Contribution of noncoding pathogenic variants to
RPGRIP1-mediated inherited retinal degeneration
Farzad Jamshidi, M.D., Ph.D.
1
, Emily M. Place, M.S.
1
, Sudeep Mehrotra, M.S.
1
,
Daniel Navarro-Gomez, B.Sc.
1
, Mathew Maher, M.S.
1
, Kari E. Branham, M.S.
2
, Elise Valkanas, B.A.
3
,
Timothy J. Cherry, Ph.D.
4
, Monkol Lek, Ph.D.
3,5
, Daniel MacArthur, Ph.D.
3,5
,
Eric A. Pierce, M.D., Ph.D.
1
and Kinga M. Bujakowska, Ph.D.
1
Purpose: With the advent of gene therapies for inherited retinal
degenerations (IRDs), genetic diagnostics will have an increasing
role in clinical decision-making. Yet the genetic cause of disease
cannot be identified using exon-based sequencing for a significant
portion of patients. We hypothesized that noncoding pathogenic
variants contribute significantly to the genetic causality of IRDs and
evaluated patients with single coding pathogenic variants in
RPGRIP1 to test this hypothesis.
Methods: IRD families underwent targeted panel sequencing.
Unsolved cases were explored by exome and genome sequencing
looking for additional pathogenic variants. Candidate pathogenic
variants were then validated by Sanger sequencing, quantitative
polymerase chain reaction, and in vitro splicing assays in two cell
lines analyzed through amplicon sequencing.
Results: Among 1722 families, 3 had biallelic loss-of-function
pathogenic variants in RPGRIP1 while 7 had a single disruptive
coding pathogenic variants. Exome and genome sequencing
revealed potential noncoding pathogenic variants in these 7
families. In 6, the noncoding pathogenic variants were shown to
lead to loss of function in vitro.
Conclusion: Noncoding pathogenic variants were identified in 6 of
7 families with single coding pathogenic variants in RPGRIP1. The
results suggest that noncoding pathogenic variants contribute
significantly to the genetic causality of IRDs and RPGRIP1-
mediated IRDs are more common than previously thought.
Genetics in Medicine (2019) 21:694–704; https://doi.org/10.1038/s41436-
018-0104-7
Keywords: Inherited retinal degeneration; Noncoding patho-
genic variants; RPGRIP1; Intronic pathogenic variants; genome
sequencing
INTRODUCTION
Inherited retinal degenerations (IRDs) are a group of
monogenic diseases that are the most common cause of
blindness in the working age population.
1
About 260 genes
have been associated with IRDs with functions
spanning almost every aspect of cellular function, from
splicing machinery, to microtubular transport and phot o-
transduction.
1
State-of-the-art clinical diagnostics using
next-generation sequencing (NGS) of known IRD genes
successfully identifies the causal pathogenic variant (PV) in
only 50 to 70% of cases.
2,3
Although copy-number changes
4
and intronic PV
5
contribute to diseas e, they are not routinely
assessed and likely contribute to the genetic causality in
a significant portion of currently unsolved cases. With
the advent of successful gene therapies for IRDs,
6
understanding such noncoding PV and developing assays
to evaluate them is of increasing importance. One example is
the autosomal recessive RPGRIP1-associated disease, which is
an attractive candidate for gene therapy with already
established success in murine
7
and canine
8
models. Yet,
almost all of the PV in RPGRIP1 have been described in the
coding region.
9
RPGRIP1 plays a critical role in opsin trafficking, outer-
segment disc organization and photoreceptor survival.
10,11
While it primarily localizes to the transition zone of rods and
cones, various of its isoforms can be found in the outer
segment, along the micr otubules as well as in the amacrine
cells of the inner plexifo rm layer.
12,13
Its largest transcript
variant, NM_020366, is composed of 3861 coding base pairs
distributed over 24 exons.
