RESEARCH ARTICLE

Retina-specific loss of Ikbkap/Elp1 causes mitochondrialdysfunction that leads to selective retinal ganglion celldegeneration in a mouse model of familial dysautonomiaYumi Ueki*, Veronika Shchepetkina and Frances Lefcort

ABSTRACTFamilial dysautonomia (FD) is an autosomal recessive disordermarked by developmental and progressive neuropathies. It iscaused by an intronic point-mutation in the IKBKAP/ELP1 gene,which encodes the inhibitor of κBkinase complex-associated protein(IKAP, also called ELP1), a component of the elongator complex.Owing to variation in tissue-specific splicing, the mutation primarilyaffects the nervous system. One of the most debilitating hallmarks ofFD that affects patients’ quality of life is progressive blindness. Todetermine the pathophysiological mechanisms that are triggered bythe absence of IKAP in the retina, we generated retina-specificIkbkap conditional knockout (CKO) mice using Pax6-Cre, whichabolished Ikbkap expression in all cell types of the retina. Althoughsensory and autonomic neuropathies in FD are known to bedevelopmental in origin, the loss of IKAP in the retina did not affectits development, demonstrating that IKAP is not required for retinaldevelopment. The loss of IKAP caused progressive degeneration ofretinal ganglion cells (RGCs) by 1 month of age. Mitochondrialmembrane integrity was breached in RGCs, and later in other retinalneurons. In IkbkapCKO retinas, mitochondriawere depolarized, andcomplex I function and ATP were significantly reduced. Althoughmitochondrial impairment was detected in all Ikbkap-deficient retinalneurons, RGCs were the only cell type to degenerate; the survival ofother retinal neurons was unaffected. This retina-specific FD modelis a useful in vivo model for testing potential therapeutics formitigating blindness in FD. Moreover, our data indicate that RGCsand mitochondria are promising targets.

KEY WORDS: IKAP/ELP1, Familial dysautonomia, Mitochondria,Retinal degeneration, Retinal ganglion cells

INTRODUCTIONFamilial dysautonomia (FD;MIM223900) is an autosomal recessivecongenital neuropathy that is caused by an intronic mutation in theIKBKAP gene, which encodes the inhibitor of κB kinase complex-associated protein (IKAP), also called elongator complex protein 1(ELP1) (Anderson et al., 2001; Dong et al., 2002; Riley et al., 1949;Slaugenhaupt et al., 2001). A point-mutation in intron 20 results in

tissue-specific exon skipping and generates an unstable mRNA,causing a loss-of-function phenotype predominantly in the nervoussystem (Cuajungco et al., 2003; Dietrich et al., 2012; Keren et al.,2010). FDpatients suffer fromcongenital andprogressive neuropathies,including reduced peripheral afferent sensory function, unstableblood pressure, hypotonia, poor growth and spinal curvature;patients often die in early adulthood owing to sudden unexpecteddeath during sleep (Axelrod, 2002; Palma et al., 2014, 2017; Rileyet al., 1949). The complete functional repertoire of IKAP/ELP1remains unresolved, but includes a key role as the scaffoldingsubunit of the six-subunit elongator complex (ELP1-6) that modifieswobble uridine subunits of tRNA during translation (Bauer andHermand, 2012; Chen et al., 2009a; Huang et al., 2005). In itsabsence, translation of codon-biased mRNAs is impaired, resultingin perturbations in levels of specific proteins (Goffena et al., 2018).Either as a direct or indirect consequence of this altered translation,Ikbkap conditional knockout (CKO) neurons exhibit impairedaxonal transport, target innervation and cell survival (Abashidzeet al., 2014; Chaverra et al., 2017; Close et al., 2006; George et al.,2013;Hunnicutt et al., 2012; Jackson et al., 2014; Johansen et al., 2008;Lefler et al., 2015; Naftelberg et al., 2016; Naumanen et al., 2008;Tourtellotte, 2016; Ueki et al., 2016).

FD is classified as a hereditary sensory and autonomic neuropathy(HSAN III), yet closer examination of both the patient and mouse-model phenotypes reveals central nervous system (CNS) pathology(Axelrod et al., 2010; Mendoza-Santiesteban et al., 2014; Ochoa,2003). Furthermore, accumulating evidence demonstrates a vitalrole of the elongator complex in the CNS: variants in ELP2 areassociated with neurodevelopmental disability (Cohen et al.,2015; Franic et al., 2015) and variants in ELP3 with amyotrophiclateral sclerosis (ALS) (Simpson et al., 2009). In addition, loss ofELP4 causes Rolandic epilepsy syndrome, autism and intellectualdisability (Addis et al., 2015; Gkampeta et al., 2014; Nguyenet al., 2010; Reinthaler et al., 2014; Strug et al., 2009). One of themajor clinical hallmarks of FD is progressive blindness, whichstarts at an early age as a result of the progressive loss of retinalganglion cells (RGCs) (Mendoza-Santiesteban et al., 2014, 2012,2017). Patients are often legally blind by their thirties. In the FDcommunity, there is mounting interest in developing treatments toameliorate the blindness by preventing the progressive RGC lossin order to improve the quality of life of FD patients.

The pathophysiological mechanisms underlying the loss ofvision have not been the focus of any study until recently, however.Our recent work with IkbkapCKOmice, which lack Ikbkap in bothCNS and peripheral nervous system (PNS) neurons, demonstratedthat loss of Ikbkap in RGCs causes their progressive death (Uekiet al., 2016), recapitulating the retinal phenotype of the FD patients(Mendoza-Santiesteban et al., 2014, 2017). Unfortunately, withthis model we were unable to analyze the consequence of IkbkapReceived 22 January 2018; Accepted 12 June 2018

Department of Cell Biology andNeuroscience, Montana State University, Bozeman,MT 59717, USA.

*Author for correspondence (yueki@montana.edu)

Y.U., 0000-0003-2898-3267; V.S., 0000-0003-0948-3689

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2018. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

loss in cell types other than RGCs, as the Ikbkap deletion wasprimarily restricted to RGCs in the retina. Moreover, Ikbkap wasdeleted in themajority of cell types in both the PNS and CNS, sowewere unable to determine whether the loss of RGCs was the director indirect consequence of loss of Ikbkap throughout the nervoussystem. These mice had a severe progressive peripheral neuropathyand CNS impairments (Chaverra et al., 2017). To overcome thesecomplications, in this study we have generated and characterized anew mouse FD retina model by conditional deletion of Ikbkapsolely in the retina, using a retina-specific Pax6-Cre, which isexpressed in the retinal progenitors; thus, all retinal cell types areaffected (Marquardt et al., 2001). This retina-specific FD modelsystem allowed us to directly interrogate the consequence of loss ofIkbkap in all retinal cell types in the context of an otherwise healthynervous system. Our data demonstrate that loss of Ikbkap solely in theretina causes RGC degeneration, yet the survival of other retinal celltypes is unaffected. In addition, therewas no developmental phenotypein the FD retinas, establishing for the first time that the loss of RGCsoccurs postnatally and is a progressive neurodegeneration as in thehuman disease.Defects in mitochondria have been implicated in many, if not

all, neurodegenerative disorders (Schon and Przedborski, 2011).Similar to FD, two other optic neuropathies that are consideredmitochondrial diseases, Leber’s hereditary optic neuropathy(LHON) and dominant optic atrophy (DOA), are also characterized byloss of vision owing to slow, progressive RGC degeneration(Carelli et al., 2009; Chevrollier et al., 2008; Kirches, 2011;Newman and Biousse, 2004; Yu-Wai-Man et al., 2011; Zannaet al., 2008). In this study, we hypothesized that the progressivedemise of retinal neurons in FD results from mitochondrialdysfunction. Our data demonstrate perturbation of mitochondrialmembrane integrity and function, as well as decreased ATPcontent in the Ikbkap CKO retinas. Interestingly, althoughmitochondria of all Ikbkap-deficient retinal neurons seem to beaffected, RGCs are the only cell type that degenerate. Together,these results suggest that RGCs and mitochondria are promisingtargets for future therapeutics to mitigate the progressive loss ofvision in FD patients.

RESULTSIKAP expression is significantly reduced in the Ikbkap CKOretinaWe previously generated Ikbkap CKO mice using a TUBA1apromoter-driven Cre (Tuba1α-Cre), which targets ∼90% ofpostmitotic RGCs but not other cell types in the retina (Uekiet al., 2016). We demonstrated that loss of IKAP in the RGCscaused a slow, progressive RGC degeneration most severely inthe temporal retina, later followed by indirect photoreceptor lossand complete retinal disorganization. In this model, however,Ikbkap was deleted from the majority of CNS neurons andapproximately 40% of PNS neurons; thus, whether all or someof these phenotypes were directly due to autonomous loss ofIkbkap in retinal cells could not be determined. In this currentstudy, we have generated a retinal-specific Ikbkap CKO mouseusing Pax6-Cre. Pax6-Cre is expressed in retinal progenitors,affecting all six types of retinal neurons and Müller glia, and itsexpression is restricted solely to the retina (Marquardt et al.,2001). Therefore, we can characterize the effect of autonomousloss of Ikbkap not only in RGCs, but in all retinal neurons. Toidentify appropriate therapeutic targets, it is important tounderstand the consequence of IKAP loss in all retinal neuronsand not only in RGCs.

