Inhibiting sialidase-induced _TGF-b1 activation attenuates pulmonary fibrosis in mice

Article abstract of DOI:10.1124/jpet.120.000258

The active form of transforming growth factor-b1 (TGF-b1) plays a key role in potentiating fibrosis. TGF-b1 is sequestered in an inactive state by a latency-associated glycopeptide (LAP). Sialidases (also called neuraminidases (NEU)) cleave terminal sialic acids from glycoconjugates. The sialidase NEU3 is upregulated in fibrosis, and mice lacking Neu3 show attenuated bleomycin-induced increases in active TGF-b1 in the lungs and attenuated pulmonary fibrosis. Here we observe that recombinant human NEU3 upregulates active human TGF-b1 by releasing active TGF-b1 from its latent inactive form by desialylating LAP. Based on the proposed mechanism of action of NEU3, we hypothesized that compounds with a ring structure resembling picolinic acid might be transition state analogs and thus possible NEU3 inhibitors. Some compounds in this class showed nanomolar IC₅₀ for recombinant human NEU3 releasing active human TGF-b1 from the latent inactive form. The compounds given as daily 0.1–1-mg/kg injections starting at day 10 strongly attenuated lung inflammation, lung TGF-b1 upregulation, and pulmonary fibrosis at day 21 in a mouse bleomycin model of pulmonary fibrosis. These results suggest that NEU3 participates in fibrosis by desialylating LAP and releasing TGF-b1 and that the new class of NEU3 inhibitors are potential therapeutics for fibrosis.

Full text of DOI:10.1124/jpet.120.000258

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Inhibiting sialidase-induced TGF-β1 activation attenuates pulmonary fibrosis in mice
1
2
1
Tejas R. Karhadkar , Thomas D. Meek , and Richard H. Gomer
1
Department of Biology,
2
Department of Biochemistry and Biophysics,
Texas A&M University
3
01 Old Main Drive, College Station, Texas, 77843-3474 USA

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
a) Running title: Sialidase-induced TGF-β1 activation and fibrosis
b) Corresponding author:
Richard H. Gomer
Department of Biology
Texas A&M University
3
01 Old Main Drive
College Station, TX 77843-3474 USA
Phone: 979-458-5745. Fax: 979-845-2891. email: rgomer@tamu.edu
c) Number of text pages: 47
Number of tables: 1
Number of figures: Main figures - 7, Supplementary figures - 10
Number of References: 105
Number of words in Abstract: 199
Number of words in Introduction: 730
Number of words in Discussion: 606
d) Abbreviations:
AMPCA – 4-amino-1-methyl-2-piperidinecarboxylic acid
DANA – 2,3-didehydro-2-deoxy-N-acetyl-neuraminic acid
2

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
ER - endoplasmic reticulum
LAP - latency associated glycopeptide
L-TGF-β1 - latent transforming growth factor-β1
MP - Methyl picolinate
PBMC - peripheral blood mononuclear cells
TGF-β1 - transforming growth factor-β1
4
2
MU-NANA - 2′-(4-methylumbelliferyl)-α-D-N-acetylneuraminic acid sodium salt hydrate
AP – 2 acetylpyridine
e) Recommended section: 1) Drug Discovery and Translational Medicine
2
3
) Inflammation, Immunopharmacology, and Asthma
) Cellular and Molecular
3

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Abstract
The active form of transforming growth factor-β1 (TGF-β1) plays a key role in
potentiating fibrosis. TGF-β1 is sequestered in an inactive state by a latency associated
glycopeptide (LAP). Sialidases (also called neuraminidases) cleave terminal sialic acids from
glycoconjugates. The sialidase NEU3 is upregulated in fibrosis, and mice lacking Neu3 show
attenuated bleomycin-induced increases in active TGF-β1 in the lungs, and attenuated pulmonary
fibrosis. Here we observe that recombinant human NEU3 upregulates active human TGF-β1 by
releasing active TGF-β1 from its latent inactive form by desialylating LAP. Based on the
proposed mechanism of action of NEU3, we hypothesized that compounds with a ring structure
resembling picolinic acid might be transition state analogues and thus possible NEU3 inhibitors.
Some compounds in this class showed nanomolar IC50s for recombinant human NEU3 releasing
active human TGF-β1 from the latent inactive form. The compounds given as daily 0.1 mg/kg –
1
mg/kg injections starting at day 10 strongly attenuated lung inflammation, lung TGF-β1
upregulation, and pulmonary fibrosis at day 21 in a mouse bleomycin model of pulmonary
fibrosis. These results suggest that NEU3 participates in fibrosis by desialylating LAP and
releasing TGF-β1, and that the new class of NEU3 inhibitors are potential therapeutics for
fibrosis.
Significance statement:
The extracellular sialidase NEU3 appears to be a key driver of pulmonary fibrosis. The
significance of this report is that we show 1) the mechanism (NEU3 desialylates the LAP protein
that keeps the profibrotic cytokine TGF-β1 in an inactive state, causing active TGF-β1 release),
2
) we then use the predicted NEU3 mechanism to identify nM IC50 NEU3 inhibitors, and 3)
these new NEU3 inhibitors are potent therapeutics in a mouse model of pulmonary fibrosis.
4

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Introduction
Fibrosis involves scar-like tissue forming in an internal organ leading to organ failure
Martinez et al., 2017). Pulmonary fibrosis has an incidence of 1 in 400 in the elderly, and there
(
are no FDA-approved therapeutics that stop the progression of this disease (Raghu et al., 2011;
Wynn and Ramalingam, 2012; Raghu and Selman, 2015). Transforming growth factor-β1 (TGF-
β1) in its active mature form is a key factor which potentiates fibrosis (Fernandez and
Eickelberg, 2012; Wolters et al., 2014). TGF-β1 is synthesized as a precursor protein made of
latency-associated peptide (LAP) and the active TGF-β1 in a complex called latent-TGF-β1
(LTGF-β1) (Hinck et al., 2016; Robertson and Rifkin, 2016). Two dimerized TGF-β1 precursor
proteins are processed to cleave active TGF-β1 from the LAP, with the LAP remaining
essentially wrapped around, and sequestering, the TGF-β1 (Robertson and Rifkin, 2016). Like
many other secreted proteins, LAP is glycosylated (Lewandrowski et al., 2006; Moremen et al.,
2
012; Travis and Sheppard, 2014; Hinck et al., 2016; Robertson and Rifkin, 2016) .
Several things can release active TGF-β1 from sequestration by the LAP. Both the
integrin αvβ6, which generates mechanical traction on the LAP, and thrombospondin-1 can
release active TGF-β1 (Crawford et al., 1998; Munger et al., 1999). Proteases such as plasmin,
thrombin, MMP-2, and MMP-9, and reactive oxygen species- mediated conformational changes
in LAP, can also release TGF-β1 from the LAP (Lyons et al., 1988; Pociask et al., 2004; Jobling
et al., 2006; Jenkins, 2008) However, it is unclear if these are the only mechanisms that cause
upregulation of active TGF-β1 in fibrotic lesions.
5

