Adrian Bot, Stefan Antohi and Constantin Bona

The Department of Microbiology, Mount Sinai Medical School, New York, USA

Received 4/8/97; Accepted 4/15/97; On-line 5/1/97

4. IMMUNE RESPONSIVENESS OF THE NEONATAL T CELLS

The classical experiment of neonatal tolerance induction following the injection of allogeneic cells (1) suggested that self-nonself discrimination is a consequence of the deletion of lymphocytes that encounter antigens during a certain neonatal time-window. However, only few things about the biology of the T cells were known at that time. More recent studies showed that rather than being unresponsive, the neonates mount immune responses that are qualitatively different as compared to adult animals (8,15). Furthermore, in certain circumstances, newborns display immune responses that resemble those of the adults (17,19). Two non-mutually exclusive models could explain the particular immune responsiveness of the neonatal T cells: the first one considers that differences in the total numbers of APC and T cells between adults and neonates account for the differences in the immune responsiveness (the quantitative model). The second, considers that the neonatal T cells respond in a functionally different manner to antigens (the qualitative model). Interestingly, there are numerous studies that support both hypotheses.

4.1. Quantitative differences between the immune systems of neonates and adults

This model is based on the following observations: (a) neonates are able to mount T cell responses to various types of antigens (8,17,19,20,46); (b) the T cell response of neonates to low doses of antigens is adult-like (8,19); (c) the T cell response of adults to high doses of antigens resembles the immune response of newborns to 'tolerogenic' doses (20) and (d) antigen presentation by professional APC (pAPC) during the neonatal period, is followed by adult-like responses (20).

The central assumption of this model is that T cells are turned-on by the pAPC, whereas non-professional APC (npAPC) that do not express enough co-stimulatory molecules, turn-off the naive T cells. The 'Danger Model', that attributes the ability of initiating the immune response to the pAPC (47), regards the DCs, the activated monocytes and B cells as pAPC, whereas the resting B cells are npAPC (20,48). Thus, the ratio between the pAPC and npAPC that present the antigen to naive specific T cells would determine the outcome of the response, i.e. priming versus tolerance. This model explains why whereas low doses of antigens produce adult-like immune responses in neonates, very high doses of antigens are tolerogenic even in adults (Fig. 3). Simply, moderate doses of antigens are sufficient in neonates to shift the overall balance toward nonprofessional presentation, because the number of APC is lower. In contrast, low doses of antigens would lead to an exclusive loading of pAPC. Thus, this quantitative model explains the observations that neonatal tolerance can be avoided by immunization with low amounts of antigens (8,15,19,45,49) or with pAPC bearing the antigen (20).

Figure 3. The quantitative model that addresses the differences of immune responsiveness between the neonates and adults. The number of professional (rectangles) and nonprofessional (circles) APC is proportionally decreased in neonates. The outcome of the immune response is regulated by the ratio between professional and nonprofessional APC that present the antigen. Thus, the inoculation of antigen in doses that are immunogenic for adults ('medium') is followed by predominant non-professional presentation and subsequent tolerance in the neonates. It can be inferred that immunization with low doses of antigens may have priming effects in newborns, whereas injection of large doses of antigen may lead to tolerance in adults as well as neonates.

The quantitative model has to take into consideration the data regarding the decreased number of class II⁺ macrophages in the peripheral lymphoid organs of neonatal organisms (50-52) and the low efficacy of the neonatal B cells as APC (53). Thus, besides the lower number of APC, the ratio between pAPC and npAPC seems to be decreased in the neonatal as compared with the adults.

A consequence of the quantitative model is that the in vitro responsiveness of the neonatal and adult T cells are the same. Numerous observations argue against this fact, although they do not discard the validity of the quantitative model.

4.2. Functional differences between the lymphocytes of neonates and adults

The model based on qualitative differences between neonatal and adult peripheral T cells is based on the following observations: (a) the susceptibility of the neonates to high zone tolerance (1); (b) the strong tendency of the neonatal T cells to produce Th2 cytokines and the decreased ability to produce Th1 cytokines following in vitro stimulation with polyclonal activators (54-56); (c) the strong tendency of the neonates to mount Th2 biased immune responses that are refractory to a subsequent switch toward Th1 responses (15,16,57).

