Efficacy of Ruxolitinib Therapy in a Patient With Severe Enterocolitis Associated With a STAT3 Gain-of-Function Mutation.

Severe enterocolitis raises diagnostic and therapeutic challenges. Empirical treatment by steroids and immunosuppressive drugs is commonly used but, to date, little attempt has been made to specifically target the mechanism underlying inflammation. Possible causes of severe enterocolitis include refractory celiac disease, autoimmune enteropathy, and common variable immunodeficiency, the 2 latter being overlapping entities of possible monogenic inheritance. Besides FOXP3 mutations that cause immune dysregulation polyendocrinopathy X-linked syndrome, additional genetic causes have recently been reported.5, 6, 7 Herein, using whole exome sequencing (WES), we identified a novel STAT3 (signal transducer and activator of transduction 3) gain-of-function (GoF) mutation in an adult patient displaying chronic diarrhea and autoimmune manifestations refractory to several lines of immunosuppressive drugs and biotherapies. Identification of the mutation enabled successful treatment with the JAK1/2 inhibitor ruxolitinib.

Methods

Ethics Statement

The study was approved by the Ile-de-France II ethical committee (Paris, France).

Characterization of Enteropathy

Histological analysis, flow cytometry analysis of isolated lymphocytes, and molecular analysis of gastrointestinal specimens were performed as described.

DNA Preparation

Genomic DNA was extracted from peripheral blood mononuclear cells isolated by Ficoll HyPaque Plus (GE Healthcare, Velizy-Villacoublay, France) using the QIAamp DNA Blood Mini Kit (Qiagen, Courtaboeuf, France).

Whole Exome Sequencing

WES was performed as previously described. Genomic DNA libraries were generated from DNA (3 μg) sheared with a Covaris S2 Ultrasonicator using SureSelectXT Library PrepKit (Agilent, Garches, France) on the Genomic Platform at the Imagine Institute. Capture by hybridization was performed using Agilent Sure Select All Exon V5 (Agilent, Les Ulis, France). Targeted exons were pulled out with magnetic streptavidin beads, polymerase chain reaction (PCR)-amplified using indexing primers and sequenced on an Illumina HiSeq2500 HT system (Illumina, San Diego, CA). Data analysis was performed with Paris Descartes University/Imagine Institute’s Bioinformatics core facilities. Paired-end sequences were mapped on the human genome reference (US National Center for Biotechnology Information build37/hg19 version) using the Burrows-Wheeler Aligner. Downstream processing was carried out with the Genome Analysis Toolkit, SAMtools, and Picard, according to documented best practices (http://www.broadinstitute.org/gatk/guide/topic?name=best-practices). Variant calls were made with the Genome Analysis Toolkit Unified Genotyper based on the 72nd version of the ENSEMBL database. Genome variations were defined using the in-house software PolyQuery, which filters out irrelevant and common polymorphisms based on frequencies extracted from public databases: US National Center for Biotechnology Information database of SNP (dbSNP), 1000 genomes, Exome Variant Server (EVS, http://evs.gs.washington.edu/EVS/), and Exome Aggregation Consortium (ExAC, http://exac.broadinstitute.org). Consequences of mutations on protein function were predicted using 3 algorithms: Polyphen2 (http://genetics.bwh.harvard.edu/pph2/), SIFT (Sorting Intolerant From Tolerant, J. Craig Venter Institute), and Mutation Taster (www.mutationtaster.org). Mutations were then ranked on the basis of the predicted impact of each variant by combined annotation-dependent depletion (CADD), and compared with the mutation significance cutoff, a gene-level specific cutoff for CADD scores (http://pec630.rockefeller.edu:8080/MSC/).

Sanger Sequencing

The mutations were confirmed by Sanger sequencing using specific primers targeting exon 13: forward: TCCGGCTACTTGGTCACCTA-3; reverse: 5-CCCCCATTCCCACATCTCTG-3.

