Use of NanoBiT and NanoBRET to characterise interleukin‐23 receptor dimer formation in living cells

Background and Purpose Interleukin‐23 (IL‐23) and its receptor are important drug targets for the treatment of auto‐inflammatory diseases. IL‐23 binds to a receptor complex composed of two single transmembrane spanning proteins IL23R and IL12Rβ1. In this study, we aimed to gain further understanding of how ligand binding induces signalling of IL‐23 receptor complexes using the proximity‐based techniques of NanoLuc Binary Technology (NanoBiT) and Bioluminescence Resonance Energy Transfer (BRET). Experimental Approach To monitor the formation of IL‐23 receptor complexes, we developed a split luciferase (NanoBiT) assay whereby heteromerisation of receptor subunits can be measured through luminescence. The affinity of NanoBiT complemented complexes for IL‐23 was measured using NanoBRET, and cytokine‐induced signal transduction was measured using a phospho‐STAT3 AlphaLISA assay. Key Results NanoBiT measurements demonstrated that IL‐23 receptor complexes formed to an equal degree in the presence and absence of ligand. NanoBRET measurements confirmed that these complexes bound IL‐23 with a picomolar binding affinity. Measurement of STAT3 phosphorylation demonstrated that pre‐formed IL‐23 receptor complexes induced signalling following ligand binding. It was also demonstrated that synthetic ligand‐independent signalling could be induced by high affinity (HiBit) but not low affinity (SmBit) NanoBiT crosslinking of the receptor N‐terminal domains. Conclusions and Implications These results indicate that receptor complexes form prior to ligand binding and are not sufficient to induce signalling alone. Our findings indicate that IL‐23 induces a conformational change in heteromeric receptor complexes, to enable signal transduction. These observations have direct implications for drug discovery efforts to target the IL‐23 receptor.


| INTRODUCTION
The interleukin-23 (IL-23) cytokine is an important pro-inflammatory mediator involved in host defence against bacterial and fungal pathogens (Kagami et al., 2010;Verreck et al., 2004). The production of IL-23 is involved in the differentiation, expansion and maintenance of several pro-inflammatory cell types, including T helper 17 (Th17), natural killer T (NKT), γδT and innate lymphoid cells (ILC) (Gaffen et al., 2014). In contrast to the protective effects of IL-23 in infection, the cytokine has been shown to be a key mediator of auto-inflammatory conditions including Crohn's disease and plaque psioriasis (Tang et al., 2012). Several anti-IL-23 biological therapies have been approved for the treatment of auto-inflammatory conditions, and a new generation of peptide IL-23 receptor antagonists is in clinical trials (Cheng et al., 2019;Chyuan & Lai, 2020). In addition to auto-inflammatory indications, the IL-23 pathway has been implicated in cardiovascular disease (Ye et al., 2020), cancer (Mirlekar & Pylayeva-Gupta, 2021) and pain (Luo et al., 2021).
IL-23 is a heterodimeric disulfide linked cytokine, belonging to the promiscuously pairing Interleukin-12 (IL-12) family, which comprises IL-12, IL-23, IL-27 and IL-35 (with IL-39 reported in mice;Tait Wojno et al., 2019;Wang et al., 2016). The two components of IL-23 are termed IL23p19 and IL12p40, the latter of which also forms half of the IL-12 cytokine (Oppmann et al., 2000). The IL-23 receptor is formed of two single transmembrane domain containing cytokine receptor chains, IL23R and IL12Rβ1, with IL12Rβ1 also forming half of the IL-12 receptor (Parham et al., 2002). It has been demonstrated through mutagenesis and X-ray crystallography that the IL23R specifically binds to IL23p19 and the IL12Rβ1 to IL12p40 (Bloch et al., 2018;Glassman et al., 2021;Schroder et al., 2015).
The binding of IL-23 to its receptor leads to the transphosphorylation and activation of janus kinase (JAK) proteins, which are constitutively associated to sites in the intracellular domains of the receptor chains. JAK2 specifically binds to IL23R and TYK2 to IL12Rβ1 (Parham et al., 2002). Once the JAK proteins are activated, further signal transduction is mediated through phosphorylation of signalling proteins including dimeric members of the signal transducer and activator of transcription (STAT) family, with STAT3 dimers being one of the most prominent transducers (Floss et al., 2013;Parham et al., 2002).
The results of IL-23 stimulation on cell fate have been well studied; however, the mechanism by which extracellular binding translates to JAK activation is limited. Studies on purified truncated proteins indicated that the IL-23 receptor is activated through a ligand induced dimerisation mechanism, whereby IL-23 is first bound to the IL23R followed by the recruitment of IL12Rβ1 to the IL-23:IL23R complex (Bloch et al., 2018). However, further studies utilising recombinant cellular systems have indicated that the receptor is activated by conformational change of pre-associated complexes (Lay et al., 2022;Sivanesan et al., 2016).
In this study, NanoLuc Binary Technology (NanoBiT) was utilised to measure the formation of IL-23 receptor complexes in the presence and absence of the cytokine, to assess if heteromers form on the cell surface prior to ligand engagement. The NanoBiT methodology relies on a split version of the deep-sea shrimp derived bioluminescent enzyme NanoLuciferase (NanoLuc). The enzyme is split into a large (18 kDa) fragment (LgBit) and smaller (1.3 kDa) high affinity (HiBit) or low affinity (SmBit) peptides (Dixon et al., 2016). Fusions of NanoBiT fragments to the termini of targets of interest can be used to monitor the formation and disruption of protein complexes through changes in the production of bioluminescence by the complemented enzyme (Dale et al., 2019).

