Genetic Evidence of an Accessory Activity Required Specifically for Cubilin Brush-Border Expression and Intrinsic Factor-Cobalamin Absorption

The parents (members of a breeding colony maintained at Michigan State University, East Lansing, MI) and 23 offspring, including 13 affected and 10 clinically normal dogs, of matings between an affected female and an obligate carrier male were studied. The disease phenotype of each was determined by monitoring puppies until 12 to 16 weeks of age without parenteral cobalamin administration for growth and laboratory abnormalities previously described12 and the same parameters for 3 to 4 weeks after parenteral cobalamin administration.

Peptide sequence was obtained from canine renal cubilin purified, separated by SDS-PAGE, and transferred to polyvinylidene difluoride membrane as described.13 In situ protease digestion and peptide sequencing were performed by the Protein Chemistry Facility of the Worcester Foundation for Experimental Biology (Shrewsbury, MA). A canine renal tubule cDNA library16 was screened with a rat and canine partial cDNA probes by standard methods.17Hybridizing clones containing overlapping inserts were sequenced on both strands by automated dideoxy termination cycle sequencing methods (ABI 373A Sequencer; Applied Biosystems, Inc, Foster, CA). Position 1 of the canine cubilin cDNA refers to the first nucleotide of the full-length cDNA; the A residue of the first ATG is nucleotide 74. (Canine CUBN nucleotide sequences reported here have been submitted to GenBank: full-length cDNA, accession no. AF137068; partial genomic DNA, accession no. AF137069.)

To locate a segregating canine CUBN variation for genotyping, the CUBN intron/exon structure was partially determined by polymerase chain reaction (PCR) and sequencing. Genomic DNA was isolated from liver or blood samples by standard methods,17and portions of the CUBN gene were amplified using various combinations of the primers 838F, 5′-AGCCTGCGTGCTGGACATAGAC-3′; 947F, 5′-GGCTGGCAAGGAAATGGATATAGT-3′; 1150R, 5′-TGGGTGGCAGCCTCCATTATTGA-3′; and 1341R, 5′-CAGCCCAACCTGATTCACACTTA-3′ from the cDNA sequence. PCR reactions of 50 μL contained 1× PCR buffer, 0.4 mmol/L of each deoxynucleotide, 0.5 μmol/L of each primer, and 500 ng of genomic DNA template. TaKaRa LA Taq polymerase (2.5 U; PanVera Corp, Madison, WI) was added after 2 minutes at 94°C, and reactions were performed for 35 cycles of 98°C for 20 seconds and 68°C for 20 minutes. Obligate carrier male and affected female DNAs were amplified with primers 947F and 1150R, the products were cloned (pCR II; Invitrogen, San Diego, CA), and 6 individual clones from each dog were sequenced. Southern blots of canine genomic DNA digested with the restriction endonucleases EcoRI, BamHI,HindIII, Pst I, Xba I, and Stu I were prepared and hybridized by standard methods17 to a canine cubilin cDNA probe (bp 947-1150).

CUBN marker genotyping was by PCR amplification of an identified intron variation (CIV), a 17-base insertion/deletion sequence, using primers CIVF, 5′-GATCACAGGCCTACAGCTCCATT-3′, and CIVR, 5′-CCAGGCCAACCAGAGATCTTCTA-3′, and producing allele-specific products of 199 and 182 bp that were separable by electrophoresis on 4% agarose gels. Reactions were 50 μL containing 1× PCR buffer, 0.37 mmol/L each deoxynucleotide, 0.25 μmol/L of each primer, and 100 ng of DNA template. Taq DNA polymerase (2.5 U; GIBCO/BRL, Bethesda, MD) was added after 5 minutes at 95°C, and reactions were performed for 28 cycles of 95°C for 30 seconds, 65°C for 1 minute, and 72°C for 2 minutes.

Canine cubilin cDNA of 11,282 bp was cloned, showing 73 bp of 5′ and 349 bp of 3′ untranslated sequence and a 10,860-bp open reading frame. The 3,620 residue deduced amino acid sequence had 83% identity with human11 and 70% identity with rat cubilin.7 The structural features identified in human and rat cubilin and conserved in the deduced dog protein included a furin cleavage site after Arg32, an N-terminal amphipathic helical pattern recently suggested to mediate membrane interactions,10 and 154 Cys residues participating in 77 domain-stabilizing disulfide bonds. Identity of the cloned sequence was further confirmed by the presence of internal peptides sequenced from purified canine renal cubilin (KIKLNEEDLGEXLHQ, IDFQQPRMATERG, KLVDLERK, PFYPNVYPGER, VTGQSGIIESSGYPT, and VGNADGPLMXR) distributed throughout the deduced protein sequence.

