Reducing expression of NAD + synthesizing enzyme NMNAT1 does not affect the rate of Wallerian degeneration Laura Conforti 1,2,3 , Lucie Janeckova 1 , Diana Wagner 2 , Francesca Mazzola 4 , Lucia Cialabrini 4 , Michele Di Stefano 4, *, Giuseppe Orsomando 4 , Giulio Magni 4 , Caterina Bendotti 5 , Neil Smyth 6 and Michael Coleman 1,2 1 The Babraham Institute, Cambridge, UK 2 Center for Molecular Medicine, University of Cologne (ZMMK), Germany 3 School of Biomedical Sciences, University of Nottingham, UK 4 Dipartimento di Patologia Molecolare e Terapie Innovative, Universita’ Politecnica delle Marche, Ancona, Italy 5 Mario Negri Pharmacological Research Institute, Milan, Italy 6 School of Biological Sciences, University of Southampton, UK Introduction The essential role of NAD + in cell metabolism and energy production has been known for over a century and the NAD + synthesizing enzymes nicotinamide mononucleotide adenylyltransferases (NMNATs) are evolutionarily ancient and present throughout evolu- tion, including archaebacteria. While in prokaryotes Keywords axon; Cre-loxP knockout; NAD(P) + ; NMNAT; Wallerian degeneration Correspondence L. Conforti, School of Biomedical Sciences, D37c, University of Nottingham, Medical School, Queen’s Medical Centre, Nottingham, NG7 2UH, UK Fax: +44 (0)115 8231476 Tel: +44 (0)115 8230142 E-mail: laura.conforti@nottingham.ac.uk *Present address School of Biomedical Sciences, University of Nottingham, Medical School, Queen’s Medical Centre, Nottingham, NG7 2UH, UK (Received 7 March 2011, revised 4 May 2011, accepted 23 May 2011) doi:10.1111/j.1742-4658.2011.08193.x NAD + synthesizing enzyme NMNAT1 constitutes most of the sequence of neuroprotective protein Wld S , which delays axon degeneration by 10-fold. NMNAT1 activity is necessary but not sufficient for Wld S neuroprotection in mice and 70 amino acids at the N-terminus of Wld S , derived from poly- ubiquitination factor Ube4b, enhance axon protection by NMNAT1. NMNAT1 activity can confer neuroprotection when redistributed outside the nucleus or when highly overexpressed in vitro and partially in Drosophila. However, the role of endogenous NMNAT1 in normal axon maintenance and in Wallerian degeneration has not been elucidated yet. To address this question we disrupted the Nmnat1 locus by gene targeting. Homozygous Nmnat1 knockout mice do not survive to birth, indicating that extranuclear NMNAT isoforms cannot compensate for its loss. Heterozygous Nmnat1 knockout mice develop normally and do not show spontaneous neurode- generation or axon pathology. Wallerian degeneration after sciatic nerve lesion is neither accelerated nor delayed in these mice, consistent with the proposal that other endogenous NMNAT isoforms play a principal role in Wallerian degeneration. Enzymes NMNAT ( EC 2.7.7.1) Abbreviations ES cell, embryonic stem cell; KO, knockout; NAMPT, nicotinamide phosphoribosyltransferase; NMNAT, nicotinamide mononucleotide adenylyltransferase; PARP1, poly(ADP-ribose) polymerase 1; SCG, superior cervical ganglia; VCP ⁄ p97, valosin-containing protein; YFP, yellow fluorescent protein. 2666 FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS and in some eukaryotes such as Drosophila only one NMNAT isoform has been found to date, in other simple eukaryotes such as yeast and in higher eukary- otes including mice and humans more than one NMNAT isoform has been identified [1]. In mammals there are three NMNAT isoforms with different tissue distribution and intracellular localization [2–4]. The location of the different isoforms could be related to specific roles played by NAD + and its metabolites as second messengers in cell signalling cascades in differ- ent environments, as recently described [5]. Higher organisms could have evolved isoform-specific domains mediating subcellular targeting and post-transcrip- tional modifications responsible for NMNAT specific functions at subcellular level [6]. Alternatively, there could be some redundancy, for example with extranu- clear NMNAT isoforms being able to compensate for the nuclear isoform. Studies describing subcellular localization of the three NMNAT isoforms are based on the overexpression of fusion proteins which could reach ectopic locations. The possibility also exists that NMNATs could localize to other compartments and act at very low levels [7]. Thus, their roles may not be restricted to the reported locations. Nuclear NMNAT1 synthesizes NAD + which is required for the activity of histone deacetylase sirtuins and as substrate of poly(ADP-ribose) polymerase 1 (PARP1). High levels of NAD + are required for life- span extension in yeast and this response is mediated by the activity of sirtuin family member Sir2p [8]. Another member of this family, SIRT1, also regulates circadian rhythm in mammals [9]. Notably, nicotin- amide phosphoribosyltransferase (NAMPT), the rate limiting enzyme in NAD + synthesis, is correlated with increased longevity in human cells [10] and is also involved in the regulation of circadian rhythm [9]. NMNAT1 interacts with SIRT1 at target gene pro- moters, regulating transcription of genes important for neuronal function [11]. Nuclear NMNAT1 also regu- lates the activity of genotoxic stress activated nuclear protein PARP1 by providing NAD + [12] and by phos- phorylation-dependent association with PARP1 [13], thus participating in cell death pathways [14,15]. Some debate still exists on the presence of endogenous NMNAT1 in the axonal compartment in neurons and on its role in axon survival [16], but targeting NMNAT1 to axons, even at low levels, does confer protection [17,18]. Extranuclear NAD + , such as that generated by Golgi-associated NMNAT2 and by mito- chondrial NMNAT3, is mainly used for energy pro- duction, as a redox cofactor and as substrate of enzymes like NAD + kinase, which converts NAD + to NADP + , and NAD + glycohydrolases that convert NAD + and NADP + to ADP-ribose, cyclic ADP- ribose and nicotinic