Nanosphere based oral insulin delivery

Journal of Controlled Release 65 (2000) 261–269 www.elsevier.com / locate / jconrel

Nanosphere based oral insulin delivery Gerardo P. Carino*, Jules S. Jacob, Edith Mathiowitz Department of Molecular Pharmacology, Physiology and Biotechnology, Brown University, Providence, RI 02912, USA Received 11 May 1999; accepted 19 August 1999

Abstract Zinc insulin is successfully encapsulated in various polyester and polyanhydride nanosphere formulations using Phase Inversion Nanoencapsulation (PIN). The encapsulated insulin maintains its biological activity and is released from the nanospheres over a span of approximately 6 h. A specific formulation, 1.6% zinc insulin in poly(lactide-co-glycolide) (PLGA) with fumaric anhydride oligimer and iron oxide additives has been shown to be active orally. This formulation is shown to have 11.4% of the efficacy of intraperitoneally delivered zinc insulin and is able to control plasma glucose levels when faced with a simultaneously administered glucose challenge. A number of properties of this formulation, including size, release kinetics, bioadhesiveness and ability to traverse the gastrointestinal epithelium, are likely to contribute to its oral efficacy.  2000 Published by Elsevier Science B.V. All rights reserved. Keywords: Oral Delivery; Insulin; Nanosphere; Polyester; Diabetes

1. Introduction Almost since the initial discovery of insulin by Banting and Best in 1922, an orally effective form of the drug has been an elusive goal for many investigators. Current dosage regimens of insulin that maintain low serum glucose levels low, in order to minimize the long-term complications of diabetes mellitus [1], comprise of up to four subcutaneous injections per day. Compliance with such demanding dosing regimens is difficult, making the development of an oral form clearly appealing. Many different strategies have been attempted to develop a biologically active oral insulin formulation. It is generally acknowledged that protection of the protein drug against degradation by the harsh *Corresponding author. Tel.: 11-401-863-2378.

conditions of the gastrointestinal tract is a major obstacle. Such a successful formulation would have to bypass the two main barriers against the oral delivery of proteins: The enzymatic barrier of the intestinal tract and the physical barrier made up of the intestinal epithelium. Permeation enhancers [2,3], protease inhibitors [4,5], enteric coatings [6–8] and polymer microsphere formulations [9,10] have all been applied towards the development of oral insulin formulations with varying degrees of success. Since our observation that nanospheres made of poly(lactide-co-glycolide) (PLGA) and the polyanhydride poly(fumaric-co-sebacic) anhydride [P(FA:SA)] can cross the intestinal epithelium within 1 to 6 h of oral administration [11], we have focussed our research efforts on applying these systems to the development of an oral insulin formulation. These nanospheres are designed to

0168-3659 / 00 / $ – see front matter  2000 Published by Elsevier Science B.V. All rights reserved. PII: S0168-3659( 99 )00247-3

262

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

contain biologically active insulin and, after oral administration, protect the protein from enzymatic degradation in the intestinal tract and facilitate systemic uptake of the insulin through the gastrointestinal epithelium.

2. Materials and methods Bovine zinc insulin (specific activity of 24 I.U / mg, United States Biosciences, Piscataway, NJ) was used as the stock insulin. A number of different polymers were used in these studies: polylactic acid (PLA, Birmingham Polymers, Inc., lot [ 501-25-1A. MW52000), poly(lactide-co-glycolide) 50:50 (PLGA, resomer 503H, Boehringer Ingelheim, MW538 000), poly-fumaric anhydride-co-sebacic anhydride 20:80 (P(FA:SA), MW57700) and fumaric acid and sebacic acid oligimers (FAO and SAO, respectively). The P(FA:SA), FAO and SAO were all synthesized in house using methods described by others [12].

