Lubricin in experimental and naturally occurring osteoarthritis: a systematic review

Abstract

Objective:

Lubricin is increasingly being evaluated as an outcome measure in studies investigating posttraumatic and naturally occurring osteoarthritis. However, there are discrepancies in results, making it unclear as to whether lubricin is increased, decreased or unchanged in osteoarthritis. The purpose of this study was to review all papers that measured lubricin in joint injury or osteoarthritis in order to draw conclusions about lubricin regulation in joint disease.

Design:

A systematic search of the Pubmed, Web of Knowledge, and EBSCOhost databases for papers was performed. Inclusion criteria were in vivo studies that measured lubricin in humans or animals with joint injury, that investigated lubricin supplementation in osteoarthritic joints, or that described the phenotype of a lubricin knock-out model. A methodological assessment was performed.

Results:

Sixty-two studies were included, of which thirty-eight measured endogenous lubricin in joint injury or osteoarthritis. Nineteen papers found an increase or no change in lubricin and nineteen reported a decrease. Papers that reported a decrease in lubricin were cited four times more often than those that reported an increase. Fifteen papers described lubricin supplementation, and all reported a beneficial effect. Eleven papers described lubricin knock-out models.

Conclusions:

The human literature reveals similar distributions of papers reporting increased lubricin as compared to decreased lubricin in osteoarthritis. The animal literature is dominated by reports of decreased lubricin in the rat anterior cruciate ligament transection model, whereas studies in large animal models report increased lubricin. Intra-articular lubricin supplementation may be beneficial regardless of whether lubricin increases or decreases in OA.

Introduction

Post-traumatic osteoarthritis (PTOA) causes significant morbidity in both humans and domesticated animal species^1,2. Acute trauma, most commonly affecting the knee, shoulder or ankle, is generally the inciting cause of PTOA in humans and results in disruption of articular cartilage, bone, synovial membrane, joint capsule, and muscle³. Surgical intervention alone does not delay the development of PTOA, and current efforts are focused on investigating potential chondroprotective therapies^4,5. Lubricin has been identified as an important joint homeostatic and cartilage lubricating molecule that may be altered in the PTOA joint^6,7. Inflammation within the joint secondary to injury and changes in the concentration and composition of lubricating molecules such as lubricin and hyaluronic acid likely occur simultaneously and may result in increased friction and perpetuation of cartilage damage⁸.

Lubricin, also known as superficial zone protein, megakaryocyte stimulating factor, and proteoglycan 4 (PRG4), is a highly glycosylated mucin-like glycoprotein that functions in boundary lubrication of the joint^9,10 and inhibits synovial membrane hyperplasia^11,12. Lubricin/PRG4 is highly conserved across species and tissues, including kidney, lung, liver, heart, brain, muscle, tendon, small intestine, periodontal ligament, ocular surface, menisci, bone, cartilage, and synovial membrane^13–19. Based upon this broad distribution across tissues and the presence of several splice variants, lubricin likely has more complex biological roles beyond joint lubrication²⁰. Perhaps the most convincing evidence of the critical role lubricin plays in joint homeostasis and lubrication are the phenotypes that result from loss of lubricin in humans and in rodent models. Camptodactyly-arthropathy-coxa vera-pericarditis syndrome (CACP) is a rare heritable condition in humans that results from a mutation in the PRG4 gene locus²¹. CACP results in precocial multi-joint failure characterized by synovial hyperplasia, finger joint contracture, abnormalities of the femur and pericarditis²². The histological effects caused by lubricin deficiency are more fully characterized in lubricin knockout (Prg4^−/−) mice. These mice are born with normal joints but, as early as 2 weeks of age, develop polyarthropathy as evidenced by abnormal protein deposition on cartilage, degeneration of the cartilage surface, synovial hyperplasia, disappearance of superficial zone chondrocytes and increased friction coefficients^11,23,24.

Although lubricin is essential for normal joint homeostasis, it is not clear how lubricin is regulated after traumatic joint injury nor what role it plays in age-related OA. Lubricin levels have been reported to decrease in rodent, guinea pig and rabbit anterior cruciate ligament transection models (ACLT)^8,25–36. Human studies examining soft tissue knee injuries have predominantly found decreased lubricin levels^14,37,38, whereas lubricin is elevated after tibial plateau fracture and in end-stage arthritis^39–43. Studies examining PTOA models and naturally occurring injuries in larger domestic species, such as dogs and horses, have revealed increased lubricin after injury^44–51. It is not understood whether these differences arise from inherent variation in individual subjects^52,53, injury type, joint biomechanics, species, tissue, assay used to measure lubricin or time of sampling as joint fluid composition fluctuates temporally post-injury.