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Neuraminic acid (also known as sialic acid) is a sugar found at the terminal ends of some
glycoproteins (Varki and Gagneux, 2012; Schwab and Nimmerjahn, 2013; Varki et al., 2015).
The presence or absence of sialic acid on these glycoconjugates has a central role in various
cellular regulatory mechanisms (Varki and Varki, 2007; Zhou et al., 2020). For instance, adding
sialic acids to the glycan residues on immunoglobulin G (IgG) confers anti-inflammatory
functions to IgG (Anthony and Ravetch, 2010). Conversely, the loss of sialic acid from the
plasma glycoprotein Serum amyloid P (SAP) attenuates the potency of SAP to inhibit the
differentiation of some human peripheral blood mononuclear cells (PBMC) to fibrosis-associated
fibroblast-like cells called fibrocytes (Cox et al., 2015).
Neuraminidases (also known as sialidases) are enzymes that catalyze a hydrolysis
reaction which removes terminal sialic acids from glycoconjugates (Miyagi et al., 2012;
Pshezhetsky and Ashmarina, 2013; Smutova et al., 2014). There are four known mammalian
sialidases (Miyagi et al., 2012; Smutova et al., 2014). Of the four, the sialidase NEU3 is
localized in endosomal structures and on the extracellular side of the plasma membrane, and
under some conditions can also be released by cells, and thus can desialylate glycoproteins and
glycolipids present outside the cell (Kopitz et al., 2001; Zanchetti et al., 2007; Miyagi and
Yamaguchi, 2012; Mozzi et al., 2015; Rodriguez-Walker and Daniotti, 2017). Sialidases,
especially NEU3, are upregulated in pulmonary fibrosis (Lambré et al., 1988; Karhadkar et al.,
2
017; van de Vlekkert et al., 2019). TGF-β1 upregulates NEU3 in human lung epithelial cells,
pulmonary fibroblasts, small airway epithelial cells, and PBMC (Karhadkar et al., 2017). In the
epithelial cells, TGF-β1 increases levels of NEU3 by decreasing the NEU3 degradation rate and
increasing NEU3 mRNA translation (Chen et al., 2020). Conversely, NEU3 upregulates TGF-β1
6

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
in human PBMCs and human pulmonary fibroblasts (Karhadkar et al., 2017). Intratracheal
delivery of bleomycin upregulates TGF-β1 in the lungs (Cutroneo et al., 2007; Yue et al., 2010),
and is a standard model of pulmonary fibrosis in mice (Blackwell et al., 2013; Moore et al.,
2
013). Mice lacking NEU3, or wild-type mice treated with sialidase inhibitors, do not upregulate
TGF-β1 in the lungs after bleomycin treatment, and show attenuated bleomycin-induced
pulmonary fibrosis (Karhadkar et al., 2017; Karhadkar et al., 2020). Together, these observations
suggested that a positive feedback loop where TGF-β1 upregulates NEU3, and NEU3
upregulates TGF-β1, may potentiate fibrosis (Karhadkar et al., 2017; Karhadkar et al., 2020).
How NEU3 upregulates TGF-β1 is unknown.
Bacterial and viral sialidases can desialylate LTGF-β1, leading to the release of active
mature TGF-β1, possibly explaining why there are elevated levels of active TGF-β1 levels
during infection with sialidase-expressing pathogens (Miyazono and Heldin, 1989; Miyazono et
al., 1991; Schultz-Cherry and Hinshaw, 1996; Carlson et al., 2010; Gratz et al., 2017). There
have been many previous attempts to develop inhibitors for viral (von Itzstein, 2007) or human
(
Cairo, 2014) sialidases. Examples of viral sialidase inhibitors include Tamiflu (oseltamivir) and
Relenza (zanamivir), which can attenuate the progression of influenza in patients (Hayden et al.,
997; Monto et al., 1999; Mäkelä et al., 2000; Moscona, 2005). Potent inhibitors for viral
neuraminidases exert poor inhibition for human neuraminidases (Hata et al., 2008; Albohy et al.,
011a). There are a variety of inhibitors for human sialidases, but all have published IC50s of ~10
1
2
µM for NEU3, making them relatively weak inhibitors (Potier et al., 1979; Magesh et al., 2006;
Albohy et al., 2010; Zou et al., 2010; Cairo, 2014; Guo et al., 2018).
7

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
In this report, we find that NEU3 desialylates LAP to release active TGF-β1. We then
identified a new class of potent NEU3 inhibitors, some with IC50s in the nM range. These
compounds inhibit bleomycin induced pulmonary fibrosis in mice, further supporting the idea
that NEU3 inhibitors may be potential therapeutics for fibrosis.
8

