Molecular Pathogenesis of Cystinosis: Effect of CTNS Mutations on the Transport Activity and Subcellular Localization of Cystinosin

Human Molecular Genetics, 2004, Vol. 13, No. 13 1361–1371 DOI: 10.1093/hmg/ddh152 Advance Access published on May 5, 2004 Molecular pathogenesis of cystinosis: effect of CTNS mutations on the transport activity and subcellular localization of cystinosin

Vasiliki Kalatzis1,*, Nathalie Nevo1,{, Ste´phanie Cherqui1,{, Bruno Gasnier3 and Corinne Antignac1,2 Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021 1Inserm U574 and 2Department of Genetics, Hoˆpital Necker-Enfants Malades, 149 rue de Se`vres, 75015 Paris, France and 3CNRS UPR 1929, Institut de Biologie Physico-Chimique, 13 rue Pierre et Marie Curie, 75005 Paris, France

Received March 10, 2004; Revised and Accepted April 26, 2004

Cystinosis is an inherited disorder characterized by defective lysosomal efflux of cystine. Three clinical forms (infantile, juvenile and ocular cystinosis) have been described according to the age of onset and severity of the symptoms. The causative gene, CTNS, encodes a seven transmembrane domain protein, cystinosin, which we recently identified as a H1-driven cystine transporter using an in vitro transport assay. In this study, we explored the relationship between transport activity and intracellular localization of cystinosin mutants and their associated clinical phenotype. Thirty-one pathogenic mutations (24 missense mutations, seven in-frame deletions or insertions) were analysed. Most of the mutations did not alter the lysosomal localization of cystinosin, although three partially mislocalized the protein independently of its C-terminal sorting motif, thus confirming the presence of an additional sorting mechanism. Sixteen of 19 mutations associated with infantile cystinosis abolished transport, whereas three of five mutations associated with juvenile or ocular forms strongly reduced transport, in agreement with the milder clinical phenotype. Five atypical, unclassified or misclassified mutations could be clarified using the transport data and additional genetic information. Overall, our data demonstrate that, excluding premature termination of cystinosin, impaired transport is the most frequent cause of pathogenicity, with infantile cystinosis generally resulting from a total loss of activity. Thus the transport assay could be used as a prognostic tool when novel mutations are identified.

INTRODUCTION Cystinosis (MIM 219800), an autosomal recessive mono- genic disease with a low incidence (1 in 200 000 live births), Lysosomes are filled with hydrolases involved in the degra- is characterized by an intra-lysosomal accumulation of cystine. dation of diverse macromolecules. Specific transporters in There exists a spectrum of disease phenotypes, but affected the lysosomal membrane then export the hydrolysed products individuals are generally grouped into three clinical forms, to the cytosol. Abnormal lysosomal storage can occur in two based on the age of onset and severity of the symptoms (4). ways: a defective hydrolase can result in the accumulation The most severe form, infantile cystinosis, generally appears of an undegraded macromolecule or a defective transporter between 6 and 12 months with a proximal renal tubulopathy can result in the accumulation of a degraded substrate. The (the Fanconi syndrome characterized by fluid and electrolyte former scenario gives rise to over 40 different clinical dis- loss, poor growth and rickets) that, in the absence of treatment, orders termed ‘lysosomal storage disorders’ (LSD), classified leads to end stage renal disease (ESRD) by 10 years. Within according to the macromolecule that accumulates (1). The the first 2 years of age, corneal cystine crystals also give rise to latter causes a subset of LSD referred to as lysosomal trans- a severe and painful photophobia. Continuous widespread port disorders (reviewed in 2), such as sialic acid storage cystine accumulation eventually leads to retinal, endocrino- diseases (3) and cystinosis. logical, hepatic, gastrointestinal, muscular and neurological

*To whom correspondence should be addressed at: Institut de Ge´ne´tique Mole´culaire de Montpellier, CNRS UMR 5535, 1919 Route de Mende, 34293 Montpellier, France. Tel: þ33 467613674; Fax: þ33 467040231; Email: [email protected] {These authors contributed equally to the work.