2.1. Phase inversion nanoencapsulation of Zn– insulin A modified phase inversion nanoencapsulation (PIN) method [13] was used to encapsulate insulin into nanospheres. The decision was made to encapsulate Zn–insulin instead of the sodium salt of the drug because of the lower solubility and slower dissolution of Zn–insulin in water. The general microencapsulation procedure used to encapsulate insulin in each of the polymers was as follows. A 20 mg / ml stock solution of zinc insulin in 100 mM Tris, pH 10 was prepared and titrated with 0.3 M HCl until the drug crystals became

soluble. Generally, 0.5 ml of insulin stock solution were used for each formulation. The insulin was recrystallized by addition of 0.05 ml of 10% ZnSO 4 to the stock insulin solution and the precipitate was immediately added to 10 ml of a 3% polymer solution in methylene chloride (w / v) The mixture was vortexed and probe-sonicated to form an emulsion and then quickly dispersed into 1000 ml of petroleum ether. This resulted in the spontaneous formation of nanospheres. The spheres were then collected by vacuum filtration, frozen and lyophilized to remove excess solvent and water. Table 1 describes the specific formulations made and tested. Iron oxide (Fe 3 O 4 ) was added to many of the formulations as an electron dense tracer for transmission electron microscopy studies (results not presented here).

2.2. Scanning electron microscopy and laser diffraction sizing Nanospheres were sputter-coated with gold–palladium and observed by scanning electron microscopy as described earlier. In addition, the size distribution of these nanospheres was measured using a Coulter Sizer LS 230.

2.3. Insulin loading and release studies The actual insulin loadings were determined by extracting triplicate aliquots containing known amounts of spheres (|10–20 mg) dissolved in 5 ml of MeCl 2 with 5 ml of extraction buffer (0.1 M phosphate-buffered saline (PBS), 0.05% sodium dodecyl sulfate (SDS), pH 8). Release studies were conducted by incubating triplicate aliquots containing known amounts of spheres in 1.0 ml of 0.1 M

Table 1 Formulations tested in vivo Polymer

Zn–insulin loading

Additives

P(FA:SA) 20:80

1.6, 3.0, 5.0 and 7.0%

none

PLGA 50:50, resomer 503H

1.6%

FAO (1:2)

PLGA 50:50, resomer 503H

1.6%

10% Fe 3 O 4

PLGA 50:50, resomer 503H

1.6%

FAO (1:2), 10% Fe 3 O 4

PLA

1.6%

FAO (1:2), 10% Fe 3 O 4

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

PBS at 378. Release fluids were collected after centrifugation of the sphere suspension at 10 000 g for 3 min. The supernatant fluid was saved for analysis and the nanospheres were resuspended in a fresh 1.0 ml volume of PBS. Release fluids were collected at hourly intervals for the first 5 h and at longer intervals (|12 h) for the duration of the release study. Insulin levels were assayed with the BCA Protein Assay (Pierce Chemical Co., Rockford, IL). The total micrograms of insulin corrected for weight of the nanosphere sample were summed over the course of the release and the percent of the total cumulative loading was also calculated.

2.4. Intraperitoneal bioactivity of insulin formulations Since the encapsulation process may denature or otherwise inactivate the loaded insulin, the bioactivity of each formulation was assessed before testing for oral activity. Rats were fasted overnight, anesthetized using methoxyflurane gas, weighed and 0.2 ml blood samples were collected by either tail vein or retroorbital bleed. Twenty-five mg of a formulation suspended in 1 ml of PBS was injected intraperitoneally. The rat was denied food and blood samples were collected at 1.5 and 4 h post-feeding. The heparinized blood samples were centrifuged at 2000 g for 5 min and plasma was removed and saved for glucose monitoring. Only formulations found to be bioactive intraperitoneally were used for further oral studies.

2.5. Glucose analysis Plasma glucose levels were obtained by using the Trinder 100 Glucose Assay (Sigma, St. Louis, MO) following the manufacturer’s protocols.

