Alternative transcripts of the SERPINA1 gene in alpha-1 antitrypsin deficiency

QT-PCR assays for detection of alternative spliced SERPINA1 isoforms

To quantify specific splicing transcripts of the SERPINA1 gene we developed a novel
QT-PCR method (Figure 1). For this purpose exon–exon boundary spanning primer and primers located in alternative
spliced exons 1B and 1C were designed. Figure 1b shows the different splicing transcripts that are detected using each of the QT-PCR
assays (1A, 1B and 1C).

Only transcript I contains the exon 1A directly joined to exon 2. Therefore, we employed
a simple strategy using a primer spanning the exon 1A to exon 2 junction, and as illustrated
in Figure 1b, the 1A assay specifically quantifies the expression of transcript I. On the other
hand, the boundary between exon 1C and exon 2 is not specific and it is included in
several transcript variants (II, III, IV and V), which makes impossible individual
quantification of transcripts containing this region. Therefore, the 1C assay amplifies
and simultaneously detects several transcripts II, III, IV and V. Similarly, the 1B
assay, which is developed to detect and quantify the boundary between exon 1B and
exon 1C, allows only the combined quantification of transcripts III and IV (Figure 1b).

Pattern of expression of SERPINA1 isoforms in human tissues

The expression levels of different SERPINA1 transcripts were measured in a panel of
various human tissues. Compared to other tissues, leukocytes showed a highest expression
of the 1A, 1B and 1C transcripts (Figure 2). The transcript I, specifically detected with the 1A assay, was mainly expressed
in leukocytes and lung tissue but it was also detected in liver, colon, spleen, prostate,
kidney or brain at much lower level (Figure 2). In addition to leukocytes, 1C transcripts were also expressed in lung, liver and
kidney and less abundantly expressed in pancreas and small intestine (Figure 2). The expression of transcripts detected by using 1B assay was almost exclusively
restricted to leukocytes, although low expression was also found in lung and spleen.
The commercial cDNA samples represent a pooled preparation from many individuals,
therefore it is not possible to know which specific cell in the tissue samples is
expressing the AAT transcripts. Further experiments using isolated cells remain to
be performed.

Figure 2. Transcription analyses of SERPINA1 transcripts in human tissues. Relative expression
levels of 1A, 1B and 1C transcripts in 16 human tissues is shown. Specific patterns
of transcript expression were detected in different tissues. Leukocytes exhibited
higher expression levels of transcripts 1A, 1B, and 1C when compared to other tissues.
The 1A transcript was found predominantly in leukocytes and lung tissue. Specific
expression of 1C transcripts were detected in lung, liver, kidney or pancreas, suggesting
that transcription in these tissues start in the transcription site located in exon
1C. Expression of 1B transcripts were almost exclusively detected in leukocytes.

SERPINA1 expression in leukocytes from peripheral blood by RT-PCR

White blood cells isolated from peripheral blood samples were used to analyze SERPINA1
expression. Since alternative splicing occurs between exons 1A, 1B and 1C, we used
forward primers located in these exons and antisense primers in exon 2 (Figure 3). We previously found a unique expression fragment 28] corresponding to a transcript, in which the exon 1A is directly spliced to join exon
2 by skipping of the exons 1B and 1C (Figure 3b). This transcript was initiated in any of the two transcription sites described
within exon 1A 7]. However, with the primer located in exon 1B, several transcription products of different
size were observed, reflecting the use of the different splicing sites located within
exons 1B and 1C (Figure 3c). These products most likely begun in the initiation site described within exon
1B 7] since no alternative transcripts including exon 1B or 1C were detected using the
primer in exon 1A. Finally, as expected, amplification using primer in exon 1C generated
only one fragment corresponding to joined exon 1C to exon 2 (Figure 3d). This region is present in all transcripts containing the exon 1C. Hence, we cannot
determine whether its expression derived from the transcription site within exon 1C
or within exons 1A and 1B. Since our analysis was performed on white blood cells,
this region most likely corresponds to transcripts starting at any of the monocyte
transcription sites. Therefore, we think that leukocytes have active 1A and 1B transcription
start sites.

Figure 3. Representative expression of alternative transcripts in blood samples detected by
RT-PCR. a Schematic graphic of the structure of the SERPINA1 gene. Primers used for RT-PCR
expression analysis are represented by arrows. RT-PCR expression analysis in leukocytes from peripheral blood samples showed evidence
for different alternative splicing event occurring within noncoding exons 1A, 1B and
1C of SERPINA1 gene. b Amplification fragment corresponding to the expression in blood samples detected
using a forward primer in exon 1A and a reverse primer in exon 2. A single band of
633 bp was detected which corresponded to a transcript containing exon 1A joined directly
to exon 2. c Bands detected by using forward primer 1B. Different expression fragments were detected
corresponding to splicing variants. The structure of these splicing variants has been
described in detail before 28]. Three different bands are clearly visualized due to the different splicing sites
used within exons 1B and 1C. The band of 712 bp corresponded to the exon 1B joined
to exon 2. The band of 816 bp corresponded to an isoform containing both exons 1B
and 1C. The upper band of 971 corresponds to a fragment with retained intron in between
exons 1B and 1C. d Using a forward primer in the exon 1C a single band of 587 bp was amplified.

