Decisive structural determinants for the interaction of proline derivatives with the intestinal H+/peptide symporter


M. Brandsch, Biozentrum, Martin-Luther-University Halle-Wittenberg, Weinbergweg 22, D-06120 Halle, Germany. Fax: + 49 345552 7230, Tel.: + 49 345552 1630, E-mail:


To elucidate the decisive structural factors relevant for dipeptide–carrier interaction, the affinity of short amide and imide derivatives for the intestinal H+/peptide symporter (PEPT1) was investigated by measuring their ability to inhibit Gly-Sar transport in Caco-2 cells. Dipeptides with proline or alanine in the C-terminal position displayed affinity constants (Ki) of 0.15–1.2 mm and 0.08–9.5 mm, respectively. There was no clear relationship between hydrophobicity, size or ionization status of the N-terminal amino acid and the affinity of the dipeptides. However, analyzing the individual peptide bond conformations of Xaa-Pro dipeptides, a striking correlation between the cis/trans ratios (trans contents 24–70%) and the affinity constants was observed. After correcting the Ki values for the incompetent cis isomers, the Ki corr values of most dipeptides were in a small range of 0.1–0.16 mm. This result revealed the decisive role of peptide bond conformation even for a transport protein that is quite promiscuous in substrate translocation. When measuring affinity constants of Xaa-Pro and Xaa-Sar dipeptides, the cis/trans ratios cannot be ignored. Lower affinities of Lys-Pro, Arg-Pro and Pro-Pro indicate that additional molecular factors affect their binding at PEPT1. The Ki values obtained for the corresponding Xaa-Ala dipeptides support this conclusion.

Potential substrates or inhibitors of peptide transport were found among Xaa-piperidides and Xaa-thiazolidides. Dipeptides with N-terminal proline displayed a very diverse affinity profile. However, in contrast to current knowledge, several Pro-Xaa dipeptides such as Pro-Leu, Pro-Tyr and Pro-Pro are recognized by PEPT1 with appreciable affinities. Binding seems mainly determined by the hydrophobicity of the C-terminal amino acid and the rigidity of the structure.








pipecolic acid


high-performance capillary electrophoresis

The H+ gradient-dependent symport system PEPT1 actively transports di- and tripeptides and a variety of peptidomimetics across the brush border membrane of intestinal epithelial cells [1–4]. During 1998, various advances were made regarding the elucidation of structural requirements for substrates to be accepted by intestinal and renal peptide transporters. Temple et al. [5] reported that 4-aminophenylacetic acid, a peptide mimetic lacking a peptide bond, represents a substrate for the intestinal peptide transporter. We have shown recently that replacing the peptide bond in Ala-Pro by a thioxo peptide bond is tolerated by PEPT1 when in trans conformation [6], and that this carrier transports amino acid aryl amides with high affinity [7]. Daniel's group [8] discovered that δ-aminolevulinic acid, which is used as an endogenous photosensitizer for photodynamic tumor treatment, represents a high-affinity substrate for PEPT1 and PEPT2. The work of this group also expanded our understanding of the minimal structural requirements of PEPT1 substrates by showing electrogenic transport of ω-amino fatty acids [9]. Several groups have reported that the reason for the higher oral bioavailability of the prodrug valacyclovir compared to acyclovir is its transport via PEPT1 [10,11].

Structural features commonly believed to be crucial for substrate–carrier interaction are molecular size, type of binding, stereo-isomerism, spatial arrangement, hydrophobicity, polarity and charge. Studies on the structural requirements of dipeptides for their recognition by a protein, ideally, should be performed by analyzing the interaction of analogous series of dipeptides with that particular protein. We decided to investigate Xaa-Pro and Pro-Xaa dipeptides and to compare their affinity constants with those of the corresponding Xaa-Ala dipeptides. The role of proline in dipeptides in their transport by peptide carriers has not been studied systematically. Reviewing the available data reveals that at least some Xaa-Pro dipeptides are good substrates for both renal and intestinal type peptide transporters [12–15]. There is almost no information available regarding the effect of N-terminal proline on the interaction of dipeptides with PEPT1. Using brush-border membrane vesicles from rabbit kidney, Miyamoto et al. [12] and Daniel et al. [14] found very low or undetectable affinities of Pro-Gly and Pro-Leu to the renal peptide transporter(s). Similarly, in Caco-2 cells, an excess of Pro-Gly (> 20 mm) inhibited the uptake of Gly-Sar by only 59% [15] and that of cephalexin by only 67% [16]. Such results have led to the prevailing view that Pro-Xaa dipeptides are in general not particularly good substrates for peptide carriers. However, having measured a 10-fold higher affinity of Pro-Pro compared with Pro-Ala in preliminary experiments, we felt compelled to include a series of Pro-Xaa dipeptides in this study.

