1. Developmental Biology
Download icon

Naa12 compensates for Naa10 in mice in the amino-terminal acetylation pathway

  1. Hyae Yon Kweon
  2. Mi-Ni Lee  Is a corresponding author
  3. Max Dorfel
  4. Seungwoon Seo
  5. Leah Gottlieb
  6. Thomas PaPazyan
  7. Nina McTiernan
  8. Rasmus Ree
  9. David Bolton
  10. Andrew Garcia
  11. Michael Flory
  12. Jonathan Crain
  13. Alison Sebold
  14. Scott Lyons
  15. Ahmed Ismail
  16. Elaine Marchi
  17. Seong-keun Sonn
  18. Se-Jin Jeong
  19. Sejin Jeon
  20. Shinyeong Ju
  21. Simon J Conway
  22. Taesoo Kim
  23. Hyun-Seok Kim
  24. Cheolju Lee
  25. Tae-Young Roh
  26. Thomas Arnesen
  27. Ronen Marmorstein
  28. Goo Taeg Oh  Is a corresponding author
  29. Gholson J Lyon  Is a corresponding author
  1. Department of Life Science and College of Natural Sciences, Ewha Womans University, Republic of Korea
  2. Laboratory Animal Resource Center Korea ResearchInstitute of Bioscience and Biotechnology, Republic of Korea
  3. Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, United States
  4. Department of Chemistry, University of Pennsylvania, United States
  5. Abramson Family Cancer Research Institute, Perelman School of Medicine, University of Pennsylvania, United States
  6. Department of Biomedicine, University of Bergen, Norway
  7. Department of Molecular Biology, New York State Institute for Basic Research in Developmental Disabilities, United States
  8. Department of Human Genetics, New York State Institute for Basic Research in Developmental Disabilities, United States
  9. Research Design and Analysis Service, New York State Institute for Basic Research in Developmental Disabilities, United States
  10. Center for Cardiovascular Research, Washington University School of Medicine, United States
  11. Center for Theragnosis, Korea Institute of Science and Technology, Republic of Korea
  12. Herman B. Wells Center for Pediatric Research, Indiana University School of Medicine, United States
  13. Department of Converging Science and Technology, KHU-KIST, Kyung Hee University, Republic of Korea
  14. Department of Life Sciences, Pohang University of Science and Technology, Republic of Korea
  15. Department of Biological Sciences, University of Bergen, Norway
  16. Department of Surgery, Haukeland University Hospital, Norway
  17. Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania, United States
  18. Biology PhD Program, The Graduate Center, The City University of New York, United States
  19. George A. Jervis Clinic, New York State Institute for Basic Research in Developmental Disabilities, United States
Research Article
  • Cited 0
  • Views 380
  • Annotations
Cite this article as: eLife 2021;10:e65952 doi: 10.7554/eLife.65952

Abstract

Amino-terminal acetylation is catalyzed by a set of N-terminal acetyltransferases (NATs). The NatA complex (including X-linked Naa10 and Naa15) is the major acetyltransferase, with 40–50% of all mammalian proteins being potential substrates. However, the overall role of amino-terminal acetylation on a whole-organism level is poorly understood, particularly in mammals. Male mice lacking Naa10 show no globally apparent in vivo amino-terminal acetylation impairment and do not exhibit complete embryonic lethality. Rather Naa10 nulls display increased neonatal lethality, and the majority of surviving undersized mutants exhibit a combination of hydrocephaly, cardiac defects, homeotic anterior transformation, piebaldism, and urogenital anomalies. Naa12 is a previously unannotated Naa10-like paralog with NAT activity that genetically compensates for Naa10. Mice deficient for Naa12 have no apparent phenotype, whereas mice deficient for Naa10 and Naa12 display embryonic lethality. The discovery of Naa12 adds to the currently known machinery involved in amino-terminal acetylation in mice.

Introduction

Amino-terminal acetylation is one of the most common protein modifications, occurring co- and post-translationally. Approximately 80% of cytosolic proteins are amino-terminally acetylated in humans and ~50% in yeast (Arnesen et al., 2009), while amino-terminal acetylation is less common in prokaryotes and archaea (Dörfel and Lyon, 2015). Amino-terminal acetylation is catalyzed by a set of enzymes, the N-terminal acetyltransferases (NATs), which transfer an acetyl group from acetyl-coenzyme A (Ac-CoA) to the free α-amino group of a protein’s N-terminus. To date, eight distinct NATs (NatA–NatH) have been identified in eukaryotes that are classified based on different subunit compositions and substrate specificities (Polevoda et al., 2009; Aksnes et al., 2019; Starheim et al., 2012). Amino-terminal acetylation has been implicated in steering protein folding, stability or degradation, subcellular targeting, and complex formation (Ree et al., 2018; Shemorry et al., 2013; Dikiy and Eliezer, 2014; Holmes et al., 2014; Scott et al., 2011). The vital role of NATs and amino-terminal acetylation in development has also emerged (Lee et al., 2018).

NatA, the major NAT complex, targets ~40% of the human proteome, acetylating Ser-, Ala-, Gly-, Thr-, Val-, and Cys N-termini after removal of the initiator methionine (Arnesen et al., 2009; Starheim et al., 2012). Human NatA consists of two main subunits, the catalytic subunit N-α-acetyltransferase 10 (NAA10) (Ard1) and the auxiliary subunit NAA15 (Nat1), and a regulatory subunit HYPK (Arnesen et al., 2005; Arnesen et al., 2010; Gottlieb and Marmorstein, 2018). NAA15 function has been linked to cell survival, tumor progression, and retinal development (Arnesen et al., 2006a; Gendron et al., 2010). In addition, Naa10-catalyzed N-terminal acetylation has been reported to be essential for development in many species (Lee et al., 2018; Wang et al., 2010; Chen et al., 2014; Linster et al., 2015; Ree et al., 2015; Feng et al., 2016; Chen et al., 2018), and although NatA is not essential in Saccharomyces cerevisiae, depletion of Naa10 or Naa15 has strong effects, including slow growth and decreased survival when exposed to various stresses (Mullen et al., 1989; Polevoda and Sherman, 2003).

NAA10 mutations were found to be associated with several human diseases characterized by severe phenotypes, including global developmental defects (Lee et al., 2018). Among these, the X-linked Ogden syndrome (OS) (Myklebust et al., 2015; Rope et al., 2011) shows the most severe pathological features such as infant lethality and has reduced NatA catalytic activity. In a S. cerevisiae model for the Naa10 Ser37Pro mutant, the mutation impairs NatA complex formation and leads to a reduction in NatA catalytic activity and functionality (Van Damme et al., 2014; Dörfel et al., 2017). Further, OS patient-derived cells have impaired amino-terminal acetylation in vivo of some NatA substrates (Myklebust et al., 2015). Over the years, many additional pathogenic NAA10 variants have been identified in NAA10 or NAA15 (Esmailpour et al., 2014; Popp et al., 2015; Casey et al., 2015; McTiernan et al., 2018; Ree et al., 2019; Støve et al., 2018; Cheng et al., 2019; Cheng et al., 2018; Johnston et al., 2019) and the collection of presenting symptoms for families with NAA10 mutations is currently referred to as Ogden syndrome or NAA10-related syndrome (Wu and Lyon, 2018).

The autosomal NAA10 homolog, NAA11 (ARD2), has been reported to be present in mice and humans, and is co-expressed with NAA10 in human cell lines (Arnesen et al., 2006b). Therefore, NAA11 could conceivably compensate when NAA10 is reduced or lacking (Lee et al., 2018). However, NAA11 was only found in testis and placenta in human and gonadal tissues in mouse, while NAA10 showed widespread expression in various tissues in embryos and adults (Pang et al., 2011). Thus, any functional redundancy or compensation might be limited to certain tissues only.

To elucidate the functional role of Naa10 during development in mice, we used two different Naa10-deficient mouse lines: one, referred to as Naa10 knockout (KO), which was previously reported specifically related to bone density in postnatal day 3 (P3) mice (Yoon et al., 2014), and another denoted as Naa10tm1a(EUCOMM)Hmgu (Naa10tm1a), generated in this study. These Naa10-deficient mice exhibit pleiotropic developmental abnormalities at a range of different ages, overlapping with some of the phenotypes seen in human disease involving NAA10 impairment. Because we did not discover major changes in the overall Nt-acetylome in Naa10 KO mice, we hypothesized that there might be a compensating gene in mice, which led us to the identification of a new paralog of Naa10, which we name Naa12. Naa12 is expressed in several organs (liver, kidney, heart, and testis) and, like Naa10, binds to Naa15 to mediate NatA activity. Furthermore, lethality was observed in Naa10 Naa12 double-KO mice, which supports the compensatory role of Naa12 in vivo. Thus, we demonstrate that Naa10 is essential for proper development and Naa12, a newly identified paralog of Naa10, can play a compensatory role in mice.

Results

Naa10 KO mice can be born, but display pleiotropic developmental defects

To explore the role of Naa10 in development, most analyses were carried out using our Naa10 KO model mice that had been generated previously (Yoon et al., 2014) using a targeting vector deleting Exon1, including the start codon, and Exon2 to Exon4 containing the GNAT domain including the acetyl-CoA binding motif, which is crucial for Naa10 function. We also generated another Naa10-deficient mouse, which we called Naa10tm1a, expressing β-galactosidase rather than the Naa10 gene (Figure 1—figure supplement 1A). Naa10 expression was deficient in Naa10tm1a mice (Figure 1—figure supplement 1B, C). Especially strong β-gal staining was observed during embryonic stages in the brain, heart, and spinal cord (Figure 1—figure supplement 1D). Male Naa10 KO (Naa10-/Y) embryos displayed mild to severe developmental defects compared to wildtype (WT) (Naa10+/Y) embryos. Some Naa10-/Y mice had lower levels of somites and developmental delay. Additionally, some Naa10-/Y embryos had a normal number of somites but were retarded in growth (Figure 1A). Some of the embryos underwent lysis or remained arrested at an earlier stage than embryonic day 10.5 (E10.5), with no turning, an abnormal trunk, and underdeveloped facial features. These phenotypes also reproduced in Naa10tm1a/Y embryos. Next, we assessed whether Naa10 is essential for viability and counted the Mendelian ratios. Both Naa10-/Y and Naa10tm1a/Y mice were under-represented after birth, while there was no significant reduction in the embryonic stage in both mouse lines (Supplementary file 1a, b). We monitored the pups daily at postnatal day 0 (P0) to postnatal day 3 (P3) and beyond, and the survival rate of Naa10-/Y mice dramatically decreased relative to either WT (Naa10+/Y and Naa10+/+) or heterozygous female (Naa10+/-) mice after the first few days of life (Figure 1B), and a few Naa10-/Y mice with postnatal lethality exhibited severe developmental defects such as craniofacial anomaly, an undeveloped lower body, whole-body edema, and ocular malformations (Figure 1C).

Figure 1 with 2 supplements see all
Deficiency of Naa10 leads to abnormal development and postnatal lethality.

(A) Naa10+/Y, Naa10-/Y, and Naa10tm1a/Y embryos at E10.5. Growth retardation (5/33, more than five somites lower or undersized compared to littermate controls), kinky trunk, and developmental arrest are shown in Naa10-/Y (4/33) and Naa10tm1a/Y (1/5). Scale bars: 500 μm. (B) The percentage lethality in newborns, comparing Naa10 wildtype (WT) (Naa10+/Y and Naa10+/+), Naa10-/+ and Naa10-/Y pups until P3, derived from matings between heterozygous females and WT males. Approximately 11.6% (10/86) of WT, 24% (13/54) of Naa10+/-, and 76.3% (29/38) of Naa10-/Y mice were found dead before P3. (C) Representative images of Naa10-/Y pups during early postnatal days compared with Naa10+/Y. Severe developmental defects such as malformations of head and lower body (one leg; black arrowheads), whole-body edema, and anophthalmia (black arrows) are shown (N = 1 each). (D) Hematoxylin and eosin (H&E)-stained heart transverse section at E14.5 and vertical section at E18.5, comparing Naa10+/Y and Naa10-/Y embryos. Naa10-/Y embryo shows a ventricular septal defect (VSD) at E14.5 and E18.5. Also, at E18.5, Naa10-/Y embryo shows atrial septal defect (ASD). Arrow indicates VSD, ASD, and double outlet right ventricle (DORV). Scale bars: 20 μm.

Congenital heart defects are one of the main causes of infant lethality, and cardiac diseases are a common developmental anomaly in OS patients (Casey et al., 2015), with some OS males dying in infancy with cardiac arrhythmias (Rope et al., 2011). Therefore, we investigated whether Naa10 KO affects cardiac development. Development of a four-chambered septated heart is normally complete at E14.5; therefore, we examined the cardiovascular system at E14.5. We identified ventricular septal defects (VSDs) in several Naa10-/Y embryos, as well as concomitant double outlet right ventricle (DORV) at E14.5 (Figure 1D, upper). VSDs and atrial septal defects (ASDs) were also observed at E18.5 (Figure 1d, bottom), and persistent truncus arteriosus (PTA) or DORV, along with concomitant membranous and muscular VSDs, were found in several of the mice that died in the first day of life (n = 6/28 examined). Given the presence of outflow tract defects and VSDs, we examined whether the ductus arteriosus had closed appropriately or not at birth. Significantly, both Naa10-/Y and Naa10-/- females (n = 3/28 examined) exhibited a patent ductus arteriosus, meaning that there is a failure of the mutant in utero cardiovascular system to adapt to adult life (birth) and close the interatrial and aorta-pulmonary trunk shunts that are required for normal fetal life (Conway et al., 2003). As murine outflow tract and VSDs are not compatible with postnatal survival (Conway et al., 2003), these data suggest that congenital heart defects in Naa10-/Y mice may explain some of their neonatal lethality (Figure 1—figure supplement 2). We also examined surviving adult mice for any possible situs inversus, but we did not observe this in any adult (>4 weeks) Naa10-/Y mice examined (n = 19). Combined, these data suggest that Naa10 mutant CHDs are mainly confined to aberrant remodeling of the great vessels of the heart, leading to pulmonary overload at birth resulting in lethality.

Some of the surviving homozygous mice (Naa10-/Y and Naa10-/-) had reduced body weight (Figure 2A). This reduced body weight continued through weaning, and some mice lost more weight as they developed progressive hydrocephaly. We observed that the smallest weight animal between the Naa10+/Y and one Naa10-/Y genotypes was almost always the Naa10-/Y genotype when the analysis was restricted to only include litters in which there was at least one of each of those genotypes living beyond 4 days of life. For example, 13 litters met this criteria from the mating (Naa10+/- × Naa10+/Y), and 12/13 of the litters had the Naa10-/Y as the lower weight (Fisher's exact test, two-tailed, p-value<0.0001). Five litters met this criteria from the mating (Naa10+/- × Naa10-/Y), and of these, all of them had the Naa10-/Y as the lower weight (Fisher's exact test, two-tailed, p-value=0.0079). Therefore, despite the known variability in weight data as a function of genetic background, environment, and stochastic variation (Pun et al., 2013), it does appear at least for ‘within-litter’ analysis that Naa10-/Y males are born at a smaller weight than Naa10+/Y males and on average remained the smallest male in the litter throughout their life.

Figure 2 with 2 supplements see all
Pleiotropic phenotypes of Naa10 knockout (KO) mice.

(A–C) Representative images of abnormalities in Naa10-/Y compared with Naa10+/Y. (A) Body weight of male (left) and female (right) versus ages was monitored from 2 weeks. The weight of Naa10-/Y and Naa10-/- mice is markedly reduced compared with that of the wildtype (WT) mice. Asterisks indicate a statistical difference calculated by Student’s t-test: *p<0.05. (B) Representative images of completely penetrant phenotypes. Hypopigmentation (Naa10+/Y, n = 243; Naa10-/Y, n = 121; Naa10tm1a/Y, n = 17) and supernumerary ribs (Naa10+/Y, n = 3; Naa10-/Y, n = 6; Naa10tm1a/Y, n = 2; E18.5) were found 100% in Naa10-deficient mice. (C) Naa10-/Y is smaller in size and has round-shaped head (Naa10+/Y 0/59, Naa10-/Y 7/33). Over time, hydrocephaly became apparent (N = 14/29 [~48%] for >P7 male Naa10-Y; N = 7/19 [~36%] for >P7 female Naa10-/-). Hydronephrosis (red arrow, Naa10+/Y 0/23, Naa10-/Y 14/29, Naa10+/+ 0/5, Naa10-/- 7/19) and abnormal genitalia (black arrow) of male (middle, Naa10+/Y 0/23, Naa10-/Y 16/29) and female (bottom, hydrometrocolpos, Naa10+/+ 0/5, Naa10-/- 7/19) are shown.

