CL-L1 and CL-K1 Exhibit Widespread Tissue Distribution With High and Co-Localized Expression in Secretory Epithelia and Mucosa

Collectin liver 1 (CL-L1, alias collectin 10) and collectin kidney 1 (CL-K1, alias collectin 11) are oligomeric pattern recognition molecules associated with the complement system, and mutations in either of their genes may lead to deficiency and developmental defects. The two collectins are reportedly localized and synthesized in the liver, kidneys, and adrenals, and can be found in the circulation as heteromeric complexes (CL-LK), which upon binding to microbial high mannose-like glycoconjugates activates the complement system via the lectin activation pathway. The tissue distribution of homo- vs. heteromeric CL-L1 and -K1 complexes, the mechanism of heteromeric complex formation and in which tissues this occurs, is hitherto incompletely described. We have by immunohistochemistry using monoclonal antibodies addressed the precise cellular localization of the two collectins in the main human tissues. We find that the two collectins have widespread and almost identical tissue distribution with a high expression in epithelial cells in endo-/exocrine secretory tissues and mucosa. There is also accordance between localization of mRNA transcripts and detection of proteins, showing that local synthesis likely is responsible for peripheral localization and eventual formation of the CL-LK complexes. The functional implications of the high expression in endo-/exocrine secretory tissue and mucosa is unknown but might be associated with the activity of MASP-3, which has a similar pattern of expression and is known to potentiate the activity of the alternative complement activation pathway.

is a "selfamplificative" cascade system, mainly found in the blood, and includes several PRMs; among which some activate a pathway known as the lectin activation pathway, via activation of serine proteases known as MBLassociated serine protease (MASP1, 2, and 3) (2,3). Mannanbinding lectin (MBL) is probably the most studied PRM of the lectin activation pathway and binds to mannoserich glycoconjugates on the surface of microorganisms, initiating complement activation. In humans, MBL deficiency may increase susceptibility toward infections in certain situations but not in general (4), most likely contributed by coincidental activation of different complement activation pathways by a given microorganism.
Collectin liver 1 (CLL1) and collectin kidney 1 (CLK1) are "MBLlike" proteins that also are found in the circulation in association with MASPs and with specificity toward mannoserich glycoconjugates and negatively charged molecules (5)(6)(7)(8)(9)(10). In the circulation they can be found as heteromeric molecules, referred to as CLLK, with superior complement abilities via MASP2 in comparison with their respective homomers (11). They are both synthesized in the liver by hepatocytes, in the adrenal glands and in the tubules of the kidney, in addition to other tissues as well (5)(6)(7). On the protein level, human CLK1 has been associated with the same tissues, while human CLL1 in the original study only was associated with hepatocytes, however, without examination or exclusion of other tissues and cells (6,7). Among normal healthy populations of different origins, they both constitute average serum concentrations of 250-450 ng/ml, with a clear correlation between levels of CLL1 and CLK1, supportive of heteromeric complexes between the two or similar regulation (7,(12)(13)(14)(15). The liver and adrenals are due to their endocrine nature and a relative high synthesis of mRNA believed to be the major organs contributing to the CLL1 and CLK1 found in the circulation. Partly due to the recently described characterization and associa tion with complement, little is known about their roles in vivo. In a recent work using mice deficient of CLK1, Wakamiya and col leagues showed that CLK1 protected mice against Streptococcus pneumonia infections induced via nasal inoculation (16). However, in another work there was no protective effect of CLK1/ CLLK in a mouse model for infection by Mycobacterium tuberculosis (17). CLK1 has been shown to bind with relative high affinity to the disaccharide Man(α12) and to negatively charged molecules, including nucleic acid ligands, and may also play a role in the opsonization of apoptotic cells by recognizing a combina tion of carbohydrate and nucleic ligands (10,18).
