Evaluation of Severe Combined Immunodeficiency and Combined Immunodeficiency Pediatric Patients on the Basis of Cellular Radiosensitivity

Abstract

Pediatric patients with severe or nonsevere combined immunodeficiency have increased susceptibility to severe, life-threatening infections and, without hematopoietic stem cell transplantation, may fail to thrive. A subset of these patients have the radiosensitive (RS) phenotype, which may necessitate conditioning before hematopoietic stem cell transplantation, and this conditioning includes radiomimetic drugs, which may significantly affect treatment response. To provide statistical criteria for classifying cellular response to ionizing radiation as the measure of functional RS screening, we analyzed the repair capacity and survival of ex vivo irradiated primary skin fibroblasts from five dysmorphic and/or developmentally delayed pediatric patients with severe combined immunodeficiency and combined immunodeficiency. We developed a mathematical framework for the analysis of γ histone 2A isoform X foci kinetics to quantitate DNA-repair capacity, thus establishing crucial criteria for identifying RS. The results, presented in a diagram showing each patient as a point in a 2D RS map, were in agreement with findings from the assessment of cellular RS by clonogenic survival and from the genetic analysis of factors involved in the nonhomologous end-joining repair pathway. We provide recommendations for incorporating into clinical practice the functional assays and genetic analysis used for establishing RS status before conditioning. This knowledge would enable the selection of the most appropriate treatment regimen, reducing the risk for severe therapy-related adverse effects.

Severe combined immunodeficiency (SCID) and combined immunodeficiency (CID) are rare genetic disorders. The incidence of SCID in Australia is 1 in 69,000 live births,¹ comparable to the reported incidence of 1 in 100,000 births worldwide. CID patients have increased susceptibility to invasive and opportunistic bacterial, viral, and fungal infections due to poor T-lymphocyte production and/or function, in addition to failure of B lymphocytes to generate functional antibodies. SCID is the extreme form of CID. Patients with this condition often present in the first year of life, with severe life-threatening infections, and consequently failure to thrive, requiring prompt intervention. SCID without treatment is usually fatal within the first year of life.²

Both SCID and CID are genetically diverse syndromes, and over 50 molecular defects resulting in these syndromes have been described.³ Approximately 30% of SCID patients have defects in V(D)J recombination (antigen receptor recombination), an essential process for the normal development of T and B lymphocytes.⁴ This process randomly combines variable (V), diverse (D), and joining (J) gene segments in lymphocytes.⁵ V(D)J defects lead to T- and B-cell lymphocytopenia and a classic T-B-natural killer⁺ SCID phenotype. The initial steps of V(D)J recombination are performed by recombination-activating genes 1 and 2 (RAG1 and 2), which induce DNA double-strand breaks (DSBs) at the recombination-specific sequences. DSBs are then repaired by the nonhomologous end-joining (NHEJ) repair pathway. Mainly, T^–/B^– SCID defects arise from mutations in RAG1 and 2⁵; however, a subgroup of patients are deficient in NHEJ genes, such as DNA-dependent protein kinase complex (DNA-PKcs-PRKDC), ligase IV (LIG4), X-ray repair cross-complementing factor 4 (XRCC4), Cernunnos (NHEJ1; alias XLF), and Artemis (DNA cross-link repair 1C, DCLRE1C).^6–12 Cells deficient in NHEJ are unable to effectively repair DNA DSBs and, therefore, are radiosensitive (RS).

Although SCID represents the extreme immunologic phenotype, highly variable degrees of immune dysfunction, ranging from CID to immune dysregulation to pancytopenia secondary to bone marrow failure, have been observed in patients with RS defects.^5,13 This dysfunction is believed to be due to variability in the functional impact of mutations involving NHEJ-repair mechanisms. Since NHEJ is also used in Ig class switch–recombination after V(D)J recombination, patients with milder defects may present with normal or markedly raised IgM but diminished or absent IgA and IgG.¹⁴

In patients with RS defects, increased susceptibility to DNA damage in somatic and hematopoietic cells may significantly affect treatment response and prognosis. Hematopoietic stem cell transplantation (HSCT), a curative treatment for patients with SCID or CID,¹⁵ may be preceded by conditioning regimens of radiomimetic drugs, such as cyclophosphamide, melphalan, busulfan, and thiotepa.^16,17 Also, transplantation may be followed by prophylaxis of graft-versus-host disease using agents that also have the potential to induce DNA DSBs.¹⁸ RS-SCID and RS-CID patients receive a higher transplantation priority and adjusted pre- and post-transplantation treatment regimens. These regimens involve reduced doses of radiomimetic drugs and/or alternative drugs reported to be efficacious in these patients.^19,20 The survival rate after HSCT for RS-SCID or RS-CID has been reported to be lower in comparison to those with other genetic types of SCID and CID.^15,21 The impact of these newer pre- and post-transplantation regimens on survival rates relative to non–RS-SCID and non–RS-CID needs further evaluation. Furthermore, although HSCT treats these patients' immunodeficiency, these patients' somatic cells retain their DNA-repair defects and susceptibility to malignant transformation.¹⁴ Accordingly, long-term, post-transplantation survivors with RS defects need to be monitored for malignancy.

There is a lack of reports of the application of functional assays for detecting RS-SCID and RS-CID. However, a range of established laboratory assays used for detecting cellular RS can be applied to RS-SCID and RS-CID screening. The predictive assays proposed to date are mainly based on the measurements of cellular DNA damage response and/or its consequences as a surrogate for the RS of an organism.^22–26 Chromosomal aberrations (mainly translocations and dicentrics) and clonogenic survival are the classic and most reliable end points for predicting individual RS.^23,24,27,28 However, they are time-consuming, and the latter often requires cell transformation for conferring indefinite proliferation and facilitate propagation, which itself changes cell properties.^29–31 Although sequencing techniques, including exome sequencing, enhance the detection of mutations in candidate genes conferring RS-SCID and RS-CID, functional assays can provide an important additional tool for establishing, for example, whether missense mutations affect functioning.

In our laboratories, we used the γ histone 2A isoform X (H2AX) assay^32,33 for detecting DNA DSBs in irradiated primary fibroblasts established from skin biopsy samples from SCID and CID patients. γ-H2AX is a de novo protein formed in response to DSB formation. When stained with fluorescently labeled phosphospecific antibody, γ-H2AX molecules can be visualized as nuclear foci at the sites of DSBs.^32,34,35 The number of γ-H2AX foci per cell increases with the radiation dose and follows for well-studied kinetics of decline (DSB repair) in a large variety of normal cells and tissues.^32,36–39 Significantly altered heterogeneous kinetics, which can be distinguished from the normal repair kinetics, have been reported in repair-deficient cells.^40,41 The assay is extremely sensitive, and it measures changes that occur quickly; the maximal response is within 1 hour after irradiation, with a decline of the signal within several hours. These properties make the γ-H2AX an attractive screening biomarker in translational research, for the assessment of clinical biodosimetry of diagnostic and therapeutic radiation and DNA-damaging chemotherapy.^33,42,43

