Generation of aptamers for the selective detection and neutralization of soluble lymphotoxin alpha
Matthew Stephens, Eman Nassef, Pierre-Yves von der Weid
TL;DR
Researchers developed DNA aptamers that specifically detect and neutralize a form of the inflammatory protein lymphotoxin alpha, offering a new tool for studying immune signaling.
Contribution
The paper introduces DNA aptamers that selectively target homotrimeric lymphotoxin alpha without affecting its heterotrimeric form.
Findings
Four aptamer candidates were generated that selectively detect LTα3 but not LTα1β2.
Optimized aptamers LTa1 and LTa5 significantly reduced LTα3-TNFR1 engagement-related cytotoxicity in vitro.
The aptamers show potential for future therapeutic use with minimal off-target effects.
Abstract
Lymphotoxin alpha (LTα) is a potent inflammatory cytokine implicated in the pathophysiology of numerous human autoimmune and inflammatory diseases. Existing as a soluble homotrimer (LTα3) or membrane-bound heterotrimer (LTα1β2), the differential and distinct functions of lymphotoxin signaling have meant that selective targeting of the cytokine with traditional pharmacological agents has proven difficult. While monoclonal antibodies that can neutralize human LTα3in vivo do exist (e.g., pateclizumab), their efficacy and subsequent use within the clinic have been limited. This may be in part perhaps due to cross-reactivity between the homotrimeric and heterotrimeric forms, leading to LTα3-independent effects. Herein, we implement a counter-Systematic Evolution of Ligands by Exponential Enrichment (SELEX) protocol to enrich aptamers targeting LTα3 but not LTα1β2. Through a combination of in…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsSynthesis and Biological Evaluation · Toxin Mechanisms and Immunotoxins · Nicotinic Acetylcholine Receptors Study
Introduction
Tumor necrosis factor alpha (TNF-α) is arguably the most well-known and investigated pro-inflammatory cytokine, which, since its discovery in 1986, has been highlighted as a key and targetable inflammatory mediator in a variety of acute and chronic inflammatory diseases.1^,^2^,^3^,^4 With the advent of anti-TNF-α monoclonal antibodies, e.g., infliximab, a positive explosion of immunosuppressive biologics has appeared and been implemented to great success, reducing disease severity in diseases such as rheumatoid arthritis,5 ankylosing spondylitis,6 as well as inflammatory bowel disease.7 Lymphotoxin alpha (LTα), formally TNF-β, is a TNF superfamily member that exists as both a soluble homotrimer and heterotrimer.8^,^9 In its soluble homotrimer form, LTα_3_ primarily interacts with TNF receptors TNFRI/TNFRII, triggering a P65/RELA mediated pro-inflammatory response in both immune and stromal cells.10 In its heterotrimeric form, LTα_1_β_2_ is transiently expressed on the cell surface of activated B and T cells, binding to its cognate receptor, TNFRIII/LTβR.11 The role of lymphotoxin, therefore, has been implicated in a variety of autoimmune disease models, including Sjögren’s disease,12 neurodegeneration,13 and many other T cell-mediated pathologies, including those driven by Th1 and Th17 cells.14 The efficacy of T cell depletion in mouse models through anti-LTα biologic therapy has been promising; however, despite the generation of highly potent anti-LTα monoclonal antibodies, several clinical concerns remain. Pateclizumab, for example, binds both soluble and membrane-bound LTα_3_ and thus not only diminishes the pro-inflammatory action of LTα_3_-TNFR1 signaling but also suppresses Th1-induced inflammation and depletes LTα_1_β_2_-expressing B cells.15 While initially attractive as a potent immunosuppressant, prolonged use drives structural alterations within lymphoid tissues, as B cell-mediated LTβR signaling is essential for their structural maintenance and immune surveillance properties.16^,^17
Aptamers are single-stranded oligonucleotides that, depending on their sequence and hydrophobicity, form complex tertiary structures in solution, binding with high specificity and selectivity toward their desired targets.18 Raised through an iterative process of refinement against their desired target in a process termed Systematic Evolution of Ligands by Exponential Enrichment (SELEX),19^,^20 aptamers possess many advantages compared to small-molecule pharmacological agents or larger antibody-based biologics, including low cost of production, reproducibility and stability, low toxicity and immunogenicity, as well as flexibility to chemical modification to refine function and tissue tropism.21^,^22^,^23 In targeting the TNF superfamily specifically, several studies have highlighted the ability for aptamers to neutralize soluble TNF-⍺,24^,^25 OX40,26 CD40L,27 RANK,28 or indeed the TNFR1 receptor itself.29 However, targeting LTα with aptamers remains unreported.
We therefore sought to generate ssDNA-based aptamers that could selectively bind LTα_3_ and, through combining counter-SELEX methodology and in silico molecular docking interactions, develop first-generation aptamers lacking cross-reactive binding toward the heterotrimeric form, LTα_1_β_2_. The generation of a selective LTα_3_ aptamer ultimately fills a void within the molecular toolbox to investigate the role of LTα signaling in multiple inflammatory diseases while limiting the off-target effects previously documented.
Results
Enrichment of ssDNA library toward LTα3 using counter-SELEX protocol
LTα plays multiple essential physiological roles during tissue homeostasis and disease. Binding of the lymphotoxin homotrimer to its cognate receptors, TNFR1 and TNFR2, drives opposing effects upon the cell, with TNFR1 ligation inducing inflammation and TNFR2 inducing pro-survival effects such as proliferation and anti-inflammatory cytokine production (Figure 1). As a heterotrimer with lymphotoxin beta, however, LT⍺ plays an essential role in lymphoid tissue homeostasis and thus presents potential clinical concerns through its unrestricted inhibition. Using a previously implemented nitrocellulose-based SELEX protocol, we sought to generate ssDNA aptamers targeting recombinant human LTα homotrimer (LT⍺3), performing counterselection against recombinant human LTα 1 beta 2 (LT⍺1_β_2) to limit cross-reactive sequences.22 In the first round of SELEX, recombinant human LT⍺ was diluted to a final concentration of 1 μM in phenol red-free Dulbecco’s modified Eagle medium (DMEM) containing 10% heat-inactivated fetal calf serum (FCS) and mixed with 100 pmol of ssDNA library containing a 40-nucleotide randomized region flanked by conserved primers. After incubation, samples were partitioned using a high-bind nitrocellulose membrane and washed thoroughly to remove low-affinity binders. Aptamers bound to their target protein immobilized on the nitrocellulose membrane were eluted by boiling the membrane in nuclease-free water, and the enriched sequences were recovered by PCR, with ssDNA recovered via lambda exonuclease digestion of the phosphorylated antisense strand. Sequence enrichment was monitored by agarose gel electrophoresis, and positive selection was halted once the recovered DNA template drove the formation of spurious products within 15 PCR cycles (>round 8). In total, 7 positive rounds of SELEX were performed, with counterselection to limit non-LTα_3_ binding aptamers performed after round 1, 3, and 6 in a similar manner, but the nonspecific target consisted of DMEM containing 10% FCS or 1 μM recombinant LT⍺1_β_2 (Schematic of SELEX, Figure 1).Figure 1. Selection of DNA aptamers that target lymphotoxin alphaSchematic diagram of the SELEX protocol, including “positive” SELEX targeting sequences that bind LT⍺3, combined with “negative” SELEX removing cross-reactive sequences that bind LT⍺1_β_2. Blunt-end cloning and bacterial cloning was followed by colony PCR of single aptamer candidates, which were analyzed by sanger sequencing. Illustration created using BioRender.
