A new molecular link between defective autophagy and erythroid abnormalities in chorea-acanthocytosis

Circulating acanthocytes (spiculated red cells) are a clinical hallmark of chorea-acanthocytosis, one of several hereditary neurodegenerative disorders known as neuroacanthocytosis syndromes (NAs).^1,2  Chorea-acanthocytosis is associated with mutations in the VSP13A gene encoding VSP13A/chorein,^2,3  a protein implicated by cultured cell expression studies in protein trafficking.^4-6  However, despite progress toward elucidation of the molecular defects responsible for chorea-acanthocytosis, the functional contribution of chorein deficiency to chorea-acanthocytosis remains unclear.^2,3 

Chorein is present in normal mature erythrocytes, but it is partially or completely absent in chorea-acanthocytosis red cells.^1,7,8  Studies comparing the human VPS13A gene product, chorein, with homologs in other organisms such as yeast and Dictyostelium discoideum suggest possible involvement of chorein in trafficking of transmembrane proteins or in vesicular sorting systems directing cargo to the prevacuolar compartment (comparable to the human late endosome).¹  A possible link between chorein and autophagic processes has been suggested by the recent report of perturbed autophagic flux with impaired accumulation of cytosolic microtubule-associated protein light chain (LC3) in VPS13a-deficient Dictyostelium discoideum and HeLa cells.⁹  Chorein was recently linked to mitochondrial clearance in yeast, supporting a role of chorein in vesicle trafficking and autophagic processes.¹⁰  Overexpression of chorein in cell lines (MRC5, CO7, and HEK) leads to its vesicular accumulation, supporting the involvement of chorein in cell vesicle trafficking.¹¹  Similar data were recently reported for Vps13C-expressing HeLa cells.¹² 

The current lack of information regarding chorein structure impedes understanding of its role in red cell homeostasis. We recently reported that chorea-acanthocytosis erythrocytes exhibit abnormal phosphorylation of the integral membrane protein and spectrin-based cytoskeletal anchor, band 3, by the Src family tyrosine kinase (SFK) Lyn.^13,14  In addition, the observation of acanthocytes in genetically engineered mice expressing constitutively active Lyn¹⁵  supports the importance of Lyn signaling in the dysregulation of protein–protein interactions underlying the membrane–cytoskeletal network that generates abnormally shaped erythrocytes.

Here, we uncover in chorea-acanthocytosis erythroid cells an association between accumulation of active Lyn and impaired autophagic flux, suggesting a link between chorein and autophagic vesicle trafficking during erythroid maturation. Although chorea-acanthocytosis is a rare disease, it represents an interesting model to explore and characterize new mechanisms involved in red cell proteostasis and erythroid maturation processes and may be useful in understanding other hematologic disorders characterized by altered in autophagy such as β-thalassemia.^16,17 

We studied 16 patients diagnosed with chorea-acanthocytosis based on clinical neurologic manifestations and confirmation by chorein immunoblot analysis and/or detection of VPS13A mutations¹⁸  (Table 1). Details are reported in supplemental Materials and methods, available on the Blood Web site.

Erythrocytes were isolated from whole blood after removal of white blood cells and platelets as previously described.^14,19-21  Red cell membrane (ghost) and cytosol fractions were obtained as previously reported.^22-24  Immunoblot^23,25  and immunoprecipitation (IP)^14,26-28  were carried out as previously reported (see supplemental Materials and methods for details). When required, the IP assays were carried out as required in the absence or presence of the GST-Lyn/SH3 domain or of the inhibitors PP2 or geldanamycin (GA).²⁹ 

Red cell ghosts (3 μg protein) were in vitro phosphorylated as previously reported.^{22,23,25,28,30}  Syk-phosphorylated membranes were used as substrates, as described in De Franceschi et al.¹⁴ 

Lyn activity was detected as previously described.^14,27 

Details are reported in supplemental Materials and methods.^23,31 

Cytosolic proteins from healthy and chorea-acanthocytosis red cells were loaded on a linear glycerol (10-40%) gradient as described in Trentin et al.³²  The tubes were centrifuged 18 hours at 100 000g in a SW60Ti rotor (Beckman) at 4°C, and 18 fractions were collected from the top of each tube.³² 

Details are reported in supplemental Materials and methods.^14,33-38 

Details are reported in supplemental Materials and methods.