What is already known
• IL-23 is a pro-inflammatory cytokine involved in pathogen defence and the development of autoimmune diseases.
• IL-23 mediates its effect through a heteromeric receptor complex composed of IL23R and IL12Rβ1.

What does this study add
• Inactive IL-23 receptor complexes can form prior to cytokine engagement.
• High affinity constraint of the receptor N-terminal domains can synthetically induce ligand-independent signalling.

What is the clinical significance
• The interleukin-23 receptor is an emerging drug target for the treatment of auto-inflammatory diseases.
• These observations have direct implications for drug discovery efforts to target the IL-23 receptor complex.

| NanoBiT BRET experiments
Transfected cells in 96-well microplates had media removed and replaced with a concentration titration of IL23-TMR in HBSS with 1 mgÁml À1 BSA with or without IL-23. The cells were incubated for an hour before 5 μl of 77-μM diluted furimazine was added to each well.
The plate was then incubated for 5 min, a white plate back was added and the BRET signal was measured on a PheraStar FS plate reader using a 450-nm (30-nm bandpass) and >550-nm filter with gains of 2800 and 3600, respectively.

| Measurement of STAT3 phosphorylation
AlphaLISA Surefire Ultra assays were carried out as previously described (Lay et al., 2022). Briefly, media was replaced with 100-μl DMEM without FBS, and the cells incubated for 3 h. Media was then replaced with 50-μl HBSS with 1 mgÁml À1 BSA, with or without IL-23 and cells incubated for a further 30 min. The assay was then carried out according to the manufacturer's instructions.

| Measurement of HaloTag fluorescence intensity
Cells were incubated with media containing 200-nM Alexa Fluor (AF) 488 HaloTag ligand for 30 min. Cells were then washed three times with 50-μl HBSS followed by the addition of 50-μl HBSS with 1 mgÁml À1 BSA and measurement of fluorescence intensity using a PheraStar FS plate reader using an excitation laser at 485 nm and emission at 520 nm.

| Data analysis
Data are presented as mean ± SEM. All experiments were performed in at least five independent experiments with triplicate or quadruplicate wells (see figure legends for details). Drug additions were randomly allocated to wells within a 96-well plate.
Scale bars were added to luminescence images using FIJI (National Institute of Health). BRET ratio values were generated in MARS (BMG Labtech) using the below equation: BRET ratio ¼ Acceptor signal Donor signal : Data were exported from the PheraStar FS reader and stored in Microsoft Excel. Further data analysis was carried out using GraphPad Prism.
BRET Ligand binding curves were fitted with the GraphPad Prism 'One site -Total and nonspecific binding' fit, which used the equation: Competition curves were fitted with the GraphPad Prism 'log (inhibitor/agonist) vs. response -Variable slope (four parameters)' fit which used the equation: Normalised binding curves for HiBit and LgBit were fit using the GraphPad Prism 'One site -Specific binding' fit using the equation: where 'B max ' is the maximum signal of the curve and '[B]' is the concentration of IL23-TMR.
The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology (Curtis et al., 2022). Statistical analysis was undertaken only for studies where each group size was at least n = 5.
A value of P < 0.05 was used to determine statistical significance. These constructs were then transiently expressed in HEK293T cells to test that they were surface expressed and that they retained the ability to be complemented by corresponding NanoBiT fragments. IL23R complemented with LgBit, with a lower signal observed for the SmBit-IL23R construct due to the lower affinity of SmBit for LgBit (Peach et al., 2021).

