Abstract
More than 20 years after the description of the two IR (insulin receptor) isoforms, designated IR-A (lacking exon 11) and IR-B (with exon 11), nearly every functional aspect of the alternative splicing both in vitro and in vivo remains controversial. In particular, there is no consensus on the precise ligand-binding properties of the isoforms. Increased affinity and dissociation kinetics have been reported for IR-A in comparison with IR-B, but the opposite results have also been reported. These are not trivial issues considering the reported possible increased mitogenic potency of IR-A, and the reported link between slower dissociation and increased mitogenesis. We have re-examined the ligand-binding properties of the two isoforms using a novel rigorous mathematical analysis based on the concept of a harmonic oscillator. We found that insulin has 1.5-fold higher apparent affinity towards IR-A and a 2-fold higher overall dissociation rate. Analysis based on the model showed increased association (3-fold) and dissociation (2-fold) rate constants for binding site 1 of IR in comparison with IR-B. We also provide a structural interpretation of these findings on the basis of the structure of the IR ectodomain and the proximity of the sequence encoded by exon 11 to the C-terminal peptide that is a critical trans-component of site 1.
- binding kinetics
- harmonic oscillator
- insulin
- insulin receptor
- insulin receptor lacking exon 11 (IR-A)
- insulin receptor with exon 11 (IR-B)
- mathematical modelling
INTRODUCTION
The IR (insulin receptor) belongs to a subfamily of receptor tyrosine kinases that comprises the type I IGF-IR [IGF (insulin-like growth factor) receptor] and the insulin-receptor-related receptor (for a review, see [1–3]). They are covalent disulfide-linked dimers made of two extracellular α-subunits that contain the ligand-binding domains and two transmembrane β-subunits that contain the intracellular tyrosine kinase. The cloning and sequencing of the IR cDNA in 1985 by two different groups revealed two transcripts that predicted receptor sequences differing by the absence [4] or presence [5] of a 12-amino-acid segment (718–729 in the numbering in [5]) at the C-terminal end of the IR α-subunit. The structure of the IR gene [6] was shown to consist of 22 exons and 21 introns. The 718–729 sequence is encoded by exon 11, which suggested that the two isoforms were generated by alternative splicing of that exon. The isoform lacking the insert is nowadays denoted IR-A, and the other isoform is denoted IR-B. The existence of alternative splicing was confirmed by PCR [7–9].
The relative abundance of the two splice variants is highly regulated in a tissue-specific manner. Thus liver expresses only IR-B, whereas muscle and placenta express both isoforms, and lymphocytes express only IR-A [7–9]. The same tissue-specific pattern was also seen in the rat [10], suggesting functional significance of this alternative splicing. However, in other insulin target tissues significant species differences have been reported. In particular, in skeletal muscle, isoform expression is approximately 95% IR-A in rat [11,12], but approximately 70% IR-B in human [7,13–15]. There are also species differences in isoform expression in adipose tissue, kidney and spleen [9,11,16,17]. IR-A is predominantly expressed in prenatal life and has a much higher affinity for IGF-II than IR-B, and IR-A thus plays an important role in fetal development and may play a role in certain cancers (for a review, see [1]). IGF-I also binds better to IR-A [18], but not as well as IGF-II [19]. IR-A is considered to be more involved in ‘mitogenic’ signalling, whereas IR-B is more involved in ‘metabolic’ signalling [1,20]. The functional consequences of this splicing remain controversial. An altered distribution of isoforms in adipocytes or muscle of patients with Type 2 diabetes was reported by some, but not others (see [1] for a review). There is also variability in the reported ligand-binding properties of the two isoforms of the IR. IR-A was initially suggested to have a 5-fold higher affinity than IR-B [21], whereas subsequent work using a variety of cell-transfection systems reported a ~2-fold difference [8,18,22,23] or no difference [24,25]. In contrast, Denley et al. [26] found IR-B to have a 2-fold higher affinity than IR-A. The dissociation rate of IR-A (by dilution) was reported to be lower than that of IR-B by McClain et al. [27], but to be higher by Gu et al. [22] and Yamaguchi et al. [18]. Internalization of IR-A has been reported to be faster than for IR-B {in CHO (Chinese-hamster ovary) cells [18,23]; in rat-1 fibroblasts [28]; and in Hela cells [20]}, whereas others found no difference in rat-1 fibroblasts [18,27]. The insulin affinity for hybrids with IGF-IR was found to be isoform-dependent by Pandini et al. [29], but not by Slaaby et al. [30] and Benyoucef et al. [19].