Peripheral blood from adult normal volunteers and from chorea-acanthocytosis patients was collected. We analyzed 10 erythroid cultures from the peripheral blood of different normal subjects and 8 erythroid cultures from 4 chorea-acanthocytosis patients (see supplemental Materials and methods for details).^22,39,40 

Details on immunoblot analysis³⁹  are reported in the supplemental Materials and methods.^22,39,40 

Details are provided in supplemental Materials and methods.⁴¹ 

The 2-way analysis of variance algorithm for repeated measures was used for data analysis. Differences of P < .05 were considered significant.

We first confirmed in a larger number of chorea-acanthocytosis patients the increased abundance of active Lyn associated with the red cell membrane (Figure 1A; supplemental Figure 1A).¹⁴  We then evaluated whether tyrosine (Tyr-) phosphorylated band 3 might affect its interaction with ankyrin in chorea-acanthocytosis. We phosphorylated control red cell membrane with exogenous Syk in the presence or absence of exogenous Lyn (supplemental Figure 1B). As shown in supplemental Figure 1B, the amount of ankyrin coimmunoprecipitated with band 3 progressively decreased as a function of increasing band 3 pTyr content and reached its minimum only after sequential phosphorylation of control red cell membranes by Syk and Lyn. Because we previously showed in chorea-acanthocytosis erythrocytes that Lyn-mediated Tyr phosphorylation of band 3 is independent from its primary phosphorylation by Syk, we immunoprecipitated band 3 from chorea-acanthocytosis red cells and observed a low level of coprecipitated ankyrin (Figure 1B). Incubation with Lyn inhibitor PP2 markedly increased ankyrin coprecipitation, suggesting that abnormal activation of Lyn affects the association between band 3 and ankyrin in chorea-acanthocytosis erythrocytes (Figure 1B). To further test this hypothesis, we generated inside-out vesicles stripped of their spectrin-based cytoskeleton (potassium iodide [KI]–stripped, inside-out vesicles [KIOVs]). We found that the amount of ankyrin retained by band 3 in KIOV exposed to increasing ionic strength was markedly reduced in chorea-acanthocytosis KIOV compared with normal KIOV (supplemental Figure 1C). These data, combined with our previous observations,¹⁴  suggest that in chorea-acanthocytosis, the abnormally activated Lyn weakens band 3–based multiprotein complexes anchoring the membrane to the skeleton with generation of acanthocytes. We then explored whether the “toxic” accumulation of membrane-associated Lyn was paralleled by accumulation of active Lyn in cytosol from chorea-acanthocytosis red cells.

Cystosol from chorea-acanthocytosis red cells showed higher levels of activated Lyn than healthy controls (Figure 1C; supplemental Figure 1D). In addition, we found that chorein coimmunoprecipitated with Lyn in healthy red cells but not in chorea-acanthocytosis erythrocytes (Figure 1D). We then evaluated the effect of either PP2 or dasatinib, a synthetic pharmacologic SFK inhibitor, on Lyn activity. As shown in Figure 1E, we found a significant reduction in Lyn activity in healthy and chorea-acanthocytosis erythrocytes by both PP2 and dasatinib. To evaluate the impact of Lyn inhibition on chorea-acanthocytosis red cell phenotype, we fractionated red cells as a function of cell hemoglobin content and cell volume (V/HC), revealing a dense cell fraction present only in chorea-acanthocytosis subjects. Thus, the increased cell hemoglobin concentration mean (CHCM) and hemoglobin distribution width (HDW) in chorea-acanthocytosis might serve as a marker of dense fraction of red cell–containing acanthocytes.^42,43  As shown in Figure 1F, PP2 significantly reduced the fraction of dense red cells in chorea-acanthocytosis (left; blue circle) as supported by significant reductions in CHCM and HDW observed only in chorea-acanthocytosis erythrocytes (supplemental Figure 1E). Dasatinib reversed acanthocytes to almost normal red cell morphology, indicating the key role of abnormally activated Lyn in the chorea-acanthocytosis hematologic phenotype (Figure 1G). No changes were observed in dasatinib-treated healthy red cells (Figure 1G).