Materials and methods

Cell culture and uptake measurements

Caco-2 cells (passages 18–73) were obtained from the German Collection of Microorganisms and Cell Cultures and cultured as described previously [7,17,18]. Uptake of [glycine-1-14C]Gly-Sar (sarcosine) (53 mCi·mmol−1, Amersham International) was measured on the sixth to seventh day after cells reached confluence [7,17,18]. The uptake buffer (1 mL) contained 25 mm Mes/Tris (pH 6.0), 140 mm NaCl, 5.4 mm KCl, 1.8 mm CaCl2, 0.8 mm MgSO4, 5 mm glucose, 10 µm[14C]Gly-Sar and increasing concentrations (0–31.6 mm) of unlabeled compounds (Sigma-Aldrich Chemie or synthesized according to standard procedures [19] in peptide chemistry).

Data analysis

Results are given as means ± SE (n = 4). Nonlinear regression analysis, calculation of inhibition constants (Ki) from IC50 values and statistical analysis was carried out as described [7,17,18,20].

High-performance capillary electrophoresis (HPCE)

Capillary electrophoretic analyses of uptake buffer solutions (1 mm, pH 6.0) were performed on a P/ACE MDQ machine equipped with a diode array detector (Beckman Coulter). Separations were carried out using a fused silica capillary (Polymicro Technologies, inner diameter 50 µm, length 40 cm, background electrolyte 0.2 m sodium borate buffer, pH 8.5). Injection was performed hydrodynamically using a pressure of 3.45 kPa for 10 s. A voltage of 30 kV was applied. The instrument was adapted for low temperature measurements and operated in a refrigerated box to keep the capillary coolant temperature at −6 °C. At this temperature, the rate of interconversion in the capillary is slow enough not to interfere with the separation and the cis/trans ratios obtained in uptake buffer samples are ‘locked’. UV detection was carried out at 200 nm. Electropherograms were analyzed as described [6,21]. The stability of dipeptides in the extracellular medium during uptake was checked using conventional HPCE (BioFocus 3000, Bio-Rad) according to the laboratory standard.

Results and discussion

Affinity of Xaa-Pro and Xaa-Ala dipeptides for PEPT1

We first investigated the concentration-dependent inhibition of [14C]Gly-Sar uptake in Caco-2 cells by dipeptides with proline in C-terminal position (Fig. 1A,B). Gly-Sar is used as reference substrate for peptide transport because of its relatively high enzymatic stability [17]. All Xaa-Pro dipeptides tested were able to displace the labeled Gly-Sar at the carrier thereby inhibiting Gly-Sar uptake. The Ki values between 0.15 mm (Ala-Pro) and 1.2 mm (Pro-Pro) (Table 1) qualify these compounds as very good or good potential substrates for PEPT1. Pro-Pro showed the lowest affinity, probably because of the imino group in the N-terminal position (see below). The lower affinity of Trp-Pro (0.54 mm) compared with, for example, Ala-Pro was not due to a lower enzymatic stability of Trp-Pro during uptake. Using HPCE, we found no evidence for extracellular hydrolysis of Tyr-Pro, Trp-Pro, Ala-Pro or Phe-Pro.

Figure 1.

Inhibition of [14C]Gly-Sar uptake by Xaa-Pro dipeptides in Caco-2 cells (A and B). Uptake of 10 µm[14C]Gly-Sar was measured for 10 min in monolayer cultures of Caco-2 cells at pH 6.0 in the absence and presence of increasing concentrations of unlabeled dipeptides (0–3.16 mm). Uptake of Gly-Sar measured in the absence of the inhibitors [252.6 ± 19.7 pmol·(10 min−1)·(mg protein−1)] was designated as 100%.