Validation and refinement of LT⍺ aptamer core sequences
The conserved flanking sequences (primers) contained within the original aptamer library are suspected to be functionally redundant, and it is desirable to produce them in their minimal sequence to save on production costs. However, it has been noted by others that removal of primers drastically alters aptamer binding capacity to their target, and we too noted early on in our testing that certain aptamer candidates synthesized without their flanking primers had minimal or no apparent binding capacity toward their target despite having survived the selection pressures, suggesting functional incorporation of these primer sequences into the aptamer itself.30^,^31 We therefore sought to test whether structural prediction tools, such as Vienna WebFold, could predict the minimum functional aptamer sequence to optimize four aptamers (LTa1, LTa5, LTa9, LTa12) and confirm the prediction through in vitro tests. Candidate sequences were synthesized with a 5′-biotin modification and tested for their specificity, selectivity, and sensitivity toward recombinant LTα_3,_ LT⍺1_β_2, or TNF-⍺ using a plate-based enzyme-linked oligonucleotide assay (ELONA). The rationale truncation involved either the removal of the 3′ primer or 5′ and 3′ flanking primers, resulting in sequences of 58 nucleotides (nt) or 40 nt in length (Figures 2A, 2D, 2G, and 2J). Binding curves were then calculated based on the ability of each of these truncation derivatives to bind immobilized recombinant LTα_3_ (Figures 2B, 2E, 2H, 2K), with cross- reactivity of each aptamer variant, as well as an aptamer scrambled sequence control, assessed against recombinant LT⍺1_β_2, or TNF-⍺ (Figures 2C, 2F, 2I, and 2L). Through rational sequence truncation of LTa1-76-nt (K_D_ = 46.91 nM), we found that LTa1-58 nt (18-nt truncation from the 3′ end) retained similar binding capacity (K_D_ = 64.3 nM) to the full-length control, but when further shortened to LTa1-40 nt (additional 18-nt truncation from the 5′ end of LTa1-58 nt), the sequence lost its binding capacity almost entirely (Figure 2B). In a similar fashion, aptamers LTa5-76 nt (K_D_ = 54.26 nM) and LTa12-76 nt (K_D_ = 50.97 nM) retained similar K_D_ values when truncated at the 3′ end (K_D_ = 58.52 and 24.42 nM, respectively) but showed a dramatic increase in K_D_ when truncated at the 5′ end (Figures 2E and 2K). LTa9-76 nt (K_D_ = 25.68 nM) demonstrated stable K_D_ values when truncated at the 3’ end (K_D_ = 18.32 nM) or at both 3′ and 5′ ends of the sequence (K_D_ = 78.3 nM, Figure 2H). Together, these data suggested a functional incorporation of the sequence or possible steric hindrance of biotin-streptavidin interactions by removal of flexible or disposable motifs within the sequence. To address the latter hypothesis, primer-truncated aptamers 1, 5, and 12 were generated with a poly-T (10 nt) tail at the 5′ end. The addition of this flexible linker restored the aptamers’ binding capacity in this plate-based binding assay, suggesting that despite their shortened length, 40-nt aptamers were still able to bind their target and that the loss in binding capacity was rather a feature of physical interference within the assay itself (Figures 2M–2O). Confirming the biological activity of the assay and the reagents used, we also calculated the binding capacity of pateclizumab to recombinant LTα_3_, LT⍺1_β_2, and TNF-⍺. These data demonstrated the specific affinity of pateclizumab for LTα_3_ (K_D_ = 0.38 nM) and LT⍺1_β_2 (K_D_ = 0.56 nM) without cross-reactivity toward TNF-⍺ (Figure 2P). Addition of the 5′ polyT to the primer-truncated aptamers highlighted that the flanking sequence was dispensable for target binding and that it was rather steric interactions of 5′-biotin and streptavidin-HRP (horseradish peroxidase) that required some flexibility of the sequence to avoid interference with binding in this plate-based assay, as previously reported. Therefore, the minimal 40-nt truncated aptamers were taken forward for further assessment.Figure 2. Characterization and binding optimization of aptamers through rationally guided sequence truncationPredicted secondary structures for full-length (76 nt), 3’-truncated (58 nt) and 5′- and 3′-truncated (40 nt) sequences of (A) LTa1, (D) LTa5, (G) LTa9 and (J) LTa12. Impact of truncation on their calculated dissociation constants to rhLTα_3_ (B, E, H, K) as well as the limited cross-reactivity of 1μM each corresponding candidate sequence to rhLTα_3_, rhLTα_1_β_2_ and rhTNFα (C, F, I, L). Rescuing effect of an 5’ poly-T (10 nt) addition on measured dissociation constants to rhLTα_3_ on candidates (M) LTa1 (N) LTa5 and (O) LTa12. (P) Representative binding curve showing the specific recognition of lymphotoxin homotrimer and heterotrimer by Pateclizumab without cross-reactivity toward TNFα. Data are represented as mean ± SD, with experiments performed twice. Dissociation constants were calculated via non-linear one-site total analysis.
