Journal of Controlled Release 61 (1999) 267–279
`
`www.elsevier.com/locate /jconrel
`
`Mechanisms controlling diffusion and release of model proteins
`through and from partially esterified hyaluronic acid membranes
`b ,*
`a
`b
`, V.J. Stella , W.N. Charman , S.A. Charman
`L.D. Simon
`aDepartment of Pharmaceutical Chemistry,University of Kansas, Lawrence, KS 66045,USA
`bDepartment of Pharmaceutics,Victorian College of Pharmacy, Monash University,381 Royal Parade, Parkville,Victoria 3052,
`Australia
`
`a,b ,1
`
`Received 9 November 1998; received in revised form 8 April 1999; accepted 28 May 1999
`
`Abstract
`
`The effects of polymer percent esterification and protein molecular weight on the diffusion of two model proteins,
`deoxyribonuclease (DNase) and ribonuclease A (RNase A), through and from partially esterified hyaluronic acid membranes
`were compared. The permeability of the polymer membranes was inversely related to the degree of polymer esterification
`and the molecular weight of the protein. Transport rates of proteins through the membranes decreased dramatically over
`narrow ranges of polymer esterification. As expected, the apparent diffusivity of the larger protein in the polymer matrix was
`more sensitive to changes in membrane hydration than that of the smaller protein. These observations demonstrated the
`dependence of the mobility of large molecular weight proteins on polymer hydration and chain relaxation. The relationship
`between protein diffusion through and release from the modified hyaluronate matrices was also investigated using RNase A
`as a model. The release profiles from fully esterified membranes showed lag behavior and varied with protein load and
`hyaluronate hydrolysis rates, while release from less esterified membranes was rapid and independent of polymer
`esterification or hydrolysis. Potential applications of modified hyaluronate matrices in the controlled delivery of proteins are
`discussed.
`1999 Elsevier Science B.V. All rights reserved.
`
`Keywords: Protein release; Protein permeability; Hyaluronic acid esters; Polymer membranes; Protein load
`
`1. Introduction
`
`Research in the area of controlled protein and
`peptide delivery is continuing to expand due to the
`
`address: susan.charman@vcp.monash.edu.au
`
`*Corresponding author. Tel.: 161-3-9903-9626; fax: 161-3-
`9903-9627.
`E-mail
`Charman)
`1
`Current address: Department of Pharmaceutics Research and
`Development, Bristol-Myers Squibb Company, New Brunswick,
`NJ 08903, USA.
`
`(S.A.
`
`increasing availability of proteins for potential thera-
`peutic use. Of particular interest has been the de-
`velopment and characterization of implantable poly-
`meric matrices for the extended release of proteins
`[1–3]. Such matrices are designed to overcome some
`of the limitations of the conventional
`intravenous
`route of administration by eliminating the need for
`frequent dosing and reducing undesirable side ef-
`fects. Both non-biodegradable and biodegradable
`polymers have been studied for use in implantable
`devices. While non-biodegradable polymers such as
`poly(vinyl alcohols) [4], poly(ethylene-co-vinyl ace-
`
`0168-3659/99/$ – see front matter
`PII: S0168-3659( 99 )00123-6
`
`1999 Elsevier Science B.V. All rights reserved.
`
`ALL 2090
`PROLLENIUM V. ALLERGAN
`IPR2019-01505 et al.
`
`(cid:211)
`(cid:211)
`

