Analysis of the nicotinamide phosphoribosyltransferase family provides insight into vertebrate adaptation to different oxygen levels during the water-to-land transition

Abstract

One of the most important events in vertebrate evolutionary history is the water-to-land transition, during which some morphological and physiological changes occurred in concert with the loss of specific genes in tetrapods. However, the molecular mechanisms underlying this transition have not been well explored. To explore vertebrate adaptation to different oxygen levels during the water-to-land transition, we performed comprehensive bioinformatics and experimental analysis aiming to investigate the NAMPT family in vertebrates. NAMPT, a rate-limiting enzyme in the salvage pathway of NAD+ biosynthesis, is critical for cell survival in a hypoxic environment, and a high level of NAMPT significantly augments oxidative stress in normoxic environments. Phylogenetic analysis showed that NAMPT duplicates arose from a second round whole-genome duplication event. NAMPTA existed in all classes of vertebrates, whereas NAMPTB was only found in fishes and not tetrapods. Asymmetric evolutionary rates and purifying selection were the main evolutionary forces involved. Although functional analysis identified several functionally divergent sites during NAMPT family evolution, in vitro experimental data demonstrated that NAMPTA and NAMPTB were functionally conserved for NAMPT enzymatic function in the NAD+ salvage pathway. In situ hybridization revealed broad NAMPTA and NAMPTB expression patterns, implying regulatory functions over a wide range of developmental processes. The morpholino-mediated knockdown data demonstrated that NAMPTA was more essential than NAMPTB for vertebrate embryo development. We propose that the retention of NAMPTB in water-breathing fishes and its loss in air-breathing tetrapods resulted from vertebrate adaptation to different oxygen levels during the water-to-land transition.

Abbreviations

And1

actinodin 1

And2

actinodin 2

GFP

green fluorescent protein

HEK

human embryonic kidney

hpf

hours postfertilization

LRT

likelihood ratio test

ME

minimum evolution

ML

maximum likelihood

MO

morpholino

MT

mutant-type

NAMPT

nicotinamide phosphoribosyltransferase

NJ

Neighbor-joining

OGD

oxygen-glucose deprivation

STD

standard control

TCA

tricarboxylic acid cycle

WT

wild-type

Introduction

One of the most important events in vertebrate evolutionary history is the transition from water to land, during which a series of critical characters changed, including fin-to-limb transition, olfactory organ innovation and remodeling of the ear [1-3]. As vertebrates transitioned to new terrestrial environments, these changes in the morphological and physiological characters were concurrent with the loss of specific genes in tetrapods [4, 1]. For example, actinodin 1 and 2 (And1 and And2), which are essential structural components of elastoidin, are present in teleost fishes and the elephant shark but were lost during tetrapod evolution. Double gene knockdown of and1 and and2 in zebrafish embryos led to the loss of lepidotrichia, which may be conducive to the fin-to-limb transition [5]. Recently, it was reported that more than 50 genes were lost in tetrapods compared to coelacanth and zebrafish genomes, with these genes likely being associated with the development of fins, otoliths, tails and some other tissues [4]. However, the detailed molecular mechanisms underlying these transitions have not been well studied. We found that the nicotinamide phosphoribosyltransferase (NAMPT) family mainly consisted of two members (NAMPTA and NAMPTB) in vertebrates resulting from second round whole-genome duplications (2R). NAMPTA existed in all classes of vertebrates. NAMPTB was only found in water-breathing vertebrates, including cartilaginous fish, ray-finned fish and lobe-finned fish but not in air-breathing tetrapods. To advance the study of the evolutionary trajectory of vertebrates during the move from water to land, we comprehensively investigated the NAMPT family.

NAMPT, also called visfatin/pre-B-cell colony-enhancing factor, is an important protein, catalyzing the first limiting reaction in the synthesis of NAD⁺ from nicotinamide. It is an ancient gene found in both prokaryotes [6] and eukaryotes [7]. NAMPT was first identified as a presumptive cytokine-like protein from a human peripheral blood lymphocyte cDNA library [8] and its rediscovery as the key enzyme in NAD⁺ generation has broadened its potential biological function [6]. NAMPT has a ubiquitous expression in several cells, tissues and organs, including bone marrow, adipose tissue, liver, skeletal muscle, immune cells and the brain [8-12]. This widespread distribution indicates a broad function of NAMPT with respect to both physiology and pathophysiology. For example, NAMPT could elevate intracellular NAD⁺ levels and increase cell resistance to genotoxic damage induced by oxidative stress, alkylating agents or ionizing radiation [13, 14]. Moreover, NAMPT has an impact on several inflammatory diseases such as acute lung injury [15], rheumatoid arthritis [16], inflammatory bowel disease [17] and ischemic brain disorders [18]. By regulating the NAD⁺ level to influence both cell viability and the inflammatory response, NAMPT may indicate a potential pharmacological target.

Recently, it was reported that the gene for NAMPT was a direct target of hypoxia-inducible factor-2α and was upregulated in hypoxic environments [16, 19]. It also had neuroprotective properties against ischemia-like oxygen-glucose deprivation (OGD) because overexpression of NAMPT significantly attenuated the noxious effect of OGD on cell viability and apoptosis by maintenance of the NAD⁺ pool [20, 21, 18]. A high level of NAMPT may thus play an important role in cellar survival in a lower oxygen environment. Water-breathing fishes confront oxygen limitation more often than do air-breathing tetrapods. Movement to terrestrial environments, with higher oxygen levels than aquatic environments, has driven vertebrates to evolve a series of strategies that enable adaption to different oxygen levels during the water-to-land transition [22]. The present study aimed to determine whether the retention of NAMPTB in water-breathing fishes and its loss in air-breathing tetrapods may be an adaptation for vertebrates to live in environments with higher oxygen levels. Previous research has focused on NAMPTA, with little information being available for NAMPTB. Therefore, we performed a comprehensive analysis including the origin, characterization, evolution, expression and, in particular, the biofunction of NAMPT family members. The results of the study will increase our understanding of the NAMPT family and allow further exploration of the molecular mechanisms underlying the adaptations of vertebrates in the transition from water to land.

Results and Discussion

NAMPTA has been reported in different vertebrate lineages such as fish [23], birds [24] and mammals [25], although nothing is known about the presence of NAMPTB. In the present study, the origination, evolution and functions of NAMPTB were analyzed for the first time in vertebrates.

