Synthesis and structure-property evaluation of cellulose ω-carboxyesters for amorphous solid dispersions

Article abstract of DOI:10.1016/j.carbpol.2012.11.049

The use of amorphous solid dispersions (ASDs) is an effective and increasingly widely used approach for solubility enhancement of drugs and drug candidates with poor aqueous solubility. Successful molecular dispersion of drugs in polymer matrices requires new polymers that are designed to meet all ASD requirements, including drug release and prevention of drug recrystallization in storage or from solution. We describe herein design and synthesis of a new series of cellulose ω-carboxyalkanoates for ASDs, by reaction of cellulose with long-chain diacids that have been monoprotected as benzyl esters at one end, and monoactivated as acid chlorides at the other. Glass transition temperatures (T_g) of these cellulose ω-carboxyesters exceed ambient temperature by at least 50 C, providing a sufficient ΔT to prevent drug mobility and crystallization. Cellulose acetate suberates and sebacates prepared in this way are extraordinary solution crystal growth inhibitors for the poorly soluble anti-HIV drug ritonavir. These new cellulose ω-carboxyesters have strong potential as ASD polymers for enhancement of drug solubility and bioavailability.

Full text of DOI:10.1016/j.carbpol.2012.11.049

Carbohydrate Polymers 100 (2014) 116–125
Contents lists available at ScienceDirect
Carbohydrate Polymers
journal homepage: www.elsevier.com/locate/carbpol
Synthesis and structure–property evaluation of cellulose ␻-carboxyesters for
amorphous solid dispersions
Haoyu Liua,b, Grace A. Ilevbarec, Benjamin P. Cherniawskia, Earl T. Ritchiea, Lynne S. Taylorc,
Kevin J. Edgara,b,∗
a Macromolecules and Interfaces Institute, Virginia Tech, Blacksburg, VA 24061, United States
b Department of Sustainable Biomaterials, College of Natural Resources and Environment, Virginia Tech, Blacksburg, VA 24061, United States
c Department of Industrial and Physical Pharmacy, College of Pharmacy, Purdue University, West Lafayette, IN 47907, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The use of amorphous solid dispersions (ASDs) is an effective and increasingly widely used approach for
solubility enhancement of drugs and drug candidates with poor aqueous solubility. Successful molec-
ular dispersion of drugs in polymer matrices requires new polymers that are designed to meet all ASD
requirements, including drug release and prevention of drug recrystallization in storage or from solu-
tion. We describe herein design and synthesis of a new series of cellulose ␻-carboxyalkanoates for ASDs,
by reaction of cellulose with long-chain diacids that have been monoprotected as benzyl esters at one
end, and monoactivated as acid chlorides at the other. Glass transition temperatures (Tg) of these cellu-
lose ␻-carboxyesters exceed ambient temperature by at least 50 ◦C, providing a sufficient ꢀT to prevent
drug mobility and crystallization. Cellulose acetate suberates and sebacates prepared in this way are
extraordinary solution crystal growth inhibitors for the poorly soluble anti-HIV drug ritonavir. These new
cellulose ␻-carboxyesters have strong potential as ASD polymers for enhancement of drug solubility and
bioavailability.
Received 15 August 2012
Received in revised form
14 November 2012
Accepted 18 November 2012
Available online 27 November 2012
Keywords:
Cellulose ␻-carboxyalkanoates
Crystal growth inhibition
Amorphous solid dispersion
Oral drug delivery
Cellulose esters
Cellulose acylation
© 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Poor drug solubility, and the low bioavailability that often
results, is a particular problem in modern oral drug therapy. The
Cellulose esters have
a
long history of effective use in
process of high throughput screening of drug candidates against
often hydrophobic receptor binding sites, with selection based on
magnitude of binding constant, leads inevitably to a plethora of
hydrophobic drug candidates. The need for purity of drug candi-
dates leads drug developers to favor highly crystalline molecules.
These two properties, hydrophobicity and high crystallinity, are
inimical to water solubility. As a result, a high proportion of new
and existing drugs and drug candidates has poor aqueous solubility
(estimated as high as 40–70%) (Hauss, 2007).
In recent years, amorphous solid dispersion has emerged as
an effective approach to enhancing drug solubility (Alonzo et al.,
2011; Konno, Handa, Alonzo, & Taylor, 2008; Newman, Knipp, &
Zografi, 2012). To create an ASD of a particular drug, the drug is
dispersed, ideally molecularly, in a polymer matrix. The creation of
a miscible drug-polymer blend leads to a high-energy, amorphous
state of the drug, where the polymer inhibits drug crystallization.
This enhances drug solution concentration by reducing the kinetic
energy barrier that must be surmounted for the drug to dissolve.
The criteria for the polymer may be summarized as follows:
pharmaceutical applications, based on their low toxicity, water per-
meability, essentially complete lack of ability to permeate through
the enterocytes into the circulation, and the possibility of tailoring
the cellulose ester structure to impart, for example, pH responsive-
ness (Edgar et al., 2001). More recently, cellulose esters have been
explored as key components of advanced drug delivery systems
addressing difficult issues like poor drug bioavailability (Friesen
et al., 2008) and nanoparticulate therapy (Mawad et al., 2009). Cel-
lulose esters are more useful in oral drug administration than in
modes like intravenous or inhalation therapy, since humans lack
the ability to metabolize and clear cellulose from the circulation.
Oral drug administration is generally preferred by patients, due
to its unsurpassed convenience, potential for outpatient therapy,
and relatively low cost, but the ability to administer drugs orally
is currently limited by physical characteristics of drugs and drug
candidates.
∗
Corresponding author at: Department of Sustainable Biomaterials, College of
•
Polymer and any decomposition products must be non-toxic.
Polymer must be miscible with the active drug, and prevent
drug crystallization during transport and storage, even in hot,
Natural Resources and Environment, Virginia Tech, Blacksburg, VA 24061, United
States. Tel.: +1 540 231 0674; fax: +1 540 231 8176.
•
E-mail address: kjedgar@vt.edu (K.J. Edgar).
0144-8617/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.carbpol.2012.11.049

H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
117
humid ambient environments, for at least several years (Hancock,
Shamblin, & Zografi, 1995).
our ability to enhance drug solubility and bioavailability by ASD
(making it effective for an even broader range of drug structures),
and to enhance the stability of the drug in the ASD, permitting
longer term stabilization of maximum solution concentrations
attainable from amorphous drug (increasing bioavailability). This
paper describes the development of methods for the synthesis
and characterization of soluble, non-crosslinked cellulose esters
containing suberate or sebacate groups, as well as the results of
preliminary structure–property investigations that demonstrate
the great promise of these new polymers for amorphous solid
dispersions of poorly soluble drugs.
•
•
Must prevent or severely retard drug crystallization after release
but prior to absorption, in aqueous solution (Alonzo, Zhang, Zhou,
Gao, & Taylor, 2010; Ilevbare, Liu, Edgar, & Taylor, 2012b).
Must release the drug effectively in the absorptive zone of the
digestive tract (Edgar et al., 2001; Posey-Dowty et al., 2007).
