Congenital Disorders of Glycosylation Type Ig Is Defined by a Deficiency in Dolichyl-P-mannose:Man 7 GlcNAc 2 -PP-dolichyl Mannosyltransferase*

Type I congenital disorders of glycosylation (CDG I) are diseases presenting multisystemic lesions including central and peripheral nervous system deficits. The disease is characterized by under-glycosylated serum glycoproteins and is caused by mutations in genes encoding proteins involved in the stepwise assembly of dolichol-oligosaccha-ride used for protein N -glycosylation. We report that fibroblasts from a type I CDG patient, born of consanguineous parents, are deficient in their capacity to add the eighth mannose residue onto the lipid-linked oligosaccharide precursor. We have characterized cDNA corresponding

Oligosaccharides N-linked to glycoproteins are initially synthesized on a lipid carrier in the endoplasmic reticulum (ER) 1 membrane by the stepwise addition of monosaccharides onto dolichol pyrophosphate to form Glc 3 Man 9 GlcNAc 2 -PP-dolichol (1,2). The oligosaccharide moiety is then transferred from lipid onto nascent proteins in the lumen of the ER. The resulting oligomannose-type oligosaccharides N-linked to glycoproteins are involved in the folding (3,4), degradation (5,6), and efficient transport (7) of glycoproteins in the secretory pathway. During their passage through the Golgi apparatus, the oligomannose-type sugar units attached to glycoproteins are often partially demannosylated and subsequently decorated with GlcNAc, galactose, sialic acid, and fucose residues (2). The resulting glycoproteins, bearing complex-type oligosaccharide, are subsequently secreted into the extracellular space or exported to the cell surface where they are involved in diverse regulatory processes (8). The importance of this widespread cotranslational modification is attested to by the identification of a rapidly growing number of diseases whose molecular origins have been linked to abnormalities in glycoprotein biosynthesis (9 -14).
Type I congenital disorders of glycosylation (CDG I) are newly described rare metabolic diseases presenting a severe clinical picture in which multisystemic lesions including central and peripheral nervous system deficits are apparent (9,11,15). Although the molecular events leading to this complicated clinical picture are multifactorial and not well understood, the disease is characterized at the biochemical level by underglycosylated serum glycoproteins of hepatic origin (16). Disruption of any one of the enzymic steps involved in the biosynthesis of lipid-linked oligosaccharides (LLO) and the transfer of the mature oligosaccharide onto protein could potentially lead to the glycoprotein hypoglycosylation that is considered to be the hallmark of CDG I. Work on the underlying genetic bases of CDG I is still at an early stage because genes encoding the proteins involved in glycoprotein biosynthesis are numerous and, at least in humans, have for the most part yet to be identified (9). Notwithstanding these difficulties, mutations in six of the genes encoding proteins of the glycosylation pathway have been shown to underlie CDG I, and these disease types have been classified as CDG I subtypes a-f as follows: (Ia, phosphomannomutase 2 (17,18); Ib, phosphomannose isomerase (19,20); Ic, dolichyl-P-Glc:Man 9 GlcNAc 2 -PP-dolichyl ␣3-glucosyltransferase: (21)(22)(23); Id, dolichyl-P-Man:Man 5 GlcNAc 2 -PP-dolichyl ␣3-mannosyltransferase (24); Ie, dolichol-P-Man synthase I (25,26); and If, the MPDU1 gene product of unknown function, (27,28). CDG I cases for which the genetic origins remain unknown are defined as CDG subtype Ix. Identification of the molecular bases of type I CDG is crucial from the clinical standpoint in that CDG Ib is treatable by administering oral mannose to the patient (29). In addition, identification of the underlying genetic defects will ultimately facilitate the design of antenatal diagnostic tests for different disease subtypes.
Here we report on the genetic deficit underlying a new subtype of CDG I. We show that skin biopsy fibroblasts from a CDG I patient have a reduced capacity to add the eighth mannose residue onto the dolichol-PP-oligosaccharide precursor required for protein glycosylation. Sequencing of cDNA derived from the patient and control cells indicated that the patient was homozygous for a point (F142V) mutation in the human homolog of the yeast ALG12 gene that encodes dolichyl-P-Man:Man 7 GlcNAc 2 -PP-dolichyl ␣6-mannosyltransferase.

