Triple-negative breast cancer (TNBC) mainly occurs in younger, premenopausal women with a more aggressive nature, i.e., it is more likely to metastasize to other organs such as the brain. Metastatic TNBC (mTNBC) has a poor overall survival of about 17.5 months, which shows its poor prognosis [1]. This is partly due to the fact that TNBC does not overexpress the common targets of breast cancer treatment including estrogen receptors, progesterone receptors, or human epidermal growth factor receptor 2 (HER-2); therefore, the common breast cancer-targeted therapies cannot be utilized in these patients. The mainstay of TNBC treatment has remained chemotherapy for decades, which, due to its non-selective nature, causes adverse reactions and toxicities in patients, leading to decreased patient compliance and increasing tumor resistance to treatment [2]. Consequently, the effort to find novel, targeted therapies for TNBC has been of special interest and attention. As an example, patients with BReast CAncer gene 1/2 (BRCA1/2) mutations undergo treatment with inhibitors of poly(ADP-ribose) polymerase (PARP) [3]. With the hope of boosting the patient’s immune system to kill cancer, one strategy has been to use immunotherapy targeting PD-1/PD-L1, cytotoxic T lymphocyte-associated antigen-4 (CTLA-4), lymphocyte-activation gene 3 (LAG-3), T cell immunoglobulin and mucin-domain containing-3 (TIM-3) and the hedgehog (Hh) and neuropilin-2 (NRP-2) signaling pathway (reviewed in [2]). So far, the most studied immunotherapy target in TNBC belongs to a group of antibodies targeting PD-1 and PD-L1 [4]. A subgroup of TNBC patients overexpressing PD-L1 were studied in the IMpassion130 Trial. The mTNBC patients received a combination of anti-PD-L1 atezolizumab and nab-paclitaxel, which improved their progression-free survival [5]. This led to the US food and drug administration (FDA) approval of atezolizumab for unresectable advanced PD-L1-positive TNBC patients [6]. The mechanism of this drug relies on the interaction of T cells and cancer cells. Following the interaction of the T cell receptor (TCR) and the cancer antigen, PD-1 expressed on cytotoxic T cells (CTLs) interacts with PD-L1 expressed on cancer cells, leading to the inhibition of the activation of CTLs and an “immune escape”. PD-L1 also plays some roles in the proliferation of cancer cells by affecting the mitogen-activated protein kinase (MAPK) pathway [7]. Therefore, blocking PD-L1 has at least two different outputs: Activating the immune system to kill cancer and inhibiting the proliferation of cancer cells ^[4][7][4,7].

TNBC with brain metastasis (TNBC-BM) is the most severe form of mTNBC to treat. This is because the BBB, a complex structure surrounding the brain, is a highly specialized and selective structure that tightly controls and regulates the delivery of necessary materials to the brain to maintain brain homeostasis ^[8][9][8,9]. Several cellular layers that comprise the BBB include the endothelial cells (ECs), pericytes, astrocytes and the basement membrane. The space between the BBB endothelial cells is sealed with more tight junctions compared to endothelial cells in other parts of the body, which makes them impermeable to hydrophilic molecules ^[9][10][9,10]. Moreover, the electrical resistance of ~1000–2000 ohm cm² in the BBB restricts the movement of ionic molecules [11]. Chemotherapeutics have bulky structures with limited access to the brain and the same applies to monoclonal antibodies targeting PD-1 or PD-L1. In addition, these molecules have animal-derived domains and, therefore, naturally inherit a structure that might cause immunogenic reactions. This has made the medicinal intervention of brain tumors challenging, leading to the failure of these options and keeping the overall survival of these patients to less than two years [12].