This qualitative model is strongly supported by studies that show phenotypic differences between neonatal and adult T cells. For example, most of the mouse neonatal CD4⁺CD8^- T cells lack Qa-2 membrane antigen and display low responsiveness to polyclonal stimulators (58). Furthermore, developmentally immature T cell subsets are present in the periphery of young rodents and they may undergo post-thymic maturation in the peripheral lymphoid organs (59,60). At present, little is known about the expression of co-receptors (i.e. CTLA-4, CD28, CD27, CD50, Fas) or cytokine receptors (i.e. TNFalpha-R) on the neonatal T cells. A functional study indicated a high tendency for the neonatal T cells to undergo apoptosis following in vitro polyclonal stimulation (56).

A recent study carried out with human neonatal T cells demonstrated that polyclonal activation by anti-CD3 mAb in the presence of anti-CD28 mAb led to Th1 or Th2 differentiation, depending on the addition of IL-2 or IL-4 (61). Interestingly, neonatal T cells were able to produce both IL-2 and IFNgamma (61), suggesting that strong co-stimulation circumvented the reduced ability of the neonatal T cells to differentiate to adult-like Th1 cells. Because the human neonatal T cells do not express detectable CD40-L (36), a direct CD28 engagement would bypass the requirements for B7.2 expression and IL-12 secretion by pAPC, that are otherwise dependent on the CD40/CD40-L interaction (39,62). This hypothesis is indirectly supported by studies which showed that production of IL-2 by polyclonally stimulated neonatal T cells of mouse origin, can be enhanced by co-stimulation with anti-CD28 or IL-6 (63). Furthermore, strong co-stimulatory factors like IL-2, IL-6 or anti-CD28 mAb partially rescued the neonatal T cells from anti-CD3 mAb induced apoptosis (56). Thus, the overall view is that the neonatal T cells are impaired in their ability to receive and transduce co-stimulatory signals, leading to predominant differentiation to Th2 cells or apoptosis.

The above mentioned studies were carried out using polyclonal stimulation of the T cells. In order to circumvent this caveat, we addressed the question regarding the in vitro responsiveness of neonatal T cells in a TCRalpha/beta transgenic system. The transgenic receptor is specific for HA 110-120 peptide in the context of I-E^d molecules and the transgenic mice express the receptor (TCR-HA) on both CD4⁺ and CD8⁺ T cells (64). If quantitative differences were solely responsible for the particular immune responsiveness of neonates, we would expect an identical behavior of neonatal and adult T cells in vitro stimulated by antigen and pAPC. Interestingly, the responsiveness profiles of neonatal and adult T cells from TCR-HA transgenic mice are distinct: (a) whereas neonatal T cells secreted large amounts of IL-4 and no IFNgamma, adult T cells secreted large amounts of IFNgamma and no IL-4; (b) proliferation ability of the neonatal T cells was very low compared to that of the adult T cells (Table 2).

a Proliferation index at least 10 times higher for adult compared to neonatal T cells;

b Following in vitro priming with HA 110-120 peptide, effector cells displayed specific lysis values between 50-70% in both cases;

c (-) represents values below the sensitivity of the ELISA assays (5 pg/ml); (++) corresponds to 150-300 pg/ml secreted by 1x10⁵ T cells after 4 days in culture with HA 110-120 peptide; (+++) corresponds to higher concentrations of cytokines (above 350 pg/ml);

d Enriched dendritic cells (DCs) were obtained from adult BALB/c splenocytes by overnight incubation of plastic-adherent cells in presence of GM-CSF (10 U/ml), followed by the recovery of the cells detached into the medium.

The overall frequency of the TCR-HA⁺ T cells in the spleens of adults and neonates were comparable, so that this quantitative factor alone cannot explain the above-mentioned differences (see below and Fig. 4). Furthermore, stimulation of the T cells in presence of enriched adult pAPC (Table 2) or various doses of HA 110-120 peptide (not shown) did not change the responsiveness profiles. Surprisingly, high CTL activities could be generated by in vitro peptide priming of both neonatal and adult TCR-HA⁺ T cells (Table 2). This indicates that high CTL activity can be generated during an exclusive Th2 response, although it is not clear if this result is valid only in the case of MHC-II restricted CTLs.