Phylogenetic analysis and shared synteny

Phylogenetic reconstruction using maximum likelihood (ML), Neighbor-joining (NJ) and minimum evolution (ME) methods resulted in similar topologies (Fig. 1). NAMPTA and NAMPTB clades were well supported in these three phylogenetic reconstructions. However, the ML algorithms were well supported for the interior branches, although there were some minor topological differences at the terminal branches. Therefore, the ML tree was selected for further study. In the ML phylogenetic tree (Fig. 1A), NAMPTA and NAMPTB clustered into independent clades generally following the evolutionary order described previously [26, 27, 4]. Tree topology analyses showed a clear distinction of NAMPTA and NAMPTB presence in vertebrates. Interestingly, the NAMPTA subfamily existed in all classes of vertebrates that were surveyed, whereas the NAMPTB subfamily was only found in water-breathing vertebrates, including cartilaginous fish, ray-finned fish and lobe-finned fish but not in any air-breathing tetrapod species. There were two rounds of whole-genome duplication events in the early evolution of vertebrates. The first occurred before the existence of the common ancestor of vertebrates, and the second occurred before the divergence of lamprey from jawed vertebrates or somewhat later [28-30]. The NAMPT duplicates present in elephant shark indicated that NAMPT was an ancient gene family. Based on the phylogenetic analysis of NAMPT family members, the results suggested that NAMPT duplicates might have arisen from second round whole-genome duplication events (2R) because it occurred prior to the divergence of cartilaginous fishes and after the divergence of jawless vertebrates [30, 31].

Figure 1

Phylogenetic tree of the NAMPT family in selected metazoan lineages. (A) Amino acid sequences of full-length NAMPT were analyzed using the ML method with the LG+I+G substitution model and 100 repetitions. Numbers above branches indicate bootstrap support values. (B) Amino acid sequences of full-length NAMPT were analyzed using NJ and ME methods under the Poisson model and bootstrap values were calculated from 2000 repetitions. Numbers above branches indicate bootstrap support values > 50% (NJ/ME). The air-breathing and water-breathing vertebrates are shown in boxes. 1R and 2R represent the first round whole-genome duplication and the second round whole-genome duplication, respectively.

We also observed that NAMPTB was absent from the tetraodon and fugu genomes. A shared genomic gene order flanking the NAMPTB loci revealed extensive conservation of synteny, and also confirmed that NAMPTB was absent from tetrapod and tetraodon genomes (Fig. 2). However, the scenarios of NAMPTB loss were different. In tetrapods, genes for NAMPTB were lost but genes flanking the regions were still conserved when comparing xenopus and human chromosomes with spotted gar, coelacanth and zebrafish chromosomes (Fig. 2A). Thus, we speculate that the loss of NAMPTB in tetrapod was probably adaptive. In tetraodon, the synteny analysis showed that an interchromosomal exchange occurred near the NAMPTB region (including SYNPR, NAMPTB, CADPS, FEZF2, C3orf14) compared to the zebrafish chromosome (Fig. 2B). After exchanging with the IP6k2 and SEC61A loci, the NAMPTB region was allocated to chromosome Un_random in tetraodon (Ensembl location: Un_random:13 539 000–13 613 200). Although the CADPS, FEZF2, C3orf14 loci were conserved, the NAMPTB and SYNPR loci were deleted. It was therefore likely that NAMPTB was lost during the interchromosomal exchange process.

Figure 2

Synteny of genes for NAMPTB in vertebrates. Strand information and different genes are indicated by arrows, and genes belonging to the same gene family are coded in the same color. NAMPTB are shown as red arrows. The nonhomologous genes are shown as gray bars. (A) Synteny of genes for NAMPTB in fishes and tetrapods. (B) An interchromosomal change took place in tetrodon NAMPTB loci compared to the zebrafish genome.

Genomic structure

In vertebrates, there are two subfamilies that belong to the NAMPT gene family called the NAMPTA and NAMPTB subfamilies. In terms of exons, the genes for NAMPTA and NAMPTB were homogeneous because both possessed 11 exons and 10 introns (Fig. 3). The intron sizes of NAMPTA or NAMPTB varied among species, whereas the exon sizes were well conserved. The deduced amino acid sequences between NAMPTA and NAMPTB also showed a high degree of identity (Table 1). NAMPTA is more conserved than NAMPTB throughout its sequence. For example, the predicted protein of zebrafish NAMPTA exhibited a high degree of identity with its homologs from humans (88%), medakas (93%) and sharks (83%). Zebrafish NAMPTB also had a high degree of identity with NAMPTB from coelacanths (77%), medakas (84%) and sharks (70%). Moreover, NAMPT showed a mid-amount of identity (61%) between paralogs in zebrafish.

Table 1. Amino acid identity of NAMPTA and NAMPTB in different species

	Mouse a	Pig a	Chicken a	Lizard a	Xenopus a	Coelacanth a	Zebrafish a	Fugu a	Tetraodon a	Cod a	Medaka a	Stickleback a	Platyfish a	Gar a	Shark a	Coelacanth b	Zebrafish b	Cod b	Medaka b	Stickleback b	Platyfish b	Gar b	Shark b
Human a	96	97	94	92	89	90	88	87	88	87	86	88	87	89	83	62	61	60	62	61	62	62	61
Mouse a		98	94	92	91	91	89	88	88	88	88	89	89	89	83	62	61	59	61	61	61	62	61
Pig a			95	92	90	91	89	88	88	88	87	89	88	89	83	62	61	59	61	60	61	62	61
Chicken a				93	90	91	89	89	89	87	88	89	88	89	83	60	60	60	61	60	61	61	61
Lizard a					89	91	88	88	88	86	88	87	89	88	83	61	60	59	61	60	62	61	60
Xenopus a						90	88	87	87	86	86	86	87	88	81	61	59	59	61	60	61	61	60
Coelacanth a							91	90	91	89	91	90	91	91	83	61	60	58	60	60	60	61	60
Zebrafish a								93	94	92	93	94	94	95	83	62	61	59	61	60	61	62	62
Fugu a									98	93	94	94	94	93	82	62	60	59	60	59	60	61	61
Tetraodon a										93	94	94	94	93	82	62	60	59	60	60	61	61	62
Cod a											94	93	93	92	82	62	60	59	60	60	60	61	61
Medaka a												93	95	94	82	61	59	58	59	59	60	60	61
Stickleback a													95	92	82	62	61	59	60	59	60	61	62
Platyfish a														93	82	61	59	58	60	59	60	60	61
Gar a															82	61	60	59	59	59	60	60	62
Shark a																63	61	60	60	60	61	61	61
Coelacanth b																	77	74	74	76	74	76	73
Zebrafish b																		81	84	85	84	82	70
Cod b																			81	82	81	77	69
Medaka b																				86	87	79	70
Stickleback b																					88	79	72
Platyfish b																						78	70
Gar b																							71

• The numbers in gray shades are the amino acid identity between NAMPA and NAMPTB.

Figure 3

Genomic structure of genes for NAMPT in different vertebrate species. Black boxes, white boxes and horizontal lines represent exons, first or last coding exons, and introns, respectively. The numbers above the boxes indicate the length of each coding exon sequence. The numbers under the horizontal line show the length of introns. The numbers under the species indicate the length of corresponding genes and proteins. (a) NAMPTA; (b) NAMPTB.