2. Apigenin (API) and Cosmosiin (COS)

Apigenin (API) and Cosmosiin (COS) are extracted from the traditional medicinal plant Salvia plebeia R. Br (SP) and API and COS share similar structures. API is a trihydroxyflavone, and COS is API 7-O-beta-D-glycoside, which also exists in an L-glycoside enantiomer form. API, together with some other flavonoids such as quercetin (QUE) and kaempferol (KMF), is the most ubiquitous plant flavonoid among more than 5000 ^[13][21]. It has a low toxicity in normal cells compared to cancer cells with antioxidant, and anti-inflammatory properties and influences the induction of apoptosis and cell cycle arrest in cancer cells. These functions are via its effect on several cellular signaling pathways such as PI3K/AKT, MAPK/ERK, and NF-κB Signaling, the Wnt/β-Catenin pathway, STAT 3 and epidermal growth factor receptor (EGFR) (reviewed in ^[14][22]). Moreover, studies have shown the efficacy of this molecule in several cancers such as breast, lung and melanoma models ^[14][22]. In a study by Choi et al. (2020), the extract of Salvia plebeia R. Br. (SPE) blocked the binding of PD-1/PD-L1 in an enzyme-linked immunoassay (ELISA). This was dose-dependent and with specific blocking with no effect on the CTLA-4/CD80 interaction; however, the potency of the SPE was less than a PD-L1-blocking antibody. The blocking effect was attributed to the ethyl acetate fraction of the extract. This fraction of the extract had about eighteen-fold higher amounts of API and eight-fold COS. At 50 mg/mL, the SPE and the ethyl acetate fraction showed an ~42% and ~63% inhibitory action on the PD-1/PD-L1 interaction. At concentrations <50 mg/mL (24 h), the SPE showed no cytotoxicity in a co-culture system containing Jurkat and aAPC/CHO-K1 cells (CHO cells engineered to express a hPD-L1 and TCR agonist). The SPE and the ethyl acetate fraction were used in a co-culture system of aAPC/CHO-K1 cells and Jurkat cells. In this system, a half-effective concentration (EC₅₀) of the PD-L1 blocking antibody was ~0.3 µg/mL in activating TCR signaling. The relevant EC₅₀ value for the SPE and the ethyl acetate fraction was ~27 and 1 µg/mL, respectively ^[15][23]. This demonstrated the importance of the ethyl acetate fraction. When humanized PD-L1-expressing MC38 cells (hPDL1-MCs) co-cultured with humanized PD-1 mouse splenocytes were exposed with the non-cytotoxic concentrations of SPE, the cell viability was significantly decreased. A co-culture of hPDL1-MCs with CTLs isolated from the tumor showed that cell death was induced by the activation of T cells; however, these effects were not compared with a blocking antibody control. In an hPD-L1 knock-in MC38 tumor-bearing humanized PD-1 mouse model, mice received 5 mg/kg of anti-hPD-1 antibody (intraperitoneal (IP)—twice a week) as the control or 100 and 300 mg/kg of oral SPE. The 100 and 300 mg/kg of SPE inhibited tumor growth by ~45% and 78%, respectively, while the efficacy of the control antibody was 88%. This treatment increased the number of CTLs and CD3+ tumor-infiltrating lymphocytes ^[15][23]. Among the seven components of SPE tested at 2 µM as single agents, API and COS showed the best improvement of T cell function, by about two-fold, and also showed a dose-dependent increased T cell function. Both molecules showed a dose-dependent blockage of the PD-1/PD-L1 interaction in an ELISA assay. In both experiments, the COS showed a more effective action. The structure–activity relationship studies confirmed that the monosaccharide group at C₇ played a major role in the observed effects. This effect of COS was specific to the PD-1/PD-L1 and did not affect the CTLA4/CD80 interaction. The COS showed a dissociation constant (K_D) of 386 and 85 µM for PD-1 and PD-L1, respectively (R² 0.9804 and 0.9866, respectively). Due to its higher binding rate, the COS had an ~4.5-fold higher affinity for PD-L1 ^[15][23]. Molecular docking using AutoDock Vina between COS and the crystal structure of hPD-1/hPD-L1 (4ZQK) predicted the binding affinities of −6.2 and −5.8 kCal/mol with PD-L1 and PD-1, respectively. The interaction site was found to have hydrogen bonds between the residues N63, D61, N58 and the glycoside of COS, in addition to a hydrophobic interaction between R131, M115, Q66, and I54 and the backbone of COS (API) ^[15][23].