Assessment of aptamer functionality toward ligand-receptor interference
We next sought to determine the ability, if any, of the generated aptamers to impede TNFR1-LT⍺ binding via a plate-based ligand-receptor interference assay. TNFR1-Fc chimera (R&D Cat no: 372-RI) was immobilized on an ELISA plate and probed with soluble LT⍺3 at increasing concentrations, which was then detected via a biotinylated anti-human LT⍺ antibody (Biolegend Cat No. 503104) and quantified through streptavidin-HRP signal development. Using this assay, we first demonstrated a dose-dependent interaction of LT⍺ with TNFR1 (Schematic - Figure 3A). Next, pateclizumab (1 μg/ml), serving as a positive control, strongly inhibited the binding interactions between LT⍺ and TNFR1, while a scrambled sequence aptamer (AptScram) had no significant impact on their interaction (Figure 3B). The concentration of pateclizumab used here was chosen based on previous literature reports to show maximal inhibition while sparing reagent use.14 Ultimately, incubation of LT⍺ with candidate aptamers (100 nM) for 1 h prior to incubation with the TNFR1-Fc chimera, significantly reduced LT⍺-TNFR1 interaction in two of the sequences tested (LTa1 and LTa5, Figures 3C and 3D). However, there was no significant difference in binding when LT⍺ was incubated with LTa9 or LTa12 (Figures 3E and 3F).Figure 3. Ligand-receptor interference assay(A) Schematic representation of the ligand-receptor interference assay in which TNFR1-Fc is adsorbed to a surface, and interaction with soluble rhLT⍺ is detected via an LT⍺-specific monoclonal antibody and a secondary antibody conjugated to HRP. Impact of (B) pateclizumab (1 μg/ml) or scrambled aptamer sequence (Scram Apt), and (C–F) 100 nM aptamer preincubation on receptor binding. Data are represented as mean ± SD of 3 independent assays, with control values duplicated between graphs. Statistical significance was determined via two-way ANOVA with Sidak’s multiple comparison post hoc tests. ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Molecular modeling of aptamer-LT⍺ interactions
Having validated the binding capacity of the aptamers to inhibit LTα_3_-TNFR1 interaction, we investigated the possible antagonizing mechanism via series of molecular docking simulations. First, we used in silico modeling to predict the secondary (Figure 4A) and tertiary (Figure 4B) structures of each individual candidate aptamer (Vienna WebFold and RNAcomposer, respectively), then performing docking simulations via PatchDock using the crystal structure of the human TNFR1 receptor-LT⍺3 complex (PDB: 1TNR, Figure 4C). For our in silico analysis, we first assessed and established the protein-protein interactions of LT⍺3-TNFR1 (PDB:1TNR) using PyMOL. According to the modeling analysis of 1TNR, we highlight that LT⍺ binds the groove formed between two LT⍺ subunits, contacting LT⍺ at residues R46, A47, N48, T49, D50, and S81, G82, K83 (Figures 4D and 4E). This intra-groove binding formed between the two subunits likely highlights the refined ability of the lymphotoxin homotrimer to interact specifically with TNFR1. These key residues (highlighted in Figures 4D and 4E, red for LT⍺ and orange for TNFR1), specifically Asp50, have been previously documented to be essential for correct engagement with the CRD2 and CRD3 loops of TNFR1, which form the core ligand-receptor interface.29^,^32 We then implemented predictive molecular docking generated using PatchDock to interrogate possible binding sites of the two aptamer sequences previously shown to limit LT⍺-TNFR1 interaction (Figure 4F). Modeling highlighted that both LTa1 and LTa5 were predicted in silico to interact with LT⍺ at residues N48, T49, D50 of R46-D50, as well as G82 of S81-K83 (sequences highlighted in bold, Figure 4E), providing some explanation as to why both aptamers could inhibit TNFR1-LT⍺ interactions. Supporting these data, aptamers LTa9 and LTa12, which previously demonstrated no ability to limit LTa-TNFR1 interaction (Figures 3E and 3F), were modeled to bind LT⍺ at sites unrelated to ligand-receptor interaction, with LTa9-lymphotoxin interactions predicted at residues N40, S41, L42, D151, G152, I153, and LTa12-lymphotoxin interactions at residues L124, Q125, H130, S131, M132 (Figure 4F).Figure 4. Predicted of interaction between LT⍺ and aptamersSchematic illustration of in silico structural modeling using Vienna Web fold and RNAcomposer for (A) 2D and (B) 3D aptamer structures, followed by predicted molecular modeling using (C) PatchDock and PyMOL. (D) Highlighted interaction residues between LT⍺ (red) and TNFR1 (orange). (E) Amino acid (aa) residues highlighted in red and orange on the model are shown in matching colors on the aa sequence of each protein, with predicted aptamer interaction sites highlighted in bold. (F) Predicated modeling interactions of TNFR1-LT⍺, LTa1 (40 nt), LTa5 (40 nt), LTa9 (40 nt), and LTa12 (40 nt) with 1TNR.