`

`268
`
`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`tate) [5], and acrylates [6] are capable of releasing
`protein over prolonged periods, they have the dis-
`advantage of requiring surgical removal following
`depletion of their drug load. Therefore, most effort
`has addressed the development of controlled release
`devices composed of biodegradable polymers such as
`polylactide or poly(lactide-co-glycolide) systems [7–
`9], polyanhydrides [10], and poly(ortho esters) [11].
`Much research in the area of controlled drug
`delivery has focused on the use of polymers to
`achieve constant or zero-order release of therapeutic
`agents over extended periods of time. However, there
`are many applications for which continuous drug
`release may not be appropriate. For example, the
`releasing hormones of gonadotropins and growth
`hormones are most effective when delivered in a
`pulsatile fashion [12,13]. Other proteins, such as
`insulin and thrombolytic agents, may show toxicity
`when administered continuously [6,14]. Controlled
`release systems designed for pulsatile rather than
`sustained, constant release are also potentially valu-
`able in the area of vaccine delivery [15] where
`multiple dose therapy often tends to reduce patient
`compliance. The rate of protein release from poly-
`mer-based delivery systems is generally dependent
`upon the diffusion rate of the protein through the
`polymeric network. Examination of the factors in-
`fluencing the diffusion of macromolecules in hy-
`drated polymers is therefore useful in determining
`the mechanisms of release from polymeric matrices
`in order to select the polymer system most appro-
`priate for the intended application.
`Hyaluronic acid is a naturally occurring polysac-
`charide found throughout the body in various tissues
`including connective tissue,
`the synovial fluid of
`joints and the aqueous humor of
`the eye [16].
`Partially esterified hyaluronic acid is currently being
`evaluated as a potential biodegradable and biocom-
`patible matrix for controlled protein and peptide
`delivery [17–22]. Preliminary studies of protein
`diffusion through hyaluronate membranes have sug-
`gested that these materials may be very well suited
`for use as pulsatile-release matrices for macromole-
`cules [22]. The membrane permeability of RNase A
`has been previously described and was shown to
`occur primarily through the free water domain of the
`polymeric matrix, consistent with the ‘free volume’
`theory of diffusion developed by Yasuda et al. [23].
`
`The diffusion characteristics of the hyaluronates have
`indicated that zero-order protein release profiles
`would be unlikely, but that these polymers could be
`useful for applications requiring combinations of
`immediate and delayed release.
`In the present study, the dependence of hyaluro-
`nate membrane permeation on molecular size of the
`permeant was evaluated by comparing the membrane
`diffusion characteristics of RNase A (molecular
`weight 13.7 kDa), deoxyribonuclease (DNase, mo-
`lecular weight 31 kDa), and thymidine (Sung et al.
`[24], reproduced with permission). To investigate
`release mechanisms,
`the rate of release of mem-
`brane-incorporated RNase A was examined under
`conditions of variable protein load and polymer
`esterification and was compared with the rate of
`protein release from compressed hyaluronate pellets.
`The relationship between protein release and poly-
`mer ester hydrolysis was also determined.
`
`2. Experimental
`
`2.1. Materials
`
`The partial benzyl esters of hyaluronic acid used
`in these studies are depicted in Fig. 1. Fully esterified
`and 77% esterified hyaluronic acid (denoted as HA
`p100 and HA p77, respectively) were supplied by
`Fidia Advanced Biopolymers, S.p.A (Abano Terme,
`
`Fig. 1. Structure of the hyaluronic acid (HA) repeating unit,
`consisting of two sugar subunits: N-acetyl glucosamine and D-
`glucuronic acid. R represents either the benzyl ester or the sodium
`salt, depending on the degree of esterification (esterification range
`of 64 to 100%). The nomenclature HA pX is used to describe
`esterified HA where X represents the degree of esterification.
`
`