As shown in Fig. 4, the binding sites for nicotinamide, ribose, phosphate or nicotinamide mononucleotide (Asp16, Arg196, Arg311, Gly353, Asp354 and Gly383) [32] were highly invariant in all of the sequences analyzed for these two subfamilies in vertebrates. The amino acid motifs surrounding catalytic residues Tyr18, Phe193, Asp219, His247, Asp279 and Asp313 [32-36] were highly conserved as well (Fig. 4). However, Gly384, a binding site for phosphate, was substituted with a serine residue (Ser) in NAMPTB in ray-finned fish but was conserved in other species. Long stretches of identical amino acids surrounding important structural and catalytic positions indicated that NAMPT was highly conserved in vertebrates. These conserved genomic structural and functional sites may also indirectly support the hypothesis that NAMPTA and NAMPTB were functionally conserved in basic nicotinamide phosphoribosyltransferase enzymatic function throughout their evolutionary history in vertebrates.

Figure 4

Alignment of the amino acid sequence of NAMPTA and NAMPTB in vertebrate species. Identical and similar amino acid residues are indicated and shaded. The residues involved in binding sites are shown as an inverted triangle. Star, catalytic residues; bar, Type-I divergence sites; two points, Type-II divergence sites. (A) NAMPTA; (B) NAMPTB. Gly384, a binding site of nicotinamide mononucleotide substituted with a serine residue (Ser) in NAMPTB in ray-finned fish is shown in a yellow box.

Model testing of selective pressures

To detect sites under positive selection after the initial divergence event, we used the site models. According to the likelihood ratio test (LRT) of site models (Table 2), the model M3 was statistically better than the null model (M0) in three individual clades, although the model M2a failed compared to the null model (M1). The selective pressure acting on NAMPT revealed that NAMPTA-fish had the lowest value of ω (0.03), NAMPTA-tetrapod had ω of 0.04 and NAMPTB-fish had the highest value of ω (0.07). Although the overall values of ω were lower than one, we found that one site likely evolved under positive selection. For the NAMPTA-fish subunit, model M8 was significantly higher than model M7 (2ΔlnL = 14.74, d.f. = 2, P < 0.01) and the site (336 V) had a posteriori probability of 0.954 when using the Bayes empirical Bayes method. In addition, we compared models M8 and M8a to test whether this site was statistically above neutrality, and found no significant difference (2ΔlnL = 0.82, d.f. = 1, P = 0.36). This may indicate that the site has been evolving under neutral evolution rather than under positive selection. For the NAMPTA-tetrapod subunit and the NAMPTB-fish subunit, neither of the models with positive selection fit the data significantly better than the null model. Thus, no positive selection was detected by these models. This analysis of selective pressure indicated that NAMPTA-fish, NAMPTA-tetrapod and NAMPTB-fish all have been generally under strong selective constraints, in accordance with their important biological roles in both physiology and pathophysiology.

Table 2. Maximum likelihood analysis using site models in NAMPTA-fish, NAMPTA-tetrapod and NAMPTB-fish

Gene	Model (number of parameters)	Paramenters estimates	lnL	LRT pairs	2ΔlnL	P value
NAMPTA-fish	M0: one ratio (19)	ω = 0.03	−7543.40
	M3: discrete (23)	p = 0.00 0.83 0.17 ω = 0.00 0.01 0.15	−7402.04	M0/M3	282.72	< 0.01
	M1a: neutral (20)	p = 0.96 0.04 ω = 0.02 1.00	−7478.23
	M2a: selection (22)	p = 0.96 0.04 0.00 ω = 0.02 1.00 36.15	−7478.23	M1a/M2a	0	NS
	M7: beta (20)	p = 0.17 q = 4.43	−7395.07
	M8: beta&ω (22)	p0 = 1.00 p = 0.19 q = 5.66 (p1 = 0.00) ω = 1.73	−7387.70	M7/M8	14.74	< 0.01
	M8a (null) (21)		−7388.11	M8/M8a	0.82	0.36
NAMPTA-tetrapod	M0: one ratio	ω = 0.04	−4562.30
	M3: discrete	p = 0.70 0.25 0.05 ω = 0.00 0.09 0.53	−4501.20	M0/M3	122.2	< 0.01
	M1a: neutral	p = 0.95 0.05 ω = 0.02 1.00	−4516.31
	M2a: selection	p = 0.95 0.05 0.00 ω = 0.02 1.00 7.69	−4516.31	M1a/M2a	0	NS
	M7: beta	p = 0.11 q = 2.04	−4502.28
	M8: beta&ω	p0 = 1.00 p = 0.11 q = 2.04 (p1 = 0.00) ω = 1.00	−4502.28	M7/M8	0	NS
NAMPTB-fish	M0: one ratio	ω = 0.07	−7786.32
	M3: discrete	p = 0.62 0.31 0.07 ω = 0.01 0.15 0.46	−7632.17	M0/M3	308.3	< 0.01
	M1a: neutral	p = 0.92 0.08 ω = 0.06 1.00	−7714.40
	M2a: selection	p = 0.92 0.04 0.03 ω = 0.06 1.00 1.00	−7714.40	M1a/M2a	0	NS
	M7: beta	p = 0.35 q = 3.47	−7632.93
	M8: beta&ω	p0 = 1.00 p = 0.35 q = 3.47 (p1 = 0.00) ω = 1.00	−7632.93	M7/M8	0	NS

• Note the models (M0, M3,M1a, M2a, M7, M8 and M8a) and the corresponding calculated likelihood. NAMPTA-fish, NAMPTA-tetrapod and NAMPTB-fish clades are shown in Fig. 5. lnL, log-likelihood; NS, the LRT is not signifiant for each pairwise comparison.

To analyze whether NAMPT are under divergent selective pressure, we analyzed selective pressure among NAMPTA-fish, NAMPTA-tetrapod and NAMPTB-fish clades. Based on comparisons of amino acid nonsynonymous with the synonymous substitution rate ratios (d_N/d_S) in the clade model C, we found that NAMPTA-fish, NAMPTA-tetrapod and NAMPTB-fish subfamilies were all subjected to stringent purifying selection in vertebrates (Table 3). In clade model C (CmC) A, CmC B and the multi-model, all estimated ω values for the divergent selection site class were less than one. These models fit the data significantly better than the null model. Even so, significant differences in selective constraints appeared when the subfamilies were compared. In 43% of the sites, selection pressure on NAMPTA showed a much lower ratio of d_N/d_S than that of NAMPTB (ω = 0.07 and ω = 0.15–0.16, respectively). The decreased constraints in NAMPTB suggested that NAMPTB was less functionally important than NAMPTA. This may explain why NAMPTB rather than NAMPTA was lost in tetrapods. After the loss of NAMPTB in tetrapods, NAMPTA-tetrapod was less constrained compared to NAMPTA-fish (p₂ = 0.428; ω₂ = 0.063; ω₃ = 0.094).

Table 3. Parameters estimation and LRTs for the clade models

Model (number of parameters)	SC0		SC1		SC2		lnL	2ΔlnL	P value
	ω0	p 0	ω1	p 1	ω2 : ω3 : ω4	p 2
CmC A (53)	0.009	0.539	1.000	0.030	ω2 = 0.149 ω3 = 0.071	0.431	−18875.559	50.312	< 0.01
CmC B (53)	0.008	0.553	1.000	0.029	ω2 = 0.071 ω3 = 0.160	0.418	−18869.732	61.966	< 0.01
Multi-clade (54)	0.009	0.544	1.000	0.028	ω2 = 0.149 ω3 = 0.063 ω4 = 0.094	0.428	−18871.746	7.626	< 0.01
M2a_rel (52)	0.102	0.401	1.000	0.026	ω2 = 0.010	0.573	−18900.715

• The models (CmC A, CmC B and multi-clade) are shown in Fig. 5. The model M2a_rel is used as the null model. The two tree partitions model is used as the null model to analyze the multi-clade model. SC, site class; lnL, log-likelihood.