Recently, it was demonstrated that CLK1deficient mice partly were protected against destructive complementmediated inflammatory responses in post ischemic kidneys and that CLK1 further promoted development of renal fibrosis in the tubules (19,20). CLK1 and CLL1 are not regulated significantly by inflam matory stimuli. Their plasma/serum levels do not correlate with increased levels of traditional inflammatory mediators, including CRP and TNFalpha (8,9).
The two collectins play apparently also important roles for embryogenesis. Deficiency of CLK1 or CLL1 leads in humans to a rare congenital developmental syndrome known as 3MC (alias Malpuech facial clefting syndrome), an effect that the two collectins share with MASP3. It has been shown that CLK1 and CLL1 may act as attractants and guidance cue for neural crest cells, although the precise mechanism for embryonic involve ment remains to be elucidated (21,22).
A functional and complete activation of the complement system involves many complement factors and is in general only associated with the effect in the circulation or at inflamed sites, where blood components gain access. Most tissue expresses certain complement components, e.g., the lung and the intes tinal system (23,24). At such sites the complement system is believed to be partially functional or to mediate activation when inflammation progresses. In the quest of elucidating the func tion of CLL1 and CLK1, we characterized their localization in human tissues. It appears that the previously characterized sites of localization in the circulation, adrenals, liver, and kidney, may have disparaged a compelling localization in especially exo crine/endocrine tissues and mucosa, suggestive of that CLL1, CLK1, and CLLK may play roles in the periphery as well.
generation of Mabs against cl-l1 and cl-K1 CLLK was purified from outdated plasma by calcium sensitive immunoaffinity chromatography as previously described (25). Purified CLLK (50 µg) was used for s.c. immunizations of outbred NMRI female mice using Gerbu as adjuvant. Three days before the fusion, mice were boosted (i.p.) with the same amount of CLLK. The fusion between spleen cells obtained from the CLLK immunized mice and myeloma cells (Sp2) was performed using polyethylene glycol essentially as described previously in Ref. (26). Positive clones were identified by ELISA using micro titer plates coated with either purified recombinant CLK1 or CLL1 expressed in CHO cells as fulllength molecules without any tags. Cells from the positive wells were recloned at least thrice by the limiting dilution method. For antibody production and subsequent purification, hybridomas were grown and allowed to express the MAb in HybridomaSFM (Invitrogen). Monoclonal antibodies were purified by means of affinity chromatography using a HiTrap Protein G HP column (GE Healthcare) under previously described conditions and elution with 50 mM glycine, pH 2.3 (27). The two antibodies, MAb 111 (antiCLK1) and MAb 1613 (CLL1), which were superior in specificity and IHC sensitivity and applied in the following studies, were both of the isotype IgG1kappa.  (Invitrogen) and MES or MOPS SDS running buffer (Invitrogen) (28). Proteins were transferred to the HybondP polyvinylidene fluoride membrane (GE Healthcare) (29). The membrane was blocked in 5% nonfat dried milk and 0.1% HSA, and incubated with primary monoclonal antibodies (0.5 µg/ml). Subsequently, the membrane was washed and incubated with HRPconjugated rabbit antimouse antibody diluted (1:20,000) accordingly to the manufacturer's recommendation (Dako, Denmark) and devel oped by means of the ECL plus Western blotting detection kit (GE Healthcare). For specificity testing of applied antibodies, 1 µl of serum was applied to the gel per 4 mm well width.

sDs-Page and Western Blotting
surface Plasmon resonance  Prior to incubation with MASP-2 and C4, plates were prepared with CL-LK and coated ligands (mannan or DNA) as above. Deposition of C4b was detected with biotin-labeled anti-C4b mAb and HRP-streptavidin. The results shown are representative of three independent experiments. Error bars refer to max and min of triplicate measurements. None of the tested mAbs interfered with ligand binding or complement activation. CL-LK binding to mannan and DNA occurs via two separate binding site, and the latter is not inhibited by mannose, whereas uncharacterized blood components inhibit both types of interaction (10, 11).

