Utilization of endogenous albumin trafficking pathways in the lungs has potential to modestly increase the lung interstitial access and absorption of drug delivery systems after inhaled administration

ABSTRACT Objectives Drug delivery systems typically show limited access to the lung interstitium and absorption after pulmonary delivery. The aim of this work was to undertake a proof-of-concept investigation into the potential of employing endogenous albumin and albumin absorption mechanisms in the lungs to improve lung interstitial access and absorption of inhaled drug delivery systems that bind albumin. Methods The permeability of human albumin (HSA) through monolayers of primary human alveolar epithelia, small airway epithelia, and microvascular endothelium were investigated. The pulmonary pharmacokinetics of bovine serum albumin (BSA) was also investigated in efferent caudal mediastinal lymph duct-cannulated sheep after inhaled aerosol administration. Results Membrane permeability coefficient values (Papp) of HSA increased in the order alveolar epithelia<small airway epithelia<microvascular endothelium, where the permeability of HSA through small airway and microvascular endothelia were approximately 4- and 28-fold higher than alveolar epithelia, respectively. Only 6.5% of the delivered BSA aerosol dose was absorbed from the lungs of sheep over 5 days, although half of the absorbed dose was absorbed via the lung lymph. Conclusion Drug delivery systems that bind endogenous albumin may show a modest increase in lung permeability and absorption after inhaled delivery compared to systems that do not efficiently bind albumin.


Introduction
The inhaled route allows drugs to be delivered directly to the site of lung resident disease.In doing so, concentrations of drug at the disease site can be maximized while limiting whole body (systemic) exposure and associated adverse effects compared to oral administration [1].The inhaled delivery of small molecule drugs, however, is limited by typically rapid absorption from lungs and limited lung residence times [2].This is important since the efficacy of some drugs is related to both the maximum drug concentration that target cells are exposed to, as well as the exposure time [3].To overcome this limitation of small molecule drugs, macromolecular, and nanometer-sized inhalable drug delivery systems have been extensively explored and developed to improve lung exposure time to lung-active drugs.These work by providing a depot for drug liberation in the lungs over a prolonged period of time [1,4].A drawback of this approach though, is that drug delivery systems tend to remain in the air spaces of the lungs and liberate drug over time [4].While this is advantageous for targets that are in contact with the air side of the lungs, exposure to targets located in the lung interstitium is more limited.For example, the primary targets for idiopathic pulmonary fibrosis and lymphangitic carcinomatosa are fibroblasts and cancer cells located within the interstitium of the lungs and lung lymphatic vessels, respectively [5,6].The development of drug delivery systems and approaches that can optimize drug exposure to the lung interstitium is therefore expected to improve the treatment of these diseases compared to conventional oral or inhaled administration of small molecule drugs.It is also expected to improve the treatment of lung diseases compared to the inhaled delivery of conventional 'nanomedicines' that do not access the lung interstitum and also increase the systemic absorption of these systems.
One approach that can be employed is the development of drug delivery strategies that allow the prolonged or controlled release of drug and which also utilize active transport mechanisms that transport solutes from the air side of the lungs into the lung interstitium.From the interstitium, these drug delivery systems can then be absorbed into the systemic circulation.Examples of transporters that are highly expressed in the lungs and which contribute to maintaining homeostasis in healthy lungs are glycoprotein 60 (gp60) that facilitate albumin transcytosis [7], the ATP-binding cassette (ABC) transporters that transport cholesterol and phospholipids [8] and the transferrin receptor [9] that mediates iron homeostasis and metabolism.As the major plasma protein that exhibits bidirectional transport across the respiratory epithelial barrier, albumin transport processes are potentially the most useful targets to enhance the lung interstitial access of drugs [10].These transporters are important in preventing the accumulation of albumin in the epithelial fluid lining the lungs (ELF) that can lead to pulmonary edema.Previous work has shown that the rate of albumin clearance from the distal airways occurs at a rate of approximately 1-2%/h.Further, up to 30% of the cleared dose is absorbed via the lung lymphatics after instillation of 100 ml 125 I-labeled albumin into lung lobes of sheep [11].This is backed by in vivo and in situ work in other animals which show that at typical rates of albumin transport, an entire dose of albumin instilled into distal lungs is absorbed from the respiratory region within 5 days [12].
As a result of the high biocompatibility of albumin and its utility as a drug carrier, several albumin nanoparticle-based nanomedicines have been successfully developed and approved as injectable therapeutics for cancer [13].Following on from this success, albumin nanoparticles have also been developed and explored as inhalable drug delivery systems.The outcomes from these studies have been nicely described in a recent review by Joshi [14].Briefly, albumin nanoparticles have exhibited good tolerability in the lungs, efficacy against a number of respiratory diseases when administered directly to the lungs and good aerosol particle properties after formulation as dry powders.Potential issues with immunogenicity in the lungs emerge, however, when albumin is chemically modified through conjugation of, for example, drugs and pharmacokinetic modifiers.
One approach that has been underexplored to date is the development of drug delivery systems that utilize endogenous albumin in the lungs to facilitate trafficking through the respiratory endothelium into the lung interstitium and subsequent absorption via the blood and lymph.Such systems could usefully take advantage of hydrophobic-binding pockets in albumin such that (1) the delivery of exogenous albumin is not required, and (2) albumin is not chemically modified, thus enhancing the safety of the formulation.To establish whether this approach has significant merit, we first need to fully understand the lung clearance kinetics of albumin using appropriate models.
Sheep are considered one of the best animal models of human lung physiology and pulmonary pharmacokinetics.Further, they are one of the few animal models where lungderived lymph fluid can be sampled via cannulation of the efferent caudal mediastinal lymph duct (CMLD).Importantly, regional lymph fluid sampling can be used to inform interstitial drug concentrations, since interstitial fluid is collected by lymphatic capillaries [15,16].This approach is especially efficient for measuring interstitial protein concentrations [17].However, while sheep were employed in the pivotal pharmacokinetic study by Berthiaume [11], 125 I-labeled albumin was applied exclusively to the lungs as a high volume (100 ml) liquid instillation into the distal airways.This is a common approach used in the literature since the focus has been on the absorption of albumin from the respiratory regions rather than the whole lungs.Data presented by this paper suggested that the 125 I labeling (used as standard in the albumin literature) could have significantly increased the rate of albumin clearance from the lungs (when comparing 125 I-labeled albumin and total protein clearance, of which albumin is the major protein).Liquid instillation is known to increase the rate of macromolecule/nanomaterial clearance from the lungs compared to aerosol or dry powder inhalation approaches [18,19].Finally, while albumin absorption from the lungs has been almost exclusively examined in the respiratory region (since albumin trafficking in this region is responsible for maintaining fluid balance), drug carriers delivered to the lungs as an aerosol or dry powder deposit throughout the lungs and on the surface of the ELF.
The objective of our study was therefore to investigate the rate and site of albumin absorption in the lungs of sheep after aerosol delivery, rather than liquid instillation, and in vitro using monolayers of several key primary human lung cells.Aerosol delivery to the lungs was used since it deposits protein in the same regions of the lung where inhaled drug carriers would be deposited.This would provide more insight into whether albumin trafficking pathways could be usefully utilized to enhance lung interstitial access and/or absorption of inhaled nanomedicines and other drug carriers when compared to dosing via liquid instillation.Further, this approach takes into account elimination of macromolecules and nanomaterials from the lungs via mucociliary clearance which typically occurs after deposition in the conducting regions of the lungs.In addition, the permeability of albumin through various pulmonary membranes has typically been evaluated in vitro or in situ using rodent cells and lung tissues or lung cancer cells, which do not appropriately reflect trafficking through individual lung barriers in humans.The apparent permeability of albumin through air-liquid interface monolayers of primary human alveolar endothelial cells, small airway epithelial cells, and liquid-liquid interface microvascular endothelial cells were therefore also investigated.This model was used since it well capitulates key endothelial and epithelial barriers in the human lungs and since transport kinetics of albumin can be specifically quantified in models of individual lung membrane barriers.

