Unraveling the mechanism of [4Fe‐4S] cluster assembly on the N‐terminal cluster binding site of NUBP1

Abstract [4Fe‐4S]2+ cluster assembly in human cytosol requires both a [2Fe‐2S] cluster chaperone being able to donate two [2Fe‐2S]2+ clusters and an electron donor providing two electrons to reductively couple the two [2Fe‐2S]2+ clusters into a [4Fe‐4S]2+ cluster. The mechanism through which the cytosolic [4Fe‐4S]2+ cluster assembly works is still not defined. Here, we show that a hetero‐tetrameric complex formed by two molecules of cluster‐reduced [2Fe‐2S]+ 2‐anamorsin and one molecule of dimeric cluster‐oxidized [2Fe‐2S]2+ 2‐GLRX32 orchestrates the assembly of a [4Fe‐4S]2+ cluster on the N‐terminal cluster binding site of the cytosolic protein NUBP1. We demonstrate that the hetero‐tetrameric complex is able to synergically provide two [2Fe‐2S]2+ clusters from GLRX3 and two electrons from anamorsin for the assembly of the [4Fe‐4S]2+ cluster on the N‐terminal cluster binding site of NUBP1. We also showed that only one of the two [2Fe‐2S] clusters bound to anamorsin, that is, that bound to the CX8CX2CXC motif, provides the electrons required to form the [4Fe‐4S]2+ cluster. Our study contributes to the molecular understanding of the mechanism of [4Fe‐4S] protein biogenesis in the cytosol.


| INTRODUCTION
In the cytosol of eukaryotes, the maturation of [4Fe-4S] proteins is carried out by a cytosolic iron-sulfur (Fe-S) cluster assembly (CIA) machinery Netz et al., 2014). An early proposed component of the CIA machinery is the cytosolic monothiol glutaredoxin GLRX3 (Braymer et al., 2021), a protein that dimerizes upon binding of two bridged [2Fe-2S] 2+ clusters ([2Fe-2S] 2+ 2 -GLRX3 2 , hereafter) (Camponeschi et al., 2020;Haunhorst et al., 2010). Among the proposed functions of GLRX3, there is its action as a [2Fe-2S] cluster chaperone by donating its two [2Fe-2S] 2+ clusters to various partner proteins (Banci, Camponeschi, et al., 2015;Frey et al., 2016). One of these is NUBP1 (Camponeschi et al., 2020), another component of the CIA machinery, acting as a scaffold for the assembly of [4Fe-4S] clusters together with its homologous NUBP2 (Stehling et al., 2008;Stehling et al., 2018). NUBP1 and NUBP2 Beatrice Bargagna and Sara Matteucci contributed equally to this work. belong to the deviant Walker A family NTPases (Grossman et al., 2019), and are both required for the CIA machinery (Stehling et al., 2008;Stehling et al., 2018). The two proteins form homo-and heterodimeric complexes, bridging a [4Fe-4S] cluster bound to a CxxC motif close to the C-termini of both proteins (Netz et al., 2012;Stehling et al., 2018). NUBP1 differs from NUBP2 as it harbors an N-terminal conserved CX 13 CX 2 CX 5 C motif not present in NUBP2, which tightly binds a further [4Fe-4S] 2+ cluster (Camponeschi et al., 2020;Netz et al., 2012;Stehling et al., 2008). This N-terminal iron-sulfur cluster is required for the CIA processes (Netz et al., 2012), but its role is not well understood. In the current working model, the Nterminal iron-sulfur cluster remains stably bound to NUBP1 during the formation of the C-terminal [4Fe-4S] 2+ cluster which is transiently bridged on the heterodimeric NUBP1-NUBP2 complex as it is transferred along the CIA machinery pathway. Thus, at variance with the C-terminal [4Fe-4S] 2+ cluster, the N-terminal Fe-S cluster binding motif of NUBP1 behaves like a final acceptor of a [4Fe-4S] cluster. The maturation of this [4Fe-4S] cluster in yeast NUBP1 has been shown to be independent of the proteins acting later in the CIA machinery (Netz et al., 2007), thus differing markedly from the maturation of [4Fe-4S] clusters into the majority of cytosolic target proteins that depends on the other components of the CIA machinery (Braymer et al., 2021;Ciofi-Baffoni et al., 2018).