In the present study, we have carefully re-examined the ligand-binding kinetics of the two IR isoforms expressed in CHO cells using a recently developed structure-based mathematical model [31] of the insulin-binding mechanism.
Insulin has been shown to contain two binding surfaces which are involved in binding to the IR: the classical binding surface and a recently mapped binding site named the novel binding surface [32,33]. The classical binding surface overlaps with the surface involved in dimer formation, and the novel binding surface overlaps with the surface involved in hexamerization of insulin [32,33]. Insulin binding to the IR exhibits negative co-operativity, manifested by curvilinear Scatchard plots and acceleration of the dissociation rate of prebound 125I-labelled insulin in an infinite dilution in the presence of unlabelled ligand (for a review, see [32]). Binding models have been proposed that imply bivalent cross-linking by site 1 and site 2 of insulin to two distinct sites on each IR α-subunit [32,34]. To explain negative co-operativity, De Meyts [32] proposed that sites 1 and 2 of the IR should have an antiparallel symmetry that allows alternative cross-linking to each pair of sites 1 and 2 of the IR. This model is supported by the crystallographic structure of the IR extracellular domain [35]. The extracellular domain of the IR consists of two leucine-rich (L1 and L2) domains separated by a CR (cysteine-rich) domain, followed by three fibronectin type III (Fn1–3) domains. The combination of photoaffinity cross-linking [36–39] and alanine scanning mutagenesis [40,41] studies suggest that the classical binding surface of insulin binds to the IR residues located in the central β-sheet of the L1 module and to a C-terminal peptide (704–719) (located next to the alternatively spliced 12-amino-acid sequence), which combine in trans to form ‘site 1’, whereas the novel binding surface binds to the less well-characterized ‘site 2’ of the IR, which most likely corresponds to residues from the C-terminal portion of L2 as well as the loops at the junction of the Fn1 and Fn2 modules [42–45].
The structure-based mathematical model of IR kinetics by Kiselyov et al. [31] allows us, for the first time, to accurately describe all of the kinetic properties of the IR using only five parameters: association rate constants a1 and a2 and dissociation rate constants d1 and d2 for site 1 and site 2 respectively, and a cross-linking constant Kcr. This model is based on an assumption that the IR conformational change upon ligand binding is analogous to a spontaneous oscillation of a one-dimensional harmonic oscillator in thermal equilibrium with the surroundings. A highly simplified scheme of the harmonic oscillator model is shown in Figure 1, in which an analogy from mechanics is used for depiction of the IR oscillator: the IR monomer is depicted as two rigid balls (representing sites 1 and 2) connected by a rigid concave rod. The force acting to restore the inactive symmetrical conformation is represented by a spring, which is assumed to be relaxed when the IR dimer is in the symmetrical (inactive) conformation. The harmonic oscillator model confirms the intuitive predictions of the symmetrical bivalent model suggested by De Meyts [32] and allows robust determination of all of the kinetic parameters.
a1 and a2 are the association rate constants for binding sites 1 and 2 respectively (M−1·s−1), and d1 and d2 are the dissociation rate constants for binding sites 1 and 2 respectively (s−1). Kcr is the cross-linking constant (s−1). S1 and S2 are the binding sites 1 and 2 of the IR. The binding of insulin to the IR is only shown for sites 1 and 2, numbered 1 and 2 respectively. It is assumed that there are no differences between site 1, numbered 1 or 3, or between site 2, numbered 2 or 4. The insulin molecule is pig insulin (PDB code 9INS). The graphics program used was PyMOL (http://www.pymol.org).
Since the C-terminal peptide (704–719) contributes to site 1 of the IR, it is not unexpected that the two IR isoforms have different kinetic properties, since the 12-amino-acid long sequence (718–729 in the numbering in [5]) encoded for by exon 11 is located right next to it. We show that the association rate constant a1 is approximately 3-fold higher for the IR-A isoform than IR-B, whereas the dissociation rate constant d1 is 2-fold increased, resulting in a 1.5-fold increase in the apparent high affinity. We explain this in terms of the recent structural information regarding the insulin–IR binding mechanism.
