TAT–RHIM: a more complex issue than expected

Regulated cell death (RCD) is an essential part of tissue development, and plays fundamental roles in host defences against pathogens and the maintenance of tissue homeostasis [1]. However, excess activation of RCD pathways can be detrimental and is associated with a wide range of human diseases. Therefore, modulation of RCD and molecules involved in signal transduction cascades has great therapeutic potential for intervention in numerous human diseases including (auto)inflammatory, neurodegenerative, infectious, and metabolic disorders [2].

Necroptosis, the best-described form of regulated necrosis, is characterised by plasma membrane rupture and the release of intracellular contents, which is canonically mediated by a signalling cascade of receptor-interacting serine/threonine protein kinases 1 and 3 (RIPK1/3) and mixed lineage kinase domain-like(MLKL) protein [3]. Three different pathways for activation of the necroptosis machinery are currently known. Early work first demonstrated that necroptosis following tumour necrosis factor receptor 1 (TNFR1) activation is dependent on RIPK1 and RIPK3. This pathway is commonly referred to as canonical necroptosis [4]. Subsequently, it was shown that necroptosis can also occur independently of RIPK1 following activation of Toll/interleukin-1 receptor (TIR) domain-containing adapter-inducing interferon-β (TRIF) or Z-DNA-binding protein 1/DNA-dependent activator of IFN-regulatory factors (ZBP1/DAI) [5–8]. RIPK3 hereby serves as the major component in the so-called necrosome, which converges multiple upstream signals to induce necroptosis. Of particular interest here is that RIPK3 is activated via interactions with proteins that contain RIP homotypic interaction motifs (RHIMs) such as RIPK1, TRIF, and ZBP1/DAI [9].

In 2012, it was first demonstrated that association between RIPK1 and RIPK3 via their RHIMs, required for necroptosis, results in the formation of fibrillary structures, termed amyloids [10]. Furthermore, the necrosome, which consists of RIPK3 and MLKL (and in certain settings RIPK1), has an amyloid-like structure [11], and intermolecular RHIM–RHIM interactions are mainly responsible for amyloid formation in necrosomes. RHIMs contain the conserved tetrad core sequence (V/I)-Q-(V/I/L/C)-G, flanked by hydrophobic sequences on both sides. Notably, the structure of the RIPK1–RIPK3–MLKL necrosome has recently been resolved by solid-state NMR. The RIPK1–RIPK3 necrosome is an amyloid-signalling complex that initiates tumour necrosis factor (TNF)α-induced necroptosis. Herein, RIPK1 and RIPK3 alternately stack (RIPK1, RIPK3, RIPK1, RIPK3, etc.) to form heterotypic β-sheets. Two such β-sheets interact along a compact hydrophobic interface featuring an unusual ladder of alternating Ser (from RIPK1) and Cys (from RIPK3) residues [12]. Despite the amyloid characteristics reported to date, it has not yet been possible to fully elucidate which minimal features characterise a functioning RHIM.

Numerous reports by ourselves and others revealed that RIPK1 is a major regulator of signalling pathways that lead to inflammation and RCD, hence RIPK1 has gained considerable interest as a drug target for a spectrum of human diseases [13], including ischemia-reperfusion injury and sepsis [14,15]. Furthermore, studies on RHIM-containing adaptors have unveiled a central role of RIPK1 in controlling several complex pathways related to decisions of cell survival and death [16,17].

RHIM-containing proteins also play a special role in the defence against viral infections. Infection of a cell by various viruses leads to RHIM-mediated cell death, in particular necroptosis [5,18,19]. To ensure their own survival or lifelong infection in the host, specific viruses carry out sophisticated molecular strategies to evade these host cell death responses. For instance, viral proteins like ICP6, encoded by herpes simplex virus 1 (HSV-1), and other members of the Herpesviridae family including the M45 protein from murine cytomegalovirus (MCMV), encode a RHIM domain that inhibits host cell death pathways by forming heteromeric amyloid fibrils with RHIM-containing proteins of the host cell [20]. This indicates that the formation of RHIM-containing amyloid structures is not necessarily accompanied by cell death, and that regulation of cellular life and death following activation of RHIM proteins is far more complicated and context-dependent. Recent atomic force microscopy studies demonstrated that RIPK3 amyloid fibrils can further assemble into large fibrillary networks during necroptosis, and revealed distinct inter-filament interactions in RIPK3 and viral M45 amyloids that contribute to preventing mammalian necroptosis [21]. Formation of viral–host heteromeric amyloid assemblies may underpin a general mechanism of viral adaptation against host immune machineries.