Table 1. Inhibition constants of Xaa-Pro and Xaa-Ala and trans contents of Xaa-Pro dipeptides.Ki values ± SE were calculated from IC50 values derived by nonlinear regression analysis of data shown in Fig. 1 for the Xaa-Pro dipeptides. Cis/trans conformers of Xaa-Pro dipeptides in cis/trans equilibrium were separated in uptake buffer samples (pH 6.0) using subzero temperature HPCE as described. From electropherograms such as shown in Fig. 2, cis and trans contents were calculated. Ki corr values were calculated by multiplying Ki with trans content/100.
Xaa-Pro dipeptideKi (mm)Trans content (%)Ki corr (mm)Xaa-Ala dipeptideKi (mm)
H-Ala-Pro-OH0.15 ± 0.0264 ± 10.10 ± 0.01H-Ala-Ala-OH0.08 ± 0.01
H-Nle-Pro-OH0.16 ± 0.0164 ± 10.10 ± 0.01H-Nle-Ala-OH0.14 ± 0.01
H-Leu-Pro-OH0.18 ± 0.0160 ± 10.11 ± 0.01H-Leu-Ala-OH0.12 ± 0.01
H-Ser-Pro-OH0.19 ± 0.0266 ± 10.13 ± 0.01H-Ser-Ala-OH0.14 ± 0.01
H-Glu-Pro-OH0.26 ± 0.0155 ± 10.14 ± 0.01H-Glu-Ala-OH0.25 ± 0.02
H-Gly-Pro-OH0.30 ± 0.0252 ± 20.16 ± 0.01H-Gly-Ala-OH0.38 ± 0.02
H-Phe-Pro-OH0.39 ± 0.0334 ± 10.13 ± 0.01H-Phe-Ala-OH0.11 ± 0.02
H-Arg-Pro-OH0.39 ± 0.0570 ± 20.27 ± 0.04H-Arg-Ala-OH0.28 ± 0.01
H-Lys-Pro-OH0.41 ± 0.0864 ± 20.26 ± 0.05H-Lys-Ala-OH0.34 ± 0.02
H-Tyr-Pro-OH0.53 ± 0.0330 ± 10.16 ± 0.01H-Tyr-Ala-OH0.09 ± 0.01
H-Trp-Pro-OH0.54 ± 0.0824 ± 10.13 ± 0.02H-Trp-Ala-OH0.16 ± 0.02
H-Pro-Pro-OH1.2 ± 0.151 ± 20.61 ± 0.06H-Pro-Ala-OH9.5 ± 0.4

Under identical conditions, the concentration-dependent inhibition of [14C]Gly-Sar uptake by the corresponding dipeptides with alanine in C-terminal position was studied (Table 1). Ki values between 0.08 mm (Ala-Ala) and 9.5 mm (Pro-Ala) were obtained. With the remarkable exception of Pro-Ala, the affinity constants of Xaa-Ala dipeptides were found to be lower than those of the Xaa-Pro dipeptides.

Using these data, we performed structure–binding analyses. Surprisingly, there was no significant relationship between the affinity of the dipeptides and their physicochemical characteristics. For example, the conclusion that might be drawn from the Xaa-Pro series, that dipeptides with aromatic amino acids in N-terminal position are of lower affinity, is not supported by the results obtained with Xaa-Ala dipeptides. Similarly, attempts failed to correlate the affinity with hydrophobicity, flexibility, molecular size or ionization status. Moreover, there was no similarity of Xaa-Pro and Xaa-Ala dipeptides with regard to affinity.

