Plasma and liver pharmacokinetics of the GalNAc-siRNA JNJ-73763989 in rAAV-HBV infected mice

JNJ-73763989 is a N-Acetylgalactosamine (GalNAc) conjugated short interfering RNA (siRNA) combination product, consisting of 2 triggers, in clinical development for chronic hepatitis B virus (HBV) infection treatment, inducing a selective degradation of all HBV mRNA transcripts . Our aim is to characterize the plasma and liver pharmacokinetics (PK) of JNJ-73763989 after intravenous (IV) and subcutaneous (SC) administration in recombinant adeno-associated (rAAV) HBV infected mice. Forty-two male rAAV-HBV infected C57Bl/6 mice received JNJ-73763989 doses of 10 mg/kg IV or 1, 3 and 10 mg/kg SC. Plasma and liver concentrations were analyzed simultaneously using nonlinear mixed-effects modeling with the NONMEM 7.4. A population PK model consisting of a two-compartment disposition model with transporter-mediated drug disposition (TMDD) including internalization to the liver compartment, linear elimination from plasma and liver, and first-order absorption following SC administration was suitable to describe both plasma and liver PK. After SC dosing, absolute bioavailability was complete and flip-flop kinetics were observed. JNJ-73763989 distributes from plasma to liver via transporter-mediated liver internalization in less than 24 hours, with sustained (>42 days) liver exposure. The saturation of transporter-mediated liver internalization was hypothesized to be due to asialoglycoprotein receptor (ASGPR) saturation. Increasing the dose decreased the relative liver uptake efficiency in mice for IV, and to a lower extent for SC administered JNJ-73763989. Lower dose levels administered SC in mice can maximize the proportion of the dose reaching the liver. Model-based simulations. Single dose deterministic simulations for IV and SC doses of 1, 3 and 10 mg kg -1 were performed to evaluate the influence of dose and administration route on the liver uptake of JNJ-73763989. Multiple dose deterministic simulations for SC monthly dose levels of 1, 3 and 10 mg kg -1 for dosing intervals ranging between 1 day to 28 days over a period of 1 year were performed to evaluate the influence of ASGPR saturation and to select the dosing regimen that maximizes JNJ-73763989 liver uptake. For multiple dose deterministic simulations, the total dose amount administered per month was maintained constant across all dosing intervals (e.g. for a dosing interval of 2 weeks, the amount administered per dosing occasion is equal to the total monthly amount divided by 2). saturation. Our findings suggest lower dose levels administered SC in mice maximize liver uptake efficiency (in terms of relative drug amounts).


INTRODUCTION
Hepatitis B Virus (HBV), a hepatotropic partially double-stranded DNA Orthohepadnavirus and member of the Hepadnaviridae family, causes an infection that attacks the liver and induces both acute and chronic liver disease. Acute HBV infection occurs after contact with body fluids of an infected host.
Approximately 2-6% of adults with acute HBV will develop chronic HBV infection, whereas patients exposed to HBV in early childhood have a 90% chance to develop chronic HBV. Chronic HBV infection remains a major global public health problem since patients may develop liver cirrhosis and/or hepatocellular carcinoma and are at high risk of death. In 2019, WHO estimated 820,000 deaths annually among the 296 million people living with chronic HBV (World Health Organization, 2022).
As any other Hepadnavirus, HBV replicates via protein-primed reverse transcription of pregenomic RNA (pgRNA) (Summers and Mason, 1982). Upon infection, circular, partially double-stranded HBV DNA is converted in the nucleus of the hepatocytes to a covalently closed circular DNA (cccDNA) that assembles into a mini-chromosome. This corresponds to a template for viral mRNA transcription, driving translation into viral proteins, including hepatitis B envelope antigen (HBeAg) and hepatitis B surface antigen (HBsAg) (Seeger and Mason, 2015). Additionally, HBV DNA integration in the host DNA occurs via DNA repair pathways, starting in early HBV infection (Zhao et al., 2020). While integrated HBV DNA is unable to produce pgRNA, representing a replicative dead-end for the virus, it is expected to influence HBV replication, persistence and pathogenesis (Tu et al., 2017). The HBeAg positive, chronic infection stage is characterized by a high HBV replication rate and high viral protein load, which results in elevated for chronic HBV only aim for functional cure, defined as sustained HBsAg loss and HBV DNA suppression when off-treatment (Ning et al., 2019). Functional cure is associated with improved clinical outcomes, suggesting prolonged patient survival (Ahn et al., 2005). However, functional cure is rarely achieved using modern antiviral treatment as it does not eliminate the risk of resurgence of the viral infection because of the nuclear persisting cccDNA (Loglio and Lampertico, 2020).