Although piebaldism has never been reported in humans with OS, all (100%) of the Naa10-/Y and Naa10tm1a/Y mice exhibited hypopigmentation on their belly (Figure 2B, upper), with this piebaldism quite varied in its extent but not appearing to correlate in any way with other phenotypes, such as hydrocephaly. Another phenotype with complete penetrance was bilateral supernumerary ribs (14 pairs of ribs instead of 13) in all Naa10-/Y and Naa10tm1a/Y mice (Figure 2B, middle and bottom, Table 1). This extra pair of ribs linking to the sternum transforms the T8 vertebrae into an anterior T7-like phenotype (Figure 2—figure supplement 1, Table 1).

Table 1
Skeletal analyses for ribs, sternebrae, and vertebrae.
Naa10+/Y
(n = 50)
Naa10+/+
(n = 10)
Naa10+/-
(n = 17)
Naa10-/Y
(n = 17)
Naa10-/-
(n = 1)
4 sternebrae7 (14.0%)1 (10%)3 (17.6%)9 (52.9%)1 (100%)
3 sternebrae27 (54.0%)8 (80%)11 (64.7%)5 (29.4%)0 (0%)
4 sternebrae but with 3/4 fusion16 (32%)1 (10%)3 (17.6%)3 (17.6%)0 (0%)
14 ribs total bilaterally0 (0%)0 (0%)0 (0%)17 (100%)1 (100%)
13 ribs total bilaterally50 (100%)10 (100%)17 (100%)0 (0%)0 (0%)
8 ribs attached to sternum bilaterally0 (0%)0 (0%)0 (0%)17 (100%)1 (100%)
7 ribs attached to sternum bilaterally50 (100%)10 (100%)17 (100%)0 (0%)0 (0%)
14 thoracic vertebrae0 (0%)0 (0%)0 (0%)17 (100%)1 (100%)
13 thoracic vertebrae50 (100%)10 (100%)17 (100%)0 (0%)0 (0%)
  1. Tabulation regarding the number of sternebrae found in skeletons, including ones in which there was partial fusions between the third and fourth sternebrae.

A majority of the Naa10-/Y and Naa10-/- mice also had four instead of the usual three sternebrae, which were sometimes fused (Table 1). Cervical vertebrae fusion was also demonstrated in Naa10-/Y mice, particularly involving C1 and C2, suggesting possible anteriorization of C2 into a C1-like phenotype (Figure 2—figure supplement 1E, F, Supplementary file 1c). The number of lumbar vertebrae remained the same, thus suggesting an anterior transformation of the first sacral vertebra to a lumbar-like phenotype. These combined observations suggest possible anterior transformations in the Naa10 mutant skeletal phenotype, with an anteriorization of C2, a T8 transformation to a T7-like phenotype with ribs connecting to the sternum, an extra pair of ribs on L1 likely due to an L1 transformation to a T13-like phenotype, and an anterior transformation of the first sacral vertebra to a lumbar-like phenotype with loss of fusion to the sacral wings.

Out of 32 Naa10-/Y that survived past the third day of life and which were then examined longitudinally, about 60% survived past 200 days of life (~7 months) (Figure 2—figure supplement 2), with some of these then developing hydronephrosis (Figure 2C, middle). They had some hollowed space in the kidney, which had been filled with fluid and their ureter was thickened already at P3 stage of prenatal development in some Naa10-/Y mice (Figure 2—figure supplement 2A). Commonly, hydronephrosis is caused by a blockage or obstruction in the urinary tract. We speculate that this swelling in Naa10 KO (Naa10-/Y and Naa10-/-) mice is likely caused by ureteral defects rather than the kidney itself. Moreover, some Naa10 KO mice displayed genital defects, such as seminal vesicle malformation and hydrometrocolpos, respectively (Figure 2C, bottom). Many Naa10-/- female mice appeared to have decreased fecundity, although they were fertile upon the first mating, and this decrease in fecundity is possibly due to the development of hydrometrocolpos (Figure 2C, bottom), which might result from structural issues, like vaginal atresia or a retained vaginal septum, although this requires further investigation. Additionally, hydrocephaly became clinically apparent with a round-shaped head (Figure 2C, upper) in ~40% of the Naa10-/Y mice that had survived past 3 days of life (Figure 2—figure supplement 2C). CT scanning of some of these mice confirmed hydrocephaly as the primary cause of their rapid deteriorating condition, usually within the first three months of life (Figure 2—figure supplement 2B, C). CT scanning did not reveal any obstructive lesions (such as a tumor) in any of the ventricles that could account for the hydrocephaly. Taken together, these results indicate that Naa10 contributes to overall development and is particularly important for viability.

Litter sizes and offspring from other matings were also investigated, as shown in Supplementary file 1d. Matings were setup between Naa10-/- females and C57bl6J WT (Naa10+/Y) males, involving 11 mating pairs with seven unique females and seven unique males. Of a total of 127 pups that were born, 37 died in the first day of life and were degraded and/or cannibalized prior to any tail sample being retrieved, thus not being genotyped. This was a relatively high death rate in the first 24 hr of life (29%), more so than with the other matings, except for the one between Naa10-/- females and Naa10-/Y males (Supplementary file 1d). However, this is substantially less than the death rate of 90% (46/51) reported for the same mating in the Lee et al., 2017 paper, and we currently do not have an explanation for this discrepancy. Of the remaining 90 pups that could be genotyped, 59 of these were Naa10+/- females and 31 were Naa10-/Y males. 7 of the 59 Naa10+/- females and 2 of the 31 Naa10-/Y males died in the first three days of life (for a total death rate in the first three days for all born pups of 46/127, or 36%), and after this time, none of the remaining Naa10+/- females died in the first 10 weeks of life (52/59, or 88% overall survival), whereas 10 of the remaining 29 Naa10-/Y males developed hydrocephaly and died in the first 10 weeks of life, for an overall survival of (19/31, or 61%). The death rate for all pups of 36% in the first three days of life is similar to the rate of 42.4% seen with the mating of Naa10-/- females with Naa10-/Y males (Supplementary file 1d), whereas this rate is higher than that seen for Naa10+/- females mated with Naa10+/Y males (15.8%) or with Naa10-/Y males (13.6%).

Naa10-deficient mice have a functionally active NatA complex

Prior experiments showed reduced in vivo protein amino-terminal acetylation of a few putative targets in patient cells (Myklebust et al., 2015). Reduced Nt-acetylomes were also observed in the Naa10 mutant yeast models (Van Damme et al., 2014). Given these prior reports, we hypothesized that pleiotropic phenotypes in Naa10-deficient mice are due to a decrease in global N-terminal acetylation. To test our hypothesis, integrated N-terminal peptide enrichment method (iNrich) (Ju et al., 2020) was used to analyze the level of protein amino-terminal acetylation in mouse embryonic fibroblast (MEF) lysates of Naa10+/Y and Naa10-/Y. Since the samples are treated with deuterated acetic anhydride prior to MS, unacetylated N-terminal site appears with +3 Da mass shift in the MS spectrum of the corresponding acetylated N-terminal site (Van Damme et al., 2011a). The peak intensity ratios of acetyl/heavy acetyl pairs represent the degree of acetylation of the N-terminal site. We found 765 acetyl/heavy acetyl pairs of N-termini throughout five replicates of Naa10+/Y and five replicates of Naa10-/Y MEFs. Except for the sites detected only in either WT or mutant, 533 N-terminal sites could be compared (see tabs called ‘N-term’ and ‘Header Key’ in Supplementary file 2a, c). Approximately 98% (n = 522) of N-termini sites showed less than 10% variation in the degree of terminal acetylation, indicating that there is no major difference in amino-terminal acetylation between Naa10-/Y and Naa10+/Y MEFs (Figure 3A). A more stringent analysis was also conducted in which peptides had to be detected in all 10 samples (i.e., tabs marked ‘N-term detected in all samples’ and ‘Header Key’ in Supplementary file 2b, c), and this resulted in 152 N-termini sites, of which only 3 (Rpl27, PPia, and Histone H1.0) had a slightly greater than 10% difference in the degree of acetylation between Naa10+/Y and Naa10-/Y. Although this was not a significant result statistically (p=0.09), it is worth noting that peptidyl-prolyl cis–trans isomerase A (PPIA), having a 10.3% decrease in amino-terminal acetylation, was previously identified with decreased amino-terminal acetylation in patient-derived B cells and fibroblasts in boys with the S37P mutation in NAA10 (Myklebust et al., 2015), along with being decreased in siNatA knockdown HeLa cells (Arnesen et al., 2009). PPIA also had decreased amino-terminal acetylation in one sample from homozygous null NAA15L314*/L314*-induced pluripotent stem cells (Ward et al., 2021).

Figure 3 with 1 supplement see all
Activity measurement of NatA from wildtype (WT) and Naa10 knockout (KO) mice.

(A) Correlation of Naa10 alteration state on amino-terminal acetylation in mouse embryonic fibroblasts (MEFs). Each dot (n = 533) represents the average amino-terminal acetylation percentage of five replicates of Naa10+/Y and Naa10-/Y, respectively. Dashed lines are the borders of ±10% difference. Except for the 10 dots, 522 of the 533 dots are within the borders. The marginal histograms show the distribution of amino-terminal acetylation data points. (B) Immunoprecipitation of Naa15. Liver tissue from WT and Naa10 KO mouse was lysed and incubated with anti-Naa15 antibody to retrieve NatA complexes. Proteins were separated by SDS-PAGE and immunoblots probed with anti-Naa15 antibody and anti-NAA10 antibody. (C) Catalytic activity of immunoprecipitated NatA. The catalytic activity of NatA precipitated from WT and Naa10 KO mouse liver tissue by anti-Naa15 was measured towards the NatA substrate peptide SESS24 in an in vitro [14C]-Ac-CoA–based acetylation assay. Control reactions were performed with no enzyme or no peptide to account for background signal. The immunoprecipitation (IP) and activity measurements were performed in three independent setups, each with three technical replicates per assay. One representative setup is shown.

Overall, given the very minor differences with amino-terminal acetylation, we measured the in vitro amino-terminal acetylation activity of NatA via immunoprecipitation of the large auxiliary subunit Naa15 from mouse tissues. This analysis showed normal expression of Naa15 in Naa10 KO liver tissue as in WT tissues (Figure 3B), and we isolated a physical complex composed of Naa15 and undefined partners that retains NatA activity from Naa10 KO tissues (Figure 3C). These data suggest that despite the loss of Naa10 in mice the NatA complex remains active, thus explaining the lack of major differences with amino-terminal acetylation.

A Naa10 paralog exists in mice

Naa10 disruption is lethal in a variety of organisms, including Drosophila melanogaster (Wang et al., 2010), C. elegans (Chen et al., 2014), and Trypanosoma brucei (Ingram et al., 2000). Given the relatively mild phenotype and no reduction of the Nt-acetylome in Naa10 KO mice, we hypothesized that there might be a yet unidentified paralog of Naa10, which can compensate for loss of function in mice. A Blast search for genomic sequences with homology to Naa10 exposed several Naa10 pseudogenes on chromosomes 2, 3, 7, 12, 15, and 18. Additionally, Southern blot analysis from C57BL/6J DNA with Naa10 cDNA probe detected bands of the expected sizes on the X chromosome (Figure 3—figure supplement 1A, B), while other bands of unexpected sizes appeared on other chromosomes 2, 5, 15, and 18. The previously identified Naa10 paralog Naa11 is located on chromosome 5; however, this paralog is only expressed in testes (Pang et al., 2011). We found a predicted gene (Gm16286, UniProt: Q9CQX6) on chromosome 18, with high similarity to Naa10, which we name Naa12, and RiboSeq and mRNA traces of this region suggest possible transcription and translation of this gene (Figure 3—figure supplement 1C). The protein sequence of Naa12 is >80% identical to Naa10 and almost 90% identical with Naa11 (Figure 4—figure supplement 1C).

Quantitative PCR (q-PCR) analysis also confirmed the expression of this transcript in all tested tissues (Figure 4—figure supplement 1A), with the expression of Naa12 unchanged in the corresponding Naa10 KO tissues. We attempted to test for Naa12 expression in mouse tissues by developing an antibody specific for Naa12 by performing a sequence alignment of the two known mNaa10 isoforms, mNaa11 and mNaa12, and selecting a unique Naa12 peptide for immunization and antibody generation (Figure 4—figure supplement 1B). After generation and affinity purification, we validated the specificity and sensitivity of this Naa12 antibody with recombinant proteins purified from bacterial hosts (Figure 4—figure supplement 1C, D). However, multiple attempts to use this antibody to detect Naa12 in mouse tissues met with conflicting results, so that we were unable to consistently detect Naa12 even in WT liver, kidney, or brain tissue lysates, which could be due to a poor antibody and/or very low expression or post-translational modification of Naa12 in these tissues, thus making it difficult to detect. Furthermore, given that this antibody was raised against a peptide at the C-terminus of Naa12, such data could not be used anyway to completely exclude the possibility of truncated non-functional mini-protein expression, although the lack of any signal with RT-PCR (Figure 4—figure supplement 2) likely means that nonsense-mediated decay occurred. Presently, the rabbit polyclonal antibody is no longer recognizing any consistent protein bands in western blotting, so we have abandoned any further attempts to use this antibody.

To test whether Naa12 has a similar enzymatic activity as Naa10, we performed a radioactive-based acetyltransferase assay using synthetic peptides (Figure 4A). Since monomeric Naa10 preferentially acetylates N-termini with acidic side chains (Foyn et al., 2013; Van Damme et al., 2011b; Liszczak et al., 2013), we used peptides representing the N-termini of γ-actin (starting DDDIA-) and γ-actin (starting EEEIA-), which are two known Naa10 in vitro substrates. Additionally, we used a peptide starting with SESSSKS-, representing an in vitro NatA complex substrate high-mobility group protein A1. As expected for the monomeric proteins, we could not detect any activity towards the SESSSKS substrate. Importantly, both Naa10 and Naa12 significantly Nt-acetylated the acidic N-terminal peptides, demonstrating the intrinsic capacity of Naa12 to catalyze amino-terminal acetylation (Figure 4A).

Figure 4 with 2 supplements see all
Characterization of Naa12.

(A) In vitro N-terminal acetyltransferase radioactive-based assay. Comparison of mouse Naa10 and Naa12 towards Naa10 peptide substrates, beta-actin (DDDIA-) and gamma-actin (EEEIA-), and the optimal NatA complex peptide substrate, SESSS-. Background control reactions were performed in the absence of either peptide or enzyme. Assays were performed in triplicate; error bars represent SEM. (B) Co-immunoprecipitation assay. HEK293 cells were transfected as indicated and lysed after 48 hr. Cell lysates were incubated with 1 μg anti-V5 antibody to precipitate V5-tagged Naa15. The isolated complexes were separated on SDS-PAGE and probed with the indicated antibodies. (C) Recombinant mouse Naa12/human Naa15 chimera complex activity. Radioactive acetyltransferase activity assay evaluating the activity of mNaa12-hNaa15 towards peptide (closed circles, ‘mNaa12-hNaa15’) and peptide chemical acetylation in the absence of enzyme (closed circles, ‘Buffer’) as well as chemical acetylation of the enzyme in the absence of peptide (open circles) assay and background (open circles). Error bars represent SD of two technical replicates. These are the same results from fraction #14 (both SESSS- and No Peptide) and both Buffer and Background used to illustrate the size-exclusion-purified mNaa12-hNaa15 complex activity in Figure 4—figure supplement 1F.