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Materials and methods
Cell isolation and culture
Human peripheral blood was collected from healthy volunteers who gave written consent
and with specific approval from the Texas A&M University human subjects review board. All
methods were performed in accordance with the relevant guidelines and regulations. Blood
collection, isolation of PBMC, and cell culture were done as described previously (Pilling et al.,
2
009a; Pilling et al., 2009b).
Mouse NEU3 expression in HEK 293-f cells
HEK 293 freestyle cells (Life Technologies, Grand Island, NY) were cultured in
5
FreeStyle 293 media (12338-018, Life Technologies). 1 x 10 cells were mixed with 2 μg of 100
μg/ml of murine Neu3 expression clone (MR223297, Origene, Rockville, MD) in 100 μl PBS
(GE Lifesciences, Marlborough, MA), and were transfected by electroporation using a 4D-
Nucleofector System (Lonza, Walkersville, MD) following the manufacturer’s protocol. The
transfected cells were kept at room temperature for 15 minutes for recovery, after which the cells
2
were cultured in 25 ml Freestyle 293 media in a humidified incubator at 37 °C with 5% CO .
After 24 hours, 400 μg/ml of G418 (345812, Calbiochem EMD, San Diego, CA) was added to
select for transfected cells. After 10 days, the cells were isolated, lysed, and c-Myc tagged
recombinant mouse NEU3 was purified using a Myc-Trap_A kit (ytak-20, Chromotek,
Hauppauge, NY) following the manufacturer’s protocol. The eluted protein was stored in 50 μl
of 10% glycerol, 100 mM glycine, 25 mM Tris-HCl pH 7.3, at 4 °C.
Human NEU3 expression in E. coli
9

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
The human NEU3 cDNA sequence was amplified from a human NEU3 ORF clone
(RC216537, Origene). The amplified NEU3 sequence was inserted between the NdeI and EcoRI
sites of the pMAL-c5X vector (N8108S, NEB, Ipswich, MA). The resulting construct encoding a
fusion of maltose-binding protein (MBP) and NEU3 was transformed into chemically competent
E. cloni cells (Lucigen, Middleton, WI) mixed in 37 °C pre-warmed Luria-Bertani (LB) broth
(#240230, BD, Franklin Lakes, NJ), grown with shaking at 37 °C, 240 rpm for 1 hour and plated
on LB agar with ampicillin (100 μg/ml). Clones were selected and grown in 100 μg/ml
ampicillin LB broth medium, and the plasmids were isolated using a GeneJET Plasmid Miniprep
Kit (#K0502, Thermo Scientific, Waltham, MA). Clones were verified by sequencing (Eton
Bioscience, San Diego, CA). The sequence verified clones were transformed into chemically
competent BL21(DE3) E. coli cells (Lucigen), grown with shaking at 37 °C, 240 rpm, for 1 hour
in pre-warmed LB broth, and plated on LB agar with 100 μg/ml ampicillin. Individual colonies
were selected and added to 10 ml pre-warmed LB broth with 100 μg/ml ampicillin and grown at
3
7 °C, 240 rpm, for no longer than 16 hours. The culture was added to 1 L of LB broth and
incubated at 37 °C, 240 rpm until OD600 ~0.6. To induce protein expression, isopropyl β- d-1-
thiogalactopyranoside (IPTG, Goldbio, St. Louis, MO) solution was added to 0.5 mM, and
incubated at 20 °C at 240 rpm overnight. Protein purification was done as described previously
(Albohy et al., 2010) with the following modifications. Bacterial cells were harvested after 16 –
1
8 hours of induction by centrifugation at 4000 × g at 4 °C. The pellet was resuspended in
resuspension buffer (20 mM Tris, pH 7.2, 300 mM NaCl, 1 mM EDTA and 0.1% Triton X-100)
supplemented with 1x protease inhibitor (PIA32965, VWR, Radnor, PA). The lysate was
sonicated three times for three minutes with the tube submerged in ice at setting 16 of a
Microson ultrasonic cell disruptor (Misonix Incorporated, Farmingdale, NY), and then clarified
1
0

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
by centrifugation at 10,000 × g for 20 minutes at 4 °C. The supernatant was loaded onto an
amylose column (New England Biolabs) equilibrated with 20 mM Tris/ 300 mM NaCl, pH 7.2.
MBP-NEU3 was eluted with running buffer containing 10% glycerol (v/v) and 20 mM maltose.
Sialidase effects on LAP assay
To determine the effect of sialidases on latency associated peptide (LAP), 400 ng/ml
human recombinant latency-associated peptide (LAP), (246-LP/CF, R&D Systems, Minneapolis,
MN) was incubated with and without 200 ng/ml of human recombinant sialidases NEU1
(TP300386, Origene), NEU2 (TP319858, Origene), NEU3 (TP316537, Origene), or NEU4
(TP303948, Origene), or C. perfringens neuraminidase (N2876-2.5UN, Sigma-Aldrich, St.
Louis, MO) in a total volume of 20 μl of 1x phosphate buffered saline (PBS; #SH30256.01, GE
Lifesciences) pH 6.9. The PBS was adjusted to pH 6.9 with 12N HCl (H613-05, Avantor
Performance Materials, Radnor Township, PA). The reaction mixtures were incubated at 37 °C
for 2 hours. Following incubation, 5 μl of 4x Laemmli’s buffer (GTX16355, GeneTex, Irvine,
CA) containing 20 mM dithiothreitol was added to the reaction mixtures, and these were heated
at 100 °C for 5 minutes. 15 μl of the heated reaction mixture was electrophorized on a 4-20%
Tris/glycine polyacrylamide pre-cast gel (12001-058, VWR, Radnor, PA) in Tris/glycine/SDS
running buffer (13.5 g Tris base, 7.2 g glycine, 5 g SDS / liter) at room temperature for 90–120
minutes.
Western blots, and staining the blots with lectin, was done as previously described (Cox
et al., 2015), and immunostaining western blots for NEU3 was done as previously described
(Karhadkar et al., 2017; Karhadkar et al., 2020). Silver staining of gels was done following
1
1

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
(Morrissey, 1981). Staining intensity was quantified by Image Lab software (Bio Rad, Hercules,
CA).
NEU3 induced L-TGF-β1 activation assay
To determine if NEU3 can cause latent transforming growth factor-β1 (L-TGF-β1) to
release active TGF-β1, 200 ng/ml recombinant human latent transforming growth factor-β1 (L-
TGF-β1) (299-LT/CF, R&D Systems) was incubated with varying concentrations of sialidases
with or without 1 mM calcium chloride in a total volume of 100 µl of PBS pH 6.9 in a 96-well
microplate. The microplates were covered with aluminum foil and incubated at 37 °C for 2
hours. Following incubation, the reaction mixtures were assayed using an active TGF-β1 ELISA
kit (DY240, R&D Systems) following the manufacturer’s protocol with the exception that the
reaction mixtures were not processed for acid treatment, so as to measure only active TGF-β1
and not total TGF-β1. The absorbance was read with a SynergyMX plate reader (BioTek,
Winooski, VT). To determine the enzyme kinetics of NEU3 with L-TGF-β1 as a substrate,
varying concentrations of L-TGF-β1 were incubated with 100 ng/ml of recombinant human
NEU3 (Origene), in a total volume of 100 μl, and the samples were processed as described
above.
Sialidase activity assay with 2′-(4-methylumbelliferyl)-α-D-N-acetylneuraminic acid sodium salt
hydrate (4MU-NANA)
Sialidase assays were done following (Potier et al., 1979; Marathe et al., 2013) with the
following modifications. PBS was adjusted to pH 6.9 with 12N HCl and bovine serum albumin
(BSA, VWR) was added to a final concentration of 100 μg/ml (PBS-BSA). Varying
1
2