MCMV M45 protein is a RHIM-containing adaptor that, like ZBP1, interacts with cellular RIPK1 and RIPK3 [22], which prompted us to generate a cell-permeable recombinant TAT–M45 fusion protein that should counteract the RHIM–RHIM interactions of RIPK1 and RIPK3 during pathophysiologic cell death processes.

When screening various cancer cell lines and primary cells from different species, we unexpectedly discovered that all tested cells were killed by recombinant TAT–M45 protein in a time- and concentration-dependent manner. Therefore, we changed our primary approach and confirmed this phenomenon by testing a selection of short synthetic TAT–RHIM peptides, ruling out the possibility that the mere presence of the TAT-protein fusion or formation of unfolded protein aggregates induced cell death. Subsequent studies showed that TAT–RHIM has a direct toxic (disruptive) effect on biological membranes. This killing effect is mediated independently of a host cell specific endogenous RHIM domain and cannot be interrupted by any known inhibitor of RCD.

MCMV DNA (Smith strain) was a kind gift from Dr. Michael Winkler (German Primate Center — Leibniz Institute for Primate Research, Göttingen, Germany). cDNA was generated using the Thermoscript RT-PCR System (Invitrogen, Thermo Fisher Scientific, Schwerte, Germany) with 5′-CAAGGGTACCATGGATCGCCAGCCCAAAGTC-3′ forward and 5′-CGGGAATTCTTAGTACAGGAGGGGATCAGCTTTAG-3′ reverse primers. The multiple cloning site of the pTAT vector (a kind gift from Prof. Steven F. Dowdy, UCSD School of Medicine, La Jolla, CA, U.S.A.) includes an N-terminal 6-histidine leader, followed by an 11 amino acid TAT protein transduction domain (PTD) and a polylinker [23]. The PCR product was cloned into the KpnI/EcoRI-cut pTAT-vector, and the resulting construct isolated from a colony harbouring the TAT–M45 cDNA was transformed into competent T7 Express Escherichia coli BL21(DE3) bacteria (New England BioLabs, Frankfurt, Germany). Expression was induced in Luria Bertani (LB) media with isopropyl-d-thiogalactopyranoside at a final concentration of 0.4 mM. After incubation for 3 h at 37°C, cells were harvested by centrifugation (6 500×g, 10 min, 4°C) followed by sonication in binding buffer (300 mM NaCl, 50 mM sodium phosphate, pH 7.0). The lysate was clarified by centrifugation (14 000×g, 20 min, 4°C), and supernatant containing TAT–M45 protein was purified under native conditions using pre-equilibrated TALON Metal Affinity Resin (BD Clontech, Heidelberg, Germany). To remove contaminating bacterial proteins, columns were washed by stepwise addition of increasing imidazole concentrations. Finally, target proteins were eluted with elution buffer (300 mM NaCl, 50 mM sodium phosphate, 200 mM imidazole, pH 7.0). Removal of salt was performed using a disposable Sephadex G-25 PD-10 desalting column (Amersham Biosciences, Freiburg, Germany) equilibrated with RPMI medium. An Amicon centrifugal filter (Sigma–Aldrich/Merck Millipore, Taufkirchen, Germany) was used to concentrate the fusion protein, which was used immediately after purification or stored at −20°C. Protein samples were analysed by gel electrophoresis followed by Coomassie Blue staining (Figure 1B).