Partial intron/exon structure of canine CUBN and a segregatingCUBN variation in an intron between bp 1079 and 1080 of the cDNA sequence were identified (Fig 1A). Sequence of the PCR products demonstrated consensus splice donor and acceptor sites at the points at which genomic sequences diverged from cDNA sequence. In the 5′ to 3′ direction, 4 introns (2.2, 0.9, 2.0, and 0.2 kb) and 3 exons (cDNA sequences 948-1079, 1080-1175, and 1176-1294) were defined. A 17-bp insertion (5′-CAGAACATTGTTTATGC-3′) was found 187 bp 5′ of the 3′ intron/exon boundary in 3 of 6 clones of the PCR-amplified 0.9-kb intron from a clinically normal, carrier male (F274) and in all 6 clones from an affected female (F284). Single hybridizing bands of 5 to 10 kb, identical in carrier and affected dog DNA, were observed on Southern blots of canine genomic DNA digested with 6 different restriction enzymes and hybridized to a probe of CUBN cDNA sequence between bp 947 and 1150, thus confirming that the identified 17-bp variation was in a unique region of the canine genome (data not shown).

Genetic linkage of canine I-GS and CUBN was investigated in 10 clinically normal and 13 affected offspring of matings between dogs F284 and F274. The disease phenotype of each dog was determined by the following criteria: affected dogs failed to gain weight after 9 to 13 weeks of age; they had low serum cobalamin concentrations, mild megaloblastic dyshematopoiesis, methylmalonic aciduria, and proteinuria; and all abnormalities other than proteinuria were reversed by parenteral cobalamin administration. In this family, canine I-GS is a fully penetrant, autosomal recessive trait.12 Therefore, affected offspring from these matings are homozygous and clinically normal littermates are heterozygous at the I-GS disease locus.

CUBN genotyping was performed by PCR amplification of allele-specific products using primers flanking the 17-bp CUBNvariation. PCR confirmed that the affected dam, homozygous at the disease locus, was homozygous for the intronic CUBNinsertion and that the clinically normal sire, heterozygous at the disease locus, was heterozygous at the CUBN locus (Fig 1B). However, 13 recombinants were detected among 23 offspring of these matings, a recombination fraction (0.56) that did not differ from the recombination fraction (0.5) expected under the hypothesis of independent segregation of the CUBN and disease loci (χ² = .39, df = 1, P = .53). A disease trait locus and a marker locus on a different chromosome, or located far from the disease locus on the same chromosome, recombine randomly (at a frequency of 0.5 in large samples) and segregate independently during meiosis. Thus, in contrast to I-GS reported in some human kindreds,4 the linkage data above indicate that canine I-GS in this kindred is not caused by mutation of CUBN or any gene within 50 recombination units either side of the canine CUBNlocus. (A recombination unit [equal to 1% recombination and called a centiMorgan] in mammals is equivalent, on average, to ∼1 Mb of DNA.)

But still, affected dogs lack apical brush-border expression of cubilin in ileal and renal epithelia,13 causing selective cobalamin malabsorption12 and selective proteinuria,8respectively. Therefore, the absence of genetic linkage of the canineCUBN marker to canine I-GS indicates there must be some other gene, distant from CUBN, the product of which is required for normal cubilin folding, exit from the ER, and/or transport to the brush-border. Because cellular physiology of humans and dogs is highly homologous, these findings suggest that human I-GS may exhibit locus heterogeneity in addition to allelic heterogeneity at the CUBNlocus. Human I-GS remains to be mapped in kindreds of various ethnic backgrounds.

There are at least 2 genes whose products associate with cubilin5,7 that are not linked to the human MGA 1locus and defects of which could explain our findings and/or be implicated in some forms of human I-GS. The endocytic receptor, megalin (gp330), is postulated to anchor cubilin in the apical membrane and to facilitate endocytosis of cubilin-ligand complexes.7Receptor-associated protein (RAP) is a molecular escort of the lipoprotein receptor-related proteins, including megalin, and may also facilitate proper folding of cubilin within the ER.18Alternatively, there may be activities of other gene products, yet to be determined, that are specifically required for ER-export competence and brush-border expression of cubilin.

The authors thank Pierre Verroust, Søren Moestrup, Donald Patterson, Paula Henthorn, and Karen Friderici for fruitful discussions; Meg Weil, Mary Lassaline, and Rebeccah Kurzhals for dog care; and Dr Lloyd Shaw (Woodstock Veterinary Clinic, Woodstock, IL) for initial referral of the propositi.