Journal of Pharmaceutical and Biomedical Analysis 172 (2019) 253–258

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis

journal homepage: www.elsevier.com/locate/jpba

Development of simple isocratic HPLC methods for siRNA

quantitation in lipid-based nanoparticles
Zongyun Huang ∗ , William P. Fish
Bristol-Myers Squibb, Drug Product Science & Technology, 1 Squibb Drive, New Brunswick, NJ 08901, USA

a r t i c l e i n f o a b s t r a c t

Article history: This paper describes the development of simple and user-friendly HPLC methods that can quantitate the
Received 8 February 2019 amount of small interfering RNA (siRNA) in lipid-based nanoparticle (LNP) formulations. The methods
Accepted 12 April 2019 have been used as alternative chromatographic approaches to the size exclusion chromatography in order
Available online 27 April 2019
to perform “fit for purpose” analysis such as determining the amount of released siRNA from LNP formu-
lations as a part of in-vitro release testing. Two HPLC conditions were optimized using reversed phase (a
Keywords:
250 × 4.6 mm Waters XSelect CSH column) and cation exchange columns (a 250 × 4.6 mm Zorbax SCX-
Oligonucleotide
300 column) maintained at 30 ◦ C with a mobile phase of 0.1 M ammonium bicarbonate aqueous solution
siRNA
Lipid nanoparticle
containing 20–30% acetonitrile. All the siRNA variants were excluded from pores in stationary phase and
Liquid chromatography led to a single peak eluted earlier than the solvent front. The impact of chromatographic conditions and
Size exclusion chromatography column configurations on method specificity, accuracy, and sensitivity have been investigated. Validation
Analytical method data for both methods in terms of precision, linearity, accuracy and sensitivity are also presented.
© 2019 Elsevier B.V. All rights reserved.

1. Introduction siRNA (free plus encapsulated) in the sample matrix. Tradition-

ally, size exclusion chromatographic (SEC) methods are used for
Small interfering ribonucleic acids (siRNAs) are short dou- siRNA determination with both advantages and drawbacks. Dou-
ble stranded RNAs which can silence the expression of specific ble and single stranded siRNA can be separated adequately using
post-transcriptional genes by the direct sequence-specific cleavage SEC columns with appropriate pore size, however, the susceptibil-
leading to translation repression and consequent degradation of ity of SEC columns to back pressure requires low flow rate resulting
messenger RNA [1,2]. SiRNAs have become a promising therapeu- in broadened peak shape and long run time. Band broadening of
tic tool for several pathological areas such as viral infections, cancer, siRNA peaks often posts challenges in quantifying siRNA at low
genetic disorders, fibrosis, and autoimmune disorders [3]. In order concentration (i.e. < 1 ␮g/mL) In addition, SEC columns are usually
for siRNA to reach the targets to show efficacy, effective drug deliv- costly and have shorter column life times when unfriendly sample
ery systems are required to facilitate transfection because siRNA is matrixes are used such as the ones containing substantial amount of
susceptible to degradation by endogenous enzymes, and is too large lipids. Nevertheless, the methodologies for determining total siRNA
(approximately 13 kDa) and too negatively charged to cross cellular amount are not restricted to SEC, provided that the alternative
membranes. Lipid-based nanoparticles (LNP) are commonly used technique demonstrates specificity, accuracy and sensitivity.
for siRNA delivery due to their effective protection from plasmatic The objective of this study is to develop a fast, user-friendly
nucleases, desirable particle size and non-immunogenicity [4–6]. and robust isocratic HPLC method with adequate sensitivity for the
Various analytical tests are required to characterize many in-vitro analysis of large numbers of samples in timely matter. This paper
features of siRNA drug products made of nanoparticles in order to discusses the course of development for such an HPLC method
assure the desired in-vivo performance. Typically, Encapsulation for the determination of total siRNA content in various sample
Efficiency, a measure of how much siRNA is inside the LNP, and in- matrixes and media.
vitro release, a measure of how much siRNA will leak from the LNP
over time, are two critical quality attributes used to monitor LNP
quality. Monitoring these attributes is challenging as there is a need 2. Experimental
to measure “free siRNA” (RNA outside of the LNP), as well as the total
2.1. Materials and reagents

∗ Corresponding author. ACS reagent grade chemicals from MilliporeSigma (Burlington,

E-mail address: zongyun.huang@bms.com (Z. Huang). MA, USA) were used unless otherwise indicated. Milli-Q purified

https://doi.org/10.1016/j.jpba.2019.04.026
0731-7085/© 2019 Elsevier B.V. All rights reserved.
254 Z. Huang, W.P. Fish / Journal of Pharmaceutical and Biomedical Analysis 172 (2019) 253–258