3. Kaempferol and Kaempferol 7-O-Rhamnoside

A variety of edible, non-medicinal and medicinal plants such as Geranii Herba produce the flavonol, KMF ^[16][24]. KMF is produced to protect plants against oxidative reactions; therefore, an inherent nature of this molecule is its antioxidant property which is important in chemoprevention and anti-inflammatory reactions. Like API, several studies have shown the anticancer properties of KMF in breast, colon and liver cancers ^[17][25].  In 2020, Kim et al. studied the active ingredients of the Geranii Herba extract and showed that it inhibited the interaction of PD-1/PD-L1 (a half inhibitory concentration of (IC₅₀) ~88 µg/mL). KMF, among its glycosylated derivatives, was the most potent one (IC₅₀~8 µM) with a dose-dependent effect; however, its IC₅₀ was higher than the controls, neutralizing antibodies and the PD-1/PD-L1 inhibitor C1 ^[18][26]. KMF or its glycosides showed no cytotoxicity (< 100 µM) on Jurkat and CHO-K1 cells but showed a dose-dependent decreased interaction of PD-1/PD-L1. Both the KMF and KOR showed a similar half-effective concentration (EC₅₀) of ~16 µM. The KOR was shown to have a K_D of 1.56 × 10⁻⁴ M. In silico molecular docking studies between KMF and the crystallographic structure of human PD-L1/PD-1 (PDB code: 4ZQK) showed that KMF and KOR attached to PD-L1 at the interaction site of PD-1 with different modes of action (i.e., binding energies of −5.4 and −5.6 kcal/mol, respectively) ^[18][26]. It was decided that the glycoside group was associated with the functional activity of KOR in blocking the PD-1/PD-L1 interaction. The binding scores were not compared with a control molecule and, therefore, it cannot be concluded whether this interaction is a strong one or not.

4. Quercetin

The other abundantly found flavonoid in fruits and vegetables such as broccoli, onion, pepper and apple is QUE. QUE has a similar backbone to KMF, API and their derivatives and, therefore, similar anti-inflammatory and antioxidant actions can be expected. Its pro-apoptotic properties, induction of cell cycle arrest and DNA damage have made QUE a good anticancer candidate ^[19][20][27,28]. Some of the suggested anticancer mechanisms of action of QUE include a decreased production of cyclooxygenase and lipoxygenase and its effect on some signaling pathways such as NF-κB, ERK, and JNK ^[21][22][23][29,30,31]. In addition, QUE, via inducing the expression of interferon-γ(IFN-γ) and interleukin-4 (IL-4) and promoting the natural killer (NK) cell function, improves the immune system ^[24][25][26][32,33,34]. Moreover, due to its anti-inflammatory effects, QUE has shown efficacy in several disease models including infection and cardiovascular disease. Jing et al., in 2021, used an ELISA system on a library of 1018 compounds and showed that 5 µM of QUE-dihydrate showed the best (80%) and dose-dependent inhibition of a PD-1/PD-L1 interaction with an IC₅₀ of ~0.2 µM ^[27][35]. At 5 µM, the QUE showed a 50% inhibition of the PD-1/PD-L1 interaction.  These results are not comparable to those of KMF, as different techniques were applied to evaluate the inhibitory actions. It was previously shown that the interaction of PD-1/PD-L1 is in the glycosylated form of the PD-L1 ^[28][36] and that the QUE inhibited the binding of these glycosylated proteins (IC₅₀ 0.5 µM). In a co-culture system of Jurkat and cancer cells (MDA-MB-231 and H460), QUE potentiated the activity of Jurkat T cells causing about a 40% cancer cell death. Furthermore, in a xenograft mouse model, 60 mg/kg of QUE inhibited tumor growth, the population of cytotoxic T cells increased and the expression of cytokines such as interferon-gamma (IFN-γ) and granzyme B in the tumor microenvironment increased to kill the tumor ^[27][35].