RNA interference (RNAi), the endogenous pathway used by short interfering RNA (siRNA) therapeutics, can be used as a tool to successfully and specifically silence gene expression by degrading specific mRNA sequences complementary to the siRNA therapeutic and reducing target protein expression (Fire et al., 1998). JNJ-73763989, a siRNA combination product in clinical development for chronic HBV treatment, induced a selective degradation of all HBV mRNA transcripts and consists of two triggers: JNJ-73763976, the S-trigger, targeting all S open reading frame (ORF) mRNA including transcripts derived from integrated DNA and cccDNA-derived transcripts, and JNJ-73763924, the X-trigger, targeting X ORF mRNA present in all cccDNA-derived transcripts (Gane et al., 2019b). Subsequently, the degradation of viral mRNA from all sources leads to decreasing levels of circulating HBV related proteins, including HBsAg, HBeAg, and HBV DNA. Ultimately, suppression of viral protein expression is aimed at removing the tolerogenic effects of high antigen load, which in turn might allow immune rejuvenation, thereby potentially increasing the likelihood to obtain chronic HBV functional cure.
In JNJ-73763989, both the S-and X-triggers are siRNAs conjugated with triantennerary Nacetylgalactosamine (GalNAc), facilitating hepatic delivery. GalNAc conjugates show a high affinity for 7 2010). While the exact mechanism of escape across the endosomal lipid bilayer membrane remains unknown, enough siRNAs enter the cytoplasm to induce robust responses in vivo (Springer and Dowdy, 2018).
Our goal was to characterize the pharmacokinetics (PK) of JNJ-73763989 in plasma and liver after IV and SC administration of both S-and X-trigger in recombinant adeno-associated (rAAV) HBV infected mice, aiming to understand the efficiency of JNJ-73763989 to enter the liver as a function of dosing regimen and route of administration.

Study design and bioanalysis methods.
Data from a preclinical in vivo study including 42 rAAV-HBV infected C57Bl/6 male mice were used. Animals, approximately 6 weeks old, with a body weight ranging from 24 to 30 g, were used to establish the rAAV-HBV infection by IV administration of rAAV-HBV virus (Beijing FivePlus, Molecular Medicine Institute, China) at a concentration of 1⋅10 11 viral genomes in 200 L phosphate buffered saline (PBS) 60 days prior to JNJ-73763989 dosing. Infection was considered stable over time as monitored by measuring HBsAg in the vehicle cohort. Mice were block randomized to JNJ-73763989 IV or SC in a 2:1 S-to-X-trigger ratio at a total (S+X) dose of 1, 3 and 10 mg/kg. JNJ-73763989 dosing formulations were freshly prepared by dissolving JNJ-73763989 in PBS (at pH 7.4). Dosing formulations were kept in the refrigerator, protected from light, and were stable for 26 days.
h for IV) post-dose, and 75 L of blood was sampled at 2 and 4h after dosing. Blood samples were centrifuged within 1 hour after sampling, and plasma was stored at -80 °C within 1 hour after centrifugation. Blood and plasma samples were always protected from direct exposure to UV light to avoid RNA self-cleavage (Ariza-Mateos et al., 2012).
At study endpoints, mice were sacrificed via decapitation after euthanasia. Blood and tissues were harvested between 1 and 42 days after initial dosing. The liver tissues were individually homogenized in homogenization buffer (1/9 w/v) using Precellys homogenizer beads at high speed for 30 seconds.
Samples were heat inactivated at 56 °C overnight and were stored at -80 °C until analysis.