We and others have demonstrated that the number and kinetics of decline of radiation-induced foci (surrogate of DSB repair) are a measure of the cellular RS in SCID and CID applications and other settings.^{6–8,40,44,45} Recently, we presented several analytical tools for improving the statistical and computational approaches to applying the assay for differential diagnostics in RS and non-RS biodosimetry in tissues and in the primary fibroblast skin culture model.^45,46

Here, we further refine the mathematical criteria of cellular RS. To describe the kinetics of γ-H2AX foci decline, we fitted the experimental data (foci counts per nucleus; at 0, 0.5, 2, 6, 24, 48, and 72 hours after irradiation) to an empirical model that assumed the presence of two repair components, slow and fast. Nonlinear regression analysis (curve-fitting) was used for evaluating the three parameters that define the repair/disappearance of foci—the rates of the slow (s) and fast (f) and the fraction (Q) of foci that are repaired by the slow process. We demonstrate the crucial role of the fraction of the slow component in identification of RS status. Meticulous analysis of the cellular response to radiation in primary fibroblast cultures derived from explanted skin, measured by the γ-H2AX assay, allowed for the prediction of RS status in five dysmorphic and/or developmentally delayed pediatric SCID or CID patients. The patients were associated with three Australian centers, and the assays were performed before, or as a part of, the assessment of the suitability of HSCT and the associated conditioning treatment. We confirmed the DNA-repair status of all patients by assessing the radiation dose–response relationship, as determined by the clonogenic survival end point. The results of NHEJ gene sequencing, and the patients' clinical presentation and response to treatment, were also taken into account.

In one patient, the γ-H2AX signature suggested either LIG4 or XLF deficiency. Sequencing of the LIG4 gene subsequently identified compound heterozygous mutations, which informed the treatment protocol. Finally, we make recommendations on how the functional assays can be incorporated into clinical practice, aiming to avoid severe therapy-related adverse effects.

Materials and Methods

Patient Selection for RS Testing

Five patients (P1 to P5) were referred for RS testing (Table 1) on the basis of clinical presentations suggestive of an underlying RS defect and/or the need for exclusion of such a defect before contemplation of the use of radiomimetic drugs. The age range at presentation was broad (3 months to 14 years). All had histories of recurrent infections, T- and B-cell lymphopenia (except P5), abnormal lymphocyte proliferative response, and abnormal B-cell functioning. Three patients (P2, P3, and P4) had pancytopenia and two of them also had features of immune dysregulation (P3 and P4). None of the patients were born to consanguineous parents. Expanded case descriptions are presented in sections P1 to P5 below.

Table 1.

Clinical Presentation and Outcome of the Patients

Patient characteristic	P1	P2	P3	P4	P5
Age at presentation	3 months	4 years	13 years	3 months	14 years
Sex	Female	Male	Male	Female	Male
Dysmorphism	No	Yes	No	No	No
Developmental delay	No	Yes	No	No	No
Infections	Pulmonary infection, urinary tract infection	Sinopulmonary infections	Gastroenteritis	Enteritis (chronic Norovirus and Enterovirus)	Sinopulmonary infections (PJ), meningitis (Pneumococcus), bacteremia (Pneumococcus), enteritis (Salmonella)
Malignancy	No	No	No	No	B-NHL
Immune dysregulation	No	No	Nephrotic syndrome, autoimmune hemolytic anemia, inflammatory lung and brain lesions	Panniculitis, hepatosplenomegaly, IgM-mediated hemolytic anemia	No
Cytopenias	Lymphopenia	Anemia, neutropenia, lymphopenia, thrombocytopenia	Anemia, neutropenia, lymphopenia, thrombocytopenia	Anemia, lymphopenia, thrombocytopenia	No
Immunologic abnormalities	T and B lymphopenia, abnormal PHA response,∗ hypogammaglobulinemia (IgA ↓↓↓, IgM ↓↓↓)	T and B lymphopenia, abnormal anti-CD3 response, hypogammaglobulinemia (IgA ↓↓↓, IgM ↓↓), abnormal vaccine antibody response (23PPV)	T and B lymphopenia, abnormal anti-CD3 response, abnormal vaccine antibody response (23PPV)	T and B lymphopenia, abnormal PHA response,∗ hypogammaglobulinemia (IgG ↓↓↓, IgA ↓↓↓)	Abnormal PHA response,∗ hypogammaglobulinemia (IgG↓↓↓, IgM ↓↓↓)
Pretransplantation conditioning regimen	Bu/Flu/ATG (RIC)	Cy/Flu/ATG (MA-RSC)	Bu/Flu/Alem† (RIC)	Treo/Flu/Alem (RIC)	Treo/Flu/ATG (RIC)
HSCT	MUD UCBT	MUD UCBT	MUD BMT	MUD BMT	MUD BMT
Significant post-transplant complications	Poor T-cell engraftment, multiple opportunistic infections, cGVHD	aGVHD (gastrointestinal), EBV reactivation, EBV-associated PTLD, multi-organ failure	Nil	Nil	aGVHD (cutaneous)
Current status	Deceased (21 months after HSCT)	Deceased (3.5 months after HSCT)	Alive (7 months after HSCT)	Alive	Alive (15 months after HSCT)

aGVHD, acute graft-versus-host disease; Alem, alemtuzumab; ATG, antithymocyte globulin; BMT, bone marrow transplantation; B-NHL, B cell non-Hodgkin lymphoma; Bu, busulfan; cGVHD, chronic graft-versus-host disease; Cy, cyclophosphamide; Dip, diphtheria vaccine; EBV, Epstein-Barr virus; Flu, fludarabine; HSCT, hematopoietic stem cell transplantation; MA-RSC, myeloablative radiosensitivity-adjusted conditioning; MUD, matched unrelated donor; 23PPV, 23-valent pneumococcal polysaccharide vaccine; PHA, phytohemagglutinin-A; PJ, Pneumocystis jiroveci; PTLD, post-transplantation lymphoproliferative disorder; RIC, reduced-intensity conditioning; Tet, tetanus vaccine; Treo, treosulfan; UCBT, umbilical cord blood transplantation.

^∗Lymphocyte proliferative response to anti-CD3 not performed.

^†Received alemtuzumab 28 days before transplantation. (↓ = reduced, > 70% of the lower limit of normal (LLN); ↓↓ = reduced, 40% to 70% of LLN; ↓↓↓ = reduced, < 40% of LLN).

P1

P1 was born to nonconsanguineous Australian Caucasian parents and presented at 3 months of age with poor weight gain and persistent lymphopenia on a background of palatal ulcer, urinary tract infection, and chest infection. Immunologic evaluation revealed marked T- and B-cell lymphopenia, normal numbers of natural killer cells, low Ig levels, and abnormal lymphocyte proliferation in response to phytohemagglutinin A (PHA). RAG1/RAG2 sequencing did not reveal any mutation. After the exclusion of an RS defect, at 6 months of age, the patient received reduced-intensity conditioning with fludarabine, busulfan, and antithymocyte globulin, followed by a matched unrelated donor umbilical cord blood transplantation. Her post-transplantation course was complicated by poor T-cell engraftment, multiple opportunistic infections, and treatment-resistant, chronic graft-versus-host disease. She succumbed to infections and died at 2 years of age.