Fig. 1. Chromatograms of a solution containing 0.1 mg/mL siRNA using a Waters CSH C18 (3.5 ␮m particle size) 15 cm × 4.6 mm column operated at different column
temperatures of (A) 60 ◦ C and (B) 25 ◦ C; flow rate: 1.0 ml min−1 ; injection volume: 10 ␮l; UV detection: 260 nm; mobile phase A: 0.1 M ammonium bicarbonate buffer, B:
acetonitrile with initial composition of 97.5:2.5 (%A:%B) followed by a 15-minute gradient from 97.5:2.5 to 80:20 (%A:%B) with 5 min hold at 20%B. Run time: 20 min. Sample
diluent: TE buffer. Peak identity: 1 = system peak attributed to TE buffer, 2 = double strand siRNA, 3 = single strand siRNA.

water (MilliporeSigma), sodium dodocyl sulfate (SDS), ammonium 25 cm × 4.6 mm i.d.) maintained at 30 ◦ C with a mobile phase of
bicarbonate, monobasic sodium phosphate, dibasic sodium phos- 0.1 M ammonium bicarbonate aqueous solution and acetonitrile
phate, phosphoric acid, HPLC grade acetonitrile and isopropyl (80:20 v/v), a flow rate of 1.0 mL/min, UV detection at 260 nm, an
alcohol were used to prepare the mobile phases and sample injection volume of 10 ␮L, and a run time of 5 min.
solutions. siRNA drug substances and formulated LNP drug prod-
uct were provided by Bristol-Myers Squibb Drug Product Science
and Technology Department (New Brunswick, NJ, USA). Tris-EDTA 2.3. Sample preparation
buffer solution (TE), phosphate buffered saline solution (PBS) and
10% TritonTM X-100 solution were used as received. A stock standard solution of siRNA was prepared at a con-
centration of approximately 5 mg/mL in purified water. Working
2.2. Chromatographic conditions for the determination of siRNA standards were prepared in TE, PBS and HEPES buffer solu-
tions by diluting the stock siRNA solution to a concentration of
A Waters Alliance HPLC system equipped with an ultraviolet 10 ␮g/mL. The LNP sample solutions were prepared by recon-
detector (Waters Corporation, Milford, MA) was used for all the stituting lyophilized LNP with Milli-Q water and diluting the
studies. Chromatographic data were recorded and processed using reconstituted LNP drug product samples in different buffer dilu-
Waters Empower software. ents to a target siRNA concentration ranging from 1 to 8 ␮g/mL.
Several chromatographic conditions for siRNA quantitation The diluent for the determination of total siRNA content in LNP
were evaluated during the course of method development and drug products is TE buffer with 1% TritonTM X-100 or SDS, in order
a number of analytical columns were used including Ascen- to solubilize the lipid components.
tis Express C18 (2.7 ␮m, 15 cm × 4.6 mm i.d., Sigma-Aldrich, St.
Louis, MO), Waters Xselect CSH C18 (3 ␮m, 15 cm × 4.6 mm i.d.;
Waters Corporation, Milford, MA, USA), and Zorbax SCX-300 2.4. Method validation
(5 ␮m, 15 cm × 4.6 mm i.d.; Agilent Technologies, Santa Clara,
CA). The details of these method conditions are discussed in The method for siRNA quantitation were validated for speci-
detail in the Results and Discussion section. The HPLC conditions ficity, linearity, accuracy, precision, sensitivity and solution
were optimized using a Waters Xselect CSH C18 column (5 ␮m, stability. Details are described in the Results and Discussion section.
 Z. Huang, W.P. Fish / Journal of Pharmaceutical and Biomedical Analysis 172 (2019) 253–258 255

Fig. 2. Example chromatograms of a solution containing 0.1 mg/mL siRNA using a Waters CSH C18 (3.5 ␮m particle size) 15 cm × 4.6 mm with different mobile phase
composition between 0.1 M ammonium bicarbonate and acetonitrile: (A) 95:5 v/v, (B) 90:10 v/v and (C) 80:20 v/v. Column temperature: 30 ◦ C; flow rate: 1.0 ml min−1 ;
injection volume: 10 ␮l; UV detection: 260 nm. Sample diluent: TE buffer. Peak identity: 1 = siRNA, 2 = system peak attributed to TE buffer.