Human Molecular Genetics, Vol. 13, No. 13 # Oxford University Press 2004; all rights reserved 1362 Human Molecular Genetics, 2004, Vol. 13, No. 13 anomalies. Juvenile cystinosis (MIM 219900) is generally RESULTS characterized by an adolescent onset of photophobia and Two assays were performed in parallel to investigate the mol- glomerular renal impairment, but not necessarily renal Fanconi ecular effects of the CTNS mutations. To analyse the effect on syndrome. Ocular non-nephropathic cystinosis (MIM 219750) the lysosomal localization of the protein, cystinosin–EGFP is characterized solely by an adult onset of mild photophobia fusion constructs containing the mutation of interest were without renal anomalies. Atypical forms, where individuals expressed in HeLa and Martin-Darby Canine Kidney present with various clinical features of severity between (MDCK) cells. The wild-type fusion protein localizes to late infantile and juvenile cystinosis also have been recently endosomes and lysosomes in these cells (Fig. 1A and B) (16). described (5,6). To assay the effect on cystine transport, mutations were intro- The gene underlying cystinosis, CTNS, is located on 17p13 duced into an expression plasmid encoding cystinosin deleted (7) and contains 12 exons, the first two of which are non- in its C-terminal targeting motif (cystinosin–DGYDQL). coding (8). The most common mutation associated with Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021 0 COS-7 cells expressing wild-type cystinosin–DGYDQL cystinosis is a 57 kb deletion (8,9) that removes the 5 35 are able to take up [ S]L-cystine from the extracellular region of the gene upstream of, and including, exon 10. medium (17). CTNS mutations have been identified in all forms of the disease and are primarily situated in the coding region (5,6,8,10–14), although three mutations also have been Effect of mutations on localization reported in the undelineated promoter region (15). It has been hypothesized that the underlying differences between The mutations examined in this study are listed according to the clinical forms is because individuals with infantile cysti- the clinical phenotype to which they are associated (Table 1). nosis harbour ‘severe’ mutations (i.e. loss of function) on Twenty-seven of the 31 mutations showed a late endosomal/ both alleles, whereas individuals with milder forms of cysti- lysosomal localization (Table 2; Fig. 1I and J and data not nosis harbour a specific, non-severe mutation (i.e. never shown). Three mutations, Q222R, IVFD343–346del and observed in cases of infantile cystinosis) either on both DVVF346–349del, led to a partial expression of cystinosin alleles or in association with a severe allele (compound het- at the plasma membrane (Fig. 1E and F and data not erozygotes) (5,11,12). shown). Deletion of the GYDQL motif completely relocalized CTNS encodes a 367 amino acid protein named cystinosin these mutant proteins to the plasma membrane (Fig. 1G and H (8). Cystinosin is a lysosomal membrane protein (16) with a and data not shown), in contrast to the wild-type protein and seven transmembrane domain (TM) topology. The C-terminal other mutants, which are partially relocalized by this deletion tail is predicted to be oriented towards the cytosol and the (Fig. 1C, D, K and L and data not shown). Finally, no fluor- highly glycosylated N-terminal region towards the lysosomal escence was observed upon transfection of cystinosin–EGFP lumen. Two signals have been identified as playing a role in harbouring the mutation M1I, suggesting that the fusion the lysosomal sorting of cystinosin: a classic tyrosine-based protein is not translated. GYDQL motif situated in the C-terminal tail and a novel conformational signal situated in the fifth inter-TM loop Effect of mutations on cystine transport (16). Owing to the relative inaccessibility of the lysosomal lumen, we expressed cystinosin at the plasma membrane, by None of the mutants tested transported cystine significantly at deletion of the GYDQL motif, and measured the cellular neutral pH (data not shown), therefore, only data obtained at uptake of cystine at acidic external pH to assay its ‘lysosomal acid pH are presented. cystine efflux’ activity. This in vitro model, equivalent to giant, inside-out lysosomes, showed that cystinosin is the Infantile cystinosis. Of the 19 mutations associated with typical lysosomal cystine transporter and that its activity is driven cases of infantile cystinosis, 14 are missense mutations and by the Hþ electrochemical gradient of the lysosomal five are in-frame deletions or insertions. Seventeen of these membrane (17). mutations are situated within (I133F, S141F, L158P, G169D, This expression assay now represents a unique tool for the W182R, Q222R, S270del, D305Y, G308R, L338P, G339R, functional analysis of CTNS mutations that do not truncate IVFD343–346del, DVVF346–349del and DVEF349–350ins) the protein. We previously showed that the missense mutation or immediately adjacent to (D205N, D205del and S298N) the G308R, situated in the sixth TM and identified in several indi- TMs. Of the remaining two mutations, N288K was situated in viduals with infantile cystinosis (5,10), is sufficient to abolish the middle of the fifth inter-TM loop and M1I affected the the transport activity of cystinosin without altering its subcel- methionine start codon. lular localization (17). This observation accounted for the Consistent with the subcellular localization data, no cystine severe phenotype associated with G308R and indicated a transport was observed with M1I (Table 2 and Fig. 2), further critical role for the sixth TM. We report here a study of 31 suggesting the absence of an alternative initiation codon. missense CTNS mutations or in-frame deletions/insertions Q222R, IVFD343–346del and DVVF346–349del, which associated with the various forms of cystinosis, and their partially altered the subcellular localization of cystinosin, effect on the subcellular localization and function of cystino- abolished cystine transport. This was also true for S141F, sin. This work provides an insight into the regions of the L158P, G169D, S270del, D205N, D205del, N288K, D305Y, protein critical to its transport activity, as well as into the G308R, L338P, G339R and DVEF349–350ins, which did not relationship between the molecular and clinical phenotypes affect the lysosomal localization of cystinosin. As cystinosin is of the disease. still present at the lysosomal membrane in these cases, the lack Downloaded fromhttps://academic.oup.com/hmg/article/13/13/1361/652333bygueston30September2021 ua oeua eeis 04 o.1,N.13 No. 13, Vol. 2004, Genetics, Molecular Human

Figure 1. Effect of cystinosis-associated mutations on the subcellular localization of cystinosin. EGFP-fused cystinosin constructs bearing the mutation of interest, were transiently expressed in HeLa (A, C, E, G, I, K) and MDCK (B, D, F, H, J, L) cells. For HeLa cells, the cystinosin–EGFP ﬂuorescence (left and right) is compared with the immunoreactivity of the endogenous epithelial membrane antigen (EMA) (middle and right). Scale bar ¼ 20 mm. For MDCK cells, the cystinosin–EGFP ﬂuorescence (left and right) is compared with the immunoreactivity of the endogenous late endosomal/lysosomal marker LAMP-2 (middle and right). Scale bar ¼ 10 mm. (A, B) Wild-type cystinosin–EGFP exhibits an intracellular, punctate distribution corresponding to late endosomes and lysosomes. (C, D) Deletion of the C-terminal GYDQL lysosomal sorting motif from wild-type cystinosin–EGFP induces a partial mislocalization to the plasma membrane with a signal remaining in the late endosomes/lysosomes. (E, F) IVFD343–346del, associated with infantile cystinosis, induces a partial mislocalization of cystinosin–EGFP to the plasma membrane, in addition to the endosomal/lysosomal localization. (G, H) Del-

etion of the GYDQL motif from cystinosin–IVFD343–346del–EGFP completely redirects the protein to the plasma membrane. (I, J) N323K does not alter the late endosomal/lysosomal localization of 1363 cystinosin–EGFP. (K, L) Deletion of the GYDQL motif from cystinosin–N323K–EGFP partially mislocalizes the protein to the plasma membrane, as is the case for wild-type cystinosin–EGFP. 1364 Human Molecular Genetics, 2004, Vol. 13, No. 13

Table 1. List of cystinosis mutations studied according to the associated phenotype

Clinical Base pair change Amino acid change Position Mutation detected on Reference phenotype second allele

Infantile 342G.A M1I N-terminus D (6) 736A.T I133F First TM W138X (18) 761C.T S141F First TM S141F and I260T on both u.d. alleles 812T.C L158P Second TM D (18), u.d. 845G.A G169D Second TM G169D (10) 883T.C W182R Second TM 1367TCGTCTTC.A (10) 952G.A D205N Second inter-TM S298N; D (10), u.d.