Protected peptides (Figures 2A,B, and 3) were synthesised in-house using standard Fmoc/tBu chemistry on a Syro II Multiple Peptide Synthesizer (MultiSyntechTech, Witten, Germany) with acid-labile TCP resin (Intavis, Tübingen, Germany) as described previously [24,25]. After deprotection, peptides were purified by gradient reversed-phase HPLC on a C18 column and their identity was confirmed using electrospray mass spectrometry (ESI-MS) and MALDI-TOF-MS. Peptide purities were ≥95% as determined by analytical reversed-phase HPLC (data not shown). However, large amounts of peptides shown in Figures 2A and 3A,B and I were synthesised commercially by ProteoGenix (Schiltigheim, France) using identical methodology.

RIPK3-expressing NIH3T3-N cells [26], L929, HT-29, and U937 cells were obtained from the American Type Culture Collection (Manassas, VA, U.S.A.). NIH3T3, L929, and HT-29 cell lines were cultured in DMEM (Gibco, Thermo Fisher Scientific) supplemented with 10% (v/v) FCS (PAN-Biotech GmbH, Aidenbach, Germany), 100 U/ml penicillin, and 100 μg/ml streptomycin (both from Merck Millipore GmbH, Darmstadt, Germany). U937 cells were cultured in RPMI 1640 media supplemented with 10% (v/v) FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, and 0.25% d-(+)-glucose (Sigma–Aldrich Chemie GmbH, Taufkirchen, Germany). All cell lines were cultured in a humidified 5% CO₂ atmosphere at 37°C. Negativity for mycoplasma was routinely checked using a MycoAlert Mycoplasma Detection Kit (Lonza, Cologne, Germany). Unless stated otherwise, assays were conducted on 1 × 10⁵ cells in 1 ml of medium.

Nlrp3-knockout (ko) mice (stock no. 021302) were purchased from Charles River (Sulzfeld, Germany). All mice used in this study, including wild-type (wt) animals, were 8-week-old males of the C57BL/6J background. The animals were housed in the Central Animal Facility of the University Hospital Schleswig-Holstein (Kiel, Germany). All mice were maintained on a standard diet and a 12 h day–night rhythm. For isolating bone marrow cells rodents were first anesthetised with isoflurane followed by euthanasia through cervical dislocation. All experiments were performed according to the Protection of Animals Act with the approval of German authorities.

Bone marrow cells were isolated as described previously [27]. Briefly, segregated cells were differentiated into BMDMs by cultivating in high-glucose RPMI 1640 medium with NEAAs (Gibco, Thermo Fisher Scientific), supplemented with 10% (v/v) FCS, 100 U/ml penicillin, 100 μg/ml streptomycin and 30% (v/v) L929-conditioned medium at 37°C with 5% CO₂ for 6 days. For experiments, cells were plated overnight for stimulation at 4 × 10⁵ cells/ml on 12-well plates in a total volume of 1 ml.

CRISPR/Cas9-edited Ripk1-, Ripk3-, and Ripk1/Ripk3/Mlkl triple deficient NIH3T3 cells were generated as described previously [15,28]. Briefly, sgRNAs for selected targets (TCCCGAAGCCTCCGCTGTCTAGG, AACCCGAGTGCCCTCGGCCCTGG and GCACACGGTTTCCTAGACGCTGG for Ripk1, Ripk3, and Mlkl, respectively, and all three for generating the Ripk1/Ripk3/Mlkl triple knockout were designed in silico using the CRISPR Design Tool (http://chopchop.cbu.uib.no/). Single-stranded sgRNA oligos were annealed and ligated into an expression plasmid bearing both the sgRNA scaffold backbone and Cas9 (pX330-U6-Chimeric_BB-CBh-hSpCas9, Addgene Plasmid no. 42230). The resulting plasmid (pSpCas9_sgRNA) was then co-transfected with the pcDNA3.1(+) vector (Invitrogen, Thermo Fisher Scientific, no. V790-20) containing a geneticin resistance gene into target cells. After single-cell cloning by serial dilution in 96-well plates (media was supplemented with 1 mg/ml G418), clones were assayed and selected for their target gene-knockout by western blotting analysis to identify stable CRISPR/Cas9-ko cell lines (data not shown). To exclude feasible off-target effects or clonal variations within the cell population we generated and analysed three different guide RNAs for each target gene(s) and observed congruent outcomes in each case (data not shown).