Figure 4—source data 1

Characterization of a potential Naa10 homolog.

https://cdn.elifesciences.org/articles/65952/elife-65952-fig4-data1-v3.pdf

Across species, Naa10 is bound to its auxiliary subunit, Naa15, which links the catalytic subunit to the ribosome to facilitate co-translational amino-terminal acetylation of proteins as they emerge from the exit tunnel (Mullen et al., 1989; Sugiura et al., 2003; Park and Szostak, 1992; Gautschi et al., 2003; Magin et al., 2017; Varland and Arnesen, 2018). Due to its high sequence similarity (Figure 4—figure supplement 1B), we suspected that Naa12 may also interact with Naa15. To test this hypothesis, we performed co-immunoprecipitation assays in HEK 293 cells. Apart from Naa10 (isoform 1, Naa10235) and Naa12, we also included the second isoform of mNaa10, mNaa10225 that has been described earlier (Arnesen et al., 2005; Park and Szostak, 1992; Kim et al., 2006) as well as Naa11. Both Naa10 isoforms as well as Naa11 and Naa12 co-precipitated with V5-Naa15 but not V5 alone, suggesting that all tested proteins could form a stable complex with Naa15 in mouse (Figure 4B). As we have previously purified the human NatA complex composed of truncated human Naa10 (residues 1–160) and full-length human Naa15 complexes that had been expressed in insect cells (Gottlieb and Marmorstein, 2018), we attempted to co-express a chimeric truncated mouse Naa12 (residues 1–160) with full-length human Naa15 complex in insect cells (human and mouse Naa15 are highly conserved with a sequence conservation of 98.2%). The complex was purified by a combination of affinity, ion exchange, and size-exclusion chromatography, and size-exclusion fractions harboring a clearly detectable band of Naa15 and a lighter band for Naa12, as determined by silver staining, were analyzed for activity towards a SESSSKS- peptide (Figure 4C, Figure 4—figure supplement 1E, F). This analysis revealed that peak fractions containing the Naa12-Naa15 complex harbored detectable amino-terminal acetylation activity towards the SESSSKS- peptide (Figure 4C, Figure 4—figure supplement 1F), thus demonstrating catalytic activity of a NatA complex with mouse Naa12.

In a mass spectrometry analysis of a similar setup to that shown in Figure 3B, NAA15 immunoprecipitates from WT or Naa10-KO mouse livers were analyzed by mass spectrometry. We found five distinct peptides derived from Naa12 (Table 2 and Supplementary file 2d). Three of these derive from the same part of the peptide sequence, RDLSQMADELRR, and all of these three peptides had one or two missed trypsin cleavages (DLSQMADELRR, RDLSQMADELR, and RDLSQMADELRR). The other two peptides, AMIENFSAK and ENQGSTLPGSEEASQQENLAGGDSGSDGK, are not the results of missed cleavages. None of these peptides are found in other sequences in the mouse genome and thus unambiguously identify Naa12 in our experiments. They have higher intensities in Naa15 IPs compared to Ctrl IPs, indicating that Naa12 is selectively enriched by Naa15 IP. Some peptides are additionally assigned to the Naa10/Naa11/Naa12 protein group as a large part of their sequences are identical. As expected, no unique Naa10 peptides are identified in the IPs from Naa10-KO mice. 12 peptides were ambiguously assigned to Naa12 or to major urinary proteins (Mup9, Mup8, Mup1, Mup17, Mup5, or Mup2), but these are as likely to be derived from Mups as from Naa12, as they have comparable intensities between Ctrl and Naa15 IPs.

Table 2
Naa10, Naa11, and Naa12 peptides identified by LC-MS/MS analysis in Naa15 IP samples from WT and Naa10-KO mouse.
Gene namePeptide sequenceLog2 LFQ intensity Naa15-IP
WT mouseNaa10-KO mouse
Naa12AMIENFSAK23.814427.5563
Naa12DLSQMADELRR25.263728.38
Naa12ENQGSTLPGSEEASQQENLAGGDSGSDGK21.29922.09
Naa12RDLSQMADELR-22.20
Naa12RDLSQMADELRR-27.77
Naa10AALHLYSNTLNFQISEVEPK26.7672-
Naa10AMIENFNAK27.3981-
Naa10DLTQMADELRR25.5107-
Naa10GNVLLSSGEACREEK25.0717-
Naa10HMVLAALENK25.5293-
Naa10NARPEDLMNMQHCNLLCLPENYQMK25.8928-
Naa10YYFYHGLSWPQLSYIAEDENGK26.5915-
Naa12;Naa11AALHLYSNTLNFQVSEVEPK-27.3833
Naa12;Naa11YYFYHGLSWPQLSYIAEDEDGKIVGYVLAK-25.2517
Naa12;Naa11;Naa10IVGYVLAK28.087325.7753
Naa12;Naa11;Naa10MEEDPDDVPHGHITSLAVK29.106929.265
Naa12;Naa11;Naa10MEEDPDDVPHGHITSLAVKR24.760521.7784
Naa12;Naa11;Naa10YVSLHVR22.861123.7383
Naa12;Naa11;Naa10YYADGEDAYAMK-27.2083
Naa12;Naa11;Naa10YYADGEDAYAMKR27.231927.1689
  1. Samples were run in technical duplicates and the average log2 LFQ intensity of the peptides is presented.

    IP: immunoprecipitation; WT: wildtype; KO: knockout; LFQ: label-free quantification.

Naa12 rescues loss of Naa10 in mice

To investigate whether Naa12 can rescue the loss of the function of Naa10 in vivo, Naa12 KO mice were generated using CRISPR technology (Singh et al., 2015). One 95-base pair deletion Δ131–225 in Naa12 was characterized in depth (Figure 5A). This mutation introduces a frameshift, leading to a termination codon at amino acid 67, which should either result in complete KO of the protein or, at best, the expression of a truncated mini-protein that would be far shorter than the usual 220 amino acid Naa12. We confirmed the deletion by PCR with genomic DNA (Figure 5B). QPCR further showed deletion of Naa12 in the tested tissues of Naa12 KO mice (Figure 5C); however, it seemed that Naa12 might be slightly expressed in testis. Due to the high similarity between Naa11 and Naa12, the expression shown in Naa12 KO testis could actually be Naa11 rather than Naa12, and this was confirmed by RT-PCR showing definite deletion (Figure 4—figure supplement 2A).

Generation of Naa12 knockout (KO) mice.

(A) Scheme of Naa12 (Gm16286, UniProt: Q9CQX6) deletion used to generate Naa12 KO mouse. 95 base pairs (131–225) were deleted. F′: genomic DNA forward primer; F′′: cDNA forward primer; R: reverse primer. (B) Genotyping of Naa12 KO mice by PCR. Wildtype (WT) allele size was 381 bp and targeted allele size was 287 bp. (C) mRNA level of Naa12 was analyzed in selected tissues by qPCR. Relative expression level of WT (white bars) and Naa12 KO (black bars) after normalizing to that of GAPDH.

Paralogs are homologous genes that originate from the intragenomic duplication of an ancestral gene. Homologs that play a compensatory role can sometimes show similar phenotypes to each other when one of them is deficient (Peng, 2019), whereas other homologs might only offer partial compensation when the primary gene is more widely expressed or has higher activity levels. We analyzed Naa12 KO mice to see if they produced similar developmental defects to those in Naa10 KO mice. KO mice for this gene were viable (Supplementary file 1e). Although there was initially a question of decreased fertility for the male mice, larger numbers of matings and litters did not bear this out (Supplementary file 1d), and necropsy and inspection of testes and seminal vesicles under a stereomicroscope did not reveal any macroscopic differences. Furthermore, the phenotypes (piebaldism and bilateral supernumerary ribs, Figure 2B) observed in Naa10 KO mice with complete penetrance were not present in Naa12 KO mice (Figure 4—figure supplement 2B). Overall, there were not any obvious phenotypes in these mice.

Matings between Naa10+/- Naa12+/+ female mice and either Naa10+/y Naa12+/- or Naa10+/y Naa12-/- males produced zero male Naa10-/y Naa12+/- progeny, while also suggesting that compound heterozygous (Naa10+/- Naa12) female mice are produced at a rate much less than predicted by Mendelian ratios (Supplementary file 1f, g). Matings between surviving compound heterozygous (Naa10+/- Naa12+/-) females and Naa10+/Y Naa12+/- males demonstrate that no live births occurred for Naa10 Naa12 double-knockout (DKO) males (Naa10-/Y Naa12-/-) (Figure 6). In addition, the average litter size was small when compared to the control (WT × WT) matings, suggesting embryonic lethality (Table 3). In order to determine whether lethality occurs during the embryonic stage, we genotyped E18.5 litters – just before birth. Consistent with our previous observations, we could not obtain any Naa10-/Y Naa12-/- embryos, and many embryos could not be genotyped because they were already in the midst of resorption (n = 23) (Figure 6). We checked an even earlier stage at E10.5 and also found zero Naa10-/Y Naa12-/- embryos, and also with far fewer resorptions at this stage (N = 3). Interestingly, we did observe Naa10-/Y Naa12+/- embryos where two of them displayed delayed developmental stage (appearing younger than E10.5) and another two embryos were lysed and had already begun degenerating (but despite this, we could at least genotype these embryos). This helps explain why only one Naa10-/Y Naa12+/- embryo was observed at E18.5. Furthermore, Naa10+/- Naa12-/- female embryos were also lysed/degenerating at E10.5 and were not observed from that day onward. Matings between compound heterozygous females and Naa10+/Y Naa12-/- males also did not yield any Naa10-/Y Naa12-/- male mice at any embryonic stage examined, and only a couple of Naa10+/- Naa12-/- female mice at early stages of development (Figure 6—figure supplement 1), and the litter sizes were even smaller, suggesting increased embryonic lethality (Table 3). Consistent with this, we noted many resorptions at E12.5 and E18.5 that could not be genotyped. The number of living postnatal compound heterozygous female mice was also considerably lower than the predicted Mendelian ratios (Figure 6, Figure 6—figure supplement 1) and the surviving Naa10+/- Naa12+/- females were smaller in size than littermate controls (Figure 7D).

Figure 6 with 5 supplements see all
Lethality in Naa10 Naa12 double-knockout (DKO) mice.

Naa10 Naa12 DKO exhibit embryonic lethality. Pedigree and genotypes of pups and embryos at E10.5 and E18.5 from Naa10+/- Naa12+/- female mice crossed to the Naa10+/Y Naa12+/- male mice.

Decreased body weight in compound heterozygous females.

(A) Male body weight for the Naa10 mice on inbred genetic background (eight backcrosses to C57bl6/J). (B) Female body weight for the Naa10 mice on inbred genetic background (eight backcrosses to C57bl6/J). (C) Male body weight for the Naa10 and Naa12 mice on mixed genetic background. (D) Female body weight for the Naa10 and Naa12 mice on mixed genetic background.

Table 3
Litter size of Naa10 × Naa12 matings.
Genotypes of Naa10; Naa12 breeders (♀ x ♂)Total number of pupsTotal number of littersAverage litter size
(pups/litters)
SD of litter size
Naa10+/+ Naa12+/+ × Naa10+/Y Naa12+/+206248.61.6
Naa10+/- Naa12+/- × Naa10+/Y Naa12+/-157324.91.5
Naa10+/- Naa12+/- × Naa10+/Y Naa12-/- *225633.61.7
  1. *This mating was performed at IBR in Staten Island, New York, whereas the other two matings were performed at Ewha Womans University, Seoul, Republic of Korea.

    SD: standard deviation.

Due to the severe embryonic lethality observed in the Naa10 Naa12 DKO male mice and the Naa10+/- Naa12-/- female mice, which was not seen in each single KO (Naa10 KO or Naa12 KO), it seems likely that, without compensation by Naa12, amino-terminal acetylation is disrupted in Naa10 Naa12 DKO mice. Together, these data support the compensatory role of Naa12 in vivo.

Genotype distribution modeling of Naa10- and Naa12-deficient offspring

The discrepancies we noted between the observed offspring genotype distributions and the expected Mendelian frequencies prompted us to examine the results from four matings (Supplementary file 1h–m) with the goal of understanding the effects of combined Naa10 and Naa12 mutations on embryonic and postnatal mortality. We created mathematical models to predict the observed genotype distribution at each age based on successive incorporation of assumptions of the lethality of specific offspring genotypes. Embryonic genotype data was obtained from two matings for which embryonic genotype data were obtained (Supplementary file 1h, i). Those matings were (1) Naa10+/Y; Naa12+/- males crossed with Naa10+/-; Naa12+/- females (Figure 6, Supplementary file 1h) and (2) Naa10+/Y; Naa12-/- males crossed with Naa10+/-; Naa12+/- females (Figure 6—figure supplement 1, Supplementary file 1i). The genotype numbering shown in Supplementary file 1h was used throughout this analysis, and the corresponding genotypes for all other crosses are aligned to have the same numbers. Each model described below adjusted the expected observed genotype frequencies at each age to account for loss of embryos or pups due to the predicted lethal effects of one or more genotypes by the method described in Materials and methods. Three stages of models (B–D) were compared with the expected Mendelian distribution (model A).

Model B assumed that the double KO male genotype 12 (Naa10-/Y; Naa12-/-) is lethal from very early in development based on the observation that this genotype was not found in any embryos or pups out of 483 obtained genotypes from all litters. Specifically, 0 out of an expected 7.9 were detected at E10.5 or earlier, 0 out of an expected 14.5 were detected at E18.5 or earlier, and 0 out of an expected 46.7 were detected by P3. Thus, the survival for genotype 12 was 0% for all ages examined.

Model C was developed from model B in two stages by incorporating separately observations that the male genotype 11 (Naa10-/Y; Naa12+/-) and the female genotype 6 (Naa10+/-; Naa12-/-) were lethal during mid to late fetal development. Based on the Mendelian model, 5 of 9.8 (51%) expected genotype 11 were detected by E10.5 but only 1 of expected 8.6 (11.6%) were identified on E12.5 or E18.5 and none were detected at P3. The five embryos that were present at E10.5 were noted to be lysed and/or developmentally delayed; the single E18.5 genotype 11 embryo was not observed to be abnormal. Based on the Mendelian model for genotype 6, 1 of 2.6 expected E8.5 embryos and 3 of 5.3 expected E10.5 embryos were identified. All three E10.5 embryos were identified as lysed. Genotype 6 was not identified after age E10.5. Cumulatively, 4 of 7.9 expected embryos detected by E10.5 and 0 of 38.8 expected embryos/pups thereafter.

Model D incorporated the assumptions of models B and C and added adjustments to the survival rates of genotype 5 (Naa10+/-; Naa12+/-) and genotype 10 (Naa10-/Y; Naa12+/+) based on the observations that these genotypes were underrepresented at late fetal ages or early postpartum. Genotype 5 was overrepresented during embryogenesis (31 identified but only 18.4 expected for all embryonic ages) but was underrepresented at P3 (17 of 42 expected) based on the expected Mendelian frequencies. A better analysis was achieved by comparing the observed genotype frequencies with those predicted by model C because the expected distributions are significantly affected by the lethal effects of the three genotypes considered in that model. In that case, the genotype overrepresentation during embryogenesis is somewhat less (31 identified but only 22.1 expected) but the underrepresentation at P3 is significantly increased (17 identified of 62 expected, or 27%). Using a model (D3) that incorporated adjusted survival rates for genotypes 12, 11, 6, and 5, we found that genotype 10 remained underrepresented in the observed postnatal offspring counts. The subsequent model (D4) incorporated adjustments to genotype 10 survival rates at E18.5 and P3 to account for this and then was slightly refined by adjusting other genotype survival rates to maximize the fit of all genotypes at all ages. The survival values for model D4 are shown in Table 4. A comparison of the observed offspring numbers with those predicted by the Mendelian distribution and model D4is shown in Figure 6—figure supplements 25.

