*Review* **Natural Carbon Nanodots: Toxicity Assessment and Theranostic Biological Application**

**Ming-Hsien Chan 1,†, Bo-Gu Chen 2,†, Loan Thi Ngo 2,3, Wen-Tse Huang 2, Chien-Hsiu Li 1, Ru-Shi Liu 2,\* and Michael Hsiao 1,4,\***


**Abstract:** This review outlines the methods for preparing carbon dots (CDs) from various natural resources to select the process to produce CDs with the best biological application efficacy. The oxidative activity of CDs mainly involves photo-induced cell damage and the destruction of biofilm matrices through the production of reactive oxygen species (ROS), thereby causing cell auto-apoptosis. Recent research has found that CDs derived from organic carbon sources can treat cancer cells as effectively as conventional drugs without causing damage to normal cells. CDs obtained by heating a natural carbon source inherit properties similar to the carbon source from which they are derived. Importantly, these characteristics can be exploited to perform non-invasive targeted therapy on human cancers, avoiding the harm caused to the human body by conventional treatments. CDs are attractive for large-scale clinical applications. Water, herbs, plants, and probiotics are ideal carboncontaining sources that can be used to synthesize therapeutic and diagnostic CDs that have become the focus of attention due to their excellent light stability, fluorescence, good biocompatibility, and low toxicity. They can be applied as biosensors, bioimaging, diagnosis, and treatment applications. These advantages make CDs attractive for large-scale clinical application, providing new technologies and methods for disease occurrence, diagnosis, and treatment research.

**Keywords:** natural carbon nanodots; tumor-targeting probes; biosensing; cancer theranostic; toxicity assessment

#### **1. Introduction**

The materials studied in the past were at the micrometer scale; however, in the past few decades, nanometer-scale development has changed, which has become an essential direction of scientific and technological development. Nano-sized materials are widely used in various fields, such as medicine, biosensor development, energy research, and catalysis. As fluorescent nanomaterials can emit light, they have the potential for application in biomarking technology, such as semiconductor-based quantum dots containing cadmium sulfide (CdS) or cadmium selenide (CdSe). However, most materials with a high quantum yield (QY, φ) on the market include heavy metal components, which pose risks related to their biological toxicity and cytotoxicity. Contamination may occur during the synthesis process, and they are not adequately recycled after use; as such, the associated harm to the environment should not be underestimated. The above shortcomings limit the application range as semiconductor quantum dots.

In recent years, nanotechnology has been widely used in the field of biomedicine [1]. Generally, when the scale of a substance drops to the nanometer level, its physical and

**Citation:** Chan, M.-H.; Chen, B.-G.; Ngo, L.T.; Huang, W.-T.; Li, C.-H.; Liu, R.-S.; Hsiao, M. Natural Carbon Nanodots: Toxicity Assessment and Theranostic Biological Application. *Pharmaceutics* **2021**, *13*, 1874. https://doi.org/10.3390/ pharmaceutics13111874

Academic Editor: Mazzucchelli Serena

Received: 30 September 2021 Accepted: 29 October 2021 Published: 5 November 2021

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

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

chemical properties also undergo tremendous changes, especially in terms of its optical properties, providing an ideal development space for applications in the field of biomarkers and optical imaging of diseases [2]. Since their inception, carbon nanomaterials have attracted attention from researchers in materials science and biomedicine, especially considering their excellent optical properties, facilitating their application in bioimaging, biomarkers, and sensors [3]. Fluorescent carbon dots (CDs) are another new type of carbon nanomaterial, with a typical particle size of about 10 nm. In addition to the advantages related to the particle size and wavelength-dependent luminescence characterizing quantum dots, fluorescent CDs also have high light stability and present no light scintillation phenomena. Their surface is easily functionalized and modified, and the preparation materials are widely available [4–6]. Wang et al. evaluated the toxicity of CDs and showed that they do not cause any abnormalities or damage to tissues and organs [7]. Kang et al. compared CDs, single-walled carbon tubes, and carbon dioxide [8]. After the cytotoxicities of silicon and zinc oxide were determined, CDs showed the lowest toxicity compared with the materials mentioned above. CDs' unique optical properties and excellent biological properties confer outstanding advantages and good development prospects in biomedical optical imaging and tumor diagnosis. In addition, the ability to introduce multiple functional groups onto the surface of CDs also provides possibilities for their modification.

Green manufacturing is a modern manufacturing model that comprehensively considers the environmental impact and resource benefits. Due to the increasing awareness of environmental protection, green manufacturing is becoming an increasingly important process in all walks of life. The chemical synthesis of CDs can be similar to that produced from various natural resources in the nanoscale process. For instance, Hsu et al. used coffee grounds to synthesize CDs by calcining in 2012. The average particle size of the synthesized CDs was about 5 nm, and they were successfully applied to biological imaging. Unlike the previous need for complex processing procedures, a rapid and straightforward synthesis method was developed. Later, Crista et al. used different organic compounds to synthesize CDs, with an average particle size of about 2.6–7.9 nm, with many carboxyl and amine groups, proving that the quantum yield varies with functional groups [9]. In 2013, Qu et al. used citric acid and ethylenediamine to synthesize CDs, leading to a quantum yield as high as 60.2% [10]. It can be speculated that naturally synthesized carbon nanomaterials should have higher biocompatibility and lower toxicity. To date, many techniques for synthesizing CDs from natural sources have been developed, including laser ablation, arc discharge, electrochemistry, thermal decomposition, ultrasonic, and microwave methods, among others. In particular, the thermal decomposition method is simple, safe, fast, and effective. Research has found that carbon nanoparticles synthesized by this method have good fluorescent performance, although the quantum yield is not high [11–13]. The optical properties, stability, and biocompatibility are still powerful enough for use in bioapplications. This review introduces and simplifies the current carbon nanocomposite synthesis techniques. We provide a brief introduction to fluorescent CDs, mainly reviewing their structural characteristics, carbon source materials, preparation methods, luminescence mechanism, and applications in the field of biomedicine. Obtaining CDs from natural resources can produce nanomaterials that are more environmentally friendly. The considered technique intends to use available (in a daily sense) materials to synthesize low-toxicity fluorescent nanomaterials. The production process does not require the use of many precursor chemicals to build natural CDs. We hope that developing a safe and straightforward production process can unite fluorescent materials and natural CDs and contribute to understanding and exploring the application of nano-fluorescent materials.

#### **2. Natural Carbon Nanodots**

CDs are a new type of carbon nanomaterials with luminescence characteristics, quasizero dimension, relatively simple preparation, abundant sources of raw materials, easy surface functionalization, low toxicity, and good biocompatibility. The fluorescence wavelength can be adjusted, and the two-photon absorption area is large. In various literature

reports, CDs have also been referred to as carbon quantum dots (CQDs), carbon nanodots (CNDs), Graphene quantum dots (GQDs), carbon nanocrystals, and so on. Due to their excellent performance, CDs show promising potential application prospects in sensing, biology, medicine, food, environment, catalysis, photoelectricity, energy, and so on. Researchers have conducted extensive scientific studies and made significant progress [14]. CDs synthesis methods can be mainly divided into two categories: One uses physical or chemical means to crack larger carbon structures (e.g., carbon nanotubes, graphite rods, graphene, carbon powder) into tiny CDs; these top-down methods include arc discharge methods, laser ablation methods, chemical oxidation methods, and so on. Small organic molecule precursors, such as sugar, citric acid, and amino acids, are used as carbon sources for these methods through functional group coupling to achieve polymerization to prepare the CDs [15–17]. The second category comprises bottom-up methods, such as electrochemical, hydrothermal, pyrolysis, microwave-assisted, and ultrasonic methods [18].

Other methods, such as template methods, neutralization reaction exothermic methods, and micro-fluidized bed technology methods, can also be used to prepare carbon dots [19–21]. The preparation of CDs well-embodies the concept of green chemistry, using cheap, environmentally friendly carbon source precursors and natural renewable, cheap raw materials as carbon sources. The resources for synthesizing CDs can be found in the natural environment, such as eggs, grass, tea leaves, leaves, silk, silkworm pupae, shrimp shells, grapefruit peel, peanut shells, coffee grounds, beer, and other materials. CDs were discovered and debuted for the first time due to their fluorescent luminescence characteristics. The luminescence mechanism of CDs has always been a critical research direction for researchers, considering factors such as the quantum size effect, surface state, functional group mechanisms, electron holes, and radiation. Their rearrangement theory has been studied in various aspects. Although a complete theoretical explanation system of the CDs fluorescence mechanism has not been formed, the absorption and fluorescence of CDs exhibit properties such as photoluminescence, chemiluminescence, electrochemiluminescence, and luminescence. The conversion of photoluminescence, peroxidase-like activity, non-toxicity, and other physical and chemical properties provides a solid and feasible theoretical basis for further research.

The main application research directions of CDs can be divided into the following categories (Figure 1):


**Figure 1.** The main application research directions of CDs. This review focuses on summarizing the biological applications of CDs, such as cell imaging, photocatalysis, optoelectronic devices, molecules sensors, electrocatalysis, and biomedicines, comprehensively. Finally, current challenges, research emphasis, and prospects of this field are also discussed.

Most carbon sources derive from exhausted petroleum resources or non-environmentally friendly manufacturing processes. Therefore, using recycled materials and bio-renewable resources to develop high-performance CDs is a critical green environmental issue. In recent years, the awareness of environmental protection has significantly risen, and, as such, naturally derived CDs have gradually received more attention.

The main factor is that they are derived from renewable and sustainable biological materials; for example, lignocellulose from dead wood, waste wood, rice straw, bagasse, wheat straw, etc. Suppose that such a raw material can be used as a carbon resource and converted into CDs with therapeutic and diagnostic value. In that case, we can imagine the result will not overwhelm the demand of the food supply chain. Still, it can also address waste disposal problems while applying the resource to produce high value-added products such as electronics, energy, and biomedicine. Table 1 summarizes the existing research on converting various natural resources into CDs, and the associated applications and emission spectra are described in detail.

*Pharmaceutics* **2021**, *13*, 1874


**Table 1.** Natural CDs and their applications.

*Pharmaceutics* **2021**, *13*, 1874








**Table1.**



#### **3. Toxicity Evaluation of Natural Carbon Nanodots**

CDs can be derived from a wide range of synthetic raw materials, low-cost, stable chemical properties, and non-toxic materials. The application of CDs in medicine and pharmacy was recently extensively studied. One of the most promising nanomaterials is carbon quantum dots (CQDs). For this purpose, the toxicity of CQDs was investigated in cells and living systems (Table 2). In 2019, L. Janus used human dermal fibroblasts to conduct a cytotoxicity study of N-doped chitosan-based CDs [129]. As shown in Figure 2, after 48 h, the cell viability was recorded as 94%. CDs were synthesized by utilizing non-toxic raw materials and removing unreactive residues during purification.

