*Article* **Preparation of Biphenyl-Conjugated Bromotyrosine for Inhibition of PD-1**/**PD-L1 Immune Checkpoint Interactions**

**Eun-Hye Kim 1,2, Masuki Kawamoto 1,3,\*, Roopa Dharmatti <sup>1</sup> , Eiry Kobatake <sup>2</sup> , Yoshihiro Ito 1,3 and Hideyuki Miyatake 1,\***


Received: 7 April 2020; Accepted: 17 May 2020; Published: 21 May 2020

**Abstract:** Cancer immunotherapy has been revolutionized by the development of monoclonal antibodies (mAbs) that inhibit interactions between immune checkpoint molecules, such as programmed cell-death 1 (PD-1), and its ligand PD-L1. However, mAb-based drugs have some drawbacks, including poor tumor penetration and high production costs, which could potentially be overcome by small molecule drugs. **BMS-8**, one of the potent small molecule drugs, induces homodimerization of PD-L1, thereby inhibiting its binding to PD-1. Our assay system revealed that **BMS-8** inhibited the PD-1/PD-L1 interaction with IC<sup>50</sup> of 7.2 µM. To improve the IC<sup>50</sup> value, we designed and synthesized a small molecule based on the molecular structure of **BMS-8** by in silico simulation. As a result, we successfully prepared a biphenyl-conjugated bromotyrosine (**X**) with IC<sup>50</sup> of 1.5 µM, which was about five times improved from **BMS-8**. We further prepared amino acid conjugates of **X** (**amino-X)**, to elucidate a correlation between the docking modes of the **amino-X**s and IC<sup>50</sup> values. The results suggested that the displacement of **amino-X**s from the **BMS-8** in the pocket of PD-L1 homodimer correlated with IC<sup>50</sup> values. This observation provides us a further insight how to derivatize **X** for better inhibitory effect.

**Keywords:** PD-1/PD-L1; immune checkpoint inhibitors; biphenyl-conjugated bromotyrosine; amino acid conjugation; **amino-X**; in silico simulation; IC<sup>50</sup>

### **1. Introduction**

Immunotherapy has recently emerged as a fourth modality for cancer therapy, together with surgery, chemotherapy, and radiation therapy [1–4]. The immunotherapy promotes T-cells to kill cancer cells by the blockade of immune checkpoint pathways [5,6]. One of the major immune checkpoint pathways is inactivated by the binding of programmed cell-death 1 (PD-1) [7], which is largely expressed on T cells, and its ligand PD-L1 [3,8,9], which is mainly expressed on antigen-presenting cells under physiological conditions but is upregulated on cancer cells [10]. PD-L1 binding to PD-1 suppresses T-cell function, including cytolytic activity, leading to downregulation of the anti-tumor immune response [2,5]. Another immune checkpoint is mediated by binding of the ligands B7-1/2

(CD80, CD86) on activated antigen-presenting cells or cancer cells to cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) on T cells, which also suppresses T-cell activity [11,12]. Identification of these immunosuppressive pathways led to the development of monoclonal antibody (mAb)-based cancer therapies that inhibit PD-1/PD-L1 or CTLA-4/B7 pathways, thereby reinvigorating the host anti-tumor immune response [2,13–17]. Among the therapies currently approved for clinical use are the anti-CTLA-4 mAb ipilimumab (Yervoy®), which was the first immune checkpoint inhibitor to demonstrate an anti-cancer effect [18,19], and the anti-PD-1 mAb nivolumab (Opdivo®) [20]. In addition to these and other approved mAb-based immune checkpoint inhibitors [21], many others are currently in clinical trials for various cancers and immune-based diseases [22–25]. *Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 2 of 17 associated protein 4 (CTLA-4) on T cells, which also suppresses T-cell activity [11,12]. Identification of these immunosuppressive pathways led to the development of monoclonal antibody (mAb)-based cancer therapies that inhibit PD-1/PD-L1 or CTLA-4/B7 pathways, thereby reinvigorating the host anti-tumor immune response [2,13–17]. Among the therapies currently approved for clinical use are the anti-CTLA-4 mAb ipilimumab (Yervoy®), which was the first immune checkpoint inhibitor to

Protein-based drugs such as mAbs have some important drawbacks, such as high production costs associated with the preparation of biologicals [26], poor tumor penetration due to their large molecular weights (~150 kDa) [27], and unexpected post-translational glycosylation patterns [28]. Small molecule drugs, which are generally orally active and can overcome many of the challenges associated with protein drugs, are therefore being pursued as attractive alternative immune checkpoint inhibitors [28,29]. demonstrate an anti-cancer effect [18,19], and the anti-PD-1 mAb nivolumab (Opdivo®) [20]. In addition to these and other approved mAb-based immune checkpoint inhibitors [21], many others are currently in clinical trials for various cancers and immune-based diseases [22–25]. Protein-based drugs such as mAbs have some important drawbacks, such as high production costs associated with the preparation of biologicals [26], poor tumor penetration due to their large molecular weights (~150 kDa) [27], and unexpected post-translational glycosylation patterns [28]. Small molecule drugs, which are generally orally active and can overcome many of the challenges

Until now, Bristol-Myers Squibb (BMS) has disclosed the patent claim [30] with structures of a number of BMS compounds, which are the potential inhibitors of the PD-1/PD-L1 pathway. Previous works have shown that one of the BMS compounds, **BMS-8**, binds directly to PD-L1 and induces formation of PD-L1 homodimers, which in turn prevents the interaction with PD-1 [31]. In the patent claims, the homogenous time-resolved fluorescence (HTRF) assay report that **BMS-8** has a sub µM order of IC50, 0.146 µM [30], with other BMS compounds [32]. In this study, however, our amplified luminescence proximity homogeneous assay (Alpha) measured the IC<sup>50</sup> of **BMS-8** as 7.2 µM. Therefore, we aimed to prepare higher affinity compounds by taking the advantage of the complex structure of **BMS-8**/PD-L1 [31] with in silico simulation [33–35]. Figure 1 shows our strategies to improve the affinity of **BMS-8**. We used fragmented structures of 3-hydroxymethyl-2-methylbiphenyl (**1**) and 3-bromotyrosine (**2**). After conjugation of **1** and **2**, a biphenyl-conjugated bromotyrosine (denoted as **X**) was synthesized. Because an amino and carboxyl group included in **X**, it could be conjugated to various amino acids. [36,37]. During the procedures, we employed in silico simulation and IC<sup>50</sup> assay to reveal molecular mechanism of the inhibition. associated with protein drugs, are therefore being pursued as attractive alternative immune checkpoint inhibitors [28,29]. Until now, Bristol-Myers Squibb (BMS) has disclosed the patent claim [30] with structures of a number of BMS compounds, which are the potential inhibitors of the PD-1/PD-L1 pathway. Previous works have shown that one of the BMS compounds, **BMS-8**, binds directly to PD-L1 and induces formation of PD-L1 homodimers, which in turn prevents the interaction with PD-1 [31]. In the patent claims, the homogenous time-resolved fluorescence (HTRF) assay report that **BMS-8** has a sub μM order of IC50, 0.146 μM [30], with other BMS compounds [32]. In this study, however, our amplified luminescence proximity homogeneous assay (Alpha) measured the IC50 of **BMS-8** as 7.2 μM. Therefore, we aimed to prepare higher affinity compounds by taking the advantage of the complex structure of **BMS-8**/PD-L1 [31] with in silico simulation [33–35]. Figure 1 shows our strategies to improve the affinity of **BMS-8**. We used fragmented structures of 3-hydroxymethyl-2 methylbiphenyl (**1**) and 3-bromotyrosine (**2**). After conjugation of **1** and **2**, a biphenyl-conjugated bromotyrosine (denoted as **X**) was synthesized. Because an amino and carboxyl group included in **X**, it could be conjugated to various amino acids. [36,37]. During the procedures, we employed in silico simulation and IC50 assay to reveal molecular mechanism of the inhibition.