human Tissue samples
Human tissue samples were obtained from the tissue bank at the Department of Pathology, Odense University Hospital (Odense, Denmark) and derived from surgically removed specimens fixed 4% phosphate buffered formaldehyde for 12-48 h. Samples were conventionally dehydrated, and subsequently embedded in paraf fin before sectioning (4-5 µm) and mounting on slides.
immunohistochemistry Paraffinembedded, formalinfixed human tissue sections were deparaffinized and rehydrated through serial wash in xylene and decreasing concentrations of ethanol. Endogen peroxidase activity was blocked by incubation with 1.5% H2O2 for 10 min. After wash in TNT buffer, the tissue sections were incubated with DAB+ (Dako) for 10 min followed by staining with hema toxylin. The final immunohistochemical analysis was carried out using "multi block" sections comprising the following nor mal tissues: the cerebellum, esophagus, fetal and adult liver, gall bladder, kidney, large intestine, lung, skeletal muscle, pancreas, parotid gland, placenta, prostate, pylorus, spleen, tonsils, thy mus, thyroid gland, rectum, small intestine, testis, and urinary bladder. The adrenal gland was derived from a patient diagnosed with pheochromocytoma.

image acquisition
Histology slides were scanned at 20× (controls) or 40× mag nification using a NanoZoomerXR (Hamamatsu Photonics, Japan). Image sections were acquired using NDP.view2 software (NanoZoomer Digital Pathology; Hamamatsu Photonics) and final JPG images were all uniformly adjusted for color saturation (+25) and light (−1) in Adobe Photoshop.
resUlTs antibody specificity, affinities, and impact on complement activation The reactivity of the applied MAbs was demonstrated by Western blotting using serum as source of antigens (Figure 1).
This analysis showed that the applied MAbs 1613 (antiCLL1) and 111 (antiK1) only reacted with protein bands correlating with the molecular weight of CLL1 and CLK1, respectively (Figure 1) (7). There was no crossreactivity of the two antibod ies, and both MAbs recognized all isoforms, ensuring detection  of all forms of CLK1 and CLL1 in the tissue sections. To fur ther validate the specificity and reactivity, the two monoclonal antibodies were analyzed by SPR using immobilized purified collectins and antibodies in fluid phase. MAb1613 bound to CLL1 (KD = 0.16 ± 0.007 nM, means ± SD) and to CLLK (KD = 0.14 ± 0.003 nM) but not to CLK1. MAb 111 bound to CLK1 (K1 KD = 5.4 ± 1.8 nM) and to CLLK (KD = 4.6 ± 1.9 nM) but not to CLL1. Again, crossreactivity was undetectable and binding affinities/avidities were of satisfactory strengths. In the characterization of the two applied MAbs, we found that they neither interfered with the binding activity of CLLK to suitable ligands nor did they modulate or inhibit the CLLKmediated complement activation via MASP2 (Figure 2).
immunohistochemical localization of cl-l1 and cl-K1 In the majority of the tested tissues we observed identical localization of CLK1 and CLL1, both in terms of tissue and cell types. Unless the difference in immunoreactivity between the two was striking, the colocalization is not commented further, neither is the absence of staining of the tissues incu bated with isotype matched control antibody. Frequently, the immunoreactivity of the CLK1 MAb (111) was stronger than that of the CLL1 MAb (1613). This may not necessarily reflect an increase in CLK1 quantity in comparison with CLL1 but may originate from the nature of the antibodies (also discussed further below). In the liver, immunoreactivity for CLK1 and CLL1 was asso ciated with hepatocytes with absent staining of Kupffer cells. Staining intensities of CLL1 was pronounced in the centrilobular hepatocytes (Figure 3).