**Figure 2.** TXT assay study on human dermal fibroblasts. Adapted from [129], MDPI, 2019.

First, we start with the selection of carbon source and the adjustment of dopants that synthesize carbon dots, and at the same time list the test models of carbon dots. Secondly, according to the selective response of the carbon dot to the biological models, the corresponding toxicity was constructed, and the actual application was investigated. As mentioned in Table 2, CQDs are usually doped with nitrogen to enhance their fluorescence quantum yield and optical performance [130,131]. CDs doped with nitrogen (such as carbon nanotubes, graphene, hollow spheres, etc.) have unique properties. They can inject electrons into the carbon substrate to change the electron and transport characteristics such as sensors, nanogenerators, etc., and are widely used and have become a research trend. Moreover, green methods that utilize natural biomass/biowaste and micro-organisms without introducing toxic chemicals as CDs precursors have been widely used to synthesize CQDs [132].



**Table 2.** *Cont*.

S. Cong et al. [122] used PC12 cells for a cytotoxicity study of CQDs obtained from a roast duck. After 36 h, the cell viability of PC12 was recorded as 91.19% at the concentration of 1 mg/mL. In addition, using CDs at a concentration of 15 mg/mL for treating nematodes did not lead to any death for 24 h. These results indicate the low toxicity of CDs, even after a long period of exposure at high concentrations. In 2021, R. Atchudan et al. [141] used *C. elegans* as an animal model for their toxicity evaluation of CQDs. As displayed in Figure 3, the CQDs synthesized from banana peel were shown to have low nematode toxicity, even at a high concentration of 200 ug/mL. These results can be explained by their utilizing non-toxic raw materials without adding any passivates or additives.

**Figure 3.** (**a**) Toxicity assay of nematode incubation under different concentrations of synthesized CQDs. Multicolor imaging of in vivo model nematode incubated with CQDs (100 μg/mL) under the excitation wavelengths of (**b**) 400 nm, (**c**) 470 nm, (**d**) 550 nm, (**e**) bright-field (BF), and (**f**) merge (overlap). Live nematodes were immobilized using 0.05% sodium azide (NaN3) for imaging under fluorescence filters. Adapted with permission from [141], Elsevier, 2021.

#### **4. Theranostic Application of Natural Carbon Nanodots**

Therefore, fluorescent materials are expected to be critical in biological applications. CDs have become the focus of attention as new nanomaterials due to their excellent light stability, fluorescence, good biocompatibility, and low toxicity. Surgery, radiotherapy, and chemotherapy are inevitable treatments for most cancer patients, but these processes have considerable side effects on the human body. However, fluorescent nanomaterials have the advantages of high fluorescence stability, low biotoxicity, and good biocompatibility. The most important thing is that they can be used to perform non-invasive targeted therapy on human lesions, exploiting their characteristics to avoid harm to the human body caused by the abovementioned treatments.

This paragraph covers the luminescence mechanism of CDs and their applications in biology, focusing on applying natural CDs in biological diagnosis and treatment. We discuss the combination of CDs with specific targeting molecules to form CD-based probes for detecting fluorescent signals. With the help of advanced optical imaging technology, real-time dynamic monitoring of molecules in cells and organisms can be carried out, and rapid immunofluorescence analysis of primary infectious disease sources can be carried out. They can provide new technologies and methods for disease occurrence, diagnosis, and treatment research [150].

#### *4.1. Bioimaging*

The discrete and diverse microstates of CDs lead to broad excitation and emission ranges [151,152]. CDs have many unique properties, and their excellent photostability can provide fluorescent information in a biological environment. The surface modification of functional groups can make CDs more helpful in applying biomarkers (Table 3). In this section, we compare bioimaging from cells to living animals, emphasizing the biological diversity to prove the generalized safety of CDs. In cell imaging applications, CDs are usually applied to HepG2 cells, HeLa cells, T24 cells, and so on [27,102,120,124]. In addition, normal cell lines have been used in trials, showing good cell compatibility [24,71]. Alam et al. treated HaCaT cells with cabbage-derived CDs and showed that, at 500 μg/mL of CDs, the cell viability was 100%. Furthermore, CDs' tunable excitation and emission show promise for normal cell imaging under the irradiation of confocal fluorescence microscopy [71]. In a cell imaging experiment, Mehta et al. considered CDs originating from apple juice and three fungal (*M. tuberculosis*, *P. aeruginosa*, and *M. oryzae*) sources. Cell compatibility was shown by feeding more than 100 mg/mL of CDs. The germination of *M. oryzae* spores also strongly indicated how CDs are good biocompatible nanocomposites at high concentrations (i.e., 400 mg/mL). The fungi appeared red, green, and blue in confocal laser microscopic images [62].

Kasibabu et al. provided pictures of *Bacillus subtilis* and *Aspergillus aculeatus* after incubating with CDs derived from papaya juice for 1–6 h. The uptake ability was shown by observing well-dispersed CDs in the cytoplasm of the fungi [90]. In vivo tests are critical standards for investigating the potential and toxicity of CDs in animals. Atchudan et al. explored colorful nitrogen-doped CDs (NCDs) derived from gooseberry by hydrothermal methods in *C. elegans* imaging. *C. elegans* presented blue, green, and red in the whole body when excited under 400 nm, 470 nm, and 550 nm. The cell viability was over 97% after incubating *C. elegans* in NCDs for 24 h from 0 to 200 μg/mL (Figure 4a–f) [96]. Cong et al. using roast duck as the source and pyrolyzing at 200 ◦C, 250 ◦C, and 300 ◦C, synthesized single-fluorescence CDs. *C. elegans* were treated with 15 mg/mL of the CDs pyrolyzed at 300 ◦C (300-CDs) for 24 h. Compared with the wild-type group, the accumulation and uptake of 300-CDs made the intestine appear blue under UV light exposure (Figure 4g–j) [122]. The murine model has also been widely used for determining the efficiency of CDs.

Ding et al. subcutaneously injected 100 μL of red-emitting CDs (R-CDs) into nude mice. Strong fluorescence at 700 nm was detected under an excitation wavelength of 535 nm, indicating the excellent penetration ability from tissues to skin. Furthermore, the mice were still alive after 10 days (Figure 5a) [22]. Liu et al. investigated how the accumulation of carbonized polymer dots (CPDs) varied with time. Most CPDs remained in the lung and liver in the early period, with negligible dispersion to the brain and heart. The metabolism of CPDs was confirmed after 4 h, as their fluorescence decreased sharply (Figure 5b) [23]. Ding et al. compared subcutaneous and intravenous injections of near-infrared emissive CDs (NIR-CDs). Through the use of subcutaneous methods, NIR-CDs were distributed into mouse skin and tissues. As for intravenously injected nude mice, fluorescence was seen in the bladder, indicating NIR-CDs elimination via urine (Figure 5c) [26]. Liu et al. tracked CDs (Fn-CDs) in Kunming mice for different periods.

**Figure 4.** *C. elegans* confocal imaging excited under the wavelengths of (**a**) 400 nm, (**b**) 470 nm, and (**c**) 550 nm; as well as (**d**) bight field; (**e**) the merged image; and (**f**) cell viability test under different concentrations of NCDs, adapted with permission from [96], ACS Publications, 2018. Bright-field (**g**) wild-type *C. elegans* imaging and (**i**) 300-CDs treated *C. elegans* imaging; UV-exposed (**h**) wild-type *C. elegans* imaging and (**j**) 300-CDs treated *C. elegans* imaging, adapted with permission from [122], Elsevier, 2019.

They concluded that, with the assistance of PL, Fn-CDs show strong fluorescence in the bladder and are eliminated after 7 h (Figure 5d) [28]. Therefore, CDs can perform well as imaging nanoparticles and be adapted to different cell lines and living animals. In addition, their multiple emission ranges, low toxicity, and small size confer their high potential in future clinical applications.

**Figure 5.** In vivo images and accumulation of CDs: (**a**) in vivo images of R-CDs, adapted with permission from [22], Royal Society of Chemistry, 2017; (**b**) metabolism of CPDs, adapted with permission from [23], Wiley, 2020; (**c**) subcutaneous and intravenous injection of NIR-CDs, adapted with permission from [26], Elsevier, 2019; and (**d**) metabolism of Fn-CDs, adapted with permission from [28], Royal Society of Chemistry, 2021.

#### *4.2. Sensors*

Inorganic ions are critical for creatures, not only for enhancing the efficiency of catalytic reactions in bio-systems but also for maintaining our fundamental life functions. However, it is harmful to have too many metal cations, which are highly toxic to human beings. Therefore, developing sensors for inorganic ions is a simple beneficial method to collate the concentration and standard. Owing to the multiple PL of CDs, we can observe the intensity variation and chelation quenching effect at other peaks in the PL image. For natural CDs sensors, Fe3+, Hg2+, and Cu2+ are the most examined targets. Shen et al. assessed HepG2 and HeLa cell images after incubating with CDs and Fe3+.

According to the decreased blue fluorescence under 405 nm irradiated light, Fe3+ had a practical quenching effect on the CDs (Figure 6a–d) [33]. Hu et al. prepared double–emitted biomass nitrogen co-doped CDs (B-NCdots) for Cu2+ probing in T24 cells. Similarly, the quenching effect was still available for Cu2+, causing a decreased intensity of blue and green, as shown in Figure 6e,f [117]. Furthermore, bio-related molecules, including peptides and drug-containing cells, are even more crucial. Some researchers designed chemically sensitive CDs to assist in directly resolving the effects of molecules by monitoring the decrease or increase in fluorescence. Liang et al. added 0.5 mM and 1 mM of glutathione and 150 μg/mL rose-red fluorescence CDs (wCDs) in L929 cells, HeLa cells, and HepG2 cells. An intense quenching effect of glutathione was observed in L929 cells and HeLa cells. However, no apparent variation occurred in HepG2 cells, which implies a distinct response of wCDs to glutathione (GSH) in various cells (Figure 7) [24].

**Figure 6.** Confocal images collected at 405 nm of (**a**) HeLa cell; (**c**) HepG2 cell incubated with CDs; adding Fe3+ and CDs to (**b**) HeLa cell; and (**d**) HepG2 cell, adapted with permission from [33], Elsevier, 2017. Confocal laser images of T24 cells excited at 405 nm and 488 nm, (**e**) treated with B-NCdots and (**f**) treated with Cu2+ and B-NCdots, adapted with permission from [117], Springer Nature, 2019.

Wang et al. constructed a glutathione assay composed of eggshell-derived CDs and Cu2+. The authors quantify without other indicators by plotting the fluorescence ratio versus the glutathione concentration (Figure 8a) [107]. Wang et al. examined how the fluorescence intensity of Shiitake mushroom-derived CDs (MCDs) varied with pH in dexamethasone-induced HeLa cell apoptosis. At higher concentrations of dexamethasone, the fluorescence under excitation at 405 nm and 488 nm was stronger, indicating a positive relationship on MCDs with increasing intracellular acidification (Figure 8c–e) [66]. As for drug probing, Zhu et al. analyzed doxorubicin (DOX), an anthracycline-based anticancer medicine, by taking advantage of the PL of plum-based carbon quantum dots (PCQDs). The dual-emitted property at the wavelengths of 491 nm and 591 nm provided a ratiometric calibration curve as a function of the DOX concentration. They also confirmed the accuracy by analyzing urine and serum samples (Figure 8b) [27]. CDs are suitable for detecting different kinds of molecules and ions. The intensity changed at a single emission peak, but the amplitude ratio of two emission peaks is valid for sensing experiments. Due to their intrinsic properties, CDs show promise in the bio-sensing field and are applied to cancer therapy.

