Development of Newly Synthesized Chromone Derivatives with High Tumor Specificity against Human Oral Squamous Cell Carcinoma

Since many anticancer drugs show severe adverse effects such as mucositis, peripheral neurotoxicity, and extravasation, it was crucial to explore new compounds with much reduced adverse effects. Comprehensive investigation with human malignant and nonmalignant cells demonstrated that derivatives of chromone, back-bone structure of flavonoid, showed much higher tumor specificity as compared with three major polyphenols in the natural kingdom, such as lignin-carbohydrate complex, tannin, and flavonoid. A total 291 newly synthesized compounds of 17 groups (consisting of 12 chromones, 2 esters, and 3 amides) gave a wide range of the intensity of tumor specificity, possibly reflecting the fitness for the optimal 3D structure and electric state. Among them, 7-methoxy-3-[(1E)-2-phenylethenyl]-4H-1-benzopyran-4-one (compound 22), which belongs to 3-styrylchromones, showed the highest tumor specificity. 22 induced subG1 and G2 + M cell population in human oral squamous cell carcinoma cell line, with much less keratinocyte toxicity as compared with doxorubicin and 5-FU. However, 12 active compounds selected did not necessarily induce apoptosis and mitotic arrest. This compound can be used as a lead compound to manufacture more active compound.


Introduction
This review is composed of four parts. The first part reviews the adverse effect of chemotherapeutic agents. The second part introduces our life-work research of development of chromone derivatives that show comparable anticancer activity and lower keratinocyte toxicity, as compared with anticancer drug. The third part describes the serious problems of neurotoxicity in G2 + M blocker. The fourth part is the summary of our major findings and future direction of chromone research.

Oral Mucositis Associated with Anticancer Drug
Oral mucositis is one of the most frequent adverse events in cancer drug therapy and hematopoietic stem cell transplantation. Oral mucositis is reported to occur in 5-50% of patients receiving standard-dose chemotherapy and 68-98% of high-dose chemotherapy related to hematopoietic stem cell transplantation [1]. Oral mucositis not only lowers the patient's QOL due to pain but also lowers the oral intake, leads to undernutrition and dehydration, and deteriorates the general condition. It also serves as a gateway for bacterial invasion and may trigger systemic infection. Decreased doses and delayed schedules in chemotherapy lead to reduced efficacy and survival rates, but currently there are no established preventive or curative methods for oral mucositis, hindering smooth cancer treatment [2].

Neurotoxicity of Anticancer Drugs
Cancer drug therapy has contributed to the improvement of survival rate and QOL by the development of cytocidal anticancer drugs and molecular targeted therapeutic agents, while the adverse effect of cancer drug therapy causes a decrease in QOL, sometimes causing the discontinuation of the drug therapy. Typical side effects include organ disorders such as bone marrow suppression, physical disorders such as nausea and vomiting, and neuropathy represented by paresthesia. Chemotherapy-induced peripheral neuropathy (CIPN) associated with an anticancer agent is not recovered quickly by a drug withdrawal like myelosuppression, and some disorders may remain for the lifetime. CIPN is a serious adverse event that interferes with the continuation of chemotherapy. It was reported that the incidence of CIPN was 68.1% within 1 month after chemotherapy, 60.0% after 3 months, and 30.0% after 6 months in a follow-up study of 4179 patients with colorectal cancer, breast cancer, gynecologic cancer, and multiple myeloma [3]. However, there are a few reports of preventive and therapeutic drugs for CIPN. Platinum, taxane, and vinca alkaloid are known as causative agents of CIPN. Different drugs have different mechanisms that cause peripheral neuropathy. For example, platinum-based cisplatin causes sensorineural deafness in the high range due to acoustic nerve damage. It has been reported that it is cumulative and that symptoms often continue for a long period of time after discontinuation of administration [4].
Oxaliplatin, a platinum drug, has acute and chronic symptoms. Acute symptoms are characterized by paresthesia around the extremities and around the lips and chronic symptoms may persist for months to years [5]. Carboplatin, a platinum drug, causes relatively few neurological symptoms when used at normal doses, and high doses may cause symptoms similar to cisplatin [6]. Paclitaxel, a taxane-based drug, mainly causes paresthesia of the extremities and is correlated with single dose and total dose [7]. Docetaxel, a taxane-based drug, causes sensory and motor disorders but is less frequent than paclitaxel [8]. Vincristine, a vinca alkaloid drug, causes sensory abnormalities in the fingers and movements within a few weeks after the start of treatment and often persists for a long time after the discontinuation of treatment [9]. CIPN pathological findings are classified into axonopathy, neuronopathy, and myelinopathy. Axonopathy is the most common disorder in CIPN. Neuronal cell body is relatively retained due to damage from thick and long axons. Clinically, glove and stocking type sensory deficits often begin at the extremities. Representative agents are microtubule inhibitors, vinca alkaloids, and taxanes. Neuronopathy is mainly cell bodies of lesions, mainly caused by cell death of dorsal root ganglion cells, and secondary damage to axons and myelin sheaths. Clinically, nerve cell bodies with short axons are also damaged, so sensory deficits occur not only on the extremities but also on the trunk and face. Representative agents are platinum agents such as oxaliplatin and cisplatin [10]. The frequency of mucositis and peripheral neuropathy of various anticancer agents was summarized from the interview form of the pharmaceutical company (Table 1).