Because it was unlikely that positive selection affected all sites over a long period of time, we used branch-site models to increase the power of our tests for positive selection. We applied these models to test the potential for positive selection at specific sites in a total of 46 branches separately. There were only 18 sites in 10 branches (A–J) where specific sites may have been subjected to positive selection (Fig. 5 and Table 4). The limited number of sites potentially affected by positive selection confirmed that NAMPT was a stable family and that purifying selection was the major force involved. This was consistent with the view that duplicates retained over long evolutionary periods were most likely a result of purifying selection [37].

Table 4. Parameters estimation and LRTs for the branch-site models

Branch	lnL		2ΔlnL	Positive selection sites (Bayes empirical Bayes)
	Alternative	Null
A	−19184.675	−19188.272	7.194**	297 H
B	−19178.361	−19182.802	8.882**	229 K, 304T, 438G, 469K
C	−19184.396	−19189.205	9.618**	271 S,
D	−19185.213	−19189.205	7.984**	338 S
E	−19183.583	−19186.787	6.408*	188 Y, 339 K
F	−19178.361	−19189.205	21.688**	99 K, 151 I, 268 Q
G	−19173.020	−19176.487	6.934**	29 N, 140 E,
H	−19185.392	−19187.965	5.146*	307 P
I	−19184.939	−19189.205	8.532**	85 D,
J	−19167.294	−19169.406	4.224*	174 K, 340 G

• Branch-site models are used to identifying the potential for positive selection in all 46 branches separately (Fig. 5). Only 10 branches (A–J) in the tree are where specific sites may have been subjected to positive selection. lnL, log-likelihood. *P < 0.05; **P < 0.01 (chi-squared test). Sites potentially under positive selection are identified using the human sequence in Fig. 2 as the reference. Sites highlighted in italic indicate sites with a posterior probability > 99%.

Figure 5

Phylogenetic tree of NAMPT used to test the roles of selection. The NAMPT family is divided into three parts: NAMPTA-fish (solid, red branches); NAMPTA-tetrapod (dotted, red branches); and NAMPTB-fish (dashed, blue branches). Lancelet is used as the outgroup (solid, black branch). Ten branches (A–J) listed in Table 4 may be subject to positive selection. Sites potentially under positive selection are on the branches. The sites highlighted in italic indicate sites with a posterior probability > 99%. (a) NAMPTA; (b) NAMPTB.

These sites were then mapped onto the sequence alignment and the tertiary structure of human NAMPT (Figs 4 and 6A). The distribution of these sites was greatly disordered but these sites were concentrated in some regions. Nine and eight positively selected sites were situated within the turns and helices, respectively, although no site was found in a strand. Interestingly, the nine positively selected sites in turns were all exposed on the surface of NAMPT. Turns of the protein molecule may play some significant biological roles in molecular recognition because they are tend to be solvent when exposed to the hydrophobic core [38]. As bioactive structures, they often interact with other molecules such as receptors, enzymes or antibodies [39]. Thus, these nine positively selected sites, which were situated in turns and exposed on the surface, may be related to substrate recognition. This was consistent with the characteristics of NAMPT as a multifunctional protein. Other sites distributed in helices may have other unknown functions. For example, they may be related to structure stabilization of proteins. A better understanding of these positively selected sites needs further investigation.

Figure 6

Localization of NAMPT sites of positive selection and functional divergence. The tertiary structure of human NAMPT with enzyme monomers in light blue and gray. (A) Localization of the positively selected sites in the model. The positively selected sites are shown as red spheres. (B) Localization of Type-I sites in the model. Type-I sites are shown as green spheres. (C) Localization of Type-II sites in the model. Type-II sites are shown as yellow spheres.

Functional divergence

Gene duplication has been described as an important source of new genes, and several models of molecular evolution can be used to explain why duplicated copies have been maintained over long evolutionary time periods, such as conservation [40], subfunctionalization [41], neofunctionalization [42] and subneofunctionalization [43]. Our previous study suggested that NAMPT duplicates were maintained mainly as a result of gene conservation. However, NAMPT is a multifunctional protein, and a complete redundancy over long evolutionary time periods is unlikely because mutational pressure will ultimately result in the nonfunctionalization of one of the duplications. To evaluate potential functional divergence between NAMPTA and NAMPTB, Type-I and Type-II functional divergence was estimated by a posteriori analysis. In Type-I functional analysis, the coefficients of divergence (θ_I = 0.446 ± 0.060) values were strongly significantly greater than 0 (P < 0.01) (Fig. 7A and Table 5), indicating amino acid site-specific selective constraints on the NAMPT protein family. After consecutively removing sites with the highest values of functional divergence until the LRT in the Type-I functional divergence was not significant (P > 0.01), we defined the critical cut-off value for the comparisons as 0.78. There were 16 predicted sites above this value. The functional divergence analysis for Type-II (θ_II) revealed a similar trend (θ_II = 0.1042 ± 0.047; P < 0.05) (Fig. 7B and Table 5). Applying an empirical cut-off value of 8.0 (representing sites with more than 89% probability of being functionally divergent), the comparison showed 10 sites above this value. If the cut-off value was decreased to 2.0, more than 25 sites showed evidence for Type-II functional divergence. These results from the analysis of Type-I and Type-II functional divergence indicated that the NAMPT family should be significantly functionally divergent as a result of differences in their evolutionary rates and amino acid properties at specific sites. This also implied that NAMPTA and NAMPTB may have been subjected to neo-/sub-functionalization on specific regions.

Table 5. Type-I and II functional divergence between NAMPTA and NAMPTB

Type-I				Type-II
θI ± SE	LRT	P	Cut-off > 0.78	θII ± SE	P	Cut-off > 8.0	Cut-off > 2.0
0.446 ± 0.060	55.97	< 0.01	16	0.1038 ± 0.046	< 0.05	10	25

• Coefficients of Type-I (θ_I) and II (θ_II) functional divergence are estimated using diverge, version 3.0.

Figure 7

Site-specific profile for functional divergence between NAMPTA and NAMPTB in vertebrate. (A) Type-I functional divergence. (B) Type-II functional divergence. Horizontal lines indicate different cut-off values.

These sites were then mapped onto the secondary and tertiary structure of human NAMPT (Figs 4 and 6B,C). Interestingly, almost all of the sites were situated far away from the active sites. Only one site (Ile 310) was adjacent to the binding site (Arg311). This suggested that NAMPTA and NAMPTB were still functionally conserved in basic nicotinamide phosphoribosyltransferase enzymatic function in the NAD⁺ salvage pathway. Thus, the functional divergence and conservation of these amino acid sites possibly resulted from the existence of long-term selective pressures.