#### *4.3. Antibacterial Activity*

Bacteria are well-known as the origins of various diseases. Recently, super bacteria have appeared globally, which cause incurable illnesses due to the abuse of antibiotics. In addition, people have come to pay more attention to the side effects of antibiotics and wish to avoid unexpected risks. Nanomedicines, especially CDs, have been taken into consideration as substitute methods. *E. coli* and *S. aureus* are the most common types of bacteria for investigating how nanomedicines or antibiotics work to induce apoptosis in bacteria. Wang et al. used CDs (ACDs) derived from *Artemisia argyi* leaves to treat *E. coli* and *S. aureus* cultures. According to the SEM images (Figure 9a–h), it can be seen that the cell walls of *E. coli* were destroyed; however, there was no distinct difference between treated and untreated *S. aureus*. This means that ACDs are selective to Gram-negative bacteria due to the structural properties of their cell walls [123]. Sun et al. synthesized chlorhexidine gluconate CDs from large to small (l-CGCDs, m-CGCD, and s-CGCDs) to determine the relationship between size and antibacterial activity. From the SEM imagery (Figure 9i), it can be seen that the rigidity of cell walls was the strongest in the control group and decreased from l- to s-CGCDs groups. These results revealed that CGCDs lead to frustration in the walls and membranes of *E. coli* and *S. aureus*. Bacterial death can be controlled by tuning the size of the CGCDs [76]. Ma et al. tested three kinds of CDs,

including osmanthus leaves-derived CDs (OCDs), milk vetch-derived CDs (MCDs), and tea leaves-derived CDs (TCDs). In Figure 10, 80% of *E. coli* and *S. aureus* were killed by OCDs at a 1 mg/mL concentration, while 70% of bacteria survived in the MCDs group. In addition, *E. coli* had stronger resistivity than *S. aureus* among these CDs. CDs are internalized into bacteria. The outer surface of bacteria is attached to CDs leading to indirect proliferating inhibition [153–155]. These results prove the natural sources are essential for the synthesis of CDs [79]. As mentioned above, the tunability of raw material and diameters primarily affect the antibacterial efficiency and selectivity of the resultant CDs. Treating the patient's wounds after surgery with CDs with editable properties can tremendously decrease the associated risks.

**Figure 7.** Confocal laser images of: (**a**) L929 cells, (**c**) HeLa cells, and (**e**) HepG2 cells in 0.5 mM GSH and 150 μg/mL wCDs; and (**b**) L929 cells, (**d**) HeLa cells, (**f**) and HepG2 cells in 1 mM GSH and 150 μg/mL wCDs, adapted with permission from [24], Royal Society of Chemistry, 2021.

**Figure 8.** (**a**) Linear calibration of GSH probing, adapted with permission from [107], Royal Society of Chemistry, 2012. (**b**) The fluorescence spectra of PCQDs with varying DOX concentrations, adapted with permission from [27], Elsevier, 2021. (**c**) Confocal images of HeLa cells treated with MCDs, (**d**) adding 10 μM dexamethasone or (**e**) 100 μM dexamethasone, at excitation wavelengths of (**c1**,**d1**,**e1**) 405 nm, (**c2**,**d2**,**e2**) 488 nm, and (**c3**,**d3**,**e3**) bright-field, adapted with permission from [66], Royal Society of Chemistry, 2016.

**Figure 9.** SEM images without ACDs of (**a**) *E. coli*, (**e**) *S. aureus*, and ACDs-treated (**c**) *E. coli*, and (**g**) *S. aureus*. Magnified SEM images in the red square (**b**) *E. coli*, (**f**) *S. aureus*; and ACDs-treated (**d**) *E. coli*, and (**h**) *S. aureus*, adapted with permission from [123], Royal Society of Chemistry, 2020. SEM images of (**i**) *E. coli* and *S. aureus* untreated and treated with 75 μg/mL and 50 μg/mL of s-CGCDs, m-CGCDs, and l-CGCDs in Luria–Bertani broth medium for 6 h, adapted with permission from [76], Elsevier, 2021.

#### *4.4. Anticancer Activity*

At present, cancers are prevalent within all age ranges. Cancers may be fatal due to unexpected syndromes as well as the disorder of living functions. Furthermore, conventional cancer therapies are long-term processes. Surgeries are straightforward methods, but recovery typically poses a challenge for patients. Even though chemotherapy seems safer, the currently used drugs lack selectivity and affinities to specific tumors. Some targeted therapies have been developed in recent years. However, they are expensive and only valid for certain types of cancers. CDs can provide great theranostic nanomedicines in cancer treatments. Scientists have attempted to eliminate cancer cells through photothermal therapy (PTT) [81,82] and photodynamic therapy (PDT) [83,87,129] to fulfill tumor targeting. Li et al. tested NIR-II emitted (900–1200 nm) CDs (CDs), adapted for 808 nm laser photothermal therapy. According to the in vivo test (Figure 11), the temperature

increased to 50 ◦C in the intratumoral environment after intravenous injection. Tumor inhibition and volume shrinkage were observed within 6 days, compared with the PBS group. No detectable damage to tissues or weight loss after the treatment confirmed the high biocompatibility of the CDs [156]. Jia et al. prepared red-light absorbing (610 nm) CDs from *Hypocrella Bambusa* (HBCUs). They found that HBCDs highly generate 1O2 under 635 laser irradiation. The reactive radicals induced apoptosis of cancer cells, which is helpful in the hypoxia tumor environment.

As shown in an in vivo experiment (Figure 12), due to the synergistic effect of PDT and PTT, the temperature at the tumor site increased to 56.4 ◦C in 10 min. Secondly, a good tumor inhibition effect was found after 14 days of therapy, even though the tumor could not be depleted thoroughly. No harmful phenomena were observed in other organs, indicating the safety of the treatment [82]. Xue et al. conducted modification with polyethylene glycol diamine (H2N-PEG-NH2), chlorin e6 (Ce6), and transferrin (Tf) on natural biomass CDs (NBCDs) to increase the targeting efficacy.

**Figure 10.** Graphs of *E. coli* and *S. aureus* incubated with varying concentrations of OCDs, TCDs, and MCDs for 24 h; alive ratio of *E. coli* and *S. aureus* calculated using UV–Vis spectroscopy methods, adapted with permission from [79], Elsevier, 2020.

**Figure 11.** (**a**) Schematic diagram of photothermal therapy of cancer in vivo; (**b**) infrared thermal images of tumor-bearing mice with intravenous (i.v.) or intratumoral (i.t.) injection of CDs (50 μL, 20 mg/mL), and PBS (50 μL); (**c**) tumor volume after treatment; and (**d**,**e**) images of tumor-bearing mice and harvested tumors after 6 days, adapted with permission from [156], ACS publications, 2019.

The resulting products, NBCD-PEG-Ce6-Tf, were shown to remain within the tumor environment for 120 h using a real-time NIR fluorescence image (Figure 13a). The mice were irradiated daily under a 650 nm laser to generate 1O2. During the 21-day process, tumor growth was stopped, and the tumors were ablated, indicating no conflict between the NBCDs and modifications (Figure 13b,c) [126]. Li et al. synthesized reactive oxygen species (ROS)-generating CDs from ginger. The CDs were harvested from HepG2 tumor inoculating mice; next, the tumor regressions were observed in the C-dot (440 μg) treated group; the tumor growth was prominently delayed, which attained only 3.7 ± 0.2 mg. In contrast, tumors in the PBS group grew up to 104 mg [84]. Boobalan et al. added 30 μg/mL of CDs into *P. aeruginosa*. They observed the destruction of cell walls due to ROS attack, in agreement with the results of ROS fluorescence detection using a fluorogenic dye, 2- ,7- -dichlorofluorescein diacetate (DCFDA) (Figure 14a–c). MDA-MB-231 breast cancer cells were treated with CDs (3.34 μg/mL). Cell apoptosis staining, acridine orange and ethidium bromide (AO/EtBr), and nuclear staining (Hoechst 33342) were applied to the

cells. The presence of orange colors and blue dots indicate cell fragmentation due to CDs (Figure 14d–g) [80]. CDs have a powerful potential in the anticancer field. Their flexibility is because the CDs are modified with various molecules, which can improve the uptake by tumor cells and increase the tumor-killing ability of the nanohybrids.

**Figure 12.** (**a**) IR thermal images post-injection; (**b**) temperature increase trends; (**c**) images of mice with different therapeutic methods; and (**d**) growth of the tumor during 14 days, adapted with permission from [82], Elsevier, 2018.

**Figure 13.** (**a**) Real-time NIR fluorescence image of mice tumors under NBCD-PEG-Ce6-Tf treatment at different time points; tumor propagating images for (**b**) NBCD-PEG-Ce6-Tf + laser-treated and (**d**) control groups; (**c**) time-dependent tumor growth curves after other treatments, adapted with permission from [126], Royal Society of Chemistry, 2018.


**Table 3.** Modifications of CDs.

#### **5. Discussion and Conclusions**

Carbon nanomaterials have been widely used in various scientific, engineering, and commercial fields, due to their high catalytic activity and good stability. Among them, the new "zero-dimensional" carbon nanomaterials, CDs, have unique optical properties, such as stable fluorescence signals, no light scintillation, adjustable excitation and emission wavelength, and low biological toxicity and biocompatibility. These advantages gradually led to the popularity of researching carbon nanomaterials, widely used in bioimaging, natural cell labeling, sensors, photocatalysis, solar cells, and light-emitting elements. This article mainly reviewed the different synthesis methods of CDs (including top-down and bottom-up methods) and their applications. Their luminescence properties can be adjusted through surface modification. They have been applied in many fields and have great potential. The function of CDs can also be modified by using various surface functional groups, allowing them to act, for example, as detectors and cleaners for different heavy metal ions or by doping with other ions. By controlling the surface light energy groups, they can be better used in the required fields.

Green chemistry is a discipline that has gradually received attention in recent years. The core concept focuses on the development of environmentally friendly chemical technologies. At the technical level, chemical technologies and methods are applied to reduce or eliminate the use and generation of hazardous substances in chemical synthesis and analysis, and recovery and reuse technologies are combined with increasing energy and material use efficiency. Green chemistry and nanotechnology have become emerging technology research and development directions in recent years. With the deepening of the concept of sustainability, combining the advantages of the two and accelerating the

expansion of their research and development applications has become a top priority. Laboratories are committed to determining the complementary relationships between green chemistry practices and nanotechnology and applying them to materials development, chemical analysis, energy, environmental, and other related fields. The preparation of CDs well embodies the concept of green chemistry. Cheap, environmentally friendly carbon source precursors and natural renewable raw materials as carbon sources for preparation, such as eggs, grass, leaves, silk, coffee grounds, beer, and other materials, have become carbon sources for the synthesis of CDs.