Studies in other cell models have shown that the fate of SFK is under strict control of the HSP90/HSP70 chaperone machinery.^44-47  Here, we found reduced expression of Cdc37 and increased expression of HSP70 in the cytosolic fraction of chorea-acanthocytosis red cells, compared with control erythrocytes, whereas HSP90 levels were similar in both cell types (Figure 2A). Because reticulocyte levels were similar in control and chorea-acanthocytosis subjects (data not shown, see also ref. ¹ ), the differences in Cdc37 and HSP70 expression are unlikely related to cell age, but rather to the chorea-acanthocytosis phenotype. Western blot analysis of red cell cytosol fractions from glycerol gradient fractionation revealed a peak of Lyn abundance in complexes of ∼250 kDa in normal red cells. In contrast, chorea-acanthocytosis erythrocytes exhibited a bimodal distribution of Lyn, with peaks at both ∼250 and ∼500 kDa, the latter suggesting presence of an aberrant complex (Figure 2B; supplemental Figure 2A). To evaluate the possible role of chaperone components in this complex, the same glycerol gradient fractions were probed with antibodies directed against HSP90, HSP70, and Cdc37. Whereas HSP90 and Cdc37 profiles were minimally if at all changed in chorea-acanthocytosis red cell cytosol, the chorea-acanthocytosis–associated change in Lyn profile was paralleled by similar substantial changes in the HSP70 profile, suggesting participation of HSP70 in formation of the aberrant Lyn complex (Figure 2B; supplemental Figure 2B). To test this hypothesis, we first evaluated the formation of a high-molecular-weight complex in the presence of GA, a specific HSP 90 inhibitor.

We found Lyn coimmunoprecipitated with HSP90 and HSP70 only in chorea-acanthocytosis red cells (Figure 2C; supplemental Figure 2B), whereas it was undetectable in HSP90 immunoprecipitates from GA-treated chorea-acanthocytosis red cell lysates (Figure 2C). In agreement, HSP90 and HSP70 were undetectable in Lyn immunoprecipitates from GA-treated chorea-acanthocytosis red cell lysates. These findings clearly suggest that active HSP90 and HSP70 are required for binding Lyn. This conclusion is supported by the absence of high molecular complex formation in lysates from both healthy and chorea-acanthocytosis red cells incubated with GA (supplemental Figure 6B). To evaluate the role of Lyn's SH3 domain in the interaction with HSP90 or HSP70, we performed competition assays using recombinant Lyn SH3 domain.^14,29,32  The Lyn SH3 domain serves as an interaction domain for protein partners and as a regulatory domain for Lyn kinase.^14,29,32  Addition of GST-Lyn/SH3 domain failed to disrupt the interaction between Lyn and HSP90 or HSP70 (Figure 2C; supplemental Figure 2B), confirming previous data in other cell models that highlighted the central role of the Lyn catalytic domain in binding HSP90.³²  These data together suggest that active Lyn in chorea-acanthocytosis erythrocytes is sequestered by chaperone machinery for either proteasomal degradation or rescue from degradation.

The modeled structure of the complex of HSP90 and Lyn^48-53  was consistent with the hypothesis of reduced Lyn accessibility to the proteasome system, resulting in Lyn accumulation to chorea-acanthocytosis red cells (supplemental Figure 3). Because previous studies have shown that SFK degradation mainly involves proteasome system,^44-46,54  we evaluated the effects of bortezomib, a selective inhibitor of 20S proteasome.^44-46,54  In chorea-acanthocytosis red cells, bortezomib prevented degradation of Lyn associated with HSP90/HSP70 in high-molecular-weight complexes (Figure 3; supplemental Figures 4 and 5). In the absence of bortezomib, the amount of Lyn associated with HSP90/HSP70 in high-molecular-weight complexes decreased progressively with increasing incubation time (Figure 3; supplemental Figure 5). These findings suggest that chaperonin machinery in chorea-acanthocytosis red cells monitors Lyn to stabilize and prevent its toxic effects related to delayed protesomal degradation.