Chemotherapy Extravasation
Systemic intravenous chemotherapy can cause multiple emergencies by local and systemic reactions. Drug extravasation is one of the most devastating complications in chemotherapy. Overall incidences of chemotherapy extravasation ranges from 0.01% to 6.5% [11][12][13], with reports of extravasation occurrence via central venous catheters ranging from 0.3% to 10.3% [14][15][16][17]. The exact incidence rate of extravasation varies greatly due to the general lack of reporting and absence of centralized registry of extravasation events. Therefore, no benchmark existed for the incidence of chemotherapy extravasations.
Extravasation is the accidental leakage of cytotoxic chemotherapy drugs that can cause severe tissue damage, tissue necrosis, blistering, or sloughing into the subcutaneous or subdermal tissue at the injection site [18,19]. Extravasated drugs are further classified into the three groups: vesicants, irritants, and nonvesicants/nonirritants, according to their potential for causing damage as (Table 2) [12,13,19,20].
Vesicant drugs have the capability to induce the formation of blisters and/or cause tissue destruction. Vesicant drugs may be subclassified into DNA-binding and non-DNA-binding compounds [20]. DNA-binding compounds are capable of producing more severe tissue damage and mainly include anthracyclines and alkylating agents. Non-DNA-binding compounds are mainly vinca alkaloids and taxanes.
Irritant drugs can cause pain at the injection site or along the vein, with or without an inflammatory reaction. Some of these agents have the potential to cause soft tissue ulcers only if a large amount of concentrated drug solution is inadvertently extravasated. Nonvesicant or nonirritant drugs, if extravasated, rarely produce acute reactions or tissue necrosis. Tissue damage related to extravasation occurs by different mechanisms [3]. First, the drug is absorbed by local cells in the tissue and binds to critical structures, causing cell death. After the endocytolysis, surrounding cells can also die through the release of the drug from nearby dead cells. The repetitive nature of this process impairs healing and may result in progressive and chronic tissue injury. Second, the drug that does not bind to cellular DNA may metabolize and be cleared, limiting the degree of tissue injury [21]. However, the literature addressing extravasation is limited to animal studies, case reports, and small human studies. Classic randomized studies in humans for the treatment of extravasations are unthinkable because of ethical reasons. On the whole, the highest possible grade Medicines 2020, 7, 50 5 of 18 of recommendation of each measure for extravasations would be low. Novel studies are clearly needed to elucidate the mechanism of chemotherapy extravasations.

Development of Newly Synthesized Chromone Derivatives with High Tumor Specificity, but Low Keratinocyte Toxicity
Our strategy to explore new compounds is composed with the following eight steps: search of natural products that shows the highest tumor specificity (Step 1); QSAR analysis of chromone-related compounds (Step 2); investigation of action mechanism (Step 3), identification of target molecules (Step 4), exploration of more activity compounds by prediction, synthesis, and confirmation (Step 5); check for adverse effects (Step 6); in vivo experiments with animals (Step 7); and clinical application (Step 8) (Figure 1).

Why We Focused on the Chromones
For the quantification of anticancer activity of text samples, it was necessary to establish the in vitro assay method ( Figure 2) [22], using human malignant and nonmalignant cells: four human oral squamous cell carcinoma (OSCC) cell lines (Ca9-22, HSC-2, HSC-3, and HSC-4), three human normal oral mesenchymal cells (gingival fibroblast HGF, periodontal ligament fibroblast HPLF, and pulp cell HPC), and two human normal oral epithelial cells (human oral keratinocyte HOK and primary human gingival epithelial cells HGEP).