3. Results and discussion ability to accurately quantitate total siRNA content, which consists
of double and single stranded siRNA and all the variants, has several
3.1. Evaluation of gradient RPLC method drawbacks especially when very diluted samples are analyzed: 1).
since siRNA and its variants are exhibited as multiple peaks in the
The analyte used for evaluation is an siRNA substance with chromatogram, the sensitivity of each individual peak is reduced;
approximate molecular masses of 13,000 and 6500 g/mole for 2). the sum of area responses of all the siRNA and variants peaks
double and single strand, respectively. It also contains numer- needs to be calculated and any changes to the peak area ratio may
ous RNA variants with similar molecular masses. The retention impact the accuracy of total siRNA determination because peak area
behavior of oligonucleotide analytes in RPLC is different from that increases with increasing retention times; 3). The run time of the
of small-molecule analytes with relatively low molecular mass gradient RPLC method is relatively lengthy (>15 min) for single ana-
(MW < 2000 g/mole). Partitioning of the analyte between station- lyte quantitation. Therefore, the analytical approach using gradient
ary and mobile phases is believed to play an important role in elution was not preferred for our particular task to quantitate total
retaining small-molecule analytes [7,8]. Oligonucleotides behave siRNA content.
more like polymeric analytes due to their high molecular mass
and their “retention” in gradient RPLC is governed through the 3.2. Development of isocratic HPLC methods
contributions of multiple effects. In RPLC, polymeric analytes,
such as oligonucleotides, tend to be adsorbed and precipitated on The method development strategy for isocratic elution is to
the immobilized hydrophobic ligands on the stationary phase. As exclude all siRNA peaks prior to the solvent front as a single peak
the organic strength (e.g. acetonitrile level) in the mobile phase using the polymeric characteristics of oligonucleotides. A similar
increases, they are desorbed and re-dissolved. The threshold of analytical approach has been used successfully for quantitation of
mobile phase strength to desorb/re-dissolve polymers primarily a polymeric impurity in asunaprevir drug product by our research
depends on the hydrophobicity of analytes. Once desorbed/re- team [12]. The isocratic conditions were evaluated with varying
dissolved, the polymer homologues with different molecular mass acetonitrile level in the mobile phase on several reversed phase
are excluded from the column at different elution times in a gra- columns. As shown in Fig. 2, a significant amount of siRNA was
dient elution [9–11]. Fig. 1 shows chromatograms of a solution re-dissolved and excluded from the C18 column when the ace-
containing 0.1 mg/mL siRNA using gradient elution varying the ace- tonitrile level in the mobile phase reached 10% v/v. Fig. 3 shows
tonitrile fraction in mobile phase from 5 to 20% for 15 min. siRNA a plot of siRNA peak area responses versus the percent acetoni-
eluted as double and single strand peaks at 5.0 and 7.3 min, respec- trile levels in the mobile phase under isocratic conditions. For
tively. Using a column temperature at 60 ◦ C, significant changes of RPLC, the total area response of siRNA increased with increasing
the peak area ratio between the double and single strand siRNA percent acetonitrile in the mobile phase and reached a plateau
peaks was observed, indicating denaturing of the double strand when mobile phase with 15% acetonitrile is used, indicating that
siRNA (Fig. 1a). This method can quantitate both double and single all the variants of siRNA were re-dissolved and excluded from
strand siRNA with adequate precision and accuracy, however, its the column. This analytical approach enables the separation of
256 Z. Huang, W.P. Fish / Journal of Pharmaceutical and Biomedical Analysis 172 (2019) 253–258