Western blot analysis showed a significant reduction in IKAPprotein in the Pax6-Cre Ikbkap CKO retinas at 1 month comparedwith littermate control retinas (Fig. 1A); on average, there wasover 70% reduction in IKAP expression in the CKO retina(Fig. 1B). Previous studies show that Pax6-Cre is expressed in theperipheral retina, but not in the central retina (Georgi and Reh,2010; La Torre et al., 2013; Marquardt et al., 2001). We alsoobserved this peripheral bias when we crossed Pax6-Cre micewith Rosa-EGFP Cre reporter mice (Fig. 1C,D). In a typical Cre+

retina, 50-85% retinal regions were GFP+ (Fig. 1C), whichcorresponds with ∼70% reduction in IKAP expression (Fig. 1A,B). When we analyzed Cre reporter retinal cross-sections at1 month, all the retinal cell types including RGCs [identified usingthe retinal markers Brn transcription factor (Brn3+) and/or RNA-binding protein with multiple splicing (RBPMS+)] expressed GFP

Fig. 1. Generation of a retinal model of FD. Pax6-Cre+;IkbkapFlox/Flox mice were generated to delete Ikbkap selectively from the retina. (A) RepresentativeIKAP western blot at 1 month. (B) Pax6-Cre+;IkbkapFlox/Flox (CKO) retinas show a 70% decrease in IKAP expression compared with Pax6-Cre−;IkbkapFlox/Flox

(Control; Ctrl) retinas at 1 month. *P=0.000016with Student’s t-test (n=5). (C-E)Pax6-Cre expressionwas analyzed on retinal (C) flatmounts and (D) cross-sectionsusing Rosa-EGFP Cre reporter mice at 1 month. Cre expression was detected in the mid-peripheral retinas and, typically, 50-85% of the retina was affectedin each eye. Images in C and D are composites. Multiple images were taken and reconstructed to show whole retinal (C) flatmount and (D) cross sections.(E)Representative images of retinal cross-sections in theCre-expressing region are shown.All cell types of the retina expressedCre (GFP+), includingRGCs (Brn3+

and/or RBPMS+). GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

2

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

(Fig. 1E), as reported in previous studies (Marquardt et al., 2001).We used Pax6-Cre Ikbkap CKO that had been crossed with theRosa-EGFP Cre reporter strain for immunohistological experiments,

in order to identify the retinal regions that were deficient inIKAP. In summary, we have established a useful mouse model ofFD blindness.

Fig. 2. Loss of IKAP in the retina causes progressive loss of RGCs. (A) At 9 months, Pax6-Cre Ikbkap CKO retinas show a decrease in the number of Brn3+

RGCs at 1 mm from the optic nerve. (B) Significant loss of Brn3+ RGCs in CKO retinas was detected in all quadrants (temporal, superior, nasal and inferior)as early as 1 month, and the loss was progressive. Rapid RGC loss occurs between 1 month and 3 months. There was no significant difference in thenumber of RGCs at P14, when retinal development has completed. P14 (n=4), 1 month (n=6), 3 months (n=4), 6 months (n=6) and 9 months (n=3). The numberof RGCs in the control and CKO retinas at the same region and agewere comparedwith Student’s t-test (*P<0.05). (C) Cross-sections of 6 month temporal retinasat 1.5 mm from the optic nerve. There was an apparent reduction in the number of Brn3+ RGCs and total RGCs (RBPMS+, pan-RGCmarker). (D) Cross-sectionsof optic nerves at 1 month (top) and 9 months (bottom) were visualized with DAPI staining. (E) There was no difference in control and CKO optic nervecircumference at 1 month. P=0.41 (not significant) with Student’s t-test (n=3). (F) By 9 months, there is a 45% decrease in CKO optic nerve circumferencecompared with controls. *P=0.018 with Student’s t-test (n=3). (G,H) There was no loss of melanopsin+ intrinsically photosensitive RGCs at 18 months.Melanopsin+ cells were counted at 1 mm from the optic nerve in each quadrant. P=0.2 (temporal), P=0.3 (superior), P=0.3 (nasal) and P=0.1 (inferior) withStudent’s t-test (n=5). Scale bars: 100 µm (A,D,G) and 50 µm (C).

3

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

Loss of IKAP induces progressive RGC degeneration,whereas other retinal cell types are unaffectedAs expected from our previous study (Ueki et al., 2016), weobserved progressive loss of RGCs owing to loss of IKAP in theretina. Staining of 9 month retinal flatmounts and 6 month retinalcross-sections with RGC markers Brn3 (Fig. 2A) and RBPMS(Fig. 2C), respectively, showed an apparent reduction in the RGCnumber in all four retinal quadrants. Quantification of RGC numberreveals a significant loss of RGCs at 1 month, followed by a rapidreduction in the RGC number over the subsequent 2 months(Fig. 2B). By 6 months, there was a 40-60% loss of RGCs in CKOretinas compared with controls. This progressive loss of RGCs is notdue to Pax6-Cre expression itself or a result of the loss of one Ikbkapallele, as there was no significant difference in RGC counts in Pax6-Cre+; IkbkapFlox/+ or Pax6-Cre−; IkbkapFlox/Flox (control) retinas at18 months (Fig. S1). To determine whether IKAP is required in thedeveloping retina, we quantified RGC number at postnatal day 14(P14), the age when retinal development has completed.Importantly, there was no significant difference in RGC numberbetween control and CKO retinas at P14, indicating that RGCdevelopment does not require IKAP (Fig. 2B). Thinning of the CKOoptic nerves (RGC axon bundles) appeared to temporally followRGC degeneration (Fig. 2D). Although there was no difference inthe thickness of the optic nerve at 1 month (Fig. 2E), CKO opticnerves at 9 months were significantly thinner compared with thoseof the controls (Fig. 2F). As observed in FD patients (Mendoza-Santiesteban et al., 2017), intrinsically photosensitive, melanopsin+

RGCs were preserved even at 18 months, despite extensivedegeneration of conventional RGCs (Fig. 2G,H). This observationsuggests that loss of IKAP differentially affects RGC subtypes.We have previously shown that IKAP is expressed in RGCs,

photoreceptors, amacrine cells and a subset of bipolar cells in adultmouse retinas (Ueki et al., 2016). Although the loss of RGCs wassignificant by 1 month in the CKO retinas (Fig. 2), histologicalanalyses of 1, 3, 6, 9, 12 and 15 month CKO retinas did not show anyadditional retinal phenotype nor were any other retinal cell typesaffected by IKAP loss (Fig. 3 and Fig. S2). Immunohistochemistry(IHC) with various cell-type markers showed no significantdifference in photoreceptor (Otx2), bipolar cell (Otx2), amacrinecell (Pax6) or Müller glial (Sox9, Sox2, GFAP) number,morphology or organization at 9 months (Fig. S2A). There was noabnormal morphology in the CKO retinas at 12 months comparedwith the controls (Fig. S2B). At 15 months, although retinal cross-sections of the Cre-affected area (middle-peripheral) show decreasedRGC number, the overall retinal morphology remains intact (Fig. 3Aand Fig. S2C). When the number of rows of photoreceptors werecounted at 0.25 mm intervals from the optic nerve head, there was nodifference in the number of photoreceptors between control andCKO retinas (Fig. 3B). In our previous study, Ikbkap−/− CNSneurons showed disrupted cilia morphology (Chaverra et al., 2017).Therefore, we analyzed photoreceptor cilia, located in between innerand outer segments. A cilia marker, PKD2L-1(red), showed noabnormality in photoreceptor cilia structure in CKO photoreceptorsat 9 months, further indicating that photoreceptors are not affected bythe absence of Ikbkap. As we also observed loss of cholinergicneurons in our other FDmodel (Chaverra et al., 2017), the number ofChAT+ cholinergic amacrine cells was counted in both the innernuclear layer (INL) and ganglion cell layer (GCL) at 6 months(Fig. 3D,E). However, we did not observe any reduction in ChAT+

amacrine cells in the CKO retina. In summary, we determined thatRGCs are the only cell type to require IKAP for their survival, eventhough Pax6-Cre is expressed in all retinal cell types.