Polymers used for ASD to date have largely been selected
from those already in approved pharmaceutical formulations,
rather than having been designed for ASD requirements. Some
of these polymers can themselves crystallize during storage
(poly(ethylene glycol)), so are not suitable from a stability per-
spective. Poly(vinylpyrrolidinone) (PVP) is an excellent polymer for
ASD of drugs, but is highly soluble in water, especially at low (gas-
tric) pH; thus overly fast release or undesired exposure of stomach
to drug can be problems. Hydroxypropyl methyl cellulose acetate
succinate (HPMCAS) is also quite effective in ASD of drugs. How-
ever HPMCAS is complex to synthesize and analyze; amounts of
four separate substituents must be controlled, and the hydrox-
ypropyl substituent carries the additional complication that it can
form oligomeric poly(hydroxypropyl) side chains of various lengths
from the cellulosic hydroxyl initiator. The overall hypothesis in the
collaborative ASD work between the Taylor and Edgar groups is
that, by understanding the fundamental science and specific per-
formance requirements for ASDs and ASD polymers, we can design,
synthesize and use polymers with superior performance in ASD
systems that meet all system requirements.
In previous work, we have reported that cellulose esters con-
taining adipate groups have useful combinations of properties as
ASD polymers (Kar, Liu, & Edgar, 2011; Liu, Kar, & Edgar, 2012). The
tetramethylene side chains of the adipate groups accentuate the
already hydrophobic nature of the cellulose esters, enhancing mis-
cibility with hydrophobic drugs and enabling slow drug release.
The carboxyl end groups not only provide specific interactions
with hydrogen bond acceptors on the drug, enhancing miscibil-
ity and stability, but also provide a mechanism for drug release
through ionization and subsequent swelling upon reaching the
neutral environment of the small intestine. In our previous studies,
cellulose acetate adipate propionate (CAAdP) was found to be an
effective inhibitor of crystal growth of the important, hydropho-
bic drug ritonavir at supersaturations lower than that generated
during the dissolution of the amorphous form (Ilevbare, Liu, Edgar,
& Taylor, 2012a; Ilevbare et al., 2012b); ritonavir is an important
metabolic inhibitor that is a key component of several drug for-
mulations used in human immunodeficiency virus (HIV) therapy
(Cameron et al., 1999) and is administered as the amorphous form.
These structure–property studies showed that both the hydropho-
bicity of the polymer and its carboxyl group content were important
contributors predicting strong inhibition of ritonavir crystalliza-
tion, and that crystal growth inhibition was highly sensitive to
these parameters; indeed, the closely related cellulose acetate adi-
pate butyrate (CAAdB) was ineffective in inhibiting ritonavir crystal
growth. In related research in our groups (Li, Konecke, Wegiel,
Taylor, & Edgar, 2012), amorphous solid dispersion in CAAdP matri-
ces has been shown not only to strongly inhibit crystallization
from either solid dispersion or aqueous solution of the biologi-
cally potent but poorly soluble flavonoid curcumin, but remarkably
also protected curcumin against the rapid chemical degradation
(retro-aldol reaction) that occurs in the absence of polymer.
The hypothesis of the current study is that further exploration
of this ASD structure–property relationship, through synthesis
of novel cellulose ␻-carboxyalkanoates bearing longer, more
hydrophobic tethers between the cellulose main chain and the
pendent carboxyl group, will lead to even more effective ASD
polymers. In this way we hoped both to expand the breadth of
2. Experimental
2.1. Materials
Microcrystalline cellulose (MCC, Avicel® PH-101, Fluka) was
dried under reduced pressure at 50 ◦C overnight prior to use. Cel-
lulose acetate propionate (CAP-504-0.2), cellulose acetate butyrate
(CAB-553-0.4) and cellulose acetate (CA 320S) were obtained from
Eastman Chemical Co. and dried overnight under vacuum at 40 ◦C
prior to use. N,N-dimethylacetamide (DMAc) and 1,3-dimethyl-2-
imidazolidinone (DMI) were purchased from ACROS Organics and
˚
dried over 4 A molecular sieves. Other purchased reagents were
used as received. Suberic acid, sebacic acid, methyl ethyl ketone
(MEK), p-toluenesulfonic acid (PTSA), triethylamine (Et3N), oxa-
lyl chloride and anhydrous tetrahydrofuran were purchased from
ACROS Organics. Toluene, benzyl alcohol, N,N-dimethylformamide
(DMF), dichloromethane, lithium chloride and sodium bicarbon-
ate were purchased from Fisher Scientific. Hydrogenolysis catalyst
20% Pd(OH)/C was obtained from Sigma–Aldrich. Ritonavir was
purchased from Attix Corporation, Toronto, Ontario, Canada.
2.2. Measurements
1H NMR spectra were acquired on INOVA 400 or Bruker Avance
500 spectrometers. Samples were analyzed as solutions in CDCl3
or DMSO-d6 (ca. 10 mg/mL) at 25 ◦C in standard 5 mm o.d. tubes.
A drop of trifluoroacetic acid was added to shift the water peak of
DMSO-d6 downfield from the spectral region of interest. 32 scans
were obtained for each sample. 13C NMR spectra were obtained on
a Bruker Avance 500 MHz spectrometer with a minimum of 5000
scans in DMSO-d6 (ca. 50 mg/mL) at 80 ◦C. DSC analyses of cellulosic
polymers were performed on a TA Instruments Q2000 apparatus.
Dry powders (ca. 5 mg) were loaded in TzeroTM aluminum pans.
Each sample was equilibrated at 30 ◦C and then heated to 180 ◦C
at 15 ◦C/min. The sample then was cooled at 50 ◦C/min to −50 ◦C.
During the second heating cycle the sample was heated to 180 ◦C
at 15 ◦C/min. All DSC measurements were verified with a dupli-
cate run. Tg values were recorded as the step-change inflection
point from 2nd heat scans. IR spectra were obtained on a Nicolet
8700 instrument. Size exclusion chromatography (SEC) was per-
formed in HPLC grade THF at 40 ◦C at flow rate 1 mL/min using a
Waters size exclusion chromatograph equipped with an autosam-
pler, three in-line 5 ␮m PLgel Mixed-C columns, and a Waters 410
refractive index (RI) detector operating at 880 nm, which was pro-
grammed to a polystyrene calibration curve. We report number
average molecular weights relative to polystyrene standards. Cel-
lulose ester solubility was tested by adding ca. 10 mg of sample into
2 mL each of various solvents. Each mixture was subjected to vor-
tex mixing for 5–10 min at room temperature, then solubility was
judged by visual examination.
Thermal analysis of the solid dispersion was carried out through
the use of a modulated differential scanning calorimeter (Model
Q2000, TA Instruments). The calorimeter was calibrated with sap-
phire and indium standards under N2. For each measurement a

118
H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
sample of approximately 2–5 mg was heated to 180 ◦C at 20 ◦C/min
to remove moisture. Then the sample was cooled to −78 ◦C at
50 ◦C/min, followed by the second heating scan to 180 ◦C at
20 ◦C/min. From this second scan the Tg was measured and the
lack of crystallization or melting transitions was confirmed. DSC
heating curves were analyzed using Universal Analysis 2000 soft-
ware (TA instruments). Powder X-ray diffraction (PXRD) data were
collected using a Bruker D8 Discover X-ray Diffractometer. Cu K␣
(ꢁ = 0.154 nm) radiation was at 40 kV/40 mA and the diffraction
profile was detected using a locked couple 2 theta scan from 10
to 50◦ scattering angle. The scan speed was 0.1 s/step with the
increment of 0.01◦. XRD samples were prepared by flattening the
lyophilized solid dispersions into powder form onto a glass slide.
under reduced pressure. The product was stored and used in this
crude form without further purification. Yield: 90.1% (0.046 mol,
14.35 g). 1H NMR (CDCl3): 1.30 (m, 8H), 1.67 (m, 4H), 2.35 (t, 2H),
2.86 (t, 2H), 5.11 (s, 2H), 7.35 (m, 5H).