EXPERIMENTAL PROCEDURES
EST Clone and Primers-The human EST (accession number AI923828, I.M.A.G.E. Clone ID 2451352) was obtained from the I.M.A.G.E. Consortium, Livermore, CA (30). The different primers used for PCR, sequencing, hybridization, or subcloning are listed in Table I.
Northern Blotting-Cells were rapidly disrupted in 7 ml of 4 M guanidine thiocyanate, and total RNA was isolated (31). RNA (20 g) was denatured, fractionated by electrophoresis, transferred, and hybridized as described previously (32). 18 S rRNA was monitored using a complementary oligonucleotide (33).
Western Blot and Isoelectric Focusing-Western blotting and isoelectric focusing of human serum proteins were performed as described previously using a rabbit polyclonal anti-transferrin antibody (34).
Cells, Cell Culture, and Metabolic Radiolabeling of Cells-Skin biopsy fibroblasts were grown in Dulbecco's modified Eagle's medium containing 2 g/liter glucose, 10% fetal calf serum, and 1% penicillin/ streptomycin. For some experiments, immortalized fibroblasts were used and were transformed by Dr. Thierry Levade (INSERM U466), using the pAS plasmid encoding the simian virus 40 large T antigen (35). EBV-transformed lymphoblasts were generated from peripheral blood mononuclear cells isolated using a Ficoll-Paque Plus gradient and were grown in RPMI 1640 medium supplemented as described above. Fibroblasts in confluent 25-cm 2 tissue culture flasks or EBV-transformed lymphoblasts were pulse-radiolabeled for 30  Isolation and Analysis of Lipid-linked and N-Linked Oligosaccharides-Cells were extracted by a modification of the method of Folch (see Refs. 36 and 37). The lower (chloroform) and upper (methanolic) phases were removed and kept, and the interphase proteins were extracted with 2 ϫ 2 ml CHCl 3 /MeOH/H 2 O, 10:10:3. The lower and 10: 10:3 phases were dried and hydrolyzed with 0.02 N HCl (36) in order to release oligosaccharides from LLO. After desalting on AG-1/AG-50 columns, oligosaccharide components were resolved by TLC on silicacoated plastic sheets (Merck) in n-propyl alcohol/acetic acid/H 2 O, 3:3:2 for 36 h (38) and detected by fluorography after spraying the dried TLC plates with En 3 hance (PerkinElmer Life Sciences).
Structural Analysis of Oligosaccharides-Oligosaccharides released from LLO or from Pronase-digested glycoproteins by endo H were analyzed by concanavalin A-Sepharose (ConA-Sepharose, Amersham Biosciences AB) chromatography as described previously (39) except that the elution buffer containing 0.5 M methyl ␣-D-mannopyranoside (Toronto Research Chemicals Inc., ON, Canada) was heated to 60°C prior to use (40). The standard N-linked Man 7 GlcNAc structure was generated by endo H digestion of glycopeptides obtained from Pronase digestion of cellular glycoproteins isolated from HepG2 cells that had been metabolically radiolabeled with 2-[ 3 H]mannose in the presence of CST and kifunensin (39).
Reduced oligosaccharides were subjected to acetolysis (41), and the resulting fragments were analyzed as described previously (42) by TLC on cellulose-coated plastic sheets (Merck). The N-linked Man 8 GlcNAc and Man 9 GlcNAc structures used to generate standard acetolysis fragments were obtained as described above except that the latter structure was derived from normal human skin biopsy fibroblasts.