1004A.G Q222R Third TM D (6) Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021 1203C.A N288K Fifth inter TM N288K (6) 1232G.A S298N Fifth inter TM D205N (10) 1252G.T D305Y Sixth TM D (5) 1261G.A G308R Sixth TM G308R; 1310 2 12G.A; (5,10) IVFD343–346del; D 1352T.C L338P Seventh TM L338P (5) 1354G.A G339R Seventh TM G339R; 564 þ 1G.A; (5,6,10,19), 1036–1037insCG; u.d. 1310 2 12G.A; D 950–952del3 D205del Second inter-TM D205del; D (5,10) 1146–1148del3 S270del Fifth TM S270del (5) 1366–1377del12 IVFD343–346del Seventh TM G308R (5) 1375–1386del12 DVVF346–349del Seventh TM D (6) 1386–1387ins12 DVEF349–350ins Seventh TM D (6) Juvenile 938C.T P200L Second inter-TM D (6) 1178A.G K280R Fifth inter TM D (11) 1308C.G N323K Sixth inter-TM N323K (11) 800–801ins9 PCS154–155ins First inter-TM PCS154–155ins (5) Ocular 928G.A G197R Second inter-TM D; 2303G.T; 2303insT (12,13,15) Atypical 463G.A V42I N-terminus V42I (5) 668G.T G110V N-terminus G110V (6) 755C.T S139F First TM 985–986insA (5) 1375G.A D346N Seventh TM 1080delC (5) Unclassiﬁed 869A.C N177T Second TM n.d. (6) 1118T.C I260T Fourth inter-TM I260T and S141F on both u.d. alleles 537–557del21 ITILELP67–73del N-terminus ITILELP67–73del; n.d.; D (5,10,14), u.d.

Nomenclature according to (20); the first nucleotide of the cDNA is position þ1. Abbreviations: D, 57 kb deletion; n.d, not determined; u.d., unpublished data. of transport was attributed to an inability to transport cystine, N323K abolished cystine transport, in contrast to the associated rather than to a decreased cell surface expression of cystino- juvenile phenotype. To examine whether these mutations sin–DGYDQL in our functional assay, as confirmed for the induced a low level of activity undetected in our assay, the G308R mutation (17). In contrast, the three remaining cystinosin constructs were expressed for 72 h (as compared mutations, I133F, S298N and W182R, retained transport with 48 h in the standard protocol) and cystine uptake time activity. I133F (not a causative mutation according to recent was extended to 20 min (compared with 10 min). These con- data, see Discussion) and S298N did not significantly alter ditions significantly improved the signal-to-noise ratio of the cystine transport (92 + 16 and 77 + 21% of wild-type wild-type cystinosin–DGYDQL leading to a mean transport cystinosin–DGYDQL activity, respectively), whereas activity of 1200 + 73% above background (n ¼ 6), compared W182R reduced the uptake to 34 + 5.9% of cystinosin– with 640 + 50% in the standard protocol (17). However, we DGYDQL activity (Table 2 and Fig. 2). did not detect cystine transport by these juvenile cystinosis- associated mutants. We also examined whether the lack of Juvenile cystinosis. Of the four mutations associated with transport activity resulted from a lack of cell surface expression juvenile cystinosis, three are missense mutations situated of the cystinosin–DGYDQL constructs used in the transport immediately adjacent to the fifth TM (K280R) or within the assay. For this, constructs bearing K280R and N323K in inter-TM loops oriented towards the lysosomal lumen (P200L addition to the GYDQL deletion and EGFP tag were expressed and N323K), and one is an insertion of three amino acids in in HeLa and MDCK cells (Fig. 1I–L and data not shown). the first inter-TM loop (PCS154–155ins). P200L and However, these recombinant proteins, which also did not lead PCS154–155ins led to a low level of cystine transport to cystine uptake (data not shown), partially localized to the (15 + 4.5 and 9.2 + 2.1% of cystinosin–DGYDQL activity, plasma membrane, as does wild-type cystinosin–DGYDQL– respectively) (Table 2 and Fig. 2). However, K280R and EGFP (16). Thus, despite the juvenile phenotype associated Human Molecular Genetics, 2004, Vol. 13, No. 13 1365

Table 2. Effect of each mutation on the subcellular localization and cystine transport activity of cystinosin

Clinical phenotype Amino acid change Subcellular localization Transport activity at acid pH No. of determinations

Infantile M1I No expression 4.0 + 4.8 4 I133F Lysosome 92 + 16 5 S141F Lysosome 2.0 + 5.3 3 L158P Lysosome 1.4 + 4.2 4 G169D Lysosome 20.82 + 1.4 3 W182R Lysosome 34 + 5.9 5 D205N Lysosome 22.1 + 2.9 3 Q222R Lysosome þ plasma membrane 2.4 + 1.3 6 N288K Lysosome 1.6 + 1.2 4

