**1. Introduction**

The active form of vitamin D<sup>3</sup> (1α,25(OH)2D3) plays essential roles in calcium and phosphate homeostasis, cellular proliferation and differentiation, and immune responses. Since it could cause hypercalcemia and hypercalciuria, its clinical utility is limited [1,2]. A huge number of vitamin D derivatives have been synthesized. Many of them have been studied in clinical trials for the treatment of type I rickets, osteoporosis, psoriasis, renal osteodystrophy, and also leukemia, pancreatic, prostate, and breast cancers [3–9]. A number of vitamin D derivatives have been approved by the FDA for clinical use in a variety of disorders, for example, 22-oxacalcitriol (Maxacalcitol) and calcipotriol (Dovonex) for treatment of psoriasis, 19-nor-1α,25(OH)2D<sup>2</sup> (Zemplar), 26,26,26,27,27,27-hexafluoro-1α,25(OH)2D<sup>3</sup> (Falecalcitriol), and doxercalciferol (Hectorol) for secondary hyperparathyroidism, and 1α(OH)D<sup>3</sup> (alfacalcidol) and eldecalcitol (Edirol) for osteoporosis. Although many vitamin D derivatives have antiproliferative activity, none have been approved for cancer treatment. So far, only a small number of clinical studies have taken place, such as EB1089 in a phase II study for pancreatic cancer [6,10], and Hectorol and Zemplar in phase I/II advanced androgen-insensitive prostate cancer trials [7,11,12]. Unfortunately, neither have produced any significant objective responses. However, a new 1α,25(OH)2D<sup>3</sup> analog, 19-nor-14-epi-23-yne-1α,25(OH)2D<sup>3</sup> (inecalcitol), is being developed for prostate cancers and chronic leukemia [13,14].

**Citation:** Yasuda, K.; Nishikawa, M.; Mano, H.; Takano, M.; Kittaka, A.; Ikushiro, S.; Sakaki, T. Development of In Vitro and In Vivo Evaluation Systems for Vitamin D Derivatives and Their Application to Drug Discovery. *Int. J. Mol. Sci.* **2021**, *22*, 11839. https://doi.org/10.3390/ ijms222111839

Academic Editor: Enikö Kallay

Received: 1 September 2021 Accepted: 26 October 2021 Published: 31 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The active form of vitamin D<sup>3</sup> (1α,25(OH)2D3) plays essential roles in calcium and phosphate homeostasis, cellular proliferation and differentiation, and immune responses. Its clinical utility is limited because it can cause hypercalcemia and hypercalciuria. [1,2]. Several thousand vitamin D derivatives have been synthesized, and many have been studied in clinical trials to treat conditions, including type I rickets, osteoporosis, leukemia, psoriasis, renal osteodystrophy, and pancreatic, prostate, and breast cancers. [3–9]. A number of vitamin D derivatives have been approved by the FDA for clinical use in a variety of disorders. These derivatives include calcipotriol (Dovonex; Leo Pharmaceuticals) and 22-oxacalcitriol (Maxacalcitol; Chugai Pharmaceuticals) for treatment of psoriasis; 19-nor-1α,25(OH)2D<sup>2</sup> (Zemplar; Abbot Laboratories; Chicago, IL, USA), 26,26,26,27,27,27 hexafluoro- (Falecalcitriol; Sumitomo Pharmaceuticals and Taisho Pharmaceuticals), and doxercalciferol (Hectorol; Bone Care Int.; Middleton, WI, USA) for secondary hyperparathyroidism; and 1α(OH)D<sup>3</sup> (alfacalcidol; Chugai Pharmaceuticals Co., Ltd.; Tokyo, Japan) and eldecalcitol (Chugai Pharmaceuticals Co., Ltd.; Tokyo, Japan) for osteoporosis. Although many vitamin D derivatives, including those approved by the FDA for treating secondary hyperparathyroidism and renal osteodystrophy, have displayed antiproliferative activity, none have been approved for cancer treatment. To date, only a limited number of clinical studies have taken place, including a phase II study of EB1089 in pancreatic cancer. [6,10]. Hectorol and Zemplar have been studied in phase I/II advanced androgen-insensitive prostate cancer trials [7,11,12]. Unfortunately, neither produced any significant objective responses. Recently, a new 1α,25(OH)2D<sup>3</sup> analog, inecalcitol, is being developed for prostate cancers and chronic leukemia [13,14].