Accumulation of active Src protected from proteasomal degradation has been shown to be linked with autophagy in other cell models.^55-57  As shown in Figure 4A, Ulk1, Atg7, Atg13, Rab5, and the cytosolic form of LC3 (LC3I) accumulated to higher levels in chorea-acanthocytosis than in healthy red cells, whereas Atg4 and Atg5 were undetectable in both healthy and chorea-acanthocytosis erythrocytes. Based on these findings, we sought candidate Src phosphorylation motifs in the sequence of Ulk1, Atg7, and Atg13. We found stringent-threshold SFK consensus sequence in (1) Ulk1 (Tyr295 for Src) and (2) Agt7 (Tyr602 for Src and Tyr184, Tyr401, and Tyr657 for Lyn). Indeed, we found increased Tyr-phsphorylation state of Atg7 but not of Ulk1 (supplemental Figure 6A). We then evaluated the possible association of Ulk1 and Atg7 with Lyn, and we found that both Ulk1 and Atg7 were present at higher levels in Lyn immunoprecipitates from chorea-acanthocytosis red cells than in healthy controls (Figure 4B-C, upper). To validate this observation, we immunoprecipitated Ulk1 and Atg7 from normal and diseased red cells and immunoblotted them with antibody to active Lyn. As shown in Figure 4B-C (lower), levels of active Lyn (P-Lyn396) coimmunoprecipitated with Ulk1 and with Atg7 were higher from chorea-acanthocytosis red cell cytosols than from healthy controls. To verify the increased activity of Lyn associated with either Ulk1 or Atg7, we immunoprecipitated equal amount of proteins and then carried out the assay for Lyn activity. As shown in Figure 4D, we found greatly increased Lyn kinase activity in both Ulk1 and Atg7 immunoprecipitates from chorea-acanthocytosis red cells compared with those from healthy controls.

Because a previous study showed that Ulk1 requires the HSP90-cdc37 complex to initiate autophagy,⁵⁸  we evaluated the presence of Ulk1 in the high-molecular-mass complexes of chorea-acanthocytosis erythrocyes. Ulk1 but not Atg7 was present together with Lyn in the high-molecular-weight cytosolic complex only in chorea-acanthocytosis (Figure 4E) but not in healthy red cells (data not shown). Treatment with GA prevented the formation of Lyn-containing high-molecular-weight complexes together with Ulk1 (supplemental Figure 6B). These data indicate the functional connection between active Lyn and Atg proteins in chorea-acanthocytosis, suggesting possible impairment of autophagy in chorein-deficient red cells.

The accumulation of Ulk1 and Atg7^58-61  in chorea-acanthocytosis red cells prompted the hypothesis that chorein functions in protein trafficking to the autophagosome. We therefore searched in the chorein primary sequence for candidate LC3-interacting region (LIR) motifs proposed to target autophagy substrates.⁶²  Alignment of the chorein sequence with the W/F/YxxL/I/V consensus identified 9 putatitive LIR sites distributed throughout the protein. Although identification of LIR motifs is not itself indicative of LIR binding propensity, the presence of serine or threonine N-terminally flanked by a negative residue is considered a crucial determinant of a functional LIR.⁶³  Based on these findings, we then explored the possible interplay between chorein and autophagy in chorea-acanthocytosis. Electron microscopic (EM) analysis revealed multivesicular bodies close to the surface membrane or in the process of extrusion as well as ring-shaped intracellular membranes and double membrane fragments (Figure 5A). These cell constituents were present in both acanthocytic and normally shaped chorea-acanthocytosis red cells but were not detected in erythrocytes from healthy controls. In addition, chorein was coimmunoprecipitated with Atg7 from healthy but not from chorea-acanthocytosis red cells, directly linking chorein to the physiologic autophagic machinery (Figure 5B). Previous studies have shown that autophagy plays a key role in vesicle trafficking and mitochondrial clearance during late stage of erythropoiesis and in reticulocytes.^16,41,64  Thus, we used MitoTracker and LysoTracker to follow mitochondrial and lysososmal clearance in the reticulocyte-enriched red cells fraction from healthy and chorea-acanthocytosis subjects incubated for 3 hours in a plasma-like medium.^41,64  As shown in Figure 5C, we found delayed mitochondria (supplemental Figure 6C) and lysosomal clearance (supplemental Figure 6D) in chorea-acanthocytosis red cells compared with those from healthy subjects. These data collectively support a perturbation of autophagic flux in chorein-deficient erythroid cells.