acid adenine dinucleotide phos- phate, all of which act as second messengers in Ca 2+ release from intracellular stores. The role of NAD + in Wallerian degeneration has emerged since the discovery of the Wld S gene, where the full coding sequence of Nmnat1 is fused to the 5¢ end of ubiquitination factor Ube4b giving rise to the Wld S protein, a modified NMNAT1 enzyme with an extended N-terminal sequence. Wallerian degenera- tion, the degeneration of axons and synapses after an injury, is delayed 10-fold by Wld S both in vivo and in vitro, in organisms as diverse as mice, rats and flies [19]. NMNAT1 enzyme activity is required for the protective phenotype [20,21] but the N-terminal sequences are also necessary to achieve full protection in vivo. NMNAT1 overexpression is not sufficient to delay axon degeneration in transgenic mice [22] and does so only weakly in Drosophila [20]. In dorsal root ganglia cultures, NMNAT1 confers protection when locally transduced into axons or when highly overexpressed [18,23]. The critical N-terminal sequence of Wld S resides within the first 16 amino acids, as their removal results in loss of neuroprotec- tive phenotype [20,21]. Interestingly, the only known binding partner of the N-terminal region, AAA ATPase valosin-containing protein (VCP⁄ p97), is a very abundant cellular protein mainly localized at the surface of membranous intracellular organelles [24,25]. It is possible that NMNAT1 is redistributed to a specific location by binding to this N-terminal region and acquires a protective function by produc- ing or overproducing NAD + at that locus. As down- regulation or rapid degradation of NMNAT2 triggers spontaneous Wallerian degeneration, the NMNAT1 component of Wld S is likely to substitute for endoge- nous NMNAT2 when this is degraded after an injury [26]. In order to evaluate the role of endogenous NMNAT1 in Wallerian degeneration, we inactivated the gene by homologous recombination. Complete inactivation of both alleles was embryonic lethal but Nmnat1 heterozygous knockout (KO) mice were born, developed as normal and showed reduced NMNAT1 mRNA, protein and enzyme activity levels. Wallerian degeneration of transected sciatic nerves proceeded at wild-type rate. These data confirm that NMNAT1 is an essential enzyme for which NMNAT2 or NMNAT3 cannot compensate and that NAD + synthesis in the nucleus is indispensable for survival. These data are also consistent with a primary role for other endoge- nous NMNAT isoforms such as NMNAT2 in main- taining axon integrity. L. Conforti et al. NMNAT1 gene inactivation and axon degeneration FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS 2667 Results Targeting the Nmnat1 gene Mouse Nmnat1 is formed by four exons and spans a 148 850 kb genomic region on distal chromosome 4. In order to allow eventual conditional deletion, we designed a targeting construct based on the vector pEASYFlox (a gift from W. Mu ¨ ller and K. Rajewsky) to insert a NEO R selection cassette flanked by loxP sites upstream of exon 1, within the promoter region, approximately 600 bp 5¢ of the start ATG. A third loxP site was placed within intron 2, between exons 2 and 3 (Fig. S1A), so that a 2.3 kb region comprising some 5¢UTR and exons 1 and 2 was in turn flanked by two loxP sites. After Cre-mediated recombination between the second and the third loxP sites, part of promoter and the first two exons of the gene would be disrupted. Even in the unlikely event that a truncated protein lacking these two exons was expressed, it would not be functional because important substrate binding sites are encoded within the first two exons. After introduction of the targeting vector into C57BL ⁄ 6 embryonic stem (ES) cells we verified correct integration of the NEO R selection cassette and of the third loxP site by southern blotting using 5¢ and 3¢ specific probes (Fig. S1). A 420 bp probe, placed 5¢ of the targeting region, recognized a 9.5 kb wild-type band on southern blots of EcoRI digested ES cell genomic DNA. In heterozygous targeted ES cells, in addition to the wild-type band, another band at approximately 3 kb was found, due to the introduc- tion of an additional EcoRI site within the NEO R cas- sette. Cointegration of the third loxP site was also verified in southern blots of ES cell genomic DNA digested with HindIII. A 3¢, 750 bp probe recognized an 8.7 kb band in wild-type and a 6.3 kb band in the correctly targeted ES cells due to the introduction of a HindIII site located immediately outside the loxP sequence (Fig. S1). We had a success rate of 0.26% in the generation of correctly targeted ES cell clones, with one clone where both NEO R cassette and third loxP site were correctly integrated out of 384 total screened. We refer to the correctly targeted allele as Nmnat1 + ⁄ 3lox . Next, we transfected Nmnat1 + ⁄ 3lox ES cells with a Cre recombinase expressing vector (pPGK-Cre-bpA, kind gift of W. Mueller). Cre in vitro excised the DNA between the loxP sites as shown in Fig. 1A. We selected only the ES cell clones with a type II deletion (Fig. 1A). Those clones became again sensitive to G418 due to the excision of the NEO R cassette. South- ern blot analysis of G418 sensitive ES cell genomic DNA digested with BamHI, using a probe located out- side the third loxP site, showed a 3.5 kb band, in addi- tion to the wild-type 13 kb band, in cells where Cre- mediated type II deletion and splicing of the first and second loxP sites had occurred (Fig. 1A,B). The tar- geted allele in these ES cells had the NEO R cassette removed and only two loxP sites remaining; therefore the cells are referred to as Nmnat1 + ⁄ 2lox . Generation of heterozygous Nmnat1 knockout mice Nmnat1 + ⁄ 2lox ES cells were injected into the blast- ocysts of 129 ⁄ J mice and chimeric mice identified by coat colour and bred to obtain a germline transmission of the mutant floxed allele. Germline transmission events were confirmed by both southern blotting and PCR of