Why We Focused on the Chromones
For the quantification of anticancer activity of text samples, it was necessary to establish the in vitro assay method ( Figure 2) [22], using human malignant and nonmalignant cells: four human oral squamous cell carcinoma (OSCC) cell lines (Ca9-22, HSC-2, HSC-3, and HSC-4), three human normal oral mesenchymal cells (gingival fibroblast HGF, periodontal ligament fibroblast HPLF, and pulp cell HPC), and two human normal oral epithelial cells (human oral keratinocyte HOK and primary human gingival epithelial cells HGEP). Tumor specificity (TS) was defined as the ratio of the mean of CC50 against normal cells to that against OSCC cells). When mesenchymal or epithelial cells were used, TSM and TSE could be obtained, respectively. TSE can be used as an index for neurotoxicity. (Figure 1). It would be the most ideal if we could use human epithelial cells as target cells. However, most of anticancer drugs show potent cytotoxicity against epithelial cells (as described later). Therefore, we used TSM value, rather than TSE at the first stage of random screening. Using this method, we found that three major polyphenols, i.e., lignin-carbohydrate complexes, tannins, and flavonoids, showed much lower TSM in comparison to popular chemotherapeutic antitumor drugs (Table 3). On the other hand, the derivatives of chromone, the backbone structure of various flavonoids such as flavonoid, flavone, flavanone, and isoflavone ( Figure 3) showed much higher TSM than the majority of polyphenols [22]. These findings encouraged us to explore more active chromone derivatives.   Tumor specificity (TS) was defined as the ratio of the mean of CC 50 against normal cells to that against OSCC cells). When mesenchymal or epithelial cells were used, TS M and TS E could be obtained, respectively. TS E can be used as an index for neurotoxicity. (Figure 1). It would be the most ideal if we could use human epithelial cells as target cells. However, most of anticancer drugs show potent cytotoxicity against epithelial cells (as described later). Therefore, we used TS M value, rather than TS E at the first stage of random screening. Using this method, we found that three major polyphenols, i.e., lignin-carbohydrate complexes, tannins, and flavonoids, showed much lower TS M in comparison to popular chemotherapeutic antitumor drugs (Table 3). On the other hand, the derivatives of chromone, the backbone structure of various flavonoids such as flavonoid, flavone, flavanone, and isoflavone ( Figure 3) showed much higher TS M than the majority of polyphenols [22]. These findings encouraged us to explore more active chromone derivatives.  1 Cited from [22].

Synthesis of Chromones, Esters, and Amides
We have focused on the following three groups of compounds (

Synthesis of Chromones, Esters, and Amides
We have focused on the following three groups of compounds ( Figure 4):   As for chromone derivatives, 3-styrylchromones (A) were synthesized by Knoevenagel condensation of the corresponding 3-formylchromones with various phenylacetic acid derivatives [23] (Figure 5).
As for esters and amides, cinnamic acid phenethyl esters (M) were synthesized by the condensation of cinnamic acid and its analogs, such as caffeic acid, ferulic acid, and p-coumaric acid, with the corresponding phenethyl alcohols. In addition, phenylpropanoid amides (O) were synthesized by the condensation of the corresponding cinnamic acid derivatives with various biogenic amines.
Piperic acid esters (N) were synthesized by the condensation of piperic acid with the corresponding alcohols. In addition, piperic acid amides (P) were synthesized by the condensation of the acid chloride of piperic acid with various amines. Piperic acid was prepared by alkaline hydrolysis of piperine.
Oleoylamides (Q) were synthesized by the condensation of oleoyl chloride, derived from oleic acid and oxalyl chloride, with the various corresponding biogenic amines ( Figure 6).
As for esters and amides, cinnamic acid phenethyl esters (M) were synthesized by the of the acid chloride of piperic acid with various amines. Piperic acid was prepared by alkaline hydrolysis of piperine. Oleoylamides (Q) were synthesized by the condensation of oleoyl chloride, derived from oleic acid and oxalyl chloride, with the various corresponding biogenic amines ( Figure 6).
Medicines 2020, 7, x FOR PEER REVIEW 12 of 19 noted that these compounds showed comparable TS values of doxorubicin (DXR) and much higher TS value than 5-FU. It was unexpected that DXR and 5-FU showed potent toxicity against human epithelial cells such as human oral keratinocyte (HOK) and human progenitor of human gingival epithelial cells (HGEP) (c/a and d/a in Table 5). We have reported previously that DXR induced apoptosis (characterized by the loss of cell surface microvilli, chromatin condensation, nuclear fragmentation, and caspase-3 activation) in these keratinocytes [54]. On the other hand, compounds 11, 22, 34, 62, and 69 showed much lower keratinocyte toxicity (Table 5).    . It is noted that these compounds showed comparable TS values of doxorubicin (DXR) and much higher TS value than 5-FU. It was unexpected that DXR and 5-FU showed potent toxicity against human epithelial cells such as human oral keratinocyte (HOK) and human progenitor of human gingival epithelial cells (HGEP) (c/a and d/a in Table 5). We have reported previously that DXR induced apoptosis (characterized by the loss of cell surface microvilli, chromatin condensation, nuclear fragmentation, and caspase-3 activation) in these keratinocytes [54]. On the other hand, compounds 11, 22, 34, 62, and 69 showed much lower keratinocyte toxicity (Table 5).