Mitochondrial membrane integrity and function is impairedin the absence of IKAPOur recent study revealed that mitochondria of embryonic IkbkapCKO dorsal root ganglia neurons are depolarized, produce elevatedlevels of reactive oxygen species (ROS), are fragmented, and do notaggregate normally at axonal branch points (Ohlen et al., 2017). Insupport of a mitochondrial deficit in FD, FD patients have a temporaloptic nerve degeneration that is reminiscent of LHON andDOA, bothof which are caused by mutations that affect mitochondrial function(Carelli et al., 2009; Chevrollier et al., 2008; Kirches, 2011; Newmanand Biousse, 2004; Yu-Wai-Man et al., 2011; Zanna et al., 2008).Therefore, we investigated the health of mitochondria in our CKOretinas. We performed transmission electron microscopy (TEM) oncontrol and CKO retinas at 1 month and 2.5 months (Fig. 4 and Fig. S3).We observed a breakdown in the mitochondrial double-membraneintegrity in CKO RGCs as early as 1 month, which is during theperiod of rapid RGC degeneration (Fig. 4). By contrast, althoughmitochondria of 2.5 month CKO amacrine cells show a similar lossof membrane integrity, mitochondria of 1 month CKO amacrine cellsdisplay no difference compared with controls (Fig. S3). Moreover,despite this slower progressive loss of membrane integrity, amacrinecells do not die in the absence of IKAP (Fig. S2A and Fig. 3D,E).This finding suggests that loss of IKAP affects different cell types atdifferent rates, and that RGCs are more vulnerable to the loss ofIKAP than are other retinal cell types.

Given the morphological disruption to mitochondrial membraneintegrity detected in TEM, we sought to characterize membraneintegrity at the molecular level. We collected retinas at 3 months andused an antibody cocktail that measures levels of mitochondrialproteins in four different mitochondrial compartments (outer andinner membrane, intermembrane space and matrix) (Fig. 5).Significant decreases in the mitochondrial proteins cyclophilin Dand cytochrome c were detected in CKO retinas at 3 months(12 weeks), supporting the loss of membrane integrity seen in TEManalyses (Fig. 4 and Fig. S3). Loss of mitochondrial integrity in theCKO retinas was progressive: at 2 weeks (P14), the timewhen retinaldevelopment is completed, there was no difference in cyclophilin Dand cytochrome c expression, but both proteins did progressivelydecline in expression in older CKO retinas (Fig. 5B,C). These datasuggest that loss of Ikbkap during development does not affectretinal cell mitochondria. Although RGCs showed morphologicalmitochondrial abnormality and cell death as early as 1 month(Figs 2B and 4), we did not see a significant difference in themitochondrial protein levels by western blot at 1 month (4 weeks).This is probably because RGCs only represent 0.5% of the totalretinal population (Jeon et al., 1998) so that RGC changes are notreflected in the total retinal lysates. As retinas aged, however, we didmeasure a significant difference in whole retinal lysates after 7-12 weeks. This decrease in mitochondrial proteins in CKO retinascorresponds well with the morphological changes in mitochondriaobserved in other cell types of the retina (Fig. S3). Our western blotanalysis suggests that the number of total mitochondria in the CKOretinas was unchanged, as the expression of mitochondrial loadingcontrol porin/voltage-dependent anion-selective channel protein 1(VDAC1) was the same between control and CKO retinas (Fig. 5A).In support of this, IHC analysis of the RGC layer shows comparablelevels of VDAC expression between control and CKO retinas at3 months (Fig. S4), also suggesting that the total retinal mitochondrialnumber is unaffected by the loss of IKAP.

To assess changes in the mitochondrial membrane potential, webriefly treated freshly isolated 3 month retinal explants withMitoTracker Red CMXRos, a dye retained by healthy mitochondria

4

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

that have a normal membrane potential. MitoTracker images of theCKO RGC layer show a dramatic reduction in fluorescence signal(Fig. 6A,B), indicating a significant loss of mitochondrial membranepotential in these cells. To identify the mechanisms underlying the

mitochondrial depolarization, we measured mitochondrial complex Ifunction at 2.5 months. Complex I was chosen because its activitywas decreased significantly in othermitochondrial optic neuropathies,such as LHON (Brown et al., 2000; Carelli et al., 1997).Wemeasured

Fig. 3. Loss of IKAPdoes not cause degeneration of retinal neurons other thanRGCs. (A) H&E staining of 15 month retinal cross-sections at central (0.5 mmfrom the optic nerve head, ONH), middle (1 mm from ONH) and peripheral (1.5 mm from ONH) retina. Central retinas, where Pax6-Cre is not expressed, do notshow any morphological differences between control and CKO mice. In the middle-peripheral retinas of CKO mice, where Pax6-Cre is active, reduction ofRGCs in the ganglion cell layer (GCL) is apparent, whereas other cell types of the retina do not display any abnormal number and morphology. (B) The number ofrows of photoreceptor nuclei in the outer nuclear layer was counted at 0.25 mm from the ONH in the temporal and nasal retinas at 15 months. There was nodifference in the photoreceptor number between control and CKO retinas. The number was compared at the same distance with Student’s t-test (n=6 for control;n=5 for CKO). For all points, P>0.05 (not significant). (C) IHC of cilia marker PKD2L-1(red) shows no abnormality in photoreceptor cilia structure betweenthe inner segment and outer segment at 9 months. (D,E) The number of ChAT+ cholinergic amacrine cells in 6 month retinas was counted at 1 mm from the ONHin each quadrant. Representative images of the temporal inner nuclear layer (INL) and GCL ChAT+ cells (red) are shown (D). There was no difference in thenumber of ChAT+ cholinergic amacrine cells in either INL or GCL (E). P>0.05 with Student’s t-test for all points (n=5). T, temporal; S, superior; N, nasal; I, inferior.Scale bars: 50 µm (A,C) and 100 µm (D).

5

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

a 22%decrease in complex I function in theCKO retinalmitochondriacompared with controls (Fig. 6C). This would be an underestimateof complex I reduction in an affected CKO cell, as the central halfof the CKO retinas were not affected by Pax6-Cre-mediatedrecombination of Ikbkap and were essentially the same as controlretinas (Fig. 1). In support of this loss in complex I activity, the totalcellular ATP concentration per retina was significantly decreased(60% decrease) in CKO retinas compared with controls (Fig. 6D).As most cellular ATP is generated in mitochondria, the resultsindicate impaired mitochondrial function in the CKO retinas. Insummary, we demonstrate that there are significant reductions inmitochondrial integrity, membrane potential and function in theretina in the absence of IKAP.

DISCUSSIONFD is caused by a point-mutation in IKBKAP/ELP1, a gene thatencodes an elongator complex subunit, IKAP/ELP1. Although FDpatients suffer from many congenital and progressive neuropathies,

progressive blindness owing to the loss of RGCs is one of the mostdebilitating phenotypes suffered by patients as they age. To determinewhether autonomous loss of IKAP function in the retina wouldrecapitulate the human FD optic neuropathy, we generated a newmouse model in which the targeted deletion of Ikbkap was restrictedsolely to the retina with Cre expression in all retinal cell types (Fig. 1)(Marquardt et al., 2001). Pax6-Crewas able to abolish 70% of IKAPexpression in the retina (Fig. 1). Interestingly, Ikbkap CKO retinasshowed normal retinal development but demonstrated progressiveRGC degeneration and optic nerve atrophy starting at 1 month of age(Fig. 2), whereas the survival of other retinal neurons was unaffected(Fig. 3). These findings indicate that not only is IKAP not required forretinal development, but in adulthood only RGCs, and not otherretinal cell types, depend on its function for survival. Finally, wedemonstrate here that Ikbkap-deficient retinal neurons, includingRGCs, exhibit a loss of mitochondrial integrity and function anddecreased levels of ATP (Figs 4-6), suggesting that mitochondrialdysfunction is one of the causes of RGC death in FD.

Fig. 4. Integrity of the mitochondrial membranewas lost in CKO RGCs. Mitochondria of RGCswere visualized using TEM at (A-C) 1 month and(D-F) 2.5 months. At both 1 and 2.5 months, CKOmitochondria (B,C,E,F) show a disrupted double-membrane structure (arrows) compared withcontrols (A,D), indicating that loss of IKAP led tomorphological impairment of mitochondria inRGCs. Representative images are shown.Asterisks indicate mitochondria. Squared regionsin A-F are shown enlarged in A′-F′.

