In vivo organ specific drug delivery with implantable peristaltic pumps

Classic methods for delivery of agents to specific organs are technically challenging and causes superfluous stress. The current study describes a method using programmable, implantable peristaltic pumps to chronically deliver drugs in vivo, while allowing animals to remain undisturbed for accurate physiological measurements. In this study, two protocols were used to demonstrate accurate drug delivery to the renal medulla. First, the vasopressin receptor-2 agonist, dDAVP, was delivered to the renal medulla resulting in a significant increase in water retention, urine osmolality and aquaporin-2 expression and phosphorylation. Second, in a separate group of rats, the histone deacetylase (HDAC) inhibitor, MS275, was delivered to the renal medulla. HDAC inhibition resulted in a significant increase in histone H3-acetylation, the hallmark for histone deacetylase inhibition. However, this was confined to the medulla, as the histone H3-acetylation was similar in the cortex of vehicle and MS275 infused rats, suggesting targeted drug delivery without systemic spillover. Thus, implantable, peristaltic pumps provide a number of benefits compared to externalized chronic catheters and confer specific delivery to target organs.

Scientific RepoRts | 6:26251 | DOI: 10.1038/srep26251 purchased from Harlan (Indianapolis, IN) and maintained on a 12 h light 12 h dark schedule. Rats were fed a normal salt diet (0.49% NaCl Teklad TD.96208) and allowed water ad libitum.
For all animal surgeries, proper aseptic technique was used, and all drapes, supplies, surgery tools and gloves were sterilized. Rats were anesthetized with 2% isoflurane and given an s.c. injection of carprofen (5 mg/kg) and buprenorphine (0.1 mg/kg) to minimize pain post surgery. Just lateral to the left rib cage, hair is shaved, and the skin prepared by three alternative wipes of 10% betadine (Purdue Pharma, Stamford, CT) and 70% ethanol (in water). A small incision is made through the skin and muscle, and the kidney exposed. The adrenal gland of the left kidney was carefully freed from the upper pole of the renal capsule before the renal pedicle is ligated with 5-0 silk suture (Ethicon, Summerville, NJ) and the kidney removed. The muscle was sutured closed with 4-0 prolene suture (Ethicon) and the skin was closed with surgical staples. The area was cleaned with 3% hydrogen peroxide in water, and the incision was given 0.25% Marcaine + 0.5% lidocaine (mixed 50/50). The animal was placed in a clean cage and allowed to recover for 7 days before the pump surgery. Animals were closely monitored after surgery and if they lost more than 20% body weight, appeared lethargic or lost righting ability, they would've been euthanized and excluded from the study; however, in this study that was not observed, thus all animals in this study were included.
Pumps, modification and pump surgery. In this study, implantable peristaltic pumps (iPrecio ® model smp-200, Tokyo, Japan) were modified for delivery of drugs to the renal medulla. These pumps are 38.7 × 19.2 × 9.7 mm and emptied weight is 7.9 g. The reservoir holds a maximum volume of 900 μ l of reagents. The pumps can be programmed to flow at rates from 1 ± 0.1 μ l/h to 30 ± 0.1 μ l/h, and before surgery pumps can be programmed to have a constant or variable flow rate depending on the end user's experimental design. The battery life of the pump depends on the flow rate, but it is predicted to pump for 6 months at 1 μ l/h or 1 week at 30 μ l/h. In our hands, we have used pumps at 30 μ l/h for 30 min followed by 9 μ l/h for over 1 month of pumping time. The refill rate will be dependent on the flow rate, so for example at 1 μ l/h the pump will need to be refilled in 37 days, while at 9 μ l/h it will need to be refilled every 4 days. Currently they cost ~$270/pump. Although they are considered "disposable" we have successfully reused pumps. In the MS275 study (see below) the pumps were new but in the dDAVP study (see below) the pumps were sterilized and reused. The sterilization protocol in listed in Supplementary Note 1. Just prior to surgery, the pumps were programmed to pump at 30 μ l/h for 30 min to  maintain catheter patency and then 9 μ l/h for the remainder of the study. Next the pumps were removed from their packaging, and the catheter cut to 6 cm in length (Fig. 1A). The end of the catheter was modified (for renal medullary specific infusion in this case) by placing a 5 mm circle of silicone sheeting and inserting a small piece of micro medical grade catheter vinyl tubing (V/1) into the catheter. The joint between catheters was sealed and secured by a small drop of superglue (Fig. 1B). The pump was then placed subcutaneously on the back, and sutured into the muscle (Fig. 1C). A small incision was made through the muscles of the abdomen, and the catheter inserted 7-8 mm into the renal capsule and adhered with Vetbond ® (Fig. 1D). The muscles were then sutured together and the skin closed with staples. The rats were allowed to recover for 2 days prior to delivery of drugs or saline. Detailed explanation of the surgery is located in Supplementary Note 1 and can be found in the Nature Protocol Exchange.