`

`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`269
`
`Italy). The HA used by Fidia in the preparation of
`the esterified derivatives was extracted from rooster
`comb and had an average approximate molecular
`weight of 160 kDa. Polymers of intermediate and
`lower percent esterification were prepared by hy-
`drolysis of
`the HA p100 polymer as described
`previously [22]. Bovine pancreatic RNase A (Type
`III-A) and its substrate, cytidine 29:39-phosphate,
`were purchased from Sigma Chemical Company (St.
`Louis, MO, USA) and were used without further
`purification. Recombinant human DNase was gener-
`ously donated by Genentech (San Francisco, CA,
`USA) and was also used without further purification.
`All other chemicals were reagent grade and were
`used as received.
`
`2.2. Validation of DNase stability – exposure to
`physiological temperature/shear stress
`
`The aqueous stability of DNase upon exposure to
`physiological
`temperature (378C) with continuous
`shear stress (solution stirring) was evaluated over
`120 h using size exclusion chromatography (SEC)
`and reducing and non-reducing SDS-PAGE. Details
`of the stability validation experiments have been
`presented elsewhere [22].
`SEC analysis was performed at ambient tempera-
`ture using a Superose 12 HR 10/30 size exclusion
`column (Pharmacia LKB Biotechnology, Uppsala,
`Sweden) with a flow-rate of 1.0 ml/min. Protein was
`detected by UV absorbance at a wavelength of 278
`nm. The mobile phase consisted of 0.1 M phosphate
`buffer (pH 7.0) containing 0.02% sodium azide.
`Under
`these conditions,
`the retention volume of
`DNase was 13.2 ml, and the limit of detection was
`approximately 1063 mg/ml.
`SDS-PAGE was conducted using a FisherBiotech
`
`Table 1
`Components used in the preparation of SDS-PAGE gels
`
`Electrophoresis System (Model FB105; Fisher Sci-
`entific, Pittsburgh, PA, USA) and 12.5% SDS-PAGE
`gels cast from a solution of 30% acrylamide stock,
`buffer, and water in proportions listed in Table 1.
`Gels were polymerized overnight using N,N,N9,N9-
`tetramethylethylenediamine (TEMED) and a 10%
`solution of ammonium persulfate. The running buffer
`consisted of 0.0225 M tris buffer, 0.2 M glycine and
`0.1% SDS. Following electrophoresis, the gels were
`stained overnight with Coomassie blue and then
`destained with a solution of 25% ethanol with 8%
`acetic acid. The molecular weight of DNase was
`estimated by use of a calibration curve of low
`molecular weight (14.4–97.4 kDa) protein standards.
`
`2.3. DNase transport and membrane permeability
`
`Partially esterified hyaluronate membranes of 76%
`esterification were prepared using a solvent extrac-
`tion method, and membranes of 64 and 73% esterifi-
`cation were prepared by a membrane hydrolysis
`method. Dimethylsulfoxide (DMSO) was used as the
`solvent for these membranes, and a full description
`of solvent extraction and membrane hydrolysis ma-
`trix preparation methods were described previously
`[22]. The permeability of DNase through these
`membranes was measured in triplicate using Side-Bi-
`Side glass diffusion cells. Prior to each experiment,
`three membrane disks (1.7 cm diameter) were cut,
`weighed, and the absence of holes confirmed by light
`microscopy (100–4003 magnification). The dry
`thickness of each membrane disk was measured
`using an Ames micrometer (n 5 3), and the disks
`were mounted on the diffusion cells, sealing with a
`thin layer of vacuum grease. A 3 ml solution of 1
`mg/ml DNase in 0.01 M phosphate buffer (pH 7.4,
`m5 0.15, containing 0.02% sodium azide) was
`
`Stacking gel
`
`Component
`
`Stacking gel buffer
`30% acrylamide
`Water
`10% APS
`TEMED
`
`Volume (ml)
`
`500
`350
`1150
`6
`4
`
`Separating gel
`
`Component
`
`Stacking gel buffer
`30% acrylamide
`Water
`10% APS
`TEMED
`
`Volume (ml)
`
`1250
`2084
`835
`18.4
`2.5
`
`(cid:210)
`