NAD⁺ formation function

To validate whether zebrafish NAMPTA and NAMPTB maintained the rate-limiting enzyme in the salvage pathway for NAD⁺ formation, we measured NAD⁺ and NADH levels, followed by calculating the average NAD⁺/NADH ratio in both control and experimental groups. Average NAD⁺/NADH ratios in the control, pCMV-tag2C-nampta and pCMV-tag2C-namptb groups were 1.14, 1.42 and 1.45, respectively (Fig. 8). The NAD⁺/NADH ratios of pCMV-tag2C-nampta and pCMV-tag2C-namptb groups significantly increased by 24.5% (P < 0.05) and 27.2% (P < 0.05), respectively, compared to the control group. There was no significant difference in NAD⁺/NADH ratios between the two experimental groups. This experimental data demonstrated that NAMPTA and NAMPTB were probably equivalent in basic nicotinamide phosphoribosyltransferase activity in the NAD⁺ salvage pathway. This concurred with the conserved genomic structure and constrained selection pressure. Western blot showed that, in the recombinant zebrafish, pCMV-tag 2C-nampta and pCMV-tag 2C-namptb proteins were expressed in experimental groups, and the molecular masses were ~ 55 kDa, which is consistent with predictions (Fig. 8).

Figure 8

The average NAD+/NADH levels and corresponding western blot detection. Columns represent the mean fluorescence intensity values. Blue columns: fluorescence intensity value of NAD⁺. Red columns: fluorescence intensity value of NADH. The error bars indicate the SEs, and the numbers above the line are the average NAD⁺/NADH ratios. **P < 0.01 compared to the control group. The corresponding western blot result is shown in the box. (a) pCMV-tag2C-nampta. (b), pCMV-tag2C-namptb.

Expression analysis

We chose the zebrafish model to examine the expression profile of the genes for NAMPTA and NAMPTB during vertebrate embryo development. In situ hybridization experiments of zebrafish embryos revealed similar expression patterns of NAMPTA and NAMPTB at early developmental stages (Fig. 9). By 24 h postfertilization (hpf), NAMPTA and NAMPTB were widely expressed across many tissues, with enhanced levels of expression in the anterior head region, including the brain, eyes, otic vesicles and somite boundaries. By 48 hpf, NAMPTA and NAMPTB were still predominantly expressed in the anterior head region, somite boundaries and pectoral fins. The expression of transcripts in the body was reduced by 72 hpf but was still high in the anterior head region, pectoral fins and intestine. The widespread distribution of NAMPT transcripts implied broad and significant functions in zebrafish embryo development. Thus, equivalent NAD⁺ biosynthetic activity and similar expression patterns between NAMPTA and NAMPTB may represent a partially redundant function in zebrafish.

Figure 9

Zebrafish NAMPTA and NAMPTB are expressed ubiquitously during development. (A) Spatio-temporal expression of zebrafish NAMPTA. (B) Spatio-temporal expression of zebrafish NAMPTB. A1 and B1, 12 hpf, lateral view; A2 and B2, 24 hpf, dorsal view; A3 and B3, 24 hpf, lateral view; A4 and B4, 48 hpf, lateral view; A5 and B5, 72 hpf, lateral view. e, Eye; mb, midbrain; hb, hindbrain; mhb, midbrain–hindbrain boundary; s, somite; fb, fin buds; in, intestine.

Knockdown and rescue experiments

We used knockdown technology to analyze the roles of NAMPTA and NAMPTB during zebrafish embryogenesis. To test the efficiency and specificity of morpholinos, four groups of mixtures [nampta-morpholino (MO) and nampta-green fluorescent protein (GFP)-wild-type (WT), nampta-MO and nampta-GFP-mutant-type (MT), namptb-MO and namptb-GFP-WT, namptb-MO and namptb-GFP-MT] were separately injected into the yolk at the one-cell stage. The nampta-MO (4 ng) and namptb-MO (4 ng) effectively blocked the expression of nampta-GFP-WT and namptb-GFP-WT, respectively, but not the expression of the mutated targets (Fig. 10A).

Figure 10

Morpholino-mediated knockdown of zebrafish NAMPTA and NAMPTB. (A) Efficiency and specificity of morpholinos. A1, nampta-MO+WT: embryos were injected with nampta-MO and a nampta-GFP-WT fusion protein expression vector. A2, nampta-MO+MT: nampta-MO and a nampta-GFP-MT fusion protein expression vector. A3, namptb-MO+WT: namptb-MO and a namptb-GFP-WT fusion protein expression vector. A4, namptb-MO+MT: namptb-MO and a namptb-GFP-MT fusion protein expression vector. (B) Morphology of the NAMPTA knockdown embryos. B1 and B3, morphants of the STD-MO (4 ng) at 25 and 30 hpf. B2, B4 and B5, morphants of the nampta-MO (4 ng) at 25, 30 and 56 hpf. (C) Zebrafish embryos were injected with morpholinos and mRNA and scored at 3 days postfertilization: STD-MO (4 ng), nampta-MO (4 ng) and nampta-MO (4 ng) plus nampta-mRNA (200 pg). White box, wild-type indicated the embryos with no obviously defects; gray box, defect type indicated the embryos with reduced head and a curvature in body shape; black box, dead embryos at 24 hpf. The number above the box indicated the total number of injected embryos.

Next, we injected nampta-MO, namptb-MO and standard-MO into zebrafish yolks separately. We detected that 4 ng of nampta-MO consistently produced specific defects during embryonic development (Fig. 10B,C). At 25 hpf, the morphants developed a significantly reduced head and a curvature in body shape compared to control embryos injected with standard control (STD)-MO. At 56 hpf, the morphants showed a more serious curvature in body shape, and lethality was observed by 5 days postfertilization. In mouse, the homozygous NAMPTA knockout (Nampta^−/−) caused lethality in embryos [12]. Similarly, NAMPTA-knockdown zebrafish died during early embryo development. Thus, NAMPTA is conserved and is essential for early embryonic development in vertebrates. However, we did not observe obvious phenotypic defects in embryos injected with namptb-MO from 4 ng to 20 ng. Therefore, it is probable that NAMPTB is not essential for embryo development. Alternatively, the phenotype may have been masked as a result of partial functional redundancy with the remaining NAMPTA transcripts. Because morpholinos are functional for 4–5 days and zebrafish embryos can tolerate low oxygen levels over a wide range, we suggest a further knockout experiment to survey the long-term functions of NAMPTB. These results are consistent with the results obtained in PAML analyses, which identified decreased constraints in NAMPTB compared to NAMPTA. It is likely that NAMPTB acts in a supporting role to NAMPTA.

For the rescue experiments, nampta-MO alone or nampta-MO plus mutant NAMPTA mRNA was co-injected into embryos at the one-cell stage. The phenotype induced by nampta-MO could be successfully rescued by 200 pg of NAMPTA mRNA (Fig. 10C). This demonstrated a direct relationship between nampta-MO mediated knockdown and the changes in phenotype.