P2

P2 was a 6-year-old dysmorphic and developmentally delayed boy born to nonconsanguineous Australian Caucasian parents. He initially presented at age 4 years 6 months with pancytopenia and bone marrow aspirate suggestive of a myelodysplastic syndrome. Over the next 9 months, he developed recurrent chest infections and his Ig levels declined. Further immunologic evaluation revealed undetectable B cells, reduced CD4⁺ T cells, and markedly reduced naive CD4⁺ and CD8⁺ T cells. Lymphocyte proliferative response to PHA was normal but to anti-CD3 was reduced. He had hypogammaglobulinemia. Although he demonstrated good antibody response to a conjugated protein vaccine (Haemophilus influenzae B), he had no response to a polysaccharide vaccine (Pneumovax). After confirmation of an RS defect, he received a myeloablative but RS-tailored conditioning regimen with fludarabine, reduced-dose cyclophosphamide, and antithymoglobulin, followed by a matched unrelated donor umbilical cord blood transplantation. Donor cell engraftment and complete donor cell chimerism in the bone marrow were achieved within 2 months after transplantation. However, his clinical course was complicated by reactivation of Epstein-Barr virus and refractory Epstein-Barr virus–associated post-transplantation lymphoproliferative disorder, eventually leading to multiorgan failure and death at 14 weeks after transplantation.

P3

P3 is a 15-year-old boy born to nonconsanguineous parents in whom culture-negative nonspecific inflammatory lesions developed in his chest and brain, 4 years after the initial onset of a progressively worsening and increasingly transfusion-dependent pancytopenia. His history included nephrotic syndrome treated with cyclophosphamide and corticosteroids at 4 years of age. Bone marrow aspirates showed features of myelodysplasia with normal cytogenetics. Immunologic evaluation revealed reduced proportions of naive CD4⁺ and CD8⁺ T cells and switched memory B cells. Lymphocyte proliferative response to PHA was normal but to anti-CD3 was abnormal. Ig levels were normal, and the patient demonstrated normal antibody responses to protein vaccines (tetanus and H. influenzae B) but abnormal response to a polysaccharide vaccine (Pneumovax). He received alemtuzumab for the treatment of his inflammatory lesions for 6 months in the lead-up to transplantation. After the exclusion of an RS defect, he underwent a reduced-intensity conditioning (fludarabine and busulfan) matched unrelated donor bone marrow transplantation (BMT). Seven months after transplantation, he is well and recovering.

P4

P4 is a 4-year-old dysmorphic, but developmentally normal, girl born to nonconsanguineous parents. She had a 2-year history of failure to thrive and parenteral nutrition–dependent chronic viral (Norovirus and Enterovirus EV71) enteritis. Initial immunologic evaluation revealed T- and B-cell lymphopenia and undetectable IgG and IgA. Her lymphocyte proliferative response to PHA was abnormal and progressively deteriorated with time. Her clinical course was complicated by immune dysregulation, including panniculitis, hepatosplenomegaly, and immune-mediated hemolytic anemia. After the exclusion of an RS defect, the patient underwent reduced-intensity conditioning (treosulfan, fludarabine, alemtuzumab) matched unrelated donor BMT. She is alive after BMT in 2012, with improved intestinal and immune function. However, features of steroid-dependent inflammatory syndrome persist.

P5

P5 is a 14-year-old boy who presented with a B-cell non-Hodgkin lymphoma on a background of multiple invasive and opportunistic infections including Pneumocystis jiroveci, pneumococcal meningitis, Salmonella enteritis, recurrent pneumococcal bacteremia, and recurrent sinopulmonary infections since 4 months of age. Immunologic evaluation revealed normal lymphocyte subpopulations and normal CD40 ligand expression but abnormal lymphocyte PHA. He had hypogammaglobulinemia and was commenced on Ig replacement therapy from 2 years of age. Further evaluation at the time of diagnosis of non-Hodgkin lymphoma revealed oligoclonal expansion of CD8⁺ T cells and undetectable marginal zone–like memory B cells. Family history was significant for an older sister who also had a history of recurrent invasive infections and who had died of progressive non-Hodgkin lymphoma that relapsed after initial treatment. After confirmation of a mild RS defect, he underwent reduced-intensity conditioning (treosulfan, fludarabine, anti-thymocyte globulin) matched unrelated donor BMT and is alive and well 15 months after transplantation.

Cell Cultures

Eleven strains of untransformed primary human fibroblasts derived from skin biopsies were used in this study (Table 2). Five SCID or CID patient–derived primary fibroblast strains with unknown RS status were obtained for diagnostic purposes from three Australian pediatric centers: P1, P2, and P3 cells were provided as a gift by Sharon Choo (Royal Children's Hospital, Melbourne, VIC, Australia), P4 cells by Paul Gray (Sydney Children's Hospital, Sydney, NSW, Australia), and P5 cells by Chris Fraser (Children's Health Services, Queensland Health, Herston, QSL, Australia), with written informed assent received from all parents/caregivers according to the procedure established by the ethics committees of the institutions for studies using human subjects.

Table 2.

Primary Skin Fibroblast Cell Cultures Used in This Study

Culture ID	Description	Source	Passage
P1	SCID	Royal Children's Hospital (Melbourne, VIC, Australia)	5
P2	CID	Royal Children's Hospital	1
P3	CID	Royal Children's Hospital	2
P4	CID	Sydney Children's Hospital (Sydney, NSW, Australia)	1
P5	CID	Queensland Health (Herston, QSL, Australia)	1
C1 (IBR3)	Wild-type	University of Sussex (Brighton, UK)	10
C2 (F02/385)	DCLRE1C null	University of Sussex	11
C3 (618BR)	Compound het Leaky DCLRE1C and P171R het	University of Sussex	17
C4 (NM720)	Compound het PRKDC A3574V het/exon16Δ het	University of Sussex	6
C5 (180BR)	LIG4 R278H Homo	University of Sussex	9
C6 (P2)	XLF E78X Homo	University of Sussex	7

CID, combined immunodeficiency; SCID, severe combined immunodeficiency.

Six reference cultures—primary skin fibroblasts from a normal donor [IBR3; normal control (C1)] and five strains from a cell collection of RS-SCID or RS-CID patients with known mutations in NHEJ genes—were provided as a gift by Dr. Penny Jeggo (University of Sussex, Brighton, UK). These reference mutant cultures were: F02/385 (DCLRE1C null; C2), 618BR (leaky DCLRE1C; C3), NM720 (PRKDC, C4), 180BR (leaky LIG4; C5), and P2 (XLF; C6). Cells were received in either live culture or frozen. They were plated into Cellstar T75 flasks (Greiner Bio One, Frickenhausen, Germany) and cultured in AmnioMAX-C100 basal medium (Gibco Life Technologies, Melbourne, VIC, Australia) at 37°C/5% CO₂. Before irradiation, cells were plated into 8-well chamber slides (Becton, Dickinson, and Company, Franklin Lakes, NJ) in concentrations needed to achieve 5 × 10⁵ cells/well on the day of irradiation (the plateau phase of growth with approximately 100% confluence).