OA studies utilizing human subjects are inherently difficult to control for due to variation in injury severity, difficulty in obtaining samples at regular intervals and difficulty in obtaining baseline or healthy control samples. Animal models, regardless of species, obviate many of these concerns. Rat and mouse models are commonly used due to the availability of uniform genetic strains, the ability to standardize injuries, rigidly control environmental factors and evaluate large sample sizes. Large animal models, while requiring more specialized infrastructure, offer more clinical relevance to humans. Rodent models and some human studies are limited by the inability to directly obtain synovial fluid (SF) from the joint, necessitating the use of SF lavage. In large animal models, arthroscopic evaluation and serial SF sampling including a baseline sample are possible due to the larger SF volumes as compared to ~1 μL in the murine knee^54,55. Body weight and joint loading in clinically relevant models, such as the dog and horse, more closely approximate that of humans, though the quadruped stance does introduce some variation in distribution of loading⁵⁶. Rodents and rabbits have been cited to spontaneously heal cartilage lesions while dogs, horses, and humans do not⁵⁷. Based on these pertinent differences, there may be advantages in evaluating lubricin and studying lubricin supplementation in large animal models of OA.

The objective of this article was to systematically review the body of literature investigating lubricin quantification in joint injury and OA in both humans and animal models with the goal of determining if lubricin increases, decreases or remains unchanged post-injury. In addition, papers that described intra-articular lubricin supplementation and/or investigated lubricin/Prg4 murine knockout models were evaluated. Finally, an assessment of citation symmetry between the papers that cited an increase or decrease in lubricin post-injury was performed, and a methodological quality assessment was applied to papers that quantified lubricin.

Methods

Author disclosure

The authors of this review are English speaking veterinarians. The laboratory of one author performs research on lubricin in rat, horse, and dog models. All papers that fit the study criteria were included in this review in an effort to avoid citation or outcome reporting bias. Papers closely related to, but not meeting all inclusion criteria, were addressed in the discussion.

Search criteria

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) were followed. All PRISMA-compliant searches were conducted on peer-reviewed literature with dates of inception prior to April 4, 2019 using the following databases: PubMed, Web of Science and EBSCOhost (Fig. 1). No language restrictions were applied in the search strategy, but only articles available in English were considered. Published studies and abstracts were included.

Fig. 1.

Selection and inclusion process for papers in this review.

Inclusion criteria

Potential studies were considered eligible for inclusion if the following PICOS criteria were met. Participants: humans with naturally occurring osteoarthritis or joint injury; animals of any species, age or sex with naturally occurring osteoarthritis or joint injury or experimentally induced joint injury; animals with or without joint disease that received intra-articular lubricin supplementation; murine lubricin/Prg4 knockout models or murine models that over-expressed lubricin/PRG4. Interventions: lubricin/PRG4 was measured in cartilage, fibrocartilage, meniscus, SF or synovial membrane, by any method; any formulation of lubricin was supplemented intra-articularly; the phenotype of a genetically modified animal that was deficient in or over-expressed lubricin was described. Comparisons: control interventions included cadaver tissue, contralateral limb if unaffected by disease, alternate area of joint if unaffected by disease, healthy animal, sham surgery, sham injection, vehicle injection, affected joint pre-injury, and wild-type animal. Outcome: reported quantification of lubricin following natural or experimental injury compared to a control or baseline; reported gross morphology, histological evaluation, immunohistochemical or immunofluorescence analysis, biochemical analysis, radiological evaluation, gene expression, or behavior analysis following supplementation of lubricin; reported on phenotype resultant from genetic manipulation of the Prg4 locus. Study Design: controlled studies that assessed lubricin concentration, expression, and/or staining in the joint of humans with joint disease or animals with naturally or experimentally occurring joint disease; controlled studies that assessed the effect of intra-articular lubricin supplementation on osteoarthritis; controlled studies that described the phenotype of lubricin deficient animals.

Exclusion criteria

Studies in which lubricin was only quantified in cell culture media or serum/plasma were excluded, though studies could contain in vitro components as long as lubricin was measured from a non-cultured tissue or SF sample. Studies in which the insult to the joint was performed post-mortem were excluded. Studies in which the focus tissue was tendon/ligament or the joint being assessed was the temporomandibular joint were excluded, as were studies primarily focusing on inflammatory arthritis or antigen-induced arthritis.

Study selection

One reviewer screened the title and abstract of all search results eliminating obviously non-compatible articles. Two reviewers independently screened the remaining titles and abstracts. Thorough reading of the text was performed when the abstract suggested inclusion criteria would be met.

Assessment of methodological quality

The opportunity for bias and the quality of the experimental design were independently assessed by two separate reviewers based on a modified version of a previously published list of criteria (Supplementary data 1)⁵⁸. If there were disagreements, they were resolved via discussion until a consensus was reached. If a particular evaluation criteria was not applicable, the final score was calculated without including that question in the total possible points achievable. Articles that utilized a contralateral limb or other non-independent control were considered casee-control studies in this review although to truly fit the category an independent control should be included.

Citation data

To gain an understanding of the impact made on this field by each of the thirty-eight papers included in Tables I and II, the number of times cited as of April 22, 2019 was determined using the Web of Knowledge search function. This value was then divided by the number of years in publication to produce a citations/year value. These values were then combined between papers with similar outcomes to obtain a sum citations/year value.

Table I.

Lubricin quantification in human joint disease. ELISA – enzyme-linked immunosorbent assay, IHC – immunohistochemistry, WB – Western blot, MS – mass spectrometry, SF – synovial fluid, OA – osteoarthritis, NA – not available, mo – months, ACL – anterior cruciate ligament, eOA – early osteoarthritis, lOA – late osteoarthritis.