Role of cis/trans conformation for the affinity of Xaa-Pro dipeptides

We then determined the ratios of cis/trans peptide bond conformers using HPCE (Fig. 2). The Xaa-Pro dipeptides existed as conformer mixtures with trans contents between 24% (Trp-Pro) and 70% (Arg-Pro) (Table 1). It is important to note that both the uptake experiments and the HPCE analyses were performed with cis/trans equilibrated uptake buffer solutions: Thunecke et al. [21] have shown that the cis/trans ratios of Xaa-Pro dipeptide conformers in aqueous ad hoc solutions are in many cases very different from the ratios in equilibrium. The interconversion towards equilibrium occurs at room temperature spontaneously within minutes until a certain dipeptide-specific cis/trans ratio is reached. The apparent affinities of Xaa-Pro dipeptides correlated well with the amount of trans conformers (Fig. 3). For example, Trp-Pro showed a relatively low affinity for PEPT1 but it also had the lowest trans content of all peptides tested. In contrast, Ala-Pro is characterized by high affinity and a high trans content (64%). Interestingly, Pro-Pro and the cationic dipeptides Lys-Pro and Arg-Pro revealed lower affinities not completely explained by their trans content. Omitting these three values from the linear regression increased the correlation coefficient drastically (r = 0.974, P < 0.0001). Because the clustering of data seen in Fig. 3 might cause overestimation of the parametric coefficient, the nonparametric Spearman's rank correlation coefficient was calculated, again without considering Pro-Pro, Lys-Pro and Arg-Pro. This more reliable calculation revealed a very high Spearman's rank correlation coefficient rs of 0.895 (P = 0.0011). Hence, different affinities of Xaa-Pro dipeptides can be explained simply in most cases by their different trans contents in aqueous solution. Having thereby established a relationship between the trans content and the apparent affinity of Xaa-Pro dipeptides, we corrected the Ki values by the respective cis/trans ratios (Table 1). After performing this procedure, the corrected Ki corr values of most Xaa-Pro dipeptides were in the small range of 0.1–0.16 mm. The carrier clearly accepts trans conformers of zwitterionic Xaa-Pro dipeptides regardless of size, hydrophobicity and aromaticity of the N-terminal amino acid. On the other hand, it now became even more evident that, for Pro-Pro, Lys-Pro and Arg-Pro, additional structural characteristics independent of peptide bond conformation apply. Interestingly, among the Xaa-Ala dipeptides, Lys-Ala and Arg-Ala also were of lower affinity. Comparison of the affinity constants of Lys-Pro and Lys-Ala with those of Nle-Pro and Nle-Ala, where the ε amino group is removed from lysine, showed that it is probably the positively charged side chain that reduced the affinity of Lys dipeptides threefold. It is possible that the carrier protein disfavours positively charged dipeptides because of electrostatic interaction of the side chain with a group close to the binding site of the carrier. Alternatively, the peptide transporter shows different affinities for different ionic species of a given electrically charged peptide substrate at a certain pH [14,17,22,23].

Figure 2.

HPCE-signals of trans and cis conformers of selected Xaa-Pro dipeptides. Aliquots of cis/trans equilibrated uptake buffer solutions (1 mm, pH 6.0) were analyzed by HPCE. The capillary coolant temperature was kept at −6 °C and a voltage of 30 kV was applied. UV detection was carried out at 200 nm.

Figure 3.

Correlation between the affinity constants of Xaa-Pro dipeptides and their trans contents in the uptake buffer. Data are taken from Table 2 (Pro-Pro, Lys-Pro and Arg-Pro excluded from regression). Dashed line, 99% confidence interval of linear regression.

After correcting the Xaa-Pro concentrations by the share of ineffective cis conformers, the Ki corr values were in the same range as it has been observed for the Xaa-Ala peptides (with the exception of Pro-Ala). Therefore, proline in the C-terminal position does not, per se, result in a lower affinity. Again, we performed linear regression analyses to quantify the indirect relation between the affinities of Xaa-Ala and Xaa-Pro dipeptides but using the Ki corr values of Xaa-Pro. This time the calculations yielded a high indirect correlation with a linear correlation coefficient r of 0.677 (P = 0.02, Pro-Xaa omitted). Spearman's nonparametric rank correlation coefficient rs was 0.720 (P = 0.0082, all 24 peptides included). Hence, it appears that the same factors determine both the affinity of Xaa-Pro and that of Xaa-Ala dipeptides for PEPT1.