As shown in Fig. 4, separation between siRNA and the solvent front

peak depends on the pore size, not the particle size, of the substrate
stationary phase particle. Therefore, Xselect CSH C18 column was
selected for siRNA quantitation due to the favorable porous prop-
erty of its packing materials. The optimized RPLC conditions for the
determination of total siRNA are described in the Experimental sec-
tion above. Mobile phase containing 20% acetonitrile was chosen
based on the data shown in Fig. 3 to ensure all the siRNA variants
are eluted prior to the solvent front and a longer length Xselect CSH
column (25 cm) was selected to increase the separation.
The elution behavior of siRNA was also studied on a strong cation
exchange column to examine the exclusion effect from both ion
and size. All the siRNA was eluted before solvent front using a Zor-
bax SCX column with a 100% aqueous buffer because the negatively
Fig. 3. Plot of peak area response of siRNA versus acetonitrile percent of mobile charged ligands on the stationary phase excluded the siRNA (a weak
phase on a Waters C18 (3.5 ␮m particle size) 15 cm × 4.6 mm column. acid) in addition to the size exclusion effect. When analyzing siRNA
in samples containing non-ionic surfactants such as TritonTM X100,
RPLC method is less advantageous due to potential interference
siRNA from all other non-polymeric components by leveraging the resulting from the strong retention and UV absorption of Triton. The
size-exclusion effect of porous stationary phase particles. The res- conditions using Zorbax SCX columns is a viable alternative for ana-
olutions between the siRNA peak and solvent front peaks were lyzing such samples because Triton is less retained on ion exchange
examined using C18 columns packed with particles of different stationary phases. An example chromatogram for LNP sample solu-
types and pore sizes including partially porous (Ascentis Express tions with Triton using SCX column were shown in Fig. 5a. The
C18, fused-core, 2.7 ␮m particle size and 90 Å pore size) and fully Triton peak eluted at 4 min and the baseline was re-equilibrated to
porous (XSelect CSH C18, 3.5 ␮m particle size and 130 Å pore size; the initial level at approximately 6 min due to a late eluting system
Aquity CSH C18, 1.7 ␮m particle size and 90 Å pore size) substracts. peak attributed to the TE buffer resulting in a final run was time

Fig. 4. Example chromatograms of siRNA sample solutions in TE buffer using 15 cm × 4.6 mm columns packed with different particles: (A) Ascentis Express C18 fused-core
particles (2.7 ␮m particle size, 90 Å pore size, partially porous), (B) Xselect CSH C18 fully porous particles (3.5 ␮m particle size, 130 Å pore size) and (C) Acquity UPLC CSH C18
fully porous particles (1.7 ␮m particle size, 130 Å pore size) in isocratic elution. Column temperature 30 ◦ C; flow rate: 1.0 ml min−1 ; injection volume: 10 ␮l; UV detection:
260 nm; mobile phase: 0.1 M ammonium bicarbonate/acetonitrile (80:20 v/v). Peak identity: 1 = siRNA, 2 = system peak attributed to TE buffer.
 Z. Huang, W.P. Fish / Journal of Pharmaceutical and Biomedical Analysis 172 (2019) 253–258 257

Fig. 5. (A) An example chromatogram of an LNP sample solution in TE buffer containing 1.2 ␮g/mL siRNA with 1% Triton using the IEC condition. (B) An example chromatogram
of an LNP sample solution containing 4 ␮g/mL siRNA with 1% SDS using the RPLC condition. RPLC conditions are described in the Experimental Section. Conditions using 15 cm
x 4.6 mm Zorbax SCX-300 column are column temperature 30 ◦ C; flow rate: 1.0 ml min−1 ; injection volume: 10 ␮l; UV detection: 260 nm; mobile phase: 0.1 M ammonium
bicarbonate/acetonitrile (70:30 v/v). Peak identity: 1 = siRNA, 2 = system peak attributed to TE buffer, 3 = Triton.

Table 1
Results for Validation and Sample Analysis.

Linear Regression

Column Type Range Correlation Coefficient

Reversed Phase 0.05 – 15 ␮g/mL 0.9995

SCX Phase 0.05 – 15 ␮g/mL 0.9998

Accuracy

% Recovery vs. siRNA Amount Expected in Sample Solution

Concentration of siRNA in Spiked Sample Solution (␮g/mL)
Reversed Phase SCX Phase
1.0 97.2 98.4
5.0 99.3 100.4
10.0 99.2 99.3

Repeatability of 6 Sample Preps for Total siRNA Determination of LNP Drug Product

%RSD (6) 1.1

Sensitivity - Injection Repeatability of 0.025 ␮g/mL siRNA in the Presence of LNP

Column Type Area (6) % RSD (6)

Reversed Phase 487 2.8

SCX Phase 543 1.9

Results for siRNA Amount in LNP Drug Product Sample Solutions in Different Diluents

Sample Type Sample Diluent Theoretical Conc. (mg/mL) Conc. Found (mg/mL) Column Used

Reconstituted Solution TE buffer/1% Triton 3.0 2.982 SCX

Frozen Solution TE buffer/1% Triton 2.0 1.984 SCX
Reconstituted Solution TE buffer/1% SDS 3.0 2.976 Reversed Phase
Frozen Solution TE buffer/1% SDS 2.0 2.014 Reversed Phase
258 Z. Huang, W.P. Fish / Journal of Pharmaceutical and Biomedical Analysis 172 (2019) 253–258

of 7 min. Fig. 5b shows an example chromatogram of LNP sample These two HPLC methods can quantitate siRNA amount in LNP sam-
solution without TritonTM X100 using the optimized RPLC condi- ples with adequate accuracy, sensitivity and precision, and reduce
tions and all the system peaks were eluted within 3.5 min, reducing the testing time significantly.
the run time to 4 min. The resolution between siRNA and the clos-
est peaks is larger than 2.0 in both chromatograms shown in Fig. 5 Acknowledgement
which is sufficient for accurate quantitation.
The authors gratefully acknowledge the valuable insights pro-
3.3. Method validation and applications vided by Sherif Badawy and Rao Mantri of Bristol Myers Squibb Co.
in their review of this article.
Several key parameters including specificity, linear, accuracy,
precision, sensitivity and solution stability were evaluated for both References
method conditions. Specificity was demonstrated by the absence of
a siRNA peak in the sample chromatograms of a placebo LNP solu- [1] C. Mello, D. Conte, Revealing the world of RNA interference, Nature 431
(2004) 338–342.
tion. Moreover, apart from the siRNA peak, no other peaks in the [2] D.P. Bartel, MicroRNAs: genomics, biogenesis, mechanism, and function, Cell
sample chromatograms exhibited absorbance at 260 nm (typical 116 (2004) 281–297.
maximum absorbance of oligonucleotides) which further verified [3] J.C. Burnett, J.J. Rossi, RNA-based therapeutics: current progress and future
prospects, Chem. Biol. 19 (2012) 60–71.
method specificity. The method sensitivity of 0.025 ␮g/mL was
[4] K. Tatiparti, S. Sau, S.K. Kashaw, A.K. Iyer, siRNA delivery strategies: a
obtained in the presence of LNP for both methods. The results of the comprehensive review of recent developments, Nanomaterials 7 (2017) 2–17.
total siRNA content determination of various samples in different [5] Y. Suzuki, H. Ishihara, Structure, activity and uptake mechanism of
sample diluents demonstrated adequate accuracy and robustness siRNA-lipid nanoparticles with an asymmetric ionizable lipid, Int. J. Pharm.
510 (2016) 350–358.
for these two techniques. The validation results for all parameters [6] G. Cavallaro, C. Sardo, E.F. Craparo, B. Porsio, G. Giammona, Polymeric
are consistent with ICH validation criteria [13] as shown in Table 1, nanoparticles for siRNA delivery: production and applications, Int. J. Pharm.
indicating the methods have acceptable analytical performance. 525 (2017) 313–333.
[7] K.B. Sentell, J.G. Dorsey, Retention mechanisms in reversed-phase
With the short run time, chromatographic simplicity and ability chromatography, J. Chromatogr. 4610 (1989) 193–207.
to handle different sample diluents, these methods offer increased [8] J.G. Dorsey, K.A. Dill, The molecular mechanism of retention in
laboratory efficiency when supporting LNP process development reversed-phase liquid chromatography, J. Chromatogr. 4610 (1989) 193–207.
[9] Y. Brun, P. Alden, Gradient separation of polymers at critical point of
and scale up efforts compared to traditional fluorescent based adsorption, J. Chromatogr. A 966 (2002) 25–40.
assays. [10] H.J.A. Philipsen, B. Klumperman, A.M. Herk, A.L. German, Critical retention
behaviour of polymers. A study on the influence of some practical parameters,
J. Chromatogr. A 727 (1996) 13–25.
4. Conclusions [11] M. Janco, D. Sykora, F. Svec, J.M.J. Frechet, J. Schweer, R. Holm, Rapid
determination of molecular parameters of synthetic polymers by
Analytical approaches to determine siRNA content in LNP drug precipitation/redissolution high-performance liquid chromatography using
“molded” monolithic column, J. Polym. Sci. Part A 38 (2000) 2767–2778.
product samples have been studied. Two user-friendly and iso- [12] Z. Huang, Y. Hu, Y. Zha, R. Perrone, J. Ruan, W.P. Fish, Determination of
cratic HPLC conditions with short run times were developed using polymeric impurities in asunaprevir drug substanceand product using size
reversed phase and cation exchange columns with a mobile phase exclusion effect of reversed-phase columns, J. Pharm. Biomed. Anal. 151
(2018) 200–208.
containing 20–30% acetonitrile. The siRNA variants were excluded
[13] International Conference on Harmonization, Guideline for validation of
from pores in stationary phase and eluted as a single peak before analytical procedures: text and methodology Q2(R1), in: International
solvent front. It was observed that the separation of siRNA peak Conference on Harmonization of Requirements for Registration of
from other components was impacted by the pore size of substract Pharmaceuticals for Human Use, 2006.

particles rather than the type and particle size of stationary phases.