Interleukin 11 therapy causes acute heart failure and its use in patients should be reconsidered

Background Interleukin 11 (IL11) was initially thought important for platelet production, which led to recombinant IL11 being developed as a drug to treat thrombocytopenia. IL11 was later found to be redundant for haematopoiesis and its use in patients is associated with unexplained cardiac side effects. Here we identify previously unappreciated and direct cardiomyocyte toxicities associated with IL11 therapy. Methods We injected recombinant mouse lL11 (rmIL11) into mice and studied its molecular effects in the heart using immunoblotting, qRT-PCR, bulk RNA-seq, single nuclei RNA-seq (snRNA-seq) and ATAC-seq. The physiological impact of IL11 was assessed by echocardiography in vivo and using cardiomyocyte contractility assays in vitro. To determine the activity of IL11 specifically in cardiomyocytes we made two cardiomyocyte-specific Il11ra1 knockout mouse models using either AAV9-mediated and Tnnt2-restricted (vCMKO) or Myh6 (m6CMKO) Cre expression and an Il11ra1 floxed mouse strain. In pharmacologic studies, we studied the effects of JAK/STAT inhibition on rmIL11-induced cardiac toxicities. Results Injection of rmIL11 caused acute and dose-dependent impairment of left ventricular ejection fraction (saline (2 µL/kg), 60.4%±3.1; rmIL11 (200 mcg/kg), 31.6%±2.0; p<0.0001, n=5). Following rmIL11 injection, myocardial STAT3 and JNK phosphorylation were increased and bulk RNA-seq revealed upregulation of pro-inflammatory pathways (TNFα, NFκB and JAK/STAT) and perturbed calcium handling. SnRNA-seq showed rmIL11-induced expression of stress factors (Ankrd1, Ankrd23, Xirp2), activator protein-1 (AP-1) transcription factor genes and Nppb in the cardiomyocyte compartment. Following rmIL11 injection, ATAC-seq identified epigenetic enrichment of the Ankrd1 and Nppb genes and stress-responsive, AP-1 transcription factor binding sites. Cardiomyocyte-specific effects were examined in vCMKO and m6CMKO mice, which were both protected from rmIL11-induced left ventricular impairment and molecular pathobiologies. In mechanistic studies, inhibition of JAK/STAT signalling with either ruxolitinib or tofacitinib prevented rmIL11-induced cardiac dysfunction. Conclusions Injection of IL11 directly activates JAK/STAT3 in cardiomyocytes to cause acute heart failure. Our data overturn the earlier assumption that IL11 is cardioprotective and explain the serious cardiac side effects associated with IL11 therapy, which questions its continued use in patients. Clinical Perspective What is new? Injection of IL11 to mice causes acute and dose-dependent left ventricular impairment IL11 activates JAK/STAT3 in cardiomyocytes to cause cell stress, inflammation and impaired calcium handling These data identify, for the first time, that IL11 is directly toxic in cardiomyocytes, overturning the earlier literature that suggested the opposite What are the clinical implications? Recombinant human IL11 (rhIL11) is used as a drug to increase platelets in patients with thrombocytopenia but this has severe and unexplained cardiac side effects We show that IL11 injection causes cardiomyocyte dysfunction and heart failure, which explains its cardiac toxicities that were previously thought non-specific These findings have immediate translational implications as they question the continued use of rhIL11 in patients around the world