S298N Lysosome 77 + 21 4 Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021 D305Y Lysosome 24.8 + 1.7 3 G308R Lysosome 25.3 + 8.5 4 L338P Lysosome 28.0 + 3.9 3 G339R Lysosome 28.0 + 3.3 3 D205del Lysosome 1.9 + 2.4 4 S270del Lysosome 21.9 + 1.0 4 IVFD343–346del Lysosome þ plasma membrane 20.57 + 1.1 4 DVVF346–349del Lysosome þ plasma membrane 27.6 + 5.8 4 DVEF349–350ins Lysosome 21.6 + 2.4 4 Juvenile P200L Lysosome 15 + 4.5 6 K280R Lysosome 20.68 + 0.9 5 N323K Lysosome 0.14 + 0.8 4 PCS154–155ins Lysosome 9.2 + 2.1 6 Ocular G197R Lysosome 20 + 7.8 6 Atypical V42I Lysosome 97 + 4.1 3 G110V Lysosome 120 + 27 3 S139F Lysosome 22.9 + 1.0 4 D346N Lysosome 61 + 11 7 Unclassiﬁed N177T Lysosome 0.70 + 4.1 4 I260T Lysosome 74 + 13 5 ITILELP67–73del Lysosome 19 + 6.1 7

Transport activity is expressed as a mean percentage of wild-type cystinosin activity + SEM. with these mutations, K280R and N323K apparently abolish the capacity of cystinosin, whereas G110V, V42I and D346N do transport activity of cystinosin in our assay. not, or only partially, alter this capacity.

Unclassified mutations. Three mutations that could not be Ocular cystinosis. The missense mutation G197R, which is classified into any group with certainty were also studied associated with several cases of ocular cystinosis and situated (Table 2). The missense mutations N177T and I260T are in the second inter-TM loop, led to a low level of cystine situated in the second TM and fourth inter-TM loop, respect- transport: 20 + 7.8% of wild-type cystinosin–DGYDQL ively, and the seven amino-acid deletion ITILELP67–73del activity (Table 2 and Fig. 2). in the N-terminal region. N177T abolished cystine transport, I260T moderately (74 + 13%) altered this activity and Atypical cystinosis. Four missense mutations associated with a ITILELP67–73del reduced cystine transport to 19 + 6.1% of combination of symptoms that could not be classified as either cystinosin–DGYDQL activity. infantile or juvenile cystinosis, were also studied (Table 2 and Fig. 2). G110V (6) and V42I (5), localized in the N-terminal DISCUSSION region, did not significantly alter cystine transport activity (120 + 27 and 97 + 4.1% of cystinosin–DGYDQL activity, The development of a functional assay for cystinosin (17) respectively), whereas D346N (5), situated in the seventh provides a tool for analysing the pathogenesis of cystinosis TM, moderately altered this activity (61 + 11%). In contrast, at the molecular level. For this, we studied 31 mutations S139F (5) abolished cystine transport and, as observed for the associated with the different clinical forms of cystinosis for juvenile K280R and N323K mutations, increased times of their effect on transport activity and subcellular localization. protein expression and cystine uptake did not give signifi- Overall, for the three defined forms of cystinosis (infantile, cantly different results. Furthermore, cystinosin–S139F– juvenile and ocular), the transport data tend to correlate with DGYDQL–EGFP partially localized to the plasma membrane the clinical data: of the 18 infantile mutations (excluding (and did not induce cystine uptake), demonstrating that M1I which prevents translation), 15 abolish cystine transport. the absence of activity was not due to the lack of surface Conversely, three of the five mutations associated with expression. We thus conclude that S139F impairs the transport juvenile or ocular forms display 9–20% of wild-type 1366 Human Molecular Genetics, 2004, Vol. 13, No. 13

cytosolic sorting machinery in the IVFD343–346del and DVVF346–349del mutants. This is demonstrated by the fact that deletion of the GYDQL motif results in the complete relocalization of these mutants to the plasma membrane. Thus, IVFD343–346del and DVVF346–349del probably interfere with the second sorting signal. The same observations were made with Q222R, which is in the third TM. Taken together, these data suggest that the second sorting signal is contained within the overall conformation of cystinosin, rather than restricted to the linear sequence in the ﬁfth inter-TM loop (16). An attractive, though speculative, mechanistic possibility

for this second signal could be that cystinosin dynamically Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021 interacts with another lysosomal protein that possesses a sorting motif. Such an interaction would drive cystinosin partially to the lysosome in the absence of the GYDQL motif.