Labeling of HSA
Fluorescent-labeled HSA was initially prepared to allow quantification of HSA transport through monolayers and uptake into cells.Fatty acid-free HSA was prepared in PBS (pH 7.4) to 5 mg/ml and ATTO 647N maleimide was made to 10 mg/ml with anhydrous dimethylformamide.A 1.3-fold molar excess of maleimide dye was added to the HSA solution and allowed to react for 1 h at room temperature in the dark to give HSA conjugated with a single ATTO 647N label on the exposed cysteine residue.This was done exclusively to restrict chemical modification of the albumin structure, which can interfere with target binding and rate of catabolism in vivo.The reaction was then passed through a PD-10 desalting column containing Sephadex G-25 resin to remove free label to give pure 647N-HSA.

Cell culture
Lonza human primary small airway epithelial cells guaranteed for air-liquid-interface (ALI) culture (SAEC, CC-2547S) and human lung microvascular endothelial cells (HMVEC-L, CC-2527) were purchased from Capsugel (NSW, Australia).An immortalized human alveolar epithelial cell line (hAELVi) [20] was purchased from InSCREENeX GmbH (Braunschweig, Germany).SAEC and hAELVi cells were cultured in Lonza S-ALI growth medium (CC-4539) and HMVEC cells were cultured in Lonza EGM-2 MV cell growth medium (CC-3202) according to the supplier's protocols.Primary cells were passaged before reaching approx.80% confluence and were used in experiments between passages 3 and 6.Tissue culture plasticware was coated with huAEC coating solution (InSCREENeX, INS-SU-1018) according to the supplier's protocol prior to culturing hAELVi cells.The cells were maintained at 37°C in a humid atmosphere of 5% CO 2 .

Animals
Female merino (Ovis aries) sheep (1-2 years of age, weighing 24-35 kg) were obtained from a commercial supplier through the Monash Animal Research Platform, Monash University for pulmonary pharmacokinetic studies.Sheep were acclimatized for at least 1 week in communal indoor pens prior to being moved into metabolism cages for a further acclimatization period of at least 1 day.Indoor housing was maintained at 20-22°C and rooms were held on a 12 h light dark cycle.Food (standard sheep pellets and chaff) and water were provided ad libitum with the exception of a fasting period of 12 h prior to surgery.Food and water were returned to sheep once they regained full consciousness from surgery.After surgery, sheep remained in metabolism cages until termination.All experimental procedures were approved by the Monash University Animal Ethics Committee and were conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes.

