https://orcid.org/0000-0002-3011-8654
Abstract
Primary immunodeficiency disorders comprise a rare group of mostly monogenic disorders caused by inborn errors of immunity. The majority can be identified by either Sanger sequencing or next generation sequencing. Some disorders result from large insertions or deletions leading to copy number variations (CNVs). Sanger sequencing may not identify these mutations. Here we present droplet digital PCR as an alternative cost-effective diagnostic method to identify CNV in these genes. The data from patients with large deletions of NFKB1, SERPING1, and SH2D1A are presented.
Introduction
Primary immunodeficiency disorders (PIDs) are a heterogeneous group of mostly, but not exclusively, inherited genetic disorders characterized by impairment of the immune system [1]. Patients with PIDs typically experience recurrent infections and autoimmunity as well as malignancies [2]. Currently there are more than 400 PID genetic disorders curated by the International Union of Immunological Societies (IUIS) expert committee on PIDs [3, 4].
There are multiple overlapping advantages (and some disadvantages) in identifying the genetic basis of PIDs and other genetic disorders. This includes certainty of diagnosis, diagnosis of family members, and preimplantation genetic diagnosis [5–7]. PIDs are also characterized by marked genetic, allelic, and phenotypic heterogeneity [1]. In many cases, similar predisposition to recurrent infections and autoimmunity may be due to mutations of different genes (genocopy, locus heterogeneity). Serial Sanger sequencing of dozens of genes was an inefficient use of valuable resources [5, 8].
The advent of next generation sequencing (NGS) has identified causative single nucleotide variants (SNVs) and copy number variants (CNVs) in PID patients [9–11]. NGS now allows the rapid identification of the genetic basis of disorders with locus heterogeneity such as patients with common variable immunodeficiency (CVID)-like disorders [12]. By definition, patients with CVID do not have a known explanation for their hypogammaglobulinemia [13–15]. Patients with causative mutations are thus deemed to have CVID-like disorders caused by their own genetic defect [15]. NGS gene panels and whole exome sequencing (WES) have become routine in the diagnostic setting.
CNV is an important pathogenic mechanism in some PID-causing genes such as C1 inhibitor deficiency. Sanger sequencing performs poorly when identifying CNV and complex rearrangements as the genetic basis of disorders. Large heterozygous deletions and complex rearrangements may erroneously be diagnosed as showing wild-type sequence using Sanger sequencing. Customized PID microarrays have been described for evaluating CNV in large gene panels [11, 16]. Multiplex ligation-dependent probe amplification (MLPA) remains an integral part of identifying CNV in monogenic PIDs. CNV analysis by MLPA can be expensive if only few patients are tested for each gene. MLPA may not be readily available for certain PID genes.
We have previously described customized testing of PIDs for SNV or small insertion/deletion variants (indels) [6, 7]. In this paper, we present droplet digital PCR (ddPCR, Fig. 1) using probe and EvaGreen DNA binding dye, as an alternative approach for CNV detection. We also discuss the potential application of ddPCR in customized CNV detection in diagnostic settings.
ddPCR workflow (adapted from Bio-Rad Droplet Digital PCR application guide).
Patients
A 61-year-old woman (Patient 1) was diagnosed with probable CVID on the basis of pan-hypogammaglobulinemia (IgG 3 NR 7–14 g/l, IgA 0.2 NR 0.8–2.4 g/l, IgM 0.3 NR 0.4–1.4 g/l), idiopathic thrombocytopenia, neutropenia, and interstitial lung abnormalities. Her daughter (Patient 2) presented at the age of 44 years. She had experienced frequent respiratory infections since childhood, recurrent labial herpes simplex, and longstanding facial skin pigmentation change. Immunological testing showed normal immunoglobulins, normal B, T, and NK cell numbers, and switched memory B cells, as well as protective levels of antibody to common pathogens (pneumococcus and haemophilus) with appropriate increase after diagnostic vaccination with polysaccharide and conjugate vaccine, respectively.
Patient 1 had NGS PID gene panel testing and was found to have a heterozygous deletion at position 4:103, 446, 575–103, 451, 152 encompassing nuclear factor kappa B subunit 1 (NFKB1) exons 2 and 3. Her daughter (Patient 2) agreed to undergo testing for the variant. ddPCR CNV detection was performed in both mother and daughter.
