Galactosylated hydroxyl‐polyamidoamine dendrimer targets hepatocytes and improves therapeutic outcomes in a severe model of acetaminophen poisoning‐induced liver failure

Abstract Toxicity to hepatocytes caused by various insults including drugs is a common cause of chronic liver failure requiring transplantation. Targeting therapeutics specifically to hepatocytes is often a challenge since they are relatively nonendocytosing unlike the highly phagocytic Kupffer cells in the liver. Approaches that enable targeted intracellular delivery of therapeutics to hepatocytes have significant promise in addressing liver disorders. We synthesized a galactose‐conjugated hydroxyl polyamidoamine dendrimer (D4‐Gal) that targets hepatocytes efficiently through the asialoglycoprotein receptors in healthy mice and in a mouse model of acetaminophen (APAP)‐induced liver failure. D4‐Gal localized specifically in hepatocytes and showed significantly better targeting when compared with the non‐Gal functionalized hydroxyl dendrimer. The therapeutic potential of D4‐Gal conjugated to N‐acetyl cysteine (NAC) was tested in a mouse model of APAP‐induced liver failure. A single intravenous dose of a conjugate of D4‐Gal and NAC (Gal‐d‐NAC) improved survival in APAP mice, decreased cellular oxidative injury and areas of necrosis in the liver, even when administered at the delayed time point of 8 h after APAP exposure. Overdose of APAP is the most common cause of acute hepatic injury and liver transplant need in the United States, and is treated with large doses of NAC administered rapidly within 8 h of overdose leading to systemic side effects and poor tolerance. NAC is not effective when treatment is delayed. Our results suggest that D4‐Gal is effective in targeting and delivering therapies to hepatocytes and Gal‐D‐NAC has the potential to salvage and treat liver injury with a broader therapeutic window.