Permeability of 647N-HSA through lung monolayers of human small airway epithelial, alveolar epithelial, and microvascular endothelial cells
To compare relative transport rates of albumin through alveolar, small airway, and microvascular membranes in the lungs, we examined the apparent permeability of albumin through primary monolayers of human SAEC, HMVECs, and hAELVi cells.Cells were trypsinized and seeded onto the upper 6.5 mm Transwell permeable supports (0.4 µm pore size tissue culture treated polycarbonate membrane inserts, growth area 0.33 cm 2 ) at 30,000 cells in 0.1 ml media.A further 0.6 ml of media was added to lower plate well.Prior to seeding cells, media was added to Transwells to equilibrate them for 1 h in the incubator.In addition, Transwells for hAELVi cells were coated with huAEC coating solution for 2 h at 37°C, and Transwells for SAEC were coated with rat tail collagen I (0.03 mg/ml) for 45 min at 37°C.Coating solutions were then removed and Transwell membranes rinsed with PBS prior to seeding cells.Tissue culture media in both the upper (apical) and lower (basolateral) chambers of the Transwells were changed every other day.After 3 days in culture, SAEC and hAELVi cells were airlifted by removing media in the upper chamber of the Tranwells and changing the medium in the lower chamber to Lonza S-ALI differentiation medium (CC-4539).Cells were exposed to an air-liquid interface for a total of 14 days before being used for permeability assays.HMVEC cells were also cultured for 14 days prior to permeability assays.
On the day of performing the permeability studies (all replicates for each cell line were performed on the same day), media was replaced with fresh media and transepithelial/endothelial electrical resistance (TEER) of the cell monolayers was measured using a Millipore Millicell ERS-2 voltameter (Merck, NSW, Australia, MERS00002) and a MERSSTX01 electrode.Only Transwells with cell monolayers having TEER values greater than 50 Ω.cm 2 (HMVEC-L), 250 Ω. cm 2 (ALI-SAEC), and 750 Ω.cm 2 (hAELVi) were used for permeability studies as used by others.To characterize the cell monolayers some Transwells were stained with anti-zonula occludens-1 (ZO-1) antibody to assess tight junctions, and ALI cell monolayers were stained with Alcian blue to assess mucus formation (see Supporting Information for methodology and data).
The apical and basolateral chambers of the Transwells were prepared for permeability assays by removing the media and washing twice with prewarmed 20 mM HEPES-HBSS buffer (0.1 ml in the apical chamber and 0.6 ml in the basolateral chamber).After a 30 min equilibration period in a shaking incubator (37°C, 55 rpm), the buffer in the upper apical chambers of the Transwells were removed and replaced with 0.1 ml buffer containing 0.5 mg/ml 647N-HSA (or FITC-dextran control), and then placed back into the shaking incubator.Samples of 0.1 ml were removed from the basolateral chamber and replaced with 0.1 ml prewarmed buffer at 15, 30, 60, 120 and 180 min.The removed samples were transferred to a black 96-well plate for analysis.Standard curves were constructed over the range of 0.1-100 µg/ml by serial dilution of 647N-HSA and FITC-Dextran to quantify the levels of these compounds in the samples removed from the Transwells.Quantitation was performed using a Tecan Spark Multimode microplate reader with Ex 646/5 and Em 664/5 for detection of 647N-HSA and Ex 485/20 and Em 535/20 for detection of FITC-dextran.TEER values and dextran control permeability data are given in the Supporting Information.

Uptake of 647N-HSA into of human small airway epithelial, alveolar epithelial, and microvascular endothelial cells
Fluorescent confocal microscopy was used to examine the cell uptake of HSA into each of the cell lines examined above to provide an indication of the extent of active cell internalization of albumin into these cells.SAEC, hAELVi, and HMVEC cells were seeded onto Millipore Millicell EZ 8-well slides (Merck, PEZGS0816) at 20,000 cells per well in 0.2 ml media and allowed to adhere overnight.Wells for hAELVi cells were coated with huAEC coating solution prior to seeding cells.After the overnight incubation, culture media was replaced with HEPES-HBSS alone or containing 0.5 mg/ml 647N-HSA and cells incubated for a 3 h.The HEPES-HBSS was removed, and the cells gently washed 3 times with cold PBS.Cells were then fixed with 4% paraformaldehyde/0.1% TritonX-100 in PBS for 20 min at room temperature, washed 4 times with PBS and then stained simultaneously with DAPI (5 mg/ml, diluted 1/ 600) and Phalloidin-AF488 conjugate (Cell Signaling Technology, 8878S, 6.6 mM in methanol, diluted 1/200) in PBS for 30 min at room temperature.Cells were rinsed briefly twice with PBS and then mounted using ProLong Diamond Antifade Mountant (Thermo Fisher Scientific, NSW, Australia, P36965).Microscopy was performed using a Leica SP8 point scanning confocal microscope with a 63× objective under glycerol and 405 nm (DAPI), 496 nm (Phalloidin AF488) and 633 nm (647N-HSA) lasers.