Combining selection analyses, functional divergence, NAMPT activity assays, expression patterns and knockdown experiments, we propose that the retention of NAMPTB in water-breathing fishes and its loss in air-breathing tetrapods may be a result of the adaptation of vertebrates to life in environments with different oxygen levels. The hypothesis is based on several reasons. First, aquatic environments exhibit much lower oxygen levels compared to terrestrial environments. Oxygen serves as the terminal electron acceptor in oxidative phosphorylation, and several enzymatic processes in vivo also required molecular oxygen as the direct substrate. Fish evolved a more efficient respiratory system (i.e. countercurrent exchange of gases in gill) than tetrapods for acquiring oxygen to maintain metabolic energy balance. NAMPT was a rate-limiting enzyme in the NAD⁺ salvage pathway, and NAD⁺ acted as a key coenzyme for energy production as a result of its involvement in the mitochondrial tricarboxylic acid cycle (TCA) and the electron transport chain. Although the use of NAD⁺ as a donor of ADP-ribose would lead to a net consumption of NAD⁺ [13, 44], upregulation of NAMPT significantly increased NAD⁺ and ATP concentrations [45, 46]. Thus, two copies of NAMPT may represent a physiologically important homeostatic mechanism, which could constantly maintain an adequate cellular NAD⁺ pool to support an efficient TCA cycle and electron transport chain that is able to match the efficient respiration of fish living in water. Second, NAMPT was a direct target of hypoxia-inducible factor-2α and would be upregulated in an hypoxic environment [16, 19]. NAMPT also acted in a neuroprotective manner against ischemia-like OGD because overexpressing NAMPT significantly attenuated the negative effect of OGD on cell viability and apoptosis by maintaining the intracellular NAD⁺ pool [20, 21]. A high level of NAMPT therefore played an important role in cell survival in a low oxygen environment. We thus suggest that the retention of NAMPTB is an adaptation for fish to live in a lower oxygen environment. However, tetrapods obtain enough oxygen from the air using lungs. Overexpression of NAMPT significantly augmented the production of reactive oxygen species and oxidative stress in human primary pulmonary artery endothelial cells and human primary lung microvascular endothelial cells by increasing the NAMPT–NADH dehydrogenase subunit 1 and NAMPT–ferritin interactions [47]. NAMPT also stimulated NADPH oxidase activity and promoted premature endothelial cell senescence by causing oxidative stress [48]. Oxidative stress has been linked to a series of pathologies by damaging lipids, proteins and DNA [49]. Thus, we suggest that the loss of NAMPTB may be an important mechanism for protecting tetrapods against oxidative stress in a higher oxygen environment.

Conclusions

The present study has demonstrated that NAMPTA and NAMPTB underwent functional divergence after duplication but were still conserved in the basic nicotinamide phosphoribosyltransferase activity. NAMPTA is conserved and essential for vertebrate embryo development, and NAMPTB likely acts in a supporting role to NAMPTA. The retention of NAMPTB in water-breathing vertebrates is an adaptation for fish to live in a low oxygen environment, and the loss of NAMPTB in air-breathing vertebrates may be an important process for tetrapod adaptation to the terrestrial oxygen environment.

Materials and methods

Sequence collection

To determine the presence the genes for NAMPT in vertebrates, we used reciprocal blast best hits to search candidate genes from different databases (with identity > 40%, overlap > 60%, e-value < 1e⁻¹⁰). The sequences of 10 strategically chosen species of major water-breathing vertebrates, including an ancient vertebrate lineage lamprey (Petromyzon marinus), an ancient ray-finned fish spotted gar (Lepisosteus oculatus), a lobe-finned fish coelacanth (Latimeria chalumnae), seven teleost fish [zebrafish (Danio rerio), cod (Gauds morhua), fugu (Takifugu rubripes), medaka (Oryzias latipes), platyfish (Xiphophorus maculates), stickleback (Gasterosteus aculeate) and tetraodon (Tetraodon nigroviridis)], and six species of air-breathing vertebrates [human (Homo sapiens),mouse (Mus musculus),pig (Sus scrofa), chicken (Gallus gallus), anole lizard (Anolis carolinensis) and xenopus (Xenopus tropicalis)] were collected from ENSEMBL (http://www.ensembl.org). A cartilaginous fish, elephant shark (Callorhinchus milii) was retrieved from the NCBI database. Protein sequences for another four outgroup species [lancelet (Branchiostoma floridae), sponge (Suberites domuncula), sea anemone (Nematostella vectensis) and choanoflagellate (Monosiga brevicollis)] were obtained by searching the NCBI protein database via blastp (Table 6). The corresponding cDNA sequences of NAMPT in vertebrates were also retrieved. If the acquired protein or cDNA sequences appeared to be partial ones, we collected the DNA sequences surrounding the best blast hits in the genome contigs and repredicted the gene structure using genescan (http://genes.mit.edu/GENSCAN.html) and gene wise [50]. Then the sequences were additionally verified as homologous with alignments and phylogenies.

Table 6. Sequence collection of NAMPT family

Gene name	Gene ID	Protein ID	Transcript ID
Human_a	ENSG00000105835	ENSP00000222553	ENST00000222553
Mouse_a	ENSMUSG00000020572	ENSMUSP00000020886	ENSMUST00000020886
Pig_a	ENSSSCG00000015435	ENSSSCP00000016366	ENSSSCT00000016815
Chicken_a	ENSGALG00000008098	ENSGALP00000013129	ENSGALT00000013144
Lizard_a	ENSACAG00000014697	ENSACAP00000014491	ENSACAT00000014786
Xenopus_a	ENSXETG00000005918	ENSXETP00000013017	ENSXETT00000013017
Coelacanth_a	ENSLACG00000004482	ENSLACP00000005040	ENSLACT00000005085
Zebrafish_a	ENSDARG00000030598	ENSDARP00000069804	ENSDART00000075320
Tetraodon_a	ENSTNIG00000018493	ENSTNIP00000021667	ENSTNIT00000021902
Fugu_a	ENSTRUG00000016695	ENSTRUP00000042715	ENSTRUT00000042859
Cod_a	ENSGMOG00000011742	ENSGMOP00000012560	ENSGMOT00000012891
Medaka_a	ENSORLG00000013189	ENSORLP00000016533	ENSORLT00000016534
Stickleback_a	ENSGACG00000018838	ENSGACP00000024911	ENSGACT00000024960
Platyfish_a	ENSXMAG00000009407	ENSXMAP00000009449	ENSXMAT00000009463
Spotted gar_a	ENSLOCG00000015835	ENSLOCP00000019495	ENSLOCT00000019527
Coelacanth_b	ENSLACG00000009065	ENSLACP00000023121	ENSLACT00000025212
Zebrafish_b	ENSDARG00000027183	ENSDARP00000105264	ENSDART00000129817
Cod_b	ENSGMOG00000009639	ENSGMOP00000010300	ENSGMOT00000010582
Medaka_b	ENSORLG00000010224	ENSORLP00000012822	ENSORLT00000012823
Stickleback_b	ENSGACG00000011613	ENSGACP00000015355	ENSGACT00000015385
Platyfish_b	ENSXMAG00000007783	ENSXMAP00000007822	ENSXMAT00000007830
Spotted gar_b	ENSLOCG00000010385	ENSLOCP00000012733	ENSLOCT00000012757
Shark_a	NW_006890189.1
Shark_b	NW_006890091.1
Lamprey	ENSPMAG00000008401	ENSPMAP00000009251	ENSPMAT00000009291
Lancelet	XM_002601731.1
Sponge	Y18901.1
Choanoflagellate	XM_001744871.1
Anemone	XM_001623904.1

• Genes were collected for NAMPT in metazoan from different databases. Elephant shark, lancelet, sponge, anemone and choanoflagellate were retrieved from the NCBI database. Other speices were retrieved from the NCBI database. a, NMPTA; b, NAMPTB.