MATERIALS AND METHODS
Chemicals and reagents
125I-labelled and non-labelled human insulin were provided by Novo Nordisk A/S, Denmark.
Cell culture
CHO cells overexpressing human IR-A or human IR-B (provided by Bo Falck Hansen, Novo Nordisk A/S, Denmark) were cultured in DMEM (Dulbecco's modified Eagle's medium; 1000 g/l glucose, sodium pyruvate, Glutamix-1, pyridoxal) with 10% FBS (fetal bovine serum), 1% penicillin/streptomycin, 1% MEM-NEAA (minimum essential medium with non-essential amino acids) and 0.45 mg/ml G418 (Gibco, Invitrogen) at 37°C in a humidified atmosphere of 5% (v/v) CO2.
Preparation of cells for binding experiments
The cells were washed in 10 ml of PBS without CaCl2 and MgCl2 (Gibco, Invitrogen) and 10 ml of 100 μM EDTA (Sigma) in PBS was added to the cells. Then, 25 ml of HBB (Hepes binding buffer) [100 mM NaCl (Merck), 5 mM KCl (Merck), 1.2 mM MgSO4 (Sigma), 1 mM EDTA (Sigma), 10 mM glucose (Sigma), 15 mM sodium acetate (Merck), 100 mM Hepes (Sigma) and 1% BSA (Amresco), pH 7.6 adjusted with NaOH (Sigma)] was added and the cell suspension was centrifuged at 200 g for 5 min at room temperature in a Heraeus, Sorvall® Labofuge 400e (AXEB Laboratory Products), and the supernatant was discarded. The cells were re-suspended in an appropriate volume of HBB, and the concentration of cells was measured using a NucleoCounter (Chemometec). The cells were centrifuged at 200 g for 5 min at room temperature, and the supernatant was discarded. The cell pellet was re-suspended in an appropriate volume of HBB to get the right cell concentration according to the assay the cells were used for.
Receptor-binding assays
All experiments were performed three times in duplicate unless otherwise stated.
Homologous competition assay
The cell suspension was incubated for 4 h at 15°C with increasing concentrations (1.67 × 10−11 M to 1.67 × 10−5 M) of unlabelled insulin, 50 μl (20000 c.p.m./50 μl) of 125I-labelled insulin, and HBB to a total volume of 0.5 ml. Duplicate aliquots were centrifuged after incubation and the bound activity was counted. Two additional aliquots were not centrifuged, but counted as the total.
Dose–response assay for accelerated dissociation
125I-labelled human insulin was added to 900 μl of the cell suspension (~15 × 106 cells/ml) to a final concentration of 75000 c.p.m./ml and incubated for 2 h at 15°C. After pre-incubation the cells were centrifuged for 5 min at 1000 g at 4°C. The cell pellet was re-suspended in the original volume of HBB, and 25 μl of the cell suspension was transferred into 15 duplicate tubes containing 500 μl of an increasing concentration of non-radioactive ligand (1.67 × 10−11 M to 1.67 × 10−5 M) starting with the tubes with the lowest concentration. The 30 reaction tubes were incubated at 15°C for 0, 7.5, 15, 30, 60 or 120 min. Sextuples of 25 μl of the cell suspension were counted as the total (no centrifugation). After the dissociation the 30 reaction tubes were centrifuged in a 5810 R centrifuge (Eppendorf, Germany) for 5 min at 3220 g at 4°C and the radioactive ligand bound was counted in a gamma counter for 30 min.
Curve-fitting using a mathematical model
The binding experiments were modelled using a mathematical model developed in our department [31]. The model is based on the hypothesis that the receptor dimer with anti-parallel disposition of the two half-receptors can be assimilated to a one-dimensional harmonic oscillator. On the basis of the assumptions of the model, the possible combinations of insulin and IR result in a set of 35 intermediaries and thus 35 differential equations that are fitted to the experimental data using the program Mathematica. The model describes the binding of insulin to IR by using only five binding parameters: a1 and a2, the association rate constant to IR sites 1 and 2 respectively; d1 and d2, the dissociation rate constant to IR sites 1 and 2 respectively; and Kcr, the receptor cross-linking constant (which is equal to the reciprocal of the time it takes for the oscillator to close). The goodness-of-fit between the experimental (Xi) and simulated (Yi) data points can be evaluated with a score function equal to 1/N Σ(Xi−Yi)/(SDi)2, where N is the number of data points and SDi is the S.D. of the experimental data point using a genetic algorithm.