Impact of aptamers on LTα-induced transcriptional responses
Having determined the ability of two candidate aptamers to selectively bind LTα and further inhibit receptor binding, we sought to determine whether they were effective in blocking LTα-TNFR1 signaling in vitro. To assess the biological impact of LTα_3_ signaling in vitro, L929 murine fibroblasts were first stimulated with rhLT⍺3 or rhLT⍺1_β_2 for 1, 4, and 24 h to establish a time course of early (LT⍺3-dependent) and late (LT⍺1_β_2-dependent) transcriptional responses (left, Figures 5B–5J). A schematic illustration of the shared TNFR1-dependent signaling pathway for LT⍺3 and TNF-⍺, as well as the unrelated LT⍺1_β_2-LTβR signaling cascades, is depicted in Figure 5A. The TNFR2-dependent response was not considered here, as human TNF-⍺ and LT⍺ are specific agonists of murine TNFR1 and do not activate murine TNFR2.33 Stimulation of L929 cells with rhLT⍺3 1 ng/ml) induced the transient, robust expression of Cxcl1 (Figure 5B), Ccl7 (Figure 5C), and Ccl2 (Figure 5D) at 1 h, which had significantly subsided by 4 and 24 h. Expectedly, stimulation of the cells with rhLT⍺1_β_2 (10 ng/ml) drove the induction of Cxcl13 (Figure 5E), Ccl21 (Figure 5F), and Ccl19 Figure 5G), which was significantly elevated only at 24 h, highlighting the late response dynamic (Figure 5C). Due to the same receptor dependency (TNFR1), rhTNF-⍺ (1 ng/ml) was used as a control. In perhaps unsurprisingly similar pattern to LT⍺3, TNF-⍺ induced an early, robust response of Cxcl1 (Figure 5H), Ccl7 (Figure 5I), and Ccl2 (Figure 5J), which significantly subsided by 4 and 24 h. Using this time course, each cytokine was then incubated with pateclizumab (1 μg/ml),14^,^15 lymphotoxin aptamer candidates (LTa1, LTa5, LTa9, or LTa12 at 100 nM), or scrambled sequence aptamer control (AptScram) for 1 h prior to addition to the cells (right, Figures 5B–5J). LT⍺3, TNF-⍺-, or LT⍺1_β_2-induced transcripts were then quantified via quantitative reverse-transcription PCR (RT-qPCR) at 1 or 24 h post-stimulation, respectively, based on peak cytokine response determined within the time course. We observed that LTa1 and LTa5 were able to significantly reduce LT⍺3-specific induction of Cxcl1, Ccl7, and Ccl2 mRNA levels (Figures 5B–5D) without cross-reactivity to either LT⍺1_β_2 (Figures 5E–5G) or TNF-⍺ induced transcripts (Figures 5H–5J). Incubation of each cytokine with pateclizmab resulted in a significant and selective reduction in LT⍺3- and LT⍺1_β_2-dependent transcripts, with TNF responses spared (Figures 5B–5J). Overall, these data highlighted the selective and potent limitation of LT⍺3-TNFR1 interaction by aptamer candidates LTa1 and LTa5.Figure 5. Functional ability of candidate aptamers to inhibit cellular transcriptional response to LT**⍺**(A) Illustration of signaling pathways induced by LT⍺/TNF⍺-TNFR1. Left: Engagement induces TNFR1 trimerization and recruitment of the adaptor protein TRADD, along with TRAF2, RIPK1, and cellular inhibitors of apoptosis (cIAP1/2), forming the membrane-associated signaling complex I. This complex activates TAK1, which in turn phosphorylates the IKK complex (IKK⍺, IKKβ, and IKKƔ/NEMO). Activated IKK phosphorylates IκB⍺, targeting it for degradation and releasing the canonical NF-κB heterodimer RelA (p65)/p50. RelA/p50 translocate to the nucleus to drive expression of acute inflammatory and survival-associated genes, exemplified here by Cxcl1, Ccl2, and Ccl7. Right: Signaling initiated by binding of LTβR by LT⍺1_β_2 recruits TRAF2, TRAF3, and cIAP1/2, forming a regulatory complex that controls the stability of NF-κB-inducing kinase (NIK). Ligand engagement leads to TRAF3 degradation and consequent NIK accumulation, enabling activation of an IKK⍺ homodimer independent of IKKβ and NEMO. Activated IKK⍺ phosphorylates p100, promoting its proteolytic processing into p52. The resulting RelB/p52 heterodimer then translocates to the nucleus, driving a distinct transcriptional program associated with lymphoid tissue organization and stromal cell activation, including Cxcl13, Ccl19, and Ccl21, and the transcription of genes involved in lymphoid tissue organization, stromal cell differentiation, and chemokine expression. (B–J) L929 murine fibroblasts were stimulated with LT⍺3 (1 ng/ml) alone or after 1 h preincubation with pateclizumab (PzMAb, 1 μg/ml), LT⍺-specific aptamers (LTa1, LTa5, LTa9, LTa12 at 100 nM), or a scrambled aptamer control (AptScram 100 nM). Temporal induction of cytokine and chemokine transcripts were determined to identify peak induction, while bar graphs show data from a single measured time point (indicated as 1 or 24 h) and the impact of specific inhibitors. Both panels display gene expression changes as mean ± SD ΔΔCt normalized to housekeeping controls. (B–D) (left) Time course and (right) 1-h qPCR analysis of (B) Cxcl1, (C) Ccl7, and (D) Ccl2 expression following LT⍺3 stimulation. (E–G) Expression of non-canonical NF-κB-associated lymphoid chemokine transcripts: (E) Cxcl12, (F) Ccl21, and (G) Ccl19 following LT⍺1_β_2 stimulation. (Left) Temporal expression kinetics and (right) alterations in 24-h peak cytokine transcript expression levels under the indicated inhibitory conditions. (H and I) (left) Time course and (right) 1-h endpoint expression of (H) Cxcl1, (I) Ccl7, and (J) Ccl2 following TNF-⍺ stimulation in the presence of indicated inhibitors. Statistical significance was determined using two-way ANOVA with Tukey’s multiple-comparison post hoc test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Impact of aptamers on lymphotoxin-induced cytotoxicity
Ligation of TNFR1 by LT⍺ induces the activation of several downstream responses, including caspase-8-induced cell death, in a similar manner to that of TNF-⍺.34 With the success of candidate aptamers LTa1 and LTa5 in reducing LT⍺-induced transcript induction, we sought to determine whether there were any cytoprotective effects. Therefore, in the presence of the transcription blocker actinomycin D, a cytotoxic dose-response curve was first generated to ascertain the cytolytic response of L929 cells to rhLT⍺3.33 Using an MTT cell viability assay, the LD50 of rhLT⍺3 was determined to be 1.171 pg/ml (Figure 6A). Using these data, we then incubated candidate aptamer sequences at fixed concentrations (100 nM) or pateclizumab (1 μg/ml) with 1 pg/ml rhLT⍺3 for 1 h at 37°C prior to stimulation of cells. The data demonstrated that incubation of rhLT⍺3 with pateclizumab rescued cellular cytotoxicity, while the scrambled aptamer had no impact (Figure 6B). Meanwhile, aptamer LTa1 (Figure 6C) and LTa5 (Figure 6D) provided a significant protective effect on rhLT⍺3-induced cell death over the 24 h incubation period. Supporting the ligand-receptor interference assay (Figure 3), in silico models (Figure 4), and transcriptional responses (Figure 5), aptamer candidates LTa9 and LTa12 demonstrated no significant impact on LT⍺-induced cytotoxicity (Figures 6E and 6F).Figure 6. Protective effect of candidate aptamer sequences on LT⍺-induced cytotoxicityStimulation of L929 cells with LT⍺ in the presence of actinomycin-D (1 μg/ml) drives apoptosis by blocking the transcription-dependent survival arm of the TNF signaling pathway(A) Calculated LD50 of actinomycin-D treated L929 fibroblasts in response to a dose-response curve of recombinant human LT⍺. Cell viability was determined using an MTT assay, with data expressed as % viability relative to unstimulated cells (control). All other cells were treated with 1 ng/ml rhLT⍺ +/− preincubation for 1 h with pateclizumab (PzMAb, 1 μg/ml) or indicated concentrations of (B) scrambled aptamer, (C) LTa1, (D) LTa5, (E) LTa9, or (F) LTa12 or control. Data are presented as mean ± SD of values from 3 independent experiments. Statistical analysis was performed using one-way ANOVA with Tukey’s post hoc test. ∗∗p < 0.01, ∗∗∗∗p < 0.0001.