Irradiations, Immunostaining, and Microscopy

The cells were irradiated with 3 Gy using a ¹³⁷cesium source (GammaCell40 Irradiator; Nordion International, Ottawa, ON, Canada), at a dose rate of 0.531 Gy/minute. At 0.5, 2, 6, 24, 48, and 72 hours after irradiation, the cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) for 20 minutes at room temperature. The cells were stored in chilled 70% ethanol until immunostaining, which was performed as described elsewhere.^36,47 Shortly before processing, the chamber slides were brought to room temperature before ethanol removal and washing with phosphate-buffered saline (prepared in-house). Then the cells were blocked with bovine serum albumin (Sigma-Aldrich, St. Louis, MO) for 30 minutes, before the primary antibody staining with mouse anti–γ-H2AX (phospho S139) mononuclear antibody (catalog no. 22551; Abcam, Cambridge, UK) for 2 hours at room temperature at a dilution of 1:500. Secondary antibody staining followed, with the use of Alexa Fluor 488 goat anti-mouse IgG (InvitroGen/Life Technologies, Mount Waverley, VIC, Australia) for 1 hour at room temperature at a dilution of 1:500, before counterstaining and mounting with either propidium iodide– or DAPI-containing Vectashield mounting medium (Vector, Burlingame, CA).

Microscopy was conducted using an Olympus FV1000 laser confocal microscope (Olympus, Tokyo, Japan). The images used for analysis were captured using a 60× water-immersion objective and compiled from a Z-stack of 6 to 10 optical sections (0.8-μm separation). A minimum of 100 cells per experimental group were analyzed. This number was dependent on cell density; a much higher number of cells were analyzed in some groups. All samples were imaged using identical microscope settings to avoid any artefactual variation in image intensity. Analysis and foci counting were conducted using the in-house JcountPro software (developed by P.L.), which is a further development of the TGI software that we have reported previously.^44–46

Colony Formation Assay

The colony formation assay was conducted in parallel in two laboratories, at the Peter MacCallum Cancer Center (East Melbourne, VIC, Australia) and the University of Sussex, using similar methodology. At Peter MacCallum Cancer Center, the cells were grown to approximately 80% confluence in 25-cm² flasks and irradiated using the ¹³⁷cesium source at a dose rate of 0.525 Gy/minute with 0, 1, 3, 5, and 7 Gy, at room temperature, using the GammaCell40 Irradiator. Cells were then washed with phosphate-buffered saline/EDTA (Sigma-Aldrich), treated with Pronase Protease (Merck Millipore, Darmstadt, Germany) and neutralized with AmnioMAX-C100 medium with added supplements, spun into a pellet with a centrifuge (model 4K15; Sigma Laboratories, Osterode, Germany) at 400 × g for 5 minutes, and resuspended in the AmnioMAX-C100 medium with added supplements. The cells were then counted, diluted with serial dilutions to required concentrations, and plated onto 9-cm Petri dishes (Greiner Bio One) containing a feeder layer that had been prepared 24 hours before each cloning assay. The feeder layer was produced with a test cell line that had a relatively high growth rate (P3, fibroblasts). These cells were grown in 75-cm² flasks to 80% confluence and irradiated under the same conditions as tested cells to a total dose of 35 Gy before being plated at a density of 6 × 10⁴ cells/plate.

Petri dishes were left to incubate for 21 days undisturbed at 37°C/5% CO₂. After 21 days the medium was removed and the plates were rinsed with 0.9% saline (made in-house) before being stained with 0.01% crystal violet solution (Sigma-Aldrich) containing 10% neutral formalin (Merck Millipore) for 20 minutes. Resulting colonies containing 50 or more individual cells were then manually counted under a stereo microscope (model SZ-60; Olympus).

γ-H2AX Data Analysis

The analysis of immunofluorescence images comprised the identification of nuclei and the calculation of the number of γ-H2AX foci in each individual nucleus.

In each experimental group, foci counts were combined to build a frequency distribution of foci per nucleus and were fitted to the Poisson distribution, essentially as described previously⁴⁵:

$$P\left(k\right)=e^{-a}\frac{a^k}{k!},$$ (1)

where P(k) is the probability (or the frequency) of nuclei with k foci and a is the average number of Poisson distribution. We termed this value the principal average, and, as we demonstrated previously, it provides a better measure of foci count in a cell population.⁴⁵

Principal average values of foci counts from two to three replicate experiments per patient were combined to calculate the average foci number at each time point and to build the foci kinetics plot. The γ-H2AX foci kinetics in each patient were then analyzed using a two-component (fast and slow) repair model:

$$N\left(t\right)=N_M\left[Q{e}^{\mathrm{-}st}+\left(1\mathrm{-}Q\right)e^{\mathrm{-}ft}\right],$$ (2)

where N_M is the maximal number of foci; Q is the fraction of the slow-repair component; f and s are the fast- and slow-repair rates, respectively; and t is the postirradiation time.

The linear-quadratic model was used for the regression analysis of clonogenic survival curves:

$$S\left(D\right)=e^{\left(\mathrm{-}\mathit{\alpha }D\mathrm{-}\mathit{\beta }{\mathit{D}}^{\mathrm{2}}\right)},$$ (3)

where S(D) is the survival fraction after the irradiation of cells with the dose D, and α and β are parameters reflecting the linear and quadratic components of the survival curve, respectively. For survival curves that indicated the presence of a minor resistant subpopulation, the second exponential term was added to the linear-quadratic equation:

$$S\left(D\right)=\left(1-F\right)e^{\left(-\mathit{\alpha }D-\mathit{\beta }{\mathit{D}}^{\mathrm{2}}\right)}+F{e}^{-\mathrm{\gamma }D}$$ (4)

The nonlinear regression analysis and the testing of statistical hypotheses were performed using Prism version 6.01 software (GraphPad Software Inc., San Diego, CA).

Sequencing Analysis

Sequencing analysis was performed at the University of Sussex using total RNA extracted from primary human fibroblasts with the RNeasy Mini Kit (Qiagen, Crawley, UK) according to the manufacturer's recommendations. cDNA was synthesized with a First Strand cDNA Synthesis Kit (Amersham Biosciences/GE Healthcare, Bucks, UK). Artemis cDNA was generated by RT-PCR and amplified by PCR in two overlapping fragments using two primer pairs to cover the entire open reading frame, Art1/Art2, and Art3/Art4 [Art1/Art2 was subsequently cloned into PCR-TOPO (Invitrogen, Warrington, UK) and individual clones analyzed]. LIG4 cDNA was amplified in two overlapping fragments using two primer pairs, F1 and R2, and F2 and R1. To amplify XRCC4 and XLF, only one primer pair was used: XRCC4-F1/XRCC4-R1 and XLF-F1(A)/XLF-R1, respectively. Amplification reactions were performed using KOD high-fidelity Taq polymerase (Millipore, Beeston, UK). Cycle conditions used were: 1 minute 95°C, 30 cycles at 94°C 30 seconds, 55°C 30 seconds, and 72°C 2 minutes, plus a final extension of 72°C 5 minutes. All PCR products were analyzed by gel electrophoresis (1% agarose) and subsequently purified using a QiaQuick PCR Purification Kit (Qiagen). All sequencing was performed on an ABI Prism 310 sequencer using BigDye fluoresceinated dye terminators (Applied Biosystems, Warrington, UK) with a series of internal primers.