| INTRODUCTION
Acute liver failure occurs due to rapid and catastrophic injury to hepatocytes with loss of function and is accompanied by coagulopathy and encephalopathy rapidly progressing to death. Viral hepatitis and drug toxicities including acetaminophen (APAP) overdose are the most common causes of acute liver failure. 1,2 APAP overdose is responsible for almost 50% of acute liver failure cases 1,2 and 7% of liver transplants, 3 representing an enormous burden on the healthcare system. APAP is an over-the-counter analgesic, but, because it is easy to access and incorporated into many medications, has been a significant source of both intentional 4 and unintentional 5 overdose at almost equivalent rates. 6 This commonly used drug becomes toxic instead of salutatory when taken in large quantities as the tertiary pathway of APAP metabolism in the liver depletes endogenous glutathione levels, allowing for the accumulation of N-acetyl-p-benzoquinone imine (NAPQI) in hepatocytes, which creates protein adducts and causes mitochondrial fission and eventual hepatocyte apoptosis. 7 Fortunately, through treatment with intravenous or oral N-acetyl cysteine (NAC) soon after overdose, liver failure can be prevented or reversed in most acute cases. 8 However, even with liver transplants and NAC therapy, the mortality rate of patients with APAP-induced acute liver failure is currently at 30%. 9 The benefits of NAC therapy when a patient arrives in the emergency room more than 14 h after the overdose are questionable and unproven. 10 These patients who are not candidates for NAC therapy due to severe overdose or late arrival at the hospital represent a significant underserved population. Intentional overdoses are high doses 11 and the symptoms of APAP-induced acute liver failure do not manifest until hours after ingestion, 2 making timely transport to the hospital unlikely in cases of unintentional overdose. Therapies that specifically target hepatocytes may help rescue liver cell function in severe cases.
A promising branch of novel hepatic therapeutics is nanoparticlemediated drug delivery. [12][13][14][15][16][17] However, very few nanoparticles actually are taken up into liver cells in a meaningful way as they are rapidly cleared from circulation by Kupffer cells (liver macrophages) through the mononuclear phagocyte system and hepatocytes through hepatobiliary clearance as part of the reticuloendothelial system. 18 The function of the reticuloendothelial system is critical for the removal of toxins and foreign bodies from the bloodstream but proves a major hindrance to the delivery of drugs to hepatocytes. Targeted nanomedicine is one way for overcoming hepatobiliary clearance by incorporating ligands to the nanoparticles' surfaces that actively bind with receptors expressed on hepatocytes, such as hepatic transferrin, G-coupled protein, and asialoglycoprotein receptors (ASGPR). [19][20][21][22] ASGPR is unique in that it is only expressed on hepatocytes and has a high binding affinity for galactose, glucose, and lactose sugars; however, there are limits to the size and dosage of particle uptake by hepatocytes via ASGPR. 23 Sugars are of particular import to dendrimer nanoparticles because the multifunctional nature of the dendrimer allows it to participate in both multivalent binding 24,25 and the glycoside cluster effect, which describes the ability of multivalent carbohydrates to bind with greater strength to their target receptors than would be predicted even after corrections for molar concentration and valence. 26 Researchers have previously shown that a linear increase in the number of monovalent ligands on the dendrimer surface results in a nonlinear increase of receptor-specific binding in the case of carbohydrate-coated dendrimers. 27,28 However, some of these studies were performed on dendrimers that may be considered toxic when delivered systemically. 29 Hydroxyl-terminated polyamidoamine (PAMAM) dendrimers are a class of nanoparticles with very few reported systemic toxicities, [30][31][32] which are further minimized above generation 3 when the end groups provide ample shielding of the ethylenediamine core. Generation 4 hydroxyl-terminated PAMAM dendrimers (D4-OH) specifically have been shown to perfuse freely through tissue, clear intact through the kidneys, and localize only in actively phagocytic macrophages and microglia in many animal models. [33][34][35] Additionally, D4-OH is far smaller than the fenestrations in the sinusoidal epithelium, allowing for free access to liver hepatocytes. 36 The diffusive and biocompatible qualities of these dendrimers make them good candidates for developing targeted drug delivery strategies, and it has already been shown that the introduction of targeting ligands on these dendrimers can modify which organs, cells, 37 and even subcellular organelles 38,39 the dendrimer localizes within upon systemic administration without impeding the dendrimers' inherent transport processes.
In this study, we have developed a generation 4 hydroxylterminated PAMAM dendrimer modified to display 12 galactose molecules on the surface. We show that the surface galactose sugars create a multivalent binding effect to ASGPR, allowing the dendrimer to selectively target and internalize in hepatocytes in vitro and in vivo.
This hepatocyte-targeting dendrimer was then conjugated to NAC and efficacy was assessed in a mouse model of severe APAP poisoning with a high rate of mortality. We believe that the addition of nontoxic sugar to the dendrimer will not change this result in hepatocytes, but the effects of long-term dendrimer accumulation in hepatocytes will require further study prior to translation.
To quantify the binding affinity of D4-Gal for ASGPR, we utilized a cell surface binding assay 40,43 on HEPG2 cells, which express on the order of 100,000 ASGPR per cell. 44 This method was chosen as it has previously been reported that binding affinity to ASGPR can change over 100-fold between free receptor systems, like surface plasmon resonance, and cell-based assays. 45 Based on the total binding isotherm and the correlated nonspecific binding quantification through the introduction of ASGPR blocking peptide, we determined the binding affinity, K D , of D4-Gal to ASGPR to be 3.66 μM ( Figure 3b, r 2 = 0.9192 total, r 2 = 0.8297 nonspecific, 95% CI: 2.025 < K D <6.562 μM, n = 3) as opposed to D4-OH, which indicated no preferential binding to ASGPR. K D of free galactose is estimated to be between 2 and 5 mM, 46 making the binding of D4-Gal to ASGPR about 1000-fold stronger than free galactose sugar. We attribute the observed improvement in binding affinity to the impact of both multivalency and the cluster glycoside effect. 24,25 Introducing the capacity to bind to ASGPR to the dendrimer was encouraging, so we next investigated the impact of galactosylation on dendrimer uptake into HEPG2 cells in culture.