`

`270
`
`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`added to the donor compartment of each diffusion
`cell, and 3 ml of buffer without protein was added to
`the receiver compartment. The temperature was
`maintained at 378C throughout all diffusion experi-
`ments using a thermostated circulating water bath,
`and both donor and receiver solutions were continu-
`ously stirred at 600 rpm by a magnetic stirring
`console upon which each diffusion cell was
`mounted. The diffusional surface area of the cells
`2
`was 0.693 cm .
`Samples were periodically removed from the
`receiver solution and replaced with fresh buffer. In
`order to monitor both protein permeation and poly-
`mer benzyl ester hydrolysis rates, each sample was
`analyzed for both DNase and benzyl alcohol by SEC
`as described in the previous section. The retention
`volume of benzyl alcohol using these chromato-
`graphic conditions was 36.1 ml. The detector was
`programmed to change the wavelength from 278 to
`205 nm between the elution of the DNase peak and
`the benzyl alcohol peak during each run. At the end
`of the experiment, the hydrated membrane thickness
`was measured.
`Protein transport data were analyzed assuming that
`initial membrane hydration was rapid compared to
`protein diffusion (confirmed in preliminary studies
`[22]), that transport through the hydrated membranes
`occurred through simple (Fickian) diffusion, and that
`the protein was not degraded during permeation. The
`apparent permeability coefficient (P ) was calcu-
`app
`lated using the initial slope of the cumulative mass of
`protein transported versus time curve and Eq. (1):
`›M/ ›t
`]]
`(A)(C )d
`is the apparent diffusivity of DNase in
`where D
`app
`the hydrated membrane, K is the membrane/ buffer
`partition coefficient, M is the cumulative mass of
`protein transferred to the receiver solution, A is the
`diffusional surface area of the membrane (0.693
`2cm ), C is the protein concentration of the donor
`d
`solution, and h is the hydrated membrane thickness.
`The membrane/buffer partition coefficient of DNase
`into the partially esterified hyaluronate membranes
`was measured at 378C using a solution depletion
`method presented previously [22].
`Membrane ester hydrolysis data were analyzed
`assuming that benzyl alcohol was released only from
`
`P 5 (D )(K) 5
`app
`app
`
`*h
`
`the hydrated polymer within the diffusional surface
`area of the membrane in the diffusion cell. The rate
`of change in membrane percent esterification was
`calculated based on the mass of exposed hydrated
`polymer and the rate of benzyl alcohol release.
`
`2.4. Preparation of RNase A-loaded hyaluronate
`membranes
`
`The solvent extraction and membrane hydrolysis
`methods used to prepare blank hyaluronate mem-
`branes for protein transport studies were not suitable
`for
`the preparation of protein-loaded membranes
`because of the potential for protein loss from the
`matrix during the solvent extraction process. There-
`fore, thin partially esterified hyaluronic acid mono-
`lithic membranes (blank and RNase A-loaded) were
`prepared using a solvent evaporation method in
`which solutions of polymer and protein (protein
`concentration variable depending on the desired final
`protein load) in 90% hexafluoroisopropanol (HFIP)
`in water were cast onto Teflon
`petri dishes and
`slowly air dried. Resulting membranes contained two
`visually distinct regions, a transparent region in the
`membrane center and a translucent region around the
`perimeter. Since the transparent region of HA p100
`membranes prepared by solvent evaporation showed
`similar permeability properties to HA p100 mem-
`branes prepared with the solvent extraction method,
`only the transparent regions of the solvent evapora-
`tion membranes were used in these studies. Further
`details of the membrane preparation method and the
`characteristics of the resulting membranes are de-
`scribed elsewhere [25]. Thicker membranes were
`prepared by casting larger volumes of the polymer /
`protein mixture onto the petri dish. The polymers
`and solvent compositions used to prepare membranes
`containing incorporated RNase A are listed in Table
`2. Also listed are the percent protein load (de-
`termined using a ninhydrin total protein assay [26])
`and dry membrane thickness, as measured by an
`Ames micrometer.
`
`2.5. Preparation of RNase A-loaded compressed
`hyaluronate pellets
`
`Protein-loaded HA p100 pellets, each composed of
`a physical mixture of polymer with either 5% or 30%
`
`(1)
`
`(cid:210)
`