14
This encodes a 1287–amino acid
Submitted 3 April 2018; accepted: 15 June 2018
Published online: 3 August 2018
1
Ocular Genomics Institute, Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston, Massachusetts, USA;
2
Department of
Ophthalmology and Visual Sciences, University of Michigan Medical School, Ann Arbor, Michigan, USA;
3
Program in Medical and Population Genetics, Broad Institute of MIT and
Harvard, Boston, Massachusetts, USA;
4
Center for Developmental Biology and Regenerative Medicine, Seattle Children’s Research Institute and University of Washington,
Department of Pediatrics, Seattle, Washington, USA;
5
Analytic and Translational Genetics Unit, Massachusetts General Hospital, Boston, Massachusetts, USA. Correspondence:
Eric A. Pierce (eric_pierce@meei.harvard.edu) Kinga M. Bujakowska (kinga_bujakowska@meei.harvard.edu)
ARTICLE © American College of Medical Genetics and Genomics
694 Volume 21
|
Number 3
|
March 2019
|
GENETICS in MEDICINE
protein that interacts with a variety of other IRD proteins
such as retinitis pigmentosa GTPase regulator (RPGR),
SPATA7, and NPHP4
15
. The expression of RPGRIP1 is
limited to the retina and testis.
16
Confident genetic diagnosis with RPGRIP1 as the causal
gene will be crucial for effective clinical trials of potential
therapies. In our analysis of IRD families with targeted panel
sequencing of coding regions of IRD-associated genes,
17
we
repeatedly noted identification of single likely pathogenic
variants in RPGRIP1 in families without PV in other IRD
disease genes. To test the hypothesis that PV in RPGRIP1 are
the likely cause of disease in these families, we performed
exome (ES) and genome sequencing (GS) to search for
noncoding PV and structural variations accounting for the
loss of function (LoF) of the second allele.
MATERIALS AND METHODS
Ethical guidelines
The study was approved by the institutional review board at the
Massachusetts Eye and Ear (Human Studies Committee MEE in
USA) and adhered to the Declaration of Helsinki. Informed
consent was obtained from all individuals on whom genetic
testing and further molecular evaluations were performed.
Clinical evaluation
All the patients in this study underwent clinical assessment by
ophthalmologists subspecializing in inherited retinal degen-
erations. The clinical characteristics are outlined in Table 1.
Sequencing
DNA was extracted from venous blood using the DNeasy
Blood and Tissue Kit (Qiagen, Hilden, Germany). All samples
underwent GEDi sequencing as described previously.
17
Exome and genome sequencing were done at the Center for
Mendelian Genomics at the Broad Institute of MIT and
Harvard using methodology described previously.
18
Sanger
sequencing was performed on ABI 3730xl using BigDye
Terminator v3.1 kits (Life Technologies, Carlsbad, CA) and
using polymerase chain reaction (PCR) primers indicated in
Supplementary Information. When PCR products were
sequenced, they were purified prior to sequencing (ExoSap-
IT, Affymetrix, Santa Clara, CA). Gel bands that were Sanger
sequenced had DNA extracted via the Zymoclean
TM
Gel
DNA Recovery Kit (Zymo Research, Irvine, CA).
Bioinformatics
Analyses of DNA sequence data were performed as described
previously.
17,19
Briefly, Burrows–Wheeler Aligner (BWA) was
used for alignment. SAMtools and custom programs were used
for single-nucleotide polymorphism and insertion/deletion calls.
19
Variants of interest were limited to polymorphisms with less than
0.005 allelic frequency in the gnomAD (http://gnomad.
broadinstitute.org/)andExAC(http://exac.broadinstitute.org/)
databases.
18
Genome copy-number analysis, with consideration
of structural changes, was done using Genome STRiP 2.0 (ref.
20
).
For the anal ysis of splicing patterns from amplicon sequencing,
STAR (version 2.5.3a) aligner
21
was used to generate an index of
the human genome (GRCh37.75.dna.primary_assembly.fa) and
to align the reads. Integrative Genomics Viewer (IGV)
22
was used
to load the aligned sequences (BAM files) and for data
visualization with Sashimi plots.
PCR, cloning, and site-directed mutagenesis
PCR was performed using PfuUltra II Fusion polymerase
(Agilent Technologies, Santa Clara, CA) on genomic DNA of
patients harboring the PV of interest (primers are listed in the
Supplementary Material). The PCR products were cloned into
pENTR Directional TOPO vector (Thermo Fisher, Waltham,
MA) and used to transform chemically competent Escherichia
coli (One Shot TOP10, Thermo Fisher, Waltham, MA).