5. Eriodictyol and Fisetin

Toxicodendron vernicifluum (TV) or Rhus verniciflua Stokes is another traditional herbal medicine native to China, India, Japan, and Korea and is a source of flavonoids and polyphenols such as eriodictyol (ERI), fisetin (FIS) and QUE ^[29][37]. ERI and FIS share a similar backbone structure to those of API, KMF, and QUE and have also shown some anticancer potential in several cancer models such as breast, colon, and pancreas cancers ^{[30][31][32][33][34]}[38,39,40,41,42]. In 2020, Li et al. showed that the extract of TV (TVE) inhibited the interaction of PD-1/PD-L1 in a dose-dependent manner (IC₅₀~26 µM in a competitive ELISA). The efficacy of TVE was attributed to the ethyl acetate fraction. TVE, at 5 µg/mL showed an ~30% inhibitory action on the interaction of CTLA4/CD80, with the ethyl acetate fraction being the most effective ^[29][37]. Among several active ingredients in TVE (e.g., ERI, FIS, protocatechuic acid, and caffeic acid), ERI and FIS showed a potent and specific blocking of the PD-L1/PD-1 interaction (IC₅₀ 0.04 µM) with no effect on the CTLA4/CD80 interaction ^[29][37]. Based on the presented results, this low IC₅₀ seemed to be higher than the IC₅₀ of the control, i.e., the PD-L1 inhibitor C1 (value not reported in the paper). Additionally, the binding affinity of these molecules needs to be studied.

6. Caffeoylquinic Acid

Caffeoylquinic acids (CQAs) are a group of phenolic molecules with a quinic acid core that is acetylated with caffeoyl groups. CQAs have shown a wide range of therapeutic activity such as antioxidant, antibacterial, anticancer, antiviral, and anti-Alzheimer’s activities (reviewed in ^[35][43]). In 2018, Han et al. compared the affinity of several mono-CQAs (e.g., 1-CQA, 3-CQA, 4-CQA and 5-CQA) and di-CQAs (e.g., 1,3-diCQA, 1,5-diCQA, 3,4-diCQA, 3,5-diCQA, and 4,5-diCQA) to the affinity of PD-1 and PD-L1. The K_D for the PD-1/PD-L1 interaction was 0.17 µM, while the CQAs showed a weaker but comparable affinity of 0.50–0.81 µM. A surface plasmon resonance competition assay showed that the mono-CQAs had a better inhibitory action on the PD-1/PD-L1 compared to di-CQAs. The IC₅₀ values of 1-, 3-, 4- and 5-CQA were ~87, 37, 38 and 45 µM, respectively ^[36][44].

7. Glyasperin C

Glyasperin C (GC) is a methoxyisoflavan derivative that has been extracted from the ethyl acetate fraction of the traditional herbal medicine, Glycyrrhiza uralensis. The bioactive compounds existing in this fraction were determined to be 10 flavonoids, 4 coumarins and 2 benzophenones. At 100 µM, a 30–65% inhibitory action on the PD-1/PD-L1 interaction was observed with these molecules with the GC showing the highest inhibitory action ^[37][45]. This makes GC another potential candidate. The backbone structure of GC shares some similarities with the previously mentioned flavonoids and, therefore, this mechanism of action could be expected.

8. Ellagic Acid

Ellagic acid (EA) is a chromene-dione derivative that has a hydrophobic moiety of two hydrocarbon rings and a hydrophilic moiety of four hydroxyl groups and two lactones. It is found in a variety of fruits, vegetables and seeds and has several medicinal activities including anticancer, neuroprotective, anti-inflammatory, antioxidant, hepatoprotective, and skin protection actions (reviewed in ^[38][46]). The fruit of Rubus coreanus Miquel, commonly known as black raspberry, has been used in traditional herbal medicine for centuries. The extract of the plant (RCE), which contains polyphenolic and flavonoid molecules such as QUE and EA, has antioxidant and anti-inflammatory effects ^[39][40][41][47,48,49]. Kim et al. in 2020, used RCE in a competitive ELISA and showed a dose-dependent inhibition of the PD-1/PD-L1 interaction (IC₅₀~84 µg/mL), vs. that of anti-PD-L1 antibody, ~1.7 µg/mL. The RCE was non-cytotoxic on aAPC/CHO-K1 and Jurkat cells at <100 µg/mL. In a co-culture system containing these two cell lines, the RCE activated TCR (EC₅₀~56 µg/mL), and at 100 µg/mL it increased the activation of T cells as indicated by an increased production of interleukin 2 (IL-2) by 1.8-fold compared to an untreated control. In a humanized PD-1 mouse model, 50 and 100 mg/kg of orally administered RCE decreased the tumor growth rate by 67% and 74%, respectively. The anti-hPD-L1 antibody at 5 mg/kg showed a 95% decreased tumor growth. None of the treatments affected the mice’s body weights ^[42][50]. EA is the major constituent of RCE. IC₅₀ of EA in blocking the PD-1/PD-L1 interaction in a competitive ELISA assay was ~23 µg/mL. A Western blot analysis showed that EA interacted with both PD-1 and PD-L1. Up to 120.9 µg/mL, EA was non-cytotoxic to Jurkat cells and showed a minor decreased viability in aAPC/CHO-K1 cells at 7.56 µg/mL. At a concentration < 7.56 µg/mL, the EA blocked the PD-1/PD-L1 interaction and showed a dose-dependent increase in IL-2 production ^[42][50].