| Nomenclature of targets and ligands
When HiBit-IL23R was expressed with LgBit-IL12Rβ1, a strong luminescent signal was observed (73.3-fold that of the mean of the receptor constructs expressed in isolation). When an equivalent pairing was expressed containing SmBit-IL23R, the signal was weaker (6.19-fold lower) but remained higher than that of the monomers complemented with exogenous protein (10.3-fold higher than the mean of receptor constructs expressed in isolation; Figure 1b). Finally, when cells expressing the complemented receptor pairs were exposed to purified HiBit and LgBit protein (Figure 1c,d), there was no increase in luminescence in either condition for HiBit-IL23R containing heteromers. There was however an increase in luminescence equivalent to that observed for LgBit-IL12Rβ1 expressed alone when purified HiBit was added to cells expressing the SmBit-IL23R containing heteromer, indicating that this construct had a lower affinity for LgBit-IL12Rβ1 than the equivalent HiBit construct.
To assess the localisation of NanoBiT complemented heteromers, luminescence images were taken of cells expressing HiBit-IL23R and LgBit-IL12Rβ1. These images demonstrated luminescence at the periphery of the cell, and at intracellular localisations, these observations were similar to NanoLuc (NL)-IL23R when imaged with IL12Rβ1 under equivalent conditions ( Figure 2).
Previously, we created and validated a TMR labelled IL-23 species termed IL23-TMR (Lay et al., 2022). To test if the NanoBiT heteromers were able to bind ligand, IL23-TMR was applied to cells expressing HiBit or SmBit-IL23R and LgBit-IL12Rβ1 before measuring a NanoLuc Bioluminescence Resonance Energy Transfer (NanoBRET) signal generated between the complemented luciferase and IL23-TMR. The luminescence signal generated from the SmBit heteromer was too low to produce a reliable BRET ratio; however, a robust BRET signal was generated using cells transfected with the HiBit heteromer, which could be abolished through the addition of unlabelled IL-23 ( Figure 3). The affinity of IL23-TMR for these binary IL-23 receptors was measured to be 234 ± 40 pM ( Figure 3b, mean ± SEM, n = 5). This value was between that of the previously measured values for IL23-TMR binding to HEK293T cells expressing NL-IL23R and IL12Rβ1 (27.0 ± 3.6 pM) and cells expressing IL23R and NL-IL12Rβ1 (647 ± 80 pM; Lay et al., 2022). To confirm that nonmodified IL-23 could also bind to the NanoBiT complemented heteromers, we carried out a NanoBRET competition assay to measure the displacement of 300 pM IL23-TMR from cells expressing HiBit-IL23R and LgBit-IL12Rβ1 by increasing concentrations of IL-23 (Figure 3c).
An IL-23 affinity of 352 ± 155 pM (mean ± SEM, n = 5) was measured, which was comparable to that measured for IL23-TMR. Whilst it was confirmed that both IL23-TMR and IL-23 bound to the Nano-BiT heteromers, there was no increase in NanoBiT complementation when IL-23 was added to either SmBit or HiBit heteromers (Figure 4).
On the contrary, the addition of IL-23 significantly (paired t test,

| Use of NanoBiT to monitor IL23R homomer formation and ligand-binding
The luminescent signal generated by homomeric pairings of IL23R was also tested (Figure 5a). A luminescent signal was generated by  Figure 5b).