Drug delivery. In this study, there were 2 different experimental protocols: 1) Vehicle (saline) compared to the vasopressin receptor-2 agonist, dDAVP (n = 4/ group). 2) Vehicle (30% DMSO in sterile saline) compared to the class 1 histone deacetylase inhibitor, MS275 (n = 5/group). At the start of each protocol, each rat was anesthetized with 2% isoflurane, and the remaining saline in the pump removed from the reservoir by inserting a 27-gauge needle syringe percutaneously and withdrawing. The pump was then filled with either vehicle, 1.1 ng/ μ l dDAVP, or 1.5 μ g/μ l MS275 and the rats were placed in a metabolic cage for urine collection. Emptying and refilling of the pumps was all done via a syringe and needle percutaneously without the need to re-open the wound and externalize the pump. The rats were also challenged with high salt diet (4.0% NaCl) during the infusion. dDAVP was delivered at 10 ng/h at a rate of 9 μ l/h to the interstitium of the renal medulla for 4 days. This rate of delivery was based upon previous studies using tethered catheters that determined 10 μ l/min infusion of 0.9% saline had no significant effect on renal hemodynamics 6 , thus we were well below this delivery rate and don't anticipate affecting renal hemodynamics. Second, at this rate, the pumps have to be refilled every 96 h, thus limiting the amount of handling incurred by the animals. MS275 was also delivered at 9 μ l/h, which resulted in a delivery of 1mg/kg/day. Urine was collected every 24 h for 4 days. Urine samples were spun 1000 g for 10 min, snap frozen and stored at − 80 °C until analysis.
Flow rate Accuracy. Flow rate accuracy was determined ex vivo and in vivo with new and previous used pumps. For ex vivo determination of flow rate accuracy, pumps were programmed (9 μ l/h or 30 μ l/h), filled with 900 μ l of sterile saline, and weighed. The pumps were kept in 200 ml of water in a 400 ml beaker at 37 °C. Over the course of 48 h the pumps were carefully hand dried and weighed 4 different times. For in vivo determination, pumps were programmed (listed in Table 1) and implanted as described above. After 12 h or 3 days, the volume remaining in the pump was determined by percutaneously withdrawing the remaining solution. From either the weight (ex vivo) or volume (in vivo) the volume pumped was plotted against time and linear regression analysis performed. The slopes calculated from these data were compared to the slope of the expected volumes. The % accuracy was calculated as ((computed slope-the expected slope)/expected slope)*100, and compared to the manufacturer's flow rate accuracy listed in their technical note (http://www.iprecio.com/technology/tabid/147/ Default.aspx) Western blots and histone extraction. Inner medullae from Protocol 1, were lysed in 10 vols/ wt of lysis buffer + protease inhibitors and phosphatase inhibitor cocktail (Thermo, Waltham, MA) as previously described 14 , and spun at 6,500× g for 10 min to pellet nuclei. The supernatant (herein referred to as lysate) from this spin was used to determine expression of AQP2 and phosphorylated AQP2.