Surgical cannulation of sheep for pharmacokinetic evaluation
Pulmonary lymphatic pharmacokinetics of BSA was then examined after aerosol delivery to the lungs of merino sheep and bioavailability was determined by comparison to an IVdosed control group.Sheep were cannulated under isoflurane anesthesia using techniques that have been previously described in detail and replicates were conducted over several weeks [18].The preparation of sheep for surgery and provision of antibiotics and analgesics were as described previously [21].Briefly, all sheep were cannulated via the right jugular vein to allow for the intravenous administration of substances and the infusion of saline (0.9% NaCl).Saline was given to all sheep during surgery and for up to 48 h after surgery to assist in cannulation of the efferent caudal mediastinal lymph duct (CMLD) and to maintain CMLD patency.All sheep were then cannulated via the CMLD via thoracotomy as previously described.Sheep that were successfully cannulated via the CMLD and where patency was maintained during the recovery period were assigned to the 'pulmonary dosing' group.Where the CMLD could not be cannulated or where patency was lost during recovery, sheep were assigned to the 'IV dosing' group that allowed calculation of pulmonary bioavailability.This ensured that there was no possible impact of thoracotomy surgery or provision of analgesics in IV versus pulmonary dosed sheep.The number of sheep that were dosed via the IV and pulmonary routes were 5 and 4, respectively.
The CMLD cannula was exteriorized through a small chest and lymph was collected continuously into heparinized tubes attached to the side of sheep using body netting.Sheep were allowed to recover from surgery for 2 to 3 days prior to dosing.

Pulmonary pharmacokinetic analysis of BSA in sheep after inhaled aerosol administration to the lungs
Sheep were used to investigate the pulmonary pharmacokinetics of albumin since they represent a highly physiologically relevant animal model of the human respiratory system.While evaluation of the pulmonary pharmacokinetics of sheep albumin in this model is the most ideal approach, this requires attachment of a label which can significantly increase cell uptake and absorption of albumin.For this reason, BSA was chosen as the test albumin since it shows good homology (92%) with sheep albumin [22] and can be quantified via ELISA without cross reactivity with sheep albumin.
Prior to BSA administration, samples of pre-dose blood (5 ml) and CMLD lymph were collected from all sheep to provide background correction in subsequent assays.Blood samples were collected via the jugular vein cannula into citrate-EDTA tubes, while lymph drained continuously into heparinized tubes.The BSA dosing solution for each sheep was prepared on the day of dosing by diluting the mass of the BSA required to give 1 mg BSA per kg body weight into 5 ml sterile saline (to give a BSA dosing solution of 5 to 7 mg/ml).Stability of the BSA solution to nebulization over 10 min was initially evaluated as described in the Supporting Information and was found to be highly stable over this time frame with no overt changes in tryptophan fluorescence or size as determined by size exclusion chromatography.
Sheep in the IV dosing group were administered BSA via the jugular vein cannula as a bolus over 5 seconds followed by the infusion of 20 ml saline to flush through dose remaining in the cannula.Blood samples (5 ml) were then collected into citrate-EDTA tubes immediately after dosing and at 0.08, 0.5, 1, 2, 4, 6, 8, 24, 48, 72, 96, and 120 h post dose.Blood samples were stored on ice until further processing.
Sheep in the pulmonary dosing group were moved into a custom-built body sheath to restrain animals during pulmonary dosing and collection of bronchoalveolar lavage fluid (BALF).Immediately before dosing, a sample of BALF was collected by infusing and aspirating 20 ml of warm sterile saline into a lung lobe via a catheter placed with the aid of a fiber-optic endoscope as previously described [21].Samples of BALF were stored on ice until further processing.
A cuffed endotracheal tube was then placed into the trachea via the nasal passage and attached to a respirator to control breathing rate during pulmonary dosing (Harvard Apparatus, MA, U.S.A.; 20 breaths per minute, 1:2 IDE ratio).BSA was then delivered to the lungs via the endotracheal tube as an aerosol using a PARI eFlow Inline vibrating mesh nebulizer (PARI, Gräfeling, Germany) until no dose remained (dosing was complete within 5-10 min).Immediately after dosing, a post-dose blood sample was collected as above.An additional BALF sample was also collected from an alternate lung lobe to the initial BALF sample.Further blood samples were collected at times described above for IV dosed sheep and additional BALF samples were collected from different lung lobes at 24, 72, and 120 h post-dose completion.CMLD lymph was also collected at the following post dose time intervals: 0-5 min, 5-30 min, 30-60 min, 1-2 h, 2-4 h, 4-6 h, 6-8 h, 8-12 h, 12-24 h, and daily thereafter.After the completion of pulmonary dosing and immediate post-dose samples had been collected, all tubing used for nebulized administration, the nebulizing bulb, and exhaled air filter were rinsed with water that was collected to quantify the mass of BSA dose not delivered into the lungs to calculate the actual dose, as previously described [23].
After the last samples had been collected, sheep were euthanized via IV Lethabarb and lungs, liver, kidney, spleen, and caudal mediastinal lymph node (CMLN) were collected and weighed.Samples of each tissue were then stored at −20°C in a standard laboratory freezer without a frost-free function until further processing and analysis.
All BALF samples were centrifuged at 4°C for 10 min (3,500 × g) to obtain cell-free BALF.Plasma and cell-free lymph were obtained via centrifugation of blood and lymph, respectively, at 4°C and 3,500 × g for 10 min.All samples were stored at −20°C along with samples of BSA standard (20 ug/ml in predose BALF, plasma or lymph) until analyzed via a commercial BSA ELISA.

