Cytomegalovirus (CMV) infects a majority of the human population and establishes a life-long persistence. CMV infection is usually asymptomatic but the virus carries pathogenic potential and causes severe disease in immunocompromised individuals. T-cell-mediated immunity plays an essential role in control of CMV infection and adoptive transfer of CMV-specific CD8+ T cells restores viral immunity in immunosuppressed patients but a role for CD4+ T cells remains elusive. Here, we analyzed in adoptive transfer studies the features and antiviral functions of virus-specific CD4+ T cells during primary murine CMV (MCMV) infection. MCMV-specific CD4+ T cells expanded upon MCMV infection and displayed an effector phenotype and function. Adoptive transfer of in vivo activated MCMV-specific CD4+ T cells to immune-compromised mice was protective during pathogenic MCMV infection and IFN-γ was a crucial mediator of this protective capacity. Moreover, co-transfer of low doses of both MCMV-specific CD4+ T cells and CD8+ T cells synergized in control of lytic viral replication in immune-compromised mice. Our data reveal a pivotal antiviral role for virus-specific CD4+ T cells in protection from pathogenic CMV infection and provide evidence for their antiviral therapeutic potential.
Immunity to viral infections relies on effector mechanisms exerted by different components of the immune system. Despite the prominent roles of cytotoxic CD8+ T cells and neutralizing antibodies in controlling viral infections, there is growing evidence for an antiviral role of CD4+ helper T cells, not only in provision of help to other effector cells, but also in exertion of direct antiviral functions either by secretion of cytokines or exertion of cytotoxicity . CD4+ T-cell-mediated protective immunity has been reported for a number of viral infections [2-5]; however, the antiviral effector mechanisms exerted by CD4+ T cells in vivo have remained poorly characterized.
Cytomegalovirus (CMV) is a β-herpes virus that infects a majority of the human population worldwide. After resolution of primary infection, this large DNA β-herpes virus establishes a life-long persistence in its host. CMV infection is usually clinically silent in healthy individuals. Nevertheless, CMV-infected immunosuppressed patients, such as hematopoietic BM or solid organ transplant recipients, suffer from CMV disease with potentially severe clinical outcome . Cellular immunity is important to control CMV infection [7, 8] and a prominent role of CD8+ T cells in preventing CMV replication and protecting against CMV disease has been established in the immunocompromised host . In contrast, despite the accumulating evidence for an importance of CD4+ T cells in CMV immune surveillance, virus-specific CD4+ T-cell responses, with respect to their development and function, remain poorly defined. In humans, the presence of CMV-specific CD4+ T cells correlates with virus control in renal transplant patients  and impaired induction of CMV-specific CD4+ T cells correlates with prolonged virus shedding in children with CMV viremia . Furthermore, CD4+ T cells have been shown to play a role in the maintenance of antiviral CD8+ T-cell response in patients undergoing allogeneic BM or solid organ transplantation [12, 13]. Interestingly, adoptive transfer of CMV-specific T-cell lines restored human CMV (HCMV) specific CD4+ T-cell immunity, which correlates with reduced virus load in stem cell transplant patients with CMV viremia , suggesting a prominent role of virus-specific CD4+ T cells in control of HCMV replication. In mice, CD4+ T cells were shown to be crucial for the control of MCMV replication in the salivary glands, an important mucosal site of virus persistence and virus transmission . Recently, we identified a mechanism by which CD4+ T cells exert their potent effector functions in the salivary glands and showed that CD4+ T-cell-derived IFN-γ induces antiviral signaling in salivary gland cells of nonhematopoietic origin . Here, we use a recently developed TCR transgenic mouse recognizing the MCMV-specific CD4+ T-cell epitope M25411–425 to study the development and antiviral functions of MCMV-specific CD4+ T cells during primary MCMV infection on a monoclonal level. The I-Ab-restricted epitope M25411–425 is an immunodominant epitope of MCMV in the C57BL/6 genetic background . We show in adoptive transfer experiments that M25-specific CD4+ T cells expand upon MCMV infection display an effector phenotype and secrete antiviral cytokines IFN-γ and TNF. In a model of adoptive T-cell therapy in immunocompromised mice, we tested the in vivo antiviral potency of M25-specific CD4+ T cells and demonstrate a protective role during acute pathogenic MCMV infection. We identified IFN-γ to be the crucial effector molecule mediating CD4+ T-cell antiviral activity in vivo. Moreover, we show that co-transfer of low doses of activated M25-specific CD4+ T cells and M38-specific CD8+ T cells synergizes to control pathogenic MCMV infection in immunocompromised mice. Our data establish a new role for virus-specific CD4+ T cells in immune surveillance of acute pathogenic CMV infection, pointing to their largely unexplored therapeutic potential in preventing CMV disease in immunocompromised hosts.