Restriction enzymes KpnI (R0142) and EcoRI (R0101) were obtained from New England BioLabs. Recombinant purified TNFα (570104) was purchased from BioLegend (Amsterdam, Netherlands). The pan-caspase inhibitor zVAD-fmk (4026865) was purchased from Bachem (Weil am Rhein, Germany) and Nec-1_s (2263-1-BV) was obtained from BioCat (Heidelberg, Germany). Isopropyl-d-thiogalactopyranoside (I6758) and primidone (P7295) were purchased from Sigma–Aldrich/Merck Millipore. Etanercept was obtained from Pfizer Pharma GmbH (Berlin, Germany). For fluorescence microscopy, we used anti-Penta-His antibody (34660, Qiagen, Hilden Germany) and TRITC-Donkey IgG anti-Mouse IgG (H + L) (715-025-150, Dianova, Hamburg, Germany). Slides were mounted using ImmunoSelect Antifading Mounting Medium with DAPI (SCR-038448, Dianova).

Phosphatidylserine exposure to the outer cell membrane of apoptotic cells or inner plasma membrane of necrotic cells and incorporation of 7-amino-actinomycin D (7-AAD) into necrotic cells was quantified by flow cytometry analysis. After stimulation of cells under the indicated conditions, staining was performed on single-cell suspensions using FITC Annexin V (BioLegend, 640906) and 7-AAD Viability Staining Solution (BioLegend, 420404) according to the manufacturer's instructions. Fluorescence was analysed using an FC-500 flow cytometer (Beckman Coulter GmbH, Krefeld, Germany).

L929 cells were seeded overnight on 8-well chamber slides at 2 × 10⁴ cells in 200 µl DMEM supplemented with 10% (v/v) FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin. Following incubation with 40 µM TAT–_M45RHIM, cells were fixed in 4.5% formalin, blocked with horse serum, permeabilised with Triton X-100, and stained for the His-Tag of TAT–_M45RHIM using an anti-Penta-His antibody and TRITC-Donkey IgG anti-Mouse IgG (H + L). Slides were mounted using ImmunoSelect Antifading Mounting Medium with DAPI. Imaging was performed using a Zeiss Axio Imager Z1 fluorescence microscope and AxioVision Rel. 4.8 software (Carl Zeiss GmbH, Jena, Germany). Figures were prepared using Fiji/ImageJ software. Greyscale images were assigned the respective pseudocolour, and channels were merged. Gamma correction (gamma-value = 0.9) was uniformly applied to prepare images for publication.

Necroptosis plays a critical role in many clinically relevant human diseases, hence its inhibition may offer promising therapeutic approach in many settings [30]. Blockade of host-specific necroptosis in infected cells by various RHIM orthologs is a conserved and highly effective mechanism employed by several herpesviruses [31]. The potential therapeutic use of a viral RHIM ortholog is, therefore, obvious. Initially, we believed that the M45 protein of MCMV was particularly promising in this regard because, in contrast with HSV-RHIM proteins, for example, it inhibits necroptosis equally in both human and murine systems [32]. However, to make M45 therapeutically useful via external delivery, a vehicle is needed to transport the protein into cells. Analyses of the native M45 protein, consisting of 1175 amino acids (aa), demonstrated that the N-terminal region of the protein, also containing the required RHIM domain, is sufficient for M45-mediated cell death suppression [16]. Thus, in order to increase the solubility of the recombinant protein, we cloned only the coding sequence of the N-terminal region (aa 1–277) of the viral protein into the TAT vector [33]. We first synthesised a recombinant TAT–M45 fusion protein consisting of 58 aa, including the truncated TAT domain of HIV-1 (11 aa) and aa 1–277 of M45 (Figure 1A). TAT–M45 was readily soluble and appeared slightly larger than 37 kDa on Coomassie Blue-stained gel (Figure 1B). Unexpectedly, starting at a concentration of 5 µM in all cell lines studied, the protein displayed pronounced cytotoxicity (Figure 1C), unlike any of the TAT fusion proteins we had synthesised and tested before [33,34]. Unfortunately, however, we failed to detect any inhibitory effect on necroptosis even at lower concentrations (Figure 1D).

Since we were initially unable to elucidate the reasons for the pronounced cytotoxicity observed for the TAT–M45 protein, we decided to generate a peptide that was as small as possible and contained only the TAT and M45 residues (Figure 2A). This TAT–_M45RHIM peptide (M_r= 4394 Da) enclosed only the N-terminal polyhistidine tag for detection via α-His-Tag antibody, the 11 aa of the TAT protein that are crucial for translocation across the plasma membrane [35], flanked by glycine spacers, and 14 aa of the M45-RHIM region, containing the conserved 4 aa RHIM core tetrad indispensable for interaction with other RHIM-containing proteins (Figure 2B). This peptide was again readily soluble, but, surprisingly, just like the TAT–M45 protein, it exhibited a time- and concentration-dependent cytotoxic effect at a concentration of ∼5 µM in all cell lines tested (Figure 2C,D). However, like the larger TAT–M45 protein, TAT–_M45RHIM peptide showed no inhibitory effect on necroptosis in human or murine cells at lower concentrations (data not shown).

Since we had never observed such cytotoxicity before for TAT proteins, we assumed that the specific M45-RHIM sequence might be responsible. We, therefore, synthesised TAT–RHIM peptides with a wide variety of different RHIM sequences. Valine (aa 58) in M45 located three positions upstream of the conserved RHIM core tetrad IQIG may be crucially involved in the interaction between RIPK3 and M45, and thus essential for the inhibitory properties of M45 in necroptosis [36]. If true, our original TAT–_M45RHIM peptide would have been too short, because the RHIM sequence we initially chose from M45 did not start until one amino acid after this valine. We, therefore, generated two TAT–RHIM variants including this valine. Unfortunately, neither differed from our original peptide (Figure 3A,B). Next, we generated an even shorter TAT–RHIM peptide without the His-Tag (denoted ΔHis-Tag). Yet again, this peptide was cytotoxic (Figure 3C) and also failed to inhibit necroptosis (data not shown). We then tried TAT–RHIM constructs with the RHIM sequences of murine RIPK1 and RIPK3 truncated and leaving out the initial spacer glycines that originally linked the His-Tag to the TAT peptide. Even with these, we could only detect cell death, but no inhibition of regulated necrosis (Figure 3D,E). ICP6 and ICP10 proteins from HSV-1 and HSV-2 are viral RHIM orthologs known to inhibit necroptosis via RHIM–RHIM interactions [32]. We originally refrained from using these RHIM sequences, since ICP6 and ICP10 prevent necroptosis only in human cells, whereas they do exactly the opposite in murine cells, which would complicate studies in vitro and would prevent their use in animal models [37]. However, given our previous findings, we wanted to investigate whether using these sequences would yield different results. As expected, the respective TAT–RHIM peptides killed murine cells like the previous ones (Figure 3F,G), but unexpectedly they also killed human cells and had no inhibitory properties on necroptosis in these cell lines (data not shown). Next, to investigate whether the combination of the TAT peptide and the RHIM domain was responsible for the cytotoxic effect, we synthesised a RHIM peptide containing only the M45-RHIM domain without the TAT moiety (Figure 3H). With a calculated molecular mass of only 1527 Da, we predicted that this peptide could be a freely diffusing biomolecule that would easily cross the cell membrane, but unfortunately, it was insoluble in water.

The inhibitory effect of M45 on necroptosis is dependent on its RHIM core tetrad IQIG. If this sequence is substituted to AAAA, M45 loses its protective effect [16]. Therefore, to investigate whether the cytotoxic effect of our peptide is also dependent on the RHIM core tetrad, we synthesised the AAAA (mutRHIM) variant of the TAT–_M45RHIM peptide. To our surprise, this peptide showed no cytotoxicity whatsoever in vitro even at concentrations up to 80 µM (Figure 3I). Based on this phenomenon, we assumed that the cytotoxicity of our peptides must be a form of RCD that involves known RHIM proteins and the necroptosis machinery.

Strikingly, the pan-caspase inhibitor zVAD-fmk, the RIPK1 inhibitors Nec-1_s and primidone, and the TNF antagonist etanercept [14,15] could not inhibit TAT–_M45RHIM-induced necrosis (Figure 4A). We then investigated whether knockout of the endogenous RHIM proteins and necroptosis effectors RIPK1 and RIPK3 alone or in combination with the terminal necroptosis effector MLKL would provide any resistance against TAT–_M45RHIM-induced cell death. Although the Ripk1, and Ripk3 single knockout and also the Ripk1/Ripk3/Mlkl triple knockout cell lines were maximally protected against TNFα-induced necroptosis, they resembled non-edited (parental) cells in terms of their reaction with TAT–_M45RHIM (Figure 4B). By comparing primary BMDMs isolated from C57BL/6 wt mice and Nlrp3-ko mice, respectively, we could also rule out any role of canonical pyroptosis [38] in TAT–_M45RHIM-induced cell death (Figure 4C).

On this basis, we assumed that TAT–_M45RHIM-induced necrosis is not a regulated process. We now believed that the peptide, with its combination of polycationic TAT moiety and hydrophobic RHIM domain predisposed to self-oligomerisation, could have a direct destructive effect on biomembranes. Similar effects have been previously described for other arginine-rich peptides [39]. By applying fluorescence microscopy, we observed that our added TAT–_M45RHIM peptide accumulates in cells after a short incubation period, and massive changes in the cell membrane and subsequent necrosis occur within 2 h (Figure 5A,B). Direct permeabilisation of phospholipid membranes by TAT–_M45RHIM was also reproducible in a biomimetic system; after addition of TAT–_M45RHIM a rapid permeabilisation of large unilammellar vesicles (LUVs) was detectable at concentrations ≥5 µM (Figure 5C). This corresponds to the concentration at which the cytotoxic effect of the peptide starts in cell culture.

In summary, our results indicate that recombinant proteins and peptides consisting of TAT and any form of RHIM sequence universally induce necrosis, unless the RHIM core tetrad is removed. This process is neither dependent on the known necroptosis machinery nor the NLRP3 inflammasome and we were not able to inhibit it by any means. The resulting loss of membrane integrity is instead caused by direct permeabilisation by TAT–RHIM derivatives.

Our initial goal was to establish a new approach for the therapeutic inhibition of necroptosis in human diseases. We drew inspiration for this from naturally occurring viruses; inflammatory cell death in viral infections usually leads to elimination of the virus from the host. To prevent this, some viruses have evolved sophisticated mechanisms, especially members of the Herpesviridae family. Cytomegaloviruses as well as HSV-1 and HSV-2 are known to encode multiple death inhibitors that are required for efficient viral replication and to prevent death of the host cell [40]. In practice, these viruses encode a viral inhibitor of caspase activation (vICA) to prevent death receptor (DR)-mediated apoptosis, but this may sensitise the cells to necroptosis. To escape this fate, the virus also encodes a viral inhibitor of RIP activation (vIRA) which disrupts the RHIM-dependent RIPK1–RIPK3 interaction that is critical for TNFα-induced necroptosis, as well as a RIPK3–RHIM-dependent step in virus-induced necrosis [41].

To this end, we aimed to generate a recombinant fusion protein that could penetrate cells efficiently and, analogous to viral RHIM proteins, inhibit necroptosis by preventing the critical RHIM–RHIM interactions during formation of the necrosome. Using the PTD TAT from HIV-1 as a vehicle, we successfully generated a variety of fusion proteins in the past that were cell-permeable and biologically active [33,34]. We predicted that MCMV M45 would be a promising viral RHIM candidate, because M45 is naturally produced during infection as a full-length protein that is cleaved after residue 277, and this truncated stretch (aa 1–277) is sufficient to inhibit necroptosis in both human and murine cells [32]. Although there is no restriction on the size or type of the delivered cargo, the capacity for successful uptake of cargoes into cells increases when the cationic PTD is attached to lower molecular mass compounds [23]. As shown in Figure 1B, we were able to purify a recombinantly expressed TAT–M45 fusion protein under native conditions, but the protein did not show the desired effect of inhibiting necroptosis (Figure 1D). On the contrary, it was instead cytotoxic at concentrations ≥5 µM (Figure 1C). Therefore, we changed our primary approach and generated a short synthetic TAT–RHIM peptide (TAT–_M45RHIM, Figure 2A), ruling out that the mere presence of the TAT protein or formation of unfolded protein aggregates induced the observed cell death. By screening a plethora of cancer cell lines and primary cells of different species, we surprisingly discovered that all tested cells were also directly killed by the TAT–_M45RHIM peptide in a time- and concentration-dependent manner (Figure 2C,D). The rapid onset of cell death suggests that no transcriptional processes are involved in the regulation of TAT–_M45RHIM sensitivity (at even higher concentrations of the peptide, most cells die within 2 h; data not shown).

RHIM-containing proteins most likely appeared early in the evolution of complex life and, while conserved among mammals, RHIM-like motifs are found in fungi and Drosophila [42,43]. In mammals including humans and mice there are exactly four RHIM-containing proteins and these are the previously mentioned RIPK1, RIPK3, DAI/ZBP1, and TRIF. In general, for RHIM–RHIM interactions, especially the RHIM core tetrad is of great importance. This feature, which as the name suggests consists of 4 aa residues, is highly conserved across mammals and their viruses [44]. We, therefore, speculate that maybe the mere presence of a RHIM motif could be responsible for the observed toxic effect of the TAT–RHIM protein/peptide. However, for the cytotoxic effect of TAT–RHIM, the origin of the RHIM sequence does not seem to play a role, since we tested almost all known viral RHIM sequences as well as ‘native’ RHIM from murine RIPK1 and murine RIPK3, and always observed the same toxic effect (Figure 3), with no inhibition of TNFα-induced necroptosis in any of the tested cell lines (data not shown). The only known viral RHIM sequence that we did not examine is that from varicella zoster virus (VZV). VZV also encodes a viral RHIM protein, namely open reading frame (ORF) 20. However, we decided not to generate a TAT–_ORF20RHIM because ORF20 is only thought to bind DAI/ZBP1 and thus prevent DAI/ZBP1-induced apoptosis in the course of VZV infection [45]. For necroptosis in general, and especially for TNFα-induced necroptosis, ORF20 should have no effect, in contrast with the other viral RHIM proteins we tested. Furthermore, human CMV (HCMV) can also block necroptosis, but not through a viral RHIM protein; rather, it acts in a way that is still not completely understood, downstream of activated (phosphorylated) MLKL [18], hence we could not generate a TAT–RHIM based on HCMV.

To exclude the possibility that the combination of the TAT and RHIM alone was responsible for the cytotoxic effect, we synthesised a RHIM peptide that encompassed only the short RHIM domain of M45 without the TAT moiety (Figure 3H), which we assumed should be taken up by cells by passive diffusion. Regrettably, this peptide was completely insoluble in water, preventing this experiment. Although the cytotoxicity of TAT–RHIM does not, according to our experiments performed so far, depend on the origin of the RHIM sequence, we nevertheless demonstrated that the RHIM core tetrad is indispensable for this effect. The mutated RHIM variant of the TAT–_M45RHIM peptide, in which the four critical amino acids of the RHIM core tetrad were replaced by AAAA, was not at all cytotoxic, even at much higher concentrations than all other TAT–RHIM variants tested (Figure 3I). Thus, it is clear that there is much left to learn about these viruses.

From the results shown in Figure 3I, we hypothesised that the cytotoxicity of all the other TAT–RHIM peptides tested previously might be a form of RCD based on interactions with endogenously expressed RHIM proteins and the necroptosis machinery. Again, we falsified this hypothesis, because neither common pharmacological inhibitors of necroptosis or other RCD processes, nor knockout of endogenously expressed RHIM proteins, had any protective effect on TAT–_M45RHIM-induced cell death (Figure 4A,B).