`

`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`271
`
`Table 2
`Characterization of esterified hyaluronic acid membranes (HA
`p100) containing RNase A (mean6SD, n 5 3) prepared by the
`solvent evaporation method using 90% v/v HFIP and 10% v/ v
`water
`
`Theoretical %
`protein load (w /w)
`
`Actual % protein
`load (w/ w)
`
`Unhydrated
`thickness (mm)
`
`5
`10
`10
`20
`30
`50
`60
`
`3.460.4
`8.661.1
`8.960.3
`19.460.5
`28.863.6
`52.962.4
`58.760.8
`
`8568
`7761
`961
`1060
`1662
`1761
`1761
`
`w/w RNase A, were prepared by direct compression
`at 8000 lbs for 1 min using a Model 2512 Carver
`tablet press (Fred S. Carver, Inc., Summit, NJ, USA).
`The pellets were 1.3 cm in diameter, approximately
`0.05 cm thickness, and weighed 80 mg each. Follow-
`ing compression, all pellets were stored in a vacuum
`desiccator at room temperature over CaSO for at
`4
`least 72 h prior to use.
`
`2.6. Release of RNase A from hyaluronate
`membranes
`
`The release rates of RNase A from HA p100 and
`HA p77 membranes were measured in triplicate
`using Side-Bi-Side
`glass diffusion cells as de-
`scribed elsewhere [25]. Samples were periodically
`removed from each compartment,
`replaced with
`fresh buffer, and analyzed for both RNase A and
`benzyl alcohol by SEC. Analysis was performed as
`described in Section 2.2 but using a flow-rate of 0.8
`ml/min. Under these conditions, the retention vol-
`umes of RNase A and benzyl alcohol were 14.5 and
`36.1 ml, respectively. The total mass of protein
`released from each membrane was taken to be the
`sum of the mass released into each of the two cell
`compartments. The percentage released was then
`calculated based on the mass of the membrane, the
`percent protein load (see Table 2), and the assump-
`tion that protein was released only from the hydrated
`membrane within the diffusional surface area of the
`cells.
`The rate of change in membrane percent esterifica-
`tion during the release experiments was calculated as
`outlined elsewhere [22], and the integrity of RNase
`
`A following release from the membranes was con-
`firmed by assay of enzymatic activity [27].
`
`2.7. Release of RNase A from compressed
`hyaluronate pellets
`
`The release of RNase A from compressed hyaluro-
`nate pellets was measured, in triplicate, by submerg-
`ing the pellets in 25 ml of 0.01 M phosphate buffer
`(pH 7.4, m5 0.15, with 0.02% sodium azide) con-
`tained in a sealed bottle. The temperature was
`maintained at 378C, and the solutions were agitated
`using a Model 25 Precision reciprocal shaking water
`bath (Precision Scientific, Chicago, IL, USA) with
`shaker speed of 100 rpm. Samples were periodically
`removed and analyzed for both RNase A and benzyl
`alcohol by SEC (see Section 2.6). The percentage of
`protein released from each pellet was calculated
`based on the mass of the pellet and the percent
`protein load. The protein load was obtained by SEC
`upon complete dissolution of the pellet. The rate of
`change in polymer percent esterification during
`protein release was calculated based on the weight,
`protein load, and initial polymer percent esterifica-
`tion of each pellet.
`
`3. Results
`
`3.1. Validation of DNase stability – exposure to
`physiological temperature/shear stress
`
`No significant changes in the DNase peak area or
`peak shape were observed in the SEC chromato-
`grams following 120 h of incubation in solution at
`378C under conditions of continuous solution stirring
`(data not shown). These results indicated that the
`apparent molecular weight of DNase remained un-
`changed during the study. Reducing and non-reduc-
`ing SDS-PAGE confirmed that the protein did not
`undergo apparent peptide bond hydrolysis or co-
`valent cross-linking. On the basis of these results, all
`conclusions from the permeability studies were based
`on the assumption that any undetected chemical
`alterations (e.g., deamidation, etc.) did not affect the
`permeability characteristics of DNase over the course
`of the experiments.
`
`(cid:210)
`