Discussion
Despite the success of TNF-⍺-targeting biologics, targeting other TNF superfamily members such as LT⍺ remains limited, with no current clinically approved inhibitors.35 This is perhaps in part due to the dominate and well-described impact of TNF-⍺ in inflammatory pathology, with many other members having specialized functions, carrying unnecessary risks without strong therapeutic gain. However, due to their nuanced function, specific targeting of superfamily members such as LT remains an unmet need, with aptamers targeting others mediators of inflammation being proposed for their ability to fill such a niche.21^,^22^,^29^,^36 Here, using a strategic counter-SELEX methodology, we have developed two ssDNA aptamers (LTa1 and LTa5) that demonstrate selective binding of the soluble homotrimeric form of LTα. Our rationalized truncation assays, while refining the functional sequence, also support previous work describing that removal of flexible, seemingly redundant sequences in aptamers can depress observed aptamer affinity through multiple steric constraints. This is exemplified in solid-state assays such as the plate-based assays used herein, where surface crowding and tight intermolecular interactions can block biotin access or limit conformational transition of either the target protein, aptamer, or streptavidin detection probe.37^,^38^,^39 Our in vitro ligand-binding and cytotoxicity assays demonstrate the ability of both LTa1 and LTa5 aptamers to limit LT⍺-TNFR1 interaction, which, through in silico docking analyses, are modeled to interact with key residues involved in ligand-receptor binding.29^,^32 While supportive, due to the complexity of DNA-protein interactions, the results from that analysis are purely speculative.40 Finally, despite their ability to reduce cytotoxicity by 26% (LTa1) or 49% (LTa5), their relative success is overshadowed by that of pateclizumab (concentration 6.67 nM), which reduced cytotoxicity >60%, highlighting the potent effect of the monoclonal antibody (Figure 5D). While further optimization of LT⍺-binding aptamers remain to be performed for in vivo tests, we suggest the difference between the aptamer and antibody inhibitors may be in part due to antibody-induced endocytosis and protein recycling within the system via proteasomal degradation pathways, a mechanism that aptamer interactions have not been previously demonstrated to induce.41 Although ssDNA aptamers offer several advantages over their protein-based antibody comparators, their biochemical and biological stability is inherently lower due to their small size and enzymatic susceptibility. Unmodified ssDNA is rapidly degraded by serum and cellular nucleases and can have its secondary and tertiary structures altered by environment-specific pH, which can result in a significant reduction in binding capacity.42 These factors, combined with rapid clearance through renal filtration in vivo, ultimately reduce the fraction of aptamers available to bind their target, especially within cellular or organismal systems. Fortunately, ssDNA aptamers are prized for their receptiveness to chemical modification, which can improve bioavailability and target interaction dynamics. Optimization of sequences through the inclusion of chemical modifications, including locked nucleic acids, phosphorothioate linkages, or hydrophobic base analogues, has been demonstrated to markedly extend aptamer half-life and improve target engagement.21^,^22 The aptamer sequences used in this study were left unmodified and therefore likely have reduced structural stability and availability under the in vitro assay conditions documented. Inclusion of such stabilizing chemical modifications in future iterations of these aptamers would be expected to improve their biochemical robustness and bioavailability, which may reduce the perceived performance gap between aptamers and antibody counterparts. As pateclizumab is a complete bivalent antibody, comparison to unmodified monoclonal aptamers may be unfair. As lymphotoxin exists as a multimer, it may be more easily targeted and inhibited by the antibody, possessing a higher avidity. Future work utilizing a single-chain or FAb fragment variant of pateclizumab, or a bivalent aptamer, may provide a more apples-to-apples comparison, enhancing the performance of these and more candidate sequences. Equally, synergy between aptamer sequences is completely ignored in this project, whereby mutualism of a polyclonal aptamer pool may enhance ligand binding and inhibition. Ultimately, future studies should seek to prioritize the in vivo testing of these and other TNF superfamily-targeted aptamers for the continued investigation of these interesting yet underappreciated facets of immunotherapeutics.
Using a counter-SELEX methodology, we have identified and characterized aptamers that can selectively target LTα_3_ with limited cross-reactivity to LTα_1_β_2_. Through in silico modeling, we identify predicted aptamer-protein interactions that limit receptor ligation, supporting corroborating in vitro ligand-receptor interaction assays. Finally, we highlight the generation and characterization of LTa1 and LTa5 as aptamer sequences that, in vitro, demonstrate a significant ability to limit LT⍺3-induced transcriptional responses and cytotoxicity. The data provide a novel immunotherapeutic opportunity for the selective investigation of LTα-TNFR signaling, and through their use, may provide a better understanding of the role of LTα in homeostasis and disease and require future validation in preclinical animal models.