All samples were analyzed for S-and X-trigger using hybridization-based liquid chromatographyfluorescence assay. The assay principle is described in Wang et al. 2016 (Wang andJi, 2016). The linear range of quantification in plasma was 2.10 ng/mL to 2100 ng/mL for S-trigger and 1.00 ng/mL to 1000 ng/mL for X-trigger, whereas the linear range for quantification in liver homogenate was 21.0 ng/g to 21000 ng/g for S-trigger and 10.0 ng/g to 10000 ng/g for X-trigger. To conduct the pharmacokinetic analysis, liver concentrations expressed in nmol‧g -1 were scaled to nmol‧mL -1 based on the calculated relative murine hepatocyte density of 1.088 mL g -1 (Sohlenius-Sternbeck, 2006;Morales-Navarrete et al., 2015).
Modeling analysis. The structural PK model of JNJ-73763989 used for the PK data analysis is schematically presented in Fig. 1. A nonlinear mixed-effects population PK modeling approach was used to jointly describe the plasma and liver concentration-time data across a 1 to 10 mg/kg SC dose range and after a 10 mg/kg IV dose administration of JNJ-73763989. A two-compartment model for plasma with an additional liver compartment, including the competition for the transporter-mediated drug disposition (TMDD) from plasma to liver, was selected to describe JNJ-73763989 PK. Both triggers were described by the same structural model, with potential differences in parameter estimates, which was judged based on visual inspection of diagnostic plots and on reduction of the objective function value (OFV) in line with -test (3.84 reduction for 1 degree of freedom and p = 0.05). The final model was selected based on This article has not been copyedited and formatted. The final version may differ from this version. lowest OFV, adequate goodness-of-fit and visual predictive check, acceptable parameter uncertainty (e.g. percentual relative standard error (RSE) ≤ 50%), physiological relevance of parameter estimate, absence of correlations, and the principle of parsimony (Dykstra et al., 2015).
In this model, after SC dosing, a linear absorption for S-and X-trigger, characterized by the first-order absorption rate constant ( ), was assumed and described as follows: where represents the amount of S or X-trigger in the dosing compartment after SC JNJ-73763989 administration. Absolute bioavailability (F) was estimated (and constrained to be between 0 and 1 using a logit transformation) for both triggers by simultaneous analysis of plasma and liver concentrations after forming the transporter-siRNA complex (RC) according to the second-order association rate (k on ) and first-order dissociation rate constant (k off ). Once the RC is formed, it is internalized into the liver, according to a first-order process represented by a liver internalization rate constant ( ) as described in This article has not been copyedited and formatted. The final version may differ from this version. Once internalized in the liver, both triggers are distributed in a liver volume ( ), and the corresponding liver concentrations ( ) are eliminated via a linear pathway, quantified by a liver elimination clearance ( ). The binding of JNJ-73763989 to the transporter was characterized by the quasi-steady-state (QSS) approximation of the TMDD model (Yan et al., 2012;Koch et al., 2017). The QSS approximation assumes that the RC is at steady state, the complex internalization rate is not negligible compared to the dissociation rate and the drug binding to the transporter is balanced by the complex dissociation and internalization as shown in Equation 3: The steady-state constant (K SS ) was estimated as shown in Equations 4-6.
represents the total (free and transporter-bound) S or X concentration and represents the total transporter concentration. Provided the binding properties of GalNAc-conjugates to ASGPR, was assumed to be equal to the murine hepatic total ASGPR concentration of 647 nM reported by Bon et al.
and was fixed in the model (Bon et al., 2017). was assumed to be consistent between both triggers and equal the transporter degradation rate, . Assuming is constant over time and = , this resulted in = ⋅ .
Both S-and X-trigger were assumed to bind to the same ASGPR binding site. Therefore, competitive binding between the two triggers was accounted for as previously described in literature, and , and , were estimated (Koch et al., 2017). The ordinary differential equations describing the S-and Xtrigger concentrations in plasma as well as the unbound transporter concentrations were as follows: Equations 7 to 9 are the product of matrix (M) and vector (G) as shown by Koch et al (Koch et al., 2017) and reported in Equation 10: Where matrix , , , , is characterized by: and vector , , , , is characterized by: This article has not been copyedited and formatted. The final version may differ from this version. Where the elimination rate constant, , defined as , was defined for the S-and X-trigger as , , and , , respectively.
After liver internalization, liver disposition is characterized by (using Equation 4): represents S or X-trigger liver amounts and , represents the first-order liver elimination rate. where is the typical population parameter, and , is assumed to be an independent and random normal individual deviation from log( ) with zero mean and a variance of . where , is the predicted plasma or liver concentration for the th concentration of the th individual for the k th trigger, and ɛ is assumed to be an independent and random normal variable representing the residual error for the log th concentration of the th individual for the k th trigger, with zero mean and a variance of .