Mutation Analysis-First strand cDNA was generated from 5 g of total RNA employing the Invitrogen Superscript TM Preamplification System using either a specific antisense primer (9 AS, see Table I) or nonspecific hexamers. For each primer couple, PCR conditions were optimized using the EST and then applied to the cDNAs. PCR products were purified with the Qiaquick PCR purification kit (Qiagen SA, France) prior to automated sequencing. The PCR products obtained with primer couple MALG12S/MALG12AS (see Table I) were submitted to AccI (Ozyme, Saint Quentin en Yvelines, France) restriction digestion followed by electrophoresis in a 6% NuSieve agarose gel (FMC BioProducts, Rockland, ME). Parental mRNA was extracted by Trizol from peripheral blood mononuclear cells isolated using a Ficoll-Paque Plus gradient. Trizol and Ficoll-Paque were used according to the manufacturer's instructions.  HIV-1-derived Lentiviral Vectors-The pSIN.PW.eGFP HIV-1-derived transfer vector (43) contains a 400-bp deletion in the promoter/ enhancer U3 region of the long terminal repeat (Fig. 5A). This so-called self-inactivating (SIN) integrated proviral vector is thus able to drive transcription from its long terminal repeat and therefore prevents promoter interference and oncogene activation. The eGFP reporter transgene is under the control of the internal human phosphoglycerate kinase promoter (hPGK). The post-transcriptional cis-acting regulatory element of the woodchuck hepatitis virus was inserted in sense orientation in order to increase transgene expression (44). The central polypurine tract associated with the central termination site was inserted into the transfer vector because it has been shown to enhance the nuclear translocation of lentiviral pre-integration complexes (43,45). cDNA encoding the human ␣6-mannosyltransferase,

RESULTS
Patient ME, Clinical Picture and Biochemical Diagnosis of CDG I-The patient (ME), a girl of healthy consanguineous Tunisian parents, was admitted to the neonatal unit for weak suckling. Clinical examination revealed generalized hypotonia and facial dysmorphism, and routine laboratory investigations showed normal blood chemistry except for hypocalcemia. Evolution of the condition of the patient was marked by persistent feeding difficulties, failure to thrive, severe psychomotor involvement, major hypotonia, and progressive microcephaly. Moreover, the patient developed frequent ear, nose, throat, and respiratory infections. Later examinations, performed at 6 months, showed normal blood chemistry, although immunologic screening showed low IgG levels (2.38 g/liter) with normal IgA and IgM. At 18 months, failure to thrive required gastrostomy, but neuroimaging of the brain remained normal.
Plasma proteins from ME were subjected to Western blot and isoelectric focusing as shown in Fig. 1A, and examination of serum transferrin revealed hypoglycosylation profiles characteristic to those observed in type I CDG (11). Phosphomannomutase and phosphomannose isomerase activities, known to be reduced in CDG types Ia and Ib, respectively, were normal in ME fibroblasts (results not shown).
Accumulation of Man 7 GlcNAc 2 -PP-dolichol in ME Fibroblasts-In order to understand the cause of glycoprotein hypoglycosylation in patient ME, skin biopsy fibroblasts were metabolically pulse-radiolabeled with 2-[ 3 H]mannose and then extracted with organic solvents to yield CHCl 3 and CHCl 3 / MeOH/H 2 O (10:10:3) lipid fractions. Oligosaccharides were released from LLO present in both these fractions and resolved by TLC as shown in Fig. 1B. Control fibroblasts yielded predominantly mature triglucosylated LLO in the 10:10:3 lipid fraction, whereas less mature species such as Man 9 GlcNAc 2 -PP-dolichol and Man 5 GlcNAc 2 -PP-dolichol were also apparent in the CHCl 3 fraction. By contrast, ME fibroblasts elaborated much less Glc 3 Man 9 GlcNAc 2 -PP-dolichol in both the 10:10:3 and CHCl 3 fractions; however, the latter fraction contained substantial amounts of an LLO whose oligosaccharide moiety migrated between Man 9 GlcNAc 2 and Man 5 GlcNAc 2 (labeled "a" in Fig. 1B). After treatment with endo H, as shown in Fig. 1C, component a yielded a product that migrated slightly slower than a Man 7 GlcNAc structure that was generated by endo H treatment of 2-[ 3 H]mannose-labeled glycoprotein material derived from HepG2 cells. However, ConA-Sepharose chromatography of oligosaccharide a and HepG2 glycoprotein-derived Man 7 GlcNAc demonstrated these two structures to be quite different (see Ref. 46 and Fig. 2A). The endo H sensitivity of oligosaccharide a on the one hand and its weak binding to the lectin column on the other hand suggested that the ␣6-linked mannose of the Man 3 GlcNAc core is substituted with an ␣3-linked mannose residue. Acetolysis, a reaction that specifically cleaves ␣6-linkages of oligosaccharide a (shown in Fig. 2B), confirmed this and demonstrated that its mannose substituents possess an isomeric configuration identical to that which is known to occur in Man 7 GlcNAc 2 -PP-dolichol under normal physiological conditions (oligosaccharide structure shown to the right in Fig. 2B  (2, 46)). Cloning of the Human Homolog of S. cerevisiae ALG12-The above results suggested that the block in biosynthesis of LLO observed in ME fibroblasts is due to inefficient addition of the eighth mannose residue. The gene encoding dolichyl-P-mannose:Man 7 GlcNAc 2 -PP-dolichyl mannosyltransferase has been cloned in yeast and corresponds to the ALG12 locus. Examination of the literature (47,48) indicated that ALG12 belongs to a family of genes encoding several putative mannosyltransferases containing the sequence HKEXRF flanked by two hydrophobic regions (motif 2, Fig. 3). We found several new members of this family in the data banks and constructed a phylogenetic tree that allowed us to delineate four subfamilies. 2 The subfamily containing Saccharomyces cerevisiae ALG12 revealed a second peptide motif (TKVEESF, motif 1, Fig. 3) found to be absent from the other subfamilies. By using this second motif we looked for human expressed sequence tags (ESTs, I.M.A.G.E Consortium, Washington, D. C.) corresponding to hALG12. One such EST was found to be large enough to contain the entire open reading frame (ORF) and was therefore completely sequenced. We obtained an identical sequence when control fibroblast cDNA was used as template. This sequence, which we deposited in the data banks (GenBank TM accession number AJ303120), encoded a highly hydrophobic protein containing 488 amino acids (see Fig. 3).