Plasmid DNA from single colonies was extracted with
miniprep kits (ZymoPURE, Zymo Research) and analyzed
by restriction enzyme digestion with BsrGI (NE Biolabs,
Ipswich, MA) and Sanger sequencing. Essential splice-site PV
were introduced by site-directed mutagenesis (QuickChange
II Site Directed mutagenesis kit, Agilent Technologies) and
verified by Sanger seque ncing. Colonies with the correct
sequence and restriction enzyme pattern were then subcloned
into the pCS2 + GW vector (kind gift from Er ica Davis) via
Gateway LR clonase II (Thermo Fisher) and similar analyses
as before was done to isolate vectors with the appropriate
inserts for transfection experiments. The final vector included
RPGRIP1 exons 11–16 including extensions into introns 10
and 16 on the 5′ and 3′ ends, which was cloned into pCS2 +
GW and used for splicing assays.
Quantitative polymerase chain reactions (qPCR)
Five nanograms (ng) of genomic DNA, 200 nM of each
primer, and 10 µl of Fast SYBR Green Maste r Mix (Life
Technologies, Grand Island, NY) were used for qPCR
reactions, which were perform ed on a Stratagene Mx3000P
instrument (Agilent Technologies) using the standard ther-
mocycling program (95 °C for 3 min, 40 cycles of 95 °C for 20
s, and 60 °C for 1 min, followed by a melting curve). The
ddCT method was used for the analysis of results where
ZNF80 was used as a reference gene and an in-house DNA
sample with wild-type RPGRIP1 (OGI081-200) used for
normalization. Each sample was tested in triplicate and the
average value was used. Standard deviation with error
propagation was used to calculate up and down errors.
Cell culture and transfections
Human embryonic kidney (HEK293T) and retinoblastoma
(WERI-Rb1) cells purchased from American Type Culture
Collection (ATCC, Manassas, VA) and maintained in RPMI
medium supplemented with 10% fetal bovine serum 1640
(Thermo Fisher). Then, 2 ml of 5 × 10
5
cells/ml were plated
into each well of a 6-well plate (Corning Inc., Corning, NY)
12 h prior to transfections; 1–5 μg of vector DNA per well was
used for transfections using a commercial reagent (Lipofecta-
mine 2000, Invitrogen, Carlsbad, CA). Cells were harvested
for RNA extraction 48 h after transfection.
JAMSHIDI et al ARTICLE
1234567890():,;
GENETICS in MEDICINE
|
Volume 21
|
Number 3
|
March 2019 695
Table 1 Clinical characteristics and PV in RPGRIP1-mediated inherited retinal degeneration patients
ID
a
Gender Age at Dx Variant 1
b
MAF
c
Variant 2
b
MAF
c
Signs
d
Method
237–523 F Early
childhood
c.3793ins4(p.V1265Gfs*19) 4×10
-
6
Exon1-2 dup N/A Keratoconus, PSC, asteroid hyalosis, ↓ CV GEDiGS
281–608 M Early
childhood
c.1615_1624del10(p.
E539Qfs*2)
4×10
-
6
Exon2 dup N/A Nystagmus, ON atrophy, ↓ CV GEDiGS
601–1236 M Infancy c.3238+1G>A 0 c.1611+27G>A 0 Nystagmus, scatter hypopigmented spots in retinal
periphery
GEDiGS
949–1907 M Infancy c.3618-1_3621del5 0 c.1468-263G>C 0 Nystagmus, peripheral atrophy, pigmentary macular
changes
GEDiGS
827–1591 M 1.5 yo c.895_896del2(p.G299Sfs*21) 0 c.2367+23delG 2.3×10
-
4
Nystagmus, macular atrophy GEDiES
1797–3128 M 15 yo c.2302C>T (p.R768*) 2×10
-
5
Exon19 del N/A GEDi
79–194 F Infancy c.3793ins4(p.V1265Gfs*19) 4×10
-
6
c.3793ins4(p.V1265Gfs*19) 4×10
-6
Peripheral pigmentary changes, bull’s eye macular changes GEDi
501–336 F 4 months c.2302C>T(p.R768*) 2×10
-
5
c.711_711del1(p.
P237Pfs*40)
0 Nystagmus GEDi
690–1378 F Infancy c.1084_1087del(p.