**Figure 1.** Strategies to improve inhibitory effect of **BMS-8**. 3-hydroxymethyl-2-methylbiphenyl (**1**) and 3-bromotyrosine (**2**) were selected as fragmented structures. A biphenyl-conjugated bromotyrosine **X** was synthesized after conjugation of **1** and **2**. We conjugated a variety of amino acids as additions, to the amino- and carboxyl-groups of **X** to reveal molecular mechanism of the inhibition. **Figure 1.** Strategies to improve inhibitory effect of **BMS-8**. 3-hydroxymethyl-2-methylbiphenyl (**1**) and 3-bromotyrosine (**2**) were selected as fragmented structures. A biphenyl-conjugated bromotyrosine **X** was synthesized after conjugation of **1** and **2**. We conjugated a variety of amino acids as additions, to the amino- and carboxyl-groups of **X** to reveal molecular mechanism of the inhibition.

#### **2. Results 2. Results**

#### *2.1. In Silico Docking Simulation and Organic Chemistry Synthesis of a Biphenyl-Conjugated Bromotyrosine Bromotyrosine*

*2.1. In Silico Docking Simulation and Organic Chemistry Synthesis of a Biphenyl-Conjugated* 

We designed a biphenyl-conjugated bromotyrosine (denoted as **X**), based on the **BMS-8**. We docked **X** into the crystal structure of **BMS-8**/PD-L1AB complex (PDB ID: 5J8O) [31] using ICM 3.8-7 software (Molsoft L.L.C., San Diego, CA, USA) [33–35], without guidance and induced fitting to avoid over-fitting. We obtained the docking score of −42.96 for **X**, which was the same order of **BMS-8**, −49.5 (Table 1). Based on the scores, we confirmed the potential of **X** for inhibition. Therefore, we synthesized **X** by the organic chemistry procedures. Scheme 1 shows the synthetic route for a biphenyl-bromotyrosine **6**. Full synthesis details are provided in Materials and Methods. The C- and N-terminals of 3-bromotyrosine (**2**) were first protected by *tert*-butyl and fluorenylmethyloxycarbonyl (Fmoc) groups, respectively, to produce the amino acid **4**, which was then reacted with 3-hydroxymethyl-2-methylbiphenyl (**1**) through the Mitsunobu reaction to yield compound **5**. Deprotection of the *tert*-butyl group in compound **5** produced the Fmoc-protected amino acid **6**. Deprotection of the Fmoc group in **6** yielded the compound **X**. Peptide conjugates were obtained by solid-state peptide synthesis using compound **6**. <sup>1</sup>H NMR spectra of the compounds are shown in Figures S1–S4. A summary of the analytical data for the synthesized compounds is given in Table S1. The analytical data indicate the successful synthesis of **X** and 29 **amino-X** derivatives consisting of 2-mers (**GX**, **XG**, **XS**, **XR**, **XA**, **XW**), 3-mers (**YXC**, **WXG**, **QXQ**, **CXA**, **RXN**, **SXR**, **NXR**, **CXR**, **GXG**, **XNL**, **XNH**, **XHP**, **XGG**), 4-mers (**XCSE**, **XGGG**), 5-mers (**WRXNN**, **ERXNK**, **WRXNQ**, **XRRRR**, **XGGGG**), 6-mer (**XGGGGG**), and 7-mers (**CERXNKM**, **FWRXNNI**). We designed a biphenyl-conjugated bromotyrosine (denoted as **X**), based on the **BMS-8**. We docked **X** into the crystal structure of **BMS-8**/PD-L1AB complex (PDB ID: 5J8O) [31] using ICM 3.8-7 software (Molsoft L.L.C., San Diego, CA, USA) [33–35], without guidance and induced fitting to avoid over-fitting. We obtained the docking score of −42.96 for **X**, which was the same order of **BMS-8**, −49.5 (Table 1). Based on the scores, we confirmed the potential of **X** for inhibition. Therefore, we synthesized **X** by the organic chemistry procedures. Scheme 1 shows the synthetic route for a biphenyl-bromotyrosine **6**. Full synthesis details are provided in Materials and Methods. The C- and N-terminals of 3-bromotyrosine (**2**) were first protected by *tert*-butyl and fluorenylmethyloxycarbonyl (Fmoc) groups, respectively, to produce the amino acid **4**, which was then reacted with 3-hydroxymethyl-2-methylbiphenyl (**1**) through the Mitsunobu reaction to yield compound **5**. Deprotection of the *tert*-butyl group in compound **5** produced the Fmoc-protected amino acid **6**. Deprotection of the Fmoc group in **6** yielded the compound **X**. Peptide conjugates were obtained by solid-state peptide synthesis using compound **6**. 1H NMR spectra of the compounds are shown in Figures S1-S4. A summary of the analytical data for the synthesized compounds is given in Table S1. The analytical data indicate the successful synthesis of **X** and 29 **amino-X** derivatives consisting of 2-mers (**GX**, **XG**, **XS**, **XR**, **XA**, **XW**), 3-mers (**YXC**, **WXG**, **QXQ**, **CXA**, **RXN**, **SXR**, **NXR**, **CXR**, **GXG**, **XNL**, **XNH**, **XHP**, **XGG**), 4-mers (**XCSE**, **XGGG**), 5-mers (**WRXNN**, **ERXNK**, **WRXNQ**, **XRRRR**, **XGGGG**), 6-mer (**XGGGGG**), and 7-mers (**CERXNKM**, **FWRXNNI**).

**Scheme 1.** Synthetic scheme for the biphenyl-conjugated bromotyrosine **6**. **Scheme 1.** Synthetic scheme for the biphenyl-conjugated bromotyrosine **<sup>6</sup>**.


**Table 1.** Docking simulation and IC<sup>50</sup> measurements of **BMS-8** and **amino-X**s.

#### *2.2. Inhibition Assays of PD-1*/*PD-L1 Binding by BMS-8 and X*

To evaluate the binding affinities of the compounds for PD-L1, we used the amplified luminescence proximity homogeneous assay (Alpha) by using the AlphaLISA® assay kit [38]. This assay is based on photoinduced energy transfer between donor and acceptor beads conjugated to PD-1 and PD-L1, respectively (Figure S6).

The AlphaLISA® assay revealed that the intermediates of **X**, compounds **1–6**, showed a few hundred µM or weaker IC<sup>50</sup> values (Figure 3). **BMS-8** inhibited the PD-1/PD-L1 interaction with IC<sup>50</sup> of 7.2 µM (Figure 2), which was weaker than that previously reported, IC<sup>50</sup> of 0.146 µM [30]. On the other hand, nivolumab showed nano-molar order of inhibition (IC<sup>50</sup> = 5.1 nM, Figure 2), corresponding to the previously reported value [39], which suggests the validity of our assay system.

*Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 4 of 17

To evaluate the binding affinities of the compounds for PD-L1, we used the amplified luminescence proximity homogeneous assay (Alpha) by using the AlphaLISA® assay kit [38]. This assay is based on photoinduced energy transfer between donor and acceptor beads conjugated to PD-

The AlphaLISA® assay revealed that the intermediates of **X**, compounds **1–6**, showed a few hundred μM or weaker IC50 values (Figure 3). **BMS-8** inhibited the PD-1/PD-L1 interaction with IC50 of 7.2 μM (Figure 2), which was weaker than that previously reported, IC50 of 0.146 μM [30]. On the

corresponding to the previously reported value [39], which suggests the validity of our assay system.

**Figure 2.** Inhibition of PD-1/PD-L1 interaction by **BMS-8 a**nd nivolumab measured by the AlphaLISA® assay. **Figure 2.** Inhibition of PD-1/PD-L1 interaction by **BMS-8** and nivolumab measured by the AlphaLISA® assay.

#### *2.3. Fragmentation of BMS-8 and Conjugation of Compounds to Prepare X 2.3. Fragmentation of BMS-8 and Conjugation of Compounds to Prepare X*

*2.2. Inhibition Assays of PD-1/PD-L1 Binding by BMS-8 and X* 

1 and PD-L1, respectively (Figure S6).