	Injury	Assay				Time of sampling	Control	Outcome
		ELISA	IHC	WB	MS
39	Tibial plateau fracture			✓		11 +/−9 days postinjury	Contralateral knee SF (n=3); healthy human knee SF (n=3)	[Lubricin] increased post-injury
40	Knee OA (varus malalignments)		✓			End stage OA	Lateral “spared compartment” cartilage in varus knee joints (n=3)	Lubricin staining increased in OA
41	Total knee arthroplasty	✓	✓	✓		End stage OA	Histologically healthy knee cartilage; cartilage from radiographically normal cadaver knees (n=3)	[Lubricin] increased in OA; lubricin staining increased in more severely affected areas of joint
42	Total knee arthroplasty				✓	End stage OAOA	Observational OA progression cohort knee SF (n=173)	Lubricin increased with decreasing joint space in age/sex-adjusted model
43	Total knee arthroplasty		✓			End stage	Calf and adult bovine cartilage (n=NA)	Lubricin staining increased in OA
59	Tibial plateau fracture			✓		12–66 mo postinjury	Contralateral knee SF (n=10)	[Lubricin] unchanged
60	Total arthroplasty, knee and hip		✓			End stage OA	Histologically normal areas of same menisci and labra (n=NA)	Lubricin staining unchanged
61	Tibial osteotomy; total knee arthroplasty; meniscectomy with OA	✓				End stage OA; chronic OA	Healthy human knee SF (n=NA)	[Lubricin] unchanged; correlated with anteroposterior laxity, full flexion angle and range of knee motion
37	ACL injury	✓				32–364 days postinjury	Contralateral knee SF (n=30)	[Lubricin] decreased post-injury
62	eOA; lOA	✓				eOA (arthroscopy) and lOA (arthroplasty)	Human cadaver knee SF (n=16)	[Lubricin] decreased in lOA
38	Meniscal injury		✓			2 days post-injury	Human cadaver menisci (n=9)	Lubricin staining decreased post-injury
14	Meniscectomy with OA	✓	✓			NA	Human menisci and SF obtained 3 months post-ACL injury (n=9)	Lubricin expression decreased in OA menisci and OA SF lavage samples

Table II.

Lubricin quantification in naturally occurring injuries and experimental animal models of osteoarthritis. ELISA – enzyme-linked immunosorbent assay, IF – immunofluorescence, WB – Western blot, PCR – polymerase chain reaction, MS – mass spectrometry, SF – synovial fluid, SM – synovial membrane, OA – osteoarthritis, OCD – osteochondrosis dissecans, wk – week, ACLT – anterior cruciate ligament transection, DMM – destabilization of the medial meniscus, MFC – medial femoral condyle, MCLT – medial collateral ligament transection, PCLT – posterior cruciate ligament transection, * – AlphaLISA.

	Species	Model	Assay					Control	Outcome (Lubricin)
			ELISA	IF/IHC	WB	PCR	MS
48	Horse	Cartilage defect			✓			Time 0 baseline sample	SF [lubricin] increased post-injury
46	Horse	Carpal osteochondral fragment model; naturally occurring carpal OA	✓	✓		✓		Contralateral sham-operated joint	SF [lubricin] increased post-injury; SM lubricin staining increased post-injury; PRG4 expression increased in SM but decreased in cartilage post-injury
49	Horse	Late stage OA; acute joint injury; OCD	*		✓			Healthy equine joints	SF [lubricin] increased post-injury (*AlphaLISA)
50	Horse	Carpal osteochondral fragment model	✓					Contralateral sham-operated joint	SF [lubricin] increased post-injury
47	Horse	Carpal osteochondral fragment model, talar impact model, stifle cartilage defect model	✓					Time 0 baseline sample	SF [lubricin] increased post-injury
51	Horse	Naturally occurring carpal and fetlock injury; acute <3 wk, chronic > 3 wk			✓			Contralateral uninjured joints	SF [lubricin] increased post-injury, with acute injury higher than chronic injury
45	Dog	ACLT	✓					Healthy canine synovial fluid	SF [lubricin] increased post-injury
44	Dog	ACLT		✓				Contralateral non-operated joint	Lubricin staining depth increased post-injury
63	Mouse	DMM		✓	✓			Non-operated mice; contralateral sham-operated joint	Lubricin staining & Prg4 expression increased post-injury
64	Sheep	MFC full thickness defect; fetal and adult					✓	Twin lamb joint, healthy adult sheep joint	Fetal: PRG4 expression increased post-injury Adult: PRG4 expression not changed post-injury
65	Sheep	ACLT/MCLT or lateral meniscectomy	✓					Contralateral unoperated joint, sham-operated joint	SF [lubricin] not changed
67	Sheep	Lateral meniscectomy		✓		✓		Sham-operated sheep joints	Lubricin staining and PRG4 expression decreased post-injury
25	Rat	ACLT	✓	✓				Sham-operated rat joint	SF [lubricin] and staining decreased post-injury
26	Rat	ACLT		✓		✓		Contralateral non-operated joint	I Lubricin staining and Prg4 expression decreased post-injury
66	Rat	Naturally occurring age-related OA	✓	✓	✓			Young, healthy rat joint	SF [lubricin], staining and protein expression decreased with age-related OA
34	Rat	ACLT	✓	✓	✓			Non-operated rat joints	SF [lubricin], staining, and protein expression decreased postinjury
27	Rat	ACLT	✓	✓		✓		Non-operated rat joints	SF [lubricin], lubricin staining, and Prg4 expression decreased post-injury
32	Rat	ACLT		✓		✓		Non-operated rat joints	Lubricin staining and expression decreased post-injury
33	Rat	ACLT		✓				Non-operated rat joints	Lubricin staining decreased post-injury
31	Rat	ACLT		✓				Non-operated rat joints	Lubricin staining decreased post-injury
35	Rat, mouse	ACLT		✓				Sham-operated mice and rat joints	Lubricin staining and expression decreased in ACLT + vehicle-treated joints post-injury
36	Mouse	ACLT		✓				Sham-operated or vehicle-injected mouse joints	Lubricin staining decreased post-injury
30	Guinea pig	ACLT	✓		✓			Non-operated guinea pig joints, contralateral nonoperated joint	SF [lubricin] and expression decreased post-injury
29	Guinea pig	ACLT; naturally occurring age-related OA	✓					Non-operated guinea pig joints	SF [lubricin] decreased post-injury
8	Rabbit	ACLT & PCLT	✓		✓			Non-operated rabbit joints, sham-operated joint	SF [lubricin] decreased at weeks 2 & 3 post-injury compared to baseline & week 1 post-injury
28	Rabbit	ACLT			✓	✓		Non-operated rabbit joints, contralateral non-operated joint	SF [lubricin] and Prg4 expression decreased post-injury