tail DNA (Fig. 1B). For PCR, primer pairs Pr1 + Pr2 and Pr3 + Pr4 were designed to amplify across the two loxP sites, and detected a 32 or 39 bp wild-type band respectively that increased to 66 and 73 bp when the loxP sites were also present. For constitutive Nmnat1 gene inactivation, we crossed Nmnat1 + ⁄ 2lox male mice with C57 ⁄ BL6 K14 Cre female mice to produce heterozygous null mice on a black background. The K14 Cre induces a full dele- tion when bred from the female as K14 is expressed in the oocyte [27]. The offspring of this cross had recom- bination between the two loxP sites; therefore the 2.3 kb floxed region had been removed and only one of the loxP sites was left behind (Fig. 1A,C). Identifi- cation of heterozygous KO (Nmnat1 + ⁄ ) ) mice was done by southern blot analysis and PCR using primers Pr1 and Pr4 (see Fig. 1A). The BamHI band shifted from 3.5 kb in the Nmnat1 floxed mice to 1.2 kb in Nmnat1 + ⁄ ) mice. The wild-type band of 13 kb was still present (Fig. 1C). PCR with primers Pr1 + Pr4 gave a 2.4 kb PCR product in wild-type (not shown), shortened to 80 bp if Cre-mediated recombination between the two loxP sites had occurred (Fig. 1C). When we intercrossed Nmnat1 + ⁄ ) mice to produce homozygous knockouts, we found no live homozygous nulls from a total of 88 offspring that were genotyped. Heterozygotes were born at the expected Mendelian ratio given the absence of homozygotes (two-thirds of total births). The remaining offspring were wild-type. Thus, Nmnat1 is essential for embryo development. Protein and mRNA expression analysis in Nmnat1 + ⁄ ) mice We tested whether NMNAT1 expression and enzyme activity were decreased in Nmnat1 + ⁄ ) mice. Because NMNAT1 gene inactivation and axon degeneration L. Conforti et al. 2668 FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS NMNAT1 is not abundantly expressed in brain, we first assessed protein levels in skeletal muscle, where the protein is expressed in higher amounts [28,29]. NMNAT1 expression was significantly reduced in het- erozygous KO mice, as shown in western blots of skel- etal muscle homogenates probed with antibody 183 [28] (Fig. 2A). Although more difficult to visualize, NMNAT1 band intensity was also reduced in western blots of Nmnat1 + ⁄ ) brain homogenates probed with antibody 183 relative to wild-type (Fig. S2A). North- ern blots from total brain RNA of Nmnat1 + ⁄ ) and wild-type mice probed with an Nmnat1 cDNA probe also showed a reduced band intensity of Nmnat1 tran- script (Fig. S2B). In agreement with expression data, total NMNAT enzyme activity was significantly reduced in brain homogenates of Nmnat1 + ⁄ ) mice rel- ative to wild-types (Fig. 2B). Despite the reduction in protein levels and enzyme activity, NAD + levels were not reduced in Nmnat1 + ⁄ ) mouse brains (Fig. 2C). In order to test whether NMNAT1 partial deletion had any influence on the expression of the other two NMNAT isoforms and to investigate any compensa- tory mechanisms, we assayed isozyme-specific mRNA expression levels in brain homogenates by real time RT-PCR (Fig. 2D). As expected, we found that NMNAT1 mRNA was greatly reduced in Nmnat1 + ⁄ ) brain homogenates (Fig. 2D, left panel). However, no significant differences were observed in NMNAT2 and NMNAT3 mRNA relative expression levels in Nmnat1 + ⁄ ) brain compared with wild-type (Fig. 2D, right panel). We also determined the enzyme activity of each NMNAT isoform in order to evaluate their relative contribution to total NAD + formation. Isoform-spe- HindIII (18 670) EcoRI (7282) 15 00014 000 16 000 17 187 HindIII (10 138) BamHI (13 390) EcoRI (16 760) Sal1 Sal1 HindIII BamHI (1) 1 BamHI 11 00080007000 Probe 4 Wild-type (BamHI band ca 13 kb) Type II del (BamHI band 3.5 kb) Type I del or Cre-mediated recombination from type II del (BamHI band 1.2 kb) 1° loxP 3° loxP –ve –ve –ve Wild-type allele LoxP insertion Wild-type allele (13 kb) Floxed allele (3.5 kb) 60 61 Wild-type allele (13 kb) KO allele (1.2 kb) Pr1 Pr2 Pr3 Pr4 Pr1 Pr4 Nmnat1 +/2lox (Floxed Nmnat1 het) Nmnat1 +/– (KO Nmnat1 het) NEO 1 2 3 4 loxP1 loxP2 loxP3 1 2 BamHI band BamHI band A BC Fig. 1. Generation of heterozygous targeted mice. (A) Representation and map of Nmnat1 targeted allele and the deletion events after Cre transfection of Nmnat1 + ⁄ 3lox ES cells. The expected change in the size of a BamHI band in genomic southern blots is shown in the diagram. (B) Southern blot and PCR analysis of genomic DNA of ES cell clones after Cre-mediated recombination (type II deletion according to the dia- gram in A). The genomic DNA was digested with BamHI and probed with probe 4. The wild-type and the recombinant band are the expected size. PCR with primer pairs Pr1 + Pr2 and Pr3 + Pr4 shows the correct placement of the two remaining loxP sites. Exactly the same result was shown in southern blot and PCR analysis of Nmnat1 + ⁄ 2lox mouse tail DNA, after blastocyst injection and coat colour screen- ing of mice. (C) PCR and southern blot analysis of tail DNA from Nmnat1 + ⁄ 2lox · C57BL ⁄ 6 K14 Cre offspring. PCR was performed with prim- ers Pr1 + Pr4 to demonstrate the correct deletion of the genomic region between the first and the third loxP site as shown by the 80-bp product formed. The 2.4 kb wild-type PCR product cannot be distinguished on this high percentage agarose gel (left panel). The fact that the new Cre-mediated recombination leaves only one loxP site is also demonstrated by the 1.2 kb BamHI specific band on a southern blot (right panel). L. Conforti et al. NMNAT1 gene inactivation and axon degeneration FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS 2669 cific NMNAT enzyme activity was determined with a biochemical discrimination assay based on the dis- tinctive metal ion sensitivity of the three isoforms (Orsomando G, Cialabrini L, Amici A, Agostinelli S, Janeckova L, Di Stefano M, Conforti L, Coleman M, Magni G, manuscript in preparation, adapted from [30,31]). In agreement with mRNA and