Excellent performance and a unique structure provide natural CDs unlimited charm and various changes. Natural CDs, combined with biological and pharmaceutical molecules of interest through surface modification, seem to be an emerging platform for imaging probes that are both diagnostic and therapeutic. The next generation of nano-molecular probes integrates a variety of fluorescent dyes, drugs, and multifunctional nanomaterials into a single nanoprobe, providing superior signal contrast, controllable transmission, and targeted drug delivery capabilities. However, before this kind of multifunctional imaging probe was used in diagnosis and treatment, there were still many challenges, such as long-term safety, risk-benefit, biocompatibility, and biodistribution, to be evaluated. In the future, we need to solve several critical scientific problems in the research of natural CDs. First of all, the uncertain chemical groups on the surface indicate that natural-synthetic CDs are a kind of unsure material. It means the method of natural mass production of high-quality CDs is still a big challenge. Secondly, due to the different sources of naturally transformed CDs, the luminescence centers of CDs are also dissimilar. Finding a suitable luminescence position is also essential to research content. Third, categorizing different natural CDs and conducting a systematic comparative analysis will be beneficial research methods. Nanotoxicology is the emerging study of potential adverse effects derived from the interaction between nanomaterials and biological systems, and it is bound to become more critical. To further clarify its physical toxicity and adjust the size and structure accordingly, it will significantly improve CDs' application performance while expanding a more comprehensive range of applications. The scale and complexity of biomedical issues have always been an enormous challenge for researchers. It is more necessary to conduct cross-disciplinary research in chemistry, physics, materials, biomedical engineering, toxicology, public health, and clinical medicine. As more research on natural CDs continues to develop, the topic will achieve breakthroughs and progress in a short period.

**Author Contributions:** Conceptualization, M.-H.C., B.-G.C., R.-S.L. and M.H.; writing—original draft preparation, M.-H.C., B.-G.C. and L.T.N.; writing—review and editing, M.-H.C., B.-G.C., L.T.N., W.-T.H. and C.-H.L.; supervision, R.-S.L. and M.H.; project administration, R.-S.L. and M.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by Genomics Research Center, Academia Sinica, Taiwan, to Michael Hsiao. This work was also financially supported by the Ministry of Science and Technology in Taiwan (MOST 109-2113-M-002-020-MY3) to Ru-Shi Liu. Academia Sinica Outstanding Postdoctoral Fellowship supported to Ming-Hsien Chan.

**Acknowledgments:** The authors would like to express their gratitude to support from Genomics Research Center, Academia Sinica, Taiwan.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**



#### **References**


### *Review* **RNAi-Based Approaches for Pancreatic Cancer Therapy**

**Min Ju Kim 1,2, Hyeyoun Chang 1, Gihoon Nam 1,2, Youngji Ko 1, Sun Hwa Kim 2, Thomas M. Roberts 1,\* and Ju Hee Ryu 2,\***


**Abstract:** Pancreatic cancer is one of the most lethal forms of cancer, predicted to be the second leading cause of cancer-associated death by 2025. Despite intensive research for effective treatment strategies and novel anticancer drugs over the past decade, the overall patient survival rate remains low. RNA interference (RNAi) is capable of interfering with expression of specific genes and has emerged as a promising approach for pancreatic cancer because genetic aberrations and dysregulated signaling are the drivers for tumor formation and the stromal barrier to conventional therapy. Despite its therapeutic potential, RNA-based drugs have remaining hurdles such as poor tumor delivery and susceptibility to serum degradation, which could be overcome with the incorporation of nanocarriers for clinical applications. Here we summarize the use of small interfering RNA (siRNA) and microRNA (miRNA) in pancreatic cancer therapy in preclinical reports with approaches for targeting either the tumor or tumor microenvironment (TME) using various types of nanocarriers. In these studies, inhibition of oncogene expression and induction of a tumor suppressive response in cancer cells and surrounding immune cells in TME exhibited a strong anticancer effect in pancreatic cancer models. The review discusses the remaining challenges and prospective strategies suggesting the potential of RNAi-based therapeutics for pancreatic cancer.

**Keywords:** RNA interference; siRNA; miRNA; pancreatic cancer; nanocarrier

#### **1. Introduction**

Pancreatic cancer is one of the most aggressive malignancies, ranking as the fourth leading cause of cancer-related deaths with a five-year survival rate of 8% [1]. If outcomes are not improved, pancreatic cancer is predicted to be the second leading cause of cancerrelated mortality by 2025 [2]. With a minimal increase in patient survival rate in past four decades, pancreatic cancer is often discovered and diagnosed at advanced stage where metastasis have already taken asymptomatically, and limited treatment options are left [3]. Because of aggressive perineural and vascular local growth, chemotherapy, radiotherapy, and molecularly target therapies are better treatment options than surgical resection. Typically, chemotherapy elongates the life span only by 8 to 16 weeks, and there is an urgent need for new therapeutic options for pancreatic cancer patients [4]. In particular, pancreatic cancer is characterized both by complicated genetic mutations and epigenetic alterations and by development of a dense tumor microenvironment that together result in a low survival rate of just ~15–25% of patients who undergo surgical resection [5]. One of the unique histological features of pancreatic cancer is a dense fibrotic stroma surrounding the tumor that comprises around 80% of the whole tumor mass [6]. In addition to the early metastasis and robustly immunosuppressive TME, this protective stromal barrier and distorted tumor blood vessels contribute greatly to poor survival by increasing the difficulty of drug delivery and distribution [7–10]. In addition, the abnormal vasculature induces

**Citation:** Kim, M.J.; Chang, H.; Nam, G.; Ko, Y.; Kim, S.H.; Roberts, T.M.; Ryu, J.H. RNAi-Based Approaches for Pancreatic Cancer Therapy. *Pharmaceutics* **2021**, *13*, 1638. https://doi.org/10.3390/ pharmaceutics13101638

Academic Editors: Sophia Hatziantoniou

Received: 20 August 2021 Accepted: 27 September 2021 Published: 8 October 2021

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

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

hypoxic environment in TME which contributes to resistance towards chemotherapy and immunotherapy [11,12]. Desmoplastic features with an abundance extracellular matrix are known to be developed by the cancer-associated pancreatic stellate cells (PSCs) in communication with the surrounding cells in the TME [13]. The PSCs that are normally quiescent in the healthy pancreas are recruited upon cancer cell communication to fuel the aggressive proliferation of tumors [14–16]. Given the clear importance of TME in tumorigenesis, approaches targeting specific features within TME have been investigated in addition to the focus on directly targeting tumor cells.

In recent decades, studies have delineated the genetic mutations unique to pancreatic cancer. For example, 94% of pancreatic tumors feature mutations along with mutations in the *TP53* (64%), *SMAD4* (21%), and *CDKN2A* (17%) tumor suppressor genes [17]. The largest challenge with the genetic aberrations of pancreatic cancer is that the *KRAS* oncogene has been 'undruggable' using the now conventional treatment approach of targeted chemical inhibitors [18]. Undruggable targets represent proteins that is pharmacologically incapable of being targeted due to elusive factors from molecular shapes and mechanism of action. For this reason, the potential to selectively inhibit undruggable genes with RNAi represents a promising strategy to halt pancreatic cancer progression. RNAi is a post-transcriptional mechanism that involves the inhibition of specific gene expression through promoting cleavage of the relevant messenger RNA (mRNA). After decades of investigation, in 2018 the first RNAi-based drug, Patisiran (ONPATTRO®) finally proved the potential of siRNA therapeutics as demonstrated by FDA approval and market entry [19]. In the following years, Givosiran (GIVLAARI®) and Lumasiran (OXLUMO®) were FDA approved in 2019 and 2020, embarking the era of RNAi-based oligo therapeutics. Thus, RNAi-based technology has presented an opportunity to regulate a set of tumorpromoting genes in cancer cells and other targets in TME. However, the clinical application of RNAi therapeutics is still hindered by poor delivery to the target organs and inefficient internalization into target cells. In this review, we discuss the prospects and challenges in RNAi-based approaches using various strategies of nanomedicine to target key genes in cancer cells or TME of pancreatic cancer.

#### **2. RNAi Targets for Pancreatic Cancer Therapy**

RNAi is a regulatory mechanism using small regulatory RNAs to prevent translation of mRNA and regulate protein expression levels. As an intrinsic mechanism for posttranscription editing, RNAi comprises two major different types of RNA molecules, siRNA, and miRNA, generated via similar mechanisms involving the dicer protein (Figure 1). Both siRNA and miRNA are short non-coding RNA strands that can silence the target mRNA in sequence-specific manner, therefore offering the advantage of greater selectivity and expanding the inhibitory action for undruggable targets. Despite numerous challenges involving delivery, safety, and potency, RNAi-based drugs have shown therapeutic potential in rare genetic diseases and complex diseases such as cancer [20–24].

siRNA is a 20 to 25 base pairs double-stranded non-coding RNA that plays a key role in RNAi machinery. Once internalized in a cellular system, the siRNA forms an activated RNA-induced silencing complex to guide pairing with a target mRNA of complementary sequence resulting in cleavage and degradation [25,26]. The ability of siRNA to silence specific gene expression involved in diseases has led to a significant effort to use it as an alternative treatment option. Despite the breakthroughs in the generation of siRNA therapeutics beyond preclinical studies, siRNA delivery into cells at the disease site has remained the major hurdle in various clinical applications. The challenges include the negatively charged nature of siRNA, immediate degradation by the serum proteins, and fast clearance from the reticuloendothelial system [27]. Furthermore, the dense fibrotic stroma and vascular barriers of pancreatic cancer add an extra layer of difficulty in successfully delivering siRNA to the tumor and surrounding cells in TME. We will use this forum to introduce siRNA-based approaches to pancreatic cancer therapy using various reagents and strategies.

miRNAs are 19–25 nucleotides long, endogenous non-coding RNAs found in living organisms as one of the key players in the RNAi machinery as shown in Figure 1 [28]. Similar to siRNA, miRNA interfere with the expression of specific proteins by binding to the mRNA strand. miRNA is a long bimolecular RNA duplex that partially binds to the untranslated region of multiple genes to interfere with the post-transcriptional expression. Contrary to the direct and specific inhibition of siRNA, miRNA often hinders the translation of several mRNAs via closely related binding sites in the 3 regions of target mRNAs [29]. Compelling evidence has demonstrated that miRNA is involved in the cellular machinery and signaling contributing to the dysregulation of cancer cell growth [30]. miRNA may function as oncogenes or tumor suppressors interfering with a set of specific protein expression. The dysregulation of miRNA is known to be prevalent across cancer types, contributing to sustaining proliferative signaling, evading growth suppressors, resisting cell death, activating invasion, and inducing metastasis and angiogenesis. For this reason, miRNA-based therapeutics that inhibit the levels of oncogenic signaling or elevate tumor suppressor functionality have enormous potential [31].

**Figure 1.** Gene silencing mechanism of siRNA and miRNA of RNAi.

#### *2.1. Pancreatic Tumor Targets*

Conventional treatment approaches are focused on the eradication of pancreatic cancer using chemotherapies. The potency of a of given procedure is related to its ability to effectively remove cancerous cells; however, many therapies often harm the normal organs causing major side effects in patients [32]. To overcome these drawbacks, targeted therapies have been advanced as the next generation of cancer therapies. The carcinogenesis of pancreatic ductal adenocarcinoma (PDAC), comprising 90% of the pancreatic cancers, involves progressive accumulation of key driver mutations, usually leading to the activation of the KRAS oncogene and loss of the tumor suppressor gene TP53, contributing to the histological evolution of pancreatic intraepithelial neoplasia to PDAC [33–35]. These genetic aberrations induce invasive adenocarcinoma and remodel the surrounding stromal structure into one more favorable for tumor formation. Because of the unique

stromal feature and frequent oncogenic mutations, wide range of pancreatic cancer models exist for investigation for varying purposes as summarized in Table 1.

