ZIP8 Zinc Transporter: Indispensable Role for Both Multiple-Organ Organogenesis and Hematopoiesis In Utero

Previously this laboratory characterized Slc39a8-encoded ZIP8 as a Zn2+/(HCO3 –)2 symporter; yet, the overall physiological importance of ZIP8 at the whole-organism level remains unclear. Herein we describe the phenotype of the hypomorphic Slc39a8(neo/neo) mouse which has retained the neomycin-resistance gene in intron 3, hence causing significantly decreased ZIP8 mRNA and protein levels in embryo, fetus, placenta, yolk sac, and several tissues of neonates. The Slc39a8(neo) allele is associated with diminished zinc and iron uptake in mouse fetal fibroblast and liver-derived cultures; consequently, Slc39a8(neo/neo) newborns exhibit diminished zinc and iron levels in several tissues. Slc39a8(neo/neo) homozygotes from gestational day(GD)-11.5 onward are pale, growth-stunted, and die between GD18.5 and 48 h postnatally. Defects include: severely hypoplastic spleen; hypoplasia of liver, kidney, lung, and lower limbs. Histologically, Slc39a8(neo/neo) neonates show decreased numbers of hematopoietic islands in yolk sac and liver. Low hemoglobin, hematocrit, red cell count, serum iron, and total iron-binding capacity confirmed severe anemia. Flow cytometry of fetal liver cells revealed the erythroid series strikingly affected in the hypomorph. Zinc-dependent 5-aminolevulinic acid dehydratase, required for heme synthesis, was not different between Slc39a8(+/+) and Slc39a8(neo/neo) offspring. To demonstrate further that the mouse phenotype is due to ZIP8 deficiency, we bred Slc39a8(+/neo) with BAC-transgenic BTZIP8-3 line (carrying three extra copies of the Slc39a8 allele); this cross generated viable Slc39a8(neo/neo)_BTZIP8-3(+/+) pups showing none of the above-mentioned congenital defects–proving Slc39a8(neo/neo) causes the described phenotype. Our study demonstrates that ZIP8-mediated zinc transport plays an unappreciated critical role during in utero and neonatal growth, organ morphogenesis, and hematopoiesis.

The mammalian SLC39 family contains 14 members of the zinc-and iron-related protein (ZIP) family, which transport essential-metal divalent cations such as zinc, iron, copper and manganese [1;2]; these 14 genes are very highly conserved between human and rodent. The ZIP transporter proteins appear to serve crucial roles in metal homeostasis and perhaps unappre-ciated important signaling pathways; for example, mutations in the human SLC39A4 gene are responsible for acrodermatitis enteropathica, zinc-deficiency (AEZ) [3][4][5]. Mutations in the human SLC39A13 gene result in a spondylocheiro dysplastic form of Ehlers-Danlos syndrome showing generalized skeletal dysplasia [6]. Endogenous functions and importance of the other twelve ZIP transporters are not yet fully understood.
Starting with genetically ''sensitive'' vs ''resistant'' inbred strains of mice and a phenotype of cadmium (Cd)-induced testicular necrosis, our lab launched a forward-genetics study to identify and characterize the major locus (Cdm) responsible for this trait [7]. By means of positional cloning, we identified Slc39a8, which encodes the metal transporter ZIP8 as the likely candidate for the Cdm locus; ZIP8 mRNA expression is robust in endothelial cells of the testicular vasculature in two Cd-sensitive mouse lines and negligible in these cells from two Cd-resistant mouse lines [8]. That the Slc39a8 gene is indeed the Cdm locus was unequivocally proven-using a bacterial artificial chromosome (BAC)-transgenic mouse line. A 168.7-kb BAC, containing only the Slc39a8 gene (plus 87 kb of 59-flanking and 18 kb of 39-flanking sequences) from a 129S6/SvEvTac (Cd-sensitive) BAC library, was inserted into the Cd-resistant C57BL/6J genome; Cd treatment produced testicular necrosis in BAC-transgenic BTZIP8-3 mice but not in non-transgenic littermates [9].
Further studies have shown that ZIP8 functions endogenously as a Zn 2+ /(HCO 3 -) 2 as well as Mn 2+ /(HCO 3 -) 2 symporter-moving both ions into the cell in an electroneutral manner [10][11][12]. Moreover, Cd 2+ is able to displace Zn 2+ , thereby entering cells that carry the functional ZIP8 transporter. A common theme in pharmacology and toxicology is that drugs and other environmental agents usurp the same receptors and transporters as those used by endogenous substrates; likewise, drugs and other environmental agents are metabolized by enzymes that normally handle endogenous substrates [13].
The next step in understanding ZIP8's endogenous functions was to generate a knockout mouse line. In building the Slc39a8 knockout construct [14], we inserted an Frt-flanked neomycinresistance mini-gene (neo) into intron 3, combined with loxP sites in introns 3 and 6. Retention of the neo gene in intron 3 leads to an intriguing hypomorph. Its phenotype is reported herein-a combination of stunted growth, severe anemia, dysregulation of hematopoiesis, multiple organs failing to develop normally in utero, and neonatal lethality.
Slc39a8(neo/neo) newborns showed malformed craniums, hypoplastic hind limbs (forelimbs less affected), and underdeveloped eyes (Fig. 1B). The spleen was virtually absent. There was substantial hypoplasia of the kidneys and liver-and lungs to a lesser extent ( Table 2). Failure of organ development was ranked as spleen most severely affected . kidney = liver . lungs smaller than normal. The Slc39a8(neo/neo) organ sizes (Fig. S1), as well as placenta, yolk sac, and total embryo (Fig. S2) were smaller than those of the wild-type or heterozygote. On the other hand, the size of Slc39a8(neo/neo) thymus, cerebrum and cerebellum was no different than wild-type ( Table 2). The Slc39a8(neo/neo) heart appeared slightly larger than wild-type or heterozygotes; this increased size might reflect the severe anemia that we found in the hypomorph. Thus, in the Slc39a8(neo/neo), we conclude that multiple organs fail to develop properly, from very early embryogenesis; in utero normal growth rate is also affected.