Figure 4. Frequency and immunophenotype of TCR-HA⁺ T cells in neonatal and adult spleen (A) or peripheral blood (B) of transgenic mice. Samples were pooled from 3-10 mice in each group, T cells were separated and triple-staining FACS analysis was carried out using reagents specific for the TCR-HA receptor, CD4 and CD8 surface molecules. Results were expressed as percentage of cells in a given phenotypic group, relative to the total number of T cells.

The study of presence and phenotype of TCR-HA⁺ T cells in the spleen or blood of transgenic mice (Figure 4) showed that: (a) the overall frequency of TCR-HA⁺ T cells in newborns and adults were comparable, around 5-15% of the total T cell population; (b) the frequency of CD4⁺ CD8^- TCR-HA⁺ T cells was significantly lower in the neonates; (c) in contrast to adult mice, most of the neonatal TCR-HA⁺ T cells were CD4-CD8^-. Thus, although the relevance of this model for the wild-type organisms is still unclear, it appears that there are important functional and phenotype differences between the peripheral T cells of newborns versus adults. Furthermore, a recent study regarding the TCR-HA transgenic mice, suggested the age-dependent accumulation of T cells with activated phenotype, in the absence of intentional antigenic stimulation (65).

Taking into account all the data mentioned above, it becomes clear that a better model has to combine both quantitative and qualitative differences between APC and T cells respectively, from adults and neonates. The quantitative aspect is still very important and can explain some peculiarities of the neonatal immune system as a whole: (a) the low magnitude of the 'adult-like' immune response obtained in certain conditions, due to the presence of a few functional lymphocytes in the periphery of neonates and (b) the susceptibility of neonates to high-zone tolerance simply because the number of specific T cells to be turned-off is significantly lower. Indeed, an analysis of the results regarding the phenotype of TCR-HA⁺ T cells, suggested that the total number of specific T cells in neonates is at least 100 times lower than in adult organisms (Fig.4 and data not shown).

4.3. Cellular immune responses of mice immunized as newborns with naked DNA

In influenza virus infection, the cell mediated immunity plays an important role in the recovery from disease (66). First, virus-specific CTL mediate the clearance of the virus from the infected lungs (67). Secondly, in the absence of CD8⁺ T cells and MHC-II expression on the lung epithelial cells, CD4⁺ virus-specific T cells are able to clear type A influenza viruses from the infected lungs (68). Whereas Th1 cells are able to mediate a protective effect, Th2 cells display rather detrimental effects on the evolution of influenza pneumonia (69). Thus, multiple arms of the cellular immunity participate to the defense reaction in influenza virus infection.

Previous studies showed that plasmid immunization of adult animals led to protective cellular responses in the influenza virus system (40,43). We hypothesized that small amounts of antigen would be inefficient in inducing high-zone tolerance even in very young animals, so that the few mature antigen specific T cells could be primed and subsequently expanded because of the persistence of antigenic stimulation. We immunized newborn and adult BALB/c mice with a plasmid encoding NP of the PR8 strain of influenza virus (NPV1) (70) according to the protocol described in Fig.1. We studied the CTL immune response against target cells infected with influenza viruses or coated with NP 147-155 peptide, that is the immunodominant K^d epitope. Interestingly, immunization of newborn mice with NPV1 primed a strong CTL response rather than inducing unresponsiveness (17,71). CTL priming by neonatal NPV1 inoculation was demonstrated by: (a) increased CTL precursor (pCTL) frequency in the spleens of animals immunized with NPV1 and boosted with PR8 virus, compared to animals inoculated with virus or plasmid only (Fig.5); (b) significant cytotoxic activity specific for various type A influenza viruses as well as for the NP 147-155 immunodominant epitope of splenocytes from mice immunized as neonates or adults with NPV1 (17,71); (c) CD8⁺ T cells from mice immunized as neonates with NPV1 secreted IFNgamma but no IL-4 when restimulated with NP 147-155 peptide, like those from adult animals immunized with live-virus (71); (d) T cells from animals immunized as newborns with NPV1 and boosted with live-virus displayed significantly enhanced proliferative abilities following antigenic stimulation, compared to cells from animals immunized with virus only (71); (e) animals immunized as newborns or adults with NPV1 showed enhanced protection in terms of pulmonary virus titer reduction and survival, subsequent to lethal infection with PR8 (H1N1) or HK (H3N2) viruses (17,71).