RESULTS
In order to compare the binding properties of the two IR isoforms, we first tested if there is a difference in the affinity of the receptors for insulin. For this purpose, binding of human insulin tracer (at equilibrium) to cells overexpressing either IR-A or IR-B in the presence of various concentrations of unlabelled insulin was analysed. As shown in Figure 2, insulin competed more efficiently when IR-A was used. Fitting of the binding data to a two-site sequential model produced the following values for the apparent high affinity (cross-linked sites 1 and 2) dissociation constant (Kd): 236±86 pM and 349±128 pM for IR-A and IR-B respectively. From this we conclude that IR-A has a 1.5-fold higher affinity in comparison with IR-B. The Kd values for the second site (uncross-linked site 1) were 1.75±5.86 nM for IR-A and 2.62±8.80 nM for IR-B, which is in agreement with previously reported data for the affinity of site 1 [24,31].
Experiments were performed with insulin as the ligand and CHO cells overexpressing either IR-A (●) or IR-B (○).The dots are the experimental data points and the lines correspond to the fitted curves using a sequential model. The apparent high-affinity (corresponding to cross-linked sites 1 and 2) dissociation constant (Kd) value for insulin when using IR-A was 236±86 pM and 349±128 pM when using IR-B. The Kd values for the second site (corresponding to the uncross-linked site 1) were 1.75±5.86 nM for IR-A and 2.62±8.80 nM for IR-B.
According to the harmonic oscillator model [31], the rate of the IR dissociation at maximum acceleration is proportional to the dissociation coefficient for site 1, and can be estimated with high accuracy as 0.5 d1. Thus measurement of the maximum dissociation rate allows us to easily compare the d1 values for the two receptor isoforms. In order to estimate the maximum dissociation of IR, we tested the rate of the insulin tracer dissociation at various concentrations of the unlabelled insulin and established that the maximum acceleration was achieved at 100 nM concentration for both receptor isoforms. As can be seen from Figure 3, the rate of the IR dissociation at maximum acceleration was higher for IR-A, and fitting of the dissociation data to a mono-exponential decay showed that the maximum dissociation rate for IR-A was approximately 2-fold higher than that for IR-B. This indicates that the d1 value, or dissociation coefficient for site 1, is 2-fold higher for IR-A than IR-B.
Prebound insulin was allowed to dissociate for various time periods (0, 7.5, 15, 30, 60 and 120 min) in dilution alone from IR-A (●) or IR-B (■), or in the presence of an excess (100 nM) of non-radioactive insulin from IR-A (○) or IR-B (□).
The Kd value for the apparent high-affinity-binding site of the IR can be calculated from the a1 and a2 (association coefficients for sites 1 and 2 respectively), d1 and d2 (dissociation coefficients for sites 1 and 2 respectively), and Kcr (cross-linking constant) according to an approximate formula [31]:
We have shown above that the d1 value for IR-A is 2-fold higher than for IR-B. At the same time, the Kd value for IR-A is 1.5-fold lower. From the formula above, it is obvious that, in order for Kd to become 1.5-fold lower (assuming d1 is 2-fold higher): (i) a1 has to increase 3-fold, (ii) d2 has to decrease 3-fold, or (iii) Kcr has to increase 3-fold.
Since the cross-linking constant describes the oscillatory properties of the whole IR molecule, it is unlikely that the presence of exon 11, which represents only a very small fraction of the IR total mass, can substantially change the value of the cross-linking constant. Thus it is likely that either a1 is increased 3-fold or d2 is decreased 3-fold in order for IR-B to have 1.5-fold lower affinity.
To test whether either the a1 or d2 parameters are affected, we observed that the rate of the IR dissociation (in the absence of unlabelled ligand) is given by d1 × d2/Kcr [31]. This means that if the d2 values are the same for the two isoforms, then the rate of the IR dissociation is expected to be 2-fold higher for IR-A, since we have already established that d1 is 2-fold higher for IR-A. As appears from Figure 3, the rate of the IR dissociation for IR-A is indeed higher than for IR-B, and fitting of the mono-exponential decay to the dissociation data shows that the difference in the dissociation rate is approximately 2-fold.