protein expression analysis, NMNAT1 enzyme activity in Nmnat1 + ⁄ ) mouse brain was about half that in wild- type (Fig. 2E). In contrast, no significant differences were observed in NMNAT2 and NMNAT3 activity (Fig. 2E). Despite the high brain mitochondrial content and energy demand, NMNAT3 enzyme activity is very low. This result was obtained in brain extracts after dis- ruption of mitochondrial membranes, excluding the possibility of an underestimation of NMNAT3 activity Level of NMNAT1 protein (arbitrary units) 0.30 0.25 0.20 0.15 *(P = 0.046) 32 A BC E D 1321 NMNAT1 (31.5 kDa) β -actin (42 kDa) NAD + levels (nmol·g –1 tissue) 0 50 100 150 200 250 300 350 N.S. C57BL/6 Nmnat1 +/– % relative expression normalised to β -ACT NMNAT1 **(P = 0.0054) 0 20 40 60 80 100 120 140 160 0 0.02 0.04 0.06 0.08 0.1 NMNAT enzyme activity (m U·mg –1 ) NMNAT1 NMNAT2 NMNAT3 N.S. N.S. *(P = 0.0212) Nmnat1 +/– Wild-type NMNAT2 N.S. NMNAT3 0 20 40 60 80 100 120 % of each isoform expressed in Nmnat1 +/– mice relative to wild-type N.S. ** NMNAT1 Nmnat1 +/– Wild-type C57BL/6 Nmnat1 +/– C57BL/6 Nmnat1 +/– 0.40 0.30 0.20 0.10 0.00 NMNAT enzyme activity (mU·mg –1 ) C57BL/6 Nmnat1 +/– *(P = 0.013) Fig. 2. NMNAT isoform expression and enzyme activity in Nmnat1 + ⁄ ) mice. (A) Western blots of skeletal muscle homogenates from Nmnat1 + ⁄ ) and C57BL ⁄ 6 mice probed with antibody 183 (the antibody also reveals a non-specific upper band). The histogram represents the integrated band intensity of the NMNAT1 band normalized to the b-actin control (n = 3, Mann–Whitney test, P = 0.046). (B) Total NMNAT activity of brain homogenates from Nmnat1 + ⁄ ) and C57BL ⁄ 6 mice. The enzyme activity is strongly reduced in the heterozygous mice with respect to wild-types (n = 9, Student’s t-test, P = 0.013). (C) NAD + levels in wild-type and Nmnat1 + ⁄ ) total brain homogenate (n = 5, Student’s t -test). (D) Left panel: NMNAT1 mRNA relative expression in Nmnat1 + ⁄ ) and wild-type brains showing strong reduction of NMNAT1 mRNA in heterozygous KO mice. Right panel: Relative mRNA expression of each NMNAT isoform in Nmnat1 + ⁄ ) compared with wild-type, showing that while NMNAT1 mRNA expression is reduced, NMNAT2 and 3 mRNA relative expression is not changed. Normaliza- tion was performed for each isoform by calculating the ratio between the expression of an individual NMNAT isoform and that of the refer- ence gene (b-actin) in wild-type samples. The arbitrary number of 100% was assigned to this ratio for one control, and NMNAT expression of the same isoform in the remaining controls and in Nmnat1 + ⁄ ) brains relative to the reference gene was compared with this number. Therefore relative mRNA expression levels can be compared between wild-type and Nmnat1 + ⁄ ) (n = 3, Student’s t -test, **P = 0.0054). (E) Determination of NMNAT isozyme activity in wild-type and Nmnat1 + ⁄ ) total brain homogenates reveals highly reduced NMNAT1 activity in heterozygous KO tissue compared with wild-type but no change in the activity of the other two isoforms. Note the very low activity of NMNAT3 in mouse brain. (n = 3, Student’s t-test, *P = 0.0212). NMNAT1 gene inactivation and axon degeneration L. Conforti et al. 2670 FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS due to lack of solubilization of mitochondria during the extraction procedure. The absence of compensatory changes in NMNAT2 and NMNAT3 when NMNAT1 is depleted supports the model of non-redundant functions for these isoforms. Despite the reduction in NMNAT1 protein levels and enzyme activity, Nmnat1 + ⁄ ) mice are healthy, indistinguishable from their wild-type littermates and have a normal lifespan, suggesting that downregulation of NMNAT1 is compatible with normal life and a healthy nervous system, although complete inactivation is lethal. Wallerian degeneration rate in Nmnat1 + ⁄ ) mice Wld S neuroprotective protein contains NMNAT1 and requires its enzyme activity to delay axon degeneration after injury, but NMNAT1 overexpression in vivo is not neuroprotective [21,22]. However, the role of endogenous NMNAT1 on the rate of Wallerian degen- eration has never been determined. To test this, we lesioned sciatic nerves of Nmnat1 + ⁄ ) mice and their wild-type littermates after crossing them with YFP-H mice [32] where some axons are labelled with the yel- low fluorescent protein (YFP). In YFP-H positive mice it is easy to follow axon degeneration in longitudinal sections of lesioned sciatic nerves observed under a fluorescent microscope [22,33]. Wallerian degeneration of the distal stump of a sciatic nerve after an injury follows a precise time course in wild-type mice. Axon fragmentation begins at around 36 h, then proceeds quickly and is complete 42 h after the lesion. In spon- taneous mutant Wld S , however, Wallerian degenera- tion is highly delayed and axon continuity is preserved up to 3 weeks from injury [33,34]. Thus we studied Wallerian degeneration in Nmnat1 + ⁄ ) mice with sciatic nerves lesioned for 30 h as a non-stringent test for accelerated Wallerian degeneration, and for 72 h as a non-stringent test for any delay in Wallerian degenera- tion. Nmnat1 + ⁄ ) X YFP-H nerves fully maintained axon integrity 30 h after sciatic nerve lesions, similar to wild-type nerves [Fig. 3A(a,b)]. All axons were com- pletely fragmented 72 h after lesion, in the same way as wild-types [Fig. 3A(d,e)]. In great contrast, axons from Wld S heterozygous mice are completely preserved at this time point [Fig. 3A(f),B]. In order to exclude an effect on the time of onset of the degenerative process, we also analysed axon degeneration in wild-type and Nmnat1 + ⁄ ) mice 36 h after sciatic nerve lesion. At this time, axon degeneration has just begun to occur in wild-types [34]. However, even at this time point, we could not detect any significant difference in the num- ber of degenerated Nmnat1 + ⁄ ) axons compared with wild-types (Fig. 3B). We conclude that NMNAT1 downregulation neither