Statistics

To develop an efficient statistical criterion for the identification of RS cells, we attempted a series of statistical comparisons of the data from each repair-deficient reference cell strain (C2 to C6) to the data from the normal control group (C1), assuming that the DNA-repair deficiency implies an increased RS. We applied various comparison approaches aiming to establish a criterion that indicates the difference with the highest degree of statistical significance. These approaches included the paired t-test of the experimental data points; testing the hypothesis that the same set of three parameters f, s, and Q (Equation 2) can describe two data sets; testing the hypothesis that the same pairs of the slow-repair parameters s and Q can describe two data sets, assuming a shared value of the fast-repair rate f; testing the hypothesis that the same value of the slow-repair fraction Q can describe two data sets, assuming a shared value of the slow-repair rate s; and testing the hypothesis that the same value of the slow-repair fraction Q can describe two data sets, assuming a fixed value of the slow-repair rate s obtained from Table 3 for each of the reference repair-deficient cell strains.

Table 3.

Results of the Nonlinear Regression of the γ-H2AX Foci Kinetics Shown in Figures 1 and 2

Parameter/Culture ID	NM	Fast-repair rate (f), per hour	Slow-repair rate (s), per hour	Slow-repair fraction (Q)
C1	41.7 ± 8.2	0.295 ± 0.085	0.0196 ± 0.0061	0.128 ± 0.045
C2	36.9 ± 2.9	0.248 ± 0.067	0.0092 ± 0.0035	0.248 ± 0.067
C3	39.6 ± 7.3	0.160 ± 0.065	0.0 (constrain >0)	0.097 ± 0.050
C4	38.9 ± 8.0	0.497 ± 0.521	0.0134 ± 0.0033	0.588 ± 0.128
C5	39.6 ± 4.7	0.200 ± 0.071	0.0285 ± 0.0020	0.455 ± 0.060
C6	36.1 ± 14.1	0.59 ± 1.85	0.0289 ± 0.0049	0.744 ± 0.291
P1	51.4 ± 9.8	0.406 ± 0.107	0.0141 ± 0.0053	0.128 ± 0.039
P2	40.2 ± 2.5	0.168 ± 0.064	0.0180 ± 0.0022	0.536 ± 0.066
P3	42.9 ± 3.7	0.303 ± 0.044	0.0236 ± 0.0026	0.172 ± 0.026
P4	28.5 ± 1.9	0.193 ± 0.026	0.0130 ± 0.0026	0.128 ± 0.018
P5	34.0 ± 9.1	0.215 ± 0.084	0.0000 ± 0.0087	0.071 ± 0.026

Best-fit values of four parameters according to Equation 2 and their standard errors are presented for fibroblasts from six reference cultures (C1–C6) and cultures from five severe combined immunodeficiency or combined immunodeficiency (P1–P5).

H2AX, histone 2A isoform X, N_M, maximal number of foci.

Results

Development of Criteria for DNA Damage Repair Capacity as a Measure of Cellular RS

DNA-repair capacity was measured by quantifying postradiation exposure formation and the decline of γ-H2AX foci, a surrogate marker of DNA DSBs. It was compared with measuring the radiation dose response for loss of clonogenic survival.

γ-H2AX Assay in Reference Cells

Figure 1 presents the experimental results on the postirradiation γ-H2AX foci kinetics in the six reference cell strains (Table 2), without (C1) and with (C2 to C6) known deficiencies in the NHEJ repair pathway. Table 3 summarizes the results from the independent nonlinear regression analysis of the experimental data for each reference culture shown in Figure 1, presented as best-fit values of parameters in the two-component repair model (Equation 2 in Materials and Methods). Data points at time 0 hour (background foci count in nonirradiated cells) were excluded from the regression analysis.

Figure 1.

Analysis of post–3-Gy irradiation (IR) γ histone 2A isoform X (H2AX) foci kinetics in six reference cultures without [C1 (normal control)] and with known deficiencies in nonhomologous end-joining repair pathway [C2 to C6 (repair-deficient mutants)]. A: C2, DCLRE1C null. B: C3, leaky DCLRE1C heterozygous. C: C4, PRKDC heterozygous. D: C5, leaky LIG4. E: C6, XLF. Circles represent the principal mean foci numbers; error bars show the SEMs of the means before IR exposure, positioned at 0 hour and after IR exposure, at 0.5 (0.25 for C5), 2, 6, 24, 48, and 72 hours. Lines demonstrate the results of regression analysis of the repair kinetics using the two-component model according to Equation 2. Each panel shows data from C1 and one of the repair-deficient mutants.

The results from the statistical comparisons (Materials and Methods) are presented as P values in Table 4. These results indicate significant differences (P < 0.05) between the normal cells (C1) and each of the repair-deficient cells (C2 to C6), for at least one of the five approaches described. The significance levels, however, vary between approaches, with usually stronger significance (lower P values) in two-parameter and one-parameter comparisons. These latter approaches assume that the slow-repair component, described by the slow-repair rate s and the fraction of the slow component Q, is affected in repair-deficient cells, with an expectation that the slow component Q will increase and/or the slow-repair rate s will decrease in repair-deficient cells. This postulation is supported by the results presented in Table 4, which indicate a statistically significant increase in the fraction of the slow-repair component in all of the repair-deficient cell strains (C2 to C6). The results presented in Table 4 from cells C2 to C6 suggest that the increase of the fraction of the slow-repair component can be considered as a criterion for the identification of RS.

Table 4.

Results of the Pairwise Comparison of the RS Controls (C2–C5) and Five SCID (P1–P5) Patients with Normal Control (C1)

Test/Culture ID	Paired t-test	Three-parameter comparison (f, s, Q)	Two-parameter comparison (s, Q), shared f	One-parameter comparison (Q), shared s	One-parameter comparison (Q), fixed s
C2	0.025∗	0.0008∗	0.0001∗	0.014∗	0.0012∗
C3	0.14	0.12	0.089	0.075	0.049∗
C4	0.054	<0.0001 (63)∗	<0.0001 (106)∗	<0.0001 (178)∗	<0.0001 (182)∗
C5	0.097	0.28	0.12	0.11	0.029∗
C6	0.11	0.043∗	0.013∗	0.031∗	0.026∗
P1	0.27	0.35	0.22	0.51	0.51
P2	0.021∗	0.015∗	0.0031∗	0.0025∗	0.0008∗
P3	0.39	0.89	0.71	0.69	0.72
P4	0.17	0.54	0.61	0.28	0.35
P5	0.24	0.21	0.29	0.17	0.22

P values (statistical significance levels) are presented for five different approaches as described in Statistics. Three-parameter comparison was performed based on independent regression analysis of each data set (Table 3). Two-parameter and one-parameter comparisons were performed by parallel regression of two data sets (control and test) with some parameters shared between two data sets or fixed as described in the text in this table. For C4 group, F distribution critical values are shown in parentheses.