We recently investigated the mechanism for the formation of the N-terminal [4Fe-4S] cluster of NUBP1 in which two [2Fe-2S] 2+ clusters, donated by [2Fe-2S] 2+ 2 -GLRX3 2 homodimer, are reductively coupled to form a [4Fe-4S] 2+ cluster, thanks to the electrons donated in vitro by glutathione (GSH) (Camponeschi et al., 2020). However, a possible physiological electron donor to form the [4Fe-4S] 2+ cluster in vivo at the N-terminal site of NUBP1 is anamorsin, which is a partner protein of GLRX3 (Saito et al., 2011), and which receives electrons from another CIA component, the NADPH-dependent diflavin reductase NDOR1 (Banci, Bertini, et al., 2013;Netz et al., 2010). This role of anamorsin is supported by the finding that the yeast homolog of anamorsin, that is, Dre2, is required for the Fe-S cluster assembly on Nbp35, but not on Cfd1, the yeast protein homologs of NUBP1 and NUBP2, respectively (Netz et al., 2010). As further support of anamorsin being the electron donor required to assemble the Nterminal [4Fe-4S] 2+ cluster of NUBP1, it has been observed that, in Arabidopsis thaliana, the anamorsin homolog Dre2 specifically interacts with the NUBP1 homolog, NBP35 (Bastow et al., 2017). On the other hand, a direct experimental evidence of anamorsin acting as electron donor to assemble the [4Fe-4S] cluster at the N-terminal site of NUBP1 is lacking.
Anamorsin is an iron-sulfur cluster binding protein, structurally composed of a N-terminal S-adenosyl methionine methyl transferase-like domain connected via a flexible linker to a largely unstructured C-terminal cytokine-induced apoptosis inhibitor 1 (CIAPIN1) domain (Banci et al., 2011;Soler et al., 2012;Song et al., 2014). The latter contains two conserved cysteinerich motifs (a CX 8 CX 2 CXC motif [M1, hereafter], followed by a CX 2 CX 7 CX 2 C motif [M2, hereafter]), each binding a [2Fe-2S] cluster (Banci et al., 2011;Matteucci et al., 2021). The [2Fe-2S] cluster bound at the M1-motif is reduced by the diflavin reductase NDOR1 and the same reaction occurs between the yeast homolog proteins Dre2 and Tah18 Netz et al., 2010). Conversely, the M2-motif has been found to bind different types of clusters in different organisms, that is, a [2Fe-2S] cluster in humans Matteucci et al., 2021) and a [4Fe-4S] cluster in yeast (Netz et al., 2016;Zhang et al., 2008;Zhang et al., 2017), and its function is still unknown in any organism (Netz et al., 2016;Zhang et al., 2017). The M2-bound cluster is, however, not able to receive electrons from the diflavin reductase NDOR1 neither in humans nor in yeast Netz et al., 2010), indicating that this Fe-S cluster does not have a redox function in the electron transfer process.
In this work, we characterized the mechanism for the assembly of the [4Fe-4S] 2+  Methods for details). After $1 hour, His 6 -tagged NUBP1 was isolated from GLRX3 and anamorsin, using Ni 2+affinity chromatography. The protein content of the collected fractions was assessed by polyacrylamide gel electrophoresis (SDS-PAGE) ( Figure S1), and the fraction containing isolated His 6 -tagged NUBP1 protein was characterized by UV-visible (UV-vis) and NMR spectroscopies, and by performing acid-labile sulfide and iron quantification. The UV-vis spectrum of isolated His 6tagged NUBP1 is characteristic of a [4Fe-4S] 2+ -bound protein, with the presence of a broad absorbance band at $410 nm ( Figure 1A). The paramagnetic 1D 1 H NMR spectrum of the isolated His 6 -tagged NUBP1 showed four hyperfine shifted signals in the 18-11 ppm spectral region ( Figure 1B), whose chemical shift values and temperature dependence ( Figure S2) are typical of βCH 2 of cysteines bound to an oxidized [4Fe-4S] 2+ cluster Bertini et al., 1994). These signals are well superimposable with those previously observed for the four N-terminal cysteines of NUBP1 bound to an oxidized [4Fe-4S] 2+ cluster (Camponeschi et al., 2020). Overall, these data indicate that the [2Fe-2S] 2+ 2 -GLRX3 2 /[2Fe-2S] + 2 -anamorsin mixture is able to form a [4Fe-4S] 2+ cluster on the N-terminal site of His 6 -tagged NUBP1. Acid-labile sulfide and iron analysis showed the presence of $0.6 [4Fe-4S] clusters per His 6 -tagged NUBP1 molecule (Table 1).
To further support that a [4Fe-4S] cluster is assembled at the N-terminal site of NUBP1 and to rule out that no [4Fe-4S] cluster is assembled at the C-terminal site of NUBP1, we mixed under anaerobic conditions untagged [2Fe-2S] 2+ 2 -GLRX3 2 and untagged, cluster-reduced [2Fe-2S] + 2 -anamorsin with a His 6 -tagged variant of NUBP1 where the two cysteines in the CPXC motif where mutated into alanines, thus containing only the Nterminal cluster binding motif (NUBP1 NT , hereafter) or with a His 6 -tagged construct of NUBP1 where the first 37 N-terminal residues were deleted, thus containing only the C-terminal CPXC cluster binding motif (NUBP1 CT , hereafter). The UV-visible spectrum of isolated His 6 -tagged NUBP1 NT showed the presence of a broad absorption band at $410 nm ( Figure S3A), indicating the binding of a [4Fe-4S] 2+ cluster to the protein. Acid-labile sulfide and iron analysis showed the presence of $0.7 [4Fe-4S] clusters per NUBP1 NT molecule (Table 1), which is comparable with what observed for full-length wild-type NUBP1. Conversely, the UV-vis spectrum of isolated His 6 -tagged NUBP1 CT did not show a significant absorption band arising from a bound Fe-S cluster ( Figure S3B). Overall, these data indicate that the [2Fe-2S] 2+ 2 -GLRX3 2 /[2Fe-2S] + 2 -anamorsin mixture is able to assemble a [4Fe-4S] 2+ cluster at the N-terminal cluster binding site of NUBP1, but not at the Cterminal site.
UV-vis spectroscopy was also performed on the [2Fe-2S] 2+ 2 -GLRX3 2 /[2Fe-2S] + 2 -anamorsin mixture before its incubation with His 6 -tagged apo NUBP1 and then after its separation from the latter. The spectrum of the GLRX3/anamorsin mixture before incubation with NUBP1 shows faint bands at 330, 420, 510, and 580 nm ( Figure 2, cyan line), which are typical of cluster-reduced  2 -bound form of anamorsin ( Figure 2A, red line). These data clearly showed that anamorsin gets oxidized after incubation with His 6 -tagged NUBP1. Moreover, the spectrum of the GLRX3/anamorsin mixture after incubation with His 6 -tagged apo NUBP1 does not show the bands at 510 and 580 nm, typical of the [2Fe-2S] 2+ clusters bound to GLRX3, indicating that GLRX3 has no bound cluster anymore and thus that it transfers its clusters to NUBP1.
Overall, these data indicate that reduced [2Fe-2S] + 2anamorsin is able to promote [4Fe-4S] 2+ assembly on the T A B L E 1 Iron and acid-labile sulfide quantification of His 6 -tagged NUBP1 after cluster transfer/assembly reaction.

Sample
Fe Since it was not possible to selectively reduce either the M1-bound or the M2-bound [2Fe-2S] cluster of anamorsin, both [2Fe-2S] clusters were reduced in the previous set of experiments, and therefore both clusters might be involved in the electron transfer process. Thus, we investigated the role of the M1-and M2-motifs in the electron transfer by performing a new set of experiments, analogous to those performed with wild-type anamorsin (WT-anamorsin, hereafter), but in the presence of: (i) a construct of anamorsin lacking the last 49 C-terminal residues, and therefore containing only the M1-motif (M1-anamorsin, hereafter), and (ii) a mutant containing only the M2-motif, as the four cysteines of the M1-motif were mutated into alanines (M2-anamorsin, hereafter). Each of the cluster-binding sites in the anamorsin variants were previously shown by us to maintain the same cluster coordination and electronic properties of the wild-type protein Matteucci et al., 2021), indicating that the two cys-rich motifs can independently bind a [2Fe-2S] cluster, and thus that the presence of one cluster does not affect the redox properties of the other cluster in the molecule. After incubation of His 6 -tagged NUBP1 with the [2Fe-2S] 2+ 2 -GLRX3 2 /[2Fe-2S] + -M1-anamorsin mixture followed by its isolation, His 6 -tagged NUBP1 showed a broad absorbance band at $410 nm in the UV-vis spectra, characteristic of the oxidized [4Fe-4S] 2+ clusterbound form of NUBP1 (Camponeschi et al., 2020) ( Figure 3A, red line). Accordingly, the UV-vis CD spectrum of His 6 -tagged NUBP1 was essentially featureless ( Figure S6), in agreement with the binding of an oxidized [4Fe-4S] 2+ cluster. Moreover, the four sharp hyperfine shifted signals, characteristic of the oxidized [4Fe-4S] 2+ cluster bound to the N-terminal site of NUBP1 (Camponeschi et al., 2020), were observed in the paramagnetic 1D 1 H NMR spectrum of His 6 -tagged NUBP1 isolated after the reaction ( Figure 3B, a). Acid-labile sulfide and iron analysis showed the presence of $0.7 [4Fe-4S] clusters per NUBP1 molecule (Table 1), which is comparable with what observed with WT-anamorsin. Therefore, these data reproduce those observed for WTanamorsin, indicating that the M1-cluster provides electrons to assemble the [4Fe-4S] 2+ cluster on the Nterminal site of NUBP1.
On the contrary, when [2Fe-2S] + -M2-anamorsin was used in the reaction, the UV-vis spectrum of the isolated His 6 -tagged NUBP1 showed absorption bands centered at 330 and at 420 nm, the latter with a shoulder at $460 nm ( Figure 3A, blue line). UV-vis CD spectrum of the isolated His 6 -tagged NUBP1 showed positive bands centered at 327 and 476 nm, and negative bands centered at 436 and 530 nm ( Figure S6). Both the UV-vis absorption and CD spectra are characteristic of a [2Fe-2S] 2+ cluster, thus indicating the transfer of a [2Fe-2S] 2+ cluster from GLRX3 to NUBP1, without the assembly of a significant amount of [4Fe-4S] 2+ cluster. The 1D 1 H NMR spectrum of isolated His 6 -tagged NUBP1 from the latter mixture is in full agreement with this result, showing indeed a broad, unresolved main signal in the 30-20 ppm region ( Figure 3B, b), typical of βCH 2 of cysteines bound to a [2Fe-2S] 2+ cluster (Banci et al., 1990;Banci et al., 2018), and very low intensity signals of βCH 2 of cysteines bound to the N-terminal [4Fe-4S] 2+ cluster in the 16-11 ppm region ( Figure 3B, b While both clusters of anamorsin are EPR-silent in their [2Fe-2S] 2+ oxidized states, having a S = 0 ground spin state ( Figure 4A, gray line), the S = 1/2 ground spin state of each of the two reduced M1-and M2-bound [2Fe-2S] + clusters have distinct EPR signals Matteucci et al., 2021), that allow to easily discriminate them. The EPR spectra of dithionitereduced [2Fe-2S] + 2 -WT-anamorsin ( Figure 4A, green line) and of the mixture composed of [2Fe-2S] 2+ 2 -GLRX3 2 and dithionite-reduced [2Fe-2S] + 2 -WTanamorsin before the incubation with apo NUBP1 ( Figure 4A, magenta line), showed, at 10 K and 1 mW, EPR features arising from two distinct rhombic EPR signals, with principal g values of 2.00, 1.96, 1.92 and of 2.01, 1.94, 1.89, that were previously assigned to the two reduced [2Fe-2S] + clusters bound to the M1 and M2 motifs of WT-anamorsin, respectively Matteucci et al., 2021), and that were observed in the EPR spectra of the two isolated clusterreduced M1-and M2-anamorsin constructs ( Figure 4A, blue and red lines, respectively). Upon incubation with   Figure 4A, black line). Conversely, the EPR signal originating from the reduced M2-bound [2Fe-2S] + cluster did not change in intensity, accounting for the majority of the EPR signal, as showed by the comparison with the EPR spectrum recorded on isolated dithionite-reduced M2-anamorsin ( Figure 4A, red line).
Analytical SEC was also applied to investigate changes in the quaternary structure of the heterotetrameric complex after its interaction with apo NUBP1. Specifically, the untagged hetero-tetrameric complex was incubated with His 6 -tagged apo NUBP1 and the products of the [4Fe-4S] cluster assembly reaction were analyzed by SEC and by SDS-PAGE before and after their separation by Ni 2+ -affinity chromatography. The mixture of the hetero-tetrameric complex and apo NUBP1 elutes as two peaks at 12.7 and 13.4 mL ( Figure 5C), corresponding respectively to a GLRX3/anamorsin hetero-complex and to His 6 -tagged NUBP1, as showed by SDS-PAGE ( Figure S1) and by analytical SEC of the two isolated species, which elute with the same elution volumes once separated by Ni 2+ -affinity chromatography ( Figure 5C, purple and magenta lines, respectively). Interestingly, after the [4Fe-4S] assembly reaction, the GLRX3/anamorsin complex elutes as a main single peak at 12.7 mL ( Figure 5C). This elution volume is lower than that observed for the starting hetero-tetrameric GLRX3-anamorsin complex, which is thus not present anymore. The apparent mass of this new peak (MW app = 142.7 kDa, Table S1) corresponds to that of the apo GLRX3/[2Fe-2S] 2 -WT-anamorsin heterodimeric complex ( Figure 5C, purple line). Accordingly, after separation, the Ni 2+ column-bound fraction containing His 6tagged [4Fe-4S] 2+ NUBP1 elutes as a main single peak at 13.4 mL, which is the same elution volume found for apo NUBP1 ( Figure 5C).

| DISCUSSION
In this work, we unraveled the mechanism of [4Fe-4S] 2+ cluster assembly on the N-terminal cluster binding site of NUBP1. We found that the two components of the CIA machinery GLRX3 and anamorsin are able to assemble the [4Fe-4S] 2+ cluster at the N-terminal site of NUBP1. We showed that [2Fe-2S] + -anamorsin acts as electron donor in the assembly of the  (Haunhorst et al., 2010;Li et al., 2012). Although our work does not address the specific sequence of events for the formation of the [4Fe-4S] 2+ cluster on NUBP1 involving the electron transfer between the M1-bound [2Fe-2S] + cluster of anamorsin and the [2Fe-2S] 2+ clusters of [2Fe-2S] 2 -GLRX3 2 , we can make some relevant considerations on this aspect. In a previous study by us, we showed that the interaction between the N-terminal domains of anamorsin and GLRX3 is essential for the formation of the complex between the two proteins, while the Fe-S cluster binding domains of [2Fe-2S] 2 -anamorsin and of [2Fe-2S] 2 -GLRX3 2 are not involved in any permanent interaction . This finding agrees with what we find now by analytical SEC, that is, that the redox state of the clusters bound to anamorsin does not impact on complex formation between the two proteins. On the other hand, it is expected that, during the [4Fe-4S] cluster assembly process, the cluster-binding CIAPIN1 domain of anamorsin approaches transiently the N-terminal cluster-binding region of NUBP1, as required to donate two electrons for the reductive coupling of the two GLRX3-donated [2Fe-2S] 2+ clusters into the [4Fe-4S] 2+ cluster. Indeed, transient interactions are typically observed in electron transfer processes (Prudêncio & Ubbink, 2004). The high structural flexibility of the linker and of the CIAPIN1 domain of anamorsin is well suited to allow the conformational changes required to drive the transient interaction between the reduced M1-bound [2Fe-2S] + cluster of anamorsin and the [2Fe-2S] 2+ clusters of GLRX3 to reduce the latter clusters and thus to form the [4Fe-4S] 2+ cluster on NUBP1. We also showed that the reduced [2Fe-2S] + cluster bound to the M2-motif of anamorsin does not have any role in the assembly of the [4Fe-4S] 2+ cluster on the N-terminal site of NUBP1, and that the electron transfer from the [2Fe-2S] + cluster bound to the M1 site occurs independently of the presence of the [2Fe-2S] cluster in the M2 site. This model is consistent with previous data showing that, in the physiological electron transfer chain composed of NDOR1 and anamorsin, the reduced FMN moiety of NDOR1 is able to transfer electrons exclusively to the cluster bound to the M1-motif of anamorsin, and not to the cluster bound to the M2-motif (Banci, Bertini, et al., 2013;Netz et al., 2010). In summary, these findings suggest a model where the [2Fe-2S] cluster bound to the M2-motif of anamorsin does not take part in the electron transfer chain required to assemble the N-terminal [4Fe-4S] 2+ cluster of NUBP1. Our data also showed that the [2Fe-2S] cluster bound to the M2-motif of human anamorsin is not involved in cluster transfer to NUBP1. In vivo data are required to address the specific functional role of M2 motif of anamorsin. The role of this second Fe-S cluster has been only investigated in yeast, where the M2-motif is essential for yeast cell viability and for cytosolic [4Fe-4S] proteins activity (Netz et al., 2016;Zhang et al., 2017). However, this functional data on yeast cannot be applied to human anamorsin as M2-motif of Dre2 binds a [4Fe-4S] cluster while anamorsin binds a [2Fe-2S] cluster, and this different cluster type can significantly differentiate the functional role of the M2-motif in the two organisms.
anamorsin is then able to receive electrons from the reduced FMN moiety of NDOR1, thus enabling it to perform its function as an electron donor within the CIA machinery. After GLRX3-driven anamorsin maturation, the next cellular step involves the interaction between two molecules of [ -cluster containing proteins, but this contrasts with the reported moderate deficiencies of cytosolic Fe-S cluster enzymes resulting from the depletion of GLRX3 in mammalian cells (Haunhorst et al., 2010). The most reasonable explanation for the relatively weak in vivo phenotype of GLRX3 is that GLRX3-dependent pathways described above by us can be likely bypassed in the cells via the activation of alternative pathways, that is, redundant systems can partially substitute for GLRX3 function in mammalian cells. This effect has been already proposed in the literature when the role of GLRX3 in providing Fe-S cluster to anamorsin was studied (Frey et al., 2016). It was indeed reported that anamorsin may be capable of acquiring Fe-S clusters from a source alternative to GLRX3, for example, the NEET family of [2Fe-2S] proteins (Tamir et al., 2015). Thus, it is possible that the same or alternative systems are activated for the formation of the N-terminal [4Fe-4S] cluster of NUBP1. A model where mitoNEET or miner1 substitute GLRX3 as [2Fe-2S] cluster donors, although still requiring experimental evidences, is possible for two reasons: 1. both mitoNEET and miner1 have molecular features similar to those of homodimeric GLRX3, which are those required to form a [4Fe-4S] cluster on NUBP1, that is, they have two easily transferable [2Fe-2S] clusters bound in a homodimer, as previously shown (Lipper et al., 2015); 2. both [2Fe-2S] clusters of mitoNEET are reduced by a complex composed by [2Fe-2S]-anamorsin and the FMN binding domain of NDOR1 via the formation of a transient complex that brings the [2Fe-2S] 2+ clusters of mito-NEET close to the reduced [2Fe-2S] + cluster bound to M1-motif of anamorsin (Camponeschi et al., 2017).

| CONCLUSIONS
This work contributes to the molecular understanding of the mechanism of [4Fe-4S] proteins biogenesis in the cytosol, showing that anamorsin and GLRX3 act as an electron donor and as a [2Fe-2S] cluster donor, respectively, in the assembly of the N-terminal [4Fe-4S] cluster of NUBP1. Our studies allowed us to propose that the dimeric [2Fe-2S] 2+ 2 -GLRX3 2 complex and two molecules of [2Fe-2S] 2 + -anamorsin form a hetero-tetrameric complex that acts as a component of the CIA machinery at its early stages. The hetero-tetrameric complex is able to provide two [2Fe-2S] 2+ clusters from the GLRX3 dimeric unit and two electrons, one from each reduced [2Fe-2S] + cluster bound to the M1-motif of anamorsin, to assemble a [4Fe-4S] 2+ cluster on the N-terminal site of NUBP1. On the contrary, the [2Fe-2S] cluster bound to the M2-motif of anamorsin is not involved in any step of this [4Fe-4S] cluster assembly process.

| Protein, iron and acid-labile sulfide quantification
Protein quantification was carried out with the Bradford protein assay, using BSA as a standard. Non-heme iron and acid-labile sulfide content was determined as described previously (Banci et al., 2011).

| Biochemical and spectroscopic UVvis, NMR and EPR methods
and discovery in structural biology," funded by the Horizon 2020 research, grant no. 871037. Open Access Funding provided by Universita degli Studi di Firenze within the CRUI-CARE Agreement.