Other Biological Actions of Chromones, Esters, and Amides
We searched other biological activities of chromones, esters, and amides (Supplementary Table S2). Table 7 listed up the most potent compounds that showed biological activity higher than positive controls. Compounds 10, 12, 15, 124, 136, 229, 231, 237, 258, and 261 scavenged the DPPH radical, more potently than ascorbic acid, a well-known antioxidant [23], suggesting its antioxidant action.  [24,27,28,30,55]. Halogen-containing compounds show more potent inhibitory activity. All compounds showed higher MAO-B-specific inhibition than positive controls and, therefore, were not likely to exert adverse effects due to MAO-A inhibition. Furthermore, they show reversible inhibition and thus were much convenient for the sudden interruption of treatment, as compared with irreversible inhibitors.
Compounds 230 and 235 inhibited the butyrylcholinesterase (BChE) more potently than neostigmine, suggesting that they may serve as lead compounds for the development of novel BChE inhibitors and candidate lead compounds for the prevention or treatment of Alzheimer's disease [55].
However, many reports, including ours, demonstrated that microtubule-targeted agents have potent neurotoxicity, adversely affecting the quality of life of patients on a long-term basis [61][62][63][64]. Iijima et al. recently reported that carboplatin (CBDCA) was highly neurotoxic (TS N = 0.11 (3.2/27.9)), calculated using the data of Table 2 in Ref. [64]. It is urgent to investigate the neurotoxicity, extravasation as well as stomatitis of chromone derivatives, esters, and amides.

Conclusions and Future Direction
We found that: (i) Chromone showed much higher tumor specificity as compared with three major polyphenols. (ii) A total 291 newly synthesized compounds of 17 groups (consisting of 12 chromones, 2 esters, and 3 amides) gave a wide range of the intensity of tumor specificity. (iii) Their tumor specificity is correlated with chemical descriptors that reflect 3D structure and electric state. It is crucial to identify the target molecules (Step 4 in Figure 1). To accomplish this, 13 C-labeled compound 22 will be prepared, using 2-hydroxyacetophenone derivatives and 13 C-dimethylformamide, or using 13 C-iodomethane as methylation agent, and then the differential incorporation of 13 C into malignant and nonmalignant cells will be investigated, with LC-MS. Compound 22, labeled with fluorescence dye (Cy3, CY5, Cy7), will be tested to detect the intracellular uptake and distribution into organelles, using confocal laser microscopy. Binding of cellular protein to and elution from chromone-attached beads may be useful to identify the binding proteins.
In order to explore more potent chromone derivatives, the following three steps will be repeated: (i) prediction by QSAR of the best fit substituents that yield the highest TS M and TSE, (ii) synthesis of compounds introduced with such predicted substituents, and (iii) confirmation of antitumor potential (Step 5). However, it is important to eliminate the compounds that show potent keratinocyte toxicity, neurotoxicity, and extravasation (Step 6), before animal experiment (Step 6) and clinical application (Step 7).
The present study demonstrated that only selected compounds that have the optimal 3D structure show the highest tumor specificity, whereas most of other analogs that have similar structure show much less tumor specificity (Supplementary Table S1). This suggests the presence of binding components or receptors for chromones. It remains to be investigated whether compounds 22 and 40 may interact with estrogen receptors, since these compounds have structural similarity with isoflavones (such as daidzein and genistein) and to some degree with tamoxifen, which have been used for the treatment of oral squamous cell carcinoma that express estrogen receptors [65][66][67]. In addition, it seems that the "para" like substitution is favorable, possibly because it mimics the structure of estrogen. It is highly probable that different groups of chromone-related compounds have different anticancer mechanisms depending on their structure ( Table 6).