A similar procedure was followed to synthesize monoben-
zyl suberoyl chloride. Yield: 90.7% (0.051 mol, 14.54 g). 1H NMR
(CDCl3): 1.34 (m, 4H), 1.66 (m, 4H), 2.36 (t, 2H), 2.86 (t, 2H), 5.12
(s, 2H), 7.35 (m, 5H).
Preparation of benzyl cellulose acetate propionate suber-
ate/sebacate (Scheme 1). CAP-504-0.2 (1.00 g, 3.56 mmol) was
dissolved in MEK (20 mL), and the solution was heated to 60 ◦C
with stirring under nitrogen. Triethylamine (0.54 mL, 3.92 mmol,
1.1 equiv.) was added to the solution. Monobenzyl sebacoyl chlo-
ride (1.11 g, 3.56 mmol, 1 equiv.) or monobenzyl suberoyl chloride
(1.02 g, 3.56 mmol, 1 equiv.) was added dropwise and allowed to
react at 60 ◦C for 20 h. The reaction mixture was diluted with
∼10 mL of acetone and filtered to remove in situ formed tri-
ethylamine hydrochloride precipitate. The remaining solution was
precipitated into water (250 mL). The precipitate was collected
and washed with isopropanol. The filtered product was redis-
solved in a minimal amount of chloroform and reprecipitated in
hexanes.
2.3. Methods
Synthesis of monobenzyl sebacate (Supplementary material,
Scheme S1). The method of English, Girard, Jasys, Martingano,
and Kellogg (1990) was adapted for the synthesis of monobenzyl
sebacate. Sebacic acid (0.25 mol, 50.56 g), toluene (200 mL), benzyl
alcohol (0.30 mol, 1.2 equiv., 32.4 g, 31.1 mL), and p-toluene sulfonic
acid (2.5 mmol, 0.475 g) were combined in a Dean–Stark apparatus
and refluxed until the desired volume of H2O (0.30 mol, 5.40 mL)
was collected from the Fischer esterification of the sebacic acid,
indicating the end of the reaction. The mixture was allowed to cool
to room temperature, then 150 mL of DI water was added to the
reaction. Using vigorous mixing, this mixture was adjusted to pH 9
with 6 M NaOH. The aqueous layer with the di-acid and mono-ester
was separated and washed twice with 50 mL of diethyl ether; these
diethyl ether washes contained primarily the dibenzyl ester by-
product, and were discarded. Diethyl ether (200 mL) was combined
with the aqueous layer which was acidified to pH of 2.0–2.5 with
6 M HCl. The ether layer containing the mono-ester was separated.
To purify further, the ether layer was washed with 1 M NaHCO3.
The ether layer then was concentrated under reduced pressure and
vacuum-dried for 1.5 h to yield the final product as white, needle-
like solid. Yield: 29.0% (16.96 g, 0.058 mol). 1H NMR (CDCl3): 1.29
(m, 8H), 1.62 (m, 4H), 2.33 (m, 4H), 5.11 (s, 2H), 7.35 (m, 5H).
A similar procedure was followed to synthesize monobenzyl
suberate (colorless oil). Yield: 47.5% (31.35 g, 0.12 mol). 1H NMR
(CDCl3): 1.33 (m, 4H), 1.63 (m, 4H), 2.33 (m, 4H), 5.11 (s, 2H), 7.34
(m, 5H).
Benzyl CAP sebacate yield. 89.1% (3.17 mmol, 1.10 g). 1H NMR
(CDCl3): 0.95–1.20 (COCH2CH3 of propionate), 1.30 (COCH2CH2CH2
CH2CH2CH2CH2CH2CO of sebacate), 1.63 (COCH2CH2CH2CH2CH2
CH2CH2CH2CO of sebacate), 2.13–2.44 (COCH2CH2CH2CH2CH2CH2
CH2CH2CO of sebacate, COCH2CH3 of propionate, and COCH3 of
acetate), 3.00–5.20 (cellulose backbone), 5.11 (CH2C6H5), 7.35
(CH2C6H5).
Benzyl CAP suberate yield. 78.4% (2.79 mmol, 0.97 g). 1H NMR
(CDCl3): 0.95–1.22 (COCH2CH3 of propionate), 1.33 (COCH2CH2
CH2CH2CH2CH2CO of suberate), 1.63 (COCH2CH2CH2CH2CH2CH2
CO of suberate), 2.10–2.46 (COCH2CH2CH2CH2CH2CH2CO of
suberate, COCH2CH3 of propionate, and COCH3 of acetate),
3.00–5.20 (cellulose backbone), 5.10 (s, CH2C6H5), 7.35 (CH2
C6H5).
A similar procedure was followed for the reaction of cellulose
acetate butyrate (CAB-553-0.4, 1.00 g, 3.26 mmol) with monoben-
zyl suberoyl/sebacoyl chloride (1 equiv.).
Preparation of benzyl cellulose acetate suberate/sebacate. CA 320S
(1.00 g, 4.19 mmol) was dissolved in DMI (20 mL), and the solu-
tion heated to 90 ◦C with stirring under N2. Triethylamine (1.29 mL,
9.22 mmol, 2.2 equiv.) was added; a condenser was used to avoid
evaporative loss of the base catalyst. Monobenzyl sebacoyl chlo-
ride (2.61 g, 8.38 mmol, 2 equiv.) or monobenzyl suberoyl chloride
(2.41 g, 8.38 mmol, 2 equiv.) was added dropwise and allowed to
react at 90 ◦C for 20 h. The reaction mixture was then filtered, and
the remaining solution was diluted with ∼10 mL of acetone. The
solution was precipitated in 300 mL of water. The precipitate was
then collected and washed with isopropanol. The filtered product
was redissolved in a minimal amount of THF and reprecipitated in
hexanes.
Synthesis of monobenzyl sebacoyl chloride (Supplementary
material, Scheme S1). The procedure was adapted from Abell,
Morris, and Litten (1990), as follows. Monobenzyl sebacate
(0.051 mol, 15.00 g) was dissolved in dichloromethane (300 mL),
then DMF (5 drops) was added. The solution was cooled in an
ice bath to 0 ◦C. Oxalyl chloride (24.42 mL, 5.6 equiv.) was added
slowly. The ice bath was removed and the reaction proceeded
at room temperature. A halt in gas evolution indicated comple-
tion, then solvent was removed under reduced pressure. Toluene
(50 mL) was added and the product concentrated azeotropically
Scheme 1. General two-step synthetic method for ␻-carboxyalkanoate derivatives of cellulose.

H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
119
Reaction of cellulose in DMAc/LiCl solution with monobenzyl seba-
coyl chloride. Microcrystalline cellulose (1.00 g, 6.17 mmol) was
dissolved in DMAc (37.5 mL) and LiCl (1.88 g) by a procedure
reported earlier (Edgar, Arnold, Blount, Lawniczak, & Lowman,
1995). The clear solution was heated to 60 ◦C and triethylamine
(1.1 equiv. or 4 equiv.) was added. Monobenzyl suberoyl chloride
(1 equiv. or 3 equiv.) was added dropwise and allowed to react
for 20 h. For 1 equiv. reaction, the mixture was precipitated in
300 mL of DI water with slow addition, filtered, and the precipitate
washed twice with 50 ◦C ethanol, then vacuum-dried. Yield: 78.6%
(4.85 mmol, 1.52 g). For the 3 equiv. reaction, after precipitating in
DI water, the aggregated product was ground using a mortar and
pestle, Soxhlet-extracted with methanol overnight, and vacuum-
dried. Yield: 94.8% (5.85 mmol, 4.09 g). Structural analysis by 1H
NMR was performed after peracylation.
structural formula of the polymer. Solubility parameter can be eval-
uated using:
ꢀ
ꢂ
ꢁ
iꢀei
ꢀE
V
v
ꢁ
ı =
=
(1)
iꢀvi
where ꢀei and ꢂvi are the additive atomic and group contribution
for the energy of vaporization and molar volume, respectively. The
contributions applicable at a temperature of 25 ◦C are presented in
the reference (Fedors, 1974). For high molecular weight polymers
that have Tg > 25 ◦C, there is a deviation between the experimentally
measured ꢀE and V and the estimated values. A small correction
v
factor was introduced to take into account the divergence in the V
values:
ꢀvi = 4n, n < 3
(2)
(3)
A similar procedure was followed to synthesize benzyl cellulose
suberate.