Analysis of the hALG12 Derived from Patient ME and the Parents-Northern blot analysis (Fig. 4A) revealed only a single ALG12 transcript of 2.5 kb even using low stringency washes. No differences in either the size or level of expression of hALG12 mRNA between normal and ME fibroblasts could be observed. We noted that three different fibroblast populations showed similar levels of mRNA expression to those observed in HepG2 cells, whatever the growth state of these human hepatocellular carcinoma cells.
The similarity of expression of the hALG12 mRNA in the fibroblasts from the patient and normal fibroblasts prompted us to sequence the hALG12 cDNA from the patient. Our sequence data indicated that the patient was homozygous for a point mutation (T571G) that caused an amino acid substitution (F142V) in a conserved region of the peptide sequence (see Fig.  3). The mutation was observed in eight independent sequences derived from two independent RNA preparations. This base substitution generates a restriction site for AccI that allowed us to confirm that the patient ME is homozygous and that the parents are both heterozygous for this mutation (see Fig. 4B). The heterozygous status of both parents was confirmed by sequencing (results not shown). The primer couple designed for PCR amplification of the region around the mutation amplifies a DNA fragment that has the same size irrespective of whether cDNA or genomic DNA was used as template, indicating that AccI digestion of the restriction fragment amplified from genomic DNA can be used for a prenatal diagnostic test.
Transduction of ME Fibroblasts with Wild Type hALG12 cDNA-Next we evaluated the effects of introducing normal hALG12 into the fibroblasts from the patient. This was undertaken by constructing two HIV-1-derived lentiviral transfer vectors as shown in Fig. 5A and described under "Experimental Procedures." One vector contained DNA encoding enhanced GFP (eGFP vector), and a second contained both wild type hALG12 cDNA, and the eGFP DNA in a bicistronic based expression system (eGFP/hALG12 vector). To facilitate these retroviral gene transfer experiments, we chose to transduce a line of the patient's fibroblasts that had been immortalized by transfection with the SV 40 large T antigen. Under the experimental conditions used, it was determined that greater than 90% of the immortalized fibroblasts became eGFP-positive irrespective of whether the eGFP or eGFP/hALG12 vector was used to transduce the cells (results not shown). Fig. 5B shows that whereas immortalized ME fibroblasts transduced with the eGFP vector generated an LLO pattern close to that observed in their non-transduced counterparts, transduction with the eGFP/hALG12 vector caused a dramatic increase in the proportion of mature LLO (Glc 3 Man 9 GlcNAc 2 -PP-dolichyl) in immortalized ME fibroblasts. Indeed, when ME fibroblasts were treated with the higher viral dose the distribution of LLO was similar to that observed in normal non-transduced cells.