accelerates nor delays axon degeneration after sciatic nerve lesion. We tested the rate of neurite degeneration after cut also in vitro, in superior cervical ganglia (SCG) cultures obtained from Nmnat1 + ⁄ ) and wild-type pups. SCG explants were allowed to extend neurites in culture for 7 days. The neurites were then cut with a scalpel per- pendicular to the direction of growth and observed at different times. Axons in wild-type SCGs remain intact 3 h after cutting, but start degenerating at 6–9 h, with degeneration complete by 24 h. Axon fragmentation in Nmnat1 + ⁄ ) SCG explants followed an identical time course (Fig. 4A,B). We determined NMNAT1 specific enzyme activity in SCG explant extracts from wild-type and Nmnat1 + ⁄ ) mice (Orsomando G, Cialabrini L, Amici A, Agostinelli S, Janeckova L, Di Stefano M, Conforti L, Coleman M and Magni G, manuscript in prepara- tion, adapted from [30,31]). Similarly to what was detected in brain, NMNAT1 activity in Nmnat1 + ⁄ ) SCG explants (0.015 mUÆmg )1 ) was half that in wild- types (0.033 mUÆmg )1 ). NAD(P) + levels in SGC whole cell extracts showed a non-significant trend towards lower levels in heterozygous null mice relative to wild- types (Fig. 4C). This could reflect a reduced level of nuclear NAD + that is masked by the activity of extra- nuclear NMNAT isoforms synthesizing high levels of NAD + in neurites. Indeed, neurite density in these cul- tures is very high, and NAD(P) + levels in wild-type SCG neurites are around double those of their corre- sponding cell bodies (L. Conforti, L. Janeckova and M. Coleman, unpublished results). Thus reduction of NAD + within nuclei remains possible. However, in agreement with the result in vivo, dowregulation of NMNAT1 expression does not affect the rate of axon degeneration in vitro. Discussion These data indicate that complete NMNAT1 gene inactivation is incompatible with the normal develop- ment of embryos, as the extranuclear isoforms NMNAT2 and NMNAT3 cannot compensate for complete loss of NMNAT1. Nmnat1 + ⁄ ) mice have reduced NMNAT1 expression and enzyme activity; however, they develop normally and their lifespan is not altered. We show that the rate of Wallerian degen- eration in vivo and in vitro in sciatic nerves and in SCG explant cultures from Nmnat1 + ⁄ ) mice is not different from wild-type. NMNAT1-generated NAD + in the nucleus is used as substrate of histone deacetylase sirtuins and L. Conforti et al. NMNAT1 gene inactivation and axon degeneration FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS 2671 PARP1. Sirtuins have been implicated in cellular pro- cesses such as ageing, transcription, apoptosis and stress resistance. Yeast Sir2 and its mammalian homo- logue SIRT1 are upregulated upon caloric restriction and this is associated with increased lifespan [8]. SIRT1 controls the activity of genes that regulate circadian rhythm and promotes the transcription of NAMPT, the rate limiting enzyme in NAD + synthesis, in a feedback loop that has been recently described [35,36]. NAD + is substrate also for nuclear PARP1, whose overactivation consequent to genotoxic stress leads to NAD + depletion in the cytoplasm and cell necrosis, demonstrating a communication between the nuclear and the cytoplasmic NAD + pool [37]. Thus, the failure of Nmnat1 homozygous null embryos to survive and develop may reflect perturba- tions in gene transcription, especially sirtuin targets, or PARP1-mediated NAD + depletion that cannot be replenished locally within the nucleus. Indeed, NMNAT1 downregulation in cell lines by small inter- fering RNA has a profound effect on transcription of a number of genes, some of which are important for neuronal maintenance and normal neuronal function [11]. Conditional homozygous inactivation of Nmnat1 in neurons in the adult mouse will be essential to understand whether and how transcriptional regula- tion affects neuronal maintenance and survival. NMNAT1 is also part of the neuroprotective protein Wld S and its enzyme activity is necessary but not suffi- cient for this protein to delay degeneration of axons after an injury in vivo [20–22]. However, in cell cultures and in Drosophila NMNAT1 overexpression is par- tially neuroprotective [20,23]. Moreover, in Drosophila, targeted disruption of NMNAT causes spontaneous axon degeneration via a chaperone activity [38,39]. We investigated the role of endogenous NMNAT1 in axon protection in heterozygous null mice where we found a strong reduction in NMNAT1 protein expression and enzyme activity, while the other two isoforms were expressed at wild-type levels and their enzyme activity Nmnat1 +/– cut t = 72h WT cut t = 72 h UNCUT WT cut t = 30 h Wld S het cut t = 72 h Nmnat1 +/– cut t = 30 h 50 µm (a) (b) (c) (f)(e)(d) 0 20 40 60 80 100 120 % intact axons 30 h 36 h 72 h Wild-type Nmnat1 +/– Wld S het N.S. N.S. N.S. A B Fig. 3. Wallerian degeneration rate in Nmnat1 + ⁄ ) mice. (A) Tibial nerves from Nmnat1 + ⁄ ) mice crossed to YFP-H with sciatic nerves lesioned for the indicated time show a wild-type rate of Wallerian degeneration with intact axons 30 h after the lesion (a, b) and completely degenerated axons 72 h after the lesion (d, e). At this time point, Wld S heterozygous axons are still completely preserved (f). Bar, 50 lm. (B) Quantification of axon degeneration at the indicated time points after sciatic nerve lesions. Note that at 36 h post-lesion, when Wallerian degeneration normally begins, the number of degenerated axons is similar in wild-type and Nmnat1 + ⁄ ) .