Clinical Perspective
What is new?
• Injection of IL11 to mice causes acute and dose-dependent left ventricular impairment • IL11 activates JAK/STAT3 in cardiomyocytes to cause cell stress, inflammation and impaired calcium handling • These data identify, for the first time, that IL11 is directly toxic in cardiomyocytes, overturning the earlier literature that suggested the opposite What are the clinical implications?
• Recombinant human IL11 (rhIL11) is used as a drug to increase platelets in patients with thrombocytopenia but this has severe and unexplained cardiac side effects • We show that IL11 injection causes cardiomyocyte dysfunction and heart failure, which explains its cardiac toxicities that were previously thought non-specific • These findings have immediate translational implications as they question the continued use of rhIL11 in patients around the world Introduction Interleukin 11 (IL11) is an elusive member of the interleukin 6 (IL6) family of cytokines, which collectively signal via the gp130 co-receptor.Following its identification in 1990 1 recombinant human IL11 (rhIL11) was found to increase megakaryocyte activity and peripheral platelet counts in mice 2 .Soon after, IL11 was developed as a therapeutic (Oprelvekin; Neumega) to increase platelet counts in patients with chemotherapy-induced thrombocytopenia, received FDA approval for this indication in 1998 and is still used to this day 3,4 .In recent years, longeracting formulations of rhIL11 have been tested in pre-clinical studies and new clinical trials of PEGylated rhIL11 in patients are anticipated 5 .
RhIL11 was also trialled to increase platelet counts in patients with von Willebrand factor deficiency, myelodysplastic syndrome, cirrhosis and sepsis, and tested as a putative cytoprotective agent in numerous other conditions, including myocardial infarction 6 [Table 1 and Suppl Table 1].However, it became apparent that IL11 is not required for basal or compensatory blood cell or platelet counts in mice or humans: IL11 is in fact redundant for haematopoiesis 7,8 .Thus, the effects of injection of high dose rhIL11 on platelets appear nonphysiological and possibly reflect non-specific gp130 activity 9,10 .
Unfortunately, injection of rhIL11 into patients has severe and hitherto unexplained cardiac side effects.Up to 20% of patients given rhIL11 (50 mcg/kg) develop atrial arrhythmias, a high proportion of individuals develop heart failure and rare cases of ventricular arrhythmias and sudden death are reported 11,12 .Furthermore, serum natriuretic peptide levels become acutely and transiently elevated in patients receiving IL11 therapy, with B-natriuretic peptide levels sometimes exceeding those diagnostic of heart failure.
While IL11 was previously thought to be cytoprotective, anti-inflammatory and anti-fibrotic in the heart [13][14][15] and other organs, recent studies by ourselves and others have challenged this premise [16][17][18] .Indeed, experiments over the last five years have questioned the earlier literature and IL11 is increasingly viewed as pro-inflammatory and pro-fibrotic.Given this large shift in our understanding of IL11 and the fact that cardiomyocytes (CMs) robustly express IL11RA, we devised experiments to determine whether IL11 is toxic to CMs and if this could explain cardiac side effects associated with IL11 therapy in patients.previously 19 .In the presence of Cre-recombinase excision of exon 4-7 results in a nonfunctional IL11 receptor.
These mice were crossed with homozygous Il11ra1 floxed females.In the second generation, mice from generation one, heterozygous for the Il11ra1 flox allele and heterozygous for the Cre, were crossed with Il11ra1 flox homozygotes to produce littermate experimental and control animals.

Viral Vector
The viral vector used in this study, AAV9-cTNT-EGFP-T2A-iCre-WPRE (VB5413), was purchased from Vector Biolabs (Malvern, PA, USA).A codon optimised Cre was delivered using an AAV9 capsid and under the control of the Tnnt2 promoter.This was linked to an enhanced green fluorescent protein (EGFP) reporter with a 2a self-cleaving linker.1x10 12 genome copies or an equivalent volume of saline were injected into the tail veins of 8 -9 week old homozygous male Il11ra1 flox mice and from this point mice were housed separately from saline-injected controls for 3 weeks prior to further experiments.

Echocardiography
Echocardiography was performed under light isoflurane anaesthesia using a Vevo3100 imaging system and MX550D linear transducer (Fujifilm Visualsonic Inc, ON, Canada).
Anaesthesia was induced with 4% isoflurane for 1 minute and maintained with 1-2% isoflurane.Mice were allowed to equilibrate to the anaesthetic after induction for 9 minutes before imaging was started.Heart rate measurement from single lead electrocardiogram (ECG) recordings were taken at the completion of the equilibration period.Measurements of ventricular ejection fraction (LVEF) was measured from m-mode images taken in the parasternal short axis (PSAX) view at midventricular level and averaged across 3 heartbeats.
Global circumferential strain (GCS) measurements were also taken from the PSAX view and analysed in a semi-automated fashion by the VevoStrain imaging software (VevoLab, version 5.5.0,Fujifilm Visualsonic).Aortic velocity time integral (VTI) was measured using pulse wave doppler in the aortic arch and an average taken from 3 heart beats.The investigator was blinded to the treatment group for all studies at both the imaging acquisition and analysis stages. qPCR The tissue was washed in ice-cold PBS and snap-frozen in liquid nitrogen.Total RNA was extracted using TRIzol (15596026, Invitrogen, MA, USA,) in RNeasy columns (74106, Qiagen, MD, USA).cDNA was synthesised using Superscript Vilo Mastermix (11755050, Invitrogen).Gene expression analysis was performed using quantitative polymerase chain reaction (qPCR) with TaqMan gene expression assay in duplicate over 40 cycles.Il11ra1: custom TaqMan assay [Suppl Raw RNAseq data and gene-level counts have been uploaded onto the NCBI Gene Expression Omnibus database and will be made available with accession number (GSE240804).

Single nuclei RNAseq
Single nuclei sequencing was performed on flash frozen LV tissue that was extracted from 8 week old male C57BL/6J mice 3 hours after injection with rmIL11 (200mcg/kg) or saline (2µL/kg).The tissue was processed according to standard protocols as previously described 20,21 .Nuclei were purified by fluorescent activated cell sorting and libraries were sequenced using HiSeq 4000 (Illumina, CA, USA) with a minimum depth of 20,000-30,000 read pairs per nucleus.
All single nuclei sequence data generated and analyzed in this study have been deposited as BAM files at the NCBI Gene Expression Omnibus database and will be made available upon request.
ATAC Seq 8 week old male C57BL/6J mice were given an IP injection with rmIL11 (200mcg/kg) or saline.
The heart was excised 3 hours after injection and flash-frozen tissue was sent to Active Motif to perform assay for transposase-accessible chromatin with sequencing (ATAC-seq) analysis.

Protein Analysis
Protein extraction was performed on flash frozen tissue using ice-cold Pierce RIPA buffer (89901, ThermoFisher) supplemented with protease inhibitors (11697498001, Roche, Basel, Switzerland) and phosphatase inhibitors (4906845001, Roche).Tissue was lysed using a Qiagen Tissue Lyser II with metallic beads for 3 mins at 30Hz.Protein quantification was performed using a Pierce bicinchoninic acid assay colorimetric protein assay kit (23225, ThermoFisher).10-20µg of protein was loaded per well and run on a 4-12% bis-tris precast sodium-dodecyl sulfate page gel (NP0323BOX, Invitrogen).Semi-dry transfer was performed using the TransBlot Turbo transfer system (1704150, BioRad, CA, USA) and the membrane was blocked in 5% bovine serum albumin (A3803, Sigma-Aldrich, MO, USA).Primary

Cardiomyocyte extraction
CMs were extracted from the heart of 12 week old male C57BL/6J mice.Mice were deeply anaesthetized with ketamine and xylazine before the heart was harvested.Cells were incubated in Tyrode solution (1mM Ca, 1mM Mg) or Tyrode solution supplemented with rmIL11 (10 ng/mL) for 2 hours before recording.Cells were paced at 1Hz (10V, 10ms pulse width).Cell recordings were made using the Cytocypher high-throughput microscope (Cytocypher BV, Netherlands) and the automated cell finding system was used to identify and take recordings from 20 individual cells per heart per experimental condition.Calcium recordings were performed by incubating CMs with Fura 2AM dye (1uM) for 20 mins before fluorescent recordings were taken.

Statistics
Statistical analyses were performed in GraphPad Prism V9.5.0 unless otherwise stated.
Normality testing was performed using the Shapiro-Wilk test.Hypothesis testing for single comparisons was done using an unpaired two ways Student's t-test for normally distributed data or by Mann-Whitney U test for non-normally distributed data.
Comparisons involving male and female mice were performed using a two-way analysis of variance (ANOVA) with Sidak's multiple comparisons testing.Changes in expression over multiple time points were analysed using a one-way ANOVA with Sidak's multiple comparisons testing.All graphs display the mean and standard error of the mean unless stated otherwise.P-values in RNA seq analysis were corrected for multiple testing using the false discovery rate (FDR) approach.A p-value and FDR of <0.05 was considered significant.

Hierarchical testing of nested data
Statistical analysis of the data from high throughput microscopy of extracted cardiomyocyte experiments were analysed using a hierarchical statistical approach 22 .This approach tests for clustering within the data as may occur due to differences in the quality of myocyte preparation on different days.This uses a two-level random intercept model of linear regression.The analysis was performed using R-studio and the data was presented as the mean and standard deviation and effective n number taking the intraclass clustering into account.well-characterised and potent inhibitor of calcium current through L-type calcium channels 24,25 and its acute upregulation may account for the changes in calcium transients seen in isolated CM preparations.

ATAC-Seq highlights AP-1 family genes
To better understand the molecular changes induced by IL11 in the heart, we performed Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) analysis.This methodology identifies regions of the genome undergoing epigenetic variation to make transcription factors binding sites more or less accessible.6].These genes belong to the activator protein-1 (AP-1) transcription factor family, which is important for cardiomyocyte stress responses, cardiac inflammation and fibrosis. 31,32Notably, the STAT3 binding motif was also highly enriched.
We revisited our bulk RNA-seq data to examine the expression of the AP-1 transcription factor family transcripts after rmIL11 injection.This revealed that almost all of the AP-

Viral-mediated CM-specific deletion of Il11ra1
To test whether the acutely negative inotropic effects of IL11 and the signature of increased CM stress are specifically mediated via IL11 activity in CMs, we proceeded to conditionally delete Il11ra1 in CMs in the adult mouse.We used an adeno-associated virus serotype 9

JAK inhibition protects against IL11-induced cardiac dysfunction
Canonical IL11 signalling through the IL11RA/gp130/JAK/STAT3 pathway has recently been implicated in the acute pro-inflammatory effects of IL11 34   cytoprotective in CMs [13][14][15] , which we challenge here.In retrospect, the use of rhIL11 in a clinical trial of patients with myocardial infarction was likely ill-founded 6 .
The powerful enrichment of the AP-1 family of transcription factors following rmIL11 injection, seen in bulk RNA-seq, snRNA-seq and ATAC-seq was unexpected and likely has detrimental effects in the mouse heart 31,37 .AP-1 family activation is not immediately downstream of IL11:IL11RA:gp130 signalling and thus, the early IL11-stimulated activation of JAK/STAT3 likely primes CMs to upregulate, activate and respond to AP-1 transcription factors.In the injured zebrafish heart, AP-1 contributes to sarcomere disassembly and regeneration 38 , which is IL11-dependent 39 , providing an evolutionary context for IL11mediated effects in the heart 40 .
Our use of two mouse models of CM-specific Il11ra1 deletion show and replicate that the effects of rmIL11 on cardiac function are via direct cardiotoxic effects on CMs and are not explained by changes in circulating volume, as has previously been suggested 36 .The models used in this study involved the administration of a single dose of rmIL11 however in clinical practice, courses of therapy can involve daily infusions of rhIL11 for up to 21 days between chemotherapy cycles which are likely to compound the effect on the heart, specifically on fibrotic pathologies that are slow to establish 30 .
There are several limitations to our study.The discrepancy between the tachycardia seen in vCMKO but not m6CMKO mice suggests a direct effect of IL11 on sinoatrial node, which is differentially deleted for Il11ra1 between the models, was not explored.We did not determine the putative roles of Rrad or Camk2d in IL11-induced contractile dysfunction, which should be investigated more fully in follow-on studies.Effects on human CMs were not examined, although we have observed conserved effects of IL11 on multiple cellular phenotypes in varied human and mouse cell types (e.g.fibroblasts, hepatocytes, epithelial cells) 16,23 .Whether endogenous IL11 is toxic to CMs and negatively inotropic in heart failure syndromes is not known and we cannot extrapolate from the data seen with acute, high dose injection of recombinant protein.The cardiac side effects associated with IL11 include arrhythmias (notably atrial fibrillation and flutter) that we did not study here.
In conclusion, we show for the first time that IL11 injection causes IL11RA-dependent, CMspecific toxicities and acute heart failure.These data may explain the serious cardiac side effects that occur with rhIL11 therapy, which have been overlooked.Our findings question the ongoing use of rhIL11, and its further development 5 , in patients with thrombocytopenia while identifying novel toxic effects of IL11 in the cardiomyocyte compartment of the heart.

Figures
Figures

Figure 2 .
Figure 2. Transcriptional changes in the myocardium following rmIL11 injection.Volcano plot of all detected genes (A) 1 hour (n=3) and (B) 3 hours (n=4) after intraperitoneal injection of rmIL11 at 200 mcg/kg.Red lines are drawn at Log2Fc of 1 and -1 and FDR of 0.05.(C) Chart of most significantly enriched KEGG terms from at 1-hour post injection of rmIL11 ranked by FDR.(D) Gene set enrichment analysis of the most highly enriched Hallmark gene sets from RNAseq data at 1 hour after injection of rmIL11 ranked by normalised enrichment score.

Figure 3 .
Figure 3. Single nuclear RNA sequencing reveals an IL11-induced cardiomyocyte stress signature.(A) UMAP embedding of all cell types from the left ventricle of male C57BL/6J mice 3 hours after intraperitoneal injection of rmIL11 (200 mcg/kg) or an equivalent volume of saline (n=5).(B) Comparison of cellular composition of the left ventricle in rmIL11 treated mice compared to saline treated mice.(C) UMAP embedding of cardiomyocyte fraction.4 distinct clusters are identified based on gene expression.(D) UMAP embedding of cardiomyocytes annotated with the treatment group.(E) UMAP embedding of cardiomyocyte fraction of saline or rmIL11 treated cardiomyocytes annotated with relative expression of Nppb, Ankrd1, and Camk2d.

Following
IL11 administration, there were a large number of loci with variation in DNA accessibility (increased, 945; reduced, 445; shrunkenLog2FC:+/-0.3, Padj<0.1)[Fig 4A & Suppl Table 5].The top twenty most differentially enriched regions [Fig 4B, C] include areas adjacent to Camk2d, Ankrd1 and Nppb, stress and calcium handling genes that we had already found to be upregulated in CMs by snRNAseq [Fig 3E, Fig 4B & Suppl Table 4].DNA motif analysis of sequences captured by ATAC-seq, revealed the most enriched motifs after rmIL11 treatment were targets for FOSL2 and JUNB transcription factors [Fig 4D & Suppl Table 1 family transcripts are upregulated in the heart after rmIL11 [Fig 4E].We then queried the snRNA-seq data and observed that Fosl2, Junb, Atf6, Jun, Atf3 and Mafg are all significantly differentially expressed in cardiomyocytes following rmIL11 injection [Fig 4E and Suppl Table4].

Figure 4 .
Figure 4. ATAC-Seq reveals a stress signature that occurs acutely in the myocardium after rmIL11 injection.(A) Number of positively and negatively enriched genomic regions identified by ATAC-Seq analysis of the myocardium 3 hours after injection of rmIL11 (n=4).(B) Top 20 most strongly enriched DNA regions in ATAC-seq analysis and adjacent genes, when present.(C) Top 20 most strongly negatively enriched DNA regions in ATAC-seq analysis and adjacent genes.(D) De novo Homer motif analysis of ATAC-seq data most highly enriched motifs in myocardial samples.(E) Heatmap of AP-1 transcription factor family members from bulk RNA sequencing data of myocardium at baseline, 1, 3 and 6 hours after rmIL11 injection.Genes differentially expressed in cardiomyocytes in single nuclear RNA sequencing data are highlighted in red.

Figure 7 .
Figure 7.The acute toxic effects of rmIL11 are mediated via JAK/STAT signalling.(A) Schematic of the pretreatment of wild type male C57BL/6J mice with JAK inhibitor (JAKi) or vehicle (Veh) 30 mins before administration of recombinant mouse IL11 (rmIL11) or saline.(B) Western blot of myocardium lysate in mice treated with a combination of ruxolitinib (30 mg/kg) or vehicle and saline or rmIL11.Membranes have been probed for pSTAT3 and STAT3 (n=3).Mice treated with a combination of vehicle (Veh), ruxolitinib (Ruxo) or tofacitinib (Tofa) and either saline or rmIL11 had an echocardiogram performed 2 hours after treatment which measured (C) left ventricular ejection fraction, (D) global circumferential strain, (E) velocity time integral (VTI) in the aortic arch and (F) heart rate (n=4).QPCR of myocardial tissue from combinations of Ruxo and rmIL11 treatments of (G) Nppb expression and (H) Rrad expression (n=3).Statistics: Comparison between groups by one-way ANOVA with Sidak's multiple comparisons test.Significance denoted as denoted *p<0.05,**p<0.01,****p<0.0001.

Table 1 . Human clinical trials registered with clinicaltrials.gov using recombinant human 107 interleukin 11.
NCT00151125Phase II Study of IL-11 (Neumega) in Von Willebrand Disease Jul-04 12 Completed Imperial College London.Animal experiments were carried out under UK Home Office Project License P108022D1 (September 2019).Wild type (WT) mice on a C57BL/6J background were purchased from Charles River (Cat#632).They were bred in a dedicated breeding facility and housed in a single room of the experimental animal facility with a 12-hour light-dark cycle and provided food and water ad libitum.The Il11ra1 floxed mouse (C57BL/6-Il11ra1 em1Cook /J, Jax:034465) has exons 4-7 of the Il11ra1 gene flanked by loxP sites as has been described Genscript) was dissolved in PBS and injected IP at a dose of 200 mcg/kg.Mice were randomly assigned to a treatment group using a random number generator and syringes for injection were prepared and blinded by a different investigator than administered the IP injection.

Table 3
This work was supported by grant funding from Wellcome Trust (203928/Z/16/Z), Foundation Leducq [16 CVD 03], the Medical Research Council (UK), The NIHR Biomedical Research Centre Imperial College London, the National Medical Research Council (NMRC) Singapore STaR award (NMRC/STaR/0011/2012) and a Goh Cardiovascular Research Award (Duke-NUS-GCR/2015/0014).For the purpose of open access, the authors have applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising.Disclosures SAC is a co-inventor on a number of patent applications relating to the role of IL11 in human diseases that include the published patents: WO2017103108, WO2017103108 A2, WO 2018/109174 A2, WO 2018/109170 A2.SAC is also a co-founder and shareholder of Enleofen Bio PTE LTD and VVB PTE LTD.