In evaluating these vitamin D derivatives, (1) affinity for vitamin D receptor, (2) affinity for vitamin-D-binding protein (DBP), (3) resistance to metabolism by CYP24A1, and (4) ability to differentiate leukemia-derived HL-60 cells into macrophages are considered to be essential properties. In addition, they must show therapeutic efficacy in animal studies. In the case of derivatives under development for cancer treatment, therapeutic efficacy will be evaluated using tumor-bearing animals. Construction of appropriate evaluation models is indispensable for developing vitamin D derivatives for pharmaceutical use. We have developed in vitro systems that can easily measure vitamin D receptor (VDR) affinity [15–19] and CYP24A1-mediated metabolism [20–24]. We have also generated genetically modified rats using genome editing as follows: *Cyp27b1-*gene-deficient rats (a type 1 rickets model animal), vitamin D receptor-gene-deficient rats, and rats harboring a mutant vitamin D receptor (R270L) gene (type II rickets model animals) [25]. We have also generated *Cyp24a1* gene-deficient rats to elucidate enzymes and metabolic pathways responsible for vitamin D derivative metabolism [26]. In this review, we describe the in vitro and in vivo systems we have developed for evaluation of vitamin D derivatives, and discuss the derivatives we have synthesized to date.

#### **2. In Vitro System to Easily Examine the Affinity for VDR of Vitamin D Derivatives**

*2.1. Measurement of Binding Affinity of Vitamin D Derivatives for VDR*

The widely used method for evaluating the binding ability of vitamin D derivatives for VDR in a cell-based assay system is a reporter assay that induces expression of luciferase (Luc) under the control of a promoter containing a vitamin D response element (VDRE) [27–29]. It is noted that it takes more than 12 h for the reporter protein to be expressed, and the direct binding between the receptor and the ligand cannot be evaluated. Although a competitive system using native VDR and tritium-labeled 1α,25(OH)2D<sup>3</sup> was widely used, it is no longer commercially available. Thus, we tried to develop a new detection system that easily evaluates the affinity of vitamin D derivatives for VDR in a short time. We focused on the split-type luciferase technology [15–19,30–37]. This system can evaluate the affinity of the ligand by increasing or decreasing the luminescence of the split-type luciferase.

### *2.2. Development of a Novel Bioluminescent Sensor to Detect and Discriminate between Vitamin D Receptor Agonists and Antagonists in Living Cells (1st Generation)*

Two chimeric fusion proteins that contained both split-luciferase and the ligand binding domain (LBD) of the VDR were constructed. This fusion protein was labeled as LucN–LBD–LucC. It contained the N-terminal domain taken from luciferase (LucN), LBD, and C-terminal domain from luciferase (LucC) from N-terminus to C-terminus. LucC– LBD–LucN has the C-terminal domain of luciferase at the N-terminus of the fusion protein (Figure 1) [15]. Unexpectedly, the LucC–LBD–LucN worked better than LucN–LBD–LucC. Luciferase activity was significantly diminished by the addition of the VDR agonists to COS-7 cells that expressed LucC–LBD–LucN. On the other hand, the VDR antagonist notably enhanced the activity of the chimeric luciferase in a dose- and time-dependent manner. Our novel model for detecting and discriminating between VDR agonists and antagonists is very useful for testing synthetic analogs of vitamin D that show reasonable affinity for normal or mutant VDRs. *Int. J. Mol. Sci.* **2021**, *22*, x FOR PEER REVIEW 4 of 16

**Figure 1.** Schematic diagrams of the biosensors to detect VDR ligands. (**A**) 1st generation. Binding of the VDR agonists to the LBD may cause a conformational change of the LBD that leads to disruption of the functional complex between Nterminal and C-terminal domains of the luciferase. In contrast, binding of the antagonist leads to the reassembly of Nterminal and C-terminal domains of the luciferase to increase the activity. (**B**) 2nd generation. Binding of VDR ligands to the biosensor may cause a conformational change of helix12 (H12) in LBD. After conformational change of LBD, the LXXLL motif interacts with LBD in the biosensor. Then, this intramolecular dynamic change of the WT biosensor leads to reconstitution of the functional complex between LucN and LucC fragments of the split luciferase. (**C**) 3rd generation. Binding of the VDR ligands to the LBD–LucC may cause a positional change of helix12 in LBD. Then, the LucN–LXXLL and LBD– LucC forms a functional complex to exhibit the luciferase activity. **Figure 1.** Schematic diagrams of the biosensors to detect VDR ligands. (**A**) 1st generation. Binding of the VDR agonists to the LBD may cause a conformational change of the LBD that leads to disruption of the functional complex between N-terminal and C-terminal domains of the luciferase. In contrast, binding of the antagonist leads to the reassembly of N-terminal and C-terminal domains of the luciferase to increase the activity. (**B**) 2nd generation. Binding of VDR ligands to the biosensor may cause a conformational change of helix12 (H12) in LBD. After conformational change of LBD, the LXXLL motif interacts with LBD in the biosensor. Then, this intramolecular dynamic change of the WT biosensor leads to reconstitution of the functional complex between LucN and LucC fragments of the split luciferase. (**C**) 3rd generation. Binding of the VDR ligands to the LBD–LucC may cause a positional change of helix12 in LBD. Then, the LucN–LXXLL and LBD–LucC forms a functional complex to exhibit the luciferase activity.