Transfer of Non-glucosylated Man 7 GlcNAc onto Glycoproteins in ME Fibroblasts-Finally, the nature of the oligosaccharides transferred onto nascent glycoproteins in ME fibroblasts was examined. Cellular glycoproteins were digested with Pronase, and the resulting glycopeptides were treated with endo H in order to release N-linked oligomannose-type oligosaccharides. Resolution of these structures by TLC (Fig. 6A) shows that control fibroblasts yield predominantly Man 9 GlcNAc and its monoglucosylated counterpart. These two N-linked structures are also prominent on ME fibroblast glycoproteins, but in these cells two additional structures are present. ConA-Sepha-rose chromatography and acetolysis indicated that these two components correspond to Glc 1 Man 7 GlcNAc and Man 7 GlcNAc and that the isomeric configuration of their mannose residues was the same as that observed in oligosaccharide a and different to that of the N-linked Man 7 GlcNAc intermediate commonly found associated with glycoproteins. Triglucosylated oligosaccharides are more efficiently transferred from dolichol onto polypep- tides; however, the direct transfer of Man 7 GlcNAc 2 from dolichol onto protein has been observed in mammalian cell lines (46). In order to test whether or not this phenomenon can occur in ME fibroblasts, we treated these cells with CST, the ER glucosidase I and II inhibitor. In the presence of this drug, oligosaccharides transferred from dolichol onto protein cannot be deglucosylated. Accordingly, when control fibroblasts are treated with CST, all N-linked oligosaccharides were found to be triglucosylated, indicating that in normal cells there is no transfer of non-glucosylated oligosaccharides from lipid onto nascent glycoproteins. By contrast, although treatment of ME fibroblasts with the glucosidase inhibitor caused the appearance of triglucosylated Man 9 -7GlcNAc species and reduced the amount of Man 7 GlcNAc, it did not lead to the disappearance of N-linked Glc 1 Man 7 GlcNAc. When similar 30-min pulse radiolabeling experiments were conducted in the presence of CST and the Golgi mannosidase I inhibitor, kifunensin, there was little change to the patterns of N-linked structures observed in either cell population (results not shown). The inability of CST to inhibit the appearance of Glc 0 -1 Man 7 GlcNAc species in ME cells demonstrates that Man 7 GlcNAc 2 can be transferred from dolichol onto protein. The inability of CST to inhibit the generation of N-linked Glc 1 Man 7 GlcNAc in ME fibroblasts can be accounted for by UDPglucose:glycoprotein glucosyltransferase-mediated monoglucosylation of Man 7 GlcNAc 2 N-linked to ER-situated misfolded glycoproteins (3). We tested this hypothesis by pretreating cells with the protein synthesis inhibitor, cycloheximide, and tunicamycin, an inhibitor of LLO biosynthesis (49). Under these conditions, as shown in Fig. 6B, 2-[ 3 H]mannose is not incorporated into newly synthesized glycoproteins. By contrast, when cells are incubated with [ 3 H]galactose, which is incorporated into UDP-glucose via the action of UDP-glucose epimerase on UDP-galactose, oligosaccharides released from glycoproteins by endo H are radiolabeled irrespective of the presence or absence of the two inhibitors. These results are compatible with the transient reglucosylation of ER-situated unfolded glycoproteins by UDP-glucose:glycoprotein glucosyltransferase. In normal fibroblasts we observed three such reglucosylated N-linked oligosaccharides migrating as Glc 1 Man 9 -7 GlcNAc (Fig. 6B). Although ME fibroblasts also generated the same Glc 1 Man 9 -8 GlcNAc species, two additional reglucosylated N-linked structures were found. A Glc 1 Man 7 GlcNAc structure was observed that, in contrast to its counterpart observed in [ 3 H]galactose-labeled normal fibroblasts, comigrated with the Glc 1 Man 7 GlcNAc observed in 2-[ 3 H]mannose-labeled ME fibroblasts. In addition, a species that migrated as Glc 1 Man 6 GlcNAc was observed in [ 3 H]galactose-labeled ME fibroblasts (Fig. 6B).
In summary, the ensemble of these results demonstrates that in ME fibroblasts three oligosaccharide structures are transferred in approximately equal amounts from dolichol onto glycoproteins; Glc 3 Man 9 GlcNAc, Glc 3 Man 7 GlcNAc, and Man 7 GlcNAc. Furthermore, the N-linked Man 7 GlcNAc structure that is observed in the cells from patient ME can be glucosylated by UDP-glucose:glycoprotein glucosyltransferase.

FIG. 4. Expression and genetic analysis of hALG12 alleles. A,
Northern blot analysis of total RNA from control, CDG Ix, and ME fibroblasts. Total RNA was extracted from pre-confluent ME, CDG Ix, and normal fibroblasts and from HepG2 hepatocellular carcinoma cells at different days of culture. The radiolabeled probe was generated by 32 P labeling of the double digestion (SalI and NotI) fragment generated from the EST. 18 S rRNA was visualized as a gel loading control. B, restriction digestion of PCR products generated from ME and parental cDNA. The PCR products obtained from the EST template or from the cDNA templates derived from normal (N) or ME fibroblasts (ME) or from parental lymphocytes were incubated either in the absence (Ϫ) or presence (ϩ) of the AccI restriction enzyme.