Often, the genes discovered to be the novel cancer-related oncogenes that promote tumor progression are found to be 'undruggable,' since they are difficult to inhibit using small drug molecules [36]. Because of lack of suitable ligand binding sites or the presence of large protein-protein interaction interfaces, various oncogenes are relatively intractable using small-molecule inhibitors [37]. A notable undruggable target, mutationally activated KRAS, is present in >90% of PDAC and represents the most frequent (>90%) and the earliest genetic alternation, found even in low-grade pancreatic intraepithelial neoplasia 1A lesions [38,39]. Moreover, KRAS activation is also a poor prognostic factor resulting resistance to traditional treatments [5,39,40]. Genetically engineered mouse models also support the initiating role of KRAS mutation in PDAC and its involvement in progression to the malignancy [41,42]. When coupled with mutational inactivation of TP53 (TP53R175H), CDKN2A, or SMAD4, pancreatic intraepithelial neoplasia formation rapidly evolves into high-frequency occurrence of metastatic PDAC [41,43,44]. In recent decades, KRAS has been an intractable challenge as the therapeutic target with no targeted inhibitors. Recently a breakthrough was made with the generation of KRASG12C inhibitors which have proved effective in non-small cell lung cancer patients in phase I clinical trials; however, only 2% of the PDAC carry KRASG12C specific mutations rather than the common KRASG12D mutation. Thus, while KRASG12C inhibitors are a great proof of concept, their existence only increases the urgency to develop alternative KRAS-targeting moieties such as RNAi [45]. In addition to the prominent undruggable KRAS mutation, other aberrations in gene expression in PDAC also emphasize the need for RNAi mediated approach for therapy. Table 2 summarizes the reported siRNA therapeutics used in pancreatic cancer therapy by targeting various 'undruggable' genes. Most importantly, the incorporation of delivery agents such as nanoparticles and extracellular vesicles have facilitated the delivery of siRNA to cancer cells.

For miRNA, there are numerous evident data from in vitro models as well as from PDAC patient samples proving dysregulated levels of miRNA in pancreatic cancer [46]. From analyzing 95 miRNAs expression levels in pancreatic cancer tissues and cell lines, eight miRNAs—miR-196a, miR-190, miR-186, miR-221, miR-222, miR-200b, miR-15b, and miR-95—were significantly up-regulated in most pancreatic cancer samples. On the other hand, miR-148a, miR-217, miR-34a and miR-375 tumor suppressor miRNAs are frequently down-regulated in PDAC [47,48]. Over the past decade, extensive research efforts have been dedicated to investigating the underlying mechanism of chemoresistance. Along with dysregulation of various genes involved in key signaling pathways, such as phosphoinositide-3-kinase (PI3K), NF-κB, and hedgehog, miRNAs have also been demonstrated to modulate the expression associated with chemoresistance in PDAC [49]. For this reason, miRNA has gained attention as a useful biomarker for PDAC prognosis and as a target for therapeutic applications. Table 3 summarizes the miRNA-based approaches for pancreatic cancer therapy now in preclinical studies.

#### *2.2. Targets in the Pancreatic Tumor Stroma and Immunosuppressive Microenvironment*

The prominent stromal structure of PDAC consists of several cellular and acellular components, which together play a critical part in disease progression [83–85]. Although substantial research studies have focused on developing small-molecule or biological inhibitors directly targeting the cancer cells interest has more recently turned to agents targeting the tumor microenvironment. The key players of PDAC microenvironment are PSCs and cancer-associated fibroblasts responsible for the desmoplasia as well as various immune cells including immunosuppressive myeloid-derived suppressor cells, tumorassociated macrophages, and regulatory T cells [86]. The unique stromal barrier is often the cause of drug resistance while tumor induced alterations of the stroma have been suggested to facilitate tumor progression and proliferation. Thus, the concept of stomal reprogramming holds considerable promise [87].


**Table 1.** Various models of pancreatic cancer for preclinical studies.

Parameters are appreciated as abundant (+++), moderate (++), minimal (+) and none (-).

**Table 2.** siRNA-based therapeutics in preclinical pancreatic cancer models.



**Table 2.** *Cont.*

Abbreviations: PDMAEMA, Poly[2 (Dimethylamino)ethyl methacrylate]; POEGMA, Polyoligo(ethylene glycol) methyl ether methacrylate; PLGA, Poly lactic-co-glycolic acid; PSPG, G2 dendrimers through disulfide linkages; PEG, polyethylene glycol; RGD, arginylglycylaspartic acid; CXCR4, C-X-C chemokine receptor type 4; BCPV, Biodegradable charged polyester-based vector; PKN3, protein kinase N3; NGF, nerve growth factor; PLK1, polo-like kinase 1; VEGF, vascular endothelial growth factor; EPAS1, endothelial PAS domain protein 1; NCOA3, nuclear receptor coactivator 3; ARHGEF4, rho guanine nucleotide exchange factor 4; CCDC88A, coiled-coil domain containing 88a; LAMTOR2, late endosomal/lysosomal adaptor, MAPK, and mTOR activator 2; mTOR, mechanistic target of rapamycin; NUP85, nucleoporin 85; WASF2, Wiskott–Aldrich syndrome protein family member 2; PD-L1, programed death-ligand 1; s.c., subcutaneous; PDX, patient-derived xenograft; I.V., intravenous; I.P., intraperitoneal; I.T., intratumoral.

> There have been numerous siRNA targets in TME and stromal or immune cells to seek therapeutic effect for pancreatic cancer. One of the unique drivers of the dense stroma in PDAC are the activated PSCs. The PSCs are normally in a quiescent state functioning as a storage of vitamin A-rich lipid droplets [88]. During the initial development of PDAC, the PSCs are switched to an activated state losing the ability to produce lipid droplets and instead, producing an excessive amount of extracellular matrix components and proinflammatory chemokines [89]. The extracellular matrix production disrupts the normal parenchyma, leading to hypovascularity, and hypoxic conditions, which in turn further activates PSCs maintaining a hypoxia-fibrosis cycle [90]. Toll-like receptors are proteins that recognize foreign pathogen-associated molecular patterns and play a key role in the innate immune system. Recently, it has been shown that PSCs express a variety of toll-like receptors contributing to the activation of proinflammatory signaling pathways [91,92]. In particular, the toll-like receptor 4 is heavily involved in the stromal development in pancreatic cancer and has been advanced as a therapeutic target in PSCs [92]. Based on this, vitamin A-coupled liposomes carrying toll-like receptor 4-silencing small hairpin RNAs were used to target PSCs, reducing fibrosis by inducing mitochondrial apoptosis [93]. Ishiwatari et al. similarly used vitamin A-coupled liposomes (VA-lip) containing siRNA to induce collagen secretion by activated PSCs [94]. The siRNA targeting the collagen specific chaperone protein gp46 in VA-lip successfully inhibited the secretion of collagen and tissue collagenases, which dissolves the existing collagen matrix. Furthermore, Han et al. used a TME-responsive nanosystem to re-educate the PSCs to return to a quiescent state [95]. The nanosystem consists of polyethylene glycol-modified polyethyleneimine-coated gold nanoparticles to co-deliver all-*trans*-retinoic acid and siRNA targeting heat shock protein 47 (hsp47). The combination strategy successfully restored homeostatic stromal function and improved the efficacy of chemotherapy in stroma-rich tumor Another molecular target in the TME is the tumor-associated macrophages, of which depletion or reprogramming can induce immunogenic attack of the tumor cells [96]. Although various targets have been explored to reprogram or deplete tumor-associated macrophages, inhibition of the colony stimulating factor 1 (CSF1) or C-C Motif Chemokine Ligand 2 (CCL2) signaling have been particularly reported [97,98]. As one of the most abundant tumor-infiltrating leukocytes in all types of cancer, tumor-associated macrophages tend to stay in an alternative M2 state rather than the classically activated M1 state [99]. The M2 phase macrophage exhibit numerous tumor-promoting properties such as angiogenic signaling and restraining adaptive immune responses arising from the interactions of dendritic cells and CD8+ T cells. To deplete M2 macrophage, Qian et al. developed a dual-targeting nanoparticle (M2NPs), whose structure and function are controlled by two targeting moieties, an αpeptide coupled with an M2 macrophage binding protein for extremely specific targeting of M2 macrophage [100]. Upon loading of an anti-colony stimulating factor-1 receptor targeted siRNA, M2NPs showed higher affinity to M2-like tumor-associated macrophages

than to tissue-resident macrophages resulting in the elimination of M2 macrophage, thus decreasing tumor size in cancer bearing mice. In a later study, colony stimulating factor-1 receptor silencing siRNA was co-delivered with PI3K inhibitor, BEZ-235 using an M2 tumor-associated macrophage targeting nanomicelle to elicit therapeutic immune responses against pancreatic cancer cells both in vitro and in tumor models [101]. The combination treatment switched the macrophage state from M2 to M1, thus, reducing tumor infiltration of myeloid-derived suppressor cells. The reprogramming of M2 tumor-associated macrophage activated antitumor immune responses and enhanced anti-pancreatic tumor effects in a Pan-02 pancreatic xenograft model.


**Table 3.** miRNAs used for preclinical pancreatic cancer therapy.

Abbreviations: miR, microRNA; ASO, antisense oligonucleotide; MUC4, mucin 4; HER2, human epidermal growth factor receptor 2; SOD2, superoxide dismutase 2; CAT, catalase; DCK, deoxycytidine kinase; RPS18, ribosomal protein S18; Acta-2, actin alpha 2; PDGFR, platelet-derived growth factor receptor; mTOR, mechanistic target of rapamycin; CDK6, cell division protein kinase 6; IRAK3, interleukin 1 receptor associated kinase 3; NRP1, neuropilin 1; SMAD7, SMAD family member 7; SOCS6, suppressor of cytokine signaling 6; C5ORF41, chromosome 5 open reading frame 41; KLF12, Krueppel-like factor 12; MAPK10, mitogen-activated protein kinase 10; EFNA1, Ephrin-A1 precursor; PDCD4, programed cell death protein 4; PTEN, phosphatase, and tensin homolog; USP9X, ubiquitin specific peptidase 9 x-linked; SHH, sonic hedgehog; Bcl-2, B-cell lymphoma 2; SIRT1, sirtuin 1; RREB1, ras responsive element binding protein 1; s.c., subcutaneous; PDX, patient-derived xenograft.