Patient 3 is a 5-year-old male child, whose mother and grandmother are affected by type 1 hereditary angioedema with C1 inhibitor deficiency (HAE). Both mother and grandmother are on stanozolol prophylaxis. Patient 3 has had mild angioedema. The mother had reduced C4 levels and reduced C1 inhibitor function. Patient 3 had reduced C4 levels (C4: 0.05 g/l, NR: 0.2–0.6 g/l) and reduced C1 inhibitor function (40%, NR > 80%). He is currently not on specific treatment.
Patient 4 was suspected of X-linked lymphoproliferative syndrome (XLP). The DNA sample was part of an interlaboratory sample exchange quality assurance programme. No clinical details were available for this patient.
Patients 3 and 4 had specimens submitted for Sanger sequencing for all protein coding exons and exon–intron boundaries. No pathogenic variants were detected in the serine protease inhibitor, clade G, member 1 (SERPING1) gene of Patient 3 whilst Patient 4 demonstrated an absence of PCR amplification of SH2 domain-containing protein 1A (SH2D1A) exon 1. We proceeded to determine the exon copy number by ddPCR and MLPA.
Method
Patient DNA
Genomic DNA was isolated from whole blood using Gentra Puregene Blood Kit (Qiagen, Valencia, CA, USA) following the manufacturer’s protocol. DNA concentration and quality were determined using a Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Patient DNA was digested with HindIII restriction endonuclease (New England BioLabs, Ipswich, MA, USA) at 37°C for 1 h prior to ddPCR.
SH2D1A MLPA
The deletion in the SH2D1A gene was verified by MLPA, following the manufacturer’s instructions (MRC-Holland). DNA from volunteer healthy donors and the patient were applied to the MLPA reaction setup using the SALSA MLPA EK1 FAM reagent kit, coupled with the SALSA MLPA Probemix, P205-B1 SH2D1A-XIAP-ITK. Ligation and amplification were performed in an Applied Biosystems SimpliAmp thermal cycler, with capillary electrophoresis of the products performed in an ABI 3100 Genetic Analyser (Applied Biosystems). MLPA data analysis was performed using GeneMarker software (SoftGenetics) with assistance by the DNA Analysis Facility, SA Pathology at the Women’s and Children’s Hospital, Adelaide, Australia.
ddPCR primer and probe design
Coordinates for the genes tested are based on GRCh38: NFKB1 (4: 102, 501, 331–102, 617, 302), SERPING1 (11: 57, 597, 387–57, 619, 171), and SH2D1A (X: 124, 346, 563–124, 373, 160). The exon copy number was evaluated in probe and EvaGreen-based ddPCR. We designed primers targeting individual exon regions with Tm around 60°C and GC% of 40–60%. Adaptor Related Protein Complex 3 Subunit Beta 1 (AP3B1, Bio-Rad assay ID: dHsaCP2500348) or Ribonuclease P/MRP Subunit P30 (RPP30) are used as reference in assays. Primers and probes were designed with Primer3Plus or Oligo Primer Analysis v.7 (Molecular Biology Insights, USA). Primer and probe specificity was verified in NCBI Primer-BLAST and SNPcheck. No HindIII cut site was present in the DNA regions subjected to PCR. Targeted intron/exons, primer/probe sequences, and amplicon length are listed in Supplementary Table 1. All oligos were reconstituted with TE and stored as 100 μM stock.
Probe-based ddPCR
The assay was performed on a QX200 Droplet Digital PCR System (Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer’s instructions. HindIII-digested patient DNA (100 ng/reaction) was added to 1× ddPCR Supermix for probes (no dUTP) (Bio-Rad), target exon primer/probe (900 nM primers, 250 nM FAM probe), reference primer/probe (900 nM primers, 250 nM HEX probe) in 25 μl final reaction volume. Twenty microlitres of the reaction was mixed with 70 μl of Droplet Generation oil for Probes (Bio-Rad) and partitioned into approximately 20 000 droplets using the Droplet Generator. The plate was sealed with a foil heat seal using PX1™ PCR plate Sealer (Bio-Rad). PCR was performed at 95°C for 10 min; 40 cycles of 95°C for 30 s, 60°C for 1 min; 98°C for 10 min before cooling to 4°C. The thermocycler ramp rate was set at 2°C/s. Droplet reading was carried out on QX200 Droplet Reader and results analysed with Bio-Rad Quantasoft Analysis Pro. CNV was determined by calculating the ratio of target concentration to the reference concentration multiplied by 2 (reference gene CNV). Droplets were reported as copies/μl with confidence intervals of 95%.