FIG. 5. Transduction of ME fibroblasts with an HIV-1-derived lentiviral vector harboring wild type hALG12.
A, HIV-1-derived lentiviral vectors containing DNA encoding enhanced GFP (eGFP) alone or both eGFP and hALG12 DNA in a bicistronic arrangement were constructed as described under "Experimental Procedures" (abbreviations are defined under "Experimental Procedures"). Fibroblasts from patient ME were immortalized and subsequently transduced with either 10 or 100 multiplicities of infection (moi) of each of the vector particles. 48 h after transduction the cells were radiolabeled. B, after mild acid hydrolysis LLO was resolved by TLC, and after elution and scintillation counting of the radioactive components, the percentages of total structures occurring as Man 7 GlcNAc 2 or Glc 3 Man 9 GlcNAc 2 were computed and are indicated below each lane. LLOs derived from a representative pulse radiolabeling of normal fibroblasts (N) are also shown. The abbreviations are as for Fig. 1B.

DISCUSSION
Here we report on a CDG I patient with a deficiency in the ability to add the eighth mannose onto LLO. Structural analysis of the lipid-linked Man 7 GlcNAc 2 that was found to accumulate in fibroblasts from this patient revealed the same isomeric configuration of mannose residues to that which has been shown to occur in the normal LLO intermediate generated during glycoprotein biosynthesis (2,46). This result allowed us to eliminate the hypothesis that the block was due to a rare "change of function" mutation potentially causing the seventh mannose residue to be added to an inappropriate position of the LLO acceptor. Additionally, it is noteworthy that in a yeast strain deficient in ALG11p, the enzyme that adds the fifth mannose to the growing oligosaccharide-lipid (50), an abnormal Man 7 GlcNAc 2 structure linked to dolichol has been observed (50) further emphasizing the necessity to carry out detailed structural analyses on the oligosaccharide species that occur in these unusual circumstances. The most likely hypothesis to explain the block in LLO biosynthesis was a deficiency in dolichyl-P-Man:Man 7 GlcNAc 2 -PP-dolichyl ␣6-mannosyltransferase. In yeast, this enzyme is thought to be encoded by the ALG12 gene (48,51). The proteins encoded by the ALG12 genes of different species are all hydrophobic, and like some membrane transporters may possess up to 10 -12 transmembrane domains. Nevertheless, the peptide sequences of the ALG12 gene products have features in common to those of the ALG3p which has been shown to have dolichyl-P-mannose:Man 5 GlcNAc 2 -PP-dolichyl mannosyltransferase activity (52), indicating that ALG12p is also likely to have mannosyltransferase activity. In the work presented here we characterized the human homolog of this gene, and the accuracy of the sequence we deposited in the data banks was attested to by the subsequent appearance of an identical "anonymous" cDNA sequence (Gen-Bank TM accession number BC001729). Interestingly, we noted the presence of a consensus glycosylation site (N(K/R)S, see Fig. 3) which was conserved in all species except yeast. Theoretical predictions of hALG12p transmembrane regions indicate that this glycosylation site is situated in an extramembrane loop sufficiently large to support glycosylation. The question of whether or not hALG12p is glycosylated will have to await experimental evidence.
When the human wild type ALG12 gene was introduced into the fibroblasts of the patient, we noted a remarkable normalization of LLO biosynthesis (Fig. 5B), and a concomitant increase in the fraction of fully mannosylated oligosaccharides was transferred from dolichol onto polypeptide (results not shown). These observations strongly suggest that the ALG12 gene is defective in the patient described here.