(n = 4, Student’s t-test). NMNAT1 gene inactivation and axon degeneration L. Conforti et al. 2672 FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS was unchanged. Since NMNAT1 activity is predomi- nant in brain (Fig. 2E) and NMNAT1 is also the most catalytically efficient isoform [31], its downregulation determines a significant reduction in total NMNAT activity in Nmnat1 + ⁄ ) mice that cannot be compen- sated by NMNAT2 and ⁄ or NMNAT3. Sorci et al. [31] reported that NMNAT2 is the predominant activity in human brain. However, these authors used human per- itumoural tissue for their determination of isoform- specific NMNAT activity, whereas we used mouse half brain homogenates. Brain has a heterogeneous cellular composition that could influence relative abundance of this enzyme activity; therefore our result is neither directly comparable nor in conflict with that described by Sorci et al. [31]. Despite NMNAT1 strong downregulation, Nmnat1 + ⁄ ) mice do not show any unusual phenotype and the rate of Wallerian degeneration in these mice or in primary neurons derived from them is unaltered. It is possible that the maintenance of normal NAD + steady state levels despite the decrease in NMNAT activity in our mutant mice underlines the lack of any adverse phenotype. The embryonic lethality of NMNAT1 full inactivation precludes the possibility of testing the rate of Wallerian degeneration in the com- plete absence of NMNAT1. However, the result obtained in heterozygous NMNAT1 KO mice suggests that extranuclear NMNAT activities predominantly control the rate of Wallerian degeneration. Accord- ingly, the two extranuclear NAD + -synthesizing iso- zymes, NMNAT2 and NMNAT3, maintain wild-type expression levels and enzyme activities in Nmnat1 + ⁄ ) mice where Wallerian degeneration after injury pro- ceeds at a wild-type rate. i 0.60 Wild-type t = 0 Nmnat1 +/– t = 0 Nmnat1 +/– t = 3 h Nmnat1 +/– t = 6 h Nmnat1 +/– t = 9 h Nmnat1 +/– t = 24 h Wild-type t = 3 h Wild-type t = 6 h Wild-type t = 9 h Wild-type t = 24 h 0.50 0.40 0.30 0.20 0.10 0.00 t = 0 h t = 3 h t = 6 h t = 9 h t = 24 h N.S. NAD(P) + (nmol·mg –1 protein) A BC Fig. 4. In vitro degeneration of injured axons in Nmnat1 + ⁄ ) SCG cultures. (A) SCG explants from C57BL ⁄ 6 and Nmnat1 + ⁄ ) mice were cul- tured for 7 days and the extended neurites were separated from the cell body mass using a scalpel. Neurites were imaged after the cut at the time points indicated. Bar, 10 lm. (B) Quantification of axon degeneration in SCG explant cultures after cut. The results show that there is a time effect (P < 0.0001) but no difference between wild-type and Nmnat1 + ⁄ ) (n = 6, two-way repeated measures ANOVA, P = 0.808). (C) NAD(P) + levels in whole SCG explant cultures from C57BL ⁄ 6 and Nmnat1 + ⁄ ) mice are similar. (n = 7, independent samples t-test, P = 0.492). L. Conforti et al. NMNAT1 gene inactivation and axon degeneration FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS 2673 This is also consistent with our observation of an increased Wld S protective potency when this protein is redistributed outside the nucleus [7,17]. Moreover, we showed lack of protection in transgenic mice overex- pressing NMNAT1 alone and in variant-Wld S trans- genic mice where an N-terminal 16 (N-16) amino acid sequence derived from Ube4b had been removed [21,22]. Interestingly, the only known property of the N-16 amino acid sequence indispensable for Wld S action is its ability to bind the abundant cellular pro- tein VCP ⁄ p97. This protein is involved in many cellu- lar activities and is particularly enriched at the surface of membranous organelles [24,25]. NMNAT activity in mammals has become more specialized by evolving several isoforms, each of them playing a particular role according to its most abundant location within the cell. Wld S protection may be the result of a fine redistribution of NMNAT1, potentially via VCP bind- ing, at a specific location inside the cell, where its enzyme activity leads to downstream events finally resulting in axon protection. Accordingly, cytoplasmic Wld S and cytoplasmic or axonally targeted NMNAT1 are all neuroprotective [7,17,18,40]. This location could match that of the endogenous extranuclear NMNAT isoform NMNAT2. NMNAT2 downregula- tion triggers spontaneous axon degeneration in pri- mary SCG neurons [26], suggesting this may be the endogenous NMNAT activity that normally controls Wallerian degeneration. NMNAT3 could also be responsible for controlling injury-induced axon degen- eration. However, the low level of NMNAT3 activity we detect in the nervous system and the lack of a phe- notype when this isoform is downregulated in neuronal cultures [26] makes it a weaker candidate. NMNAT2 is rapidly degraded after an injury and its rapid degra- dation could trigger axon degeneration. However, the more stable Wld S protein, when present, or an abnor- mal targeting of NMNAT1 itself [17] could preserve the injured axons by substituting for NMNAT2 [26]. The results presented here argue against functional redundancy of the three mammalian NMNAT iso- forms. NMNAT2 and 3 cannot compensate for loss of NMNAT1 when this isozyme is completely inactivated, leading to the lack of viability of null NMNAT1 KO mice. In addition, there is no upregulation of NMNAT2 or 3 in Nmnat1 + ⁄ ) mice, where NMNAT1 is highly downregulated. In cultured SCGs, NMNAT1 and 3 cannot compensate for loss of NMNAT2 trig- gered by RNA interference or by axon injury [26]. Indeed, the low level of NMNAT3 activity in brain suggests its main functions may be predominant in other tissues [30]. However, the various isoforms could compensate for each other when redistributed to a dif- ferent location. For instance NMNAT1 appears to compensate for loss of NMNAT2 when it reaches ectopic location by high overexpression or by re-target- ing, therefore conferring protection to axons after cut [17,18,22,26]. Despite the role for other NMNAT isoforms such as NMNAT2 in controlling axonal integrity, a related role for NMNAT1 remains possible in the absence of data from homozygous null mice. In particular, it is possible that the level of this enzyme activity in heterozygous KO could remain above a threshold level needed to significantly modify axon degeneration after an injury. The availability of NMNAT1 floxed mice will enable us to address this question in a future study by generat- ing conditional KOs