Histology of Tissues among the Three Genotypes
Fetal yolk sac and liver showed histological differences in hematopoietic islands, as described later. GD18.5 and PND1 lung, kidney, heart, and other tissues listed in Table 2 and Fig. S2 showed no obvious histological differences among the three genotypes (data not illustrated), other than organ size.
In PND1 tissues (Fig. 2, bottom), ZIP8 mRNA was strikingly decreased in all tissues of Slc39a8(neo/neo) that were examined. The mRNA differences (Fig. 2) were confirmed at the protein level by Western immunoblots in whole embryo/fetus, yolk sac, liver, lung, kidney and heart (Fig. 3). ZIP8 protein levels appeared to follow mRNA levels quite closely in all samples; densitometric readings ( Fig. 3C) further confirmed the Western blot data.
Note that sometimes ZIP8 mRNA and protein levels were similar in the Slc39a8(+/+) and Slc39a8(+/neo) samples, while being much less in the Slc39a8(neo/neo); in other instances, ZIP8 mRNA and protein were similarly diminished in the Slc39a8(+/ neo) and Slc39a8(neo/neo) samples, while being much higher in the Slc39a8(+/+) wild-type (Figs. 2 & 3). Most likely, this phenomenon represents transvection in the heterozygote-an epigenetic phenomenon of activation or repression that results from interaction between an allele (usually recessive) on one chromosome and the corresponding allele on the homologous chromosome [15][16][17][18][19]. This can happen specifically to one cell type, and can vary even as a function of age. Although transvection is somewhat rare, it has also been demonstrated in Slc39a4(+/-) mice [20].
It is intriguing that low ZIP8 mRNA and protein in an organ and the effect of low ZIP8 on organogenesis do not go hand-inhand. In other words, low ZIP activity and small organ size were seen in liver, lung and kidney; low ZIP expression yet normal organ size were observed in cerebrum, cerebellum, thymus and heart ( Table 2 & Figs. 2 & 3). The severe pale phenotype was seen in Slc39a8(neo/neo) mice but did occur occasionally in Slc39a8(+/neo) mice. This observation might reflect the relative importance of endogenous ZIP8 function in different organs. The various compensatory mechanisms (in scenarios of ZIP8 loss) in different organs might also affect the final outcome of organ development.
Previous work from this lab [9] showed ZIP8 mRNA levels on Northern blots to be higher in adult mouse lung and kidney than in liver or heart. In contrast, we did not find such higher ZIP8 mRNA or protein levels in lung and kidney of newborn mice (Fig. 2), which may suggest distinct roles of ZIP8 in developing tissues and adult tissues.
As well as the band at the expected molecular weight for ZIP8 protein on Western immunoblots (50.082 kDa), additional heavier bands were ofttimes seen ( Fig. 3A & B), depending on the tissue-most likely representing products of posttranslationally-modified ZIP8 protein, e.g. one or several events of glycosylation [10]. However, the intensity of these additional bands always appeared to correlate across genotypes with the intensity of the 50-kDa band (representing the non-glycosylated protein). To reduce clutter, we have omitted most of the higher-MW bands in Fig. 3. The highest number of observed bands (six), showing a direct correlation of the band intensities with genotype, is best illustrated in GD13.5 embryo (Fig. 3A, lower right). Ample evidence exists for multiple glycosylations of at least four ZIP proteins: ZIP4 [10;21], ZIP5 [21;22], ZIP8 [10], and ZIP14 [23].