ꢀvi = 2n, n ≥ 3
Peracylation. Benzyl cellulose suberate or sebacate (300 mg) was
dissolved in 6 mL of pyridine, 25 mg 4-dimethylaminopyridine,
and 5 mL of acetic or propionic anhydride. After stirring for 24 h
at 80 ◦C, the product was precipitated into 250 mL water, col-
lected by filtration, and then washed several times with water.
The crude product was re-dissolved in 15 mL of CH2Cl2. This
solution was added slowly with rapid stirring to 250 mL of
ethanol. After filtration and washing with excess ethanol sev-
eral times, the sample was dried under vacuum at 40 ◦C. 1H
NMR of peracetylated benzyl cellulose sebacate (CDCl3): 0.94–1.35
(COCH2CH2CH2CH2CH2CH2CH2CH2CO of sebacate), 1.50–1.72
(COCH2CH2CH2CH2CH2CH2CH2CH2CO of sebacate), 1.82–2.25
(COCH3 of acetate), 2.32 (COCH2CH2CH2CH2CH2CH2CH2CH2CO of
sebacate), 3.00–5.20 (cellulose backbone), 5.11 (CH2C6H5), 7.33
(CH2C6H5).
where n is the number of main chain skeletal atoms in the smallest
repeating unit of the polymer.
Crystal growth rate of ritonavir. Crystal growth rate was charac-
terized by measuring the rate of desupersaturation in the presence
of seed crystals. Crystal growth rate experiments were performed in
the presence and absence of pre-dissolved cellulose derivatives and
the initial ritonavir concentration was 10 ␮g/mL. Polymer concen-
trations of 5 ␮g/mL were used and all experiments were performed
in triplicate. The solubility of crystalline ritonavir is known not
to change in 5 ␮g/mL polymer solutions. The rate of desupersat-
uration of ritonavir was measured using an SI Photonics (Tuscon,
AZ) UV spectrometer coupled to a fiber optic probe (path-length
5 mm) at a constant temperature of 37 ◦C. Data were acquired at
5 s time intervals within wavelengths ranging from 200 to 450 nm.
The ritonavir peak was detected at 240 nm. Supersaturated solu-
tions were generated by adding a small volume of pre-dissolved
ritonavir in methanol to 100 mM sodium phosphate buffer, pH 6.8.
Prior to addition of solubilized ritonavir, seed crystals were added
to the buffer and allowed to equilibrate at 37 ◦C. Data collection
began immediately after generation of supersaturation. An over-
head stirrer was used to stir the solution at a speed of 400 rpm. The
slope of the concentration vs. time curve over the first 2 min of the
experiment was taken as the initial crystal growth rate. In order to
mitigate particle scattering effects, second derivatives (SIMCA P+
V. 12 software (Umetrics Inc., Umea, Sweden) of the spectra were
taken for the sample data.
Pd(OH)2/C hydrogenolysis of the benzyl cellulose ester (Scheme 1).
To
a solution of 500 mg benzyl cellulose mixed ester dis-
solved in 50 mL of anhydrous THF, 500 mg palladium hydroxide,
20 wt% Pd (dry basis) on carbon was added. A hydrogen bal-
loon was attached to the flask, and the solution stirred overnight
under H2 at room temperature. The mixture was filtered
through Celite and concentrated. The product was dissolved in
dichloromethane and precipitated into hexanes. The precipitate
was collected and dried under vacuum. For benzyl CA 320S
suberate and sebacate esters, a second hydrogenolysis reaction
was required to achieve complete cleavage of the benzyl ester
bond.
CAP sebacate yield. 96.1% (1.39 mmol, 0.45 g). 1H NMR (d6-
DMSO): 0.80–1.10 (COCH2CH3 of propionate), 1.22 (COCH2CH2
3. Results and discussion
CH2CH2CH2CH2CH2CH2CO
of
sebacate),
1.45
(COCH2CH2CH2CH2CH2CH2CH2CH2CO of sebacate), 2.05–2.52
(COCH2CH2CH2CH2CH2CH2CH2CH2CO of sebacate, COCH2CH3 of
propionate, and COCH3 of acetate), 2.80–5.30 (cellulose backbone).
Preparation of amorphous solid dispersions by co-precipitation.
Powder solid dispersion systems of 1/3 (w/w) ritonavir/cellulose
␻-carboxyalkanoate (50 mg/150 mg) were prepared as follows:
accurately weighed quantities of polymer and drug were dissolved
in 32 mL acetone (or THF if the polymer was not soluble in acetone),
then the homogeneous solution was added into 150 mL water with
vigorous stirring. The organic solvent was removed under reduced
pressure by rotary evaporation at 40 ◦C. The aqueous drug/polymer
mixture was freeze-dried at −53 ◦C for 2 days to obtain solid pow-
der.
Solubility parameter (SP) of cellulose ω-carboxyalkanoates. SP cal-
culations were used to compare the relative hydrophobicities of the
polymers. SP values were estimated using the method proposed by
Fedors (1974). The method is based on group additive constants
and the contribution of a large number of functional groups was
evaluated, therefore this method requires only knowledge of the
3.1. Synthesis and characterization of cellulose
␻-carboxyalkanoates
In our previous work on synthesis of cellulose ester adipates,
we found that reaction of cellulose esters with adipic anhydride
could lead to crosslinked, insoluble products because of reaction
of multiple cellulose chains with poly(adipic anhydride) contam-
inant in the adipic anhydride (Kar et al., 2011). Though we later
developed a carefully designed procedure that circumvented these
issues and permitted use of adipic anhydride reagent to afford solu-
ble cellulose ester adipates (Liu et al., 2012), we found in the interim
that there was a more forgiving, if slightly less direct approach.
We found that conversion of adipic acid into a mono-functional,
mono-protected reagent provided a useful route for synthesis of
cellulose ester adipates for our initial structure–property studies.
Fischer esterification of the diacid with benzyl alcohol afforded
a mixture of diester, monoester, and starting diacid, from which
the monoester could be readily isolated in reasonable yield; this
was then smoothly converted using oxalyl chloride/DMF to the

120
H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
monoacid chloride. Commercially available cellulose esters were
reacted with the product monobenzyl adipoyl chloride to gener-
ate protected cellulose ester adipates. Finally, mild hydrogenation
removed the benzyl protective group to afford the cellulose adipate
esters.
(DMI). MEK is preferred over acetone due to its higher boiling point,
permitting the reaction to be run under more energetic conditions
(60 ◦C). The nature of this condensation reaction is such that there
should be considerable flexibility with regard to reaction solvent.