Results

The electronic data search identified 787 potentially relevant papers (Fig. 1). 683 papers, including ninety-one reviews, were excluded by title and abstract evaluation. The remaining 104 were assessed for eligibility. Ultimately, sixty-two papers satisfied the inclusion criteria, including twelve using human subjects and fifty using animal models [Fig. 2(A)]. Of the animal papers, twenty-six measured lubricin post-injury, fifteen examined lubricin supplementation in animal models, and eleven papers utilized a genetic model.

Fig. 2.

Summary of studies selected for inclusion divided by species and study type. A) Total number of studies included in Tables I–IV showing that the majority of studies investigated animal models. B) Total number of studies included in Tables I–IV divided by species and study type. Human studies made up 19% of all included studies. Rodent studies were primarily supplementation and genetic model studies. C) Total number of studies that quantified lubricin (Tables I and II) divided by human and animal. Human studies comprised 31% of these studies. D) Studies that quantified lubricin divided by species displaying the diversity in animal models that have been used to quantify lubricin.

Twelve papers fit the inclusion criteria for human studies (Table I). Five showed an increase in lubricin post-injury^39–43, three showed no change^59–61, and four showed a decrease [Fig. 3(A)]^14,37,38,62. Of the studies that showed an increase in lubricin, four characterized knee OA at the point of total arthroplasty and one examined tibial plateau fractures at approximately 11 days post-injury [Fig. 2(A)]. The three studies that found no change in lubricin had the most variation in injury type, which included tibial plateau fractures, arthroplasties and tibial osteotomies to treat end-stage joint disease, and meniscectomies with concurrent OA. Of the manuscripts that showed a decrease in lubricin, three characterized ACL or meniscal injuries within 1 year of injury, while one study examined knees at the time of arthroscopy (early OA) and arthroplasty (end-stage OA). Lubricin was measured in human studies primarily via ELISA and immunohistochemical staining, with infrequent use of immunoblotting and mass spectroscopy. Methodological assessment revealed variation in quality scores across human studies, with the percentage of total possible points ranging from 33 to 63% (Supplementary data 2). The majority of variation in quality scores between human studies was attributed to documentation of inclusion/exclusion criteria and the availability of appropriate controls which also resulted in overall lower scores for human papers compared to animal papers.

Fig. 3.

Frequency distributions representing the number of manuscripts reporting a decrease, increase, or no change in lubricin organized by (A) injury type in humans (B) animal species, and (C) OA animal model with ACLT being compared against all other models.

Fifty papers fit the inclusion criteria for animal studies. Thirtyseven used rodent models, with a distribution of eighteen mouse studies, 15 rat studies and one study using both mice and rats. Sixteen described non-rodent models, with six horse models, three sheep models, two studies each using dog, guinea pig and rabbit models and one Yucatan mini-pig model [Fig. 2(B)]. Twenty-six papers quantified lubricin post-injury (Table II). Nine showed an increase in lubricin concentration, staining, or expression postinjury [Fig. 3(A)]^44–51,63. One study found mixed results with an increase in PRG4 expression in a cartilage defect model in fetal sheep but no change in adult sheep⁶⁴. One study found no change in lubricin concentration in SF following ACLT or lateral meniscectomy in sheep⁶⁵. Fifteen studies found a decrease in lubricin concentration, staining, or expression post-injury^{8,25–36,66,67}. The animal studies demonstrated more homogeneous and higher quality scores than the human studies, with the percentage of total possible points ranging from 71 to 94%, acknowledging that it is difficult to make direct comparisons between human studies and experimental animal studies (Supplementary data 3). The majority of the variation in quality scores between animal studies was attributed to the application of allocation sequences and the blinding of outcome assessors.

