Atopic diseases including allergic rhinitis and asthma are inflammatory conditions that have increased in prevalence over the past two decades 1 . The inflammatory response to common environmental allergens during allergy and asthma has been extensively studied in the past years, and has clearly determined the pivotal role of T cell activation, with a predominant Th2 cytokine production 2 3 . T regulatory (Treg) cells, characterized by the production of anti-inflammatory cytokines such as IL-10 and TGF-β 4 5 are considered as responsible for the normal tolerance against auto-antigens and external antigens such as allergens 6 . Accordingly, a deficiency in Treg counts and activation was found in autoimmune diseases and allergic conditions, notably during allergen exposure 7 8 and exacerbations of severe asthma 9 . However although this Th2/Treg imbalance applies both for allergic rhinitis and asthma, it is remarkable that despite a same atopic background and allergen exposure, some subjects will develop both rhinitis and asthma whereas other will display rhinitis only. We hypothesize since several years that T cell activation is different between both conditions and with others we previously described a Th1 activation in asthma that was absent in non asthmatic allergy in blood, induced sputum and broncho-alveolar lavages 10 11 12 . However, the role of allergen in the tuning of T cell activation in allergic rhinitics with and without asthma was not explored yet.

Allergen-induced T cell activation depends on signals delivered from antigen presenting cells (APCs) through the antigen-specific T cell receptor as well as additional co-stimulatory signals provided by engagement of so-called co-receptors on APCs and T cells 13 . Major T cell co-receptors are CD28, inducible costimulatory molecule (ICOS) and cytotoxic T lymphocyte antigen (CTLA)-4. They belong to the immunoglobulin gene superfamily and display various kinetics of expression. CD28 is a constitutive co-stimulatory receptor binding CD80 and CD86 on APCs, delivering important signals for T cell activation and survival. Ligation of CD28 promotes the production of IL-4 and IL-5 and provides resistance to apoptosis and long-term expansion of T-cells. As CD28, ICOS is a positive regulator of T cell activation which is up-regulated on activated T-cells. ICOS was initially shown to selectively induce high levels of IL-10 and IL-4, but is also able to stimulate both Th1 and Th2 cytokine production in vivo 14 .

CTLA-4 is also a CD80/CD86-binding protein. It is up-regulated on activated T cells and delivers mainly an inhibitory signal, playing an important role in maintenance of peripheral tolerance 15 . Indeed, it was shown in murine Treg cells, that CTLA-4 controlled homeostasis and suppressive capacity of regulatory T cells 16 .

Co-receptors thus represent important potential targets for therapeutic immunomodulation. Indeed the blockade of CD28 and CTLA-4 agonists are tested for their ability to prevent graft rejection 17 , and in animal models, ICOS inhibition prevented allergic inflammation 18 . However, the actual role of co-receptors in the context of asthma and allergy in humans is still unexplored.

The objective of this study was therefore to compare the pattern of T cell activation between allergic rhinitics and asthmatics upon allergen stimulation and to assess the role of co-receptors CD28, ICOS and CTLA-4 in this process.

The results of our ex vivo study strongly suggest a contrasted picture of T cell activation in allergic rhinitis and asthma, with distinct patterns of Th1, Th2 and Treg profiles and expression of ICOS, CD28 and CTLA-4 co-receptors.

Indeed, we showed that in asthma, IFN-γ production was constitutive, did not increase upon allergen stimulation, and was not blocked by any of the anti-co-receptor antibodies. Similarly, the constitutive defect of Treg and CTLA-4 expression seen in asthmatics and not enhanced in non allergic asthmatics after allergen stimulation was not modified after co-receptors blockade. The Th1/Treg imbalance in asthma is therefore constitutive and independent of allergen presentation.

The constitutive Th1 activation in asthma was demonstrated before 10 12 21 . It could result from the intrinsic defect in the CTLA-4+ and Treg populations as CTLA-4, known to be involved in tolerance induction 22 , could prevent the asthmatic inflammation by inducing T cells to differentiate in T regulatory cells. Recently, we have showed during in vivo studies a lower proportion of Treg cells in blood from severe refractory asthmatics compared to controls, which was even deeper during exacerbations, both in blood and induced sputum 9 . Herein we show that this lower proportion of Treg is present in milder stages of asthma. Relevantly, Treg cells were higher in mild than in moderate asthma whatever the allergic status. This results are concordant with the primary Treg cell deficiency suggested in asthma and allergy 23 . That the Th1/Treg imbalance is similar in allergic and non allergic asthma suggests that it is a characteristic of asthma independent of allergy, possibly triggered by infectious agents or non specific substances such as pollutants but it must be precised that asthmatics included in the present study were controlled and did not experienced any recent exacerbation. Another hypothesis would be that the Th1/Treg imbalance in asthma is really intrinsic and independent of any external aggression.