To prepare higher affinity compounds based on **BMS-8**, we first considered a scenario that smaller groups of **BMS-8**, compounds **1**–**6** (Scheme 1), showed better inhibitory effect for PD-1/PD-L1 PPI. The docking scores of the compounds, however, were larger than that of **BMS-8** (−49.5), suggesting pooper inhibition effect. Actually, AlphaLISA assay revealed that the IC50 values were a few hundred μM, which were much weaker than that of **BMS-8** (7.2 μM) (Figure 3). Therefore, we considered the next scenario of conjugation of compounds; we conjugated To prepare higher affinity compounds based on **BMS-8**, we first considered a scenario that smaller groups of **BMS-8**, compounds **1**–**6** (Scheme 1), showed better inhibitory effect for PD-1/PD-L1 PPI. The docking scores of the compounds, however, were larger than that of **BMS-8** (−49.5), suggesting pooper inhibition effect. Actually, AlphaLISA assay revealed that the IC<sup>50</sup> values were a few hundred µM, which were much weaker than that of **BMS-8** (7.2 µM) (Figure 3). *Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 5 of 17

compound **4** and compound **1** to prepare biphenyl-bromotyrosine (**X**), which resembled **BMS-8**

**Figure 3**. Docking scores, IC50 values and measurements of compounds **1–6**. All compounds showed larger scores than that of **X** (score = −42.96) with a few hundred μM of IC50 values. **Figure 3.** Docking scores, IC<sup>50</sup> values and measurements of compounds **1–6**. All compounds showed larger scores than that of **X** (score = −42.96) with a few hundred µM of IC<sup>50</sup> values.

 ten sity IC50 = 1.5 μM Therefore, we considered the next scenario of conjugation of compounds; we conjugated compound **4** and compound **1** to prepare biphenyl-bromotyrosine (**X**), which resembled **BMS-8** except the terminal amino- and carboxyl-groups. In turn, **X** showed a docking score of −42.96, comparable to that of **BMS-8** (−49.5). In fact, **X** inhibited PD-1/PD-L1 PPI with IC<sup>50</sup> = 1.5 µM (Figure 4), which was five times better than that of **BMS-8** (7.2 µM).

**Figure 4.** AlphaLISA assay of **X**. **X** shows IC50 = 1.5 μM with docking score = −42.96.

The binding mode of the BMS compounds and derivatives to PD-L1 has previously been revealed by X-ray crystallography [31,40–42]. BMS compounds induces transient homodimerization of PD-L1AB on the binding, which masks the binding site for PD-1 located in the homodimerization interface. We docked **amino-X**s to the crystal structure of **BMS-8**/PD-L1AB complex (PDB ID: 5J8O) [31], using ICM 3.8-7 software (Molsoft L.L.C., San Diego, CA, USA) [33–35], without guidance and induced fitting to avoid over-fitting. After the docking, we calculated the root mean square deviation (RMSD) of distances between atoms in compound **BMS-8** and **X**, excluding Cα, NH2, and COOH

X (m M )

*2.4. Docking Simulation and Inhibition Assay of Amino-Xs* 

In

atoms (Figure 5).

**Figure 3**. Docking scores, IC50 values and measurements of compounds **1–6**. All compounds showed larger

scores than that of **X** (score = −42.96) with a few hundred μM of IC50 values.

*Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 5 of 17

Br HO

C om pound1 (m M )

O O NH2

<sup>O</sup> Br O O HN O

OH OH Br

<sup>O</sup> O

C om pound5 (m M ) C om pound6 (m M )

C om pound3 (m M ) C om pound4 (m M )

HO

<sup>O</sup> Br HO

O HN O O

> Br OH O <sup>N</sup> <sup>H</sup> O O

O NH2

C om pound2 (m M )

**Figure 4.** AlphaLISA assay of **X**. **X** shows IC50 = 1.5 μM with docking score = −42.96. **Figure 4.** AlphaLISA assay of **X**. **X** shows IC<sup>50</sup> = 1.5 µM with docking score = −42.96.

#### *2.4. Docking Simulation and Inhibition Assay of Amino-Xs 2.4. Docking Simulation and Inhibition Assay of Amino-Xs*

Com pound Score IC50 (**μ**M )

Compound1 -35.82 309.1

Compound2 -10.05 N/V

Compound3 -0.1608 N/V

Compound4 -11.77 1061

Compound5 -29.9 896.9

Compound6 -40.31 1418

The binding mode of the BMS compounds and derivatives to PD-L1 has previously been revealed by X-ray crystallography [31,40–42]. BMS compounds induces transient homodimerization of PD-L1AB on the binding, which masks the binding site for PD-1 located in the homodimerization interface. We docked **amino-X**s to the crystal structure of **BMS-8**/PD-L1AB complex (PDB ID: 5J8O) [31], using ICM 3.8-7 software (Molsoft L.L.C., San Diego, CA, USA) [33–35], without guidance and induced fitting to avoid over-fitting. After the docking, we calculated the root mean square deviation (RMSD) of distances between atoms in compound **BMS-8** and **X**, excluding Cα, NH2, and COOH atoms (Figure 5). The binding mode of the BMS compounds and derivatives to PD-L1 has previously been revealed by X-ray crystallography [31,40–42]. BMS compounds induces transient homodimerization of PD-L1AB on the binding, which masks the binding site for PD-1 located in the homodimerization interface. We docked **amino-X**s to the crystal structure of **BMS-8**/PD-L1AB complex (PDB ID: 5J8O) [31], using ICM 3.8-7 software (Molsoft L.L.C., San Diego, CA, USA) [33–35], without guidance and induced fitting to avoid over-fitting. After the docking, we calculated the root mean square deviation (RMSD) of distances between atoms in compound **BMS-8** and **X**, excluding Cα, NH2, and COOH atoms (Figure 5). *Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 6 of 17

**Figure 5**. Root mean square deviation (RMSD) calculation between amino-X and **BMS-8** bound to PD-L1AB homodimer. After docking of amino-X (in this case, WXG), we calculated RMSD between a part of **WXG** (excluding Cα, amino-group and carboxyl-group) and the corresponding part of **BMS-8**, as shown by the red dotted-rectangle. Table 1 shows the docking scores and RMSD values for **amino-X**s docked to PD-L1AB. Also, the **Figure 5.** Root mean square deviation (RMSD) calculation between amino-X and **BMS-8** bound to PD-L1AB homodimer. After docking of amino-X (in this case, WXG), we calculated RMSD between a part of **WXG** (excluding Cα, amino-group and carboxyl-group) and the corresponding part of **BMS-8**, as shown by the red dotted-rectangle.

IC50 values for the **amino-X**s are listed in Table 1. As a result, they suggested some positive

correlations. The IC50 values of the 1–2-mer **amino-X**s showed moderate correlations with both the RMSDs (CC 0.67, Table 2) and the scores (CC 0.40, Table 2). However, these correlations weakened as the number of conjugated amino acids increased (RMSD from 0.67 to 0 and CC 0.40 to −0.20, Table 2). These results suggest that the current in silico docking worked better for **amino-X**s conjugated with shorter amino acids. To discuss the correlations further, we compared the docking structures of **X** (IC50 = 1.5 μM), **XG** (IC50 = 2.1 μM), and **GX** (IC50 = 448.5 μM). **Table 1.** Docking simulation and IC50 measurements of **BMS-8** and **amino-X**s**.** Table 1 shows the docking scores and RMSD values for **amino-X**s docked to PD-L1AB. Also, the IC<sup>50</sup> values for the **amino-X**s are listed in Table 1. As a result, they suggested some positive correlations. The IC<sup>50</sup> values of the 1–2-mer **amino-X**s showed moderate correlations with both the RMSDs (CC 0.67, Table 2) and the scores (CC 0.40, Table 2). However, these correlations weakened as the number of conjugated amino acids increased (RMSD from 0.67 to 0 and CC 0.40 to −0.20, Table 2). These results suggest that the current in silico docking worked better for **amino-X**s conjugated with shorter amino acids.

**Amino Acid Length Sequence Score RMSD ( ) IC50 (μM) - BMS-8** −49.5 - 7.2 **Table 2.** Correlation coefficients (CC) for IC<sup>50</sup> vs. Score and IC<sup>50</sup> vs. RMSD.


**YXC** −37.1 0.46 465.0 **WXG** −50.6 0.48 404.8 CC values were calculated by the Microsoft Excel.

> **QXQ** −37.7 0.73 1961.0 **CXA** −42.0 0.48 665.0

> **NXR** −50.3 0.46 982.0 **CXR** −41.5 0.54 550.0 **GXG** −43.6 0.39 676.0 **XNL** −43.0 0.58 855.0 **XNH** −40.7 0.50 313.0 **XHP** −33.5 0.55 359.0 **XGG** −36.1 0.57 6505.0

**3** 

To discuss the correlations further, we compared the docking structures of **X** (IC<sup>50</sup> = 1.5 µM), **XG** (IC<sup>50</sup> = 2.1 µM), and **GX** (IC<sup>50</sup> = 448.5 µM).