STAT3 Luciferase Reporter Assay of STAT3 N401D Allele

Mutations into human wild-type STAT3 sequence (NM_139276; Origene) were introduced using the GENEART Site-Directed Mutagenesis System (Invitrogen, Thermo Fisher Scientific, Waltham, MA) and the following primers: c.1201A>G forward: ATGGAAGAATCCAACGACGGCAGCCTCTC TG; c.1201A>G reverse: CAGAGAGGCTGCCGTCGTTGGATTCTTCCAT; c.1175A>G forward: TGGGCACAAACACAAGAGTGATGAACATGGA; c.1175A>G_reverse: TCCATGTTCATC ACTCTTGTGTTTGTGCCCA according to the manufacturer’s instructions. Each mutation was confirmed by direct sequencing of the STAT3 insert (Eurofins, Luxembourg). STAT3 inserts were, then, subcloned into pLVX-EF1alpha-IRES-mCherry lentiviral expression vector (transfer vector) (Clontech, Saint-Germain-en-Laye, France) via the restriction site Not1 (New England Biolabs, Evry, France) and the correct orientation of the insert was confirmed by sequencing. Lentiviral particles encoding the different mutants were generated by transfecting HEK293T cells with transfer plasmid, packaging expressing plasmid psPAX2 (Addgene, Cambridge, MA), VSV-G envelope expressing plasmid PMD2.G (Addgene) using Lipofectamine 2000 (Invitrogen, Thermo Fisher Scientific). Six hours after transfection, cells were washed and new medium without antibiotics was added for 60 hours. The recombinant virus–containing medium was filtered and used to transduce HEK293T cells, stably expressing STAT3-responsive firefly luciferase reporter (Qiagen, Courtaboeuf, France) in the presence of polybrene (4 μg/mL). M-cherry–positive cells were sorted by FACS. Comparable expression of STAT3 protein following transduction of each lentiviral construct was assessed by western blot using mouse anti-STAT3 antibody (124H6, 1:1000; Cell Signaling, Saint Quentin Yvelines Cedex, France), and rabbit anti-gapdh (14C10, 1:1000; Cell Signaling). STAT3 reporter activity was evaluated on 10 ng/mL interleukin (IL)-6 stimulation for 48 hours using the Dual-luciferase reporter assay system (E1910, Promega, Charbonnières-les-Bains, France) according to the manufacturer’s recommendations.

Gene Expression Analysis

For analysis of SOCS3 transcription, Epstein-Barr virus (EBV) immortalized B-cell lines derived from the patient, from 2 healthy donors, and from 1 patient carrying the previously described GoF p.T716M mutation in STAT3 DNA binding domain, were stimulated for 20 hours with 10 ng/mL IL21 (R&D Systems, Lille France) or with 5 ng/mL IL6 (R&D Systems). Cells (1 × 106) were lysed in RLT plus buffer (Qiagen) and RNA extracted using RNAeasy kit (Qiagen). For cytokine mRNA expression studies, intestinal tissues were immediately placed in RNA later (ThermoFisher Scientific) and stored at –80 oC. RNA was extracted using RNAeasy kit (Qiagen). Quantitative reverse transcriptase PCR was performed on Applied Biosystems 7300 Real-Time PCR apparatus as described using TaqMan universal PCR Master Mix and Taqman probe assays (ThermoFisher Scientific, Montigny-Le-Bretonneaux France): RPLP0 (Hs99999902_m1); SOCS3 (Hs02330328_s1); GzmB (Hs01554355_m1); TNFalpha (Hs99999043_ml); IFNG (Hs99999041_m1); IL6 (Hs00985639_m1); IL10 (Hs00961622_m1); IL12A (Hs009673447_m1); IL17A (Hs00174383_m1); IL21 (Hs00222327_m1); IL22 (Hs01574154_m1); IL23 (Hs00372324_m1). Results were expressed as relative expression 2–ΔCt normalized to the housekeeping ribosomal Protein Large PO (RPLPO) gene.