Studies that reported an increase in lubricin post-injury utilized diverse models, which included destabilization of the medial meniscus (1), ACLT (2), cartilage defect (3), carpal osteochondral fragmentation (3), naturally occurring injury/OA (3), osteochondrosis dissecans (1), and a talar impact model (1). Three studies that reported an increase in lubricin utilized more than one model. Studies that reported a decrease in lubricin post-injury were overrepresented by the ACLT model (13), with other models including naturally occurring OA (2), and lateral meniscectomy (1). One study that reported a decrease in lubricin utilized more than one model. Lubricin was measured primarily via immunohistochemistry (IHC) or immunofluorescence (IF) staining of articular cartilage and/or synovial membrane and ELISA, with less frequent use of immunoblotting and qRT-PCR.

The rodent, rabbit, and guinea pig studies revealed a decrease in lubricin post-injury 93% of the time, while 100% of the horse and dog studies revealed an increase in lubricin post-injury [Fig. 2(B)]. Studies that used sheep models represented all three lubricin outcomes. Studies that used ACLT as a model reported a decrease in lubricin post-injury 83% of the time, while the percentage decreased to 23% if any other model was evaluated [Fig. 2(C)]. These findings are confounded by the fact that 83% of ACLT studies were performed in rodent, rabbit, and guinea pig models. Eleven of the papers that quantified lubricin post-injury also evaluated lubricin post-injury in the face of an ancillary treatment (Supplementary data 4).

The number of citations per year for the thirty-eight papers included in Tables I and II are represented in Fig. 3(B). Human studies that reported a decrease in lubricin were cited twice as often as those that reported an increase. Animal studies that reported a decrease in lubricin were cited six times more often than those that reported an increase. The papers that reported no change in lubricin post-injury were infrequently cited.

Fifteen papers described intra-articular lubricin supplementation with two of the papers also quantifying lubricin post-injury (Table III)^{6,26,27,68–73,74–79}. Of the papers that described intraarticular lubricin supplementation, five used mouse models, nine used rat models, and one used a Yucatan mini-pig model. Nine used ACLT as the injury model, three used destabilization of the medial meniscus, and one study each used the following: a mouse TGF-β receptor mutant model and rat models of ovariectomy, destabilization of the medial meniscus, and intra-articular monosodium urate monohydrate crystal injection. The treatments included PRG4 overexpression and intra-articular injection of a truncated recombinant human lubricin construct (LUB:1) or full-length recombinant human lubricin, lubricin derived from cultured human synoviocytes, or purified human SF lubricin. All papers found a beneficial effect of lubricin supplementation based on histological, morphological, functional and/or gene expression outcome measures.

Table III.

Lubricin supplementation in the OA joint. ACLT – anterior cruciate ligament transection, HDV – helper dependent adenoviral vector, OA – osteoarthritis, OARSI – osteoarthritis research society international, PTOA – post-traumatic osteoarthritis, IL-1ra – interleukin 1 receptor antagonist, IA – intraarticular, DMM – destabilization of the medial meniscus, AC – articular cartilage, DNIIR – dominant-negative mutant of TGF|3R2 (OA model), OVX – ovariectomized rat model, rhPRG4 – recombinant human lubricin/proteoglycan 4, MMP-13 – matrix metalloproteinase 13, TRAP – tartrate-resistant acid phosphatase, HSL – human synoviocyte lubricin, HSFL – human lubricin, GAG – glycosaminoglycans, CTX-II – carboxy-terminal telopeptides of type II collagen, MT – meniscal tear, ASI – aggrecanase specific inhibitor, MUM – monosodium urate monohydrate, hrs – hours, HA – hyaluronic acid, PBS – phosphate buffered saline.