Outer medulla and Cortex from Protocol 2, were lysed with lysis buffer + protease inhibitors, and phosphatase inhibitor cocktail, and centrifuged at 6,500× g for 10 min at 4 °C to pellet nuclei. Histones were then extracted from the nuclear pellet by acid extraction with 5 volumes of 0.2 N hydrochloric acid, overnight at 4 °C. The histone sample was then spun at 6,500 g for 10 min at 4 °C to pellet debris, and the protein concentration of the histone supernatant and original lysate determined by Bradford assay (Quickstart ® , Biorad, Carlsbad, CA), and samples stored at − 20 °C until used in Westerns as previously described 14,15 . Antibodies used in the study were

Results
Flow rate accuracy of the pumps in both an in vivo and ex vivo setting was determined using a variety of programs in the pumps. As shown in Table 1, our calculated flow rate accuracy for new pumps used in vivo programmed to deliver at 9 or 15 μ l/h was 1.3% and − 3.4% respectively, and was similar to the − 2.6% accuracy reported by the manufacturer in an ex vivo environment ( Table 1). The manufacturer did not provide any accuracy measurements in an in vivo setting. Furthermore, the flow rate accuracy was within an overall average ± − 2.2% compared to the expected flow rate, suggesting accurate flow rates with new pumps in vivo. Reused pumps were also evaluated in ex vivo and in vivo settings. Again, ex vivo and in vivo recordings of flow rate accuracy ranges for the various programs were −2.3 to 0.9% of the expected flow rate. Finally, with a subset of the in vivo pumps, there were 2 varied flow rate protocols: 1. 30 μ l/h for 30 min + 9 ul/h for 3 days, and 2. 15 μ l/h for 12 h + 9 ul/h for 12 h. The flow rate accuracy was −5.3 ± 2.5% and −1.3 ± 2.7% respectfully, demonstrating the versatility and accuracy for varied delivery in vivo (Table 1). For comparison, the manufacturer of osmotic mini pumps (Alzet, Cupertino, CA) reports that osmotic mini pumps have a variation of infusion rate < 10%. Compared to vehicle infused rats (n = 4), the rats receiving dDAVP (n = 4) gained significantly more body mass; the change in mass of the saline group was 17.1 ± 1.1 g, while the dDAVP group gained 23.7 ± 1.4 g (P = 0.01). In addition, dDAVP treatment caused a significant reduction in water intake (P group < 0.001 P diet < 0.001 P interaction < 0.001 P subjects matching < 0.001) ( Fig. 2A) and urine flow (P group < 0.01 P diet < 0.0001 P interaction < 0.0001 P subjects matching < 0.0001) (Fig. 2B) and a significant increase in urine osmolality (P group = 0.01 P diet < 0.001 P interaction = 0.19 P subjects matching = 0.13) (Fig. 2C). Free water clearance for the saline infused group was − 1.3 ± 0.06 ml/min and this was significantly greater than the dDAVP infused group −1.6 ± 0.07 ml/min (P = 0.02). Food intake, sodium, potassium and chloride excretion and creatinine clearance were similar between the groups (Fig. S1).
To confirm catheter position, each kidney was dissected at the insertion site (Fig. S2A), and catheter placement was determined by tracing the track left by the catheter (Fig. S2B). Proper placement was seen in all 8 animals, with the catheter track ending at the junction between the outer (OM) and inner medulla (IM) (Fig. S2B). To further confirm vasopressin receptor-2 activation by dDAVP, AQP2 phosphorylation was measured at serine 261, 264 or 269 16 of the IM (Fig. 3A). AQP2 expression and phosphorylation of serine 264 and 269 were significantly increased and phosphorylation of serine 261 significantly reduced in the dDAVP treated rats, consistent with previous reports 16 (P AQP2 = 0.03, P p261 = 0.007, P p264 = 0.03, P p269 = 0.02) (Fig. 3A). Next, to determine if intramedullary interstitial infusion with the pumps leads to systemic or non-specific "spillover", the histone deacetylase inhibitor, MS275 17,18 was delivered to the renal medulla. The hallmark of HDAC inhibition is hyper-histone H3 acetylation 19 . As shown in Fig. 3B, intramedullary infusion of MS275 resulted in a significant increase in H3 acetylation in the OM of the kidney (P = 0.02), but not the cortex (P = 0.11).

Discussion
Chronic catheterization is used in both basic science research with animal models, and in the clinic to deliver drugs and other agents. An unfortunate side effect of chronic catheter use is biofilm accumulation and infection 8 , because they must be externalized to a pump. Thus, there is a need to improve on current methodologies. In the current study, small programmable, implantable peristaltic pumps were used to deliver drugs to the renal medulla. The medulla of the kidney plays a critical role in the regulation of whole body water homeostasis. This is predominantly regulated through the actions of the hormone vasopressin acting on the renal collecting duct to promote water channel (aquaporin-2) expression, phosphorylation and apical surface expression resulting in water retention (anti-diuresis) 16 . In the current study, we present the use of implantable pumps to deliver the vasopressin receptor-2 agonist, dDAVP to the interstitial region of the renal medulla. dDAVP is a well-described agonist used to stimulate aquaporin-2 expression and subsequent water retention in rodents 16 and to treat diabetes insipidius in humans 20 . As predicted, delivery of dDAVP to the renal medulla led to an increase in all of the hallmark signs of water retention. There were no incidences of infection in either the kidney, or around the pump. This experiment confirms that implantable pumps are ideal for delivery of drugs to specific organs than the classical used of externalized catheters.