Expansion, activation, and phenotype of M25-specific CD4+ T cells during acute MCMV infection
To study the characteristics of MCMV-specific CD4+ T cells during primary MCMV infection, we used naive CD4+ T cells isolated from the TCR transgenic mouse line M25-II with specificity for the immunodominant CD4+ T-cell epitope M25411–425 [17, 18]. We transferred 105 naive CD45.1+ CD4+ T cells to C57BL/6 (B6) mice 1 day prior to MCMV infection and followed the expansion of M25-specific CD4+ T cells in the blood of recipient mice at different time points post infection (p.i.) (Fig. 1A). M25-specific CD4+ T cells, detected by expression of the CD45.1 congenic marker, expanded during the first days of infection, reaching the peak at day 8 p.i., thereafter contracted and remained stable at low frequencies up to 1 year after infection (data not shown). Furthermore, frequencies and total numbers of M25-specific CD4+ T cells in various organs were analyzed at day 8 p.i. (Fig. 1B) when the abundance of M25-specific CD4+ T cells in the total CD4+ T-cell pool was the most prominent in peripheral tissues such as lungs and liver. Analysis of the activation status of M25-specific CD4+ T cells in inguinal LNs (iLNs), blood, and lungs revealed a CD44hiCD62Llow phenotype, consistent with activated effector CD4+ T cells (Fig. 1C, upper panel). This was further confirmed by high expression of CD11a and CD49d surface markers (Fig. 1C, middle panel). Moreover, assessment of the proliferative capacity of M25-specific CD4+ T cells (Fig. 1C, lower panel) showed high expression of the proliferation marker Ki67 and low expression of the antiapoptotic molecule Bcl-2, indicating a high proliferative activity of M25-specific CD4+ T cells 8 days p.i.
We further analyzed the phenotype of M25-specific CD4+ T cells in more detail by assessing the expression levels of CD127 (IL-7R-α chain) and Klrg-1 (Fig. 2A). In comparison with endogenous CD4+ T cells, M25-specific CD4+ T cells were CD127low in all organs tested and showed elevated Klrg-1 expression levels, the phenotypic features of short-lived effector T cells. Moreover, we analyzed the expression levels of PD-1, Tim-3, and LAG-3 (Fig. 2A), all of which are known to negatively regulate T-cell effector function during certain chronic infections with highly replicative viruses . We have not observed any differences in expression levels of Tim-3 and LAG-3 on M25-specific CD4+ T cells compared with those of endogenous CD4+ T cells, with the exception of PD-1 being upregulated on M25-specific CD4+ T cells compared with endogenous CD4+ T cells in the lungs and iLNs. Considering that PD-1 is upregulated on activated effector T cells during a number of acute viral infections [20, 21], elevated levels of PD-1 on M25-specific CD4+ T cells likely indicate an activated, rather than exhausted phenotype. Further, follicular Th cells have recently been described to be elevated in the context of chronic viral infections and virus-specific effector CD4+ T cells during acute LCMV infection were reported to be composed of Th1 and Tfh effector cells, with Th1 cells expressing high levels of PSGL-1 and Ly6C and Tfh cells expressing low levels of both markers . We analyzed the expression levels of these markers (Fig. 2B) and found that the majority of M25-specific CD4+ T cells in the blood and lungs were PSGL-1hi Ly6Chi, resembling the Th1 phenotype, whereas in the iLNs the major part of M25-specific CD4+ T cells was PSGL-1low Ly6Clow, resembling the Tfh phenotype.