EvaGreen ddPCR
HindIII-digested patient DNA (100 ng/reaction) was added to 1× QX200 ddPCR EvaGreen Supermix (Bio-Rad), 150 nM of target primers, and 50 nM reference primers in 25 μl reaction. Simplex reactions were performed with 50 nM target or reference primers. QX200™ Droplet Generation Oil for EvaGreen (Bio-Rad) was used in droplet generation. PCR thermal cycling (ramp rate 2°C/s) was performed at 95°C for 5 min; 40 cycles of 95°C for 30 s, 60°C for 1 min; 4°C for 5 min; 90°C for 5 min before cooling to 4°C.
Data analysis
ddPCR data were processed with QuantaSoft Analysis Pro software (Bio-Rad, USA). The CNV results from the SH2D1A and SERPING1 assays were automatically calculated with the Copy Number Variation (CNV) experimental setting. In the NFKB1 assay, the concentrations (copies/μl) of exon 2, exon 3, AP3B1, and RPP30 were obtained from QuantaSoft following analysis. Two ratios were calculated by dividing the concentration of exon by concentration of two separate reference genes. Multiply the ratios by two copies per diploid genome to obtain CNV for each exon.
where C[exon] is concentration of exon 2 or 3, C[AP3B1] is the concentration of AP3B1, and C[RPP30] is the concentration of RPP30.
Results
In the NFKB1 EvaGreen assay target and reference were performed in separate wells. The QuantaSoft software computed positive droplets from all wells as Poisson 95% confidence interval. Two primer sets of references (AP3B1 and RPP30) gave similar results. NFKB1 exons 2 and 3 copy numbers were calculated from dividing concentration of positive droplets in target wells by those in reference wells and multiplying by 2 (Table 1). The mother (Patient 1) is heterozygous for exon 2 and 3 deletion. The daughter (Patient 2) does not have the variant.
NFKB1 exons 2 and 3 CNV results in normal donor and patients
| Copy number | |||
|---|---|---|---|
| NFKB1 exon | Healthy donor | Patient 1 (mother) | Patient 2 (daughter) |
| 2 | 1.84 | 0.88 | 1.94 |
| 3 | 2.1 | 0.92 | 1.96 |
| Copy number | |||
|---|---|---|---|
| NFKB1 exon | Healthy donor | Patient 1 (mother) | Patient 2 (daughter) |
| 2 | 1.84 | 0.88 | 1.94 |
| 3 | 2.1 | 0.92 | 1.96 |
Patient 1 is heterozygous for exons 2 and 3 deletion. Healthy donor and Patient 2 have normal CNV.
NFKB1 exons 2 and 3 CNV results in normal donor and patients
| Copy number | |||
|---|---|---|---|
| NFKB1 exon | Healthy donor | Patient 1 (mother) | Patient 2 (daughter) |
| 2 | 1.84 | 0.88 | 1.94 |
| 3 | 2.1 | 0.92 | 1.96 |
| Copy number | |||
|---|---|---|---|
| NFKB1 exon | Healthy donor | Patient 1 (mother) | Patient 2 (daughter) |
| 2 | 1.84 | 0.88 | 1.94 |
| 3 | 2.1 | 0.92 | 1.96 |
Patient 1 is heterozygous for exons 2 and 3 deletion. Healthy donor and Patient 2 have normal CNV.
We evaluated SERPING1 exon copy number in Patient 3 with probe ddPCR (Fig. 2). Figure 2A depicts a representative 2-D plot of an exon CNV detection in the AP3B1-HEX/SERPING1-FAM assay where four clusters were located spatially at right angles to each other. The top left cluster were FAM-positive droplets representing exonic template. The bottom right cluster were HEX-positive droplets of the reference template. On the top right corner was the cluster of FAM- and HEX-positive droplets. Negative droplets with no templates were located in the cluster at the bottom left. Figure 2B showed the results obtained post ddPCR analysis with QuantaSoft software. Two copies of all eight exons of the SERPING1 gene were detected in the healthy donor. The assay detected heterozygous exon 4 deletion in Patient 3. Other exons were present in two copies in the HAE patient. The result was also confirmed with MLPA from an external genetic testing service.
