Mutations in mitochondrial
carrier family gene SLC25A38
cause nonsyndromic autosomal
Duane L Guernsey1, Haiyan Jiang1, Dean R Campagna2,
Susan C Evans1, Meghan Ferguson3, Mark D Kellogg4,
Mathieu Lachance5, Makoto Matsuoka1, Mathew Nightingale1,
Andrea Rideout3, Louis Saint-Amant5, Paul J Schmidt2,
Andrew Orr6, Sylvia S Bottomley7, Mark D Fleming2,
Mark Ludman8, Sarah Dyack8, Conrad V Fernandez9&
Mark E Samuels10
The sideroblastic anemias are a heterogeneous group of
congenital and acquired hematological disorders whose
morphological hallmark is the presence of ringed sideroblasts—
bone marrow erythroid precursors containing pathologic iron
deposits within mitochondria. Here, by positional cloning,
we define a previously unknown form of autosomal recessive
nonsyndromic congenital sideroblastic anemia, associated
with mutations in the gene encoding the erythroid specific
mitochondrial carrier family protein SLC25A38, and
demonstrate that SLC25A38 is important for the
biosynthesis of heme in eukaryotes.
Sideroblastic anemias are hematological disorders characterized by
the presence of ringed sideroblasts1,2. Only one gene until now has
been known to cause nonsyndromic congenital sideroblastic anemia
(CSA) when mutated: the X-linked gene ALAS2, which encodes the
erythroid-specific, mitochondrial-localized first enzyme in the heme
biosynthetic pathway3. However, there are numerous unexplained
sporadic cases of CSA in the literature, as well as reports of families
with CSA consistent with autosomal recessive inheritance1,4. We
ascertained three families from the Canadian Maritime provinces,
each with one child affected with CSA (Fig. 1a, Table 1 and
Supplementary Fig. 1a online). One tested affected individual was
negative for any mutation in ALAS2. The occurrence of an affected
female in one of the families, without evidence of skewed X inactiva-
tion, as well as the absence of a disease phenotype in any parent
was inconsistent with autosomal dominant or X-linked recessive
inheritance. Although not formally related, the families derive from
a local subpopulation isolate, consistent with a possible genetic
founder effect. Consequently, we carried out a SNP-based genome-
wide scan in 11 individuals, including the three CSA-affected
individuals, their parents and one unaffected sibling in each of two
families. We identified an B3-Mb segment of chromosome 3 in
which the three affected individuals shared a homozygous haplotype
consisting of 299 consecutive SNPs identical by state, suggesting a
region of identity by descent (Supplementary Table 1 online). This
locus was more than three times as large as any other homozygous
region. No evidence of linkage was observed to any known locus
related to heme or iron metabolism. The chromosome 3 linked
region contains 32 annotated genes according to RefSeq and the
UCSC browser. After sequencing most or all of 15 genes in the
interval, we detected a homozygous 117R4X (726C4T) stop
codon in exon 4 of the gene SLC25A38 in all three affected individuals
(Table 1, Supplementary Table 2 and Supplementary Fig. 1b online).
