and CD4+ cDC subsets. Thus, even with the exogenous addition of FL, the environment of the FL–/– mouse cannot, at least in the short term, support the development of all DC subsets. The treatment of FL–/– mice with muFL or huFL increased resistance to levels of untreated C57BL/6 mice, but transfer of purified adult pDC to these mice was less effective (Fig. 8). It is possible that the synergism between pDC and cDC subsets, particularly CD8+ DC, enhances the response to HSV-1. Likewise, we cannot rule out that FL-increased pDC and FL-increased NK cells act synergistically. Moreover, administration of FL to mice increases DC numbers in all organs. It is unlikely that transferred mature adult pDC are as effective in seeding organs and subsequently responding to potential systemic HSV-1 infection as are the pDC derived from FL-induced precursors in vivo.
We conclude that IFN type I, and perhaps other mediators immediately released by virus-infected pDC, controls local viral replication, attracts more pDC to local inflamed tissue and induces a second wave of pDC by an FL-dependent mechanism important to control the dissemination of virus to other organs. This mechanism allows enough time to initiate protective specific immune responses that may rely on additional DC or cell subsets.
Materials and methods
Animals and reagents
C57BL/6, 129Sv/Ev, IFN-α/β receptor gene-deleted (A129) 9 and FL gene-deleted (FL–/–) 26 mice were either bred under specific pathogen-free conditions at the Labortierkunde, University of Zurich (Switzerland) or bought from Harlan (The Netherlands). Litters were of mixed gender.
The HSV-1 F strain was originally obtained from B. Roizmann (University of Chicago, USA) and propagated on Vero cells. For all experiments, virus particles were used after purification by ultracentrifugation on a sucrose density gradient, and the virus titer (PFU) was determined as described 36, 37.
Human rIFN-α was a gift from M. A. Horisberger (Novartis, Basel, Switzerland), or the same material was obtained commercially (PBL Biomedical Laboratories).
huFL was a gift from Amgen (Seattle, USA) and was prepared in endotoxin-free phosphate-buffered saline (PBS). muFL was a gift from M. Becher (University of Zurich, Switzerland).
The Bavarian Nordic strain of MVA (MVA-BN: European Collection of Animal Cell Cultures; no. V000083008) was used for some of the vaccination experiments 5. MVA-BN was tested in highly immune-deficient mice with IFN receptor and RAG genes deleted 38 and proved to be safe, as no virus replication occurred and the mice survived for >100 days 5.
Treatment with rIFN-α, MVA-BN, FL and infection with HSV-1 of newborn mice
For analysis of the ‘long-term effects’ of IFN-α, mice were treated subcutaneously (s.c.) with two doses each of 105 IU rIFN-α within 24 h of birth and 12 h later (day 0/1 treatment). For analysis of the short-term effects of IFN-α, mice were injected at day 6 with two doses each of 105 IU rIFN-α within 8 h (day 6 treatment).
Newborn mice were injected s.c. within 24 h of birth and for six consecutive days with 1 µg huFL or muFL in 50 µL PBS as described 4.
C57BL/6 mice at 6–8 wk of age were injected s.c. with 10 µg huFL in 100 µL PBS for six consecutive days.
For challenge experiments, animals were infected at day 7 by intraperitoneal (i.p.) inoculation of 50 µL HSV-1 suspension. We euthanized infected animals when they were terminally ill or at 21 days post infection. All experiments were conducted at least three times, and different doses of virus were used as indicated in the text.
Immunofluorescent labeling of DC
The mAb, the fluorescent conjugates and the multicolor labeling procedure have been described elsewhere 4. For sorting cDC and pDC from spleens, anti-CD11c (HL3)-allophycocyanin (APC) conjugate and anti-CD45RA (clone 14.8)-PE conjugate (BD Bioscience) were used. For surface phenotype analyses, the following mAb were used: anti-CD11c (HL3)-APC, anti-CD45RA (clone 14.8)-PE, anti-CD8α-FITC, anti-CD4-FITC, anti-MHC class II (2G9)-FITC, anti-MHC class II (M5/114)-PE, anti-CD19 (1D3)-FITC, anti-B220 (RA3–6B2)-APC, anti-CD3 (17A2)-PE and anti-NK1.1 (PK136)-FITC.
Flow cytometric sorting and analysis
For sorting cDC and pDC populations, an ARIA (Becton Dickinson) was used. Cell analyses were performed on a FACSCaliburPLUS instrument (Becton Dickinson) as described 4, using four fluorescent channels for the immunofluorescent staining (FL1 for FITC, FL2 for PE, FL3 for PI, FL4 for APC).
DC isolation procedure
The procedure is described elsewhere 4. Briefly, spleens of 8–21 neonates were digested with collagenase and DNAse, following EDTA treatment at room temperature. The next steps were performed at 4°C. Light-density cells were collected after density cut with Nycodenz 1.077 g/cm3 (Nycomed, Oslo, Norway). The cells were incubated with an antibody depletion cocktail, which contained anti-CD3 (KT3), anti-Thy1 (T24/31.7), anti-CD19 (ID3), anti-GR-1 (RB6–8C5) and anti-erythrocyte (TER-119) antibodies, and were incubated with anti-rat Ig Dynabeads M450 (Dynal). For purification and segregation of the populations of cDC and pDC, cells were labeled with anti-CD11c (HL3)-FITC and anti-CD45RA (clone 14.8)-PE mAb, and the distinct populations were sorted using the ARIA instrument.
Spleen, bone marrow or liver cells (2.5 × 106/mL) depleted of red blood cells, or bone marrow cells (2.5 × 106/mL), spleen or liver cells (1 × 106/mL) depleted of CD3- (KT3), Thy1- (T24/31.7), CD19- (ID3), TER-119- and GR-1-positive (IA8) cells were stimulated in vitro with CpG-2216 (1 µM; TIB MOLBIOL, Berlin, Germany), HSV-1 or MVA-BN at concentrations shown to induce maximal IFN-α production in pDC, in a 96-well plate. Supernatants were harvested after 18–24 h of culture and analyzed by ELISA for IFN-α production 12.
Detection of serum FL
Mice were treated with rIFN-α at day 0/1 and bled at different time points to determine the FL concentration in the serum by ELISA (R&D Systems), according to the manufacturer's recommendations.
Spleen cells of FL-treated, 6–8-week-old C57BL/6 mice were obtained by enzyme digestion and density cut and stained with anti-CD11c (HL3)-APC and anti-CD45RA (clone 14.8)-PE mAb. Sorted pDC, cDC, CD11c-negative/CD45RA-positive and double-negative cells in a concentration of 5 × 106 cells per 50 µL were injected i.p. into 6-day-old mice 5.
This work was supported by the Kanton of Zurich, the Swiss Science foundation, Bavarian Nordic, Germany, and the European Commission FP6 ‘Network of Excellence’ initiative under contract no. LSHB-CT-2004–512074 DC-THERA.