The human LARGE gene from 22q12.3-q13.1 is a new, distinct member of the glycosyltransferase gene family

Abstract

Meningioma, a tumor of the meninges covering the central nervous system, shows frequent loss of material from human chromosome 22. Homozygous and heterozygous deletions in meningiomas defined a candidate region of >1 Mbp in 22q12.3-q13.1 and directed us to gene cloning in this segment. We characterized a new member of the N-acetylglucosaminyltransferase gene family, the LARGE gene. It occupies >664 kilobases and is one of the largest human genes. The predicted 756-aa N-acetylglucosaminyltransferase encoded by LARGE displays features that are absent in other glycosyltransferases. The human like-acetylglucosaminyltransferase polypeptide is much longer and contains putative coiled-coil domains. We characterized the mouse LARGE ortholog, which encodes a protein 97.75% identical with the human counterpart. Both genes reveal ubiquitous expression as assessed by Northern blot analysis and in situ histochemistry. Chromosomal mapping of the mouse gene reveals that mouse chromosome 8C1 corresponds to human 22q12.3-q13.1. Abnormal glycosylation of proteins and glycosphingolipids has been shown as a mechanism behind an increased potential of tumor formation and/or progression. Human tumors overexpress ganglioside GD3 (NeuAcalpha2,8NeuAcalpha2, 3Galbeta1,4Glc-Cer), which in meningiomas correlates with deletions on chromosome 22. It is the first time that a glycosyltransferase gene is involved in tumor-specific genomic rearrangements. An abnormal function of the human like-acetylglucosaminyltransferase protein may be linked to the development/progression of meningioma by altering the composition of gangliosides and/or by effect(s) on other glycosylated molecules in tumor cells.

Figure 1

Structure of the human LARGE gene. The extent of the deletions in tumors 11 and 119A is shown on the left side. The dashed bar indicates the location of centromeric breakpoints of deletions (4). Markers KI-1186 and KI-844 are retained and deleted, respectively, in both tumors. Cosmids shown in this figure represent only a fraction of all cosmids identified in the course of contig construction. The exact length of cosmid clones and extent of overlap between each cosmid step in the primary contig has not been determined. Clones sequenced from this region are shown in the sequenced contig, which is composed of cosmids, bacterial artificial chromosomes, and one PAC, indicated by prefixes “c”, “b”, and “dJ”, respectively. The vertical bar for each sequenced clone is proportional to the amount of sequence data generated from each clone and is indicated in parentheses and in kilobase pairs, next to the clone name. Clones cE140F8, cN120B6, cN2E9, and bK566G5 are not yet fully sequenced. Accession numbers for sequenced clones are bK282F2, AL008630; cB1D7, Z82173; bK1216H12, AL008715; bK566G5, AL023577; cE140F8, Z82179; cN37D7, Z73421; cN120B6, Z73987 and Z73988; cN4F11, Z69943; cB33D11, AL008640; cE78G1, Z70288; cN117F11, Z97354; cE95B1, Z69042; cN38E12, Z68287; dJ75E8, Z76736; cN20A6, Z69713; cN7A10, Z68324; cN2E9, Z68685, Z68686, and Z68286; cN73A10, Z49866; cN13E1, Z54073; cN53F3, Z77853; cN74G7, Z69715; cN32F9, Z73429; cN116A5, Z69925; and cE110C7, Z68223. “Ter” indicates the direction of the telomere. The human LARGE gene is composed of 16 exons shown by filled rectangles on the right side. The sizes (in base pairs) for each exon (ex) and intron (int) are shown in parentheses.

Figure 2

Expression pattern of the LARGE gene in human and mouse. Dark field autoradiograms illustrate the gene expression in mouse embryo (A) and adult brain (B) by using mRNA in situ hybridization. The antisense probe of the mouse Large cDNA sequence was hybridized to a sagittal section of a day-17.5 mouse embryo (A) and to a coronal section from mouse adult brain (B). Note the ubiquitous pattern of gene expression with a strong signal in heart (ha), central nervous system structures such as cerebral cortex (cx), hippocampus (hip), olfactory lobe (ol), trigerminal ganglion (trig), and spinal cord (sc) as well as in diaphragm (dia) and duodenum (du). As a negative control, the sense probe was hybridized to contiguous sections (C and D). (E) Ubiquitous expression of the LARGE gene in human tissues as an ≈4.5-kb transcript. The entire cDNA was used as probe on Northern blot containing poly(A)⁺ selected mRNA from human adult tissues (MTN 7760–1, CLONTECH): Lanes: 1, heart; 2, brain; 3, placenta; 4, lung; 5, liver; 6, skeletal muscle; 7, kidney; 8, pancreas.

Figure 3

Alignment of predicted amino acid sequences of the three LARGE genes from human (h.), mouse (m.), and C. elegans (c.). Identity and similarity are indicated by black and gray boxes, respectively. White boxes indicate nonconservative amino acid changes and dashes (−) indicate gaps. The positions of exon/intron borders of the human and C. elegans genes are indicated by vertical arrows above the human sequence and below the C. elegans sequence. Four protein domains of the human LARGE protein predicted by computer-assisted sequence analysis are marked: a signal peptide (amino acids 1–24), a transmembrane domain (amino acids 20–28), and two coiled-coil domains (amino acids 55–90 and 422–441). The portion of the human Large sequence between two asterisks (∗) shows sequence similarity with the part of the human i-β-1,3-N-acetylglucosaminyltransferase (amino acids 91–408, acc. AF029893).

Figure 4

FISH of PAC 396N1 to mouse metaphase spreads. (a) The green arrow indicates hybridization of PAC 396N1 to mouse chromosome 8 (red arrowhead). (b) PAC 396N1 was localized to chromosome 8C1 by using an inverted 4,6-diamino-2-phenylindole banding pattern.