Table 4
Model Dgenotype survival by age.
#*GenotypeE8.5 (%)E10.5 (%)E12.5 (%)E18.5 (%)Postnatal (%)
12Naa10-/Y; Naa12-/-00000
11Naa10-/Y; Naa12+/-403510100
6Naa10+/-; Naa12-/-4033000
5Naa10+/-; Naa12+/-10010010010035
10Naa10-/Y; Naa12+/+1001001005555
All Others100100100100100
  1. *Genotype number according to Supplementary file 1h.

    E: embryonic day.

The observations of reduced survival for selected Naa10/Naa12 mutants suggests that Naa10 is the more dominant function (e.g., is able to provide Naa12 functions more successfully than Naa12 can provide Naa10 functions) but that two copies of Naa10 are required to replace complete loss of Naa12 in females, possibly due to X-linked inactivation of Naa10 during development. The stochastic nature of X-linked inactivation in time and space may make Naa10 functionality somewhat unpredictable during development in a background having a mixture of Naa10 and Naa12 mutations.

Statistical examination of weight data in Naa10- and Naa12-deficient mice

To determine whether Naa10 and Naa12 are essential for viability and development, we examined the survival, weights, and growth rates of 688 Naa10 and Naa12 KO and WT mice. The genotypes of mice examined are listed in Supplementary file 1n. To avoid potential survival biases, only weights taken during the first 180 days were included. Growth curves are shown in Figure 7. Age and age-squared (the quadratic term) are both entered in the analyses; the quadratic term shows the degree to which the effect of age itself changes with age.

Supplementary file 1o shows the results in which the weight of Naa10 mice in grams is regressed upon age, Naa10 KO status, and their interaction. Unsurprisingly, age predicts weight for males and females strongly, with growth slowing with age (first column). Though a strong negative effect of the KO is seen in for both males and females (second column), when both age and KO status are modeled together (third column), this effect all but disappears in females. Moreover, in females there is no interaction of KO status with age (fourth column), suggesting that the Naa10 KO status itself has no significant effect on the growth rate in females. For males, however, the main effect of the KO remains when age is included in the model (third column) and the interaction is significant (fourth column), indicating that the Naa10 KO both reduces weight of males overall and lowers the rate of growth.

Results of analyses of mixed-genetic background Naa10/Naa12 mice are shown in Supplementary file 1p. Effects of age and KOs on weight comprise the upper portion of the table, while the lower portion shows their effect on the rate of weight gain. Among females, a significant reduction of weight (above, second column) and in the rate of growth (below, first column) is seen among mice heterozygous for the Naa10 KO. There were no homozygous Naa10 KO female mixed-breed mice available to analyze as the matings were not setup to yield any such mice (so breeding patterns, not mortality in utero, are the reason for this absence). No significant effect on growth rate is seen for heterozygous or homozygous Naa12 KO (above, third column) or for their interactions with age (below, second column), and only the effects of the heterozygous Naa10 KO and its interaction with age are seen in the full model (below, third column). Thus, the Naa12 KO, whether heterozygous or homozygous, does not appear to reduce the weight or growth rate of females, while a heterozygous Naa10 KO is sufficient to reduce both weight and growth rate. Interestingly, when modeled together, both the Naa10 and the Naa12 KOs significantly reduced weight (above, fourth column) and the interaction of the Naa10 and the Naa12 heterozygous KOs significantly reduced weight (above, fifth column). As no female mice were both KO for Naa10 and homozygous KO for Naa12, the effect of the interaction of those two factors could not be determined. The triple interaction of heterozygous Naa10 KO, Naa12 KO, and age was weakly significant, suggesting that the presence of both KOs affects growth rate above and beyond the effects of each KO independently (below, fourth column). No males with KOs of both Naa10 and Naa12 were born, so no test of their interaction was possible. An effect was seen for the Naa10 KO on weight when modeled with age and age2 (second column), and the significant interaction of the Naa10 KO with age and age2 (third column) shows that the Naa10 KO in males reduces the growth rate. As with females, no significant effect of a Naa12 KO, whether heterozygous or homozygous, was seen in males, nor is there a significant interaction with age (fourth column). When the interactions of age with both Naa10 and Naa12 KO status are entered in one model, Naa10 alone is seen to reduce growth rates (fifth column).

Discussion

We have shown that Naa10 deficiency results in pleotropic developmental defects in two different Naa10-deficient mouse models. Similar to infant mortality in some OS males, the lethality of Naa10 KO mice increased dramatically in pups in the first three days of life (Figure 1B). Defects in kidney, brain, pigmentation (piebaldism), and ribs were observed during embryonic or early postnatal stages in some mice (Figure 2B, C). These observed phenotypes overlap with some of the phenotypes found in surviving humans with OS, including supernumerary vertebrae and hydrocephaly, although piebaldism has not been reported to date in any humans. However, the puzzling lack of embryonic lethality in the Naa10 KO mice prompted us to discover Naa12 as a possible compensatory NAT, with Naa10-like amino-terminal acetylation activity (Figure 4A), with an interaction between Naa15 and Naa12 (Figure 4B), and with enzymatic activity in a chimeric complex with human NAA15 (Figure 4C). In addition, co-immunoprecipitation of endogenous Naa15 from Naa10 KO mouse tissues followed by mass spectrometry analysis (Table 2) and amino-terminal acetylation assays (Figure 3C) fully supports that the endogenous Naa12-Naa15 complexes produces NatA activity. Finally, we found genetic proof of the compensatory activity of Naa12 in mice when we observed embryonic lethality in in Naa10 Naa12 DKO male and Naa10+/- Naa12-/- female mice (Figure 6). This compensation by Naa12 explains the mouse proteomics data indicating normal amino-terminal acetylation in Naa10 KO mice (Figure 3A). We have confirmed the expression of Naa12 in various tissues using qPCR (Figure 4—figure supplement 1A).

Gene duplication has long been believed to be a major driving force in evolution that provides genetic novelty in organisms. Paralogous genes, originating by small-scale or whole-genome duplication, overlap functional roles for each other and can completely or partially compensate for the loss of the duplicate gene (Peng, 2019; Veitia, 2017). There is not yet any human reported with complete KO for NAA10. There is one published truncating variant in the C-terminal portion of NAA10 in a male patient with microphthalmia (Cheng et al., 2019), but unfortunately there are no cell lines available from this family to confirm whether any truncated NAA10 protein is expressed, as was shown with a splice-site mutation in a Lenz microphthalmia family (Esmailpour et al., 2014). NAA10 was also identified in screens for essential genes in human cell lines (Blomen et al., 2015; Wang et al., 2015), so it seems unlikely that an unknown NAA10-like paralogous gene exists in humans, other than the already known NAA11.

The pleiotropic phenotypes shown in Naa10 KO mice, including hypopigmentation and supernumerary ribs with a penetrance of 100%, were not observed in the Naa12 KO mice. Naa10 itself has been described to have N-ɛ-acetyl-activity towards internal lysine residues of proteins involved in various disease- and development-related signaling pathways (Lee et al., 2018), although its acetylation of some substrates is controversial (Magin et al., 2016; Vo, 2020). Since the Nt-acetylome appears to be globally intact in MEFs from Naa10 KO mice (Figure 3A), it is possible that the presented phenotypes could be due to the loss of Naa10-specific N-ɛ-acetyl-activity or non-catalytic roles of Naa10 (Aksnes et al., 2019). Alternatively, the quantitative expression of Naa10 and Naa12 might be different within or between tissues, which might then explain why there is clearly a phenotype for Naa10+/- Naa12+/- female mice (not born at Mendelian ratios and the few that are born are usually much smaller) but no apparent phenotype in Naa10+/+ Naa12-/- female mice. It seems likely that the mechanism cannot be simply additive between two equally expressed proteins, because if the expression of each protein is theoretically set at an arbitrary unit of 10, then Naa10+/- Naa12+/- female mice might possibly have half as much of each protein, so that the total dose of both proteins together would be 10, instead of 20. Likewise, the total dose of both proteins together would also be predicted to be 10 in a Naa10+/+ Naa12-/- female. Yet, the Naa10+/- Naa12+/- female mice have a phenotype, whereas the Naa10+/+ Naa12-/- female mice do not (Figure 6). Therefore, other explanations could include different tissue-specific dosages of each protein, different expression between different tissues, possible X-chromosome skewing for the X-linked Naa10 in different tissues, or different functions of the two enzymes, including Naa10-specific N-ɛ-acetyl-activity or non-catalytic roles of Naa10 (Aksnes et al., 2019). These questions remain unanswered and are worth exploring in future studies. It is worth highlighting that X-chromosome inactivation could certainly be one explanation, given that males that are Naa10+/Y with Naa12+/- or Naa12-/- show expected survival rates, whereas females that are Naa10+/- with Naa12+/- show ~35% survival but with Naa12-/- show 0% survival (Table 4).

There are several clinical features that were presented in the original description of OS (Rope et al., 2011) which can now be better understood in light of the phenotypes found in the KO mouse model. For example, all of the affected children in the first families with OS were noted to have large and, in some cases, persistently open fontanels (Myklebust et al., 2015; Rope et al., 2011). For one child (family 1, individual II-1), CT scanning revealed cerebral atrophy with enlarged ventricles, and in another child (family 1, individual III-4), there was evidence on magnetic resonance imaging (MRI) of ‘moderate lateral and third ventricular dilatation without identified cause.’ Lastly, all of the children had respiratory depression and apneic episodes, along with varying course of hypotonia and/or hypertonia (including documented hyperreflexia in at least one case [family 2, individual III-2]). In retrospect, it seems that these clinical features could be consistent with mild hydrocephaly in these probands with OS, which resolved over time. This is also consistent with the ventriculomegaly reported in several female OS probands with missense mutations in NAA10, along with ventriculomegaly in one other male proband who died in the first week of life, with generalized hypotonia and lack of spontaneous respirations (Saunier et al., 2016). One of the female patients with an Arg83Cys mutation in Naa10 (#9 in Table 1 of that paper) was reported as having intraventricular hemorrhage in the occipital horn, hypoxic-ischemic encephalopathy, and a ventriculo-peritoneal shunt. It is possible that this sequence of events is compatible with hydrocephaly with clinical signs and symptoms that required the placement of the shunt.

There are additional cardiac and skeletal features that are also worth re-examining in light of these new findings. In some of the original cases of OS, there were varying levels of pulmonary valve stenosis detected on echocardiography, along with some documentation of pulmonary hypoplasia (Rope et al., 2011). For example, individual III-7 in family 1 was found on echocardiography to have small persistent ductus arteriosus, a mildly decreased left ventricular systolic function, an abnormal appearing aortic valve, an enlargement of the right ventricle, decreased right ventricular systolic function, and persistence of the foramen ovale. Individual III-6 from this same extended family was found on echocardiography to have a thickened bicuspid aortic valve and mild pulmonary hypertension. One of the OS female patients with an Arg83Cys mutation in NAA10 was reported to have ‘supernumerary vertebrae’ (Saunier et al., 2016). Prompted by our findings of supernumerary ribs in the mice, we obtained an MRI report for this patient, in which the radiologist concluded that there appeared to be 25 distinct vertebrae, as opposed to the usual 24, with a suggestion of a 13th rib, at least on the right. The report went on to state that “the vertebrae represent seven cervical vertebrae, 13 rib-bearing thoracic vertebrae, and five lumbar vertebrae, and the L1 vertebra is mildly dysmorphic, with a suggestion of anterior breaking.” In addition, chest and abdominal X-rays from two of the brothers in generation VI of a family with microphthalmia demonstrated the presence of 13 rib-bearing thoracic vertebrae, alongside the dramatic scoliosis in both individuals. Four other females carrying mutations in Naa10 were reported as having either pectus carinatum or excavatum (Saunier et al., 2016), one of the boys with OS (family 1, individual III-4) was noted to have pectus excavatum, and retrospective review of some of the clinical photographs appears to show mild pectus excavatum in individual III-6 of the same family. Studies of human populations have shown that the levels of transition may be shifted cephalad, resulting in 23 mobile vertebrae, or shifted caudad, resulting in 25 presacral vertebrae. Such variations may occur in 2–11% of the population (Bornstein and Peterson, 1966). In addition, the number of ribs can also vary in mice as a result of teratogenic and genetic influences (Mclaren and Michie, 1958; Chernoff and Rogers, 2004). However, the complete penetrance for supernumerary ribs in the Naa10-deficient mice, along with the presence of extras ribs in some of the patients, suggests that there is a pathway common to humans and mice that is altered by mutations involving NAA10.

Several mouse mutants show similar cardiac or skeletal phenotypes to the Naa10-deficient mice. Pax3 mutants phenocopy our Naa10-/- mutants as Pax3+/- adults exhibit 100% piebaldism and exhibit neural crest (NC)-related PTA/DORV with concomitant VSDs (Conway et al., 1997a; Conway et al., 1997b; van den Hoff and Moorman, 2000; Olaopa et al., 2011). Pax3 systemic nulls also have skeletal defects due to abnormal somite morphogenesis (Henderson et al., 1999; Dickman et al., 1999). Moreover, Pax3 cKOs demonstrated that NC-specific deletion is sufficient to cause DORV/VSDs and death at birth (Olaopa et al., 2011; Koushik et al., 2002), and that restricted deletion within the neuroepithelium causes congenital hydrocephalus (Zhou and Conway, 2016). While Pax7 systemic deletion does not cause NC-associated defects, it does exhibit overlapping expression, and Pax3-Pax7 compound heterozygous mice develop hydrocephalus (Zhou and Conway, 2016), suggesting combinatorial function. Hox C8-/- mice exhibit an extra rib and an extra rib articulating with the sternum (Le Mouellic et al., 1992; Juan and Ruddle, 2003), and an unfused sacral vertebra, which lead to 27 presacral vertebrae (van den Akker et al., 2001), as seen in our model. Hox A4-/- mice described in Horan et al., 1994 show cervical fusions of C2/C3, a rib on C7 not fully penetrant and sternal defects with bone ossification anomalies. Hox A5-/- mice display numerous cervico-thoracic defects such as a rib process coming from the seventh cervical vertebra, an increase in the number of sternebrae and total number of ribs (Jeannotte et al., 1993). Both Hox A4-/- and A5-/- mice exhibit an extra rib articulating with the sternum. Hox D3-/- mice are the only Hox gene mutation leading to cervical fusion of both the atlas and axis (Condie and Capecchi, 1993). Hox A9-/- mice have anteriorization of both sacral and lumbar parts, with an extra pair of ribs at the lumbar level. Hox A9-/- mice do not have any relevant sternal defect (Fromental-Ramain et al., 1996). Hox B9-/- mice have an extra rib articulating with the sternum and 14 pairs of rib (Chen and Capecchi, 1997). These phenotypes, especially Hox C8, share common features with the Naa10-deficient mice. This phenotype is also close to the Rpl38-/- phenotype (Kondrashov et al., 2011), except for the sacral fusion described in Rpl38-/- mice. Interestingly, it was shown that Hox genes were dysregulated in this genotype. The skeletal findings and comparison to other mutant mice suggest a pattern consistent with a homeotic anterior transformation hypothesis.

The developmental role of Naa10 in mice has been previously described (Lee et al., 2017). Lee et al. reported embryonic lethality at E12.5–14.5 and beyond (due to placental defects), hydrocephaly, postnatal growth retardation, and maternal effect lethality in Naa10 KO mice and suggested that genomic imprinting dysregulation is associated with those developmental phenotypes. In the present study, hydrocephaly and postnatal growth retardation were also apparent, but embryonic lethality was not observed, which prompted the search for and discovery of Naa12. The previous paper (Lee et al., 2017) did not report the piebaldism, homeotic anterior transformation, hydronephrosis, and genital defects (such as seminal vesicle malformation and hydrometrocolpos), nor did it explain the cause of death in the first day of life, which is at least partly due to congenital heart defects, as reported herein. A more recent paper from the same group reported that conventional and adipose-specific Naa10p deletions in mice resulted in increased energy expenditure, thermogenesis, and beige adipocyte differentiation in the surviving mice (Lee et al., 2019), although the authors do not comment on whether any of the male mice used in that study starting at age 5 weeks ended up developing hydrocephaly and/or hydronephrosis, which we have observed in older mice. Although the Lee et al., 2017 paper reported a very high maternal effect lethality rate of 90% (46/51) (otherwise stated as a survival rate of 10% [5/51]) for newborns in matings following Naa10-/- female and C57BL/6J WT male intercrossing, this rate was only 29% (37/127) in this same mating herein in the first 24 hr of life and with a total death rate in the first three days for all newborns of 46/127, or 36% (Supplementary file 1d), with this result deriving from a larger number of mating pairs, litters, and pups. Although this rate of 36% is higher than that seen with matings involving Naa10+/- females (15.8% and 13.6%) (Supplementary file 1d), the explanation for this ~20% difference in survival in the first three days of life could involve differences in maternal care provided by the Naa10+/- and Naa10-/- females, but this would have to be investigated in future studies, involving detailed behavioral and cognitive assessment of the dams.

