In vivo determination of collecting lymphatic vessel permeability to albumin: a role for lymphatics in exchange

Abstract

While it is well established that the lymphatic vasculature is central to fluid and solute homeostasis, how it accomplishes this task is not well defined. To clarify the basic mechanisms underlying basal fluid and solute homeostasis, we assessed permeability to rat serum albumin () in mesenteric collecting lymphatic vessels and venules of juvenile male rats. Using the quantitative microfluorometric technique originally developed for blood capillaries, we tested the hypothesis that as a consequence of venules and collecting lymphatics sharing a common embryological origin, their would not differ significantly. Supporting our hypothesis, the median collecting lymphatic (3.5 ± 1.0 × 10⁻⁷ cm s⁻¹, N= 22) did not differ significantly from the median venular (4.0 ± 1.0 × 10⁻⁷ cm s⁻¹, N= 8, P= 0.61). For collecting lymphatics the diffusive permeability (P_d= 2.5 × 10⁻⁷ cm s⁻¹) was obtained from the relationship of apparent and pressure. While the measured , P_d and estimated hydraulic conductivity of collecting lymphatics and venules were similar, the contribution of convective coupling differs as a result of the higher hydrostatic pressure experienced by venules relative to collecting lymphatics in vivo. In summary, the data demonstrate the capacity for collecting lymphatics to act as exchange vessels, able to extravasate solute and filter fluid. As a consequence these data provide experimental support for the theory that prenodal lymphatic vessels concentrate intraluminal protein.

Introduction

The lymphatic system is known to be vital for fluid homeostasis, lipid absorption and immune surveillance under conditions of health (Marble et al. 1934; Mayerson, 1963; Gnepp, 1984). Less widely understood is the role of the lymphatic system in disease states such as oedema, inflammation, cancer and obesity (Alitalo et al. 2005; Harvey, 2008; Stanton et al. 2008). A recent upsurge in interest, probably reflecting the development of novel molecular markers specific for lymphatic vessels and realization of their importance to disease, has led to significant advances in this field (Cueni & Detmar, 2008). Knowledge of lymphatic vessel function, though, still lags behind that of the blood vessels given the difficulty in studying these structures that are sensitive to both mechanical and chemical stress (Mayerson, 1963). While lymphangiogenesis, lymphatic pumping and lymph node function are receiving current attention, other elusive aspects of the lymphatic system that deserve attention still remain.

One aspect of lymphatic function relates to its role in exchange, in particular, lymphatic vessel permeability to solute. Lymphatic permeability was studied in the 1960s predominantly as a means for gaining insight into blood capillary permeability. The few experiments performed during that era concluded that the lymphatic system retains all solutes with molecular mass larger than 2300–6000 (Mayerson, 1963). Since radiometric methods used at that time had limited sensitivity compared to present day microfluorometric methods that facilitate quantitative measurement of solute flux in living tissue (Huxley et al. 1987), we decided to use the latter approach to test the hypothesis that lymphatic vessels possess a macromolecular permeability analogous to that of the blood microvessels, particularly venules, under basal conditions.

Blood microvessels are permeable to solute and water. In health, microvessel permeability is influenced by tissue function and can be modestly regulated both positively and negatively. Microvessels possess a barrier of finite permeability to provide adequate perfusion of the tissues with nutrients, oxygen and lipids. However, it is widely taught that the processes controlling exchange, whether water or solute, are passive and that only under inflammatory states does permeability change – usually an increase whereupon materials contained in the vascular compartment leak into the tissue space, thereby compromising function. Furthermore, microvascular exchange has been represented pictorially in multiple texts by a single exchange vessel, a capillary, connecting a single arteriole with a single draining venule (Ratnoff, 1983; Boron & Boulpaep, 2005). In this oversimplified model, filtration of fluid occurs on the arterial (high pressure) end of the capillary and reabsorption of 90% of this filtrate occurs on the venous side of the capillary (Guyton, 1971; Boron & Boulpaep, 2005). In opposition to this view, many experimental studies have provided evidence that filtration occurs throughout the microvascular network of vessels, with reabsorption only occurring during special states (e.g. increased tissue hydrostatic pressure) or in encapsulated organs (Renkin, 1986; Michel & Phillips, 1987; Levick, 1991). Appreciation of this discrepancy is important because the mental picture of a single capillary replacing a microvessel network obscures the clinically important role that the missing constituents, the lymphatic vessels, play in the establishment and maintenance of proper fluid homeostasis. Thus, the fact that the microvessels filter approximately 50% of the circulating plasma proteins per day – all of which is returned to the circulation by the lymphatic system – reinforces this assertion (Drinker, 1937; Gnepp, 1984; Renkin, 1986).

The focus of this manuscript, the collecting lymphatic vessels, share important features with veins, including the presence of valves, smooth muscle and chronic exposure to a low hydrostatic pressure environment. Notably, the developmental origin of lymphatic endothelium has recently been shown to be the veins (Srinivasan et al. 2007). Given their morphological similarity and common embryological origin, we hypothesized that the permeability of collecting lymphatic vessels and venules to albumin would not differ in a given tissue. Additionally, given that collecting lymphatics exhibit a greater hydrostatic pressure than microlymphatics, we posited that collecting lymphatics were a potential site for solute exchange (Zweifach & Prather, 1975). Here, for the first time, we measure basal permeability values to rat serum albumin for in vivo rat mesenteric collecting lymphatics. To accomplish this task we developed and validated a reliable method for cannulation and perfusion of collecting lymphatics in rat mesentery.