Affinity and cis/trans ratios of Xaa-Sar dipeptides

Besides the Xaa-Pro dipeptides, Xaa-Sar dipeptides also exist as a mixture of cis and trans conformers. Therefore, sarcosine-containing dipeptides were included in this study (Table 2). Gly-Sar was found to exist in the uptake medium at 61% in trans conformation. Transformation of the Michaelis–Menten constant, Kt, of Gly-Sar of 0.83 ± 0.04 mm as described above for the Xaa-Pro dipeptides resulted in a Kt corr of 0.51 ± 0.03 mm. It is important to note that this correction is of no significant consequence for the calculation of Ki values of inhibitors from IC50 values using the formula established by Cheng and Prusoff [20] as long as the Gly-Sar concentrations used are very much lower than the Kt value of Gly-Sar.

Table 2. Inhibition constants of Xaa-Pro derivatives and analogues.Ki values ± SE were calculated from IC50 values derived from inhibition plots of the uptake of 10 µm[14C]Gly-Sar in Caco-2 cells in the presence of increasing concentrations (0–31.6 mm or concentration of maximal solubility) of competing derivatives.
DerivativeStructure            Ki uncorr (mm)
H-Ala-Pro-OHinline image0.15 ± 0.02
(trans content 64 ± 1%)
H-Ala-d-Pro-OHinline image15 ± 2
H-Ala-pyrrolidideinline image> 30
H-Phe-pyrrolidideinline image> 30
H-Ile-pyrrolidideinline image> 20
H-Ala-piperidideinline image13 ± 2
H-Ile-piperidideinline image5.6 ± 0.6
H-Ala-thiazolidideinline image11 ± 2
H-Ile-thiazolidideinline image6.3 ± 1.6
H-Ala-Pip-OHinline image0.06 ± 0.01
(trans content 58 ± 3%)
H-Gly-Sar-OHinline image0.83 ± 0.04
(trans content 61 ± 1%)
H-Sar-Sar-OHinline image> 30
H-Sar-Pro-OHinline image2.5 ± 0.1
(trans content 62 ± 2%)
H-Phe-Sar-OHinline image0.21 ± 0.01
(trans content 60 ± 2%)
H-Ala-Sar-OHinline image0.25 ± 0.01
(trans content 59 ± 1%)

Sar-Sar showed no affinity to PEPT1 (Table 2). Sar-Pro, however, inhibited the Gly-Sar transport with a Ki = 2.5 ± 0.1 mm (Ki corr = 1.6 ± 0.1 mm). Phe-Sar (Ki corr = 0.13 ± 0.01 mm) and Ala-Sar (Ki corr = 0.15 ± 0.1 mm) were competitors with threefold higher affinity to PEPT1 than Gly-Sar. These data are in good agreement with the results shown in Table 1. Among the Xaa-Ala dipeptides devoid of proline residues, Gly-Ala represents the compound with the lowest affinity measured the Ki being threefold to fivefold higher than that of Phe-Ala and Ala-Ala.

Affinity and conformation of Xaa-Pro analogues

In the course of these experiments, we also investigated several new Xaa-Pro analogues with respect to their interaction with PEPT1 (Table 2). Derivatives such as amino acid amides of the pyrrolidide and thiazolidide type are known peptidase inhibitors currently under investigation for possible clinical applications [24,25]. Xaa-pyrrolidides showed no affinity to PEPT1. Next, the sulfur-containing amino acid thiazolidides were tested. Surprisingly, in contrast to their pyrrolidide counterparts, Ala-thiazolidide and Ile-thiazolidide inhibited Gly-Sar uptake with apparent Ki values of 11 and 6 mm (Table 2). Similar Ki values were found for Ala-piperidide and Ile-piperidide, were the five membered rings of Xaa-pyrrolidides are extended by another CH2 group. This result raised the question as to whether Xaa-Pip dipeptides are recognized. Indeed, Ala-Pip apparently interacts with high affinity with PEPT1 (Ki = 0.06 ± 0.01 mm). Regarding the peptide bond conformation, Ala-Pip existed in uptake medium at 58% in trans conformation (Ki corr = 0.03 ± 0.005 mm). These results emphasize the importance of the C-terminal COOH group for a high affinity of PEPT1 for these compounds. However, it is interesting that for several decarboxylated derivatives, some interaction with the carrier can be found. Especially in these cases, it is important to note that competition with Gly-Sar at the binding site of the carrier does not necessarily mean that these compounds are indeed transported into the cell [9]. Their actual H+-dependent translocation and/or accumulation, therefore, still has to be proven.