The reasons for the differences between the studies in regards to maternal effect lethality and in utero lethality are unknown at present. Whilst Lee et al. deleted Naa10 exons 2–6 (Lee et al., 2017), the current study deleted Naa10 exons 1–4 or used an allele Naa10tm1a expressing β-galactosidase instead of the Naa10 gene (Figure 1—figure supplement 1D), and there was not any significant embryonic lethality in either line (Supplementary file 1a, b). All three of these mouse models were made using 129Sv/Ev ES cells, and all three are nulls lacking Naa10 protein. It is the case that the previous study used the Cre/loxP system to generate the Naa10 KO mice, where a floxed Naa10 female mouse was crossed with the Ella-Cre transgenic male mouse expressing Cre recombinase for germ line deletion of loxP-flanked Naa10, whereas our mice were made using standard gene-targeting methods without the use of Cre recombinase, but it is not clear how this would have resulted in embryonic lethality, particularly as these mice were only used after ‘at least six generations of backcross with C57BL/6 mice,’ which are noted by the authors to be the substrain C57BL/6JNarl, first established at the Animal Center of National Research Institute from the Jackson Laboratory (JAX) in 1995. The explanation for differences in embryonic lethality might be more likely due to different combinations of modifying alleles that are present in the different C57BL/6J substrain genetic backgrounds, rather than differences in our model systems, and future plans will address this after back-crossing more than 20 generations to C57BL/6J (imported annually from JAX) to achieve an entirely inbred line. The impact of genetic background is supported by the observation that additional null alleles on mixed genetic backgrounds, made during the process of generating missense mouse models for OS, have far less penetrance for a range of the various phenotypes, including much less perinatal lethality (unpublished observations).

In conclusion, our study provides strong evidence that Naa10, the catalytic subunit of NatA, is critical for normal development in mice. Furthermore, this study explains the puzzle regarding the lack of complete embryonic lethality in the Naa10 KO mice due to the discovery of a second mouse Naa10 paralog, which, unlike Naa11, is expressed in the heart as well as other tissues. Taken together, our findings suggest that the newly identified Naa12 can functionally rescue Naa10 loss and act as a catalytic subunit in mouse NatA complexes.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Gene (Mus musculus)Naa10GenBankMGI:MGI:1915255
Gene (M. musculus)Naa15GenBankMGI:MGI:1922088
Gene (M. musculus)Naa11GenBankMGI:MGI:2141314
Gene (M. musculus)Naa12This paperGm16286, UniProt: Q9CQX6Provided by corresponding author, Gholson J. Lyon
Genetic reagent (M. musculus)Naa10-/-Nature Communication Yoon et al., 2014Provided by corresponding author, Goo Taeg Oh
Genetic reagent (M. musculus)Naa12-/-This paperGm16286, UniProt: Q9CQX6Provided by corresponding author, Gholson J. Lyon
Cell line (Homo sapiens)HEK293 (normal, embryonic kidney cells)ATCCCRL-1573
Biological sample (M. musculus)Primary mouse embryonic fibroblastsThis paperFreshly isolated from mouse embryos (E13.5)
AntibodyAnti-Naa10 (rabbit polyclonal)AbcamCat# ab155687(1:1000)
AntibodyAnti-Naa10 (rabbit polyclonal)Protein TechCat# 14803-1-AP(1:3000)
AntibodyAnti-Naa10 (rabbit monoclonal)Cell SignalingCat# 13357(1:1000)
AntibodyAnti-Naa10 (goat polyclonal)Santa CruzCat# sc-33256(1:1000)
AntibodyAnti-Naa10 (rabbit polyclonal)Santa CruzCat# sc-33820(1:1000)
AntibodyAnti-Naa11 (rabbit polyclonal)Novus BiologicalsCat# NBP1-90853(1:1000)
AntibodyAnti-Naa15/NARG1 (mouse monoclonal)AbcamCat# ab60065(1:1000)
AntibodyAnti-NAA15 (rabbit polyclonal)Biochemical Journal (reference 12 in this paper)
Arnesen et al., 2005
(1:2000)
Provided by author Thomas Arnesen,
AntibodyAnti-NAA50 (rabbit polyclonal)LifeSpan BioSciencesCat# LS-C81324-100(1:3000)
AntibodyAnti-FLAG (rabbit polyclonal)Sigma-AldrichCat# F7425(2 μg/mL)
AntibodyAnti-GAPDH (mouse monoclonal)AbcamCat# ab9484(1:3000)
AntibodyAnti-actin (goat polyclonal)Santa CruzCat# 1615(1:3000)
AntibodyAnti-GST (mouse monoclonal)GenScriptCat# A00865(1 µg/mL)
AntibodyAnti-V5 (mouse monoclonal)Life TechnologiesCat# R960-25(1:1000)
AntibodyAnti-Naa12 (rabbit polyclonal)This paperGm16286, UniProt: Q9CQX6C-terminus (aa191-205: QENLAGGDSGSDGKD-C) conjugated to OVA by PrimmBiotech
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmNaa10-Exon2/3_FThis paperPCR primersctcttggccccagctttctt
Provided by corresponding author, Goo Taeg Oh
Sequence-based reagentmNaa10-Exon3/4_RThis paperPCR primerstcgtctgggtcctcttccat
Provided by corresponding author, Goo Taeg Oh
Sequence-based reagentmNaa11_FThis paperPCR primersaccccacaagcaaagacagtg
Provided by corresponding author, Goo Taeg Oh
Sequence-based reagentmNaa11_RThis paperPCR primersagcgatgctcaggaaatgctct
Provided by corresponding author, Goo Taeg Oh
Sequence-based reagentmNaa12(Gm16286)_FThis paperPCR primersacgcgtatgctatgaagcga
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmNaa12(Gm16286)__RThis paperPCR primersccaggaagtgtgctaccctg
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmNaa15_FThis paperPCR primersgcagagcatggagaaaccct
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmNaa15_RThis paperPCR primerstctcaaacctctgcgaacca
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmNaa50_FThis paperPCR primerstaggatgccttgcaccttacc
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmNaa50_RThis paperPCR primersgtcaatcgctgactcattgct
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmGAPDH_FThis paperPCR primersaggtcggtgtgaacggatttg
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmGAPDH_RThis paperPCR primerstgtagaccatgtagttgaggtca
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmACTB_FThis paperPCR primersggctgtattcccctccatcg
Provided by corresponding author, Gholson J. Lyon
Sequence-based reagentmACTB_RThis paperPCR primersccagttggtaacaatgccatgt
Provided by corresponding author, Gholson J. Lyon
Software, algorithmZen 3.0 SRZEISSVersion 16.0.1.306Black 64bit edition
OtherAlcian Blue 8GXSigma-AldrichCat# A52680.03%
OtherAlizarin RedSigma-AldrichCat# A55330.05%
OtherHematoxylinSigma-AldrichCat# MHS80
OtherEosinSigma-AldrichCat# HT110116

Mice

All experiments were performed in accordance with guidelines of International Animal Care and Use Committee (IACUC) of Ewha Womans University (protocol #18-012), Cold Spring Harbor Laboratory (CSHL) protocol #579961-18, and Institute for Basic Research in Developmental Disabilities (IBR) (protocol #456). At CSHL and IBR, any matings that required genotyping were screened on a daily basis by animal husbandry staff, with notation of how many newborn pups were present each morning, but with paw tattoo and tail genotyping not being performed until day 3 of life, so as to not disturb the litters and thus to not increase the risk for maternal rejection of the litter. The stock of C57BL/6J was replenished annually from Jackson Laboratory so as to avoid genetic drift from the JAX inbred line.

Generation of Naa10-deficient mice

Request a detailed protocol

The Naa10 KO mice were generated as previously described (Yoon et al., 2014). Naa10tm1a [B6;129P2-Ard1tm1a(Eucomm)Gto/J] (Naa10tm1a) mice, used for Naa10 reporter mouse, were generated using standard method based on a standard gene-targeting in E14 embryonic stem (ES) cells (129/Sv) by using a targeting vector from EUCOMM. Correctly targeted ES clones were used for blastocyst microinjection and generation of chimeric mice. Chimeric mice were crossed to C57BL/6J mice, and then the progeny were backcrossed to C57BL/6J for more than 10 generations. The Naa10-deficient mice used in the weight analyses were derived from mice backcrossed eight times to a C57BL/6J inbred genetic background, and this was confirmed with genome scanning at the Jackson Laboratory, showing heterozygosity for only one marker for 129S1/SvImJ out of 290 autosomal markers tested, thus giving a percentage of C57BL/6J of 99.66%.

Generation of Naa12 (Gm16286, UniProt: Q9CQX6) KO mice

Request a detailed protocol

The mice were made using standard methods by microinjection of CRISPR reagent mix into zygotes obtained from the mating of B6D2F1 females (i.e., 50% C57BL/6J, 50% DBA/2J [D2]) females to inbred C57BL/6J males. The guide RNA was produced and validated from Sigma using a Cel1-nuclease assay, and the most active guide was selected, which was Naa12_0_125 (C9587), with a target sequence of GAGCGTTTCACAGCCAGCG and including the targeting cr-RNA sequence and the tracrRNA portion. The indels were transmitted by breeding again to inbred C57BL/6J males, and the resulting progeny were interbred on a mixed genetic background of approximately 12.5% DBA/2J (D2)/87.5% C57BL/6J, for use in the reported experiments, including the weight analyses. Progeny from these mice have been backcrossed to C57BL/6J for more than 10 generations, with no discernible new phenotypes emerging. Genomic DNA was isolated from paw and tail. DNA was screened for mutations using PCR and Surveyor assay (Qiu et al., 2004), followed by Sanger sequencing of selected clones and the use of CRISP-ID (Dehairs et al., 2016) to identify putative deletions.

Primers for mice genotyping

Request a detailed protocol

The primers used for Naa10 KO and Naa10tm1a genotyping were Naa10-F: 5′-cctcacgtaatgctctgcaa-3′, Naa10-neo-F: 5′-acgcgtcaccttaat-atgcg-3′, Naa10-R: 5′-tgaaagttgagggtgttgga-3′, Naa10tm1a-F: 5′-gcacactctctgaattggac-3′, Naa10tm1a-neo-F: 5′-ggccgcttttctggattcat-3′, and Naa10tm1a-R: 5′-gcaggggaataaggcattgg-3′. The primers used for Naa12 KO were Naa12 Surveyor F: 5′-gctccacctcgctaacctgg-3′, Naa12 Surveyor R: 5′-gccagatgacctgatgaacatgc-3′ and HEX-Naa12 Surveyor F: 5′-gctccacctcgctaacctgg-3′.

Antibodies

The following antibodies were used: rabbit anti-Naa10 (Abcam #ab155687), rabbit anti-Naa10 (Protein Tech #14803-1-AP), rabbit monoclonal anti-NAA10 (Cell Signaling, #13357), goat anti-Naa10 (Santa Cruz, #sc-33256), rabbit anti-Naa10 (Santa Cruz, #sc-33820), rabbit anti-Naa11 (Novus Biologicals; #NBP1-90853), mouse anti-Naa15/NARG1 (Abcam; #ab60065), rabbit polyclonal anti-NAA15 (Arnesen et al., 2005), rabbit anti-Naa50 (LifeSpan BioSciences; #LS-C81324-100), rabbit anti-FLAG (Sigma; #F7425), mouse anti-GAPDH (Abcam; #ab9484), goat anti-actin (Santa Cruz, #1615), mouse anti-GST (GenScript; #A00865), and mouse anti-V5 (Life Technologies; #R960-25). The antibody against the potential mNaa10 paralog mNaa12 (Gm16286, UniProt: Q9CQX6) was raised in rabbits after immunization with a synthetic peptide of the Naa12 C-terminus (aa191-205: QENLAGGDSGSDGKD-C) conjugated to OVA by PrimmBiotech.

Alcian Blue and Alizarin Red co-staining of skeletons

Request a detailed protocol

After the skin and internal organs were removed, embryos were fixed in 95% ethanol (EtOH) for 4 hr, then in 100% acetone for overnight. Embryos were stained with 0.03% Alcian Blue 8GX in ethanol/acetic acid (4:1 v/v) for overnight and kept in 1% KOH for 2 days until they became clearly visible, followed by staining with 0.05% Alizarin Red in 1% KOH for 4 hr. After washing with 100% glycerol/1% KOH (1:1 v/v), skeletons were kept in 100% glycerol.

Isolation and imaging of mouse embryos

Request a detailed protocol

Timed matings were performed either by using the presence of a vaginal plug to assess fertilization. The morning vaginal plug was designated E0.5. Pregnant mice were sacrificed at several time points after conception. The embryos were isolated in ice-cold PBS with 1% FBS and washed three times in ice-cold PBS. Embryos were imaged using a Zeiss Axiozoom V16 with Zen software and merged 50 slides between Z-stack intervals.

β-Galactosidase staining

Request a detailed protocol

Isolated E10.5 embryos were incubated in fixation solution (4% paraformaldehyde) at 4°C for 25 min. Samples were washed in ice-cold PBS and then incubated in permeabilization solution (PBS containing 0.01% Na deoxycholate, 0.02% Nonidet-P40, 2 mM MgCl2) for 20 min at 4°C. Subsequently, samples were incubated in β-gal staining solution (PBS containing 1 mg/mL X-Gal, 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 0.02% Nonidet-P40, 2 mM MgCl2) at 37°C overnight. Following β-gal staining, samples were washed with PBS and incubated in fixation solution at 4°C for storage.

Hematoxylin and eosin staining

Request a detailed protocol

Isolated kidney tissues at E18.5 and P3 were fixed with 4% paraformaldehyde at 4°C for overnight and embedded in paraffin. Samples were sectioned at 8 μm thick and stained with hematoxylin (MHS80, Sigma) and eosin (HT110116, Sigma) for morphology.

Cloning

Request a detailed protocol

Full-length mouse Naa10 and Naa12 (Gm16286, UniProt: Q9CQX6) expression vectors were separately constructed using a pMAL-c5x vector. In both cases, the catalytic subunit contained an N-terminal uncleavable MBP-tag. Bacterial expression vectors of mNATs were cloned from cDNA generated from mouse liver or testes. mRNA was isolated using the Oligotex direct mRNA kit (Qiagen) according to the manufacturer's recommendations. 1 µg RNA was reverse transcribed with Superscript IV reverse transcriptase (Thermo Fisher) and Oligo dT(18) primer. The PCR product was digested and cloned into BamHI restriction sites of pGEX-4T1 (GE Healthcare), pMAL-p5X and p3xFLAG-CMV10 (Sigma-Aldrich) using standard techniques. All constructs were sequenced to validate correct insert and orientation.

Primers for cloning

Request a detailed protocol

cDNA was amplified using the primers CCG GGA TCC ATG AAC ATC CGC AAT and CTG GGA TCC CTA GGA GGC AGA GTC AGA for mNaa10 variants, CCG GGA TCC ATG AAC ATC CGC AAT GC and CTG GGA TCC CTA GGA GAT GGA ATC CAA GTC for mNaa11, CCG GGA TCC ATG AAC ATC CGC CGG and CTG GGA TCC CTA GGA GGC GGA CCC TAG for mNaa12.

