Identifying Factors of Microparticles Modified with ArginineDerivatives That Induce Phenotypic Shifts in MacrophagesAlexander J. Dunn-Sale† and Kaitlin M. Bratlie*,†,‡,§

†Department of Chemical & Biological Engineering and ‡Department of Materials Science & Engineering, Iowa State University,Ames, Iowa 50011, United States§Ames National Laboratory, Ames, Iowa 50011, United States

*S Supporting Information

ABSTRACT: Macrophages are key players in the progression of many diseases,ranging from rheumatoid arthritis to cancer. Drug delivery systems have thepotential not only to transport payloads to diseased tissue but also to influencecell behavior. Here, poly(N-isopropylacrylamide-co-acrylic acid) (pNIPAm-co-AAc) microparticles were modified with 14 different arginine derivatives. Theseparticles were then incubated with interleukin-4 or lipopolysaccharide-stimulatedmacrophages or naive macrophages (RAW 264.7). The phenotypic state of themacrophages was assessed by measuring arginase activity, tumor necrosis factor-α(TNF-α) secretion, and nitrite production. Partial least-squares analysis revealedmaterial properties and descriptors that shifted the macrophage phenotype for thethree cell conditions in this study. Material descriptors relating to secondarybonding were suggested to play a role in shifting phenotypes in all threemacrophage culture conditions. These findings suggest that macrophageresponses could be altered through drug delivery vehicles, and this method could be employed to assist in screening potentialcandidates.

KEYWORDS: macrophage phenotypes, TNF-α, polymer properties

1. INTRODUCTION

Material properties are known to alter cellular responses.1−8 Inparticular, surface modifications can alter cytokine andmolecular secretion profiles.6,8−11 In the case of macrophages,changes in the amount of secreted molecules can signal a shiftin their phenotype. Macrophages exist on a spectrum, with oneend being pro-inflammatory, denoted M1, and the other beingpro-angiogenic, termed M2. Alternatively, these cells can bethought of as occupying a color wheel, as proposed by Mosserand Edwards,12 in which the different activations overlap witheach other, blending into “shades” of macrophage phenotypes.Newer nomenclature has been proposed by Murray et al.13 tomore accurately characterize the in vitro state of themacrophage in which the state is characterized by the molecularactivator. The pro-angiogenic macrophages can be activated byexposure to interleukin-4, M(IL-4), and the pro-inflammatorycells can be achieved through incubation with lipopolysacchar-ide M(LPS). The exact material properties that promotepolarization to specific phenotypes remain poorly defined.Tumor-associated macrophages (TAMs) can exist anywhere

on the spectrum between M1- and M2-like macrophages.However, they are typically polarized toward an M2-like state.14

The phenotype is currently thought to change from an M1state during cancer initiation to the predominant M2phenotype during tumor progression,15,16 creating a permissiveenvironment for the tumor that encourages angiogenesis,matrix remodeling, and metastasis. Altering the phenotype of

these cells can be achieved in vitro and in vivo through proteindelivery. Systemic delivery of proteins is generally not feasibledue to short half-lives and a low maximum tolerated dose of500 ng/kg in the case of IL-12.17 These high doses led to thedevelopment of a local delivery system consisting of the proteinencapsulated in biodegradable polylactic acid to polarize M2TAMs to an M1 phenotype.18,19 On the other side of this coin,excessive M1-like macrophage presence is implicated ininflammatory diseases like arthritis.20,21 Using microspheresto alter macrophage phenotype in combination with deliveringchemo- or immunochemotherapies may improve outcomes.Arginine is generally metabolized by two enzymes: nitric

oxide synthase (NOS), which produces citrulline and reactivenitrogen intermediates, and arginase, which produces ornithineand urea.22,23 These enzymes are differently up-regulated inpolarized mouse macrophages, with NOS being dominant inM1 cells and arginase in M2 macrophages.24 Many researchersused the arginase/NOS ratio to describe the phenotypicpopulation of macrophages.25−28 Several of the intermediateproducts in the conversion of arginine to nitric oxide or ureahave been linked to diseases related to inflammation, whichimplicate macrophages in these diseases. Cardiomyopathy29 hasbeen associated with carnitine deficiency. Supplementation with

Received: January 22, 2016Accepted: May 13, 2016Published: May 13, 2016

Article

pubs.acs.org/journal/abseba

© 2016 American Chemical Society 946 DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

creatine has improved outcomes for patients with Huntington’sdisease30 and type 2 diabetes.31 At the other end of thespectrum, nitroarginine exhibits neuroprotective effects inParkinson’s disease.32 These are molecules that are inter-mediate products in the conversion of arginine to nitric oxideor urea. Theses diseases are related to inflammation, whichimplicates macrophages in these diseases. Using arginine and itsderivatives to alter the phenotypic profile of macrophages hasnot been explored, particularly in using these molecules assurface modifiers.In this study, we have examined the effect of arginine

derivatives as polymer modifiers in macrophage response.Arginine is known to have a role in immune responses.22

Furthermore, several of its derivatives are known to impactNOS production and function. Both 2-amino-3-guanidinopro-pionic acid and nitroarginine inhibit NOS.33,34 Carnitinesuppresses the production of NO and the protein expressionof inducible nitric oxide synthase (iNOS).35 In addition tothese molecules, several products in arginine metabolism werestudied, including creatine, which is produced througharginine/glycine amidinotransferase22,23 and citrulline. Otherarginine derivatives are thought to be pro-angiogenic, such as 3-guanidinopropionic acid, which shifts mitochondrial metabo-lism from a glycolytic to oxidative mechanism,36 andacetylcarnitine, which protects against oxidative stress gen-erated by mitochondrial metabolism.37 Acetylcarnitine has alsobeen used to decrease iNOS protein expression in multiplesclerosis patients when delivered orally.38 Additional argininederivatives were explored to examine how these modifiers altermacrophage responses. In addition to M(LPS) and M(IL-4)cells, naive macrophages were also tested. Correlations betweenthe molecules detected and material properties were examinedand used to determine what parameters promote differentmacrophage phenotypes under different activation conditions.