Effect of SERPINA1 mutations on alternative transcripts expression

Three QT-PCR assays (1A, 1B and 1C, Figure 1) were applied on the peripheral blood samples from 33 AAT deficient subjects and
7 controls with normal MM genotype of AAT (Figure 4a). Subjects were carriers of heterozygous or homozygous variants of AAT (Z, S, MMalton,
MProcida) or specific null alleles (QOMattawa, QOPorto, QOMadrid, QOBrescia or MVarallo)
(Table 1). Notably, controls showed similar expression levels of transcripts detected either
by 1A, 1B and 1C assays, although expression of the transcript I, detected by the
1A assay, was slightly higher as compared to 1B and 1C transcripts (Figure 4b). When compared to controls, ZZ patients showed no difference in the transcript
levels. Interestingly, the level of 1A transcript appeared higher in cases carrying
either one or two deficient AAT alleles (Figure 4b), and even higher in null cases caused by the splicing mutations but not by stop
gain mutations (Figure 4a). Nevertheless, most cases with deficient alleles, such as heterozygous MMalton
or MProcida alone or in combination with S or Z alleles, showed no significant changes
in expression of the 1B and 1C transcripts (Figure 4b). This supports the notion that Z and other deficiency mutations do not affect the
transcription of AAT. However, when compared to MM controls, null cases showed reduced
levels of transcripts. Transcripts 1B and 1C were significantly reduced in carriers
of either one or two null alleles with splicing mutations (QOPorto and QOMadrid) affecting
the splicing donor site of intron 1C. In two cases carrying these two splicing mutations
affecting both alleles, 1C transcripts were almost completely absent (Figure 4a). Null alleles caused by stop-gain mutations (QOMattawa and QOBrescia) showed substantial
decrease in all transcripts. However, the null allele Mvarallo in combination with
Z did not display any change in transcript levels.

Figure 4. Transcriptional analyses in cases with normal, deficient or null alleles. a Relative expression of the AAT transcripts detected with the 1A, 1B or 1C assays
in blood samples of subjects with AAT deficiency with deficient or null variants and
normal samples (MM). bBox plots of the levels of expression of the 1A, 1B and 1C transcripts in groups of normal
cases (MM), ZZ patients, cases with one or two deficient alleles other than Z (DEF),
and cases with null alleles (NULL). Median expression is represented by the horizontal line within the box. Cases with null alleles showed statistically significant differences (asterisk) in the expression of 1B and 1C transcripts compared with expression in MM cases.

Correlation between expression of different transcripts and they association with
AAT protein levels

The expression of transcripts detected with 1B and 1C assays showed a significant
correlation (r = 0.83, p  0.0001) in all analyzed cases. This finding suggests that
(1) both assays are detecting the same transcription products and (2) the expression
detected with the 1C assay comes from the monocyte transcription initiation sites
but not from the hepatocyte specific transcript initiating in 1C exon. In contrast,
the expression of 1B and 1C transcripts did not correlate with the different human
tissues analyzed, indicating that the 1C expression, at least in lung, liver, kidney
or pancreas tissues (Figure 2) comes from the hepatocyte transcription start site. The expression of 1A transcript
did not correlate with 1B (r = 0.27, p = 0.104) or 1C (r = 0.25, p = 0.138) expression.
Notably, in peripheral blood cells the 1A transcript was expressed at a higher level
than 1B and 1C transcripts.

In all subjects we found no significant correlation between serum levels of AAT protein
and mRNA of AAT by using 1A or 1C assays. In the other hand, AAT mRNA detected by
the 1B assay showed a weak correlation with serum levels of AAT protein (r = 0.34,
p = 0.04) (Figure 5). When 1B assay was applied for cases with deficient AAT alleles, levels of mRNA
well correlated with serum AAT levels (r = 0.50, p = 0.016). Notably, this correlation
did not exist in null cases.

Figure 5. Correlation between expression level of transcripts and serum AAT levels in AAT deficient
cases. Panels show correlation of expression of 1B transcripts (upper row), 1A (middle row) or 1C transcripts (lower row) with AAT levels in serum. Significant correlation was only found for 1B assay expression,
both when all cases were analyzed together and in cases with deficient alleles, but
not in cases with null alleles. ns Not statistically significant.