Quantification of BSA in plasma, lymph, BALF and organ samples
BSA concentrations in biological samples were determined using a commercial ELISA kit (F030, Cygnus Technologies).The assay linear range was 0.5 to 32 ng/ml).
Tissue samples (and blank tissue to provide a matrix for standard curve generation, 1-3 g) were dissociated in 5 ml Tris buffered saline (TBS, pH 7.4) in Miltenyi 'M' tubes using a Miltenyi GentleMacs homogenizer and then centrifuged at 20,000 × g for 10 min.The supernatant was collected and centrifuged again at 20,000 × g for 10 min to remove any remaining particulate material.
All reagents, standards, controls and samples were brought to room temperature and the wash solution prepared as described.The samples were then processed as per the manufacturer's protocol with some modifications.Samples were diluted in Tris buffered saline (pH 7.4) to the following dilutions to bring samples within the standard curve range: IV plasma, 1:1000; pulmonary plasma, 1:20; lymph, 1:100; BALF, 1:100.Microtiter plates were then measured for absorbance at 450/650 nm using a plate reader.
The urea correction method was used to account for the collection and dilution of BALF samples in saline and determine the actual BSA concentration in BALF, as previously described.Further, BALF concentrations are presented as the % difference compared to the sample collected immediately post dose.

Calculation of membrane permeability coefficients (P app )
P app values (in cm/sec) were calculated by (dC/dT x V R )/(A x C 0 ), where V R is the volume of the basal chamber, A is the monolayer surface area, C 0 is the initial concentration in the apical chamber at time 0, and dC.dT is the steady state linear rate of change in concentration in the basal chamber (in ug/sec).The rate of albumin transport was calculated by linear regression analysis.

Pharmacokinetic analysis and statistics
Non-compartmental pharmacokinetic parameters (including elimination rate constant, k; half-life; area under the curve, AUC; apparent volume of distribution, Vz; plasma clearance, Cl; maximum plasma concentration and time to maximum plasma concentration, Cmax, and Tmax, respectively) were calculated using PKSolver [24].Bioavailability was calculated by 100 × (AUC pulm /AUC IV ).The cumulative proportion of the pulmonary dose recovered in lymph over time was calculated by determining the mass of BSA recovered in each sample and converting this to a % of the delivered dose.The % recovered at each time was added to the % recovered at all prior times to give the cumulative curve.
Statistical analysis was performed using GraphPad Prism, with statistical significance determined at a level of p < 0.05.Concentrations of plasma BSA between pulmonary and IV administration, and between plasma and lymph concentrations were compared via two-way ANOVA followed by Sidak's multiple comparison test.Differences in organ biodistribution were compared between IV and pulmonary delivery via unpaired T test.P app values between cell lines were statistically compared via one-way ANOVA with Tukey's multiple comparison test.

Permeability of HSA through monolayer membranes of lung alveolar, small airway epithelial and microvascular cells
All monolayers exhibited the required TEER values and dextran exclusion, indicative of the capacity for paracellular transport, reflected these TEER values (see Supporting Information).P app values showed that the apical to basolateral transport of HSA occurred significantly slower through hAELVi cell monolayers compared to SAEC (approx.4-fold slower, p = 0.035) and HMVEC (approx.28-fold, p = 0.0018) monolayers (Figure 1).Transport through HMVEC monolayers was most efficient at a P app of 1.4 × 10 −5 cm/sec.

Uptake of HSA into lung alveolar, small airway epithelial, and microvascular cells
To provide further information about the cellular trafficking of albumin through these cell lines, the uptake of HSA was examined via confocal microscopy after 3 h incubation to match the time course of the permeability study (Figure 2).Evidence of cellular uptake was observed in the form of punctate bodies present throughout the cell cytoplasm in all cell lines examined, with no clear evidence of a difference in the extent of cellular internalization between cell lines.