From measurements of albumin flux at multiple hydrostatic pressures, we determined the diffusive permeability to albumin (P_d) and the convective coupling of albumin flux to transmural fluid flux of collecting lymphatic vessels from rat mesentery. Utilizing these data, estimates of volume flux of the collecting lymphatics over a range of physiological pressures were determined. We found, consistent with our hypothesis, that permeability to albumin does not differ between venules and collecting lymphatic vessels. Preliminary results from this study have been presented in abstract form (Scallan & Huxley, 2007).

Nomenclature

To avoid confusion the nomenclature and structure of the lymphatic vessels will be defined. Interstitial fluid formed from the capillary filtrate enters blind-ended sacs, composed only of an endothelial layer, tethered to the interstitial matrix. These 10–60 μm diameter sacs are called initial (or terminal) lymphatics. The fluid, thereafter called lymph, moves into lymphatic vessels of a similar diameter, termed microlymphatics (or lymphatic capillaries), consisting of an endothelial layer and basement membrane (Vainionpääet al. 2007). Since the phrase ‘lymphatic capillary’ is sometimes applied to initial lymphatics, and ‘lymphatic capillaries’ may be much larger than traditional capillaries carrying whole blood, this ambiguous phrase will be avoided in this manuscript. Microlymphatic vessels then carry lymph towards the larger collecting lymphatic vessels. Collecting lymphatics (50–200 μm diameter), the focus of this study, are composed of an endothelial layer, a basement membrane, smooth muscle cells, and contain endothelial valves (von der Weid & Zawieja, 2004). The smooth muscle layer is unique in that it possesses both tonic and phasic contractile activity (Schmid-Schönbein, 1990; von der Weid & Zawieja, 2004). The phasic activity gives rise to spontaneous contractions that aid in propelling lymph along the intervalvular segments of the lymphatic vessel, called lymphangions. After passing through larger collecting lymphatic ducts, lymph is finally propelled to the thoracic duct, emptying into the left subclavian vein. More comprehensive reviews can be found elsewhere (Gnepp, 1984; Schmid-Schönbein, 1990).

Methods

Ethical approval

Animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Missouri and conducted in accordance with the National Institutes of Health's Guide for the Care and Use of Laboratory Animals. All animals were killed after the experiment by an overdose of anaesthetic followed by bilateral pneumothorax and aortic transection in agreement with the protocol.

General surgical preparation

All experiments were performed in situ on juvenile male (35–55 days, 200–250 g) Sprague–Dawley rats (Hilltop Lab Animals, PA, USA) housed three to a cage for a minimum 1 week acclimation period prior to the experiment. For these initial studies, juvenile males were chosen for their lack of visceral fat, which facilitated location and subsequent cannulation of collecting lymphatics. One venule or collecting lymphatic was cannulated per animal.

Following induction of anaesthesia with Inactin (i.p., thiobutabarbital, Research Biochemicals Int., MA, USA; 128 mg (kg body weight)⁻¹), surgical exteriorization of the mesentery and continuous superfusion (2–3 ml min⁻¹) with mammalian Krebs at 37 ± 0.5°C, a vessel was located under the Zeiss dissection microscope. The animal on a custom, heated Plexiglas board was then transferred to a Leitz Diavert inverted microscope and the suffused preparation was allowed to equilibrate for approximately 30 min prior to vessel cannulation and data collection.

Venular and lymphatic microvessel classification

To facilitate comparison of our data with previously published measurements of mammalian venular permeability (Rumbaut & Huxley, 2002; Sarelius et al. 2006), only relatively straight, unbranched venules with brisk flow and fewer than two adherent leukocytes were used. Vessels enveloped by adipose tissue were avoided because the fat cells make cannulation nearly impossible. Venules were identified by size, paucity of smooth muscle and converging flow on either side of the vessel segment as described by Chambers & Zweifach (1944). Collecting lymphatic vessels were identified by the presence of valves, spontaneous contractions and their transparent contents (Zweifach & Prather, 1975). Leukocytes were often visualized in collecting lymphatics, which were not studied if more than two adherent cells were seen.

Solutions

Mammalian Krebs

All superfusion and perfusion solutions were prepared fresh and used on the same day. Unless otherwise stated all materials were purchased from Sigma (MO, USA). The Krebs solution consisted of (in mmol): 141.4 NaCl, 4.7 KCl, 2.0 CaCl₂·2H₂O, 1.2 MgSO₄, 1.2 NaH₂PO₄·H₂O, 5.0 d-glucose, 3.0 NaHCO₃ and 1.5 Na Hepes. The pH of the solution was 7.4 ± 0.05 at 37°C.

Krebs–BSA

Dialysed (see online Supplemental Methods) bovine serum albumin (BSA, Sigma cat. no. A7906) was added to the Krebs solution to achieve a final concentration of 1 mg ml⁻¹ on the day of the experiment. The osmolarity, measured by freezing point osmometry, was generally between 292–297 mosmol l⁻¹ after the addition of BSA.