Peptide competition assay

Request a detailed protocol

To determine the specificity of the mNaa12 antibody, a peptide competition assay was performed using the same peptide as used for immunization (aa 191–205: QENLAGGDSGSDGKD-C). 100 µg antibody were bound to 50 mg peptide-coupled CNBr-Sepharose (10 mg peptide/g Sepharose) in PBS + 0.2% Triton X-100 for 1 hr at 4° C on an orbital shaker. The beads were pelleted by centrifugation at 2.700 × g for 3 min at 4°C and 250 µL of the antibody-depleted supernatant diluted in 5 mL TST for detection (1:100 final antibody dilution). Western blots of mouse lysates were probed with the depleted antibody or untreated antibody as control (1:100 dilution in TST).

Cell lines

Request a detailed protocol

HEK293 cells were purchased from ATCC, authenticated via STR profiling, and confirmed mycoplasma free.

Co-immunoprecipitation assay

Request a detailed protocol

Protein-protein interaction studies were performed in HEK293 cells. Briefly, 8 × 105 cells were seeded per well in 6-well plates. After 24 hr, cells were co-transfected with pcDNA3.1/V5-His-mNaa15 and p3xFLAG-CMV10-Naa10235 (isoform 1), -Naa10225 (isoform 2), -Naa11, or -Naa12 or the corresponding empty vectors. Cells were lysed after 48 hr in 200 µL PBS-X per well and cellular debris pelleted at 20.800 × g for 10 min at 4°C. 350 µL of the generated lysate was incubated with 1 µg anti-V5 antibody for 1 hr at 4°C, followed by a 30 min incubation with 30 µL protein-A Sepharose (Sigma-Aldrich). Protein complexes were washed three times by centrifugation (2.700 × g, 2 min) and eluted in 30 µL 2×SDS sample buffer.

Proteins were separated by SDS-PAGE and transferred onto a nitrocellulose membrane (Amersham Protran 0.2 µM NC) by immunoblotting. The membrane was blocked in 5% non-fat dry milk and incubated overnight with rabbit polyclonal anti-NAA15 (Arnesen et al., 2005) (1:2000, BioGenes) and rabbit monoclonal anti-NAA10 (anti-ARD1A, 1:1000, Cell Signaling, #13357) diluted in 1× PBS containing 1% non-fat dry milk and 0.1% Tween. The immunoblots were washed and incubated for 1 hr at room temperature (RT) with HRP-linked secondary antibody donkey anti-rabbit IgG (GE Healthcare, NA934). The HRP-signal was detected using SuperSignalTM West Pico PLUS Chemiluminescent Substrate Kit (Thermo Scientific) and ChemiDocTM XRS+ system (Bio-Rad) and visualized by ImageLab Software (Bio-Rad).

Immunoprecipitation of Naa15 to form NatA complex

Request a detailed protocol

For immunoprecipitation of Naa15, 90–120 mg liver tissue from a WT- and Naa10 KO mouse was lysed in 500 µL IPH lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM) NaCl, 5 mM EDTA, 0.5% NP-40, 1× complete EDTA-free protease inhibitor cocktail (Roche) using Kontes Pellet Pestle Motor and incubated on ice for 40 min. Cell debris was pelleted by centrifugation (17,000 × g, 4°C, 10 min) and the supernatants transferred to new Eppendorf tubes. The protein concentration was determined by BCA Protein Assay Kit (Thermo Scientific) and the tissue lysates were subsequently diluted with IPH lysis buffer to an equal protein concentration of 25 µg/µL. The WT- and Naa10 KO tissue lysates were then divided in two, whereof one half was mixed with 15 µg of anti-Naa15 antibody and the other half with 15 µg of anti-V5 antibody as a negative control. The mixtures were incubated at 4°C for 3 hr on a rotator. Afterwards, 180 µL of Protein A/G magnetic beads (Millipore) pre-washed in IPH lysis buffer was added to each sample and incubated overnight. Then, the magnetic beads were washed three times in IPH lysis buffer and two times in 1× acetylation buffer (100 mM Tris-HCl pH 8.5, 2 mM EDTA, 20% glycerol) prior to being resuspended in 90 µL of 2× acetylation buffer and used in a [14C]-Ac-CoA-based acetylation assay.

[14C]-Ac-CoA-based acetylation assay of immunoprecipitated samples

Request a detailed protocol

Three positive replicates were prepared for each IP sample containing 10 µL IP beads, 200 µM synthetic oligopeptide SESS24 (BioGenes), 100 µM [14C]-Ac-CoA (Perkin-Elmer), and dH2O to a final volume of 25 µL. In addition, two replicates for each IP sample were prepared without synthetic oligopeptide as negative controls. The samples were incubated at 37°C for 45 min in a thermomixer with shaking at 1400 rpm. Finally, the magnetic beads were isolated and 23 µL of the supernatant transferred to P81 phosphocellulose filter disks (Millipore). The filter disks were washed three times for 5 min in 10 mM HEPES buffer (pH 7.4) and air dried. To determine the amount of incorporated [14C]-Ac, the filter disks were added to 5 mL Ultima Gold F scintillation mixture (Perkin-Elmer) and analyzed by a Perkin-Elmer TriCarb 2900TR Liquid Scintillation Analyzer.

Proteomics sample preparation

Request a detailed protocol

Immunoprecipitation of Naa15 from a WT- and Naa10 KO mouse was performed as described above. Bound proteins were eluted from the magnetic beads using 60 µL of elution buffer (2% SDS, 100 mM Tris-HCl pH 7.6, 0.1 M DTT) and heated for 5 min at 95°C. The eluates were processed for LC-MS/MS analysis using filter-aided sample preparation (FASP) method (Wiśniewski et al., 2009). The eluted protein mixtures were mixed with UA buffer (8 M urea, 100 Mm Tris-HCl pH 8.0) and centrifuged through Microcon 30 kDa MWCO filters followed by Cys-alkylation with 50 mM iodoacetamide dissolved in UA buffer. Afterwards, the buffer was exchanged with 50 mM ammonium bicarbonate through sequential centrifugation, proteins were trypsinized (Sequencing Grade Modified Trypsin, Promega), and digested peptides were collected by centrifugation. Peptides were acidified using 5% formic acid and desalted using C18-stagetips according to protocol (Rappsilber et al., 2007). Briefly, 40 µg peptides from each sample were loaded onto C18-stagetips pre-conditioned with buffer A (1% formic acid). The C18-stagetips were then washed with buffer A, before peptides were eluted with buffer B (80% acetonitrile [ACN], 1% formic acid). The final eluate was concentrated by Speedvac to evaporate ACN and diluted to desired volume with 5% formic acid.

Mass spectrometric analysis for immunoprecipitate

Request a detailed protocol

1 µg of the peptide samples were injected into an Ultimate 3000 RSLC system (Thermo Scientific) connected to a Q-Exactive HF mass spectrometer (Thermo Scientific) equipped with EASY-spray nano-electrospray ion source (Thermo Scientific). Trapping and desalting was performed with 0.1% TFA (flow rate 5 µL/min, 5 min) on a pre-column (Acclaim PepMap 100, 2 cm × 75 µm ID nanoViper column, 3 µm C18 beads). Peptides were separated on an analytical column (PepMap RSLC, 50 cm × 75 µm i.d. EASY-spray column, 2 µm C18 beads) during a biphasic ACN gradient with a flow rate of 200 nL/min. Solvent A (0.1% FA [vol/vol] in water) and B (100% ACN) were used for the following gradient composition: 5% B for 5 min, 5–8% B for 0.5 min, 8–24% B for 109.5 min, 24–35% B for 25 min and 35–80% B for 15 min, 80% B for 15 min, and conditioning with 5% B for 20 min. The mass spectrometer was operated in data-dependent mode to automatically switch between full-scan MS and MS/MS acquisition. MS spectra (m/z 375–1500) were acquired with a resolution of 120,000 at m/z 200, automatic gain control (AGC) target of 3 × 106, and maximum injection time (IT) of 100 ms. The 12 most intense peptides above an intensity threshold (50,000 counts, charge states 2–5) were sequentially isolated to an AGC target of 1 × 105 and maximum IT of 100 ms and isolation width maintained at 1.6 m/z, before fragmentation at a normalized collision energy of 28%. Fragments were detected in the orbitrap at a resolution of 15,000 at m/z 200, with first mass fixed at m/z 100. Dynamic exclusion was utilized with an exclusion time of 25 s and ‘exclude isotopes’ enabled. Lock-mass internal calibration (m/z 445.12003) was used. Raw files were processed with MaxQuant v. 1.6.17.0 (Cox and Mann, 2008) and searched against a database of Swiss-Prot annotated mouse protein sequences (retrieved 22.06.2018) in which the NAA12 sequence was added manually, and with a reverse decoy database. MaxQuant was run with default settings. Peptide and protein identifications were filtered to a 1% false discovery rate (FDR). Minimum peptide length was set to 7. Modifications included in protein quantification were oxidation (M), Nt-acetylation, acetylation (K), and phosphorylation (STY). Other parameters: match between runs – true, matching time window – 0.7 min, alignment time window – 20 min, find dependent peptides – true, mass bin size – 0.0065. Protein and peptide intensities were quantified by label-free quantification (LFQ) (Cox et al., 2014). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD026684.

Whole-body CT scanning

Request a detailed protocol

CT scans were acquired on a Nanoscan PET/CT scanner from Mediso using Nucline v2.01 software. All mice were kept sedated under isoflurane anesthesia for the duration of the scan. Scans were acquired with an X-ray tube energy and current of 70 kVp and 280 uA, respectively. 720 projections were acquired per rotation, for three rotations, with a scan time of approximately 11 min, followed by reconstruction with a RamLak filter and voxel size 40 × 40 × 122 µm. For ex vivo analyses, mouse heads were fixed in 10% formalin buffered saline, followed by scanning and reconstruction with 1440 projections per revolution. Cranial volume was measured using VivoQuant software (v2.50patch2) using the spline tool to manually and accurately draw around the circumference of the cranium on multiple stepwise 2D slices.

Integrated N-terminal peptide enrichment (iNrich) assay

Request a detailed protocol

iNrich assays were performed as described (Ju et al., 2020). MEFs were made from E13.5 embryos, using standard techniques, with DMEM media supplemented with 10% fetal bovine serum (FBS), L-glutamine, and penicillin/streptomycin. Cells were harvested by trypsinization, washed twice with ice-cold phosphate-buffered saline (PBS, pH 7.4; Gibco), and resuspended in ice-cold lysis buffer containing 0.2 M EPPS (pH 8.0), 6 M guanidine, 10 mM TCEP (Thermo Fisher Scientific), and 40 mM 2-chloroacetamide (Sigma-Aldrich). After 10 min of incubation on 95°C, cells were lysed by ultrasonication by a BranSonic 400B. The proteins from the cell lysate were isolated by transferring supernatant after centrifugation at 12,000 g for 10 min at 4°C. The protein concentration of the collected supernatant was determined by bicinchoninic acid (BCA) protein assay. After enrichment of the N-terminal peptides, the peptide samples were analyzed by LC–MS/MS on an LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific) without further fractionation. Mass spectrometry data were uploaded to PRIDE under project name: Naa10 mutant mouse N-terminome LC-MS, project accession: PXD026410. Data analysis used unpaired, equal variance algorithm for Student’s t-test.

RNA and protein isolation and assays

Request a detailed protocol

70–120 mg tissues were lysed in 5 μL/mg tissue RIPA buffer (Sigma) with 1× Complete protease inhibitors and 1 U/μL Superase In RNase inhibitor (Thermo Scientific) using Fisherbrand Pellet Pestle Cordless Motor. Afterwards, homogenization debris was removed by centrifugation at 20.800 × g for 10 min at 4°C. Protein concentration was determined using APA assay (Cytoskeleton Inc) and 50 μg total protein were separated on SDS-PAGE followed by western blot. Membranes were stained with anti-Naa10, anti-Naa15, and anti-GAPDH antibodies (all Abcam).

For RNA purification, 30 μL clarified lysates were mixed with 70 μL RNase free water and RNA isolated using the RNeasy Mini Kit (Qiagen) according to the manufacturers recommendations, including on-column Dnase digest. 1 μg RNA was reverse transcribed using the TaqMan Reverse transcription kit and gene level detection performed using SYBR Green Master Mix (all Thermo Scientific). Relative expression was normalized to GAPDH and ACTB.

For the characterization of the mNaa12 antibody, tissue was lysed in 2 µL per mg tissue PBS-X (PBS + 0.2% [v/v] Triton X-100 + 1× Complete protease inhibitor cocktail). 10–200 µg lysate were subjected to SDS-PAGE and western blot.

Primers for mice qPCR

Request a detailed protocol

The following primers pairs were used: mNaa10-Exon2/3 F: 5′-ctcttggccccagctttctt-3′ and mNaa10-Exon3/4 R: 5′- tcgtctgggtcctcttccat-3′, mNaa11-F: 5′-accccacaagcaaagacagtg-3′ and mNaa11-R: 5′-agcgatgctcaggaaatgctct-3′, mNaa12(Gm16286)-F: 5′-acgcgtatgctatgaagcga-3′ and mNaa12(Gm16286)-R: 5′-ccaggaagtgtgctaccctg-3′, mNaa15-F: 5′-gcagagcatggagaaaccct-3′ and mNaa15-R: 5′-tctcaaacctctgcgaacca-3′, mNaa50-F: 5′-taggatgccttgcaccttacc-3′ and mNaa50-R: 5′-gtcaatcgctgactcattgct-3′, mGAPDH-F: 5′-aggtcggtgtgaacggatttg-3′ and mGAPDH-R: 5′-tgtagaccatgtagttgaggtca-3′, mACTB-F: 5′-ggctgtattcccctccatcg-3′ and mACTB-R: 5′-ccagttggtaacaatgccatgt-3′.

Expression and purification of WT mouse, Naa10, and Naa12

Request a detailed protocol

All constructs were expressed in Rosetta (DE3)pLysS competent Escherichia coli cells. Cells were grown in LB-media to OD600 0.6–0.7 prior to inducing protein expression with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 18°C for ~16 hr. All subsequent purification steps were carried out at 4°C. Cells were isolated by centrifugation and lysed in lysis buffer containing 25 mM Tris, pH 8.0, 150 mM NaCl, 10 mM β-mercaptoethanol (β-ME), 10 µg/mL phenylmethanesulfonylfluoride (PMSF), and DNase. The lysate was clarified by centrifugation and incubated with amylose agarose resin (New England Biolabs) for 1 hr before washing the resin with ≥100 column volumes of lysis buffer and then eluted with 10-column volumes of lysis buffer supplemented with 20 mM maltose. The resulting eluent was pooled and concentrated to ~10 mg/mL (30 kDa concentrator; Amicon Ultra, Millipore) such that 500 µL was loaded onto a Superdex 200 Increase 10/300 GL gel filtration column (GE Healthcare). The gel filtration run was performed in sizing buffer containing 25 mM HEPES, pH 7.0, 200 mM NaCl, and 1 mM TCEP. After confirming the purity of the peak fractions at ~14 mL by denaturing SDS-PAGE (15% acrylamide), peak fractions were concentrated to 0.6 (6.1 µM) WT mouse Naa10 and 0.3 mg/mL (3.5 µM) WT mouse Naa12, as measured by UV280 (Nanodrop 2000; Thermo Fisher Scientific), and stored at 4°C.

Expression and purification of recombinant mNaa12 (1–160)-hNaa15 constructs

Subcloning

Request a detailed protocol

Both full-length and truncated (1–160) mouse Naa12 were amplified from the pMAL-c5x Naa12 plasmid using Q5 HF Master Mix (NEB), AAAACCCGGGTATGAACATCCGCCGGGCTCGGC as the forward primer, and either AAAAGGTACCCTAGGAGGCGGACCCTAGGGTCTG (full-length) or AAAAGGTACCTCACCGTCTCAGCTCATCGGCCATCTG (1-160) as the reverse primer. An Spodoptera frugiperda (Sf9) pFastBac dual vector containing the sequence for the N-terminally 6xHis-tagged human Naa15 and truncated human Naa10 (residues 1–160) sequences was digested using KpnI-HF (NEB) and XmaI (NEB) to remove the human Naa10 sequence. The PCR product was also digested using the same restriction enzymes and ligated into the corresponding restriction sites using Mighty mix (Takara) using standard techniques. Both constructs were sequenced to validate the insert sequence and directionality.