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
concentrations of bacteriailly expressed recombinant human NEU3 were prepared in PBS-BSA.
A final concentration of 200 μM 4MU-NANA (M8639, Sigma-Aldrich) was added to the NEU3
in a total volume of 100 μl in a 96-well plate. Where indicated, inhibitors were added for 30
minutes before adding the 4MU-NANA. The reactions were incubated at 37 °C for 30 minutes,
and the fluorescence was measured at 37 °C in a prewarmed SynergyMX plate reader with
excitation at 360 nm and emission at 440 nm. The fluorescence in the absence of sialidases was
subtracted from all readings.
Inhibition of NEU3 induced L-TGF-β1 activation
2
,3-didehydro-2-deoxy-N-acetyl-neuraminic acid (DANA, #252926-10MG) was
purchased from EMD-Millipore (Burlington, MA), Oseltamivir phosphate (Tamiflu, ) was
purchased from (#1479304) was purchased from Sigma, (St. Louis, MO), 2-acetyl pyridine
(
2AP, sc-254121) was purchased from Santa Cruz Biotechnology (Dallas, TX) and methyl
picolinate (MP, sc-228575) was purchased from Santa Cruz. 2-piperidinecarboxylic acid, 4-
amino-1-methyl (AMPCA, #A00285-13785-026) was synthesized by Sundia (Shanghai, China)
following the scheme shown in Supplementary Fig. 1. These compounds were dissolved in water
to 20 mM. All stocks were stored at 4 °C and used within 2 weeks of preparation. Dilution series
of the compounds were made in PBS pH 6.9. 100 µl of diluted compound was added to the well
of a 96 well plate and then 50 µl of 400 ng/ml recombinant human sialidase NEU3
(TP316537,Origene) in PBS pH 6.9 was added to each well (50 µl PBS pH 6.9 for the control),
and the plate was incubated for 30 minutes at 37 °C. 50 µl of 800 ng/ml L-TGF-β1 (299-LT/CF,
R&D Systems) in PBS pH 6.9 was then added to the well. The plate was covered with aluminum
1
3

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
foil and incubated for 2 hours at 37 °C, and released active TGF-β1 was assayed as described
above.
Inhibition of NEU3 induced IL-6 accumulation
Dilutions of the above compounds were made in serum-free medium prepared as previously
described (Pilling et al., 2009b; Karhadkar et al., 2020), and 50 µl of diluted compound was
added to the well of a 96 well plate and then 50 µl of 400 ng/ml sialidase in serum-free medium
was added to each well (50 µl serum-free medium for control), and the plate was incubated for
5
3
0 minutes at 37 °C in a humidified incubator with 5% CO
2
. 100 µl of PBMC at 1 x 10 cells/ml
4
was then added to the well to make the total volume in a well 200 μl with 1 x 10 cells/well, and
2
this was incubated at 37 °C in a humidified incubator with 5% CO for 24 or 48 hours. IL-6 in
the culture supernatant was then assayed with an IL-6 ELISA kit (430501, BioLegend, San
Diego, CA) following the manufacturer’s protocol, reading absorbance with a SynergyMX plate
reader (BioTek). At least three different donors were used for each assay.
Mouse model of pulmonary fibrosis
To determine if the NEU3 inhibitors affect pulmonary fibrosis in mice, 8-10 week old
2
6–30 g male C57BL/6 mice (Jackson Laboratories, Bar Harbor, ME) were given an
oropharyngeal aspiration of 3 units/kg bleomycin (2246-10, BioVision Incorporated, Milpitas,
CA) in 50 µl of 0.9% saline to induce symptoms of pulmonary fibrosis or oropharyngeal saline
alone as a control as previously described (Pilling and Gomer, 2014; Pilling et al., 2014; Cox et
al., 2015). Starting 10 days after bleomycin had been administered, some of the bleomycin-
treated mice were given daily intraperitoneal injections of 100 μl of PBS or 1 mg/kg 2-acetyl
1
4

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
pyridine, 1 mg/kg methyl picolinate, or 0.1 mg/kg of 2-piperidinecarboxylic acid, 4-amino-1-
methyl in 100 µl PBS. All mice were weighed daily for 21 days after bleomycin was
adminstered. At day 21, mice were euthanized by CO inhalation, and bronchoalveolar lavage
2
(BAL) fluid and cell-spots of BAL cells were obtained as previously described (Daubeuf and
Frossard, 2012; Pilling and Gomer, 2014; Pilling et al., 2014; Cox et al., 2015). The liver, heart,
kidneys, spleen, white fat tissue, and brown fat tissue were removed and weighed. The total cells
from BAL cell-spots were quantified as described previously (Daubeuf and Frossard, 2012;
Pilling and Gomer, 2014). The lungs were removed and inflated with Surgipath frozen section
compound (3801480, Leica, Buffalo Grove, IL) and preserved at -80 °C. 6–10 µm cryosections
of lungs were placed on glass slides (48311-703, VWR) and were air dried for 48 hours before
use. The mouse experiments were performed in accordance with the recommendations in the
Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The
Texas A&M University Animal Use and Care Committee approved the protocol (IACUC 2017-
0
414).
Histology, immunohistochemistry, and BAL analysis
Air dried cryosections were stained with hematoxylin and eosin or with Sirus red to
detect collagen, or with 1 μg/ml anti-active TGF-β1 antibodies (AB-100-NA, R&D Systems,
Minneapolis, MN) as described previously (Pilling et al., 2007; Karhadkar et al., 2017). The
staining was quantified with ImageJ (Schneider et al., 2012; Jensen, 2013).
Immunohistochemistry on BAL cell-spots and cryosections was performed as described
previously (Pilling and Gomer, 2014; Pilling et al., 2014; Cox et al., 2015; Karhadkar et al.,
2
020) using anti-CD11b (101202, clone M1/70 BioLegend) to detect blood and inflammatory
1
5