9. Heterocyclic Compounds

Lung et al. (2020) used the ZBC natural product dataset (180,000 molecules) and 5J89, the dimer structure of the PD-L1 IgV domain protein data bank, to perform a virtual molecular docking screening and contact fingerprint analysis. The top 22 selected molecules were subject to in vitro testing using an AlphaLISA PD-1/PD-L1 binding assay and two molecules, i.e., ZINC67902090 ((3S,3aR,6S,6aR)-N6-[4-(3-fluorophenyl)-pyrimidin-2-yl]- N3-(2-pyridylmethyl)-2,3,3a,5,6,6a-hexahydrofu) and ZINC12529904 (1-isopropyl-3-[(3S,5S)-1-methyl-5-[3- (2-naphthyl)-1,2,4-oxadiazol-5-yl]pyrrolidin-3-yl]urea), inhibited the interaction by 30 and 40%, respectively. The ZINC12529904 was more potent than the ZINC67902090 in increasing the PD-L1 dimerization ^[43][51].

10. Gramicidin S

Gramicidin S (GS) is an antibiotic produced by the bacterium, Bacillus brevis, which is active against some bacteria and fungi. GS is an amphiphilic molecule with a stable β-sheet with hydrophilic and hydrophobic residues. Consequently, due to the amphiphilic properties of the interaction surface of PD-L1 with PD-1, Sun et al. used GS as an anti-PD-L1 candidate ^[44][52]. The GS showed a weak inhibitory action on the PD-1/PD-L1 interaction (~7%), while a synthesized derivative of GS, namely, Cyclo(-Leu-DTrp-Pro-Thr-Asp-Leu- DPheLys(Dde)-Val-Arg, showed a high potency of 95% at 20 µM and a low IC₅₀ of 1.42 µM ^[44][52]. In a B16F10 tumor-bearing mouse model, 40 mg/kg of GS (IP) plus anti-CD8 antibody reduced the tumor volume and tumor weight by~55% and 65%, respectively, while this molecule increased the level of CD3+ T cells and CD8+ CTLs ^[44][52].

11. Rifabutin (RIF)

RIF is a macrocyclic antibiotic mostly known as a treatment for tuberculosis. Using an AlphaLISA human PD1–PDL1 binding assay, Patil et al. (2018) screened RIF together with 19 other FDA-approved macrocyclic molecules for their inhibitory action on the PD-1/PD-L1 interaction. The positive control was an anti-human PD1 antibody with an IC₅₀ of 400 ng/mL. In this assay, at 50 µM, rifampin showed an inhibitory action of 48%. Then, the efficacy of rifampin was compared with four other orally available molecules of this class: RIF, 3-formyl rifamycin, rifamycin SV, and rifapentine. The RIF and rifapentine showed the highest inhibition by ~68% and 52%, respectively. The RIF, rifampin and rifapentine all showed a dose-dependent inhibition of the PD-1/PD-L1 interaction, while the best IC₅₀ belonged to the RIF (25 μM). Based on molecular docking studies, RIF formed a stable complex via several hydrogen bonding and π–π interactions ^[45][53].