6

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

The retinal phenotype observed in our CKO mice recapitulatesthe human FD pathology. Although in the PNS, IKAP deficiencyclearly affects neuronal development (Abashidze et al., 2014;George et al., 2013; Hunnicutt et al., 2012; Jackson et al., 2014), wedid not observe any developmental defects in the CKO retinas. Thisfinding suggests that PNS and CNS neurons have differentsusceptibility to the loss of IKAP, and that there are clearly twodistinct stages of neuronal loss in the disease: developmental versusadult progressive degeneration. Analysis of the vision of FD patients

revealed that loss of the RGCs and optic nerve fiber, especially inthe maculopapillary region (temporal retina), is the cause of theprogressive blindness in FD (Mendoza-Santiesteban et al., 2014,2017). As also seen in our CKO mouse retinas, survival of retinalneurons other than RGCs is not compromised in the retinas of FDpatients. Interestingly, a recent report on postmortem human FDretinas showed degenerating mitochondria in the temporal portionof the optic nerve head pre-laminar region (Mendoza-Santiestebanet al., 2017). This observation corresponds well with our findingthat mitochondrial morphology and function are impaired in retinalneurons in the absence of Ikbkap (Figs 4-6). In addition, RGCdegeneration was subtype-specific, with melanopsin+ RGCs beingresistant to death even in the absence of Ikbkap (Fig. 2), confirmingour previous finding in a mouse model in which Ikbkap was deletedfrom the majority of the CNS and PNS (Ueki et al., 2016).Melanopsin+ RGCs are also preserved in the retinas of human FDpatients (Mendoza-Santiesteban et al., 2017). Our previous analysisconducted in an FD mouse model, in which Ikbkap was deleted inthe CNS and PNS using a Tuba1α-Cre, also showed progressiveRGC degeneration (Chaverra et al., 2017; Ueki et al., 2016). Withthat model, however, we were unable to ascertain whether thedemise of RGC was the direct or indirect consequence of loss ofRGC Ikbkap, given that the Tuba1α-Cre Ikbkap CKO modelexhibited widespread PNS and CNS neurodegeneration, whichcould indirectly affect the integrity of the neural retina (Chaverraet al., 2017; Ueki et al., 2016). To overcome these complications, wegenerated a retina-specific Ikbkap CKO and demonstrate here thatRGCs autonomously require Ikbkap for their survival. In summary,our retina-specific Ikbkap CKO is an excellent FD model to studymechanisms of RGC degeneration and to test potential therapeuticsin otherwise healthy mice.

Why neurons die in the absence of IKAP is not fully resolved, norhave we thoroughly identified its function(s) in neurons. In othersystems, loss of IKAP also impairs target innervation, exocytosis,cytoskeletal organization and axonal transport, which might directlyand/or indirectly lead to cell death (Abashidze et al., 2014; Chaverraet al., 2017; Close et al., 2006; George et al., 2013; Hunnicutt et al.,2012; Jackson et al., 2014; Johansen et al., 2008; Lefler et al., 2015;Naftelberg et al., 2016; Naumanen et al., 2008; Tourtellotte, 2016).IKAP/ELP1 is the scaffolding subunit of the six-subunit elongatorcomplex, which is required for the translation of codon-biasedgenes. The wobble uridine (U34) of tRNAs that recognize both AA-and AG-ending codons is modified by the addition of both athiol (s2) and a methoxy-carbonyl-methyl (mcm5). This doublemodification enhances the translational efficiency of AA-endingcodons (Bauer and Hermand, 2012; Chen et al., 2009b; Huang et al.,2005). Importantly, these specific wobble uridine modifications intRNA are reduced in FD patients (Karlsborn et al., 2014) and FDmice (Goffena et al., 2018). One important class of mRNAs that wediscovered were codon-biased, and hence their proteins wereexpressed at lower levels in Ikbkap CKO peripheral neurons, arethose that repair DNA damage (Goffena et al., 2018). Interestingly,mitochondrial DNA can also be damaged in neurodegenerativedisease (Van Houten et al., 2016) and, if unrepaired, can lead to anelevation in ROS and trigger apoptosis. Furthermore, in thedeveloping PNS, we have shown that Ikbkap−/− neurons diethrough a p53/activated caspase-3-mediated apoptosis (Georgeet al., 2013). Whether this death results from the direct or indirecteffects of impaired tRNA modification is still unresolved, however.Finally, a study in yeast shows that wobble uridine modification bythe elongator complex (Elp1-6) is required for mitochondrialfunction under stress conditions (Karlsborn et al., 2014; Tigano

Fig. 5. Progressive loss of mitochondrial membrane integrity in the CKOretinas. Retinas of the indicated ages (2, 4, 7 and 12 weeks) were collectedand western blot analysis performed using the mitochondrial membraneintegrity antibody cocktail. The cocktail included protein markers of the fourmitochondrial compartments. (A) Representative blot. There was no differencein the expression of protein markers of the four mitochondrial compartments at2 weeks (P14, at the completion of retinal development) between control andCKO retinas. A significant reduction in cyclophilin D and cytochrome c in CKOretinas was detected as the retina ages, suggesting progressive disruption inmitochondrial structure. IM, inner membrane; OM, outer membrane; IMS,intermembrane space. Equal amounts of retinal lysate proteins were loaded.(B) Western blot quantitation for cyclophilin D. There was a progressivedecrease in cyclophlin D and a significant decrease in its expression wasdetected in CKO retinas at 7 weeks. *P<0.05 with Student’s t-test; 2 weeks(P=0.26; n=5), 7 weeks (P=0.031; n=3 for control; n=6 for CKO) and 12 weeks(P=0.018; n=4). (C) Significant reduction in cytochrome c expression wasobserved in CKO retinas compared with controls by 12 weeks. *P<0.05 withStudent’s t-test; 2 weeks (P=0.31; n=5), 7 weeks (P=0.080; n=3 for control;n=6 for CKO) and 12 weeks (P=0.016; n=4).

7

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

et al., 2015). Thus, although it is clear that the loss of IKAP andelongator function cause intracellular stress, including severeimpairments in mitochondrial morphology and function,elucidating the exact cellular and molecular pathways connectingthe loss of elongator function to mitochondrial damage andneuronal death in FD requires future study.Recently, we demonstrated that mitochondria of Ikbkap CKO

embryonic PNS neurons were fragmented, had disrupted membranepotential and had increased ROS (Ohlen et al., 2017). In this newstudy, we show that mitochondria of Ikbkap-deficient retinalneurons are also morphologically and functionally impaired(Figs 4-6), demonstrating that mitochondrial dysfunction occursin both developing and adult neurons in the PNS and CNS in theabsence of IKAP. Our TEM analysis revealed disruption ofmitochondrial membranes in Ikbkap-deficient retinal neurons(Fig. 4 and Fig. S3). Mitochondrial translocation of the p53 tumorsuppressor protein has been shown to disrupt inner and outermitochondrial membrane integrity (Wolff et al., 2008). In line withthis, given the elevated p53 activity in PNS neurons (George et al.,2013), it is possible that p53 might contribute to the breakdown inmitochondrial membrane integrity in Ikbkap CKO retinal neurons.Furthermore, we demonstrate here that Ikbkap CKO retinas had

decreased complex I activity and ATP levels compared with controlretinas, which might explain the poor growth, rhabdomyolysis andreduced muscle tone observed in FD patients (Axelrod, 2002; Rileyet al., 1949). Intriguingly, a muscle biopsy conducted on an FDpatient revealed a severe impairment of mitochondrial complex I, IIIand IV activity (A.S., personal communication). Decreased ATPlevels have also been detected in the optic nerve of amouse glaucomamodel, which had experienced RGC degeneration owing to elevatedintraocular pressure (Baltan et al., 2010). Interestingly, reduced levelsof ATP are a common hallmark of the major neurodegenerativedisorders, including Alzheimer’s disease, Parkinson’s disease andALS (Pathaket al., 2013).Although totalmitochondrial protein levels

were not reduced in the IkbkapCKO retina, ATP levels were reducedby 60% which is commensurate with the 70% loss of IKAP protein.What is surprising, however, is that despite Cre being expressed inapproximately 90% of retinal cells in the peripheral retina, only 40-60% of the RGCs die and none of the other retinal cell types die. Thisis similar to observations made in the embryonic dorsal root gangliawhere approximately half of the tropomyosin receptor kinase A(TrkA+) nociceptors and thermoreceptor subpopulation die in theabsence of IKAP, yet the TrkC+ subset (which comprise theproprioceptors) do not die. Thus, for reasons we do not currentlyunderstand, different neuronal populations have a differentialdependency on the IKAP protein for their survival. Although theseresilient neurons can survive, theymight not be as healthy as Ikbkap+/+

cells, as evidenced by the breached mitochondrial morphology inseveral amacrine cells (Fig. S3 and Fig. 3D,E); thus, these neuronsmight produce less ATP, which explains the 60% reduction in ATPlevels observed in the CKO retina. Given the normal cellularvariability in ATP production, our measurement of total retinal ATPcontent would not accurately reflect any changes in ATPconcentration per single cell. What is intriguing, however, is thatrecent work has demonstrated that, although ATP concentrations in ahealthy cell are in the millimolar range, only micromolarconcentrations of ATP are required for bioenergetics. By contrast,an intracellular concentration of ATP that is 1000-fold higher thanrequired enables it to function as a hydrotrope to keep proteins insolution and dissolve protein aggregates that form within the cytosol(Patel et al., 2017). As protein aggregation is a common hallmark ofall major neurodegenerative diseases, and has been shown to occur inother elongator mutant systems (Laguesse et al., 2015), it isconceivable that the reduction in ATP we demonstrate in Ikbkap−/−

neurons might exacerbate intracellular stress by triggering proteinaggregation. Hence, these reducedATP levels might contribute to theprogressive degeneration of several cell types in FD as patients age, atopic to be pursued in future studies.