2.6. Oral bioactivity of insulin formulations Two types of experiments were conducted to assess the oral activity of the various insulin formulations. The first was a simple oral bioactivity test in which fasted rats were anesthetized with methoxyflurane and weighed. Baseline blood samples (approximately 100 ml) were taken by either tail vein or

263

retroorbital bleed. While still under anesthesia, nanospheres in 2 ml of suspension buffer were fed to each rat through a stainless steel orogastric tube. Repeat blood samples were taken at 1.5, 3, 4 and 5.5 h post-feeding. The blood samples were centrifuged and the plasma samples were collected for glucose analysis. The other feeding experiment was a glucose tolerance test which differed from the above experiment only in that the rats received a subcutaneous glucose load of 800 mg glucose / kg of rat in the form of a 5% glucose solution in sterile saline. This glucose tolerance test was meant to model a slightly hyperglycemic state in these otherwise normal rats and allow for the better detection of an insulin effect [14].

3. Results and discussion

3.1. Nanospheres Fig. 1 shows the typical appearance of a insulinloaded nanosphere formulation produced by PIN. In all formulations, even those not pictured here, the PIN method formed nanospheres ,5 mm in size and with no signs of unencapsulated zinc insulin crystals visible on the scanning electron micrographs. In general, PIN produced nanospheres with a diameter of less than 5 mm. Coulter sizing of the specific insulin containing formulation made of 1.6% insulin in FAO:PLGA with 10% Fe 3 O 4 added showed that 80% of the spheres were smaller than 1 mm. However, when considered on a volume percent basis, 25% of the volume was made of spheres less than 4 mm in size and 10% of spheres were less than 1.6 mm.

3.2. Insulin release The P(FA:SA) 20:80 formulations had a very rapid release of the loaded insulin. The vast majority of the insulin (.90%) was out within 2 h, likely to rapid hydration and degradation of the spheres. As a result, these formulations made exclusively of P(FA:SA) were not expected to result in an effective oral insulin delivery system. The three insulin formulations (3, 5 and 7% insulin) made of PLGA with

264

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

6 h. Unfortunately, it is quite likely that this insulin released during the first hour would be unavailable to intestinal uptake as it would be released in the stomach and degraded. A formulation which had a smaller initial release and more prolonged release during the crucial 1 to 6 h time period after feeding would theoretically be more advantageous. Such a release was obtained with the PLA formulation. With this formulation, approximately 24% of the encapsulated insulin was released within the first hour, 35% by 2 h, 42% by 4.5 h and 53% by the end of 22 h with a total recovery of 85.6% of the loaded insulin after extraction. The two formulations made of polymer blends (PLGA:FAO) also showed more favorable release kinetics and are shown in Fig. 2. As seen in the figure, the FAO:PLGA, 10% Fe 3 O 4 formulation releases approximately 70% of the theoretically loaded insulin by 6 h and slowly continues to release while the formulation without Fe 3 O 4 only releases about 35% within this time period. Both formulations release a large amount of their insulin in the first hour of release, but a smaller percentage of that observed in the nanospheres made solely of PLGA.

3.3. Oral efficacy of insulin formulations

Fig. 1. Scanning electron micrographs of nanospheres produced by PIN. Bar53 mm.

10% iron oxide showed similar release characteristics to one another. There was a rapid release of 50–65% in the first hour and almost total release by

Every insulin formulation was found to be bioactive upon intraperitoneal injection, thus showing that phase inversion nanoencapsulation can be used to effectively encapsulate proteins. However, out of all the formulations tested, only the 1.6% insulin in FAO:PLGA formulations with and without an Fe 3 O 4 additive proved to be orally effective. Despite its favorable release kinetics and a similar insulin dosage (20.0 I.U.), the PLA formulation was not successful in lowering the glucose levels in any of the five experimental rats over the first 4.25 h (see Table 2). As stated earlier, glucose reductions were observed in rats fed the FAO:PLGA formulations with and without Fe 3 O 4 . Two fasted 300 gram rats fed 19.2 I.U. of insulin in the FAO:PLGA, 10% Fe 3 O 4 spheres both had reduced glucose of subcritical levels (,60 mg / dl) at 4 h and then began to return to normal at about 6 h. A 300 g rat which received the FAO:PLGA spheres with no Fe 3 O 4 also had a