^∗P < 0.05.

γ-H2AX Assay in Test Patient Cells

We applied a similar analysis using a two-component repair model in fibroblasts derived from the five SCID or CID patients with unknown RS status (Table 2). The experimental results on γ-H2AX foci kinetics in these cell lines and the regression curves are presented in Figure 2. Table 3 summarizes the best-fit parameter values from the independent regression analysis of each cell line data set, and Table 4 shows the results (P values) from the comparisons of the parameters between normal control cells (C1) and each of the five cell cultures from the SCID or CID patients, using the same statistical criteria as described for reference cells. These results indicate a significantly increased fraction of the slow-repair component Q compared with that in control cells in P2 only. Accordingly, this patient was classified as RS.

Figure 2.

Comparative analysis of post–3-Gy irradiation (IR) γ histone 2A isoform X (H2AX) foci kinetics in two reference cultures, C1 (normal control) and C2 (DCLRE1C null), and the five cell strains from patients with unknown radiosensitive status (P1 to P5). A–E: The repair kinetics of each patient's cell strain is shown in a separate panel, alongside with kinetics for the normal (C1) and DCLRE1C-deficient (C2) reference cells for comparison. Circles represent the principal mean foci numbers; error bars show the SEMs of the means before IR exposure, at 0 hour, and after IR exposure, at 0.5, 2, 6, 24, 48, and 72 hours. Lines demonstrate the results of regression analysis of the repair kinetics using the two-component model according to Equation 2. F: Kinetics of each cell culture (C1, C2, and P1 to P5). G: Representative laser confocal microscopy images of fibroblasts from normal (C1) and DCLRE1C-deficient (C2) donors and from the patient (P2) immunostained with anti–γ-H2AX antibody (green channel). Nuclei are counterstained with propidium iodide (red). An increased number of γ-H2AX foci is evident at 6 and 24 hours after IR in C2 and P2 cells compared with C1 cells.

The results presented in Table 3 for the s and Q parameter values indicate a large degree of ambiguity (interdependence) between these two parameters for both reference cell lines (C1 to C6) and cells from SCID or CID patients. This ambiguity is reflected in the large SEMs and the lack of an overall trend in differences in these parameters between normal and RS cells. This finding follows, to a large extent, from the limited number of experimental data points (four) in the interval between 6 and 72 hours. Although the parameters of the repair kinetics are derived from the regression analysis of the whole data set, the later data points are more crucial for the evaluation of the slow-repair component. This compromised precision, in addition to the interdependence of the slow-repair rate s and the fraction of slow component Q, prevented the development of a single universal numerical criterion, such as a fixed threshold value for the fraction of slow-repair component Q, that would allow for distinguishing between normal cells (below threshold) and RS cells (above threshold). Therefore, we attempted a comparison based on pairwise regression, as described in Statistics for the last two approaches. The results presented in Table 4 from one-parameter comparisons, with slow-repair rate s fixed at the value for the test data set, allow for the detection of statistically significant differences, whereas other criteria fail to detect such differences, as in cases C3 and C5.

To develop a universal criterion for the identification of RS status, we exploited graphical presentation of the one-parameter comparison for the fraction of the slow-repair Q with fixed values of the slow-repair rate s (described in Statistics). We performed the nonlinear regression analysis of the experimental γ-H2AX foci kinetics for control cells C1 with a range of fixed values s, and plotted the obtained values of the fraction of the slow-repair component Q as a curve on the graph using Q versus s axes (Figure 3A). This curve reflects the ambiguity between Q and s and allows for the specification of a threshold value of Q at a given value of s. For further validation of this approach, we plotted on this graph the best-fit values obtained from seven independent experiments conducted at the University of Sussex for this and other (48BR) control cell cultures (Figure 3A). These results fit well within SEMs of the threshold curve and demonstrate the interdependence between Q and s. For the prediction of RS status, we therefore generated a threshold curve (Figure 3).

Figure 3.

The radiosensitive (RS) map. A: The thick line represents Q values obtained from the nonlinear regression analysis (Equation 1) of the γ histone 2A isoform X (H2AX) foci kinetics of normal control cell culture C1 (Figure 1) using a range of fixed s values, with closed circles representing the best-fit s and Q values; the thin lines indicate SEMs of these Q values. The area between these lines can be defined as a normal zone. Open circles represent best-fit s and Q values from nonlinear regression analysis of the γ-H2AX foci kinetics of normal control cells (C1 and 48BR) obtained in seven independent experiments. The dashed line represents the boundary of the normal zone calculated as a set of best-fit Q values obtained from the regression of the kinetics data from the same seven independent experiments using a range of fixed s values, based on the kinetics data from the same seven independent experiments. B: The individual data points are best-fit s and Q values of the independent nonlinear regression analysis of the γ-H2AX foci kinetics of all of the investigated reference and patient cell cultures (C1 to C6, P1 to P5). Lines represent the normal zone corresponding to the dashed line in A. Data points above the normal zone indicate increased cellular RS status. C: The individual data points are best-fit s and Q values of the nonlinear regression analysis of the foci kinetics reported in seven RS–severe combined immunodeficiency or RS–combined immunodeficiency patients (DCLRE1C deficient: DB333,⁴¹ CJ176,⁴¹ AA554⁷; LIG4 deficient: 495GOS,⁴⁸ 411BR,⁴⁹ 2303⁴⁹; XLF deficient: F07/402⁴⁸). Lines represent the normal zone.

The results of the independent regression analysis of all reference and patient cell cultures in the present study (s and Q values; Table 3) are presented as individual data points in Figure 3B, which we termed an RS map. This RS map allows for the graphical comparison of Q values for normal control cells C1 (threshold curve) and cells with unknown RS status at a fixed value of s that is specific for these cells, thus providing classification of a data set (a patient) as RS if the s/Q data point is located above the threshold curve on the map. According to this criterion, repair-deficient cultures C2 to C6 and P2 are classified as RS, and P1, P3, P4, and P5 are on the edge of normal RS zone with a trend for increased RS. This classification matches the RS status identified from Table 4.

Figure 3C is a further validation of the RS algorithm and shows the best-fit Q and s values obtained from nonlinear regression analysis of the γ-H2AX foci kinetics from seven primary fibroblast cultures obtained from RS-SCID or RS-CID patients with determined deficiencies in NHEJ factors, which have been previously characterized by our group.^7,41,48,49 Since all data points were located in the RS zone (above the threshold curve), all of these patients were confirmed as RS.

Comparison of γ-H2AX Results with the Clonogenic Survival Assay

Although it is the gold standard in measuring cellular RS, the clonogenic survival assay has limited applicability in SCID and CID patients, since primary cultures, especially those with the RS phenotype, may not survive in the assay conditions. Nonirradiated C4 cells (PRKDC mutated) and C6 cells (XLF mutated) cells, when plated at low density, did not form countable colonies, so the survival curve experiments could not be conducted.