`

`272
`
`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`p90 membranes were too low to measure. Size
`exclusion analysis indicated that DNase permeated
`the polymer matrices intact with no detectable
`alterations in protein size.
`The apparent permeability coefficients (P ) ob-
`app
`tained from the transport data for DNase are listed in
`Table 3 along with the values determined previously
`for RNase A. Also listed are the apparent diffusivity
`coefficients (D ), the percent membrane hydration,
`app
`and the membrane/buffer partition coefficient (K),
`where D
`is defined as the ratio of P
`to K. The
`app
`app
`value of P
`reported for DNase in HA p76 mem-
`app
`branes and for RNase A in HA p90 membranes was
`based on the limit of detection of the SEC assay. The
`apparent permeability and apparent diffusivity of
`both proteins increased as initial membrane percent
`esterification decreased.
`Fig. 3 illustrates the relationship between the
`logarithm of the apparent permeabilities of DNase
`and RNase A and percent membrane esterification.
`The graph indicates that the diffusion of the larger
`protein, DNase, through the hyaluronate membranes
`was one to two orders of magnitude slower than the
`diffusion of the smaller protein, RNase A. Both
`proteins
`showed abrupt decreases
`in membrane
`permeability over relatively narrow ranges of poly-
`mer esterification.
`The gradual decrease in membrane percent esteri-
`fication due to hydrolysis of benzyl ester bonds
`within the hydrated polymer matrix was monitored
`throughout each permeability experiment and the
`initial hydrolysis rates were not significantly in-
`fluenced by the percent esterification of the polymer
`
`Fig. 2. Cumulative percent DNase transported across HA p76
`(m), HA p73 (s) and HA p64 (^) membranes. Filled symbols
`designate membranes prepared by the solvent extraction method
`and open symbols represent membranes prepared by the mem-
`brane hydrolysis method. Error bars represent the standard devia-
`tion of the mean for n 5 3 measurements.
`
`3.2. DNase transport and membrane permeability
`
`rate of DNase across partially
`The transport
`esterified hyaluronate membranes was determined
`and the results are shown in Fig. 2. DNase diffusion
`through HA p76 membranes was significantly slower
`than that through HA p73 and HA p64 membranes.
`The transport rates of DNase through HA p84 and
`
`Table 3
`Transport parameters for DNase and RNase A through partially esterified hyaluronic acid membranes (mean6SD, n 5 3)
`
`Membrane
`
`a
`
`a
`
`a
`
`b
`
`b
`
`p90
`p84
`p76
`p73
`p64
`
`% increase
`thickness
`
`(% H)
`
`50
`160
`290
`360
`640
`
`2
`Diffusivity (cm /s)
`8
`(D 310 )
`app
`
`Partition
`coefficient (K)
`
`2
`Permeability (cm /s)
`7
`(P 310 )
`app
`
`RNase A
`
`c
`
`,0.02
`1.660.1
`3.660.4
`4.161.4
`10.264.2
`
`DNase
`
`RNase A
`
`DNase
`
`RNase A
`
`DNase
`
`d
`n.d.
`n.d.
`,0.01
`0.0660.02
`0.1960.06
`
`c
`
`7.460.8
`5.260.3
`4.560.5
`4.661.3
`4.061.7
`
`n.d.
`n.d.
`34.260.6
`29.260.3
`11.661.8
`
`,0.01
`0.860.03
`1.660.1
`1.960.4
`4.060.2
`
`n.d.
`n.d.
`,0.03
`0.1860.05
`0.2260.06
`
`c
`
`a Prepared using the solvent extraction method [22].
`b Prepared using the membrane hydrolysis method [22].
`c Below the limit of quantitation.
`d Not determined.
`
`

`

`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`273
`
`Fig. 3. Effect of polymer percent esterification on the apparent
`permeability of RNase A (d,s) and DNase (j,h) through
`partially esterified hyaluronic acid membranes. Filled symbols
`designate membranes prepared by the solvent extraction method
`and open symbols represent membranes prepared by the mem-
`brane hydrolysis method. Error bars represent the standard devia-
`tion of the mean for n 5 3 measurements.
`
`Fig. 4. Effect of solute molecular weight on the relationship
`between the natural logarithm of diffusivity and reciprocal hydra-
`tion (i.e. adherence of diffusion behavior to free volume theory)
`for thymidine (mw 240) (j, data from Ref. [24]), RNase A (mw
`13,700) (d), and DNase (mw 31,000) (m). Error bars represent
`the standard deviation of the mean for n 5 3 measurements.
`
`studied. Preliminary experiments
`range
`in the
`showed that DNase did not affect the ester hydrolysis
`rate of the polymer (unpublished data).
`As illustrated in Fig. 4 the diffusion behavior of
`DNase across hyaluronate membranes of various
`percent esterification was tested against ‘free vol-
`ume’ theory developed by Yasuda et al. [23] accord-
`ing to Eq. (2):
`
`were linear, suggesting that the diffusion behavior of
`each of
`these solutes in the hydrated partially
`esterified hyaluronate membranes was consistent
`with free volume theory. The slope of the free
`volume plot for thymidine was relatively shallow,
`but the relationship between ln D and 1/H became
`app
`increasingly more steep with increasing molecular
`weight of the permeant.
`
`*
`ln D 5 ln D 2 Y (1/H 2 1)
`0
`
`(2)
`
`3.3. Release of RNase A from hyaluronate
`membranes and compressed hyaluronate pellets
`
`where D is the diffusivity of the solute in the
`hydrated polymer matrix, D is the diffusivity of the
`0
`*
`solute in the solvent, Y is a constant characteristic
`of the solute and the solvent, and H is the polymer
`membrane hydration. The results for DNase were
`compared with those obtained for RNase A [22] and
`for the small hydrophilic molecule, thymidine (Sung
`et al. [24], reproduced with permission). Membranes
`used for thymidine studies were prepared using a
`solvent evaporation method essentially the same as
`that described in Section 2.4. The plots of ln D
`app
`versus 1/H for DNase, RNase A, and thymidine
`
`The effect of membrane percent protein load on
`the release of RNase A from HA p100 membranes is
`presented in Fig. 5. Membranes with lower protein
`load (5 to 30% RNase A) showed an initial lag phase
`followed by a release phase during which complete
`release was obtained. Membranes with higher protein
`load (50 to 60% RNase A) showed burst effects,
`releasing approximately 10 to 20% of their protein
`loads within the first 30 min. In the case of the 50%
`load, this initial burst was followed by a lag time and
`a secondary release phase during which the remain-
`
`