| Low complementation between HiBit-IL-23R
and LgBit-VEGFR-2 To check whether complementation could occur between HiBit-IL23R and an unrelated LgBit-tagged receptor expressed on the surface of HEK293T cells, we investigated the association between the IL23R and LgBit-tagged VEGFR-2, which we have previously show to form heterodimers with its co-receptor neuropilin-1 (Peach et al., 2021). As

| Impact of heteromerisation on the binding affinity of purified LgBit for HiBit-IL23R
To assess whether receptor heteromerisation alters the conformation of the N-terminal regions of the partner receptor proteins, we measured whether this changed the affinity of N-terminal NanoBiTtagged receptors for purified NanoBiT complementation partners.
We first investigated the effect of applying increasing concentrations of purified HiBit or LgBit to cells expressing the HiBit-IL23R, SmBit-IL23R or LgBit-IL12Rβ1, respectively ( Figure 9).
The HiBit-IL23R had an affinity of 15.2 ± 1.6 nM (n = 5) for purified LgBit, which was 21.7-fold weaker than the value quoted for HiBit alone (Dixon et al., 2016). The affinity of purified LgBit for cells expressing the SmBit-IL23R was also tested; however, no luminescence signal was detected over the concentration range used; this observation is in accordance with the reported 190-μM affinity of SmBit for LgBit (Dixon et al., 2016). When the affinity of purified HiBit control protein for LgBit-IL12Rβ1 was measured it was found that whilst there was an increase in luminescence, this signal did not saturate over the concentration range tested (up to 1 μM) (Figure 9b).
To probe if the affinity of the receptor conjugated NanoBiT tags was influenced by heteromerisation, assays were carried out to measure the affinity of HiBit-IL23R for purified LgBit protein when expressed in isolation or co-expressed with untagged IL12Rβ1 and in the presence and absence of IL-23 ( Figure 10). The affinity of LgBit for HiBit-IL23R expressed in isolation, with addition of 5-nM IL-23 was 25.9 ± 7.9 nM (n = 5), which was not statistically different from the HiBit dissociation constant measured in the absence of IL-23 (unpaired t test). In contrast, when HiBit-IL23R was co-expressed with IL12Rβ1, the affinity of LgBit was decreased, to a similar degree both in the presence and absence of IL-23, to a level where there was no clear saturation in luminescence signal at the highest concentration used in the assay.