ZIP8 mRNA Length between the Two Alleles
Are there splicing differences in ZIP8 mRNA between the two alleles? We developed a PCR-based strategy to examine the entire Slc39a8 transcript; we devised appropriate primers (Table S2) in order to divide the cDNA (corresponding to the mRNA) into 22 overlapping PCR fragments (Fig. S3). No differences in PCR fragment size were seen among the Slc39a8(+/+), Slc39a8(+/neo) and Slc39a8(neo/neo) genotypes. We conclude that the ZIP8 mRNA length stays the same in the two alleles and among three genotypes.

ZIP8 mRNA Levels in Mouse Fetal Fibroblast (MFF) and Fetal Liver Cultures
Comparing MFF cultures (Fig. 4, left) and fetal liver cultures (Fig. 4, right), ZIP8 mRNA was significantly diminished in Slc39a8(neo/neo) cells. These data demonstrate that MFF and fetal liver-derived cell cultures represent a faithful in vitro system for studying ZIP8 function.
These observations are consistent with the pale color seen in Slc39a8(neo/neo) homozygotes but not Slc39a8(+/neo) heterozygotes or Slc39a8(+/+) wild-type (Fig. 1B). Compared with that in Slc39a8(+/+) wild-type-the red blood cell count, hemoglobin levels and hematocrit were all statistically significantly decreased in Slc39a8(neo/neo) PND1 pups (Fig. 5C). The volume density of hematopoietic cells and islands (expressed as percentage of total cell number) in GD16.5 liver was more than 3-fold lower in Slc39a8(neo/neo) than Slc39a8(+/+) pups (Fig. 5C), another observation consistent with the anemia. These data strongly implicate an association of the strikingly decreased ZIP8 mRNA and protein levels with a severe anemia in utero and in neonates.
Comparing Slc39a8(+/+) and Slc39a8(+/neo) with Slc39a8(neo/ neo) fetuses ( Table 3 & Fig. S5), we found that blood smears of the hypomorph showed: more erythroblasts; greater numbers of nucleated red cells (including binucleated and cells with micronuclei); hyper-condensation of the nuclei in erythroid precursors; mature neutrophils; and a suggestive premature maturation of both erythroid and myeloid precursors. Counting 1,000 cells for each animal, we found nucleated red cells + erythroblasts in whole blood averaged far fewer in Slc39a8(+)-containing animals than in Slc39a8(neo/neo) homozygotes; this dramatic increase in Slc39a8(neo/neo) nucleated red cells and erythroblasts persisted from GD16.5 through PND1 ( Table 3). The percent erythroid-toleukocyte ratio in Slc39a8(neo/neo) was correspondingly decreased. The number of megakaryocytes was similar in fetal and newborn liver of all three genotypes. These data indicate that smears of peripheral blood of the Slc39a8(neo/neo) hypomorph are abnormal, consistent with anemia.
No striking differences among the three genotypes were found for the myeloid and lymphoid lineages. The CD3e-positive population was not gated, when plotted against CD11b *P-values were calculated using one-way ANOVA followed by the post-hoc least-significant-differences (LSD) test; numbers of organs ranged from 5 to 40 per group. Of all the organs surveyed, no differences were found between Slc39a8(+/+) pups and Slc39a8(+/neo) pups. In Slc39a8(neo/neo) pups, the weights of liver, lung, kidney and spleen were significantly lower than those from Slc39a8(+/+) and Slc39a8(+/neo) pups. doi:10.1371/journal.pone.0036055.t002  6B); instead, the CD3e-positive population was gated when plotting CD3e against forward scatter to ensure that the cell size was consistent with lymphocytes. Comparing T cells by plotting them vs forward scatter or vs CD11b, the difference was insignificant. The CD11b-positive gate (Fig. 6B) shows a small, but significant, increase in the proportion of CD11b-positive cells among the Slc39a8(neo/neo) samples, suggesting that ZIP8 deficiency might affect myelogenesis in some manner. There could be an alternative explanation for the observed increase in CD11b + cells-such as a mild inflammatory process in the liver. However, as noted elsewhere in this text, histology of liver showed no inflammation; moreover, serum ALT and AST levels ( Table  S3) were not elevated in Slc39a8(neo/neo) pups. However, CD3epositive cells (representing T cells) were not different among the three genotypes. We conclude that erythroid cells from Slc39a8(neo/neo) liver ( Fig. 6A) are blocked in an early maturity stage (shown by the higher percentage of more immature cells in Region 2, plus the lower percentage of cells that are at more advanced differentiated stage in Region 5). These results match our observations in the blood smears ( Table 3 & Fig. S5). This erythroid developmental blockade is likely the reason for in utero anemia in the Slc39a8(neo/ neo) hypomorph.