In allergic groups, we demonstrated a Th2/Treg imbalance inducible upon allergen stimulation. That Th2 activation was not seen in non allergic patients and could be broken by CD28 and ICOS blockade indicates that it is really the cognate allergen presentation by antigen presenting cells that was responsible for it. IL-13 secretion was suppressed also by blocking CTLA-4, indicating that in peripheral cells (1) Th2 activation cannot be considered globally, Th2 cytokines being regulated distinctly, and (2) CTLA-4 being not only involved in tolerance but also in inflammation. This result is concordant with Lordan and al., who showed that allergen-induced production of IL-5 and IL-13 by PBMC from allergic asthmatics could be inhibited by blocking CTLA-4 receptor with CTLA-4-Ig 24 . Regarding the allergen-induced Treg defect in allergics, other co-receptors than these tested are likely involved, among which PD1 is a candidate 25 . Indeed, Meiler et al. recently demonstrated in PBMC from allergic patients that the suppressive effect of IL-10 secreting T cells was partially inhibited by blocking CTLA-4 or PD-1 co-receptors, whereas blocking both receptors simultaneously had an additive effect 26 .

The association of allergy with ICOS over-expression before any allergen stimulation suggests a non specific priming of T cells towards the Th2 pathway in allergic subjects. Indeed ICOS was clearly related to Th2 activation, as shown by anti-ICOS stimulation results. Numerous studies using animal models of airways inflammation have showed that ICOS-mediated signalling was essential for induction of Th2 cytokines 27 28 . Indeed inhibition of ICOS suppresses allergic lung inflammation and Th2 cytokines production in mice models 29 . However in other models ICOS engagement induces tolerance and inhibits the allergic inflammation. These distinct actions of ICOS seem related to the density of ICOS molecules per cell, with inflammation being related to a high density of co-receptors and tolerance induction to a lower number of ICOS molecules per cell 30 .

That in R ICOS expression does not increase after allergen stimulation by contrast with the AR group could result from a maximal expression of ICOS in R whereas it is still inducible in AR. Indeed the basal level of ICOS expression is lower in the latter group than in the former. This relative defect in ICOS expression in AR patients could result from the constitutive Th1/Treg imbalance of asthmatics that by a Th1-driven "anti-Th2" effect would decrease ICOS expression.

Under allergen stimulation, CD28 expression increased significantly in R and AR, and blockade of CD28 decreased the Th2 cytokine production, indicating the involvement of CD28 in Th2 cell activation in allergy. It is noteworthy that although significant statistically, the proportion of CD28+ cells could not increase in high proportion, as most T cells constitutively expressed CD28 in all groups. CD28 is a crucial co-receptor for inducing T cell cytokine production 31 , and was showed to be involved both in Th1 and Th2 activation. CD28 blockade is proposed as an immunosuppressive strategy to prevent graft rejection, and is experimented in various inflammatory diseases. However the practical use of CD28 blockade was refrained by the agonist action of some anti-CD28 antibodies encountered in clinical trials 32 .

Our study provides new insights into the hypothesis of Treg cell deficiency as a paradigm for allergic diseases, by showing a constitutive Treg cell deficit in asthma whatever the allergic status and an inducible Treg deficit in allergy, whatever the presence of asthma. As a consequence, the Treg cell deficiency is the highest in asthmatic allergics after allergen stimulation. This distinction between allergy and asthma contradicts our previous hypothesis of a gradient of Treg cell deficiency from allergy to asthma 23 , and better suggests that the abnormalities seen in both diseases could be juxtaposed and independent, as showed by the multiple linear regression analysis.

Recently an in vivo study showed no difference in the number of Treg cells between asthmatics and controls, whereas FOXP3 protein expression within Treg cells was significantly decreased in asthmatic patients 33 . Our study was performed in blood ex vivo and therefore might not fully reflect the in vivo and in situ reality. However many studies showed that blood compartment was relevant to the in situ inflammation as far as T cells and allergy were concerned 21 , and the mechanistic studies proposed here cannot be assessed in situ in humans. They can be performed in vivo in animals, but the relevance to real asthma would also be uncertain.

In conclusion, allergy is associated to a constitutive ICOS over-expression and inducible CTLA-4 under-expression with Th2/Treg imbalance, when a constitutive CTLA-4 under-expression and Th1/Treg disequilibrium appears as a hallmark of asthma. Both profiles are mixed in allergic asthma, and one can argue that asthma would occur in allergic subjects only if the unknown conditions leading to the constitutive Th1 activation are present. Still missing in the puzzle is the stimulus inducing the Th2 activation present in non allergic asthma 3 . Lastly, our results demonstrate that although targeting one type of T cell activation only would be a pitfall in allergic asthma, there is a rationale to develop strategies based on targeting co-receptors in allergy.

In conclusion, our work adds significant insights into the immune mechanism involved in allergy and asthma and states the rationale for new diagnosis and/or therapeutic strategies in these pathologies.

The authors declare that they have no competing interests.

All the authors have contributed significantly to the research and preparation of the manuscript, and they approve its submission.