In the kidney, immunoreactivity for CLK1 and CLL1 was especially associated with the tubular system, with the most pronounced staining of the distal tubules (Figure 3), in com parison with proximal tubules. CLL1 immunoreactivity was for some distal tubules distinctly associated with the brush border. Immunoreactivity for both collectins was also associated with the epithelial cells lining the Bowman's capsules, whereas immu noreactivity in the glomerulus itself mainly was associated with CLK1 and only minimally with CLL1. CLK1 immunoreactivity in the glomerulus was associated morphologically appeared to include both podocytes and mesangial cells.
In the lung, CLK1 immunoreactivity was associated with alveolar macrophages, type I and II pneumocytes (Figure 3). CLL1 immunoreactivity appeared only to be associated with alveolar macrophages. In the thyroid gland, cuboidal epithelial cells lining the base membrane of thyroid follicles and parafollicular cells (Ccells) were associated with immunoreactivity for both CLK1 and CLL1 (Figure 4). Most pronounced staining was observed for CLK1.
In the pancreas CLK1 and CLL1 immunoreactivity was associated with the islets of Langerhans and the pancreatic epithelial acinar cells and ducts (Figure 4). Within the islets, the vast majority of cells stained positive, indicating for sure that insulinproducing cells (beta cells) were associated with immunoreactivity and also most likely glucagonproducing cells (alpha cells) as well.
In the adrenal tissue section (Figure 4), derived from a patient diagnosed with pheochromocytoma, the histology was slightly unclear. However, as immunoreactivity for both CLK1 and CLL1 was associated with nearly all cells, it was deducted that the majority of adrenal cells, including both medullary and cortical cells, are associated with the two collectins, similar with previous findings for the localization of CLK1 (7).
In the gall bladder, immunoreactivity for both CLK1 and CLL1 was associated with columnar epithelial cells of the muco sal folds, with increasing intensity toward the luminal side of the folds (Figure 5). Various cell types in the lamina propria stained weakly positive. We observed only scattered staining of cells in the muscularis and serosa layers.
In the duodenum, immunoreactivity for CLK1 and CLL1 was associated with epithelial cells in both the mucosa and sub mucosa. In the mucosal luminal membrane of the villi, especially columnar cells (enterocytes) stained positive (Figure 5). Further and intense immunoreactivity of the mucosa was associated with the crypts of Lieberkuhn, whereas the muscularis externa was only weakly positive for staining. In the submucosa, immuno reactivity was associated with cells of the Brunner's glands.
In the colon, CLK1 and CLL1 immunoreactivity was domi nantly associated with mucosa and especially with columnar epithelial cells in the crypts of Lieberkuhn (tubular glands) (Figure 5). In the lamina propria, the staining was scattered and associated with various cells types. Within the layers of the mus cularis externa and submucosa, staining was also associated with endothelial cells, best illustrated for the localization of CLK1.
In the testis, CLK1 immunoreactivity was associated with germinal epithelial cells lining the tunica propria of the seminif erous tubules, spermatogonia (type A and B), and spermatocytes (primary and secondary) (Figure 6). These cells were embedded in the less immunoreactive Sertoli cells. In the interstitium between seminiferous tubules, Leydig cells and endothelial cells of capillaries stained weakly positive. CLL1 immunoreactivity was weak in comparison with that of CLK1, but the pattern of the two collectins followed each other. In the prostate, CLK1 and CLL1 immunoreactivity was associated with epithelial cells of the prostatic glands, with staining of both acini and ducts (Figure 6). The staining was associated with both columnar pseudostratified and involuted luminal epithelial cells; however, with most intense staining of the basal epithelial cells. In the stroma, scattered staining was observed in various cells types, including staining of endothelial cells. In the ducts, secretory vesicles and concretized material stained weakly positive, especially for CLK1.
In the corpus uterus, CLK1 and CLL1 immunoreactivity was localized to the epithelial cells in the endometrial glands, glandular ducts, and at luminal surface, with comparable stain ing of glandular structures in both the stratum functionalis (compactum and spongiosum) and stratum basale (Figure 6). Both stratified columnar and ciliated cells in the glands stained intensely. In the stroma, the immunoreactivity was moderate but associated with the majority of cells, with a pronounced staining of endothelial cells of the capillaries.