Sf9 cells were grown to a density of 1 × 106 cells/ml and infected using the amplified baculoviruses to a multiplicity of infection (MOI) of ~1–2. Because the full-length mNaa12 construct did not produce protein, cells transfected with mNaa121-160/hNaa15 were grown at 27°C and harvested 48 hr post infection. All subsequent purification steps were carried out at 4°C. Following centrifugation of the cells, the pellet was resuspended and lysed in buffer containing 25 mM Tris, pH 8.0, 500 mM NaCl, 10 mM Imidazole, 10 mM β-ME, 10 µg/mL PMSF, DNase, and complete, EDTA-free protease inhibitor tablet (Roche). The lysate was clarified by centrifugation and incubated with nickel resin (Thermo Scientific) for 1 hr before washing the resin with ~125 column volumes of lysis buffer and then eluted with 10-column volumes of elution buffer (25 mM Tris, pH 8.0, 500 mM NaCl, 200 mM imidazole, 10 mM β-ME). Eluted protein was diluted to a final salt concentration of 200 mM NaCl and loaded onto a 5 mL HiTrap SP ion-exchange column (GE Healthcare). The protein was eluted in the same buffer with a salt gradient (200 mM to 1 M NaCl) over the course of 20 column volumes. Using the resulting peak fractions, the remainder of the purification was performed as described for the recombinant monomeric mNaa10 and mNaa12. However, resulting size-exclusion fractions were analyzed by denaturing SDS-PAGE using a 12% acrylamide gel, which was then silver stained (Bio-Rad) according to the manufacturer's instructions.

In vitro radioactive acetyltransferase assays with recombinant protein

Request a detailed protocol

For recombinant mNaa12 and mNaa10 constructs, the assays were carried out in 40 mM HEPES, pH 7.5, 200 mM NaCl, where reactions were incubated with 150 nM of the gel-filtration purified WT mouse Naa10 or Naa12 in a 30 µL reaction volume containing each 250 µM substrate peptide and radiolabeled [14C]acetyl-CoA (4 mCi/mmol; PerkinElmer Life Sciences) for 12 min (Naa12) or 13 min (Naa10) at 25°C. Respective time points were selected to ensure detection of sufficient activity within the linear range as determined by a time-course experiment. The substrate peptides used in the assay correspond to the first seven amino acids of β-actin (DDDIAAL-), γ-actin (EEEIAAL-), or the in vivo NatA complex substrate high-mobility group protein A1 (SESSS-), along with C-terminal positively charged residues for capture to the anion exchange paper. Background control reactions were performed in the absence of enzyme or in the absence of substrate peptide to ensure that any possible signal due to chemical acetylation was negligible. Each reaction was performed in triplicate.

To quench the reaction, 20 µL of the reaction mixture was added to negatively charged P81 phosphocellulose squares (EMD Millipore), and the paper disks were immediately placed in wash buffer (10 mM HEPES, pH 7.5). The paper disks were washed three times, at 5 min per wash, to remove unreacted acetyl-CoA. The papers were then dried with acetone and added to 4 mL of scintillation fluid, and the signal was measured with a PerkinElmer Life Sciences Tri-Carb 2810 TR liquid scintillation analyzer. The counts per minute were converted to molar units using a standard curve of known [14C]acetyl-CoA concentrations in scintillation fluid.

Full peptide sequences:

  • β-actin: NH2-DDDIAALRWGRPVGRRRRPVRVYP-COOH

  • γ-actin: NH2-EEEIAALRWGRPVGRRRRPVRVYP-COOH

  • High-mobility group protein A1: NH2-SESSSKSRWGRPVGRRRRPVRVYP-COOH

For mNaa12-hNaa15, reactions were carried out similar to the monomeric mNaa12 and mNaa10, with the following exceptions: reactions were prepared by combining 21 µL of the respective fraction or sizing buffer with 5 µL of 10X buffer (500 mM HEPES, pH 7.5) to yield a buffer composed of 50 mM HEPES, pH 7.5, 140 mM NaCl, 0.7 mM TCEP, and 250 µM of each substrate upon reaction initiation. The reactions were allowed to incubate overnight at ambient temperatures (~25°C) and then quenched as described above. Control reactions were conducted in parallel as described above without conversion to molar units. Two technical replicates of the reactions were performed.

Statistical analyses

Request a detailed protocol

Significant differences (p<0.05) are indicated by asterisks. Weight analyses were performed using generalized estimating equations (GEEs) (Zeger and Liang, 1986), an extension of generalized linear models that adjusts for the effects of autocorrelation resulting from multiple measurements, and implemented within version 15.1 of Stata (StataCorp 2017).

Genotype distribution analyses and modeling

Request a detailed protocol

Genotype distributions for several Naa10/Naa12 KO crosses were analyzed and models were created to estimate the number of the live (or at least intact) embryos or pups that are expected to be observed based on the assumptions and rules that follow. (1) Genotype survival rates are the fractional value, from 0 to 1 (or 0–100%), of the expected Mendelian fraction for that genotype in the cross being evaluated. (2) Genotype survival rates cannot exceed 1 (or 100%). (3) Genotype survival rates can decrease with age but not increase. (4) WT genotypes (Naa10+/Y; Naa12+/+ and Naa10+/+; Naa12+/+) are expected to have 100% survival at all ages because the models predict the number of embryos or pups relative to WT survival. Reductions in overall in litter sizes for crosses were estimated through other calculations. (5) The biological basis for a reduced survival rate assumes that loss of one or more copies of either Naa10 or Naa12 removes or reduces functions that are required for successful embryonic development or postnatal life. Reduced survival rates for non-WT genotypes were estimated based on differences (delta) between the expected number of embryos or pups based on the Mendelian proportion (or the current best model) and the observed number of embryos or pups. Separate comparisons were made using deltas for each specific age and for the cumulative numbers at each age. (6) Genotype frequencies for each model were calculated as described in the section below. (7) The fit between a model and the observed data was determined by calculating the relative standard deviation (SD) for the deltas across all genotypes, for example, the SD across genotypes divided by the number of animals observed (either age-specific or cumulative). Each model was evaluated at each age by minimizing the relative SD for all genotypes at that age and over all ages. The final model (D4) was created by refining the assumptions for model D3 in a sequential series of comparisons of survival rates for genotypes 12, 11, 6, 10, and 5 in that order.

Genotype frequency calculations

Request a detailed protocol

The models described adjust the expected observed genotype frequencies at each age to account for loss of embryos or pups due to the presumed lethal effects of one or more genotypes. The models account directly for the effect of genotype-specific mortality by reducing the number (or frequency) observed for that genotype in the sample and thus increasing the expected proportion of other genotypes. This also indirectly implies a larger theoretical litter size at conception, which can be used to determine the theoretical litter sizes had there been no mortality in the affected genotypes. The predicted proportion for each genotype is calculated at each age as the genotype Mendelian frequency multiplied by the fractional genotype survival at that age divided by the expected total fractional survival (i.e., one minus the sum of all genotype fractional losses). The formula is

For all ‘n’ possible genotypes:

Gx=MxSx/(1[(1S1)M1+(1S2)M2++(1Sn)Mn)]

where

  • Gx = model genotype fractional value (frequency) for genotype ‘x’ (Gx value from 0 to 1),

  • Mx = Mendelian fractional value (frequency) for genotype ‘x’ for the cross, and

  • Sx = fractional survival between 0 and 1.

(1 – Sn) * Mn is the fractional reduction due to survival < 100% for genotype ‘n’, for example, when Sn = 1 (e.g., 100% survival), the loss is zero; when Sn = 0 (e.g., 0% survival), the loss is Mn or the entire Mendelian fraction.

Note that the sum of all Gx for all genotypes at any age is always equal to 1 (or 100%).

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files. Mass spectrometry data were uploaded to PRIDE under Project Name: Naa10 mutant mouse N-terminome LC-MS, Project accession: PXD026410 and Project Name: mNaa10-KO liver immunoprecipitation, Project accession: PXD026684.

The following data sets were generated
    1. Ju S
    2. Lee C
    (2021) PRIDE
    ID PXD026410. Naa10 mutant mouse N-terminome LC-MS.
    1. Ree R
    2. Arnesen T
    (2021) PRIDE
    ID PXD026684. mNaa10-KO liver immunoprecipitation.

References

    1. Mclaren A
    2. Michie D
    (1958)
    Factors affecting vertebral variation in mice. 4. experimental proof of the uterine basis of a maternal effect
    Journal of Embryology and Experimental Morphology 6:645–659.

Decision letter

  1. Hening Lin
    Reviewing Editor; Cornell University, United States
  2. Philip A Cole
    Senior Editor; Harvard Medical School, United States

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Acceptance summary:

The manuscript describe the phenotype when Naa10 and Naa12 are inactivated and how Naa12 compensates Naa10 deficiency, reducing the phenotype observed when Naa10 is inactivated. The finding of a new N-acetyltransferase will be useful to the field of N-terminal acetylation.

Decision letter after peer review:

Thank you for submitting your article "Naa12 rescues embryonic lethality in Naa10-Deficient Mice in the amino-terminal acetylation pathway" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Philip Cole as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Summary:

Mutations in the X-linked gene Naa10, which encodes a major protein N-terminal acetyltransferase, are known to be causative in Ogden syndrome, a genetic disorder associated with infantile death. The manuscript reports that hemizygous knockout of Naa10 in male mice does not cause complete embryonic lethality despite the fact that it is responsible for the acetylation of about 50% of all the proteins. This leads the authors to discover a paralogous mouse gene Naa12 and demonstrate that Naa12 can compensate for Naa10 loss of function and that null mutations in both genes lead to complete embryonic lethality. This is an important finding. The authors also provided a thorough account of the variety of development abnormalities in the Naa10 hemizygous mice at all stages of development. Importantly, the authors identified several phenotypes in the mice that upon further analysis, we also noted in the some patients.

All reviewers are enthusiastic about the finding of a new N-terminal acetyltransferase and think that this is a major strength of the manuscript. In addition, the phenotypical characterization of Naa10 is also useful to the field.

Essential revisions:

Some of the revision recommendations requires additional experiments. The reviewers and the Reviewing Editor have discussed and agree that these additional experiments should only take a few weeks to complete.

1) Carry out an in vitro acetyltransferase assays with Naa12-Naa15 to validate the proposed mechanistic model.

2) Check whether Pax3 and HoxC8 proteins' abundance/stability are affected in NAA10 KO mouse tissue or MEF.

3) Obtain experimental evidence to support that N-terminal acetylation is lost in the double knockouts. If such evidence is not possible or will take too much experimental effort, please tone down the claim that the complete machinery for the process of amino-terminal acetylation of proteins in mouse development is identified.

4) Revise the text according to the comments by all three reviewers to make it easier for readers to understand.

Reviewer #1 (Recommendations for the authors):

1) It is important to show that N-terminal acetylation is lost in the double knockouts. Only with that, the authors can conclude that they have identified the "the complete machinery for the process of amino-terminal acetylation of proteins in mouse development."

2) Naa12 is new, so if not done yet, the sequence needs to be deposited into Genbank.

3) The presentation needs to be polished.

i) The title "Naa12 rescues embryonic lethality in Naa10-Deficient 1 Mice in the amino-terminal acetylation pathway" is misleading. When I saw the title, I got the impression that Naa10-dficient 1 mice show embryonic lethality. I would suggest to change it to indicate that Naa10 and Naa12 have redundant roles in embryonic development. Also, "Naa10-Deficient 1 Mice" needs to be changed to "Naa10-deficient mice."

ii) In the impact statement "Mice doubly deficient for Naa10 and Naa12 display embryonic lethality…", the word "doubly " is unnecessary.

iii) Too many acronyms, which make the reading a bit difficult. The terms NTA and Nt-acetylation could be avoided.

iv) At the end of page 9, please cite the sequence alignment in Figure S6.

v) On page 12, "Naa12 may rescue loss of Naa10 in mice" could be more assertive.

vi) Overall, I feel that the authors could polish the manuscript so that the salient points could be conveyed more easily to readers.

Reviewer #2 (Recommendations for the authors):

Overall this is a very good paper, but some issues need further attention.

– Authors present data showing that mNAA12 can acetylate some peptides at their N-end, however these reactions are performed with just the monomer and NatA activity is catalysed in vivo by the heterodimer Naa10-Naa15, showing an N-terminal acetylation in vitro activity different to Naa10 and being more similar to the observed in cells. Therefore, it is important to characterize Naa12-Naa15 in vitro enzymatic activity assessing its specificity using diverse NatA described peptidic substrates and compare it with the regular NatA enzymatic complex that presents NAA10 as catalytic subunit.

– There is a clear phenotype when mNAA10 is inactivated however, there is not a dramatic defect in protein N-terminal acetylation as it has been observed in other organisms like yeast and Arabidopsis thailiana. It has been observed by several authors that some proteins stability is affected when they are not N-terminal acetylated influencing the biological activity of these proteins. Authors comment that Pax3 and HoxC8 proteins regulate some of the biological activities deregulated in mice when mNAA10 is inactivated. As both proteins are NatA substrates and their stability could be affected when mNAA10 is inactivated it would be interesting to assess their abundance in NAA10KO embryos, mice tissue and/or MEFs in comparison with the observed in wild-type samples. These results could bring some light to the molecular mechanisms that are governing the effects observed in these mice.

Reviewer #3 (Recommendations for the authors):

Line 55 of the abstract "Male mice lacking Naa10 show no globally apparent in vivo NTA impairment and, surprisingly, do not exhibit embryonic lethality". This statement is misleading as written (and appears again in the discussion) as later in the paper, the authors describe that homozygous mice were underrepresented at birth, and describe embryonic phenotypes and embryos that underwent lysis by 10.5. Suggest rephrasing to.…do not exhibit complete embryonic lethality.

The fate of the Naa10 mice could be written more clearly. As it stands, there is a vagueness as to the number of mice that were not born, that die as early neonates and the number that survived to adulthood. Authors should consider adding a Kaplan Myer survival curve. Statements like "Some of the Naa10 hemizygous mice survived to 4 months is.." is not helpful.

Given the phenotypic effect of Naa12+/- Naa10 +/-, females, it may have been interesting to access any effects in acetylation?

https://doi.org/10.7554/eLife.65952.sa1

Author response

Essential revisions:

Some of the revision recommendations requires additional experiments. The reviewers and the Reviewing Editor have discussed and agree that these additional experiments should only take a few weeks to complete.

1) Carry out an in vitro acetyltransferase assays with Naa12-Naa15 to validate the proposed mechanistic model.

We added these analyses in the manuscript, and this does add new evidence showing that the complex has activity.

2) Check whether Pax3 and HoxC8 proteins' abundance/stability are affected in NAA10 KO mouse tissue or MEF.

We have attempted to perform the experiments, but the results were not compelling, as described below.

3) Obtain experimental evidence to support that N-terminal acetylation is lost in the double knockouts. If such evidence is not possible or will take too much experimental effort, please tone down the claim that the complete machinery for the process of amino-terminal acetylation of proteins in mouse development is identified.

Double knockouts show early embryonic lethality (Table 4), so if we want to confirm the N-terminal acetylation level of double knockout, the embryos have to be isolated before E8.5 day. We have not yet been able to obtain sufficient material for these experiments. Therefore, we toned down the wording regarding complete machinery, as recommended.

4) Revise the text according to the comments by all three reviewers to make it easier for readers to understand.

We have made these changes.

Reviewer #1 (Recommendations for the authors):

1) It is important to show that N-terminal acetylation is lost in the double knockouts. Only with that, the authors can conclude that they have identified the "the complete machinery for the process of amino-terminal acetylation of proteins in mouse development."