Blood Chemistry Analyses
Consistent with the severe anemia and hematopoietic dysregulation described above, plasma iron levels and total iron-binding capacity (TIBC) were more than 40-fold decreased in Slc39a8(neo/ neo), compared with Slc39a8(+/neo) and Slc39a8(+/+) newborns (Table S3). To the contrary, total bilirubin (sum of conjugated plus non-conjugated) was very low and not different among the three genotypes-indicating an absence of liver pathology. In neonatal mice, serum transaminase levels are known to be highly variable, with ALT levels ranging between 100 and 600 and AST levels between 1,000 and 5,000 Units/L [27]. Serum ALT and AST levels ( Table S3) were not elevated in the Slc39a8(neo/neo), consistent with no evidence of hepatocellular or extrahepatic (especially muscle) cellular damage.

Divalent Cation Uptake in MFF and Fetal Liver-Derived
Cultures. Studying MFF cultures derived from the three genotypes, we found the Slc39a8(neo) allele to be associated with decreased zinc (Fig. 7A), cadmium (Fig. 7B), and iron (Fig. 7C) uptake. Fetal liver cells (which include hepatocytes and hematopoietic progenitor cells, as well as interstitial cells) were cultured separately from MFFs; again, we found the Slc39a8(neo) allele was associated with decreased iron uptake (Fig. 7D). As were noted for some samples with ZIP8 mRNA (Fig. 2) and protein (Fig. 3) levels, the divalent cation uptake in the Slc39a8(+/neo) heterozygote exhibited a gene-dose behavior in zinc-exposed MFFs (Fig. 7A) and was not statistically different from the Slc39a8(neo/ neo) in the other three panels. These discrepancies might reflect the phenomenon of transvection.
Divalent Metal Content in PND1 Tissues. ICP-MS analysis (Fig. 8A) revealed that presence of the Slc39a8(neo) allele was statistically significantly associated with decreased zinc content in liver and heart, but not lung or kidney. This lack of difference for zinc in kidney is not surprising, however, because-like many other transporters in the renal proximal and distal tubules-it may take several weeks following parturition for ZIP8 to be fully expressed and fully functioning in the kidney [28].
The Slc39a8(neo) allele was also associated with lower iron content in liver and lung but not kidney or heart (Fig. 8B). Again, the Slc39a8(+/neo) phenotype resembled the Slc39a8(+/+) wildtype in some tissues, whereas the Slc39a8(+/neo) phenotype resembled more closely the Slc39a8(neo/neo) genotype in others.
We also carried out ICP-MS analysis for copper and manganese content in these same tissues, and no significant differences among the three genotypes were found: for copper, values ranged between 15.5 and 29 ng/mg tissue for liver, 2.9 and 4.2 for lung, 4.3 and 6.2 for kidney, and 4.9 and 6.1 for heart. In addition, we measured manganese content, which was undetectable (,1.0 ng/mg) in liver, lung, kidney and heart (not shown).