*Vitamin D Receptor Agonists and Antagonists* 

cancers, and immune disorders.

*2.3. Development of a Highly Sensitive In Vitro System to Detect and Discriminate between* 

LBD–LucN proteins expressed in *Escherichia coli (E. coli)* cells [17]. It should be noted that this system could be completed within 30 min, and its activity was unchanged after 10 freeze–thaw cycles. This highly sensitive and convenient system would be quite useful to screen VDR ligands with therapeutic potential for osteoporosis, renal osteodystrophy,

Patients with type II rickets showing the R274L mutation caused a 1000-fold reduction in the binding activity for 1α,25(OH)2D<sup>3</sup> and remarkably lowered vitamin-D-related gene expression [38]. It is Arg274, located in LBD of VDR, that is responsible for attaching 1α,25(OH)2D3. This happens by a formation of an additional hydrogen bond with 1αhydroxyl of 1α,25(OH)2D3. LucC–LBD (R274L)–LucN was constructed to investigate vitamin D ligands of high affinity for the mutant VDR (R274L). A total of 5 out of the 33 vitamin D analogs tested showed much higher binding for the mutant VDR (R274L) than the vitamin D hormone. The highest binding activity was shown by 2α-(2-(tetrazol-2-yl)ethyl)-(AH-1). These analogs might be considered as future drug candidates against HVDRR that is caused by the mutant VDR (R274L) [16].

#### *2.3. Development of a Highly Sensitive In Vitro System to Detect and Discriminate between Vitamin D Receptor Agonists and Antagonists*

We have established an in vitro screening system for VDR ligands using the LucC– LBD–LucN proteins expressed in *Escherichia coli (E. coli)* cells [17]. It should be noted that this system could be completed within 30 min, and its activity was unchanged after 10 freeze–thaw cycles. This highly sensitive and convenient system would be quite useful to screen VDR ligands with therapeutic potential for osteoporosis, renal osteodystrophy, cancers, and immune disorders.

### *2.4. Design of a Biosensor Based on Split Luciferase for Detection of VDR Ligands (2nd Generation)*

The model we developed is very useful for a fast investigation of VDR ligands. However, the sensitivity of our biosensor (LucC–LBD–LucN) is not as high as expected. LBD is known to interact via the LXXLL motif with transcription coactivators, such as SRC-1, TIF-2, or DRIP-205 to initiate vitamin-D-related gene expression, when binding natural VDR ligands. This is why we anticipated that it is the LXXLL motif that changes the enzymatic profile of luciferase–LBD biosensors. This is why LucN–LBD–LucC and not LucC–LBD–LucN was used as a basing fragment. We created a new biosensor consisting of the LBD (121–427 aa) of VDR, N- and C-terminal of firefly luciferase fragments (LucN (1–415 aa) and LucC (416-550 aa)), the LXXLL peptide sequence, and peptide sequence (Gly-Gly-Gly-Gly-Ser (GGGGS)) × 3 as the flexible linker [18]. This construct we labeled as LucN–LXXLL–(GGGGS) × 3–LBD–LucC WT biosensor and WT means the wild-type of LBD (Figure 1). Light intensity of luciferase is low when natural VDR ligands are absent. The luciferase light intensity is immediately and remarkably increased when the ligand is bound to the WT biosensor. To sum up, we have successfully created a very sensitive biosensor which shows the increase in light intensity when binding VDR agonists.