Considering that CMV-specific CD4+ T cells were shown to produce the effector cytokines IFN-γ and TNF [10, 17], as well to exert MHC-II-restricted cytolytic activity in vitro during latent CMV infection , we analyzed the cytokine secretion profile of M25-specific CD4+ T cells at day 8 p.i. and found that the majority of M25-specific CD4+ T cells secreted IFN-γ and TNF in response to M25 peptide ex vivo restimulation (Fig. 2C). The majority of M25-specific CD4+ T cells producing IFN-γ also degranulated in response to M25 peptide compared to the unstimulated control, as assessed by CD107a surface expression. Moreover, M25-specific CD4+ T cells expressed elevated levels of granzyme B, as compared to endogenous CD4 T cells (Fig. 2E). Of note, concomitant secretion of effector cytokines and the ability to degranulate was significantly higher in M25-specific CD4+ T cells compared to endogenous CD4+ T cells (Fig. 2C), all together indicating that the majority of virus-specific effector CD4+ T cells are polyfunctional.
Taken together, these data show that during primary MCMV infection, virus specific-CD4+ T cells expand upon cognate antigen encounter, differentiate to short-lived effector T cells and home to peripheral tissues where they display a Th1-like effector phenotype, secrete IFN-γ and TNF effector cytokines, and degranulate upon antigen encounter.
Antiviral functions of M25-II cells during primary MCMV infection
Considering the effector phenotype and function of M25-specific CD4+ T cells, we sought to directly dissect the antiviral role of virus-specific CD4+ T cells in protection from pathogenic MCMV infection. We modified a previously described method of adoptive transfusion of T cells from MCMV immune mice to partially hemoablated recipients (Fig. 3) . M25-specific CD4+ T cells were activated in vivo by transferring CD4+ T cells from naive M25-II mice to recipient mice followed by MCMV infection (Fig. 3A and Supporting Information Fig. 1A). On day 8 p.i., in vivo primed M25-specific CD4+ T cells were isolated from the lungs of infected donor mice, purified by FACS sorting, and transferred in graded numbers to sublethally irradiated recipient mice (Fig. 3A and Supporting Information Fig. 1B). One day after the cell transfer, mice were infected subcutaneously with MCMV and virus loads were assessed at day 12 p.i. While recipient mice that received 104 or 105 MCMV-primed M25-specific CD4+ T cells could not control the infection (Fig. 3B), recipient mice that received 106 MCMV-primed M25-specific CD4+ T cells controlled viral replication significantly better (Fig. 3B).
This protective effect was prominent in all organs tested. In order to verify that observed protective effect was driven by antigen specificity of virus-specific CD4+ T cells, we activated LCMV-specific CD4+ T cells (SMARTA) in vivo with a low dose of LCMV-WE infection, transferred 106 LCMV-immune SMARTA cells to sublethally irradiated recipients that were infected with MCMV 1 day posttransfer, and assessed the viral loads at day 12 p.i. The group of mice that was transfused with 106 LCMV-immune SMARTA cells could not control MCMV replication (Supporting Information Fig. 2). Taken together, these data clearly indicated a direct protective role for MCMV-specific CD4+ T cells in control of pathogenic MCMV replication in an antigen-dependent manner.
Mechanism of CD4+ T-cell-mediated control of pathogenic MCMV infection
Considering our observation that virus-specific CD4+ T cells can mediate systemic protection from pathogenic MCMV infection, we sought to identify the underlying antiviral mechanism exerted by M25-specific CD4+ T cells. We first tested whether CD4+ T-cell-mediated secretion of IFN-γ plays a physiological role in antiviral MCMV immunity (Fig. 4). To this end, using the same approach as above, we transferred in vivo activated M25-specific CD4+ T cells to sublethally irradiated IFNγR−/− or B6 mice. Strikingly, in the case when recipients could not sense IFN-γ, CD4+ T-cell mediated virus control was diminished in all the organs tested, indicating a vital role for CD4+ T-cell secreted IFN-γ in antiviral immunity toward MCMV infection (Fig. 4A).
Next, we tested whether perforin-dependent cytotoxicity exerted by M25-II CD4+ T cells plays a relevant role for control of MCMV replication in vivo. To obtain M25-specific CD4+ T cells selectively deficient in perforin production, M25-II mice were crossed to perforin-deficient mice (Prf−/−). In vivo activated Prf−/−xM25-II CD4+ T cells were transferred into sublethally irradiated recipients followed by MCMV infection and the virus loads were assessed at day 12 p.i. (Fig. 4B). A protective effect of in vivo activated M25-specific CD4+ T cells deficient in perforin production was present in all tested organs (Fig. 4B). Nevertheless, the protective effect of perforin-deficient M25-specific CD4+ T cells in lungs (Fig. 4B), although present, seemed reduced when compared with that of perforin-sufficient M25-specific CD4+ T cells (Fig. 4A), indicating that a contribution of perforin-mediated cytotoxicity exerted by M25-specific CD4+ T cell plays a less prominent role than IFN-γ-mediated protection. Taken together, these data establish a pivotal role for CMV-specific CD4+ T cells in immuosurveillance of MCMV infection, manifested by exertion of direct effector mechanisms via production of IFN-γ and to a lesser extent by perforin-mediated cytotoxicity.