To explore the possible impact of impaired autophagy in chorea-acanthocytosis during erythroid maturation, we studied erythropoiesis using CD34⁺ cells from healthy and chorea-acanthocytosis subjects. As shown in Figure 6A, culture day 14 cell production was significantly reduced in chorea-acanthocytosis compared with healthy controls, corresponding to the late stage of erythropoiesis. Morphologic analyses revealed signs of dyserythropoiesis characterized by bi- and trinucleated cells, hypocondensated nuclei, and greater accumulation of vesicles compared with healthy cells (Figure 6A, lower; see supplemental Figure 6E for quantitation and supplemental Figure 7A for primary FACS data). Electron microscopy of chorea-acanthocytosis erythroid precursors at 14 days of culture showed (1) erythroblasts with pyknotic nuclei, containing variable amounts of cytoplasmic constituents including mitochondria, small and large vesicular elements, and profiles of endoplasmic reticulum (Figure 6B, upper); and (2) cells apparently in the process of losing large portions of cytoplasm and differentiated, acanthocyte-like red cell–containing organellar remnants (Figure 6B, lower). The chorea-acanthocytosis maturation profile at culture day 14 revealed decreased pro-erythroblasts and intermediate erythroblasts associated with increased late erythroblasts compared with healthy controls. This was associated with increased apoptosis of chorea-acanthocytosis erythroblasts compared with healthy controls (Figure 6C-D) and accumulation of active Lyn compared with healthy cells (Figure 6D). These data indicate the presence of dyserythropoiesis associated with accumulation of vesicles and organellar remnants in chorein-deficient erythroblasts.

Because erythroid maturation requires efficient autophagic flux, we evaluated expression of lysosmal-associated membrane protein LAMP1 and p62, a key autophagy component of inclusion bodies and autophagosomes that recognizes damaged proteins.^65,66  As shown in Figure 7A, chorea-acanthocytosis erythroblasts accumulated LAMP1 and p62 to higher levels than did healthy cells, consistent with impaired autophagy. Immunofluorescent analysis revealed more numerous LAMP1 punctae in chorea-acanthocytosis erythroblasts than in healthy cells (Figure 7B; supplemental Figure 7B). LysoTracker staining confirmed that chorea-acanthocytosis erythroblasts exhibited greater retention of lysossomes than healthy cells (Figure 7C). Reduced mitochondrial clearance in chorea-acanthocytosis erythroblasts was also confirmed by MitoTracker staining, in agreement with findings in chorea-acanthocytosis reticulocyte enriched red cell fractions (supplemental Figure 7C). These data collectively indicate an impairment of autophagic flux in chorea-acanthocytosis erythroid cells, resulting in dyserythropoiesis, delayed clearance of mitochondria and lysososmes, and accumulation of autophagy proteins and active Lyn in chorea-acanthocytosis cells.

In this paper, we show for the first time that chorea-acanthocytosis red cells accumulate activated Lyn and autophagy-related proteins to deleterious levels, resulting in retention of membrane remnants and multivesicular bodies. These findings were associated with delayed clearance of mitochodria and lysosome in chorea-acanthocytosis reticulocyte-enriched fractions, indicating impaired autophagy with accumulation of toxic Lyn in chorein-deficient erythroid cells. We also presented the first evidence of the importance of chorein in late-stage erythropoiesis, influencing both autophagy and lysosomal clearance.

In chorea-acanthocytosis erythrocytes, increased levels of active Lyn were not restricted to membrane fractions as in control red cells. Membrane-associated hyperactivated Lyn phosphorylated band 3, weakening the membrane’s interaction with the ankyrin-based multiprotein complex (Figure 1). This finding parallels our previous report showing the effects of Lyn-dependent tyrosine phosphorylation on the interaction of band 3 with β-adducin in chorea-acanthocytosis erythrocytes. Taken together, our data strengthen the proposed role of hyperactive Lyn in altering plasma membrane attachment of multiple cytoskeletal proteins and leading to generation of acanthocytes in chorea-acanthocytosis. Remarkably, the inhibition of Lyn by either PP2 or dasatinib largely reversed the chorea-acanthocytosis hematologic phenotype (Figure 1F-G). In addition, we found that Lyn coimmunoprecipitaed with chorein in healthy erythrocytes. Collectively, these data suggest a perturbation of proteostasis in chorea-acanthocytosis red cells, resulting in accumulation of active Lyn. Indeed, cytosolic active Lyn is stabilized in 500-kDa multiprotein cytosolic complexes that include HSP70 and 90. The presence of HSP70 in high-molecular-weight complexes only in chorea-acanthocytosis red cells most likely reflects to its role as an HSP promoting proteasomal protein degradation.⁴⁶  These data are consistent with evidence from other cell models, in which levels of active SFKs are strictly monitored by HSPs.^45,54  Indeed, the interaction of HSP90 with Fgr and Lck detected in rabbit reticulocytes has suggested a role for HSP in the metabolic fate of erythroid SFKs.^67,68  Moreover, treatment with HSP90 inhibitor GA blocked HSP90 interaction with active Lyn by a mechanism independent of the Lyn-SH3 domain, indicating the involvement of HSP90 in monitoring the active form of Lyn.⁴⁴  In addition, GA also prevented the formation of Lyn-containing high-molecular-weight complexes in chorea-acanthocytosis erythrocytes, further supporting the functional interaction between HSPs and active Lyn in chorea-acanthocytosis (supplemental Figure 5B).

Because the proteasome system is the major route of degradation for HSP client proteins in many cell types,⁴⁶  we evaluated the impact of the proteasome inhibitor bortezomib on Lyn fate in both healthy and chorea-acanthocytosis red cells. Although bortezomib prevented Lyn degradation in both chorea-acanthocytosis and healthy erythrocytes, Lyn degradation in chorea-acanthocytosis erythrocytes was slower than in cells, favoring Lyn accumulation in chorea-acanthocytosis. Thus, perturbation of intracellular signaling associated with accumulation of active Lyn in chorea-acanthocytosis erythrocytes can generate acanthocytes and might be a sign of impaired quality control mechanisms for cellular protein. This is in agreement with the reported reduction in haptoglobin levels in chorea-acanthocytosis patients,³  suggesting that abnormal Lyn-mediated signal transduction might impact in vivo red cell homeostasis in chorea-acanthocytosis subjects.

Studies in other cell models have shown that accumulation of SFKs might be related to perturbation of protein quality control related to autophagy.^56,57,69  Indeed, chorea-acanthocytosis red cells revealed accumulation of autophagy-related proteins Ulk1, Atg7, Atg13, and cytosolic LC3 compared with healthy erythrocytes. A substantial proportion of active Lyn was also largely associated with Ulk1 and Atg7 in chorea-acanthocytosis, but not in healthy red cells. It is of note that the Atg7 Tyr phosphorylation state but not Ulk1 was also increased in chorea-acanthocytosis red cells. Collectively, these findings may reflect functional regulation of Ulk1 and Atg7 through Tyr phosphorylation of their consensus SFK phosphorylation sites or, alternatively, could reflect the ability of Lyn association with Ulk1 and Atg7 to promote their trafficking to the autophagosome, as previously reported for Src in a different cell model.⁵⁶  The observation of chorein coimmunoprecipitation with Atg7 only in healthy erythrocytes links chorein to autophagy. We therefore propose that that chorein may function as an autophagic partner for active Lyn, preventing the accumulation of active Lyn and the perturbation of intracellular signaling. Indeed, the EM evidence of membrane remnants and multivesicular bodies only in chorea-acanthocytosis red cells of both acanthocytic and nonacanthocytic morphologies further supports the proposed link between chorein and autophagy and the involvement of chorein in erythroid maturation processes. Previous studies have shown that mitochondrial clearance is part of autophagy during the erythroid maturation processes and involves Ulk1 with the HSP90–cdc37 complex, as well as with Atg7.^16,41  Chorea-acanthocytosis red cells exhibited accumulation of Ulk1 and Atg7 in concert with delayed mitochondrial and lysosomal clearance in reticulocyte-enriched red cell fractions, supporting the proposed contribution of chorein to autophagic flux. It is of note that Atg4, which has been linked to fusion of autophagosomes to lysosomes during erythropoiesis,⁷⁰  has not been found in chorea-acanthocytosis red cells, suggesting that the absence of chorein may affect initiation of autophagy related to Ulk1- or Atg-7–linked formation of autophagosomes.^16,41  Indeed, chorea-acanthocytosis erythroid precursors exhibit dyserythropoiesis associated with EM-documented retention of vesicles, mitochondria, and profiles of endoplasmic reticulum, along with accumulation of active Lyn (Figure 6). In addition, the accumulation of p62, delivering proteins to autophagosomes,⁷¹  and of lysosomal membrane protein LAMP1,⁶⁶  further links the absence of chorein with impaired autophagy and accumulation of proteins such as Lyn that require controlled degradation. Indeed, late-stage chorea-acanthocytosis erythroblasts exhibited delayed clearance of mitochondria and lysosomes, indicating deficient autophagy during erythroid maturation in chorea-acanthocytosis (supplemental Figure 7). Although mitochondrial clearance during erythroid maturation is mainly mediated by autophagy, our data do not exclude the involvement of alternative autophagy-independent pathway(s) for mitochondrial degradation.⁷² 