| DISCUSSION
It has previously been suggested that inactive complexes of the IL-23 receptor may associate in the absence of ligand (Lay et al., 2022;Sivanesan et al., 2016). In this study, heteromer formation in the absence of ligand was observed using the luminescence signal generated when N-terminal domains of the receptor are labelled with corresponding NanoBiT fragments and co-expressed in HEK293T cells. The lack of subsequent STAT3 phosphorylation observed for SmBit-IL23R:LgBit-L12Rβ1 heteromeric complexes confirms that heteromerisation alone is not sufficient to induce signalling. Furthermore, we F I G U R E 7 Low complementation between HiBit-IL-23R and LgBit-VEGFR-2. Luminescence generated when HEK293T cells transiently transfected with LgBit-conjugated IL12Rβ1 or VEGFR-2 constructs was incubated with either purified HiBit (50 nM) or cotransfected with HiBit-IL-23R. The luminescence generated by LgBit-VEGFR-2 and LgBiT-IL12Rβ1 in the absence of purified HiBit is also shown. Data are mean ± SEM from four independent experiments, each performed in triplicate. with an 18 amino acid truncated linker. Data are mean values generated in eight (b) or five (c) independent matched experiments with previously generated wildtype data for comparison; overall mean and SEM are also plotted. A one-way ANOVA was used to measure the statistical significance of differences. * signifies p <0.05.
confirmed that IL-23 could bind to these receptor heteromers using NanoBRET and phospho-STAT3 transduction assays. Taken together with the observation that treatment of HiBit-IL23R:LgBit-L12Rβ1 or SmBit-IL23R:LgBit-L12Rβ1 heteromeric complexes with IL-23 did not lead to an increase in luminescence, these data indicate that ligandinduced dimerisation is not the sole factor governing receptor activation.
Whilst we could not completely rule out the possibility that HiBit: LgBit interactions were contributing to complex formation, we found that the affinity of HiBit and LgBit for their corresponding NanoBiT fragment was greatly reduced when fused to IL-23 receptor monomers. The effect was most marked for LgBit-IL12Rβ1, which did not generate a saturable luminescence signal with increasing concentrations of purified HiBit over the concentration range tested, having a likely affinity in the order of 100-1000 nM, making this the limiting affinity for NanoBiT complementation driven heteromer formation.
These data strongly suggest that the NanoBiT tags are not driving the interactions observed between IL23R and IL12Rβ1.
Utilising a NanoBiT BRET binding assay, in which light generated from complemented receptors was used to excite fluorophore conjugated IL-23 (IL23-TMR), it was possible to monitor binding interactions of IL-23 solely to pre-formed heteromers. This approach demonstrated an affinity of pre-formed IL23R-IL12Rβ1 heteromers for IL23-TMR similar to that previously measured for IL23-TMR binding to NanoLuc tagged receptor heteromers. Interestingly, the equilibrium dissociation constant was between that of IL23-TMR binding to cells expressing NL-IL23R and IL12Rβ1 and those expressing NL-IL12Rβ1 and IL23R. The hypothesis previously suggested for the difference in these affinities, was that tagging IL12Rβ1 with NanoLuc led to greater steric interference in the binding of the ligand (Lay et al., 2022). A NanoBiT heteromer binding affinity mid-way between these values suggests that the position of the complemented luciferase interferes less in binding than NanoLuc fused to IL12Rβ1, but more than NanoLuc fused to IL23R. As the affinity of IL23-TMR binding to NanoBiT complemented heteromers is equivalent to the affinities measured for the interaction of the probe with NanoLuc-tagged heteromers, and that IL-23 does not induce heteromerisation, it is most likely that heteromers are pre-formed prior to ligand binding in both experiments.
We previously observed that binding of IL-23 to its receptor leads to a change in the BRET ratio between N-terminal donor and acceptor tags, and hypothesised that this was due to a ligand induced change in the proximity or orientation of the N-terminal domains (Lay et al., 2022). By utilising the NanoBiT system, which is less influenced by orientation than BRET due to the importance of dipole-dipole interactions for resonance energy transfer, we were able to focus primarily on ligand induced changes in the proximity of the N-terminal F I G U R E 9 HiBit-IL23R has a greater affinity for purified LgBit protein than SmBit-IL23R. (a) The luminescence signal generated when a concentration titration of LgBit was applied to cells expressing either HiBit or SmBit conjugated IL23R. (b) The luminescence signal generated when increasing concentrations of purified complementary NanoBiT fragment was applied to cells transiently expressing HiBit-IL23R or LgBit-IL12Rβ1 in isolation. Data are mean with SEM from five or six (SmBit-IL23R) independent experiments, each performed in triplicate.
F I G U R E 1 0 Heteromerisation but not ligand engagement decreases the affinity of exogenous LgBit for HiBit-IL23R. The luminescence signal generated by the application of increasing concentrations of purified LgBit protein to cells expressing HiBit-IL23R in the presence or absence of co-expressed IL12Rβ1 and with or without IL-23. Data are mean ± SEM from five independent experiments, each performed in triplicate.
domains of the receptor (Machleidt et al., 2015). The lack of a change in luminescence when ligand was added to NanoBiT complemented complexes, indicated that IL-23 induces a change in orientation rather than proximity of the N-terminal domains. It has been shown that both the N-terminal and C-terminal domains of the IL-23 receptor undergo a shift in position on ligand binding (Lay et al., 2022;Sivanesan et al., 2016). These observations led to the suggestion that the IL-23 receptor is activated through a 'scissor-like' conformational change in a similar manner to the Erythropoietin receptor in which the C-terminal domains of the receptor are brought into closer proximity allowing trans-phosphorylation of C-terminally associated JAK proteins (Sivanesan et al., 2016). Our observation of ligand independent signalling when the N-terminal domains of the receptor are constrained in specific orientations is consistent with this hypothesis. It is possible that this finding indicates that the wildtype IL-23 receptor could be activated by a synthetic bitopic ligand capable of constraining the N-terminal domains of the receptor in an active conformation.