Labelled protein

The macromolecular probe used in this study, rat serum albumin (RSA, Sigma cat. no. A6272), was bound to the Alexa-488 fluorophore (Invitrogen, CA, USA) by modifying the manufacturer's protocol (see Supplemental File). The perfusate solution contained 10% (w/v) labelled RSA (1 mg) with 9 mg of unlabelled RSA and was brought to a volume of 1 ml with freshly made Krebs solution so that the final total protein concentration was 10 mg ml⁻¹. A washout solution was made, with unlabelled protein, at an identical total protein concentration. Both solutions of 10 mg ml⁻¹ RSA possessed an oncotic pressure of 4.1 cmH₂O calculated from the Landis–Pappenheimer equation (Landis & Pappenheimer, 1963).

Assays for total protein and albumin concentration of lymph

The concentration of albumin and total protein were measured in samples of lymph, peritoneal fluid and plasma (N= 4). Lymph was obtained after cannulation with an empty siliconized single lumen micropipette (tip diameter, 25 μm) of collecting lymphatic vessels in rat mesentery covered with mineral oil. Peritoneal fluid was obtained by swabbing the mesentery with a haematocrit tube immediately following the midline incision. Approximately 20–60 μl of blood-free lymph and peritoneal fluid were obtained using these methods. Plasma was obtained by centrifuging ∼3 ml of blood (taken by intracardiac puncture) for 10 min at 3220 g.

Total protein concentration of each sample was determined (Supplemental File) by the micro BCA protein assay, while albumin concentration was determined by the Albumin Blue 580 fluorescence assay (Kessler et al. 1997). All solutions were made fresh on the day of use. The albumin or total protein concentration was then used to calculate the albumin or total protein oncotic pressures, respectively, for each sample using the Landis–Pappenheimer equations (Landis & Pappenheimer, 1963).

Measurement of microvessel protein flux

The method for determining solute permeability (P_s) to proteins in mammalian blood microvessels and its limitations is described in several publications (Huxley et al. 1987; Rumbaut & Huxley, 2002; Sarelius et al. 2006). In short, collecting lymphatics and venules studied here were cannulated with custom-made pipettes and perfused with either fluorescently labelled RSA (dye) or unlabelled RSA (washout) in Krebs solution. Details of the pipette fabrication and cannulation procedures are to be found in the Supplemental file. In experiments using theta pipettes the vessel was perfused at constant, selected pressures under user control. Switching a valve changed the perfusate from the washout solution to the dye solution for a time sufficient to make measurements of fluorescence intensity without altering pressure. Additionally, this system allowed the native vessel pressure (P_lumen, Table 1) to be measured prior to assessment of solute flux by keeping the fluorescent dye front from moving within the pipette. Pressure in spontaneously contracting collecting lymphatics is constantly changing so only diastolic pressure (during relaxation) was recorded. Importantly, one must understand that these native pressure measurements (Table 1) were made before measuring solute flux to establish the native pressure range for collecting lymphatics. To measure solute flux the pipette pressure was raised above that of the vessel in order to elicit flow of the perfusate. During the solute flux measurements, the perfusion pressure of the theta pipette was measured when there was zero flow through the non-perfusing side. This pressure was plotted in Fig. 3, but was not reported in Table 1.

Table 1.

Basal permeability (P_s) values for rat serum albumin and measured vessel characteristics

Date	Vessel type	Diameter (μm)	Plumen (cmH2O)	(×10−7 cm s)−1
02/01/08	CL1	113	1	3.6
06/03/08	CL2	76	10	4.0
08/06/08	CL3	101	∼0	8.8
08/11/08	CL4	88	2.5	16.7
08/12/08	CL5	113	∼0	1.5
08/27/08	CL6	101	5	4.8
09/01/08	CL7	139	10	2.7
09/05/08	CL8	139	16	1.0
09/09/08	CL9	101	6	1.5
09/24/08	CL10	139	11	2.7
09/25/08	CL11	126	9	7.0
10/20/08	CL12	139	12	2.7
12/05/08	CL13	88	5	3.2
12/17/08	CL14	88	9	6.4
01/21/09	CL15	101	2	15.2
01/23/09	CL16	88	∼0	4.0
2/05/09	CL17	50	5	3.3
02/06/09	CL18	126	10	9.9
03/02/09	CL19	113	10	0.2
03/05/09	CL20	63	5	3.1
03/10/09	CL21	88.2	8.5	4.3
03/11/09	CL22	88.2	8.5	3.2
08/26/08	V1	63	20	6.1
09/03/08	V2	38	23	6.2
09/26/08	V3	50	17	3.3
04/17/09	V4	25	17	3.9
09/23/09	V5	38	21	4.1
09/23/09	V6	38	17	9.1
09/23/09	V7	38	20	3.4
09/23/09	V8	25	21	0.5
Mean ±s.e.m.	CL	103 ± 5*	7 ± 1*	5.0 ± 0.9
	V	39 ± 4	20 ± 1	4.6 ± 0.9
Median ± MAD	CL			3.5 ± 1
	V			4.0 ± 1

P_s values are ×10⁻⁷ cm s⁻¹; CL, collecting lymphatic; V, venule; P_lumen, native vessel pressure (measured during relaxation phase for lymphatics); m.a.d., median absolute deviation; *P < 0.01 versus venules.