We have decided to remove this sentence, as the reviewer is correct that there might be even more unidentified N-terminal acetyltransferase enzymes in mice. For now, we simply present the data showing embryonic lethality for mice doubly deficient in Naa10 and Naa12. We did add a concluding sentence to the abstract, which is more toned down.

2) Naa12 is new, so if not done yet, the sequence needs to be deposited into Genbank.

When looking for homolog sequences, we did a blast-search using the sequence for Naa10. That led to the automatically annotated gene identifier on chromosome 18, which we named Naa12: Gm16286. This predicted gene translates to NM_001384178.1 (predicted gene mRNA) and NP_001371107.1 (N-α-acetyltransferase 11-like protein). This means that Naa12 is already in Genbank, however, with a different name and the status "predicted". We cannot find a way for us to re-annotate existing sequences, so we assume it will be re-annotated by the curators of Genbank after our paper is published. We modified the following sentence to read: " We found a predicted gene (Gm16286, UniProt: Q9CQX6) on chromosome 18, with high similarity to Naa10, which we name Naa12, and RiboSeq and mRNA traces of this region suggest possible transcription and translation of this gene.”

3) The presentation needs to be polished.

We have edited throughout with this in mind.

i) The title "Naa12 rescues embryonic lethality in Naa10-Deficient 1 Mice in the amino-terminal acetylation pathway" is misleading. When I saw the title, I got the impression that Naa10-dficient 1 mice show embryonic lethality. I would suggest to change it to indicate that Naa10 and Naa12 have redundant roles in embryonic development. Also, "Naa10-Deficient 1 Mice" needs to be changed to "Naa10-deficient mice."

We have changed the title to: Naa12 compensates for Naa10 in mice in the amino-terminal acetylation pathway.

ii) In the impact statement "Mice doubly deficient for Naa10 and Naa12 display embryonic lethality…", the word "doubly " is unnecessary.

Removed.

iii) Too many acronyms, which make the reading a bit difficult. The terms NTA and Nt-acetylation could be avoided.

We replaced NTA and Nt-acetylation with amino-terminal acetylation throughout the manuscript. We have also tried throughout to expand on acronyms.

iv) At the end of page 9, please cite the sequence alignment in Figure S6.

We inserted this.

v) On page 12, "Naa12 may rescue loss of Naa10 in mice" could be more assertive.

We changed this to "Naa12 rescues loss of Naa10 in mice".

vi) Overall, I feel that the authors could polish the manuscript so that the salient points could be conveyed more easily to readers.

We have edited the manuscript throughout, attempting to clarify reading.

Reviewer #2 (Recommendations for the authors):

Overall this is a very good paper, but some issues need further attention.

– Authors present data showing that mNAA12 can acetylate some peptides at their N-end, however these reactions are performed with just the monomer and NatA activity is catalysed in vivo by the heterodimer Naa10-Naa15, showing an N-terminal acetylation in vitro activity different to Naa10 and being more similar to the observed in cells. Therefore, it is important to characterize Naa12-Naa15 in vitro enzymatic activity assessing its specificity using diverse NatA described peptidic substrates and compare it with the regular NatA enzymatic complex that presents NAA10 as catalytic subunit.

We have attempted to prepare recombinant Naa12/Naa15 complex from Sf9 insect cells for biochemical studies, but were unfortunately unable to prepare sufficient levels of complex for detailed enzymatic analysis. Nonetheless, we were able to prepare a sufficient amount of recombinant complex to analyze fractions from a final size exclusion chromatograph purification. This analysis demonstrated that the Naa12/Naa15 complex harbored detectable amino-terminal acetylation activity toward the SESSSKS- peptide (Figure 4C), thus demonstrating catalytic activity of a NatA complex with mouse Naa12.

– There is a clear phenotype when mNAA10 is inactivated however, there is not a dramatic defect in protein N-terminal acetylation as it has been observed in other organisms like yeast and Arabidopsis thailiana. It has been observed by several authors that some proteins stability is affected when they are not N-terminal acetylated influencing the biological activity of these proteins. Authors comment that Pax3 and HoxC8 proteins regulate some of the biological activities deregulated in mice when mNAA10 is inactivated. As both proteins are NatA substrates and their stability could be affected when mNAA10 is inactivated it would be interesting to assess their abundance in NAA10KO embryos, mice tissue and/or MEFs in comparison with the observed in wild-type samples. These results could bring some light to the molecular mechanisms that are governing the effects observed in these mice.

We included the references to Pax3 and Hoxc8 as possible future directions, and it would be an entire new paper to demonstrate this mechanism. Nonetheless, to attempt to address this for the reviewer, we have performed Western blotting for Pax3 and Hoxc8. It is not easy to detect Pax3 level with western blot, so B16F10 (mouse melanoma cell line) stably expressing pax3 was used as a control. We have detected a slight decline only in NAA10KO E10.5 embryos, but this is very small and likely not significant, and this is only semi-quantitative at best with Western blotting, and so we are not prepared to publish this result until we have confirmed it with further experiments, including mass spectrometry. For Hoxc8, signal detection was not successful using one antibody that we purchased, and we will have to try again for a future paper. Of course, this does not specifically address the N-terminal acetylation of either protein, but N-terminal peptides from Pax3 and Hoxc8 are not present in our mass spectrometry data (included here and also unpublished data). As such, we cannot determine if there are differences in their N-terminal acetylation levels, and we are beginning now various approaches to try to determine this.

Reviewer #3 (Recommendations for the authors):

Line 55 of the abstract "Male mice lacking Naa10 show no globally apparent in vivo NTA impairment and, surprisingly, do not exhibit embryonic lethality". This statement is misleading as written (and appears again in the discussion) as later in the paper, the authors describe that homozygous mice were underrepresented at birth, and describe embryonic phenotypes and embryos that underwent lysis by 10.5. Suggest rephrasing to.…do not exhibit complete embryonic lethality.

We modified this to read: "Male mice lacking Naa10 show no globally apparent in vivo amino-terminal acetylation impairment and do not exhibit complete embryonic lethality."

The fate of the Naa10 mice could be written more clearly. As it stands, there is a vagueness as to the number of mice that were not born, that die as early neonates and the number that survived to adulthood. Authors should consider adding a Kaplan Myer survival curve.

We have added such a curve, and have referred to it in the Results.

Statements like "Some of the Naa10 hemizygous mice survived to 4 months is.." is not helpful.

We have quantified this in the Results, with statements such as: “Out of 32 Naa10-/Y that survived past the third day of life and which were then examined longitudinally, about 60% survived past 200 days of life (~7 months)”.

Given the phenotypic effect of Naa12+/- Naa10 +/-, females, it may have been interesting to access any effects in acetylation?

This is not an easy experiment to perform, as we have spent several years attempting to demonstrate decreased acetylation in KO males, with no success, leading us to the characterization of Naa12. The compound heterozygous Naa12+/- Naa10 +/- females are born in small numbers, and we have now quantified this in detail in a new section titled “Genotype distribution modeling of Naa10- and Naa12-deficient offspring”. It is outside the scope and timeframe for the current paper to attempt to quantify amino-terminal acetylation levels in the compound heterozygous females, and we are likely going to focus instead on dying (or just before dying) doubly deficient embryos, where the degree of acetylation difference will likely be greater.

https://doi.org/10.7554/eLife.65952.sa2

Article and author information

Author details

  1. Hyae Yon Kweon

    Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    Contribution
    Formal analysis, Investigation, Writing - original draft, Writing - review and editing
    Contributed equally with
    Mi-Ni Lee
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1534-0767
  2. Mi-Ni Lee

    1. Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    2. Laboratory Animal Resource Center Korea ResearchInstitute of Bioscience and Biotechnology, Chungbuk, Republic of Korea
    Present address
    Laboratory Animal Resource Center, Korea Research Institute of Bioscience and Biotechnology, Chungbuk, Republic of Korea
    Contribution
    Supervision, Investigation, Project administration, Writing - review and editing
    Contributed equally with
    Hyae Yon Kweon
    For correspondence
    minilee@kribb.re.kr
    Competing interests
    No competing interests declared
  3. Max Dorfel

    Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, Woodbury, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Seungwoon Seo

    Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Leah Gottlieb

    1. Department of Chemistry, University of Pennsylvania, Philadelphia, United States
    2. Abramson Family Cancer Research Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
    Contribution
    Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Thomas PaPazyan

    Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, Woodbury, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  7. Nina McTiernan

    Department of Biomedicine, University of Bergen, Bergen, Norway
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1749-6933
  8. Rasmus Ree

    Department of Biomedicine, University of Bergen, Bergen, Norway
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  9. David Bolton

    Department of Molecular Biology, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, United States
    Contribution
    Formal analysis, Writing - review and editing
    Competing interests
    No competing interests declared
  10. Andrew Garcia

    Department of Human Genetics, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  11. Michael Flory

    Research Design and Analysis Service, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  12. Jonathan Crain

    Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, Woodbury, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  13. Alison Sebold

    Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, Woodbury, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  14. Scott Lyons

    Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, Woodbury, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  15. Ahmed Ismail

    Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, Woodbury, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  16. Elaine Marchi

    Department of Human Genetics, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  17. Seong-keun Sonn

    Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  18. Se-Jin Jeong

    Center for Cardiovascular Research, Washington University School of Medicine, Saint Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6375-5334
  19. Sejin Jeon

    Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3819-3421
  20. Shinyeong Ju

    Center for Theragnosis, Korea Institute of Science and Technology, Seoul, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5483-4690
  21. Simon J Conway

    Herman B. Wells Center for Pediatric Research, Indiana University School of Medicine, Indianapolis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  22. Taesoo Kim

    Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3902-1058
  23. Hyun-Seok Kim

    Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  24. Cheolju Lee

    1. Center for Theragnosis, Korea Institute of Science and Technology, Seoul, Republic of Korea
    2. Department of Converging Science and Technology, KHU-KIST, Kyung Hee University, Seoul, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8482-4696
  25. Tae-Young Roh

    Department of Life Sciences, Pohang University of Science and Technology, Pohang, Republic of Korea
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5833-0844
  26. Thomas Arnesen

    1. Department of Biomedicine, University of Bergen, Bergen, Norway
    2. Department of Biological Sciences, University of Bergen, Bergen, Norway
    3. Department of Surgery, Haukeland University Hospital, Bergen, Norway
    Contribution
    Supervision, Writing - review and editing
    Competing interests
    No competing interests declared
  27. Ronen Marmorstein

    1. Department of Chemistry, University of Pennsylvania, Philadelphia, United States
    2. Abramson Family Cancer Research Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
    3. Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
    Contribution
    Supervision, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4373-4752
  28. Goo Taeg Oh

    Department of Life Science and College of Natural Sciences, Ewha Womans University, Seoul, Republic of Korea
    Contribution
    Funding acquisition, Project administration, Writing - review and editing
    For correspondence
    gootaeg@ewha.ac.kr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1104-1698
  29. Gholson J Lyon

    1. Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory, Woodbury, United States
    2. Department of Human Genetics, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, United States
    3. Biology PhD Program, The Graduate Center, The City University of New York, New York, United States
    4. George A. Jervis Clinic, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, United States
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    gholsonjlyon@gmail.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5869-0716

Funding

National Research Foundation of Korea (2020R1A3B2079811)

  • Goo Taeg Oh

National Research Foundation of Korea (2017RIDIAB03032286)

  • Mi-Ni Lee

National Research Foundation of Korea (2020RICIC1007686)

  • Mi-Ni Lee

National Institute of General Medical Sciences (R35GM133408)

  • Gholson J Lyon

National Institute of General Medical Sciences (R35GM118090)

  • Ronen Marmorstein

Research Council of Norway (Project 249843)

  • Thomas Arnesen

National Institutes of Health (Project 249843)

  • Simon J Conway

Norwegian Health Authorities of Western Norway (912176)

  • Thomas Arnesen

Norwegian Health Authorities of Western Norway (F-12540-D11382)

  • Thomas Arnesen

Norwegian Cancer Society (PR-2009-0222)

  • Thomas Arnesen

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

GTO thanks Huiju Jo for supporting mouse genotyping. GJL would like to thank Ryan Driscoll for mouse genotyping, Fatima Inusa for embryo dissection, Jemima Kadima for mouse husbandry, Leyi Li for transgenic mouse assistance, Melissa Nashat, Jodi Coblentz, Rachel Rubino, and Lisa Bianco for veterinary advice, and Vicky Brandt for editorial assistance.

This work was supported by National Research Foundation of Korea (NRF) grants funded by the Korean Government (2020R1A3B2079811 (GO), 2017RIDIAB03032286 (ML), and 2020RICIC1007686 (ML)). Research reported in this publication was also supported by the National Institute of General Medical Sciences of the National Institutes of Health (NIH) under Award Numbers R35GM133408 (GJL) and R35GM118090 (RM), and NIH grant HL148165 (SJC). The work was also supported by the Research Council of Norway (Project 249843), the Norwegian Health Authorities of Western Norway (projects 912176 and F-12540-D11382), and the Norwegian Cancer Society (PR-2009-0222). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Funding was also provided by the Stanley Institute for Cognitive Genomics at Cold Spring Harbor Laboratory, the George A Jervis Clinic and the Department of Human Genetics, Laboratory of Genomic Medicine at the New York State Institute for Basic Research in Developmental Disabilities (IBR), New York State Office for People with Developmental Disabilities. Part of the work was carried out at the Proteomics Unit at University of Bergen (PROBE).

Ethics

Animal experimentation: All experiments were performed in accordance with guidelines of International Animal Care and Use Committee (IACUC) of Ewha Womans University (protocol #18-012), Cold Spring Harbor L aboratory (CSHL) protocol # 579961 18, and Institute for Basic Research in Developmental Disabilities (IBR) (protocol #456).

Senior Editor

  1. Philip A Cole, Harvard Medical School, United States

Reviewing Editor

  1. Hening Lin, Cornell University, United States

Publication history

  1. Preprint posted: December 20, 2020 (view preprint)
  2. Received: December 20, 2020
  3. Accepted: August 5, 2021
  4. Accepted Manuscript published: August 6, 2021 (version 1)
  5. Accepted Manuscript updated: August 11, 2021 (version 2)
  6. Version of Record published: August 19, 2021 (version 3)

Copyright

© 2021, Kweon et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 380
    Page views
  • 47
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Alessandro Bonfini et al.
    Research Article

    The gut is the primary interface between an animal and food, but how it adapts to qualitative dietary variation is poorly defined. We find that the Drosophila midgut plastically resizes following changes in dietary composition. A panel of nutrients collectively promote gut growth, which sugar opposes. Diet influences absolute and relative levels of enterocyte loss and stem cell proliferation, which together determine cell numbers. Diet also influences enterocyte size. A high sugar diet inhibits translation and uncouples ISC proliferation from expression of niche-derived signals but, surprisingly, rescuing these effects genetically was not sufficient to modify diet's impact on midgut size. However, when stem cell proliferation was deficient, diet's impact on enterocyte size was enhanced, and reducing enterocyte-autonomous TOR signaling was sufficient to attenuate diet-dependent midgut resizing. These data clarify the complex relationships between nutrition, epithelial dynamics, and cell size, and reveal a new mode of plastic, diet-dependent organ resizing.

    1. Developmental Biology
    2. Physics of Living Systems
    Yonghyun Song, Changbong Hyeon
    Research Article Updated

    Spatial boundaries formed during animal development originate from the pre-patterning of tissues by signaling molecules, called morphogens. The accuracy of boundary location is limited by the fluctuations of morphogen concentration that thresholds the expression level of target gene. Producing more morphogen molecules, which gives rise to smaller relative fluctuations, would better serve to shape more precise target boundaries; however, it incurs more thermodynamic cost. In the classical diffusion-depletion model of morphogen profile formation, the morphogen molecules synthesized from a local source display an exponentially decaying concentration profile with a characteristic length λ. Our theory suggests that in order to attain a precise profile with the minimal cost, λ should be roughly half the distance to the target boundary position from the source. Remarkably, we find that the profiles of morphogens that pattern the Drosophila embryo and wing imaginal disk are formed with nearly optimal λ. Our finding underscores the cost-effectiveness of precise morphogen profile formation in Drosophila development.