Fig. 6. Loss of IKAP causesmitochondrial membrane depolarization, impairs complex I function and decreases cellular ATP levels. (A) Freshly isolated3-month old retinas were treated briefly with MitoTracker RedCMXRos and Hoechst (nuclear marker), fixed and flatmounted for confocal imaging. Representativeimages of the RGC layer are shown. The signal intensity of MitoTracker Red CMXRos correlates with healthy mitochondrial membrane potential. Scale bar:50 µm. (B) ImageJ analysis revealed a 50% decrease in MitoTracker Red CMXRos signal intensity in the CKO RGC layer compared with controls, indicating thattherewas significant loss of mitochondrial membrane potential in the cells of the RGC layer in CKOmice. *P=0.00026 with Student’s t-test (n=5). (C) Mitochondriaof 2.5 month control and CKO retinas were isolated and mitochondrial complex I function was measured. CKO retinas had a 22% reduction in complex I activitycompared with controls. *P=0.0025 with Student’s t-test (n=5). (D) Total ATP of the control and CKO retinas was measured from the total retinal lysates at2.5 months. Total ATP content of the CKO retina was 40% that of the control. *P=0.0053 with Student’s t-test (n=5).

8

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

Two retinal disorders, LHON andDOA, and FD share remarkablesimilarity in their disease progression: all are characterized by loss ofvision owing to slow, progressiveRGCdegeneration (Fig. 2) (Carelliet al., 2009; Chevrollier et al., 2008; Kirches, 2011; Mendoza-Santiesteban et al., 2017; Newman and Biousse, 2004; Ueki et al.,2016; Yu-Wai-Man et al., 2011; Zanna et al., 2008). Both LHONand DOA are mitochondrial diseases: the genes that are mutatedfunction in mitochondria. LHON is generally caused by a point-mutation in one of three mitochondrial genes, all three of whichencodemitochondrial complex I subunits (Howell et al., 1991; Johnset al., 1992; Wallace et al., 1988); the LHON retina has decreasedcomplex I activity and reduced ATP synthesis that results inprogressive RGC degeneration. In DOA, mitochondrial fusion isimpaired. Interestingly, in both LHON and DON, melanopsin-containing RGCs are spared, despite the extensive degeneration ofconventional RGCs (La Morgia et al., 2011, 2010; Moura et al.,2013) as seen in FD mouse retinas (Fig. 2) (Ueki et al., 2016) andFD patients (Mendoza-Santiesteban et al., 2017). Mitochondrialdysfunction is also implicated in glaucoma, which is characterizedby RGC death (Kim et al., 2015; Osborne and del Olmo-Aguado,2013). Together, we conclude that mitochondrial dysfunction is oneof the causes of RGC degeneration in FD.To our surprise, despite their mitochondrial integrity and function

also being impaired due to loss of Ikbkap, cell types other thanRGCs were spared from death in the CKO retina (Fig. 2). Our TEManalysis revealed that the mitochondrial membrane integrity ofamacrine cells was also progressively disrupted, but at a reducedpace and scale compared with that of RGCs (Fig. S3). Significantreduction in mitochondrial membrane proteins, complex I functionand ATP content was observed from total retinal lysates ormitochondria of CKOs, compared with controls (Figs 5 and 6).This observation suggests that mitochondria of all retinal cells thatexpress Cre (50-85%) in the CKO retina (Fig. 1) are affected, asRGCs only account for 0.5% of retinal cell populations (Jeon et al.,1998). In support of this, mutations in the mitochondrial gene/protein in LHON and DOA also cause degeneration of RGCs, butnot of the other retinal neurons (Yu-Wai-Man et al., 2011). WhyRGCs of all retinal cell types are the most vulnerable to loss ofIkbkap is unknown. One explanation is that the RGCs have a highenergy demand and are vulnerable to mitochondrial dysfunctionowing to their unique anatomy: RGC axons are only myelinatedbeyond the lamina cribosa (where the optic nerve starts), andtherefore propagation of action potentials is less efficient andrequires more energy in the unmyelinated prelaminar region of theRGC axons (Morgan, 2004). The brain consumes 20% of the basalmetabolic rate (Clarke and Sokoloff, 1999), with the visual systembeing one of the most energy-consuming systems. Moreover, theretina is one of the tissues with the highest oxygen demand (Ameset al., 1992; Anderson and Saltzman, 1964; Niven and Laughlin,2008). Cytochrome c oxidase (COX) is a terminal enzyme of themitochondrial electron-transport chain and is necessary for ATPproduction (Wikstrom, 1977). Therefore, the level and activity ofCOX reflects energy demand. In retina, strong COX expression andactivity is detected in the nerve fiber layer, the unmyelinated RGCaxon bundle within the retina, and in the prelaminar and laminarregions of the optic nerve (Andrews et al., 1999). Theseunmyelinated regions of RGC axons make them especiallyvulnerable to energy deprivation (Andrews et al., 1999; Morgan,2004), causing RGCs to degenerate under metabolic dysfunction. Infact, in FD, LHON and DOA patients, RGCs of the moremetabolically taxed temporal retinas are the first to die, beforepan-retinal RGC degeneration is observed (Mendoza-Santiesteban

et al., 2014, 2017;Wakakura et al., 2009; Yu-Wai-Man et al., 2014).In addition, among all the retinal neurons, IKAP expression ishighest in the RGCs (Ueki et al., 2016), suggesting that RGCsmightbe more dependent on normal IKAP function than other retinalneurons. In summary, our data clearly indicate that RGCs are thecell type to target with potential treatments for FD blindness.

Whether the loss ofmitochondrial integrity and function observedin our CKO retinas is a direct cause of RGC death or not remainscurrently unresolved. However, our work on Ikbkap-deficient dorsalroot ganglia (DRG) neurons demonstrates that improvement ofmitochondrial function in FD DRG neurons leads to increasedneuronal survival (Ohlen et al., 2017), suggesting that mitochondriaare promising targets for preventing RGC loss in FD. Given theremarkably similar phenotypes between FD, LHON and DOA,understanding the mechanisms of RGC loss in FD, and identifyingpotential therapeutics, might help mitigate progressive RGC loss notonly in FD, but also in other mitochondrial optic neuropathies.

MATERIALS AND METHODSMiceAll mice were housed in the Animal Resource Center at Montana StateUniversity and protocols were approved by the Montana State UniversityInstitutional Animal Care and Use Committee. Retina-specific Ikbkap CKOmice were generated by crossing Ikbkap floxed (International KnockoutMouse Consortium) and αPax6 promoter-driven Cre (Pax6-Cre) mice(Marquardt et al., 2001), which were a gift from the original founders (DrsPeter Gruss and Ruth Ashery-Padan). Pax6-Cre is expressed in the retinalprogenitors, allowing Ikbkap deletion from all the retinal neurons andMüller glia (Marquardt et al., 2001). Both male and female mice were usedat the age indicated, and littermate Pax6-Cre−;Ikbkapflox/flox mice were usedas controls. To analyze Cre expression in the retina and to determineaffected retinal regions, Pax6-Cre mice were crossed to Rosa-EGFP Crereporter mice (Jackson Laboratory, stock #004077).

Western blotsRetinas were collected at the age indicated and standard westernblot procedures were performed. An equal amount of total protein wasloaded in each well. Primary antibodies used were as detailed: anti-IKAP(1:2000; Anaspec #AS-54494), mitochondrial membrane integrity cocktail(1:1000; Abcam #ab110414) and anti-β-actin (1:10,000; Santa CruzBiotechnology #sc-47778). Quantitation of blots was performed usingImageJ and values were normalized using a loading control. Statisticalanalysis was performed using Student’s t-test. Data are consideredsignificant when P<0.05.