The results of these experiments in fibroblasts from reference cultures C1, C2, C3, and C5 and SCID or CID in P1 to P5 irradiated with a range of radiation doses (survival curves) are presented in Figure 4. Experimental survival curves were approximated by nonlinear regression analysis using the linear-quadratic model (Equation 3). Since resistant cell subpopulations were observed in survival curves for fibroblasts from the C2, C3, C5, and P2 groups, an additional exponential component was added to the linear-quadratic equation (Equation 4). To identify the RS status of the cells, we calculated for each group the sensitization factor (SF) as an inverse value of the dose-modification factor.⁵⁰ The SF was calculated as a ratio of doses that reduce the survival level to 10% in D₁₀(control) and D₁₀(test) groups; SF = D₁₀(control)/D₁₀(test). Values of SF, along with P values for testing the hypothesis that SF = 1, are detailed in Figure 4.

Figure 4.

A: Radiation dose–response curves for cell survival of four reference cell cultures (C1, normal control; C2, C3, and C5, cells with known deficiencies). B–F: The five SCID patient cell cultures with unknown radiosensitivity (RS) status. Lines are generated by nonlinear regression using the linear-quadratic model. Sensitization factors (SFs) and significance versus normal control cells (C1) are as follows: C2, SF = 4.24 ± 0.92, P < 0.0003; C3, SF = 1.64 ± 0.20, P < 0.009; C5, SF = 6.26 ± 1.09, P = 0.0001; P1, SF = 1.50 ± 0.30, P = 0.074; P2, SF = 2.68 ± 0.79, P < 0.006; P3, SF = 1.18 ± 0.09, P = 0.082; P4, SF = 1.01 ± 0.10, P = 0.92; and P5, SF = 1.19 ± 0.53, P = 0.698.

A marked and statistically significant increase in SF in mutant reference cultures compared with normal control cells (C1) was evident in DCLRE1C-deficient cells (C2; SF = 4.24), leaky LIG4 cells (SF = 6.26), and in P2 cells (SF = 2.68). An intermediate increase in RS was observed in C3 cells with a leaky DCLRE1C deficiency (SF = 1.64) and in P1 (SF = 1.50) and P5 (SF = 1.19), but the differences in P1 and P5 were not statistically significant. No statistically significant differences compared with normal control cells were detected in P3 and P4.

The differences obtained from the analysis of postirradiation γ-H2AX kinetics are consistent with the RS status of the SCID or CID patients' cells detected on the clonogenic survival assay. In P1 and P5, a trend for increased RS was evident from the γ-H2AX data set and RS map (Figures 2 and 3); however, the difference is not statistically significant, in agreement with nonsignificant trends for the clonogenic survival end point.

Genetic Analysis in SCID and CID Patients

Genetic analysis of cDNA derived from the five SCID or CID patients for DCLRE1C, XLF, LIG4, and XRCC4 genes identified mutations in P2 and P5 (Figure 5). P2 carried two heterozygous mutations in LIG4 (c.73C>T=p.R25X and c.845A>T=p.H282L). P5 carried a heterozygous loss of exon 11 in Artemis-DCLRE1C.

Figure 5.

Identification of mutational changes in cDNA in P2 and P5. A: Mutational changes in LIG4 (P2). Dye-terminator sequence figures illustrating the c.73C>T = p.R25X (in the presented figure, C was automatically selected) and c.845A>T = p.H282L mutational change in P2 and respective wild-type sequence from C1. B: Locations of the identified mutational changes in LIG4 in relation to important domains. C: Mutational changes in DCLRE1C (P5). Dye-terminator sequence figures illustrating the Δexon 11 mutational change in P5 and respective wild-type sequence from C1.

Patient Response to HSCT

Summarized patient information is presented in Table 1, and expanded clinical case descriptions are included in Patient selection for RS testing. All patients underwent reduced-intensity (P1, P3, P4, and P5) or myeloablative RS-adjusted conditioning (P2) regimens inclusive of fludarabine combined with either reduced-dose busulfan, cyclophosphamide, or treosulfan, and serotherapy (alemtuzumab or antithymocyte globulin) before HSCT. All three patients who received matched unrelated donor BMTs (P3, P4, and P5) time after treatment. P2 received an umbilical cord blood transplant and achieved complete donor cell engraftment and chimerism within 2 months of transplantation. However, he succumbed to multiple post-transplantation complications 1 month later. P1 received an umbilical cord blood transplant and did not achieve satisfactory T-cell engraftment. She experienced chronic graft-versus-host disease and multiple opportunistic infections and died 21 months after transplantation.

Discussion

NHEJ is a main mechanism of DNA DSB repair in mammalian cells since it can occur throughout the cell cycle independently of the availability of a sister chromatid as a homologous template.^51,52 It is also an essential process in the development of the immune response with the purpose of generating enormous antibody diversity of Ig and T-cell receptor genes, thus enabling defense against a wide variety of specific pathogens.⁴ Consequently, deficiencies in NHEJ proteins lead to RS-SCID or RS-CID because DSBs are left unrepaired, misrepaired, or repaired with a substantial delay. DSBs are the most significant DNA lesions; unrepaired DSBs can cause cell death, and misrepaired DSBs cause genomic instability and lead to mutations. Therefore, RS-SCID or RS-CID patients are likely to overreact to some pre- and post-transplantation agents used in standard HSCT.^16–18

Severe or mild mutations in NHEJ factors impair postirradiation γ-H2AX kinetics; moreover, defects in early, late, or both early and late DSB rejoining have been associated with distinct repair profiles.^6,7,14,40 Analysis of γ-H2AX foci in irradiated primary fibroblasts derived from SCID patients with an RS-like phenotype has been used in the laboratory of Dr. Peggy Jeggo for diagnostic purposes, to predict the response to HSCT conditioning treatment based on the γ-H2AX repair curves (ie, DNA DSB-repair capacity).^6,7,14

For instance, cells with defects in LIG4 or XLF exhibit dramatically slow DSB repair but eventually are capable of rejoining all DSBs due to residual protein function. Cells defective in Artemis-DCLRE1C initially repair DSBs with the same kinetics as do normal cells but fail to rejoin DSBs completely. DNA-PKcs-PRKDC mutants have defects in both the fast and slow kinetics of DSB repair.^14,40 Although generally the DSB-repair profiles are described intuitively, an objective quantitative analysis can be applied. Recently we reported a new mathematical approach to examine the γ-H2AX foci repair on the basis of their frequency distributions and calculation of average foci per cell values in the major Poisson-distributed subpopulation, yielding what we termed the principal average.⁴⁵ Such an approach avoids distortions related to the presence of proliferating cells in the population⁵³ and the detection of subpopulations of cells with defective foci formation and repair. Comparison of the principal average between RS and non-RS cells allowed for a better resolution of differences and a more reliable identification of minor DSB-repair defects.

Here, to inform clinical decisions associated with HSCT, we have further refined the analysis of postirradiation γ-H2AX foci kinetics and selected analytical criteria that are quantitative and appropriate for the detection of cellular RS status of primary fibroblast cultures from five SCID or CID patients from three Australian medical centers. We compared their cells with reference cells that had been derived from donors without or with NHEJ deficiencies. We first attempted to differentiate the kinetics of γ-H2AX foci in cells with known deficiencies in NHEJ factors from the kinetics of normal cells, and to investigate the correlation between these deviations and the RS status determined using the clonogenic survival assay. We demonstrated that the increase in the fraction of the slow-repair component is a common feature of all repair-deficient cell strains (C2 to C6). This finding correlated with the increased RS determined using the clonogenic survival assay and indicates that the increased fraction of the slow-repair component may be considered as a sufficient condition for the classification of cells as RS.