Next, we speculated that transduced RHIM peptides in increasing concentrations (i.e. supersaturation) may form intermolecular amyloid aggregates through self-assembly. If so, these homo-oligomers could associate with cellular membranes, directly or indirectly altering plasma membrane integrity, and inducing amyloid-induced lysis, as described previously [46]. This hypothesis would also be consistent with the discovery that functional RHIM amyloids in mammals are composed of hetero-oligomeric fibrils containing mixtures of dissimilar RHIM proteins. However, pathologically accumulated amyloids are predominantly constituted from homo-oligomeric fibrils [47,48]. Involvement of the NLRP3 inflammasome in this process, which is potentially activated by amyloids [49], was ruled out by our results showing in Figure 4C. In contrast, we demonstrated that TAT–_M45RHIM rapidly leads to loss of cell membrane integrity. In an in vitro set-up with LUVs, this disruptive effect on biomembranes was found to be independent of any types of (RHIM-containing) cellular proteins and pathways, but TAT–_M45RHIM could directly permeabilise biomembranes (Figure 5C).

In summary, our plan to design a specific, therapeutically applicable inhibitor of necroptosis failed unexpectedly. Instead, we created a substance that can efficiently destroy all cells that encounter it. Even the sturdiest tumour cell lines of various origins, including murine lymphoblast J558, human osteosarcoma U-2 OS, and human glioma U-87 MG, which have the greatest possible intrinsic resistance to RCD, were killed by TAT–_M45RHIM and its derivatives (data not shown) in a manner distinct from known forms of (regulated) necrosis. However, this discovery could lead to a new field of application for these peptides. Induction of immunogenic necrotic cell death in malignant tumours represents a potentially important development for oncologic therapies [50]. Resistance of tumour cells to different forms of RCD often results in tumour cells that are refractory to chemotherapeutic agents [51]. Remarkably, TAT_M45–RHIM-induced necrosis in all tested cell lines. Thus, it would conceivable that it could do so in malignant tumours, and our peptides are available for such studies.

All data and reagents are available from the authors upon request.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by Dr. Werner Jackstädt-Stiftung, Fresenius Medical Care, Germany and Else Kröner-Fresenius Stiftung (to S.K.). This work was additionally funded by the Medical Faculty of CAU Kiel, Germany (to B.K. and S.K.).

Stefan Krautwald: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing—original draft, project administration, writing—review and editing. Benedikt Kolbrink: Formal analysis, investigation, visualization, writing—original draft, writing—review and editing. Theresa Riebeling: Data curation, formal analysis, validation, investigation, visualization, methodology. Nikolas K. Teiwes: Formal analysis, investigation, visualization, methodology. Claudia Steinem: Formal analysis, validation. Hubert Kalbacher: Data curation, validation, investigation, methodology. Ulrich Kunzendorf: Supervision, validation.

We thank Maike Noetzel, Katja Bruch and Neele May Garling for excellent technical assistance. Furthermore, we thank the former lab members Federica De Zen, Christine Dewitz, and Caroline Moerke for their excellent help at the initial phase of the study.

7-AAD

7-amino-actinomycin D

BMDMs

bone marrow-derived macrophages

DR

death receptor

HCMV

human CMV

HIV-1

human immunodeficiency virus 1

HSV-1

herpes simplex virus 1

LUVs

large unilammellar vesicles

MCMV

murine cytomegalovirus

MLKL

mixed lineage kinase domain-like protein

ORF

open reading frame

PTD

protein transduction domain

RCD

regulated cell death

RHIM

RIP homotypic interaction motif

TAT

transactivator

TIR

Toll/interleukin-1 receptor

TNF

tumour necrosis factor

TNFR1

tumour necrosis factor receptor 1

vICA

viral inhibitor of caspase activation

VZV

varicella zoster virus

ZBP1

Z-DNA-binding protein 1