In the skin, CLK1 and CLL1 immunoreactivity was associ ated with the sweat glands and ducts (Figure 7). CLK1 staining was further associated with the basal layer of the epidermis. In the sweat glands and ducts, especially epithelial cells, stained positive, while myoepithelial cells only stained weakly positive.
Within the inner duct, staining was associated with the luminal part and eventual content in the duct. Staining of sporadically distributed and nonidentifiable cells in the dermis was also observed.
In the partoid salivary glands, immunoreactivity for CLL1 and CLK1 was associated with both epithelial glandular (acini) and epithelial ductal cells (Figure 7). Immunoreactivity was localized dominantly to the epithelial cells constituting the salivary ducts and less with the secretory acini. However, the majority of serousproducing epithelial cells were associated with immunoreactivity. Basal epithelial cells of mucin pro ducing stained weakly positive. All three kinds of ducts: intercalated (minor), intralobulated (striated), and major showed equal and dominant immunoreactivity. The immunoreactivity of CLL1 in the salivary serous glands was superior to that of CLK1.
In the fullterm (mature) placenta, CLK1 and CLL1 immu noreactivity was mainly associated with the syncytiotrophoblast layer, at the border of maternal and fetal circulation, and weakly with the underlying cytotrophoblasts associated with a villus (Figure 7).
Various levels of CLK1 and CLL1 immunoreactivity were also found to be associated with the following tissues: the thymus, spleen, tonsil ( Figure S1 in Supplementary Material), esophagus,

cl-K1 and cl-l1 mrna abundancies
To compare protein localization by IHC with site of synthesis for the two collectins, data were retrieved from the three major RNA expression databases HPA, GTEx, and FANTOM5 RNA. For comparison, expression levels were normalized and gradu ated into four categories based on a logarithmic division (Tables S1 and S2 in Supplementary Material). Levels of immuno reactivity were visually validated by three independent persons and categorized similarly ( Table 1). mRNA transcripts encoding CLK1 was detectable in all tested tissues and the major sites of synthesis, grouped in the "high" category, comprised the adrenals, gallbladder, and liver. In all the tested tissues, there was only minimal variance, in terms of a single category shift, i.e., "high" to "medium, " between CLK1 mRNA levels and immunoreactivity, therefore we considered there to be an excellent accordance between site of CLK1 synthesis and protein localization. mRNA transcript encoding CLL1 was not detected in as many tissues as CLK1. Some tis sues were categorized with "extremely low/absent" number of CLL1 transcripts. However, by IHC quite a lot of these tissues were found to be associated with immunoreactivity, albeit in a "low" degree. The major site of CLL1 mRNA synthesis was the liver and placenta and in these tissues the protein was also readily detected. Similar to the observations for CLK1 and using the same criteria, there appeared in general to be accord ance between site of CLL1 synthesis and protein localization (discussed further below).
To associate the protein localization with an eventual local functionality of the two collectins, mediated via the presence of MASPs, localization of MASPs (and MAps), and synthesis of their respective mRNAs were evaluated by the same approach. As the major RNA expression databases currently do not take alternatively splicing of the MASP genes into consideration, data were retrieved and gathered from the previous work by Thiel and colleagues and Garred and colleagues (30)(31)(32)(33). MASP3 expres sion appeared to both overlap and being as widely distributed as the two collectins, whereas the other MASPs and MAp had a restricted pattern of tissue localization, with the liver being a tissue of major synthesis and/or detection: an observation, which also applied for MBL.