2. MATERIALS AND METHODSFourteen chemically unique functional groups were coupled topolymers to assess how they affect macrophage activation. Allexperiments had at least four replicates, and error bars indicate thestandard deviation. All materials were purchased from Sigma (St.Louis, MO) and used as received unless otherwise indicated. Freshdeionized water (Milli-Q, Barnstead Nanopure, Thermo Scientific,Waltham, MA) was used throughout this study.Modifiers that were coupled to p(NIPAm-co-AAc) particles were 2-

amino-3-guanidinopropionic acid, 3-guanidinopropionic acid, nitro-

arginine, creatine (Fisher, Pittsburgh, PA), carnitine, citrulline, 5-hydroxylysine, acetylglutamine, N-carbamyl-α-aminoisobutyric acid,acetylcarnitine, 2,4-diaminobutyric acid, acetylornithine, albizziin, andarginine (Amresco, Solon, OH).

2.1. Polymer Synthesis. 2.1.1. p(NIPAm-co-AAc) Particle Syn-thesis. Modification of p(NIPAm-co-AAc) particles and their synthesishave been described previously.5,8 Briefly, NIPAm (2.4 g), N,N-methylenebis(acrylamide) (0.16 g), and 157 μL of AAc (J.T. Baker,Center Valley, PA) were dissolved in 100 mL of H2O and stirredunder N2 in a 250 mL round-bottom flask for 30 min at 70 °C. Next,15 mL of 13.3 mg/mL K2S2O8 was added to the flask. The reactionwas allowed to proceed for 3.5 h, after which the suspension wasslowly cooled to room temperature, filtered with P5 grade filter paper,and dialyzed for 48 h in Milli-Q water. The particles were freeze-driedusing a lyophilizer (Labconco, Kansas City, MO, 4.5 L).

In a 15 mL tube, 7.5 mL of phosphate buffered saline (PBS, dilutedfrom a 10× solution, Fisher Scientific, to 0.1 M, pH 7.4), 16.7 μL ofethylenediamine (Eda), 1.5 mL of 5% w/v p(NIPAm-co-AAc)particles, and 75 mg of 1-ethyl-3(3-(dimethylamino)propyl)-carbodiimide hydrochloride (EDC) were vortexed and incubatedovernight at room temperature. The particles were dialyzed againstH2O and lyophilized. The modifiers in Figure 1 were conjugated to theEda−p(NIPAm-co-AAc) particles through the carboxylic acid groupon the modifiers in Figure 1 by combining 10 mg of Eda−p(NIPAm-co-AAc) particles with 2 mg of modifiers and 10 mg of EDC in 2 mL ofPBS at room temperature and stirring overnight. The particles weredialyzed against H2O and lyophilized.

2.1.2. ζ-Potential. Milli-Q water was neutralized to pH 7 with HClor NaOH to ensure that the ions in water would not interfere with theζ-potential of particles. A low salt concentration was confirmed bymeasuring conductivities at <0.2 mS/cm. A 100 μL aliquot of 1% w/vof particles was added into 5 mL of water, and ζ-potential wasmeasured with a Zetasizer Nano Z (Malvern).

2.2. Cell Viability. RAW 264.7 macrophages were cultured at 37°C with 5% CO2 in 10% fetal bovine serum, 100 U/L penicillin, and100 μg/L streptomycin in Dulbecco’s modified Eagle’s medium(DMEM high glucose; Thermo Scientific), to be referred to ascomplete medium (CM). RAW 264.7 cells were seeded in a 24-wellplate at 1.25 × 105 cell/cm2 in the presence of 5 μg/mL LPS39,40 or 25ng/mL IL-441 (eBioscience Inc., San Diego, CA) and incubated 24 h at37 °C in 5% CO2. A control set of experiments was not activated. Afteractivation, the medium was replaced with fresh CM, and particles wereadded to the wells (100 particles per cell). The cells were incubated foran additional 24 h. All particles were sterilized by washing three timesin 70% ethanol, followed by three washes in sterile H2O andcentrifugation at 10 000g for 10 min. In a control experiment, cellswere incubated without particles. A control of particle and IL-4 or LPSin the absence of cells was also conducted. The medium in each wellwas collected and stored at −20 °C.

Figure 1. Chemical structures of all molecules used for the modification of p(NIPAm-co-AAc) particles. The letters are used as labels in the followingfigures for convenience.

ACS Biomaterials Science & Engineering Article

DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

947

A (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide(MTT) assay was used to determine cell viability in the presence ofthe modified particles. A total of 50 μL of MTT (5 mg/mL in DIH2O) and 500 μL of CM was added to each well, and the plate wasincubated for 2 h at 37 °C in 5% CO2, after which time 425 μL ofmedia was aspirated. The resulting formazan crystals were dissolved in500 μL of dimethylsulfoxide (DMSO, Fisher, Pittsburgh, PA). Theabsorbance was measured at 540 nm with a reference at 690 nm usinga plate reader (BioTek Synergy HT Multidetection Microplate Reader,Winooski, VT).2.3. Measurement of Tumor Necrosis Factor-α Production.