Substituent DS values in these protected cellulose seba-
cate/suberate products were calculated by integrating the cellulose
backbone and benzylic hydrogen peaks together and calculat-
ing the ratio to the phenyl protons (Fig. 1; sample calculation
Appendix 1). As shown in Table 1, sebacate DS in cellulose acetate
propionate sebacate (CAP Seb) responded in straightforward fash-
ion to an increase in monobenzyl sebacoyl chloride (BnSeCl)
used, affording nearly 3-fold increase in DS (Seb) with a three-
fold increase in acid chloride. We observe that ␻-carboxyalkanoyl
DS values decrease slightly for the CAB and CAP polymers as
the diacid chain gets longer. This can be rationalized by steric
hindrance. Similarly, CAP derivatives consistently have higher ␻-
carboxyalkanoyl DS values than the corresponding CAB derivatives,
with the obvious culprit being poorer accessibility of the OH groups
on cellulose in the presence of the longer butyrate chains. CA 320S
has less bulky substituents and higher hydroxyl content, thus by
using 2 equiv. monobenzyl sebacoyl chloride per CA 320S AGU,
the DS of benzyl-protected carboxyester can be up to 0.57. All
these benzyl-protected intermediate products are quite soluble in
organic solvents, for example in THF.
Hydrogenolysis of the benzyl ester groups on protected cel-
lulose adipates, for example on protected CAAdP, can be slightly
sluggish, occasionally requiring longer reaction times or re-
hydrogenation. In contrast, the longer and more flexible suberate
and sebacate chains are more suitable for heterogeneous catalytic
hydrogenation, so that all benzyl cellulose suberate and sebacate
hydrogenation reactions were facile and straightforward, achieving
clean hydrogenolysis of the benzyl ester. Use of Pd(OH)2/C catalyst
enables reaction completion overnight at room temperature on the
gram scale. Both 1H NMR and 13C NMR spectra (Fig. 2a and b) show
the disappearance of benzylic and aromatic signals, without any
observable residual substitution. In larger scale reactions it may be
necessary to perform the hydrogenation reaction twice to obtain
complete hydrogenolysis.
We planned to follow a similar approach for the synthesis of
cellulose ester suberates and sebacates (suberic and sebacic anhy-
drides would have large and unfavorable (9 and 11 membered,
respectively) ring sizes and to our knowledge are unknown). We
were however cognizant that the synthesis of the monoprotected,
monofunctional reagents required could be more complicated due
to solubility issues, and that often reaction of longer chain reagents
with cellulose can be difficult (Edgar, 2009). Indeed, we found that
using the same methodology we employed for monobenzyl adipate
synthesis gave exceptionally low yields of monobenzyl suberate
and monobenzyl sebacate. We investigated stoichiometric as well
as work up issues to improve the yields. In the monobenzyl adi-
pate synthesis, a molar ratio of benzyl alcohol to di-acid of 1.5 was
most effective, in spite of the excess benzyl alcohol that might be
expected to generate an excessive amount of dibenzyl adipate. We
were also concerned that the strongly basic conditions (NaOH) used
in the adipate workup might be reducing yield due to ester hydrol-
ysis. We reduced the molar ratio of benzyl alcohol to diacid to 1.2,
to minimize generation of the dibenzyl suberate or sebacate. By
extracting residual diacid with aqueous 1 M NaHCO3, we were able
to avoid excessive hydrolysis (NaOH) or poor separation of lay-
ers (Na2CO3), while still ensuring quantitative removal of diacid
(crucial since residual diacid would be converted to a diacid chlo-
ride in the ensuing step, leading to polymer crosslinking). 1H NMR
confirmed that the product monobenzyl esters, obtained in mod-
erate yields, were pure and uncontaminated by dibenzyl esters or
unreacted diacid.
Subsequent conversion of monobenzyl suberate and monoben-
zyl sebacate to the corresponding monoacid chlorides went very
smoothly using oxalyl chloride/DMF, affording more than 90% yield
in each case of crude products that were not purified, but were
shown by 1H NMR to have the desired monoprotected, monofunc-
tional structures (Supplementary materials, Figs. S3 and S4). Both
products were reactive yellow liquids that were stable if stored at
4 ◦C.
Reactions of these monofunctional reagents with commercially
available cellulose esters in common organic solvents were facile,
with no sign of crosslinking. Three different commercially avail-
able cellulose esters, cellulose acetate propionate (CAP-504-0.2),
cellulose acetate butyrate (CAB-553-0.4) and cellulose acetate (CA
320S), were selected due to their solvent solubility and high num-
ber of hydroxyls available for reaction (see Table 1 for details on
substituent DS); this also gave us a range of hydrophobicity that
would facilitate our structure–property studies. CAP-504-0.2 and
CAB-553-0.4 were reacted with the acid chlorides in methyl ethyl
ketone (MEK). The low DS CA we used is insoluble in MEK, so
instead was dissolved in warm 1,3-dimethyl-2-imidazolidinone
It was important to determine the relationship between cellu-
lose ␻-carboxyalkanoate structure and thermal properties. Tg in
particular is important in ASDs; the high Tg values for most known
cellulose esters are advantageous in ASD formulations, since they
ensure that the formulation Tg will remain well above ambient
temperature even when containing a substantial portion of drug
(which may be a plasticizer) and at high ambient humidity (water
will definitely be a plasticizer). Keeping the formulation Tg well
above ambient temperature ensures that the matrix is in the glassy
state, preventing drug migration through the matrix that could lead
to crystallization (Shelton et al., 2009). From the thermogram we
are also able to observe whether the polymer itself has a tendency
to crystallize. Thermal properties of cellulose suberate and seba-
cate mixed esters after deprotection are summarized in Table 1.
Table 1
Cellulose suberate/sebacate mixed-esters syntheses, DS values, and properties.
Sample
Starting CE
Carboxy-ester
Solvent
Temp
(◦C)
Molar ratioa
DS
(COOH)
DS
(Other)
DS
(Total)
Tg (◦C)
Mn
(g/mol)
DP
1
2
3
4
5
6
7
CAP-504-0.2
CAP-504-0.2
CAP-504-0.2
CAB-553-0.4
CAB-553-0.4
CA 320S
Sub
Seb
Seb
Sub
Seb
Sub
Seb
MEK
MEK
MEK
MEK
MEK
DMI
DMI
60
60
60
60
60
90
90
1
1
3
1
1
2
2
0.27
0.24
0.67
0.26
0.22
0.63
0.57
Pr 2.09 Ac 0.04
Pr 2.09 Ac 0.04
Pr 2.09 Ac 0.04
Bu 1.99 Ac 0.14
Bu 1.99 Ac 0.14
Ac 1.82
2.39
2.37
2.80
2.39
2.37
2.45
2.39
120
116
78
89
89
16.7
18.8
19.1
20.9
22.6
21.2
25.0
52
58
47
60
65
63
73
101
117
CA 320S
Ac 1.82
Abbreviations. Sub, suberate; Seb, sebacate; Ac, acetate; Pr, propionate; Bu, butyrate.
a
Mol monobenzyl acid chloride per mol anhydroglucose unit.

H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
121
Fig. 1. 1H NMR of benzyl CAP-504-0.2 sebacate (DS Se = 0.24) in CDCl3.
Fig. 2. 1H NMR (a) and 13C NMR (b) of CAP-504-0.2 sebacate (DS Se = 0.24) in d6-DMSO.

122
H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
Tg values of the products are generally 40–80 ◦C lower than the
starting cellulose esters due to the plasticizing effect of the poly-
methylene chains of the pendent ␻-carboxyalkanoyl groups. DSC
results also showed that Tg depression increased with increasing
chain length and increasing DS. Even so, Tg values of these depro-
tected cellulose ␻-carboxyalkanoates are at least 50 ◦C higher than
the highest anticipated ambient temperatures, indicating adequacy
for ASDs. Degree of polymerization (DP) data for CAP and CAB
derivatives indicate that little or no cellulose chain degradation
occurred during acylation and deprotection, consistent with the
relatively mild conditions applied. It should be noted that the DP of
CAP Seb (DS Se = 0.24) slightly exceeds the DP of the starting CAP
(Supplementary material, Table S1); it is reasonable to speculate
that the carboxyl-containing cellulosic polymers can self-aggregate
in THF solvent (Kar et al., 2011). For CA derivatives, the DPs are
reduced to less than half that of the starting CA-320S, presum-
ably due to the higher reaction temperatures employed (90 ◦C) in
acylation of CA.