In the lamprey genome, we detected two NAMPT fragments [ENSEMBL: ENSPMAG00000003842 and ENSPMAG00000008401], although the former was too short to be informative. We also detected two NAMPT fragments [ENSTNIG00000006019 and ENSTNIG00000018493] in the tetraodon genome, and the former only encoded 221 amino acids. Additional genomic structure and phylogenetic analysis showed that ENSTNIG00000006019 was a partial retrocopy of ENSTNIG00000018493. Accordingly, these two gene fragments, ENSPMAG00000003842 and ENSTNIG00000006019, were excluded from further analysis.

Phylogenetic analysis and shared synteny

Multiple protein sequence alignments were generated using clustalw with the default parameters (http://align.genome.jp). The bases with ambiguity were manually inspected and removed to optimize the alignment. We used prottest, version 3.0 [51] to obtain a substitution model that best fit the data (LG+I+G). The ML tree was obtained with phyml, version 3.0 [52] and the bootstrap values were calculated from 100 repetitions. Moreover, additional NJ and ME trees were also constructed with mega, version 5.0 [53] under the Poisson model and bootstrap values were calculated from 2000 repetitions. To confirm the existence of NAMPTB to establish syntenic relationships between genomes, we collected the genes flanking NAMPTB at their genomic loci in a selection of representative vertebrate species from ENSEMBL and the Synteny Database (http://syntenydb.uoregon.edu/synteny_db/). Additionally, genomic organization (exon/intron structures) of each gene was manually checked by cDNA-to-genome comparison.

Model testing of selective pressures

The nonsynonymous to synonymous substitution rate ratio (ω = d_N/d_S) provided a method to detect selection pressure at the protein level, with ω < 1, = 1 and > 1, indicating purifying selection, neutral evolution and positive selection, respectively [54]. To avoid synonymous substitution saturation, we did not use the duplicate-specific gene phylogeny but, instead, a species phylogeny based on published studies [26, 4]. To explore the vertebrate adaptation to different oxygen levels during the water-to-land transition, we divided the NAMPT family into three parts (Fig. 5): NAMPTA in water-breathing vertebrates (NAMPTA-fish); NAMPTA in air-breathing tetrapods (NAMPTA-tetrapod); and NAMPTB in water-breathing vertebrates (NAMPTB-fish). First, we employed the site models to tested on individual clades to identify sites under positive selection after the initial divergence event. These models assumed variable selective pressure among sites but fixed selective pressure among branches in the phylogeny. We used two pairs of models, that compared M1a (Nearly Neutral) with M2a (Positive Selection) and M7 (beta) with M8 (beta &ω) to test for positive selection. The M0 (one ratio) to M3 (discrete) comparison aimed to test ω among sites. Additionally, we compared the null hypothesis M8a (NSsites = 8, fix_omega = 1, omega = 1) with M8 to test whether the site class > 1 was statistically different from neutrality.

In addition, we tested whether NAMPTA-fish, NAMPTA-tetrapod and NAMPTB-fish had experienced significant differences in selective constraints after gene duplication. Selection analysis of the NAMPT family was performed using the CmC (model = 3; NSsites = 2) [55, 56]. This model allowed variation in d_N/d_S among sites, with a portion of sites evolving under divergent selection between clades of a gene family. First, the entire NAMPTA clade (CmC A) or the entire NAMPTB clade (CmC A) was set as the foreground branches (Fig. 5). The model M2a_rel (NSsites = 22) [57] was used as the null model to test for the presence of divergently selected sites in the lineages between the NAMPTA and NAMPTB clades. Additionally, we used another developed ‘multi-clade’ approach [58], which separated the NAMPTA-tetrapod branches from the NAMPTA-fish and NAMPTB-fish branches (Fig. 5). The two tree partitions model was used as the null model. We employed nested LRTs on the species phylogeny, and the null distribution for this LRT should follow a chi-squared distribution with one degree of freedom.

It was unlikely that positive selection affect all sites over a long period of time. It might occur in particular stages of evolution or in particular branches. We thus used the branch-site model A (model = 2; NSsites = 2) [59] to identify the potential for positive selection along the lineage of interest that affected only a few sites. This mode allowed variation in d_N/d_S both among lineages and among sites. The null model was model A but with ω₂ = 1 fixed (fix_omega = 1 and omega = 1). All branches (46 branches in Fig. 5) after duplication were separately tested to understand which sites showed signatures of positive selection. The Bayes empirical Bayes method for calculating the posterior probability of positively selected sites was also used for the results of branch-site model A. These positively selected sites were then mapped onto the tertiary structure of human NAMPT (Protein Data Bank code: PDB 2e5b) [35] using pymol (http://www.pymol.org/).

Functional divergence analysis

Gene duplications provide raw materials for functional innovations, and changes in amino acid sites may result in altered functional constraints on subsequent evolution. To analyze the relationship between NAMPT protein family evolution and functional divergence, two types of functional divergence (Type-I and Type-II) were tested with diverge, version 3.0 [60, 61]. Type-I functional divergence occurred shortly after gene duplication, and it resulted in site-specific changes in evolutionary rates between paralogous clusters [61]. These Type-I sites were well conserved in one duplicate cluster but were highly variable in the other duplicate cluster. However Type-II functional divergence occurred in late phases after gene duplication, and it also resulted in changes to paralogous group-specific amino acid properties at individual sites [62]. These Type-II sites were highly conserved in both paralogous clusters but varied between the two clusters. The coefficients of Type-I and Type-II functional divergence (θ_I and θ_II) were also calculated. Amino acid sites likely to have undergone Type-I or -II divergence were detected by a site-specific posteriori analysis. For Type-I functional divergence, a LRT test was used to assess changes in site-specific evolutionary rates. The cut-off value for the posterior probability was defined after consecutive removal of the highest scoring sites from the alignment until the θ_I values became not significant (P > 0.05) [63].For Type-II, a Z-score test was used to identify residues with a radical shift of amino acid physiochemical properties [60]. In the present study, we considered the sites with a cut-off > 8.0 as potential sites crucial for Type-II functional divergence.