The mutation was heterozygous in all six parents, as expected, and the
unaffected siblings were both homozygous wild type at this position of
Subsequently, we examined 41 additional subjects with familial or
sporadic CSA in whom ALAS2 mutations had previously been
excluded. We identified multiple additional biallelic SLC25A38 muta-
tions in three families plus nine individuals from this cohort (Table 1
and Supplementary Fig. 1a,b). Where feasible, segregation of the
mutation plus examination of short tandem repeat markers in the
vicinity of SLC25A38 supported linkage to the gene. In total we found
11 variants, including three stop codons, two frameshifts and one
splice acceptor site mutation. We also identified a mutation of the
presumptive stop codon predicted to result in an extended C termi-
nus, as well as four missense alleles. Three of the missense mutations,
encoding 30G4E, 134R4H and 187R4P, were predicted to have
deleterious effects on protein function by SIFT, PolyPhen, PANTHER
and Align-GVGD (Supplementary Methods online). The fourth
missense variant, 209D4H, may affect splicing as it involves the
conserved guanine nucleotide at the 3¢ terminus of exon 5 (http://
spliceport.cs.umd.edu/). The mutations that recurred in multiple
families were on different microsatellite haplotypes, consistent with
independent events rather than common ancestry. None of the
presumptive disease-associated alleles was found in 96 European or
145 Maritime Canadian Caucasian population sample controls.
Received 8 December 2008; accepted 17 February 2009; published online 3 May 2009; doi:10.1038/ng.359
1Department of Pathology, Dalhousie University Halifax, Nova Scotia, Canada.2Department of Pathology, Children’s Hospital Boston and Harvard Medical School,
Boston, Massachusetts, USA.3Maritime Medical Genetics Service, IWK Health Centre, Halifax, Nova Scotia, Canada.4Department of Laboratory Medicine, Children’s
Hospital Boston, Boston, Massachusetts, USA.5Department of Pathology, Le Groupe de Recherche sur le Syste `me Nerveux Central (GRSNC), Universite ´ de Montre ´al,
Montre ´al, Quebec, Canada.6Department of Ophthalmology and Visual Sciences, Dalhousie University, Halifax, Nova Scotia.7Department of Medicine, Hematology-
Oncology Section, University of Oklahoma College of Medicine, Oklahoma City, Oklahoma, USA.8Department of Pediatrics, Division of Medical Genetics, IWK Health
Centre and Dalhousie University, Halifax, Nova Scotia, B3K 6R8, Canada.9Department of Pediatrics, Division of Hematology and Oncology, IWK Health Centre and
Dalhousie University, Halifax, Nova Scotia, Canada.10Centre de Recherche du CHU Ste-Justine, Universite ´ de Montre ´al, Montre ´al, Quebec, Canada. Correspondence
should be addressed to M.E.S. (email@example.com).
NATURE GENETICS VOLUME 41 [ NUMBER 6 [ JUNE 2009 651
© 2009 Nature America, Inc. All rights reserved.
SLC25A38 is highly and preferentially expressed in transferrin
receptor (CD71) positive erythroid cells (Supplementary Fig. 2
online). It is a member of the SLC25 family of inner mitochondrial
membrane transporters, which share an N-terminal mitochondrial
targeting signal and six transmembrane helices (Supplementary
Fig. 1b)5,6. Most SLC25 proteins promote the exchange of one
metabolite for another across the inner mitochondrial membrane7.
They are subdivided into three major groups—keto acid, amino acid
and adenine nucleotide carriers—each of which shares conserved
amino acids within transmembrane helices 2, 4 and 6 that are thought
to provide contact points that determine substrate specificity8.
SLC25A38 is predicted to encode an amino acid carrier, particularly
because of a conserved arginine-asparate (RD) dipeptide sequence
present in transmembrane helix 4 (so-called contact point II, CPII).
Three of the missense mutations that we identified are predicted to be
in close proximity to CPII (Supplementary Fig. 3a–c online). One
missense, R187P, is in the conserved arginine in the RD dipeptide of
CPII itself, supporting the functional significance of the missense
sequence variants found in this cohort of individuals with CSA.
In order to confirm a role for SLC25A38 in erythropoiesis, we
carried out mRNA knockdown experiments by injecting antisense
morpholinos into zebrafish embryos. The zebrafish has two presumed
functional SLC25A38 orthologs, called by us slc25a38a and slc25a38b,
that are 60.1% and 61.9% identical to human SLC25A38, respectively
(Supplementary Fig. 4 online). Simultaneous injection of morpho-
linos directed at both genes led to an anemic phenotype in morphant
zebrafish embryos (Fig. 1b), similar to, although not as severe as, an
anti-alas2 knockdown done as a positive control. Incomplete heme
reduction was possibly due to limitations on the total amount of
morpholinos that can be injected without nonspecific disruption
Taken together, these genetic and functional data provide definitive
evidence that germline loss-of-function mutations in SLC25A38 cause
a nonsyndromic, autosomal recessive form of CSA in humans. In fact,
SLC25A38 seems to be a relatively common cause of CSA; the
SLC25A38 mutations we detected in 12 non-Maritime probands
account for 17% of the complete cohort of 72 cases (41 non-Maritime
analyzed here plus an additional 31 previously found to have ALAS2
mutations). Some fraction of the affected individuals without
observed mutations in either gene could still carry undetected patho-
genic variants in upstream regulatory or intronic sequences. No
genetic conditions in human or other mammals have previously
been associated with variation in SLC25A38 (ref. 9).