Measurement of basal
Indeed, it has previously been shown that a chimeric IL-23 receptor, with extracellular domains switched for anti-mCherry or GFP nanobodies, could be activated by the application of a bitopic ligand consisting of a mCherry-GFP fusion (Engelowski et al., 2018).
We also examined if heteromers containing more than one IL23R molecule would associate on the cell surface in the absence of ligand.
We found that complexes containing HiBit-IL23R or SmBit-IL23R and LgBit-IL23R gave half the signal of complexes containing HiBit or SmBit-IL23R and LgBit-IL12Rβ1. We also examined if these homomeric complexes could engage ligand, finding that when co-expressed with IL12Rβ1, the majority of homomers were disrupted; however, the remaining complexes could engage the IL23-TMR. As IL23R and IL12Rβ1 must be present to form the high affinity IL-23 binding site (Lay et al., 2022), a multimeric species consisting of at least two IL23R molecules and IL12Rβ1 must be engaging IL-23 in this situation.
Taken together, our results indicate that multimeric IL-23 receptor species containing multiple IL23R can be induced to form with no reduction in ligand binding affinity; therefore, the formation of higher order receptor structures, as is the case for other closely related receptors (Ward et al., 1994), cannot be ruled out.
Our finding that the affinity of purified LgBit for HiBit-IL23R was significantly decreased when co-expressed with IL12Rβ1 indicates that IL12Rβ1 shields the N-termini of HiBit-IL23R in the IL-23 receptor complex. As the affinity of LgBit is reduced to a similar degree both in the presence and absence of IL-23, it is likely that in both conformations the N-terminal domain of IL23R is in close association with IL12Rβ1.
We also evaluated whether increasing expression of HT-IL12Rβ1 Numerous examples of membrane-spanning blocking peptides have been reported that act as allosteric inhibitors of complex formation (Stone & Deber, 2017). Such molecules could be useful tools to further study heteromer formation and a potential novel therapeutic tactic to antagonise IL-23 signalling at the heteromeric interface.
A crucial piece of information for this approach is the location of the formation of heteromeric receptor species. In our study, we observed luminescence at intracellular foci and on the cell membrane when imaging cells expressing HiBit-IL23R and LgBit-IL12Rβ1. Other cytokine receptors such as the growth hormone receptor have been shown to be exported from the endoplasmic reticulum as a dimer (Gent et al., 2002); however, studies on IL23R have shown that the molecule is internalised after ligand stimulation and recycled to the cell surface in the absence of IL12Rβ1 (Sun et al., 2020).
Whilst single transmembrane receptors were originally thought to be activated through dimerisation, the growing list of receptors that are heteromeric and inactive prior to ligand binding suggests that many receptors may be activated by a conformational rather than oligomeric mechanism, for example, ligand induced rotation (Maruyama, 2015;Westerfield & Barrera, 2020). Our study demonstrates that when co-expressed on HEK293T cells, IL23R and IL12Rβ1 associate in the absence of ligand to form inactive complexes and are therefore unlikely to be activated by dimerisation alone. The precise mechanism by which these heteromers are activated remains to be elucidated. Whilst the structure of IL-23 in complex with the purified and truncated IL23R domain 1-3 (Bloch et al., 2018) and more recently with IL12Rβ1 domain 1 (Glassman et al., 2021) have been solved, structural elucidation of the remainder of IL12Rβ1 was not possible, and both studies indicated that the affinity of purified truncates of IL12Rβ1 was much lower for IL-23 than when the receptor is expressed in the membrane of living cells (Lay et al., 2022). Cryo-EM of liganded and unliganded receptor complexes solubilised in a lipid membrane would be a useful next step in the defining the conformational change that leads to receptor activation. Structural studies utilising mutagenesis, conformational restraint, crystallography and molecular modelling have been used to investigate the activation mechanisms of related cytokine receptors (Brown et al., 2005;Lu et al., 2006;Poger & Mark, 2010;Seubert et al., 2003). Techniques such as these could be applied to IL-12 family cytokine receptors to further understand receptor activation.

| CONCLUSION
The IL-23 cytokine and its receptor are important drug targets for the treatment of auto-inflammatory conditions. Although receptor antagonists are undergoing clinical trials, anti-IL-23 biologics are currently the only licensed therapeutics to specifically target the pathway. To successfully target the receptor, further information is needed on the activation mechanism by which IL-23 binding translates to signalling.
Conflicting hypotheses have been suggested in which the receptor is

RIGOUR
This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Design and Analysis, and as recommended by funding agencies, publishers and other organisations engaged with supporting research.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.