	Species	Model	Treatment	Outcome following lubricin supplementation
6	Mouse	ACLT	Prg4 overexpression in cartilage (Col21a promoter) or Prg4 via HDV	Prg4 overexpression decreased signs of age-related and post-traumatic OA, including preservation of cartilage volume, decreased OARSI score, increased expression of Col10a1 and Mmp13 and decreased functional impairment; gene transfer with HDV-Prg4 protected against PTOA and age-related OA based on decreased OARSI score and increased cartilage volume
68	Mouse	ACLT	Prg4 over-expression in cartilage (Col21a promoter) or Prg4 ± Il-1Ra via HDV	Prg4 overexpression reduced age-related and post-traumatic OA; Prg4 expression inhibited transcriptional programs that promote cartilage catabolism and hypertrophy; IA injection of HDV expressing Prg4 protected against PTOA
69	Mouse	ACLT	IA HDV-Prg4 or 10mabHDV-Prg4	Both treatments improved OARSI scores and increased cartilage volume and surface area; 10mabHDV-Prg4 treated joints had lower OARSI scores and larger bone area covered by cartilage than HDV-Prg4 treated joints
70	Mouse	ACLT or DMM	IA HDV-NFκB-Il1ra and/or HDV-EF1-Prg4	Combined therapy increased expression of pro-anabolic and cartilage matrix genes; decreased catabolic gene and inflammatory mediator expression; preserved AC volume and surface area; prevented thermal hyperalgesia
71	Mouse	DNIIR	Prg4 overexpression via Col2a1 promoter	Fewer missed steps; decreased cartilage fibrillation and OARSI scores; increased cartilage thickness
72	Rat	OVX	IA rhPRG4 at day 0or14	Decreased cartilage degeneration, OARSI score, proteoglycan loss, and staining for collagen X and MMP-13; increased lubricin expression; normalized bone remodeling; inhibited elevation of TRAP and Osterixpositive cells; rescued thickening of calcified cartilage zone; decreased vascular invasion of calcified cartilage
73	Rat	DMM	IA rhPRG4 1 or 3x/wk	Decreased cartilage degeneration and cartilage lesion/degeneration width; facilitated boundary lubrication and decreased synovial cell adhesion; 3x/wk dosing increased benefit
74	Rat	ACLT	IA HSL or rhPRG4 or purified HSFL	Decreased OARSI scores; increased lubricin staining and Prg4 expression; decreased fibrillation at cartilage surface with more superficial chondrocytes; decreased collagen type II degradation; no difference between treatments
75	Rat	ACLT	IA HSL	Equalized weight distribution between hind legs; increased lubricin mRNA expression; increased lubricin staining to level of sham-operated joints; increased GAG content; decreased urinary CTX-II; OARSI score not different
76	Rat	MT	IA rhPRG4 ± ASI	Decreased pain 4 wks post-injury and for 3 wks post-dose cessation; combined therapy decreased pain more than monotherapy
77	Rat	MUM	IA rhPRG4	At 6 h: supplementation normalized weight bearing and myeloperoxidase activity. At 24 h: no difference
78	Rat	ACLT	IA purified HSFL ± HA	Decreased radiographic and histologic OA scores, increased SF [lubricin]; HA provided no added benefit
27	Rat	ACLT	IA rhPRG4	Increased lubricin staining in articular cartilage post-injury; decreased caspase-3 and MMP-13 staining
26	Rat	ACLT	IA purified HSFL ± exercise	Increased lubricin staining in articular cartilage; increased SF [lubricin]; decreased urinary CTX-II levels and caspase-3 staining
79	Pig	DMM	IA rhPRG4 ± HA	Decreased macroscopic OA scores; increased SF [lubricin]; decreased serum IL-1β; no difference between PBS and rhPRG4 + HA

Eleven papers utilized a murine Prg4 knock-out model (Table IV)^{11,23,24,80–87}. Four described the phenotype of knock-out mice at varying ages without any treatment. Two papers assessed the effects of genetically restoring lubricin expression at varying timepoints, including at pre-conception,1 week, 2 weeks, 2 months and 6 months of age. Five papers examined the effects of treatments, including cyclic loading, lubricin and systemic parathyroid hormone administration, on Prg4 null mice.

Table IV.

Phenotype and treatment effect of lubricin-null mouse models. Mo – month, AC – articular cartilage, SZ – superficial zone, wk – week, COF – coefficient of friction, MFC – medial femoral condyle, OA – osteoarthritis, WT – wild type, HTZ – heterozygote, IA – intra-articular, PTH – parathyroid hormone, SDF-1 – stromal cell-derived factor 1, FGF-2 – fibroblast growth factor-2, CTX-II – carboxy-terminal telopeptides of type II collagen.

	Model	Treatment	Outcome/Phenotype
11	Prg4 null mice	None	Normal joints at birth. At 2–4 mo: abnormal protein deposition on AC; decreased SZ chondrocytes; synovial hyperplasia; abnormal calcification of tendons/sheaths
23	Prg4 null mice	None	Normal joints at birth. At 2 wks: irregular cartilage surface with disruption of parallel orientation of collagen fibrils; absence of lamina splendens. At 1 mo: increase in joint COF
24	Prg4 null mice	None	At 2 wks: Roughened cartilage surface. At 16 wks: decreased pericellular proteoglycans; increased COF. Adult: lower elastic modulus in articular cartilage
80	Prg4 null mice	None	At 10 wks: decreased live chondrocyte volume fraction in MFC; increased cartilage thickness in MFC, increased caspase-3 positive cells in MFC superficial and upper intermediate cartilage zones
81	Prg4 null mice	Prg4 genetically restored at pre-conception or 3 wk, 2 mo, or 6 mo of age	Rx: no joint disease with pre-conception restoration of Prg4; 3 wk restoration: increased caspase-3 activation and histological evidence of OA; 2 & 6 mo restoration: unchanged from Prg4 null littermates
82	Prg4 null mice	Prg4 genetically restored at7or14 days of age	Rx: restored joints unchanged from null joints in COF and histology; restored cartilage had decreased numbers of peroxynitrite and caspase-3 positive cells
83	Prg4 null mice	Cyclic loading	COF of null joints higher than WT or HTZ; Rx: upto4h of cyclicjoint loading increased null joint COF while WT and HTZ COF remained unchanged. At 26 h: all COF increased (larger increases in null and HTZ joints)
84	Prg4 null mice	IA human lubricin	Null joints showed mitochondrial dysregulation-mediated caspase activation and increased levels of peroxynitrite and superoxide; Rx: treatment prevented caspase-3 activation in AC and decreased whole joint friction
85	Prg4 null mice	PTH	Null mice showed increased peripheral blood neutrophils; decreased marrow B-lymphocytes; decreased SDF-1; Rx: PTH normalized findings
86	Prg4 null mice	PTH	Young: decreased growth plate hypertrophic zone height, trabecular bone, and serum bone formation markers; Adult: decreased trabecular and cortical bone, decreased tarsal range of motion, decreased liver and bone marrow FGF-2 mRNA; Rx: Young: no change; Adult: PTH increased bone mass and normalized FGF-2 mRNA
87	Prg4 null mice	PTH	Acellular material deposited on AC and menisci; increased AC degradation; increased serum [CTX-II]; decreased articular chondrocyte apoptosis; increased SDF-1 expression in synovium; irregularly contoured subchondral bone; Rx: PTH induced a secondary deposit overlaying the acellular material on the AC