#### **3. RNAi Delivery Strategies for Pancreatic Cancer Therapy**

#### *3.1. Nanocarrier-Mediated RNAi Therapy*

Over the last several decades of nanotechnology advancement, several applications and products containing nanomaterials have become available in the field of medicine [102]. Nanotechnology for medical purposes has been termed nanomedicine and is defined as the use of nanomaterials for diagnosis, monitoring, prevention, and treatment of diseases [103]. Nanomaterials can be applied in nanomedicine largely in three areas: diagnosis, drug delivery, and regenerative medicine. As discussed earlier, the direct injection of naked, unmodified siRNA was not found to be effective as siRNA will faces challenges of RNA degradation, short half-life, and circulation time, and poor biodistribution on systemic injection. Direct chemical modifications on the 2 carbon of the RNA ribose ring have been shown to prolong the half-life of siRNA without altering its functionality [104]. However, the modified siRNA has still shown discouraging results in clinical trials due to poor delivery to target tissues. For this reason, nanocarrier-mediated strategies have emerged as tool for therapy by acting as a protective delivery agent for siRNA. To provide the best opportunity to penetrate and accumulate within the cancer cells of the tumor, the size of delivery agent should range from 10 to 200 nm [105]. This size enables the nanoparticle to

passively localize in the tumor sites via 'enhanced permeability and retention effect' (EPR effect) which arises from the leaky tumor vascular structure [106] (Figure 2). Nanoparticles that are over 10 nm in size are too large to penetrate healthy vessels and tight-gap junctions [107]. Several passively targeted nanomedicines—including Doxil™, Abraxane™, Marqibo™, DaunoXome™, and Onivyde™—are currently in clinical use, suggesting that the EPR effect can drive therapeutic outcomes [108]. Although EPR effect is the underlying mechanism for nanomedicine, active targeting of nanoparticles with ligands and modalities are also being used as strategies to increase the target ability of RNAi therapeutics. Active targeting can significantly increase the quantity of drug delivered to the target cell compared to free drug or passively target nanosystems [109]. Ligands that target transferrin receptors or nicotinic acetylcholine receptors or even antibodies greatly enhance targeting of nanomedicine. In addition to nanoparticles, naturally occurring extracellular vesicles have also gained attention as a possible RNAi delivery agent also capable of biological communication with the tumor by harnessing cytokines and signaling proteins. Here, we will discuss some recent studies have demonstrated the potential of siRNA therapeutics using nanoparticles and extracellular vesicles for effective siRNA delivery and release into the tumor cell.

**Figure 2.** Drug delivery in vasculature of normal and tumor tissue.

#### 3.1.1. Lipid-Based Nanoparticle Delivery

Lipid-based nanoparticles are siRNA delivery agents consisting of pH-sensitive cationic lipid components. The lipids are amphiphilic and capable of self-assembly into micelles or liposomes in conjunction with siRNA via electronic charge interactions under acidic conditions [110]. Lipid-based nanoparticles such as liposomes are a suitable carrier for both small-molecule drug and nucleic acid delivery because of their excellent biocompatibility, biodegradability, low toxicity and immunity, structural flexibility, and ease of mass-production [111]. Decades after the discovery of RNAi, Patisiran (ONTATTRO™), which is a lipid-based nanoparticle formulated with siRNA targeting transthyretin, was FDA approved in 2018 as the first siRNA drug for the treatment of transthyretin-type amyloid polyneuropathy [112]. Because *KRAS* is so frequently mutated and active in cancer, especially in pancreatic cancer, *KRAS* is often the molecular target of siRNA therapeutics.

In 2019, Sasayama et al. generated a lipid nanoparticle formulation (LNPK15) to deliver KRAS-targeting siRNA to MIA PaCa-2 cells in vitro and in tumors in vivo achieving food protein knock down suggesting antitumor efficacy [50]. The group developed a PEGylated lipid-based nanoparticle named LNPK15 using two novel cationic lipids, SST-01 and SST-31, encapsulating KRAS siRNA that showed long circulation and efficient siRNA delivery to tumor sites. The *KRAS*-targeting LNPK15 showed strong antitumor efficacy in tumor xenograft mice of MIA PaCa-2 pancreatic cancer although the in vitro knock down activity of lipid-based nanoparticle was rather weak compared to its effect in tumors.

Although PEGylation with 3.3% 1,2-distearoyl-sn-glycero-3-phosphatidylethanolamine-N-(polyethylene glycol-2000) (PEG-DSPE) provides benefit in terms of long durations, it could be disadvantageous for intracellular delivery and endosomal escape of siRNA, which explains the lower activity in cell lines. The in vitro activity was rescued with lipase treatment for lipid nanoparticle hydrolysis for enhanced knock down of *KRAS*.

Rao et al. have demonstrated *KRAS* down-regulation by a lipoplex encoding a short hairpin RNA via systemic injection in a pancreatic cancer model [51]. The lipoplex contained a plasmid encoding a short hairpin RNA that executes selective silencing of the mutated KRAS gene with tumor cell entry facilitated by facilitated by fusogenic liposomes. The bifunctional short hairpin RNA design is designed for more rapid silencing, higher efficacy, and greater durability of siRNA action compared to conventional short hairpin RNA [113]. The lipoplex containing the *KRAS*-targeting bi-short hairpin RNA was systemically injected in female athymic Nu/Nu mice bearing PANC1 xenografts and showed successful delivery. In addition to the *KRAS* mutation as therapeutic target for RNAi, other genes such as protein kinase N3 have been investigated as targets for siRNA in pancreatic cancer. Kaufmann's group generated a liposomal formulation Atu027) for protein kinase N3 siRNA delivery from neutral fusogenic and polyethylene glycol-modified lipid components (for improved pharmacokinetic properties, cellular uptake, and efficient siRNA release in cells [52]. The siRNA-lipoplexes were capable of accumulating in tumor endothelial cells in vivo and halted tumor growth in two different xenograft mouse models [114]. As protein kinase N3 has previous been identified as a downstream effector of the PI3K signaling pathway, reduction of protein kinase N3 expression with Atu027 resulted in changes in tumor lymphatic vasculature, suggesting an anti-lymph angiogenesis therapeutic strategy for pancreatic cancer [115].

In addition to the demonstration of therapeutic potential of siRNA, miRNA has also been investigated in pancreatic cancer therapy. Pramanik et al. selected miR-34a and miR-143/145 for PDAC therapy based on their involvement in cancer stem cell survival and expression of KRAS and effectors. The miRNAs were formulated in lipid-based nanovector and intravenously injected in MiaPaCa-2 subcutaneous and orthotopic models where they effected tumor growth inhibition and increased apoptosis [82]. The success of miRNAbased therapeutics in preclinical studies has led some candidates to enter human clinical trials for cancer and other diseases [116–118]. The first ever miRNA-based clinical trial was with the miR-122 targeting called drug Miravirsen, which features the use of an LNA-based miRNA inhibitor, and has gone on to Phase II with for Hepatitis C virus infected liver cells [119]. Another miRNA mimicking drug MRX34, a lipid-based nanoparticle, was tested in clinical trials to treat advanced solid tumors [120]. However, the clinical trials involving miRNA-based drugs have largely been terminated because of severe adverse events arising from immune responses after treatments [121]. The two major discontinued miRNA-drug clinical trial results indicated that more than 30 genes in immunogenic and cytokine signaling pathways were identified to be affected by the miRNA machinery [122]. This results in the immune-related adverse events in phase I studies of miRNA-based drug clinical trials.

#### 3.1.2. Metal Nanoparticle USE for Delivery

For centuries, the noble metal gold has been studied first as a bulk material and now impacting nanomedicine [123]. In particular, gold nanoparticles (AuNPs) have shown important roles in the vaccine field as both an adjuvant and a carrier, by reducing toxicity, and increasing stability for storage [124]. Outside of vaccine applications, Lei et al. have shown that AuNPs are capable of siRNA delivery in vivo as nanocarriers for pancreatic tumor models [54]. As the nervous microenvironment has been recognized as a novel niche for PDAC progression and metastasis, nerve growth factor (NGF) and brain derived growth factor (BDGF) are rising targets for pancreatic cancer [125]. The gold nanoclusterassisted delivery of NGF targeting siRNA (GNC-siRNA) allowed efficient NGF gene silencing and pancreatic tumor treatment in subcutaneous, orthotopic, and patient-derived

xenograft models. To add the biomimetic property, the AuNP nanoflowers (Au@BSA) were coated with bovine serum albumin (BSA) and conjugated with survivin-targeting siRNA for efficient delivery and inhibition of gene expression in BxPC-3 pancreatic cancer cells [126]. Other than AuNPs, iron oxide nanoparticles have also gained attention as a stable agent for thermal and immunotherapies for cancer [127]. Mahajan et al. developed superparamagnetic iron oxide nanoparticles coupled with siRNA directed against pololike kinase-1 (PLK1), serving a dual purpose by both delivering siRNA to the tumor and facilitating non-invasive assessment of targeting efficiency via molecular imaging [56]. With uMUC1-specific ligand, EPPT1 and the myristoylated polyarginine peptide ligand and dextran, superparamagnetic iron oxide nanoparticle uptake was specific to pancreatic cancer and induced a drastic decrease in tumor volume in a syngeneic orthotopic model driven by PLK1 silencing. In addition to published reports of siRNA delivery using AuNPs, other inorganic nanoparticles such as graphene oxide and carbon nanotubes use were demonstrated in pancreatic cancer cells and models. Feng et al. developed a multifunctionalized monolayer graphene oxide as delivery agent for HDAC1 and KRAS siRNA in MIA PaCa-2 pancreatic cancer cells [128]. The dual gene silencing induced apoptosis, proliferation inhibition and cell cycle arrest and when treated in combination with near infrared light thermotherapy showed tumor volume growth inhibition for pancreatic in vivo models. Furthermore, Anderson et al. formulated a functionalized single walled carbon nanotubes as a siRNA carrier for in vitro gene therapy [129]. With KRAS siRNA in the carbon nanotube, PANC-1 cells showed great knockdown of KRAS gene expression.

#### 3.1.3. Polymer-Based Nanoparticle Delivery

McCarroll et al. has developed a star-shaped polymeric nanoparticle for delivery of siRNA into pancreatic cancer cells in vitro and in vivo. The star polymers are cost-effective nanoparticles that self-assemble with siRNA to produce a small (<50 nm) delivery system [58]. These star polymers, containing different lengths of cationic poly(dimethylaminoethyl methacrylate) side-arms and varied amounts of poly[oligo(ethylene glycol) methyl ether methacrylate], delivered siRNA targeting *β*III-tubulin to orthotopic pancreatic tumors [57]. Using this approach *β*III-tubulin expression was silenced by 80% and tumor growth progression was delayed when star polymers were administered in orthotopic models. Often, polyethylene glycol coating offers nanoparticles increased compatibility with aqueous environments. prolonging circulation time and reducing immunogenicity.