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

265

Fig. 2. Release of insulin from two FAO:PLGA formulations.

reduction in glucose levels, but did not reach critical levels (see Table 3) This, however, was also taken as a positive result as the fasting level was quite a bit higher, yet still within the normal range, in this animal. This was extremely exciting as all three animals fed this formulation had a reduction of glucose levels. These results can be used to calculate an

Table 2 Blood glucose levels in rats fed 42.2 I.U. insulin in PLA nanospheres; N55 Time (h)

Average blood glucose (mg / dl)6S.E.M.

0

97.469.1

1.5

95.667.2

3.25

92.463.2

4.25

100.267.5

approximate pharmacological availability by utilizing Eq. (1): AUC0 – 6 oral 3 (weight / dose) oral f 5 ]]]]]]]]] AUC0 – 6 i.p. (weight / dose) ip

(1)

Eq. (1) gives you the pharmacological efficacy of an insulin formulation based on percentage deviations from fasting glucose levels where, f is the pharmacological efficacy of the dose versus an i.p. dose, the AUCs 0 – 6 are the areas under the reduction in serum glucose levels on a percent basis from 0 to 6 h and the weight / dose are determined experimentally. The average deviations of plasma glucose levels measured in the two experimental rats are graphed versus time and the trapezoid rule is used to calculate the AUCs 0 – 6 (see Fig. 3). Using the above graph, the oral AUC is calculated to be 126.9%. In a previously published paper looking at intraperitoneally administered zinc insulin,

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

266

Table 3 Blood glucose levels in rats fed 19.2 I.U. insulin in FAO:PLGA (1:2) nanospheres Time (h)

Blood glucose (mg / dl) Rat 1 19.2 I.U. insulin 100 mg FAO:PLGA, 10% Fe 3 O 4

Blood glucose (mg / dl) Rat 2 19.2 I.U. insulin 100 mg FAO:PLGA, 10% Fe 3 O 4

Blood glucose (mg / dl) Rat 3 19.2 I.U. insulin 100 mg FAO:PLGA, 0% Fe 3 O 4

0

75

73

103

1

71

47

86

4

57

45

74

6

62

82

95

the AUC i.p. was found to be 258% at a dose of 4 I.U. in 270 gram rats [6]. Combining this with our results, Eq. (1) yields: 126.9% h 3 (300 g / 19.2. I.U.) f 5 ]]]]]]]]] 258% h (270 g / 4.0 I.U.) where f 511.4%5oral efficacy.

3.4. Glucose tolerance testing In an attempt to make the testing of the insulin

formulations easier and also to better model a fed state, the glucose tolerance test was begun. In these experiments, fasted rats were fed either the experimental formulation or control solutions while simultaneously receiving a subcutaneous injection of glucose (800 mg / kg). Hopefully, reductions of blood glucoses would be when this glucose challenge was present. The 1.6% insulin in FAO:PLGA (1:2), 10% Fe 3 O 4 formulation was tested in greatest detail because its reduction of glucose levels to subcritical levels seemed most promising.

Fig. 3. Top: Average plasma glucose levels in two rats fed 19.2 I.U. of insulin in FAO:PLGA, 10% Fe 3 O 4 nanospheres produced by PIN. Bottom: Percent deviation of plasma glucose levels from fasting levels in same rats.

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

Fig. 4 shows the results of a number of experiments comparing the blood glucose levels in rats fed 20 I.U. of encapsulated insulin versus control rats which just received the glucose load and either a sham fed of suspension buffer or insulin solution [11]. There is a definite reduction in glucose levels over both sham fed and soluble insulin fed animals. The difference is most extreme when compared to the sham fed animals that show statistically significantly higher glucose levels at all timepoints tested. The animals given the encapsulated formulation did not have any increase in blood glucose even when given

267

the large subcutaneous dose of 800 mg / kg. This shows that this formulation is bioactive and implies that it may be useful in the every day control of glucose levels in the face of the continuous influx of glucose and possibly other carbohydrates.