To this end, we developed a novel and WT biosensor of high sensitivity by examining three types of LXXLL peptides (NHPMLMNLLKDN, LTEMHPILTSLLQNGVDHV, and LSETHPLLWTLLSSTEGDSM) that interact with the LBD in response to 1α,25(OH)2D<sup>3</sup> or synthetic VDR agonists. The COS-7 cells that expressed each type of biosensor were treated with 1α,25(OH)2D<sup>3</sup> (100 nM) and then the luminescence was measured 90 min later. Among the 10 biosensors we constructed, one showed a reduction in intensity of light in response to 1α,25(OH)2D3. Seven biosensors showed an excellent increase in light intensity. Our best biosensor showed the light intensity ca. one-third of that of full-length native luciferase of firefly. Quite unexpectedly, 25(OH)D3, as the low-affinity VDR ligand, also enhanced the intensity of light in a concentration-dependent manner. The half maximal relative intensity of light was recorded at 1 nM of 1α,25(OH)2D<sup>3</sup> and at 20 nM of 25(OH)D3, respectively. We then compared the binding activity of 1α,25(OH)2D<sup>3</sup> and 25(OH)D<sup>3</sup> for the mutant VDR (R274L). As previously mentioned, the substitution of Arg274 to Leu causes a 1000-fold decrease in affinity of 1α,25(OH)2D3. As expected, in the R274L biosensor the concentration–response curve of 1α,25(OH)2D<sup>3</sup> was very similar to that of 25(OH)D3. Thus, the biosensor system we developed may be very useful in elucidating novel vitamin D analogs as drug candidates against type II rickets resulting from VDR mutation, such as R274L.

#### *2.5. Development of a Novel Two-Molecule System with a Highly Sensitive Biosensor (3rd Generation)*

In the next step, we developed a two-molecule system named LXXLL + LBD biosensor, as shown in Figure 1, with a combination of two components [19]. The two plasmids were co-transfected and two proteins were co-expressed in COS-7 cells. The LXXLL + LBD biosensor-expressing COS-7 cells were treated with 100 nM of 1α,25(OH)2D3, and luciferase light intensity was measured at 90 min after treatment. Among all combinations of LXXLL + LBD biosensor, relative light intensity of A1 + B1 [19] was the highest in all combinations. The relative light intensity of combination A1 + B1 was approximately a 90- to 100-fold increase in response to 100 nM of 1α,25(OH)2D3. It should be noted that the detection limit was 0.005 nM (5 pM) of 1α,25(OH)2D3, indicating that the sensitivity of LXXLL + LBD biosensor is higher than that of our previous biosensors. [15–18]. Our LXXLL + LBD biosensor might be used for the measurement of 1α,25(OH)2D<sup>3</sup> and 25(OH)D<sup>3</sup> in the plasma.

#### **3. In Vitro Evaluation of CYP24A1-mediated Metabolism of Vitamin D Derivatives**

#### *3.1. Expression of Rat or Human CYP24A1 in E. coli Cells*

The rat *Cyp24a1* cDNA was cloned from the rat kidney cDNA library [39], and the isolated cDNA clone contained the open reading frame consisting of 514 amino acids. Since the amino acid sequence showed less than 40 % homology with already known CYPs, the new CYP family name, CYP24, was given to this vitamin-D-24-hydroxylase.

The molecular mechanism of *CYP24A1* gene regulation is quite complicated, and many factors are tissue-specifically involved in the expression of *CYP24A1* [40–44]. These facts strongly suggest that CYP24A1 is a physiologically essential enzyme that regulates the level of the active form of vitamin D.

When the deduced amino acid sequence from its cDNA was compared to that aminoterminal amino acid sequence of the CYP24A1 purified from rat kidney, it was found that the mature form of rat CYP24A1 lacks amino-terminal 32 amino acids. These results suggest that amino-terminal 32 amino acids function as a mitochondrial targeting signal, which is removed after translocation of CYP24A1 to mitochondria. We have successfully expressed the mature forms of rat and human CYP24A1 in *E. coli* cells to reveal their enzymatic properties [20–24].

#### *3.2. Construction of a CYP24A1 Enzyme System Containing Adrenodoxin (ADX) and NADPH-Adrenodoxin Reductase (ADR)*

The mitochondrial P450 system consists of three components: CYP, ADX, and ADR. Electrons are sequentially transferred from NADPH through ADR and ADX to CYP24A1 (Figure 2). Thus, CYP24A1-dependent activity was measured in an in vitro reconstituted system containing purified ADX and ADR proteins. On the other hand, in a whole-cell system, co-expression of mature forms of CYP24A1, ADX, and ADR in *E. coli* is required. We have demonstrated that the *E. coli* expression system is quite useful to investigate enzymatic properties of CYP24A1. Using this *E. coli* expression system, we have determined kinetic parameters of CYP24A1 in the metabolism of the native vitamin D and various vitamin D derivatives, and revealed their metabolic pathways [45–58].