Predictive value of the transport assay Figure 2. Effect of cystinosis-associated mutations on the transport activity of Before considering the discrepancies between biochemical and cystinosin. Cystine transport is expressed as a percentage of wild-type cystinosin–DGYDQL activity. Error bars correspond to the SEM (the number of deter- clinical data, it is important to assess the level of confidence minations is given in Table 2). Diamonds, triangles and the circle represent the we can place in our functional assay. First, transport data mutations associated with infantile, juvenile and ocular cystinosis, respectively. are in good agreement with topological and phylogenetic Squares represent the four mutations associated with atypical forms of the information on cystinosin: as observed for many other mem- disease. Three mutations that could not be classified unambiguously to a clinical phenotype are omitted from the graph. Note that the mutations S298N and brane proteins, transport activity is less tolerant of amino W182R allow a significant level of cystine transport that contrasts with the acid substitutions when they affect conserved residues or resi- severe phenotype. This is also the case for I133F, indicated in italics because dues located in TMs. Among 23 substitutions analysed recent data suggest that this mutation is in fact not causative (see Discussion). (excluding M1I), 14 abolish transport, and all but one Conversely, the mutations N323K and K280R abolish cystine transport, (N323) of these essential residues are strictly conserved despite their association to a milder juvenile phenotype. The same holds true for S139F, which is associated with an atypical form that is less severe than across eukaryotic species (Fig. 3). With respect to the position the infantile, but more severe than the juvenile phenotype. on the topological model, 12 of the 14 deleterious substitutions are located within or at the border of the TMs; the remaining two, N288K and N323K, are located in the last cytosolic and activity. Therefore, the hypothesis that infantile cystinosis lumenal loops, respectively (Fig. 4). is caused by two loss-of-function alleles, whereas Second, from a more quantitative perspective, the value of patients with milder forms are homozygous or compound 20% of wild-type transport activity obtained for the ocular heterozygotes for a specific, partially active allele (5,11,12), G197R mutation is probably relevant, as it is consistent with holds true in most cases. However, there are some exceptions. previous biochemical data on cystinotic lysosomes. Indeed, Six mutations retain a full or substantial (.30% of wild-type) cystine transport is reduced 2-fold in lysosomes from leuko- level of activity, despite their association to a severe cytes of individuals heterozygous for an infantile mutation phenotype (infantile and atypical). Conversely, for two other (22), implying that a residual activity of 50% should be clini- mutations associated with juvenile cystinosis, no transport cally silent. Moreover, similar measurements made on lyso- activity could be detected. The lack of transport in the somes from patients with ocular cystinosis yielded a cystine S139F mutant is also surprising because the carrier, reported transport corresponding to 19 + 10% of wild-type, whereas as atypical due to a late onset of renal Fanconi syndrome values for those with infantile cystinosis were in the 0–5% at 3 years, only presents with moderate chronic renal range (23). Therefore, our assay provides a good estimate of insufficiency at 22 years of age (5). These discrepancies will the effect of mutations on transport. In the future, its sensi- be discussed later. tivity should be increased to allow precise comparison of With regards to subcellular localization, only three mutants with low residual activities (e.g. G197R versus mutations, Q222R, IVFD343–346del and DVVF346– P200L and PCS154–155ins). 349del, which are associated with infantile cystinosis, partially The predictive potential of our assay becomes evident relocalize cystinosin to the plasma membrane. We have pre- when transport data for the three unclassified mutations viously shown that the sorting of cystinosin to lysosomes ITILELP67–73del, N117T and I260T, the infantile mutation requires two sorting signals: a classical tyrosine-based motif I133F and the atypical mutation G110V, are compared with (GYDQL) in the C-terminal tail and a conformational signal clinical and genetic data. ITILELP67–73del has been described delineated to a region of the fifth inter-TM loop (16). Mutation in homozygotes and compound heterozygotes with infantile of either signal results in partial redirection to the plasma cystinosis (10,14). However, we detected this mutation in the membrane, whereas a complete relocalization is achieved heterozygous state in two individuals with juvenile cystinosis: by mutating both signals. Although deletions in the seventh originally, in a case where the second mutated allele has not TM could have prevented exposure of the GYDQL motif to yet been identified (5) and, more recently, in another case, the cytosolic compartment, this motif is recognized by the segregating with the 57 kb deletion (unpublished data). Our Human Molecular Genetics, 2004, Vol. 13, No. 13 1367 Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021

Figure 3. Amino acid sequence alignment of eukaryotic cystinosin homologues. Orthologous amino acid sequences from Homo sapiens (SWISSPROT accession no. O60931), Mus musculus (accession no. P57757), Drosophila melanogaster (accession no. Q9VCR7), Caenorhabditis elegans (accession no. Q09500), Sac- charomyces cerevisiae (accession no. P17261) and Arabidopsis thaliana (accession no. P57758) were aligned using the CLUSTAL W software (21). Black shading indicates amino-acid identities and grey indicates similarities. The position of missense mutations that abolish cystine transport are indicated by ﬁlled circles, and those that allow partial or total cystine transport are indicated by open circles. Overlining indicates the positions of the TMs. 1368 Human Molecular Genetics, 2004, Vol. 13, No. 13 Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021