We compared the binding modes of **BMS-8** and **X** in the pocket of PD-L1AB homodimer (Figure 6). **BMS-8**, with IC<sup>50</sup> of 7.2 µM (Figure 2), binds the pocket with a hydrogen bind to Q66<sup>A</sup> and a hydrophobic interaction with V68<sup>A</sup> (Figure 6A), respectively. On the other hand, **X** forms a hydrogen bond with the hydroxy group of the side chain of Y56A, which stabilizes the binding (Figure 6A), with IC<sup>50</sup> of 1.5 µM (Figure 4). The superposition of **X** onto **BMS-8** showed an RMSD displacement of 0.40 Å (Figure 6B) We conclude that binding of **X** would not markedly impede PD-L1 homodimerization, which is consistent with its relatively low IC<sup>50</sup> value of 1.5 µM (Figure 4). These results suggest that we can improve an IC<sup>50</sup> value by substituting the six-membered group of **BMS-8** with some proper groups, leading to rearrangement of interactions around it. Besides, smaller displacement of biphenyl-bromotyrosine portion shown by RMSD is preferable for higher affinity.

**Figure 6.** Docking conformations of **BMS-8** and **X**. (**A**) The docking modes of **BMS-8** and **X** were revealed by the X-ray crystallography and in silico docking simulation, respectively, which the 2D binding pictures. The 2D figures show that biphenyl portions of the ligands bind into the pocket by hydrophobic interactions shown in light-green color. In contrast, the amino cation at the six-membered ring of **BMS-8** makes a hydrogen bond with the sidechain of Q66<sup>A</sup> in cyan color. In addition, the six-membered ring makes hydrophobic interaction with V68A. On the other hand, amino-group of bromo-tyrosine in **X** makes a hydrogen bonding to the hydroxyl group of Y56<sup>A</sup> colored in cyan, without other hydrophobic interaction, as shown in the 2D picture. (**B**) **BMS-8** and **X** without Cα, NH<sup>2</sup> , and COOH superposed each other with RMSD of 0.40 Å.

Modeling of **XG** identified two potential hydrogen bonds between the N-terminal of **XG** and the side chain of Q66<sup>A</sup> and between the carboxyl group of Gly and R125<sup>B</sup> in the side chain (Figure 7A). The RMSD between **XG** and **BMS-8** was 0.28 Å (Figure 7B), which suggested that the IC<sup>50</sup> value of

**XG** would be similar to that of **X**. Indeed, **XG** had a measured IC<sup>50</sup> for PD-1/PD-L1 binding of 2.1 µM (Figure 7C). **X** and **XG** potentially have the inhibitory effect for PD-1/PD-L1 interaction because K<sup>D</sup> between PD-1 and PD-L1 are reported as 6.4 µM [43]. *Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 9 of 17 *Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 9 of 17

**Figure 7.** In silico binding mode of **XG**. (**A**) Behavior of **XG** in the binding pocket of the PD-L1AB homodimer. (**B**) Superposition of **XG** onto **BMS-8**. The RMSD for displacement was 0.28 Å. (**C**) IC50 of **XG** for PD-1/PD-L1 binding. **Figure 7.** In silico binding mode of **XG**. (**A**) Behavior of **XG** in the binding pocket of the PD-L1AB homodimer. (**B**) Superposition of **XG** onto **BMS-8**. The RMSD for displacement was 0.28 Å. (**C**) IC<sup>50</sup> of **XG** for PD-1/PD-L1 binding. homodimer. (**B**) Superposition of **XG** onto **BMS-8**. The RMSD for displacement was 0.28 Å. (**C**) IC50 of **XG** for PD-1/PD-L1 binding. **GX** docking into the binding pocket of the PD-L1 homodimer revealed two hydrogen bonds

**GX** docking into the binding pocket of the PD-L1 homodimer revealed two hydrogen bonds formed between **GX** amino groups and carbonyl group of Y123B (Figure 8A). As a result, the calculated RMSD between **GX** and **BMS-8** was 0.52 Å (Figure 8B), which was larger than the RMSD of **X** and **XG**. This observation suggests that **GX** binding might sterically hinder PD-L1 homodimerization, leading to poorer inhibition of PD-1/PD-L1 binding. Consistent with this, the measured IC50 for **GX** was 448.5 μM (Figure 8C), which was several hundred times higher than those for **X** and **XG** ( Figure 4; Figure 7C). It is possible that the larger displacement of **X** of **GX** caused to deform the pocket of the PD-L1 homodimer, leading to the weaker inhibition of **GX** than those of **X GX** docking into the binding pocket of the PD-L1 homodimer revealed two hydrogen bonds formed between **GX** amino groups and carbonyl group of Y123<sup>B</sup> (Figure 8A). As a result, the calculated RMSD between **GX** and **BMS-8** was 0.52 Å (Figure 8B), which was larger than the RMSD of **X** and **XG**. This observation suggests that **GX** binding might sterically hinder PD-L1 homodimerization, leading to poorer inhibition of PD-1/PD-L1 binding. Consistent with this, the measured IC<sup>50</sup> for **GX** was 448.5 µM (Figure 8C), which was several hundred times higher than those for **X** and **XG** ( Figure 4; Figure 7C). It is possible that the larger displacement of **X** of **GX** caused to deform the pocket of the PD-L1 homodimer, leading to the weaker inhibition of **GX** than those of **X** and **XG**. formed between **GX** amino groups and carbonyl group of Y123B (Figure 8A). As a result, the calculated RMSD between **GX** and **BMS-8** was 0.52 Å (Figure 8B), which was larger than the RMSD of **X** and **XG**. This observation suggests that **GX** binding might sterically hinder PD-L1 homodimerization, leading to poorer inhibition of PD-1/PD-L1 binding. Consistent with this, the measured IC50 for **GX** was 448.5 μM (Figure 8C), which was several hundred times higher than those for **X** and **XG** ( Figure 4; Figure 7C). It is possible that the larger displacement of **X** of **GX** caused to deform the pocket of the PD-L1 homodimer, leading to the weaker inhibition of **GX** than those of **X** and **XG**.

**Figure 8.** In silico binding mode of **GX**. (**A**) Behavior of **GX** in the binding pocket of the PD-L1AB **Figure 8.** In silico binding mode of **GX**. (**A**) Behavior of **GX** in the binding pocket of the PD-L1AB homodimer. (**B**) Superposition of **GX** onto **BMS-8**. The RMSD for displacement was 0.52 Å RMSD. (**C**) IC50 of **GX** for PD-1/PD-L1 binding. **Figure 8.** In silico binding mode of **GX**. (**A**) Behavior of **GX** in the binding pocket of the PD-L1AB homodimer. (**B**) Superposition of **GX** onto **BMS-8**. The RMSD for displacement was 0.52 Å RMSD. (**C**) IC<sup>50</sup> of **GX** for PD-1/PD-L1 binding.

homodimer. (**B**) Superposition of **GX** onto **BMS-8**. The RMSD for displacement was 0.52 Å RMSD.