However, there has been biosafety issues from PEG recognizing antibodies contributing to immune responses and mitigation of therapeutic efficacy of PEGylated drugs. Despite these concerns, these nanoparticles have shown improved efficacy with polyethylene glycol fabrication when systemically administered [130]. Furthermore, Kataoka's group have developed a polyethylene glycol coated calcium phosphate hybrid micelle system for siRNA delivery in pancreatic cancer [60,61]. First, the PEG-*b*-charge-conversional polymer/calcium phosphate hybrid micelles were systemically administered to EL1-Luc/TAg transgenic mice carrying luciferase targeting siRNA to ensure RNAi activity. As vascular endothelial growth factor (VEGF) is a key pro-angiogenic molecule in tumor progression, VEGF specific siRNA was then therapeutically administered with the help of the hybrid micelle. After systemic administration of purified nanoparticles, the reduction of expression level VEGF mRNA was evaluated in tumor tissues and found to be correlated with the pancreatic tumor growth rates [60]. Because of major issues concerning toxicity, safety, and clearance of inorganic nanoparticles, biodegradable and biocompatible organic materials have been sought out as a nanocarrier option for various drugs [131]. Poly (lactic-co-glycolic acid) (PLGA) is one of the most investigated biodegradable polymeric nanoparticles, and already has won FDA approval for medical applications [132]. Zorde Khvalevsky et al. synthesized a local prolonged siRNA delivery system, siG12D-LODER™, with PLGA for protection and gradual release of drug over time [62]. An siRNA against the mutated KRASG12D was encapsulated in LODER and implanted in a pancreatic xenograft and an orthotopic model, leading to decreased KRAS levels, as well as inhibition

of cell proliferation and epithelial-mesenchymal transition over 155-day period. The group later injected the siRNA encapsulated LODER into tumors of locally advanced pancreatic cancer patients in combination with chemotherapy in a clinical trial study. Pan et al. also designed a PLGA/poloxamer nanoparticle loaded with EPAS1 siRNA and tested it on BxPC-3 pancreatic cancer cells in a mouse model of pancreatic cancer [133]. The endothelial PAS domain protein, also known as hypoxia-inducible factor-2, is a transcription factor known to regulate tumor cell adaption to hypoxic microenvironments and to accelerate metastatic spread [63]. The study demonstrated that the PLGA/poloxamer nanoparticle successfully inhibited VEGF and CD34 expression, therefore halting the growth of pancreatic tumors.

Oupicky's group developed a different type of polyplex for nucleic acid delivery using cholesterol-modified polymeric CXCR4 antagonist nanoparticles to improve pancreatic cancer therapy [64,65]. The nuclear receptor coactivator-3 (NCOA3) is a critical modulator of musin expression that is critical in pancreatic cancer progression. The siRNA targeting NCOA3 was formulated along with a cholesterol-modified polymeric CXCR4 antagonist in a dual-function polyplex for delivery to tumors to inhibit CXCR4 chemokine receptor to treat metastatic pancreatic cancer [64]. Additionally, triple targeted nanoparticles featuring the CXCR4 antagonist for cancer stromal blockade action, as well as anti-miR-210 and siRNA targeting mutated *KRAS* were developed. The triple action-nanoparticle modulated the desmoplastic TME via inactivating PSCs and promoting infiltration of cytotoxic T cells [65].

Another organic material used for siRNA delivery is a biodegradable charged polyesterbased vector which has demonstrated no significant toxicities on blood vessels, tissues, or organs even at high dosage [134]. When *KRAS*-targeting siRNA is formulated in this delivery vector and peritumorally injected in pancreatic xenograft models, significant decreases in tumor growth and local invasion were observed [66]. Many polysaccharidederived nanoparticles such as chitosan nanoparticles are being investigated for nucleic acid drug delivery across diseases [135]. In particular, chitosan oligosaccharide lactate can be assembled with folic acid and polyethylene glycol into nanoparticles that are capable of encapsulating siRNA and intravenously localizing in tumors in pancreatic orthotopic models [67]. This formulated nanoparticle containing siRNA silencing genes *LAMTOR2, mTOR,* and *NUP85* have shown clinical potential by inhibiting retroperitoneal invasion and dissemination, therefore improving the prognosis of the mice. Overall, these studies indicate promising strategies of nanoparticle usage for siRNA therapeutics for pancreatic cancer therapy.

Other than the lipid and polymeric nanoparticles, peptides have also been investigated as siRNA delivery materials due to their biocompatibility and biodegradability for minimal toxicity [136]. In addition, peptides also have intrinsic property to act as ligands for targeting specific receptors. An arginine-glycine-aspartic acid-peptide is well-known to specifically bind to the integrin receptors overexpressed on the surface of cancerous cells [137]. Bhatia's group formulated arginine-glycine-aspartic acid peptides with polyethylene glycol to encapsulate siRNA to generate a tumor-penetrating nanocomplex [68]. When a KRAS siRNA was intravenously delivered by this tumor-penetrating nanocomplex into a genetically engineered PDAC mouse model, the growth of pancreatic tumors arising from allografted KP D8-175 cells were significantly delayed. Similarly, pancreatic tumor inhibition was seen when miR-34a was delivered in nanocomplexes formulated with a tumor-targeting and penetrating bifunctional CC9 peptide [81]. The nanoparticle facilitated the cellular uptake of miR-34a and induced cell cycle arrest and blocked angiogenesis in xenograft model using PANC-1 cells. Another type of tumor suppressor miRNA, miR-150 was administered to pancreatic cancer cells in a PLGA-based nanoformulation incorporating polyethyleneimine. The cells showed a significant decrease in MUC4 expression and downstream signaling proteins such as human epidermal growth factor receptor 2 (hEGFR2), therefore reducing tumor growth, clonogenicity, motility, and invasion [74]. Due to these and other successes, the application of various nanoparticles is still being investigated in preclinical and clinical trials for pancreatic cancer therapy.

#### 3.1.4. Extracellular Vesicle-Mediated Delivery

Exosomes or extracellular vesicles are roughly 100 nm spherical vesicles, which are secreted by all cells. Initially, exosomes were considered to be garbage bags for cells [138]. However, recent findings demonstrating that exosomes contain a variety of bioactive molecules reflecting parental cells and play a crucial role in cell-cell signaling, have drawn much attention to exosome research [139]. In addition, exosomes exhibit many advantages as delivery systems including low immunogenicity, multi-functionality, and disease-targetability [140]. Notably, exosomes can be internalized by cells via a fusion mechanism, leading to efficient delivery of drugs into the target cell cytoplasm [141]. Some reports have indicated that exosomes expressing viral fusogen fused significantly with the target cell membrane in acidic conditions compared with native exosomes [142]. Notably, Liu et al. showed that viral fusogen decorated and programed death-ligand 1 siRNA-loaded, macrophage-derived extracellular vesicles substantially elevated tumorspecific CD8<sup>+</sup> T cell immune responses in CT26 tumor-bearing mice through efficient programed death-ligand 1 gene silencing [143]. Although packing drug candidates into extracellular vesicles is still challenging, these findings indicate that the exosome has great potential as a drug delivery platform due to strengths mentioned above [144].

Several studies suggested that extracellular vesicles can effectively transfer RNAi into the cancer cells, eliciting substantial antitumor effects. For example, extracellular vesicles containing miRNA-134 enhanced the therapeutic effects of anti-heat shock protein 90 drugs in breast cancer [145]. Similarly, glucose-regulated protein 78 (GRP78) siRNA-loaded extracellular vesicles could sensitize resistant tumor cells to sorafenib [146]. Another study found that extracellular vesicle delivery of let-7a miRNA had potent antitumor effects on breast cancer cells breast cancer cells, and that extracellular vesicles expressing GE11 or AS1411, to target epidermal growth factor or nucleolin, respectively, on breast cancer cells, could enhance the effects of let-7a miRNA by enhancing delivery of their payload to the tumor cells [117,147]. Furthermore, taking advantage of fact that extracellular vesicles can penetrate the blood-brain barrier via transcytosis, systemic administration of extracellular vesicles harboring miRNA-21, miRNA-164b, miRNA-124a, or vesicular endothelial growth factor siRNA significantly retarded brain tumor growth, improving survival [148–151].

There have also been attempts to treat PDAC using exosomal siRNA. Kalluri et al. reported that mesenchymal stem cell-derived exosomes loaded with KRASG12D siRNA (termed iExosomes) provoked effective antitumor responses in pancreatic cancer models [72]. They also found that iExosomes showed a long half-life due to high expression of exosomal CD47, facilitating evasion of immunological clearance. The production protocol of iExosomes has been successfully established to generate clinical grade-exosomes that have entered phase 1 clinical trials for the treatment of PDAC [ClinicalTrials.gov identifier: NCT03608631] [152]. Zhou et al. recently developed engineered exosomes loaded with galectin-9 siRNA to alleviate immunosuppressive properties of tumor-associated macrophages [73]. Co-delivery of galectin-9 siRNA and the immunogenic chemotherapy oxaliplatin by exosomes elicited successful tumor control via immune activation. Collectively, these studies indicate that exosome-mediated gene silencing could be attractive for the treatment of PDAC.

#### *3.2. Combination Therapy*

Many studies have shown that the combination of siRNA and other therapeutic agents can exhibit synergistic effects to overcome multiple drug resistance by enhancing both silencing and chemotherapy activity [153,154]. Continuous administration of inhibitors often leads to the development of resistance via acquired mutations or epigenetic alterations. In combination therapy, the approach is designed to reverse the acquired genetic or epigenetic mechanism with the RNAi drugs instead of additional chemical inhibitors which result in increased toxicity. There have been numerous reports on therapeutic efficacy when existing therapeutic agents are combined with the concomitant use of RNAi drugs [155]. There is also evidence indicating that a combination strategy can also reduce the side effects of individual therapeutics by enhancing the therapeutic index [156].

As one of the most common cancer types with very low survival rate, pancreatic cancer is uniquely identified by its dense surrounding tumor microenvironment that limits the therapeutic effect of existing drugs. Both the dense stroma and acquired resistance to conventional therapies are triggered from dysregulated expression of cancer-associated genes. To target these key genes, RNAi-based approaches have gained attention both to elicit direct antitumor effects and to alter the nature of tumor microenvironment to remove barriers for chemotherapeutic action. Thus, several studies have sought to achieve antitumor therapeutic effects using KRAS siRNA in combination with chemotherapies and other small-molecule drugs. Zeng et al. reported that the combination of KRAS siRNA and arsenic trioxide effectively inhibited pancreatic tumor growth by exerting a synergistic effect of siRNA-induced cell cycle blockage and arsenic-induced apoptosis [69]. Wang et al. showed the combination of KRAS siRNA and gemcitabine exhibited improved inhibition of cell proliferation, cell cycle arrest, increased apoptosis, and suppression of tumor progression without any toxicity. In addition, they demonstrated that the anticancer effect of KRAS siRNA contributes to the low IC50 value of siRNA/gemcitabine-loaded liposomes [53]. Golan et al. tested that a biodegradable implant termed siG12D-LODER™ loaded with siRNA drug against KRASG12D and gemcitabine for pancreatic cancer. Of the 15 patients, 12 patients did not show tumor progression, 10 out of 12 patients showed stable disease, and 2 patients had partial response, yielding an increased average patient survival rate [157]. In addition, several attempts have been made to knock down several other genes highly associated with pancreatic cancer. Li et al. used a peptide-binding nanovehicle for co-delivery of paclitaxel and siRNA targeting TR3, an orphan nuclear receptor, finding that the co-delivery system exhibited a higher loading efficiency of siRNA, markedly inhibited tumor growth, and induced cancer cell apoptosis [70]. McCarroll et al. also identified βIII-tubulin as a novel mediator for chemoresistance and metastases in pancreatic cancer and used siRNA to increase chemosensitivity [158]. When βIII-tubulin expression is inhibited in orthotopic pancreatic cancer model in the presence of gemcitabine, tumor growth, and metastases were reduced in vivo. Additionally, siRNAs have been used to alter the immunosuppressive microenvironment and to reduce the stromal physical barrier to improve antibody or chemotherapeutic drug actions [159]. Yoo et al. combined an siRNA targeting the programed death-ligand 1 in a magnetic nanocarrier with gemcitabine to activate immune responses against tumor cells while noninvasively monitoring the therapeutic response via magnetic resonance imaging [71].