Figure 5. Frequency of the PR8 specific pCTLs in the spleen of mice immunized as newborns or adults with NPV1, at one month after the completion of immunization. Some mice were boosted with live-PR8 virus one week previous to pCTL estimation. Control mice were immunized only with live-virus one week before the sacrification. The frequency of pCTLs was estimated by limiting dilution analysis. Results were expressed on a log₁₀ scale, as mean of pCTL frequency x 10^-5 in groups of 3-4 mice.

Mice immunized as newborns with NPV1 displayed significant virus-specific CTL immunity at least 6 months after the completion of immunization (Fig.6 and ref.71). Interestingly, the kinetics of CTL induction was slower in the case of animals immunized with NPV1 as newborns, indicating that the expansion of the memory pool paralleled the development of the T cell repertoire. The plasmid as detected by PCR, persisted at least 1 month but not more than 3 months in mice inoculated at birth (Fig.6 and ref.71), indicating that the continuous antigenic exposure primed newly emerging T cells and expanded the memory pool.

Figure 6. Kinetics of the total number of PR8 specific pCTLs in spleens of animals immunized as newborns or adults with NPV1. The number of pCTLs was estimated from the frequency obtained by limiting dilution analysis and the total number of splenocytes. Results were expressed as mean of total pCTL/spleen x 10^-2 in groups of 3-4 mice. Control mice were inoculated with the plasmid that lacks the NP open reading frame (CP) or with live PR8 virus, one week previous to the sacrification. PCR results regarding the persistence of NPV1 at the site of injection were considered positive (+) if at least one mouse in a particular group was positive and negative (-) if no mouse was positive. Each PCR group consisted of 6-10 mice. ND - not done.

Thus, the CTL response of neonates and adults to NPV1 were qualitatively identical, although some quantitative differences occurred. Whereas NP encodes immunodominant CTL epitopes, HA encodes the major B and Th epitopes of influenza virus (reviewed in ref.72). We investigated the T helper response of mice immunized as neonates or adults with pHA that expressed HA of WSN virus. It was already mentioned that pHA immunization induced HI antibodies specific for WSN virus, in animals inoculated as neonates or adults. The study of the isotypes profiles was carried out by RIA and showed: (a) the predominance of IgG₂ antibodies in mice immunized as neonates or adults with pHA, compared to adult mice immunized only with virus (Fig.7A and ref.44); (b) the predominance of IgG₂ antibodies in mice immunized as adults with pHA and boosted with WSN virus; (c) the predominance of IgG₁+IgG₃ antibodies in mice immunized as newborns with pHA and boosted with WSN virus.

These observations suggested that whereas pHA immunization of adult mice led to the development of Th1 cells, immunization of neonates was followed by a mixed Th1/Th2 response: the Th1 profile predominated in mice that were not exposed to WSN virus and the Th2 response predominated after the exposure to live WSN virus. The study of IFNgamma and IL-4 secretion by T cells from mice immunized as neonates or adults with pHA and boosted with live-virus, supported this hypothesis: while the T cells from mice immunized as adults produced significantly more IFNgamma than IL-4, the T cells from mice neonatally immunized produced comparable amounts of IFNgamma and IL-4 (Fig.7B and ref.44). Thus, antigenic exposure during the early postnatal stage of development led to a qualitatively different Th response comprising virus-specific Th2 commited precursors, that differentiated to Th2 effector cells after restimulation with antigen.

Figure 7. The helper profile of the WSN specific T lymphocytes primed by pHA inoculation of newborn or adult mice. A: The ratio between IgG₂ and IgG₁+IgG₃ WSN specific antibodies in the sera of mice immunized with pHA one month previously and boosted or not with the live-virus one week previous to the blood harvest. The concentration of the specific antibodies bearing a particular isotype was estimated by RIA using isotype-specific reagents. B: The concentration of IFNgamma and IL-4 in culture supernatants of WSN restimulated T cells, from mice immunized with pHA as newborns or adults and boosted with live-virus one week before the sacrification. The concentration of cytokines was assessed by ELISA and the results were expressed as means of triplicate estimations.