Figure 3. Juvenile rat mesenteric collecting lymphatic permeability to albumin () plotted as a function of hydrostatic pressure (cmH₂O) to determine the contribution of (pressure-driven) convective coupling to solute flux.

The predicted relationship (continuous line) was obtained from simultaneous solution for P_d (cm s⁻¹) and L_p(1 −σ) (cm s⁻¹ cmH₂O⁻¹) in eqn (2). The dashed line marks the effective oncotic pressure (σΔπ, cmH₂O). The apparent permeability is equal to the diffusive permeability (P_d) at zero net filtration pressure (intersection of continuous and dashed lines) and the limiting slope is equal to L_p(1 −σ).

Direct measures of solute flux (J_s, mmol s⁻¹) were made over an area of vessel defined by a rectangular diaphragm (width ≥ 4 vessel diameters; length = 8 vessel diameters). When dye-labelled solute filled the vessel lumen there was an initial step increase in fluorescence intensity (I_o), followed by a gradual, but linear, increase in intensity as fluorescent probe accumulated in the interstitial space over time (dI_f/dt) in addition to the signal generated from the fluorophore flowing through the vessel lumen. Apparent solute permeability (P_s, cm s⁻¹) was calculated from the equation relating solute flux (J_s, mmol s⁻¹) to surface area (S, cm²) at a constant concentration difference (ΔC, mmol ml⁻¹) (Huxley et al. 1987):

(1)

Vessels were assumed to be circular and to have a volume-to-surface area ratio of D/4.

Triplicate measures of flux were attempted in each vessel. In some vessels, solute flux was measured at different hydrostatic pressures to assess the relationship between convective and diffusive flux. Previously, we and others (Curry, 1984; Huxley et al. 1987) have shown that when convective water movement contributes to J_s, measures of P_s overestimate the true diffusive permeability (P_d). These terms are related by the equation:

(2)

where Pé is the Péclet number:

(3)

Equation (2) is a modified form of the Patlak, Goldstein and Hoffman equation (Patlak et al. 1963). Assuming a homoporous barrier and a small constant interstitial oncotic pressure,

(4)

where ΔP (cmH₂O) is the hydrostatic pressure difference between the vessel lumen and the interstitium (P_c−P_i), L_p is the hydraulic conductivity (cm s⁻¹ cmH₂O⁻¹), σ is the unit-less reflection coefficient for solute and Δπ (cmH₂O) is the oncotic pressure difference (π_c−π_i) exerted across the vessel wall. Plotting P_s against hydrostatic pressure and simultaneously solving for P_d and L_p(1 −σ) from eqn (4) yields a curve describing the data. When the net filtration pressure (ΔP−σΔπ) is equal to zero, the y-intercept becomes P_d. The limiting slope of this line is L_p(1 −σ) and is labelled in Fig. 3 (Huxley et al. 1993).

Volume flux (J_v, cm³ s⁻¹) across a semi-permeable membrane of surface area (S) is described by the modern form of the Starling equation:

(5)

where L_p, ΔP, σ and Δπ are as defined above. However, this equation describes J_v during transient (i.e. when pressure is changing) rather than steady-state conditions. Another equation has been developed that describes steady-state J_v (Michel & Phillips, 1987):

(6a)

Equation (6a) can be rewritten to group the unknown terms on one side of the equation:

(6b)

where π_c is the colloid osmotic pressure in the vessel. Neither eqn (5) nor (6) assume large hydrostatic pressures or large Pé values.

Statistical analyses

Prism (GraphPad Software, CA, USA) software was used for all statistics. To determine whether the P_s to albumin in collecting lymphatic vessels differed significantly from the venular , the non-parametric Mann–Whitney U test was performed at a 95% confidence level to compare the medians because data from both groups were left skewed (i.e. not normally distributed). Distribution normality was determined by the D’Agostino–Pearson K² test. The Kolmogorov–Smirnov two-sample test was performed (Daniel, 1990) to compare the distributions of collecting lymphatic and venular values with 95% confidence. To compare the medians of all three skewed data sets, the non-parametric one-way ANOVA Kruskal–Wallis test was performed at a 95% confidence level. Medians were reported as median ± median absolute deviation (MAD). Means were compared with the Student's unpaired t test and reported as means ± standard error of the mean (s.e.m.) to compare the present permeability data with published values. For comparisons of total protein and albumin concentration differences between peritoneal fluid and lymph, Student's paired t tests were performed and the means ±s.e.m. were reported.

Power analysis performed beforehand indicated that to detect a 20% change between the two groups (1 × 10⁻⁷ cm s⁻¹), with a 95% confidence (significance level P < 0.05), approximately 21 animals were needed in each group (Neter et al. 1990).

Results

Raw tracings of solute flux

Figure 1 illustrates a raw data tracing of the fluorescence intensity of the probe (Alexa-488 conjugated RSA) as a function of time from experiments on two separate animals. The tracing on the left is from a collecting lymphatic vessel perfused at a pressure of 5 cmH₂O and the tracing on the right is from a similar vessel perfused at a pressure of 15 cmH₂O. Visible spontaneous contractions, shown in the expanded view as fluctuations in the fluorescence intensity, are evident from the tracing in A, while the tracing in B lacks these fluctuations. Both patterns were encountered in the present study only after cannulation, since prior to cannulation all collecting lymphatic vessels exhibited spontaneous contractions.