E362NAfs*12)
0 c.767C>Gp.S256* 0 Nystagmus GEDi
GEDi is the genetic eye disease (GEDi) diagnostic test
17
CV color vision, del deletion, Dup duplication, MAF minor allele frequency, N/A not available, ON optic nerve, PSC posterior subcapsular cataract, ES exome sequencing, GS genome sequencing, yo years old
a
IDs are presented as Ocular Genomics Institute (OGI) family numbers followed by the individual patient number
b
RPGRIP1 PV in each of two alleles, with protein alteration in parentheses
c
Minor allele frequency based on the Genome Aggregation Database (gnomAD)
d
All patients exhibited diminished visual field, visual acuity, and electroretinography signals in addition to attenuated vessels and bone spicules on fundoscopy. Additional finding are indicated in the table
ARTICLE JAMSHIDI et al
696 Volume 21
|
Number 3
|
March 2019
|
GENETICS in MEDICINE
RNA isolation and complementary DNA synthesis
Cells were lysed with TRIzol (Thermo Fisher). After
1-Bromo-3-chloropropane or chloroform (Sigma-Aldrich, St.
Louis, MO) treatment, the aqueous phase was transferred to
mRNeasy columns with DNase I digestion performed on-
column (Qiagen). Quantification was performed via
NanoDrop (Thermo Fisher) and 500 ng of RNA was
converted to complementary DNA (cDNA) using oligo (dT)
primers and SuperScript II (Thermo Fisher).
Splicing assay and amplicon sequencing
The mutant, control, and wild-type vectors descri bed above
M1: c.1615_1624de110
OGI-281
1.5
645
646
608
1
0.5
0
M2: dup[ex2]
I
II
645
[=],[M2]
646
[M1],[M2]
608
Sequenced
Mutant
Aligned
Reference
f
Intron2
135bp gap
Intron2 Intron1
21297360-21297372 21290000-2129003221297200-21297224
2
RPGRIP1
608
[M1],[M2]
608
500bp
300bp
1000bp
645 646 293T
FR PCR
F
Cntrl PCR
21280000 21285000 21290000 21295000 21300000
RPGRIP1
23
R
de
c
ab
JAMSHIDI et al ARTICLE
GENETICS in MEDICINE
|
Volume 21
|
Number 3
|
March 2019 697
were transfected into HEK293T and WERI-Rb1 cells. Two
days posttransfection, cDNA was generated as described and
reverse transcription polymerase chain reaction (RT-PCR)
performed amplifying the flanking exons of the point PV of
interest. The PCR products were purified by DNA-clean and
concentrator kits (Z ymo Research). Amplicon sequencing
was then performed at the Massachusetts General Hospital
Center for Computational and Integrative Biology, where the
PCR products were fragmented using sonication and
sequenced with the standard NGS pipeline. Visualization
and analysis of the data were performed as described under
bioinformatics.
Epigenetic features surrounding the RPGRIP1 locus
ATAC-Seq and ChIP-Seq were performed according to
previously published methods.
23,24
Briefly, human retinal
tissue was obtaine d from donors 25–65 years old with a
postmortem interval <8 h (Lions Vision Gift, Portland, OR).
Approximately 20,000 nuclei were isolated for ATAC-Seq and
the transposition reaction was performed for 60 min at 37 °C.
ChIP-Seq was performed on approximately 25 million cells
per reaction using the following antibodies: CTCF (Abcam
AB70303); H3K4me2 (Abcam AB7766 lot GF160184-1; RRI D
AB_2560996); H3K27ac (Abcam AB4729, lot GF150367-1;
RRID AB_2118291); CRX (Santa Cruz, B11X, lot E1409);
OTX2 (Abcam, AB21990, lot GR242019-1); NRL (Abcam,
AB137193, lot GR104520-2); RORB (Diagenode, pAb-001-
100, lot HS-0010); MEF2D (Greenberg lab, 2373). All
sequencing was performed on an Illumina NextSeq500 to a
depth of >10 M reads. Mapping, alignment, and normal-
ization of reads was performed as previously described.
24
Genome tracks were displayed using the University of
California–Santa Cruz Genome Browser.
RESULTS
Genetic analysis of 1722 IRD probands who underwent
targeted exon sequencing of known IRD genes
17
revealed 3
patients with biallelic loss-of-function (LoF) PV in RPGRIP1
and 7 with only one LoF change in this gene (Table 1, Fig. S1).
In the latter 7 families no other significant PV in RPGRIP1 or
other IRD genes were identified. Because they were all
diagnosed with an early-onset IRD, a characteristic presenta-
tion of RPGRIP1 disease,
25
further testing was performed to
search for noncoding and structural variants in RPGRIP1 or
other IRD genes through exome and genome sequencing.