Assay for 5-Aminolevulinic Acid Dehydratase (ALAD) Activity
Altered zinc transport in utero, leading to the phenotype described in the Slc39a8(neo/neo) hypomorph, is likely to be caused by inactivation of one or more critical zinc-dependent enzymes and/or transcription factors. One outstanding candidate might be ALAD, also called porphobilinogen synthase (PBGS), an enzyme requiring zinc as a cofactor; in a state of severe zinc deficiency, ALAD could thus become the rate-limiting step, resulting in a lack of hemoglobin synthesis (Fig. S7). . Western immunoblot analysis of ZIP8 protein (50.082 kDa) in the same tissues described in Fig. 2. (A) Whole embryo/fetus and yolk sac; (B) Four tissues at PND1. Antibody to b-actin protein (ACTB; 41.7 kDa) was used as the lane-loading control. To maximize contrast, exposure times ranged from 1 min to 6 h. (C) Densitometric semi-quantification of Western immunoblots: whole embryo/fetus, yolk sac and four tissues at PND1. The Slc39a8(+/+) wild-type ZIP8 protein/ACTB protein ratio (left-most in each panel) and protein expression levels were expressed as linear fold-changes-relative to the normalized wild-type control. doi:10.1371/journal.pone.0036055.g003 Figure 4. Relative ZIP8 mRNA levels in untreated Slc39a8(+/+), Slc39a8(+/neo) and Slc39a8(neo/neo) MFF and fetal liver cultures. As in Fig. 2, the Slc39a8(+/+) wild-type ZIP8 mRNA/GAPDH mRNA ratio (left-most in each panel) and mRNA expression levels were expressed as linear fold-changes-relative to the normalized wild-type control. The MFFs were prepared from GD14.5 fetuses whose livers had been removed. Livers were trypsinized and cultured until they approached ,80% confluency (,2 weeks). *P,0.0001, compared with (+/+) or (+/ neo) genotypes. These data represent duplicate results from three independent experiments. doi:10.1371/journal.pone.0036055.g004

ZIP8-Mediated Hematopoiesis Essential In Utero
PLoS ONE | www.plosone.org We analyzed ALAD activity in pooled PND1 livers from the three genotypes. For each set of livers, native activity (nonactivated assay) was compared with zinc-activated activity; this can be expressed as the Relative Index (R.I.) of ALAD activity: ALADspecificactivity(zinc-activatedassay) = ALADspecificactivity(non-activatedassay) .
The R.I. should be a sensitive indicator of differences in enzyme activity due to low levels of zinc in liver: the larger the R.I., the greater the difference in ALAD activity between the zinc-activated and non-activated systems. Therefore, the larger the R.I., the greater the effect of zinc deficiency.
Two sets of pooled livers per genotype were incubated (Fig. 9); a maximal effect of zinc enhancement of ALAD was found at 15 min, compared with an incubation time of 30 min (data not shown) when enzyme kinetics approached saturation levels, and zinc activation was noticeably less. We tested differences in variously added zinc concentrations between 8 and 400 mM ZnCl 2 , and we chose to add 18 mM zinc (which is the physiologically normal plasma zinc level) for the zinc-activated assay. We found the R.I. to be greatest around 18 mM added zinc, with decreases below native activity when zinc concentrations exceeded 50-60 mM.
These data indicate that the zinc-dependent ALAD appears not to represent the underlying cause of the Slc39a8(neo/neo) dysregulation in hematopoiesis.

Attempt to Rescue Slc39a8(neo/neo) Homozygotes via Breeding with BTZIP8-3 Mice
Finally, we wished to confirm that the observed phenotype (anemia and newborn lethality) was indeed due to the neo-cassettemediated down-regulation of Slc39a8 gene expression during mouse embryo development. This lab had previously created the BAC-transgenic BTZIP8-3 mouse line, which carries three extra copies of the 129S6/SvEvTac Slc39a8 gene in addition to its normal diploid number of C57BL/6J Slc39a8 genes. The transgenic BAC allele (168,722 bp) appears to have all the necessary regulatory elements to ensure proper tissue and temporal expression of the Slc39a8 gene [9]. We postulated that addition of three Slc39a8 BAC transgenes (located elsewhere in the genome) might be able to rescue the observed phenotype during in utero and early postnatal development.
There is a small fraction of events that do not fall into any of the gates R2 through R5, which do, however, meet the conditions for classification as TER119 + . doi:10.1371/journal.pone.0036055.t005 multiple-organ failures in development, and in utero and neonatal lethality is indeed due to lowered ZIP8 caused by the Slc39a8(neo/ neo) genotype.