Materials and methods
Chemicals and reagents
The 76-nt ssDNA library GCGGATGAAGACTGGTGT-n40-GCCCTAAATACGAGCAAC, as well as individual aptamer clones used in this manuscript, were ordered from Integrated DNA Technologies Inc. The sequences (5′–3′) were as follows: LTa1-76, GCGGATGAAGACTGGTGTAATCTTTTGTTGTTTCCTATAACGTTCTCTCTTCATCGTGGCCCTAAATACGAGCAAC; LTa1-58, GCGGATGAAGACTGGTGTAATCTTTTGTTGTTTCCTATAACGTTCTCTCTTCATCGTG; LTa1-40, AATCTTTTGTTGTTTCCTATAACGTTCTCTCTTCATCGTG; LTa5-76, GCGGATGAAGACTGGTGTGAGTTAATTGTTTCGAAAGCTTGTATGACACCATTACTATGGCCCTAAATACGAGCAAC; LTa5-58, GCGGATGAAGACTGGTGTGAGTTAATTGTTTCGAAAGCTTGTATGACACCATTACTATG; LTa5-40, GAGTTAATTGTTTCGAAAGCTTGTATGACACCATCATG; LTa9-76, GCGGATGAAGACTGGTGTAAGACTGTAGGGAGGCAGAGGCTTCTTAATCTAGACTGGTGCCCTAAATACGAGCAAC; LTa9-58, GCGGATGAAGACTGGTGTAAGACTGTAGGGAGGCAGAGGCTTCTTAATCTAGACTGGT; LTa9-40, AAGACTGTAGGGAGGCAGAGGCTTCTTAATCTAGACTGGT; LTa12-76, GCGGATGAAGACTGGTGTCCCGTTTCCTTTCTAAGCTATGACATTGTAACACTTTTTAGCCCTAAATACGAGCAAC; LTa12-58, GCGGATGAAGACTGGTGTCCCGTTTCCTTTCTAAGCTATGACATTGTAACACTTTTTA; LTa12-40, CCCGTTTCCTTTCTAAGCTATGACATTGTAACACTTTTTA. A scrambled control aptamer (Apt-Scram) sequence was GCCGGAATATGTCGCTTTACGGGTCCTGGGCCGGGGTGCG.
SELEX
Aptamers were developed as previously described.22 For the first round of SELEX, the initial naive pool of sequences (100 pmol) was dissolved and denatured at 95°C in SELEX buffer (20 mM TrisHCl, 150 mM NaCl, and 1 mM MgCl_2_) for 10 min and allowed to cool to room temperature (RT) for 15 min to allow folding. In positive selection rounds, aptamers were added to a low-bind Eppendorf tube containing a 100 μL of DMEM (phenol red-free, supplemented with 1% v/v FCS) (Millipore Sigma D6434) and 1 μM human recombinant LTα (R&D Systems 211-TBB/CF). Aptamer-target solutions were incubated at 37°C under mild agitation (300 rpm, orbital shaker) for 1 h to allow for binding. After incubation, aptamer-LTα complexes were adsorbed onto a 0.5 × 2 cm nitrocellulose membrane (400 ng/cm^2^ binding capacity, 0.4 μm pore size) to separate protein-aptamer complexes from the unbound aptamer fraction. The membrane was washed three times for 5 min per wash in 1 mL of SELEX buffer to remove low-affinity binders. Retained sequences were eluted from the membrane by heating to 95°C in 200 μL nuclease-free dH_2_O for 10 min. Recovered aptamers were subsequently amplified using a Taq polymerase PCR mastermix (Promega M7502) with specific primers: forward, 5′-GCGGATGAAGACTGGTGT-3′; reverse, 5′-/5Phos/GTTGCTCGTATTTAGGGC-3′. The amplification protocol consisted of an initial denaturation step at 98°C, followed by 15 cycles of 95°C for 30 s, 55°C–59°C for 15 s, 72°C for 30 s, and a final extension 72°C for 5 min. Amplification was confirmed by agarose gel electrophoresis. ssDNA was recovered for further rounds of SELEX via lambda exonuclease treatment. Enrichment of the recovered sequences were assessed via gel recovery and band intensity and halted once template input caused sporadic products identified as a smear within the 15 cycles of amplification. This ultimately resulted in 7 rounds of positive selection and 3 rounds of counter/negative selection.
ssDNA recovery
Briefly, in a total reaction volume of 50 μL, 5 μg of dsDNA was incubated with 5 μL lambda exonuclease reaction buffer (10×) and 1 μL (5 units) of Lambda Exonuclease for 30 min at 37°C. The reaction was halted by heat inactivation at 95°C for 10 min in 10× v/v SELEX buffer. ssDNA was recovered using the Oligonucleotide Cleanup protocol with the Monarch PCR & DNA Cleanup Kit, as per the manufacturers guidelines (New England Biolabs #T130). To improve the affinity of selected aptamers, the incubation time was gradually shortened but 10 min per round and the number of washes was increased by 1 additional wash per round. After 3 positive selection rounds, aptamers underwent a negative selection round to improve the specificity of the final pool (Figure 1). In the negative selection round, aptamers were incubated with DMEM + 10% FCS or in the presence of 1 μM rhLTα_1_β_2_ (R&D Systems 8884-LY-025/CF). Aptamer-protein complexes recovered onto the nitrocellulose membrane were discarded, while the unbound fraction was retained and amplified. A final positive selection against LTα immediately followed to confirm successful recovery of LT⍺-specific sequences. Educated aptamer pools were then sequenced.
Bacterial cloning and aptamer sequencing
To identify individual aptamer sequences, the aptamer pool was separated using bacterial cloning, as previously described.22 Simply, PCR amplicons from round 7 were ligated into the PJET1.2 blunt-end plasmid cloning kit (Thermo Fisher Scientific K1231), as per the manufacturers guidelines, before plasmids were transfected via heat shock into DH10B competent cells (Thermo Fisher Scientific EC0113). Transformed cells were subcultured for 1 h in SOC media prior to being spread on agar plates containing 100 μg/ml ampicillin, alongside positive and negative controls, and incubated at 37°C overnight. Resultant colonies were picked and underwent colony PCR using specific primers: forward 5′-CGACTCACTATAGGGAGAGCGGC-3′, reverse 5′-AAGAACATCGATTTTCCATGGCAG-3′, to confirm aptamer insertion within the PJET1.2 plasmid vector. Vector amplicons were then purified using a PCR cleanup kit (Monarch T1030S) to remove extraneous PCR reagents before being sequenced at the University of Calgary’s Center for Genomics and Informatics using the PJET1.2 forward sequencing primer.
Dot blot assay
Dot blots were performed to rapidly screen the ability of selected aptamer clones to bind LT⍺. Recombinant LT⍺ or control protein (LT⍺1_β_2) was diluted to 0.1 μM in PBS, and a 0.5 μL droplet was spotted onto a nitrocellulose membrane. After allowing the protein to dry onto the membrane, the membrane was blocked with 2% BSA in PBST (PBS +0.1% Tween 20) at RT for 1 h, followed by incubation with biotinylated aptamers (100 nM) diluted in 2% BSA PBST for 30 min. Membranes were then washed 3 × 10 min with PBST before being incubated with streptavidin-HRP (1 μg/ml) in 2% BSA PBST for 30 min. Membranes were washed again in PBST before being visualized with enhanced chemiluminescence (ECL) substrate on a Bio-Rad ChemiDoc.