Figure 1.

Raw data tracings from two different rat mesenteric collecting lymphatics A, collecting lymphatics perfused at low pressures exhibit spontaneous contractions at a frequency of ∼15 min⁻¹ (see ∼90 s tracing in inset). In this case, perfusion pressure was 5 cmH₂O. B, upon perfusion at a higher pressure of 15 cmH₂O, spontaneous contractions are greatly attenuated so that only noise is evident (see ∼60 s inset).

Basal albumin permeability of collecting lymphatics and venules

Table 1 reports the apparent albumin permeability values of all vessels in this study along with their respective native pressure and diameter measurements. The of collecting lymphatics and venules did not differ significantly whether analysed as means (P= 0.80, 5.0 ± 0.9 × 10⁻⁷ cm s⁻¹ (N= 22) vs. 4.6 ± 0.9 × 10⁻⁷ cm s⁻¹ (N= 8)) or medians (P= 0.61, 3.5 ± 1 × 10⁻⁷ cm s⁻¹vs. 4.0 ± 1 × 10⁻⁷ cm s⁻¹) consistent with our hypothesis. As anticipated from the work of Zweifach & Prather (1975), collecting lymphatic native pressure (P_lumen) correlated with diameter (P= 0.03, Table 1). When P_lumen and diameter of collecting lymphatics was compared to that of venules, lymphatic pressure was significantly lower (P < 0.0001) and diameter was significantly greater (P < 0.0001).

Comparing frequency distributions for all 22 collecting lymphatic values reported in Table 1 (Fig. 2A) and 26 rat mesenteric venular values (Fig. 2B) (Rumbaut, 1998; Rumbaut & Huxley, 2002) provides additional support for the hypothesis. Both frequency distributions were strikingly similar and left skewed. The Kolmogorov–Smirnov two-sample test (Daniel, 1990) concluded that these two distributions did not differ (P= 0.37). Since the distributions are not statistically normal (P < 0.0001), it is more accurate to compare the medians instead of the means of the two populations (Neter et al. 1990; Rumbaut & Huxley, 2002). The Mann–Whitney U test was performed and demonstrated that the median collecting lymphatic ( = 3.5 ± 1 × 10⁻⁷ cm s⁻¹, N= 22) and median venular ( = 5.6 ± 2 × 10⁻⁷ cm s⁻¹, N= 26) values for albumin permeability did not differ significantly (P= 0.12). When we tested for any differences between all three groups by performing the Kruskal–Wallis test, none were observed (P= 0.24). Finally, when we compared our measured venular values to those previously reported (Rumbaut, 1998; Rumbaut & Huxley, 2002), we discovered that they did not differ (P= 0.31).

Figure 2.

Frequency distribution of basal apparent permeability (, ×10⁻⁷ cm s⁻¹) values of juvenile male rat mesenteric collecting lymphatics (A) and venules (B) Venular data are redrawn from a dissertation by Rolando Rumbaut (Rumbaut, 1998; with permission) and do not contain any venular measures from this study. All data are plotted as a percentage of the total.

Convective coupling of albumin flux

All apparent values from collecting lymphatics were plotted in Fig. 3 to illustrate the dependence of solute flux on hydrostatic pressure. The dashed line demarcates the effective osmotic pressure (σΔπ= 3.85 cmH₂O) of the perfusion solution when σ is 0.94. Where it meets the continuous line is the true diffusive permeability to albumin (P_d= 2.5 × 10⁻⁷ cm s⁻¹). The limiting slope of the continuous line is equal to L_p(1 −σ) = 1.3 × 10⁻⁸ cm s⁻¹ cmH₂O⁻¹, and is used to calculate the Péclet number. Additionally, the L_p of the pathway conducting both solute and water can be estimated from this term if σ is known.

Predicted solute and volume flux from the albumin concentration of lymph and peritoneal fluid

Table 2 summarizes the measured albumin and total protein concentration of plasma, peritoneal fluid and lymph from four animals. The calculated oncotic pressures for total protein and albumin, given for each compartment, were then used to predict the relationships illustrated in Fig. 4.

Table 2.

Measured total protein and albumin concentration of lymph, peritoneal fluid and plasma

	Plasma	Peritoneal fluid	Lymph
Total protein (mg ml−1)	41.6 ± 2.4	31.3 ± 0.9	32.4 ± 1.1
Albumin (mg ml−1)	26.2 ± 0.8	9.5 ± 1.5#	13.2 ± 1.5#
Albumin/total protein ratio (%)	63.4 ± 3.7	30.3 ± 4.5 #	40.9 ± 5.1#
Total protein oncotic pressure* (cmH2O)	23.9 ± 1.7	7.2 ± 1.4†	9.4 ± 1.0†
Albumin oncotic pressure* (cmH2O)	11.9 ± 0.4	3.9 ± 0.6#	5.5 ± 0.7#

*Oncotic pressure is calculated from the Landis–Pappenheimer equation for total protein or albumin, respectively. ^†,# Paired t tests between peritoneal fluid and lymph; ^†P < 0.05, ^#P < 0.1. Values are reported as means ±s.e.m. from 4 animals.