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
macrophages, anti-CD11c (M100-3, clone 223H7, MBL International, Woburn, MA) to detect
alveolar macrophages and dendritic cells, anti-CD45 (103102, clone 30-F11, BioLegend) for
total leukocytes, anti-Ly-6G (127602, clone 1A8, BioLegend) to detect neutrophils, and anti-Ly-
6
C (128001, clone HK1.4, BioLegend) to detect a variety of inflammatory immune system cells,
and with isotype-matched irrelevant antibodies (BioLegend) as controls. SDS-PAGE analysis of
BAL, and nanodrop protein assays, were performed as described previously (Pilling et al., 2003;
Karhadkar et al., 2017; Karhadkar et al., 2020). Western blots stained for albumin were done as
described previously (Karhadkar et al., 2020).
Hydroxyproline assay
Hydroxyproline assays were performed as described previously (Karhadkar et al., 2020)
with the exception that approximately half lobes of lungs frozen in OCT were cut off in three
pieces, thawed, and washed three times with 0.5 ml PBS to remove OCT by centrifugation at
2
000 × g for 5 minutes in pre-weighed Eppendorf tubes. After the last centrifugation, the tubes
were kept inverted for 5 minutes to allow PBS to blot onto blotting paper, and the tissue was then
weighed. Tissues were then processed using a hydroxyproline quantification kit (MAK008-1KT,
Sigma-Aldrich) following the manufacturer’s directions.
Statistics
Data were analyzed by t-test or ANOVA using Prism 7 (GraphPad, La Jolla, CA).
Significance was defined as p < 0.05. IC50 curves for enzyme assays were generated using the
Dose Response - Inhibition curve fitting model with variable slope with Prism 7. Based on
previously analyzed confidence intervals and historical data, and using G*Power power analysis
1
6

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
software (Faul et al., 2007; Faul et al., 2009), a sample size of 3 mice per group was determined
sufficient to detect the effect sizes anticipated in the present experiments.
1
7

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Results
Recombinant human NEU3 desialylates recombinant human LAP and releases active TGF-β1
Neuraminidases from influenza virus and bacteria remove sialic acid from the LAP
protein, which causes the LAP to release active TGF-β1 (Schultz-Cherry and Hinshaw, 1996;
Carlson et al., 2010; Gratz et al., 2017). To determine the effect of recombinant human sialidases
on recombinant human LAP (rhLAP), we incubated recombinant human sialidases (made in
HEK293 cells) or C. perfringens neuraminidase (Cox et al., 2015) with rhLAP. Western blots of
the rhLAP were stained with Sambucus nigra lectin to detect sialic acids (Shibuya et al., 1987;
Feng et al., 2013). NEU3, NEU4, and C. perfringens neuraminidase significantly decreased sialic
acid on rhLAP (Figs. 1A and B).
We previously found that the addition of recombinant human NEU3 (rhNEU3) to human
immune cells, pulmonary fibroblasts, and small airway epithelial cells causes them to upregulate
extracellular and intracellular TGF-β1 (Karhadkar et al., 2017). To determine if NEU3 causes
recombinant human latent TGF-β1 (rhL-TGF-β1) to release active TGF-β1, we incubated rhL-
TGF-β1 with rhNEU3, and measured released active TGF-β1. rhNEU3 did cause a release of
active TGF-β (Fig. 1C). In the presence or absence of calcium (which potentiates the activity of
some viral sialidases (Dimmock, 1971; Shtyrya et al., 2009)), rhNEU1 or rhNEU2 did not cause
significant release of active TGF-β1 (Supplementary Figs. 2 A and B). rhNEU3 and C.
perfringens neuraminidase released active TGF-β1, and NEU4 caused some release of TGF-β1
(Supplementary Figs. 2 A and B). Together, these results suggest that NEU3 removes sialic acid
from the LAP protein and efficiently releases active TGF-β1 from rhL-TGF-β1.
1
8

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
The standard assay for sialidases is cleavage of the fluorogenic substrate 2′-(4-
Methylumbelliferyl)-α-D-N-acetylneuraminic acid (4MU-NANA) (Potier et al., 1979). At
concentrations of rhNEU3 up to 100 ng/ml (1.94 nM), and using 4MU-NANA with no enzyme
as a control, we were unable to detect statistically significant activity of rhNEU3 on 4MU-
NANA. Compared to commercially available HEK-293 cell-synthesized rhNEU3, we were able
to purify much higher amounts of a bacterially expressed fusion of the 42 kDa maltose binding
protein with the 51 kDa human NEU3 (brhNEU3; Supplementary Figs. 3 A and B). The ~93 kDa
brhNEU3 contained a contaminant band that stained weakly for NEU3 on western blots
(
Supplementary Figs. 3A and B), so the concentration of bhrNEU3 was estimated by comparison
of the ~93 kDa band to BSA standards on silver-stained gels. The brhNEU3 caused a release of
active TGF-β1 from rhL-TGF-β1, albeit with somewhat lower efficacy than rhNEU3 (Figs. 1 C
and D). Compared to rhL-TGF-β1 as a substrate, a much higher concentration of brhNEU3 was
required to obtain detectable cleavage of 4MU-NANA (Figs. 1, C, D, and E).
A previous study with 4MU-NANA as the substrate for NEU3 reported a K
m
of 45 ± 3
-1
-1
m
μM and a kcat of 0.33 ± 0.02 min and thus a catalytic efficiency kcat / K of 0.007 ± 0.006 µM
-1
min (Albohy et al., 2010). Using rhL-TGF-β1 as the substrate for rhNEU3, and, measuring the
rate of TGF-β1 release for different concentrations of rhL-TGF-β1 (Figs. 1 F and G), we
-5
-1
observed a K
and thus a kcat / K
β1 compared to 4MU-NANA suggest a higher affinity of NEU3 for rhL-TGF-β1 compared to
MU-NANA as a substrate. Together, these results suggest that rhL-TGF-β1 can be used as a
better substrate than 4MU-NANA for a more sensitive assay to determine NEU3 activity.
m
of 1.5 ± 0.5 nM and a kcat of (4.0 ± 0.4) x 10 min (both mean ± SEM, n = 4),
-1
-1
m
m m
of 0.026 ± 0.008 µM min . The lower K and higher kcat / K for rhL-TGF-
4
1
9

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Design of NEU3 inhibitors
We designed possible inhibitors of human NEU3 based in part on the structure of 2,3-
didehydro-2-deoxy-N-acetyl-neuraminic acid (DANA) (Fig. 2A), a known inhibitor of NEU3 (Ki
2
=
30 μM) (Albohy et al., 2011b). DANA contains a double bond, of which the sp -hybridized
carbon-2 (circled) may be a structural mimic of the putative oxacarbenium ion intermediate
OCI) (Figs. 2, A and B) of the chemical mechanism of enzymatic catalysis (Albohy et al.,
010). As DANA lacks the positive charge of the OCI, a better mimic of this putative reaction
(
2
intermediate is likely afforded by the tetrahydropyridine-1-ium-2-carboxylate (THPC) (Fig. 2C)
2
which provides the positive charge with retention of sp -hybridized carbon-2. Consequently, we
2
pursued substituted picolinic acids, which contain both a cationic nitrogen atom and sp -
hybridization at carbon-2 (Fig. 2D), as well as a substituted pipecolinic acid (compounds 3-5 in
Table 1).
Compounds 3-5 inhibit NEU3 release of TGF-β1
To test the hypothesis that compounds 3-5 inhibit NEU3, we assayed their effects in the
rhNEU3 - rhL-TGF-β1 release of active TGF-β1 reaction (Table 1 and Supplementary Fig. 4).
Picolinates like 2-acetyl pyridine (2AP) and methyl picolinate (MP) had higher potencies of
inhibiting rhNEU3 activity compared to DANA and oseltamivir phosphate (Tamiflu) (Table 1,
Supplementary Figs. 4A – D and F – I), which supported our rationale for potential NEU3
inhibitors. The fully-saturated pipecolinic acid compound, 4-amino-1-methyl-2-piperidine
carboxylic acid (AMPCA), was also a very potent inhibitor of NEU3 (Table 1, Supplementary
2
Fig. 4 E and J) despite its lack of an adjacent sp -hybridized carbon as in the case of the
2
0

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
picolinate inhibitors. These results suggest the cationic nitrogen and an adjacent keto or
carboxylate group provided the essential structural elements to effect potent inhibition of
rhNEU3, possibly via mimicry of the enzymatic transition state structure, and that these
inhibitors are significantly improved over DANA and oseltamivir when rhL-TGF-β1 is the
substrate.
Compounds 3-5 inhibit NEU3-induced extracellular accumulation of IL-6 by human immune
cells
rhNEU3 upregulates extracellular accumulation of IL-6 from human PBMC (Karhadkar
et al., 2020). To determine if other sialidases have an effect on IL-6 production by PBMC, we
cultured human PBMC with recombinant human NEU1 – 4, and measured extracellular
accumulation of IL-6 in media supernatants after 24 and 48 hours. At 48 hours, compared to
controls, only rhNEU3 significantly upregulated extracellular accumulation of IL-6
(Supplementary Fig. 5). We measured the effects of inhibitors in a NEU3-induced IL-6 release
by human immune cells assay (Table 1 and Supplementary Fig. 6). DANA, Tamiflu, 2AP, and
MP all inhibited rhNEU3 in the cell-based NEU3-induced IL-6 release assay (Table 1,
Supplementary Fig. 6 A – D and F – I). Compared to the other compounds, the pipecolinate
AMPCA had an increased inhibition potency (Table 1, Supplementary Figs. 6 E and J). These
results suggest that some picolinates and a pipecolinate can be used as inhibitors against NEU3
in cell-based or in-vivo assays. The differences in inhibition potencies in the cell-based
extracellular IL-6 accumulation assay and in-vitro TGF-β1 based assay may be due to the
presence or absence of factors that interact with NEU3 and/or the inhibitors in the cell-based IL-
6
assay compared to the buffer-based TGF-β1 release assay.
2
1

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
NEU3 inhibitors inhibit mouse NEU3
To determine if compounds 3-5 would be effective in mice, we tested them against
recombinant mouse NEU3 in the NEU3-induced release of active TGF-β1 assay. The IC50 of
2
AP was 0.8 ± 1.7 µM, MP was 3.2 ± 1.6 nM, and AMPCA was less than 1 nM (Supplementary
Figs. 7 A – F). These results suggest that the three selected NEU3 inhibitors are inhibitors of
mouse NEU3.
AMPCA attenuates weight loss of mice after bleomycin treatment
To determine if NEU3 inhibitors affect bleomycin-induced pulmonary fibrosis, C57BL/6
mice were treated with an oropharyngeal aspiration of saline or bleomycin and then starting 10
days after saline or bleomycin, mice were given daily intraperitoneal injections of 1 mg/kg 2AP
or MP, or 0.1 mg/kg AMPCA. There was no discernable effect of any of the compounds on the
appearance or behavior of the mice. As previously observed, compared to saline treated-control
mice, bleomycin-treated control mice had lower body weights at day 21 after bleomycin
aspiration (Fig. 3A). In saline-treated mice, 2AP, MP, and AMPCA did not significantly affect
body weights. AMPCA attenuated the bleomycin-induced weight loss (Fig. 3A). 2AP, MP, and
AMPCA did not significantly affect liver, heart, kidneys, spleen, white fat or brown fat weights
as a percent of total body weight (Supplementary Figs. 8 A – F). There was no discernable effect
of any of the compounds on the appearance of any internal organs other than the lungs, which
had large white patches with bleomycin alone, and a normal pink color with the inhibitors
without bleomycin. With bleomycin, the lungs of mice treated with 2AP or AMPCA were a
2
2

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
normal pink color, and showed occasional small white patches in mice treated with MP.
Together these results suggest that AMPCA attenuates bleomycin-induced weight loss in mice.
AMPCA attenuates protein increase in the BAL fluid at day 21 after bleomycin treatment
Bleomycin causes an increase in protein levels in the bronchoalveolar lavage fluid (BAL)
from mouse lungs (Parker and Townsley, 2004; Kulkarni et al., 2013; Pilling and Gomer, 2014).
To determine if inhibiting NEU3 affects these BAL protein levels, we measured the protein in
the BAL as a measure of lung tissue damage post bleomycin treatment. As observed previously
(Parker and Townsley, 2004; Kulkarni et al., 2013; Pilling and Gomer, 2014; Karhadkar et al.,
2
020), at day 21, compared to saline-treated control mice, bleomycin-treated control mice
showed upregulated BAL protein levels (Fig. 3B-D). Compared to saline-treated control, the
saline-treated 2AP, MP, or AMPCA injected mice had no significant difference in the levels of
proteins in the BAL as assayed by either gel densitometry (Fig. 3B and C) or nanodrop (Fig. 3D).
AMPCA treatment significantly attenuated the bleomycin-induced increase of protein levels in
the BAL (Fig. 3B-D). Although 2AP and MP seemed to decrease BAL protein levels in
bleomycin-treated mice, the effect was not statistically significant. As reported previously
(Parker and Townsley, 2004; Karhadkar et al., 2020), the bleomycin-induced protein in BAL
appeared to be mainly albumin, which was confirmed by an immunoblot with anti-albumin
antibodies (Supplementary Figs. 9 A and B). No significant differences were observed in the
BAL volumes from the mice in the different groups (Fig. 3E). We previously observed that
Tamiflu treatment, but not DANA treatment, attenuated bleomycin-induced increased protein
levels in the BAL (Karhadkar et al., 2017). Together, these results suggest that some NEU3
inhibitors can attenuate bleomycin-induced upregulation of protein levels in the BAL.
2
3

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
NEU3 inhibitors attenuate the increased number of inflammatory cells in the BAL at day 21 after
bleomycin treatment
Bleomycin aspiration also causes an upregulation of inflammatory cells in the BAL at
day 21 (Murray et al., 2010; Pilling and Gomer, 2014; Pilling et al., 2014; Cox et al., 2015).
Compared to saline-treated control mice, the bleomycin-treated control mice showed upregulated
BAL cell counts (Fig. 4A). The saline-treated control mice and saline-treated and 2AP, MP, or
AMPCA injected mice showed similar BAL cell counts (Fig. 4A). Compared to saline-treated
mice, bleomycin-treated and MP or AMPCA injected mice showed upregulated BAL cell counts,
but these were significantly less compared to bleomycin-treated control mice (Fig. 4A). In the
BAL of bleomycin-treated control mice, as previously observed (Karhadkar et al., 2020),
bleomycin increased the numbers of CD11b+ cells (Fig. 4 B), CD11c+ cells (Fig. 4C), CD45+
cells (Fig. 4D), and Ly6G positive cells (Fig. 4E). Significant differences were not observed for
Ly6C positive cells (Fig. 4F). The BAL of saline-treated control mice, and saline-treated and
2
AP, MP, or AMPCA treated mice, showed similar counts of the above cell types (Fig. 4B – F).
Compared to BAL from saline-treated mice, bleomycin-treated 2AP or AMPCA injected mice
showed upregulated counts of CD11b positive cells (Fig. 4B), CD11c positive cells (Fig. 4C),
and CD45 positive cells (Fig. 4D) in the BAL, but these counts were significantly less compared
to bleomycin-treated control mice. The MP injected mice did not show significantly increased
counts of the above cell types in the BAL of saline-treated or bleomycin-treated mice (Figs. 4B –
F). Together, these data suggest that NEU3 inhibitors attenuate the bleomycin-induced increase
of inflammatory cell counts in the BAL.
2
4

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Lung sections were also stained to detect inflammatory cells. As previously observed
Karhadkar et al., 2020), bleomycin increased the counts of CD11b+, CD11c+, and CD45+ cells
(
remaining in the lungs after BAL (Figs. 5 A – C), but did not affect the counts of Ly6G+ or
Ly6C+ cells (Figs. 5 D – E). 2AP, MP, or AMPCA did not affect the counts of CD11b+ and
CD11c+ cells (Figs. 5 A and B), but 2AP and MP significantly counteracted the increase in
CD45+ cells in the lungs after BAL at day 21 (Fig. 5C). Together, these results suggest that
some NEU3 inhibitors attenuate the upregulation of some leukocytes remaining in the lungs after
BAL at day 21 post bleomycin treatment.
NEU3 inhibitors decrease fibrosis
To determine if inhibiting NEU3 can decrease bleomycin-induced fibrosis, lung sections
were stained with hematoxylin and eosin to detect tissue, picrosirius red to detect total collagen,
and hydroxyproline levels were measured in the lung tissue lysates. As previously observed
(Adamson and Bowden, 1974; Izbicki et al., 2002; Walters and Kleeberger, 2008; Moore et al.,
2
013), bleomycin treatment caused fibrosis in the mouse lungs at day 21 (Figs. 6 A – D and
Supplementary Fig. 10). Compared to saline treatment, lungs from mice exposed to bleomycin
and then treated with MP showed increased Sirius red staining and hydroxyproline levels at day
2
1 (Figs. 6B – D). The Sirius red staining and hydroxyproline levels from bleomycin treated and
then 2AP, MP, or AMPCA treated mice was significantly less than the bleomycin-treated control
mice (Figs. 6C and D and Supplementary Fig. 10). The bleomycin-treated and then 2AP or MP
or AMPCA treatment showed reduced staining for active TGF-β1 in the lung sections compared
to bleomycin-treated control mice (Figs. 7A and B). These results suggest that NEU3 inhibitors
reduce bleomycin-induced fibrosis and TGF-β1 upregulation in mouse lungs.
2
5