Immunohistochemistry, RGC counts and H&E stainingMice were euthanized with CO2 and the eyes were marked with a greentattoo dye on the temporal surface. IHC and confocal imaging wereperformed as described (Ueki et al. 2016). Primary antibodies used were asdetailed: anti-GFP (1:5000; Abcam #ab13970), anti-Brn3 (1:250; SantaCruz Biotechnology #sc-6026), anti-RBPMS (1:500; PhosphoSolutions#1832-RBPMS), anti-melanopsin (1:5000; Advanced Targeting SystemsAB-N38), anti-PKD2L-1 (1:500; Millipore #AB9084) and anti-ChAT(1:200; Millipore #AB114P). For Brn3+ or melanopsin+ RGC counts onflatmounts, confocal images were taken at 1 mm from the optic nerve headin temporal, nasal, superior and inferior retinas, and the number of Brn3+

cells in each image was counted manually. The number of cells in 1 mm2 ofthe retina was calculated and plotted. The optic nerve was carefully cut awayfrom the eyecup after fixation, rinsed in phosphate-buffered saline (PBS)and cryoprotected overnight in PBS plus 20% sucrose. Optic nerves werethen embedded in OCT and cryostat-sectioned transversely at 16 μm. DAPIstaining was performed to visualize the optic nerve, the circumference ofeach optic nerve was measured and plotted. Statistical analysis wasperformed using Student’s t-test. Data are considered significant whenP<0.05. Hematoxylin and Eosin (H&E) staining and photoreceptorcounting were performed according to Ueki et al. (2016).

9

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

Transmission electron microscopyEyes were collected and fixed in Karnovsky’s fixative overnight at 4°C,washed in Ringer’s saline and treated with 2% osmium tetroxide in 0.1 Mpotassium-sodium phosphate buffer (PSPB) for 4 h at room temperature.Eyes were rinsed three times with PSPB and the cornea and lens werecarefully removed. The remaining eyecups were cut into four equal piecesunder a dissecting microscope. Eyecups were dehydrated in a gradedconcentration series of ethanol (50-100%) and then in propylene oxide (PO).Infiltration of eyecups was performed in PO/Spurr’s resin (2:1 vol:vol, 2:2vol:vol, then 100% resin; overnight at 4°C for each). Each eyecup quarterwas embedded in Spurr’s resin and polymerized in a 70°C oven overnight.Sectioning was performed using an ultramicrotome with a diamondknife, and sections were collected and placed on mesh copper grids.Sections were then stained with uranyl acetate and lead citrate. TEMimaging was carried out using a LEO 912 (Zeiss) operating at 100 kVaccelerating voltage. Sections from the peripheral retina were imaged andretinal cell types were identified by morphology, location and nuclear size(Jeon et al., 1998).

Mitotracker treatmentRetinas were isolated and incubated with 200 nM MitoTracker RedCMXRos (Thermo Fisher Scientific) and Hoechst 33342 (Thermo FisherScientific) in neurobasal media at 37°C for 30 min. Chloromethyl X-rosamine (CMXRos) does not accumulate in depolarized mitochondria.Retinas were washed with PBS and fixed with 4% PFA for 5 min atroom temperature. After fixation, retinas were washed and then flatmountedon a slide. Confocal imaging was performed immediately. Images (singlefocal plane) of the RGC layer at 1.75 mm from the optic nerve head werecaptured and the intensity of MitoTracker Red CMXRos quantified usingImageJ.

Complex I activityRetinas were pooled and gently lysed with a Dounce homogenizer (20-25 strokes). The mitochondrial fraction was isolated according toDimauro et al. (2012). The concentration of the mitochondrial proteinswas measured using a BCA protein assay kit (Thermo Fisher Scientific)and a 50 µg aliquot of mitochondrial protein was used for the assay.Mitochondrial complex I activity was measured using Complex IEnzyme Activity Microplate Assay Kit (Abcam). Each experiment wasrun in duplicate. The amount of complex I activity in CKO retinas wasplotted compared with control retinas. Statistical analysis was performedusing Student’s t-test.

ATP assayThe amount of ATP was measured using a StayBrite Highly Stable ATPBioluminescence Assay Kit (BioVision). Each experiment was run induplicate. ATP per retina was calculated and plotted compared with controlretinas. Statistical analysis was performed using Student’s t-test.

AcknowledgementsWe thank Susan Brumfield in the Department of Plant Sciences and Plant Pathologyat Montana State University for technical assistance with EM.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: Y.U., F.L.; Methodology: Y.U.; Formal analysis: Y.U., V.S.;Investigation: Y.U.; Writing - original draft: Y.U.; Writing - review & editing: Y.U., F.L.;Supervision: F.L.; Funding acquisition: F.L.

FundingThis work was supported by the National Institutes of Health (R01NS086796 andR43 EY025458) and the Dysautonomia Foundation (F.L.).

Supplementary informationSupplementary information available online athttp://dmm.biologists.org/lookup/doi/10.1242/dmm.033746.supplemental

ReferencesAbashidze, A., Gold, V., Anavi, Y., Greenspan, H. and Weil, M. (2014).

Involvement of IKAP in peripheral target innervation and in specific JNK andNGF signaling in developing PNS neurons. PLoS ONE 9, e113428.

Addis, L., Ahn, J. W., Dobson, R., Dixit, A., Ogilvie, C. M., Pinto, D., Vaags, A. K.,Coon, H., Chaste, P., Wilson, S. et al. (2015). Microdeletions of ELP4 areassociated with language impairment, autism spectrum disorder, and mentalretardation. Hum. Mutat. 36, 842-850.

Ames, A., III, Li, Y. Y., Heher, E. C. andKimble, C. R. (1992). Energymetabolism ofrabbit retina as related to function: high cost of Na+ transport. J. Neurosci. 12,840-853.

Anderson, B., Jr and Saltzman, H. A. (1964). Retinal oxygen utilization measuredby hyperbaric blackout. Arch. Ophthalmol. 72, 792-795.

Anderson, S. L., Coli, R., Daly, I. W., Kichula, E. A., Rork, M. J., Volpi, S. A.,Ekstein, J. and Rubin, B. Y. (2001). Familial dysautonomia is caused bymutations of the IKAP gene. Am. J. Hum. Genet. 68, 753-758.

Andrews, R. M., Griffiths, P. G., Johnson, M. A. and Turnbull, D. M. (1999).Histochemical localisation of mitochondrial enzyme activity in human optic nerveand retina. Br. J. Ophthalmol. 83, 231-235.

Axelrod, F. B. (2002). Hereditary sensory and autonomic neuropathies. Familialdysautonomia and other HSANs. Clin. Auton. Res. 12 Suppl. 1, I2-I14.

Axelrod, F. B., Hilz, M. J., Berlin, D., Yau, P. L., Javier, D., Sweat, V., Bruehl, H.and Convit, A. (2010). Neuroimaging supports central pathology in familialdysautonomia. J. Neurol. 257, 198-206.

Baltan, S., Inman, D. M., Danilov, C. A., Morrison, R. S., Calkins, D. J. andHorner, P. J. (2010). Metabolic vulnerability disposes retinal ganglion cell axonsto dysfunction in a model of glaucomatous degeneration. J. Neurosci. 30,5644-5652.

Bauer, F. and Hermand, D. (2012). A coordinated codon-dependent regulation oftranslation by Elongator. Cell Cycle 11, 4524-4529.

Brown, M. D., Trounce, I. A., Jun, A. S., Allen, J. C. and Wallace, D. C. (2000).Functional analysis of lymphoblast and cybrid mitochondria containing the 3460,11778, or 14484 Leber’s hereditary optic neuropathymitochondrial DNAmutation.J. Biol. Chem. 275, 39831-39836.

Carelli, V., Ghelli, A., Ratta, M., Bacchilega, E., Sangiorgi, S., Mancini, R.,Leuzzi, V., Cortelli, P., Montagna, P., Lugaresi, E. et al. (1997). Leber’shereditary optic neuropathy: biochemical effect of 11778/ND4 and 3460/ND1mutations and correlation with the mitochondrial genotype. Neurology 48,1623-1632.

Carelli, V., La Morgia, C., Valentino, M. L., Barboni, P., Ross-Cisneros, F. N. andSadun, A. A. (2009). Retinal ganglion cell neurodegeneration in mitochondrialinherited disorders. Biochim. Biophys. Acta 1787, 518-528.

Chaverra, M., George, L., Mergy, M., Waller, H., Kujawa, K., Murnion, C.,Sharples, E., Thorne, J., Podgajny, N., Grindeland, A. et al. (2017). The familialdysautonomia disease gene IKBKAP is required in the developing and adultmouse central nervous system. Dis. Model. Mech. 10, 605-618.

Chen, C., Tuck, S. and Bystrom, A. S. (2009a). Defects in tRNA modificationassociated with neurological and developmental dysfunctions in Caenorhabditiselegans elongator mutants. PLoS Genet. 5, e1000561.

Chen, Y.-T., Hims, M. M., Shetty, R. S., Mull, J., Liu, L., Leyne, M. andSlaugenhaupt, S. A. (2009b). Loss of mouse Ikbkap, a subunit of elongator,leads to transcriptional deficits and embryonic lethality that can be rescued byhuman IKBKAP. Mol. Cell. Biol. 29, 736-744.