Thus we have demonstrated that IR-A binds to insulin with 1.5-fold higher affinity when compared with IR-B, and our kinetic analysis indicates that this difference is caused solely by the perturbation of the site 1 rate constants, namely the presence of exon 11 leads to a 2-fold decrease of d1, and a 3-fold decrease of a1.
In order to verify the conclusions above, we decided to estimate the a1, a2, d1, d2 and Kcr parameters for the two IR isoforms by curve-fitting the harmonic oscillator mathematical model to the dose–response of the accelerated dissociation for the two IR isoforms. As can be seen from Figure 4, the dose–response for IR-A is characterized by faster dissociation in comparison with IR-B. The data are also represented in Figure 5, in which the dissociation is plotted as a function of time instead of concentration. Since fitting results might be affected by the initial choice of the parameter values, we performed ten independent parameter estimations for each IR isoform using a genetic algorithm (as described in [31]) with randomly selected values of the initial parameters. The average values of the parameters and their S.D. values are shown in Table 1. As can be seen in Table 1, some parameters have large S.D. values, which imply that a relatively large range of parameter values allow us to make a good fit of the model to the experimental data. Thus, when comparing the parameter values between the two IR isoforms, the uncertainty of the parameter estimation (represented by the S.D.) should be taken into account. To characterize the significance of the difference between the parameter values for IR-A and IR-B, we calculated a significance factor, defined as the difference between the parameter values divided by the sum of the parameter S.D. value. As can be seen from Table 1, the a1 parameter values for the two isoforms differ from each other by almost four S.D. values (column named Significance), the d1 parameter values for the two isoforms differ from each other by approximately 1.5 S.D. values, and all of the other parameters values for the two isoforms differ from each other by much less than 1 S.D. From this we can conclude that only the parameters of site 1 (a1 and d1) differ substantially between the two isoforms, namely the presence of exon 11 leads to a 1.6-fold decrease of d1, and a 2.3-fold decrease of a1, which is in good agreement with our previous kinetic analysis.
Cells expressing IR-A (A) or IR-B (B) were pre-incubated with 125I-labelled insulin for 2 h, and then pre-bound 125I-labelled insulin was allowed to dissociate in dilution alone or dilution in the presence of increasing amounts of non-radioactive ligand for 0 min (●), 7.5 min (■), 15 min (▲), 30 min (▼), 60 min (◆) or 120 min (○). The data were plotted as bound 125I-labelled insulin divided by bound 125I-labelled insulin at zero concentration non-radioactive ligand and zero time, as a function of concentration of added unlabelled ligand. Each experiment was performed in duplicate and repeated three times on different days.
Cells expressing IR-A (A) or IR-B (B) were pre-incubated with 125I-labelled insulin for 2 h, and then pre-bound 125I-labelled insulin was allowed to dissociate for various time points in the presence of 0 nM (○), 1.67 × 10−11 M (□), 5.1 × 10−11 M (△), 1.67 × 10−10 M (▽), 5.1 × 10−10 M (●), 1.67 × 10−9 M (■), 5.1 × 10−9 M (▲), 1.67 × 10−8 M (▼) or 5.1 × 10−8 M (◆) non-radioactive insulin. The data in this Figure are the same data presented in Figure 4, although are presented here as the amount of bound 125I-labelled ligand normalized to the amount bound ligand at zero time and concentration as a function of time.
The significance parameter describes how significant the difference between the IR-A and IR-B isoforms for a particular parameter is and is calculated as the difference between the two parameters divided by the sum of their S.D. values. In addition to the significance parameter, the IR-A/IR-B ratios as well as its S.D. values are given.
DISCUSSION
We have shown in the present study that IR-A has a slightly higher affinity (1.5-fold) than IR-B. This agrees well with some, but not all, previous findings [8,18,22]. The higher association rate constant a1 that we determined agrees with the only study that looked at association rate, which was found to be higher for IR-A [18]. As for the dissociation rate, our observation of a higher off-rate by dilution alone agrees with the data of Gu et al. [22] and Yamaguchi et al. [18], but not McClain et al. [27]. Only one study previously examined the dissociation rate in the presence of unlabelled insulin and, similar to the present study, found the IR-A isoform to dissociate faster [22]. Discrepancies between the present study and previous reports could be due to differences in assay conditions, such as the nature of receptor preparation, buffer, pH and temperature. It would be interesting to extrapolate these studies to include a more physiologically relevant temperature. However, internalization of the IR at 37°C is significantly increased in comparison with 15°C, which makes data interpretation difficult unless one uses a more complicated model, which can take into account the higher degree of internalization.