Tumor necrosis factor-α (TNF-α) present in the LPS, IL-4, or naivecell supernatant was determined by commercially available immuno-assay kits (eBioscience Inc., San Diego, CA) and performed asdescribed by the manufacturer.2.4. Arginase Activity. After 24 h incubation with particles in

section 2.2, cells were washed with 500 μL of PBS. In each well, 100μL of cell lysis buffer (150 μL protease inhibitor cocktail (Amresco,Solon, OH) and 15 μL of Triton X-100 (Acros Organics) filled to 15mL with DI water) was added, and the plates were incubated on ice for10 min. Next, 25 μL of the cell lysate was transferred to a 96-well plate,and 25 μL of 10 mM MnCl2 (Fisher) and 50 mM Tris (Fisher) wasadded to each well. Incubation at 55 °C for 10 min activated thearginase. A solution of 50 μL of 1 M arginine (pH 9.7) was added toeach well, and the plate was incubated at 37 °C for 20 h. The activityof arginase was determined by measuring the amount of arginineconverted to urea. In 96-well plates, 5 μL of either the urea-containingsamples or standards was added to 200 μL of a 1:2 mixture of solution1 (1.2 g of o-phthaldialdehyde (Alfa Aesar, Ward Hill, MA), 1 L ofH2O, and 500 μL of HCl (Fisher, Pittsburgh, PA)) and solution 2 (0.6g of N-(napthyhl)ethylenediamine dihydrochloride (Acros Organics),5 g of boric acid (Fisher, Pittsburgh, PA), 800 mL of H2O, and 111 mLof sulfuric acid (Fisher, Pittsburgh, PA), diluted to 1 L with H2O).

42

The plates were incubated for 30 min, and the absorbance in each wellwas measured at 540 nm with a reference at 630 nm using a platereader.2.5. Griess Reagent Assay. Nitrite concentration in the

supernatant collected in section 2.2 was quantified through a Griessreagent assay. A standard curve was generated through serial dilutionsof 100 μM NaNO2. To each well were added 150 μL of sample orstandard, 130 μL of DI H2O, and 20 μL of Griess reagent. The platewas incubated for 20 min, and the absorbance of each well wasmeasured with a plate reader at 448 nm with a reference of 690 nm.TNF-α, arginase, and nitrite concentrations were determined at thesame time point of 24 h after introduction of the particles.2.6. Statistics and Data Analysis. Statistical analysis was

performed using XLSTAT statistical software (New York, NY).Statistical significance of the mean comparisons was determined by atwo-way ANOVA. Pair-wise comparisons were analyzed with Tukey’shonest significant difference test. Differences were consideredstatistically significant for p < 0.05. The macrophage responsesmeasured above were analyzed by performing partial least-squares(PLS) regression on the material descriptors shown in Table 1. Thesedescriptors were based on descriptors defined by Bicerano.43

3. RESULTS AND DISCUSSION

3.1. Synthesis, ζ-Potential, and Cell Viability. Function-alized p(NIPAm-co-AAc) particles were synthesized andmodified with the molecules shown in Figure 1 through thecarboxylic acid on the modifier. The ζ-potential of the modifiedand unmodified p(NIPAm-co-AAc) is in Figure 2. There was arange in ζ-potentials for the particles from −4.2 to −20.8 mV.Cytocompatibility of these materials was tested through anMTT viability assay. Macrophages were activated with LPS orIL-4 to simulate M1- and M2-like cells. After being incubatedfor 24 h, the medium was replaced and the functionalizedparticles were added to each well. Metabolic activity of the cellswas assessed after an additional 24 h incubation period. Naive

cells, denoted M(0), were also assessed. The results of theviability assay are shown in Figure 3. All tested conditions were>70% viable, with most particles and activation conditionsresulting in 90% viability of the macrophages. The largestdecrease in viability came from B, D, G, I, and J for M(LPS), Hfor M(IL-4), and L for M(0).

Table 1. Thirty-Eight Molecular Descriptors for the SurfaceModifications Used in This Studya

IDstructuredescriptor descriptions

1 R freely rotating bonds

2 HD H-bond donors

3 HA H-bond acceptors

4 Nsp2 number of sp2 carbon atoms

5 N1°C Number of 1° carbon atoms

6 1X connectivity index 1, calculated based on the molecular graphof the modifier

7 1Xv connectivity index 2, calculated based on the electronicconfiguration of the molecular graph of the modifier

8 0X atomic index 1, calculated based on the number of non-hydrogen atoms to which a non-hydrogen atom is bound

9 0Xv atomic index 2, calculated based on the electronicconfiguration of each non-hydrogen atom

10 1ξintensive connectivity index 1, ξ =

N1 X1

11 1ξvintensive connectivity index 2, ξ =ν ν

N1 X1

12 0ξintensive atomic index 1, ξ =

N0 X0

13 0ξvintensive atomic index 2, ξ =ν ν

N0 X0

14 N number of non-hydrogen atoms

15 NC number of carbon atoms

16 NH number of hydrogen atoms

17 NO number of oxygen atoms

18 NN number of nitrogen atoms

19 NCH2 number of CH2

20 NK NK = 5Namide + 4Nhydroxyl − 3Nether − 5NCC + 3Nsulfone

21 Ndc Ndc = 19NN + 12N(side group O, −S−) + 52Nsulfone − 14Ncyc

22 Ngroup Ngroup = 12Nhydroxyl + 12Namide + 2N(non‑amide −(NH)− unit) −N(alkyl ether −O−) − NCC + 4N(non‑amide −(CO)− next to a nitrogen)+ 7N(−(CO)− in carboxylic acid, ketone, or aldehyde) +2N(other −(CO)−)

23 NMV NMV = −5Nsulfone + 3Nester + 5NCC − 11Ncyc

24 NHη NHη = −7Namide + 6Nether + 5Nsulfone +3N(six‑membered nonterminal aromatic rings) + 8NCC + 4 N4° carbon atoms+ 2N3° carbon atoms

25 NUR NUR = −2NOH − 3Ncyc + 4N(six‑membered aromatic rings)

26 Nds Nds = 2NCH2− 7NCC + 3N(six‑membered aromatic rings) − 12NCO2

27 NPS NPS = −3Nsp3carbon atoms + 6Ncarbonyl groups

28 NvdW NvdW = Nmenonarb + Nalamid

c + NOH − 4Ncyc + 2NCC

29 Vw Vw = 3.861803°X + 13.7484351Xv, van der Waals volume

30 ξNOH ξ = + −N N NNNOH

0.125N O H

31 M molecular weight

32 α polarizability

33 V molar volume

34 log p octanol−water partition ratio

35 log D(pH 5.5) distribution coefficient at pH 5.5

36 log D(pH 7.4) distribution coefficient at pH 7.4

37 ΔHv enthalpy of vaporization

38 γ surface tensionaThe structure descriptors are defined by Bicerano.43 bNmenonar is thenumber of methyl groups attached to nonaromatic atoms. cNalamid isthe total number of linkages between amide and nonaromatic atoms.

ACS Biomaterials Science & Engineering Article

DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

948

3.2. TNF-α Secretion from Activated and NaiveMacrophages. Activated and naive macrophages wereexposed to functionalized p(NIPAm-co-AAc) particles, andthe concentration of secreted TNF-α was measured (Figure4A). LPS was used to stimulate that M1-like phenotype. Thecontrol in which no particles were added to the cells contained4.03 ± 0.51 ng/mL. This was statistically higher (p < 0.05) thanthe unmodified p(NIPAm-co-AAc) particles and the particlesmodified with ethylenediamine of 2.49 ± 0.29 and 2.55 ± 0.25ng/mL, respectively. Compared to both the unmodified andEda-modified particles, modifications D and I−M resulted instatistically higher TNF-α secretion (p < 0.05). Of these,albizziin increased TNF-α secretion ∼2-fold. Modifications B,E, G, and H decreased the TNF-α production (p < 0.05). Themost significant decrease was observed for 3-guanidinopro-pionic acid of ∼40%. Details of multiple comparisons for thesedata are in Table S1.

To explore the M2-like phenotype, macrophages werestimulated with IL-4. The control without particles resultedin 2.69 ± 0.08 ng/mL TNF-α measured in the supernatant.The unmodified p(NIPAm-co-AAc) and Eda-modified particlesresulted in a decrease in TNF-α by 1.78 ± 0.13 and 1.68 ± 0.12ng/mL, respectively (p < 0.05), similar to the M(LPS) cells.Modifications A, I−K, and N resulted in increased TNF-αsecretion (p < 0.05), while modifications B, C, E, F, and Hresulted in a decrease (p < 0.05) compared to the unmodifiedand Eda particles. Modifications C−G and K resulted inincreases in TNF-α secretion, and particles modified with I, J,L, and M decreased the TNF-α concentration detectedcompared to the unmodified and Eda particles. 2-Amino-3-guanidinopropionic caused the largest increase in TNF-αsecretion of ∼200%, and 3-guanidinopropionic acid caused themost significant decrease of ∼40%. Details of multiplecomparisons for these data are in Table S2.Macrophages without activation were also incubated with the

modified particles, and the secreted TNF-α was assessed. Thecontrol, unmodified p(NIPAm-co-AAc), and Eda particles were

Figure 2. ζ-Potentials for modified p(NIPAm-co-AAc) particles. Datarepresent the mean value of three replicates for each sample ±standard deviation. Eda = ethylenediamine, pN = unmodifiedp(NIPAm-co-AAc).

Figure 3. Viability assay showing the cytotoxicity of all particles withLPS or IL-4 activated or naive RAW 264.7 cells. The % viability wasnormalized to RAW 264.7 cells treated with LPS, IL-4, or naive cellswithout particles. Data represent the mean value of four replicates foreach sample ± standard deviation. Eda = ethylenediamine, pN =unmodified p(NIPAm-co-AAc).

Figure 4. Molecular expression of (A) TNF-α and (B) urea/nitrite bymacrophages in respose to functionalized p(NIPAm-co-AAc) particlesand stimulation with LPS or IL-4. Naive cells are also shown. Datarepresent the mean value of four replicates for each sample ± standarddeviation. Eda = ethylenediamine, pN = unmodified p(NIPAm-co-AAc), PC = positive control of cells without materials.

ACS Biomaterials Science & Engineering Article

DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

949

not statistically different with 1.04 ± 0.11, 1.31 ± 0.11, and 1.12± 0.13 ng/mL TNF-α detected. The largest increase in TNF-αwas found for modifications F and K, with an increase of∼350%. The largest decrease was observed for modifications I,J, and L, resulting in an 80% decrease in TNF-α compared withthe unmodified or Eda particles. Details of multiplecomparisons are in Table S3.Taken together, modifications I, J, and K lead to increases in

TNF-α for both M(LPS) and M(IL-4), while modifications B,E, and H lead to decreases in TNF-α secretion. The mostdramatic decreases in TNF-α secretion for both M(LPS) andM(IL-4) were in the presence of the 3-guanidinopropionic acidmodification, which reduced TNF-α secretion by ∼40%compared to unmodified and Eda-modified particles.3.3. Urea/Nitrite for Activated and Naive Macro-

phages. Arginase activity was measured by exposing celllysate to arginine and assessing its conversion to urea after 20 hincubation. Increased arginase activity is suggestive of an M2-like phenotype. Increased iNOS indicates an M1-likephenotype. The activity of iNOS was indirectly measuredthrough a Griess assay in which nitrites, the stable form of NO,are quantified. NO results from reactive nitrogen intermediates,which are synthesized by iNOS. Arginine is generallymetabolized through one of these pathways, thus the ratio ofurea/nitrite can assess the extent of polarization toward eitherphenotype.25,26 These results are shown in Figure 4B. Thenitrite and urea concentrations are shown as Figure S1A,B.M(IL-4) cells were found to have 447 ± 68 mg/μmol urea/

nitrite. Cells incubated with A−D and I−L increased theamount of urea/nitrite measured (p < 0.05), while particlesmodified with E−H decreased the levels of urea/nitrite (p <0.05). Measured levels of urea/nitrite ranged from 252 to 1023mg/μmol. There was no statistical difference between theM(IL-4) cells in the absence of particles and M(LPS) cellsincubated with unmodified and Eda-modified p(NIPAm-co-AAc) particles: 421 ± 100 and 366 ± 70 mg/μmol urea/nitrite,respectively. Details of multiple comparisons for urea/nitrite forM(IL-4) are in Table S4.M1-like cells were also studied using M(LPS)-stimulated

cells. No statistical differences were observed between thecontrol and cells incubated with unmodified and Eda-modifiedparticles: 62 ± 5, 69 ± 6, and 70 ± 7 mg/μmol urea/nitrite,respectively. Particles E−H, K, and M resulted in elevated urea/nitrite levels. Only particles A and I decreased urea/nitrite (p <0.05) compared to the positive control, unmodified, and Eda-modified particles. Details of multiple comparisons for urea/nitrite for M(LPS) are provided in Table S5. In comparingtrends between M(LPS) and M(IL-4), only modification Kresulted in increasing urea/nitrite levels and no modificationswere able to decrease urea/nitirite in both M(LPS) and M(IL-4) cells. It is important to note that these results may beparticular to murine macrophage cells. Human macrophagestransport arginine through system y+L transporters,44 whereas acationic amino acid transporter has been suggested as anarginine transporter in mice.45

Naive macrophages were also tested for urea/nitrite levels.The levels of nitrites detected for all M(0) cells were lower thanthe standard curve. Statistical analysis on these values is notreasonable.3.4. Material Parameters That Influence Macrophage

Phenotypic Shifts. Secretion of TNF-α in the absence ofparticles followed the expected trend, in which the highestlevels were observed for M(LPS), followed by M(IL-4) and

M(0).6,46−48 With the exception of modification A, M(LPS)cells secreted more TNF-α than the M(IL-4) cells when theywere incubated with particles. The M(0) cells exhibited a rangeof responses to the particles, secreting less TNF-α in thepresence of modifications A, C, J, Eda, and unmodifiedp(NIPAm-co-AAc) than M(LPS) cells. Increased TNF-α wasobserved for modification M compared to M(LPS) cells. Theincrease in TNF-α secretion may be a result of LPS tolerancedeveloped in the M(LPS) cells, which leads to a decreasedproduction of TNF-α.49 Another possibility is that IL-10, asuppressor of TNF-α, may be secreted. Previous findings havedemonstrated a decrease of TNF-α secretion in M(LPS) cellsby a factor of 21.4 ± 2.5 in the presence of 0.1 ng/mL IL-10.50

Arginase/iNOS is a measure of macrophage polarization,25,26

with high levels denoting M2-like states and low levelsindicating a more M1-like state. Here, we measure this usingurea/nitrite levels. In the case of M(IL-4), both the control andthe cells incubated with particles resulted in higher urea/nitritelevels, which is expected.The aim of this work was to identify material properties that

influence M1- and M2-like phenotypes. The trends betweenmaterial properties and cell responses were quantitativelymodeled using material descriptors listed in Table 1. Thesedescriptors were based largely on structural descriptors43 alongwith physiochemical properties. PLS was performed on themeasured data using principle component analysis values ofFigure 4A,B as a function of the molecular descriptors in Table1. The results of the model are presented in Figure 5, where thehorizontal axis represents TNF-α secretion and the vertical axisis urea/nitrite. Cells that secrete high levels of TNF-α andexhibit low levels of urea/nitrite are M1-like, while the converseare M2-like. Both of these regions are outlined in Figure 5. Inaddition, the specific descriptors that promote either of thesetwo extremes are labeled on the plots. For M(IL-4) cells(Figure 5A), increasing H-bond donors promoted a shifttoward an M1-like phenotype, while increasing H-bondacceptors, Ndc, log D (5.5), and log D (7.4) retained the M2-like phenotype. Log D is the octanol−water distributioncoefficient, which can be measured at any pH, and Ndc is acorrection factor for the dielectric constant. It bears mentioningthat the inverse of these material descriptors would result in ashift toward the opposite phenotype. Based on theseparameters, polarity of the molecule appears to be an importantfactor in altering M(IL-4) polarization. It should be noted thatsurface tension had a negligible effect on either TNF-α or urea/nitrite.A larger number of material properties were identified as

altering M(LPS), shown in Figure 5B. Increasing the number offreely rotating bonds, aliphatic carbons, van der Waalsinteractions (calculated using NνdW), the number of carbonand hydrogen atoms, and the enthalpy of vaporizationpromotes an M2-like phenotype. On the other end of thespectrum, primary carbon atoms, the number of sp2 carbonatoms, NMVa correction term for molar volume, NURacorrection term for molar Rao function, ξNOHa correctionterm for thermal conductivity, log p, log D (7.4), and log D(5.5) promoted the M1-like phenotype. All of these parameterssuggest that hydrophobicity plays a dominant role in alteringM(LPS) cells.Since the levels of nitrites detected for all M(0) cells were

lower than the standard curve, material properties influencingphenotypic shifts were not analyzed because statistical analysison these values is not reasonable.

ACS Biomaterials Science & Engineering Article

DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

950

Interesting similarities exist for the material properties thatpromote phenotypic shifts in all the conditions examined here.Descriptors related to hydrophobicity, van der Waalsinteractions, and H-bonding were found to be related tomacrophage polarization for M(LPS) and M(IL-4). Theconventional dogma surrounding ligand−receptor bindinginvolves all of these interaction forces additively combining tostabilize the ligand−receptor pair.51 Very likely, serum proteinspresent in the culture media adsorb in different configurationson the particle surface, thus enabling different interactions withsurface receptors. The complexity of macrophage activation andreprogramming through surface receptors prevents predictingwhich receptors these particles are interacting with; however,examining the impact of material descriptors on alteringmacrophage response has the potential to improve materialsystems for implanted biomaterials as well as drug deliverysystems.Based on the observations described above and in Figure 5,

three molecules were selected to alter the macrophagephenotype: lysine, ornithine, and ureidopropionic acid. Thesemolecules were conjugated to Eda-modified particles asdescribed in section 2.1.1. The concentration of secreted

TNF-α and urea/nitrite in response to incubation with themodified particles is shown in Figure 6.

Ureidopropionic acid was predicted to promote urea/nitriteand TNF-α secretion in M(IL-4) cells and decrease secretion inM(LPS). Factors that would increase urea/nitrite and TNF-αsecretion in M(IL-4) cells would be increased distributioncoefficients at pH 5.5 and 7 and a reduction in H-bondacceptors. Compared to albizziin, ureidopropionic acid hasfewer H-bond acceptors and higher distribution coefficients atpH 5.5 (−4.0 vs −2.62, respectively). The distributioncoefficients at pH 7 are decreased for ureidopropionic acid(−4.4 vs −4.01). The distribution coefficients at pH 5.5 and 7are very similar when compared to those of albizziin tocitrulline, ergo an increase in urea/nitrite and TNF-α secretionwould be expected for ureidopropionic acid compared tocitrulline, as is observed in Figure 5A,B, particularly sincecitrulline has an additional H-bond donor compared toureidoprionic acid, much like albizziin. In the case of M(LPS)cells, factors that cause cells to secrete decreased levels of urea/nitrite are the distribution coefficients at pH 5.5 and 7. In the

Figure 5. Analysis of molecular descriptors to identify polymercharacteristics that promote M1- or M2-like profiles. Pearson’scorrelation between molecular descriptors that promote urea/nitrite(y-axis) and TNF-α (x-axis) for (A) IL-4 and (B) LPS. Relevantdescriptors that promote M1- or M2-like responses are labeled on thefigure.

Figure 6. Molecular expression of (A) TNF-α and (B) urea/nitrite bymacrophages in respose to functionalized p(NIPAm-co-AAc) particlesand stimulation with LPS or IL-4. Naive cells are also shown. Datarepresent the mean value of four replicates for each sample ± standarddeviation.

ACS Biomaterials Science & Engineering Article

DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

951

case of M(LPS) cells, the higher distribution coefficient forureidopropionic acid compared to that of either albizziin orcitrulline would cause a decrease in the urea/nitrite and TNF-αlevels, as was observed. The number of freely rotating bonds isalso lower for ureidopropionic acid compared to albizziin andcitrulline.The difference between lysine and ornithine is an additional

aliphatic CH2 group in lysine, which also results in an additionalfreely rotating bond for lysine. The other parametersinfluencing urea/nitrite production are similar for the twomodifiers. This resulted in predictions that lysine would resultin increased nitrite/urea secretion compared to ornithine. Theadditional CH2 group results in a 2-fold increase in urea/nitritefor the two modifiers in the case of M(LPS) cells. In examiningthe response of M(IL-4) cells, no difference between lysine-and ornithine-modified particles was expected for urea/nitriteand TNF-α levels since the differences in their structures didnot result in differences in the material descriptors identified inFigure 5A,B.

4. CONCLUSIONSIn summary, we have examined how a library of 14 surfacemodifiers alters macrophage response and have correlated thoseresponses to material descriptors that promote shifts inphenotypes. Similarities between all three macrophage treat-ment conditions exist in that intermolecular nonbondinginteractions play an important role in altering macrophageresponses. The differences between these three conditionssuggest the possibility that material properties can be exploitedto manipulate macrophage responses, possibly reprogrammingthese cells without exogenous protein delivery. Semiquantita-tive methods such as those employed in this study have thepotential to both improve screening for materials that promoteM1- or M2-like phenotypes in macrophages and add to thecurrent library of knowledge pertaining to how materialsinfluence cellular responses.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acsbiomater-ials.6b00041.

Details of multiple comparisons for TNF-α for M(LPS),M(IL-4), and M(0) cells and for urea/nitrite for M(LPS)and M(IL-4) cells; nitrite and urea measured forM(LPS), M(IL-4), and M(0) cells (PDF)

■ AUTHOR INFORMATIONCorresponding Author*Tel: 515-294-7304. Fax: 515-294-5444. E-mail: kbratlie@iastate.edu.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the National Science Foundationunder Grant No. CBET 1227867 and the Roy J. CarverCharitable Trust Grant No. 13-4265. The authors alsoacknowledge support from NSF ARI-R2 (CMMI-0963224)for funding the renovation of the research laboratories used forthese studies.

■ REFERENCES(1) Acharya, A. P.; Dolgova, N. V.; Clare-Salzler, M. J.; Keselowsky,B. G. Adhesive substrate-modulation of adaptive immune responses.Biomaterials 2008, 29, 4736−50.(2) Schutte, R. J.; Xie, L.; Klitzman, B.; Reichert, W. M. In vivocytokine-associated responses to biomaterials. Biomaterials 2009, 30,160−8.(3) Anderson, J. M.; Rodriguez, A.; Chang, D. T. Foreign bodyreaction to biomaterials. Semin. Immunol. 2008, 20, 86−100.(4) Nilsson, B.; Ekdahl, K. N.; Mollnes, T. E.; Lambris, J. D. The roleof complement in biomaterial-induced inflammation. Mol. Immunol.2007, 44, 82−94.(5) Wang, D.; Phan, N.; Isely, C.; Bruene, L.; Bratlie, K. M. Theeffect of polymer chemistry and macrophage phenotype on particleinternalization. Biomacromolecules 2014, 15, 4102−10.(6) Wang, D.; Bratlie, K. M. Influence of Polymer Chemistry onCytokine Secretion from Polarized Macrophages. ACS Biomater. Sci.Eng. 2015, 1, 166−74.(7) Bygd, H. C.; Akilbekova, D.; Munoz, A.; Forsmark, K. D.; Bratlie,K. M. Poly-l-arginine based materials as instructive substrates forfibroblast synthesis of collagen. Biomaterials 2015, 63, 47−57.(8) Bygd, H. C.; Forsmark, K. D.; Bratlie, K. M. Altering in vivomacrophage responses with modified polymer properties. Biomaterials2015, 56, 187−97.(9) Petersen, L. K.; Xue, L.; Wannemuehler, M. J.; Rajan, K.;Narasimhan, B. The simultaneous effect of polymer chemistry anddevice geometry on the in vitro activation of murine dendritic cells.Biomaterials 2009, 30, 5131−42.(10) Ariganello, M. B.; Simionescu, D. T.; Labow, R. S.; Lee, J. M.Macrophage differentiation and polarization on a decellularizedpericardial biomaterial. Biomaterials 2011, 32, 439−49.(11) Rodriguez, A.; Meyerson, H.; Anderson, J. M. Quantitative invivo cytokine analysis at synthetic biomaterial implant sites. J. Biomed.Mater. Res., Part A 2008, 89, 152−9.(12) Mosser, D. M. The many faces of macrophage activation. J.Leukocyte Biol. 2003, 73, 209−12.(13) Murray, P. J.; Allen, J. E.; Biswas, S. K.; Fisher, E. A.; Gilroy, D.W.; Goerdt, S.; et al. Macrophage Activation and Polarization:Nomenclature and Experimental Guidelines. Immunity 2014, 41, 14−20.(14) Mantovani, A.; Locati, M. Tumor-associated macrophages as aparadigm of macrophage plasticity, diversity, and polarization: lessonsand open questions. Arterioscler., Thromb., Vasc. Biol. 2013, 33, 1478−83.(15) Biswas, S. K.; Sica, A.; Lewis, C. E. Plasticity of MacrophageFunction during Tumor Progression: Regulation by Distinct MolecularMechanisms. J. Immunol. 2008, 180, 2011−7.(16) Qian, B.-Z.; Pollard, J. W. Macrophage diversity enhances tumorprogression and metastasis. Cell 2010, 141, 39−51.(17) Gollob, J A.; Mier, J. W.; Veenstra, K.; McDermott, D. F.;Clancy, D.; Clancy, M.; et al. Phase I trial of twice-weekly intravenousinterleukin 12 in patients with metastatic renal cell cancer or malignantmelanoma: ability to maintain IFN-gamma induction is associated withclinical response. Clin. Cancer Res. 2000, 6, 1678−92.(18) Hill, H.; Conway, T.; Sabel, M.; Jong, Y.; Mathiowitz, E.;Bankert, R.; et al. Cancer immunotherapy with interleukin 12 andgranulocyte-macrophage colony-stimulating factor-encapsulated mi-crospheres: coinduction of innate and adaptive antitumor immunityand cure of disseminated disease. Cancer Res. 2002, 62, 7254−63.(19) Watkins, S. K.; Egilmez, N. K.; Suttles, J.; Stout, R. D. IL-12rapidly alters the functional profile of tumor-associated and tumor-inflitrating macrophages in vitro and in vivo. J. Immunol. 2007, 178,1357−62.(20) Hofkens, W.; Schelbergen, R.; Storm, G.; van den Berg, W. B.;van Lent, P. L. Liposomal Targeting of Prednisolone Phosphate toSynovial Lining Macrophages during Experimental Arthritis InhibitsM1 Activation but Does Not Favor M2 Differentiation. PLoS One2013, 8, e54016.