Figure 4.

Predicted solute and volume flux of mesenteric collecting lymphatics over pressure using the average concentration differences reported inTable 2 Positive and negative values indicate fluid filtration and reabsorption, respectively. A, solute flux per unit surface area and concentration difference (J_s/SΔC) is plotted as a function of hydrostatic pressure using eqn (2). The dashed line is the effective oncotic pressure calculated from the measured albumin concentration using the Landis–Pappenheimer equation (Landis & Pappenheimer, 1963). Where it intersects the continuous line is the true diffusive permeability, P_d. B, volume flux per unit surface area on the ordinate as a function of hydrostatic pressure (abscissa) is calculated by substitution of P_d and L_p into eqn (5) (continuous line, transient state) and eqn (6b) (dashed curve, steady state). Reflection coefficient (σ) was assumed equal to 0.94 (Kendall & Michel, 1995). The effective oncotic pressure for each condition is labelled at the x−intercept. The dashed line parallel to the steady state graph is to illustrate how transient state volume flux would appear when interstitial protein is negligible as occurs in many experimental, but not in vivo, conditions. The dotted line tangent to the steady state graph is only to show where the effective oncotic pressure lies. The limiting slope of each line is the hydraulic conductivity, L_p. C, idealized schematic of predicted collecting lymphatic solute and volume flux for the concentration difference in Table 2. Lighter shading corresponds to a lower albumin concentration and vice versa. The lymph rat serum albumin (RSA) concentration is believed to increase along the vessel length as water is filtered.

Predicted solute flux per unit surface area per concentration difference (J_s/SΔC) was plotted over pressure in Fig. 4A. To accomplish this, P_d and J_v/S (obtained below) were substituted into eqn (2). The limiting slope of this line is equal to L_p(1 −σ).

The volume flux per unit surface area (J_v/S) as a function of pressure was predicted for transient and steady-state conditions by substituting P_d and L_p into eqns (5) and (6b) (shown in Fig. 4B). The calculated oncotic pressure difference for albumin (Δπ) was substituted into eqn (5), while the calculated oncotic pressure of lymph albumin (π_L) was appropriately substituted into eqn (6b). Both the steady-state and transient state curves are shown over a wide range of pressures. The x-intercepts correspond to the transmural effective oncotic pressures and are labelled. The limiting slope of all lines is equal to L_p.

Discussion

The present study was a test of the null hypothesis that the albumin permeability of collecting lymphatic vessels would not differ from venules. To our surprise we accepted our hypothesis, since the P_s medians between the two groups did not statistically differ. Additionally, we noted a remarkable similarity in the distributions of the data sets, which were both left-skewed. From the plot of apparent permeability data over pressure we were able to estimate the true diffusive permeability (P_d) to albumin of rat mesenteric collecting lymphatics.

Methodological limitations for the measurement of solute flux

Measurement of solute flux is determined from the step change in fluorescence intensity upon filling of the vessel and subsequent change in intensity over time. This measurement could be compromised if there were to be photobleaching of the dye during the measurement period. Further, photobleaching of the dye is associated with free radical formation, which itself could alter barrier properties.

Photobleaching of the fluorescent probe was evaluated on the same equipment as used for experiments by exciting a sample of the conjugated probe on a standard glass slide with a spacer (thickness of 150 μm) and coverslip. Emission of the fluorescent probe, measured over 4 h of constant excitation in vitro, did not vary, demonstrating the photostability of this labelled macromolecular probe. During experiments in vivo, the longest time the interstitial probe was excited was 30 min; the probe in the vessel lumen was only excited for seconds because flow carried it out of the measuring window.

All experiments were performed on juvenile male rats (< 60 days) to take advantage of their lack of perivascular adipose tissue. To calculate the effective oncotic pressure for the perfused albumin solution, σ of albumin must be known for collecting lymphatics. As this value has never been measured, we assumed that it was equal to that of rat mesenteric venules (0.94, Kendall & Michel, 1995). Finally, for all graphs of flux we assumed that P_i was equal to an atmospheric pressure of zero.

Similarity of collecting lymphatic and venular permeability

The median collecting lymphatic (3.5 ± 1 × 10⁻⁷ cm s⁻¹) did not differ significantly from the of venules in this study (4.0 ± 1 × 10⁻⁷ cm s⁻¹) or venules from a previous study (5.6 ± 2 × 10⁻⁷ cm s⁻¹, Rumbaut & Huxley, 2002). Strikingly, even the frequency distributions of lymphatic and venular P_s did not differ (Fig. 2). No correlation was observed between and vessel diameter or time of year, ruling out seasonal variation of solute permeability in the rat (data not shown). However, as for venules, a correlation was found between collecting lymphatic and the perfusion pressure at the time of the flux measurement, a result of the coupling of solute flux to volume flux.