Discussion

Combining human and animal studies, nineteen reported an increase or no change in lubricin in joint injury or OA, while nineteen reported a decrease in lubricin. Given the heterogenous nature of OA, with distinct phenotypes based on disease etiology (traumatic, metabolic and inflammatory) and individual characteristics (obesity, age and activity level), variability in OA models and disease likely contributes to significant variation in study outcomes⁸⁸. Studies that evaluated rodent, rabbit, or guinea pig models (n = 18) found a decrease in lubricin in the PTOA joint 93% of the time, while papers investigating horses and dogs (n = 8) found an increase in lubricin 100% of the time. Similarly, studies in ACLT models reported a decrease in lubricin in the PTOA joint 83% of the time, while studies reporting on any other injury model reported a decrease in lubricin only 23% of the time. Injury type was associated with lubricin quantification outcome in human studies, with 3/4 of the studies that cited a decrease in lubricin reporting on a population of soft tissue injuries such as ACL tears and meniscal damage, while 4/5 of the studies that cited an increase in lubricin reported on a population with end-stage OA. Studies measuring lubricin used one or more of the following assays: immunohistochemistry/immunofluorescence (n = 21), ELISA (n = 18), immunoblotting (n = 9), polymerase chain reaction (PCR) (n = 11), or mass spectrometry (n = 2).

Of the thirty-eight papers that quantified lubricin, only four animal studies and one human study used three separate assays, permitting evaluation of multiple tissue types, including SF, articular cartilage and/or synovial membrane. Of these studies, two found an increase, and three found a decrease in lubricin post-injury. Immunohistochemical staining of articular cartilage and/or synovial membrane was more common in studies that reported a decrease in lubricin, while studies that reported an increase in lubricin were more likely to use ELISAs to determine the concentration in SF. This may be due to the fact that, in large animal models, where increased lubricin in PTOA joints is more frequently reported, undiluted SF aspirates can be easily obtained, while SF lavage is typically necessary to obtain measurements in rodent and small animal models. Even when normalizing for parameters such as urea, assays evaluating synovial lavage fluid are likely less accurate than evaluating direct SF aspirates⁵⁴. Approximately half of the animal studies quantified lubricin at more than two timepoints, providing more information about temporal variation as compared to most human studies.

Although a previously described methodological assessment instrument for assessing the quality of drug studies was adapted for use in this review⁵⁸, an ideal instrument is not available to compare the diverse study designs represented. Therefore, human and animal studies were assessed separately. Overall, the scores were higher in the animal studies than the human studies, highlighting the challenges with obtaining ideal human control tissues. For example, Musumeci et al. 2014 was penalized for the quality of the control population since patients with recent ACL injuries (~3 months post-traumatic rupture) were included as normal controls in comparison to patients with Kellgrene–Lawrence grade 2–3 knee OA. The authors concluded that lubricin is decreased in OA. In our opinion, ACL injury ~3 months prior to sampling would constitute a sub-acute traumatic injury, and it should be qualified that lubricin was decreased in chronic OA as compared to sub-acute ACL rupture rather than healthy knees.

Manuscripts that report a decrease in lubricin in OA are cited 4 times more often than those that do not, despite an approximately equal distribution of studies showing an increase or no change in lubricin. An explanation for this citation asymmetry is that loss of lubricin provides a direct motivation for evaluating the effects of lubricin supplementation in vivo. Lubricin supplementation resulted in improvement in all fifteen studies in which it was evaluated. Five studies used transgenic mice to evaluate long-term overexpression of Prg4, resulting in downregulation of transcriptional programs that promote cartilage catabolism and hypertrophy and upregulation of pro-anabolic and cartilage matrix genes. Four studies used purified human SF lubricin, and seven used variants of recombinant human lubricin. Histologic findings associated with lubricin supplementation included reduced OARSI scores, cartilage fibrillation, proteoglycan loss, and synovial cell adhesion and preservation of cartilage volume. Fourteen of the fifteen studies assessing lubricin supplementation were performed in rodents, with nine studies utilizing an ACLT model. Future studies in large animal models will be needed to determine if lubricin supplementation remains beneficial in models that report an increase in lubricin post-injury.