| Pharmacokinetics of intravenous Gal-D-Cy5 in vivo in healthy mice
Based on these promising in vitro results, we investigated the in vivo hepatocellular targeting of D4-Gal conjugate and compared it to D4-OH, which is known to minimally distribute to the liver and is cleared rapidly through the kidneys due to its size 49  Gal-D-Cy5 clears from nonhepatic tissue at rates comparable to that of unmodified D-Cy5 with no tissue other than kidneys demonstrating uptake of greater than 0.1% of the initial injected dose 48 h after administration. There is statistically significantly greater uptake of Gal-D-Cy5 over D-Cy5 in the healthy brain, heart, lungs, and kidneys, which may be due to sugar receptors in these organs, but the difference was not enough to observe differences in tissue imaging (n = 6 for all treatments and times). (h) Gal-D-Cy5 also clears more rapidly from serum, potentially due to its high accumulation in liver tissue (n = 6 for all treatments and times) The high specificity of Gal-D-Cy5 to liver hepatocytes, with relatively rapid clearance from the rest of the body makes it a desirable candidate for drug delivery to hepatocytes. We further wanted to

| Systemic Gal-NAC conjugate (Gal-D-NAC) treatment improved long-term survival in a mouse model of severe APAP poisoning
To evaluate the in vivo efficacy of our dendrimer therapy, we conjugated NAC to the surface of D4-Gal (Gal-D-NAC) through a glutathione-sensitive disulfide linker that would release once internalized in the hepatocytes but remain stable in solution and in plasma. 58 Systemic NAC therapy is already the standard-of-care for the clinical presentation of APAP poisoning but becomes ineffective at later time points or with increasingly large doses of APAP as is shown by the treatment nomogram. 10 We hypothesized that if NAC could be delivered more rapidly and directly to the hepatocytes that need it, the treatment window for severe APAP poisoning could increase, reducing mortality and the need for liver transplants. We also evaluated serum levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST), enzymes in hepatocytes that catalyze the reaction from either alanine or aspartate to glutamate or pyruvate, and ALT is highly liver-specific while AST is expressed throughout the body. 62 These enzymes are released into the bloodstream soon after APAP overdose due to the breakdown of hepatocytes, typically peaking soon after injury as hepatocytes are actively dying and come down over time. 63 Taken together these data indicate that D4-Gal is a powerful new tool for accessing and treating hepatocytes in a targeted manner. D4-Gal specifically targets hepatocytes very effectively and is retained in the cells, which may enable it to release the drug in a sustained manner. While specific mechanisms of action will need to be studied in the future, Gal-D-NAC clearly has a significant impact on mortality in severe APAP poisoning along with improvement in liver injury seen on histology. This is an early proof-of-concept study to demonstrate that Gal-D-NAC can be reliably synthesized and that it specifically targets hepatocytes delivering the drug at the site of action and producing efficacy in a model of APAPinduced liver injury. Since similar localization is seen irrespective of the mechanism of injury, D4-Gal can potentially be used to deliver other drugs or biologics to injured hepatocytes in other liver disorders.

Synthesis of Gal-D-NAC
The detailed synthetic procedures for Gal-D4-NAC are presented below:
F I G U R E 7 Hematoxylin and eosin staining of liver sections at 48 and 96 h after APAP exposure. (a) Sections from healthy control, APAP exposed untreated and APAP exposed Gal-D-NAC-treated mice at 48 h post exposure. Arrows indicates nuclear pyknosis (white arrow), cellular vacuolation (black), and cytoplasmic eosinophilia (blue) (b) Sections from healthy control, APAP exposed untreated and APAP exposed Gal-D-NAC-treated mice at 96 h post exposure. (c) Semi quantitative scoring of extent of cell death (pyknosis), hypereosinophilia, and vacuolation was graded and combined score is depicted in graph. Gal-D-NAC-treated animals had lower areas of cell death/necrosis, decreased vacuolation and decreased areas of hypereosinophilia indicating improvement in degree of injury with treatment (*p < 0.05). Healthy animal did not show these features. Scale 100 μm. Detailed scoring is provided in the supplement.

| Cell and animal studies
The details about the materials and methods for the cell and animal studies can be found in the supporting information S1.

| CONCLUSIONS
D4-Gal has significant potential as an intracellular hepatocyte-specific delivery vehicle in the crowded field of liver-targeted nanoparticles.
There are nanosystems with stronger binding affinities to ASGPR, 69 higher accumulation in the liver, 70

DATA AVAILABILITY STATEMENT
The data are provided in the supporting information. Additional data are available on request.