Discussion
Zinc plays a critical role in intracellular signal transduction [29], cell cycle and proliferation [30], processes involving development and differentiation [31], and maintaining normal function of numerous transcription factors [32]. In the human or rodent, there are almost 100 zinc-dependent enzymes [33] and more than 2,000 zinc-dependent transcription factors [34]. Because these enzymes and factors carry out important functions throughout development-often exerting cell-specific effects on morphogenesis, growth, and differentiation, the embryo's ability to control zinc homeostasis becomes essential from the blastocyst stage onward [35]. In the present study we have described an intriguing phenotype in the Slc39a8(neo/neo) mouse that provides valuable insight into the importance of ZIP8-mediated zinc uptake in utero. Among many and possibly redundant zinc transporters in the adult animal, ZIP8 appears to be indispensible for embryo development; a deficient ZIP8 transporter results in failure of multiple-organ development and dysregulated hematopoiesis in utero.
The critical role of ZIP8 in hematopoiesis is the most unique aspect of the present study. There is an increase in ZIP8 expression in whole embryo between GD11.5 to GD16.5, whereas ZIP8 expression in yolk sac decreases between GD11.5 and GD13.5 (Figs. 2 & 3); this coincides with the time during which early-embryo hematopoiesis is known to transition (Fig. S4) from yolk sac to fetal liver [24]. Therefore, these data are consistent with an hypothesis that ZIP8 is functioning during hematopoiesis-first in yolk sac, and later as hematopoiesis in fetal liver. There is also evidence suggesting that the placenta is important in hematopoiesis [26]. Considering the pronounced pale appearance of Slc39a8(neo/neo) embryos or fetuses, the anemia parameters that were measured in GD16.5 fetuses and PND1 neonates (Fig. 5) might in fact be an under-extrapolation of the degree of anemia in the developing embryo/fetus. Iron deficiency following ZIP8 loss could be a major reason for the anemia phenotype. Fe 2+ is essential for heme synthesis and erythropoiesis. The current understanding is that cells take up iron via the Fe 3+ -transferrin complex: the Fe 3+ -transferrin complex can bind to specific receptors present on the cell surface, which then induce endocytosis [36]. To date, there is no divalent (Fe 2+ ) transporter identified that is indispensable for iron uptake in embryonic and fetal hematopoietic organsincluding yolk sac and fetal liver. Our study suggests the existence of such a pathway.
The pronounced anemia phenotype observed in Slc39a8(neo/ neo) mice may also be an outcome of dysregulation of Zn 2+dependent transcriptional events in hematopoiesis. This idea is consistent with the emerging consensus that Zn 2+ acts as a ''second-messenger'' in cell-cell signaling [37] including the central nervous system [38;39] and immune system [40]; to date, no one has published evidence in support of Zn 2+ as a second-messenger during in utero hematopoiesis.
Thymic atrophy, lymphopenia, and compromised cell-and antibody-mediated responses that cause increased rates of infections of longer duration are the immunological hallmarks of zinc deficiency in adult humans and mice [41]; however, in Slc39a8(neo/neo) fetuses and newborns, we did not observe any decrease in thymus weight ( Table 2)-suggesting that chronic zinc deficiency in adults is distinctly different from what is seen here in our mouse in utero model, with regard to effects of ZIP8 deficiency on hematopoiesis.
Whereas a visibly pale and smaller Slc39a8(neo/neo) homozygote is apparent from GD11.5 onward (Fig. 1), we found in utero lethality and resorption did not become significant until GD18.5 ( Table 1). The precise cause of death between GD18.5 and 48 h remains unclear; however, the chronic anemia in utero, plus the serious underdevelopment of several organ systems (spleen, liver, kidney and lung) are most likely contributing factors.
Decreased total zinc and iron content occurs in many PND1   Slc39a8(neo/neo) tissues (Fig. 8). It is highly likely that decreases in zinc and iron load in the developing tissues (in utero) are more severe than those measured in PND1, because in the surviving PND1 pups, compensatory pathways that could counter the effect of ZIP8 loss may already have been selected for (e.g. up-regulation of other zinc transporters). The data in Table 6 confirm unequivocally that the neocassette-mediated dysregulation of Slc39a8 gene expression is responsible for the observed phenotype during mouse embryo development. If additional transgenic Slc39a8 alleles are present in the offspring [e.g. as occurs in Slc39a8(neo/neo)_BTZIP8-3(+) pups], the mice are viable and fertile. This experiment strongly argues that the lack of a functional ZIP8 is both necessary and sufficient for all the defects associated with this hypomorphic phenotype.