As predicted from our previous study with very bulky acid
chlorides like pivaloyl chloride (Xu, Li, Tate, & Edgar, 2011), regiose-
lectivity was quite limited. After peracetylation of benzyl cellulose
sebacate (DS Seb = 1.95), acetyl methyls at O-6 (2.10 ppm), O-
2 (1.97 ppm) and O-3 (1.90 ppm) could still be observed by 1H
NMR (Supplementary material, Fig. S14). Table S2 (Supplemen-
tary material) summarizes the total/partial DS values according to
corresponding reaction feed ratio and substitution type.
The 13C NMR spectrum of peracetylated cellulose sebacate (DS
Seb = 0.55) further confirmed the lack of regioselectivity of this
reaction (Fig. 3). The hydrogenated product is readily soluble in
DMSO. Acetyl carbonyl carbons at 170.0, 169.1 and 168.8 ppm were
correlated in heteronuclear multibond correlation spectroscopy
(HMBC) with AGU protons at the sites of acetylation at O-6, O-
3 and O-2, respectively (Iwata, Azuma, Okamura, Muramoto, &
Chun, 1992). HMBC is not sensitive enough to provide the exact
distribution of sebacate group with much lower substitution, but
fortunately, the sebacoyl carbonyl carbons demonstrate as well
three different peaks near downfield around 171–174 ppm.
3.2. Regioselectivity
3.3. Effect of newly synthesized polymers on solution crystal
growth inhibition of ritonavir
It was of interest to explore the regioselectivity of reaction
of cellulose with the monoprotected long chain diacid chlorides,
since substitution regioselectivity has been shown to have power-
ful influence on properties like solubility, thermal properties, and
optical properties (Tasaka et al., 2006) to name just a few such
relationships (Fox, Li, Xu, & Edgar, 2011). To investigate regioselec-
tivity, microcrystalline cellulose was reacted in DMAc/LiCl solution
with monobenzyl suberoyl/sebacoyl chlorides. The benzyl cellu-
lose suberate and sebacate products from acylation in DMAc/LiCl
are only sparingly soluble in DMSO. To obtain a well-resolved NMR
spectrum, peracetylation was performed to provide products that
were readily soluble in chloroform. DS (␻-carboxyester) was cal-
culated based on the ratio of the integrals of benzyl ring protons
to those of the benzyl methylene and the cellulose backbone over-
lapping from 3.00 to 5.20 ppm. At a ratio of 1 equiv. BnSeCl per
AGU, a DS of 0.55 was obtained with a reaction time of 20 h at
60 ◦C. Using 3 equiv. BnSeCl per AGU under the same conditions,
a higher DS (1.95) was achieved. Benzyl cellulose sebacate with
high ␻-carboxyester substitution tends to self-aggregate in water,
therefore grinding and extra washing steps are necessary to remove
the impurities before peracetylation.
Supersaturating dosage forms, such as amorphous solids, are
attractive for improving the delivery of poorly water soluble drugs
since they result in a solution with a higher thermodynamic activ-
ity which may enhance the absorption of a drug relative to a
saturated solution (Warren, Benameur, Porter, & Pouton, 2010).
However, in order to maintain supersaturation, crystallization
must be prevented and hence the presence of an effective crystal
growth inhibitor in solution is desirable to prolong supersatura-
tion. While polymers are often used to fulfill this role (Alonzo
et al., 2012; Lechuga-Ballesteros & Rodriguez-Hornedo, 1995a;
Lechuga-Ballesteros & Rodriguez-Hornedo, 1995b; Zimmermann
et al., 2009), there is little mechanistic understanding of the prop-
erties of a polymer that make it a good inhibitor of crystal growth
for a given compound. In a previous study (Ilevbare et al., 2012b),
the effectiveness of a group of chemically diverse polymers, includ-
ing a number of novel cellulose ␻-carboxyalkanoates, at inhibiting
the crystal growth of ritonavir from solution was quantified. It was
found that moderately hydrophobic polymers were more effective
crystal growth inhibitors of the hydrophobic drug, ritonavir, than
Fig. 3. 13C NMR of fully-substituted cellulose acetate sebacate (DS Se = 0.55) in d6-DMSO.

H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
123
Table 2
DS and SP values for cellulose derivatives and their ranking.
Polymer abbreviation
DS
(CO2H)
DS
(Other)
DS
(Total)
DS (CO2H)
Rank
SP
SP
Rank
(MPa1/2
)
CP Adp
0.48
0.67
0.63
0.57
0.85
0.81
0.33
0.67
0.27
0.25
0.24
0.26
0.22
Pr 1.68
Ac 1.82
Ac 1.82
Ac 1.82
Ac 0.04; Pr 2.09
Ac 0.14; Bu 1.99
Ac 0.04; Pr 2.09
Ac 0.04; Pr 2.09
Ac 0.04; Pr 2.09
Ac 0.14; Bu 1.99
Ac 0.04; Pr 2.09
Ac 0.14; Bu 1.99
Ac 0.14; Bu 1.99
2.16
2.49
2.45
2.39
2.98
2.84
2.46
2.80
2.39
2.38
2.37
2.38
2.35
6
3
4
5
1
2
7
3
8
23.28
22.95
22.62
22.36
21.27
20.86
20.56
20.41
20.19
20.05
19.94
19.84
19.62
1
2
3
4
5
6
7
8
9
10
11
12
13
CA 320S Adp
CA 320S Sub
CA 320S Seb
CAP Adp 0.85
CAB Adp 0.81
CAP Adp 0.33
CAP Seb 0.67
CAP Sub
CAB Adp 0.25
CAP Seb 0.24
CAB Sub
10
11
9
CAB Seb
12
Abbreviations: Ac, acetate; Pr, propionate; Bu, butyrate; Adp, adipate; Sub, suberate; Seb, sebacate.
either highly hydrophilic or highly hydrophobic polymers. In addi-
tion, the novel cellulose-based polymers containing a higher DS
of ionizable carboxylic acids were better inhibitors relative to less
ionized cellulose polymers. We have evaluated the effectiveness of
several of the newly synthesized cellulose ␻-carboxyalkanoates in
inhibiting crystal growth of ritonavir from supersaturated solution,
in comparison with that of previously synthesized and reported
cellulose derivatives (Kar et al., 2011). Ritonavir was chosen as the
model compound due to its very low aqueous solubility. Aqueous
insolubility results primarily from high log P (lipophilicity) and/or
high melting point, Tm (representing lattice energy). Ritonavir is a
highly lipophilic compound with a log P value of 5.98 and melting
point of 121 ◦C (Baird, Van Eerdenbrugh, & Taylor, 2010; Murdande,
Pikal, Shanker, & Bogner, 2010), yielding an equilibrium solubility
of 1.3 0.10 ␮g/mL, at pH 6.8 (ritonavir is un-ionized at this pH)
and 37 ◦C.