Co-transfer of virus-specific CD4+ and CD8+ T cells enhances protection from MCMV infection
Taking into account the protective role of activated MCMV-specific CD4+ T cells in protection from MCMV disease and the well-established protective role of virus-specific CD8+ T cells , we sought to test whether the co-transfer of a low, otherwise nonprotective dose of virus-specific CD4+ T cells together with low doses of virus-specific CD8+ T cells would lead to enhanced MCMV control in immunocompromised mice. This setup is of particular importance for human T-cell immunotherapy, considering the technical difficulties of isolating large numbers of virus-specific T cells from human peripheral blood. To this end, as a source of MCMV-specific CD8+ T cells, we used CD8+ T cells isolated from the previously described TCR transgenic MAXI mouse with specificity for the CD8+ T-cell epitope of the M38 MCMV protein . First, we tested the antiviral potency of in vivo activated adoptively transferred M38-specific CD8+ T cells in immunocompromised mice subsequently infected with MCMV (Supporting Information Fig. 3). Doses of 106 and 105 M38-specific CD8+ T cells provided antiviral protection in peripheral tissues of immunocompromised mice, whereas 104 M38-specific CD8+ T cells did not confer protection in any of the organs tested except in the spleen. Of note, none of the transferred doses of M38-specific CD8+ T cells could mediate protection in the salivary glands, further confirming previous findings of the crucial role of CD4+ T cells in mucosal immunity toward CMV in this particular tissue [15, 16]. Next, to test whether the co-transfer of M25-specific CD4+ T cells with M38-specific CD8+ T cells would synergize in antiviral immunity, graded numbers of in vivo activated, FACS sorted M25- or M38-specific cells were transferred in combination or alone into sublethally irradiated mice, followed by MCMV infection (Fig. 5A). Interestingly, the co-transfer of a nonprotective dose of 104 activated M25-specific CD4+ T cells together with a low dose of 104 activated M38-specific CD8+ T cells enhanced MCMV control in different peripheral tissues, such as spleen and lungs compared with control groups (no transfer or CD4+, CD8+ T cells transferred alone) (Fig. 5B). Immune surveillance in salivary glands remained critically dependent on CD4+ T cells, since only a protective dose of 106 activated M25-II cells could mediate control of lytic MCMV replication (Fig. 5C). Thus, low doses of MCMV-specific CD4+ T cells together with MCMV-specific CD8+ T cells mediate in a synergistic manner potent control of lytic virus replication during pathogenic MCMV infection in immunocompromised mice.