Among these 7 families, three of the second-allele PV were
predicted to be copy-number changes and three were intronic
PV (Table 1). In one family, OGI-578, we did not validate
second pathogenic variants in RPGRIP1 (Fig. S1).
ES and GS studies showed that three families have
structural variants as the second PV in RPGRIP1. Analysis
of ES data showed that patient OGI-1797-3128 had a
predicted deletion of exon 19 (Table 1), which was confirmed
by qPCR (Fig. S2). Exon 19 has 139 nucleotides, and thus its
deletion is predicted to lead to a frameshift resulting in a
premature stop codon and likely subsequent nonsense-
mediated decay (NMD).
26
In OGI-281-608, the gain of a
copy of exon 2 was detected through structural analysis
20
in
addition to coverage-based predictions of GS data (Fig. 1).
The stru ctural change led to misalig nment of paired-end
reads (Fig. 1c, d), which indicated a tandem duplication. This
PV was validated by qPCR (Fig. 1b), and the predicted
breakpoint confirmed through PCR and Sanger sequencing
(Fig. 1e, f). Sanger sequencing also identified 135 bp of
missing DNA upstream of the breakpoint, sugge sting a
possible complex rearrang ement as the causal event.
27
Given
that exon 2 is 133 bp, its tandem duplication would lead to a
LoF allele.
Patient OGI-237-523 similarly had a tandem copy-number
gain but in both exons 1 and 2 detected by ES (Fig. 2a, b). The
5′ breakpoint was mapped 2 Kb upstream of exon 1 (Figs. 2c,
S3). However, given that the second copy would maintain 2
Kb upstream of the exon 1, which would include a proximal
presumed promoter (Fig. 2c), we questioned whether a second
transcriptional start site (TSS) within the duplicated 5′
upstream region would exclude the mutan t exons 1′ and 2′
thus leading to normal transcription. Therefore, we hypothe-
sized that perhaps a critical RPGRIP1 regulatory domain
exists outside of the 2-Kb region upstream of exon 1. Review
of transcriptome data of normal human retina
28
revealed an
Fig. 1 Exon 2 duplication in OGI-281. (a) Pedigree of the family showing deceased parents and the three siblings all of whom were analyzed.
(b) Quantitative polymerase chain reaction (qPCR)-based copy-number results along the first three exons of RPGRIP1. All three siblings have a duplication of
exon 2 in an RPGRIP1 allele. Exons 1 and 3 are not affected. The bottom panel shows the locations of RPGRIP1 exons based on the NM_020366 transcript.
(c) Integrative Genomics Viewer
22
(IGV) view of the sequenced genome sequencing (GS) reads where the duplication was discovered for OGI-281-608. The
bottom of the figure shows the location of exon 2 of RPGRIP1. The gray thick arrows correspond to expected paired-end reads. The green thick arrows are
mapped reads that have aligned abnormally and hint to a PV. (d) Schematic explanation of the how genomic duplication would lead to the abnormal paired-
end reads seen in (c). The gray region is the area of hypothetical duplication, while the green thick arrows are the paired-end reads that will align
abnormally. The top of the figure shows what is actually sequenced in the mutant sample, while the bottom shows how such sequenced reads would map
to a reference wild-ty pe (WT) model. The aligned paired-end reads of the mutant will have a greater distance between them and will point away from one
another as seen in (c). The dark and light green hash lines correspond to the aligning sequenced of the paired-end reads. The primers used for (e) are shown
as blue and red arrows. Their directionality is indicated relative to the mutant (top) and WT models (bottom). (e) Polymerase chain reaction (PCR) across the
predicted duplication breakpoint using primers represented in (d). Presence of a tandem duplication would yield a product while its absence would lead to
no amplification as the primers would be pointing away from each other. The predicted duplication is present in all OGI-281 family members while it is
absent in HEK293 T cells. The control (Cntrl) PCR on the bottom was done to ensure that larger products could be amplified from all samples thus ensuring
DNA fragmentation or quality was not a confounding factor. (f) Sanger sequencing identifying the exact breakpoint (black arrowhead) using OGI-281-608
PCR product from (e). There is a 135-bp deletion, upstream of the breakpoint.
ARTICLE JAMSHIDI et al
698 Volume 21
|
Number 3
|
March 2019
|
GENETICS in MEDICINE