We found the hALG12 cDNA of the patient to be homozygous for a single point mutation that leads to replacement of a Phe residue by Val in the encoded protein. Several lines of argument suggest that this mutation is the underlying cause of the Man 7 GlcNAc 2 -PP-dolichyl accumulation observed in fibroblasts obtained from patient ME. First, examination of the size of the hALG12 mRNA transcript from the fibroblasts of the patient revealed it to be identical to that observed for the transcript present in other cell lines, indicating normal processing and the absence of an important deletion. Second, this mRNA transcript appeared to be as abundant as that observed in control fibroblasts, signifying on the one hand that both alleles are expressed and on the other hand an absence of mRNA instability. Third, analysis of the human EST banks allowed us to identify 28 separate human ESTs containing the region of the mutation, but we were unable to detect the T571G base change in any of these sequences, suggesting that the mutation is not simply a common polymorphism. Fourth, the F142V replacement occurs in one of the small highly conserved patches toward the NH 2 terminus of ALG12p. In fact, in all species for which sequence data are available, we noted that the position of the replaced Phe residue in the ALG12p is invariably occupied by the aromatic amino acids Phe or Tyr (Schizosaccharomyces pombe) and that this position is next to a completely conserved Tyr residue. At present the role of this pair of highly conserved aromatic amino acids in ALG12p function is not understood, but our observations suggest that they are situated in, or near, a transmembrane domain. Indeed, several point mutations in glycosyltransferases that lead to CDG type I have been noted to occur in the transmembrane regions of these proteins. 2 To conclude, the above arguments strongly favor our assertion that the point mutation identified in the hALG12 cDNA of patient ME is responsible for the accumulation of Man 7 GlcNAc 2 -PP-dolichyl observed in skin biopsy fibroblasts obtained from this subject.
As observed for other types of CDG I, the observed mutation is leaky (18,22,24) because small amounts of fully mannosylated LLO can be detected in cells from the patient. This leakiness allows a substantial transfer of Glc 3 Man 9 GlcNAc 2 from dolichol onto nascent polypeptides in the fibroblasts of the patient. We also observed that N-linked Glc 3 Man 7 GlcNAc 2 and Man 7 GlcNAc 2 account for the remaining two-thirds of the total oligosaccharide transferred onto protein from oligosaccharide lipid. The relative abundance of these latter two species linked on the one hand to lipid and on the other hand N-linked to polypeptide suggests the following: first, the lipid-linked Man 7 GlcNAc 2 is poorly glucosylated; second, the small amount of lipid-linked Glc 3 Man 7 GlcNAc 2 that is formed can be more efficiently transferred to protein than its non-glucosylated counterpart. The observation that substantial amounts of nonglucosylated Man 7 GlcNAc 2 can be transferred from lipid onto protein was surprising considering that human skin biopsy fibroblasts deficient in dolichol-P-Glc:Man 9 GlcNAc 2 -PP-dolichol transferase do not appear to be able to transfer Man 9 GlcNAc 2 directly onto protein (22). However, this phenomenon has been observed in mouse F9 teratocarcinoma cells (46). It would be interesting to look for mutations in the mouse ALG12 gene in the F9 cell line. To summarize, it appears that in addition to leading to only a partial block in hALG12p function, the truncated Man 7 GlcNAc 2 intermediate that accumulates can be transferred directly onto protein. Despite these relieving factors, it is apparent the clinical picture for patient ME is severe, comprising both central nervous system and peripheral deficits. At present it is unclear whether or not a deficiency in hALG12p will invariably lead to a severe type of CDG I; only characterization of other patients with the same enzymic deficit will answer this question. However, one compounding factor may contribute to the severity of the symptoms observed in this case. In fact, observations made with a yeast strain deficient in ALG12 led investigators to suspect that misfolded glycoproteins bearing N-linked Man 7 GlcNAc 2 are poor substrates for ER-associated glycoprotein degradation (48). Although these observations have yet to be extended to mammalian cells, it is possible that in fibroblasts from ME such a phenomenon could exacerbate the ER accumulation of glycoproteins that may result from the misfolding of hypoglycosylated polypeptides (53). Under these conditions, the resulting stress generated in the ER could have profound effects on the cellular homeostasis of such cells. Furthermore, it has been shown in vitro that the "processing" isomers of Man 7 GlcNAc (generated by ER mannosidases) that are substrates for UDPglucose:glycoprotein glucosyltransferase have only 15% of the acceptor activity of Man 9 GlcNAc (54). However, we show here that the "biosynthetic" Man 7 GlcNAc structure that is transferred directly from dolichol onto protein in cells from patient ME can be glucosylated by UDP-glucose:glycoprotein glucosyltransferase, presumably as part of the glycoprotein "quality control" system.
To conclude, our results indicate that a F142V replacement in hALG12p is the cause of inefficient addition of the eighth mannose residue onto Man 7 GlcNAc 2 -PP-dolichol during glycoprotein biosynthesis in a patient with type I CDG. The insufficiency in this step of the pathway for lipid-linked oligosaccharide biosynthesis defines a new subtype of the disease that we suggest should be called type Ig.