In conclusion, we show for the first time that chorea-acanthocytosis red cells exhibit accumulation of toxic active Lyn accompanied by accumulation of autophagy-related proteins, resulting in retention of membrane remnants and multivesicular bodies in chorea-acanthocytosis erythrocytes (supplemental Figure 8). The delayed proteasomal degradation of Lyn via high-molecular-weight complex formation with HSPs and autophagy-related protein Ulk1 represents a novel element of proteostasis in normal and chorea-acanthocytosis red cells. The link between impaired autophagy and chorein deficiency in human erythroid cells enhances our understanding of chorea-acanthocytosis and of chorein's homeostastic role promoting intracellular degradation of cytotoxic and/or damaged proteins and organellar constituents (supplemental Figure 8). The chorea-acanthocytosis–associated impairment of autophagy leads to dyserythropoietic features. The accompanying delayed clearance of mitochondria and lysosomes and the accumulation of active Lyn dysregulates red cell homeostasis and underlies the acanthocytosis of syndromic chorea-acanthocytosis.

The authors thank the patients and control subjects for donating their blood and Sara Santos Franco who contributed to some preliminary data on chorea-acanthocytosis erythropoiesis. This paper is dedicated to Glenn Irvine, who strongly supported research on NA disorders and devoted his energies to generating an international research network on NA.

This work was supported by Telethon grant GPP13005 (L.D.F.); an NA advocacy grant (L.D.F.); Fondo Universitario di Ricerca (L.D.F.); Federal Ministry of research and education Germany grant ERARE12-081-European Multidisciplinary Initiative on Neuroacantocytosis (EMINA)-2, under the frame of E-Rare-2, the ERA-Net for Research on Rare Diseases (A.H.); and the Doris Duke Charitable Foundation (S.L.A.). Western blot analysis of chorein was performed by G. Kwiatkowski and B. Bader, with financial support of NA advocacy and European Agency–net E-Rare consortium EMINA (BMBF01 GM1003) in the laboratories of H. Kretzschmar/A. Giese (Neuropathology) and A. Danek (Neurology) at Ludwing-Maximilians-Universitat Munich (Germany).

Contribution: L.D.F. and A.M.B. designed the experiments and analyzed the data; S.L.A. contributed to study design and the writing of the manuscript; F.L., A. Siciliano, and A.M. performed the experiments; A.K.S. performed the protein structure analysis; M.B. carried out in silico analysis; A.M.B., E.T., and F.Z. performed part of the Lyn analysis, analyzed the data, and contributed in writing the manuscript; D.B. and C.Z. performed electron microscopy and ultrastructural analysis; A.D., R.H.W., A.H., B.B., and A. Storch performed neurological diagnosis and provided blood samples; and all authors reviewed the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lucia De Franceschi, Department of Medicine, University of Verona, AOUI-Verona, Policlinico GB Rossi, P.Le L. Scuro 10, 37134 Verona, Italy; e-mail: lucia.defranceschi@univr.it.