Currently, miRNA-based drugs are also being investigated in combination with chemotherapeutic drugs to enhance therapeutic efficacy in pancreatic cancer [160]. Although the chemotherapeutic gemcitabine remains a cornerstone of the PDAC treatment, patients acquire resistance within weeks of treatment [161]. To overcome the resistance, several novel therapeutic approaches as well as new entrapment designs with nanoparticles are currently being investigated. Patel et al. encapsulated miR-155 in extracellular vesicles from gemcitabine-resistant pancreatic cancer cells to reduce deoxycytidine kinase (gemcitabine-metabolizing gene) and increase superoxide dismutase 2 and catalase (ROSdetoxifying genes) expression in the presence of chemotherapy [75]. The deoxycytidine kinase reduction by miR-155 delivered by the extracellular vesicles led to abrogation of acquired resistance of gemcitabine in pancreatic cancer cells. Another miRNA that was combined with gemcitabine for pancreatic cancer treatment is miR-345 which was delivered in polymeric dual delivery nanoscale device [80]. The temperature and pH-responsive pentablock copolymer system enhanced/restored the miR-345 levels, making xenograft pancreatic cancer tumors more susceptible to gemcitabine. Immunohistochemical analysis revealed that the miR-345 induced significant down-regulation of desmoplastic reaction, as well as improving SHH, Gli-1, MUC4, and Ki67 levels. Because oncogenic miRNAs are responsible for triggering the cancerous features, antisense oligonucleotides are used to inhibit the activity of commonly dysregulated miRNAs [162]. Among numerous oncogenic

miRNAs, miR-21 is one of the earliest identified and most abundant cancer-promoting miR-NAs and hence has been targeted in cancer [163]. Li et al. developed a combined therapy using an miR-21 antisense oligonucleotide and gemcitabine co-delivered by a nanoparticle made from polyethylene glycol-polyethyleneimine-magnetic iron oxide [78]. With a coating of anti-CD44v6 single-chain variable fragment for targeted delivery, the miR-21 antisense oligonucleotide and gemcitabine dual-treatment nanoparticle up-regulated programed cell death protein 4 and phosphatase and tensin homolog (PTEN) expression, thus suppressing epithelial-mesenchymal transition, cell proliferation, migration, and invasion of pancreatic cancer cells. This combination therapy also showed a synergistic antitumor effect in a MIA PaCa-2 pancreatic xenograft model.

Other than gemcitabine, doxorubicin, and the hedgehog inhibitor, GDC-0944 have been combined with miRNA for pancreatic cancer therapy. Chen et al. developed a new chimeric peptide composed of plectin-1 for PDAC-specific targeting coupled with argininerich RNA-binding motifs for miR-212 delivery [79]. The reduction of USP9X expression by miR-212 enhanced the therapeutic effect of doxorubicin to induce apoptosis and autophagy of PDAC cells in vitro as well as in patient-derived pancreatic xenograft model. Kumar et al. selected miR-let7b as a therapeutic target because of inherently low expression miR-let7b in PDAC tumors with elevated hedgehog levels [164]. Using mPEG-block-poly(2-methyl-2 carboxyl-propylenecarbonate-graft-dodecanol-graft-tetraethylene-pentamine copolymer, miR-let7b and the hedgehog inhibitor, GDC-0944 were co-formulated into micelles, which inhibited tumor growth of an ectopic xenograft model by triggering apoptosis. Instead of combination with chemotherapy, miR-21 and miR-221 antisense oligonucleotides were used to target stem-cell-like cancer cells [77]. The miR-21 and miR-221 antisense oligonucleotides were given to sorted side population cells from gemcitabine-resistant cells and inhibited the growth and metastasis when both miRNAs were given. Schnittert et al. demonstrated a novel peptide-based nanocomplex for miR-199a antisense oligonucleotide delivery into human-derived PSCs [76]. The miR-199a antisense oligonucleotide inhibited PSC differentiation into cancer-associated fibroblasts and reduced the size of 3D heterospheroids made of PSCs and cancer cells. Taken together, these studies have shown that the combination of siRNA and chemotherapy not only has excellent anticancer effects, but also reduces toxicity and inhibits resistance.

#### **4. Challenges and Future Prospects**

Even with numerous efforts and ongoing investigations of RNAi-based therapeutics, pancreatic cancer remains as one of the most challenging with devastating malignancies with an extremely poor prognosis [165]. The successful delivery of degradation-prone nucleotide-based drugs is still the challenging goal in the field of RNAi therapeutics, requiring alternative strategies and reagents. One of the most promising mechanisms to consider in addressing the hurdles of intracellular RNAi delivery to gain cellular entry is macropinocytosis. The significance of macropinocytosis as a delivery pathway is that it is broadly applicable across multiple cell types, cargos, and delivery systems. Exploitation of macropinocytosis for therapeutic delivery to cancer cells is uniquely advantageous because macropinocytosis plays a key role in sustaining cancer cell growth within the nutrient-poor by functioning as a nutrient supply route [166–170]. Frequently observed oncogenic mutations, such as *KRAS, PI3K*, and *PTEN*, have been demonstrated to greatly enhance macropinocytosis in cancer cells [171–174]. Thus, it is not surprising to find that numerous delivery systems, such as lipids, polymers, and peptides, use macropinocytosis to enter a broad range of cancer cells [72,175–180]. Another important aspect of employing macropinocytosis in RNAi delivery is that macropinocytic entry delivers cargo to late endosomes and multivesicular bodies, which are particularly favorable destinations for nucleic acids [181–183]. Whether the siRNAs are delivered by lipids or nanoparticles, the intracellular trafficking of siRNAs begins in early endosomal vesicles [184]. The early endosomes subsequently fuse with sorting endosomes, which in turn transfer the content to the late endosomes. The acidic late endosomes later then relocated to the lysosomes where

abundant nucleases exist. Thus, late endosomes and multivesicular bodies are where most endosomal escape is reported to occur before lysosomal degradation, a major barrier to the intracellular delivery of RNAi therapeutics [185]. These reports are especially interesting considering that cancer cells use macropinocytosis for nutrient scavenging, which requires lysosomal degradation of proteins into amino acids. The studies to elucidate the exact mechanism of macropinocytic delivery have encountered numerous difficulties due to the lack of specific inhibitors, varying rates among different cell types, and high sensitivity to external stimuli, and others [186,187]. However, our rapidly growing understanding of macropinocytosis presents unique opportunities to design better strategies to target macropinocytosis for intracellular delivery of RNAi.

One major hurdle specific to pancreatic cancer is the stromal barrier that hinders the therapeutic action of current and RNAi-based drugs within tumor cells or TME. Repeated reports and studies demonstrate that the tumor's genetic aberrations cause the development of dense stromal environment and suppressive immune response, emphasizing the need for RNAi for TME reprogramming in addition to direct attack on tumor cells [188]. Targeting the infiltrated immune cells present in stromal structure is an alternative approach to cancer therapy using immunotherapeutic options. The use of nanoparticles has led to successful delivery of RNAi drugs into tumor cells; however, the uptake rate remains low for clinical significance. Similar to the immunotherapeutic drug agents, RNAi has the potential to take action on immune cells to re-sensitize the suppressive immune response as an alternative cancer therapeutic strategy [189]. Genetic mutations of signaling proteins and transcription factors triggering the immunosuppression are potential RNAi drug targets with benefits over chemical inhibitors and conventional therapies. Furthermore, the application of RNAi drugs with current checkpoint inhibitors has great potential in clinical translation to patients who developed resistance to immunotherapy.

Although milestones of RNAi-based drugs have proved clinical translation of oligo therapeutics and has opened numerous application possibilities, the systemic delivery challenge remains for specific disease targets. When designing a delivery strategy for oligonucleotide, the cargo and moiety of drug often jeopardizes the safety profile and cause adverse toxicities upon accumulation. Though RNAi-based drugs are naturally occurring biological materials, the chemical modifications and fabrication is inevitable and accompanies safety concerns. For this reason, strategies to use siRNA conjugation with more transparent and simpler structures and designs is under investigation across academia and industry for its advantages in clinical translation.

#### **5. Conclusions**

The potential to regulate selective gene expression using RNAi-based moieties holds promise for pancreatic cancer treatment. Numerous results demonstrating antitumor effects, chemosensitization, anti-metastatic activity and activation of immunogenicity against cancer cells have been shown in preclinical and even in early clinical stages. Pancreatic cancer, one of the cancer types with the lowest survival rate, is uniquely identified with dense tumor microenvironment serving as a barrier to conventional drugs. Dysregulated expression of cancer-associated genes and key signaling pathways are often the driver for acquired resistance and the discouraging outcomes of current therapies. For this reason, the RNAi-based approaches have gained attention to elicit antitumor effect or altering the nature of TME to remove barriers for chemotherapeutic action. From the studies mentioned above, it appears that RNAi-based drugs hold great hope in pancreatic cancer therapy [190]. In addition to the RNAi-based drug discovery and development, it is crucial to develop a model or system that accurately reflects the biology of human pancreatic cancer with clinically similar features. Orthotopic pancreatic cancer mouse models with an intact and functional fibrotic stroma, and genetically engineered mouse models with functional immune system and 3D tissue models have been developed and are continually being improved to advance pancreatic cancer therapy [191]. In particular, the use of 3D cultures is an innovative approach that narrows the gap between traditional 2D cell culture and animal

models allowing the manipulation of individual cell population involved in TME specific to the disease type [192]. Several groups have demonstrated spheroids, hydrogel structures, and microfluidic designs to mimic the TME nature of pancreatic cancer [193–197]. These models, used in parallel with clinically relevant in vivo models, can inform the design of novel RNAi-based drugs. Since personalized medicine and targeted therapies are the next generation of cancer treatments, the unique pathophysiology of pancreatic cancer needs to be addressed. In particular, the dense fibrotic stroma presenting a large barrier is a strong consideration in introducing a novel RNAi-based approach for pancreatic cancer therapy. Despite the heterogeneity and complexity of pancreatic cancer, we believe that versatility of RNAi-based drugs with numerous targets hold promising potential for pancreatic cancer therapy in combined with chemotherapy or immunotherapy.

**Author Contributions:** Conceptualization, M.J.K., T.M.R. and J.H.R.; writing—original draft preparation, M.J.K., H.C., G.N., Y.K., writing—review and editing, G.N., T.M.R. and J.H.R.; supervision, T.M.R., S.H.K. and J.H.R. All authors discussed, edited, and approved the final version of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Korea (grant number: HI19C0753) and the Intramural Research Program of KIST.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