Interaction of Pro-Xaa dipeptides with PEPT1

The effect of N-terminal proline on the interaction of dipeptides with the intestinal peptide transporter is currently unknown. A series of such peptides was included in this investigation (Fig. 4A,B). In general, the Pro-Xaa dipeptides displayed Ki values between 0.5 and > 20 mm; these values were much more diverse and of lower affinity for PEPT1 than Xaa-Pro dipeptides (Table 3). Whereas the interaction of the carrier protein with Pro-Ser, Pro-Glu and Pro-Gly is neglectable, Pro-Leu, Pro-Tyr, Pro-Pro and Pro-Phe are good inhibitors of Gly-Sar uptake. Hence, it is not the case that the intestinal peptide transporter does not generally interact with Pro-Xaa dipeptides. Binding of these dipeptides by PEPT1 seems largely to be determined by the hydrophobicity of the C-terminal amino acid. That further structural parameters play a role in Pro-Xaa recognition is shown by the eightfold higher affinity for Pro-Pro than for Pro-Ala (Table 1). It is tempting to speculate that this drastic difference is due to the much higher rigidity of the Pro-Pro molecule.

Figure 4.

Inhibition of [14C]Gly-Sar uptake by Pro-Xaa dipeptides in Caco-2 cells (A and B). Uptake of 10 µm[14C]Gly-Sar was measured for 10 min in monolayer cultures of Caco-2 cells at pH 6.0 in the absence and presence of increasing concentrations of unlabeled dipeptides (0–31.6 mm or concentration of maximal solubility). Uptake of Gly-Sar measured in the absence of the inhibitors [250.3 ± 17.1 pmol·(10 min−1)·(mg protein−1)] was taken as 100%.

Table 3. Inhibition constants of Pro-Xaa dipeptides.Ki values ± SE were calculated from IC50 values obtained by nonlinear regression analysis of data shown in Fig. 4.
Pro-Xaa dipeptideKi uncorr (mm)
H-Pro-Leu-OH0.47 ± 0.04
H-Pro-Tyr-OH0.73 ± 0.02
H-Pro-Pro-OH1.2 ± 0.1
(trans content 51 ± 2%
H-Pro-Phe-OH1.9 ± 0.1
H-Pro-Arg-OH2.5 ± 0.1
H-Pro-Lys-OH3.2 ± 0.6
H-Pro-Hpr-OH8.5 ± 0.5
H-Pro-Ala-OH9.5 ± 0.4
H-Pro-Asp-OH9.8 ± 0.8
H-Pro-Ser-OH14 ± 2
H-Pro-Glu-OH20 ± 1
H-Pro-Gly-OH22 ± 2

In summary, the results of the present study provide the following evidence: (a) the decisive factors for the interaction of Xaa-Pro dipeptides with the intestinal peptide carrier are the peptide bond conformation and, to a lesser extent, the state of ionization of the side chain (other molecular features such as size and hydrophobicity seem to be irrelevant) and therefore, when measuring concentration-related constants such as IC50,Ki and Kt values of Xaa-Pro dipeptides and derivatives, the cis/trans ratio must be considered; (b) the affinity constants of (trans) Xaa-Pro dipeptides are not significantly different from those of Xaa-Ala dipeptides; (c) potential substrates or inhibitors of peptide transport were found among Xaa-piperidides and -thiazolidides; and (d) for interaction of PEPT1 with Pro-Xaa dipeptides, hydrophobicity of the C-terminal amino acid and rigidity of the structure seem to be the main determinants.


This work was supported by Land Sachsen-Anhalt grant 2217 A/0085G and fellowship (to F. T.) and by the Fonds der Chemischen Industrie. We thank Christa Langer for excellent technical assistance, Dr Angela Stöckel-Maschek and Dr Jürgen Faust for providing the Xaa-thiazolidides and Dr Mike Schutkowski for inspiring discussions.


  1. Note: web page available at