(**C**) IC50 of **GX** for PD-1/PD-L1 binding. The **X** portion of **BMS-8** without Cα, NH2, COOH atoms formed hydrophobic interactions in the crystal structure (PDB ID: 5J8O), with residues I54A, Y56A, V68A, M115A, I116A, S117A, A121A, D122A, I54B, Y56B, M115B, I116B, S117B, A121B, D122B, and Y123B of the PD-L1 homodimer (Figure 9A). The space-filling representation of **X** shows the adherent interaction mode to the binding pocket (Figure 9B,C). The intermediate compounds of **BMS-8**, compounds **1**–**6** (Scheme 1) showed a poor ability to The **X** portion of **BMS-8** without Cα, NH2, COOH atoms formed hydrophobic interactions in the crystal structure (PDB ID: 5J8O), with residues I54A, Y56A, V68A, M115A, I116A, S117A, A121A, D122A, I54B, Y56B, M115B, I116B, S117B, A121B, D122B, and Y123B of the PD-L1 homodimer (Figure 9A). The space-filling representation of **X** shows the adherent interaction mode to the binding pocket (Figure 9B,C). The intermediate compounds of **BMS-8**, compounds **1**–**6** (Scheme 1) showed a poor ability to inhibit PD-1/PD-L1 binding (Figure 3), which was probably due to insufficient hydrophobic filling of the compounds in the binding pocket. The **X** portion of **BMS-8** without Cα, NH2, COOH atoms formed hydrophobic interactions in the crystal structure (PDB ID: 5J8O), with residues I54A, Y56A, V68A, M115A, I116A, S117A, A121A, D122A, I54B, Y56B, M115B, I116B, S117B, A121B, D122B, and Y123<sup>B</sup> of the PD-L1 homodimer (Figure 9A). The space-filling representation of **X** shows the adherent interaction mode to the binding pocket (Figure 9B,C). The intermediate compounds of **BMS-8**, compounds **1**–**6** (Scheme 1) showed a poor ability to inhibit PD-1/PD-L1 binding (Figure 3), which was probably due to insufficient hydrophobic filling of the compounds in the binding pocket.

the compounds in the binding pocket.

inhibit PD-1/PD-L1 binding (Figure 3), which was probably due to insufficient hydrophobic filling of

*Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 10 of 17

**Figure 9.** Schematic drawing and space-filling representation of **BMS-8** binding in the binding pocket of the PD-L1 homodimer. In (**A–C)**, violet represents **X** without Cα, NH2, COOH atoms. (**A**) Binding mode of **BMS-8** in the pocket of the PD-L1AB homodimer (PDB ID:5J8O). Yellow and cyan represent PD-L1A and PD-L1B side chains, respectively. (**B**) Space-filling representation of **BMS-8** bound to the surface of PD-L1A (yellow) and contact area with PD-L1B (cyan). (**C**) Space-filling representation of **BMS-8** bound to the surface of PD-L1B (cyan) and contact area with PD-L1A (yellow). **Figure 9.** Schematic drawing and space-filling representation of **BMS-8** binding in the binding pocket of the PD-L1 homodimer. In (**A–C**), violet represents **X** without Cα, NH<sup>2</sup> , COOH atoms. (**A**) Binding mode of **BMS-8** in the pocket of the PD-L1AB homodimer (PDB ID:5J8O). Yellow and cyan represent PD-L1<sup>A</sup> and PD-L1<sup>B</sup> side chains, respectively. (**B**) Space-filling representation of **BMS-8** bound to the surface of PD-L1<sup>A</sup> (yellow) and contact area with PD-L1<sup>B</sup> (cyan). (**C**) Space-filling representation of **BMS-8** bound to the surface of PD-L1<sup>B</sup> (cyan) and contact area with PD-L1<sup>A</sup> (yellow).

Collectively, our results suggest that the larger displacement of **amino-Xs** from **BMS-8** prevents PD-L1A/PD-L1B homodimer formation. The docking simulations suggest that **X** and **GX** promote homodimerization of PD-L1, resulting in low IC50 values, whereas the larger displacement of **amino-Xs** prevents PD-L1 homodimer formation and increase the IC50 values. Collectively, our results suggest that the larger displacement of **amino-Xs** from **BMS-8** prevents PD-L1A/PD-L1<sup>B</sup> homodimer formation. The docking simulations suggest that **X** and **GX** promote homodimerization of PD-L1, resulting in low IC<sup>50</sup> values, whereas the larger displacement of **amino-Xs** prevents PD-L1 homodimer formation and increase the IC<sup>50</sup> values.

The results of this study advance our understanding of how small molecule compounds could be rationally designed to inhibit PD-1/PD-L1 interactions with high affinity. In silico docking simulations have typically shown that target proteins have stable binding pockets during ligand binding, even allowing for some local flexibility of the side chains within the pockets [37,44]. In that scenario, binding scores generally correlate well with experimentally determined inhibitor activity [45]. However, binding of **X** and **amino-X** in the PD-L1 pocket occurs through strict interactions, indicating that even a slight displacement of the **X** conformation leads to deformation of the PD-L1 homodimer, which deceases the inhibitory effect. Consistent with this, the **amino-X**s with shorter amino acid conjugates showed moderate positive correlations between the measured IC50 values and RMSDs in the no template/flexible docking mode, whereas the correlation was weakened by further amino acid addition. The results of this study advance our understanding of how small molecule compounds could be rationally designed to inhibit PD-1/PD-L1 interactions with high affinity. In silico docking simulations have typically shown that target proteins have stable binding pockets during ligand binding, even allowing for some local flexibility of the side chains within the pockets [37,44]. In that scenario, binding scores generally correlate well with experimentally determined inhibitor activity [45]. However, binding of **X** and **amino-X** in the PD-L1 pocket occurs through strict interactions, indicating that even a slight displacement of the **X** conformation leads to deformation of the PD-L1 homodimer, which deceases the inhibitory effect. Consistent with this, the **amino-X**s with shorter amino acid conjugates showed moderate positive correlations between the measured IC<sup>50</sup> values and RMSDs in the no template/flexible docking mode, whereas the correlation was weakened by further amino acid addition.

#### **3. Materials and Methods 3. Materials and Methods**

#### *3.1. Materials for Organic Chemistry Synthesis 3.1. Materials for Organic Chemistry Synthesis*

Sodium chloride (NaCl), lysozyme, monosodium phosphate (NaH2PO4), imidazole, glycerol, reduced glutathione, oxidized glutathione, methanol, dimethyl sulfoxide (DMSO), trifluoroacetic acid (TFA), tert-butyl acetate, perchloric acid (HClO4), hydrochloric acid (HCl), sodium carbonate, ethyl acetate, sodium sulfate, hexane, sodium hydrogen carbonate (NaHCO3), acetone, triphenyl phosphine (Ph3P), anhydrous dichloromethane (CH2Cl2), and anhydrous tetrahydrofuran (THF) were purchased from Wako Pure Chemical Industries Ltd. (Osaka, Japan). 3-Bromo-tyrosine, 3 hydroxymethyl-2-methylbiphenyl, and diisopropyl azodicarboxylate (DIAD; 40% in toluene, approximately 1.9 mol L–1) were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Magnesium sulfate and CH2Cl2 were purchased from Junsei Chemical Co., Ltd. (Tokyo, Japan). Deuterochloroform (CDCl3) was purchased from Isotec, Inc. (Miamisburg, OH, USA), and *N*-[(9Hfluoren-9-ylmethoxy) carbonyloxy] succinimide (Fmoc-Osu) was purchased from Watanabe Chemical Industries, Ltd. (Hiroshima, Japan). Sodium chloride (NaCl), lysozyme, monosodium phosphate (NaH2PO4), imidazole, glycerol, reduced glutathione, oxidized glutathione, methanol, dimethyl sulfoxide (DMSO), trifluoroacetic acid (TFA), tert-butyl acetate, perchloric acid (HClO4), hydrochloric acid (HCl), sodium carbonate, ethyl acetate, sodium sulfate, hexane, sodium hydrogen carbonate (NaHCO3), acetone, triphenyl phosphine (Ph3P), anhydrous dichloromethane (CH2Cl2), and anhydrous tetrahydrofuran (THF) were purchased from Wako Pure Chemical Industries Ltd. (Osaka, Japan). 3-Bromo-tyrosine, 3-hydroxymethyl-2-methylbiphenyl, and diisopropyl azodicarboxylate (DIAD; 40% in toluene, approximately 1.9 mol L−<sup>1</sup> ) were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Magnesium sulfate and CH2Cl<sup>2</sup> were purchased from Junsei Chemical Co., Ltd. (Tokyo, Japan). Deuterochloroform (CDCl3) was purchased from Isotec, Inc. (Miamisburg, OH, USA), and *N*-[(9H-fluoren-9-ylmethoxy) carbonyloxy] succinimide (Fmoc-Osu) was purchased from Watanabe Chemical Industries, Ltd. (Hiroshima, Japan).

#### *3.2. Synthesis of a Biphenyl-Conjugated Bromotyrosine 3.2. Synthesis of a Biphenyl-Conjugated Bromotyrosine Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 11 of 17

**(S)-tert-Butyl 2-amino-3-(3-bromo-4-hydroxyphenyl) propanoate (3)**. A suspension of 3 bromotyrosine (**2**; 1.0 g, 3.9 mmol) in tert-butyl acetate (16 mL, 92 mmol) was cooled to 0 °C, and stirred for 30 min. HClO4 (0.5 mL, 7.7 mmol) was then slowly added to the suspension at 0 °C, and the reaction mixture was warmed to 25 °C and stirred for 24 h. The mixture was washed with water and 1N HCl, and the aqueous phase was brought to pH 9 using sodium carbonate and then extracted with ethyl acetate. The resulting organic phase was washed with water and dried with sodium sulfate. The solvent was evaporated under reduced pressure, yielding an oily compound. This crude product was washed with cold hexane and then dried under reduced pressure to yield compound **3** (0.57 g, 47%). 1H-NMR (400 MHz, CDCl3): δ = 1.41 (s, 9H, –OC(CH3)3), 2.73 (dd, 1H, J = 14.4, 8.0 Hz, HOPh(Br)–CH2CH(NH2)–), 2.93 (dd, 1H, J = 13.6, 5.2 Hz, HOPh(Br)–CH2CH(NH2)–), 3.57 (dd, 1H, J = 7.2, 5.6 Hz, HOPh(Br)–CH2CH(NH2)–), 3.70 (m, 3H, HOPh(Br)–CH2CH(NH2)–), 6.70 (d, 1H, J = 8.0 Hz, aromatic ring), 6.94 (dd, 1H, J = 8.4, 2.0 Hz, aromatic ring), 7.26 (d, 1H, J = 1.6 Hz, aromatic ring). **(S)-tert-Butyl 2-amino-3-(3-bromo-4-hydroxyphenyl) propanoate (3)**. A suspension of 3-bromotyrosine (**2**; 1.0 g, 3.9 mmol) in tert-butyl acetate (16 mL, 92 mmol) was cooled to 0 ◦C, and stirred for 30 min. HClO<sup>4</sup> (0.5 mL, 7.7 mmol) was then slowly added to the suspension at 0 ◦C, and the reaction mixture was warmed to 25 ◦C and stirred for 24 h. The mixture was washed with water and 1N HCl, and the aqueous phase was brought to pH 9 using sodium carbonate and then extracted with ethyl acetate. The resulting organic phase was washed with water and dried with sodium sulfate. The solvent was evaporated under reduced pressure, yielding an oily compound. This crude product was washed with cold hexane and then dried under reduced pressure to yield compound **3** (0.57 g, 47%). <sup>1</sup>H-NMR (400 MHz, CDCl3): δ = 1.41 (s, 9H, –OC(CH3)3), 2.73 (dd, 1H, J = 14.4, 8.0 Hz, HOPh(Br)–CH2CH(NH2)–), 2.93 (dd, 1H, J = 13.6, 5.2 Hz, HOPh(Br)–CH2CH(NH2)–), 3.57 (dd, 1H, J = 7.2, 5.6 Hz, HOPh(Br)–CH2CH(NH2)–), 3.70 (m, 3H, HOPh(Br)–CH2CH(NH2)–), 6.70 (d, 1H, J = 8.0 Hz, aromatic ring), 6.94 (dd, 1H, J = 8.4, 2.0 Hz, aromatic ring), 7.26 (d, 1H, J = 1.6 Hz, aromatic ring). **(S)-tert-Butyl 2-amino-3-(3-bromo-4-hydroxyphenyl) propanoate (3)**. A suspension of 3 bromotyrosine (**2**; 1.0 g, 3.9 mmol) in tert-butyl acetate (16 mL, 92 mmol) was cooled to 0 °C, and stirred for 30 min. HClO4 (0.5 mL, 7.7 mmol) was then slowly added to the suspension at 0 °C, and the reaction mixture was warmed to 25 °C and stirred for 24 h. The mixture was washed with water and 1N HCl, and the aqueous phase was brought to pH 9 using sodium carbonate and then extracted with ethyl acetate. The resulting organic phase was washed with water and dried with sodium sulfate. The solvent was evaporated under reduced pressure, yielding an oily compound. This crude product was washed with cold hexane and then dried under reduced pressure to yield compound **3** (0.57 g, 47%). 1H-NMR (400 MHz, CDCl3): δ = 1.41 (s, 9H, –OC(CH3)3), 2.73 (dd, 1H, J = 14.4, 8.0 Hz, HOPh(Br)–CH2CH(NH2)–), 2.93 (dd, 1H, J = 13.6, 5.2 Hz, HOPh(Br)–CH2CH(NH2)–), 3.57 (dd, 1H, J = 7.2, 5.6 Hz, HOPh(Br)–CH2CH(NH2)–), 3.70 (m, 3H, HOPh(Br)–CH2CH(NH2)–), 6.70 (d, 1H, J = 8.0 Hz, aromatic ring), 6.94 (dd, 1H, J = 8.4, 2.0 Hz, aromatic ring), 7.26 (d, 1H, J = 1.6 Hz, aromatic ring).

**hydroxyphenyl)propanoate (4)**. A suspension of **3** (0.5 g, 1.6 mmol) and NaHCO3 (0.27 g, 3.2 mmol)

**(S)-tert-Butyl 2-({[(9H-fluoren-9-yl)methoxy]carbonyl}amino)-3-(3-bromo-4-**

in water (20 mL) was cooled to 0 °C. Fmoc-Osu (1.1 g, 3.2 mmol) in acetone (40 mL) was added to the suspension slowly, and the reaction mixture was then stirred at 25 °C for 15 h. The solvent was removed and washed with 1N HCl and water. After drying under vacuum, the crude product was purified by column chromatography on silica gel (eluent: ethyl acetate/hexane = 1:3 v/v) to yield compound **4** (0.71 g, 84%). 1H-NMR (400 MHz, CDCl3): δ = 1.42 (s, 9H, –OC(CH3)3), 3.00 (d, 2H, J = 5.6 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.21 (t, 1H, J = 7.2 Hz, HOPh(Br)– CH2CH(NHCOOCH2CH–)–), 4.33 (dd, 1H, J = 10.4, 7.2 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.43–5.00 (m, 2H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.29 (d, 1H, J = 8.0 Hz, HOPh(Br)– CH2CH(NHCOOCH2CH–)–), 5.43 (s, 1H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 6.91 (d, 1H, J = 8.4 Hz, aromatic ring), 6.96 (d, 1H, J = 9.2 Hz, aromatic ring), 7.26–7.33 (m, 3H, aromatic ring), 7.40 (dd, 2H, J = 7.4, 7.4 Hz, aromatic ring), 7.57 (dd, 2H, J = 6.2, 6.2 Hz, aromatic ring), 7.76 (d, 2H, J = 7.2 Hz, aromatic ring); high resolution mass spectrometry (HRMS) calculated for C28H28BrNO5 ([M + H]+): 538.1224, found: 538.1224. **hydroxyphenyl)propanoate (4)**. A suspension of **3** (0.5 g, 1.6 mmol) and NaHCO3 (0.27 g, 3.2 mmol) in water (20 mL) was cooled to 0 °C. Fmoc-Osu (1.1 g, 3.2 mmol) in acetone (40 mL) was added to the suspension slowly, and the reaction mixture was then stirred at 25 °C for 15 h. The solvent was removed and washed with 1N HCl and water. After drying under vacuum, the crude product was purified by column chromatography on silica gel (eluent: ethyl acetate/hexane = 1:3 v/v) to yield compound **4** (0.71 g, 84%). 1H-NMR (400 MHz, CDCl3): δ = 1.42 (s, 9H, –OC(CH3)3), 3.00 (d, 2H, J = 5.6 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.21 (t, 1H, J = 7.2 Hz, HOPh(Br)– CH2CH(NHCOOCH2CH–)–), 4.33 (dd, 1H, J = 10.4, 7.2 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.43–5.00 (m, 2H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.29 (d, 1H, J = 8.0 Hz, HOPh(Br)– CH2CH(NHCOOCH2CH–)–), 5.43 (s, 1H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 6.91 (d, 1H, J = 8.4 Hz, aromatic ring), 6.96 (d, 1H, J = 9.2 Hz, aromatic ring), 7.26–7.33 (m, 3H, aromatic ring), 7.40 (dd, 2H, J = 7.4, 7.4 Hz, aromatic ring), 7.57 (dd, 2H, J = 6.2, 6.2 Hz, aromatic ring), 7.76 (d, 2H, J = 7.2 Hz, aromatic ring); high resolution mass spectrometry (HRMS) calculated for C28H28BrNO5 ([M + H]+): 538.1224, found: 538.1224. **(S)-tert-Butyl 2-({[(9H-fluoren-9-yl)methoxy]carbonyl}amino)-3-(3-bromo-4-hydroxyphenyl) propanoate (4)**. A suspension of **3** (0.5 g, 1.6 mmol) and NaHCO<sup>3</sup> (0.27 g, 3.2 mmol) in water (20 mL) was cooled to 0 ◦C. Fmoc-Osu (1.1 g, 3.2 mmol) in acetone (40 mL) was added to the suspension slowly, and the reaction mixture was then stirred at 25 ◦C for 15 h. The solvent was removed and washed with 1N HCl and water. After drying under vacuum, the crude product was purified by column chromatography on silica gel (eluent: ethyl acetate/hexane = 1:3 v/v) to yield compound **4** (0.71 g, 84%). <sup>1</sup>H-NMR (400 MHz, CDCl3): δ = 1.42 (s, 9H, –OC(CH3)3), 3.00 (d, 2H, J = 5.6 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.21 (t, 1H, J = 7.2 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.33 (dd, 1H, J = 10.4, 7.2 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.43–5.00 (m, 2H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.29 (d, 1H, J = 8.0 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.43 (s, 1H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 6.91 (d, 1H, J = 8.4 Hz, aromatic ring), 6.96 (d, 1H, J = 9.2 Hz, aromatic ring), 7.26–7.33 (m, 3H, aromatic ring), 7.40 (dd, 2H, J = 7.4, 7.4 Hz, aromatic ring), 7.57 (dd, 2H, J = 6.2, 6.2 Hz, aromatic ring), 7.76 (d, 2H, J = 7.2 Hz, aromatic ring); high resolution mass spectrometry (HRMS) calculated for C28H28BrNO<sup>5</sup> ([M + H]+): 538.1224, found: 538.1224.

**(S)-tert-Butyl 2-({[(9H-fluoren-9-yl)methoxy]carbonyl}amino)-3-{3-bromo-4-[(2-methyl-1,1'-**

*Int. J. Mol. Sci.* **2019**, *20*, x FOR PEER REVIEW 11 of 17

**(S)-tert-Butyl 2-amino-3-(3-bromo-4-hydroxyphenyl) propanoate (3)**. A suspension of 3 bromotyrosine (**2**; 1.0 g, 3.9 mmol) in tert-butyl acetate (16 mL, 92 mmol) was cooled to 0 °C, and stirred for 30 min. HClO4 (0.5 mL, 7.7 mmol) was then slowly added to the suspension at 0 °C, and the reaction mixture was warmed to 25 °C and stirred for 24 h. The mixture was washed with water and 1N HCl, and the aqueous phase was brought to pH 9 using sodium carbonate and then extracted with ethyl acetate. The resulting organic phase was washed with water and dried with sodium sulfate. The solvent was evaporated under reduced pressure, yielding an oily compound. This crude product was washed with cold hexane and then dried under reduced pressure to yield compound **3** (0.57 g, 47%). 1H-NMR (400 MHz, CDCl3): δ = 1.41 (s, 9H, –OC(CH3)3), 2.73 (dd, 1H, J = 14.4, 8.0 Hz, HOPh(Br)–CH2CH(NH2)–), 2.93 (dd, 1H, J = 13.6, 5.2 Hz, HOPh(Br)–CH2CH(NH2)–), 3.57 (dd, 1H, J = 7.2, 5.6 Hz, HOPh(Br)–CH2CH(NH2)–), 3.70 (m, 3H, HOPh(Br)–CH2CH(NH2)–), 6.70 (d, 1H, J = 8.0 Hz, aromatic ring), 6.94 (dd, 1H, J = 8.4, 2.0 Hz, aromatic ring), 7.26 (d, 1H, J = 1.6 Hz, aromatic ring).

**(S)-tert-Butyl 2-({[(9H-fluoren-9-yl)methoxy]carbonyl}amino)-3-(3-bromo-4 hydroxyphenyl)propanoate (4)**. A suspension of **3** (0.5 g, 1.6 mmol) and NaHCO3 (0.27 g, 3.2 mmol) in water (20 mL) was cooled to 0 °C. Fmoc-Osu (1.1 g, 3.2 mmol) in acetone (40 mL) was added to the suspension slowly, and the reaction mixture was then stirred at 25 °C for 15 h. The solvent was removed and washed with 1N HCl and water. After drying under vacuum, the crude product was purified by column chromatography on silica gel (eluent: ethyl acetate/hexane = 1:3 v/v) to yield compound **4** (0.71 g, 84%). 1H-NMR (400 MHz, CDCl3): δ = 1.42 (s, 9H, –OC(CH3)3), 3.00 (d, 2H, J = 5.6 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.21 (t, 1H, J = 7.2 Hz, HOPh(Br)– CH2CH(NHCOOCH2CH–)–), 4.33 (dd, 1H, J = 10.4, 7.2 Hz, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.43–5.00 (m, 2H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.29 (d, 1H, J = 8.0 Hz, HOPh(Br)– CH2CH(NHCOOCH2CH–)–), 5.43 (s, 1H, HOPh(Br)–CH2CH(NHCOOCH2CH–)–), 6.91 (d, 1H, J = 8.4 Hz, aromatic ring), 6.96 (d, 1H, J = 9.2 Hz, aromatic ring), 7.26–7.33 (m, 3H, aromatic ring), 7.40 (dd, 2H, J = 7.4, 7.4 Hz, aromatic ring), 7.57 (dd, 2H, J = 6.2, 6.2 Hz, aromatic ring), 7.76 (d, 2H, J = 7.2 Hz,

*3.2. Synthesis of a Biphenyl-Conjugated Bromotyrosine* 

**(S)-tert-Butyl 2-({[(9H-fluoren-9-yl)methoxy]carbonyl}amino)-3-{3-bromo-4-[(2-methyl-1,1' -biphenyl-3-yl)methoxy]phenyl}propanoate (5).** To a solution of **4** (0.1 g, 0.19 mmol), 3-hydroxymethyl-2-methylbiphenyl (1; 39 mg, 0.20 mmol), and triphenyl phosphine (57 mg, 0.20 mmol) in anhydrous THF (10 mL), DIAD (0.1 mL, 0.22 mmol) was added at 0 ◦C under argon, and the reaction mixture was stirred at 0 ◦C for 12 h under argon. The organic phase was extracted with CH2Cl<sup>2</sup> and dried over anhydrous magnesium sulfate. The solvent was then evaporated under reduced pressure, with the temperature kept below 30 ◦C. The crude product was purified by column chromatography on silica gel (eluent: ethyl acetate/hexane = 1:4 v/v) to yield compound **5** (0.09 g, 66%). <sup>1</sup>H-NMR (400 MHz, CDCl3): δ = 1.46 (s, 9H, –OC(CH3)3), 2.27 (s, 3H, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 3.05 (d, 2H, J = 5.6 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.23 (t, 1H, J = 7.6 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.34 (dd, 1H, J = 6.8, 6.8 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.46-4.56 (m, 2H, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.13 (s, 2H, Biphenyl(CH3)-CH2OPh(Br)-CH2CH(NHCOOCH2CH–)–), 5.37 (d, 1H, J = 7.6 Hz, Biphenyl(CH3)–CH2OPh(Br)-CH2CH(NHCOOCH2CH–)–), 6.92 (d, 1H, J = 8.0 Hz, aromatic ring), 7.05 (d, 1H, J = 7.2 Hz, aromatic ring), 7.19–7.45 (m, 14H, aromatic ring), 7.53 (d, 1H, J = 6.8 Hz, aromatic ring), 7.60 (dd, 2H, J = 6.4, 6.4 Hz, aromatic ring), 7.77 (d, 2H, J = 7.2 Hz, aromatic ring); HRMS calculated for C42H40BrNO<sup>5</sup> ([M + H]+): 718.2163, found: 718.2164. **biphenyl-3-yl)methoxy]phenyl}propanoate (5)**. To a solution of **4** (0.1 g, 0.19 mmol), 3 hydroxymethyl-2-methylbiphenyl (**1**; 39 mg, 0.20 mmol), and triphenyl phosphine (57 mg, 0.20 mmol) in anhydrous THF (10 mL), DIAD (0.1 mL, 0.22 mmol) was added at 0 °C under argon, and the reaction mixture was stirred at 0 °C for 12 h under argon. The organic phase was extracted with CH2Cl2 and dried over anhydrous magnesium sulfate. The solvent was then evaporated under reduced pressure, with the temperature kept below 30 °C. The crude product was purified by column chromatography on silica gel (eluent: ethyl acetate/hexane = 1:4 v/v) to yield compound **5** (0.09 g, 66%). 1H-NMR (400 MHz, CDCl3): δ = 1.46 (s, 9H, –OC(CH3)3), 2.27 (s, 3H, Biphenyl(CH3)– CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 3.05 (d, 2H, J = 5.6 Hz, Biphenyl(CH3)–CH2OPh(Br)– CH2CH(NHCOOCH2CH–)–), 4.23 (t, 1H, J = 7.6 Hz, Biphenyl(CH3)–CH2OPh(Br)– CH2CH(NHCOOCH2CH–)–), 4.34 (dd, 1H, J = 6.8, 6.8 Hz, Biphenyl(CH3)–CH2OPh(Br)– CH2CH(NHCOOCH2CH–)–), 4.46-4.56 (m, 2H, Biphenyl(CH3)–CH2OPh(Br)– CH2CH(NHCOOCH2CH–)–), 5.13 (s, 2H, Biphenyl(CH3)-CH2OPh(Br)-CH2CH(NHCOOCH2CH–)– ), 5.37 (d, 1H, J = 7.6 Hz, Biphenyl(CH3)–CH2OPh(Br)-CH2CH(NHCOOCH2CH–)–), 6.92 (d, 1H, J = 8.0 Hz, aromatic ring), 7.05 (d, 1H, J = 7.2 Hz, aromatic ring), 7.19–7.45 (m, 14H, aromatic ring), 7.53 (d, 1H, J = 6.8 Hz, aromatic ring), 7.60 (dd, 2H, J = 6.4, 6.4 Hz, aromatic ring), 7.77 (d, 2H, J = 7.2 Hz, aromatic ring); HRMS calculated for C42H40BrNO5 ([M + H]+): 718.2163, found: 718.2164.

**(S)-2-({[(9H-fluoren-9-yl)methoxy]carbonyl}amino)-3-{3-bromo-4-[(2-methyl-1,1'-biphenyl-3-yl)methoxy]phenyl}propanoic acid (6)**. A solution of **5** (3.9 g, 5.42 mmol) in anhydrous CH2Cl2 (36 mL) was stirred at 0 °C under argon for 15 min. TFA (1.3 mL, 16.6 mmol) was added dropwise to the solution at 0 °C, and the reaction mixture was stirred at 25 °C under argon. After 6 h, TFA (1.5 mL, 19.5 mmol) was added to the reaction mixture, which was then stirred at 25 °C for 18 h under argon. The solvent was removed under reduced pressure, with the temperature kept below 40 °C. The crude product was purified by column chromatography on silica gel (eluent: CH2Cl2/methanol = 97:3 v/v) to yield compound **6** (3.2 g, 85%). 1H-NMR (400 MHz, CDCl3): δ = 2.24 (s, 3H, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 3.05 (dd, 1H, J = 14.0, 6.0 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 3.15 (dd, 1H, J = 14.8, 5.2 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.21 (t, 1H, J = 6.8 Hz, Biphenyl(CH3)– CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.36 (dd, 1H, J = 6.8, 6.8 Hz, Biphenyl(CH3)– CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.46 (dd, 1H, J = 10.0, 7.2 Hz, Biphenyl(CH3)– **(S)-2-({[(9H-fluoren-9-yl)methoxy]carbonyl}amino)-3-{3-bromo-4-[(2-methyl-1,1'-biphenyl-3 yl)methoxy]phenyl}propanoic acid (6)**. A solution of **5** (3.9 g, 5.42 mmol) in anhydrous CH2Cl<sup>2</sup> (36 mL) was stirred at 0 ◦C under argon for 15 min. TFA (1.3 mL, 16.6 mmol) was added dropwise to the solution at 0 ◦C, and the reaction mixture was stirred at 25 ◦C under argon. After 6 h, TFA (1.5 mL, 19.5 mmol) was added to the reaction mixture, which was then stirred at 25 ◦C for 18 h under argon. The solvent was removed under reduced pressure, with the temperature kept below 40 ◦C. The crude product was purified by column chromatography on silica gel (eluent: CH2Cl2/methanol = 97:3 v/v) to yield compound **6** (3.2 g, 85%). <sup>1</sup>H-NMR (400 MHz, CDCl3): δ = 2.24 (s, 3H, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 3.05 (dd, 1H, J = 14.0, 6.0 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 3.15 (dd, 1H, J = 14.8, 5.2 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.21 (t, 1H, J = 6.8 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.36 (dd, 1H, J = 6.8, 6.8 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.46 (dd, 1H, J = 10.0, 7.2 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 4.66

CH2CH(NHCOOCH2CH–)–), 5.23 (d, 1H, J = 8.4 Hz, Biphenyl(CH3)-CH2OPh(Br)– CH2CH(NHCOOCH2CH–)–), 6.91 (d, 1H, J = 8.8 Hz, aromatic ring), 7.03 (d, 1H, J = 7.6 Hz, aromatic ring), 7.21–7.55 (m, 15H, aromatic ring), 7.74 (d, 2H, J = 7.2 Hz, aromatic ring); HRMS calculated for

C38H32BrNO5 ([M + H]+): 662.1537, found: 662.1520.

(dd, 1H, J = 13.2, 6.0 Hz, Biphenyl(CH3)–CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.09 (s, 2H, Biphenyl(CH3)-CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 5.23 (d, 1H, J = 8.4 Hz, Biphenyl(CH3)-CH2OPh(Br)–CH2CH(NHCOOCH2CH–)–), 6.91 (d, 1H, J = 8.8 Hz, aromatic ring), 7.03 (d, 1H, J = 7.6 Hz, aromatic ring), 7.21–7.55 (m, 15H, aromatic ring), 7.74 (d, 2H, J = 7.2 Hz, aromatic ring); HRMS calculated for C38H32BrNO<sup>5</sup> ([M + H]+): 662.1537, found: 662.1520.

#### *3.3. Solid-State Peptide Synthesis*

**Amino-X**s were synthesized using an automated peptide synthesizer (MultiPep CF, INTAVIS Bioanalytical Instruments AG, Cologne, Germany). The synthetic protocol for glycine-conjugated peptide **XG** was as follows: Fmoc-protected glycine attached to a polystyrene resin (Fmoc-Gly NovaSyn TGT, Merck KGaA, Darmstadt, Germany) was deprotected by piperidine (20% in *N*-methylpyrrolidone (NMP). The resulting resin was reacted with **6** (99 mg, 0.14 mmol), 1-[bis(dimethylamino)methylene]-1H-benzotriazolium 3-oxide hexafluorophosphate (HBTU; 150 µL, 0.5 M in *N,N*-dimethylformamide (DMF), *N*-methylmorpholine (45 µL, 4.0 M in DMF) in NMP (8 µL) for 45 min. After washing, the *N*-α-protecting group of Fmoc in compound **6** was deprotected by piperidine (20% in NMP). Finally, the obtained peptide was cleaved from the resin using TFA (95% in water), yielding **XG**. Other peptides were synthesized using a similar method. (*S*)-2-amino-3-[3-bromo-4-{(2-methyl-1,10 -biphenyl-3-yl)methoxy}phenyl]propanoic acid (**X**) was obtained by deprotection of Fmoc in **6** using piperidine (20% in NMP).

#### *3.4. Characterization*

The synthesized compounds were identified using <sup>1</sup>H NMR spectroscopy (JNM–ECZ400R, JEOL Ltd., Tokyo, Japan) and HRMS (QSTAR Elite, AB SCIEX, Framingham, MA, USA).