When the values are plotted against hydrostatic pressure, one can determine the true diffusive permeability (P_d) under the special condition of zero net filtration pressure. We found that collecting lymphatic P_d was 2.5 × 10⁻⁷ cm s⁻¹ after adjusting the graph for the effective osmotic pressure of our perfusate. While not very different from the P_d to bovine serum albumin (BSA) for frog mesenteric venules (2.3 ± 0.25 × 10⁻⁷ cm s⁻¹, Curry et al. 1990), our collecting lymphatic P_d must be compared to rat mesenteric venular P_d, for which no reported measures exist. The scarcity of mammalian P_d data underscores the need for future experiments to elucidate P_d to albumin in rat mesenteric microvessels to form a complete description of the network permeability, inclusive of all vessel types.

The Péclet number describes the ratio of convective to diffusive transport of solute. In these experiments albumin, a carrier of free fatty acids, hormones and drugs, was used because it is the major determinant of oncotic pressure in intact, autoperfused tissues. At the mean hydrostatic pressure for collecting lymphatics (7 cmH₂O, Table 1) the calculated Pé= 0.36, meaning that ∼40% of albumin transport was mediated by convection through pathways/structures through which water and solute travel together and ∼60% of albumin crosses by pressure-independent mechanisms. If the permeability properties of rat mesenteric venules and collecting lymphatics are similar, the venular Pé value would be ∼1 at their greater hydrostatic pressure of 20 cmH₂O. In contrast, Pé for isolated rat skeletal muscle venules at 20 cmH₂O is 0.13 (Sarelius et al. 2006). It is notable that the difference between these two values probably reflects the fact that of skeletal muscle venules is an order of magnitude greater than mesenteric venular , indicative of the difference in metabolic demand of the two tissues (Sarelius et al. 2006) or the difference in permeability to BSA relative to RSA as a result of differing charges (Rumbaut, 1998; Bingaman et al. 2003).

Others have reported P_s to BSA (14 ± 7 × 10⁻⁷ cm s⁻¹, Price et al. 2008) from a confluent tube of lymphatic endothelial cells (D=∼100 μm). Though this is from a vessel with a diameter similar to that of rat mesenteric collecting lymphatics, it probably represents microlymphatic vessel permeability due to the lack of basement membrane, smooth muscle and pericytes. In support of this notion, from a single microlymphatic vessel cannulated in vivo we measured a of 12 × 10⁻⁷ cm s⁻¹ (D= 37.8 μm, P_lumen= 3 cmH₂O). While cell culture is a valuable system for studying the molecular mechanisms regulating permeability, caution should be exercised when interpreting these P_s values.

Influence of pressure on collecting lymphatic spontaneous contractions

Figure 1 demonstrates the effect of hydrostatic pressure, and hence flow, on the spontaneous contraction frequency of collecting lymphatics. Studies show (Gashev et al. 2002) that with increased flow, there is a decrease in spontaneous contraction frequency mediated in part by nitric oxide. Others (Zhang et al. 2007) have demonstrated that in the absence of flow, elevation in pressure alone is sufficient to diminish contraction amplitude. Figure 1A shows that a collecting lymphatic vessel perfused at a relatively low pressure possesses spontaneous contractions compared to a vessel in Fig. 1B with a higher perfusion pressure lacking spontaneous contractions. Both mechanisms are probably at play. After cannulation, approximately half of the P_s recordings exhibited spontaneous contractions, but were absent in the other half. Since solute flux is affected by pressure as a result of convective coupling, the values between contracting and non-contracting vessels differed, although the direct effect of spontaneous contractions versus a contribution of flow cannot be extracted from these data. While some vessels lacked spontaneous contractions, this occurred invariably after cannulation and only at relatively high pressure and/or flow.

Assumptions for modelled solute and volume flux over pressure

With knowledge of L_p and P_d the graph of volume flux against pressure can be constructed once we know the reflection coefficient (σ) of the vessel and the oncotic pressure difference (Δπ=π_L−π_i). As a first approximation, given the similarity in solute permeability parameters P_s and P_d, we assumed that σ for albumin does not differ between the two vessel types. The value that has been measured for rat mesenteric venules is 0.94 (Kendall & Michel, 1995). The other parameter that must be known is the oncotic pressure difference across the lymphatic vessel wall (π_L−π_i), for which the reported values vary considerably. Therefore, the albumin concentrations of lymph, peritoneal fluid and plasma were measured directly (Table 2), facilitating the calculation of the oncotic pressure difference for albumin. Ideally, interstitial fluid should be sampled instead of peritoneal fluid, but the latter has been shown to be only modestly more concentrated than the former (Barber et al. 1990). For all graphs of flux, the interstitial pressure (P_i) is assumed to equal zero.

Estimation of collecting lymphatic vessel hydraulic conductivity (L_p)

Although these studies were aimed at determining permeability to solute, the question remains: what is the magnitude of the ‘hydraulic permeability coefficient’, L_p, and does that differ from rat mesenteric venular L_p? To address this, we examined the limiting slope of Fig. 3, which defines L_p(1 −σ), and assumed that collecting lymphatic σ was identical to the rat mesenteric venular σ of 0.94 (Kendall & Michel, 1995). The calculated L_p is 2.2 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹, similar to basal values measured in rat mesenteric venules (2.4 ± 0.2 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹ (Kendall & Michel, 1995); 2.6 ± 0.3 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹ (Kim et al. 2005); 2.4 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹ (Rumbaut et al. 2000)). To validate this estimation, experiments are needed to measure hydraulic conductivity for rat mesenteric collecting lymphatics, which will provide the remaining permeability values for both albumin (σ) and water (L_p).

