Human O-GlcNAc transferase (OGT): genomic structure, analysis of
splice variants, fine mapping in Xq13.1
Dagmar Nolte, Ulrich Mu¨ller
Institut fu¨r Humangenetik, Justus-Liebig-Universita¨t, Schlangenzahl 14, 35392 Giessen, Germany
Received: 9 July 2001 / Accepted: 22 August 2001
O-linked N-acetylglucosamine (O-GlcNAc) transferase (OGT)
catalyzes polypeptide glycosylation by the addition of N-
acetylglucosamine to serine and threonine residues. This process is
referred to as O-GlcNAcylation and has been observed to occur on
many nuclear and cytosolic proteins. These include RNA-
polymerase II, nuclear pore proteins, neurofilaments, microtubule-
associated proteins such as tau, synapsin I, and both onco- and
tumor suppressor proteins. O-GlcNAcylation is a dynamic process
and appears to play an important regulatory role in the cell com-
parable to phosphorylation. In some cases, O-GlcNAcylation and
phosphorylation even have reciprocal functions on the same mol-
ecule, e.g., in the regulation of transcription by modification of
RNA polymerase II (reviewed by Wells et al. 2001).
OGT has been reported to be a heterotrimer composed of two
catalytic 110-kDa subunits and of one 78-kDa subunit. There is
evidence that the 78-kDa subunit originates from the larger subunit
by a combination of alternative RNA splicing and specific prote-
olysis (Kreppel et al. 1997). An important characteristic of OGT
are tetratricopeptide repeats (TPR) in its N-terminal portion. TPRs
are composed of subunits of 34 amino acids that contain the
loosely conserved residues W-L-G-Y-A-F-A-P (Lamb et al. 1995;
Das et al. 1998). Crystal structure analysis suggests that each re-
peat forms a pair of antiparallel ␣-helices that act as superhelical
structure in protein-protein interactions (Das et al. 1998).
OGT is highly conserved in all eukaryotes tested, including
cenorhabditis, mouse, rat, and humans. While the transcripts re-
ported are highly homologous in their 3Ј portions among the vari-
ous species, there appear to be discrepancies at the 5Ј ends. Thus,
the purported full-length human OGT (hOGT) cDNA codes for a
protein of 920 amino acids (Lubas et al. 1997). This is significantly
smaller than the protein of 1037 amino acids encoded by the full-
length rat cDNA (Kreppel et al. 1997). OGT is ubiquitously ex-
pressed, with the highest levels of mRNA detected in human pan-
creas (Lubas et al. 1997). There are several alternative transcripts
of OGT. Four transcripts of 8.0 kb, 6.0 kb, 4.2 kb, and 1.7 kb have
been described in the rat (Kreppel et al. 1997), and transcripts of
9.3 kb, 7.9 kb, 6.3 kb, and 4.4 kb were reported in humans (Lubas
et al. 1997). OGT is encoded by one gene that has been assigned
to the proximal long arm of the X Chromosome (Chr) (Xq13.1) in
humans (Nemeth et al. 1999; Shafi et al. 2000).
In order to clarify the apparent differences in size between rat
and human OGT transcripts, we analyzed the human gene at both
the genomic and transcript level. We describe the complete geno-
mic structure of human OGT, give the composition of its alterna-
tive transcripts, and describe its precise location in Xq13.1 with
respect to flanking genes.
The DNA of a considerable portion of Xq13.1 has been se-
quenced (AL590763), and 18 exons corresponding to the reported
full-length hOGT cDNA (Lubas et al. 1997) were identified in
silico (exons 6–23 of Fig. 1). We screened a human fetal brain 5Ј
stretch plus cDNA library (Clontech, Cat. HL5015b) with a 490-bp
probe corresponding to exons 8–11 (Fig. 1). Three different clones
with inserts of 1.6 kb, 2.0 kb, and 2.3 kb were obtained. The 2.0-kb
insert revealed an additional 700-bp 5Ј sequence, but did not con-
tain the published start codon. Comparison of this 700-bp cDNA
sequence with the genomic sequence (AL590763) revealed addi-
tional four exons (exons 1–4 of Fig. 1), with exon 1 containing an
ATG start codon. These four exons have not been described as part
of hOGT previously.
In an attempt to isolate phage clones with larger inserts of
hOGT cDNA, we screened the human fetal brain large insert
cDNA library (Clontech, Cat. HL5504U), using exons 1–3 (Fig. 1)
as a probe. We purified seven different phage clones with insert
sizes ranging from 4.2 kb to 5.6 kb. Analysis of the cDNA inserts
revealed alternative splicing of exon 2. This exon exists as a 181-
bp variant (exon 2a) and a truncated in-frame variant of 151 bp
(exon 2b), in which the first 10 amino acids encoded by exon 2a
are deleted. We also discovered phage inserts with and without the
3.27-kb intervening sequence (IVS4) between exons 4 and 6 (Fig.
1). Presence and absence of this IVS was independent of the type
of exon 2 used. Although this IVS4 (Fig. 1) does not include an
open reading frame (ORF), several ESTs (i.e., T78105, L44477)
have been mapped to this region. Since IVS4 can be transcribed
(also see below), we alternatively refer to it as exon 5 (Fig. 1).
In order to analyze all splice variants of OGT, we performed
Northern blot analyses using a commercially available blot (Clon-
tech, Cat # 7760-1). Hybridization with a probe derived from
exons 13–16 (Fig. 1) revealed five transcripts of 9.5 kb, 8.0 kb, 6.4
kb, 4.4 kb, and 4.2 kb (Fig. 2A). This corresponds to the previously
described alternative transcripts (Kreppel et al. 1997; Lubas et al.
1997). Unlike previous studies that reported one transcript of about
4.4 kb (Lubas et al. 1997) or about 4.2 kb (Kreppel et al. 1997)
only, we detected two transcripts of 4.2 kb and 4.4 kb, possibly
owing to a higher resolution. Expression was highest in pancreas,
with the 9.5-kb and 8.0-kb transcripts being most abundant. In
heart, brain, skeletal muscle, kidney, lung, and liver the overall
expression was less and most pronounced for the 9.5-kb and 6.4-kb
transcripts (Fig. 2A).
Using exons 1–3 as a probe (Fig. 2B), a hybridization pattern
comparable to that obtained with a probe derived from exons 13–
16 (Fig. 2A) was detected. The exception was very faint hybrid-
ization to the 4.4-kb and 4.2-kb transcripts. The results clearly
demonstrate that the human 9.5-kb, 8.0-kb, and 6.4-kb transcripts
contain the proximal exons that we reported here (see above).
A strikingly different hybridization pattern was obtained with
a probe generated from EST T78105 in the distal portion of exon
5 that is devoid of an ORF (Fig. 1; see above). This probe detects
only the two largest transcripts of 9.5 kb and 8.0 kb (Fig. 2C).
Another probe from a more proximal region of exon 5 resulted in
The nucleotide sequence data reported in this paper have been submitted to
GenBank and have been assigned accession number AJ315767.
Correspondence to: U. Mu¨ller; E-mail: ulrich.mueller@humangenetik.
Mammalian Genome 13, 62–64 (2002).
© Springer-Verlag New York Inc. 2002
Incorporating Mouse Genome