Results

Case Report

A Caucasian woman followed for diarrhea since the age of 5 months was referred to our adult center in 2011 at the age of 19. Severe intestinal villous atrophy refractory to gluten-free diet was diagnosed at 3 years. At 6 years, she developed neutropenia, thrombopenia with anti-platelet antibodies, and Hashimoto thyroiditis with anti-thyroglobulin antibodies. At 8 years, severe diarrhea relapsed with subtotal duodenal atrophy and colonic apoptotic lesions. No antibodies against gliadin or enterocytes were detected, but serum immunoglobulin (Ig)G was low (4g/L<5.5g/L) with normal IgM, IgA, and IgE. Between 3 and 18 years, she was successively treated by steroids, tacrolimus, and azathioprine, but had frequent Clostridium difficile infections. Growth hormone treatment failed to improve the severe growth retardation (−2.5 SD). In 2011, she had moderate diarrhea on azathioprine with normal duodenal architecture and mild colonic inflammation. CD4, B, and natural killer lymphopenia in peripheral blood was ascribed to prolonged treatment by azathioprine. Frequency of CD25highFoxp3+ cells among peripheral CD4+ T cells was 4% (normal: 4%–20%). Phenotyping of isolated duodenal lymphocytes showed predominance of CD3+CD8+TCRαβ+ cells among intraepithelial lymphocytes and comparable frequency of CD4+ and CD8+ TCRαβ+ cells in the lamina propria. Expression of the activating natural killer receptor NKG2C on 10% and 5% of lymphocytes in epithelium and lamina propria, respectively, suggested moderate activation. Molecular analysis of gastric, duodenal, and colonic biopsies revealed polyclonal TCRγ, IgH, and Ig Kappa repertoires. No anti-AIE75 KD antibodies were detected and serum IgG1 (2.4 g/L< 4g/L) and IgG3 (0,04g/L <0.17g/L) were low. Between 2012 and 2014, the patient had severe viral infections, including primary parvovirus B19, cytomegalovirus infections and EBV reactivation. She received high doses of intravenous immunoglobulins and azathioprine was switched to budesonide. Adalimumab, introduced in 2014 because of increased serum concentrations of tumor necrosis factor (TNF)-α, decreased diarrhea and allowed 8 kg weight gain. However, clinical relapse occurred and was not controlled by golimumab, a distinct TNF-α antibody.

Diagnosis of STAT3 GoF Mutation

Disease severity and very early onset suggested a monogenic disease. Accordingly, WES identified a de novo heterozygous missense variant, c.1201A>G, in exon 13 of STAT3 (Figure 1A, Supplementary Table 1), which was confirmed by Sanger sequencing (Figure 1B and C). Both germline loss-of-function mutations and germline and somatic GoF STAT3 mutations have been described.5, 6, 10, 11, 12 Notably, activating germline STAT3 mutations can cause multiorgan autoimmunity and enteritis.5, 6 The c.1201A>G variant was absent in all public databases and in our in-house database (12,925 exomes). This variant replaces the highly conserved asparagine in position 401 to aspartic acid (p.N401D, Figure 1D and E) within the DNA binding domain. This change was predicted to be damaging by all algorithms (Polyphen2 [0.949], Sift [0.02], MutationTaster). CADD score was very high (29.7). As for other previously reported GoF STAT3 mutations within the DNA binding domain,5, 6 we predicted that the mutated protein will retain increased AGG-element binding activity following activation and therefore increased transcriptional activity. Expression of the N401D mutant, in HEK293T cells expressing STAT3-responsive element, resulted in increased luciferase reporter activity compared with wild-type STAT3 on IL6 stimulation and comparable to the known STAT3 GoF allele K392R5, 6 (Figure 1F). Accordingly, transcription of Suppressor of cytokine signaling 3 (SOCS3), a major target of STAT3, was significantly increased on stimulation by IL21 or by IL6 in the EBV cell line from the patient compared with EBV lines derived from 2 controls and comparable to the EBV line from the know STAT3 GoF T716M carrier patient (Figure 1G). Importantly, IL6- or IL21-induced transcription was significantly reduced on inhibition of JAK1 signaling with ruxolitinib (Figure 1G).

Figure 1

Novel STAT3 GoF mutation. (A) WES analysis. (B, C) Sanger sequencing confirming de novo mutation. (D) Schematic representation of STAT3 domains. (E) Conservation of N401 among orthologs. (F) STAT3 transcriptional activity following 48-hour activation with IL6 (10 ng/mL) of HEK293T cells containing STAT3-responsive luciferase and stably transduced with empty vector (EV), wild-type (WT), and mutant STAT3. Results represent the mean of 5 independent experiments (**P < .01, ***P < .001, 1-way analysis of variance). (G) Reverse transcriptase PCR of STAT3-target SOCS3 in IL21 or IL6 activated EBV cell lines treated or not with ruxolitinib (RX). Results are shown as relative expression normalized to RPLPO housekeeping gene and represent the mean of 4 independent experiments. IL6 vs IL6 + RX (*P < .01, Mann-Whitney), control vs STAT3 mutants (**P < .01, ****P < .0001, 1-way analysis of variance).