Predicted solute and volume flux from measured albumin concentration difference

Table 2 provides evidence that lymph taken from a typical rat mesenteric collecting lymphatic can possess a greater concentration of albumin than that found in surrounding peritoneal fluid. Since interstitial fluid is slightly less concentrated than peritoneal fluid (Barber et al. 1990), we can confidently state that the albumin concentration in lymph is greater than the interstitial concentration, at least for vessels in rat mesentery. The total protein lymph-to-plasma ratio of 77% agrees well with previous studies reporting 70% in the rat tail (Aukland et al. 1984). More interestingly, the albumin-to-total protein ratio for each compartment differed, suggesting non-uniform albumin handling within each compartment.

Figure 4A shows that albumin flux (J_s) is outwardly directed since ΔC is a positive value, where the positive direction is from the vessel to the interstitium. Finding in vivo evidence supporting solute extravasation is intriguing since collecting lymphatics are believed to only absorb protein and fluid (Guyton, 1971; Ratnoff, 1983; Boron & Boulpaep, 2005). Figure 4B indicates an outwardly directed volume flux in both the transient (continuous line) and steady-state (dashed curve) conditions. The steady-state relationship shows that constitutive reabsorption of fluid by collecting lymphatics cannot occur, which confirms that proposed by Michel for microvessels (Michel & Phillips, 1987). However, collecting lymphatics of this study were observed to contract spontaneously at frequencies of 10–15 min⁻¹; consequently, a steady-state pressure is never attained. Thus, description of physiological collecting lymphatic volume flux is better represented by the transient graph of volume flux on pressure (Fig. 4B). While filtration of fluid occurs over most of the pressures observed in this study, at low pressures (<1.5 cmH₂O) reabsorption of fluid is possible.

Physiological significance of collecting lymphatic permeability properties

The measured values of albumin flux led us to accept our hypothesis that rat mesenteric collecting lymphatics are no different than venules with respect to albumin permeability, which supports the proposal that the cardinal vein is the origin of lymphatic endothelium (Srinivasan et al. 2007). Further, our data show that a positive concentration difference is possible (Table 2) such that solute moves from the vessel lumen towards the interstitium, a result that opposes the current expectations that lymphatic vessels should only absorb protein and fluid. While the concentration differences for lymphatic vessels of every organ may not be positive, these data demonstrate that the capacity for lymphatic solute exchange exists. A consequence of an outwardly directed and convectively coupled albumin flux is that during increased lymphatic hydrostatic pressure a greater amount of albumin is extravasated, increasing the interstitial albumin concentration and likelihood for oedema formation. Therefore, additional studies are needed to elucidate the regulation of lymphatic solute permeability in the context of oedema formation.

If the ideal graphs of volume flux represent the behaviour of most rat mesenteric collecting lymphatic vessels, they have important implications for oedema and overall fluid balance. For instance, collecting lymphatic hydrostatic pressure is elevated during lymphoedema and according to Fig. 4B enhances fluid filtration that, if sufficient, is expected to produce a low-protein oedema. Further, a higher hydrostatic pressure inhibits spontaneous contractions so that volume flux follows the steady-state curve (Fig. 4B) for which fluid reabsorption is no longer possible. Although the volume flux predictions are approximations, and it is probable that not all collecting lymphatic vessels possess the same oncotic pressure difference, the albumin concentration data in Table 2 show that it is likely that macromolecules in lymph of prenodal collecting lymphatics are more concentrated than in the surrounding interstitial fluid. Together our solute flux data and volume flux estimates support the theory that prenodal collecting lymphatics concentrate their luminal protein (Jacobsson & Kjellmer, 1964; Brace et al. 1977; Takahashi et al. 1997) by preferentially losing water over solute (depicted in Fig. 4C). Experiments to determine collecting lymphatic hydraulic conductivity are needed to fully delineate the conditions that produce oedemas of lymphatic origin.

Glossary

Abbreviations

BSA

bovine serum albumin

ΔC

transmural concentration difference

D

vessel diameter

J_s

solute flux

J_v

volume flux

L_p

hydraulic conductivity

ΔP

transmural pressure difference

P_c

microvessel lumen pressure

P_d

diffusive permeability

P_i

interstitial pressure

P_lumen

native vessel pressure

P_s or P^RSA_s

apparent solute permeability (to RSA)

Pé

Péclet number

RSA

rat serum albumin

S

surface area

Δπ

transmural oncotic pressure difference

π_c or π_L

microvessel or lymphatic oncotic pressure

π_i

interstitial oncotic pressure

σ

reflection coefficient

Author contributions

The idea of measuring a finite collecting lymphatic permeability was conceived by J.P.S., who designed the experiments and collected, analysed and interpreted the data. V.H.H. provided assistance and guidance throughout the project, especially with respect to the technique of cannulation and the concepts behind measuring microvascular permeability. V.H.H. & P.S. critically revised and approved the final version of the manuscript. Experiments were performed at the University of Missouri at Columbia in the Center for Diabetes and Cardiovascular Health, and the National Center for Gender Physiology.

Supplemental Methods

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors