In Vivo Adeno-Associated Viral Vector–Mediated Genetic Engineering of White and Brown Adipose Tissue in Adult Mice

Male ICR mice (8–12 weeks old), C57Bl/6J mice (9–13 weeks old), and B6.V-Lep^ob/OlaHsd (ob/ob) and BKS.Cg-+ Lepr^db/+ Lepr^db/OlaHsd (db/db) mice (8 weeks old) were used. Animal care and experimental procedures in this study were approved by the Universitat Autònoma de Barcelona Ethics Committee in Animal and Human Experimentation.

Single-stranded AAV vectors were produced by triple transfection of human embryonic kidney 293 cells and purified by a CsCl-based gradient method (27). Transgenes used were:

• Enhanced green fluorescent protein (GFP) driven by 1) the cytomegalovirus (CMV) early enhancer/chicken beta actin (CAG) promoter; 2) the CAG promoter with the addition of four tandem repeats of the mirT122a sequence (5′CAAACACCATTGTCACACTCCA3′), the mirT1 sequence (5′TTACATACTTCTTTACATTCCA3′) or both, cloned in the 3′ untranslated region of the expression cassette; 3) the short version of the adipocyte protein 2 (mini/aP2) promoter (28,29); or 4) the short version of the uncoupling protein-1 (mini/UCP1) promoter (30,31);

• Hexokinase 2 (HK2) driven by 1) the CMV promoter, 2) the mini/aP2 promoter, or 3) the mini/UCP1 promoter;

• Placental-derived secreted alkaline phosphatase (SeAP) driven by the mini/aP2 promoter;

• Vascular endothelial growth factor (VEGF)-164 driven by the mini/UCP1 promoter; and

• Red fluorescent protein (RFP), driven by the CMV promoter.

A noncoding plasmid carrying the CMV promoter, the mini/aP2 promoter, or the mini/UCP1 promoter and a multicloning site was used to produce null particles.

Mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg). For the intraepididymal (intrae) WAT delivery, a laparotomy was performed. To distribute the vector in the whole depot, each epididymal fat pad was injected twice with 50 µL AAV solution. For intrainterscapular (intrai)BAT and intrainguinal (intrai)WAT administrations, a longitudinal incision in the skin at the interscapular or inguinal area was performed, respectively. To distribute the vector in the whole depot, each interscapular BAT (iBAT) or inguinal WAT (iWAT) received four injections of 10 µL AAV solution using a Hamilton syringe. For the systemic administration, AAV vectors were diluted in 200 µL saline and injected into the lateral tail vein.

Tissues were fixed for 12–24 h in 10% formalin, embedded in paraffin, and sectioned. Sections were incubated overnight at 4°C with a goat anti-GFP antibody (Abcam, Cambridge, MA), with a rabbit anti-RFP antibody (Abcam), with a goat anti-CD105 antibody (R&D Systems Inc., Minneapolis, MN) or with a mouse anti–α-smooth muscle actin (SMA) antibody (Sigma-Aldrich, Saint Louis, MO). A biotinylated donkey anti-goat antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), a biotinylated goat anti-rabbit antibody (Pierce Biotechnology, Inc., Rockford, IL), or a biotinylated horse anti-mouse antibody (Vector Laboratories, Burlingame, CA) was used as a secondary antibody. Streptavidin Alexa Fluor 488 (Molecular Probes; Life Technologies Corp., Carslbad, CA) was used as fluorochrome, and Hoechst (Sigma-Aldrich) was used for nuclear counterstaining. Alternatively, an ABC peroxidase kit (Pierce) was used, and sections were counterstained in Mayer’s hematoxylin. Neutrophils were stained using a Naphthol AS-D Chloroacetate (Specific Esterase) kit (Sigma-Aldrich).

Tissues were disrupted in 1 mL lysis buffer (50 mmol/L Tris, 1% Nonidet P40, 0.25% sodium deoxycholate, 150 mmol/L NaCl, 1 mmol/L EDTA, in PBS, pH 7.4, sterile filtered) with a tissue homogenizer and incubated for 10 min at room temperature. After incubation, samples were centrifuged at 14,000 rpm for 10 min. GFP content was measured by luminescence in 100 µL supernatant using a 488-nm excitation wavelength and 512-nm emission wavelength (Flx800; BioTek Instruments, Inc., Winooski, VT). GFP content values were corrected by total protein content of the sample.

AAV-transduced adipocytes were isolated using a modification of the Rodbell method (32). Isoflurane-anesthetized mice were killed by decapitation, and epididymal WAT (eWAT) was minced and digested at 37°C in Krebs-Ringer bicarbonate HEPES buffer (KRBH) containing 4% BSA (fatty acid-free), 0.5 mmol/L glucose, and 0.5 mg/mL collagenase type II (C6885; Sigma-Aldrich) during 35–45 min. Fat cells were isolated by gentle centrifugation and washed three times with fresh collagenase-free KRBH without glucose.

Total RNA was obtained from isolated adipocytes or adipose tissue samples using QIAzol Lysis Reagent (Qiagen, Hilden, Germany) and RNeasy Lipid Tissue Minikit (Qiagen). To eliminate residual viral genomes (vg), total RNA was treated with DNAseI (Qiagen). For RT-PCR, 1 µg RNA was reverse-transcribed using the Transcriptor First Strand cDNA Synthesis kit (Roche Diagnostic GmbH, Roche Applied Science, Mannheim, Germany). Quantitative PCR was performed in a LightCycler 480 II (Roche) using the LightCycler 480 SYBR Green I Master kit (Roche). The sequences of the sense and antisense oligonucleotides primers used were: GFP, 5′AAGTTCATCTGCACCACCG3′, 5′TCCTTGAAGAAGATGGTGCGC3′; RFP, 5′GCGGCCACTACACCTGCGAC3′, 5′TCGGCGTGCTCGTACTGCTC3′; HK2, 5′GAAGGGGCTAGGAGCTACCA3′, 5′CTCGGAGCACACGGAAGTT3′; SeAP, 5′CGGCTGTTGGGCACTGA3′, 5′GGAAGGTCCGCTGGATTGA3′; VEGF164, 5′AGACAGAACAAAGCCAGAAATCAC3′, 5′CACGTCTGCGGATCTTGGAC3′; platelet endothelial cell adhesion molecule 1 (PECAM1), 5′CTGGTGCTCTATGCAAGCCTC3′, 5′CGGTGCTGAGACCTGCTTT3′; adiponectin, 5′TGTTCCTCTTAATCCTGCCCA3′, 5′CCAACCTGCACAAGTTCCCTT3′; leptin, 5′GAGACCCCTGTGTCGGTTC3′, 5′CTGCGTGTGTGAAATGTCATTG3′; resistin, 5′AACTCCCTGTTTCCAAATGC3′, 5′AGCAGCTCAAGACTGCTGTG3′; interleukin (IL)-6, 5′CATGGATGCTACCAAACTGGAT3′, 5′CCAGGTAGCTATGGTACTCCAGA3′, IL-1β, 5′TGTAATGAAAGACGGCACACC3′, 5′TCTTCTTTGGGTATTGCTTGG3′; and monocyte chemoattractant protein-1 (MCP-1), 5′CCCAATGAGTAGGCTGGAGA3′, 5′TCTGGACCCATTCCTTCTTG3′. Data were normalized with 36B4 values (33).

Glucose uptake in isolated adipocytes was measured at different insulin concentrations. Isolated adipocytes were obtained by collagenase digestion of eWAT from fed mice, as described above. The adipocyte suspension (250 μL) was incubated with KRBH plus 4% BSA (fatty acid-free), 10 mmol/L of the deoxy-d-glucose, 0.4 μCi of 2-[1-³H]deoxy-d-glucose (2DG) (PerkinElmer, Waltham, MA), and different insulin concentrations for 5 min. Afterward, adipocytes and incubation medium were separated through silicon oil (Sigma-Aldrich) in polypropylene tubes, and radioactivity in the adipocyte samples was assessed by liquid scintillation counting. The results were expressed as picomoles of 2DG per 10⁶ cells per min.