Herpes viral infections cause severe disease in immunocompromised individuals with impaired T-cell immunity, such as in AIDS patients or transplant recipients undergoing immunosuppressive treatment [27, 28]. Beside the appreciated importance of virus-specific CD8+ T-cell immunity, an increasing number of reports support the importance of virus-specific CD4+ T cells in reducing the viral and disease burden in such individuals, indicating that virus-specific CD8+ T cells alone are not sufficient to prevent disease manifestation. Moreover, AIDS patients with low CD4+ T-cell counts are particularly susceptible to herpes virus reactivation with severe clinical symptoms, further confirming the importance of CD4+ T cells in prevention of viral reactivation . The reconstitution of T-cell immunity in immunocompromised patients undergoing herpes virus reactivation through infusion of ex vivo expanded virus-specific T cells further points toward a protective role of CD4+ T cells for a number of herpes viral infections such as varicella Zoster virus, CMV, and epstein-Barr virus [14, 30-32]. Studies of CMV-specific CD4+ T cells during primary HCMV infection are scarce due to absence of clinical symptoms and difficulties to detect the initial timing of infection. Moreover, analyses of CMV-specific CD4+ T-cell responses on an epitope-specific level are missing, due to technical difficulties tracking virus-specific CD4+ T-cell responses irrespective of their functional capacities. In a mouse model of pathogenic MCMV infection of hemoablated recipients, CD4+ T cells were considered not to display direct antiviral activity  and thus were generally not considered for T-cell immunotherapy protocols. However, studies of HCMV-specific CD4+ T cells in humans [34, 35] and recent clinical studies  have indicated a possible protective role of CMV-specific CD4+ T-cell-mediated immunity. In this study, using a recently developed TCR transgenic mouse recognizing an MCMV-specific CD4+ T-cell epitope within the M25 protein, we characterized the development, phenotype, and antiviral functions of virus-specific CD4+ T cells during primary CMV infection on a monoclonal level. We show that M25-specific CD4+ T cells display an effector phenotype and exert direct antiviral activity toward pathogenic MCMV infection in immunocompromised mice in an IFN-γ-dependent mechanism. Moreover, we show that low doses of virus-specific CD4+ T cells together with activated virus-specific CD8+ T cells synergize in mediating protective immunity in immunocompromised MCMV-infected mice. Furthermore, also larger doses of M25-specific CD4+ T cells alone provided protection against MCMV infection. One reason for the requirement of relatively large numbers of CD4+ T cells to afford protection might be a limiting in vivo presence of MCMV antigen-presenting MHC class II positive APCs.
Our data provide direct evidence for a role of MCMV-specific CD4+ T cells in adoptive T-cell immunotherapy of CMV disease. Studies of primary CMV infection in CMV seronegative patients undergoing renal transplantation from seropositive donors show that HCMV-specific CD4+ T cells proliferate, display a phenotype of recently activated T cells, and produce IFN-γ and TNF effector cytokines . In contrast, a very recent study showed that the functionality of HCMV-specific CD4+ T cells may be impaired during primary infection, since CMV-specific CD4+ T cells isolated from pregnant women undergoing primary CMV infection produced lower amounts of IL-2 and displayed reduced functional avidity compared with CMV-specific CD4+ T cells during chronic infection . However, it is questionable whether the reduced ability to produce IL-2 during primary viremia is indicative of impaired function as low IL-2 production is a commonly observed phenotype of CD4+ T cells during the acute viraemic stage of an infection. We now characterized the development and phenotype of MCMV-specific CD4+ T cells during primary CMV infection and identify that MCMV-specific CD4+ T cells, comparable to other acute viral infections, expand upon the cognate antigen encounter, show features of activated short-lived effector T cells at the peak of expansion, migrate to peripheral tissues, secrete the effector cytokines IFN-γ and TNF, and degranulate, with the majority of CMV-specific CD4+ T cells being multifunctional.
There is growing evidence for a direct effector role of virus-specific CD4+ T cells in controlling CMV infection and disease. The appearance of HCMV-specific IFN-γ-producing CD4+ T cells in asymptomatic recipients of CMV-positive allotransplants was specifically correlated with containment of viral control  and CD4+ T-cell lines specific for certain HCMV CD4+ T-cell antigens were shown to be cytolytic [23, 35]. Interestingly, a very recent study reported the existence of high numbers of multifunctional pp65-specific CD4+ T cells in bronchoalveolar lavage-derived lung mononuclear cells of lung transplant recipients with chronic CMV infection, which correlated with mucosal viral control . However, in mice, with the exception of the salivary gland tissue, CD4+ T cells were so far considered not to exert direct antiviral functions in other organs upon MCMV infection . These studies relied on experimental protocols where effector cell populations comprising total CD4+ T cells were isolated from MCMV-immune donor mice and were used for subsequent adoptive transfer experiments . Considering the low frequency of endogenous IFN-γ-producing MCMV-specific CD4+ T effector cells in peripheral tissues , in particular in the spleen, this protocol presumably did not retrieve sufficient numbers of IFN-γ-producing virus-specific CD4+ T cells. The fact that only a high number of M25-specific CD4+ T cells mediated antiviral protection is in line with this assumption and represents a challenging factor for human immunotherapy protocols, which largely rely on enrichment of virus-specific CD4+ T cells from CMV-immune donor peripheral blood. Nevertheless, our findings that the co-transfer of activated M25-specific CD4+ T cells together with activated M38-specific T cells mediate potent protection from pathogenic MCMV infection, already when co-infused in otherwise not protective doses (Fig. 5), suggest a possibility for exploration of virus-specific CD4+ T cells in T-cell immunotherapy of pathogenic CMV infection.