Duplex AP3B1-HEX/SERPING1-FAM target detection with probe-based ddPCR. (A) Four distinct clusters in the 2D droplet plot of the assay. (B) The copy number of SERPING1 exon 1–8 in healthy donor and Patient 3 were determined with FAM/HEX probes. Two copies of SERPING1 exon 1–8 were detected in healthy donor. One copy of exon 4 was detected in the patient. The error bars indicate Poisson 95% confidence intervals.
The SH2D1A MLPA result showed absence of fluorescence signal from exon 1 probe pairs indicating exon 1 was absent in Patient 4 (Fig. 3). We performed dye-based ddPCR duplex assay on Patient 4 and a female healthy donor. A representative 2-D plot showed the droplet clusters were manually assigned in the dye-based ddPCR assay of SH2D1A exon 2 and AP3B1 (Fig. 4A). The fluorescence intensity of EvaGreen dye is directly proportional to amplicon size. The cluster with the lowest fluorescence signal were the droplets with no template, following by droplets with AP3B1 (72 bp), SH2D1A exon 2 (146 bp), and a combination of both templates (72 and 146 bp).
Ratio chart of MLPA analysis using the SALSA MLPA kit P205-B1 SH2D1A-XIAP-ITK probemix. Ratios of the relative fluorescence of each probe are expressed as box-and-whisker plots for healthy donors, and as red points for Patient 4. Note that female healthy donor data were normalized for the X-linked genes, X-linked inhibitor of apoptosis (XIAP) and SH2D1A. The dashed lines represent normal ratio thresholds (0.8–1.2). The absence of fluorescence from the probe pair to detect SH2D1A exon 1, verifies the hemizygous exon deletion in the male patient specimen.
Duplex AP3B1/SH2D1A target detection with EvaGreen ddPCR. (A) Clusters were assigned manually in the 2D droplet plot. The ddPCR was performed in the same well consisting of SH2D1A exon 2 and AP3B1 primers. (B) The healthy female donor has 2 copies of SH2D1A gene. Patient 4 has hemizygous exon 1 deletion. No deletion was detected in exons 2–4. The error bars indicate 95% Poisson confidence intervals.
The ddPCR result concurred with the MLPA result. SH2D1A exon 1 could not be detected in ddPCR (Fig. 4). Both assays showed SH2D1A exons 2–4 were present in the patient (Figs 3 and 4). Diploid SH2D1A copy number was detected in the female healthy donor in both assays.
Discussion
PIDs are characterized by marked genetic, allelic, and phenotypic heterogeneity. NGS-based gene panels or WES testing with targeted analysis have become routine diagnostic approaches for PIDs with locus heterogeneity. For PID patients with a classical phenotype of a monogenic defect, Sanger sequencing and MLPA remain effective methods for rapidly identifying causative SNV and CNV, respectively.
Given the difficulties identifying CNVs, several laboratories have implemented customized PID microarrays to detect SNVs and CNVs [11, 16]. CNVs have been reported in these settings although certain CNVs have yet to undergo functional studies to establish causality [17].
We have used ddPCR as an alternative method to determine CNV. The patient with NFKB1 deletion was confirmed with ddPCR following gene panel testing. Exon deletions were detected in a XLP patient and a HAE patient following normal Sanger sequencing results. Both deletions were identified by MLPA and ddPCR. We found ddPCR is comparable to MLPA for detecting microdeletions in our patients.
Patient 1 has a heterozygous NFKB1 exon deletion. Her daughter (Patient 2) does not have the loss-of-function (LOF) variant. NFKB1 deletions confer high risk of late onset immune failure with hypogammaglobulinaemia and autoimmunity [18, 19]. Knowledge that she does not carry the familial NFKB1 variant has provided Patient 2 and her physicians with reassuring prognostic information.
HAE is an autosomal dominant disorder where heterozygotes are symptomatic. More than 700 variants in SERPING1 have been described in HAE [20]. The SERPING1 intronic region consists of repetitive Alu-elements that are prone to deletions, duplications, and rearrangements [21]. Large gene rearrangements were found in 10–15% of well-characterized patient cohorts [20, 22, 23]. These SERPING1 gene deletions and duplications have been detected by targeted NGS [24], MLPA [20, 23, 25, 26], probe-based gene dosage [27], exon quantitation technique (EQT) [28], and Quantitative Multiplex PCR of Short Fluorescent Fragments (QMPSF) [22]. Our HAE patient is a heterozygote for exon 4 deletion. The deletion was detected by both probe-based and EvaGreen ddPCR assays as well as MLPA confirmation by an external commercial laboratory.