NAD⁺/NADH level determination

Human embryonic kidney (HEK) 293T cells were obtained from Wuhan Xiao (Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, Hubei, China) and cultured in DMEM (HyClone, Logan, UT, USA) supplemented with 10% FBS (HyClone) at 37 °C in an environment containing 5% CO₂. The entire coding region of zebrafish NAMPTA and NAMPTB were obtained by RT-PCR with the primers listed in (Table 7). The PCR products were digested with BamHI/HindIII or EcoRI/SalI, then cloned into the corresponding site of pCMV-Tag 2C (Agilent Technologies Inc., Santa Clara, CA, USA). The constructs were verified by sequencing and transfected into HEK 293T cells using VigoFect (Vigorous, Beijing, China). The plasmid pCMV-Tag 2C without the gene for NAMPT was used as a control group and each sample was repeated three times.

Table 7. Primers used in the present study

Vector	Name	Sequence	Enzyme
pCMV-Tag 2C	nampta_F	ATCGGGATCCCCATGGAGAAACACAGAGAAGCC	BamHI
	nampta_R	CGATAAGCTTGTCAGAGCAGCAGATCCTGC	HindIII
	namptb_F	ATCGGAATTCGAATGATGGCAGCTCAGGATTTC	EcoRI
	namptb_R	CGAT GTCGACGTTAGTGCACGCCATTCATTATG	SalI
pCAEGFP-N1 (WT)	nampta_F	ATCGAAGCTTGGGAAGATGGAGAAACACAGAGA	HindIII
	nampta_R	CGATGGATCCCGGACCGTCTTGTCGTATTTGACTTTT	BamHI
	namptb_F	ATCGAAGCTTTCGATGATGGCAGCTCAGGA	HindIII
	namptb_R	CGATGGATCCCGAACTGCTTTGATACGGATTGG	BamHI
pCAEGFP-N1 (MT)	nampta_F	ATCGAAGCTTGGcAAcATGGAcAtACAgAGAGA	HindIII
	namptb_F	ATCGAAGCTTTCGATGATGcCAGgTCAcGtTTTgA	HindIII
Psp64poly(A)	nampta_F	ATCGAAGCTTATGGAaAAgCAtAGgGAgGCC	HindIII
	nampta_R	CGATGGATCCTCAGAGCAGCAGATCCTGCA	BamHI
	namptb_F	ATCGAAGCTTATGATGGCtGCaCAaGAcTTtAAT	HindIII
	namptb_R	CGATGGATCCTTAGTGCACGCCATTCATTATG	BamHI

• Restriction enzyme cutting sites are underlined. Mismatched nucleotides are indicated by lowercase letters.

The incubated cells were collected after 24 h under the standard conditions described above. The cell samples were washed with cold PBS and lysed with NAD⁺/NADH lysis buffer followed by centrifugation at 12 000 g. for 5 min. The supernatants were collected for the NAD⁺/NADH assay and the residual solution was collected for western blot analysis. NAD⁺/NADH levels of control and experimental groups were determined with a Fluorimetric NAD⁺/NADH Ratio Assay Kit (AAT Bioquest, Sunnyvale, CA, USA). The assays were conducted in accordance with the manufacturer's instructions and the fluorescence increase of each sample was monitored with a fluorescence plate reader at excitation/emission = 540/590 nm.

Western blot

Recombinant proteins from the transfected HEK293T cells were extracted with lysis buffer and then boiled in 5 × protein loading buffer (Beyotime, Shanghai, China). The protein extracts were loaded and fractionated by SDS/PAGE, then transferred to a poly(vinylidene difluoride) membrane (Millipore, Billerica, MA, USA). The target proteins were blocked in 5% nonfat dry milk in TBST [TBS (pH 7.4) : 7.3 g NaCl and 3.03 g Tris-Base diluted in 1,000 mL H2O, TBST: TBS supplemented with 0.1% Tween 20] at room temperature and probed with the prepared primary antibody (Monoclonal ANTI-FLAG^® M2; Sigma, St Louis, MO, USA) and secondary antibody [horseradish peroxidase-labeled goat anti-(mouse IgG) (H+L); Beyotime]. The immunoblot was visualized using the ECL plus a western blotting detection system (Amersham Biotech, Little Chalfont, UK).

Whole-mount in situ hybridization

In situ hybridization to visualize gene spatial-temporal expression was performed using digoxigenin-labeled antisense RNA, as described previously [64, 65]. The embryos were fixed at appropriate developmental stages from 12 to 72 hpf in 4% paraformaldehyde overnight and underwent dehydration through successive dilutions of methanol. Antisense probes for in situ hybridization were synthesized from the cloned 3′ UTR cDNA fragments: NAMPTA (GenBank: XM_002661340.2, nucleotides 1558–1890) and NAMPTB (GenBank: NM_212668.1, nucleotides 1587–1960).

Knockdown and rescue experiments

Antisense translation-blocking MOs targeting the start codon of the zebrafish gene for NAMPTA (nampta_MO, GGCTTCTCTGTGTTTCTCCATCTTC), gene for NAMPTB (namptb_MO, ATTGAAATCCTGAGCTGCCATCATC) and a STD morpholino (STD_MO, CCTCTTACCTCAGTTACAATTTATA) were designed by Gene Tools (Philomath, OR, USA). To test the efficiency of nampta-MO and namptb-MO, a 177-bp NAMPTA and a 330-bp NAMPTB cDNA fragment including the target region of the respective MO were subcloned into pCAEGFP-N1 (Clontech, Palo Alto, CA, USA) to construct wild-type NAMPTA and NAMPTB recombinant plasmids (nampta-GFP-WT and namptb-GFP-WT). To validate the specificity of these two morpholinos, the zebrafish GFP-tagged mutated NAMPTA and NAMPTB recombinant plasmids (nampta-GFP-MT and namptb-GFP-MT) were also restructured by inducing five mismatched nucleotides at the MO target regions (Table 7).

For the rescue experiments, full-length wild-type cDNAs of zebrafish NAMPTA were subcloned into the Psp64 poly(A) vector (Promega, Madison, WI, USA). To avoid quenching by nampta-MO, five mismatched nucleotides were also induced in the forward primer to generate NAMPTA cDNA (Table 7). 5′ capped sense NAMPTA mRNA were synthesized using the Ampticap SP6 High Yield message maker kit (Ambion, Austin, TX, USA). Different amounts of capped mRNA mixed with nampta-MO were co-injected into yolk at the one-cell stage to obtain an optimal rescue effect.

Acknowledgements

We thank Professor Wuhan Xiao (Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, Hubei, China) for supplying HEK 293T cells and Professor Z. Yin (Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, Hubei, China) for providing the experiment platform and technical assistance. We also thank LetPub (http://www.letpub.com) for linguistic assistance during the preparation of the manuscript submitted for publication. This work was supported by the Pilot projects (Grant No. XDB13020100).

Author contributions

SH and CF conceived and designed the study. CF, LG and ZZ collected the data and performed the experiments. XG provided technical support. SH and CF analyzed the data. CF wrote the manuscript. SH and XG revised the paper. All authors read and accepted the final version of the manuscript submitted for publication.