Pulmonary lymphatic pharmacokinetics of BSA in sheep
The plasma concentration vs time profiles of BSA after IV and pulmonary administration in sheep are presented in Figure 3a, and pharmacokinetic parameters are shown in Table 1.After IV administration, BSA exhibited 3-compartment plasma elimination pharmacokinetics over 5 days with slow clearance from plasma.The calculated elimination halflife was 2.5 days.After pulmonary administration, BSA was absorbed slowly into plasma with a T max of 2.5 days and was then eliminated from plasma with an estimated elimination half-life of 6 days, 2-fold slowed than the elimination halflife determined after IV administration.The apparent bioavailability of the pulmonary dose calculated from the plasma profiles was approximately 3 and 15% when calculated to 5 days and infinity, respectively.
In contrast to the plasma pharmacokinetic profiles, concentrations of BSA in CMLD lymph peaked approximately 6 h after pulmonary dosing and appeared to plateau from 1 to 5 days (Figure 3b).Concentrations of BSA were in lymph were also an order of magnitude higher than in plasma (p < 0.0001).Cumulative recovery of BSA in CMLD lymph also showed a relatively constant appearance of BSA in lymph over 5 days, with no evidence of lymphatic uptake slowing over this time (Figure 3c).Total lymphatic recovery of the BSA dose over 5 days was 3.6%, which when combined with plasma bioavailability over this time, suggests that up to 6.5% of the BSA dose was absorbed from the lungs over 5 days, with equivalent absorption occurring via the blood and lymph.In spite of the slow absorption of BSA from the lungs after pulmonary administration, 70% of the delivered dose was cleared from the BALF by 24 h after dosing and 98% by 72 h, likely via a combination of mucociliary clearance and uptake into lung tissue, although these mechanisms were not specifically examined in this study (Figure 4a).By 5 days after dosing however, BALF concentrations were below the level of quantification.After 5 days, 1% of the delivered dose was recovered in the lungs of sheep administered BSA via nebulized inhalation (Figure 4b).In contrast, lung distribution of BSA after IV administration was 2-fold lower compared with after pulmonary administration (p = 0.0004).The biodistribution of BSA in the liver, spleen, and kidneys were low.While there was a trend toward lower concentrations of BSA in the liver and kidneys after pulmonary administration, biodistribution in these organs did not differ significantly between IV and pulmonary dosed sheep (Figure 4b and c).This may have reflected ongoing clearance of BSA from these organs 5 days after IV administration compared to ongoing accumulation of BSA in these organs after pulmonary administration.Interestingly, the biodistribution of BSA into the caudal mediastinal lymph node was significantly higher (approximately 5-fold, p = 0.018) after IV administration compared to pulmonary administration (Figure 4c).