Chevrollier, A., Guillet, V., Loiseau, D., Gueguen, N., Pou de Crescenzo, M.-A.,Verny, C., Ferre, M., Dollfus, H., Odent, S., Milea, D. et al. (2008). Hereditaryoptic neuropathies share a common mitochondrial coupling defect. Ann. Neurol.63, 794-798.

Clarke, D. D. and Sokoloff, L. (1999). Circulation and energy metabolism in thebrain. In Basic Neurochemistry: Molecular, Cellular and Medical Aspects, 6th edn.(ed. G. J. Siegel, B. W. Agranoff, R. W. Albers, S. K. Fisher and M. D. Uhler), pp.637-699. Philadelphia : Lippincott-Raven Publishers.

Close, P., Hawkes, N., Cornez, I., Creppe, C., Lambert, C. A., Rogister, B.,Siebenlist, U., Merville, M.-P., Slaugenhaupt, S. A., Bours, V. et al. (2006).Transcription impairment and cell migration defects in elongator-depleted cells:implication for familial dysautonomia. Mol. Cell 22, 521-531.

Cohen, J. S., Srivastava, S., Farwell, K. D., Lu, H.-M., Zeng, W., Lu, H., Chao,E. C. and Fatemi, A. (2015). ELP2 is a novel gene implicated inneurodevelopmental disabilities. Am. J. Med. Genet. A 167, 1391-1395.

Cuajungco, M. P., Leyne, M., Mull, J., Gill, S. P., Lu, W., Zagzag, D., Axelrod,F. B., Maayan, C., Gusella, J. F. and Slaugenhaupt, S. A. (2003). Tissue-specific reduction in splicing efficiency of IKBKAP due to the major mutationassociated with familial dysautonomia. Am. J. Hum. Genet. 72, 749-758.

Dietrich, P., Alli, S., Shanmugasundaram, R. and Dragatsis, I. (2012). IKAPexpression levels modulate disease severity in a mouse model of familialdysautonomia. Hum. Mol. Genet. 21, 5078-5090.

Dimauro, I., Pearson, T., Caporossi, D. and Jackson, M. J. (2012). A simpleprotocol for the subcellular fractionation of skeletal muscle cells and tissue. BMCRes. Notes 5, 513.

Dong, J., Edelmann, L., Bajwa, A. M., Kornreich, R. and Desnick, R. J. (2002).Familial dysautonomia: detection of the IKBKAP IVS20(+6T –> C) and R696P

10

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

mutations and frequencies among Ashkenazi Jews. Am. J. Med. Genet. 110,253-257.

Franic, S., Groen-Blokhuis, M. M., Dolan, C. V., Kattenberg, M. V., Pool, R., Xiao,X., Scheet, P. A., Ehli, E. A., Davies, G. E., van der Sluis, S. et al. (2015).Intelligence: shared genetic basis between Mendelian disorders and a polygenictrait. Eur. J. Hum. Genet. 23, 1378-1383.

George, L., Chaverra, M.,Wolfe, L., Thorne, J., Close-Davis, M., Eibs, A., Riojas,V., Grindeland, A., Orr, M., Carlson, G. A. et al. (2013). Familial dysautonomiamodel reveals Ikbkap deletion causes apoptosis of Pax3+ progenitors andperipheral neurons. Proc. Natl. Acad. Sci. USA 110, 18698-18703.

Georgi, S. A. and Reh, T. A. (2010). Dicer is required for the transition from earlyto late progenitor state in the developing mouse retina. J. Neurosci. 30,4048-4061.

Gkampeta, A., Fidani, L., Clarimon, J., Kalinderi, K., Katopodi, T., Zafeiriou,D. and Pavlou, E. (2014). Association of brain-derived neurotrophic factor(BDNF) and elongator protein complex 4 (ELP4) polymorphisms with benignepilepsy with centrotemporal spikes in a Greek population. Epilepsy Res. 108,1734-1739.

Goffena, J., Lefcort, F., Zhang, Y., Lehrmann, E., Chaverra, M., Felig, J.,Walters, J., Buksch, R., Becker, K. G. and George, L. (2018). Elongator andcodon bias regulate protein levels in mammalian peripheral neurons. Nat.Commun. 9, 889.

Howell, N., Bindoff, L. A., McCullough, D. A., Kubacka, I., Poulton, J., Mackey,D., Taylor, L. and Turnbull, D. M. (1991). Leber hereditary optic neuropathy:identification of the same mitochondrial ND1 mutation in six pedigrees.Am. J. Hum. Genet. 49, 939-950.

Huang, B., Johansson, M. J. and Bystrom, A. S. (2005). An early step in wobbleuridine tRNA modification requires the Elongator complex. RNA 11, 424-436.

Hunnicutt, B. J., Chaverra, M., George, L. and Lefcort, F. (2012). IKAP/Elp1 isrequired in vivo for neurogenesis and neuronal survival, but not for neural crestmigration. PLoS ONE 7, e32050.

Jackson, M. Z., Gruner, K. A., Qin, C. and Tourtellotte, W. G. (2014). A neuronautonomous role for the familial dysautonomia gene ELP1 in sympathetic andsensory target tissue innervation. Development 141, 2452-2461.

Jeon, C.-J., Strettoi, E. and Masland, R. H. (1998). The major cell populations ofthe mouse retina. J. Neurosci. 18, 8936-8946.

Johansen, L. D., Naumanen, T., Knudsen, A., Westerlund, N., Gromova, I.,Junttila, M., Nielsen, C., Bottzauw, T., Tolkovsky, A., Westermarck, J. et al.(2008). IKAP localizes to membrane ruffles with filamin A and regulates actincytoskeleton organization and cell migration. J. Cell Sci. 121, 854-864.

Johns, D. R., Neufeld, M. J. and Park, R. D. (1992). An ND-6 mitochondrial DNAmutation associated with Leber hereditary optic neuropathy. Biochem. Biophys.Res. Commun. 187, 1551-1557.

Karlsborn, T., Tukenmez, H., Chen, C. and Bystrom, A. S. (2014). Familialdysautonomia (FD) patients have reduced levels of the modified wobblenucleoside mcm(5)s(2)U in tRNA. Biochem. Biophys. Res. Commun. 454,441-445.

Keren, H., Donyo, M., Zeevi, D., Maayan, C., Pupko, T. and Ast, G. (2010).Phosphatidylserine increases IKBKAP levels in familial dysautonomia cells.PLoSONE 5, e15884.

Kim, K.-Y., Perkins, G. A., Shim, M. S., Bushong, E., Alcasid, N., Ju, S.,Ellisman, M. H., Weinreb, R. N. and Ju, W.-K. (2015). DRP1 inhibition rescuesretinal ganglion cells and their axons by preserving mitochondrial integrity in amouse model of glaucoma. Cell Death Dis. 6, e1839.

Kirches, E. (2011). LHON: mitochondrial mutations and more. Curr. Genomics 12,44-54.

La Morgia, C., Ross-Cisneros, F. N., Sadun, A. A., Hannibal, J., Munarini, A.,Mantovani, V., Barboni, P., Cantalupo, G., Tozer, K. R., Sancisi, E. et al.(2010). Melanopsin retinal ganglion cells are resistant to neurodegeneration inmitochondrial optic neuropathies. Brain 133, 2426-2438.

La Morgia, C., Ross-Cisneros, F. N., Hannibal, J., Montagna, P., Sadun, A. A.and Carelli, V. (2011). Melanopsin-expressing retinal ganglion cells: implicationsfor human diseases. Vision Res. 51, 296-302.

La Torre, A., Georgi, S. and Reh, T. A. (2013). Conserved microRNA pathwayregulates developmental timing of retinal neurogenesis. Proc. Natl. Acad. Sci.USA 110, E2362-E2370.

Laguesse, S., Creppe, C., Nedialkova, D. D., Prevot, P.-P., Borgs, L.,Huysseune, S., Franco, B., Duysens, G., Krusy, N., Lee, G. et al. (2015). Adynamic unfolded protein response contributes to the control of corticalneurogenesis. Dev. Cell 35, 553-567.

Lefler, S., Cohen, M. A., Kantor, G., Cheishvili, D., Even, A., Birger, A., Turetsky,T., Gil, Y., Even-Ram, S., Aizenman, E. et al. (2015). Familial Dysautonomia(FD) human embryonic stem cell derived pns neurons reveal that synapticvesicular and neuronal transport genes are directly or indirectly affected byIKBKAP downregulation. PLoS ONE 10, e0138807.

Marquardt, T., Ashery-Padan, R., Andrejewski, N., Scardigli, R., Guillemot, F.and Gruss, P. (2001). Pax6 is required for the multipotent state of retinalprogenitor cells. Cell 105, 43-55.