The in vivo basal glucose utilization index was determined as previously described (34). Briefly, 4 µCi 2DG (PerkinElmer) was mixed in BSA-citrate buffer. A flash injection of radiolabeled mix was administered into the jugular vein of anesthetized (ketamine and xylazine) fed mice at time zero. The specific blood 2DG clearance was determined by the Somogyi procedure in 25 μL blood samples (tail vein) obtained 1, 15, and 30 min after injection. Tissue samples were removed 30 min after injection. The glucose utilization index was determined by measuring the accumulation of radiolabeled compounds. The amount of 2DG-6-phosphate per milligram of protein was divided by the integral of the concentration ratio of 2DG to unlabeled glucose measured. Because values were not corrected by a “discrimination constant” for 2DG in glucose metabolic pathways, the results are expressed as the index of glucose utilization, in picomoles per milligram of protein per minute.

Circulating human SeAP was quantified in 5 µL serum using the Tropix Phospha-Light System (Applied Biosystems, Inc.). Serum adiponectin was measured with a mouse/rat RIA Kit (Millipore, Billerica, MA).

eWAT and iBAT stroma vascular fraction, hepatic nonparenchymal cells, and peripheral blood mononuclear cells were isolated as previously described (35,36). F4/80 (BioLegend, San Diego, CA), CD3em, and NK-1.1 (BD Pharmigen, Franklin Lakes, NJ) antibodies were used. Flow cytometric analyses were performed on a FACSCanto flow cytometer (BD Biosciences, Franklin Lakes, NJ).

All values are expressed as mean ± SEM. Differences between groups were compared by the Student t test. Differences were considered significant at P < 0.05.

To asses the in vivo transduction efficiency of WAT by AAV vectors, AAV of serotypes 1, 2, 4, 5, 6, 7, 8 or 9 encoding the GFP under the control of the ubiquitous promoter CAG were injected into the eWAT in adult mice. Transduction efficiency of WAT was evaluated 2 weeks postinjection, given that at this time point AAV vectors have been reported to mediate robust transgene expression in most tissues. AAV8 and AAV9 vectors transduced the vast majority of adipocytes in eWAT, outperforming AAV1, 2, 4, 5, 6, and 7 vectors (Fig. 1A and B and Supplementary Fig. 1A). Similarly, local administration of AAV8 and AAV9-CAG-GFP vectors into the iWAT mediated extensive transduction of white and beige adipocytes in this depot (Fig. 1C and D).

The intraeWAT and intraiWAT administration of AAV8 or AAV9 vectors also resulted in transduction of the liver and heart (Supplementary Fig. 1B and data not shown). To restrict transgene expression to adipose tissue, we used murine mini/aP2, which allowed overcoming the limited cloning capacity of AAV vectors (4.7 kb). This promoter comprises the basal promoter and adipocyte-specific enhancer of aP2 (28,29). The intraeWAT administration of AAV8 or AAV9-mini/aP2-GFP vectors mediated specific transduction of white adipocytes (Supplementary Fig. 1C), with no detectable GFP expression in the liver and heart (Supplementary Fig. 1D) and iBAT (data not shown).

To demonstrate that in vivo AAV-mediated transduction of adipocytes may be a suitable model to study adipose function, AAV9 vectors encoding the murine HK2, the main glucose phosphorylating enzyme in fat cells, under the control of the ubiquitous CMV promoter (AAV9-CMV-HK2) or an equal dose of the AAV9-CMV-null vector, were delivered intraeWAT. Adipocytes isolated from AAV9-CMV-HK2–treated mice showed overexpression of HK2 compared with adipocytes obtained from AAV9-CMV-null–injected mice (Fig. 1E). Accordingly, in vitro insulin-stimulated 2DG uptake was increased in AAV9-CMV-HK2–transduced adipocytes compared with adipocytes from AAV9-CMV-null–treated animals (Fig. 1F). Furthermore, animals injected intraeWAT with AAV9-mini/aP2-HK2 vectors showed a greater increase (about threefold) in 2DG uptake in vivo than AAV9-mini/aP2-null–treated animals (Fig. 1G). As expected, no differences in 2DG uptake were detected between groups in iBAT and heart (Fig. 1G).

Finally, to test the ability of AAV-engineered eWAT to secrete proteins of interest into the bloodstream, human SeAP was used as reporter. AAV9-mini/aP2-SeAP vectors were injected bilaterally intraeWAT, and a long-term follow-up of circulating SeAP was performed. At 2 weeks after AAV vector delivery, high levels of SeAP were detected in serum. Levels rose progressively up to day 30 and thereafter remained stable for the duration of the study (>140 days) (Fig. 1H).