Figure 4. Position of each class of transport-altering mutations on the topological model of cystinosin. Mutations are categorized into three classes according to whether they abolish (red symbols), severely inhibit (,30% of wild-type; blue symbols) or only moderately, or do not, alter (30% of wild-type; green symbols) cystine transport. Stars represent point mutations, dashes correspond to deletions of 1 amino acid, and triangles indicate insertions of three (position 154) or four (position 349) amino acids. Black circles represent the predicted N-glycosylation sites. Circles with a light blue centre represent the lysosomal sorting signals: the GYDQL motif in the C-terminal tail and the critical residuals of the conformational signal in the fifth inter-TM. Numbers correspond to the position of predicted N-glycosylation sites and of the extremities of TMs. data showing that ITILELP67–73del leads to 19% residual This is consistent with the fact that we never detected this transport activity, as well as the fact that no other mutation mutation in the French patients we have studied to date (108 associated with infantile cystinosis is located in the N-terminal patients), whereas we did detect the other missense mutation, region of cystinosin, further reinforce the conclusion that this L158P, identified in this same French-Canadian study (18). mutation is associated with juvenile cystinosis. The confusion Finally, the atypical mutation G110V was detected in the surrounding the clinical status of ITILELP67–73del illustrates homozygous state in a patient who presented with early and the difficulties associated with classifying cystinosis patients severe features of the Fanconi syndrome and developed into various clinical forms owing to heterogeneous criteria ESRD at 13 years. However, even in the absence of treatment, employed by clinicians world-wide. the only extra-renal disorders that she developed were mild N177T was also identified in an individual with juvenile ocular anomalies with photophobia occurring at 23 years. cystinosis, and the second mutated allele has not yet been We show here that the G110V substitution does not affect identified (6). The fact that N177T abolishes cystine transport cystine transport. However, this mutation involves the last suggests that it represents a severe mutation. This would be nucleotide of exon 6 and hence doubles as a splice site consistent with its position in the second TM, which houses mutation. Indeed, we have shown previously that it affects two other infantile cystinosis-associated mutations that the splicing of exon 6 and leads to a frameshift (at amino abolish cystine transport. Therefore, the second mutation acid 111) and a truncated protein (S124X) (6). Thus, our segregating in this individual, which may lie in the regulatory data demonstrate that it is not the amino acid substitution regions of CTNS and hence so far escaped detection, is likely but the splicing defect that gives rise to the associated pheno- to be the mild mutation. type. These data support our hypothesis that inter-tissue varia- The last unclassified mutation, I260T, was identified in the bility in the expression of the splicing defect is responsible for homozygous state in conjunction with S141F in a patient with the atypical character of cystinosis in this patient. infantile cystinosis (unpublished data). S141F abolishes cystine transport, whereas I260T only moderately affects Discrepancies between molecular and clinical phenotypes cystine transport (74%). Subsequent to these results, I260T was reported as a polymorphism (refSNP ID: rs161400) in In contrast to the aforementioned mutations, seven of the the public database of single nucleotide polymorphisms mutations that we tested show significant discrepancies (http://www.ncbi.nlm.nih.gov/SNP). Therefore, only S141F between transport and clinical data. There are two types of contributes to the clinical phenotype of the patient. discrepancies: some mutations associated with a milder clini- I133F, identified in a French-Canadian population (18), has no cal phenotype abolish cystine transport, and others associated effect on cystine transport, in apparent contradiction with its with a severe clinical phenotype have little effect on transport. reported association to an infantile phenotype. However, recent The first category consists of the atypical mutation S139F data have revealed that I133F is not the causative mutation in (5) and the juvenile mutations K280R and N323K (11). As these families (J. McGowan-Jordan, personal communication). mentioned earlier, we have delineated the cystine transport Human Molecular Genetics, 2004, Vol. 13, No. 13 1369 threshold for the appearance of clinical anomalies at 20% of identified. On the other hand, the rarer cases of discrepancies wild-type. We cannot exclude the possibility that the lack of between transport and clinical data might unravel unsuspected activity in these three mutants is only apparent and results from aspects of the pathogenic cascade. an insufficient sensitivity of our assay (e.g. transport activities 5% may be undetectable in our standard protocol). However, we doubled the sensitivity of our assay by increasing the MATERIALS AND METHODS duration of protein expression and cystine uptake, but did CTNS mutations studied not detect residual activity in the mutants. In the case of the residues S139 and K280, the lack of transport in the mutants Mutations that do not induce gross alterations of the reading is consistent with the strict conservation of these amino frame (missense mutations and in-frame deletions or acids during evolution (although K280R is a conservative insertions) were studied exclusively (Table 1). Of a total of 31 such mutations, 19 were associated with infantile cystino- change, an arginine is never observed in other eukaryotic Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021 sequences) (Fig. 3). Thus, an alternative possibility should sis, four with juvenile cystinosis, one with ocular cystinosis be considered to explain this category of discrepancies. We and four with atypical forms of the disease. The remaining speculate that, in vivo, cystinosin associates with another three mutations could not be unambiguously associated to a TM protein of the lysosome, and that this association stabilizes clinical phenotype. Firstly, the missense mutation N177T the conformation of the mutant S139F, K280R and N323K was found in two sisters with juvenile cystinosis, but the polypeptides. Such mutations would thus be less deleterious mutation on the second allele has not yet been detected (6). in the patient than in our in vitro assay, which is based on N177T could thus represent the severe allele of a compound an ectopic expression of cystinosin at the plasma membrane. heterozygote. Secondly, I260T was detected on the same As mentioned earlier, a heteromeric association with another allele as S141F, in the homozygous state, in an individual lysosomal protein would also explain the effect of other with infantile cystinosis (unpublished data), raising the possi- mutations on the intracellular localization of cystinosin. bility that it is not a causative mutation. Finally, the in-frame The second category of discrepancies concerns the infantile deletion ITILELP67–73del was initially observed in a patient mutations W182R and S298N (10), and the atypical ones V42I with juvenile cystinosis (5), but the mutation on the second and D346N (5). Unlike G110V, these mutations do not disrupt allele was not identified, and other groups reported the same consensus splice site sequences located at exon–intron junc- mutation, in the homozygous or compound heterozygous tions, and therefore are not thought to affect splicing. state, in individuals with infantile cystinosis (10,14). Recently, However, the possibility that they may alter or introduce this mutation was observed in a second patient with juvenile exonic splicing elements [such as enhancers or silencers; cystinosis, harbouring the 57 kb deletion on the second (24)] cannot be excluded. Similarly, it cannot be ruled out allele (unpublished data). that, as discussed earlier for I133F, one or more of these mutations are not, in fact, causative mutations, in particular Generation of constructs containing CTNS mutations W182R, S298N and D346N, which were detected in the compound heterozygous state. Conversely, it is conceivable that The introduction of all CTNS mutations was performed using other functional aspects than intracellular localization or trans- the QuikChange site-directed mutagenesis kit (Stratagene) and port are affected by these mutations. For example, the mutated verified by sequencing the region around the introduced cystinosin might be degraded at a higher rate in vivo, but this mutation (primer sequences available upon request). For the effect would be negligible in vitro when the protein is transi- localization studies, CTNS mutations were introduced into ently overexpressed in COS cells. Alternatively, interactions the pEGFP-N1 expression plasmid (Clontech) encoding a with other components of the lysosomal matrix or lumen wild-type cystinosin–EGFP fusion protein (16). For the important for the biological function of cystinosin might be cystine transport assay, CTNS mutations were introduced disrupted by the mutations in vivo. In this perspective, it into the pcDNA3.1/Zeoþ expression plasmid (Invitrogen) might be interesting to note that V42 is adjacent to an encoding cystinosin deleted in the C-terminal lysosomal N-glycosylation site at position 41, although its substitution sorting motif GYDQL [cystinosin–DGYDQL; (17)]. to isoleucine is not predicted to prevent recognition by oligo- Additional constructs (cystinosin–DGYDQL–EGFP bearing saccharyltransferase (25). These hypotheses serve as a remin- the mutations Q222R, IVFD343–346del, DVVF346–349del, der that a simplified in vitro model cannot recapitulate all the K280R, N323K and S139F) were also generated by site- subtleties and complexities at work in an entire organism. A directed mutagenesis and assayed for their effect on subcellu- further investigation of these discordant mutants could thus lar localization and transport activity. provide deeper insights into the molecular or cellular mechan- isms that modulate disease severity. In conclusion, this study shows that when the CTNS Cell culture and transfection mutation does not prevent the synthesis of a full polypeptide, For the localization studies, HeLa and MDCK cells were impaired transport is the most frequent cause of pathogenicity. grown in minimal essential medium (Life Technologies) Furthermore, the level of transport inhibition often correlates supplemented with 10% foetal calf serum (FCS) and 2 mM 5 with the severity of symptoms: transport is abolished by 15 L-glutamine. Approximately 2 10 cells were cultured in of the 17 mutations associated with classical infantile cystino- 35 mm wells containing glass coverslips and transfections sis (excluding M1I and I133F). This transport assay thus (with wild-type or modified cystinosin–EGFP) were carried has potential prognostic interest when novel mutations are out by incubating the cells with 2 mg plasmid and 4 ml 1370 Human Molecular Genetics, 2004, Vol. 13, No. 13