Further analysis of the receptor kinetics demonstrated that the presence of exon 11 in IR-B affects only site 1 of the receptor by decreasing the a1 and d1 rate constants approximately 3- and 2-fold respectively. This can be explained in the light of the IR ectodomain structure [46]. In the sequence of IR-B, exon 11 (718–729 in the numbering in [5]) is located next to the C-terminal peptide (residues 704–719). The C-terminal peptide, together with the L1 module, form site 1 of the receptor and, as can be seen from the crystal structure of the IR ectodomain (IR-A), the exon 11 residues are expected to be exposed on the surface of the IR in the vicinity of site 1. Docking of insulin into the IR by means of molecular dynamics simulations shows that insulin can get access to site 1 only from the outside of the receptor, and, from the inside, the access is expected to be blocked by the residues connecting the two subunits [47]. This means that the exon 11 residues are expected to obstruct binding of insulin to site 1 and thus decrease the rate of association to this site when compared with the IR isoform lacking the exon. By the same reasoning, once insulin is bound to site 1, the residues of exon 11 are expected to obstruct the dissociation of insulin and therefore decrease its dissociation rate from site 1. Although it does not seem possible to predict quantitatively how the association and dissociation rate constants of site 1 are expected to be affected by the presence of exon 11, the proposed explanation appears to be in excellent qualitative agreement with the experimental data shown in the present study.
Thus the results of the present study provide a structural insight into the mechanism of the regulation of IR binding by exon 11.
It is of interest that IR-A has a slightly higher dissociation rate, whereas it is considered by some authors [20] to be more ‘mitogenic’ than IR-B. Previous work has suggested that mitogenicity of insulin analogues correlates with a lower dissociation rate from the IR and an increased residence time [48,49]. Another important factor in this case may be linked to internalization. Jensen et al. [50] have shown that an insulin mimetic peptide, S597, that is not internalized, is a much weaker mitogen than insulin and a poor activator of the MAPK (mitogen-activated protein kinase)/ERK (extracellular-signal-regulated kinase) cascade, suggesting that signalling from the endosome is critical for MAPK/ERK activation and mitogenesis [50,51]. Although the data in the literature about the respective internalization rates of IR-A and IR-B are equivocal, we believe that the elegant recent studies of Giudice et al. [20] using fluorescent-tagged ligands and receptors convincingly demonstrate an enhanced internalization of IR-A compared with IR-B in transfected HeLa cells, with a resulting enhanced activation of ERK1/2 by IR-A, whereas IR-B was more potent than IR-A in activating Akt.
Much remains to be done to establish the functional significance of the alternatively spliced forms of the IR and the hybrids thereof.
AUTHOR CONTRIBUTION
Pierre De Meyts designed the study, Louise Knudsen performed the experiments and wrote the paper, Vladislav V. Kiselyov developed the mathematical model and performed the computer curve-fitting of the data. All authors discussed the results and conclusions.
FUNDING
This work was supported by the Danish Ministry of Science, Technology and Innovation (industrial Ph.D. scholarship to L.K.) and Corporate Research Affairs (CORA) within Novo Nordisk A/S (to L.K.).
Acknowledgments
We thank Bo Falck Hansen for providing the CHO cells and Jesper Bøggild Kristensen for the radioactive labelling of insulin.
Footnotes
Abbreviations: a1, association rate constant of site 1; a2, association rate constant of site 2; CHO, Chinese-hamster ovary; d1, dissociation rate constant of site 1; d2, dissociation rate constant of site 2; ERK, extracellular-signal-regulated kinase; HBB, Hepes binding buffer; IGF, insulin-like growth factor; IGF-IR, type I IGF receptor; IR, insulin receptor; IR-A, IR lacking exon 11; IR-B, IR with exon 11; Kcr, cross-linking constant; MAPK, mitogen-activated protein kinase
- © The Authors Journal compilation © 2011 Biochemical Society
References
December 2011
Volume: 440 Issue: 3
Actions
Open in Utopia Docs