Upon intraiBAT administration of AAV8 or AAV9-CAG-GFP vectors, numerous GFP⁺ brown adipocytes were detected throughout the depot (Fig. 2A). Quantification of transduction efficiency showed that AAV8 and AAV9 were as efficient as AAV7 and superior to AAV1, AAV2, and AAV5 vectors in transducing iBAT (Fig. 2B).

The intraiBAT administration of AAV7, AAV8, or AAV9 vectors also resulted in gene transfer to the liver and heart (Supplementary Fig. 2A). To restrict transgene expression to iBAT, the mini/UCP1 promoter was used. The mini/UCP1 promoter comprises the basal promoter and the adipose-specific enhancer of rat UCP1 (30,31). The intraiBAT administration of AAV8 or AAV9-mini/UCP1-GFP achieved efficient transduction of brown adipocytes (Fig. 2C), abolished transgene expression in the heart, and mediated very marginal liver transduction (Supplementary Fig. 2B), in agreement with previous reports (31).

Animals treated locally with AAV8-mini/UCP1-HK2 showed increased in vivo 2DG uptake in iBAT compared with AAV8-mini/UCP1-null–treated animals (Fig. 2D). As expected, no difference was found between groups in the 2DG uptake in eWAT and heart (Fig. 2D). Similarly, animals injected intraiBAT with AAV9 vectors encoding the isoform 164 of murine VEGF (AAV9-mini/UCP1-VEGF164) showed marked overexpression of VEGF164 (Fig. 2E) and increased vessel density in this fat depot, as revealed by increased expression of PECAM1, a commonly used endothelial cell marker (36,37) (Fig. 2E), and increased immunostaining with CD105, a marker of proliferating endothelial cells (38) or with the pericyte marker α-SMA (39) (Fig. 2F and Supplementary Fig. 2C).

The systemic delivery of AAV8 or AAV9 vectors transduces a wide variety of tissues (40,41). However, transduction of WAT and BAT by intravascular administration of these serotypes has never been reported before. After intravascular administration of 5 × 10¹² vg of AAV8 or AAV9-CAG-GFP vectors to lean mice, white and brown fat depots of the whole body were transduced although transduction efficiencies in the different depots varied with mouse strain (Fig. 3A–D). Lower systemic doses of vectors (5 × 10¹⁰ and 5 × 10¹¹ vg) also mediated gene transfer to WAT and BAT, although at reduced efficiency (Supplementary Fig. 3A and B). No sex differences were observed in AAV9-mediated transduction of the different adipose depots, except for gonadal fat pads (Supplementary Fig. 3C). Importantly, the intravascular administration of AAV8 or AAV9 vectors to diabetic obese ob/ob mice (Fig. 4A and B) or db/db mice (Fig. 4C and D) also resulted in the genetic engineering of WAT and BAT, with efficiencies similar to those attained in lean mice (Fig. 3C and D).

Noticeably, the intravascular administration of AAV8 or AAV9 vectors did not result in infiltration of WAT, BAT, or the liver with macrophages, T lymphocytes, or natural killer cells (Fig. 5A, B and Supplementary Fig. 4A), nor was neutrophilic infiltration detected in the liver (Supplementary Fig. 4B). Accordingly, no changes in the expression profile of adipokines and proinflammatory cytokines were detected in WAT (Fig. 5C), and circulating levels of adiponectin remained normal in AAV8- and AAV9-treated animals (Fig. 5D). Moreover, intravascular delivery of AAV8 or AAV9 vectors did not lead to transduction of macrophages in eWAT, iBAT, the liver or peripheral blood (data not shown).

We then tested whether the mini/aP2 and mini/UCP1 promoters could be used to restrict transgene expression to white and/or brown adipocytes after systemic delivery of AAV vectors. The intravascular administration of AAV9-mini/aP2-GFP or AAV9-mini/UCP1-GFP vectors attained highly adipose-specific GFP expression in white and/or brown adipocytes, respectively (Fig. 6A and B), with no detectable transgene expression in the heart and very marginal transduction of the liver, which presented only a few scattered GFP⁺ hepatocytes (Supplementary Fig. 5A). Similar results were obtained for the AAV8 serotype (data not shown). However, the level of transgene expression mediated by the mini/aP2 or mini/UCP1 promoters was much lower than that afforded by the CAG promoter (Supplementary Fig. 5B). Nevertheless, systemic administration of AAV9-mini/UCP1-VEGF164 resulted in increased expression of VEGF164 and PECAM1 (Fig. 6B and C) and increased vessel density (Fig. 6D and E) in iBAT.