Materials and methods
This study was performed in agreement with the guidelines of the animal experimentation law (SR 455.163; TVV) of the Swiss Federal Government. The protocol was accepted by the Cantonal Veterinary Office of the canton Zurich, Switzerland (Permit numbers 145/2008, 109/2011).
Mice, viruses, and peptides
C57BL/6J, IFN-γR−/−, C57BL/6N-Tg(TCRaM25,TCRbM25)424Biat(M25-II), and C57BL/6N-Tg(TcraM38,TcrbM38)329Biat (Maxi) mice were housed in the Rodent Center HCI of ETHZ under specific pathogen-free conditions. M25-II mice were crossed with PKOB mice to obtain M25-II transgenic cells deficient in perforin production. Recombinant MCMV Δm157 was previously described  and produced in C57BL/6 mouse embryonic fibroblasts (MEFs). Viral titers were quantified by standard virus plaque assay . Mice were infected i.v. with MCMV Δm157 at a dose of 5 × 106 PFU or subcutaneously into the foot pad using 105 PFU of MCMV Δm157. The MCMV-derived M25aa411–425 peptide was purchased from NeoMPS (Strasbourg, France).
All listed fluorochrome-conjugated antibodies were purchased either from Biolegend (Lucerna-Chem AG, Luzern, Switzerland) or from BD Biosciences (Allschwil, Switzerland): anti-CD4, anti-CD45.1, anti-CD44, anti-CD62L, anti-CD11a, anti-CD43, anti-Ki67, anti-Bcl2, anti-CD127, anti-Klrg-1, anti-PD-1, anti-LAG-3, anti-Tim3, anti-PSGL1, anti-Ly6C, anti-TNF, anti-IFN-γ, anti-CD107, and anti-granzyme B.
Lymphocyte stimulations and flow cytometry
Lymphocytes were harvested from spleen, liver, lung, LNs, and salivary gland as previously described [17, 41]. Cytokine production was measured by intracellular cytokine staining following peptide stimulation as described in . Briefly, lung CD4+ T cells were stimulated with 3 μg/mL of M25aa411–425, in the presence of brefeldin A (10 μg/mL, Sigma Aldrich) at 37°C for 6 h. Thereafter, cell surface staining was performed for 20 min at 4°C using directly conjugated antibodies, followed by fixation and permeabilization using Fix/Perm solution (FACSLyse diluted to 2× concentration and 0.05% Tween 20) for 10 min at room temperature. After a washing step, cells were stained for 20 min at 4°C with directly labeled anti-IFN-γ and anti-TNF antibodies. Flow cytometric analyses were done on a LSRII flow cytometer (BD Bioscience) with FACS DIVA software (BD Bioscience). Flow cytometry standard (FCS) data were analyzed with FlowJo software (Treestar, San Carlos, CA, USA).
Adoptive T-cell transfusion in immunocompromised mice
C57BL/6J donor mice were transfused with magnetically sorted (MACS) CD4+ or CD8+ T cells isolated from naive CD45.1+ M25-II or CD45.1+ MAXI mice, respectively. Mice were infected with 5 × 106 MCMVΔm157 i.v. 1 day posttransfer and lung lymphocytes were isolated from the donor mice at day 8 p.i. CD45.1+ CD4+ or CD45.1+ CD8+ T cells were FACS sorted from lung lymphocytes and transfused to sublethally total body γ-irradiated (750 rad) C57BL/6J and IFNγR−/- mice. Recipient mice were infected subcutaneously 1 day post transfer with 105 MCMVΔm157. Irradiated mice were treated with a 1:250 dilution of 24% Borgal (Intervet, Boxmeer, Netherlands) in drinking water. FACS sorting was executed on a FACSAriaTM (BD Bioscience) cell sorter supported by FACS DIVA software.
We would like to thank the members of the Oxenius lab for critical discussion. We are very grateful to Nathalie Oetiker and Franziska Wagen for excellent technical assistance and FACS sorting. This work was supported by the Swiss National Science foundation (grants 310030_129751 and 310030_146140 to AO) and ETH Zurich.
Conflict of interest
The authors declare no financial or commercial conflict of interest.