DiscUssiOn
The present work describes the localization of CLL1 and CLK1 in human tissues as determined by immunohistochemistry and summarizes further their mRNA tissue profiles derived from transcriptome databases. Both CLL1 and CLK1 were demon strated in epithelial cells in a variety of tissue throughout the human body. Of all the tested MAbs, MAbs 1613 (antiCLL1) and 111 (antiCLK1) had the best sensitivity and specificity. The two MAbs demonstrated excellent immunoreactivity in the three tis sues, the liver, kidney, and adrenals, wherein human CLK1 and CLL1 localization previously have been demonstrated (6,7). The major positive cell types comprised, hepatocytes, renal epithelial cells of tubules, and medullary and cortical cells of the adrenals. In addition to the adrenals, tissues from other endocrine glands, i.e., pancreas, demonstrated a similar con vincing excellent immunoreactivity for both collectins, derived from cells in the islets of Langerhans and epithelial cells of the ducts. Exocrine tissues of the digestive system comprising the gallbladder, duodenum, colon, and also partly the stomach and esophagus were also associated with epithelial and mucosal immunoreactivity for generally both collectins. Other exocrine tissues, comprising sexspecific organs, such as the testis, pros tate, and uterus, had also excellent to moderate immunoreactivity for both collectins. Among all the analyzed tissues, the testis and uterus appeared to be the two tissues with the relative highest CLK1 immunoreactivity. Again, epithelial cells and mucosa in the uterus were the major source of immunoreactivity, whereas immunoreactivity in testis was associated with germinal epithe lial cells of the tubules. Among the tissues analyzed, the highest CLL1 immunoreacti vity was observed in the liver, followed by the kidney and parotid gland, wherein immunoreactivity was also associated with epithelial cells. Our observation of CLK1 synthesis in various tissues falls in line with previous work by Wakamiya and colleagues, who by immunofluorescence techniques demonstrated partly overlapping localization of CLK1 in murine tissues, using a polyclonal antimouseCL K1 antibody (34). In general, we observed a stronger staining of CLK1 than of CLL1. This may reflect that CLK1 is more abundant than CLL1, although their levels in the circulation are approximately the same (12), or it may simply be a matter of affinities of the applied MAbs in combination with availability of antigen epitopes on the fixed and embedded tissue sections.
Retrieval and normalization of mRNA levels from three transcriptome databases demonstrated accordance between site 1 | Levels of RNA expression: The symbols "+++," "++," "+," and "-"denote high, medium, low, and absent/extremely low expression, respectively, based on the criteria established in Figures S1 and S2   of synthesis and protein localization. The only tissue, wherein there appeared to be a notable difference, was for CLL1 in the salivary gland. By IHC CLL1 localization was judged to be medium, whereas CLL1 mRNA appeared to be absent. Other CLL1specific antibodies showed varying staining of particular the salivary gland (data not shown), making us hypothesize that the observed disagreement could reflect some sort of uncha racterized alternative splicing of CLL1 in this tissue. Throughout the IHC staining it was evident that within the majority of the tissues, the localization pattern of two col lectins was identical; meaning that exactly the same cells in a given tissue demonstrated immunoreactivity for both collec tins. This is best exemplified when two neighbor sections were mounted and processed, as illustrated, e.g., with the corpus uterus ( Figure 6). Thus, in the majority of tissues there is opportunity for the making of CLLK heteromeric complexes, and as previously described, this structure also appears to be the most thermodynamic stabile conformation (11). In some of the tissues, comprising the thyroid gland, skeletal muscle, skin, urinary bladder, and partly the testis and esophagus, CLK1 appeared to be present in large excess in comparison with CLL1, as judged by immunoreactivity; it is likely that CLK1 homomers will be the dominating form in these tis sues. The precise distribution of homomers vs. heteromers in different tissues should not be judged by immunoreactivity and remains thus to be characterized in detail. Although the heteromers are eminent in their association with MASP2 and C4b deposition, in comparison with the homomers, it is worth emphasizing that both types of homomers interact well with MASP1/3, and may mediate downstream complement activation via those alone (8,11).