FuGENE 6 (Roche Molecular Biochemicals) for 48 h, For each CTNS mutation, mean values + standard error of according to the manufacturer’s recommendations. For the the mean (SEM) were derived from at least three independent cystine transport assay, COS-7 cells were grown under 5% transfections. CO2 in glucose-rich Dulbecco’s modified Eagle medium containing Glutamax-I (Life Technologies) supplemented with 7.5% FCS. Approximately 1 106 cells were transfected ACKNOWLEDGEMENTS with 5 mg plasmid (wild-type or modified cystinosin– This work was supported by Association Française contre les DGYDQL or, as a negative control, with pcDNA3.1/Zeoþ) Myopathies, Vaincre les Maladies Lysosomales, Association by electroporation as previously described (17). Electropo- pour le De´veloppement du Rein Artificiel, Association pour rated cells were diluted into 2 ml of culture medium, divided l’Information et la Recherche sur les Maladies Geńe´tiques into four aliquots and cultured in a 24-well plate. Reńales and Fondation pour la Recherche Me´dicale (PhD grant to S.C.). Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021

Immunofluorescence studies REFERENCES Transfected HeLa cells were washed twice with phosphate- buffered saline (PBS) on ice, and incubated with a 1:50 1. Neufeld, E.F. (1991) Lysosomal storage diseases. Annu. Rev. Biochem., 60, 257–280. dilution [in 0.1% bovine serum albumin (BSA)/PBS] of the 2. Mancini, G.M., Havelaar, A.C. and Verheijen, F.W. (2000) Lysosomal primary antibody EMA, directed against epithelial membrane transport disorders. J. Inherit. Metab. Dis., 23, 278–292. antigen (DAKO), for 30 min at 48C. Coverslips were then 3. Verheijen, F.W., Verbeek, E., Aula, N., Beerens, C.E., Havelaar, A.C., fixed with 4% paraformaldehyde for 20 min, incubated with Joosse, M., Peltonen, L., Aula, P., Galjaard, H., van der Spek, P.J. and Mancini, G.M. (1991) A new gene, encoding an anion M 50 m NH4Cl for 10 min and blocked using 1% BSA/10% transporter, is mutated in sialic acid storage diseases. Nat. Genet., 23, donkey serum/PBS for 1 h at room temperature. Transfected 462–465. MDCK cells were fixed with 4% paraformaldehyde for 4. Gahl, W.A., Thoene, J. and Schneider, J. (2001) Cystinosis: a disorder of 20 min, incubated with 50 mM NH4Cl for 10 min and permea- lysosomal membrane transport. In Scriver, C.J., Beaudet, A.L., Sly, W.S. bilized using 0.1% Triton X-100 for 10 min. Coverslips were and Valle, D. (eds), The Metabolic and Molecular Bases of Inherited Disease. McGraw-Hill, New York, pp. 5085–5108. blocked in 1% BSA/0.1% saponin/10% donkey serum/PBS, 5. Attard, M., Jean, G., Forestier, L., Cherqui, S., van’t Hoff, W., Broyer, M., and incubated with a 1:500 dilution (in 1% BSA/0.1% Antignac, C. and Town, M. (1999) Severity of phenotype in cystinosis saponin/PBS) of the primary antibody AC17, directed varies with mutations in the CTNS gene: predicted effect on the model of against canine LAMP-2 (kindly provided by A. LeBivic, cystinosin. Hum. Mol. Genet., 8, 2507–2514. 6. Kalatzis, V., Cohen-Solal, L., Cordier, B., Frishberg, Y., Kemper, M., Marseille, France), for 1 h at room temperature. Finally, all Nuutinen, E.M., Legrand, E., Cochat, P. and Antignac, C. (2002) coverslips were incubated using a 1:400 dilution (in 1% Identification of 14 novel CTNS mutations and characterization of seven BSA/PBS) of donkey anti-mouse-Cy3 secondary antibody splice site mutations associated with cystinosis. Hum. Mutat., 20, (Immunotech) for 1 h at room temperature, and mounted 439–446. using Fluoroprep (bioMe´rieux). Fluorescence was visualized 7. The Cystinosis Collaborative Research Group (1995) Linkage of the gene for cystinosis to markers on the short arm of chromosome 17. using a ZEISS LSM 510 laser scan microscope (Carl Zeiss) Nat. Genet., 10, 246–248. equipped with a plan apochromat 63.2 oil immersion objec- 8. Town, M., Jean, G., Cherqui, S., Attard, M., Forestier, L., Whitmore, S.A., tive and with a visible argon laser (488 nm) and two helium Callen, D.F., Gribouval, O., Broyer, M., Bates, G.P., van’t Hoff, W. neon lasers (543 and 633 nm). Images were collected sequen- and Antignac, C. (1998) A novel gene encoding an integral membrane protein is mutated in nephropathic cystinosis. Nat. Genet., 18, 319–324. tially and processed using Adobe Photoshop 5.5. 9. Touchman, J.W., Anikster, Y., Dietrich, N.L., Maduro, V.V., McDowell, G., Shotelersuk, V., Bouffard, G.G., Beckstrom-Sternberg, S.M., Gahl, W.A. and Green, E.D. (2000) The genomic region encompassing the nephropathic cystinosis gene (CTNS ): complete sequencing of a Cystine uptake assay 200-kb segment and discovery of a novel gene within the common cystinosis-causing deletion. Genome Res., 10, 165–173. Unless otherwise indicated, cystine uptake was assayed 48 h 10. Shotelersuk, V., Larson, D., Anikster, Y., McDowell, G., Lemons, R., after transfection as previously described (17). Briefly, culture Bernardini, I., Guo, J., Thoene, J. and Gahl, W.A. (1998) CTNS wells were washed twice with 500 ml uptake buffer A (5 mM mutations in an American-based population of cystinosis patients. D-glucose, 140 mM NaCl, 1 mM MgSO ,20mM potassium Am. J. Hum. Genet., 63, 1352–1362. 4 11. Thoene, J., Lemons, R., Anikster, Y., Mullet, J., Paelicke, K., phosphate pH 7.4) and then incubated with 200 mlbufferB Lucero, C., Gahl, W., Schneider, J., Shu, S.G. and Campbell, H.T. (identical to buffer A but with potassium phosphate adjusted (1999) Mutations of CTNS causing intermediate cystinosis. Mol. Genet. 35 to pH 5.6) supplemented with 40 mM or 1 mCi [ S]L-cystine Metab., 67, 283–293. (50–200 mCi/mmol; Amersham) for 10 min, unless otherwise 12. Anikster, Y., Lucero, C., Guo, J., Huizing, M., Shotelersuk, V., stated. The cells were rinsed twice with 500 mlofchilled Bernardini, I., McDowell, G., Iwata, F., Kaiser-Kupfer, M.I., Jaffe, R. et al. (2000) Ocular nonnephropathic cystinosis: clinical, biochemical, and buffer A, lysed with 200 ml of 0.1 N NaOH and the accumu- molecular correlations. Pediatr. Res., 47, 17–23. lated radioactivity was counted by liquid scintillation in 13. Kleta, R., Anikster, Y., Lucero, C., Shotelersuk, V., Huizing, M., Emulsifier-Safe cocktail (Packard) using a Tri-Carb 2100 TR Bernardini, I., Park, M., Thoene, J., Schneider, J. and Gahl, W.A. liquid scintillation analyser (Packard). Transport measurements (2001) CTNS mutations in African American patients with cystinosis. Mol. Genet. Metab., 74, 332–337. were performed in triplicate and expressed as percentage of 14. Heil, S.G., Levtchenko, E., Monnens, L.A., Trijbels, F.J., wild-type (cystinosin–DGYDQL) transport after subtracting Van der Put, N.M. and Blom, H.J. (2001) The molecular basis of the background uptake observed in mock-transfected cells. Dutch infantile nephropathic cystinosis. Nephron, 89, 50–55. Human Molecular Genetics, 2004, Vol. 13, No. 13 1371