RESULTS
The final analysis dataset consisted of 262 concentration-time observations collected from 42 male rAAV-HBV infected mice, of which 178 observations originated from plasma and 84 observations from liver. The two-compartment PK model including TMDD from plasma to liver as described in the methods section adequately described the JNJ-73763989 disposition as can be observed in Fig. 2 (Supp. Fig. 1), although liver JNJ-73763989 concentrations after 1 mg/kg SC administration were slightly overpredicted.
, and were found to be trigger-specific. All other parameters were consistent between both triggers. JNJ-73763989 showed rapid distribution from plasma to liver with sustained liver exposure due to a slow liver elimination. The liver elimination rate constants for S-and X-trigger (0.00225 h -1 and 0.00188 h -1 , respectively) were found to be markedly lower than the corresponding plasma elimination rate constants, with S-and X-trigger liver half-life of 12.8 days and 15.3 days, respectively.
Plasma-to-liver transport was characterized by a TMDD model. differed approximately 2-fold between the triggers, where X-trigger showed a higher affinity for ASGPR ( , = 143.2 nM and , = 73.7 nM). Fig. 3 (left) shows the relative occupancy of the transporter as a function of increasing trigger concentrations, showing the differences in transporter affinity for S-and X-trigger if administered as a monotherapy. An increased relative occupancy can be observed at after 10 mg/kg IV (100%) distribution (see Methods section) significantly improved the model fit (Δ = −16.02). Betweensubject variability was only found significant for the liver volume of distribution ( 33.0% coefficient of variation (CV)). Liver residual variability was relatively low (10.3% and 9.7% CV for S-and Xtrigger, respectively) compared to plasma residual variability (29.1% and 38.3% CV for S-and X-trigger, respectively).
SC administered JNJ-73763989 liver uptake efficiency is increased compared to IV administered drug as illustrated in Fig. 4 (upper panels). Simulations indicated that the relative liver uptake for 1 and 10 mg/kg JNJ-73763989 decreased from 46.0 % to 16.2 % for S-trigger and from 45.9 % to 10.7 % for X-trigger after IV administration compared with a decrease from 60.1 % to 50.1 % for S-trigger and from 61.5 % to 51.4 % for X-trigger after SC administration, respectively. Fig. 4 (lower panels) presents the actual dose that is reaching the liver, which increases upon increasing the dose, but this increase is relatively higher for SC compared to IV administration. The absolute drug amount reaching the liver after 3 mg/kg SC administration is greater than that after 10 mg/kg IV administration for both triggers.
Model-based simulations of the mouse liver concentration-time after multiple dose regimens of monthly 1, 3 and 10 mg/kg SC administered JNJ-73763989 are shown in Figure 5. Simulations show less than dose proportional liver PK upon multiple dosing, especially with higher doses associated with longer intervals in this simulation setup. Interestingly, for the total doses currently investigated in this mouse model (Table 3), the liver at SS after 10 mg/kg monthly dosing increased from monthly to daily dosing (ca. 20%), whereas for the two lower dose (1 and 3 mg/kg) an increase (from monthly to weekly dosing) is followed by a slight decrease (from weekly to daily dosing). This observation is due to different saturation effects across different dose levels.

DISCUSSION
A two-compartment plasma PK model with TMDD describing liver internalization, linear distribution to a non-specific peripheral compartment, linear elimination from plasma and liver, with first-order  (Fig. 2).