Routine PCR failed to amplify SH2D1A exon 1 in Patient 4. We suspected Patient 4 had a hemizygous exon 1 deletion. The deletion confirmed by MLPA and ddPCR is predicted to lead to loss of function due to absence of the translation initiation codon.
The SERPING1 and SH2D1A EvaGreen assays were developed as described in McDermott et al. [29]. The intensity of EvaGreen fluorescence emission is directly proportional to amplicon size. Amplicons of different length amplified in a single well produced distinct populations of fluorescence signal enabling quantification of multiple target species in the well. Good target and reference amplicon separation was observed in the SH2D1A assay. We found dye-based SERPING1 duplex assay problematic due to the need for assay optimization and/or redesigning primers. Presence of GC rich regions near the start codon in SERPING1 exon 2 hampered the ability to reliably quantify exon 2 CNV in EvaGreen assay. Interestingly we did not encounter such issues in the probe-based assay. We decided to perform the NFKB1 assay for each amplicon in separate wells to ensure efficient PCR amplification. Using two sets of target and reference primers to evaluate CNV in EvaGreen assays can enhance certainty in variant calling.
ddPCR has several advantages over other approaches. ddPCR measures the absolute number of target amplicons by counting fluorescence positive water-in-oil droplets on a microfluidic platform and is less prone to error compared to quantitative PCR because there is no need for standard curves. Healthy donors are not required to establish a normal range. EvaGreen assays do not rely on fluorescence probes or primers for quantifying target and reference genes, making it cost-effective compared to probe-based assays. Whilst MLPA is the gold standard for CNV evaluation, we found ddPCR EvaGreen assay is more cost-effective for detecting CNV in patients with rare diseases in countries with small population such as New Zealand.
As with many diagnostic techniques, the EvaGreen assay is unsuitable for detecting CNV where pseudogenes are present. Attempts to quantify CNV in DNA cross-link repair 1C (DCLRE1C) were unsuccessful due to the presence of the DCLRE1CP1 sequence. Sequence similarity between the gene and its pseudogene is an issue when designing primers and probes. It is unlikely prob-based assays would be able to differentiate between a gene and its pseudogenes.
Dyer et al. indicated structural variants or CNV were often identified in dominantly inherited PIDs or X-linked recessive genes [16]. CNV evaluation should be performed in cases with high index of suspicion for monogenic disorders with no pathogenic SNV or indels. Chromosomal microarray analysis has successfully detected large gene deletion in patients suspected of DOCK8 deficiency or DiGeorge patients with 22q11 microdeletion. Gene panel NGS and MLPA are more suitable for exonic deletion/duplication. ddPCR is appropriate for small genes (up to 10–12 exons) or family segregation study once the CNV is confirmed with other diagnostic approaches.
Abbreviations
- CNV
copy number variant
- CVID
common variable immunodeficiency
- ddPCR
droplet digital polymerase chain reaction
- HAE
hereditary angioedema
- MLPA
multiplex ligation-dependent probe amplification
- NFKB1
nuclear factor kappa B subunit 1
- NGS
next generation sequencing
- PID
primary immunodeficiency disorder
- SERPING1
serine protease inhibitor, clade G; member 1
- SH2D1A
SH2 domain-containing protein 1A
- SNV
single nucleotide variant
- WES
whole exome sequencing
- XLP
X-linked lymphoproliferative syndrome
Acknowledgements
We acknowledge the support of Auckland District Health Board and LabPLUS for this work. This is a diagnostic study and consent was obtained by clinicians at the time the request was made.
Funding
This manuscript was not funded.
Conflict of interest
The authors declare that they do not have conflicts of interest.
Author contributions
S.-T.W. designed ddPCR assays and wrote the first draft. J.M. performed ddPCR assays. A.Q. performed MLPA. H.L. and R.A. provided clinical details. All authors contributed to writing and proofreading of manuscript. All authors approved the final manuscript.
Data availability
There are no additional data. Supplementary files contain ddPCR primer sequences.
References