To further illustrate the presence of heteromers in different tissues we tried to establish a proximity ligation assay using antibodies usable on formalin fixed sections but without con vincing results. By using purified and fixed CLLK it appeared that even the best combination of antibodies partly shadowed for each other in proximity ligation assays and were only capable of detecting the very high oligomers of CLLK, with a sensitivity of only 0.2 µg of purified CLLK per ml immobilized onto polylysinetreated object glasses (data not shown).
The overlapping localization of the two collectins in the same cells justifies, with a few exceptions, possible assembly and presence of the heteromeric CLLK in most tissues. Based on a combination of previous observations and unpublished results by our laboratory, it appears that the oligomeric state of CLLK depends on the relative content of CLL1 and the ratio of the CLK1a/d isotypes (11). As all of our antiCLK1 MAbs recognize the two isotypes equally well, the immunohi stochemical results does not per se allow us to deduct any final conclusions on the variability of oligomers in different tissues. However, tissues with a relative large expression of CLL1 could potentially favor assembly of CLLK into large oligomers, rang ing from 2 to 6 subunits.
We have previously demonstrated that the binding activity of CLK1 and CLLK, and hence also their complement activating ability, in serum/plasma is inhibited by unknown factors (11). This has made it difficult to comprehend the role of the two col lectins as bona fida activators of complement. In the light of the widespread presence of CLK1 and CLL1 in various tissues, it is possible that binding activity in the periphery, in the absence of inhibitory blood components, may be more efficient.
As the hitherto described biological functions of CLK1 and CLL1, in terms of complement activation or involvement in embryogenesis, appear to rely on MASPs it is relevant to investigate colocalization with MASPs in the periphery. However, there is a lack of suitable antibodies specific for the three products of the MASP1 gene, MASP1/3, and MAp44, but the summarized mRNA profile presented in Table 1 shows that MASP3 synthesis, in contrast with all other MASPs and MAps, appears to overlap greatly with the localization of CLK1 and CLL1. Thus, it is likely that the role of CLK1 and CLL1 in the periphery is mainly mediated via MASP3, which was recently shown to activate profactor D to factor D, and thereby potentiate the alternative pathway and amplification loop (35)(36)(37). Although (pro) factor D mainly is synthesized in adipose tissue (hence the alias "adipsin") various tissues synthesize minor amounts of profactor D, which could be a target for MASP3 in complex with CLK1/L1/LK, and thereby potentiate the complement amplification loop in the periphery, upon encounter and binding of collectins to suitable (microbial) ligands. Alternatively, the two collectins may in the periphery, and in parallel with MBL and C1q, exert some of their functions by interacting with the metalloproteases bone morphogenic protein 1 and tolloidlike proteases, involved in extracellular matrix assembly and growth factor signaling (38). Interactions between CLK1, L1, or LK with these metallo proteases remain to be investigated.
In the light of our (co)localization of CLK1 and CLL1 to peripheral tissues it appears that the previously focus on their roles in the circulation, liver, and kidney, may have disparaged a compelling localization in especially exocrine/endocrine tissues and mucosa, suggestive of that CLL1, CLK1, and CLLK may play roles on epithelial surfaces in general and in tissue characte rized by a high degree of exocytosis. The localization of CLK1 and CLL1 reminds also in many ways of the localization of the collectin surfactant protein D (39).
aUThOr cOnTriBUTiOns SH and MH designed the study and carried out: antibody development, characterization, immunohistochemistry, data analysis, computational bioinformatics, and wrote the paper, on which all authors commented. JA and KB carried out antibody development, characterization, and immunohistochemistry. EH, ON, and HS participated in designing and performing the immuno histochemistry and analyzing data. KS carried out development of antibodies. AS and JG carried out SPR analysis and analyzed data.

acKnOWleDgMenTs
The authors thank Anette Holck Draborg, Department of Cancer and Inflammation Research, University of Southern Denmark for critical reading of the manuscript and for giving valuable comments. The authors thank Lisbeth Mortensen, Department of Pathology, Odense University Hospital for technical advice relating to immunohistochemistry.