Aptamer truncation binding assay
Candidate sequences were synthesized with 5′ biotinylation during synthesis by Integrated DNA Technologies in their full length (including both primer binding regions) or lacking one or both of their flanking primer binding sites. For binding kinetics determination, rhLTa_3_ was immobilized on Immulon 4HBX ELISA plates at a concentration of 100 nM in PBS. After overnight incubation in the fridge, the residual buffer was removed and replaced with 2% BSA in PBST (0.1% Tween 20) for 1 h at RT on a plate rocker. Plates were then incubated with truncated biotinylated aptamers (1000, 300, 100, 30, 10, 3, 1, 0 nM) for 1 h at RT. Plates were then washed with PBST 3× before being incubated with 1 μg/ml streptavidin-HRP in 2% BSA PBST for 30 min at RT on a rocker protected from light. Wells were washed again 3× with PBST before being incubated with TMB substrate (Biolegend Cat No. 421101) for 30 min in the dark. Th reaction was halted by addition of 0.1 M HCL, and 450 nm absorbance was read on a SpectraMaxi3x plate reader.
Molecular docking
To explore the predicted interactions between aptamers and LT⍺, the tertiary structure of the selected aptamers was predicted as described, and complex models of 1TNR and Apt1, Apt5, Apt9, and Apt12 were obtained through molecular docking simulations. Based on the nucleic acid sequence of each 40-nt aptamer, the three-dimensional (3D) structure was constructed as follows. First, the secondary structures were generated in the Vienna web fold server using the following parameters: DNA parameters (Matthews model, 2004), temperature 4°C, salt concentration 0.15 M, with all other setting remaining as default. The lowest-energy secondary structure, including the dot bracket format, was used as the input to predict the tertiary structure. Next, the tertiary RNA structure was constructed using the RNAComposer web server, after which the model was converted to a 3D DNA model using the ModeRNA web server. For molecular docking predictions, a 3D render of the LTα monomer was created by modifying the published crystal structure of the soluble human TNF receptor-TNF-β complex (1TNR) in PyMOL (Figure 6). The generated aptamer 3D renders were then used to predict interactions with various LTα-incorporating proteins using the PatchDock server with standard parameters. Chain-chain interface residues were identified in PyMOL 3.1.3.1 using a custom Python script. In brief, all unique chains in the structural model were enumerated, and every possible chain pair was analyzed. For each pair, residues with at least one atom located with 5.0 Å of any atom in the partner chain were defined as interface residues. These residues were visually highlighted in PyMOL using stick representations, with residues from the first chain colored red and those from the second chain colored blue. The unique residue identifiers for each interface were extracted, sorted by chain and residue number, and written to a text file for reference. This approach provided a visual depiction of predicted intermolecular contacts.
TNFR1 binding interference assay
This protocol was adapted from methodologies previously described.43 Briefly, 96-well ELISA plates (Greiner Cat No. 07000100) were coated with TNFR1-Fc fusion protein (R&D Cat No. 372-RI-050) at 1μg/ml in PBS and incubated overnight at 4°C. Extraneous binding sites were blocked with 5% BSA diluted in PBS at RT for 2 h on an orbital shaker. Specified concentrations of rhLT⍺ (1–1000 ng/ml) (R&D Systems 211-TBB/CF) were then incubated in a final volume of 100 μL of SELEX buffer for 1 h with 100 nM of each candidate aptamer sequence, after which the entire mixture was added to the TNFR1-Fc-coated plates for an additional 2 h at RT on an orbital shaker. Plates were washed gently three times with 200 μL PBST per well (0.1% Tween 20). LT⍺ bound to TNFR1-Fc was detected using mouse monoclonal antibodies to LT⍺ (Biolegend Cat No. 503104) and HRP-conjugated goat anti-mouse IgG antibodies (Jackson Immuno Cat No. 115-035-003 diluted 1:5000) for 30 min at RT. Plates were again washed three times with PBST (0.1% Tween 20), followed by incubation with TMB substrate solution (Biolegend Cat No. 421101) for 30 min in the dark. The reaction was halted by the addition of 0.1 M HCL, and absorbance was read at 450 nm on a SpectraMaxi3x plate reader.
qPCR of L929 murine fibroblast response to cytokine stimulation
All cells were grown at 37°C in 5% CO_2_ in complete DMEM (+10% heat-inactivated FBS +1% pen/strep). Prior to stimulation, L929 cells were grown to 80% confluence in T75 flasks, enzymatically dissociated with 0.25% trypsin-EDTA, and then seeded at a density of 30,000 cells/cm^2^ in a 12-well plate. The following day, recombinant cytokines (LT⍺ [1 ng/ml], LT⍺1_β_2 [10 ng/ml], or TNF-⍺ [1 ng/ml]) were incubated with pateclizumab (1 μg/ml), folded aptamer candidates (100 nM), or control scrambled sequence (AptScram 100 nM) for 1 h in SELEX buffer at 37°C. Cytokine controls (i.e., those without inhibitors) were incubated under the same conditions in SELEX buffer. Just prior to stimulation, spent media containing dead cells and debris was removed, and cells gently washed with warmed 1× PBS. Cytokine stimulations, with or without inhibitors, were applied for 1, 4, or 24 h and harvested by removing the media and immediately lysing the cells in Qiagen RLT buffer, then stored at −20°C until processing. For mRNA preparation, the QIAGEN’s RNeasy Mini Kit was used according to the manufacturer’s protocol (Qiagen, Cat No. 74104). mRNA was converted to cDNA using iSCRIPT (BioRad Cat No 1708841) and amplified using PowerUp SYBR Green Master Mix, with 1 ng/μL used as template for each reaction. qPCR data were analyzed using the comparative Ct (ΔΔCt) Livak method.44 Primers used herein were purchased from Origene unless otherwise stated: mCxcl1: For 5′-TCCAGAGCTTGAAGGTGTTGCC-3′, Rev 5′- AACCAAGGGAGCTTCAGGGTCA-3’. mCcl7: For 5′-CAGAAGGATCACCAGTAGTCGG-3′, Rev 5′-ATAGCCTCCTCGACCCACTTCT-3’. mCcl2: For 5′- GCTACAAGAGGATCACCAGCAG-3′, Rev 5′- GTCTGGACCCATTCCTTCTTGG-3’. mCxcl13: For 5′- CATAGATCGGATTCAAGTTACGCC-3′, Rev 5′- GTAACCATTTGGCACGAGGATTC-3’. mCcl21: For 5′-GGTTCTGGCCTTTGGCATC-3′, Rev 5′-AGGCAACAGTCCTGAGCCC-3’.45 mCcl19: For 5′-TCGTGAAAGCCTTCCGCTACCT-3′, Rev 5′-CAGTCTTCGGATGATGCGATCC-3’. mGapdh: For 5′- CATCACTGCCACCCAGAAGACTG-3′, Rev 5′-ATGCCAGTGAGCTTCCCGTTCAG-3’.