15. Phornphutkul, C., Anikster, Y., Huizing, M., Braun, P., Brodie, C., 20. Antonarakis, S.E. and The Nomenclature Working Group (1998) Chou, J.Y. and Gahl, W.A. (2001) The promoter of a lysosomal Recommendations for a nomenclature system for human gene mutations. membrane transporter gene, CTNS, binds Sp-1, shares sequences with Nomenclature Working Group. Hum. Mutat., 11, 1–3. the promoter of an adjacent gene, CARKL, and causes cystinosis if 21. Thompson, J.D., Higgins, D.G. and Gibson, T.J. (1994) CLUSTAL mutated in a critical region. Am. J. Hum. Genet., 69, 712–721. W: improving the sensitivity of progressive multiple sequence 16. Cherqui, S., Kalatzis, V., Trugnan, G. and Antignac, C. (2001) The alignment through sequence weighting, position-speciﬁc gap targeting of cystinosin to the lysosomal membrane requires a tyrosine- penalties and weight matrix choice. Nucl. Acids Res., 22, based signal and a novel sorting motif. J. Biol. Chem., 276, 13314–13321. 4673–4680. 17. Kalatzis, V., Cherqui, S., Antignac, C. and Gasnier, B. (2001) Cystinosin, 22. Gahl, W.A., Bashan, N., Tietze, F., Bernardini, I. and Schulman, J.D. the protein defective in cystinosis, is a H(þ)-driven lysosomal cystine (1982) Cystine transport is defective in isolated leukocyte lysosomes from transporter. EMBO J., 20, 5940–5949. patients with cystinosis. Science, 217, 1263–1265. 18. McGowan-Jordan, J., Stoddard, K.L.P., Orrbine, E., McLaine, P., 23. Gahl, W.A. and Tietze, F. (1987) Lysosomal cystine transport in Town, M., Goodyer, P., MacKenzie, A. and Heick, H. (1999) Molecular cystinosis variants and their parents. Pediatr. Res., 21, 193–196. analysis of cystinosis: probable Irish origin of the most common French 24. Categni, L., Chew, S.L. and Krainer, A.R. (2002) Listening to silence Canadian mutation. Eur. J. Hum. Genet., 7, 671–678. and understanding nonsense: exonic mutations that affect splicing. Downloaded from https://academic.oup.com/hmg/article/13/13/1361/652333 by guest on 30 September 2021 19. Rupar, C.A., Matsell, D., Surry, S. and Siu, V. (2001) A G339R mutation in Nat. Rev. Genet., 3, 285–298. the CTNS gene is a common cause of nephropathic cystinosis in the south 25. Marshall, R.D. (1972) Glycoproteins. Annu. Rev. Biochem., 41, western Ontario Amish Mennonite population. J. Med. Genet., 38, 615–616. 673–702.