where the NMNAT1 gene is inac- tivated only in neurons at postnatal stages, overcoming the embryonic lethality of a complete null mutant. In conclusion, NMNAT1 is indispensable for the normal development of the embryo and NMNAT2 and 3 cannot compensate for its loss. Decreased NMNAT1 activity in heterozygous null mice, however, does not affect the rate of Wallerian degeneration, sug- gesting that endogenous NMNAT1 does not have a primary role in axon maintenance. Materials and methods Construction of the targeting vector We determined the genomic sequence of the entire mouse Nmnat1 coding region and used this to design a targeting vector based on the plasmid pEASYFlox (a gift from W. Mu ¨ ller and K. Rajewsky). The positive selection marker, G418 ⁄ neomycin (NEO R ), is flanked by two loxP sites. To maximize the likelihood of achieving complete gene inacti- vation, we chose to delete a region comprising the first and second exons, including some 5¢ UTR where the promoter is located. This region was amplified by PCR with SalI tagged primers and cloned into the SalI site of the targeting vector. Two additional homology regions, a 5¢ 2.3 kb region and a 3 ¢ 4.6 kb region, were then obtained by PCR using primers tagged with NotI ⁄ Bam HI and HindIII sites respectively and cloned into the respective restriction sites of pEASYFlox. We confirmed the absence of PCR and cloning artefacts by sequencing all coding regions, the loxP sites and most non-coding regions. The genomic locus, the completed targeting vector and the recombination events are shown in Fig. S1. The primer pair sequence was as follows: 5¢ homology arm (NotI and BamHI site underlined and italics) 5¢-AGGAAAAAA GCGGCCGCACACTTACAGCCTGAG GCG-3¢,5¢-CGC GGATCCACTCCAAGGATACACTCC GA-3¢;3¢ homology arm (HindIII site underlined and ital- ics) 5¢-GGCCC AAGCTTATATATTTGCCTAGGAGGGT NMNAT1 gene inactivation and axon degeneration L. Conforti et al. 2674 FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS C-3¢,5¢-GGCCCAAGCTTAAGACAGTGTGGAGGAGA CCT-3¢. The targeted region (SalI site underlined and ital- ics) was 5¢-CAACGC GTCGACCCATGTGCTGAAAGCT TGGT-3¢,5¢-ACTGGC GTCGACTTGAATGTCTTAGTG ACTGGG-3¢. All primers were purchased by Sigma- Genosys, Haverhill, UK. All chemicals were obtained by Sigma-Aldrich, Gillingham, UK, unless otherwise stated. ES cell electroporation and isolation of a double recombinant clone for blastocyst injection The 18 kb targeting vector was linearized with NotI and elec- troporated into Bruce 4 ES cells (from C57BL ⁄ 6 strain, kind gift of K. Rajewsky and A. Egert). ES cell clones were posi- tively selected 24 h post electroporation with 0.2 mgÆmL )1 G418. Negative selection of random integration was per- formed by addition of 2 · 10 )6 m ganciclovir to the medium. We picked 384 clones among the ones that were resistant to both selection agents. Southern blot analysis showed that only one clone contained the entire targeting vector correctly integrated at both homology arms of the genomic locus. This clone was electroporated again in vitro with a Cre expression vector (pPGK-Cre-bpA, kind gift of K. Rajewsky and W. Mu ¨ ller). This allowed us to delete the NEO R gene and leave a loxP flanked region amenable to conditional or con- ventional deletion. One subclone was then isolated that had lost the NEO R cassette and contained a ‘floxed’ targeted locus (Fig. 1A). We designed primers spanning the two loxP sites (Pr1, Pr2, Pr3, Pr4, see Fig. 1A) to confirm the presence and the integrity of the loxP sites in the floxed clone after Cre-mediated deletion. The PCR across the loxP sites con- firmed the presence of both loxP sites in the targeted clone (Fig. 1B). Furthermore, sequencing of the PCR products confirmed that the loxP sites were correct. Generation of targeted mice The Bruce 4 targeted ES cell clone containing the floxed locus was used for injection into BALB ⁄ c derived blast- ocysts. Chimeric mice, originally identified by coat colour, were then confirmed by southern blotting (see Fig. 1B). Chimeric mice were backcrossed to C57BL ⁄ 6 mice and the transmission of the mutant allele to the progeny was revealed by coat colour analysis and southern blotting. Nmnat1 + ⁄ ) mice were obtained by crossing the floxed Nmnat1 male chimerics to female C57 ⁄ BL6 K14 Cre mice to produce heterozygous null mice on a black background [27]. Southern blot analysis demonstrated that about 50% of the offspring are heterozygous for the full deletion allele. The heterozygous mice were then intercrossed in an attempt to generate homozygous null mutants. Animal work was performed in accordance with the relevant German and UK government animal welfare legislation under licenses K13, 11 ⁄ 00 (Cologne, Germany) and 80 ⁄ 1778 and 80 ⁄ 2254 (Cambridge, UK). Preparation and analysis of DNA from ES cells, mice and embryos Genomic DNA was isolated using standard protocols [21]. For southern blot analysis, genomic DNA from ES cells was digested with EcoRI or HindIII and analysed with a 420 bp 5¢ probe and a 750 bp 3¢ probe located outside the targeted region (Fig. S1) and generated by PCR from genomic DNA with the following primer pairs: 3¢ probe, 5¢-AAT ATTTGGAA TTAGGTAA GTGT-3¢, 5¢-GTGTAAAAGACACTGTGATG-3¢;5¢ probe, 5¢-TGT CTTAAAATGCACTTCAAAC-3¢,5¢-GTCGAGTTGCCA TGCAGAG-3¢. Another 450 bp probe (called probe 4, Fig. 1A) obtained by mouse genomic DNA PCR with the primers 5¢-GGCCCAAGCTTATATATTTGCCTAG GAGGGTC-3¢ and 5¢-TCAGACATTTATAAGTTTCG GG-3¢ was used on southern blots of tail genomic DNA digested with BamHI to identify both Nmnat1 floxed mice and Nmnat1 heterozygous KO mice. PCR screening of those mice used the following primers spanning loxP site 1 and loxP site 2: Pr1, 5¢-TCGGAGTGTATCCTTG GAGT-3¢; Pr2, 5¢-ACCAAGCTTTCAGCACATGG-3¢; Pr3, 5¢-CCCAGTCACTAAGACATTCAA-3¢; Pr4, 5¢-GA CCCTCCTAGGCAAATATA-3¢. Western blotting, NMNAT enzyme activity assay and NAD(P) + level determination Western blotting of sagittally divided half brains was per- formed