To further strengthen the potential of AAV vectors to genetically modify adipose tissue when administered systemically and given the reduced transgene expression attained by the mini/aP2 and mini/UCP1 promoters, we also investigated the use of the CAG promoter in conjunction with tissue-specific micro-RNA target (miRT) sequences in an attempt to obtain high expression levels in adipose tissue and de-target transgene expression from off-target organs. Upon intravascular administration of AAV9-CAG-GFP vectors containing targets of the liver-specific miR122a (AAV9-CAG-GFP-miRT122) (42), the heart-specific miR1 (AAV9-CAG-GFP-miRT1) (43) or both (AAV9-CAG-GFP-doublemiRT), high levels of GFP expression were observed in white and brown adipocytes, whereas GFP production in the liver and/or heart, respectively, was nearly completely abolished (Fig. 7).

In this study, we have shown that local or systemic administration of AAV vectors, especially serotypes 8 and 9, leads to robust genetic engineering of white and brown adipocytes in adult mice. The systemic administration of AAV8 or AAV9 vectors mediated transduction of whole-body adipose depots in lean and obese diabetic mice. However, depending on the strain, variable transduction efficiencies among depots were observed. Distinct WAT depots are not equivalent, because developmental, morphological, molecular, and functional differences among them have been reported for mice and humans (44,45). The differences observed in transduction efficiencies between depots may be also due to their distinct degree of vascularization. Indeed, retroperitoneal and inguinal depots are less vascularized than the rest of the fat pads, being that inguinal depots have the lowest density of blood vessels (46). Although high vascular density and blood flow have been described in the mesenteric depot (46), this fat pad also contains a large amount of structural and extracellular proteins that modulate cell–cell and cell–matrix adhesion and endothelial permeability (47), which may impair genetic engineering of this depot by systemic administration of AAV vectors. Moreover, the delivery route may also substantially influence transduction efficiency. In this regard, when AAV vectors were administered intradepot to eWAT, iWAT, and iBAT, efficient transduction of adipocytes was observed.

Local administration of AAV7, AAV8, and AAV9 vectors also resulted in transduction of nonadipose tissues because these serotypes can cross endothelial barriers and reach the bloodstream (41). The use of the mini/aP2 and mini/UCP1 promoters or the liver and heart-specific mirT sequences prevented to a great extent transgene expression in off-target organs and afforded long-term transgene expression after local or systemic delivery of AAV vectors. Moreover, overexpression of HK2 or VEGF164 mediated by AAV-mini/aP2 or AAV-mini/UCP1 resulted in WAT- or BAT-specific increased in vivo glucose uptake or vascular density, respectively. Similar results have been observed using transgenic mice overexpressing VEGF164 in BAT (36,37) or the glucose phosphorylating enzyme glucokinase in WAT (32), highlighting the potential of AAV vectors to achieve genetic engineering of adipose tissue in adult mice. Thus, this gene transfer methodology may be an attractive alternative to the use of adipose-specific transgenic/knock-out mouse models, especially when high-throughput screening of genes is required. Furthermore, the aP2 promoter, which is frequently used to achieve adipocyte-specific expression of genes of interest in transgenic models, conveys transcription in cell types other than adipocytes, such as macrophages and cardiomyocytes (48). Conversely, the mini/aP2 promoter precluded transduction of the heart, and AAV8 and AAV9 vectors did not mediate genetic engineering of macrophages, consistent with the poor tropism of these serotypes for this cell type (35,49). Additional advantages of the strategy presented here over the use of transgenic animals are that AAV vector production can be done in a relatively short period of time (a few weeks), discrete depots of choice can be engineered by intradepot administration, and the same vector construct can be tested in different animal models of obesity and diabetes, including large-animal models (50).

Numerous experimental and clinical studies have documented that AAVs, in contrast to other types of vectors, such as adenoviruses, have very mild proinflammatory potential due to their inability to efficiently trigger innate immune responses (51,52). One recent study, however, reported infiltration and production of proinflammatory cytokines in the liver after intravascular administration of AAV vectors in mice, although these responses were boosted by double-stranded and not by single-stranded AAV vectors (53). Our results clearly show that intravascular administration of single-stranded AAV8 or AAV9 vectors does not lead to inflammation of the liver or adipose tissue.