Our estimation of the fraction of the slow-repair component is based on the use of the two-component empirical repair model to approximate the kinetics of γ-H2AX foci in the interval up to 72 hours. Our choice of this model is dictated by the number of available experimental points (six) and the minimal number of parameters (four) that is sufficient for the approximation of these data. The one-component repair model proved to be insufficient for this purpose. On the other hand, increasing the complexity of the model by adding, for example, a nonrepaired component, will result in overparameterization of the model. We do not subtract the background number of foci, which affects to some extent the analysis of the slow component. This method of empirical analysis is valid for diagnostic purposes but may not fully reflect the biological slow-repair process. Although the two-component mathematical repair model is supported by a biological model of biphasic DNA DSB repair that assumes the presence of a subset of more complex DNA DSB that require end-processing by Artemis nuclease before end-joining,^40,41 we do not draw a direct parallel between this subset of DNA DSBs and the fraction of slow-repair component Q in our model. We rather consider the parameters Q and s as empirical measures that allow for the identification of RS status from the experimental γ-H2AX foci kinetics within a certain time interval. Apart of the ambiguity of these parameters, their values are expected to depend on the duration of this interval, as it has been reported for DNA DSB repair parameters.⁵⁴

Analysis clearly revealed one patient (P2) with an increased slow-repair component that was classified as RS. This finding was confirmed by clonogenic survival assay, with SF = 2.68. No statistically significant increase in the fraction of the slow repair was observed in cells from P1, P3, P4, and P5; however, P1 cells demonstrated increased RS in clonogenic survival assay, with SF = 1.50. Summarizing these observations, although we suggest that the increased fraction of the slow-repair component is an indication of RS, the lack of such does not guarantee a normal RS status, as in P1. It is worthwhile to note, however, that a trend of the increased fraction of the slow repair could be seen by visual inspection of γ-H2AX foci kinetics in this patient (Figure 2). The statistical significance of this deviation could have been achieved with more data points. In this study, we approximated six experimental points with an equation that contains four free parameters, thus limiting the number of degrees of freedom to 2. A more extensive data set of γ-H2AX foci kinetics would be more reliable identification of RS status. We introduced an RS map, which allows for the graphical identification of the RS status and is useful in the interpretation of the interplay between the rate and the fraction of the slow-repair component that can arise from not only the data noise but also the real intervariability of these parameters in cells with various degrees of RS.

The clinical presentation of RS-SCID or RS-CID arising from a deficiency in the NHEJ component LIG4, XLF, DCLRE1C, or PRKDC is heterogeneous and summarized by Woodbine et al¹⁴ in relation to the current knowledge of the structure and role of the protein in NHEJ. Patients' susceptibility to infection, the severity of their immunologic deficiency, and dysregulation may be variable at presentation and may evolve with time. A number of supporting clinical features may help to identify possible RS.

Microcephaly, growth delay, and dysmorphic facial features are commonly observed in LIG4 syndrome patients,^19,49 and the severity of microcephaly has been suggested to correlate with the degree of immunodeficiency.⁵⁵ However, one patient with SCID phenotype secondary to LIG4 mutation displayed no microcephaly or growth delay.⁵⁶

History of malignancy should raise suspicion of RS, as lymphoid tumors have been observed in a number of LIG4-deficient patients, usually those with milder immunodeficiency.^57–60

Artemis-DCLRE1C defect is the most common cause of RS-SCID and RS-CID.⁶¹ DCLRE1C-null patients present with severe immunodeficiency that is more marked than in most LIG4- or XLF-deficient patients, who show reduced but residual T and B cells (CID or pancytopenia). Microcephaly, growth delay, or dysmorphism has not been reported in DCLRE1C-null patients. More recently, patients with hypomorphic mutations in DCLRE1C resulting in phenotypic variability ranging from Omenn syndrome⁶² to Epstein-Barr virus–associated lymphomas^63,64 have been described.

There are very few opportunities for pediatricians faced with decision making regarding HSCT for SCID or CID to access RS assays. Based on the findings from the present study, in SCID and CID patients with malignancy, myeloid lineage cytopenias, dysmorphism, and/or developmental delay, we recommend the following approach to predicting clinical sensitivity to HSCT conditioning treatment.

Establishing Primary Fibroblast Culture

The essential starting point is to establish a primary fibroblast culture, to provide cells and DNA for the following studies.

Assessing RS Genotype

DNA sequenced for mutations in the following panel of genes: RAG1, RAG2, LIG4, XLF, DCLRE1C, XRCC4, and PRKDC. However, mutations at these loci can be associated with a leaky phenotype (eg, reference cultures C3 and C5). Moreover, there are cases that show mild cellular RS based on findings from the γ-H2AX and survival assays, but no genetic defects in these genes have been identified (P1). Therefore, there must be other unknown genetic causes of RS-SCID and RS-CID that are not covered by the panel.

Assessing Cellular RS Phenotype

The γ-H2AX response to ex vivo exposure to IR provides the most convenient assay for assessing RS, by comparing patient-derived cells with normal fibroblast cultures. The shape of a postirradiation γ-H2AX curve can serve as a signature of a particular deficiency and raises a warning signal of possible NHEJ defects.⁴⁰ This report provides a suitable approach to analysis of the data. We plan to make an RS map algorithm freely available online for diagnostic purposes. Although the γ-H2AX assay provides immediate (within a few days) information on the likely RS status of a patient, the clonogenic survival assay should be undertaken. If the patient-derived primary fibroblast cultures clone successfully when plated at low density, radiation survival curves provide a direct and straightforward indication of the extent of RS, to complement the γ-H2AX results.

Allocation to Either the Non-RS, Mild or Possible RS, or RS Category with Corresponding Treatment Options

In our series, P2, who was confirmed to have a LIG4 defect, presented with deteriorating pancytopenia secondary to bone marrow failure and clinical and laboratory evidence of CID in addition to dysmorphism and developmental delay. Rapid and correct identification of his RS defect allowed for the adjustment of pre- and post-transplantation treatment. P5 had slightly abnormal radiation-repair kinetics. The intensity of his pretransplantation conditioning regimen was reduced. Although he had a single mutation on the Artemis gene, we hypothesized that there may have been an association between his genotype and his susceptibility to infection and malignancy.

Conclusion

RS defects should be considered in patients with features of CID (with or without immune dysregulation) and a history of malignancy, or pancytopenia and/or bone marrow failure, or dysmorphism and/or developmental delay. The absence of supporting features does not rule out RS defect. The early and correct identification of RS allows for essential modifications to treatment regimens and prioritization of definitive treatment, including HSCT. Although sequencing techniques have provided an important route toward diagnosis, functional assays are becoming an increasingly important supportive tool.