Inhibition of LT⍺-induced cytotoxicity
An LD50 of the recombinant lymphotoxin was first determined. Briefly, the mouse fibroblast cell line L929 was seeded in a 96-well plate at 10,000 cells/well and allowed to recover for 24 h. The next day, the media was replaced with fresh complete media containing 1 μg/ml actinomycin-D (Sigma-Aldrich, A9415). Using a serial dilution starting at 100 ng/ml of LT⍺, cells were stimulated for 24 h, after which the media was removed and replaced with 5 mg/ml MTT solution for 1 h. Formazan crystal formation was monitored during this time, and after 1 h the media was removed and crystals dissolved in 100 μL DMSO. Optical density 570 nm (OD_570_) readings were used to determine cell viability, and values were normalized to the untreated replicate well values. For aptamer inhibition assays, each LT⍺-aptamer candidate was diluted in DMEM +10% FCS +1% pen/strep (stock concentration 1000 u/ml) to final concentrations of 1000, 800, 600, 300, 150, 100, 50, and 25 nM. Pateclizumab (1 ug/ml) served as a positive inhibitory control. Aptamers and/or pateclizumab were incubated with LT⍺ for 1 h at 37°C prior to cell stimulation, and cell viability was determined as done previously for LD50 determination.
Statistical analyses
Data are expressed as the mean ± standard deviation (SD). Experiments were performed with a minimum of three biological replicates, with technical triplicates included within each experiment. Multiple analyses were performed using one-way ANOVA with Sidak’s post hoc test where indicated. Non-linear analyses for K_D_ determinations were performed using a one-site non-linear regression in GraphPad Prism 10. A p value of 0.05 was considered significant: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001.
Data and code availability
All data are represented within this paper. Original datasets can be obtained from the corresponding author upon reasonable request.
Acknowledgments
This study was supported by the Lymphedema Research and Education Program, Snyder Institute for Chronic Diseases, 10.13039/100012866Cumming School of Medicine, University of Calgary, to P.-Y.v.d.W. This work was also partly supported by a 10.13039/100008459University of Calgary 10.13039/100023789VPR 10.13039/100027448Catalyst grant awarded to P.-Y.v.d.W. and M.S. The authors would like to thank other members of the lab for their input during project development, including mentorship support of E.N., specifically Shan Liao, Keith Keane, Peter Brothers, and Simon Roizes. The authors also thank the staff of the core facilities of the Snyder Institute for Chronic Diseases.
Author contributions
Conceptualization, M.S. and E.N.; methodology, M.S. and E.N.; validation, M.S. and E.N.; investigation, M.S. and E.N.; writing – original draft, M.S.; writing – review & editing, M.S., E.N., and P.-Y.v.d.W.; funding acquisition, P.-Y.v.d.W. and M.S.; resources, P.-Y.v.d.W.; supervision, M.S. and P.-Y.v.d.W.
Declaration of interests
The authors have declared that no competing interests exist.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Beutler B.Cerami A.Cachectin (tumor necrosis factor): a macrophage hormone governing cellular metabolism and inflammatory response Endocr. Rev.919885766328623510.1210/edrv-9-1-57 · doi ↗ · pubmed ↗
- 2Peppel K.Crawford D.Beutler B.A tumor necrosis factor (TNF) receptor–Ig G heavy chain chimeric protein as a bivalent antagonist of TNF activity J. Exp. Med.174199114831489166052510.1084/jem.174.6.1483 PMC 2119031 · doi ↗ · pubmed ↗
- 3Maini R.N.Elliott M.Brennan F.M.Williams R.O.Feldmann M.TNF blockade in rheumatoid arthritis: implications for therapy and pathogenesis APMIS 1051997257263916446710.1111/j.1699-0463.1997.tb 00567.x · doi ↗ · pubmed ↗
- 4van Deventer S.J.H.Tytgat G.N.J.van Hogezand R.A.Treatment of Crohn’s disease with anti–tumor necrosis factor chimeric monoclonal antibody (c A 2)Gastroenterology 1091995129135779701110.1016/0016-5085(95)90277-5 · doi ↗ · pubmed ↗
- 5Maini R.St. Clair E.W.Breedveld F.Furst D.Kalden J.Weisman M.Smolen J.Emery P.Harriman G.Feldmann M.Lipsky P.Infliximab (chimeric anti–tumour necrosis factor alpha monoclonal antibody) versus placebo in rheumatoid arthritis patients receiving concomitant methotrexate: a randomised phase III trial Lancet 3541999193219391062229510.1016/s 0140-6736(99)05246-0 · doi ↗ · pubmed ↗
- 6Gorman J.D.Sack K.E.Davis J.C.Treatment of ankylosing spondylitis by inhibition of tumor necrosis factor αN. Engl. J. Med.3462002134913561198640810.1056/NEJ Moa 012664 · doi ↗ · pubmed ↗
- 7Rutgeerts P.Van Assche G.Sandborn W.J.Wolf D.C.Geboes K.Colombel J.F.Reinisch W.EXTEND Investigators Kumar A.Lazar A.Adalimumab induces and maintains mucosal healing in patients with Crohn’s disease: data from the EXTEND trial Gastroenterology 142201211021111.e 22232643510.1053/j.gastro.2012.01.035 · doi ↗ · pubmed ↗
- 8Browning J.L.Ngam-ek A.Lawton P.De Marinis J.Tizard R.Chow E.P.Hession C.O’Brine-Greco B.Foley S.F.Ware C.F.Lymphotoxin beta, a novel member of the TNF family that forms a heteromeric complex with lymphotoxin on the cell surface Cell 721993847856791665510.1016/0092-8674(93)90574-a · doi ↗ · pubmed ↗