Mendoza-Santiesteban, C. E., Hedges, T. R., III, Norcliffe-Kaufmann, L.,Warren, F., Reddy, S., Axelrod, F. B. and Kaufmann, H. (2012). Clinical

neuro-ophthalmic findings in familial dysautonomia. J. Neuroophthalmol. 32,23-26.

Mendoza-Santiesteban, C. E., Hedges, T. R., III, Norcliffe-Kaufmann, L.,Axelrod, F. and Kaufmann, H. (2014). Selective retinal ganglion cell loss infamilial dysautonomia. J. Neurol. 261, 702-709.

Mendoza-Santiesteban, C. E., Palma, J.-A., Hedges, T. R., III, Laver, N. V.,Farhat, N., Norcliffe-Kaufmann, L. and Kaufmann, H. (2017). Pathologicalconfirmation of optic neuropathy in familial dysautonomia. J. Neuropathol. Exp.Neurol. 76, 238-244.

Morgan, J. E. (2004). Circulation and axonal transport in the optic nerve. Eye 18,1089-1095.

Moura, A. L. A., Nagy, B. V., La Morgia, C., Barboni, P., Oliveira, A. G. F.,Salomao, S. R., Berezovsky, A., de Moraes-Filho, M. N., Chicani, C. F.,Belfort, R.Jr et al. (2013). The pupil light reflex in Leber’s hereditary opticneuropathy: evidence for preservation of melanopsin-expressing retinal ganglioncells. Invest. Ophthalmol. Vis. Sci. 54, 4471-4477.

Naftelberg, S., Abramovitch, Z., Gluska, S., Yannai, S., Joshi, Y., Donyo, M.,Ben-Yaakov, K., Gradus, T., Zonszain, J., Farhy, C. et al. (2016).Phosphatidylserine ameliorates neurodegenerative symptoms and enhancesaxonal transport in a mouse model of familial dysautonomia. PLoS Genet. 12,e1006486.

Naumanen, T., Johansen, L. D., Coffey, E. T. and Kallunki, T. (2008). Loss-of-function of IKAP/ELP1: could neuronal migration defect underlie familialdysautonomia? Cell Adh. Migr. 2, 236-239.

Newman, N. J. and Biousse, V. (2004). Hereditary optic neuropathies. Eye 18,1144-1160.

Nguyen, L., Humbert, S., Saudou, F. and Chariot, A. (2010). Elongator - anemerging role in neurological disorders. Trends Mol. Med. 16, 1-6.

Niven, J. E. and Laughlin, S. B. (2008). Energy limitation as a selective pressure onthe evolution of sensory systems. J. Exp. Biol. 211, 1792-1804.

Ochoa, J. G. (2003). Familial dysautonomia (Riley-Day syndrome) may beassociated with epilepsy. Epilepsia 44, 472.

Ohlen, S. B., Russell, M. L., Brownstein, M. J. and Lefcort, F. (2017). BGP-15prevents the death of neurons in a mouse model of familial dysautonomia. Proc.Natl. Acad. Sci. USA 114, 5035-5040.

Osborne, N. N. and del Olmo-Aguado, S. (2013). Maintenance of retinal ganglioncell mitochondrial functions as a neuroprotective strategy in glaucoma.Curr. Opin.Pharmacol. 13, 16-22.

Palma, J.-A., Norcliffe-Kaufmann, L., Fuente-Mora, C., Percival, L., Mendoza-Santiesteban, C. and Kaufmann, H. (2014). Current treatments in familialdysautonomia. Expert Opin Pharmacother. 15, 2653-2671.

Palma, J.-A., Norcliffe-Kaufmann, L., Perez,M. A., Spalink, C. L. andKaufmann,H. (2017). Sudden unexpected death during sleep in familial dysautonomia: acase-control study. Sleep 40, zsx083.

Patel, A., Malinovska, L., Saha, S., Wang, J., Alberti, S., Krishnan, Y. andHyman, A. A. (2017). ATP as a biological hydrotrope. Science 356, 753-756.

Pathak, D., Berthet, A. and Nakamura, K. (2013). Energy failure: does it contributeto neurodegeneration? Ann. Neurol. 74, 506-516.

Reinthaler, E. M., Lal, D., Jurkowski, W., Feucht, M., Steinbock, H., Gruber-Sedlmayr, U., Ronen, G. M., Geldner, J., Haberlandt, E., Neophytou, B. et al.(2014). Analysis of ELP4, SRPX2, and interacting genes in typical and atypicalrolandic epilepsy. Epilepsia 55, e89-e93.

Riley, C. M., Day, R. L., Greeley, D. McL. and Langford, W. S. (1949). Centralautonomic dysfunction with defective lacrimation; report of five cases. Pediatrics3, 468-478.

Schon, E.A. andPrzedborski, S. (2011).Mitochondria: the next (neurode)generation.Neuron 70, 1033-1053.

Simpson, C. L., Lemmens, R., Miskiewicz, K., Broom, W. J., Hansen, V. K., vanVught, P. W. J., Landers, J. E., Sapp, P., Van Den Bosch, L., Knight, J. et al.(2009). Variants of the elongator protein 3 (ELP3) gene are associated with motorneuron degeneration. Hum. Mol. Genet. 18, 472-481.

Slaugenhaupt, S. A., Blumenfeld, A., Gill, S. P., Leyne, M., Mull, J., Cuajungco,M. P., Liebert, C. B., Chadwick, B., Idelson, M., Reznik, L. et al. (2001). Tissue-specific expression of a splicing mutation in the IKBKAP gene causes familialdysautonomia. Am. J. Hum. Genet. 68, 598-605.

Strug, L. J., Clarke, T., Chiang, T., Chien, M., Baskurt, Z., Li, W., Dorfman, R.,Bali, B., Wirrell, E., Kugler, S. L. et al. (2009). Centrotemporal sharp wave EEGtrait in rolandic epilepsy maps to Elongator Protein Complex 4 (ELP4).Eur. J. Hum. Genet. 17, 1171-1181.

Tigano, M., Ruotolo, R., Dallabona, C., Fontanesi, F., Barrientos, A., Donnini, C.and Ottonello, S. (2015). Elongator-dependent modification of cytoplasmictRNALysUUU is required for mitochondrial function under stress conditions.Nucleic Acids Res. 43, 8368-8380.

Tourtellotte,W.G. (2016). Axon transport and neuropathy: relevant perspectives onthe etiopathogenesis of familial dysautonomia. Am. J. Pathol. 186, 489-499.

Ueki, Y., Ramirez, G., Salcedo, E., Stabio, M. E. and Lefcort, F. (2016). Loss ofIkbkap causes slow, progressive retinal degeneration in a mouse model of familialdysautonomia. eNeuro 3, ENEURO.0143-16.2016.

VanHouten, B., Hunter, S. E. andMeyer, J. N. (2016). Mitochondrial DNA damageinduced autophagy, cell death, and disease. Front. Biosci. 21, 42-54.

11

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms

Wakakura, M., Fujie, W. and Emoto, Y. (2009). Initial temporal field defect in Leberhereditary optic neuropathy. Jpn. J. Ophthalmol. 53, 603-607.

Wallace, D. C., Singh, G., Lott, M. T., Hodge, J. A., Schurr, T. G., Lezza, A. M.,Elsas, L. J., II and Nikoskelainen, E. K. (1988). Mitochondrial DNA mutationassociated with Leber’s hereditary optic neuropathy. Science 242, 1427-1430.

Wikstrom, M. K. (1977). Proton pump coupled to cytochrome c oxidase inmitochondria. Nature 266, 271-273.

Wolff, S., Erster, S., Palacios, G. and Moll, U. M. (2008). p53’s mitochondrialtranslocation and MOMP action is independent of Puma and Bax and severelydisrupts mitochondrial membrane integrity. Cell Res. 18, 733-744.

Yu-Wai-Man, P., Griffiths, P. G. and Chinnery, P. F. (2011). Mitochondrial opticneuropathies - disease mechanisms and therapeutic strategies. Prog. Retin. EyeRes. 30, 81-114.

Yu-Wai-Man, P., Votruba, M., Moore, A. T. and Chinnery, P. F. (2014). Treatmentstrategies for inherited optic neuropathies: past, present and future. Eye 28,521-537.

Zanna, C., Ghelli, A., Porcelli, A. M., Karbowski, M., Youle, R. J., Schimpf, S.,Wissinger, B., Pinti, M., Cossarizza, A., Vidoni, S. et al. (2008). OPA1mutations associated with dominant optic atrophy impair oxidativephosphorylation and mitochondrial fusion. Brain 131, 352-367.

12

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033746. doi:10.1242/dmm.033746

Disea

seModels&Mechan

isms