ACS Biomaterials Science & Engineering Article

DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

952

(21) Jain, S.; Tran, T.-H.; Amiji, M. Macrophage repolarization withtargeted alginate nanoparticles containing IL-10 plasmid DNA for thetreatment of experimental arthritis. Biomaterials 2015, 61, 162−77.(22) Bronte, V.; Zanovello, P. Regulation of immune responses by L-arginine metabolism. Nat. Rev. Immunol. 2005, 5, 641−54.(23) Morris, S. M. Arginine metabolism: boundaries of ourknowledge. J. Nutr. 2007, 137, 1602S−1609S.(24) Mantovani, A.; Sica, A.; Sozzani, S.; Allavena, P.; Vecchi, A.;Locati, M. The chemokine system in diverse forms of macrophageactivation and polarization. Trends Immunol. 2004, 25, 677−86.(25) Lumeng, C. N.; Bodzin, J. L.; Saltiel, A. R. Obesity induces aphenotypic switch in adipose tissue macrophage polarization. J. Clin.Invest. 2007, 117, 175−84.(26) Allen, J. E.; Loke, P. Divergent roles for macrophages inlymphatic filariasis. Parasite Immunol. 2001, 23, 345−52.(27) Murphy, B. S.; Sundareshan, V.; Cory, T. J.; Hayes, D.; Anstead,M. I.; Feola, D. J. Azithromycin alters macrophage phenotype. J.Antimicrob. Chemother. 2008, 61, 554−60.(28) Rath, M.; Muller, I.; Kropf, P.; Closs, E. I.; Munder, M.Metabolism via Arginase or Nitric Oxide Synthase: Two CompetingArginine Pathways in Macrophages. Front. Immunol. 2014, 5, 1−10.(29) Amat di San Filippo, C.; Taylor, M. R. G.; Mestroni, L.; Botto,L. D.; Longo, N. Cardiomyopathy and carnitine deficiency. Mol. Genet.Metab. 2008, 94, 162−6.(30) Ferrante, R.; Andreassen, O.; Jenkins, B.; Dedeoglu, a;Kuemmerle, S.; Kubilus, J.; et al. Neuroprotective effects of creatinein a transgenic mouse model of Huntington’s disease. J. Neurosci. 2000,20, 4389−97.(31) GUALANO, B.; DE SALLES PAINNELI, V.; ROSCHEL, H.;ARTIOLI, G. G.; NEVES, M.; DE SA PINTO, A. L.; et al. Creatine inType 2 Diabetes. Med. Sci. Sports Exercise 2011, 43, 770−8.(32) Arimoto, T.; Bing, G. Up-regulation of inducible nitric oxidesynthase in the substantia nigra by lipopolysaccharide causes microglialactivation and neurodegeneration. Neurobiol. Dis. 2003, 12, 35−45.(33) Grant, S. K.; Green, B. G.; Stiffey-Wilusz, J.; Durette, P. L.;Shah, S. K.; Kozarich, J. W. Structural requirements for humaninducible nitric oxide synthase substrates and substrate analogueinhibitors. Biochemistry 1998, 37, 4174−80.(34) Jang, J. J.; Ho, H. K.; Kwan, H. H.; Fajardo, L. F.; Cooke, J. P.Angiogenesis is impaired by hypercholesterolemia: role of asymmetricdimethylarginine. Circulation 2000, 102, 1414−9.(35) Koc, a; Ozkan, T.; Karabay, a Z; Sunguroglu, a; Aktan, F. Effectof L-carnitine on the synthesis of nitric oxide in RAW 264·7 murinemacrophage cell line. Cell Biochem Funct 2011, 29, 679−85.(36) Oudman, I.; Clark, J. F.; Brewster, L. M. The Effect of theCreatine Analogue Beta-guanidinopropionic Acid on Energy Metab-olism: A Systematic Review. PLoS One 2013, 8, e52879.(37) Hagen, T. M.; Ingersoll, R. T.; Wehr, C. M.; Lykkesfeldt, J.;Vinarsky, V.; Bartholomew, J. C.; et al. Acetyl-L-carnitine fed to oldrats partially restores mitochondrial function and ambulatory activity.Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 9562−6.(38) Calabrese, V.; Scapagnini, G.; Ravagna, a.; Bella, R.; Butterfield,D a.; Calvani, M.; et al. Disruption of Thiol Homeostasis andNitrosative Stress in the Cerebrospinal Fluid of Patients with ActiveMultiple Sclerosis: Evidence for a Protective Role of Acetylcarnitine.Neurochem. Res. 2003, 28, 1321−8.(39) Ricote, M.; Li, A. C.; Willson, T. M.; Kelly, C. J.; Glass, C. K.The peroxisome proliferator-activated receptor-gamma is a negativeregulator of macrophage activation. Nature 1998, 391, 79−82.(40) Guastadisegni, C.; Minghetti, L.; Nicolini, A.; Polazzi, E.; Ade,P.; Balduzzi, M.; et al. Prostaglandin E2 synthesis is differentiallyaffected by reactive nitrogen intermediates in cultured rat microgliaand RAW 264.7 cells. FEBS Lett. 1997, 413, 314−8.(41) Motran, C. C.; Díaz, F. L.; Gruppi, A.; Slavin, D.; Chatton, B.;Bocco, J. L. Human pregnancy-specific glycoprotein 1a (PSG1a)induces alternative activation in human and mouse monocytes andsuppresses the accessory cell-dependent T cell proliferation. J.Leukocyte Biol. 2002, 72, 512−21.

(42) Levinson, S. S. Kinetic centrifugal analyzer and manualdetermination of serum urea nitrogen, with use of o-phthaldialdehydereagent. Clin. Chem. 1978, 24, 2199−202.(43) Bicerano, J. Prediction of Polymer Properties; Marcel Dekker Inc.:New York, 2002.(44) Rath, M.; Muller, I.; Kropf, P.; Closs, E. I.; Munder, M.Metabolism via Arginase or Nitric Oxide Synthase: Two CompetingArginine Pathways in Macrophages. Front. Immunol. 2014, 5, 1−10.(45) Manner, C. K.; Nicholson, B.; MacLeod, C. L. CAT2 argininetransporter deficiency significantly reduces iNOS-mediated NOproduction in astrocytes. J. Neurochem. 2003, 85, 476−82.(46) Akilbekova, D.; Philiph, R.; Graham, A.; Bratlie, K. M.Macrophage reprogramming: influence of latex beads with variousfunctional groups on macrophage phenotype and phagocytic uptake invitro. J. Biomed. Mater. Res., Part A 2015, 103, 262.(47) Barth, K a.; Waterfield, J. D.; Brunette, D. M. The effect ofsurface roughness on RAW 264.7 macrophage phenotype. J. Biomed.Mater. Res., Part A 2013, 101, 2679−88.(48) Varin, A.; Mukhopadhyay, S.; Herbein, G.; Gordon, S.Alternative activation of macrophages by IL-4 impairs phagocytosisof pathogens but potentiates microbial-induced signalling and cytokinesecretion. Blood 2010, 115, 353−62.(49) Docke, F. R. W.; Bundschuhrt, D. S.; Hartung, T.; Wendel, A.;Volk, H. In vitro prevention and reversal of lipopolysaccharidedesensitization by IFN-gamma, IL-12, and granulocyte-macrophagecolony-stimulating factor. J. Immunol. 1997, 158, 2911−8.(50) Bogdan, C.; Vodovotz, Y.; Nathan, C. Macrophage deactivationby interleukin 10. J. Exp. Med. 1991, 174, 1549−55.(51) Seong, S.-Y.; Matzinger, P. Hydrophobicity: an ancient damage-associated molecular pattern that initiates innate immune responses.Nat. Rev. Immunol. 2004, 4, 469−78.

ACS Biomaterials Science & Engineering Article

DOI: 10.1021/acsbiomaterials.6b00041ACS Biomater. Sci. Eng. 2016, 2, 946−953

953