Lubricin is essential for boundary lubrication of articular cartilage; however, recent literature suggests that lubricin also has important biological functions, including modulating inflammation and protecting chondrocytes^12,89. Lubricin, via its many O-linked glycans, is able to support contact loading and protect cartilage from shear stress⁹⁰. Lubricin exists bound to the articular cartilage, likely via complexes with type II collagen, fibronectin, and cartilage oligomeric matrix protein (COMP), and it exists unbound in the SF as a reservoir⁹¹. Quantification and functional analysis of both cartilage-bound and free lubricin is an essential direction for future research. Approximately half of the mass of lubricin is determined by extensive O-linked glycosylation which may be modified in inflammation as evidenced by altered glycosylation patterns observed between normal joints, inflammatory arthritis, and joints with osteochondral fragments^7,9,89. Lubricin inhibits inflammation by competitively binding to CD44 at the same site as hyaluronic acid, thereby decreasing synoviocyte proliferation and reducing expression of MMP1, MMP3, MMP13, IL-6 and IL-8 while increasing expression of TIMP2 and COX2^92,93. Lubricin binds to TLR2, TLR4, and TLR5 in an HA-independent manner and functions as an antagonist to reduce inflammation^94,95. Lubricin/PRG4 also has important roles in joint development, with early cartilage and synovial membrane progenitor cells expressing Prg4⁹⁶. Recent data suggests that Prg4-expressing superficial zone chondrocytes may be a progenitor cell reservoir for deeper layers of cartilage⁹⁷. A better understanding of lubricin’s anti-inflammatory, chondroprotective and lubricating functions will improve our ability to select appropriate patient populations, indications and timing for lubricin therapy.

Several studies were excluded from analyses due to absence of a control as defined by the inclusion criteria but still warrant discussion. Atarod et al. reported on an ovine ACLT model and found an increased SF lubricin concentration at two- and 4-weeks post-injury compared to 20-weeks post-injury; however, baseline pre-injury data was not available⁹⁸. Similarly, Catterall et al. reported on acute knee injury in humans, demonstrating that lubricin was increased at the time of study enrollment (mean 15.2 days post-injury) as compared to at the time of arthroscopic repair (mean 47.6 days post-injury)⁹⁹. Elsaid et al. reported on a decrease in SF lavage lubricin concentration from three to 7 weeks post-ACLT, but no comparison to a baseline or contralateral limb was included²⁶. All of these papers interpreted the data as a decrease in lubricin in PTOA; however, in the absence of a pre-injury or healthy control comparison, the change demonstrated could be a return to normal level following an acute increase in lubricin post-injury.

Variation in the type of assays and reagents used to quantify lubricin and the lack of any commercially available, validated ELISAs are likely to contribute to discrepancies between studies. Immunohistochemistry is only semi-quantitative and cannot differentiate between increased lubricin that is produced locally within the articular cartilage and lubricin that has penetrated deeper into fibrillated, injured cartilage devoid of a surface zone. ELISAs are quantitative; however, ELISAs only measure free lubricin in the SF, which may be a mixture of secreted lubricin and lubricin that has become dissociated from articular cartilage. ELISAs, immunoblotting and IHC/IF can all be affected by variation in antibody reactivity to different lubricin glycosylation patterns¹⁰⁰ and non-specific binding. PCR can provide supportive information; however, gene and protein concentrations do not always correlate well. More advanced glycomics assays, including MALDI-TOF mass spectrometry are becoming available; however, their expense limits high-volume sample analysis. Ideally, studies combine and find agreement across several methodologies including ELISA, IHC, and gene expression to increase study robustness while incorporating immunoblots to detect any degradation or fragmentation of the lubricin species being measured. Additionally, it is important that studies publish the details of their methodologies to assist with transparency and reproducibility. Particularly, listing the antibody used is essential as processing or cleavage of lubricin can result in conflicting changes in concentration when using different antibodies⁵⁹.

Conclusion

Similar proportions of studies reported an increase or no change in lubricin in OA as compared to a decrease in lubricin; however, papers that reported a decrease in lubricin were cited 4 times more frequently. Large animal models were over-represented in studies that reported increased lubricin in OA, while rodent models were over-represented in studies that reported decreased lubricin in OA. Human studies were approximately evenly distributed between those reporting an increase or decrease in lubricin. Lubricin null mice demonstrate the importance of lubricin in joint homeostasis and attest how even a brief lubricin deficit can have long term degradative effects. Lubricin supplementation ameliorates age-related and post-traumatic OA based on gross, histological, and gene expression parameters. In addition to its lubricating function, lubricin has biological roles that may benefit the joint in OA, explaining why lubricin supplementation may be beneficial regardless of how lubricin content varies between models. Additional research is required to assess the value of lubricin supplementation in large animal joint injury models and to understand the biological roles of lubricin supplementation in the OA joint.

Fig. 4.

Proportion of studies reporting a decrease, increase or no change in lubricin (A) in human and animal models, and (B) proportion of total citations per year since year of publication. There are an equal number of papers describing an increase or no change in lubricin as there are describing a decrease in lubricin post-injury; however, manuscripts reporting a decrease in lubricin were over-represented by 4-fold (2-fold for human, 6-fold for animal).