Much of the research on fat cell function and differentiation published to date has relied on the use of primary adipocyte cultures and immortalized adipocyte precursor cell lines in vitro. However, not all precursor cells undergo differentiation, and mature adipocytes derived from these cell lines present remarkable phenotypic differences compared with wild-type adipocytes, such as aneuploidy, altered pattern of fat deposition, aberrant gene expression, and decreased insulin sensitivity (12–15). Gene transfer to the vast majority of these in vitro models is also inefficient, and primary adipocytes cannot be maintained in culture for long periods of time. In our study, AAV8 and AAV9 vectors mediated efficient, long-term transduction of white and brown adipocytes in vivo. Furthermore, the intraeWAT administration of AAV9-CMV-HK2 vectors led to increased glucose uptake when measured in vitro in isolated adipocytes. Thus, in vivo genetic engineering of adipose tissue by AAV vectors, followed by isolation of the engineered adipocytes, may be a useful approach to perform gene function studies in fat cells in vitro, because it would allow the use of efficiently transduced, bona fide adipocytes. Moreover, the isolation of in vivo transduced adipocytes weeks or months after vector delivery would also allow in vitro studies that require long-term genetic modification.

AAV-mediated genetic engineering of adipose tissue may also open up new opportunities for the development of future therapeutic strategies to treat obesity and type 2 diabetes. To this end, the safety and efficacy of AAV-mediated gene transfer has been extensively studied in humans, with encouraging results in the liver, muscle, central nervous system, and retina (24). Thus, AAV-mediated gene therapy approaches targeting obesity-specific alterations, such as WAT inflammation and hypoxia or thermogenesis, may constitute attractive therapeutic strategies.

The high secretory capacity shown by AAV-modified WAT may be exploited for the development of new gene therapy strategies for diseases in which supply of the therapeutic agent into the bloodstream is needed for treatment/cure, such as hemophilia (24), diabetes (50) or lysosomal storage diseases (40). Our data suggest that AAV-modified WAT can act as a “therapeutic pump” and, therefore, become an attractive target site in alternative to the liver and muscle, given its accessibility and the possibility of easy surgical removal of WAT if unexpected events occurred. This possibility may be of significant importance for patients who are not eligible for liver-directed gene transfer because of underlying hepatic disease, such as hemophilia patients with liver failure, cirrhosis, or liver cancer due to hepatitis C.

In summary, our findings demonstrate that AAV vectors offer great potential for the long-term postnatal genetic modification of murine adipose tissue in vivo. AAV-mediated genetic engineering of WAT and BAT represents a technological advance that may prove useful for the study of physiopathology of metabolic diseases as well as for the development of new therapeutic approaches for obesity and type 2 diabetes.

This work was supported by grants from Ministerio de Economía y Competitividad, Plan Nacional I+D+I (SAF 2011-24698) and Generalitat de Catalunya (2009 SGR-224 and ICREA Academia), Spain, and from European Commission DG Research FP6 (CLINIGENE, LSHB-CT-2006-018933). C.M. and E.C. were recipients of a predoctoral fellowship from the Ministerio de Educación, Cultura y Deporte, Spain.

No potential conflicts of interest relevant to this article were reported.

V.J. designed experiments, wrote and edited the manuscript, generated reagents, and performed experiments. S.M., E.C., C.M., I.E., C.J., A.R., and T.F. generated reagents and performed experiments. S.F. wrote and edited the manuscript, generated reagents, and performed experiments. F.B. designed experiments and wrote and edited the manuscript. F.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Virginia Haurigot (Universitat Autònoma de Barcelona, Spain), Federico Mingozzi (Genethon, Evry, France), Malcolm Watford (Rutgers University, New Brunswick, NJ), and Miquel Garcia (Universitat Autònoma de Barcelona, Spain) for helpful discussions, and Marta Moya (Universitat Autònoma de Barcelona, Spain), Xavier Leon (Universitat Autònoma de Barcelona, Spain), Maria Molas (Universitat Autònoma de Barcelona, Spain), and Manuela Costa (SCAC, Universitat Autònoma de Barcelona, Spain) for technical assistance.