`

`274
`
`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`Fig. 5. Effect of percent protein load and membrane thickness on
`the release of RNase A from HA p100 membranes containing 5%
`(90 mm) ('), 10% (85 mm) (^), 10% (10 mm) (m), 20% (10 mm)
`(h), 30% (20 mm) (j), 50% (20 mm) (\), and 60% (20 mm)
`(.) RNAse A. Error bars represent the standard deviation for
`n 5 3 measurements.
`
`Fig. 6. Effect of percent protein load and membrane thickness on
`polymer ester hydrolysis rates of HA p100 membranes for
`membranes containing 5% (90 mm) ('), 10% (85 mm) (n), 10%
`(10 mm) (m), 20% (10 mm) (h), 30% (20 mm) (j), 50% (20 mm)
`(,), and 60% (20 mm) (.) RNase A. Error bars represent the
`standard deviation for n 5 3 measurements.
`
`ing protein load was released. While the membrane
`loaded with 60% RNase did not show a lag phase,
`the initial burst was followed by a slower secondary
`release phase, and complete release of the protein
`from the membranes was achieved.
`The rate of change in membrane percent esterifica-
`tion for the protein-loaded HA p100 membranes of
`varying thickness is plotted in Fig. 6. The average
`hydrolysis rate increased with decreasing membrane
`thickness, possibly indicating a surface, rather than
`bulk, hydrolysis reaction which is dependent on the
`extent of membrane hydration. Studies conducted by
`others have shown that the rate of benzyl alcohol
`permeation through the matrix is rapid relative to the
`rate of hydrolysis and that the partition coefficient of
`benzyl alcohol in HA p100 is low [28]. Preliminary
`studies have also demonstrated that the presence of
`the protein in the polymeric matrix did not influence
`the rate of polymer ester hydrolysis. For membranes
`showing lag behavior, neither the length of the lag
`phase nor the rate of protein release appeared to be
`related to the percent protein load, but both were
`related to the rate of membrane ester hydrolysis (see
`
`therefore,
`Figs. 7 and 8, respectively) and were,
`correlated to the polymer membrane thickness. Re-
`gardless of protein load or membrane thickness, the
`length of the lag time corresponded to the time
`required for fully esterified hyaluronic acid to hydro-
`lyze to approximately 80% esterification (Fig. 9). In
`the case of
`the 50% nominal protein load,
`the
`beginning of the secondary protein release phase
`corresponded to 80% polymer esterification. These
`observations indicated that the molecularly dispersed
`protein remained entrapped within the matrix until
`the hyaluronate ester hydrolyzed to the critical point
`of approximately 80% esterification, after which the
`membrane became sufficiently hydrated to allow
`protein diffusion and release. Average lag times
`increased with increasing membrane thickness, con-
`sistent with the decreased hydrolysis rate of thicker
`membranes. The relationship between lag time and
`membrane thickness was non-linear, and the values
`for lag time appeared to approach a plateau of 350 to
`400 h for membrane thickness near 100 mm.
`For all membranes examined, the release rates of
`protein into the two half-cell compartments of the
`
`