There are two mutually exclusive scenarios that could define the kinetics of antigen exposure following the neonatal inoculation with plasmid vaccines: (a) plasmid inoculation leads to immediate and continous exposure of neonatal and post-neonatal T cells to the antigen or (b) the exposure occurrs later on, when the T cell repertoire is more developed. Three lines of evidence support the first scenario: (a) mice immunized with NPV1 at birth mounted enhanced CTL responses when boosted with virus three weeks later (Fig.5 and refs.17,71); (b) the Th response of mice immunized as neonates with pHA and boosted with virus 1 month later is 'newborn-like', comprising a significant Th2 component (Fig.7 and ref.44); (c) it was previously shown that the protein synthesis and antigen accumulation begins after a short lag-interval of 24-72 hours following the intracellular delivery of the expression vectors by various means (40,73-75). Thus, it is more likely that an early continuous exposure to low doses of antigens occurred following neonatal inoculation with mammalian expression vectors. A consequence of this model would be that increasing the exposure of the neonatal immune system to antigens expressed by plasmids leads to the induction of unresponsiveness. In spite of the fact that we have not encountered such a phenomenon in the case of plasmids expressing influenza proteins that were inoculated in a wide dose-range (10 µg - 100 µg/dose), another group noted in certain circumstances neonatal tolerance following the inoculation of plasmids expressing the circumsporozoite antigen of the malaria parazite (D.M. Klinman, personal communication). This observation supports the model of early antigen exposure following neonatal immunization with plasmids, but suggests that there are some important parameters that may determine whether the immunization is effective or not: (a) the efficacy of transcription and the adjuvant-like properties (76,77) of the vector; (b) the route, schedule and dose of inoculation; (c) the genetic background of the animal; (d) the type of protein encoded by the expression vector (i.e. secreted or cellular); etc.. In contrast to adults, increasing the vaccine dose or the transcriptional efficacy of the vector may lead to rather deleterious effects in neonates, like T cell unresponsiveness. However, the tremendous efficiency of DNA immunization in generating Th1 immune responses not only in adults but to a certain extent in neonates as well (Fig.7 and ref.44), correlates well with both the notion that a continued stimulation with antigen must occurr in order to maintain the T cell expression of the beta-2 chain of IL-12R (78,79) and that bacterial DNA displays Th1 promoting adjuvant properties (76).

In conclusion, the data regarding neonatal immunization with naked DNA support the concept that functionally mature T cells are present during the early period of life, but do not rule out possible qualitative differences between the dominant populations of peripheral T cells in neonates and adults, respectively. Thus, it becomes clear that the classical view of self-nonself discrimination as consequence of neonatal tolerance (1-3) has restricted validity. Instead, neonatal lymphocytes are able to mount effector and memory responses in certain conditions. It can be inferred from these data that some forms of self-nonself discrimination must follow the T cell priming events. Indeed, besides the central tolerance that occurs in the thymus (10), peripheral tolerance assures that autoreactive T cells which escaped in the periphery (reviewed in ref.80) are eventually deleted or anergized. The best evidence to support such a model is that animals deficient in Fas/Fas ligand, IL-2 or CTLA-4 (81-83), develop early in their postnatal life fatal autoimmune diseases. A model that assimilates the responsiveness of the neonatal T cells and the self-nonself discrimination mechanisms, is schematically shown in Figure 8.

Figure 8. Two alternative models addressing the neonatal responsiveness of peripheral T cells. A: The classical model views the inability of the neonatal T cells to respond to antigens as the basis of the 'learned' process of self-nonself discrimination. This would be accomplished by the deletion of all T cells that encounter their antigens during the neonatal time-window. B: The revised model takes into account the ability of neonatal T cells to mount immune responses in certain circumstances. The inactivation of autoreactive T cells that escape into periphery would be accomplished by 'downstream' mechanisms mediated through Fas/Fas-L, IL-2/IL-2R and B7/CTLA-4 interactions.

In essence, an important idea emerged from these studies, namely that there are potentially effective immunization strategies, that do not require a live or live-attenuated vector, yet they are still able to induce broad humoral and cellular responses even during the earliest stages of postnatal development.