The phenotypic similarity between affected individuals with
SLC25A38 and ALAS2 mutations raised the possibility that the
autosomal recessive form of the disorder might equally result from
a heme biosynthetic defect. To assess this possibility, we examined the
phenotype of the yeast Saccharomyces cerevisiae with a germline
deletion in the putative SLC25A38 ortholog YDL119c (Supplemen-
tary Figs. 4 and 5a online). Yeast lacking YDL119c grew well on the
fermentable carbon source dextrose, but only poorly aerobically on
glycerol (Fig. 1c,i). This typically indicates a defect in respiration,
which requires intact mitochondria and heme. Furthermore, the
deletion strainwas unable to
(Fig. 1c,ii), indicating that the defect is likely in heme biosynthesis,
as nitroprusside reduction requires heme-dependent cell-surface fer-
rireductase FRE1 (ref. 10). Nitroprusside reduction could be rescued
by supplementation of the medium with either glycine or 5-amino-
levulinic acid (ALA), a precursor and the product, respectively, of the
reaction catalyzed by ALAS. Furthermore, in minimal medium,
biochemical analyses documented a 6.7-fold reduction in the amount
of total cellular ALA in the yeast deletion strain (Fig. 1c,iii), suggesting
that loss of YDL119c impairs ALA biosynthesis in vivo. Only a modest
effect on total cellular glycine levels was observed. Our results are
consistent with proteomics studies documenting that YDL119c pro-
tein is in the mitochondrial proteome and with a genome-wide
mutagenesis screen in which this gene knockout showed a classic
petite phenotype5,11–13. The protein product of SLC25A38 was not
identified in a large-scale mouse mitochondrial proteomic study,
probably because bone marrow was not among the tissues analyzed14.
The first enzymatic step in heme biosynthesis is the condensation of
glycine and succinyl-CoA to form ALA, mediated by the products of
slc25a38a + slc25a38b
SD-NP SD-ALA-NP SD-Gly-NP
Metabolite per 106 cells (µmol)
Figure 1 SLC25A38 phenotypes in human, fish and yeast. (a) Photo-
micrograph of index case (D01-06434) bone marrow aspirate stained with
Prussian blue showing ringed sideroblasts (dark blue inclusions represent
iron deposits). Microscope Olympus BX 51, photographed at ?1000
magnification, daylight filter. (b) Zebrafish embryos stained for hemoglobin at
48 hours postfertilization (hpf) following microinjection with water control or
antisense morpholino reagents targeting fish orthologs slc25a38a,
slc25a38b, slc25a38a plus slc25a38b or alas2. (c) Heme biosynthetic
defect in yeast deficient in the presumptive SLC25A38 ortholog, YDL119c.
(c,i) Wild-type yeast strain BY4741 (WT), isogenic YDL119c knockout yeast
(ydl119cD, D), ydl119cD transformed with empty vector (pPJS209, D+V)
and ydl119cD transformed with a vector expressing wild-type YDL119c
(pPJS211, D+I) cells were plated on minimal medium plus dextrose (SD)
or glycerol (SG) and grown aerobically for 4 d. From top to bottom, plated
dilutions represent 5 ? 102, 5 ? 101and 5 ? 100cells. (c,ii) Strains in c,i
were plated on SD supplemented with 35 mM sodium nitroprusside (SD-NP)
and 50 g/l 5-aminolevulinic acid (SD-NP-ALA) or 5 mM glycine (SD-NP-Gly)
and grown for 3 d. Blue coloration of colonies results from the heme-
dependent reduction of nitroprusside. (c,iii) Total metabolites were harvested
from wild-type (WT) or ydl119cD (D) yeast grown overnight in SD media and
analyzed by LC-MSMS to determine total cell ALA or glycine concentrations
(n ¼ 9). Results are expressed as the mean ± s.e.m. and the statistical
significance was calculated by the Student’s t-test (*P o 0.001). Full
methods are provided in Supplementary Methods.
652 VOLUME 41 [ NUMBER 6 [ JUNE 2009 NATURE GENETICS
© 2009 Nature America, Inc. All rights reserved.
either the ALAS1 (generally expressed) or ALAS2 (erythroid specifically Download full-text
expressed) genes in humans. The decrease in ALA levels in the
SLC25A38 orthologous yeast deletion strain, which is expected to
have normal yeast ALAS enzymatic activity, is consistent with
decreased substrate availability for the reaction. Succinyl-CoA is
generated in mitochondria at high levels, making it an unlikely target
of SLC25A38 action. By contrast, little is known about the mechanisms
of glycine transport into mitochondria7. We hypothesize that
SLC25A38 facilitates ALA production by importing glycine into
mitochondria or by exchanging glycine for ALA across the mitochon-
drial inner membrane. No specific SLC25 family member or other
protein has yet been demonstrated to carry out mitochondrial glycine
or ALA transport.
It remains to be seen whether an equivalent function to that of
SLC25A38 is also mediated by a dedicated gene in nonerythroid cells.
No other SLC25 family member is particularly closely related to
SLC25A38, thus no obvious candidate could be defined among the
uncharacterized family members (Supplementary Fig. 5b). Concei-
vably, the equivalent metabolic transport step occurs passively in
nonerythroid cells, which have a low requirement for ALA to support
a much smaller heme synthetic demand. Alternatively, another char-
acterized amino-acid transporter in this gene family, or completely
uncharacterized family member, might import glycine adequate for
the needs of nonerythroid cells.
Note: Supplementary information is available on the Nature Genetics website.