Supplementary Table 1

Variants Identified by WES

Inheritance mode	Gene	Variant	Nucleotide change	Aminoacid change	Allele origin	Polyphen score	Sift score	Mutation Tester score	CADD score
Autosomal recessive	ZNF721	rs781885961	c.619A>G	p.N207D	Mother	0.84	0.39	T	9.1
		rs781941891	c.485A>G	p.E162G	Father	0.9	0	T	23
	TNFRSF10A	-	c.239G>T	p.R80L	Mother	0.07	0.62	T	0
		-	c.629+5G>C	splicing	Father	-	-	-	5.8
	SPTBN5	-	c.7327C>T	p.H2443Y	Mother	0.73	0.94	T	8.5
		rs377709735	c.8377C>T	p.R2793W	Father	0.95	0.02	T	27.5
	ANKRD11	rs368831797	c.6194T>C	p.F2065S	Mother	0.7	0	DC	27.4
		rs770511968	c.5230C>G	p.H1744D	Father	0.89	0.32	DC	19.7
	MUC16	rs765644259	c.14387C>G	p.T4796R	Mother	-	-1	T	0
		rs201767175	c.42181C>A	p.P14061T	Father	0.99	-1	T	19
De novo	CAPN15/SOLH	rs780136051	c.2624C>T	p.A875V	-	0.97	0.09	DC	24
	STAT3	-	c.1201A>G	p.N401D	-	0.43	0	DC	29.7

NOTE. ANKRD11: not expressed in gut nor in immune cells (nasopharynx). CAPN15 or SOLH: homolog to Drosophila small optic lobes (sol) gene: expressed in spleen, brain, lung and kidney, function poorly described. Bold text concerns the gene of interest.

DC, Disease Causing; T, Tolerated.

Therapeutic Efficacy of Ruxolitinib

Following validation of the STAT3 GoF mutation, and because the patient progressively developed uncontrolled enterocolitis (Figure 2A and B: subtotal duodenal villous atrophy and severe colitis with massive lymphocytic infiltrate and glandular apoptotic bodies), ustekinumab and tocilizumab were tested but without efficacy. In contrast, diarrhea stopped rapidly after introduction of ruxolitinib 10 mg twice a day in April 2017. Ruxolitinib was increased to 15 mg twice a day in July 2017 and budesonide discontinued after 2 months. In November 2017, the patient had gained 5 kg and completed mucosal healing (Figure 2C and D); SOCS3 transcription was markedly reduced in duodenum and colon, indicating efficient blockade of STAT3 signaling; IFN-γ and Granzyme B transcripts were decreased in duodenum and colon and IL-12 and TNF-α transcripts in colon (Figure 2E and F, Supplementary Figure 1). One year later the patient is still in remission.

Figure 2

Induction of mucosal healing by ruxolitinib. (A, B) HES staining (×100) and immunohistochemistry staining (×200) of CD3+, CD8+, GzmB+ lymphocytes before and after 6 months of ruxolitinib (RX). (C and D) Reverse transcriptase PCR quantification of indicated cytokines in patient’s duodenum (C) and colon (D) before (BRu) and after ruxolitinib (ARu). Results are compared with histologically normal duodenal (n = 4) or colonic biopsies (n = 4) (CT), and with duodenal biopsies from 5 active celiac disease (ACeD) (C) and colonic biopsies from 5 active Crohn disease (CrD) (D).

Discussion

Herein, WES identified a novel STAT3 GoF mutation as the cause of a severe enterocolitis. In keeping with previous reports of STAT3 GoF mutations,5, 6, 11 the patient displayed other autoimmune symptoms, primary hypogammaglobulinemia and short stature. Following functional validation of the mutation, the patient was successively treated with ustekinumab and tocilizumab to prevent activation of STAT3 downstream IL23 and IL6 receptor, respectively. Both treatments were inefficient despite 1 case report of enterocolitis with STAT3 GoF mutation clinically improved by tocilizumab. In contrast, ruxolitinib induced rapid and complete remission, indicating that blocking STAT3 activation simultaneously downstream different cytokines may be necessary to control intestinal inflammation. This case is reminiscent of the previous success of this inhibitor in patients with GoF mutation in JAK1, which encodes the kinase upstream STAT3. Importantly, somatic STAT3 GoF mutations are observed in many lymphoproliferative disorders and 1 patient with germline STAT3 GoF mutation developed large lymphocytic leukemia at 14 years. Ruxolitinib treatment may help to prevent such complication in our patient.