`

`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`275
`
`Fig. 7. Relationship between the lag time observed prior to
`release of RNase A from HA p100 membranes containing low
`protein load and polymer ester hydrolysis rate. Error bars repre-
`sent the standard deviation for n 5 3 measurements.
`
`Fig. 8. Relationship between the rate of RNase A release from
`HA p100 membranes and polymer ester hydrolysis rate. Error bars
`represent the standard deviation for n 5 3 measurements.
`
`Fig. 9. Plot of percent protein released from HA p100 membranes
`of varying thickness and containing varying protein load as a
`function of polymer percent esterification for membranes con-
`taining 5% (90 mm) ('), 10% (85 mm) (n), 10% (10 mm) (m),
`20% (10 mm) (h), 30% (20 mm) (j), and 50% (20 mm) (,)
`RNase A. Error bars represent the standard deviation for n 5 3
`measurements.
`
`the
`indicating that
`diffusion cells were similar,
`protein was not concentrated at either membrane
`surface during preparation. All protein released
`retained full enzymatic activity, indicating that the
`structural integrity of the active site of the protein
`was maintained following release. Protein release
`from compressed hyaluronate pellets
`(data not
`shown) was immediate, complete and was indepen-
`dent of both protein load and polymer percent
`esterification, indicating that the compressed matrix
`did not present a significant barrier
`to protein
`diffusion.
`
`4. Discussion
`
`The apparent permeability and diffusivity data
`described here for DNase and reported previously for
`RNase A [22] suggest that the diffusivity of proteins
`through esterified hyaluronic acid membranes is
`largely controlled by the decreased hydrophobicity
`and increased hydration of the hyaluronate polymers
`
`

`

`276
`
`L.D. Simon et al. / Journal of Controlled Release 61(1999)267–279
`
`with lower percent esterification. The diffusion of
`large molecules through the matrix would be ex-
`pected to become more rapid upon sufficient mem-
`brane hydration and polymer chain relaxation. Both
`proteins
`showed abrupt decreases
`in membrane
`permeability over relatively narrow ranges of poly-
`mer esterification (Fig. 3). However,
`relative to
`RNase A,
`the decrease in permeability of DNase
`occurred at a lower percent membrane esterification,
`indicating that a greater degree of membrane hydra-
`tion and relaxation was required for significant
`diffusion of the larger protein through the polymer
`network.
`The diffusion characteristics of DNase further
`extend and support the previous findings [19,22,24]
`that diffusion through partially esterified, hydrated
`hyaluronate membranes is consistent with Yasuda’s
`free volume theory [23]. The apparent relationship
`between molecular weight and the slope of the free
`volume plot of ln D
`versus 1/H (Fig. 4) further
`app
`demonstrates the dependence of the diffusion of
`larger molecules on the degree of polymer hydration
`and chain relaxation. However, from the large values
`of the membrane/buffer partition coefficients for
`DNase, and the increase in K with increasing mem-
`brane percent esterification (see Table 3), a second
`mechanism involving hydrophobic interaction be-
`tween the polymer and the protein may have also
`contributed to the transport characteristics of DNase
`through the hydrated membranes. The values of the
`membrane/buffer partition coefficients for RNase A
`determined in previous experiments were lower and
`did not indicate significant protein interaction with
`the polymer [22].
`The characteristics of RNase A and DNase diffu-
`sion through partially esterified hyaluronate mem-
`branes may be useful
`in predicting the release
`behavior of macromolecules from these systems.
`Based on the steep changes in protein membrane
`permeability over narrow ranges of polymer esterifi-
`cation, a simple, zero-order or square-root-of-time
`protein release profile would not be expected, par-
`ticularly in more highly esterified polymers. Rather,
`initial release from highly esterified hyaluronates
`would more likely be slow to negligible, while fast
`release, or ‘burst’ behavior, would be predicted from
`less esterified, more hydrophilic hyaluronates. Fur-
`thermore, due to the gradual decrease in polymer
`
`esterification over time in an aqueous environment,
`protein release from fully esterified matrices would
`be expected to proceed more rapidly upo