We gratefully acknowledge the participation of all the families with this genetic
disorder. We thank P. Drapeau for advice on zebrafish physiology and experimen-
tation and P. Wise and E. Wasson for expert technical assistance. The following
agencies provided funding for this project: Genome Canada (M.E.S.), Genome
Atlantic (M.E.S.), Nova Scotia Health Research Foundation (M.E.S.), Nova Scotia
Research and Innovation Trust (M.E.S.), IWK Health Centre Foundation (M.E.S.),
Dalhousie University (M.E.S.), Capital Health Research Fund (M.E.S.), Fonds de la
Table 1 Clinical and biochemical features of CSA-affected individuals with SLC25A38 mutations
SubjectAncestry Familial? SexAge Hb (g/dl)
Subject reference numbers correspond to those in Supplementary Figure 1a. Age given is the age at the time of referral for study with corresponding laboratory values. Familial refers
to occurrence of multiple affected sibs in the same nuclear family, not genetic versus other etiology per se. Hb, hemoglobin; MCV, mean cell volume; Tf sat, transferrin saturation; ND, not
determined; m, increased. In all cases, the Hb values were lower, and the Tf Sat and ferritin levels greater than the age-adjusted norms for the referring laboratory. Superscript ‘T’ indicates chronic
transfusion at the time of study. All subjects had microcytic red blood cell parameters prior to the initiation of transfusion. Mutations of both gene alleles are given, as are the corresponding amino-
aFor data on first three subjects, see Supplementary Methods.
Recherche en Sante ´ de Que ´bec (L.S.-A.), NIH K01 DK074410 (P.J.S.) and NIH
R01 DK080011 (M.D.F.), The US Department of Veterans Affairs (S.S.B.),
University of Oklahoma Health Sciences Center Provost’s Fund (S.S.B.), and the
Oklahoma Center for Advancement of Science and Technology (S.S.B.).
D.L.G. oversaw molecular genetic studies. H.J. performed statistical mapping
analyses. D.R.C. performed haplotype studies. S.C.E. performed molecular genetic
analyses. M.F. participated in clinical ascertainment of the patients. M.D.K.
developed and performed mass spectrometric ALA and glycine analyses. M.L.
performed zebrafish knockdown studies. M.M. performed molecular genetic
analyses. M.N. performed molecular genetic analyses. A.R. participated in clinical
ascertainment of the patients. L.S.-A. oversaw zebrafish studies. P.J.S. analyzed the
yeast phenotype. A.O. performed haplotype studies. S.S.B. obtained institutional
review board approval and consents, collected clinical data and samples and
edited the manuscript. M.D.F. oversaw haplotype determinations and yeast genetic
studies and co-wrote the manuscript. M.L. and S.D. participated in clinical
studies of the study subjects. C.V.F. ascertained the study subjects and performed
clinical studies to determine the phenotype. M.E.S. oversaw molecular genetic
studies and co-wrote the manuscript.
Published online at http://www.nature.com/naturegenetics/
Reprints and permissions information is available online at http://npg.nature.com/
1. Bottomley, S.S. Curr. Hematol. Rep. 5, 41–49 (2006).
2. Fleming, M.D. Semin. Hematol. 39, 270–281 (2002).
3. Bottomley, S.S., May, B.K., Cox, T.C., Cotter, P.D. & Bishop, D.F. J. Bioenerg.
Biomembr. 27, 161–168 (1995).
4. Camaschella, C. Br. J. Haematol. 143, 27–38 (2008).
5. Palmieri, F. et al. Biochim. Biophys. Acta 1757, 1249–1262 (2006).
6. Haitina, T., Lindblom, J., Renstrom, T. & Fredriksson, R. Genomics 88, 779–790
7. Palmieri, F. Pflugers Arch. 447, 689–709 (2004).
8. Kunji, E.R. & Robinson, A.J. Biochim. Biophys. Acta 1757, 1237–1248 (2006).
9. Palmieri, F. Biochim. Biophys. Acta 1777, 564–578 (2008).
10. Lesuisse, E., Casteras-Simon, M. & Labbe, P. Anal. Biochem. 226, 375–377 (1995).
11. Elstner, M. et al. Mol. Biotechnol. 40, 306–315 (2008).
12. Perocchi, F. et al. PLoS Genet. 2, e170 (2006).
13. Reinders, J., Zahedi, R.P., Pfanner, N., Meisinger, C. & Sickmann, A. J. Proteome Res.
5, 1543–1554 (2006).
14. Pagliarini, D.J. et al. Cell 134, 112–123 (2008).
NATURE GENETICS VOLUME 41 [ NUMBER 6 [ JUNE 2009 653
© 2009 Nature America, Inc. All rights reserved.