4. Conclusions A specific formulation which is able to reduce glucose levels in normal fasted rats and in rats given a subcutaneous glucose load has been described. This formulation was 1.6% zinc insulin in FAO:P-

Fig. 4. Glucose tolerance test on rats fed 20 I.U. insulin in FAO:PLGA, 10% Fe 3 O 4 nanospheres. Figure reprinted with permission from the publisher [11].

268

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

LGA (1:2), 10% Fe 3 O 4 . This formulation was shown to be at least 11.4% as effective as intraperitoneal injections of zinc insulin. This is very promising in that it shows that the development of an oral insulin (or oral protein) formulation based on thermoplastic nanospheres is possible. We hypothesize that each of the components of the formulation contributes to the overall efficacy of the formulations. The polyanhydride and Fe 3 O 4 seems to be necessary, possibly to increase the bioadhesive properties of the spheres [15,16], allowing for greater interaction between the nanospheres and the GI epithelium and possibly uptake of the nanospheres [11]. Insulin formulations made from solely from polyanhydrides did not work, possibly due to the rapid release observed with these formulations. We hypothesize that the PLGA contributes to the physical stability of the nanospheres and also slows release improving the efficacy. However, favorable release kinetics alone were shown not to be sufficient in obtaining a formulation with oral efficacy as evidenced by the PLA formulations which did not reduce glucose levels despite slow release. In this case, the lack of oral efficacy of PLA formulations may be explained by the fact that PLA nanospheres have been shown to not exhibit any intestinal uptake after oral administration [17]. This supports the notion that nanosphere uptake is an important mechanism by which insulin-loaded formulations of this type exhibit efficacy after oral administration. Ultimately, this ability to be taken up by the gastrointestinal epithelium may be the most important predictor of which formulations will lead to successful oral drug delivery systems. Even more exciting than the results presented here is the hypothesis that with more homogeneously sized nanospheres the oral efficacy may prove to be even better. Size has been shown to be an extremely important parameter affecting nanosphere uptake. The absolute upper size limit of nanospheres that have demonstrated intestinal uptake seems to be about 10 mm [18], but the amount of uptake certainly increases as the nanospheres get smaller [19]. With this in mind, if it is assumed that only the nanospheres less than 4 mm in size (25% of our formulation by a volume percent basis) crossed over the epithelium and were therefore active, the oral efficacy of these spheres could be as high as 45%. If only the spheres less than 1.6 mm in size were taken up,

the efficacy would approach 100% of an intraperitoneal dose. Obviously, future experiments should be conducted with smaller, more homogeneously sized nanospheres to examine if improved oral efficacies can indeed be accomplished in this way. More detailed studies directly measuring insulin levels after oral administration are being conducted. Each component of a formulation adds a level of complexity to the fabrication process and therefore should be determined to be crucial to the activity of the formulation in order to be included in future formulations. These studies do seem to show that each piece is required, but more systematic studies will be conducted to support this. In addition, future experiments using diabetic animals will also be conducted. It is much more difficult to affect the blood glucose levels in normal animals than in diabetic ones, so useful formulations may not appear so in normal animals. This, however, adds further significance to the positive results described here as no other published oral, polymer based insulin formulations have ever shown reduction in glucose levels in normal rats.

References [1] DCCT, The effect of intensive treatment on the development and progression of long-term complications in insulin-dependent diabetes mellitus, New Engl. J. Med. 8329 (1993) 977–986. [2] M. Mesiha, F. Plakogiannis, S. Vejosoth, Enhanced oral absorption of insulin from desolvated fatty acid–sodium glycocholate emulsions, Int. J. Pharm. (1994) 213–216. [3] A. Fasano, S. Uzzau, Modulation of intestinal tight junctions by zona occludens toxin permits enteral administration of insulin and other macromolecules in an animal model, J. Clin. Invest. 99 (1997) 1158–1164. [4] A. Yamamoto et al., Effects of various protease inhibitors on the intestinal absorption and degradation of insulin in rats, Pharm. Res. 11 (1994) 1496–1500. [5] M. Morishita, I. Morishita, K. Takayama, Y. Machida, T. Nagai, Site-dependent effect of aprotinin, sodium caprate, Na 2 EDTA and sodium glycocholate on the intestinal absorption of insulin, Biol. Pharm. Bull (1993) 68–72. [6] E. Touitou, A. Rubinstein, Targeted enteral delivery of insulin to rats, Int. J. Pharmaceut. (1986) 95–99. [7] I. Morishita, M. Morishita, K. Takayama, Y. Machida, T. Nagai, Enteral insulin delivery by microspheres in three different formulations using Eudragit l100 and S100, Int. J. Pharm. (1993) 29–37. [8] G. Gwinup, A.N. Elias, E.S. Domurat, Insulin and C-peptide

G.P. Carino et al. / Journal of Controlled Release 65 (2000) 261 – 269

[9]

[10]

[11]

[12]

[13]

[14]

levels following oral administration of insulin in intestinalenzyme protected capsules, Gen. Pharmac. 22 (1991) 243– 246. C. Damge, C. Michael, M. Aprahamian, P. Couvreur, J.P. Devissaguet, Advantage of a new colloidal drug delivery system in the insulin treatment of Streptozotcin-induced diabetic rats, Diabetologica 29 (1986) 531A. P. Couvreur, V. Lenaerts, B. Kante, M. Roland, P. Speiser, Oral and parenteral administration of insulin associated to hydrolyzable nanoparticles, Acta. Pharm. Technol. 26 (1980) 220–222. E. Mathiowitz et al., Biologically erodable microspheres as potential oral drug delivery systems, Nature 386 (1997) 410–414. A. Domb, R. Langer, Polyanhydrides. I. Preparation of high molecular weight polyanhydrides, J. Poly. Sci. (1987) 3373– 3386. D. Chickering, J. Jacob, E. Mathiowitz, Poly(fumaric-cosebacic) microspheres as oral drug delivery systems, Biotechnol. Bioeng. 52 (1996) 96–101. G.F. Tutwiler, T. Kirsch, G. Bridi, A pharmacologic profile of McN-3495 [N-(1-methyl-2-pyrrolodinylidene)-N9-phenyl1-pyrrolidinecarboximidamide], a new, orally effective hypoglycemic agent, Diabetes 27 (1978) 856–867.

269

[15] C. Santos et al., Correlation of two bioadhesion assays, in: Proceedings of the 24th International Symposium on Controlled Release of Bioactive Materials, Vol. 24, 1997, pp. 261–262. [16] C.A. Santos et al., Poly(fumaric-co-sebacic anhydride): a degradation study as evaluated by FTIR, DSC, GPC and X-Ray diffraction, J. Control. Rel. 60 (1999) 11–22. [17] G.P. Carino, J.S. Jacob, C. James Chen, Camilla A. Santos, Benjamin A. Hertzog, Edith Mathiowitz, Bioadhesive, Bioerodible Polymers for Increased Intestinal Uptake, in: Edith Mathiowitz, Donald E. Chickering III, Claus-Michael Lehr (Eds.), James Swarbrick (Series Ed.), Bioadhesive Drug Delivery Systems: Fundamentals, Novel Approaches and Development, Drugs and the Pharmaceutical Sciences Series, Marcel Dekker, Inc., New York, 1999, pp. 459–475. [18] T.H. Ermak, E.P. Dougherty, H.R. Bhagat, Z. Kabok, J. Pappo, Uptake and transport of copolymer biodegradable microspheres by Rabbit Peyer’s patch M cells, Cell Tissue Res. 279 (1995) 433–436. [19] P. Jani, G.W. Halbert, J. Langridge, A.T. Florence, The uptake and translocation of latex nanospheres and microspheres after oral administer to rats, J. Pharm. Pharmacol. 41 (1989) 809–812.