Ritonavir crystal growth rate in the absence and presence of
the novel cellulosic polymers was determined using the method
previously described (Ilevbare et al., 2012b). The novel cellulose
derivatives used in this report and some of their properties are pre-
sented in Table 2. Solubility parameters (SP), estimated according
to the method of Fedors (1974), were used to rank relative poly-
mer hydrophobicity. The solubility parameter provides a numerical
estimate of the intermolecular forces within a material, and can be
a good indication of solubility, particularly for non-polar materials
such as polymers. The hydrophobicity ranking is presented, where
the least hydrophobic polymer, CP Adp is ranked #1 and #13 rep-
resents the most hydrophobic polymer, CAB Seb. The DS values of
carboxyl-containing substituents (DS (CO2H)) of the new cellulose
esters are also listed and ranked in Table 2.
high DS of the suberate and sebacate substituents (more ionizable
carboxylic acids), combining the relatively more hydrophilic cel-
lulose ester (CA320S) with the longer chain ␻-carboxyalkanoate
(e.g. suberate and sebacate) resulted in polymers with an optimal
level of hydrophobicity, thus making CA320 Sub and CA320 Seb
very effective crystal growth inhibitors. This further supports the
importance of hydrophobicity as a key factor influencing polymer
effectiveness.
3.4. ASDs of ritonavir in cellulose ␻-carboxyalkanoate matrices
We prepared ASDs of the model drug ritonavir in cellulose ␻-
carboxyalkanoates by co-precipitation of a solution of drug and
polymer in a common solvent (acetone or THF) into a common
non-solvent, water, followed by solvent removal by rotary evap-
oration and lyophilization to remove water. Spray drying was not
practical due to the small amounts of each novel polymer avail-
able. ASD samples were characterized by PXRD to determine the
Fig. 4 compares the growth rate ratio of ritonavir (10 ␮g/mL,
S of 7.6) in the absence of polymer (Rg0) to that in the presence
of polymer (Rgp). Polymers are arranged in order of increas-
ing hydrophobicity (L to R). Polymers with an effectiveness ratio
greater than 1 (Rg0/Rgp > 1) are considered effective crystal growth
inhibitors. With the exception of the data for the CA 320S seba-
cates and suberates, we have previously reported effectiveness
ratios for the other cellulose ␻-carboxyalkanoates. Out of the 13
cellulose ␻-carboxyalkanoates presented in Fig. 4, 6 were ineffec-
tive, while 7 were effective at inhibiting crystal growth, albeit to
varying extents. Polymers located in the middle portion of Fig. 4
(i.e. moderately hydrophobic polymers (20.56–22.62 MPa1/2)) were
the most effective crystal growth inhibitors. In the previous study,
CAP Adp 0.85 and CAB Adp 0.81 were identified as the most effec-
tive cellulose-based polymers (Ilevbare et al., 2012b). However,
two of the new cellulose ␻-carboxyalkanoate polymers, CA320S
Sub and CA320S Seb, were significantly more effective at inhibi-
ting the crystal growth of ritonavir. Together with the relatively
Fig. 4. Crystal growth rate ratio of ritonavir at an initial ritonavir concentration
of 10 ␮g/mL. The data is arranged in order of hydrophobicity: least hydrophobic to
most hydrophobic (left to right). Crystal growth rate experiments were performed in
triplicate. Each column is an average of the effectiveness ratio and error bars indicate
one standard deviation. The y-axis is a ratio of the growth rate of ritonavir in the
absence of polymer to growth rate of ritonair in the presence of polymer (5 ␮g/mL).
Polymers with a ratio > 1 are considered effective crystal growth inhibitors. With
the exception of CA 320S Sub and CA 320S Seb, the effectiveness growth ratios for
all cellulose ␻-carboxyalkanoates were previously reported (Ilevbare et al., 2012b).

124
H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
Fig. 5. PXRD pattern of ritonavir crystalline powder and amorphous solid dispersions prepared by co-precipitation with cellulose ␻-carboxyalkanoates.
crystallinity of the ritonavir in the ASD (Fig. 5). PXRD patterns of all
polymer/RTV 3/1 ASD were devoid of the diffraction peaks evident
in the XRD pattern of crystalline ritonavir, indicating that ritona-
vir is completely amorphous in these blends. To further confirm
their amorphous nature, DSC thermograms (Fig. S17, Supplemen-
tary material) of the ASDs showed single, broad glass transitions,
whereas the melting endotherm of crystalline ritonavir at ∼125 ◦C
was not observed. Based on our previous work and that of others,
we expect that the aqueous solution concentrations of ritonavir
released from these ASDs will be significantly increased vs. that
from crystalline drug. Determination of solubility enhancement
(supersaturation) versus time and other drug release studies will
be discussed in a forthcoming paper.
enhancement, its generality with respect to drug structure, and the
mechanism of solubilization in forthcoming reports.
Acknowledgments
This project was supported primarily by a grant from the
National Science Foundation (NSF, grant number DMR-0804501).
We thank Eastman Chemical Company for their kind donation of
the commercial cellulose esters used in this work, and the Macro-
molecules and Interfaces Institute and the Institute for Critical
Technologies and Applied Science at Virginia Tech for their finan-
cial, facilities, and educational support. We also thank Alison Shultz
for SEC analyses, Rick Caudill for XRD experiments and Bruce Orler
for some of the DSC measurements (all from Virginia Tech).
4. Conclusions
Appendix A. Appendix 1. The degree of substitution (DS)
values were determined according to the following
equation
Successful pathways were developed for the synthesis of cel-
lulose ␻-carboxyalkanoates by reaction of cellulose, CAP, CAB and
CA with monobenzyl suberoyl and monobenzyl sebacoyl chlorides.
Barriers to acid chloride synthesis were overcome to develop rea-
sonablyefficientprocesses. Condensationof acid chlorideswith cel-
lulose esters can be carried out in common organic solvents under
mild conditions, and appears to be capable of affording a wide range
of DS ␻-carboxyalkanoates. Even though a relatively bulky benzyl-
protected long chain diacid was incorporated, virtually no regiose-
lectivity of substitution was observed. Debenzylation of these long
chain benzyl cellulose ␻-carboxyalkanoates was facile under mild
conditions in the presence of a palladium catalyst. The resulting
carboxyl-containing mixed esters are well suited as ASD polymers
for oral drug delivery, and most of them form amorphous disper-
sions with the hydrophobic drug ritonavir. All of them possess Tg
values adequate to prevent drug recrystallization in the solid phase.
Most importantly, CA 320S suberate and sebacate are at least 5-fold
more effective as inhibitors of ritonavir crystallization from aque-
ous solution than other cellulosic and non-cellulosic ASD polymers.
We will further explore these novel cellulose ␻-carboxyalkanoates
with respect to the degree and duration of solution concentration
7Iphenyl
DS ω-carboxyester =
5Ibackbone+benzylic hydrogen − 2Iphenyl
Appendix B. Supplementary data
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/j.carbpol.
2012.11.049.
References
Abell, A. D., Morris, K. B.,
& Litten, J. C. (1990). Synthesis and depro-
tection of [1-(ethoxycarbonyl)-4-[(diphenylmethoxy)carbonyl]-1-methyl-2-
oxobutyl]triphenylphosphonium chloride: A key intermediate in the Wittig
reaction between a cyclic anhydride and a stabilized ylide. Journal of Organic
Chemistry, 55, 5217–5221.
Alonzo, D. E., Gao, Y., Zhou, D., Mo, H., Zhang, G. G. Z., & Taylor, L. S. (2011). Dis-
solution and precipitation behavior of amorphous solid dispersions. Journal of
Pharmaceutical Sciences, 100, 3316–3331.

H. Liu et al. / Carbohydrate Polymers 100 (2014) 116–125
125
Alonzo, D. E., Raina, S., Zhou, D., Gao, Y., Zhang, G. G. Z., & Taylor, L. S. (2012).