The finding of strikingly lowered serum triglyceride levels (Table S3) is intriguing, in light of a recent genome-wide association (GWA) study showing that a single-nucleotide polymorphism (SNP) in the SLC39A8 gene-causing a missense (p.Ala391Thr) mutation-is associated with low HDL-cholesterol, elevated blood pressure, increased body mass index, and abnormal natriuric peptide levels; 8% of the eight populations studied carried the Thr391 mutation [42]. The role of ZIP8 in lipid metabolism therefore warrants further investigation.
In a case-control GWA study in Spain of 476 schizophrenia patients and 447 control subjects, and then further corroborated in a second sample comprising 4,069 cases and 15,128 control subjects of European origin, a different highly significant nonsynonymous SNP within the SLC39A8 gene was described; the allelic frequency of this missense mutation is 0.34 in Europeans [43]. Moreover, a de novo deletion affecting SLC39A8 and 10 additional genes was reported in a patient with nonsyndromic mental retardation {354}. These clinical findings suggest that ZIP8 might provide a critical function within the central nervous system (CNS). Although we found strikingly lowered ZIP8 mRNA (Fig. 2) and protein in cerebrum and cerebellum of the Slc39a8(neo/neo) hypomorph, the brain was not decreased in size ( Table 2), but we did note a misshapen cranium. Thus, a possible role of ZIP8 in the CNS should be studied further.
To elucidate the role of ZIP8 in specific cell types, conditional Slc39a8(-/-) knockout lines need to be generated. Recently, using either the Slc39a8(neo) allele or an Slc39a8(f) floxed allele, we have successfully generated tissue-specific knockout lines in which ZIP8 expression can be ablated in hepatocytes or in renal proximal tubular epithelial cells (unpublished data); creation of alveolar epithelial-specific and neuron-specific conditional knockout lines are underway. These lines should help provide insight into the function of ZIP8 in heart disease and CNS disorders.
Finally, this study demonstrates the value of characterizing a hypomorphic allele instead of a conventional knockout mouse. ZIP8 is expressed as early as the gastrula stage [45] and in visceral endoderm at GD7.5 [46]. In fact, ZIP8 is used as a potential indicator of cell differentiation (self-renewal-related signaling) in embryonic stem cells [47]. Hence, it is very likely that a Slc39a8(-/-) global conventional knockout would die very early during development-making studies of the function of this gene virtually impossible-whereas the Slc39a8(neo/neo) pup, surviving until the neonatal period, provides us with a larger window of time for studying ZIP8 function during in utero and neonatal growth, multi-organ development, and hematopoiesis. Any of the mouse lines described in this paper are available to interested colleagues.

Animals
All cloning details were described previously [14]. For all in utero studies, the morning on which the vaginal plug was found is considered gestational day-0.5 (GD0.5). Individual Slc39a8(+/+), Slc39a8(+/neo) and Slc39a8(neo/neo) whole embryos/fetuses, placentas, and yolk sacs were collected and genotyped at various time-points between GD11.5 and GD18.5; on postnatal day 1 (PND1), we also collected numerous tissues from genotyped pups for further analysis. All mouse experiments were conducted in accordance with the National Institutes of Health standards for the care and use of experimental animals and the University Cincinnati Medical Center Institutional Animal Care and Use Committee [protocol #11-09-12-01; approved 4 Sept 2011 R 3 Sept 2014].

Methods and Techniques
All methods and any associated references are available in the Supplementary Data S1 online. Figure S1 Comparison of size of six PND1 organs among the three genotypes. Inserted ruler is measured in cm and mm. The Slc39a8(neo/neo) organ can sometimes be seen as more pale than that in the heterozygote or wild-type. (TIF) Figure S2 Comparison of size of GD13.5 placenta, yolk sac and whole embryo among the three genotypes. Again, sometimes the Slc39a8(neo/neo) appears more pale than the heterozygote or wild-type. (TIF)