Discussion
Albumin removal from the respiratory region of the lungs is an important mechanism in maintaining lung homeostasis.For this reason, the mechanisms of albumin absorption from the lungs have been closely investigated in the literature using a range of in vitro, in situ and in vivo techniques.What is not clear however, is whether the lung interstitial access and systemic absorption of inhaled drug delivery systems and nanomedicines can be enhanced by 'hitchhiking' albumin transporters after binding to endogenous albumin lining the lung ELF.The purpose of this study was therefore to undertake a focused investigation into the lung absorption and clearance of albumin in primary lung cell monolayers and in sheep after inhaled deposition throughout the lungs (to mimic the location of deposited inhaled drug delivery systems) to understand whether there is likely to be any benefit in developing inhalable drug delivery systems that bind to albumin.While sheep were used as a highly relevant model of the human respiratory system, it should be pointed out that as a pharmacokinetic model for inhaled drugs and macromolecules sheep differ from humans by virtue of their longer trachea and ruminant digestive system which can alter the absorption of drugs cleared from the lungs via the mucociliary escalator.
One of the most significant barriers to macromolecule and nanomaterial absorption in the lungs is the epithelial lining of the lungs [25].From here, systemic absorption may occur via either the microvasculature or the lymphatics.Despite extensive literature data suggesting that albumin is mainly cleared from the lungs in the alveolar region compared to the airways [12,26], our data showed that the rate of HSA absorption was 4-fold higher through small airway monolayers compared to alveolar monolayers (Figure 1).Confocal microscopy images also showed no obvious evidence of a difference in punctate bodies containing HSA after 3 h, suggesting a similar extent of active transcytosis in both cell lines (Figure 2).Of note, transcytosis is well known to be the major mechanism by which albumin is reabsorbed from the airspaces of the lungs through the alveolar epithelium, with transport occurring via both cavaeolin-dependent (eg.gp60, megalin) and independent  pathways [27][28][29][30][31].This has been well characterized in the alveolar epithelium from rodents, large animals and humans.Albumin transport through small airway epithelial cells has not, to our knowledge, been previously examined.As expected based on the literature, the transport of albumin through the microvascular endothelium was considerably more efficient, since this is not a major barrier to albumin absorption from the lungs [25].The pulmonary microvasculature though, is also known to express gp60 which has a major role in the reabsorption of albumin from the lungs back into systemic circulation [7,32,33].The discrepancy between the confocal images and P app values suggest that the transport of HSA through the cell monolayers may have occurred via a combination of transcellular and paracellular transport.However, this was not specifically investigated here since it has been the subject of numerous previous studies that have identified that both processes are involved in albumin trafficking through pulmonary membranes [12].Albumin transport through the alveolar epithelium though, occurs mainly via an active process since the very tight junctions between these cells limit the passive movement of macromolecules.Airway epithelia however, have lower resistance to the passage of macromolecules and it is in these regions that lymphatic vessels are mainly found.The respiratory region of the lungs has a much higher surface area (approx.75 m 2 ) compared to the conducting region, and is therefore responsible for the bulk absorption of albumin from the airspaces back into systemic circulation.The passage of nanosized drug delivery systems from the airside of the lungs to the interstitium, however, may occur through either the respiratory or conducting regions of the lungs after binding endogenous albumin.
For this reason, it was necessary to examine the absorption of albumin from the entire lungs in vivo.Previous in vivo and in situ work have shown that the absorption of albumin from the lungs occurs at a rate of approximately 1-2%/h in vivo, and up to 25%/h in in situ lung preparations [11,34].These approaches, however, have examined lung absorption after liquid instillation of albumin, which can significantly enhance the lung absorption rates of macromolecules compared to nebulized administration, which does not change fluid and protein dynamics in the lungs, better mimicking the behavior of endogenous albumin [19].For this reason, BSA was delivered to the lungs of sheep in this study via nebulized administration which delivers the dose throughout the lungs, including the alveolar region.Further, the concentration of albumin in the ELF is approximately 3-4 mg/ml [35].Given results of our previous work estimating the volume of ELF in 30 kg sheep to be approximately 400 ml, the nominal delivery of 1 mg/kg BSA would not have significantly increased the concentration of albumin in the lungs, which could also change the kinetics of BSA elimination [23].
The results of this work in sheep showed that 70% of the BSA dose was cleared from lung BALF within 24 h, approximately 2-to 3-fold faster than absorption rates previously reported after liquid instillation.It should be noted though that elimination of albumin from the BALF does not only relate to absorption per se.The calculated absorption of BSA from the lungs was limited, however, with only 6.5% absorbed over 5 days.Absorption appeared to occur equally via both the blood and lung lymph.The plasma half-life of BSA after pulmonary administration (6 days) was approximately 2-fold higher than after IV administration (2.5 days), suggesting that absorption was still occurring from the lungs after the 5-day experimental period.This is despite only 1.5% of the BSA dose remaining in the lungs after 5 days.To this end, the bioavailability of BSA when extrapolated to infinity was 15%.It should be acknowledged that the plasma half-life of albumin is approximately 2 weeks in sheep and 3 weeks in humans [36][37][38].This suggests that over the 5-day experiment undertaken here, either the true terminal elimination phase occurred after 5 days, or BSA was eliminated from plasma more rapidly than sheep albumin.Previous work in sheep showed that the terminal elimination phase of 125 I-labeled sheep albumin occurred after approximately 2-3 days, suggesting that over the 5-day time course of our study we should have been able to capture the true elimination phase of BSA after IV dosing [36].The long circulating half-life of albumin (and antibodies) is a function of their ability to bind neonatal Fc receptors (FcRN) that effectively protect these proteins from catabolic degradation [37,38].Species differences in the binding affinity of antibodies with these receptors can lead to changes in antibody pharmacokinetics.As an example, antibodies containing mouse-derived Fc regions are cleared from humans more rapidly than antibodies containing human-derived Fc regions due to reduced binding affinity of mouse Fc with human FcRN [39].In contrast, antibodies with human-derived Fc regions show more prolonged plasma exposure in rats compared to in humans due to a higher affinity of human Fc regions with rat FcRN [40][41][42].To our knowledge, the binding affinity of BSA with sheep FcRN or gp60 has not been established but should presumably closely match that of the sheep receptors based on the high sequence homology between sheep and bovine albumin.
Despite this, the higher plasma half-life of BSA after pulmonary administration compared to IV administration likely led to an overestimation of total extrapolated plasma bioavailability.This suggests that the removal of albumin from the lungs via absorption is more limited after aerosol administration compared to what has been previously reported after liquid instillation.While the proportion of the aerosolized BSA dose cleared from the lungs via a combination of mucociliary elimination and catabolism could not be determined, it is evident that the majority of the albumin dose delivered to the lungs was cleared via these pathways, as for most inhaled macromolecules and nanomaterials.Further, while nebulization can denature albumin and change its pharmacokinetics, we found no evidence of an overt change in albumin structure by measuring tryptophan fluorescence or size exclusion profiles over the nebulization period used here (up to 10 mins, see Supporting Information).The extent of BSA denaturation during nebulization however, was previously shown to decrease as protein concentration increased, up to 1.5 mg/ml [43].The concentration of BSA used here was approx.6 mg/ml.At the air-liquid interface however, protein aggregation can occur.This is more likely to occur after nebulized administration of proteins since the lungs are not flooded with liquid that can mix with the ELF.While aggregation may have occurred in the lungs in this study, it was not possible, or feasible, to identify the extent to which this occurred.Regardless, the study designed allowed us to mimic the behavior of albumin present at the air-liquid interface which is more likely to bind to inhaled drug delivery systems.Importantly, assuming that whole lung ELF in a 30 kg sheep may contain up to 1.4 g of albumin (based on our previously estimated ELF volume of 400 ml in 30 kg sheep and ELF albumin concentration of 3-4 mg/ml), up to 17 mg of albumin may have been absorbed from the lungs per day in the present model.Hence, while the proportion of the albumin dose absorbed from the lungs was low in this case, the absolute mass of albumin removed from the lungs of sheep (and presumably humans) on a daily basis may be reasonable.
Of interest was the recovery of 3.5% of the BSA dose in lung lymph over 5 days (approximately 50% of the absorbed dose).Previous work in sheep and perfused dog lungs suggested that albumin absorption from the lungs after liquid instillation was instead, primarily via the blood [11,44].This is despite the greater permeability of lymphatic versus vascular endothelium due to the presence of a fenestrated endothelium and lack of a basement membrane which facilitates the efficient passive uptake of macromolecules.This is interesting given that liquid instillation would likely increase lymphatic flow rates that would in turn increase albumin uptake via the lymph.The discrepancy with the present study is likely derived from the different modes of administration and delivery of albumin to the entire lungs after nebulization.Specifically, liquid instillation to the distal region of the lungs would bring albumin in contact mainly with small airways and alveoli which have a limited lymphatic supply.In contrast, aerosol administration delivers albumin to the entire lungs, including the trachea and larger airways which have a denser lymphatic supply and more limited vascular supply, better favoring absorption via the lymph.
While the proportion of the BSA dose absorbed via the lung lymph seemed low, the maximum proportion of a nebulized dose of a lymph-absorbable nanomaterial (22 kDa PEGylated dendrimer) or PEGylated liposome was previously reported to be 0.3% over 5-7 days using this sheep model [18,21].The limited lymphatic access of these nanomaterials was suggested to be due to a combination of the poor lymphatic supply around the alveolar region (compared to the larger airways and pleura), combined with higher flow rates of blood through pulmonary microvasculature, providing a greater driving force for vascular absorption than lymphatic absorption [45].This suggests that albumin more effectively accesses the lung lymphatics from the air spaces of the lungs compared to these drug delivery systems.This may be due to efficient absorption of albumin from the airways that contain a denser lymphatic supply than the alveolar regions, despite previous suggestions that macromolecule and nanomaterial absorption occurs primarily through  pneumocytes [45,46].Further, the lung lymph concentration of BSA was 10-fold higher than in systemic venous blood.Altogether, this suggests that drug delivery systems that bind albumin or which are complexed to albumin, would likely have a greater capacity for lymphatic drug targeting and concentration in the lung lymph compared to the inhaled delivery of inherently 'lymph targeted' nanomaterials that do not efficiently access the lung interstitium.