Distribution. The estimated central volume of distribution ( ), 1.99 mL for a 26.7 g mouse, was very similar to the mouse plasma volume (around 1.56 mL according to FELASA guidelines). Following IV or SC administration, both S-and X-triggers distributed rapidly to the liver. Furthermore, the model predicted a liver weight of 0.80 g, consistent with the liver weight observed in C57BL/6 mice (3% of the total body weight) (Kushida et al., 2011). The non-linear liver distribution, driven by saturable ASGPRmediated hepatocyte internalization, was characterized by competitive binding between S-and X-trigger ( Fig. 3). The X-trigger affinity was found to be approximately 2-fold higher relative to S-trigger affinity for the transporter. Based on Fig. 3 (left panel), it was determined that >90 % relative occupancy is achieved at plasma concentrations greater than 843 nM and 230 nM for S-and X-trigger respectively, which are exceeded immediately after 10 mg/kg IV dosing, resulting in 100% relative occupancy at C max after IV. During the period where plasma concentrations are above those values, transport into the hepatocyte would be at close to maximum capacity, thus limiting the efficiency of liver transport due to transporter saturation. The after 10 mg/kg JNJ-73763989 IV administration exceeded , by 1443-fold and , by 721-fold, whereas after 10 mg/kg JNJ-73763989 SC administration did not Elimination. The elimination (clearance) of JNJ-73763989 from plasma was relatively fast and was quantified through a non-specific linear process, which accounts for all possible mechanisms of JNJ-73763989 elimination, except the liver disposition. After IV administration of JNJ-73763989, the alpha and beta plasma half-lives for S-trigger were estimated to be 6.39 min and 64.2 min respectively, and 3.77 min and 59.4 min, respectively, for X-trigger. Liver half-life was 12.8 days and 15.3 days for S-and Xtrigger, respectively, which is substantially longer than the plasma half-life.
The JNJ-73763989 PK features described above have important consequences in defining the optimal route of administration, dose level and dosing regimen as described below.

Route of administration.
The results of the mice study confirmed higher JNJ-73763989 liver exposure after SC administration, relative to IV administration at the same dose level, which is explained by the flip-flop phenomenon and its effect on transporter saturation. Moreover, JNJ-74763989 is administered as SC injections in clinical phase studies, enhancing liver uptake (Gane et al., 2019a).
The limited number of ASGPR can be readily saturated with the high plasma concentrations achieved after IV administration of JNJ-73763989. This may lead to considerable dose wastage and a reduction of liver uptake efficiency relative to the SC route, whereas the prolonged absorption provides relatively lower plasma concentrations leading to reduced ASGPR saturation but longer-lasting liver penetration (Fig. 4). Dose level. Given the ASGPR-mediated disposition, the increase in JNJ-73763989 liver exposure with dose became less than dose-proportional, as shown in Fig. 4. This effect is clearly more prominent following IV dosing compared to SC administration. Furthermore, the difference in the fraction of the dose that reaches the hepatocytes between IV and SC dosing decreased with dose (Fig. 4) et al., 2021;McDougall et al., 2022). This behavior is typically associated with a less than dose proportional increase in liver exposure and more than dose proportional increase in plasma exposure. Interestingly, this finding of the percentual dose recovery in the liver increasing in a less than dose-proportional manner is also consistent with that of McDougall et al. (McDougall et al., 2022).
Dosing regimen. Repeated dosing according to the simulated dosing regimens leads to JNJ-73763989 liver accumulation in mice. Since higher doses lead to increased transporter saturation, lower at SS is typically observed for less compared to more frequent dosing for a given total monthly dose. Moreover, longer dosing intervals lead to more fluctuation in the time course (peak-to-trough) of JNJ-73763989 liver concentrations. In this context, daily dosing will lead to smaller fluctuations in the JNJ-73763989 liver concentration-time profile but slightly larger liver accumulation than weekly, biweekly or monthly dosing ( Fig. 5 and Table 3). Interestingly, with the simulated multiple dosing regimen, daily dosing leads to accumulation of JNJ-73763989 in plasma, subsequently leading to a less efficient liver uptake as expected in the absence of plasma accumulation. Consequently, weekly dosing of JNJ-73763989 in mice may allow to combine the smallest liver concentration fluctuations with negligible plasma accumulation.
In conclusion, JNJ-73763989 is a GalNAc-siRNA combination product consisting of S-and X-triggers, saturating the transporter at sufficiently high concentrations, which is limiting liver uptake. Complete bioavailability and slower plasma kinetics after SC absorption in this mouse model allow to limit transporter saturation, thereby increasing relative drug amounts reaching the liver. Model-based simulations can aid in the optimization of dose and regimen combination in case there is transporter This article has not been copyedited and formatted. The final version may differ from this version.       This article has not been copyedited and formatted. The final version may differ from this version.  Time, months ver concentration, nmol/g liver 10 mg/kg 3 mg/kg 1 mg/kg