Characterizing the impact of hydroxypropylmethyl cellulose on the growth and
nucleation kinetics of felodipine from supersaturated solutions. Crystal Growth
and Design, 12, 1538–1547.
Alonzo, D. E., Zhang, G. G., Zhou, D., Gao, Y., & Taylor, L. S. (2010). Understanding the
behavior of amorphous pharmaceutical systems during dissolution. Pharmaceu-
tical Research, 27, 608–618.
Baird, J. A., Van Eerdenbrugh, B., & Taylor, L. S. (2010). A classification system to assess
the crystallization tendency of organic molecules from undercooled melts. Jour-
nal of Pharmaceutical Sciences, 99, 3787–3806.
Cameron, W., Japour, A. J., Xu, Y., Hsu, A., Mellors, J., Farthing, C., et al. (1999). Ritona-
vir and saquinavir combination therapy for the treatment of HIV infection. AIDS,
13, 213–224.
Edgar, K. J. (2009). Direct synthesis of partially substituted cellulose esters. In K. J.
Edgar, C. M. Buchanan, & T. Heinze (Eds.), Polysaccharide materials: Performance
by design (pp. 213–229). Washington, DC: American Chemical Society.
Edgar, K. J., Arnold, K. M., Blount, W. W., Lawniczak, J. E., & Lowman, D. W. (1995). Syn-
thesis and properties of cellulose acetoacetates. Macromolecules, 28, 4122–4128.
Edgar, K. J., Buchanan, C. M., Debenham, J. S., Rundquist, P. A., Seiler, B. D., Shelton,
M. C., et al. (2001). Advances in cellulose ester performance and application.
Progress in Polymer Science, 26, 1605–1688.
Kar, N., Liu, H., & Edgar, K. J. (2011). Synthesis of cellulose adipate derivatives.
Biomacromolecules, 12, 1106–1115.
Konno, H., Handa, T., Alonzo, D. E., & Taylor, L. S. (2008). Effect of polymer type
on the dissolution profile of amorphous solid dispersions containing felodipine.
European Journal of Pharmaceutics and Biopharmaceutics, 70, 493–499.
Lechuga-Ballesteros, D., & Rodriguez-Hornedo, N. (1995a). Effects of molecular
structure and growth kinetics on the morphology of l-alanine crystals. Inter-
national Journal of Pharmaceutics, 115, 151–160.
Lechuga-Ballesteros, D., & Rodriguez-Hornedo, N. (1995b). The influence of additives
on the growth kinetics and mechanism of l-alanine crystals. International Journal
of Pharmaceutics, 115, 139–149.
Li, B., Konecke, S., Wegiel, L., Taylor, L. S., & Edgar, K. J. (2012). Both solubility and
chemical stability of curcumin are enhanced by solid dispersion in cellulose
derivative matrices, under review.
Liu, H., Kar, N., & Edgar, K. J. (2012). Direct synthesis of cellulose adipate derivatives
using adipic anhydride. Cellulose, 19, 1279–1293.
Mawad, D., Mouaziz, H., Penciu, A., Méhier, H., Fenet, B., Fessi, H., et al. (2009). Elab-
oration of radiopaque iodinated nanoparticles for in situ control of local drug
delivery. Biomaterials, 30, 5667–5674.
Murdande, S. B., Pikal, M. J., Shanker, R. M.,
& Bogner, R. H. (2010). Solubil-
ity advantage of amorphous pharmaceuticals: II. Application of quantitative
thermodynamic relationships for prediction of solubility enhancement in
structurally diverse insoluble pharmaceuticals. Pharmaceutical Research, 27,
2704–2714.
English, A. R., Girard, D., Jasys, V. J., Martingano, R. J., & Kellogg, M. S. (1990). Orally
effective prodrugs of the beta-lactamase inhibitor sulbactam. Journal of Medici-
nal Chemistry, 33, 344–347.
Fedors, R. F. (1974). Method for estimating both the solubility parameters and molar
volumes of liquids. Polymer Engineering and Science, 14, 147–154.
Fox, S. C., Li, B., Xu, D., & Edgar, K. J. (2011). Regioselective esterification and etheri-
fication of cellulose: A review. Biomacromolecules, 12, 1956–1972.
Friesen, D. T., Shanker, R., Crew, M., Smithey, D. T., Curatolo, W. J., & Nightingale, J.
A. (2008). Hydroxypropyl methylcellulose acetate succinate-based spray-dried
dispersions: An overview. Molecular Pharmacology, 5, 1003–1019.
Hancock, B. C., Shamblin, S. L., & Zografi, G. (1995). Molecular mobility of amorphous
pharmaceutical solids below their glass transition temperatures. Pharmaceutical
Research, 12, 799–806.
Newman, A., Knipp, G., & Zografi, G. (2012). Assessing the performance of amorphous
solid dispersions. Journal of Pharmaceutical Sciences, 101, 1355–1377.
Posey-Dowty, J. D., Watterson, T. L., Wilson, A. K., Edgar, K. J., Shelton, M. C., & Linger-
felt, L. R., Jr. (2007). Zero-order release formulations using a novel cellulose ester.
Cellulose, 14, 73–83.
Shelton, M. C., Posey-Dowty, J. D., Lingerfelt, L. R., Kirk, S. K., Klein, S., & Edgar, K.
J. (2009). Enhanced dissolution of poorly soluble drugs from solid dispersions
in carboxymethylcellulose acetate butyrate matrices. In K. J. Edgar, T. Heinze,
& T. Liebert (Eds.), Polysaccharide materials: Performance by design (pp. 93–113).
Washington, DC: American Chemical Society.
Hauss, D. J. (2007). Oral lipid-based formulations. Advanced Drug Delivery Reviews,
59, 667–676.
Ilevbare, G., Liu, H., Edgar, K. J., & Taylor, L. (2012). Inhibition of solution crystal
growth of ritonavir by cellulose polymers—Factors influencing polymer effec-
tiveness. CrystEngComm, 14, 6503–6514.
Ilevbare, G. A., Liu, H., Edgar, K. J., & Taylor, L. S. (2012). Understanding polymer
properties important for crystal growth inhibition—Impact of chemically diverse
polymers on solution crystal growth of ritonavir. Crystal Growth and Design, 12,
3133–3143.
Iwata, T., Azuma, J., Okamura, K., Muramoto, M., & Chun, B. (1992). Preparation
and NMR assignments of cellulose mixed esters regioselectively substituted by
acetyl and propanoyl groups. Carbohydrate Research, 224, 277–283.
Tasaka, K., Umeda, H., Kuzuhara, N., & Yajima, T. (2006). Cellulose ester film,
manufacturing, optical retardation film, optical compensation sheet, elliptic
polarizing plate, and image display. US Patent 7,038,744.
Warren, D. B., Benameur, H., Porter, C. J. H., & Pouton, C. W. (2010). Using poly-
meric precipitation inhibitors to improve the absorption of poorly water-soluble
drugs: A mechanistic basis for utility. Journal of Drug Targeting, 18, 704–731.
Xu, D., Li, B., Tate, C., & Edgar, K. J. (2011). Studies on regioselective acylation of
cellulose with bulky acid chlorides. Cellulose, 18, 405–419.
Zimmermann, A., Millqvist-Fureby, A., Elema, M. R., Hansen, T., Muellertz, A., & Hov-
gaard, L. (2009). Adsorption of pharmaceutical excipients onto microcrystals
of siramesine hydrochloride: Effects on physicochemical properties. European
Journal of Pharmaceutics and Biopharmaceutics, 71, 109–116.