as described previously [22]. Sagittally divided half brains were homogenized in five volumes of RIPA buffer [phosphate-buffered saline (PBS) containing 1% NP40, 0.5% deoxycholate, 0.1% sodium dodecylsulphate]. High-speed supernatant was diluted to approximately 0.5 mgÆmL )1 total protein according to the Bradford assay (BioRad, Hemel Hempstead, UK) and fractionated by standard SDS ⁄ PAGE. After semidry blotting (BioRad, Hemel Hempstead, UK), nitrocellulose membranes (Bio- Rad) were blocked in PBS plus 0.02% Tween-20 and 5% low-fat milk powder before incubation with primary anti- body and then horseradish peroxidase conjugated second- ary antibody (1 : 3000; Amersham Biosciences, Little Chalfont, UK). Proteins were visualized using the ECL detection kit (Amersham Biosciences, Little Chalfont, UK) according to the manufacturer’s instructions. For quantification, western blot band intensities were deter- mined with image j software and normalized to b-actin. NAD + and NAD(P) + levels were determined in brain or whole cell extracts by HPLC identification or by a fluori- metric cyclic reaction as described previously [41,42]. Total NMNAT enzyme activity was determined as described earlier [41]. Tissue was suspended in six volumes of 50 mm Hepes, pH 7.4, 0.5 mm EDTA, 1 mm MgCl 2 , 1mm phenylmethylsulphonyl fluoride and 0.02 mgÆmL )1 each of leupeptin, antipain, chymostatin and pepstatin, L. Conforti et al. NMNAT1 gene inactivation and axon degeneration FEBS Journal 278 (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS 2675 [...]... adding 4 lL of 50 mm NMN and stopped by the addition of a half-volume of ice-cold 1.2 m HClO4 After 10 min at 0 °C, the mixture was centrifuged and 135 lL of supernatant was neutralized by the addition of 36 lL of 0.8 m K2CO3 NMNAT activity was calculated after HPLC identification and quantification of the product (NAD+) One unit of enzyme was defined as the amount capable of producing 1 lmol of NAD+ per... information The following supplementary material is available: Fig S1 Targeting of the Nmnat1 gene Fig S2 NMNAT1 expression analysis in wild-type and Nmnat1+ ⁄ ) mouse brain This supplementary material can be found in the online version of this article NMNAT1 gene inactivation and axon degeneration Please note: As a service to our authors and readers, this journal provides supporting information supplied by the. .. D4Cole1e chimeric protein and overexpression of Rbp7 in the slow Wallerian degeneration (WldS) mouse Proc Natl Acad Sci USA 97, 11377–11382 Fernando FS, Conforti L, Tosi S, Smith AD & Coleman MP (2002) Human homologue of a gene mutated in the slow Wallerian degeneration (C57BL ⁄ Wld(s)) mouse Gene 284, 23–29 Di Stefano M, Galassi L & Magni G (2010) Unique expression pattern of human nicotinamide mononucleotide... dilutions of cDNA were performed for each set of primers to establish PCR efficiencies Relative expression ratios in comparison with the b-actin reference gene were determined as described in [43] and statistical analysis was performed using the t-test 2676 Nerve lesion Nerve lesions to assess the rate of Wallerian degeneration were performed as described in [7,21] in 2–10 months old wild-type or Nmnat1+ ... Genazzani AA (2008) Characterization of NAD uptake in mammalian cells J Biol Chem 283, 6367–6374 Pfaffl MW, Horgan GW & Dempfle L (2002) Relative expression software tool (REST) for group-wise comparison and statistical analysis of relative expression results in real-time PCR Nucleic Acids Res 30, e36 Buckmaster EA, Perry VH & Brown MC (1995) The rate of Wallerian degeneration in cultured neurons from... separation from the cell body mass using a scalpel, and the degeneration of the isolated neurites was followed at different time points for 24 h after cut Bright field images were acquired on a microscope (IX8I; Olympus, Southendon-Sea, UK) coupled to a digital camera (U-TV 0.5XC; Olympus, Southend-on-Sea, UK) using analysis software (Soft Imaging Systems GmbH, Muenster, Germany) Axon degeneration was... Coleman MP (2005) The progressive nature of Wallerian degeneration in wildtype and slow Wallerian degeneration (WldS) nerves BMC Neurosci 6, 6 Eckel-Mahan K & Sassone-Corsi P (2009) Metabolism control by the circadian clock and vice versa Nat Struct Mol Biol 16, 462–467 Nakahata Y, Sahar S, Astarita G, Kaluzova M & Sassone-Corsi P (2009) Circadian control of the NAD+ salvage pathway by CLOCK-SIRT1 Science... (2011) 2666–2679 ª 2011 The Authors Journal compilation ª 2011 FEBS 2677 NMNAT1 gene inactivation and axon degeneration 23 24 25 26 27 28 29 30 31 32 33 L Conforti et al et al (2007) NAD(+) and axon degeneration revisited: Nmnat1 cannot substitute for Wld(S) to delay Wallerian degeneration Cell Death Differ 14, 116–127 Araki T, Sasaki Y & Milbrandt J (2004) Increased nuclear NAD biosynthesis and SIRT1 activation... Laboratories, Peterborough, UK) The fluorophore used was YFP Axon degeneration was quantified as described in [21] by counting all (intact and fragmented) fluorescent axons and calculating the percentage of intact axons in three different fields per nerve explant examined Statistical analysis was performed using the t-test unless specified otherwise in the figure legends Assessment of axon degeneration in SCG cultures... Morreale for critically reading the manuscript and Dr Gloria Esposito for helpful advice This work was funded by the BBSRC, the Centre for Molecular Medicine of the University of Cologne (ZMMK) grant NG3 and the Mario Negri Pharmacological Research Institute (Milan) References 1 Magni G, Amici A, Emanuelli M, Orsomando G, Raffaelli N & Ruggieri S (2004) Structure and function of nicotinamide mononucleotide . Reducing expression of NAD + synthesizing enzyme NMNAT1 does not affect the rate of Wallerian degeneration Laura Conforti 1,2,3 ,. The embryonic lethality of NMNAT1 full inactivation precludes the possibility of testing the rate of Wallerian degeneration in the com- plete absence of