Conclusion
These data collectively suggest that albumin transport from the lungs may enhance the trafficking of inhaled drug delivery systems and nanomaterials that non-covalently bind albumin into the lung interstitium.This is likely to be particularly efficient if the dose of the inhaled drug delivery system is small, binding affinity to albumin is high and the size of the system is small (<30 nm).The role of size stems from the work of others showing that for nanoparticles that bind albumin and other proteins in the lungs, increased hydrodynamic size through protein binding can slow absorption from the lungs [47].Albumin absorption appeared to occur via the airways and the alveolar region based on in vitro permeability studies and in vivo work in sheep.Efficient penetration of albumin through the airways from the lumen facilitates the absorption of albumin from the interstitium into both the vasculature and lung lymphatics.This appears to provide an effective means by which albumin-bound drug delivery systems (that are normally poorly absorbed from the lungs) can gain better access to the lung lymphatics.This is likely to be important for enhancing the lymphatic presentation of inhaled nanomedicinal immunomodulators to more directly target and treat immune diseases in the lungs.Further work is clearly needed though, to establish whether nanomaterials that exhibit high affinity for albumin show more efficient lung interstitial access and absorption from the lungs after inhaled administration when compared to equivalent systems that do not bind albumin.Further, given that lung disease and inflammation can significantly alter lung clearance mechanisms [4,48,49], albumin concentrations in the ELF and pulmonary pharmacokinetics, it would be of interest to examine the utility of 'albumin hitchhiking' of drug delivery systems in various models of lung disease.

Figure 3 .
Figure 3. Plasma and CMLD lymph profiles of BSA after IV and pulmonary administration to sheep.(a) Plasma concentration vs time profiles after IV and pulmonary administration.(b) Plasma and CMLD lymph concentration vs time profiles after pulmonary administration.(c) Cumulative % recovery of BSA in CMLD lymph over 5 days.Plasma and lymph concentrations were normalized to a dose of 1 mg/kg.Data represent mean ± SD (n = 4-5).*P < 0.0001 between plasma and CMLD lymph concentrations.

Figure 4 .
Figure 4. Biodistribution of BSA after IV or pulmonary administration to sheep.(a % Change in BSA concentration of cell-free BALF in sheep over time compared to immediately after the completion of pulmonary dosing.(b)Biodistribution of BSA in whole organs and tissues 5 days after via intravenous or pulmonary dosing.(c) Mass normalized biodistribution of BSA (% dose/g tissue) 5 days after IV or pulmonary dosing.It was not feasible to collect all lymph nodes in the lungs, so biodistribution of BSA in whole lung lymph nodes was not determined (ND).Data represent mean ± SD (n = 4-5).*P < 0.05 compared to IV dosing via unpaired T-test (% dose/organ lung: IV vs pulmonary p = 0.0004; % dose/g lung: IV vs pulmonary p = 0.0002; % dose/g LN: p = 0.018).

Table 1 .
Plasma pharmacokinetic parameters of BSA after IV or pulmonary administration to sheep.