Histone deacetylase inhibitors have emerged as novel therapeutic interventions for the treatment of not only cancer, but also cardiovascular and neurological diseases 21 . Specific organ or tumor delivery of HDAC inhibitors may also help prevent undesirable side effects, such as hyponatremia that have been reported with systemic HDAC inhibitor interventions 22,23 . In the current study, the Class I HDAC inhibitor, MS275 was used to demonstrate the specificity of delivery that can be achieved with implantable pumps. The gold standard endpoint for confirmation of HDAC inhibition is an increase in histone H3-acetylation. It was confirmed that MS275 infusion of the medulla resulted in an increase in histone H3-acetylation in the outer medulla but not the cortex, indicating the extreme specificity of delivery associated with this method.
In preliminary studies, during optimization of the surgery protocol we found only 3 rats with signs of infection (walling off of the implant with localized infection around pump). After improving sterilization techniques of AQP2 phosphorylation of serine 261 was significantly reduced in dDAVP treated rats. *represents P < 0.05 compared to saline infused rats (n = 4/group) as determined by unpaired, two tailed Student's t-test. (B) To demonstrate specificity of the renal intramedullary interstitial infusion, MS275-induced histone 3 acetylation status was determined in outer medullary (OM) and cortical samples (n = 5/group). MS275 resulted in a significantly higher H3 acetylation compared to vehicle infused rats in the OM, but there were no significant differences in acetylation status in the cortical samples. *Represents P < 0.05 compared to vehicle infused rats as determined by unpaired, two tailed Student's t-test.
implants and gas sterilization of tubing and silicone (see Supplementary Note 1), no other signs of infection have been found to date. Although in this study the pumps were only in the animal for 6 days (2 day recovery, 4 day experiment), we have successfully used pumps in animals for 19 days (personal observation). Thus, like all survival surgery including chronic catheters, excellent aseptic technique is critical for preventing infection. Furthermore, Tan et al. 13 reported maintaining peristaltic pumps in rats for 30 days, while measuring blood pressure, without any adverse side effects. Tan et al. 13 also demonstrated that when implantable peristaltic pumps are set to the same flow rate as osmotic minipumps to delivery angiotensin II subcutaneously (150 ng/kg/min), implantable peristaltic pumps resulted in a more rapid, and maintained increase in blood pressure compared to osmotic minipumps 13 . Likewise, Kuroki et al. 24 demonstrated very similar results when they compared implantable pumps to osmotic minipumps and externalize catheters over a 2-week infusion protocol (subcutaneous, angiotensin II, 150 ng/kg/min); blood pressure increases were greater with implantable pumps than chronic catheters or osmotic mini pumps.
To conclude, implantable peristaltic pumps provide a number of advantages over current chronic methods. The advantages over osmotic mini pumps include: 1) easy to interchange infusate while pump is implanted, 2) programmable flow rates allowing flexibility of experimental protocol, 3) able to be sterilized and reused in future experiments because of the long battery life (depending on the perfusion rate, batteries can last to pump for 6 months with a 1 μ l/h delivery rate), and 4) controlled organ specific delivery, avoiding complications of systemic administration. The advantages over chronic externalized catheters include; 1) no restraining of the animal after surgery, 2) maintenance with standard cages and husbandry with no need for special housing or connection to external pumps, 3) reduced amount of drug needed to fill the pump. These advantages improve the quality of data collected, and because the pump and catheters are implanted subcutaneously there is less risk of infection and loss of catheter patency that is typically observed in models with externalized catheters. However, in experimental designs where small volumes of drug can be delivered at low rates (< 10 μ l/h) for up to 1 or 2 weeks, osmotic mini pumps are advantageous over implantable pumps. In the present study implantable peristaltic pumps were used to infuse the medullary region of the kidney, but the catheters can easily be modified to infuse blood vessels, other organs or even specific regions of organs such as a particular lobe of the brain, or tumors. This innovative approach has broad applications for studying targeted drug delivery or tissue specific knockdown (e.g. siRNA), and is an extreme improvement over current methods.