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Discussion
In this report, we observed that the human sialidase NEU3 desialylates human LAP and
causes release of active TGF-β1. This is in agreement with the observation that LAP
desialylation by viral and bacterial sialidases (Carlson et al., 2010; Gratz et al., 2017) causes a
conformational modification of the LAP which results in disruption of the interaction between
LAP and active TGF-β1, releasing active TGF-β1 (Biernacka et al., 2011). Compared to the
more conventional mechanism of a cell sensing a signal, and activating a gene expression
pathway to make and secrete a responding signal, the desialylation of LAP and release of active
TGF-β1 is a much more rapid response. This then suggests that fibrosis is associated with a
mechanism that originally evolved to rapidly respond to an insult such as a sialidase-expressing
pathogen, with the extracellular NEU3 mimicking and thus amplifying the pathogen signal.
NEU1 may also be involved in potentiating pulmonary fibrosis (Luzina et al., 2016), but
this sialidase did not significantly desialylate LAP, cause TGF-β1 release, or induce IL-6
accumulation by PBMCs, suggesting that it may act by other mechanisms to promote fibrosis.
NEU2 also did not significantly desialylate LAP, cause TGF-β1 release, or induce IL-6
accumulation by PBMCs. NEU4 did desialylate LAP and caused modest amounts of TGF-β1
release and IL-6 accumulation, but since this sialidase is present in lysosomes (Seyrantepe et al.,
2
004) or mitochondria and endoplasmic reticulum (Yamaguchi et al., 2005; Bigi et al., 2010), it
either does not normally desialylate LAP, or NEU4 from necrotic or otherwise lysed cells may
normally desialylate LAP.
2
6

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
We used a predicted reaction intermediate from the NEU3 reaction mechanism (Albohy
et al., 2010; Albohy et al., 2011a) to test the idea that picolinic acid-like compounds containing a
cationic nitrogen replacing the putative positively-charged oxygen of the reaction intermediate
could act as NEU3 inhibitors. In both the TGF-β1 release from LAP and the IL-6 production by
PBMC assays, we observed significantly improved inhibition potency of AMPCA compared to
DANA or oseltamivir (Table 1). Since some tumors have upregulated NEU3 (Kakugawa et al.,
2
002; Nomura et al., 2006; Ueno et al., 2006; Kawamura et al., 2012), and since mice lacking
Neu3 are resistant to colitis-associated colon tumor formation (Yamaguchi et al., 2012), AMPCA
or similar picolinate compounds may useful to treat NEU3-associated tumors.
In support of the idea of using picolinate compounds against NEU3-associated tumors, a
previous study found antiproliferative and therapeutic effects of picolinic acid on tumor cells
(Ruffmann et al., 1987). Picolinic acid shows cytotoxic activity against virus-infected cells and
also decreases viral replication (Fernandez-Pol and Johnson, 1977; Fernandez-Pol et al., 2001).
Picolinic acid potentiates the antimicrobial activities of various fluoroquinolones by chelating
metal ions (Cai et al., 2006; Shimizu and Tomioka, 2006). Picolinic acid functions as a second
signal in enhancing interferon-γ induced macrophage activation (Varesio et al., 1990; Melillo et
al., 1994), and has neuroprotective, immune regulatory, and tumor suppression effects in human
neurons (Guillemin et al., 2007). These reports and our studies suggest that picolinic acid, a
metabolite of tryptophan degradation (Varesio et al., 1990; Badawy, 2017), and picolinates may
be useful as therapeutics.
2
7

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
NEU1 and NEU2 did not significantly desialylate LAP or release active TGF-β1 from
rhL-TGF-β1, and thus this assay was incompatible for determining inhibition potencies of our
compounds against NEU1 and NEU2. The extracellular pH of lesions in lung fibrosis ranges
from 6.45 to 6.95 (Jones et al., 2015). Our repeated efforts to assay NEU1, NEU2, and NEU4 in
this pH range with the substrate 4MU-NANA were unsuccessful. Thus, it is unclear if our
compounds inhibit sialidases other than NEU3. One test of the specificity of the inhibitors would
be to test them for their ability to inhibit fibrosis in mice lacking NEU3, but since mice lacking
NEU3 do not develop fibrosis in the bleomycin model (Karhadkar et al., 2020), this cannot be
done and we thus cannot rule out the possibility that the inhibitors inhibit fibrosis through some
mechanism not involving NEU3.
As with AMPCA, DANA and Tamiflu attenuated bleomycin-induced protein
upregulation in the BAL at day 21 (Karhadkar et al., 2017). However, 2AP or MP did not
significantly attenuate bleomycin-induced protein upregulation in BAL at day 21. These
disparate results may be due to dose effects, or due to effects that DANA, Tamiflu, 2AP, MP,
and AMPCA have on targets other than NEU3.
We previously observed that NEU3 is upregulated in the mouse bleomycin model, and
that inhibition of sialidases with DANA or oseltamivir, or loss of NEU3, attenuate bleomycin
induced pulmonary fibrosis in mice (Karhadkar et al., 2017; Karhadkar et al., 2020). The human
and mouse NEU3 inhibitors 2AP, MP, and AMPCA all inhibited pulmonary inflammation,
levels of active TGF-β1 in the lungs, and pulmonary fibrosis in the mouse bleomycin model. The
Uniprot sequences P01137 for human L-TGF-β1 and P04202 for mouse L-TGF-β1 show 90%
2
8

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
amino acid sequence similarity and conserved glycosylation sites (Pruitt et al., 2007; Areström et
al., 2012; Doğan et al., 2016; Consortium, 2018). A reasonable explanation for the ability of
NEU3 inhibitors to decrease bleomycin-induced increased levels of active TGF-β1 is that mouse
NEU3 also desialylates mouse LAP to release mouse TGF-β1. Together, these results support the
hypothesis that a positive feedback loop where NEU3 desialylates LAP to release active TGF-β1,
and the active TGF-β1 causes lung epithelial cells to increase translation of NEU3 mRNA and
decrease degradation of NEU3 protein (Chen et al., 2020), contributes to pulmonary fibrosis, and
suggest that NEU3 inhibitors are potential therapeutics for pulmonary fibrosis.
2
9

JPET Fast Forward. Published on November 3, 2020 as DOI: 10.1124/jpet.120.000258
This article has not been copyedited and formatted. The final version may differ from this version.
Acknowledgements
We thank Darrell Pilling for assistance with BAL sample collection, and the LARR staff
at Texas A&M for animal care. We also thank Kristen Consalvo, Wensheng Chen, and Darrell
Pilling for comments on the manuscript.
Authorship contributions
Participated in research design: Karhadkar, Meek, and Gomer
Conducted experiments: Karhadkar
Performed data analysis: Karhadkar, Meek, and Gomer
Wrote or contributed to the writing of the manuscript: Karhadkar, Meek, and Gomer
3
0